Structural Role of Fluoride in Aluminophosphate SolâGel Glasses

10402

J. Phys. Chem. B 2007, 111, 10402-10412

Structural Role of Fluoride in Aluminophosphate Sol-Gel Glasses: High-Resolution Double-Resonance NMR Studies Long Zhang,†,‡ Carla C. de Araujo,† and Hellmut Eckert*,† Institut fu¨r Physikalische Chemie, Westfa¨lische Wilhelms-UniVersita¨t Mu¨nster, Corrensstrasse 30, D-48149 Mu¨nster, Germany, and Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, P.O. Box 800-211, 201800 Shanghai, China ReceiVed: April 6, 2007; In Final Form: June 29, 2007

The local structure of Na-Al-P-O-F glasses, prepared by a novel sol-gel route, was extensively investigated by advanced solid-state NMR techniques. 27Al{19F} rotational echo double resonance (REDOR) results indicate that the F incorporated into aluminophosphate glass is preferentially bonded to octahedral Al units and results in a significant increase in the concentration of six-coordinated aluminum. The extent of Al-F and Al-O-P connectivities are quantified consistently by analyzing 27Al{31P} and 27Al{19F} REDOR NMR data. Two distinct types of fluorine species were identified and characterized by various 19F{27Al}, 19F{23Na}, and 19F{31P} double resonance experiments, which were able to support peak assignments to bridging (Al-FAl, -140 ppm) and terminal (Al-F, -170 ppm) units. On the basis of the detailed quantitative dipoledipole coupling information obtained, a comprehensive structural model for these glasses is presented, detailing the structural speciation as a function of composition.

1. Introduction Fluoride phosphate glasses combine the favorable optical properties of pure fluoride glasses with good mechanical and thermal stabilities of analogous oxidic systems. Such characteristics include a strong glass-forming tendency,1,2 the possibility of incorporating large concentrations of transition metal and rare earth ions,3-5 low phonon energies,6 a wide optical transparency window (from UV to near IR),7 low linear and nonlinear refractive indices,8,9 and tailorable spectroscopic properties by varying the phosphate content.10 As a result, fluoride phosphate glasses are attractive candidates for many applications, including optical fibers,11 amplifiers and lasers,12 optical frequency upconverters,13 as well as optical limiters.14 Among these glasses, aluminum fluoride phosphate glasses are particularly promising as they combine these optical properties with excellent chemical durability and mechanical strength.2,15-17 The goal of tailoring the particular properties of these attractive glasses requires a comprehensive understanding of their structural organization. Previous structural investigations are based on the interpretation of Raman/IR or standard magic-angle-spinning (MAS) NMR results.18-22 However, vibrational band assignments are quite ambiguous because of the overlapping components from phosphorus and aluminum polyhedra in different coordination states.18-20 While ordinary 27Al MAS NMR spectroscopy is able to distinguish between aluminum in four-, five-, and six-coordination, it is unable to provide an unambiguous distinction between oxygen and fluorine coordination. Likewise 31P chemical shifts are unable to differentiate between phosphorus-fluorine and phosphorus-oxygen bonding. On the basis of the limited power of these standard structural characterization techniques, a number of conflicting suggestions have been made concerning the structure of fluoride aluminophos* Author to whom correspondence should be addressed. Phone: 49-2518329161. Fax: 49-251-8329159. E-mail: [email protected]. † Westfa ¨ lische Wilhelms-Universita¨t Mu¨nster. ‡ Chinese Academy of Sciences.

phate glasses.17-20 For example, Gan19b and Videau19a proposed the formation of [AlF4] units crosslinking between polyphosphate chains. Other authors assume that F is predominantly attached to Al(VI) units.18,20 On the basis of X-ray photoelectron spectroscopic data, Brow et al. 18c propose that F preferentially forms F-Al bonds until each Al site is bound to a maximum of three F atoms. Fluoride incorporation beyond this limit proceeds via the formation of F-P bonds. Solid-state nuclear magnetic resonance (NMR) techniques belong to the most powerful characterization tools for disordered materials, due to their well-proven ability to provide a wealth of local structural information.23-25 In particular, advanced techniques such as rotational echo double resonance (REDOR),26-30 rotational echo adiabatic passage double resonance (REAPDOR)31,32 and heteronuclear correlation (HETCOR),33-35 which utilize through-space internuclear dipole-dipole coupling, have provided quantitative information on structural connectivity and spatial proximity beyond the first coordination sphere in glasses.25 In this contribution, we use these techniques, for the first time, to investigate comprehensively the structure of NaAl-P-O-F glasses, prepared by a novel sol-gel method recently developed in our laboratory.36 On the basis of the detailed quantitative dipole-dipole coupling information obtained, we present a comprehensive structural model for these glasses. 2. NMR Methodology The experimental strategy used in the present manuscript includes the quantitative analysis of high-resolution 19F, 23Na, 27Al, and 31P NMR solid-state NMR spectra. However, a key part of this study is based on the quantitative analysis of internuclear magnetic dipole-dipole interactions. To this end, we have previously developed the REDOR technique into an experimental tool for site resolved heteronuclear dipolar coupling measurements in glasses.37-39 Figure 1 shows the two S{I} REDOR pulse sequences used in the present study.27,29 A

10.1021/jp072725w CCC: $37.00 © 2007 American Chemical Society Published on Web 08/15/2007

Fluoride in Aluminophosphate Sol-Gel Glasses

J. Phys. Chem. B, Vol. 111, No. 35, 2007 10403

Figure 1. NMR pulse sequences used in the present study: (a) sequence used for S{I} REDOR, I ) 1/2; (b) sequence used for S{I} REDOR, I ) 3/2; (c) sequence used for S{I} REAPDOR, I ) 3/2 and 5/2; (d) sequence used for CPMAS-HETCOR.

normalized difference signal ∆S/S0 ) (S0 - S)/S0 is measured in the absence (intensity S0) and the presence (intensity S) of the dipolar interactions of the observed nuclei S with the second nuclear species I. Determination of ∆S/S0 under systematic variation of the number of rotor cycles N yields the so-called REDOR curve, in which the data are plotted as a function of dipolar evolution time NTr (Tr being the duration of one rotor period). For isolated pairs of spin-1/2 nuclei, these curves possess a universal shape, allowing the straightforward determination of the magnetic dipole-dipole coupling constant.26,29 In contrast, the analysis of S{I} REDOR curves in inorganic glasses is complicated by multispin interactions, distance distributions, and interference by nuclear electric quadrupolar couplings. We have previously shown that in the case of S{I} REDOR experiments with I ) 1/2 nuclei, the problem can be simplified by confining the REDOR data analysis to the initial curvature, where ∆S/S0 e 0.2.38,39 In this limit, the REDOR curve is found to be independent of specific spin system geometries and can be approximated by a simple parabola:

4 SI ∆S ) M (NTr)2 S0 3π2 2

(1)

The curvature of this parabola is closely related to the vanVleck moment MSI 2 ) M2(S{I}), a quantity that can be used to characterize the average dipole-dipole coupling strengths that the S nuclei experience from the magnetic moments of their neighboring I nuclei. The approach yields satisfactory results also in amorphous and strongly disordered systems where the order and geometry of the spin systems is unknown and possibly ill-defined. In those cases where the dipolar dephasing of the observed spins occurs in the local field of I > 1/2 nuclei such as 23Na (I

) 3/2), several additional complications enter. First of all, the different possible Zeeman states mI for the I nuclei differ in the respective sizes of their z components and hence generate dipolar fields of different magnitudes at the observed spins.37,40,41 Second, for strong nuclear electric quadrupolar coupling, the anisotropic broadening of the |1/2> T |3/2> “satellite transitions” produces large resonance offsets, which reduce the efficiency of π pulses in causing population inversion. In the limit of very large first-order quadrupolar splitting (rf nutation frequency ν1 , CQ, the quadrupolar coupling constant), pulses applied to the I nuclei in the REDOR sequences will affect only the central |1/2> T |-1/2> coherences. In this regime, only those S spins that are coupled to I nuclei in Zeeman states with |mI | ) 1/2 are expected to yield a REDOR response. Detailed simulations have led to the conclusion that it is desirable in such cases to minimize the number of π pulses applied to the quadrupolar nuclei,37 making the REDOR sequence of Figure 1b the method of choice. As we have previously shown,37 the initial curvature analysis discussed above can be extended to systems containing I ) 3/2 nuclei by using the expression

∆S 1 2 ) (2 + 18f1)MSI 2 (NTr) S0 15π2

(2)

where the efficiency factor f1 (0 e f1 e 1) accounts for the extent to which the dipolar coupling of S spins to I spins in their outer Zeeman states still influences the REDOR response. Again, eq 2 is valid for the initial regime (0 e ∆S/S0 e 0.2) only. The whole data analysis procedure can thus be summarized as follows. On the basis of the nuclear electric quadrupolar coupling Hamiltonion parameters determined experimentally, a universal REDOR curve is computed (using the SIMPSON

10404 J. Phys. Chem. B, Vol. 111, No. 35, 2007

Zhang et al.

TABLE 1: Experimental Conditions Used in the Double Resonance NMR Experiments combination 27

19

Al{ F} REDOR Al{31P}REDOR 19 F {31P} REDOR 19 F{27Al}REAPDOR 19 F{23Na}REDOR 31 P{23Na} REDOR 27

ν0(S)a (MHz)

ν0(I)b (MHz)

τ90°(S)c (µs)

τ90°(I)d (µs)

130.3 104.3 470.4 376.6 470.4 162.0

470.4 162.0 202.5 104.3 132.3 105.9

2.5 3.0 2.2 2.2 2.2 3.0

2.2 3.0 3.0 4.0 1.8 3.0

τa (I)e (µs)

20

Rf (s)

NSg

2 2 10 10 10 60

128-512 256 128-256 128-256 128-256 64

a S-spin resonance frequency. b I-spin resonance frequency. c S-spin-90° pulse length. d I-spin-90° pulse length. e Adiabatic passage mixing time in REAPDOR. f Relaxation delay. g Number of scans.

code42) using explicitly those experimental conditions under which the REDOR data were taken. This simulation curve is fitted to a parabola (eq 2), resulting in the appropriate f1 value, which is then applicable for the analysis of the experimental data set. With the use of the f1 value determined in this fashion, the experimental data are fitted to eq 2, resulting in an experimental second moment. This value can then be compared with calculations based on structural models using the wellknown van Vleck equation:43

MSI 2

)

() µ0

2

4

4π 15

I(I + 1)p2γS2γI2

∑S rIS-6

(3)

where γI and γS are the gyromagnetic ratios of the nuclei I and S involved, and rIS is the internuclear distance. In the present study, we will use this approach to analyze the dipolar field created by 23Na at the observed 31P and 19F nuclei. The sequence of Figure 1b is just a special case of the more general REAPDOR (rotational echo adiabatic passage double resonance) experiment (Figure 1c), in which the central pulse applied in the middle of the dipolar evolution period creates adiabatic mixing of the various I-Zeeman states under the influence of MAS. In the case of REAPDOR, the optimum pulse-length for generating a maximum difference signal amounts to one-third of the rotor period.44 In the present application, REAPDOR is used for qualitative studies of the local dipolar fields created by the spin-5/2 27Al nuclei at the 19F sites. 3. Experimental Section 3.1. Materials and Methods. Glass samples were prepared using a novel sol-gel procedure recently developed in our laboratory.36 Aluminum lactate (98%, Fluka), disodium fluorophosphate (Na2PO3F) (98%, Sigma-Aldrich), H2PO3F (4 M, freshly prepared from 70% H2PO3F solution, Sigma-Aldrich), and solid H3PO4 (98%, Fluka) were used as precursors. The pH value of the precursor mixture was adjusted within the range of 2-4, using nitric acid (1 M, puriss. p.a., Fluka) and ammonia (1 M, diluted from concentrated ammonia solution, Aldrich) solutions and controlled within 0.01 units by a pH meter (WTW pH 320, Germany). In a typical preparation, 0.004 mol (1.176 g) of aluminum lactate was dissolved in 10 mL of distilled water, followed by the addition of the appropriate amounts of Na2PO3F (and H2PO3F), H3PO4, and/or sodium acetate (NaAc) to reach the desired compositions. Upon air-drying at a temperature between ambient and 50 °C, transparent colorless Na-Al-PO-F xerogels were formed. After annealing the gel-samples at a temperature around 400 °C for 4 h in a Heraeus muffle furnace, transparent homogeneous glasses were obtained. Their noncrystalline state was confirmed by the absence of any sharp X-ray powder diffraction peaks (Guinier method, Cu KR1 radiation).

3.2. NMR Studies. 27Al, 23Na, 31P, and 19F single resonance MAS NMR spectra were recorded at resonance frequencies of 130.3 (27Al), 132.3 (23Na), 202.5 (31P), and 470.4 MHz (19F) on a Bruker DSX-500 spectrometer. 23Na, 27Al, and 31P measurements were conducted with a 4 mm MAS NMR probe with MAS rotation frequencies between 10 and 15 kHz, while the 19F measurements were taken with a 2.5 mm MAS NMR probe at a MAS rotation frequency of 25-30 kHz. Typical acquisition parameters for single-pulse measurements were pulse length 5.0 µs (90° flip angle) for 31P, 1.0 µs (30° flip angle) for 27Al and 23Na, and 2.0 µs (90° flip angle) for 19F. Recycle delays were chosen at 90, 1, 1, and 10 s for 31P, 27Al, 23Na, and 19F, respectively. Chemical shifts are referenced to a 1 M Al(NO3)3 aqueous solution, a 1 M NaCl aqueous solution, and a 85% H3PO4 and CFCl3 solution, respectively. 27Al{31P} and 31P{23Na} REDOR, as well as 19F{27Al} REAPDOR experiments, were carried out on a Bruker DSX400 spectrometer, using a 4 mm Bruker triple resonance probe at spinning frequencies of 10-15 kHz. All the other REDOR experiments involving 19F nuclei utilized a 2.5 mm Bruker X-19F/1H double resonance probe in a Bruker DSX-500 spectrometer at spinning frequencies within the range 15-35 kHz. Table 1 summarizes the specific conditions used in the present study. All the S{I ) 1/2} REDOR experiments were measured with the pulse sequence of Figure 1a. In the 27Al{31P} REDOR measurements, a compensation scheme described in ref 39 was used to correct for the effect of experimental imperfections. 31P{23Na } and 19F{23Na} REDOR measurements were conducted with the pulse sequence of Figure 1b. The π-pulses applied to the spin-1/2 nuclei were phase cycled according to the XY-4 scheme.28 19F{27Al} REAPDOR measurements were performed using the pulse sequence of Figure 1c. To increase the number of data points for the parabolic fit used for the M2 determination, some REDOR curves were measured at two or three different spinning speeds. 19F{23Na} CPMAS NMR experiments were carried out on a Bruker DSX-500 spectrometer, using a 2.5 mm X-19F/1H double probe at a spinning frequency of 25 kHz. The pulse sequence for a two-dimensional version of the CPMAS experiment is sketched in Figure 1d. For the experiments in this study, the CP matching condition was optimized by maintaining the 23Na spin-lock pulse power constant and varying the 19F pulse power. The nutation behavior was carefully characterized in order to determine the rf field strengths at each Hartmann-Hahn matching condition. Specifically, the nutation frequencies for the 23Na spin-lock and the contact pulse of 19F were 6.3 and 31.3 kHz, respectively. A contact time of 4.5 ms was used. 4. Results, Data Analysis, and Interpretation 4.1 Single-Pulse 27Al, 19F, and 31P MAS NMR. Figures 2-5 summarize the 27Al (left), 19F (middle), and 31P (right) MAS NMR spectra of the sol-gel derived fluoride aluminophosphate glasses along the composition lines Na/Al/P/F 2:x:2:1, 2:2:x:1,



Figure 2. 130.3 MHz 27Al (left), 470.4 MHz 19F (middle), and 202.5 MHz 31P (right) NMR spectra of the Na-Al-P-O-F glasses along the composition line Na/Al/P/F ) 2:x:2:1. Dashed curves indicate spectra of corresponding fluoride-free glass. Spinning sidebands are indicated with asterisks.

Figure 3. 130.3 MHz 27Al (left), 470.4 MHz 19F (middle), and 202.5 MHz 31P (right) NMR spectra of the Na-Al-P-O-F glasses along the composition line Na/Al/P/F ) 2:2:x:1. Spinning sidebands are indicated with asterisks.

Figure 4. 130.3 MHz 27Al (left), 470.4 MHz 19F (middle), and 202.5 MHz 31P (right) NMR spectra of the Na-Al-P-O-F glasses along the composition line Na/Al/P/F ) 2:2:2:x. Spinning sidebands are indicated with asterisks.

2:2:2:x, and x:2:2:x, respectively. At the external magnetic field strength of 11.7 T, the large majority of the 27Al NMR spectra are characterized by the presence of three asymmetric resonances near 41, 12, and -14 ppm (at 11.7 T), which are assigned in the usual manner to Al(IV), Al(V), and Al(VI) environments, respectively. Although the lineshapes reveal the influence of second-order quadrupolar perturbations, approximate peak area ratios could be obtained by fitting these spectra to superpositions of three Gauss-Lorentz components (see Table 2). For a few glasses, precise isotropic chemical shift and second-order

quadrupolar effect (SOQE) information were obtained by analyzing triple-quantum MAS NMR45 spectra (data not shown). Figures 2-5 illustrate that the glass composition exerts a large influence upon the aluminum speciation. First of all, the trends evident in Figures 2 and 3 illustrate that for all glasses with P/Al ratios of unity or larger, the relative concentration of Al(VI) units increases systematically with increasing P/Al ratio. This latter effect was already noted in fluoride-free sodium aluminophosphate glasses46-48 and seems to apply equally to fluoride containing glasses. In contrast, for the phosphorus-poor


Zhang et al.

Figure 5. 130.3 MHz 27Al (left), 470.4 MHz 19F (middle), and 202.5 MHz 31P (right) NMR spectra of the Na-Al-P-O-F glasses along the composition line Na/Al/P/F ) x:2:2:x. Spinning sidebands are indicated with asterisks.

TABLE 2:

27Al

NMR Parameters of the Na-Al-P-O-F Glasses under Studya AlO4

Na/Al/P/F 2:2:1:1 2:1:2:1 2:2:4:1 2:2:2:0.5 2:2:2:1 2:2:2:2 1.5:2:2:1.5 1:2:2:1 0.5:2:2:0.5 2:2:2:0 2:1:2:0

δiso/(δc)

(ppm)

(55.5) (41.1) (40.1) (44.9) (43.8) 47.4/(44.7) (41.5) 44.2/(40.1) (39.4) 51.2/(46.0) 50.1/(44.7)

SOQEb

AlO5 (MHz)

n.m.d n.m.d n.m.d n.m.d n.m.d 3.2 n.m.d 3.4 n.m.d 3.3 3.9

Pc (%) 19 6 19 50 39 26 31 45 52 51 51

δiso/(δc)

(ppm)

(28.8) (10.6) (8.9) (12.2) (12.1) 16.7/(11.7) (10.7) 14.4/(11.2) (11.7) 24.5/(16.5) 19.1/(13.9)

SOQEb (MHz) n.m.d n.m.d n.m.d n.m.d n.m.d 3.1 n.m.d 3.5 n.m.d 3.5 3.9

AlO6 Pc (%) 33 9 15 13 13 15 10 16 13 24 16

δiso/(δc)

(ppm)

-4.5 (-13.9) (-14.6) (-14.7) (-13.5) -9.7/(-13.7) (-11.5) -11.0/(-14.0) (-12.9) -4.8/(-10.3) -10.5/(-14.0)

SOQEb (MHz)

Pc (%)

n.m.d

48 85 66 37 48 59 59 39 35 25 33

n.m.d n.m.d n.m.d n.m.d 2.2 n.m.d 2.8 n.m.d 3.0 2.8

aThe errors for parameters δisο (δ), SOQE, and relative ratio (P) were less than (0.3 ppm, (0.1 MHz, and (2%. b SOQE (defined as C (1 + Q η2/3)1/2) calculated from the center of gravity of the measured 27Al MQ-MAS spectra. c Calculated from the simulation by DMFIT.49 δc peak position at 11.7 T. d n.m.: not measured

glasses with compositions Na/Al/P/F ) 2:2:2:0 and 2:2:1:1, significantly more positive 27Al resonance shifts are observed for the three distinct Al coordination states, indicating that P-O-Al linkages do not constitute major structural features in these glasses. Figures 2, 4, and 5 illustrate further that in glasses having a constant P/Al ratio, the fraction of six-coordinate aluminum is systematically increased as a function of increasing F/Al ratio. This compositional trend may suggest that the fluoride incorporated into the structure of these glasses is preferentially coordinated to the Al(VI) sites. This suggestion is further supported by various double resonance NMR experiments to be discussed further below. The 19F MAS NMR spectra are characterized by three resonances near -140, -170, and -195 ppm (see Table 3 for the relative percentages). On the basis of qualitative REDOR studies reported previously,36 the -140 ppm signal has been assigned to an F species bridging between two Al-polyhedra, while the -170 ppm signal is attributed to a terminal Al-bound fluoride. The ratio of bridging to nonbridging fluoride species appears to be sensitively influenced by the glass composition.

In addition, glasses along the series Na/Al/P/F ) x:2:2:x show a third resonance at -195 ppm, the assignment of which is not clear at the present time. The near-absence of a 19F NMR signal near -75 ppm indicates that almost no phosphorus-fluorine bonds are formed in these glasses. A very weak F-P peak in 19F spectra around -75 ppm is only seen in the 2:2:4:1 sample, (the one with the largest P/F ratio). The 31P MAS NMR spectra show broad unresolved signals, characterized by substantial compositional variations of their average chemical shifts (see Table 3). As shown by our previous studies, these chemical shifts are influenced by the extent of both P-O-Al and P-O-P linkage formation: denoting the local phosphorus environment by the symbol Q(n)mAl the resonance is shifted toward lower frequencies with increasing values of either n or m or both.46-48 In contrast, the replacement of P-O-Al linkages by P-ONa+ units containing nonbridging oxygen atoms results in resonance displacements toward more positive chemical shifts. The latter environment appears to dominate particularly in the glass with the highest Na/P ratio (composition Na/Al/P/F ) 2:2: 1:1). The same trend is observed with increasing x along the compositional series x:2:2:x (see Figure 5). This series can be



Figure 6. 27Al{19F} REDOR studies carried out on three representative glasses. In each individual trace, the upper and bottom curves represent the 27Al signals without and with 19F irradiation during the evolution period of 0.267 ms, respectively. The right shows the REDOR dephasing curves of these glasses as well as of crystalline AlF3 and an oxide-free fluoride glass. Solid, dashed, and dotted curves are parabolic fits to the data of AlF3, the glass with Na/Al/P/F ) 2:2:2:2, and the glass with Na/Al/P/F ) 2:2:2:1, respectively.

TABLE 3:

19F

and 31P NMR Parameters of the Na-Al-P-O-F Glasses under Studya 19F-peak

Na/Al/P/F 2:2:1:1 2:1:2:1 2:2:4:1 2:2:2:0.5 2:2:2:1 2:2:2:2 1.5:2:2:1.5 1:2:2:1 0.5:2:2:0.5 2:2:2:0 2:1:2:0 c

Pb

(%)

16 22 86 36 41 40 55 42 48

19F-peak

1

δiso

(ppm)c

-140.1 -139.1 -139.0 -139.2 -139.7 -140.9 -139.3 -138.9 -138.8

Pb

(%)

84 78 14 64 59 57 40 31 24

19F-peak

2

δiso

(ppm)c

-171.1 -169.5 -169.9 -168.3 -170.7 -171.1 -171.2 -170.2 -169.3

Pb

(%)

3 5 27 28

3

δiso

31

(ppm)c

-193.6 -193.8 -195.1 -194.1

δiso

P (ppm)c

-8.9 -16.5 -21.2 -14.5 -15.2 -12.5 -15.5 -20.1 -22.6 -10.2 -12.5

aThe errors for parameters δisο (δ) and relative ratio (P) were less than (0.3 ppm and (2%. b Calculated from the simulation by DMFIT.49 Center of gravity.

viewed to reflect the successive modification of AlPO4 glass by the network modifier sodium fluoride. 4.2. 27Al{19F} and 27Al{31P} REDOR. The pronounced influence of the F/Al ratio on the relative fraction of sixcoordinated aluminum evident in Figures 2-5 suggests that the fluorine species are preferentially linked to octahedral aluminum, representing mixed Al(F,OP)6 local environments. This hypothesis has been further explored by 27Al{19F} REDOR experiments (see Figure 6, left). Clearly, the signals attributed to Al(IV) species for all these glasses are virtually unaffected by the 19F REDOR irradiation, indicating that the fluorine atoms are remote from Al(IV). In contrast, the signals assigned to Al(VI) units are significantly attenuated by the 19F irradiation, confirming that some Al(VI) units are in close proximity to F. On the basis of these results, and the documented absence of P-F bonding, we can conclude that the incorporated fluorine in these solgel derived glasses are almost exclusively coordinated to octahedral aluminum sites (a small amount of fluorine might also be ligated to pentahedral units). Figure 6 (right) further summarizes the 27Al(VI){19F} REDOR curves of some representative glasses, from which experimental second moment values M2(27Al{19F}) are derived using eq 1. Since the strong

dipolar interactions produce rather steep REDOR curves, these second moment values are subject to large experimental errors. As Table 4 indicates, this situation leads to a significant underestimation of the M2(27Al{19F}) value for the crystalline model compound AlF3; the experimental value is only 60% of that expected from the crystal structure. This experimental scaling factor will be accounted for when drawing the chemical conclusions on the glass structure as discussed below. Table 4 indicates that the M2(27Al{19F}) values increase with increasing F/Al ratio, suggesting that the average number of F atoms bonded to the Al(VI) units increases accordingly. Figure 7 shows site-resolved 27Al{19F} REDOR data on a glass with composition Na/Al/P/F ) 1:2:2:1, suggesting that in this particular sample, a small amount of of the Al(V) units may also be bonded to fluoride. All of the above results are consistent with complementary 27Al{31P} REDOR experiments conducted on these glasses. Figure 8 shows the site-resolved 27Al{31P} REDOR curves of Na/Al/P/F ) 2:1:2:1 glass, as well as that of its F-free analogue (Na/Al/P/F ) 2:1:2:0). While essentially no difference is observed for the Al(IV) units, the REDOR curve pertaining to the Al(VI) units indicate a significant weakening of the 27Al-

10408 J. Phys. Chem. B, Vol. 111, No. 35, 2007 TABLE 4: Second Moment Values M2 (27Al{31P}) and M2 (27Al{19F}) Which Are Deduced from 27Al{31P} and 27Al{19F} REDOR Dephasing Curves for Al(IV) and Al(VI) Unitsa Al(VI)

sample AlF3 Al(PO3)3 2:1:2:0 2:1:2:1 2:2:2:0.5 2:2:2:1 2:2:2:2 2:2:4:1 1:2:2:1

Al(IV) M2 (27Al{31P}) [106rad2/s2]

M2 (27Al{31P}) [106rad2/s2]

5.2 5.3 5.2 5.4 5.1 5.2 5.2

6.4/6.46b 6.2 4.9 5.5 5.2 4.4 5.3 5.1

M2 (27Al{19F}) [106rad2/s2] 710/1214b 138 91 131 197 148 69

a For the 27Al{31P} REDOR studies, typical errors for the M2-values are less than (10%. For the 27Al{19F} REDOR studies, no statistical error estimate can be given, as systematic errors dominate (see text). b Calculated van-Vleck second moment based on the crystal structure.

Figure 7. Site selective 27Al{19F} REDOR curves measured for a glass of composition Na/Al/P/F ) 1:2:2:1. 31P

dipolar coupling in the fluoride-containing glass (see Figure 8, right). This result confirms that the F species are primarily associated with mixed Al(F,OP)6 local environments. As discussed earlier, these mixed units are likely to arise from the reaction of P-bonded F species with either hydrated Al(VI) or with Al(IV) units during the gel-to-glass conversion process.36 Table 4 summarizes all the M2(27Al{31P}) values measured for various glass compositions indicating that this conclusion is generally applicable. 4.3. 19F{31P}, 19F{27Al}, and 19F{23Na} Double Resonance. As shown in the 19F MAS NMR spectra (Figure 2-5, middle) of the sol-gel derived glasses, two 19F NMR signals near -140 and -170 ppm are observed, while P-bonded species around -70 ppm (see Figure 2d) are, in most glasses, completely absent. The 19F{31P} REDOR curves (see Figure 9) further confirm that the spectral components at -140 and -170 ppm are not attributable to P-bonded species. Compared with the result for the P-bound species measured in crystalline Na2PO3F (Figure 9a), the 19F{31P} REDOR curves measured in the glasses portray much weaker heteronuclear dipolar coupling and substantially smaller second moments. Since dipolar interactions to 31P are nevertheless detectable in these curves, we assign these two resonances to fluorine atoms that are bound to aluminum species, which in turn are linked to phosphate units completing the six-

Zhang et al. coordinate aluminum coordination environment (i.e., P-OAl(VI)-F fragments). Parts b and c of Figure 9 illustrate further that the REDOR attenuation for the -140 ppm resonance is significantly more pronounced than that of the -170 ppm resonance. In fact M2(19F{31P}) for the -140 ppm signal is about twice as large as that measured for the -170 ppm signal (see Figure 9c). On the basis of this finding, it is reasonable to assign the -140 and -170 ppm signals to bridging Al-F-Al and terminal Al-F species, respectively. The larger M2(19F{31P}) value for the bridging fluoride can be understood easily because a larger total number of phosphate units can be brought (via Al-O-P linking) into its vicinity (compared to the nonbridging fluoride). This 19F peak assignment is nicely confirmed by 19F{27Al} REAPDOR experiments (Figure 10). Clearly, the -140 ppm signal assigned to the bridging F-species (two directly bonded Al atoms) shows a significantly more pronounced 19F{27Al} REAPDOR effect than the -170 ppm signal assigned to the terminal F species (one directly bonded Al atom). Furthermore, Figure 10b demonstrates that for each of the two fluorine sites, the strength of this interaction is independent of the glass composition, as one would expect. Compared to this result, analogous 19F{23Na} REDOR experiments reveal just the opposite order of interaction strength with sodium: Figure 11 shows representative results for a glass of composition Na/Al/ P/F ) 2:2:2:1, note that the terminal Al-bonded F atoms show a significantly stronger interaction with sodium than the bridging F species. With the use of the procedure described in reference 37, the dipolar second moments M2(19F{23Na}) could be extracted by analyzing the REDOR curve (Figure 11b) via eq 2. On the basis of the experimental 23Na quadrupolar coupling constant of 2.0 MHz and the experimental conditions chosen, an f1-value of 0.1015 was determined via SIMPSON simulations, yielding M2(19F{23Na}) values of 297 × 106 rad2/s2 and 94 × 106 rad2/s2 for the terminal and the bridging fluorine species, respectively. Analysis of a glass sample of composition Na/Al/ P/F ) 1:2:2:1 yields rather similar results (data not shown). On the basis of the applicable efficiency factor of 0.0934, we obtain M2(19F{23Na}) ) 221 × 106 rad2/s2 and 70 × 106 rad2/ s2 for the terminal and the bridging fluorine species, respectively. In both cases, M2(19F{23Na}) for the terminal fluoride species is about three times larger than the corresponding value for the bridging fluoride. This behavior can be rationalized by simple charge considerations. As the F bonded octahedral aluminum species have a net negative charge, Coulomb-interactions with Na+ ions should be favorable. In contrast, the bridging F species carry a charge that is formally positive, tending to repel Na+ ions. The above findings are confirmed further by a 19F{23Na} CPMAS-HETCOR experiment shown in Figure 12: Note that in comparison with the regular 1D MAS 19F spectra, the -140 ppm signal (bridging F) is significantly attenuated relative to the -170 ppm signal (terminal F), also suggesting a reduced 19F-23Na dipolar coupling strength for the former species. Finally, severe spinning sideband overlap problems have precluded an unambiguous analysis of specific REDOR experiments regarding the -195 ppm resonance; thus, no conclusive assignment is possible at the present time. Figure 5 suggests that this signal is observed particularly in those glasses having small Al(VI) concentrations. On the basis of this observation and the fact that some Al(V) species also interact with fluoride (see Figure 7), we may speculate that this signal reflects some fluoride species bound to Al(V) units. However, further double resonance work at higher spinning speeds will be necessary to confirm this hypothesis.



Figure 8. 27Al{31P} REDOR dephasing curve of glass with composition Na/Al/P/F ) 2:1:2:1 and comparison with the data measured for its F-free analogue (glass with composition Na/Al/P ) 2:1:2).

Figure 9. 19F{31P} REDOR studies of (a) crystalline Na2PO3F and (b) glass with Na/Al/P/F ) 2:2:2:1. In each individual trace, the upper and bottom curves represent the 19F signals without and with 31P irradiation during the evolution period of 0.267 ms. Spinning sidebands are indicated with asterisks. (c) 19F{31P} REDOR curves of Na2PO3F and the two fluorine species in the glass, including approximate parabolic fits to eq 1.

4.4. 31P{23Na} REDOR. Figure 13 shows representative results from 31P{23Na} REDOR on glass with composition 2:2: 2:1. The dipolar analysis of this curve with eq 2 leads to M2 ) 12 × 106 rad2/s2 which is significantly smaller compared to the value for the corresponding fluoride-free 2:2:2:0 glass (M2 ) 20 × 106 rad2/s2) and NaPO3 glass (M2 ) 22 × 106 rad2/s2).48 Previously published 31P{23Na} REDOR studies of fluoridefree ternary sodium aluminophosphate glasses have shown that M2(31P{23Na}) depends linearly on the Na/P ratio only and is virtually unaffected by the presence of aluminum.48 On the basis of these results, the weakening of the 23Na-31P dipole-dipole coupling in the present material must be attributed to the additional presence of the fluoride component in this glass. The result obtained here is nicely consistent with the strong 19F23Na dipolar couplings detected in particular for the terminal

fluoride species: incorporation of fluoride into a sodium aluminophosphate glass creates new anionic sites attracting Na+ ions, thereby diminishing the interaction of the latter with the phosphorus species. Note that this kind of information is only available from dipolar spectroscopy, as the 31P chemical shifts measured in these glasses cannot be interpreted unambiguously. 5. Discussion and Conclusions All the NMR results summarized above confirm consistently that in the sol-gel prepared Na-Al-P-O-F glasses with the compositions considered here, the fluorine exists as bridging (Al-F-Al) and terminal (Al-F) units associated almost exclusively with mixed Al(OP,F)6 units. The ratio of bridging and terminal fluoride species depends on glass composition in


Zhang et al.

Figure 10. (a) 19F{27Al} REAPDOR of glass with composition Na/ Al/P/F ) 2:2:2:1. In each individual trace, the upper and bottom curves represent the 19F signals without and with 27Al irradiation during the evolution period of 0.267 ms. Spinning sidebands are indicated with asterisks. (b) 19F{27Al} REAPDOR dephasing curves measured for the two F sites at two glass compositions.

Figure 11. (a) 19F{23Na} REDOR of glass with composition Na/Al/ P/F ) 2:2:2:1. In each individual trace, the upper and bottom curves represent the 19F signals without and with 23Na irradiation during the evolution period of 0.267 ms. (b) 19F{23Na} REDOR dephasing curves measured for the two F sites in the glass with Na/Al/P/F ) 2:2:2:1

a complicated manner: bridging F species appear to be favored in glasses having both low F/Al ratios and high P/Al ratios, and in most samples no P-F bonds are formed, except for a small amount in a sample with the highest P/F content. Furthermore, the analysis of the 27Al{31P} and 27Al{19F} REDOR data indicates that both the concentration of Al(VI) units and the average number of F bonded to these Al(VI) units (CNAl-F) increase with increasing the fluorine content in these glasses. From the M2(27Al{19F}) and M2(27Al{31P}) values listed in Table 4, we can quantify the average number of Al-O-P bonds (CNAlOP) and Al-F bonds (CNAl-F) per Al(VI) unit, using the following expressions:

CNAlOP ) 6M2(27Al{31P})/6.4

(4)

CNAl-F ) 6M2(27Al{19F})/710

(5)

Here, the M2 values are given in units of 106 rad2/s2 and the denominator values 6.4 and 710 (in units of 106 rad2/s2) are the experimental reference values of M2(27Al{31P}) and M2 (27Al{19F}) for fully connected Al(OP)6 and AlF6 environments, respectively, as measured on the crystalline model compounds Al(PO3)3 and AlF3. Equations 4 and 5 imply that these M2 values scale linearly with the number of Al-O-P or Al-F linkages; this is true provided no large variations in the corresponding internuclear distances occur. Table 5 summarizes the calculated average numbers of phosphorus (CNAlOP) and fluorine (CNAl-F) ligated to Al(VI) units. As expected, the sum of CNAlOP and CNAl-F is close to 6 for each glass, further confirming the validity of this particular approach. On the basis of these data summarized and the respective 19F MAS NMR

Figure 12. 19F{23Na} CPMAS-HETCOR spectrum measured for a glass with composition Na/Al/P/F ) 2:2:2:1.

spectra, we can estimate the F/Al ratio in these glasses, using the expression

F/Al )

* [Al(VI)]CNAl-F

2 [Al - F-Al] + [Al-F - ]

(6)


J. Phys. Chem. B, Vol. 111, No. 35, 2007 10411 increase in the concentration of six-coordinated aluminum. The combined interpretation of 27Al{31P} and 27Al{19F} REDOR NMR data allows a detailed quantification of Al-F and AlO-P connectivities. The formation of octahedral aluminum species with terminal Al-F bonding creates new anionic sites, resulting in a significant redistribution of the sodium ions in these glasses. Acknowledgment. Financial support from Deutsche Forschungsgemeinschaft (Grants Ec168/3-1 and Ec168/4-2) is gratefully acknowledged. We thank Ms. Wilma Pro¨bsting for the thermoanalytical characterization. Carla C. de Araujo thanks the NRW Graduate School of Chemistry for a personal stipend, and L. Zhang thanks the SFB458 for support as a visiting scientist to the WWU Mu¨nster. References and Notes

Figure 13. 31P{23Na}REDOR curve (∆S/S0 vs NTr) measured for a glass with composition Na/Al/P/F ) 2:2:2:1. A parabolic fit to eq 2 (data range ∆S/S0 < 0.25) is shown.

TABLE 5: The Coordination Numbers CNAlOP and CNAlF, and the F/Al Ratio from the Nominal Composition and Calculated for the Na/Al/P/F Glasses F/Alb F/Al (nominal sample (calculated- F-loss Na/Al/P/F CNAlOPa CNAl-Fa composition) glass) (%) AlF3 Al(PO3)3 2:1:2:0 2:1:2:1 2:2:2:0.5 2:2:2:1 2:2:2:2 2:2:4:1 1:2:2:1

6 6 6 4.64 5.15 4.88 4.13 4.97 4.78

1.16 0.77 1.11 1.66 1.25 0.58

1 0.25 0.5 1 0.5 0.5

0.87 0.22 0.38 0.74 0.40 0.38

13 12 24 30 20 24

a Obtained by the relationships in eqs 4 and 5. b F/Al ratio calculated by the relationship in eq 6.

Here the average number of Al-F bonds (CN*Al-F) per Al(VI) unit was estimated by taking ((6 - CNAlOP) + CNAl-F)/2. In eq 5, [Al(VI)] is the fractional contribution of Al(VI), obtained from the 27Al MAS NMR spectra. [Al-F-Al] and [Al-F-] represent the relative concentrations of bridging and terminating fluorine atoms, obtained from the corresponding 19F NMR spectra. Table 5 illustrates that the F/Al ratios calculated for all of the glasses tend to lie in the vicinity of the nominal compositions, although they consistently appear to be somewhat lower than batched. Most likely this discrepancy arises from a certain amount of fluoride loss encountered upon sample calcination; this loss can be estimated at around 20-30%. Furthermore, the above analysis neglects the small amount of fluoride connected to the Al(V) units. As the overall Al(V) concentration is low, and the corresponding REDOR effect is small, the error associated with this simplification is estimated to be below 5%. In summary, the local structure and site-connectivities of a series of sol-gel prepared Na-Al-P-O-F glasses were investigated extensively using high-resolution advanced solidstate NMR techniques. Detailed structural speciations for aluminum and fluorine were derived from high-resolution 27Al and 19F NMR spectra and supported by various quantitative and qualitative double resonance NMR experiments. Incorporation of fluorine atoms into aluminophosphate glasses leads to the formation of mixed Al(OP,F)6 units and results in a significant

(1) Poulain, M.; Poulain, M.; Lucas, J.; Brun, P. Mater. Res. Bull. 1975, 10, 243. (2) Matecki, M.; Poulain, M. J. Non-Cryst. Solids, 1983, 56, 111. (3) Matecki, M.; Jordery, S.; Lucas, J. J. Mater. Sci. Lett. 1992, 11, 1431. (4) Matecki, M.; Duhamel, N.; Lucas, J. J. Non-Cryst. Solids 1995, 184, 273. (5) Binnemans, K.; Van Deun, R.; Gorller-Walrand, C.; Adam, J. L. J. Non-Cryst. Solids 1998, 238, 11. (6) Kayne, C. B.; Weber, M. J. Phys. ReV. B 1977, 16, 3259. (7) Kumar, B. Mater. Res. Bull. 1981, 16, 179. (8) Weber, M. J.; Layne, C. B.; Saroyan, R. A.; Milam, D. Opt. Commun. 1976, 18, 171. (9) Stokowski, S. E.; Martain, W. E.; Yarema, S. M. J. Non-Cryst. Solids 1980, 40, 481. (10) Galant, V. E.; Makarenko, N. M.; Petrovskii, G. T.; Urusovskaya, L. N. Fiz. Khim. Stekla 1981, 8, 603. (11) Zuo, X.; Itoh, K.; Toratani, H. J. Non-Cryst. Solids 1997, 215, 11. (12) See, for example, (a) Petrov, V.; Griebner, U.; Ehrt, D.; Seeber, W. Opt. Lett. 1997, 22, 408. (b) Philipps, J.; Topfer, T.; EbendorffHeidepriem, H.; Ehrt, D.; Sauerbrey, R. Appl. Phys. B 2000, 72, 99. (c) Kopf, D.; Kaertner, F. X.; Keller, U.; Weingarten, K. J. Opt. Lett. 1995, 20, 1169. (13) See, for example, (a) Adam, J. L.; Duhamel-Henry, N.; Allain, J. Y. J. Non-Cryst. Solids 1997, 213 and214, 245. (b). Poirier, G.; Valdimir, A. J.; de Araujo, C. B.; Messaddeq, Y.; Ribeiro, S. J. L.; Poulain, M. J. Appl. Phys. 2003, 91, 10221. (c) Philipps, J.; Topfer, T.; EbendorffHeidepriem, H.; Ehrt, D.; Sauerbrey, R. Appl. Phys. B 2002, 74, 233. (14) Poirier, G.; de Araujo, C. B.; Messaddeq, Y.; Ribeiro, S. J. L.; Poulain, M. J. Appl. Phys. 2002, 91, 10221. (15) Tick, P. A. J. Am. Ceram. Soc. 1983, 66, 716. (16) Tick, P. A. Phys. Chem. Glasses 1984, 25, 149. (17) Sun, K. H. U.S. Patent 2,511,225. (18) (a) Brow, R. K.; Tallant, D. R.; Osborne, Z. A.; Yang, Y.; Day, D. E. Phys. Chem. Glasses 1991, 32, 188. (b) Brow, R. K.; Osborne, Z. A. J. Mater. Res. 1992, 7, 1892. (c) Brow, R. K.; Osborne, Z. A. Surf. Interface Anal. 1996, 24, 91. (19) (a) Videau, J.-J.; Portier, J.; Piriou, B. J. Non-Cryst. Solids 1982, 48, 385. (b) Gan, F.; Jiang, Y.; Jiang, F. J. Non-Cryst. Solids 1982, 52, 263. (20) (a) Santos, L. F.; Almeida, R. M.; Tikhomirov, V. K.; Jha, A. J. Non-Cryst. Solids 2001, 284, 43. (b) Karmakar, B.; Kundu, P.; Dwivedi, R. N. Mater. Lett. 2002, 57, 953. (21) (a) Mo¨ncke, D.; Ehrt, D.; Velli, L. L.; Varsamis, C. P. E.; Kamitsos, E. I. Phys. Chem. Glasses 2005, 46, 67. (b) Ehrt, D. Proc. SPIE-Int. Soc. Opt. Eng. Breckinridge, J. B., Marker, A. J., III, Eds.; 1992, 1761, 213. (22) (a) Yano, T.; Mizuno, J.; Shibata, S.; Yamane, M.; Inoue, S.; Onoda, Y. J. Non-Cryst. Solids 1997, 213/214, 345. (23) Eckert, H. Prog. Nucl. Magn. Reson. Spectrosc. 1992, 24, 159. (24) Eckert, H. In Encylopedia of NMR Spectroscopy; Grant, D., Harris, R. K., Eds.; Wiley-Interscience: New York, 1996. (25) Eckert, H.; Elbers, S.; Epping, J. D.; Janssen, M.; Kalwei, M.; Strojek, W.; Voigt, U. Top. Curr. Chem. 2005, 246, 195. (26) Gullion, T. Magn. Reson. ReV. 1997, 17, 83. (27) Gullion, T.; Schaefer, J. J. Magn. Reson. 1989, 81, 196. (28) Gullion, T.; Schaefer, J. AdV. Magn. Reson. 1989, 13, 58. (29) Garbow, J. R.; Gullion, T. J. Magn. Reson. 1991, 95, 442. (30) Pan, Y.; Gullion, T.; Schaefer, J. J. Magn. Reson. 1990, 90, 330. (31) Gullion, T. J. Magn. Reson. A 1995, 117326. (32) Chopin, L.; Vega, S.; Gullion, T. J. Am. Chem. Soc. 1998, 120, 4406.

10412 J. Phys. Chem. B, Vol. 111, No. 35, 2007 (33) Fyfe, C. A.; Grondey, H.; Mueller, K. T.; Wong-Moon, K. C.; Markus, T. J. Am. Chem. Soc. 1992, 114, 5876. (34) Caravatti, P.; Bodenhausen, G.; Ernst, R. R. Chem. Phys. Lett. 1982, 89, 363. (35) Egan, J. M.; Wenslow, R. M.; Mueller, K. T. J. Non-Cryst. Solids 2000, 261, 115. (36) Zhang, L.; de Araujo, C. C.; Eckert, H. Chem. Mater. 2005, 17, 3101. (37) Strojek, W.; Kalwei, M.; Eckert, H. J. Phys. Chem. B 2004, 108, 7061. (38) Bertmer, M.; Eckert, H. Solid State Nucl. Magn. Reson. 1999, 15, 139. (39) Chan, J. C. C.; Eckert, H. J. Magn. Reson. 2000, 140, 170. (40) Schmidt, A.; MacKay, R. A.; Schaefer, J. J. Magn. Reson. 1992, 96, 644. (41) Hudalla, C.; Eckert, H.; Dupree, R. J. Phys. Chem. 1996, 100, 15986.

Zhang et al. (42) Bak, M.; Rasmussen, J. T.; Nielsen, N. C. J. Magn. Reson. 2000, 147, 296. (43) Van Vleck, J. H. Phys. ReV. 1948, 74, 1168. (44) Ba, Y.; Kao, H. M.; Grey, C. P.; Chopin, L.; Gullion, T. J. Magn. Reson. 1998, 133, 104. (45) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995, 117, 12779. (46) Kirkpatrick, R. J.; Brow, R. K. Solid State Nucl. Magn. Reson. 1995, 5, 9. (47) Lang, D. P.; Alam, T. M.; Bencoe, D. N. Chem. Mater. 2001, 13, 420 and references therein (48) Zhang, L.; Eckert, H. J. Phys. Chem. B 2006, 110, 8946. (49) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calv, S.; Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Magn. Reson. Chem. 2002, 40, 70.

Structural Role of Fluoride in Aluminophosphate SolâGel Glasses

Recommend Documents

Structural Role of Fluoride in Aluminophosphate SolâGel Glasses