Incorporation of PbF2 into Heavy Metal Oxide Borate Glasses

Incorporation of PbF2 into Heavy Metal Oxide Borate Glasses. ... C , 2012, 116 (10), pp 6434–6445 ... with composition 26.66B2O3-16GeO2-4 Bi2O3-(53...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/JPCC

Incorporation of PbF2 into Heavy Metal Oxide Borate Glasses. Structural Studies by Solid State NMR Roger Gomes Fernandes,† Jinjun Ren,‡ Andrea S. S. de Camargo,† Antonio Carlos Hernandes,† and Hellmut Eckert*,†,‡ †

Instituto de Física de São Carlos, Universidade de São Paulo, Av. Trabalhador São-carlense 400, São Carlos, SP, CEP 13566-590, Brazil ‡ Institut für Physikalische Chemie, WWU Münster, Corrensstraße 30, D 48149 Münster, Germany S Supporting Information *

ABSTRACT: A series of heavy metal oxide (HMO) glasses with composition 26.66B2O3-16GeO2-4 Bi2O3-(53.33-x)PbOxPbF2 (0 ≤ x ≤ 40) were prepared and characterized with respect to their bulk (glass transition and crystallization temperatures, densities, molar volumes) and spectroscopic properties. Homogeneous glasses are formed up to x = 30, while crystallization of β-PbF2 takes place at higher contents. Substitution of PbO by PbF2 shifts the optical band gap toward higher energies, thereby extending the UV transmission window significantly toward higher frequencies. Raman and infrared absorption spectra can be interpreted in conjunction with published reference data. Using 11B and 19F high-resolution solid state NMR as well as 11B/19F double resonance methodologies, we develop a quantitative structural description of this material. The fraction of four-coordinate boron is found to be moderately higher compared to that in glasses with the same PbO/B2O3 ratios, suggesting some participation of PbF2 in the network transformation process. This suggestion is confirmed by the 19F NMR spectra. While the majority of the fluoride ions is present as ionic fluoride, ∼20% of the fluorine inventory acts as a network modifier, resulting in the formation of four-coordinate BO3/2F− units. These units can be identified by 19F{11B} rotational echo double resonance and 11B{19F} cross-polarization magic angle spinning (CPMAS) data. These results provide the first unambiguous evidence of B−F bonding in a PbF2-modified glass system. The majority of the fluoride ions are found in a lead-dominated environment. 19F−19F homonuclear dipolar second moments measured by spin echo decay spectroscopy are quantitatively consistent with a model in which these ions are randomly distributed within the network modifier subdomain consisting of PbO, Bi2O3, and PbF2. This model, which implies both the features of atomic scale mixing with the network former borate species and some degree of fluoride ion clustering, is consistent with all of the experimental data obtained on these glasses.



INTRODUCTION Heavy metal oxide glasses (HMOG) have received much attention because of their high density, high linear, and nonlinear refractive indices enabling their use in various applications in optics and optelectronics.1−5 The use of the network former B2O3 ensures a wide glass-forming region enabling the incorporation of other constituents such as GeO2 and Bi2O3 for tailoring physical property combinations to application demands. The incorporation of fluoride ions into these glasses is of interest because it presents an additional opportunity for optical bandgap tuning and for using these glasses as host materials for luminescent rare-earth ions aiming at potential laser applications.6−10 Furthermore, a number of lead fluoroborate glass compositions have garnered interest as fast ion conductors, based on anionic fluoride mobility.11−15 In addition, such glasses have shown sub-Tg crystallization phenomena induced by externally applied electric fields.13 In an effort to optimize physical properties, the glass-forming regions and local structural properties of HMOG have been extensively © 2012 American Chemical Society

investigated by a large variety of bulk experimental and spectroscopic techniques. In addition, the structural modification of borate networks by incorporated lead fluoride and other heavy metal fluoride salts is an interesting subject from an academic point of view. On the basis of a number of indirect evidences, it has been suggested that the borate network is modified by the fluoride component, resulting in the formation of boron−fluorine bonds; however, no definitive proof has been provided.11−17 Thus far, F-1s X-ray photoelectron spectra in binary PbF2−B2O3 glasses have given the most concrete evidence for B−F bond formation.18 On the other hand, high resolution solid-state NMR presents an ideally suited technique for addressing such structural issues.19 Both isotropic chemical shifts, measured by magic-angle spinning (MAS) and magnetic dipole−dipole interactions, measured by Received: December 30, 2011 Revised: February 19, 2012 Published: February 21, 2012 6434

dx.doi.org/10.1021/jp212610g | J. Phys. Chem. C 2012, 116, 6434−6445

The Journal of Physical Chemistry C

Article

Table 1. Physical Properties of the Glasses under Study

a

sample

Tg (±2 °C)

Tx (±2 °C)

Tx − Tg (±3 °C)

ρ(g/cm3)

Vm (cm3/mol)

λc (nm)

Egap (eV)

BPBG BPBG15 BPBG30 BPBG40a

333 297 272 262

476 468 358 291

143 171 86 29

6.519 6.623 6.558 6.487

26.43 26.54 27.31 27.90

363.5 343.4 321.7

3.35 3.54 3.77

Partially crystalline.

Optical absorption was measured on a Thermo Scientific Evolution 60 spectrophotometer, within a wavelength range from 200 to 1100 nm, with 0.5 nm resolution. Solid State NMR Measurements. Solid state 1-D 11B MAS and 2-D triple-quantum (TQ)-MAS NMR experiments were carried out at 160.485 MHz on a Bruker DSX 500 spectrometer, using the three-pulse z-filtering sequence.24,25 For the 1-D spectra 15° pulses of 0.54 μs length were used with a relaxation delay of 0.5 s at a spinning frequency of 12 kHz. For the 2-D spectra, the pulse lengths of the two hard excitation and reconversion pulses were 4.5 and 1.4 μs, at a nutation frequency of 100 kHz. For detection of single-quantum coherence a soft pulse of 9 μs length applied at a nutation frequency of 31.2 kHz was used. The evolution time was incremented by 16.67 μs, and 200 increments were recorded. Chemical shifts are referenced to BF3·OEt2. Solid state 19F MAS NMR measurements were done on a Bruker Avance 400 console interfaced with a 4.65 T magnet. Data were recorded at a 19F resonance frequency of 188.357 MHz, using 90° pulses of 1 μs length, a relaxation delay of 20 s, and a MAS spinning rate of 25 kHz. Chemical shifts are referenced to CFCl3, using AlF3 (δ = −172 ppm) as a secondary standard. The spatial distribution of the fluoride ions was measured by 19F Hahn spin echo decay spectroscopy, using 90° and 180° pulses of 3.0 and 6.0 μs length, respectively. To ensure that the data are not influenced by fluoride ion motion, these measurements were done at 184 K. Following the approach of ref 26, the data were analyzed by fitting the echo amplitudes at short evolution times 2t1 to Gaussians

double resonance techniques, such as rotational echo double resonance (REDOR), and cross-polarization are highly sensitive to the local environments.20 As a matter of fact, a number of solid state NMR studies have dealt with the structural modification of borate glasses by alkali fluorides.21−23 In contrast, to date, no NMR studies addressing the structural modification of borate networks by lead fluoride have as yet appeared. In this contribution we present, to the best of our knowledge, the first multinuclear high-resolution NMR study of a PbF2- modified borate glass system. On the basis of 11B and 19F single and double resonance experiments on a series of HMO glasses with composition 26.66B2O3-16GeO2-4Bi2O3-(53.33-x)PbO-xPbF2, we develop a quantitative description of the local structure of these glasses. In addition, the mutual interaction between the fluoride and the borate species in these glasses is addressed by various 19F/11B double resonance experiments, which probe the presence of magnetic dipole−dipole interactions between these two nuclear species, giving evidence of direct bonding and/or spatial proximity. Finally, experiments measuring the strength of homonuclear 19F−19F magnetic dipole−dipole couplings are used to probe the spatial distribution of the fluoride ions. The results outline a general strategy for the structural elucidation of fluoride containing borate glasses.



EXPERIMENTAL SECTION Sample Preparation and Characterization. Glasses of composition 26.66B2O3-16GeO2-4Bi2O3-(53.33-x)PbO-xPbF2 (0 ≤ x ≤ 40) (see Table 1) were prepared by the traditional melt-quenching technique. The starting materials B2O3, GeO2, Bi2O3, PbO, and PbF2 (Alfa-Aesar, purity >99.99%) were thoroughly mixed in 30 g batches and heated at 850 °C for 15 min in a Pt crucible and stirred to ensure homogeneity. The melt was subsequently poured onto preheated graphite molds and annealed for 12 h near their glass transition temperature (to release mechanical stress) and then slowly cooled to room temperature. Homogeneous transparent samples were obtained for PbF2 contents up to x = 30. For these samples, the glassy state was verified by the absence of sharp X-ray powder diffraction peaks, using a Rigaku-Ultima IV diffractometer, with Cu Kα radiation, operated at normal incidence and a scan rate of 2°/min from 20° to 80°. The sample with x = 40 was partially crystalline, containing β-PbF2. Densities were measured by the Archimedes method in distilled water, using a Mettler Toledo AG 285 balance (sensitivity 10−4 g). Glass transition and crystallization temperatures were recorded under a N2 atmosphere, using a TA Instruments DSC 2910 differential scanning calorimeter, operated at a heating rate of 10 °C/min. FTIR spectra were recorded on a Bruker Vertex 70 spectrometer in the range from 400 to 1500 cm−1 on pressed KBr pellets. Micro-Raman spectra were obtained with a confocal Raman microscope Witec model Alpha-300S A/R. The samples were excited using an Ar+ laser at 514.5 nm (Melles Griot, Model 35-LAL-515-230). All spectra were collected with an integration time of 80 s from 100 to 1500 cm−1.

I = I0 exp{− 2t1)2 M2 /2}

(1)

where M2 is the van Vleck dipolar second moment characterizing the average magnitude of the homonuclear 19F−19F magnetic dipole dipole interactions. To investigate atomic connectivities and spatial proximities among the various 19F and 11B local environments, a number of 19F/11B double resonance experiments were carried out. Among these, the rotational echo double resonance (REDOR) experiment27 reintroduces the heteronuclear magnetic dipole−dipole coupling between two nuclear species (denoted as X and Y here), which is normally averaged out by MAS, by the effect of coherent 180° pulse trains applied to these nuclei during the rotor period. The experiment is conducted in two steps. In the first step, the signal of the observe nucleus X is detected by a rotor synchronized Hahn spin echo sequence, producing a signal with intensity S0. In the second step, the X−Y dipolar interaction is recoupled by applying π pulse trains to the Y spins during the dipolar recoupling time N × Tr (the product of the number of rotor cycles and the rotor period). As a result of this recoupling, the signal of the observed nuclei is decreased to an intensity S. A plot of the normalized difference signal ΔS = (S0 − S)/S0 against N × Tr constitutes a “REDOR curve”, from which the dipolar coupling strength can be quantified. In the present study, both 19F{11B} and 11B{19F} REDOR experiments were carried out. The 19F{11B} REDOR experiments were done at a spinning rate of 30 kHz on a Bruker 6435

dx.doi.org/10.1021/jp212610g | J. Phys. Chem. C 2012, 116, 6434−6445

The Journal of Physical Chemistry C

Article

PbF2-free glass the absorption edge is determined by excitation of valence electrons within Pb−O−Pb linkages, which, owing to the high PbO content, occur in ample concentrations in this glass. With increasing x (and concomitantly decreasing PbO content), the concentration of these linkages decreases significantly, resulting in the observed shift of the absorption edge. Figure S1 (Supporting Information) summarizes the infrared absorption spectra. Tentative peak assignments, supported by relevant literature data on lead borate glasses,16,17,28,29 are summarized in Table S2a. The IR spectra (see Figure S1) are dominated by stretching modes associated with various types of three-coordinate boron species (broad band in the 1120−1470 cm−1 region) and four-coordinate boron species (doublet feature in the 805−1120 cm−1 region). The systematic change in the relative intensities of these features suggests that the fraction of four-coordinate boron increases with increasing PbF2 content. Raman spectral assignments are summarized in Table S2b. Scattering peaks at 540 cm−1 and below can be assigned to stretching modes associated with Ge−O bonds, while the IR absorption peak near 400 cm−1 may be attributed to Pb−O−Pb vibrations. The spectra of all the glasses look rather similar, giving no evidence of the structural influence of the lead fluoride component. In particular, no specific bands involving vibrations of the fluorine atoms can be identified. Solid State NMR. High-Resolution 11B MAS NMR and 11 19 B{ F} REDOR. Figure 2 presents the 11B solid state MAS

DSX 500 spectrometer (Larmor frequencies 470.6 and 160.4 MHz, respectively), using the pulse sequence of Gullion and Schaefer.27 Solid π pulses of 3.3 and 3.7 μs length were employed for 19F and 11B, respectively, and a relaxation delay of 20 s was used. The 11B{19F} REDOR experiments were carried out on a Bruker DSX 400 spectrometer (Larmor frequencies of 376.63 and 128.42 MHz, respectively) using π pulses of 6.0 μs length for 19 F and of 6.0 and 5.9 μs length for 11B(III) and 11B(VI) respectively. As the nutation frequencies for the 11B nuclei within the two coordination states are quite different owing to the large difference in nuclear electric quadrupolar coupling strengths, separate experiments had to be done for observing the B(III) and B(V) difference signals. To explore the possibility of detecting F-bonded borate species, 11B{19F} cross-polarization (CP)-MAS experiments were carried out. These experiments were done at a spinning frequency of 14 kHz, using a relaxation delay of 20 s. Good spin-lock and Hartmann−Hahn matching conditions were found at nutation frequencies of 65.8 and 50 kHz for 11B and 19 F, respectively, and the 11B power level was subjected to a linear ramp down to a nutation frequency of 32.9 kHz. 1024 scans were accumulated with a relaxation delay of 0.5 s, under systematic variation of the contact time. 11B{19F} 2D CPheteronuclear correlated (HETCOR) experiments were undertaken with contact times of 50, 200, and 5000 μs and with 100 increments of the pre-Hartmann−Hahn contact evolution time in steps of 5 μs. Reverse 19F{11B} experiments were also attempted but found to be unsuccessful, owing to difficulties in keeping the magnetization of the 11B source nuclei spin-locked under acceptable Hartmann−Hahn matching conditions. These experiments might be also more difficult owing to the weaker 11 B−11B (as compared to the 19F−19F) homonuclear dipolar interactions.



RESULTS, DATA ANALYSIS, AND INTERPRETATION General Characterization. Table 1 summarizes the glass compositions and physical characteristics of the samples studied. As indicated by Figure 1, successive replacement of

Figure 2. 11B MAS NMR spectra and their two-component simulations for the four glasses studied.

NMR spectra. The spectra are influenced by both the chemical shift and nuclear electric quadrupolar interactions. For the three-coordinate (B(III)) species a broad structured line shape is observed. Those boron species experience strong electric field gradients produced by the anisotropic local oxygen environment, resulting in strong second-order quadrupolar coupling effects. For the four-coordinate (B(IV)) species, only a sharp signal near 1 ppm is observed, as the environment is much more symmetrical and the influence of the quadrupolar interaction upon the line shape is very weak. Table 2 summarizes the line shape parameters obtained using the DMFIT simulation routine.30 As the PbF2 content of the glasses is increased, the fraction N4 of the four-coordinate boron increases significantly, until a saturation value is reached for the sample with the highest PbF2 content, which is partially

Figure 1. Optical absorption spectra of the glasses under study, indicating the shift in the UV absorption edge with increasing PbF2 content.

PbO by PbF2 results in a significant shift of the UV absorption edge toward lower wavelength, thereby increasing the optical bandgap and the UV transmission window (see Table 1). In the 6436

dx.doi.org/10.1021/jp212610g | J. Phys. Chem. C 2012, 116, 6434−6445

The Journal of Physical Chemistry C

Article

Table 2. 11B Line-Shape Parameters Obtained from the Spectra of the Glasses Studied B(III) BPBG BPBG15 BPBG30 BPBG40

B(IV)

fraction (±1%)

δCSiso (±0.5 ppm)

η (±0.05)

Cq (±0.05 MHz)

fraction (±2%)

δCSiso (±0.5 ppm)

54 43 41 42

18.5 18.4 18.2 18.0

0.32 0.28 0.25 0.25

2.52 2.52 2.54 2.55

46 57 59 58

1.2 1.1 1.2 1.0

crystalline. Compared to the N4 values measured in binary lead borate glasses with the same PbO/B2O3 ratios,31,32 the corresponding values of the PbF2 containing glasses are found to be significantly higher, suggesting that this additive participates at least partly in the network transformation process. The trend observed here is opposite to that seen in Na2B4O7−Na3AlF6−TiO2 glasses, where N4 was found to decrease with increasing F/B ratio.23 To resolve further details in the 11B MAS NMR spectra, 2-D triple quantum experiments, which remove anisotropic secondorder quadrupolar broadening effects, were carried out. Figure 3

REDOR curves were obtained for the two boron species. Figure 4b illustrates that, aside from the short-time behavior (to be discussed below), the REDOR curves for the B(III) and the B(IV) species appear to be very similar overall, suggesting comparable dipolar interaction strengths with the 19F nuclei. The moderate dephasing rates indicate that the majority of the B(III) and the B(IV) species are not directly bonded to fluorine atoms. However, closer inspection of the REDOR data does indicate that a small fraction of the B(IV) species may be having a B−F bond, resulting in a rather steep rise of the difference signal for the shortest evolution times measured. In contrast, no such peculiarity is observed for the B(III) species, which is consistent with the absence of B−F bonding in this particular environment. High-Resolution 19F MAS NMR and 19F{11B} REDOR Results. Figure 5 summarizes the 19F MAS NMR data obtained on these samples. All of the spectra can be consistently deconvoluted into three line-shape components whose parameters are summarized in Table 3. In the partially crystallized sample, an additional sharp signal near −38 ppm is observed, indicating crystallization of β-PbF2. In the glasses, the position of the dominant signal is found in the vicinity of the β-PbF2 peak (albeit much broader), suggesting an assignment to fluoride ions in an amorphous, lead-dominated local environment. For glasses containing 15 and 30 mol % PbF2, the significant line width (fwhm) of this component clearly indicates a wide chemical shift distribution, as is typical for fluoride species in the amorphous state. A second, rather distinct component is identified near −107 to −112 ppm. As discussed below on the basis of various double resonance experiments, this signal is assigned to fluorine bound to boron atoms, specifically to four-coordinated ones in the glass network. Finally, all the line shape deconvolutions, in particular the one conducted for the partially crystallized sample suggest the presence of a third spectral component near −60 ppm. We tentatively assign this site to fluoride species that are also present in a lead-dominated environment but occur closer to the atoms in the oxide-based framework than those contributing to the −40 ppm peak. While for the samples containing 15 and 30 mol % PbF2 the deconvolution suggested in Figure 5 may appear somewhat arbitrary (and alternatively a wide distribution of shifts may be considered), the presence of a line shape component near −60 ppm is clearly suggested by the spectrum of the partially crystallized sample containing 40 mol % PbF2. While a significant fraction of the fluorine content in this sample is present as β-PbF2, some fluoride in a Pb-dominated environment remains in the amorphous state in this sample. Figure 6 summarizes the 19F{11B} REDOR results, which clearly indicate that the different types of fluorine species in the glass experience drastically different 19F−11B dipolar coupling strengths. As expected, the signal of crystalline PbF2 in sample BPBG40 yields the expected zero REDOR effect, whereas the majority fluorine signal present in both of the glass samples shows clear evidence of 19F−11B dipolar coupling. While in this case no distinction was made in the analysis between the

Figure 3. Sheared 2D-11B TQMAS spectrum of glass BPBG30. The cross section plotted horizontally corresponding to the anisotropic F2 dimension is the regular MAS NMR spectrum, while the cross section plotted vertically corresponds to the isotropic F1 dimension. Spinning sidebands are indicated by asterisks.

shows a representative result. While the overall spectroscopic resolution is indeed dramatically improved in the isotropic F1 dimension, the spectra reveal no further site differentiation between different types of B(III) and/or B(IV) species. In particular, these spectra give no evidence for the formation of three- or four-coordinated boron species bound to fluorine atoms.33 To address this question, Figure 4 presents 11B{19F} REDOR data. Part a shows that both the signals of B(III) and B(IV) are attenuated upon recoupling the magnetic dipole− dipole interactions by π pulses applied to the 19F channel, documenting qualitatively that both of the boron species are in spatial proximity to 19F nuclei. Parts b−d show full REDOR curves, obtained by systematically incrementing the dipolar evolution time, i.e., the number of rotor cycles times the rotor period, during which recoupling is done. By simulating the experimental line shapes as superpositions of the individual B(III) and B(IV) components introduced in Figure 2, separate 6437

dx.doi.org/10.1021/jp212610g | J. Phys. Chem. C 2012, 116, 6434−6445

The Journal of Physical Chemistry C

Article

Figure 4. 11B{19F} REDOR NMR data on the glasses under study: (a) 11B line shapes of BPBG30 in the absence (blue curve) and the presence (green curve) of dipolar dephasing, indicating that both types of B environments are affected by dipolar interactions with 19F nuclei. (b) REDOR difference signals ΔS/S0 versus dipolar evolution time NTr, for the B(III) and the B(IV) species. (c) REDOR difference signals ΔS/S0 versus dipolar evolution time NTr for the B(IV) species in the three glasses studied. (d) REDOR difference signals ΔS/S0 versus dipolar evolution time NTr for the B(III) species in the three glasses studied. Sample denominations are 1: BPBG15 (black); 2: BPBG30 (red); and 3: BPBG40 (green).

dominant component near −40 ppm and the one near −60 ppm when constructing the REDOR curves, the REDOR difference spectra shown at a dipolar evolution time of 400 μs do indicate a significantly stronger signal attenuation due to dipolar recoupling for the low-frequency wing of that signal compared to the high-frequency wing. This observation suggests a correlation between the 19F chemical shift and the spatial proximity of the fluoride species to the borate network; possibly the chemical shift change across this line shape may reflect different numbers of Pb2+ ions in the first coordination sphere of these fluoride species. Still, the fact that even the fluoride species contributing to the −40 to −50 ppm resonance experience significant magnetic dipole−dipole interactions with the 11B nuclei is consistent with atomic-scale mixing of the fluoride ions and the borate network and the absence of macroscopic chemical segregation effects. {19F}11B CPMAS and 2-D Heteronuclear Correlation. Finally, Figure 7 presents contact time dependent {19F}11B CPMAS NMR data on the F-containing glass samples. First of all, under the Hartmann−Hahn matching conditions chosen, only the B(IV) resonance can be detected. The failure of observing the B(III) resonance does not, however, necessarily indicate the absence of spatial proximity between these boron units and fluorine. Rather, systematic 11B spin-locking experiments conducted on this sample indicate that under the Hartmann−Hahn matching conditions chosen the magnetization

associated with the 11B(III) resonance cannot be kept spin-locked for reasonable contact times on the order of 1 ms. This wellknown effect can be attributed to the level crossings during the rotor cycle, which are critically influenced by the strength of the nuclear electric quadrupolar interaction in relation to the B1 amplitude and the spinning frequency chosen.34 In contrast to the B(III) signal, that of B(IV) is easily observed in experiments with a contact time of 50 μs. At contact times this short, crosspolarization only occurs in the presence of very strong dipole− dipole interactions, as is the case for directly bonded nuclei. Therefore, the 11B NMR signal observed in such short contact time experiments must be assigned to an F-bonded fourcoordinated boron species. As already suggested by the 11 19 B{ F} REDOR results, only a minor fraction of the fourcoordinated boron atoms are F-bonded. In the CPMAS experiments a distinction between F-bonded and non-F-bonded 11B nuclei can be made by measuring the 11B signal build-up rate, which depends on the 19F−11B dipolar coupling strengths, in variable contact time experiments. Indeed, Figure 7a indicates a bimodal signal build-up curve for the glass containing 15 mol % PbF2. The initially extremely rapid signal build-up at contact times