Network Connectivity and Crystallization in the Transparent

Subscriber access provided by UNIV OF LOUISIANA

C: Plasmonics; Optical, Magnetic, and Hybrid Materials

Network Connectivity and Crystallization in the Transparent Ferroelectric Nanocomposite LaBGeO

5

Alexander Liam Paterson, Ulrike Werner-Zwanziger, and Josef Wilson Zwanziger J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.9b02430 • Publication Date (Web): 15 Apr 2019 Downloaded from http://pubs.acs.org on April 15, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Network Connectivity and Crystallization in the Transparent Ferroelectric Nanocomposite LaBGeO5 Alexander L. Paterson,† Ulrike Werner-Zwanziger,† and Josef W. Zwanziger∗,†,‡ †Department of Chemistry, Dalhousie University, Halifax, NS B3H 4R2, Canada ‡Clean Technologies Research Institute, Dalhousie University, Halifax, NS B3H 4R2, Canada E-mail: [email protected]

1

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 51

Abstract The LaBGeO5 glass-ceramic composite is a transparent ferroelectric nanocomposite (TFN) material that has come under attention for its ferroelectric properties, and for its straightforward synthesis. The LaBGeO5 crystals-in-glass can be formed through controlled devitrification of the glass, as well as through laser irradiation. Recent works have established that both the borate and germanate environments have significant differences in short-range order between the crystalline and glass forms of LaBGeO5 . However, there is little available data on the connectivity of the different structural units. We present here a comprehensive nuclear magnetic resonance (NMR) spectroscopy study of the connectivity of crystalline LaBGeO5 and LaBGeO5 glass. Through a combination of

11

B,

11

B{ 10B},

17

O, and

139

La NMR spectroscopies we identify specific structural

motifs in the glass. In particular, several structures are positively identified in the glass that are not present in the crystal, including BØ2 O− , BØ2 O− –BØ− 4 , and GeO6 . We present evidence in support of the presence of both

5/6

Ge – O – 4Ge and

5/6

Ge – O – B

connectivities. The differences in short-range order and connectivity between the LaBGeO5 glass and crystal suggest a heterogeneous nucleation mechanism.

Introduction Ferroelectric glass-ceramics are composite materials in which a ferroelectric crystalline phase is nucleated within a glass matrix. One example of a ferroelectric glass-ceramic is the LaBGeO5 composition. As the LaBGeO5 glass can be congruently (i.e., isochemically) crystallized, it has become a particular target for investigation into transparent ferroelectric nanocomposite materials. Many different compositions (e.g., LiNbO3 , BaTiO3 ) can produce glass-ceramics that are transparent with ferroelectric crystallites. 1,2 However, many of these compositions produce multiple distinct crystalline phases upon devitrification, complicating their characterization and analysis. The apparently simple crystallization behaviour of LaBGeO5 has made it the target of many investigations. The LaBGeO5 crystal has been

2


Page 3 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


well-characterized, with its structure (Fig. 1), 3,4 second-harmonic generation efficiency, 5,6 ferroelectric moment, 7 and Curie temperature 6–8 having been reported in the literature. Similar efforts have been undertaken with regards to the glass and glass-ceramic composite,

Figure 1: The room-temperature crystal structure of LaBGeO5 . 3 The presented view is down the crystallographic c axis. The unit cell is indicated by black lines. BØ− 4 tetrahedra are 2− coloured dark green, while GeØ2 O2 tetrahedra are purple. LaO9 polyhedra are omitted for clarity. Lanthanum ions are light green. The oxygen atoms are labelled O1 through O5, as per Kaminskii et al. 3 and various physical properties of both have been reported. 9 Recently, laser-written singlecrystal LaBGeO5 architectures in LaBGeO5 glass have been reported. 10,11 To understand fully the mechanism and dynamics of single-crystal formation within a glassy matrix, it is necessary to have a thorough understanding of the glass structure and how it relates to the

3



Page 4 of 51

crystal structure. There are a number of structural studies of the LaBGeO5 glass reported in the literature. In 2000, Kratochv´ılová et al. reported Raman spectra of the glass. 12 Based upon comparison to the crystal, they concluded that the structure of the glass was similar to the crystal, albeit disordered. In 2004 Gupta et al. reported the first NMR study of the LaBGeO5 glass, presenting

11

B MAS NMR spectra definitively proving the presence of both BO3 and BØ− 4

13 units in the glass, whereas the crystal only has BØ− Sigaev et al. revisited the 4 units.

Raman spectrum of LaBGeO5 in 2010, concluding that their data confirmed the presence of BO3 , but also that all other structural units were unchanged from the crystal. 14 Further, they concluded that the LaBGeO5 glass is composed of “crystal-like” nanoregions, where the boron is fourfold coordinate, connected by regions where the boron is threefold coordinated. The above studies neglect to consider possible structural changes in either the germanium or lanthanum environments. Five- and six-coordinate germanium-oxygen polyhedra are known to form in germanate crystals, and have been shown to be present in alkali and alkalineearth germanate glasses. 15–18 The LaBGeO5 crystal structure contains only germanium tetrahedra. 3 Via neutron diffraction, we have shown that high-coordinate germanium is present in lanthanum borogermanate glasses. 19 This dramatic change in the germanium environment between the glass and the crystal suggests the possibility of radically different glass and crystal structures, on both the short-range and intermediate-range connectivity. However, only the short-range order of the LaBGeO5 glass has been studied. NMR spectroscopy offers the possibility of probing the connectivity between structural units, including between high-coordinate germanate units. We present here a study of the structural differences between the glass and crystal phases of the boron, germanium, and lanthanum local environments through a combination of 11

B{ 10B},

17

O, and

139

La NMR spectroscopies.

for the study of borate glasses. 20–22

11

11

11

B,

B NMR spectroscopy is a mature technique

B{ 10B} rotational-echo double-resonance (REDOR)

experiments can probe the connectivity of the borate network in borate glasses. 23

4


139

La

Page 5 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


NMR spectroscopy has become more accessible with the advent of the wideband, uniform rate, smooth truncation Carr-Purcell Meiboom-Gill (WCPMG) pulse sequence, 24 but no comprehensive model of the relationship between glass spectra and properties has yet been reported in the literature.

17

O NMR spectroscopy is a very powerful technique for the

study of oxide glass structure, and has previously been used to establish the presence of high-coordinate germanium in sodium germanate glasses. 15,25,26

73

Ge NMR studies of glass

are not generally considered to be feasible, as previous attempts have produced inconclusive results. 27 With this comprehensive probe into the short-range order of the glass we intend to produce reasonable constraints on the structure of the glass in order to inform future mechanistic studies of the formation of LaBGeO5 glass-ceramics.

Experimental Procedures Sample Synthesis Crystalline Ge 17O2 was synthesized by hydrolysing germanium ethoxide (Ge(OC2 H5 , ≥ 99.95 %, Sigma-Aldrich) with 40 %

17

O-enriched H2 17O (Cortecnet). The ethoxide and

enriched water were combined under nitrogen at a molar ratio of 1:2 in a sealed container and allowed to react for 3 d. The product was then heated to 350 ◦C under nitrogen for approximately 3 d for improved crystallinity. The identity of the crystalline product was confirmed to be trigonal (quartz structure) GeO2 via powder X-ray diffraction (pXRD). Crystalline H3 B 17O3 was synthesized by hydrolysing a 1 mol dm−3 solution of borane THF (BH3 OC4 H8 , Sigma-Aldrich) with 40 %

17

O-enriched H2 17O. The enriched H2 17O was slowly

added to 0 ◦C borane THF solution under dry nitrogen atmosphere. Excess THF and evolved hydrogen gas were removed under vacuum. The final polycrystalline product was verified to be H3 BO3 via both pXRD and

11

B MAS NMR spectroscopy. LaBGeO5 glass was synthesized by

combining stoichiometric amounts of the enriched Ge 17O2 and H3 B 17O3 powders with La2 O3 (99.99 %, Sigma-Aldrich), thoroughly grinding the mixture in a mortar and pestle, and then 5



Page 6 of 51

melting the mixture in a 95/5 Pt/Au crucible at 1250 ◦C in a box furnace in air. The melt was quenched by pouring it onto a brass plate preheated to 400 ◦C. Partially crystallized samples were produced by placing the powdered sample in a furnace pre-heated to 765 ◦C for short lengths of time (approx. 4 min). Observations (i.e., pXRD, NMR) were taken, and afterwards the sample was returned to the furnace for further crystallization.

17

O NMR spectra were

collected when sufficient crystallization was observed in the powder via

11

B MAS NMR.

Reported samples are labelled LBG-G, for the original glass, and LBG+8, LBG+12, LBG+24, and LBG+60, where the number in the label is the total time spent crystallizing, in minutes. An additional sample was prepared by heating the LBG+60 sample for an additional 3 d; this further increased the crystal fraction, but removed the

17

O, preventing the collection of

17

O

NMR spectra. This sample is labelled LBG+3d. LaB3 O6 crystal powder was synthesized by combining stoichiometric amounts of La2 O3 and B2 O3 and heating in a platinum crucible for 12 h at 850 ◦C. LaBGeO5 samples were produced with natural abundance reagents (La2 O3 (99.99 %), GeO2 (≥ 99.99 %), B2 O3 (≥ 98 %), all from Sigma Aldrich) for

11

B{ 10B} NMR

experiments, but otherwise followed the synthesis described above.

Sample Characterization PXRD was carried out using a Rigaku Ultima IV X-ray diffractometer equipped with a copper anode X-ray tube, a diffracted beam monochromator, and a scintillation detector. Diffractograms were collected from finely-ground samples in air at room temperature. 2θ was incremented with steps of 0.05° and a 2 s dwell time. Density of the glass sample was measured using Archimedes’ method, using absolute ethanol as the working fluid. Density measurements were controlled for variations in temperature.

Nuclear Magnetic Resonance Spectroscopy 11

B MAS NMR spectra were collected on a 16.4 T (224.63 MHz

11

B frequency) Bruker

Avance NMR spectrometer. Spectra were collected at a spinning speed of 25 000(5) Hz using 6


Page 7 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a 2.5 mm ZrO2 rotor. Single pulses of 0.54 µs duration (15° tip angle, νrf = 77 kHz) and an optimized recycle delay of 5 s were used to ensure quantitative excitation. 512 scans were collected for each sample.

11

B background from the probe was subtracted from the

experimental spectrum after careful phasing and intensity adjustment. Probe background spectra were collected immediately following the collection of sample spectra, using the same experimental conditions.

11

B chemical shifts were referenced to solid NaBH4 , with a

resonance at −42.1 ppm relative to the primary shift reference (BF3 · Et2 O, 0 ppm). 28 10

B MAS NMR spectra were collected on a 16.4 T (75.24 MHz

10

B frequency) Bruker

Avance NMR spectrometer. Spectra were collected using a HXY 3.2 mm probe head operating in HX mode.

10

B Hahn echo spectra were collected at a spinning speed of 10 000(5) Hz to

allow for the observation of many spinning sidebands. Pulse lengths for the π/2 and π pulses were optimized to be 6 µs and 10 µs. Spectra were collected using a 20 s recycle delay and 256 scans. 17

O MAS and triple-quantum magic angle spinning (3QMAS) NMR spectra were collected

on both 9.4 T (54.24 MHz

17

O frequency) and 16.4 T (94.91 MHz

17

O frequency) Bruker

Avance NMR spectrometers. All spectra were collected at a spinning speed of 14 000(5) Hz using 4 mm ZrO2 rotors. MAS spectra at 16.4 T were collected using single pulses of 0.52 µs duration (10° tip angle, νrf = 53 kHz) and optimized relaxation delays of 0.5 s. MAS spectra at 9.4 T were collected using single pulses of 0.83 µs duration (12° tip angle, νrf = 40 kHz).

17

O

MAS spectra generally required between 8192 and 16384 scans to yield spectra of sufficient quality. 17

O 3QMAS spectra were collected using a three-pulse sequence with full echo acquisition. 29

Split-t1 spectra at 16.4 T were collected using optimized excitation and conversion pulses of 6.0 µs and 1.5 µs (νrf = 53 kHz) and a 22 µs 180° selective pulse (νrf = 22 kHz). The same parameters were used for crystalline and glassy samples. 24 slices in the F1 dimension were collected, with 1152 transients per slice. All

17

O NMR spectra were referenced to tap water

(0 ppm). The 3QMAS spectra were referenced using the Cz convention. 30,31

7



Static

139

(56.52 MHz

La spectra were collected at 16.4 T (98.91 MHz

Page 8 of 51

139

La frequency) and 9.4 T

139

La frequency) using the WCPMG pulse sequence. 32 The WCPMG sequence

combines wideband, uniform rate, smooth truncation (WURST) 33 shaped pulses with the Carr-Purcell Meiboom-Gill (CPMG) signal enhancement protocol. 34,35 50 µs WURST-80 pulses were swept over 500 kHz at a rate of 10 MHz ms−1 . Between 100 and 200 echoes were collected per transient, depending on the spin-spin relaxation time (T2 ) of the sample. Due to the breadth of the peaks, spectra were collected piece-wise by moving the transmitter in increments of 100 kHz and added together, i.e., variable-offset cumulative spectrum (VOCS) collection. 36,37 Each sub-spectrum was collected by co-adding 512 transients, with an optimized 139

La spectra were referenced to a 1 mol dm−3 solution of LaCl3

relaxation delay of 0.5 s. (0 ppm). 11

B{ 10B} REDOR spectra were collected using the conventional pulse sequence. 38 Spectra

were collected using a 3.2 mm HXY probe head, with ZrO2 rotors spinning at 10 000(5) Hz. REDOR experiments were conducted on a sample with natural abundance oxygen.

11

B π/2

and π pulses were set to 3.10 µs and 6.25 µs and were optimized to provide a reasonable balance between optimum excitation for the BO3 and BØ− 4 environments. optimized using Hahn echo experiments.

11

B pulses were

10

B π pulses were optimized via maximizing the

REDOR difference, resulting in a 12.5 µs pulse length. REDOR pulse powers were 80 kHz for

11

B and 40 kHz for

10

B. Recycle delays were optimized via spin-lattice relaxation (T1 )

saturation recovery experiments, and were set to five times the longest T1 observed in a given sample. These delays ranged from 50 s to 225 s. The number of scans collected varied depending on the sample, and ranged from 8 to 48. Spectra of all nuclides studied were fitted using the int2quad module of the Dmfit 20150521 39 software package. Fits were broadened using Lorentzian functions for crystalline environments, and Gaussian functions for amorphous environments. The glass components of 139

La spectra were fitted using the Czjzek distribution, 40–42 which is discussed in more detail

below.

8


Page 9 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Density Functional Theory Calculations First-principles density functional theory (DFT) calculations were performed using the Quantum ESPRESSO code (https://www.quantum-espresso.org/), version 5.1.3. 43,44 In order to calculate NMR observables, the gauge-including projector augmented wave (GIPAW) method as implemented in version 5.1 of the QE-GIPAW package was used. 45,46 Custom projector augmented wave (PAW) datasets using the Perdew, Burke, and Ernzerhof (PBE) exchange and correlation functional, 47 based upon the Jollet, Torrent, and Holzwarth set, 48 were constructed to avoid the issue of PAW sphere overlap. 49 PAW datasets were generated using atompaw. 50 An optimized plane-wave cutoff energy of 60 Rydberg was used with a shifted 6 × 6 × 6 Monkhorst-Pack k-point grid. The latter corresponds to a k-point spacing . Results from calculations on the experimental structure 3 are presented. For of 0.025 A −1

the electric field gradient (EFG) calculations, nuclear quadrupole moments were taken from Pyykkö. 51 To convert the chemical shielding values provided by GIPAW to experimentallyobservable chemical shifts, calculated shielding values were compared to experimental shifts for known systems. Details of the conversion are reported in the supporting information.

Raman Spectroscopy Unpolarized Raman spectra were collected using a Nicolet NXR 9600 Fourier-transform Raman spectrometer. A 1064 nm laser was used to excite the samples. The spectra were collected with a spectral resolution of 2 cm−1 . The glass spectrum was collected by summing 256 scans at a power of 400 mW. The crystal spectrum was collected by summing 128 scans at a power of 70 mW. The glass spectrum is of a single piece of glass, while the crystal spectrum was collected from a powdered sample.

9



Results X-Ray Diffraction The X-ray diffraction results for LBG-G, LBG+12, and LBG+60 are shown in Fig. 2. The

2500

Normalized Intensity / arb. units

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 51

2000

1500

1000

500

0 10

20

30

40

50

60

2θ / ° Figure 2: X-ray diffractograms of: glassy LBG-G, top; partially crystallized LBG+12, middle; and mostly crystallized LBG+60, bottom. Diffractograms are offset by 1000. The intensities of the LBG+12 and LBG+60 diffractograms are normalized such that the most intense peak has a value of 1000. The intensity of the LBG-G diffractogram is normalized to 100 for ease of comparison to the partially crystallized samples. The expected position and intensities of the reflections of crystalline LaBGeO5 are marked with squares (PDF 41-0659). 52 diffractogram for the glass sample, LBG-G, shows no sharp diffraction peaks, consistent with an amorphous material. The diffraction patterns of LBG+12 and LBG+60 are consistent with the diffraction pattern of LaBGeO5 as reported by Rulmont and Tarte. 52 Despite the 10


Page 11 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


substantial difference in the level of crystallization between these two samples (see below), the diffractograms appear similar. Both show diffraction peaks from a single crystalline phase, identified as LaBGeO5 , though there is some slight variation in both peak intensity and position. Based on Rietveld refinement of the diffractograms (see Supporting Information), both the a and c lattice constants are somewhat smaller (approx. 1 %) in the partially crystallized samples than what is reported in the literature. 3,4,52 The density of the LBG-G glass sample was found to be 4.89(2) g cm−3 , consistent with previous literature reports on this composition. 9,19

Raman Spectroscopy The Raman spectra of the LaBGeO5 crystal and glass are presented in Fig. 3. The spectrum of the LaBGeO5 crystal is very similar to previous literature reports, and its peaks have been assigned to various vibrational modes. 14,52–55 The particular region of interest for this study is between 775 cm−1 to 900 cm−1 , which has been assigned to both BØ− 4 and GeO4 tetrahedral vibrations. Notably, there is no significant Raman intensity in the crystal between 630 cm−1 to 775 cm−1 . The Raman spectrum of the LaBGeO5 glass has a broad, asymmetric, and featureless peak spanning 630 cm−1 to 950 cm−1 . Like the spectrum of the crystal, our spectrum of the glass is consistent with literature reports. 14,53 The large peak in the glass encompasses the intensity found in the crystal, but also has significant intensity centred at approximately 745 cm−1 ; the crystal has no corresponding vibrational modes in this region. There is a moderate intensity peak in the glass spectrum located at approximately 545 cm−1 that is also absent from the crystal spectrum.

11

B NMR Spectroscopy

11

B MAS NMR spectra of LBG-G, LBG+12, and LBG+60 are shown in Fig. 4. The spectrum

of the glass sample, LBG-G, has two peaks: one centred around 15 ppm, and a second peak centred around 1 ppm. The position, breadth, and relative intensity of these peaks are 11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 51

Intensity / arb. units


300

400

500

600 700 800 -1 Raman shift / cm

900

1000

1100

Figure 3: Raman spectra of the LaBGeO5 crystal (solid line) and LaBGeO5 glass (dotted line). Some of the Raman intensity in the glass corresponds to crystal modes, in particular the sharp mode at approx. 803 cm−1 . However, the glass also has substantial intensity which does not correspond to known LaBGeO5 environments.

12


Page 13 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


LBG-G LBG+12 LBG+60

30

25

20

15 5 10 11B chemical shift / ppm

0

-5

-10

Figure 4: 11B MAS NMR spectra of: glassy LBG-G, top; partially crystallized LBG+12, middle; and mostly crystallized LBG+60, bottom. consistent with previous reported spectra from this composition. 13,19 The peak centred around 15 ppm is consistent with the presence of BO3 units, while the peak centred around 1 ppm 56 is consistent with the presence of BØ− Based upon 3QMAS spectra of lanthanum 4 units.

borogermanate glasses, we have previously identified this three-coordinate boron species CS = 18.3(2) ppm, as a single BØ2 O− . 19 The isotropic chemical shift of the BO3 peak, δiso

was obtained by fitting the peak of the

11

B MAS NMR spectrum of LBG-G with a single

quadrupolar environment, subject to Gaussian broadening. This shift is consistent with the CS chemical shift of the BØ2 O− peak in crystalline LaB3 O6 , δiso = 17.9(2) ppm (Fig. S5), as

well with literature reports of chemical shifts of three-coordinate borate units, 57 including those published with our 3QMAS spectra of LaBGeO5 glass (Tables 1 and 2). 19 The characterization of the BØ− 4 environments is less straightforward. The chemical shift range of the BØ− 4 structural unit is small, with chemical shift changes in response to changes in composition typically on the order of 2 ppm to 3 ppm. 59 Additionally, broadening of the NMR peaks as a consequence of the structural disorder inherent in glassy systems can obscure 13



Page 14 of 51

Table 1: 139La, 10B, 11B and 17O NMR parameters of environments in the glass. 139 La and 11B values are from experiments on the LBG-G sample, and parentheses next to these values indicate experimental uncertainties. 17O values are either means taken from our experiments on the LaBGeO5 crystal (Table 2), or means taken from the literature. For experimental data, experimental uncertainties are indicated. For literature data, standard deviations are indicated. Nucleus Environment

CS 3QMAS Label δiso / ppm

139

437(30)*

La

10 11

17

B B

BO4 BO3 BO4

4.2(2)** This work 1.0(4) 2.6(2) 0.8(4)

This work 19 19

O

Ge – O – 4Ge Ge – O – B 4 B – O – 4B 4 Ge – O – 5/6Ge O666 (rutile GeO2 ) 4 Ge – O – La B – O – La O666 (Na4 Ge9 O20 ) 4

* **

‡

Sources

† 4

†

0.8(8) 18.3(2) 2.0(3)

PQ / MHz

a b c d e f g h

63(12) 88(5) 95(2) 124(30) 156(6) 189(4) ‡ 220 246.5(5)

6.7(7) 15,25 5.2(2) This work 6.1(3) This work 6.5(5) 15,25 7.35(2) 15,25 5.2(2) This work 58 3(1) 3.75(5) 15

CS This value is the position parameter of the peak, and does not correspond directly to δiso for 139La. This value is the Czjzek distribution parameter σ, not PQ . Both are expressed in MHz. 17 O values for environments not observed in the LaBGeO5 crystal are averages from literature reports. The uncertainties reported are an estimate of the precision of these values, and not an estimate of the range, which can vary. CS values of the two B – O – La environments in LaBO3 . Average of the calculated δiso

14


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


15

‡

†

*

0.50(4) 0.54(5) 0.22(3) 0.20(3) 0.75(6)

5.2(2) 5.2(2) 5.2(1) 5.2(1) 6.1(3)

86.7(5) 0.9(4)

PQ / MHz

86(3) 93(3) 185(3) 191(3) 96(3)

CS * δiso / ppm

5.3(5) 5.4(5) 5.1(5) 5.3(5) 5.6(5)

PQ / MHz

3QMAS

111 116 233 239 98

215 2

CS * δiso / ppm

−87.1 0.4 41.3 −5.1 −5.1 −4.5 −4.6 −5.8

CQ / MHz

0.32 0.97 0.84 0.60 0.52 0.14 0.19 0.77

ηQ

Calculated −88.6 0.5 45.8 −5.4 −5.3 −4.5 −4.7 −6.4

PQ / MHz

Experimental 139La, 11B, and 17O chemical shifts are referenced to 1 mol dm−3 LaCl3 (0 ppm), NaBH4 (−42.1 ppm), and tap water (0 ppm), respectively. Details on the referencing of the chemical shifts obtained from DFT calculations are provided in the supporting information. 139 La NMR parameters are consistent with those reported in the literature. 24 73 Ge NMR experiments were not carried out. Calculated values are included for completeness.

5.0(2) 5.0(2) 5.2(1) 5.1(1) 5.6(2)

85(2) 92(2) 186.4(5) 192.5(5) 95(2)

0.3(2) 0.8(2)

85.5(5) 0.8(4)

La† B Ge‡ O1 O2 O3 O4 O5

210(25) 1.3(2)

ηQ

CS * Site δiso / ppm CQ / MHz

MAS or WCPMG

Table 2: Parameters used to fit experimental data, and parameters from DFT calculations, for 139La, 11B, and 17 O NMR of crystalline LaBGeO5 . Chemical shift anisotropy parameters are not reported, but generally are expected to be a minor contribution to the lineshape of the peak. For MAS and WCPMG spectra uncertainties derived from the fit are provided; for 3QMAS spectra, fundamental uncertainty due to the resolution of the experiment are provided.

Page 15 of 51 The Journal of Physical Chemistry


Page 16 of 51

fine details. Here, the BØ− 4 peak in the glass spectrum resonates more positively than that in the crystal by 0.7(2) ppm. Our REDOR results provide some insight into the changes in the BØ− 4 environment, and are discussed below. The spectrum of the most crystalline sample, LBG+60 (Fig. 5), shows a dominant peak in 3 the region attributed to BØ− 4 units, consistent with the crystal structure of LaBGeO5 . There

is minor residual BO3 and BØ− 4 intensity from the initial glass component, but the change in the symmetry and breadth of the BØ− 4 peak allows for the chemical shift parameters of the crystalline BØ− 4 peak to be determined. The spectrum of a sample with intermediate crystallization, LBG+12, is a combination of the spectra of LBG-G and LBG+60. The fitted 11

B MAS NMR spectrum of LBG+12 is shown in Fig. 5. All partially crystallized samples

Expt.

Total fit Crystal

Glass

30

25

20

15 10 5 0 11B chemical shift / ppm

-5

-10

Figure 5: 11B MAS NMR spectrum of partially crystallized LBG+12. The parameters for the crystal fit are found in Table 2, while the parameters for the glass fit are found in Table 1. can be fitted with three peaks: two from LBG-G and one from LBG+60. The crystalline fraction of the partially crystallized samples can be determined by subtracting the spectrum of the glass, LBG-G, and integrating the residual signal, or by fitting the crystalline and glass contributions of the BØ− 4 peak directly. Other than the relative intensity of the glass 16


Page 17 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


component and the crystal component, no fitting parameters are required to be changed. Furthermore, the ratio of BO3 /BØ− 4 in the glass component, α, does not appear to change due to crystallization. This suggests that there are no changes to the structure of the bulk glass during crystallization, and that α can be considered to be constant for a given glass sample. The level of crystallization of the samples is reported in Table 3.

139

La NMR Spectroscopy

139

La WCPMG NMR spectra of LBG-G, LBG+12, and LBG+60 are shown in Fig. 6.

The spectrum of the most crystalline sample, LBG+60, is consistent with our previously

LBG-G

LBG+12

LBG+60 8000

0

4000 139La

-4000

-8000

chemical shift / ppm

Figure 6: 139La WCPMG NMR spectra of: glassy LBG-G, top; partially crystallized LBG+12, middle; and mostly crystallized LBG+60, bottom. As the spectra were collected using WCPMG, the signal intensity is concentrated into spikelets. The intensity envelope of the spikelets recreates the powder pattern of the static NMR spectrum. reported spectrum of the LaBGeO5 crystal. 24 However, the spectrum shows poor definition of the quadrupolar features, indicating that the crystallites have limited long-range order. Furthermore, there is a substantially narrower impurity peak present in the spectrum of 17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Time / min

0 8.0(5) 12(1) 24(2) 60(3)

Sample

LBG-G LBG+8 LBG+12 LBG+24 LBG+60


18 0.65(3) 0.55(3) 0.46(3) 0.34(3) 0.18(3)

[BO3 ] 0.35(3) 0.45(3) 0.54(3) 0.66(3) 0.82(3)

[BO4 ]

Integration

0.63(1) 0.54(1) 0.44(1) 0.33(1) 0.17(1)

[BO3 ]

Fit

0.37(1) 0.46(1) 0.56(1) 0.67(1) 0.83(1)

[BO4 ]

11

0 0.15(1) 0.29(1) 0.48(1) 0.72(1)

B Fit

c

0 0.13(5) 0.26(5) 0.55(5) 0.78(5)

0.0(3) 0.3(2) 0.5(2) 0.7(1) 0.88(7)

O Integration Gupta et al. 13

17

0.00(6) 0.15(5) 0.28(5) 0.47(4) 0.73(3)

Eq. (7)

Table 3: Fraction of the sample composed of LaBGeO5 crystallites as determined by various means. Integrated BO3 and BØ− 4 intensities are reported, as well as intensities determined by peak fitting. Values of c as determined by fitting 11B MAS NMR and integrating 17O MAS NMR data are compared to calculated values.

The Journal of Physical Chemistry Page 18 of 51

Page 19 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


LBG+60, LBG+12, and all other partially crystallized samples. This peak is consistent with that of LaBO3 . 24 Relative intensities are not quantitative in WCPMG spectra, as the signal enhancement due to the CPMG loops depends on the T2 of each different environment. However, the signal due to LaBO3 is consistently quite small, and no indication of LaBO3 is seen in either the pXRD diffractograms (Fig. 2) or

11

B MAS NMR spectra (Fig. 4). As

such, we conclude that the amount of LaBO3 present in these samples is very small (≤ 3 %). This is consistent with our Rietveld refinements of or pXRD data (Fig. S1) and thus we will generally neglect it as a factor in our analysis. The

139

La WCPMG spectrum of the glass

sample, LBG-G, is fit with a Czjzek distribution. 40–42 A fit of the

139

La WCPMG spectrum of

LBG+12, processed such that the CPMG spikelets form a continuous lineshape, is presented in Fig. 7. It is important to note that the breadth parameter of the Czjzek distribution,

Expt. Total fit Glass Crystal LaBO3 8000

4000 139La

0

-4000

-8000


Figure 7: 139La WCPMG NMR spectrum of partially crystallized LBG+12. The experimental data is identical to that shown in Fig. 6, but by summing the CPMG echoes the spectrum becomes a continuous lineshape. Fitting parameters for the crystal LaBGeO5 peak are reported in Table 2, and for the glass in Table 1. The spectral parameters for both LaBGeO5 and LaBO3 are previously reported in the literature. 24 σ, is not directly comparable to the well-defined CQ , ηQ , or PQ of crystalline materials. If 19



Page 20 of 51

there were significant changes to the second coordination sphere of the LaBGeO5 glass during crystallization, we would expect to observe changes in either σ or the position parameter between samples. No change in the

139

La peak attributed to the glass is observed during

crystallization, and so we conclude that the bulk glass does not undergo significant changes during the crystallization process.

11

B{10 B} REDOR NMR Spectroscopy

As has been previously reported, BO3 environments cannot be effectively excited in NMR spectroscopy. 23 With integer nuclear spin (I = 3),

10

B

10

B spectra lack a narrow “central

transition”, hence the spectra consist only of the quadrupolar broadened satellite transitions similar to those present in non-integer-spin nuclei. 60 For environments with large quadrupolar coupling constants, such as BO3 , detecting appreciable signal following a single pulse is nearly impossible. In contrast, for environments with small quadrupolar coupling constants, such as BØ− 4 , a single pulse can potentially excite all of the satellite transitions. Figs. S7 and S8 shows the

B Hahn echo NMR spectra of LaBGeO5 glass and crystal. Only the BØ− 4

10

environments are excited and both spectra have similar widths. In short, a single pulse, − centred on the BØ− 4 resonance, will effectively excite the BØ4 spins without exciting the

BO3 nuclei, and with similar efficiency. Hence,

11

B{ 10B} REDOR can be used to selectively

− − probe BØ− 4 –BØ4 and BO3 –BØ4 connectivity in glass and crystals.

We present

11

B{ 10B} REDOR curves of LBG-G, crystalline LaBGeO5 (LBG-X), and

− LaB3 O6 in Fig. 8. The REDOR curves of the BØ− 4 –BØ4 environment of crystalline LBG-X

and the BO3 –BØ− 4 environment of LaB3 O6 are used to interpret the behaviour of the similar environments in LaBGeO5 glass. REDOR experiments evaluate the parameter ∆S/S0 , i.e., the difference between the integrated intensities of the peak of interest when dipolar coupling is removed via MAS, and when the dipolar coupling is reintroduced, normalized to the intensity when the dipolar coupling is removed (S0 ). Typical analysis of REDOR spectra of amorphous samples involving nuclei with half20


Page 21 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8: 11B{ 10B} REDOR NMR curves for crystalline LaB3 O6 , crystalline LaBGeO5 and LaBGeO5 glass. Filled data points indicate data from crystals. Open data points indicate − data from the glass. Diamonds indicate BØ− 4 –BØ4 environments. Triangles indicate BO3 – − BØ4 environments. a) The REDOR curves at long time scales. b) The REDOR curves at short time scales. 21



Page 22 of 51

integer spins involves fitting ∆S/S0 in the region 0 ≤ ∆S/S0 ≤ 0.2 to extract the value of the van Vleck second moment. 23,61–64 At these short time scales, the REDOR lineshape of spin I = 1/2 pairs follows the quadratic “universal lineshape”. 23,62 However, our curves are effectively linear at both long and short time scales. Such problems are known for REDOR pairs that involve one or two quadrupolar nuclei 65 and prevent us from a detailed quantitative analysis of the REDOR curves. Regardless, there is value in a qualitative comparative analysis. The rate of change of a REDOR curve is indicative of the strength of the dipolar interaction between the observed and indirect nuclide. For an ensemble of spins such as a glass, this strength is affected primarily by the average distance between the nuclides. Hence a decrease in the rate of change of the REDOR curve can be attributed to a decrease in the dipolar interaction, which in turn is attributed to a decrease in the proximity of the indirect and the observe nuclides. For this interpretation to be reliable between different systems, the proportion of excited spins must be similar. This is not difficult to achieve for central transitions of the

11

field strength. Our

B environments at 16.4 T, but it is not obviously true for

10

B at the same

10

B Hahn echo NMR spectra of our samples (Figs. S6–S8) are of similar

breadth, and are fitted with similar values of CQ (reported in Table S4). By moving the experimental excitation frequencies, we additionally confirmed that the

B BØ− 4 environment

10

is consistently fully excited with a single pulse across our three samples. Therefore we can meaningfully compare differences in intensity between the REDOR curves. − − The REDOR curves of the BO3 –BØ− 4 and BØ4 –BØ4 environments in the LBG-G sample

are of non-zero intensity, are similar to each other, and are persistent over long length scales (50 ms). They are of lower magnitude than the curves for the equivalent environments in the LBG-X and LaB3 O6 samples. From this we conclude that the BØ− 4 environment in LaBGeO5 glass has both BO3 and BØ− 4 neighbours. We also conclude that due to the reduced magnitude − − of the ∆S/S0 curve for the BØ− 4 environment in LBG-G, the BØ4 –BØ4 environment is

less likely to occur in LBG-G than in LBG-X. In LBG-X, the single crystallographic BØ− 4

22


Page 23 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


− site has two BØ− 4 neighbours; hence in the glass we expect that BØ4 has, on average,

significantly fewer than two BØ− 4 neighbours. Similarly, in the LaB3 O6 crystal structure, both crystallographic BO3 sites have two BØ− 4 neighbours, and so we conclude that due to the reduced magnitude of the BO3 –BØ− 4 curve in LBG-G that, on average, BO3 has − significantly fewer than two BØ− 4 neighbours. With the relative magnitudes of the BO3 –BØ4 − and BØ− 4 –BØ4 ∆S/S0 curves in the LBG-G glass being similar at all time scales, it seems − likely that the BO3 and BØ− 4 environments have similar numbers of BØ4 neighbours.

17

O NMR Spectroscopy

17

O MAS NMR spectra of LBG-G, LBG+12, and LBG+60 collected at 16.4 T are shown

in Fig. 9. All spectra show two major distinct signal regions, with maxima at 72 ppm and

LBG-G

*

*

LBG+12

*

*

LBG+60

*

* 350

* 300

250 17O

* 200

150

100

50

* 0

-50


Figure 9: 17O MAS NMR spectra of: glassy LBG-G, top; partially crystallized LBG+12, middle; and mostly crystallized LBG+60, bottom. Spinning sidebands are indicated by *. The first negative spinning sidebands from the non-bridging oxygen environments in the LBG+60 sample partially overlap the peak resulting from the bridging oxygen environments. The same is true for the LBG-G sample; additionally, the first positive spinning sidebands from the bridging oxygen in said sample overlap with the peak from the non-bridging oxygen.

23



Page 24 of 51

170 ppm for LBG+60. These are fully resolved in the spectrum of the mostly crystallized LBG+60 with distinct shapes, while the two features in the spectrum of the LBG-G glass are broad, overlap and mostly Gaussian shaped. The

17

O MAS spectrum of the LBG-G glass

sample can be subtracted from any of the spectra of the partially crystallized samples, using scaling factors that match the intensity of the edge at approximately 230 ppm. The derived difference spectra for the partially crystallized samples have the same shape and are similar to the spectrum of the most crystallized sample, LBG+60, a fact that was also observed for the

11

B MAS and

139

La WCPMG NMR spectra. As stated above, the independence of the

spectral portion attributed to LaBGeO5 crystallites from the level of crystallization in both 11

B MAS and

139

La WCPMG NMR spectra, as well as the

17

O MAS spectra, indicates that

neither the bulk glass structure nor the crystallite structure change during the crystallization process. The relative integration of these spectral decomposition into glassy and crystalline environments can be used to estimate the crystallite fraction (Table 3). The values determined by

17

O NMR agree well with those from

11

B NMR (Table 3). For further analysis, since even

the spectrum of the most crystalline sample, LBG+60, shows a small amount of residual glass intensity, we subtract the scaled spectrum of the glass sample from the spectrum of LBG+60 and fit the difference, in order to minimize error during the fitting of the

17

O NMR

spectrum of the LaBGeO5 crystal. There are five crystallographically distinct oxygen sites in the LaBGeO5 crystal structure. 3 They are summarized in Table 4. O1 and O2 link GeO4 and BØ− 4 tetrahedra, and are Table 4: The distinct oxygen environments in the LaBGeO5 crystal. The use of ’ indicates a second distinct atom. The labels correspond to the structure reported by Kaminskii et al. 3 Oxygen Type Neighbours O1 O2 O3 O4 O5

BO BO NBO NBO BO

Ge, B, La, La’ Ge, B, La, La’ Ge, La, La’ Ge, La, La’ B, B’, La

24


Page 25 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


considered bridging oxygen (BO), as they link two glass-forming elements (Ge and B). O1 and O2 are four-coordinate, singly bound to B and Ge, and loosely bound to two different La. O3 and O4 link GeO4 tetrahedra with LaO9 polyhedra, and are three-coordinate: singly bound to Ge, and singly bound to two different La. O3 and O4 are considered non-bridging oxygen (NBO), as they link a network former (Ge) with La, which often acts as a network modifier. O5 links two corner-sharing BØ− 4 tetrahedra, is three-coordinate (one La and two different B neighbours), and is considered a BO. After the glass component of the spectrum is removed, the integration of the two major peaks with maxima at 72 ppm and 170 ppm of the

17

O MAS spectrum of LBG+60 returns a 3:2 intensity ratio. A small impurity peak at

approximately 213 ppm is attributed to LaBO3 based upon pXRD and

139

La NMR data and

DFT calculations. The peak centred at 72 ppm is attributed to the three BO sites, while the peak centred at 170 ppm is attributed to the two NBO sites. The two NBO peaks are not resolved from one another; similarly, all three of the BO peaks are strongly overlapping. The 3QMAS spectrum of LBG+60 is presented in Fig. 10. As in the MAS spectrum, the NBO peak pair is resolved from the BO triplet. The 3QMAS spectrum shows some separation of the two NBO sites, while only two distinct features seem to resolve for three BO sites. Nevertheless, using the procedure described by Millot and Man, 31 reasonable estimates CS (and upper bounds) of the PQ and δiso values of a given environment can be extracted. The CS derived values support that the two NBO sites have similar CQ , ηQ , and δiso values and that

the three BO sites are not substantially different from one another. The limited resolution of the 3QMAS spectrum is a result of the fairly rapid decay of the 3QMAS echo. To aid in the interpretation of the

17

O NMR spectra of the LaBGeO5 crystal, first

principles calculations were carried out on the experimental structure. 3 The results of the calculations for the

17

calculations returned

O, 17

139

La and

11

B NMR parameters are presented in Table 2. DFT

O EFG parameters in good accord with those extracted from the

3MQAS spectrum. The relative positions of the NBO and BO calculated

17

O chemical shifts

CS CS CS are correct (with δiso (NBO) > δiso (BO)), but the calculated value of δiso (NBO) is at least

25


17O

Page 26 of 51

0 50

100 150 200 250 300 350 400

275 250 225 200 175 150 125 100 17O

75

50

25

0

50

25

0

MAS dimension / ppm

isotropic dimension / ppm

(a) Without fit.

17O

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



0 50

100 150 200 250 300 350 400

275 250 225 200 175 150 125 100 17O

75

MAS dimension / ppm (b) With fit.

Figure 10: 17O 3QMAS NMR spectrum of mostly crystallized LBG+60. The diagonal is the correspondence between the MAS and isotropic dimensions, and indicates the position where peaks free from the quadrupole interaction would appear. Fitted peaks are overlaid on experimental data; the parameters of the fitted peaks are reported in Table 2. The top left and bottom right peaks are spinning sidebands. The projection of the isotropic dimension (y axis) is processed to remove the contributions from the spinning sidebands.

26


Page 27 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


20 ppm greater than any possible experimental value. Similarly, the calculated values of CS δiso (BO) are at least 10 ppm greater than any possible experimental value. The calculated CS δiso values for the BO sites are within 20 ppm of each other, with the two Ge–O–B sites

having similar values. The relative separation of the two Ge–O–La sites is 6 ppm. Similarly, the relative separation of the two Ge–O–B sites is 5 ppm. This is lower than the resolution limit of our 3QMAS spectrum, and hence it is unsurprising that there is difficulty in resolving individual environments. CS The results of our GIPAW calculations imply that we should expect similar values of δiso ,

CQ , and ηQ for a given pair of Ge–O–B or Ge–O–La sites. Both the CQ and ηQ values of the B–O–B site are greater than of any other sites. The ηQ values of the NBO sites are much lower than those of the BO sites. For all sites, the chemical shift anisotropy (CSA) contribution to the breadth of the peak is expected to be substantially less than the quadrupolar contribution, hence we neglect it in our fits. With these trends, we can generate a reasonable set of fitting parameters that simultaneously fit the 16.4 T and 9.4 T MAS, Hahn echo, and 3QMAS spectra. The fit of the 16.4 T 3QMAS spectrum, and of the 16.4 T and 9.4 T MAS spectra, are shown in Figs. 10 and 11. The parameters for this fit are reported in Table 2. No similar fit can be constructed for the

17

O MAS NMR spectrum of the LaBGeO5 glass. There are simply too many possible

environments, and insufficient resolution, to provide a quantitative estimate of any parameter. However, based upon the 3QMAS spectrum of the glass (Fig. 12) a qualitative model of the glass structure can be constructed. The presence of some structural elements are known to be present in the glass from other experiments (i.e., B–O–La); others are expected due to the structure of the LaBGeO5 crystal (i.e., Ge–O–La, Ge–O–B); yet others can be inferred based upon data in the literature. The resulting qualitative interpretation of the glass structure is discussed in detail below. It is noteworthy that we do not observe any

17

O intensity in the 450 ppm to 600 ppm

region in any spectrum. Intensity in this region would indicate the presence of OLa4 or OLa6

27



Page 28 of 51

† Expt. Total fit O1 (4Ge-O-4B)

O2 (4Ge-O-4B) O3 (4Ge-O-La) O4 (4Ge-O-La) O5 (4B-O-4B) 300

250

200

150

100 17O

gipaw_compromise_v7_FINAL.fxml

Expt.

50

0

-50

-100

-150


(a) Spectrum at 16.4 T.

†

Total fit O1 (4Ge-O-4B) O2 (4Ge-O-4B)

O3 (4Ge-O-La) O4 (4Ge-O-La) O5 (4B-O-4B)

250

200

150

100 17O

50

0

-50

-100

-150


(b) Spectrum at 9.4 T.

Figure 11: 17O MAS NMR spectra of the mostly crystallized LBG+60. The parameters of each peak are provided in Table 2. A small impurity is marked with †, and is consistent with the presence of LaBO3 .

28


Page 29 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 12: 17O 3QMAS NMR spectrum of the glass LBG+G. The diagonal is the correspondence between the MAS and isotropic dimensions, and indicates the position where peaks free from the quadrupole interaction would appear. The projection of the isotropic dimension (y axis) is processed to remove the contributions from spinning sidebands.

29



Page 30 of 51

environments as in crystalline La2 O3 . 66 The presence of these environments could indicate partitioning of the glass into La-rich regions. 67 As we do not observe any such intensity, we conclude that La3+ is regularly distributed in the glass.

Discussion Crystallization The pXRD,

11

B MAS NMR,

139

La WCPMG NMR, and

17

O MAS and 3QMAS results all

support a common conclusion regarding the formation of LaBGeO5 crystals from the glass: there are no substantial structural changes in either the glass or the crystal in the partially crystallized samples. In other words, the glass fraction of a partially crystallized LaBGeO5 sample has the same structure as the parent glass, while the crystalline fraction is unmodified LaBGeO5 crystal. Furthermore, the quantitative

11

B and

17

O MAS NMR results provide

the same crystallization fraction (within error), with partially crystallized spectra being combinations of the crystal and glass spectra. There are some noticeable differences in the pXRD and

139

La WCPMG NMR results of

the different samples. Both become somewhat sharper as the crystallization fraction increases. This change is indicative of larger crystallites, but not of a substantial change in either the short-range order (from NMR) or long-range order (from pXRD) of the LaBGeO5 structure. At the highest crystallization levels, a small amount of LaBO3 is observed, first by

139

La

WCPMG NMR, then by pXRD (Fig. S1). The LBG-G glass likely has a small excess of B2 O3 , allowing for non-stoichiometric crystallization in small quantities. However, LaBO3 is only detected when the sample is already substantially crystallized. Finally, in the partially crystallized samples, the a and c lattice constants are somewhat lower than those reported in the literature. Based upon the amount of strain determined by our Rietveld refinement (Table S1), along with the elastic tensor of the LaBGeO5 crystal, 55 we conclude that the LaBGeO5 crystallites are under compressive stress, estimated to be on the order of about 30


Page 31 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4 GPa. The data above provide insight into the end result of the nucleation and growth of the LaBGeO5 crystallites, but do not effectively probe the dynamics of the nucleation mechanism. The most interesting structural evolution during nucleation and growth occurs at the interface between the glass and the nucleus. By definition the nuclei are extremely small, with radii on the order of nanometres. The faction of the total sample volume made up by the interface between nuclei and glass is extremely small, and hence effectively undetectable via either NMR or pXRD. While we do not directly probe the changes in structure during nucleation, knowledge of the structure and the extent of the crystallization of the resulting glass-ceramic has practical significance. The ability to estimate the degree of crystallization of a LaBGeO5 glass-ceramic composite via NMR spectroscopy is desirable, as estimating volume fraction via techniques such as optical microscopy may prove difficult if the crystallites are sufficiently small such that the material is transparent. Gupta et al. have previously reported a model which, through the use of

11

B MAS NMR spectroscopy, allows for an estimate of the crystallite fraction. 13 However,

when we apply their model to our

11

B MAS NMR results, the model disagrees with our direct

modelling of the BØ− 4 peak, as well as with the crystallization fraction estimated by our

17

O

MAS results. A small correction to the model of Gupta et al., when applied to our presented samples, leads to values which agree very well with our experimental results. The details of the correction follow. We use the same notation as Gupta et al., where square brackets indicate the relative intensity of the species within. In the LaBGeO5 crystal, [BO3 ] = 0 and [BO4 ] = 1. In the glass, both species will be present, though the precise ratio will depend on the exact 19 composition (e.g., an excess of La2 O3 will increase BO3 at the expense of BØ− In a 4 ).

partially crystallized sample, [BO3 ]expt + [BO4 ]expt = 1

31


(1)


Page 32 of 51

and [BO4 ]expt = [BO4 ]cryst + [BO4 ]glass .

(2)

As the crystal contains no BO3 units,

[BO3 ]expt = [BO3 ]glass . We assume that the ratio of BO3 /BØ− 4 in the glass, α =

(3) [BO3 ]glass , [BO4 ]glass

does not change as

crystallization progresses. This assumption is well-founded based on our

11

B,

139

La, and

17

O

NMR data, none of which show any change in the structure of the LaBGeO5 bulk glass during crystallization. With the assumption that α is constant, the fraction of the four-coordinate boron intensity attributed to the crystallites can be calculated knowing only the fraction of the overall intensity attributed to the three-coordinate site. The fraction of the crystallites in the partially crystallized sample, c, is given by

c=

[BO4 ]cryst . [BO4 ]cryst + [BO4 ]glass + [BO3 ]glass

(4)

With Eq. (2), this can be rewritten to remove the (potentially) inaccessible [BO4 ]cryst term: [BO4 ]expt − [BO4 ]glass [BO4 ]expt − [BO4 ]glass + [BO4 ]glass + [BO3 ]glass [BO4 ]expt − [BO4 ]glass c= [BO4 ]expt + [BO3 ]glass

c=

(5)

As [BO3 ]glass and [BO4 ]glass are related by α, Eq. (5) can be written as [BO ]

3 glass [BO4 ]expt − α c= [BO4 ]expt + [BO3 ]glass

(6)

With Eqs. (1) and (3), Eq. (6) is reduced to

c = [BO4 ]expt − 32

[BO3 ]expt . α


(7)

Page 33 of 51

Equation (7) is used to generate the crystallite fraction reported in Table 3. It agrees very well with the fraction determined by fitting the BØ− 4 peak directly, and by integration of

17

O

MAS spectra. The only difference between our derivation and the derivation of Gupta et al. is the introduction of the [BO3 ]glass term in Eq. (4). However, this term is required to accurately account for the total intensity of the glass, and prevent an overestimation of the crystal fraction. For comparison, our experimental results, the results of our equation, and the results from the equation of Gupta et al. are plotted in Fig. 13.

1.0 0.8

0.6

c

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


0.4 ¹¹B ¹⁷O Gupta This work

0.2 0.0

0.3

0.4

0.5

0.6

0.7

0.8

0.9

[BO4]

Figure 13: A comparison of the crystal fraction, c, of the partially crystallized samples, as determined from both 11B and 17O MAS NMR data, and the crystal fraction predicted by model calculations. Solid points represent experimental data, while open points represent calculated values. Open circles are values calculated using the equation of Gupta et al., 13 while open squares are calculated using Eq. (7). The utility of Eq. (7) is primarily in the (relatively) accessible data it requires. Fitting the 33



Page 34 of 51

BØ− 4 peak directly may not be possible except at high magnetic field strengths and spinning speeds, while the

17

O enrichment required for

17

O MAS NMR is prohibitively expensive for

routine studies. The ratio of BO3 to BØ− 4 can be quickly determined through

11

B MAS

NMR at moderate field strengths and spinning speeds, or potentially by Raman spectroscopy, allowing for routine characterization of the crystal fraction of LaBGeO5 TFN materials.

Glass Structure There are two Raman bands in the spectrum of the glass which support differences in the crystal and glass structures. In particular, our Raman spectrum supports the presence of boron coordination changes, as well as high-coordinate germanium, in the glass. The band at approximately 545 cm−1 has been previous attributed to either the presence of 68 three-coordinate boron 14 or medium-range order units containing BØ− neither of which are 4,

present in the LaBGeO5 crystal. 3 Henderson et al. have reported a Raman band attributed to

5/6

Ge in alkali germanate glasses at a Raman shift of approx. 744 cm−1 . 69 While there is

no resolved peak observed at this shift in our Raman spectrum of the LaBGeO5 glass (Fig. 3), there is substantial intensity present. No intensity is observed at this shift in the spectrum of the crystal. Hence the substantial intensity at approx. 744 cm−1 in the glass can plausibly be attributed to the presence of high-coordinate germanium species, consistent with both our previously reported neutron diffraction data, 19 and as we will discuss below, our

17

O NMR

spectra. The structure of the LaBGeO5 crystal is well understood. Three structural units are 2– present: BØ− tetrahedra, and LaO9 polyhedra (Table 4). From this, 4 tetrahedra, GeØ2 O2

three oxygen bonding motifs are apparent: the two bridging oxygen environments, B–O–B and Ge–O–B; and the non-bridging oxygen environment, Ge–O–La. These three motifs are observed in the

17

O MAS and 3QMAS NMR spectra of the LaBGeO5 crystal. By comparison

to the crystal, some of the glass intensity can be assigned to the Ge–O–B environment CS CS (δiso ≈ 88 ppm), the B–O–B environment (δiso ≈ 95 ppm), and the Ge–O–La environment

34


Page 35 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


CS (δiso ≈ 189 ppm). However, this only accounts for a small portion of the intensity displayed,

since the

17

O MAS signal in the glass has spread to more positive chemical shift values.

For both homonuclear (e.g., Ge–O–Ge) and heteronuclear (e.g., Ge–O–B) environments, we must consider differences in coordination number, for both the oxygen and the next nearest neighbours. From our previous neutron diffraction study of the germanium coordination in lanthanum borogermanate glasses, we know that high-coordinate germanium species (i.e., five- and/or six-coordinate) exist in the LaBGeO5 glass. 19 Six-coordinate germanium in glass cannot be distinguished from a possible five-coordinate germanium environment by either spectroscopic or diffraction-based techniques, and hence both five- and six-coordinate germanium must be considered. 16,26,69 The notation

5/6

Ge indicates either or both envi-

ronments. Oxygen bonded to high-coordinate germanate species have been observed via 17

O NMR spectroscopy. Du and Stebbins have observed differences in the

17

O parameters

between 4Ge – O – 4Ge and 4Ge – O – 6Ge oxygen sites in binary sodium germanate crystals. 15 4

Ge – O – 5/6Ge environments have been observed in similar binary sodium germanate glasses,

as well as alkali germanosilicate glasses. 15,26,70 Intensity in the region between 80 ppm to 150 ppm has been attributed to five- and/or six-coordinate germanium in sodium germanate glasses, 15 with an average chemical shift of approximately 120 ppm. There is intensity present in the BO region of our 3QMAS NMR spectrum of the LBG-G glass sample, consistent with the established presence of

5/6

Ge in this composition.

The oxygen sites in the LaBGeO5 crystal structure have well-defined coordination numbers, including interactions from fixed numbers of lanthanum neighbours (Table 4). 3 These interactions are unlikely to be the same in the glass as in the crystal. In particular, it is very plausible that a given oxygen environment in the glass will have fewer lanthanum neighbours than in the crystal. 19 This slight reduction in coordination would slightly broaden the

17

O

NMR peaks to more positive chemical shifts. 71 In addition to observing 4Ge – O – 5/6Ge environments in sodium germanate glasses, Du and Stebbins have observed differences between the 3B – O – 3B and 3B – O – 4B environments, as

35



Page 36 of 51

well as the 3B – O – Si and 4B – O – Si environments, in sodium borosilicate glasses. 56 Lee et al. have reported

17

O 3QMAS spectra of binary borogermanate glasses, establishing the relative

positions of 4Ge – O – 4Ge, 4Ge – O – 3B, and 3B – O – 3B environments. 72 Changes in borate coordination are expected to shift

17

CS O δiso by approximately 10 ppm, for both Si – O – B and

B – O – B environments. 56 While there have been no studies examining the relative positioning of 3B – O – Ge and 4B – O – Ge in borogermanate glasses, we expect that the trends established in borosilicate glasses would extend to borogermanate glasses. Oxygen “triclusters” (i.e., oxygen triply bonded to network forming cations such as Ge4+ , labelled O666 ) are known to exist in the rutile crystal phase of GeO2 , and should be considered as a possible environment in germanate glasses. This environment is located at CS δiso = 152.2(5) ppm in crystalline rutile GeO2 , with a large PQ . 15

With the above literature data, we can establish that in binary germanate glasses, the 17

O chemical shift increases as the neighbouring germanate coordination increases (i.e.,

CS 4 CS 4 CS δiso ( Ge – O – 4Ge) < δiso ( Ge – O – 5/6Ge) < δiso (O666 )). The opposite is true with borates,

with the

17

O chemical shift increasing as the neighbouring borate coordination decreases (i.e.,

CS 4 CS 3 CS 3 δiso ( B – O – 4B) < δiso ( B – O – 4B) < δiso ( B – O – 3B)). The

17

O chemical shifts of borates

are generally greater than those of germanates, for both BO and NBO species. With these trends, we can now identify the sites present in the glass. Our identification of the species present in the LaBGeO5 glass is shown in Fig. 14, with numerical parameters presented in Table 1. The 4Ge – O – B site from the LaBGeO5 crystal, labelled “b”, lies very close to the top-right edge of the glass spectrum. If 4Ge – O – 4Ge environments were present, we would expect the resulting signal to be present about the line labelled “a”, as Ge – O – B environments are expected to be downfield of Ge – O – Ge environments. As there is minimal intensity in the region labelled “a”, we conclude that 4Ge – O – 4Ge links are not a substantial contributor to the glass network. Lacking resolution of the bridging oxygen sites in our 3QMAS spectrum, we cannot distinguish between environments that would be attributed to the differently coordinated

36


Page 37 of 51


0

17O

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


50 100

a c

150 d

200 e

250

f

300 350 400

b

g h 250

200

150 100 17O MAS dimension / ppm

50

0

Figure 14: 17O 3QMAS NMR spectrum of the glass LBG-G. Horizontal lines indicate mean expected positions, as well as approximate expected widths in the MAS dimension, of oxygen environments which are expected to be present in LaBGeO5 . Parameters for each environment are discussed in Table 1. Solid lines are environments present in the LaBGeO5 crystal. Double lines are environments which are plausibly or possibly present based upon data from the literature. Dotted lines are environments which are unlikely to be present, again based upon data from the literature. The environments shown are: a) typical 4Ge – O – 4Ge, from literature; 15,25 b) 4Ge – O – B, from Table 2; c) 4B – O – 4B, from Table 2; d) 4Ge – O – 5/6Ge, from literature; 15,25 e) O666 as in rutile GeO2 , from literature; 15,25 f) 4Ge – O – La, from Table 2; g) B – O – La environment; h) O666 as in Na4 Ge9 O20 , from literature. 15

37



borate units. However, due to the presence of 3B and

Page 38 of 51

5/6

Ge we can predict that the Ge–O–B

and B–O–B peaks in the glass would be broadened to more positive chemical shifts than those in the crystal. The 4B – O – 4B environment from the LaBGeO5 crystal, labelled c, would have the lowest positive chemical shift of the possible B – O – B environments; the same is true of the 4Ge – O – 4B environment (“b”) for the Ge – O – B environments. Based upon the response of

17

CS O δiso to changes in borate coordination in borosilicate glasses, we

would expect the chemical shifts of the 4B – O – 3B and 3B – O – 3B environments in our glass sample to be approximately 10 ppm and 20 ppm more positive, respectively, than that of the 4

B – O – 4B environment. Similarly, we would expect the shift of the 4Ge – O – 3B environment

to be approximately 10 ppm more positive than that of the 4Ge – O – 4B environment. Hence CS the intensity region between δiso of 90 ppm to 100 ppm can be attributed to 4Ge – O – B CS environments, while the δiso range of 100 ppm to 120 ppm can be attributed in part to

B – O – B links of various coordination. This interpretation is supported by our REDOR results, which establish the presence of 4B – O – 3B environments. 3B – O – 3B environments, not detectable by the

11

B{ 10B} REDOR experiment but plausibly present in the glass, may

also contribute to intensity present between 100 ppm to 120 ppm, especially closer to 120 ppm in the

17

O MAS dimension.

Ge – O – n B and n B – O – n B species can explain much of the major intensity observed

4

between 50 ppm to 120 ppm in the

17

O NMR spectra of the glass, but there remains residual

low intensity between 120 ppm to 150 ppm which is best explained by CS ments. The δiso values of

5/6

Ge – O – X environ-

5/6

Ge – O – 4Ge in various sodium germanate glasses range from

approximately 100 ppm to 150 ppm (environment “d” in Fig. 14). 15,26 This somewhat overlaps with the expected positions of the B – O – B interactions. As the

17

CS O δiso of borates is

CS 5/6 generally more positive than that of germanates, it is also possible that δiso ( Ge – O – 4Ge) CS 5/6 < δiso ( Ge – O – B) and that both of these species contribute intensity present between

120 ppm to 150 ppm. However, only 20 % to 40 % of the germanium in LaBGeO5 glass is expected to be present as high-coordinate germanium. 19 As we have previously discussed,

38


Page 39 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


there are many energetically plausible Hence we would expect the

5/6

Ge – O – X configurations in the LaBGeO5 system. 19

17

O intensity attributable to

5/6

Ge – O – X environments to be

fairly broad, and not of particularly high intensity. This is consistent with the behaviour of the 120 ppm to 150 ppm range in our 3QMAS spectrum of the glass. In summary, the bridging oxygen environments in the glass are consistent with B–O–B and Ge–O–B environments of 3 and 4 fold coordinated boron and 4–6 fold coordinated germanium species. Individual environments cannot be resolved due to a combination of the wide range of configurations afforded by the variety of boron and germanium coordinations. There is a substantial presence of non-bridging oxygen (NBO) and the most drastic difference between the

17

O NMR spectra of the glass and crystal is the intensity to higher

ppm values in the spectra of the glass. The intensity maximum of the NBO peak in the glass is located at 199 ppm in the 16.4 T MAS spectrum (Fig. 9). 4Ge – O – La NBO environments similar to those seen in the spectrum of the LaBGeO5 crystal account for the low-shift side of the broad peak in the

17

O spectrum of the glass (environment “f” in Fig. 14). Assuming

that the Ge–O–La NBO peak in the glass can be described by a peak near the same position as the similar peak in the crystal, an additional peak centred at approximately 215 ppm can explain the increased intensity at higher chemical shifts. This position is consistent with the 3B – O – La NBO environment in LaBO3 , as observed in the LBG+60

17

O MAS spectrum

(Fig. 9). Additionally, this chemical shift is consistent with the B–O–La environments predicted by DFT calculations on LaBO3 and LaB3 O6 , and is consistent with the general trend of borate

17

O shifts being more positive than germanate shifts. The presence of the

B–O–La environment is supported by the

11

B NMR results discussed above.

Finally, the possible presence of oxygen triclusters needs to be considered. In our spectrum of the LBG glass, there is no indication of rutile-like O666 . Another oxygen tricluster geometry has been observed in crystalline Na4 Ge9 O20 whose

17

O resonances are located at significantly

CS more positive chemical shift (δiso = 246.5(5) ppm) and with a much lower PQ (3.75(5) MHz). 15

Such resonances would overlap with the expected B–O–La environment near to the leftmost

39



Page 40 of 51

edge of the glass signal in the quantitative MAS spectrum. Lacking resolution, and without any obvious shoulder in the appropriate chemical shift region, it is not possible to conclusively determine the presence or absence of Na4 Ge9 O20 -like O666 triclusters; however, if they are present, they likely exist only in low amounts. The NMR results support a LaBGeO5 glass structure that is both radically different than that of the crystal and highly interconnected. The

11

B and

17

O NMR spectra of

the glass support the presence of many species not featured in the crystal, e.g., BØ2 O− , ( 5/6Ge – O – 4Ge), and ( 5/6Ge – O – B), some of which are expected from our previously reported neutron diffraction data. 19 The change in breadth of the

139

La peak in the glass support

structural changes in the La environment, which are also supported by neutron diffraction data. 19 The BØ− 4 glass peak is of more positive

11

B chemical shift than that of the crystal,

implying changes in the nature of both its oxygen neighbours and the second coordination sphere. Our

11

B{ 10B} REDOR experiments support this interpretation, indicating that there

− are fewer BØ− 4 –BØ4 links in the LaBGeO5 glass than in the crystal, as well as the presence of 4 4 BO3 –BØ− 4 links. The absence of Ge – O – Ge and La – O – La environments, and the presence

of both 4Ge – O – B and plausibly

5/6

Ge – O – B environments, support a highly interconnected

glass network. There are no indications of ordered, “crystal-like” environments present in the glass. Direct NMR probes of internuclear distances are difficult to carry out for this composition, due to the spectroscopic properties of

139

La and

73

Ge. The presence of high-coordinate

germanium provides a plausible explanation for the difference in glass-forming ability between the LaBGeO5 and the LaBSiO5 compositions. The LaBSiO5 crystal is isostructural to the LaBGeO5 crystal, 73 but its precise composition has recently been shown to be outside the lanthanum borosilicate glass forming region when alumina contamination is avoided. 74 If six-coordinate germanium were not present, there would be no obvious structural difference between the LaBGeO5 and LaBSiO5 compositions to explain this difference in behaviour.

40


Page 41 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusions We present

11

B MAS,

11

B{ 10B} REDOR,

139

La WCPMG,

17

O MAS and

17

O 3QMAS NMR

data to provide insight regarding the structure of the glass and to probe possible structural changes during crystallization. All NMR spectra indicate substantial structural differences 2– between the glass and the crystal. The crystal is composed of BØ− 4 tetrahedra, GeØ2 O2

tetrahedra, and LaO9 polyhedra, with three oxygen environments: 4B – O – 4B, 4Ge – O – 4B, and 4Ge – O – La. Certain structural units positively identified in the glass (e.g., BØ2 O− , BØ2 O− –BØ− 4 , GeO6 ) are absent from the crystal. Other environments (e.g.,

5/6

Ge – O – B)

are plausibly present as well, while Ge and La clustering (La – O – La and 4Ge – O – 4Ge) are notably absent. The diverse range of Ge–O–B environments, as well as the absence of La – O – La and 4Ge – O – 4Ge environments, suggests a highly interconnected and homogeneous glass network. The crystallization of the LaBGeO5 system is consistent with heterogeneous nucleation, which is itself consistent with a glass structure significantly different than the isochemical crystal structure. No structural changes in either the glass or crystal were observed for a range of partially-crystallized samples. A revised equation for calculating the crystallite fraction, c, from moderate-resolution by high-resolution

11

B and

11

B MAS NMR data is presented, supported

17

O MAS NMR spectra.

Acknowledgement Financial support from NSERC (Canada Grant Number RGPIN 261987) is gratefully acknowledged. We thank Prof. Mark Obrovac and Prof. Alex Speed of Dalhousie University for access to an X-ray diffractometer and for the synthesis of H3 B 17O3 , respectively. Compute Canada is thanked for access to computational resources.

41



Page 42 of 51

Supporting Information Available Results of Rietveld refinements of pXRD data. Details of the conversion from calculated chemical shielding to experimental chemical shift.

10

B NMR spectra and parameters.

11

B

NMR spectra of LaBO3 and LaB3 O6 .

References (1) Tanaka, K.; Kuroda, H.; Narazaki, A.; Hira, K.; Soga, N. Second Harmonic Generation in BaTiO3 Film Crystallized on Tellurite Glass Surface. J. Mater. Sci. Lett. 1998, 17, 1063–1065. (2) Bescher, E.; Xu, Y.; Mackenzie, J. D. New Low Temperature Multiphase Ferroelectric Films. J. Appl. Phys. 2001, 89, 6341–6348. (3) Kaminskii, A. A.; Mill, B. V.; Belokonov, E. L.; Butashin, A. V. Growth, Structure and Spectroscopy of Lanthanum Borogermanate Crystals LaBGeO5 . Neorg. Mater. 1990, 26, 1105–1107. (4) Belokoneva, E. L.; David, W. I. F.; Forsyth, J. B.; Knight, K. S. Structural Aspects of the 530 ◦C Phase Transition in LaBGeO5 . J. Phys.: Condens. Matter 1997, 9, 3503. (5) Takahashi, Y.; Benino, Y.; Fujiwara, T.; Komatsu, T. Second Harmonic Generation in Transparent Surface Crystallized Glasses With Stillwellite-Type LaBGeO5 . J. Appl. Phys. 2001, 89, 5282–5287. (6) Takahashi, Y.; Kitamura, K.; Benino, Y.; Fujiwara, T.; Komatsu, T. LaBGeO5 Single Crystals in Glass and Second-Harmonic Generation. Mater. Sci. Eng. B 2005, 120, 155–160. (7) Stefanovich, S.; Mill, B. V.; Butashin, A. V. Ferroelectricity and Phase Transitions in Stillwellite LaBGeO5 . Sov. Phys. Crystallogr. 1992, 37, 513–515. 42


Page 43 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(8) Uesu, Y.; Horiuchi, N.; Osakabe, E.; Omori, S.; Strukov, B. A. On the Phase Transition of New Ferroelectric LaBGeO5 . J. Phys. Soc. Jpn. 1993, 62, 2522–2523. (9) Sigaev, V. N.; Stefanovich, S. Y.; Sarkisov, P. D.; Lopatina, E. V. Lanthanum Borogermanate Glasses and Crystallization of Stillwellite LaBGeO5 : I. Specific Features of Synthesis and Physicochemical Properties of Glass. Glass Phys. Chem. 1994, 20, 392– 397. (10) Stone, A.; Jain, H.; Dierolf, V.; Sakakura, M.; Shimotsuma, Y.; Miura, K.; Hirao, K.; Lapointe, J.; Kashyap, R. Direct Laser-Writing of Ferroelectric Single-Crystal Waveguide Architectures in Glass for 3D Integrated Optics. Sci. Rep. 2015, 5, 10391. (11) Knorr, B.; Veenhuizen, K.; Stone, A.; Jain, H.; Dierolf, V. Optical Properties and Structure of Er:LaBGeO5 Laser-Induced Crystals-in-Glass. Opt. Mater. Express 2017, 7, 4095–4110. (12) Kratochvilova, I.; Kamba, S.; Gregora, I.; Petzelt, J.; Sigaev, V. N.; Smelyanskaya, E. N.; Molev, V. I. Vibration Properties of Pb5 Ge3 O11 and LaBGeO5 Glasses and Crystallised Glasses. Ferroelectrics 2000, 239, 39–46. (13) Gupta, P.; Jain, H.; Williams, D. B.; Kanert, O.; Kuechler, R. Structural Evolution of LaBGeO5 Transparent Ferroelectric Nano-Composites. J. Non-Cryst. Solids 2004, 349, 291–298. (14) Sigaev, V. N.; Lotarev, S. V.; Orlova, E. V.; Golubev, N. V.; Koltashev, V. V.; Plotnichenko, V. G.; Komandin, G. A. Structure of Lanthanum-Borogermanate Glass With Stillwellite Composition According to Vibrational Spectroscopy Data. Glass Ceram. 2010, 67, 105–108. (15) Du, L.-S.; Stebbins, J. F. Oxygen Sites and Network Coordination in Sodium Germanate Glasses and Crystals: High-Resolution Oxygen-17 and Sodium-23 NMR. J. Phys. Chem. B 2006, 110, 12427–12437. 43



Page 44 of 51

(16) Hoppe, U.; Kranold, R.; Weber, H.-J.; Hannon, A. C. The Change of the Ge–O Coordination Number in Potassium Germanate Glasses Probed by Neutron Diffraction With High Real-Space Resolution. J. Non-Cryst. Solids 1999, 248, 1–10. (17) Hannon, A. C.; Di Martino, D.; Santos, L. F.; Almeida, R. M. Ge-O Coordination in Cesium Germanate Glasses. J. Phys. Chem. B 2007, 111, 3342–3354. (18) Alderman, O. L. G.; Hannon, A. C.; Feller, S.; Beanland, R.; Holland, D. The Germanate Anomaly in Alkaline Earth Germanate Glasses. J. Phys. Chem. C 2017, (19) Paterson, A. L.; Hannon, A. C.; Werner-Zwanziger, U.; Zwanziger, J. W. Structural Differences between the Glass and Crystal Forms of the Transparent Ferroelectric Nanocomposite, LaBGeO5 , from Neutron Diffraction and NMR Spectroscopy. J. Phys. Chem. C 2018, 122, 20963–20980. (20) van W¨ ullen, L.; Schwering, G.

11

B-MQMAS and

29

Si – { 11B} Double-Resonance NMR

Studies on the Structure of Binary B2 O3 – SiO2 Glasses. Solid State Nucl. Magn. Reson. 2002, 21, 134–144. (21) Prasad, S.; Clark, T. M.; Sefzik, T. H.; Kwak, H.-T.; Gan, Z.; Grandinetti, P. J. SolidState Multinuclear Magnetic Resonance Investigation of Pyrex®. J. Non-Cryst. Solids 2006, 352, 2834–2840. (22) Aguiar, P. M.; Kroeker, S. Boron Speciation and Non-Bridging Oxygens in High-Alkali Borate Glasses. J. Non-Cryst. Solids 2007, 353, 1834–1839. (23) Chen, B.; Werner-Zwanziger, U.; Zwanziger, J. W.; Nascimento, M. L. F.; Ghussn, L.; Zanotto, E. D. Correlation of Network Structure With Devitrification Mechanism in Lithium and Sodium Diborate Glasses. J. Non-Cryst. Solids 2010, 356, 2641–2644. (24) Paterson, A. L.; Hanson, M. A.; Werner-Zwanziger, U.; Zwanziger, J. W. Relating

44


139

La

Page 45 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Quadrupolar Coupling Constants to Polyhedral Distortion in Crystalline Structures. J. Phys. Chem. C 2015, 119, 25508–25517. (25) Hussin, R.; Holland, D.; Dupree, R. Does Six-Coordinate Germanium Exist in Na2 O – GeO2 Glasses? Oxygen-17 Nuclear Magnetic Resonance Measurements. J. NonCryst. Solids 1998, 232, 440–445. (26) Lee, S. K.; Lee, B. H. Atomistic Origin of Germanate Anomaly in GeO2 and NaGermanate Glasses: Insights From Two-Dimensional

17

O NMR and Quantum Chemical

Calculations. J. Phys. Chem. B 2006, 110, 16408–16412. (27) Michaelis, V. K.; Kroeker, S.

73

Ge Solid-State NMR of Germanium Oxide Materials:

Experimental and Theoretical Studies. J. Phys. Chem. C 2010, 114, 21736–21744. (28) Harris, R. K.; Becker, E. D.; De Menezes, S. M. C.; Granger, P.; Hoffman, R. E.; Zilm, K. W. Further Conventions for NMR Shielding and Chemical Shifts IUPAC Recommendations 2008. Solid State Nucl. Magn. Reson. 2008, 33, 41–56. (29) Brown, S. P.; Wimperis, S. Two-Dimensional Multiple-Quantum MAS NMR of Quadrupolar Nuclei: A Comparison of Methods. J. Magn. Reson. 1997, 128, 42–61. (30) Man, P. P. Scaling and Labeling the High-Resolution Isotropic Axis of Two-Dimensional Multiple-Quantum Magic-Angle-Spinning Spectra of Half-Integer Quadrupole Spins. Phys. Rev. B 1998, 58, 2764. (31) Millot, Y.; Man, P. P. Procedures for Labeling the High-Resolution Axis of TwoDimensional MQ-MAS NMR Spectra of Half-Integer Quadrupole Spins. Solid State Nucl. Magn. Reson. 2002, 21, 21–43. (32) O’Dell, L. A.; Schurko, R. W. QCPMG Using Adiabatic Pulses for Faster Acquisition of Ultra-Wideline NMR Spectra. Chem. Phys. Lett. 2008, 464, 97–102.

45



Page 46 of 51

¯ Freeman, R. Stretched Adiabatic Pulses for Broadband Spin Inversion. J. (33) Kupce, E.; Magn. Reson., Ser. A 1995, 117, 246–256. (34) Carr, H. Y.; Purcell, E. M. Effects of Diffusion on Free Precession in Nuclear Magnetic Resonance Experiments. Phys. Rev. 1954, 94, 630. (35) Meiboom, S.; Gill, D. Modified Spin-Echo Method for Measuring Nuclear Relaxation Times. Rev. Sci. Instrum. 1958, 29, 688–691. (36) Medek, A.; Frydman, V.; Frydman, L. Central Transition Nuclear Magnetic Resonance in the Presence of Large Quadrupole Couplings: Cobalt-59 Nuclear Magnetic Resonance of Cobaltophthalocyanines. J. Phys. Chem. A 1999, 103, 4830–4835. (37) Massiot, D.; Farnan, I.; Gautier, N.; Trumeau, D.; Trokiner, A.; Coutures, J. P. and

71

Ga

69

Ga Nuclear Magnetic Resonance Study of β-Ga2 O3 : Resolution of Four- and

Six-Fold Coordinated Ga Sites in Static Conditions. Solid State Nucl. Magn. Reson. 1995, 4, 241–248. (38) Gullion, T.; Schaefer, J. Rotational-Echo Double-Resonance NMR. J. Magn. Reson. 1989, 81, 196–200. (39) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calvé, S.; Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Modelling One-and Two-Dimensional Solid-State NMR Spectra. Magn. Reson. Chem. 2002, 40, 70–76. (40) de Lacaillerie, J.-B. d.; Fretigny, C.; Massiot, D. MAS NMR Spectra of Quadrupolar Nuclei in Disordered Solids: The Czjzek Model. J. Magn. Reson. 2008, 192, 244–251. (41) Czjzek, G.; Fink, J.; Götz, F.; Schmidt, H.; Coey, J.; Rebouillat, J.-P.; Lienard, A. Atomic Coordination and the Distribution of Electric Field Gradients in Amorphous Solids. Phys. Rev. B 1981, 23, 2513.

46


Page 47 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(42) Le Caër, G.; Brand, R. General Models for the Distributions of Electric Field Gradients in Disordered Solids. J. Phys.: Condens. Matter 1998, 10, 10715. (43) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I. et al. QUANTUM ESPRESSO: A Modular and Open-Source Software Project for Quantum Simulations of Materials. J. Phys.: Condens. Matter 2009, 21, 395502. (44) Giannozzi, P.; Andreussi, O.; Brumme, T.; Bunau, O.; Nardelli, M. B.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Cococcioni, M. et al. Advanced Capabilities for Materials Modelling With Quantum ESPRESSO. J. Phys.: Condens. Matter 2017, 29, 465901. (45) Pickard, C. J.; Mauri, F. All-Electron Magnetic Response With Pseudopotentials: NMR Chemical Shifts. Phys. Rev. B 2001, 63, 245101. (46) Profeta, M.; Benoit, M.; Mauri, F.; Pickard, C. J. First-Principles Calculation of the 17

O NMR Parameters in Ca Oxide and Ca Aluminosilicates: the Partially Covalent

Nature of the Ca- O Bond, a Challenge for Density Functional Theory. J. Am. Chem. Soc. 2004, 126, 12628–12635. (47) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865. (48) Jollet, F.; Torrent, M.; Holzwarth, N. Generation of Projector Augmented-Wave Atomic Data: A 71 Element Validated Table in the XML Format. Comput. Phys. Commun. 2014, 185, 1246–1254. (49) Zwanziger, J. W. Computation of NMR Observables: Consequences of ProjectorAugmented Wave Sphere Overlap. Solid State Nucl. Magn. Reson. 2016, 80, 14–18.

47



Page 48 of 51

(50) Holzwarth, N. A. W.; Tackett, A. R.; Matthews, G. E. A Projector Augmented Wave (PAW) Code for Electronic Structure Calculations, Part I: Atompaw for Generating Atom-Centered Functions. Comput. Phys. Commun. 2001, 135, 329–347. (51) Pyykkö, P. Year-2008 Nuclear Quadrupole Moments. Mol. Phys. 2008, 106, 1965–1974. (52) Rulmont, A.; Tarte, P. Lanthanide Borogermanates LnBGeO5 : Synthesis and Structural Study by X-Ray Diffractometry and Vibrational Spectroscopy. J. Solid State Chem. 1988, 75, 244–250. (53) Coussa, C.; Martinet, C.; Champagnon, B.; Grosvalet, L.; Vouagner, D.; Sigaev, V. In Situ Raman Spectroscopy of Pressure-Induced Changes in LaBGeO5 Glass: Hysteresis and Plastic Deformation. J. Phys.: Condens. Matter 2007, 19, 266220. (54) Hrubá, I.; Kamba, S.; Petzelt, J.; Gregora, I.; Zikmund, Z.; Ivannikov, D.; Komandin, G.; Volkov, A.; Strukov, B. Optical Phonons and Ferroelectric Phase Transition in the LaBGeO5 Crystal. Phys. Status Solidi B 1999, 214, 423–439. (55) Paterson, A. L.; Zwanziger, J. W. Anisotropic Stress in Laser-Written LaBGeO5 GlassCeramic Composites. J. Appl. Phys. 2018, 124, 083106. (56) Du, L.-S.; Stebbins, J. F. Solid-State NMR Study of Metastable Immiscibility in Alkali Borosilicate Glasses. J. Non-Cryst. Solids 2003, 315, 239–255. (57) Kroeker, S.; Stebbins, J. F. Three-Coordinated Boron-11 Chemical Shifts in Borates. Inorg. Chem. 2001, 40, 6239–6246. (58) Wang, S.; Stebbins, J. F. Multiple-Quantum Magic-Angle Spinning

17

O NMR Studies

of Borate, Borosilicate, and Boroaluminate Glasses. J. Am. Ceram. Soc. 1999, 82, 1519–1528. (59) Martens, R.; M¨ uller-Warmuth, W. Structural Groups and Their Mixing in Borosilicate

48


Page 49 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Glasses of Various Compositions–an NMR Study. J. Non-Cryst. Solids 2000, 265, 167–175. (60) MacKenzie, K. J. D.; Smith, M. E. Multinuclear Solid-State NMR of Inorganic Materials; Elsevier, 2002. (61) Bertmer, M.; Eckert, H. Dephasing of Spin Echoes by Multiple Heteronuclear Dipolar Interactions in Rotational Echo Double Resonance NMR Experiments. Solid State Nucl. Magn. Reson. 1999, 15, 139–152. (62) Bertmer, M.; Z¨ uchner, L.; Chan, J. C. C.; Eckert, H. Short and Medium Range Order in Sodium Aluminoborate Glasses. 2. Site Connectivities and Cation Distributions Studied by Rotational Echo Double Resonance NMR Spectroscopy. J. Phys. Chem. B 2000, 104, 6541–6553. (63) Chan, J. C. C.; Bertmer, M.; Eckert, H. Site Connectivities in Amorphous Materials Studied by Double-Resonance NMR of Quadrupolar Nuclei: High-Resolution11B ↔27Al Spectroscopy of Aluminoborate Glasses. J. Am. Chem. Soc. 1999, 121, 5238–5248. (64) van Vleck, J. H. The Dipolar Broadening of Magnetic Resonance Lines in Crystals. Phys. Rev. 1948, 74, 1168–1183. (65) Eckert, H.; Elbers, S.; Epping, J. D.; Janssen, M.; Kalwei, M.; Strojek, W.; Voigt, U. Top. Curr. Chem.; Springer Berlin Heidelberg, 2004; pp 195–233. (66) Angeli, F.; Charpentier, T.; Molières, E.; Soleilhavoup, A.; Jollivet, P.; Gin, S. Influence of Lanthanum on Borosilicate Glass Structure: A Multinuclear MAS and MQMAS NMR Investigation. J. Non-Cryst. Solids 2013, 376, 189–198. (67) Schaller, T.; Stebbins, J. F.; Wilding, M. C. Cation Clustering and Formation of Free Oxide Ions in Sodium and Potassium Lanthanum Silicate Glasses: Nuclear Magnetic Resonance and Raman Spectroscopic Findings. J. Non-Cryst. Solids 1999, 243, 146–157. 49



(68) Winterstein-Beckmann, A.; Möncke, D.; Palles, D.; Kamitsos, E. I.; Wondraczek, L. Structure–property correlations in highly modified Sr, Mn-borate glasses. J. Non-Cryst. Solids 2013, 376, 165–174. (69) Henderson, G. S.; Soltay, L. G.; Wang, H. M. Q Speciation in Alkali Germanate Glasses. J. Non-Cryst. Solids 2010, 356, 2480–2485. (70) Du, L.-S.; Peng, L.; Stebbins, J. F. Germanosilicate and Alkali Germanosilicate Glass Structure: New Insights From High-Resolution Oxygen-17 NMR. J. Non-Cryst. Solids 2007, 353, 2910–2918. (71) Ashbrook, S. E.; Smith, M. E. Solid State 17 O NMR—An Introduction to the Background Principles and Applications to Inorganic Materials. Chem. Soc. Rev. 2006, 35, 718–735. (72) Lee, S. K.; Kim, H. N.; Lee, B. H.; Kim, H.-I.; Kim, E. J. Nature of Chemical and Topological Disorder in Borogermanate Glasses: Insights from B-11 and O-17 Solid-State NMR and Quantum Chemical Calculations. J. Phys. Chem. B 2010, 114, 412–420. (73) Chi, L.; Chen, H.; Zhuang, H.; Huang, J. Crystal Structure of LaBSiO5 . J. Alloys Compd. 1997, 252, L12–L15. (74) Trégouët, H.; Caurant, D.; Majérus, O.; Charpentier, T.; Lerouge, T.; Cormier, L. Exploration of Glass Domain in the SiO2 – B2 O3 – La2 O3 System. J. Non-Cryst. Solids 2017, 476, 158–172.

50


Page 50 of 51

Page 51 of 51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Graphical TOC Entry

Solid-state 17O NMR is used to determine structure and connectivity in the transparent ferroelectric nanocomposite LaBGeO5 .

51


Network Connectivity and Crystallization in the Transparent

Recommend Documents