Maximum Incorporation of Oxygen Radicals, O - American Chemical

C was treated in pressurized 95% Ar-5% O2 gas confined in a hot isostatic pressing (HIP) furnace, while samples D, E, and. F were treated by the HIP u...
0 downloads 0 Views 72KB Size
Chem. Mater. 2003, 15, 1851-1854

1851

Maximum Incorporation of Oxygen Radicals, O- and O2-, into 12CaO·7Al2O3 with a Nanoporous Structure Katsuro Hayashi,*,† Satoru Matsuishi,†,‡ Naoto Ueda,‡ Masahiro Hirano,† and Hideo Hosono†,‡ Transparent Electroactive Materials Project, Exploratory Research for Advanced Technology (ERATO), Japan Science and Technology Corporation (JST), KSP C-1232, Kawasaki 213-0012, Japan, and Materials and Structures Laboratory, Tokyo Institute of Technology, Yokohama 226-8503, Japan Received October 1, 2002. Revised Manuscript Received February 12, 2003

The effects of oxygen partial pressure (pO2) up to 400 atm on the generation of active oxygen radicals in 12CaO‚7Al2O3 (C12A7) was examined using a hot isostatic pressing technique. Concentrations of superoxide (O2-) and oxygen anion radicals (O-), which were analyzed from a combination of electron paramagnetic spin resonance (EPR) and Raman spectroscopy, increased with pO2. The total concentration of the radicals reached 1.7 × 1021 cm3, which is comparable to the theoretical maximum, confirmed that dominant radical incorporation process is O2- (cage) + O2 (atmosphere) f O- (cage) + O2- (cage). The radical incorporated C12A7 has a large magnetic susceptibility that obeys Curie’s law down to 2.5 K without magnetic ordering, although a weak exchange interaction was suggested by the line narrowing of the EPR signal.

Introduction 12CaO‚7Al2O3 (C12A7),1-10 a constituent of alumina cement, has a specific nanoporous structure that features 12 cages with a diameter of ∼0.4 nm embedded in a positively charged framework of [Ca24Al28O64]4+ in a cubic unit cell (2 chemical formula per unit cell with a lattice dimension of 1.1989 nm). The two remaining O2- ions, referred to as “free oxygen”, are entrapped in * To whom correspondence should be addressed. † Japan Science and Technology Corporation. ‡ Tokyo Institute of Technology. (1) Bartl, H.; Scheller, T. Neues Jahrb. Mineral. Monatsh. 1970, 35, 547-552. (2) Nurse, R. W.; Welch, J. H.; Majumdar, A. J. Trans. Br. Ceram. Soc. 1965, 64, 323-332. Imlach, J. A.; Glasser, L. S. D.; Glasser, F. P. Cem. Concr. Res. 1971, 1, 57-61. Zhmoidin, G. I.; Chatterjee, A. K. Cem. Concr. Res. 1984, 14, 386-396. Edmonds, R. N.; Majumdar, A. J. Cem. Concr. Res. 1988, 18, 473-478. (3) Jeevaratnam, J.; Glasser, F. P.; Glasser, L. S. D. J. Am. Ceram. Soc. 1964, 47, 105-106. (4) Recently, incorporation of H- was also found: Hayashi, K.; Matsuishi, S.; Kamiya, T.; Hirano, M.; Hosono, H. Nature 2002, 419 462-465. (5) Hosono, H.; Abe, Y. Inorg. Chem. 1987, 26, 1192-1195. (6) Sto¨sser, R.; Nofz, M.; Gessner, W.; Schro¨ter, C.; Kranz, G. J. Solid State Chem. 1989, 81, 152-164. (7) Hayashi, K.; Hirano, M.; Matsuishi, S.; Hosono, H. J. Am. Chem. Soc. 2002, 124, 738-739. (8) Lemonidou, A. A.; Vasalos, I. A. Appl. Catal. 1989, 54, 119138. Pant, K. K.; Kunzru, D. Ind. Eng. Chem. Res. 1997, 36, 20592065. (9) Lacerda, M.; Irvine, J. T. S.; Glasser, F. P.; West, A. R. Nature 1988, 332, 525-526. Irvine, J. T. S.; West, A. R. Solid State Ionics 1990, 40-41, 896-899. (10) Li, Q.-X.; Hayashi, K.; Nishioka, M.; Kashiwagi, H.; Hirano, M.; Torimoto, Y.; Hosono, H.; Sadakata, M. Appl. Phys. Lett. 2002, 80, 4259-4261. Li, Q.-X.; Hayashi, K.; Nishioka, M.; Kashiwagi, H.; Hirano, M.; Torimoto, Y.; Hosono, H.; Sadakata, M. Jpn. J. Appl. Phys. 2002, 41, L530-L532. Hayashi, K.; Li, Q.-X.; Nishioka, M.; Matsuishi, S.; Torimoto, Y.; Hirano, M.; Sadakata, M.; Hosono, H. Electrochem. Solid-State Lett. 2002, 5, J13-J16. Li, Q.-X.; Hosono, H.; Hirano, M.; Hayashi, K.; Nishioka, M.; Kashiwagi, H.; Torimoto, Y.; Sadakata, M. Surf. Sci. 2003, 527, 100-112.

the cages.1 The formal charge per cage is +1/3. Thus, the entrapped oxygen ions are weakly bound electrostatically in the cage. In addition, when a free oxygen ion is replaced by two monovalent anions, the nanoporous cage may be stabilized due to charge delocalization. On the basis of these findings, it is expected that the free oxygen in C12A7 will be replaced by a significant amount of monovalent anions (X-). The resulting compound, [Ca24Al28O64]4+‚4(X-), where all the free oxygen ions are replaced by X- ions, indicates that the maximum concentration of the anion involved in the lattice cavity is 4 ions per unit cell, that is, 2.32 × 1021 cm-3. In fact, it is already known that a large amount of OH- ions2 or halogen anions (F- or Cl-)3 are easily incorporated in the lattice.4 In 1986, one of the present authors discovered that the superoxide radical (O2-) occurs in the as-prepared sample with a concentration of ∼1019 cm-3 using electron paramagnetic spin resonance (EPR).5 We recently further demonstrated that, in addition to O2-,5-7,11-13 the oxygen anion radical (O-), which is the most active oxygen species,7,11-13 is simultaneously formed by heating C12A7 ceramics at 1350 °C and subsequent cooling in a dry oxygen atmosphere with a total concentration of 6 × 1020 cm-3.7 We proposed that eq 1, in which molecular O2 in the atmosphere diffuses into the lattice followed by an electron transfer from free O2- ion to the O2, is the major process responsible for the generation of these oxygen radical ions.7

O2- (cage) + O2 (atmosphere) f O- (cage) + O2- (cage) (1)

10.1021/cm020959g CCC: $25.00 © 2003 American Chemical Society Published on Web 03/20/2003

1852

Chem. Mater., Vol. 15, No. 9, 2003

Hayashi et al.

Table 1. Conditions of Sample Preparation

sample

apparatus

A

tube furnace tube furnace HIP HIP HIP HIP

B C D E F a

total pressure (atm)

gas

pO2 (atm)

pH2O (atm)

20% O2-80% N2

1

0.2

3 × 10-4

100% O2

1

1

3 × 10-4

5% O2-95% Ar 20% O2-80% Ar 20% O2-80% Ar 20% O2-80% Ar

80 1 × 102 5 × 102 2 × 103

4 20 1 × 102 4 × 102

(>5 × 10-3)a (>6 × 10-3)a (>3 × 10-2)a (>0.1)a

The value is not directly measured and is estimated from the H2O impurities contained in the loading gas and in the sample.

The incorporation of such a high concentration of active oxygen radicals O2- and O- would provide a unique opportunity for useful applications such as a powerful oxidization catalyst,8 an oxygen ion conductor,9 and a field-mediated O- emitter.10 The magnetic properties due to the magnetic interaction among the high concentration of the paramagnetic oxygen radicals are also of interest. In this study, we have treated C12A7 in pressurized oxygen pressure, with pressures up to 400 atm, and found that as the pressure increased, the total concentrations of O2- and O- incorporated in the lattice increased up to 1.7 × 1021 cm-3, which corresponds to ca. three-fourths of the theoretical maximum. Experimental Section C12A7 ceramics were prepared by a solid-state reaction of highly pure CaCO3 and γ-Al2O3 powders at 1350 °C for 6 h in dry oxygen (pO2 ) 1 atm, pH2O ) 6 × 10-5 atm). The resultant ceramics were composed of small grains ∼5 µm in size and had ∼80% of the theoretical density (2.68 g‚cm-3 3). The samples were further heated at 1350 °C for 2 h under various pO2 atmospheres listed in Table 1. They were then cooled at a rate of 200 °C/h under the same atmospheric conditions. Samples A and B were treated in flowing dry air and dry oxygen atmospheres in a tube furnace, respectively. Sample C was treated in pressurized 95% Ar-5% O2 gas confined in a hot isostatic pressing (HIP) furnace, while samples D, E, and F were treated by the HIP using 80% Ar-20% O2 gas. Sample G was sintered at 1350 °C for 6 h in air (pO2 ) 0.2 atm, pH2O ) 4 × 10-2 atm). EPR measurements were conducted at ∼9.7 GHz (X-band) using a Bruker E580 spectrometer at 77 K. Spin concentrations were determined from the second integral of the spectrum using CuSO4‚5H2O as a standard with an accuracy of (20%. The first derivative line shape of the powder patterns, I, with respect to the magnetic field, H, were computersimulated by eqs 2 and 3,11

1 I(H) ) 4

∫ ∫ 2π

0

π

0

f′(Hr - H, ∆H) sin θ dθ dφ

(2)

where θ and φ are the polar angles, ∆H is the peak-to-peak line width of a Lorentzian or Gaussian function f′, and Hr is described by

Hr )

hν0 (g 2 sin2 θ cos2 φ + g222 sin2 θ sin2 φ + β 11 g332 cos2 θ)-1/2 (3)

where gii (i ) 1, 2, 3) are the g values of the three principal (11) Lunsford, J. H. Adv. Catal. 1972, 22 265-344. (12) Soria, J.; Martı´nez-Arias, A.; Conesa, J. C.; Munuera, G.; Gonza´lez-Elipe, A. R. Surf. Sci. 1991, 251/252, 990-994. (13) Che, M.; Tench, A. J. Adv. Catal. 1983, 32, 1-148.

Figure 1. Experimental EPR spectra for C12A7 ceramics treated in atmospheres with pO2 ) 0.2 atm (a), pO2 ) 10 atm (c), and pO2 ) 400 atm (e) together with computer-simulated spectra for pO2 ) 0.2 (b) and pO2 ) 10 (d). EPR spectra were measured at 77 K. Solid curves in (b) and (d) are superpositions of the calculated spectra of O2- (point-dashed line) and O(dashed line) components. axes for the radicals, h is the Plank constant, β is the Bohr magneton, and ν0 is the microwave frequency. Raman spectra were measured with Fourier transformation (FT)-Raman spectrometer at room temperature using a YAG laser (λ ) 1.064 µm) as the excitation light source. Magnetization was measured over a temperature range of 2.5-300 K and in a magnetic field up to 5 × 104 Oe using a superconducting quantum interference device (SQUID) magnetrometer.

Results and Discussion Figure 1 shows the experimental EPR spectra of samples A, C, and F and the simulated spectra for samples A and C. The experimental spectra for samples A, B, and C were reproduced by the superposition of the calculated powder patterns for O2- and O-, where Lorentzian line shape was employed as it yielded a better reproduction than the Gaussian line shape. The g tensors, which are assumed independent of the radical concentrations, were obtained using a least-squares fit as gxx ) 2.002 ((0.001), gyy ) 2.008 (0.001), and gzz ) 2.074 (0.001) for O2- and gxx ) gyy ) 2.036 (0.005) and gzz ) 1.994 (0.01) for O-.7 (The ∆H changes as a function of the radical concentration are shown in Figure 4). Further, the concentrations of O- and O2- radicals were separately evaluated for samples A, B, and C. While the presence of O2- in C12A7 was already established in 1986,5 the existence of O- has been

Maximum Incorporation of Oxygen Radicals

Figure 2. Raman spectra of C12A7 ceramics treated in varying atmospheres with pO2 ) 0.2 atm (sample A) and pO2 ) 400 atm (sample F). Spectrum of C12A7 sintered in air (pO2 ) 0.2, pH2O ) 6 × 10-2 atm (sample G) is shown for comparison. The inset is the relationship between the relative intensity of the 1128-cm-1 band to the 510-cm-1 band in the Raman spectrum and the concentration of O2- evaluated from EPR analysis.

confirmed from very broadened EPR spectra, which is always superposed with the signal of O2-.7 However, only the above sets of g tensors can consistently reproduce a serious of spectra for A, B, and C as well as those in the previous study.7 Line shapes similar to those of samples A and C were reported in halogendoped CeO2 and analyzed to be the coexistence of O2and O-.12 The magnitude of the gzz component of the O2- ion gives a measure of the cation charge at the adsorption site.13 The gzz of 2.074 in O2- implies that the O2- is attached to a Ca2+ ion. Although an O2adsorbed on Al3+ is expected to give a shoulder around gzz ) 2.036, the superposition of two O2- signals with gzz values of 2.074 and 2.034 do not reproduce a series of the observed spectra well. Further, the absence of resolved hyperfine splitting due to 27Al nuclear rules out this possibility. The existence of O- is also evidenced by high-density emission of O- ion from C12A7 by applying an electric field.10 When the pO2 is increased above 10 atm (samples D, E, and F), the spectral shape lost the g anisotropy and became a nearly single Lorenzian shape as in Figure 1d, which made it very difficult to separately evaluate each radical concentration from the EPR spectra. In such cases, each concentration of O- and O2- was estimated by combining Raman data with EPR data as described below. Figure 2 shows Raman spectra of samples A, F, and G. Raman bands located at 200-1000 cm-1 arise from the lattice framework, which is composed of tetrahedrally coordinated Al3+ ions.14 The intensity of a sharp band peaking at 1128 cm-1, which is not observed in sample G, markedly increased with the oxygen partial pressure in the heat treatment. Because of this observation and the agreement of the peak position with the reported one,13 this band is assigned as the O2- stretching mode. The inset of Figure 2 shows the plot of the Raman intensities of the 1128-cm-1 band relative to the 510-cm-1 band as a function of the O2- concentration obtained from the EPR spectra for samples A, B, and C. The linear relationship allows the O2- concentration in samples D, E, and F to be estimated. Then, the (14) McMillan, P.; Piriou, B. J. Non-Cryst. Solids 1983, 55, 221242.

Chem. Mater., Vol. 15, No. 9, 2003 1853

Figure 3. Plots of O2-, O-, and their total concentrations in C12A7 ceramics estimated from Raman spectroscopy and EPR against the oxygen partial pressure in the heat treatment atmosphere.

concentration of O- was determined by subtracting the O2- concentration from the total spin concentration obtained from EPR. Significant changes were not observed by FT-Raman and EPR for the samples placed in an ambient atmosphere for several months. This observation suggests that the active oxygen radicals are stably entrapped in the cage. Figure 3 plots the concentrations of total spin (i.e., oxygen radicals), O2- and O- radicals as a function of pO2. Due to the discontinuity between sample B and C, the data may be divided into two groups: one for samples A and B, which were processed in a flowing gas atmosphere, and the other for samples C, D, E, and F, treated in HIP. The concentrations increase with pO2 within each group and the total spins approach the theoretical maximum value of 2.3 × 1021 cm-1 above 100 atm. The highest total concentration in the present experiment was 1.7 × 1021 cm-3. The concentration of O- was as high as 5 × 1020 cm-3, which to the best of our knowledge is the highest concentration of O- among all materials reported to date. In the preceding paper,7 eq 1 was proposed as the tentative mechanism. According to this model, the driving force of the reaction increases with partial oxygen pressure. The formation of the oxygen radical should be suppressed in the presence of water because O2- may be replaced with OH- in the cage.2 The water vapor included in the loading gas is also pressurized in the HIP treatment, which may significantly influence the oxygen radical formation. The discontinuity of B and C in Figure 3 may be explained by the differences in pH2O between the flowing gas and the pressured gas for HIP. Further, if eq 1 is the main reaction path for active oxygen generation, the concentration ratio [O-]/[O2-] should be unity for the active oxygen generation during cooling. The concentration ratio [O-]/[O2-] obtained in the present study ranges from 0.25 to 0.60. This result along with the observed increase of the oxygen radical concentration with pO2 may validate eq 1 if it is assumed that O2- is more stable than O-.11,13 Figure 4 is a plot of the line width, ∆H, of EPR spectra, against the total spin concentration. In samples A, B, and C the effective line width of the two superposed signals was estimated from (∆Hs [O2-] + ∆Ho [O-])/([O2-] + [O-]), where ∆Hs and ∆Ho are the ∆H’s for O2- and O-, respectively. As the shape may be regarded as Lorentzian for samples D, E, and F, ∆H

1854

Chem. Mater., Vol. 15, No. 9, 2003

Figure 4. Variations in the line widths of EPR spectra with total radical concentration incorporated into C12A7. Curves are to guide the eye.

was read directly from the spectra. The value of ∆H increased with the spin concentration up to 1.0 × 1021 cm-3, and the dependence turned to a decrement in the higher concentration range. The increase of ∆H with the spin concentration is obviously attributed to a dipoledipole interaction between oxygen radicals, while the narrowing suggests an exchange interaction between them arising from a partial overlap between the semioccupied orbitals of O2- and/or O-. Figure 5 shows the magnetization (M)-magnetic field (H) characteristics of sample F (radical concentration of 1.7 × 1021 cm-3). The M-H curves at several temperatures were well described by the Brillouin function15 with S ) 1/2 among 2.5-300 K, which indicates that the paramagnetism observed here was due to the oxygen radicals with S ) 1/2. The number of S ) 1/2 spins evaluated from the M-H measurements is in agreement with that from EPR. The temperature (15) Kittel, C. Introduction to Solid State Physics, 6th ed.; Wiley: New York, 1986.

Hayashi et al.

Figure 5. Magnetization-field characteristics of C12A7 ceramics (sample F) treated in the atmospheres with pO2 ) 400 atm measured at 2.5, 4.2, and 77 K. The horizontal axis is normalized by temperature. The dashed line corresponds to the Brillouin function with S ) 1/2.

dependence of the paramagnetic susceptibility obeys Curie’s law down to 2.5 K, indicating magnetic ordering does not occur in this temperature range. Conclusion It is demonstrated that almost a theoretical maximum concentration of oxygen radicals can be incorporated into C12A7 cages by a heat treatment in high-pressure oxygen greater than 100 atm. This fact clarifies that the upper limit of anion incorporation is defined by the positive charge of the lattice framework. Concentrations of O2- and O- are estimated by the combination of Raman spectroscopy and EPR. The maximum concentration of O- reached 5 × 1020 cm-3, which is the highest concentration in all materials as far as we know. Although a large magnetization, which is due to the oxygen radicals, is observed in the oxygen radicalincorporated C12A7 in a temperature range of 2.5-300 K, magnetic ordering is not detected down to 2.5 K. CM020959G