Synthesis and Characterization of Nano-Aerogels

6.4. Conclusion

The zirconia nanomaterials were synthesized for the first time by a one-step sol-gel route in CO₂ using zirconium alkoxides and acetic acid. Both mesoporous monolith and nanoparticles were produced. The as-prepared materials exhibited a high surface area (up to 399 m²/g) and porosity, while the calcined materials demonstrated tetragonal and/or monoclinic nanocrystallites. The mesoporous monoliths exhibited H2 hysteresis loops from N₂ physisorption, while the nanospheres exhibited H3 loops. DSC analysis showed that the nanoparticles gave a very sharp exothermal peak. In situ FTIR was found to be a valuable technique for studying the morphology evolution of the polycondensation reaction. The success in synthesizing ZrO₂ nanostructures within the present study, along with our former research on SiO₂ and TiO₂, shows the promise of using CO₂ for producing metal oxide nanomaterials. The low surface tension during scCO₂ drying allows the formation of mesoporous oxides that have potential application as porous ceramics for SOFC.

Chapter 7. Kinetics Study on Direct Sol-Gel Reactions in CO₂ by Using In Situ ATR-FTIR Spectrometry

In this chapter, a chemometric method is used to extract pure-component spectra and the concentration profiles from the in situ ATR-FTIR spectra, which were obtained during the sol-gel process of TEOS, TIP and ZBO with acetic acid in CO₂. The precursor conversions and the metal complex concentration profiles under different synthesis conditions are discussed.

7.1. Introduction

In previous chapters, synthesis of SiO₂, TiO₂ and ZrO₂ nanomaterials in CO₂ were described; however, the kinetics were not discussed. A kinetics study requires rather detailed knowledge of the chemical reactions, e.g. the step-by-step reaction scheme, and the concentration profiles of reactants, intermediates and/or products. By mixing TEOS with acetic acid without a solvent, the possible sol-gel reaction steps have been proposed by Sharp using NMR results:^267

Substitution:

≡SiOR + HOAc ⇌ ≡SiOAc + ROH (7.20)

Esterification:

HOAc + ROH ⇌ AcOR + H₂O (7.21)

Hydrolysis:

≡SiOR + H₂O ⇌ ≡SiOH + ROH (7.22)

Condensation:

≡SiOH + ≡SiOH ⇌ ≡SiOSi≡ + H₂O (7.23)

Condensation:

≡SiOH + ≡SiOR ⇌ ≡SiOSi≡ + ROH (7.24)

Condensation:

≡SiOH + ≡SiOAc ⇌ ≡SiOSi≡ + HOAc (7.25)

Carboxylation:

≡SiOH + HOAc ⇌ ≡SiOAc + H₂O (7.26)

The chemistry of titanium alkoxides reacting with acetic acid in conventional solvents is also reasonably well understood. Doeuff et al. explained the formation of the Ti hexamer complex through modification, esterification, hydrolysis and condensation.^232 The reactions can be generally written as:

Modification:

Ti(OR)₄ + HOAc → Ti₆(OAc)_m(OR)_n + ROH (7.27)

Esterification:

ROH + HOAc ⇌ ROAc + H₂O (7.28)

Hydrolysis:

Ti₆(OAc)_m(OR)_n + H₂O → Ti₆(OAc)_m(OR)_n-x(OH)_x + ROH (7.29)

Oxolation:

Ti₆(OAc)_m(OR)_n-x(OH)_x → Ti₆O_x(OAc)_m(OR)_n-2x + ROH (7.30)

Where, Ti₆O_x(OAc)_m(OR)_n-2x is the hexamer complex.

Zirconium alkoxide reacting with carboxylic acid has also been studied in conventional solvents and the products were characterized by single crystal XRD.^268-270 According to the scheme provided by Kickelbick,^268 the reaction pathways can be written as:

Modification:

Zr(OR)₄ + HOAc → Zr_m(OR)_4m-n(OAc)_n + ROH (7.31)

Esterification:

HOAc + ROH ⇌ ROAc + H₂O (7.32)

Hydrolysis:

Zr_m(OR)_4m-n(OAc)_n + H₂O → Zr_m(OR)_4m-n-x(OAc)_n(OH)_x + ROH (7.33)

Oxolation:

Zr_m(OR)_4m-n-x(OAc)_n(OH)_x → Zr_mO_x(OR)_4m-n-2x(OAc)_n + ROH (7.34)

Further condensation

Zr_mO_x(OR)_4m-n-2x(OAc)_n → macromolecules (7.35)

It is noted that water is required for the sol-gel processes. At the initial stage of the reaction, water was generated only through the esterification reaction. With the reactions proceeding, water can also be generated from condensation.

To date, no kinetics study on the reactions between the alkoxides and acetic acid has been reported. For a detailed kinetics study, the concentration profiles of the reactants or products are indispensable. On-line monitoring of a chemical process using vibrational spectrometry, e.g. in situ ATR-FTIR or Raman, is of increasing importance due to the real time analysis of the components in the reactor. However, the collected spectra are often complicated by peak overlapping, and hence cannot be directly analyzed using a conventional peak height or peak area measurement. For instance, the previous in situ ATR-FTIR spectrometry results in Chapters 4-6 have shown that the significant peaks of the precursors are overlapped severely with those of the products.

In order to calculate the component concentrations from the complicated spectra, mathematical and statistical methods, known as chemometrics, have been used.^271 There are two categories of chemometric methods for resolving the spectrometry data: i.e. calibration and self-modeling curve resolution (SMCR). The least squares regression method has also been extensively used in the calibration methods.^{151, 271} However, in situations when there is an unknown component, or the pure component spectrum for the reference component is not available, the calibration methods cannot resolve the mixture spectra.^272 In these situations, e.g. the sol-gel process for SiO₂ TiO₂ and ZrO₂, the SMCR methods are attractive.

The SMCR method uses chemometrics to extract a set of concentration profiles and spectra of pure components from a set of mixture spectra, e.g. in situ IR spectra, without any prior knowledge about the system.^{273, 274} The first SMCR method was proposed by Wallace in 1960.^275 A simple-to-use interactive self-modeling mixture analysis (termed SIMPLISMA), was proposed by Windig of the Eastman Kodak Company in 1991.^{272, 276} The SIMPLISMA technique has been an effective SMCR method for data analysis in various chemical processes using UV, IR, Raman and NMR spectrometry.^277-285

To describe the chemometric SIMPLISMA technique, consider a reaction system with n components, in which a set of IR spectra collected at different reaction times can be written as a matrix, D. As D is a function of the concentration and the absorbance (B_pure) of each component, it can be expressed as:

D = C ×B (7.36)

where D is the known rw matrix transformed from a set of in situ IR spectra, C is the to-be-determined rn matrix representing concentration coefficients, B is the to-be-determined nw matrix representing n pure component spectra, r is the number of spectra, and w is the index of wavenumbers.^286

In the IR spectra, the absorbance will change at certain wavenumbers due to the reaction proceeding. The SIMPLISMA method tries to find the highest absorbance standard deviation from the mean value in the IR spectra. The “purity”, q_j, is calculated as:

(7.37)

where, μ_j is the mean value of absorbance at wavenumber j, σ_j is the standard deviation at wavenumber j, and α is an empirical parameter called offset, which is added to the mean value in order to suppress the noise in the region where the absorbance is close to zero. Typical values for α range from 1-5 % of the maximum μ_j for the conventional method; for the second-derivative method as described later, however, α is normally selected as 20 %.^{276, 287}

A pure variable is defined as the wavenumber where only one component contributes to the absorbance. The first ‘pure variable’ is selected as the wavenumber that has the highest q_j value. Then other pure variables are selected, which are independent of the first variable. The absorbances of the pure variables from all components in matrix D are used to calculate matrix C using the linear correlation of the Beer-Lambert law described in Chapter 3. With the known matrices D and C, Matrix B can be calculated using a least-squares regression method.^{272, 276, 286}

In the second-derivative method, the calculated matrix B was used to recalculate a new C denoted as C’ using a least square approximation. Thus the matrix C’ provides the second-derivative concentrations. Consequently, the second-derivative pure component spectra B’ can be obtained using matrices D and C’. It was found that the second-derivative method was effective when the conventional pure-variable approach could not give satisfactory results.^{276, 287} It should be noted that SIMPLISMA cannot calculate the absolute concentrations of the components. In the case when quantitative solution is necessary, an independent calibration of the system should be performed for every component of interest. The advantages of SIMPLISMA modeling include: knowledge about the pure component spectra are not required, the purity and resolved spectra are visualized, and the overlapped IR peaks can be resolved.^272 More details about SIMPLISMA modeling are described in Appendix 9.

In this chapter, the SIMPLISMA method is used for modeling the in situ IR spectra collected during the direct sol-gel process in CO₂, and the resulting concentration-time curves are discussed. In addition, a curve-fitting method was also used to study the in situ IR spectra. The curve-fitting procedure consists of modeling an experimental spectrum using a sum of Gauss peak functions with a least residual sum of squares. The residual sum of squares refers to the residual sum of squared differences between the calculated spectrum and the original experimental data.