- Email: [email protected]
Template evaporation method for controlling anatase nanocrystal size in ordered macroporous TiO2
Journal of Colloid and Interface Science 290 (2005) 201–207 www.elsevier.com/locate/jcis
Template evaporation method for controlling anatase nanocrystal size in ordered macroporous TiO2 Chiara Dionigi a,∗ , Gianluca Calestani b , Tiziano Ferraroni b , Giampiero Ruani a , Leonarda F. Liotta c , Andrea Migliori d , Petr Nozar a , Dimitros Palles a a Istituto per lo Studio dei Materiali Nanostrutturati—Sezione di Bologna, Consiglio Nazionale delle Ricerche, via P. Gobetti, 101, I-40129 Bologna, Italy b Dip. Chimica Generale ed Inorganica, Università di Parma, viale delle Scienze, I-43100 Parma, Italy c Istituto per la Microelettronica e Microsistemi—Sezione di Bologna, Consiglio Nazionale delle Ricerche, via P. Gobetti, 101, I-40129 Bologna, Italy d Istituto per lo Studio dei Materiali Nanostrutturati—Sezione di Palermo, Consiglio Nazionale delle Ricerche, via Ugo La Malfa, 153, 90146 Palermo, Italy
Received 15 June 2004; accepted 11 April 2005 Available online 1 June 2005
Abstract The importance of pure-phase titanium oxide materials as catalysts, sensors, and photonic band-gap materials has been growing steadily. Recently, more attention has been focused on nanostructured titanium oxide showing controlled and periodic porosity on a nanometric scale. The nanocrystal size control of porous nanostructured titanium oxide in an anatase form is a crucial step for the organic template method. Simple template removal by evaporation in an inert atmosphere is reported in this article and compared with the calcination technique usually reported in the literature. The proposed method allows the formation of a double-porous (macro and meso) anatase phase. We demonstrate that it highly preserves the macropore order into a titanium oxide material and induces narrowly dispersed mesopores by controlling the nanocrystal size that is kept around 6 nm. For the proposed method, polystyrene beads are particularly suitable as templates, being evaporated in the temperature range of anatase existence. The final high surface area makes the materials appealing for applications as photocatalysts or sensors. 2005 Elsevier Inc. All rights reserved. Keywords: Template evaporation; Titanium oxide nanocrystals; Periodic porosity; Polystyrene beads
1. Introduction The potentialities of nanostructured materials in several applications have attracted wide interest in a considerable number of research fields. In the past 5 year new methods for fabrication of nanostructured inorganic materials made of a periodic array of spherical metal oxide shells have appeared in the literature [1–12]. These methods are based on the use of a template principally made of crystalline nanobead structure with regular periodicity [13]. Removing the template after the infiltration of the nanostructured voids with a metal oxide pre* Corresponding author.
E-mail address: [email protected] (C. Dionigi). 0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.04.033
cursor results in a metal oxide inverse structure showing a periodic array of spherical macro pores. The size of such macropores is controlled by the diameter of the templating beads and they are ordered and fully interconnected in three dimensions [3–12]. The interconnection between the macropores is generated by the contact point of the original template beads and depends on their coordination number in the arrangement of the colloidal crystal. Since beads spontaneously aggregate in a close-packed structure where each bead is in contact with 12 others, 12 is the maximum number of pore interconnections allowed for a compact bead structure. During the template elimination mesoporosity is also produced in the shell of the spherical cavity by the partial sintering of oxide nanoparticles during the thermal process leading to template removal. The optimization and the complete control of the mesoporosity
202
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
and hence of the internal surface depend on the dimensions, shape, and sintering conditions of the metal oxide nanoparticles. In the present paper, we report a new method for nanocrystal size and mesopore size control in highly ordered macroporous titanium oxide. The proposed method is based on the elimination of the polymeric template by evaporation in inert atmosphere. Such evaporation does not overheat the sample and minimizes local explosions. Indeed, when the template is removed by calcinations local overheating can determine uncontrolled crystal growth and phase transitions [15]. In nitrogen atmosphere, polystyrene starts to evaporate at around 400 ◦ C [14]. At the same temperature a controlled sintering of anatase phase nanoparticles takes place. Consequently, the macropore shells result resistant and selfstanding on a long-range scale. The structure does not collapse keeping the order created by the template. Polystyrene bead aggregates seem to be the most suitable template for macroporous anatase. The final high surface area can make the resulting doubleporous anatase appealing for applications as photocatalysts or sensors.
2. Experimental details Monodisperse polystyrene (PS) beads were prepared by surfactant-free emulsion polymerization of styrene, as reported elsewhere [16]. For template fabrication, a PS bead suspension was centrifuged. A sample of 30 ml of the PS bead dispersion was poured into a polypropylene test tube and centrifuged at a rate of 350–1050g for 10 h. The sediment was dried at 70 ◦ C for 4 to 12 h in order to create necks between adjacent beads and, consequently, to enlarge the contact points between the neighboring beads to generate macropores interconnections. The shape, the diameter of beads, and the quality of the surface of aggregated samples were determined on samples self-assembled by centrifugation by scanning electron microscopy (SEM). The inverse titania structures of the template were prepared by modifying the procedures described in the literature [2–6] and nitrogen–oxygen atmosphere was used during template elimination. Samples of bare beads prepared by centrifugation were infiltrated with an organometallic precursor (titanium(IV) ethoxide, Aldrich) by immersing them in a solution of the organometallic precursor in ethanol (70– 100%) for at least 3 days. The surface of the infilled bead crystals was fast washed in pure ethanol and the samples were kept in humid air for at least 7 days to let the titanium ethoxide hydrolyze. Two different procedures were utilized for template removal. In the first one the sample was slowly heated (0.75 ◦ C/min) to 450 ◦ C in nitrogen atmosphere to cre-
ate inverse structures and kept at this temperature for 5 h, then cooled for 2 h and half to room temperature and finally heated again for 2 h and half to 450 ◦ C in oxygen atmosphere. The samples were kept in this condition for 4– 5 h. In the second procedure the sample was heated with the same thermal program (0.75 ◦ C/min) to 450 ◦ C in air and kept under the same conditions for 9–10 h. The samples prepared by the two procedures are indicated in the discussion as NO and A respectively; intermediate samples, resulting from the sole nitrogen treatment in the first procedure, are indicated by N. Elemental analysis was performed using a Carlo Erba EA 1108 automated analyzer. Specific surface area (SSA) and pore size distribution measurements were performed by using a Carlo Erba Sorptomatic 1900 Instrument. Surface areas were determined by physical adsorption–desorption of N2 at −196 ◦ C, using the BET equation [17]. Mesopore size distributions were obtained from the desorption branch of the isotherm curve by BJH method [18]. The volume of micropores was calculated from the t-plot analysis of the adsorption isotherm [19]. The samples were degassed at 250 ◦ C under 10 µTorr for 3 h before isotherm measurements. Thermogravimetric analysis (TGA) and differential scanning calorimetry analysis (DSC) were performed with Perkin–Elmer analyzers (TGA7 and DSC7, respectively) in both nitrogen and air. A temperature ramp of 10 ◦ C/min was used for TG and DSC analyses. X-ray diffraction (XRD) patterns were collected using CuKα radiation with a Thermo ARL X’tra powder diffractometer. The spectra were collected using a step size of 0.05◦ in 2θ and a counting time ranging from 2 to 5 s per step. The assignment of the crystalline phases was based on the ICSD data base (TiO2 , anatase No. 93098) [ICSD]. Raman measurements were performed in the back-scattering geometry, with a resolution of 2 cm−1 , using a BRUKER FT-Raman RFS100 and Nd-YAG laser as excitation source (λexc = 1064 nm). The morphology of the inverse structures of the titanium oxide samples were checked by scanning electron microscopy (SEM). SEM analyses of bead and titanium oxide structures were performed using a Philips XL30 electron microscope operating at 30 kV accelerating voltage. The surface of the samples was metallized by sputtering of a 3–5 nm thick layer of gold.
3. Results and discussion The reported method exploits PS evaporation in nitrogen atmosphere. For comparison, a thermal treatment in an oxidant atmosphere is also illustrated. In Fig. 1, DSC analyses in oxidizing conditions (air atmosphere—air) and inert conditions (nitrogen atmosphere—N2 ) of PS colloidal crystals impregnated with titanium ethoxide are reported.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
Fig. 1. DSC analyses of PS colloidal crystals impregnated with titanium ethoxide in nitrogen atmosphere (N2 ) and in air atmosphere (air).
Fig. 2. Thermogravimetric analysis of PS colloidal crystals impregnated with titanium ethoxide in nitrogen atmosphere (N2 ) and in air atmosphere (air).
The nitrogen curve shows an endothermic process in the temperature range 330–500 ◦ C (Fig. 1—N2 ). The correspondent thermogravimetric analysis (Fig. 2—N2 ) shows an initial smooth weight loss, reasonably the residual evaporating ethanol, the maximum weight loss appears in the temperature range 400–500 ◦ C. According to the literature, the active process in such conditions is the PS evaporation process [14]. In air atmosphere, the active process is exothermic in the range 230–500 ◦ C and can be divided into two steps (Fig. 1—air). The first step, 230–350 ◦ C, corresponds to a gentle weight loss, Fig. 2—air, and can be correlated with the burning of the residual titanium ethoxide. In the temperature range of 350–500 ◦ C, the second step occurs; the weight loss is steep and relates to the combustion of PS. The weight loss in air up to 600 ◦ C is 2–3% more then the weight loss in nitrogen. However, the nitrogen-treated sample shows a further weight loss, when successively treated in oxygen in the same temperature range. Elemental analyses performed on thermally treated samples, show that the carbon content varies notably, depending on the atmosphere used for the thermal treatment. For the
203
Fig. 3. Raman spectra of A and NO samples.
N sample, it results to be quite large (31% by moles), in accordance with their black color. After the successive oxygen treatment (NO sample) the carbon content decreases to 1.4%, in agreement with the TGA measurements. This value is only slightly larger that those found by elemental analysis of sample A (0.8% by moles). The presence of high carbon content in the N sample can be attributed mainly to the production of carbon residuals by thermal degradation of the organic template in secondary reactions that take place during the evaporation process. A similar origin can be excluded for the samples treated in oxidizing atmosphere, owing to the catalytic activity of titania in the carbon oxidation reaction. The adsorption capability of high-surface TiO2 samples towards the carbon–oxygen compounds [20] seem to be the most probable origin of the carbon detected by elemental analysis in NO and A samples. In any case, since carbon–oxygen compounds are usually present in the ambient, we are not able to discriminate between carbon impurity produced by the template elimination processes or by CO2 adsorption in uncontrolled atmosphere. If the difference in surface area and consequently the difference in reactivity toward carbon oxides is taken into account, the elemental composition of the final samples can be considered equivalent, in spite of the differences in the proposed template elimination processes. Phase composition and mean particle size was determined for all the samples by Raman spectroscopy and X-ray powder diffraction (XRPD). For what concerns the phase composition, owing to the remarkable difference among the characteristic peaks of anatase and rutile structures, Raman spectroscopy proved to be a rapid and reliable technique to characterize TiO2 samples. Raman spectra of A and NO samples are reported in Fig. 3. In all cases the results are consistent with those expected for almost pure anatase samples, in agreement with the XRPD patterns reported in Fig. 4. As can be seen, the anatase form of TiO2 is the unique crystal phase observed by XRPD in the black samples obtained by evaporation of the organic template in nitrogen atmosphere (N sample), indicating that the carbon impurities
204
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
Table 1 Physicochemical properties of titania samples Sample
BET surface area (m2 /g)
Average mesopore diametera (nm)
Total mesopore volumea (cm3 /g)
Micropore volumeb (cm3 /g)
N NO A
172 103 65
2.5 5.5 10.8
0.14 0.33 0.29
0.11 – –
a Determined by BJH method. b Determined from t -plot analysis.
Fig. 4. XRPD patterns of N, NO, and A samples.
are amorphous. No structural change is produced by further treatment in oxygen, if the sample is cooled down to room temperature before changing it to gas atmosphere (NO sample). The appearance of more crystalline features of anatase phase was observed in the sample A, directly calcined in air, as it is evident by the sharpness of the peaks (Fig. 4). It is worth saying that when pure oxygen is directly introduced at 450 ◦ C immediate and simultaneous combustion of the remaining polystyrene vapors and, consequentially, of the whole carbon residual can take place, producing a local overheating that results in the formation of the rutile phase [15]. Information on the average grain size was extracted by broadening both the XRPD and Raman spectra. Regarding XRPD, the average grain dimension D, calculated from the
line broadening of the most intense reflection by using the Scherrer equation [21], is of about 6.0, 6.5, and 14.0 nm for N, NO, and A samples, respectively. In the case of Raman spectroscopy on nanostructured systems, analysis of the characteristics of the observed Raman active modes such as full width at half maximum (FWHM) and energy, allows to determine the mean particle size if smaller than 20–25 nm in diameter—for larger sizes the Raman is no longer able to distinguish the results from a macroscopic crystal [22]. We use the energy and FWHM parameters of the anatase peaks around 630 and 140 cm−1 to determine the mean crystal sizes (D) of samples prepared in different atmospheres. By the peak fitting we calculated typical FWHM for sample A of 30.3 and 12.8 cm−1 , corresponding to crystal sizes D in the range 8.9 < D < 9.5 nm and typical FWHM for NO sample of 36.5 and 15.0 cm−1 related to a crystal size of 7.3 < D < 7.8 nm. The Raman results agree to a large extent with those obtained by XRPD. The fair discrepancy in the nanocrystal dimensions of A samples could be justified by considering
Fig. 5. N2 adsorption (filled) and desorption (unfilled) isotherms of N (2), NO (Q), and A (") samples. The inset is a magnification of the adsorption isotherms in the low-relative-pressure region, P /P0 0.4.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
205
Fig. 7. Detailed SEM image of a TiO2 inverse opal structure. Fig. 6. BJH pore size distribution curves of N, NO, and A samples, calculated from the desorption branch of the nitrogen isotherms.
the crystal size polydispersion that renders both calculation models not completely appropriate. However, in spite of this discrepancy, both techniques indicate that the NO sample is characterized by a smaller grain size. Physicochemical properties of titania samples, determined by nitrogen adsorption measurements, are listed in Table 1. In Fig. 5 the adsorption–desorption isotherms at different relative pressure (P /P0 = 0–1) are displayed. NO and A samples have a specific surface areas of 103 and 65 m2 /g, respectively, and behave as mesoporous solids exhibiting isotherms type IV with the H2 hysteresis loop, according to the IUPAC classification [23]. Totally different appears the solid texture of the N sample which has a surface area of 172 m2 /g and shows isotherm type I with H2 hysteresis loop, that is typical of some microporosity [22]. As indicated in the inset of the figure, in this case the adsorption isotherm rises rapidly at low relative pressure P /P0 , because of the strong interaction between micropores and nitrogen-adsorbed molecules. Once micropores are filled, the adsorption continues on the external surface. On the contrary, in the case of NO and A samples at low relative pressure the prevailing process is the formation of a monolayer of adsorbed molecules and the isotherms rise moderately in comparison with the N sample. Fig. 6 displays the BJH pore size distributions of the titania samples, calculated from the desorption branch of the nitrogen isotherms. The calculated average pore diameters are reported in Table 1. It appears that the two different procedure utilized for template removal induce different porosity in NO and A samples. Indeed, the NO sample has a pore volume of 0.33 cm3 /g and shows a very narrow pore size distribution, centered at 5.5 nm, according to the mild and controlled treatment in nitrogen/oxygen atmospheres. Conversely, the A sample was characterized by a larger pore size distribution centered at 10.8 nm and a lower mesopore volume, 0.29 cm3 /g. The invasive calcination in air appears responsible of surface area sintering and pores enlargement.
Interestingly, the intermediate N sample (resulting from nitrogen treatment only) shows a predominant microporosity (average pore size of 2.5 nm and a micropore volume of 0.11 cm3 /g; see Table 1). In this case, the high content of carbon present in the sample appears to influence the resulting morphological properties. It should be noted that N shows the highest surface area. This finding could be ascribed to both, the contribution to the surface area of amorphous carbon and/or the length of the thermal treatment (only 5 h at 450 ◦ C instead of 10 h at the same temperature) that favors the attainment of less sintered materials. Moreover, the covering or filling of the mesopores by carbon could be responsible of the observed weak mesoporosity (total volume of 0.14 cm3 /g; see Table 1). The morphological modifications occurring during the two different procedures for template removal were also studied by SEM characterizations. They nicely agree with the structural and morphological properties, so far discussed. SEM analysis showed for all the samples the formation of an inverse structure, where regular spherical cavities, delimited by thin TiO2 shells, are interconnected in three dimensions (Fig. 7). However differences in morphology between NO and A samples were systematically detected in lower magnification SEM images. An example is shown in Fig. 8. The surface of A sample fragments appears bumpy and full of craters, as it can be noted in both extended and detailed images (Fig. 8a). During the combustion of the template, rapid gas development likely generates local explosions that produce such craters. Owing to the presence of these extended irregular cavities, the material breaks in an irregular way; in the fragmentation process these defects constitute additional energy minima and the fracture do not propagate exclusively along a preferential lattice plane, as can be noted in Fig. 8b. On the contrary, the surface of NO samples appears smooth and relatively defect-free, so the original order of the template crystal seems to be preserved in the inverse structure over a long-range. The ordered structure of bead shells breaks preferentially along the most dense (111) plane of the cubic lattice, as it is shown in Fig. 8c. The higher de-
206
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
4. Summary Template removal is an easy method to obtain double porous anatase materials with a low amount of impurity. In this paper we propose template evaporation in inert atmosphere as a new method for template removal. The evaporation process in nitrogen atmosphere is followed by a short treatment in oxygen atmosphere to eliminate the carbon residual. The results demonstrate that the evaporation process is a less invasive treatment in comparison with the reported calcinations in air. Because of the absence of heat evolution during the template removal, local explosions, and domain fragmentation are avoided by the evaporation process that is able to conserve at maximum the original bead order in the inverse structure. The inverse structure obtained in this way shows on a long-range scale regular, well-ordered, intact, and interconnected macropores that result in intense-colored light reflection. When the evaporation process is applied, polystyrene is particularly suitable as template material, being evaporated contemporary to the sintering of TiO2 nanoparticles. A fundamental advantage of the method we propose, when applications of the inverse structures as photocatalysts or sensors are concerned, is the possibility of a better grain size control and, consequently, of a narrow mesopore size dispersion. Typical particle dimensions ranging around 6 nm are obtained by the two-step template removal and this, even in the presence of the sintering process leading to the formation of the inverse structure, allows the final surface area to be kept above 100 m2 /g.
Acknowledgments We thank Professor Pietro Moggi and Dr. Sonia Morselli for their help in preliminary measurements of surface area at a single point.
References [1] [2] [3] [4] [5] Fig. 8. SEM analysis of the surface of (a, b) A sample and (c) NO sample.
gree of order observed by SEM in NO samples agrees with the intense colored light reflection observed systematically in these samples. The template and the impregnation process being the same for both A and NO samples, the differences observed can be attributed only to template removal.
[6] [7] [8] [9] [10] [11] [12] [13]
J. Wijnhoven, W. Vos, Science 281 (1998) 802. B. Gates, D. Quin, Y. Xia, Adv. Mater. 11 (1999) 466. S.H. Park, B. Gates, Y. Xia, Adv. Mater. 11 (1999) 462. A. Stein, Micropor. Mesopor. Mater. 44 (2001) 227–239. B.T. Holland, C.F. Blanford, T. Do, A. Stein, Chem. Mater. 11 (1999) 795–805. D. Wang, R.A. Caruso, F. Caruso, Chem. Mater. 13 (2001) 364–371. Q.-B. Meng, C.-H. Fu, Y. Einaga, Z.-Z. Gu, A. Fujishima, O. Sato, Chem. Mater. 14 (2002) 83–88. E. Yablonovich, Phys. Rev. Lett. 58 (1987) 2059. B. Gates, D. Quin, Y. Xia, Adv. Mater. 11 (1999) 466. S.-Y. Lin, E. Chow, V. Hietala, P.R. Villeneuve, J.D. Joannopoulos, Science 282 (1998) 274. S. Noda, A. Chutinan, M. Imada, Nature 407 (2000) 608. S.H. Park, B. Gates, Y. Xia, Adv. Mater. 11 (1999) 462. P. Nozar, D. Di Domenico, C. Dionigi, M. Losurdo, M. Muccini, C. Taliani, Adv. Mater. 14 (2002) 1023–1027.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
[14] K. Othmer, Encyclopedia of Chemical Technology, vol. 22, fourth ed., Wiley–Interscience, New York, 2000; H. Strohm, P. Löbmann, J. Mater. Chem. 14 (2004) 138–140. [15] T. Ferraroni, Laurea thesis, University of Parma, Italy, 2002. [16] C. Dionigi, P. Nozar, D. Di Domenico, G. Calestani, J. Colloid Interface Sci. 275 (2004) 445–449. [17] S. Brunauer, P.H. Emmet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [18] E.P. Barret, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951) 373.
207
[19] B.C. Lippens, J.H. DeBoer, J. Catal. 4 (1965) 319. [20] U. Diebold, Surf. Sci. Rep. 48 (2003) 53. [21] H.P. Klug, L.E. Alexander, X-Ray Diffraction Procedures for Polycrystalline and Amorphous Materials, Wiley, New York, 1954. [22] S. Kelly, F.H. Pollak, M.J. Tomkiewicz, Phys. Chem. B 101 (1997) 2730. [23] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 57 (1985) 603.
Template evaporation method for controlling anatase nanocrystal size in ordered macroporous TiO2 Chiara Dionigi a,∗ , Gianluca Calestani b , Tiziano Ferraroni b , Giampiero Ruani a , Leonarda F. Liotta c , Andrea Migliori d , Petr Nozar a , Dimitros Palles a a Istituto per lo Studio dei Materiali Nanostrutturati—Sezione di Bologna, Consiglio Nazionale delle Ricerche, via P. Gobetti, 101, I-40129 Bologna, Italy b Dip. Chimica Generale ed Inorganica, Università di Parma, viale delle Scienze, I-43100 Parma, Italy c Istituto per la Microelettronica e Microsistemi—Sezione di Bologna, Consiglio Nazionale delle Ricerche, via P. Gobetti, 101, I-40129 Bologna, Italy d Istituto per lo Studio dei Materiali Nanostrutturati—Sezione di Palermo, Consiglio Nazionale delle Ricerche, via Ugo La Malfa, 153, 90146 Palermo, Italy
Received 15 June 2004; accepted 11 April 2005 Available online 1 June 2005
Abstract The importance of pure-phase titanium oxide materials as catalysts, sensors, and photonic band-gap materials has been growing steadily. Recently, more attention has been focused on nanostructured titanium oxide showing controlled and periodic porosity on a nanometric scale. The nanocrystal size control of porous nanostructured titanium oxide in an anatase form is a crucial step for the organic template method. Simple template removal by evaporation in an inert atmosphere is reported in this article and compared with the calcination technique usually reported in the literature. The proposed method allows the formation of a double-porous (macro and meso) anatase phase. We demonstrate that it highly preserves the macropore order into a titanium oxide material and induces narrowly dispersed mesopores by controlling the nanocrystal size that is kept around 6 nm. For the proposed method, polystyrene beads are particularly suitable as templates, being evaporated in the temperature range of anatase existence. The final high surface area makes the materials appealing for applications as photocatalysts or sensors. 2005 Elsevier Inc. All rights reserved. Keywords: Template evaporation; Titanium oxide nanocrystals; Periodic porosity; Polystyrene beads
1. Introduction The potentialities of nanostructured materials in several applications have attracted wide interest in a considerable number of research fields. In the past 5 year new methods for fabrication of nanostructured inorganic materials made of a periodic array of spherical metal oxide shells have appeared in the literature [1–12]. These methods are based on the use of a template principally made of crystalline nanobead structure with regular periodicity [13]. Removing the template after the infiltration of the nanostructured voids with a metal oxide pre* Corresponding author.
E-mail address: [email protected] (C. Dionigi). 0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2005.04.033
cursor results in a metal oxide inverse structure showing a periodic array of spherical macro pores. The size of such macropores is controlled by the diameter of the templating beads and they are ordered and fully interconnected in three dimensions [3–12]. The interconnection between the macropores is generated by the contact point of the original template beads and depends on their coordination number in the arrangement of the colloidal crystal. Since beads spontaneously aggregate in a close-packed structure where each bead is in contact with 12 others, 12 is the maximum number of pore interconnections allowed for a compact bead structure. During the template elimination mesoporosity is also produced in the shell of the spherical cavity by the partial sintering of oxide nanoparticles during the thermal process leading to template removal. The optimization and the complete control of the mesoporosity
202
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
and hence of the internal surface depend on the dimensions, shape, and sintering conditions of the metal oxide nanoparticles. In the present paper, we report a new method for nanocrystal size and mesopore size control in highly ordered macroporous titanium oxide. The proposed method is based on the elimination of the polymeric template by evaporation in inert atmosphere. Such evaporation does not overheat the sample and minimizes local explosions. Indeed, when the template is removed by calcinations local overheating can determine uncontrolled crystal growth and phase transitions [15]. In nitrogen atmosphere, polystyrene starts to evaporate at around 400 ◦ C [14]. At the same temperature a controlled sintering of anatase phase nanoparticles takes place. Consequently, the macropore shells result resistant and selfstanding on a long-range scale. The structure does not collapse keeping the order created by the template. Polystyrene bead aggregates seem to be the most suitable template for macroporous anatase. The final high surface area can make the resulting doubleporous anatase appealing for applications as photocatalysts or sensors.
2. Experimental details Monodisperse polystyrene (PS) beads were prepared by surfactant-free emulsion polymerization of styrene, as reported elsewhere [16]. For template fabrication, a PS bead suspension was centrifuged. A sample of 30 ml of the PS bead dispersion was poured into a polypropylene test tube and centrifuged at a rate of 350–1050g for 10 h. The sediment was dried at 70 ◦ C for 4 to 12 h in order to create necks between adjacent beads and, consequently, to enlarge the contact points between the neighboring beads to generate macropores interconnections. The shape, the diameter of beads, and the quality of the surface of aggregated samples were determined on samples self-assembled by centrifugation by scanning electron microscopy (SEM). The inverse titania structures of the template were prepared by modifying the procedures described in the literature [2–6] and nitrogen–oxygen atmosphere was used during template elimination. Samples of bare beads prepared by centrifugation were infiltrated with an organometallic precursor (titanium(IV) ethoxide, Aldrich) by immersing them in a solution of the organometallic precursor in ethanol (70– 100%) for at least 3 days. The surface of the infilled bead crystals was fast washed in pure ethanol and the samples were kept in humid air for at least 7 days to let the titanium ethoxide hydrolyze. Two different procedures were utilized for template removal. In the first one the sample was slowly heated (0.75 ◦ C/min) to 450 ◦ C in nitrogen atmosphere to cre-
ate inverse structures and kept at this temperature for 5 h, then cooled for 2 h and half to room temperature and finally heated again for 2 h and half to 450 ◦ C in oxygen atmosphere. The samples were kept in this condition for 4– 5 h. In the second procedure the sample was heated with the same thermal program (0.75 ◦ C/min) to 450 ◦ C in air and kept under the same conditions for 9–10 h. The samples prepared by the two procedures are indicated in the discussion as NO and A respectively; intermediate samples, resulting from the sole nitrogen treatment in the first procedure, are indicated by N. Elemental analysis was performed using a Carlo Erba EA 1108 automated analyzer. Specific surface area (SSA) and pore size distribution measurements were performed by using a Carlo Erba Sorptomatic 1900 Instrument. Surface areas were determined by physical adsorption–desorption of N2 at −196 ◦ C, using the BET equation [17]. Mesopore size distributions were obtained from the desorption branch of the isotherm curve by BJH method [18]. The volume of micropores was calculated from the t-plot analysis of the adsorption isotherm [19]. The samples were degassed at 250 ◦ C under 10 µTorr for 3 h before isotherm measurements. Thermogravimetric analysis (TGA) and differential scanning calorimetry analysis (DSC) were performed with Perkin–Elmer analyzers (TGA7 and DSC7, respectively) in both nitrogen and air. A temperature ramp of 10 ◦ C/min was used for TG and DSC analyses. X-ray diffraction (XRD) patterns were collected using CuKα radiation with a Thermo ARL X’tra powder diffractometer. The spectra were collected using a step size of 0.05◦ in 2θ and a counting time ranging from 2 to 5 s per step. The assignment of the crystalline phases was based on the ICSD data base (TiO2 , anatase No. 93098) [ICSD]. Raman measurements were performed in the back-scattering geometry, with a resolution of 2 cm−1 , using a BRUKER FT-Raman RFS100 and Nd-YAG laser as excitation source (λexc = 1064 nm). The morphology of the inverse structures of the titanium oxide samples were checked by scanning electron microscopy (SEM). SEM analyses of bead and titanium oxide structures were performed using a Philips XL30 electron microscope operating at 30 kV accelerating voltage. The surface of the samples was metallized by sputtering of a 3–5 nm thick layer of gold.
3. Results and discussion The reported method exploits PS evaporation in nitrogen atmosphere. For comparison, a thermal treatment in an oxidant atmosphere is also illustrated. In Fig. 1, DSC analyses in oxidizing conditions (air atmosphere—air) and inert conditions (nitrogen atmosphere—N2 ) of PS colloidal crystals impregnated with titanium ethoxide are reported.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
Fig. 1. DSC analyses of PS colloidal crystals impregnated with titanium ethoxide in nitrogen atmosphere (N2 ) and in air atmosphere (air).
Fig. 2. Thermogravimetric analysis of PS colloidal crystals impregnated with titanium ethoxide in nitrogen atmosphere (N2 ) and in air atmosphere (air).
The nitrogen curve shows an endothermic process in the temperature range 330–500 ◦ C (Fig. 1—N2 ). The correspondent thermogravimetric analysis (Fig. 2—N2 ) shows an initial smooth weight loss, reasonably the residual evaporating ethanol, the maximum weight loss appears in the temperature range 400–500 ◦ C. According to the literature, the active process in such conditions is the PS evaporation process [14]. In air atmosphere, the active process is exothermic in the range 230–500 ◦ C and can be divided into two steps (Fig. 1—air). The first step, 230–350 ◦ C, corresponds to a gentle weight loss, Fig. 2—air, and can be correlated with the burning of the residual titanium ethoxide. In the temperature range of 350–500 ◦ C, the second step occurs; the weight loss is steep and relates to the combustion of PS. The weight loss in air up to 600 ◦ C is 2–3% more then the weight loss in nitrogen. However, the nitrogen-treated sample shows a further weight loss, when successively treated in oxygen in the same temperature range. Elemental analyses performed on thermally treated samples, show that the carbon content varies notably, depending on the atmosphere used for the thermal treatment. For the
203
Fig. 3. Raman spectra of A and NO samples.
N sample, it results to be quite large (31% by moles), in accordance with their black color. After the successive oxygen treatment (NO sample) the carbon content decreases to 1.4%, in agreement with the TGA measurements. This value is only slightly larger that those found by elemental analysis of sample A (0.8% by moles). The presence of high carbon content in the N sample can be attributed mainly to the production of carbon residuals by thermal degradation of the organic template in secondary reactions that take place during the evaporation process. A similar origin can be excluded for the samples treated in oxidizing atmosphere, owing to the catalytic activity of titania in the carbon oxidation reaction. The adsorption capability of high-surface TiO2 samples towards the carbon–oxygen compounds [20] seem to be the most probable origin of the carbon detected by elemental analysis in NO and A samples. In any case, since carbon–oxygen compounds are usually present in the ambient, we are not able to discriminate between carbon impurity produced by the template elimination processes or by CO2 adsorption in uncontrolled atmosphere. If the difference in surface area and consequently the difference in reactivity toward carbon oxides is taken into account, the elemental composition of the final samples can be considered equivalent, in spite of the differences in the proposed template elimination processes. Phase composition and mean particle size was determined for all the samples by Raman spectroscopy and X-ray powder diffraction (XRPD). For what concerns the phase composition, owing to the remarkable difference among the characteristic peaks of anatase and rutile structures, Raman spectroscopy proved to be a rapid and reliable technique to characterize TiO2 samples. Raman spectra of A and NO samples are reported in Fig. 3. In all cases the results are consistent with those expected for almost pure anatase samples, in agreement with the XRPD patterns reported in Fig. 4. As can be seen, the anatase form of TiO2 is the unique crystal phase observed by XRPD in the black samples obtained by evaporation of the organic template in nitrogen atmosphere (N sample), indicating that the carbon impurities
204
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
Table 1 Physicochemical properties of titania samples Sample
BET surface area (m2 /g)
Average mesopore diametera (nm)
Total mesopore volumea (cm3 /g)
Micropore volumeb (cm3 /g)
N NO A
172 103 65
2.5 5.5 10.8
0.14 0.33 0.29
0.11 – –
a Determined by BJH method. b Determined from t -plot analysis.
Fig. 4. XRPD patterns of N, NO, and A samples.
are amorphous. No structural change is produced by further treatment in oxygen, if the sample is cooled down to room temperature before changing it to gas atmosphere (NO sample). The appearance of more crystalline features of anatase phase was observed in the sample A, directly calcined in air, as it is evident by the sharpness of the peaks (Fig. 4). It is worth saying that when pure oxygen is directly introduced at 450 ◦ C immediate and simultaneous combustion of the remaining polystyrene vapors and, consequentially, of the whole carbon residual can take place, producing a local overheating that results in the formation of the rutile phase [15]. Information on the average grain size was extracted by broadening both the XRPD and Raman spectra. Regarding XRPD, the average grain dimension D, calculated from the
line broadening of the most intense reflection by using the Scherrer equation [21], is of about 6.0, 6.5, and 14.0 nm for N, NO, and A samples, respectively. In the case of Raman spectroscopy on nanostructured systems, analysis of the characteristics of the observed Raman active modes such as full width at half maximum (FWHM) and energy, allows to determine the mean particle size if smaller than 20–25 nm in diameter—for larger sizes the Raman is no longer able to distinguish the results from a macroscopic crystal [22]. We use the energy and FWHM parameters of the anatase peaks around 630 and 140 cm−1 to determine the mean crystal sizes (D) of samples prepared in different atmospheres. By the peak fitting we calculated typical FWHM for sample A of 30.3 and 12.8 cm−1 , corresponding to crystal sizes D in the range 8.9 < D < 9.5 nm and typical FWHM for NO sample of 36.5 and 15.0 cm−1 related to a crystal size of 7.3 < D < 7.8 nm. The Raman results agree to a large extent with those obtained by XRPD. The fair discrepancy in the nanocrystal dimensions of A samples could be justified by considering
Fig. 5. N2 adsorption (filled) and desorption (unfilled) isotherms of N (2), NO (Q), and A (") samples. The inset is a magnification of the adsorption isotherms in the low-relative-pressure region, P /P0 0.4.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
205
Fig. 7. Detailed SEM image of a TiO2 inverse opal structure. Fig. 6. BJH pore size distribution curves of N, NO, and A samples, calculated from the desorption branch of the nitrogen isotherms.
the crystal size polydispersion that renders both calculation models not completely appropriate. However, in spite of this discrepancy, both techniques indicate that the NO sample is characterized by a smaller grain size. Physicochemical properties of titania samples, determined by nitrogen adsorption measurements, are listed in Table 1. In Fig. 5 the adsorption–desorption isotherms at different relative pressure (P /P0 = 0–1) are displayed. NO and A samples have a specific surface areas of 103 and 65 m2 /g, respectively, and behave as mesoporous solids exhibiting isotherms type IV with the H2 hysteresis loop, according to the IUPAC classification [23]. Totally different appears the solid texture of the N sample which has a surface area of 172 m2 /g and shows isotherm type I with H2 hysteresis loop, that is typical of some microporosity [22]. As indicated in the inset of the figure, in this case the adsorption isotherm rises rapidly at low relative pressure P /P0 , because of the strong interaction between micropores and nitrogen-adsorbed molecules. Once micropores are filled, the adsorption continues on the external surface. On the contrary, in the case of NO and A samples at low relative pressure the prevailing process is the formation of a monolayer of adsorbed molecules and the isotherms rise moderately in comparison with the N sample. Fig. 6 displays the BJH pore size distributions of the titania samples, calculated from the desorption branch of the nitrogen isotherms. The calculated average pore diameters are reported in Table 1. It appears that the two different procedure utilized for template removal induce different porosity in NO and A samples. Indeed, the NO sample has a pore volume of 0.33 cm3 /g and shows a very narrow pore size distribution, centered at 5.5 nm, according to the mild and controlled treatment in nitrogen/oxygen atmospheres. Conversely, the A sample was characterized by a larger pore size distribution centered at 10.8 nm and a lower mesopore volume, 0.29 cm3 /g. The invasive calcination in air appears responsible of surface area sintering and pores enlargement.
Interestingly, the intermediate N sample (resulting from nitrogen treatment only) shows a predominant microporosity (average pore size of 2.5 nm and a micropore volume of 0.11 cm3 /g; see Table 1). In this case, the high content of carbon present in the sample appears to influence the resulting morphological properties. It should be noted that N shows the highest surface area. This finding could be ascribed to both, the contribution to the surface area of amorphous carbon and/or the length of the thermal treatment (only 5 h at 450 ◦ C instead of 10 h at the same temperature) that favors the attainment of less sintered materials. Moreover, the covering or filling of the mesopores by carbon could be responsible of the observed weak mesoporosity (total volume of 0.14 cm3 /g; see Table 1). The morphological modifications occurring during the two different procedures for template removal were also studied by SEM characterizations. They nicely agree with the structural and morphological properties, so far discussed. SEM analysis showed for all the samples the formation of an inverse structure, where regular spherical cavities, delimited by thin TiO2 shells, are interconnected in three dimensions (Fig. 7). However differences in morphology between NO and A samples were systematically detected in lower magnification SEM images. An example is shown in Fig. 8. The surface of A sample fragments appears bumpy and full of craters, as it can be noted in both extended and detailed images (Fig. 8a). During the combustion of the template, rapid gas development likely generates local explosions that produce such craters. Owing to the presence of these extended irregular cavities, the material breaks in an irregular way; in the fragmentation process these defects constitute additional energy minima and the fracture do not propagate exclusively along a preferential lattice plane, as can be noted in Fig. 8b. On the contrary, the surface of NO samples appears smooth and relatively defect-free, so the original order of the template crystal seems to be preserved in the inverse structure over a long-range. The ordered structure of bead shells breaks preferentially along the most dense (111) plane of the cubic lattice, as it is shown in Fig. 8c. The higher de-
206
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
4. Summary Template removal is an easy method to obtain double porous anatase materials with a low amount of impurity. In this paper we propose template evaporation in inert atmosphere as a new method for template removal. The evaporation process in nitrogen atmosphere is followed by a short treatment in oxygen atmosphere to eliminate the carbon residual. The results demonstrate that the evaporation process is a less invasive treatment in comparison with the reported calcinations in air. Because of the absence of heat evolution during the template removal, local explosions, and domain fragmentation are avoided by the evaporation process that is able to conserve at maximum the original bead order in the inverse structure. The inverse structure obtained in this way shows on a long-range scale regular, well-ordered, intact, and interconnected macropores that result in intense-colored light reflection. When the evaporation process is applied, polystyrene is particularly suitable as template material, being evaporated contemporary to the sintering of TiO2 nanoparticles. A fundamental advantage of the method we propose, when applications of the inverse structures as photocatalysts or sensors are concerned, is the possibility of a better grain size control and, consequently, of a narrow mesopore size dispersion. Typical particle dimensions ranging around 6 nm are obtained by the two-step template removal and this, even in the presence of the sintering process leading to the formation of the inverse structure, allows the final surface area to be kept above 100 m2 /g.
Acknowledgments We thank Professor Pietro Moggi and Dr. Sonia Morselli for their help in preliminary measurements of surface area at a single point.
References [1] [2] [3] [4] [5] Fig. 8. SEM analysis of the surface of (a, b) A sample and (c) NO sample.
gree of order observed by SEM in NO samples agrees with the intense colored light reflection observed systematically in these samples. The template and the impregnation process being the same for both A and NO samples, the differences observed can be attributed only to template removal.
[6] [7] [8] [9] [10] [11] [12] [13]
J. Wijnhoven, W. Vos, Science 281 (1998) 802. B. Gates, D. Quin, Y. Xia, Adv. Mater. 11 (1999) 466. S.H. Park, B. Gates, Y. Xia, Adv. Mater. 11 (1999) 462. A. Stein, Micropor. Mesopor. Mater. 44 (2001) 227–239. B.T. Holland, C.F. Blanford, T. Do, A. Stein, Chem. Mater. 11 (1999) 795–805. D. Wang, R.A. Caruso, F. Caruso, Chem. Mater. 13 (2001) 364–371. Q.-B. Meng, C.-H. Fu, Y. Einaga, Z.-Z. Gu, A. Fujishima, O. Sato, Chem. Mater. 14 (2002) 83–88. E. Yablonovich, Phys. Rev. Lett. 58 (1987) 2059. B. Gates, D. Quin, Y. Xia, Adv. Mater. 11 (1999) 466. S.-Y. Lin, E. Chow, V. Hietala, P.R. Villeneuve, J.D. Joannopoulos, Science 282 (1998) 274. S. Noda, A. Chutinan, M. Imada, Nature 407 (2000) 608. S.H. Park, B. Gates, Y. Xia, Adv. Mater. 11 (1999) 462. P. Nozar, D. Di Domenico, C. Dionigi, M. Losurdo, M. Muccini, C. Taliani, Adv. Mater. 14 (2002) 1023–1027.
C. Dionigi et al. / Journal of Colloid and Interface Science 290 (2005) 201–207
[14] K. Othmer, Encyclopedia of Chemical Technology, vol. 22, fourth ed., Wiley–Interscience, New York, 2000; H. Strohm, P. Löbmann, J. Mater. Chem. 14 (2004) 138–140. [15] T. Ferraroni, Laurea thesis, University of Parma, Italy, 2002. [16] C. Dionigi, P. Nozar, D. Di Domenico, G. Calestani, J. Colloid Interface Sci. 275 (2004) 445–449. [17] S. Brunauer, P.H. Emmet, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [18] E.P. Barret, L.G. Joyner, P.P. Halenda, J. Am. Chem. Soc. 73 (1951) 373.
207
[19] B.C. Lippens, J.H. DeBoer, J. Catal. 4 (1965) 319. [20] U. Diebold, Surf. Sci. Rep. 48 (2003) 53. [21] H.P. Klug, L.E. Alexander, X-Ray Diffraction Procedures for Polycrystalline and Amorphous Materials, Wiley, New York, 1954. [22] S. Kelly, F.H. Pollak, M.J. Tomkiewicz, Phys. Chem. B 101 (1997) 2730. [23] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 57 (1985) 603.