Perturbations to 27Al Electric Field Gradients in Nanocrystalline α

Jan 14, 2010 - Crystal Growth Centre, Anna UniVersity, Chennai 600 025, India, and Central NMR .... National Chemical Laboratory and University of Mad...
0 downloads 0 Views 2MB Size
J. Phys. Chem. B 2010, 114, 1775–1781

1775

Perturbations to 27Al Electric Field Gradients in Nanocrystalline r-Al2O3 Studied by High-Resolution Solid-State NMR V. Sabarinathan,† S. Ramasamy,*,† and S. Ganapathy*,‡ Crystal Growth Centre, Anna UniVersity, Chennai 600 025, India, and Central NMR Facility, National Chemical Laboratory, Pune 411008, and CAS in Crystallography & Biophysics, UniVersity of Madras, Chennai 600025, India ReceiVed: August 3, 2009; ReVised Manuscript ReceiVed: December 10, 2009

27

Al solid-state NMR has been employed to study the perturbations to 27Al electric field gradients for the aluminum environments in nanocrystalline R-alumina. Triple quantum magic angle spinning experiments show that the octahedral aluminum coordination remains unchanged down to 12 nm, although severe perturbations to 27Al electric field gradients are noticed at 28 nm and below. 3Q-MAS and SATRAS experimental data of nano R-alumina have been analyzed through extensive spectral simulations to probe 27 Al electric field gradients of aluminum in the grains and grain boundaries. While the aluminum in the grains has a unique field gradient tensor, the same octahedrally coordinated aluminum environments in the grain boundaries suffer a distribution of electric field gradients. This is evidenced by data analysis of both 3Q-MAS and SATRAS spectra. By invoking the Gaussian isotropic model, in which the (CQ, ηQ) parameter space is discretely sampled by the Czjzek distribution, we have been able to analyze the 27Al SATRAS spectra of nanocrystalline R-alumina samples having grain sizes of 52, 28, 20, and 12 nm. Good agreement between experimental and simulated spectra has led to the quantitative determination of grain and grain boundary components in these materials. Introduction The thermal decomposition of aluminum hydroxides (gibbsite, bayerite, and boehmite) yields a series of structurally ordered intermediate alumina phases, collectively referred to as “transitional aluminas”.1,2 R-Alumina is the end member of the transition alumina series (η, χ, γ, κ, δ, θ, R-Al2O3)1,3-6 and has a unique crystallographic identity.7 R-Al2O3 is particularly stable and irreversible when annealed at high temperatures (>1100 °C). It has good thermal conductivity, high temperature and wear resistance, and excellent size and shape capability and possesses high strength and stiffness. Endowed with these material characteristics, and with an ability to fine-tune its particle size, R-Al2O3 is among the most widely used materials in catalysis8 and advanced ceramics.9 When the particle size falls in the nanoscale, the influence of the surface and interface leads to marked changes in its physical, chemical, and mechanical properties.10-12 These are well-documented for nanocrystalline R-Al2O3.13,14 Often, such changes which occur in nanomaterials are correlated to the presence of dangling bonds and a softening of the lattice in the grain boundaries.15-17 In so far as the aluminum environments are concerned, severe perturbations are expected to occur for the aluminum sites at the grain boundary interface compared to the same aluminum environment in micrograins. In this regard, molecular-level insights can be gathered by studying 27 Al electric field gradients as a prelude to understanding the electronic structure of nanomaterials, which largely determine the new material properties, such as the enhanced catalytic * To whom correspondence should be addressed. E-mail: sinna_ [email protected] (S.R.); [email protected] (S.G). † Anna University. ‡ National Chemical Laboratory and University of Madras.

activity of R-Al2O3 when used as a metal support18,19 and the increased sintering behavior when used as an advanced ceramic material.20 27 Al solid-state NMR has been widely used to determine the aluminum coordination and probe the local structural characteristics of aluminum environments in bulk alumina materials.21 Aluminum in four (tetra)-, five (penta)-, and six (hexa)coordinated environments can be distinguished from 27Al magic angle spinning (MAS) experiments owing to their large chemical shift dispersion (ca. 0-70 ppm). For the bulk R-Al2O3, which has a rhombohedral structure with space group R3c,7 previous MAS studies show that the aluminum is octahedrally coordinated and its electric field gradient tensor is affected by the crystallinity, defect, disorders, and dynamic processes of powdered R-Al2O3 samples.22,23 In the study of nano R-Al2O3 materials, 27Al solid-state NMR offers new opportunities to probe the aluminum environments in the grains and grain boundaries. Multiple quantum magic angle spinning (MQMAS)24 is especially suitable for studying the aluminum coordination and the 27Al electric field gradients under isotropic conditions since second-order quadrupolar broadening is removed in this experiment. Additionally, by using satellite transition magic angle spinning (SATRAS),25 the relative population of aluminum environments residing as bulk and surface species can be quantitatively determined. The present work deals with a detailed 27Al solid-state NMR study of nanocrystalline R-Al2O3 having grain sizes in the range of 52-12 nm, using triple quantum (3Q), satellite transition (ST), and central transition (CT) MAS techniques at the magnetic field of 7.01 T. Our attention is specifically focused on delineating the effects of the grain size on the 27Al electric field gradient tensor through the experimentally determined 27Al quadrupole interaction parameters (CQ, ηQ) and isotropic chemical shifts (δiso) for the aluminum sites existing within the grains

10.1021/jp907469n  2010 American Chemical Society Published on Web 01/14/2010

1776

J. Phys. Chem. B, Vol. 114, No. 5, 2010

Figure 1. XRD spectra of R-alumina (a) annealed at 1200 °C/2 h (0 h) and ball milled for 5 (b), 10 (c) 20 (d), 25 (e), and 52 h (f).

and grain boundaries. Although solid-state NMR of nano R-Al2O3 has been reported earlier,26 the present work extends studies of R-Al2O3 down to 12 nm and employs 3Q-MAS for the first time. A more detailed and quantitative characterization of the 27Al electric field gradients at the 52-12 nm scale is provided through a combination of 3Q, ST, and CT MAS techniques. Materials and Methods Al(OH)3 was first prepared by chemical precipitation method using aluminum nitrate as the source. The sample was then heated to 1200 °C for 1 h in air to yield R-alumina. The powder XRD spectrum confirmed that this end material is highly crystalline and conforms to the rhombohedral structure (see Supporting Information). A high-energy ball milling technique was used to further reduce the grain size and get R-alumina in the nanoscale. For the ball milling, alumina ball and vials were used, with a ball to sample weight ratio of 10:1 and rotation at 500 rpm. The use of alumina balls instead of iron balls also eliminated the presence of paramagnetic iron impurities in the final samples, which can affect the 27Al solid-state NMR spectra. The milling hours were 0, 5, 20, and 52 h. The XRD spectra of nano R-Al2O3 samples prepared by high-energy ball milling are shown in Figure 1. All of these spectra can be indexed and matched to the standard (JCPDS # 89-7717), confirming thereby that all of the milled samples conform to rhombohedral R-Al2O3. The average grain size of these milled samples was determined from the Scherrer formula27 after fitting the powder XRD spectra using PEAKFIT software. The grain size of the 0, 5, 20, and 52 h milled R-Al2O3 samples was found to be 52, 28, 20, and 12 nm, respectively. The starting material having a grain size of 52 nm was taken as bulk R-Al2O3. 27 Al solid-state NMR experiments were performed on a Bruker AV-300 NMR spectrometer at the Larmor frequency of 78.172 MHz, corresponding to the magnetic field of 7.01 T. NMR experiments were carried out at ambient probe temperature (296-299 K). The spinning speed was kept at 10 kHz and was controlled to within (1 Hz using a Bruker pneumatic controller. The spectra were gathered in three different acquisition modes, namely, central transition (-1/2 T + 1/2) (CTMAS), satellite transition ((5/2 T (3/2, (3/2 T (1/2), and triple quantum (3Q-MAS). Single-pulse excitation was employed to acquire CT and SATRAS spectra. The 27Al rf pulse length was chosen to be very short, taking into account the nutation behavior28 of the quadrupolar spins. The acquisition

Sabarinathan et al. parameters used were CT-MAS: pulse width ) 0.60 µs, recycle delay ) 1 s, spectral width ) 62500 Hz, number of scans ) 1000 and SATRAS: pulse width ) 0.6 µs, recycle delay ) 1 s, spectral width ) 2 MHz, number of scans ) 4000. 27 Al 3Q-MAS experiments were carried out using a threepulse sequence29 incorporating z-filter to symmetrize the coherence transfer pathways. Rotor synchronization was employed during the triple quantum evaluation period to eliminate spinning sidebands appearing along the isotropic dimension and to improve the signal-to-noise ratio. Absorption-mode 2D spectra were gathered using the hypercomplex procedure.30 For 3QMAS, the acquisition parameters used were 3Q excitation pulse length ) 2.6 µs; 3Q to 0Q conversion pulse length ) 1 µs; z-filter pulse length ) 10.5 µs; rf field for the hard pulses ) 100 kHz; rf field for the soft pulse ) 23.8 kHz; F2 spectral width ) 15 kHz; acquisition data points ) 1024; and number of scans ) 120. The 2D data matrix was apodized using an exponential (LB ) 10-50 Hz) window function along the acquisition dimension prior to Fourier transformation. No window function was used in the other dimension. The apodized data was Fourier transformed and sheared to align the 2D spectrum in such a way that projections parallel to F2 and F1 axes yielded the MAS spectrum and isotropic spectrum, respectively. In all of the experiments, 27Al chemical shift values were externally referenced to [Al(OH)6]3+. Computer Simulations of Solid-State NMR Spectra. For quantitative analysis of 27Al satellite transition and triple quantum MAS spectra, extensive computer simulations of 27Al solid-state NMR spectra were undertaken in the present study. The simulation of solid-state NMR spectra for half-integer quadrupolar spins experiencing first- and second-order quadrupolar interactions was carried out using DMFIT.31 This simulation environment allows the inclusion of crystalline and amorphous components and further invokes a distribution of electric field gradients using the Gaussian isotropic model (GIM)32 based on the Czjzek distribution.33 Additionally, a Gaussian distribution for the isotropic chemical shifts can be included for the final spectral simulations. In our study of nano R-Al2O3 samples, the final fit of the satellite transition and triple quantum MAS spectra should reflect a distribution of the quadrupole interaction parameters and the isotropic chemical shift.33 To a good degree, this could be accomplished in regular simulations. While DMFIT facilitates simulation of MAS and MQ-MAS spectra with the inclusion of the Czjzek distribution,34 it is somewhat restrictive for simulating satellite transition MAS spectra, owing to the limitation on the maximum number of sidebands in the spectrum. For the simulation of satellite transition MAS spectra, we have employed the following strategy to alleviate the above restriction in the SATRAS simulations. First, the simulation of the satellite transition MAS spectrum corresponding to one unique value of the field gradient tensor was carried out. Here, the simulation procedure calculates the spinning sideband manifold and the associated second-order shifts for the m f m - 1 satellite transition, together with the central transition in the middle, for a powder sample. For the nanomaterial, the electric field gradient distribution must be considered since all of the aluminum sites in the grain boundaries are not governed by a unique value of the field gradient tensor. For the modeling of the EFG distribution, a simple Gaussian distribution, such as that used in the 93Nb solid-state NMR studies of perovskite materials,35,36 can be used. However, we have chosen to use the Czjzek distribution as it allows a variation of both CQ and ηQ. We have resorted to using this distribution function in view

Perturbations to

27Al

Electric Field Gradients

J. Phys. Chem. B, Vol. 114, No. 5, 2010 1777

Figure 2. 27Al MAS spectra of 52 nm R-alumina showing the satellite transitions and the associated spinning sidebands (left) and the central transition (right). Experimental spectra are shown at the bottom (a,c) and the computer simulated spectra at the top (c,d). The expansions in (a) and (b) depict the second-order effect on the satellite sidebands, which is more pronounced in the 230-240 kHz range, as shown.

of its wide acceptance in other studies37,38 and the consistency it provides in the simulation of 3Q-, ST-, and CT-MAS spectra using the same mathematical model to describe the field gradient distribution. For the Czjzek function,34 the probability distribution for the electric field gradient parameters νQ ans ηQ, namely, [P(ηQ, ηQ)], is given by

P(νQ, ηQ) )

1

√2πσd

νQd-1η(1

-

{

ηQ2 /9)exp

-

νQ2(1 + ηQ2 /3) 2σ2

}

where σ, the only adjustable parameter, denotes the width of the distribution and d is taken to be 5. Here, νQ is the quadrupolar frequency which is related to the quadrupole coupling constant (CQ) [νQ ) 3CQ/2I(2I - 1)], and ηQ is the asymmetry parameter. Using the above expression, the field gradient distribution was mapped by choosing a grid of points in which the (νQ, ηQ) parameter space was adequately sampled at close intervals. This was done by taking 300 such sampling points in the region of CQ ) 1.8-3.0 MHz and ηQ ) 0.3-0.8. For each of these (νQ, ηQ) values so determined, 180 SATRAS spectra were simulated, each corresponding to a particular value of (νQ, ηQ). The spectral intensities were then weighted by the corresponding value P(νQ, ηQ) before all of these spectra were added to obtain the fully simulated spectrum. This simulation yields the SATRAS spectrum for the grain boundary component. Simulations were carried out individually for the grain and grain boundary contributions, and the resulting spectra were added to yield the final SATRAS spectrum for the nanomaterial. This spectrum was iteratively fitted to the experimental SATRAS spectrum with the grain to grain boundary ratio as the adjustable parameter. Results and Discussions 27 Al Solid-State NMR of 52 nm r-Al2O3. 27Al ST- and CTMAS spectra of 52 nm R-Al2O3 are shown in Figure 2. Its sheared 3Q-MAS spectrum is shown in Figure 3a. The shearing was done using the ratio R ) 19/12, as compared to R ) 3, which has been used in other studies.35,36 The same shearing procedure has been used in the DMFIT program. Furthermore, it allowed us to represent the 3Q-MAS spectra in the unified ppm scale,39 in which the δCS axis lies along the diagonal and the quadrupole-induced shift (QIS) axis has a slope of -10/17.

Using this shearing and scaling method, we were able to assess the relative contribution from δCS and δQIS by a simple graphical analysis of sheared 2D spectra. Projection of the 2D data yields the isotropic and MAS spectra, as shown in Figure 4. The isotropic spectrum displays a single aluminum resonance at 16 ppm, evidencing octahedral coordination (AlVI).40 The SATRAS spectrum exhibits numerous spinning sidebands whose envelope depicts pronounced quadrupolar “powder pattern” features which can be easily recognized in the spectrum. The manifold of the spinning sidebands arises due to the averaging of the first-order quadrupole interaction associated with the inner ((1/2 T (3/ 2) and outer ((3/2 T (5/2) satellite transitions. The sideband envelope at once evidences an axially symmetric field gradient tensor. Moreover, a high degree of spatial order within the grain structure of 52 nm R-Al2O3 gives rise to characteristic sideband intensities for the steps and shoulders of the powder spectrum. The SATRAS spectrum also exhibits second-order frequency shifts on the satellite sidebands (Figure 2 insert). The spectral simulations are shown in Figure 2. The quadrupole and chemical shielding parameters derived from “best-fit” spectral simulation of ST- and 3Q-MAS spectra are given in Table 1. Importantly, our simulations emphasize the importance of including chemical shielding anisotropy, which is often neglected. The unequal intensity for the inner satellites noticed by Scholz et al.,41 also evident in Figure 3, is attributable to a significant contribution from anisotropic chemical shielding to the observed line shape. The computer simulations yield the values of 2.40 MHz and 16 ppm for the quadrupole coupling and the isotropic chemical shift, respectively, and a value of 0.05 for ηQ. From the computer simulations, we determine the second-order quadrupole shift, measured more easily in the inner satellite sidebands in the 230-240 kHz range (Figure 2 inset), to be 13.1 Hz. This is in very good agreement with the simulated value of 13.2 Hz. Since the quadrupole coupling is not large, no second-order line shape features are noticeable in the CT-MAS or 3Q-MAS spectra. Nevertheless, as shown in Figure 3a, the sheared 3Q-MAS spectrum is satisfactorily simulated using the above-determined parameters. Values of the quadrupole and chemical shielding parameters determined for 52 nm R-alumina are very close to those reported in previous powder42 and single-crystal43,44 studies. From the 27Al observations, we therefore conclude that 52 nm R-alumina exhibits spectral characteristics which are identical to that of bulk R-alumina.

1778

J. Phys. Chem. B, Vol. 114, No. 5, 2010

Sabarinathan et al.

Figure 3. 27Al sheared 3Q-MAS spectra of 52 (a), 28 (b), 20 (c), and 12 nm (d) R-alumina. Contours from experimental and simulated spectra are shown in red and black, respectively. δCS denotes the chemical shift axis and δQIS the quadrupole-induced shift axis of slope -10/17.

Figure 4. 27Al isotropic (left) and MAS (right) projection spectra extracted from sheared 2D 3Q-MAS data of 52 (a), 28 (b), 20 (c), and 12 nm (d) R-alumina. Vertical dashed lines are drawn at the peak maxima to show the 4 ppm upfield shift for the isotropic resonance in nano R-alumina.

TABLE 1: Quadrupolar and Chemical Shift Parameters of 52 nm r-Alumina CQ (MHz)

ηQ

δCS (ppm)

reference

2.40 ( 0.031 2.38 ( 0.01 2.38 2.403 ( 0.015 2.30 ( 0.04

0.05 ( 0.001 0.00 ( 0.03

16.0 ( 0.48 16.0 ( 0.2 14 18.8 ( 0.3 7.4 ( 0.4

this work 40 42 43 44

0.009 ( 0.013 0.08

Al Solid State NMR of 28 nm, 20 and 12 nm r-Al2O3. We show in Figure 3b-d the sheared 3Q-MAS spectra of these nano R-alumina samples, and in Figure 4b-d, we show the corresponding isotropic and MAS projection spectra. All of these spectra display a single resonance along the isotropic dimension, evidencing thereby that the aluminum coordination (AlVI) remains unchanged at all grains sizes down to 12 nm. Strikingly, the isotropic projection spectra show a significant upfield shift 27

(4.1 ppm) at the peak maximum for the R-Al2O3 sample with smaller grain size. Incidentally, this shift begins to occur when the grain size is 28 nm. Despite the large width (ca. 20 ppm) of the isotropic peaks, we see a clear shift of the isotropic peak to the low-field side for the nanocrystalline R-alumina samples. A change in aluminum coordination (i.e., six to five to four, namely, octahedral to penta-coordinate to tetrahedral) is known to cause significant downfield shifts for the 27Al resonance in MAS spectra. Although aluminum is mainly octahedral in the nano R-alumina materials that we have studied, the octahedral coordination is severely perturbed in the grain boundaries, which are large (85%) (see below), to cause the observed shift. For the rationalization of the observed 3Q-MAS results, we consider that the mechanically milled nano R-alumina material is composed of grains and grain boundaries.45 The grain structure is close to the structure of the starting material (52 nm), whereas

Perturbations to

27Al

Electric Field Gradients

J. Phys. Chem. B, Vol. 114, No. 5, 2010 1779

Figure 5. 27Al experimental satellite transition MAS spectra (blue) of 28 nm R-alumina, compared with computer simulations (red) for the 50:50 (a), 40:60 (b), 30:70 (c), 25:75 (d), and 20:80 (e) grain to grain boundary ratios. Corresponding difference spectra (×10) are shown on the right.

the grain boundary structure is close to that of an amorphous phase. While the aluminum environment in the grains is wellcharacterized by a unique value for the electric field gradient tensor, akin to that observed at 52 nm, the grain boundary component is better characterized by a distribution of the field gradients. The electric field gradient distribution is modeled using the Czjzek distribution introduced in a previous section. By adapting this model, computer simulations of 3Q-MAS spectra were carried out using DMFIT. The arithmetic addition of 3Q-MAS spectra simulated for the grain and grain boundary components, in the estimated ratio, leads to the “best-fit” contours shown in Figure 3. The contours are dispersed over a range of PQ [CQ(1 + η2/3)1/2] and δiso values, and the overall span in the two-dimensional plot is satisfactorily matched by the simulations. A close inspection of the experimentally determined contour shape reveals severe perturbation to the electric field gradients as the contour is elongated along the δQIS axis, especially for the sample with a grain size of 28 nm and below. This is a consequence of the dominance of the electric field gradient distribution from grain boundaries over that from the grains. It may be noted that in the DMFIT simulations, the distribution concerns only CQ, not CQ and ηQ together, as the value for the asymmetry is held at a constant value of 0.61 in the simulation software. The small discrepancy between the experimental and simulated 3Q-MAS spectra of Figure 3 may be attributed to this limitation, which could not be overcome in our 3Q-MAS simulations. The satellite transition MAS spectra of 28 nm R-alumina is shown in Figure 5. Compared to the spectrum of the 52 nm sample (Figure 2), the loss of the sharp quadrupolar powder pattern features is at once apparent. These spectra display a featureless but asymmetric satellite sideband envelope. The distribution of electric field gradients that we have alluded to leads to the loss of these characteristic line shape features, in the same manner that they are seen in the 3Q-MAS spectra. It

may be further noticed that while there is a dramatic difference between the 52 and 28 nm samples, the SATRAS spectral pattern of R-alumina down to 12 nm is quite similar. We consider that SATRAS sideband intensities are determined by relative contributions from the grains and grain boundaries in the same manner as we have considered for the rationalization and analysis of 3Q-MAS spectra. Using the simulation procedure that we have developed, which is described in the previous section, the electric field gradient distribution was mapped uniformly for the grain boundary component in the ranges of 1.8-3.6 MHz and 0.3-0.8 MHz for CQ and ηQ, respectively. For the aluminum in the grains, the quadrupole and chemical shift parameters determined for 52 nm R-Al2O3 (CQ ) 2.40 MHz, ηQ ) 0.05, δCS ) 16.0 ppm) were used. The grain and grain boundary components were simulated and arithmetically added to generate the final SATRAS simulations. The experimental spectra were iteratively matched to the computer simulated spectra using the grain to grain boundary ratio as the fitting parameter. From the “best-fit” simulated spectra, the percentage of the grain to grain boundary ratio is found to be 75:25. It may be noted that when the field gradient distribution is confined to CQ only (ηQ ) 0.61), the spectral simulations lead to an overestimate of the grain boundary component by 3%, significant to alter the SATRAS spectrum. It may be noted that an accurate determination of the grain boundary contribution cannot be made from 3Q-MAS spectral simulations alone since CQ and ηQ were not jointly sampled. Moreover, 3Q-MAS experiments suffer from nonuniform excitation of triple quantum coherences, as well as their conversion of the observable single-quantum central transition frequencies,46 across a range of quadrupole couplings which are inherent at these small grain sizes. Hence, our SATRAS experimental finding may be considered significant as they serve to bring out the dominance of the grain boundaries and their estimation to within 2%. Simulations of the other R-alumina

1780

J. Phys. Chem. B, Vol. 114, No. 5, 2010

TABLE 2:

Sabarinathan et al.

27

Al Electric Field Gradient Distribution Parameters Determined from SATRAS and 3Q-MAS Spectral Analysis SATRAS CQ (MHz)

a

3Q-MAS %

ηQ

CQ (MHz)

b

amplitude %

ηQ

grain size (nm)

G

GB

G

GB

G

GB

G

GB

G

GB

G

GB

52 28-12

2.40 2.40

1.8 - 3.6 1.8 - 3.6

0.05 0.05

0.3 - 0.8 0.3 - 0.8

85 25

15 75

2.40 2.40

3.65 3.65

0.05 0.05

0.61 0.61

72 27

28 73

a Denotes the quadrupole coupling (MHz) for the grain (G) and the range spanned in the Czjzek distribution for the grain boundary (GB) with σ ) 3.65 MHz; 180 sampling points having CQ and ηQ values in the range shown were taken for the simulation of the GB component. b Here, CQ is indicated as the quadrupole coupling (MHz) for the grain (G) and as the width of the CQ distribution for the grain boundary (GB).

Figure 6. Comparison experimental (left) and “best-fit” simulated (right) 27Al satellite transition MAS spectra of 52 (a), 28 (b), 20 (c), and 12 nm (d) R-alumina.

samples with grain sizes of 20 and 12 nm show that the grain to grain boundary ratio remains the same in these samples as well. The dominance of the grain boundary component can be contrasted to the 52 nm R-Al2O3, where grain component clearly dominants. The computer simulated SATRAS spectra, using the “best-fit values given in Table 2, are shown along with the experimental spectra in Figure 6. Finally, it may be noted that a high contribution from grain boundaries (ca. 80%) was estimated for 20 nm R-alumina in a previous study.23 Our SATRAS spectral analysis has enabled us to make a more accurate estimate for the grain and grain boundary components down to 12 nm. It is clear that surface effects of the grain boundary play a major role compared to the asymmetry effect present in the bulk. We wish to point out that for the nano R-alumina, the distribution in the grain size may be considered to be small and confined within (2 nm of the average grain size that has been determined from powder XRD spectra. This is much smaller than the contribution from grain boundaries, which far outweighs the electric filed gradients contribution from the grains contributing to the observed 27Al spectral intensities. We therefore believe that our conclusions will remain unaltered and our estimates for the two components are not in serious error.

Although our 27Al solid-state NMR studies show that the Al environment remains in tact in the octahedral coordination, it is likely that further reduction of the grain size (below 12 nm) in R-alumina may possibly lead to a lower coordination for Al in the nanomaterials.47 However, such sub-nanoscale R-alumina was difficult to prepare by the methods that we have used. Conclusions Nanocrystalline R-alumina was prepared by thermal decomposition of synthetically prepared Al(OH)3 at 1200 °C followed by high-energy ball milling. The grain size was controlled by the milling time. 27Al solid-state NMR results show that the aluminum environment at 52 nm is the same as that in bulk R-alumina, characterized by a unique value for the field gradient tensor exhibiting axial symmetry. At low grain sizes, the aluminum sites held in the grain boundaries are seen to suffer a large perturbation to the electric field gradients. This is shown to lead to a distribution of electric field gradients. Graphical analysis and DMFIT simulations of 3Q-MAS spectra evidence a large grain boundary contribution to the observed spectra of 28, 20, and 12 nm R-alumina. SATRAS spectral analysis has enabled quantitative determination of the relative proportion of

Perturbations to

27Al

Electric Field Gradients

grain and grain boundary components together with the relevant electric field gradient distribution parameters. 27Al solid-state NMR results also show that the grain boundary component is already quite high (75%) at 28 nm, and it remains unchanged as the grain size is lowered to 12 nm. It is likely that at much smaller grain sizes, lattice softening11 could further change the surface nature in the grain boundaries, possibly leading to a change in aluminum coordination. Acknowledgment. We thank the Department of Science and Technology, New Delhi, (Scheme No. SR/S5/NM-58/2002) for financial support and the Emeritus Scientist Scheme (VS and SG:21(0701)/07/EMR-II; SR: SR: 21(0714)/08/EMR-II) Council of Scientific and Industrial Research, New Delhi. Supporting Information Available: Powder X-ray diffraction spectra, Transmission Electron Micrographs, ball milling data, and plots of Czjzek distribution. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Souza Santos, P.; Souza Santos, H.; Toledo, S. P. Mater. Res. 2000, 3 (4), 104–114. (2) Wefers, K.; Misra, C. Oxides and Hydroxides of Aluminum; ALCOA Laboratories: Pittsburgh, PA, 1987; p 20. (3) Wefers, K. In Alumina Chemicals; Hart, L. D., Ed.; Wiley: New York, 1990; p 13. (4) Ollivier, B.; Retoux, R.; Lacorre, P.; Massiot, D.; Ferey, G. J. Mater. Chem. 1997, 7, 1049–1056. (5) Kao, H.-C.; Wei, W.-C. J. Am. Ceram. Soc. 2000, 83 (2), 362– 368. (6) Rankin, G. A.; Merwin, H. E. J. Am. Chem. Soc. 1916, 38, 568– 588. (7) Ishizawa, N.; Miyata, T.; Minato, I.; Marumo, F.; Iwai, S. Acta Crystallogr. 1980, B36, 228–230. (8) Oberlander, R. K. In Applied Industrial Catalysis; Leach, B. E., Ed.; Academic: Orlando, FL, 1984; Vol. 3, p 63. (9) Kolar, D. Pure Appl. Chem. 2000, 72 (8), 1425–1448. (10) Maier, J. Nat. Mater. 2005, 4, 805–815. (11) Wagemaker, M.; Borghols, W. J. H.; Mulder, F. M. J. Am. Chem. Soc. 2007, 129, 4323–4327. (12) Wagemaker, M.; Mulder, F. M.; Van der Ven, A. AdV. Mater. 2009, 21, 1–7. (13) Hale, J. M.; Aurox, A.; Perotta, A. J.; Navrotsky, A. Science 1997, 277, 788–791. (14) McHale, J. M.; Navrotsky, A.; Perrotta, A. J. J. Phys. Chem. B 1997, 101, 603–613. (15) Phillpot, S. R.; Wolf, D.; Gleiter, H. J. Appl. Phys. 1995, 78, 847– 861. (16) Schiø´tz, J.; Ditolla, F. D.; Jacobsen, K. W. Nature 1998, 391, 561– 563. (17) Palosz, B.; Stelmakh, S.; Grzanka, E.; Gierlotka, S.; Palosz, W. Kristallographie 2007, 222, 580–594.

J. Phys. Chem. B, Vol. 114, No. 5, 2010 1781 (18) Molina, R.; Poncelet, G. J. Phys. Chem. B 1999, 103 (51), 11290– 11296. (19) Huang, C.-L.; Wang, J.-J.; Huang, C.-Y. Mater. Lett. 2005, 59, 3946. (20) Martinez-Huerta, M. V.; Gao, X.; Tian, H.; Wachs, I. E.; Fierro, J. L. G.; Banares, M. A. Catal. Today 2006, 118 (3-4), 279–287. (21) MacKenzie, K. J. D.; Smith, M. E. Multinuclear solid-state NMR of inorganic materials; Pergamon: Amsterdam, The Netherlands, 2002. (22) Skibsted, J.; Chr, N.; Bildsoe, N.; Jakobsen, H. J. J. Agric. Res. 1991, 95, 88–117. (23) Pecharroman, C.; Sobrados, I.; Iglesias, J. E.; G-Carreno, T.; Sanz, J. J. Phys. Chem. B 1999, 103, 6160–6170. (24) Frydman, L.; Harwood, J. S. J. Am. Chem. Soc. 1995, 117, 5367– 5368. (25) Jakobsen, H. J.; Skibsted, J.; Bildsbe, H.; Nielsen, N. C. J. Magn. Reson. 1989, 85, 173–180. (26) Zielinski, P. A.; Schulz, R.; Kaliaguine, S.; Van Neste, A. J. Mater. Res. 1993, 8 (11), 2985–2992. (27) Cullity, B. D. Elements of X-ray Diffraction; Addison-Wesley: Reading, MA, 1977; p 81. (28) Nielsen, N. C.; Bildsøe, H.; Jakobsen, H. J. J. Magn. Reson. 1992, 98, 665–673. (29) Amoureux, J. P.; Fernandez, C.; Steuernagel, S. J. Magn. Reson., Ser. A 1996, 123, 116–118. (30) States, D. J.; Haberkorn, R. A.; Ruben, D. J. J. Magn. Reson. 1982, 48, 286–292. (31) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve`, S.; Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Magn. Reson. Chem. 2002, 40, 70–76. (32) Czjzek, G.; Fink, J.; Gotz, F.; Schmidt, H. Phys. ReV. B 1981, 23 (6), 2513–2530. (33) Hung, I.; Trébosc, J.; Hoatson, G. L.; Vold, R. L.; Amoureux, J.P. J. Magn. Reson. 2009, 201 (1), 81–86. (34) Czjzek, G. Phys. ReV. B 1982, 25 (7), 4908–4910. (35) Vijayakumar, M.; Hoatson, G. L.; Vold, R. L. Phys. ReV. 2007, B75, 104104. (36) Vold, R. L.; Hoatson, G. L.; Vijayakumar, M. Phys. ReV. 2007, B75, 134105. (37) Neuville, D. R.; Cormier, L.; Massiot, D. Geochim. Cosmochim. Acta 2004, 68 (24), 5071–5079. (38) Body, M.; Legein, C.; Buzar, J.-Y.; Silly, G. Eur. J. Inorg. Chem. 2007, 1980–1988. (39) (a) Amoureux, J. P.; Fernandez, C. Solid State NMR 1998, 10, 211– 223. (b) Amoureux, J. P.; Huguenard, C.; Engelke, F.; Taulelle, F. Chem. Phys. Lett. 2002, 356, 497–504. (40) Skibsted, J.; Chr, N.; Bildsoe, N.; Jakobsen, H. J. J. Magn. Reson. 1991, 95, 88–117. (41) Scholz, G.; Staˆsser, R.; Klein, J.; Silly, G.; Buzare, J. Y.; Laligant, Y.; Ziemer, B. J. Phys.: Condens. Matter 2002, 14, 2101–2117. (42) Kraus, H.; Prins, R.; Kentgens, A. P. M. J. Phys. Chem. 1996, 100, 16336–16345. (43) Woo, A. J. Bull. Korean Chem. Soc. 1999, 20 (10), 1205–1208. (44) Le Caer, G.; Cadogan, J. M.; Brand, R. A.; Dubois, J. M.; Guntherodt, H. J. J. Phys. F: Met. Phys. 1984, 14, L73–L78. (45) Birringer, R. Mater. Sci. Eng., A 1989, 117, 33–43. (46) Amoureux, J.-P.; Fernandez, C. Solid State NMR 1998, 10, 211– 223. (47) Sreeja, V.; Smitha, T. S.; Deepak, A.; Ajithkumar, T. G.; Joy, P. A. J. Phys. Chem. C 2008, 112, 14737–14744.

JP907469N