PCCP View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PAPER
Cite this: DOI: 10.1039/c5cp02095f
View Journal
Intermediate length scale organisation in tin borophosphate glasses: new insights from high field correlation NMR G. Tricot,*a A. Saitohb and H. Takebeb The structure of tin borophosphate glasses, considered for the development of low temperature sealing glasses or anode materials for Li-batteries, has been analysed at the intermediate length scale by a combination of high field standard and advanced 1D/2D nuclear magnetic resonance techniques. The nature and extent of B/P mixing were analysed using the 11B(31P) dipolar heteronuclear multiple quantum coherence NMR sequence and the data interpretation allowed (i) detecting the presence and analysing the nature of the B–O–P linkages, (ii) re-interpreting the 1D
31
P spectra and (iii) extracting the
proportion of P connected to borate species. Interaction between the different borate species was analysed using the
11
B double quantum–simple quantum experiment to (i) investigate the presence and
nature of the B–O–B linkage, (ii) assign the different borate species observed all along the composition Received 10th April 2015, Accepted 7th July 2015
line and (iii) monitor the borate network formation. In addition,
119
Sn static NMR was used to investigate
the evolution of the chemical environment of the tin polyhedra. Altogether, the set of data allowed
DOI: 10.1039/c5cp02095f
determining the structural units constituting the glass network and quantifying the extent of B/P mixing. The structural data were then used to explain the non-linear and unusual evolution of the glass
www.rsc.org/pccp
transition temperature.
1. Introduction Tin(+II) borophosphate (SnBP) glasses have attracted much attention during the last few years1–6 due to their (i) low glass transition temperature (Tg), (ii) wide glass forming composition range, (iii) good chemical and thermal durability and (iv) nontoxicity compared to standard low-Tg PbO-based glasses. The SnBP system has thus been considered for the development of low temperature sealing glasses or anode materials for Li-batteries.1,3 Macroscopic properties like Tg, thermal expansion coefficient, refractive index, and density have shown to be significantly affected by the formulation. The 67SnO–xB2O3–(33 x)P2O5 composition line has been widely investigated4,5 and presents a non-linear evolution of the macroscopic properties when P is replaced by B. Recently, owing to the preparation of the complete composition line (from x = 0 to 33), the non-linear evolution has been fully analysed and decomposed into three regions.5 While the first two regions show usual behaviours (strong and moderate evolution of the properties), the third region presents a second increase of Tg and density, whereas other borophosphate-based a
LASIR UMR-CNRS 8516, Universite´ de Lille 1, Villeneuve d’Ascq F-59655, France. E-mail: [email protected] b Graduate School of Science and Engineering, Ehime University, Matsuyama, 3 bunkyo-cho, 790-8577, Japan
This journal is © the Owner Societies 2015
systems generally experience constant or decreasing values at high B2O3 contents.7–12 In order to understand this behaviour, the structure has been previously analysed by Raman, infra-red and 1D 31P and 11B MAS-NMR.5 While the local order has been correctly described, no clear information has been obtained concerning the intermediate length scale organisation, i.e. the extent and nature of B/P mixing, although this information is usually considered as the key parameter governing the evolution of the properties. In this paper, new insights into the intermediate length scale organisation of 67SnO–xB2O3–(33 x)P2O5 glasses (0 r x r 33) have been obtained using advanced correlation NMR performed at very high magnetic field (18.8 T). The P/B connectivity scheme has been analysed using the 2D 11B/31P correlation map edited with the 11B(31P) dipolar heteronuclear multiple quantum coherence (D-HMQC) NMR technique,13,14 which allows highlighting and characterising the formation of the mixed P–O–B linkages. Progressive formation of the borate network by the P/B substitution has been monitored using the 11B double quantum–single quantum (DQ–SQ) sequence,15 which probes the formation and nature of the B–O–B bonds. In order to follow the evolution of the tin polyhedra coordination, 119 Sn NMR experiments were recorded using the static spin echo technique. Altogether, the structural results lead to the first detailed description of the medium range organisation and its evolution all along the P/B substitution. The structural model
Phys. Chem. Chem. Phys.
View Article Online
Paper
was finally used to discuss the non-linear Tg evolution, including the unusual Tg increase at high B2O3 amounts.
2. Experimental
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
2.1
Material synthesis and characterisations
x)P2O5 (0 r x r 33) glasses have been 67SnO–xB2O3–(33 prepared using the standard melt-quenching method. Appropriate mixtures of reagent grade SnO, P2O5 and B2O3 have been mixed, placed in a vitreous carbon crucible and heated up for 60 min at 950–1100 1C in a silica tube furnace under flowing argon in order to avoid oxidation of Sn2+ to Sn4+. The melts have then been cooled down to 400 1C in the silica tube before being cast in a carbon mould. The amorphous character of each sample has been confirmed by X-ray diffraction analysis. Tg have been determined by differential thermal analysis (DTA) using a TG-DTA8120 Rigaku machine. 25–40 mg of each sample, placed in a platinum pan, have been analysed from room temperature to 550 1C at a heating rate of 10 1C min 1. The errors in Tg were estimated to be 5 1C.
PCCP
to liquid H3PO4 (85%), solid NaBH4 and solid SnO2 at 0, and 604.3 ppm, respectively.
3. Results 3.1
1D/2D NMR analyses
The 31P and 11B MAS-NMR experiments have been performed on a 18.8 T Bruker Avance III spectrometer at 324 and 256.8 MHz, respectively. The experiments have been performed using a prototype 3.2 mm 11B/31P measurement probe equipped with a B free stator operating at a spinning frequency (nrot) of 20 kHz. The 1D 31P MAS-NMR acquisitions have been recorded with a 2 ms pulse length (p/6 pulse angle), a radiofrequency field (rf) strength of 40 kHz, 16–64 transients and a recycle delay (rd) of 120 s. The 1D 11B MAS-NMR experiments have been acquired with a 1 ms pulse length (p/12 pulse angle), a rf strength of 40 kHz, 1024–2048 transients and a rd of 2 s. The 11B/31P correlation maps have been edited using the D-HMQC sequence13–14 using a selective 11B p/2-t-p-t spin echo sequence pulse (p pulse length of 18 ms) modulated by two 31P p/2 pulses of 4.5 ms; the 11 31 B/ P dipolar interaction has been recoupled using the SR4 pulse scheme16 by irradiating the 31P channel with a 40 kHz rf field (2 nrot) during 1 ms. As recently demonstrated in the case of an alkali borophosphate, such a short recoupling time allows discussing the B/P spatial proximity highlighted by our experiment as evidenced for P–O–B chemical connectivity.17 The 2048 40 acquisition points have been recorded using a rotor synchronised t1-increment (50 ms), 256 transients and a rd of 2 s. The 11B/11B 2D maps have been recorded using the 11B DQ–SQ pulse sequence.15 The 2048(t2) 100(t1) acquisition points have been recorded under rotor-synchronised conditions with a p pulse length of 25 ms, 256 transients and a rd of 2 s. The 2D coherences have been created using a 10 kHz (0.5 nrot) and a 400 ms length BR212 pulse scheme.18 The 119Sn static NMR experiments were recorded at 149.1 MHz on a 9.4 T Bruker Avance II spectrometer using a 5 mm static probe. A p/2-t-p/2-t spin-echo sequence was used to record the signal with a 2 ms pulse length, a 50 ms echo delay, 3–5k transients and a rd of 25 s. The 31P, 11B and 119Sn chemical shift values have been referred
Phys. Chem. Chem. Phys.
Glass properties
The glass transition temperatures measured on the complete glass system are reported in Fig. 1 (empty squares). Tg continuously increases from 268 1C for the pure phosphate (x = 0) to 350 1C for the pure borate (x = 33) sample. However, Tg does not increase linearly and different regions can be distinguished by comparing the actual Tg and the linear Tg obtained by assuming a linear evolution from the pure phosphate to the pure borate samples (Fig. 1 dotted line). The Tg Tglin parameter evolution (Fig. 1, full circles) clearly shows the presence of three regions with (i) important (0 r x r 5), moderate (6 r x r 15) and low (18 r x r 33) deviations from the linear Tg values, indicating that the mixed network has different impacts on the macroscopic properties depending on the glass composition and the glass network organisation. 3.2
2.2
42.06
1D MAS-NMR
The 1D 31P MAS-NMR experiments (Fig. 2a) provide limited information about the phosphate speciation, due to very broad signals. While the global chemical shift appears toward positive values, no clear numbering or assignment can be derived from the 1D analysis. The 11B MAS-NMR spectra (Fig. 2b) provide interesting information about the local order of the borate species. Clear distinction between B4 and B3 species can be made through the chemical shift values (B4: 5/+3 ppm and B3: +20/+8 ppm). The spectra show that boron ions enter first as B4 species before being replaced by trigonal groups at higher B2O3 contents. The high resolution permitted by the high magnetic field also allows distinguishing between different B4 groups (denoted as B4(1), B4(2), B4(3) and B4(4)) through different signals at 3, 0.5, 1.5 and 3 ppm. 119The Sn static NMR spectra (Fig. 3) show very broad signals due to the large chemical shift anisotropy (CSA). Similar features can be observed for the low (x = 0, 5)
Fig. 1 Glass transition temperature (open squares), linear evolution (dotted line) and deviation from linear evolution (full circles) versus the B2O3 glass content.
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Fig. 2
Paper
31
P (a) and
11
B (b) MAS-NMR spectra performed at 18.8 T on the 67SnO–xB2O3–(33
Fig. 3 119Sn static NMR spectra performed at 9.4 T on the 67SnO–xB2O3– (33 x)P2O5 composition line.
and intermediate (x = 12) samples, whereas broader signals are observed in the case of the high B2O3 glass contents (x = 25 and 33). 3.3
Correlation NMR
The investigation of the P–O–B linkages has been performed using the 11B(31P) D-HMQC sequence. The 2D maps obtained for x = 3, 7, 18 and 25 samples are reported in Fig. 4. The 2D maps are accompanied by the 1D MAS-NMR 11B and 31P spectra (denoted as i and iii) and by the horizontal and vertical 2D spectrum projections (denoted as ii and iv). The 2D spectrum
This journal is © the Owner Societies 2015
x)P2O5 composition line.
obtained for the very low B2O3 amount sample (x = 3) presents a correlation signal indicating the presence of the P–O–B4(1) linkage. The 1D 31P MAS-NMR and 2D projection (Fig. 4a(iii and iv)) show significant differences indicating that some P within the glass matrix are not connected to any B, in a good agreement with the low B2O3 amount of the sample. The 2D map obtained for the x = 7 sample displays a complex pattern composed by three correlation signals indicating the presence of P–O–B4(1), P–O–B4(2) and P–O–B3 linkages. For this sample again, some P5+ are not connected to B3+ ions, as indicated by the comparison between Fig. 4b(iii) (1D 31P MAS-NMR) and Fig. 4b(iv) (2D map 31P projection). Slight differences can be observed between the three correlation signals in the 2D maps, indicating that the phosphate species attached to each borate ion are different. However, these differences are weak (less than 2 ppm for the chemical shift) and, to avoid any over-interpretation, the three signals have been gathered into a single gaussian line (Fig. 4b(iv)) that will be considered as the signature of the P attached to the borate groups in the following. At higher B2O3 contents (x = 18 and 27), the 2D maps exhibit similar patterns. The maps present intense correlation signals involving P and B3 signals accompanied by correlation signals indicating the presence of P–O–B4(2) and P–O–B4(3) linkages. The presence and nature of B–O–B bonds have been investigated using the 11B DQ–SQ NMR technique. The 2D maps obtained for the x = 3, 7, 18 and 30 samples are reported in Fig. 5. No signal is observed in the 2D map of the x = 3 sample (Fig. 5a), indicating that the boron ions are completely dissolved into the phosphate network. Four different signals can be observed in the x = 7 sample 2D map. Three on-diagonal signals indicate auto-correlation for the B4(1), B4(2) and B3 species accompanied by two off-diagonal signals involving the B4(2) and B3 units. The two 2D maps obtained for the x = 18 and
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
PCCP
Fig. 4 2D 11B(31P) D-HMQC spectra obtained for x = 3 (a), 7 (b), 18 (c) and 25 (d) samples. The 2D maps are accompanied with 1D 11B and 31P MAS-NMR spectra (i and iii, respectively) and the 11B and 31P 2D projections (ii and iv, respectively).
30 samples display similar correlation schemes as B4(3) and B3 on- and B4(3)/B3 off-diagonal correlation signals.
4. Discussion 4.1
Tg evolution
While the three-domain evolution of the properties, depicted in Fig. 1, is a common feature in the case of B/P substitution in borophosphate glasses,7–10 the final increase of Tg in high B2O3 content glasses has not been widely reported, indicating that the system experiences an unusual structural evolution. If the MGFE is always positive (the actual Tg values are always higher than the linear values), it turns out that the maximum deviation occurs in low B2O3 glasses (region (I), x o 5) and higher B2O3 glasses exhibit lesser differences between the actual and the linear Tg values. These evolutions are governed by specific structural rearrangements that need to be highlighted in order to understand how the structure rules the macroscopic properties of these materials. New insight into the intermediate length scale organisation and the B/P mixing can be derived from the presented 1D/2D NMR investigation. 4.2
The phosphate speciation
As depicted in Fig. 2, the 1D 31P MAS-NMR spectra show a progressive evolution of the global chemical shift toward positive values, indicating the presence of depolymerised units.
Phys. Chem. Chem. Phys.
However, no clear numbering and assignment can be derived due to the broad and unresolved signals. The 2D 11B/31P correlation maps (Fig. 4) indicate clear correlations resulting from the formation of mixed P–O–B linkages. The latter involve different B4 groups and B3 units also participate in the mixed linkages at higher B2O3 contents. The NMR signature of the P connected to B units (denoted as QnmB in the following with n and m being the number of connected P and B atoms as previously used in the case of aluminophosphate glasses)19 can be easily derived from the 2D map by plotting the vertical projections. All the 31P projections are reported in Fig. 6a with representative fits allowing for a precise determination of the chemical shift and the full width at half maximum (fwhm) values of the PB units. These parameters have then been used as input data to deconvolute 1D 31P MAS-NMR using the dmfit software.20 In addition to the QnmB signals, Q1 (connected to 1 other P atom) and Q0 (isolated phosphate group) signals, that constitute the base formulation (x = 0) have been added to build a proper fit. If the presence of Q1 ( 19 ppm) is expected from the pure pyrophosphate structure, a small amount of Q0 ( 10 ppm) is also observed that could originate from a minor oxidation of Sn2+ to Sn4+ or to a slight deviation from the pyrophosphate composition. All the NMR parameters (chemical shifts, full widths at half maximum, and relative proportions) deduced from the deconvolutions are gathered in Table 1 and the representative simulations of the 1D 31P MASNMR experiments are reported in Fig. 6b.
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Paper
Fig. 5 2D 11B DQ–SQ spectra obtained for x = 3 (a), 7 (b), 18 (c) and 30 (d) samples. The 2D maps are accompanied with 1D 11B (i) and 11B 2D horizontal and vertical projections (ii and iii, respectively).
Fig. 6 31P projections of the 2D D-HMQC maps, accompanied with the fits (dotted lines) used to deconvolute the 1D MAS-NMR spectra (a), deconvolution of the representative 1D 31P MAS-NMR spectra using the parameters deduced from the 2D D-HMQC experiments (b).
As reported in Table 1, the chemical shift values of the QnmB species always lie between the Q0 and Q1 groups. Addition of a P–O–B linkage inducing a shielding effect, the QnmB species can be finally assigned to Q0mB and the presence of Q1mB that should produce a more negative chemical shift10,21–23 than the Q1 unit
This journal is © the Owner Societies 2015
can be reasonably ruled out. Moreover, we propose that the species is attached to a single B3+ ion, leading to a final assignment of Q01B, since no clear change in the chemical shift of this species can be observed when B2O3 increases. This assignment allows understanding the formation mechanism
Phys. Chem. Chem. Phys.
View Article Online
Paper
PCCP
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Table 1 31P NMR parameters obtained using the gaussian model. Chemical shifts (diso), full widths at half maximum (fwhm), relative proportions (rel. prop.) are given with errors of 0.2 ppm, 0.2 ppm, 1%, 0.05 MHz and 0.05, respectively
x
Qx
0
1
Q Q0
18.6 9.1
12.2 10.5
94.0 6.0
1
Q1 Q0 Q0B
19.4 10.7 15.6
12.5 11.4 11.1
3
Q1 Q0 Q0B
20.1 10.0 15.2
5
Q1 Q0 Q0B
7
Q1 Q0 Q0B
diso/ppm
fwhm/ppm
Rel. prop./%
x
Qx
Rel. prop./%
18.5 9.7
12.0 10.6
4.0 23.8
Q0B Q1 Q0
12.5 19.0 7.9
11.0 12.0 10.4
72.2 4.0 10.1
20
Q0B Q1 Q0
11.5 19.2 7.5
10.8 12.0 10
84.9 5.0 2.0
16.3 32.8 50.9
25
Q0B Q1 Q0
11.0 19.2 7.5
10.8 12.0 10
93.0 3.0 1.5
15.9 28.1 56
30
Q0B Q1 Q0 Q0B
10.8 19.1 7.5 10.5
10.7 12.0 10.0 11.0
95.5 2.0 2.0 96.0
10
Q Q0
64.1 22.9 13.0
15
12.2 10.8 11.2
37.0 27.9 35.1
20.4 10.0 15.1
11.3 10.5 11.4
18.9 9.2 13.6
12.1 10.8 11.0
of the P–O–B linkages: borate units directly interact with P2O74 groups by breaking a P–O–P bond and forming a P–O–B linkage. The Qn relative proportions, reported in Fig. 7, also support this formation mechanism. When B2O3 is introduced, the proportion of Q1 sites, consumed in the P–O–B formation, is significantly reduced in favor of the Q01B increase. The Q01B species experiences a non-linear evolution composed by an important and a moderate increase in the 0 r x r 5 and 5 r x r 17 compositional ranges, followed by an almost constant proportion in the 17 r x r 30 B2O3 range. This evolution has been compared to the statistical dimeric distribution derived from a pure random mixing of B and P (Fig. 7 dotted lines) forming dimers. The comparison clearly indicates that P and B elements preferentially interact at low amounts (x r 5) to create the mixed linkages. It has also to be observed that the phosphate speciation does not evolve after the B2O3
Fig. 7 Evolution of the phosphate speciation in the borophosphate mixed network. The proportion of PB calculated from the dimeric statistical distribution is reported using a dotted line.
Phys. Chem. Chem. Phys.
diso/ppm
fwhm/ppm
1
content reaches 17 mol%, where Q01B species fully dominate the phosphate network. Moderate formation of Q0 sites up to x = 5 is also observed before this site proportion decreases and completely disappears above 15 mol% of B2O3. 4.3
The borate speciation
The borate speciation can been precisely analysed owing to the highly resolved 11B NMR spectra obtained at 18.8 T. In addition to the usual progressive replacement of B4 by B3 species when the B2O3 content increases, our experiments also highlight a complex speciation for the tetra-coordinated boron though the presence of four different B4 sites. Quantification and evolution of the different borate species all along the composition line have been obtained by deconvolution of the spectra. All the 11B NMR parameters for both B4 and B3 have been gathered in Table 2 and the evolution of the relative proportions versus the B2O3 content is reported in Fig. 8. B4(1) is the main species at very low B2O3 amounts and decreases beyond 5 mol% of B2O3 when B4(2) starts to increase. At x = 10, B4(2) decreases and B3 and B4(3) start to dominate the borate network. B4(3) increases slowly up to the pure borate whereas B3 experiences a significant increase. B4(4) appears at x = 15 and increases moderately up to the pure borate composition but is always present as minor species. The information can also be obtained by the B3 speciation, in spite of the broad signal coming from the remaining second order quadrupolar effect. At low B2O3 amounts (3 r x r 7), only one signal is detected in the experiments with a chemical shift evolving from 15 to 18.7 ppm. The second B3 signal is observed in the x = 12 composition. For higher B2O3 contents, both signals are present with constant chemical shifts and with an increasing relative proportion of the second B3 signal. Due to the lack of additional information about the B3 signals, B3(1) and B3(2) were treat as a single component in Fig. 8 to discuss the evolution of the global B3 species all along the B/P substitution.
This journal is © the Owner Societies 2015
View Article Online
PCCP
Paper
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Table 2 11B NMR parameters, B4 and B3 units have been simulated using the gaussian/lorentzian and quadrupolar models available in the dmfit software.18 Chemical shifts (diso), full widths at half maximum (fwhm), relative proportions (rel. prop.), the quadrupolar constant (CQ) and asymmetry parameters (ZQ) are given with errors of 0.2 ppm, 0.2 ppm, 1%, 0.1 MHz and 0.1, respectively
x
B4
B3
1
B4(1) B4(2)
diso/ppm
CQ/MHz
ZQ
3
86.3 10.1 1.7
B3(1)
15.1
2.5
0.5
1.9
1.7 1.6 1.9
60.6 28.8 4.7
B3(1)
16.7
2.5
0.5
5.9
3.3 0.9 1.1
1.7 2.1 1.9
21.8 49.4 5.2
B3(1)
18.7
2.6
0.5
23.6
B4(1) B4(2) B4(3)
2.5 0.2 1.4
2.2 2.0 2.0
3.2 29.3 15.2
B3(1) B3(2)
18.7 17.7
2.8 2.8
0.6 0.5
27.7 24.6
15
B4(1) B4(2) B4(3) B4(4)
1.9 0.0 1.4 3.4
2.1 1.9 2.0 1.9
1.3 17.7 20.5 1.8
B3(1) B3(2)
18.8 17.9
2.8 2.8
0.6 0.5
30.5 28.2
20
B4(1) B4(2) B4(3) B4(4)
1.8 0.1 1.5 3.5
2.0 1.9 2.1 1.8
0.5 10.5 24.4 2.4
B3(1) B3(2)
18.8 17.9
2.7 2.7
0.5 0.5
34.4 27.9
25
B4(1) B4(2) B4(3) B4(4)
1.7 0.2 1.5 3.2
2.0 1.9 2.1 2.2
0.6 8.0 23.3 4.9
B3(1) B3(2)
18.7 17.9
2.8 2.8
0.5 0.6
40.8 22.4
30
B4(1) B4(2) B4(3) B4(4)
1.7 0.2 1.5 3.2
2.0 1.9 2.1 2.2
0.2 6.8 23.6 6.6
B3(1) B3(2)
18.9 18.0
2.7 2.6
0.5 0.5
31.9 30.9
diso/ppm
fwhm/ppm
Rel. prop./%
3.4 1.2
1.6 1.6
97.1 2.9
B4(1) B4(2) B4(3)
3.4 1.2 0.1
1.6 1.6 1.8
5
B4(1) B4(2) B4(3)
3.3 1.1 0.5
7
B4(1) B4(2) B4(3)
12
Fig. 8 B3 and B4(x) species evolution versus the B2O3 glass content. The amounts are expressed in terms of borate units present in the glass network.
This journal is © the Owner Societies 2015
Rel. prop./%
Beyond the proportion of the borate species and their evolution, the key information lies in the chemical environment and the B/P and B/B mixing they are involved in. This information can be derived and discussed from the correlation NMR results presented in the previous section and summarised in Fig. 9. This figure presents the 11B MAS-NMR spectra (Fig. 9(i)) accompanied by relevant 11B analyses extracted from the 11B(31P) D-HMQC (Fig. 9(ii), revealing the B involved in B–O–P linkages) and the 11B DQ–SQ (Fig. 9iii and iv, revealing the B involved in B–O–B linkages) correlation maps. The first B4 site, denoted as B4(1), presents a chemical shift of 3 ppm and is generally assigned to the B(OP)4 structural unit.21–23 While our results fully support this assignment at low B2O3 amounts (Fig. 9a(i–iii)), the presence of autocorrelation signals in the 11B DQ–SQ projection of the x = 7 sample (Fig. 9a(iv)) suggests the presence of the B4(1)–O–B4(1) linkage that cannot fit with the usual B(OP)4 assignment, but is rather indicating a B(OP)3(OB4) structure, as previously shown in lithium borophosphate by 11B(31P) REDOR experiments.10 Therefore, in
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
PCCP
Fig. 9 Comparison of the 11B MAS-NMR analysis of B4(1) (a), B4(2) (b), B4(3) (c) and B4(4) (d): (i): 1D MAS-NMR, (ii): 11B projection of D-HMQC maps, (iii) and (iv): 11 B projection of the DQ-DQ maps.
order to include both species under the same nomenclature (since they probably cannot be distinguished because of similar chemical shift values), we propose that B4(1) is a tetra-coordinated boron connected to P but also B4 units leading to a general B(OX4)4 assignment with X = P, B. The second site (B4(2): 0.5 ppm) is involved in the B–O–P bond (Fig. 9b(ii)), B–O–B4 (Fig. 9b(iii)), but also in B–O–B3 as indicated by Fig. 9b(iv) and by the off-diagonal correlation signals observed in the 2D 11B DQ–SQ map (Fig. 5b). This new species, which originates from the replacement of a tetracoordinated species (P/B4) by a trigonal and planar B3 unit, is the first species involving the combination between the three glasses forming polyhedra (P, B4 and B3) and can be assigned to the B(OX)4–n(OB3)n (X = P, B4) species. The third B4 species also appears to be connected to P, B4 and B3 units, as shown by Fig. 9c(ii–iv). Therefore, we proposed that significant changes in the chemical environment compared to B4(2) can only originate from an interaction with the Sn2+ ions, leading to the B(OX)n(OB3)m(OSn)p assignment. Finally, Fig. 9d show that the fourth B4 site is attached to B3, but is not involved in B4–O–P and B4–O–B4 linkages anymore (Fig. 9d(ii and iii)). This last site can thus be tentatively assigned to the B(OB3)m(OSn)p species. Even if two B3 species have been detected in the 1D MAS-NMR 11B experiments, no clear description can be made from our experiment owing to the broadness signal due to the remaining second order quadrupolar effect. The only information that can be derived is that both B3 units are connected to P, B4 and also B3 units (Fig. 4 and 5) and that the B3–O–B3 linkages become the most important structural feature at high B2O3 amounts. 4.4
The tin polyhedra 119
Sn static NMR spectra have been simulated using the The Haeberlen convention24 to determine the isotropic chemical
Phys. Chem. Chem. Phys.
Table 3 119Sn NMR parameters. Isotropic chemical shifts (diso), DCSA and ZCSA are defined using the Haeberlen convention.22 The values are given with errors of 2 ppm, 5 ppm and 0.1, respectively
x
diso/ppm
0 5 12 25 33
789.9 785.4 741.7 589.4 490.0
dCSA/ppm 670 674 688 728 760
ZCSA 0.32 0.38 0.35 0.37 0.33
shift (diso), the chemical shift anisotropy (dCSA) and the anisotropy parameter (ZCSA) for each sample (Table 3). The values confirm the observation and allow distinguishing between the low- and intermediate B2O3 content (x = 0, 5 and 12) and high B2O3 (x = 25 and 33) content samples. Our results indicate a modification of the chemical environment of tin between the two domains. However, the origin of this modification (evolution of the coordination state, changes in the next nearest neighbours) cannot be derived from our experiments. 4.5
The mixed glass former effect
As shown by the correlation NMR, the mixing of P and B units gives rise to a very complex structure composed of three P (Q1, Q0 and Q01B) and at least six B (4B4 and 2B3) species. The unusual Tg evolution (Fig. 1) is discussed as follows in the light of the structural evolution. At x = 0, the glass structure is formed by dimeric entities P2O74 compensated by Sn2+ ions. When B2O3 is inserted at low amounts (x o 5 mol%), B3+ ions react with dimeric Q1 to create 2 Q01B units by breaking P–O–P and forming P–O–B linkages. The mixed network is also based on B(OP)4 groups (no B4–O–B4 linkage as shown in Fig. 5a) that change the glass former topology. The combination of both
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Paper
Fig. 10 Dominating structural units in the glass network at low (a), intermediate (b) and high (c) B2O3 contents.
information indicates that B2O3 insertion lead to the formation of B4(OP0)4 structural units surrounded by Sn2+ ions. Evolution of the charge compensation by Sn2+ ions of this new species is certainly at the origin of the creation of the pure Q0 site observed in the 31P NMR analysis. This composition range is thus characterised by an important creation of P–O–B bonds, more important than expected from the pure statistical distribution (Fig. 7) and this structural parameter is certainly at the origin of the important increase of the Tg observed in the region of x r 5. This conclusion is also supported by the maximum values of B4(1) (Fig. 7) and the end of the important increase of Q01B (Fig. 8) units observed for the x = 5 composition. At medium B2O3 contents (7 r x r 15), new structural features appear and replace the B–O–P based structural units. As illustrated in Fig. 8, B4(2) dominates the borate speciation at the beginning of the composition range before being replaced by B3 and B4(3) when B2O3 increases. As discussed in the previous section, all these three units are characterised by the presence of B–O–B linkages involving both B4 and B3 units. The phosphate network also experiences significant changes. First, the structure only contains a few original dimeric species and the Q0, created though the reorganisation of the Sn2+ ions, also decreases and disappears. At the end of the compositional range, the phosphate speciation is mainly formed by Q01B entities. It is also noticeable that the Q01B evolution is now close to the evolution expected from the statistical distribution. The disappearance of B4(OP0)4 structural units at the expense of B–O–B based species, simplification of the phosphate speciation (from 3 to a single species) and the appearance of B3 units (supposed to create a less reticulated network) impact the Tg evolution but in a smaller extent than in the low B2O3 amount range, resulting in the lower Tg increase observed in Fig. 1. It is also noticeable that in the 5–10 mol% range (where significant amounts of Q0 sites are present), B3(1) is the main tri-coordinated borate unit (Table 2). The appearance and development of B3(2) in the x Z 10 samples (whit very low Q0 proportions) suggest that
This journal is © the Owner Societies 2015
B3(2) requires a higher charge compensation than B3(1). Such an analysis may support meta- and pyro-borate (mono- and di-anionic species) assignments for the B3(1) and B3(2) signals, respectively. At high B2O3 amounts (15 r x r 33), the borate network is dominated by B3 and B4(3) species, accompanied by B4(4) units in low amounts. We proposed that the main structural breakthrough of this compositional range has to be related to the phosphate speciation. Indeed, after reaching 100% Q01B, the extension of the P/B mixing would require the formation of Q02B. There is no trace in the 31P 1D and 11B(31P) D-HMQC NMR experiments of such signals that should present a shielded chemical shift compared to Q01B. Therefore, we proposed that after having created Q01B units, B3+ ions do not interact with P anymore but with other B species and Sn2+ ions. This compositional range is thus characterised by the formation of a tin borate network that induces the unusual Tg increase observed in Fig. 1. This scenario is also supported by the static 119Sn NMR spectra lineshape evolution (Fig. 3) between the low/intermediate B2O3 content region, where Sn is attached to P, and the high B2O3 content region, where Sn–O–B linkages are created. Our study finally showed that B/P mixing occurs all along the composition line in this high modifier content formulation. This mechanism is therefore different from the one previously observed in the 66.6Na2S–xB2S3–(33.3 x)P2S5 glass system,25 that presents only a low degree of B–S–P connectivity. This difference indicates that the anion nature has a clear and major impact on the extent and nature of mixing in the glass network.
5. Conclusions The intermediate length scale structure of the complete 67SnO– xB2O3–(33 x)P2O5 composition line has been investigated using a high field correlation NMR protocol. All along the B/P substitution, 31P and 11B(31P) D-HMQC experiments show that the phosphate network structure evolved from dimer species
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
(P2O74 ) to isolated P connected to 1 B3+ ions (Q01B). The 11B MAS-NMR experiments indicate the presence of 4 tetracoordinated and 2 tri-coordinated species all along the B/P substitution. 11B(31P) D-HMQC and 11B DQ–SQ NMR experiments were used to tentatively assigned each species in terms of Bx–O–P and Bx–O–Bx bonds they are involved in. The complete set of data was then used to discuss the non-linear Tg evolution observed in the system. We determined that, at low B2O3 amounts (x r 5), the Tg increase can be explained by the preferential formation of P–O–B linkages leading to the replacement of P2O74 by Q01B and B(OP)4 species (Fig. 10a). At higher B2O3 amounts (7 r x r 15), the appearance of B4–O–B4, B4–O–B3 and B3–O–B3 linkages and the complete disappearance of dimeric species in benefit of Q01B species lead to a complex but less reticulated structure that contributes to the Tg increase but in a lower extent than at low B2O3 amounts (Fig. 10b). Finally, at high B2O3 amounts (15 r x), we showed that higher BP mixing that should produce Q02B units does not occur. Therefore, we proposed that the unusual Tg increase originate from the direct interactions between the B3+ and Sn2+ ions (Fig. 10c).
Acknowledgements G.T. would like to thank Region Nord pas de Calais, Europe (FEDER), CNRS, University of Lille and TGIR-RMN-THC FR3050 CNRS for funding and the anonymous reviewers for guidance and insightful comments.
References ¨rjesson, J. Power Sources, 1 C. Gejke, E. Zanghellini and L. Bo 2003, 119–121, 576–580. 2 A. Hayash, T. Konishi, K. Tadanaga, T. Minami and M. Tatsumisago, J. Non-Cryst. Solids, 2004, 345–346, 478–483. 3 T. Konishi, A. Hayashi, K. Tadanaga, T. Minami and M. Tatsumisago, J. Non-Cryst. Solids, 2008, 354, 380–385. 4 J. W. Lin, M. L. Schmitt, R. K. Brow and S. W. Yung, J. NonCryst. Solids, 2010, 356, 1379–1384. 5 A. Saitoh, G. Tricot, P. Rajbhandari, S. Anan and H. Takebe, Mater. Chem. Phys., 2015, 149–150, 648–656.
Phys. Chem. Chem. Phys.
PCCP
6 H. Takebe, T. Kobateke and A. Saitoh, Phys. Chem. Glasses: Eur. J. Glass Sci. Technol., Part B, 2013, 54, 182–186. 7 T. Tsuchiya and T. Moriya, J. Non-Cryst. Solids, 1980, 38–39, 323–328. 8 D. Zielnok, C. Cramer and H. Eckert, Chem. Mater., 2007, 19, 3162–3170. 9 B. Raguenet, G. Tricot, G. Silly, M. Ribes and A. Pradel, Solid State Ionics, 2012, 208, 25–30. 10 B. Raguenet, G. Tricot, G. Silly, M. Ribes and A. Pradel, J. Mater. Chem., 2011, 21, 17693–17704. 11 R. Christensen, J. Byer, G. Olson and S. W. Martin, J. NonCryst. Solids, 2012, 358, 826–831. 12 Q. Jiang, H. Zeng, Z. Liu, J. Ren, G. Chen, Z. Wang, L. Sun and D. Zhao, J. Chem. Phys., 2013, 139, 124502. 13 Z. Gan, J. Magn. Reson., 2007, 184, 39–43. 14 G. Tricot, J. Trebosc, F. Pourpoint, R. Gauvin and L. Delevoye, Annu. Rep. NMR Spectrosc., 2014, 81, 145–184. 15 Q. Wan, B. Hu, O. Lafon, J. Trebosc, F. Deng and J.-P. Amoureux, J. Magn. Reson., 2009, 200, 251–260. 16 A. Brinkmann and A. Kentgens, J. Am. Chem. Soc., 2006, 128, 14758–14759. 17 G. Tricot, B. Raguenet, G. Silly, M. Ribes, A. Pradel and H. Eckert, Chem. Commun., 2015, 51, 9284–9286. 18 A. Brinkmann and M. Eden, J. Chem. Phys., 2004, 120, 11726–11745. ¨llen, G. Tricot and S. Wegner, Solid State Nucl. 19 L. van Wu Magn. Reson., 2007, 32, 44–52. ´, 20 D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calve B. Alonso, J.-O. Durand, B. Bujoli, Z. Gan and G. Hoatson, Magn. Reson. Chem., 2002, 40, 70–76. 21 S. Elbers, W. Strojek, L. Koudelka and H. Eckert, Solid State Nucl. Magn. Reson., 2005, 27, 65–76. 22 D. Raskar, M. Rinke and H. Eckert, J. Phys. Chem. C, 2008, 112, 12530–12539. 23 D. Larink, H. Eckert, M. Reichert and S. W. Martin S, J. Phys. Chem. C, 2012, 116, 26162–26176. 24 U. Haeberlen, in Advances in Magnetic Resonance, Suppl. 1, ed. J. S. Waugh, Academic Press, New York, 1976. 25 D. Larink, H. Eckert and S. W. Martin, J. Phys. Chem. C, 2012, 116, 22698–22710.
This journal is © the Owner Societies 2015
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PAPER
Cite this: DOI: 10.1039/c5cp02095f
View Journal
Intermediate length scale organisation in tin borophosphate glasses: new insights from high field correlation NMR G. Tricot,*a A. Saitohb and H. Takebeb The structure of tin borophosphate glasses, considered for the development of low temperature sealing glasses or anode materials for Li-batteries, has been analysed at the intermediate length scale by a combination of high field standard and advanced 1D/2D nuclear magnetic resonance techniques. The nature and extent of B/P mixing were analysed using the 11B(31P) dipolar heteronuclear multiple quantum coherence NMR sequence and the data interpretation allowed (i) detecting the presence and analysing the nature of the B–O–P linkages, (ii) re-interpreting the 1D
31
P spectra and (iii) extracting the
proportion of P connected to borate species. Interaction between the different borate species was analysed using the
11
B double quantum–simple quantum experiment to (i) investigate the presence and
nature of the B–O–B linkage, (ii) assign the different borate species observed all along the composition Received 10th April 2015, Accepted 7th July 2015
line and (iii) monitor the borate network formation. In addition,
119
Sn static NMR was used to investigate
the evolution of the chemical environment of the tin polyhedra. Altogether, the set of data allowed
DOI: 10.1039/c5cp02095f
determining the structural units constituting the glass network and quantifying the extent of B/P mixing. The structural data were then used to explain the non-linear and unusual evolution of the glass
www.rsc.org/pccp
transition temperature.
1. Introduction Tin(+II) borophosphate (SnBP) glasses have attracted much attention during the last few years1–6 due to their (i) low glass transition temperature (Tg), (ii) wide glass forming composition range, (iii) good chemical and thermal durability and (iv) nontoxicity compared to standard low-Tg PbO-based glasses. The SnBP system has thus been considered for the development of low temperature sealing glasses or anode materials for Li-batteries.1,3 Macroscopic properties like Tg, thermal expansion coefficient, refractive index, and density have shown to be significantly affected by the formulation. The 67SnO–xB2O3–(33 x)P2O5 composition line has been widely investigated4,5 and presents a non-linear evolution of the macroscopic properties when P is replaced by B. Recently, owing to the preparation of the complete composition line (from x = 0 to 33), the non-linear evolution has been fully analysed and decomposed into three regions.5 While the first two regions show usual behaviours (strong and moderate evolution of the properties), the third region presents a second increase of Tg and density, whereas other borophosphate-based a
LASIR UMR-CNRS 8516, Universite´ de Lille 1, Villeneuve d’Ascq F-59655, France. E-mail: [email protected] b Graduate School of Science and Engineering, Ehime University, Matsuyama, 3 bunkyo-cho, 790-8577, Japan
This journal is © the Owner Societies 2015
systems generally experience constant or decreasing values at high B2O3 contents.7–12 In order to understand this behaviour, the structure has been previously analysed by Raman, infra-red and 1D 31P and 11B MAS-NMR.5 While the local order has been correctly described, no clear information has been obtained concerning the intermediate length scale organisation, i.e. the extent and nature of B/P mixing, although this information is usually considered as the key parameter governing the evolution of the properties. In this paper, new insights into the intermediate length scale organisation of 67SnO–xB2O3–(33 x)P2O5 glasses (0 r x r 33) have been obtained using advanced correlation NMR performed at very high magnetic field (18.8 T). The P/B connectivity scheme has been analysed using the 2D 11B/31P correlation map edited with the 11B(31P) dipolar heteronuclear multiple quantum coherence (D-HMQC) NMR technique,13,14 which allows highlighting and characterising the formation of the mixed P–O–B linkages. Progressive formation of the borate network by the P/B substitution has been monitored using the 11B double quantum–single quantum (DQ–SQ) sequence,15 which probes the formation and nature of the B–O–B bonds. In order to follow the evolution of the tin polyhedra coordination, 119 Sn NMR experiments were recorded using the static spin echo technique. Altogether, the structural results lead to the first detailed description of the medium range organisation and its evolution all along the P/B substitution. The structural model
Phys. Chem. Chem. Phys.
View Article Online
Paper
was finally used to discuss the non-linear Tg evolution, including the unusual Tg increase at high B2O3 amounts.
2. Experimental
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
2.1
Material synthesis and characterisations
x)P2O5 (0 r x r 33) glasses have been 67SnO–xB2O3–(33 prepared using the standard melt-quenching method. Appropriate mixtures of reagent grade SnO, P2O5 and B2O3 have been mixed, placed in a vitreous carbon crucible and heated up for 60 min at 950–1100 1C in a silica tube furnace under flowing argon in order to avoid oxidation of Sn2+ to Sn4+. The melts have then been cooled down to 400 1C in the silica tube before being cast in a carbon mould. The amorphous character of each sample has been confirmed by X-ray diffraction analysis. Tg have been determined by differential thermal analysis (DTA) using a TG-DTA8120 Rigaku machine. 25–40 mg of each sample, placed in a platinum pan, have been analysed from room temperature to 550 1C at a heating rate of 10 1C min 1. The errors in Tg were estimated to be 5 1C.
PCCP
to liquid H3PO4 (85%), solid NaBH4 and solid SnO2 at 0, and 604.3 ppm, respectively.
3. Results 3.1
1D/2D NMR analyses
The 31P and 11B MAS-NMR experiments have been performed on a 18.8 T Bruker Avance III spectrometer at 324 and 256.8 MHz, respectively. The experiments have been performed using a prototype 3.2 mm 11B/31P measurement probe equipped with a B free stator operating at a spinning frequency (nrot) of 20 kHz. The 1D 31P MAS-NMR acquisitions have been recorded with a 2 ms pulse length (p/6 pulse angle), a radiofrequency field (rf) strength of 40 kHz, 16–64 transients and a recycle delay (rd) of 120 s. The 1D 11B MAS-NMR experiments have been acquired with a 1 ms pulse length (p/12 pulse angle), a rf strength of 40 kHz, 1024–2048 transients and a rd of 2 s. The 11B/31P correlation maps have been edited using the D-HMQC sequence13–14 using a selective 11B p/2-t-p-t spin echo sequence pulse (p pulse length of 18 ms) modulated by two 31P p/2 pulses of 4.5 ms; the 11 31 B/ P dipolar interaction has been recoupled using the SR4 pulse scheme16 by irradiating the 31P channel with a 40 kHz rf field (2 nrot) during 1 ms. As recently demonstrated in the case of an alkali borophosphate, such a short recoupling time allows discussing the B/P spatial proximity highlighted by our experiment as evidenced for P–O–B chemical connectivity.17 The 2048 40 acquisition points have been recorded using a rotor synchronised t1-increment (50 ms), 256 transients and a rd of 2 s. The 11B/11B 2D maps have been recorded using the 11B DQ–SQ pulse sequence.15 The 2048(t2) 100(t1) acquisition points have been recorded under rotor-synchronised conditions with a p pulse length of 25 ms, 256 transients and a rd of 2 s. The 2D coherences have been created using a 10 kHz (0.5 nrot) and a 400 ms length BR212 pulse scheme.18 The 119Sn static NMR experiments were recorded at 149.1 MHz on a 9.4 T Bruker Avance II spectrometer using a 5 mm static probe. A p/2-t-p/2-t spin-echo sequence was used to record the signal with a 2 ms pulse length, a 50 ms echo delay, 3–5k transients and a rd of 25 s. The 31P, 11B and 119Sn chemical shift values have been referred
Phys. Chem. Chem. Phys.
Glass properties
The glass transition temperatures measured on the complete glass system are reported in Fig. 1 (empty squares). Tg continuously increases from 268 1C for the pure phosphate (x = 0) to 350 1C for the pure borate (x = 33) sample. However, Tg does not increase linearly and different regions can be distinguished by comparing the actual Tg and the linear Tg obtained by assuming a linear evolution from the pure phosphate to the pure borate samples (Fig. 1 dotted line). The Tg Tglin parameter evolution (Fig. 1, full circles) clearly shows the presence of three regions with (i) important (0 r x r 5), moderate (6 r x r 15) and low (18 r x r 33) deviations from the linear Tg values, indicating that the mixed network has different impacts on the macroscopic properties depending on the glass composition and the glass network organisation. 3.2
2.2
42.06
1D MAS-NMR
The 1D 31P MAS-NMR experiments (Fig. 2a) provide limited information about the phosphate speciation, due to very broad signals. While the global chemical shift appears toward positive values, no clear numbering or assignment can be derived from the 1D analysis. The 11B MAS-NMR spectra (Fig. 2b) provide interesting information about the local order of the borate species. Clear distinction between B4 and B3 species can be made through the chemical shift values (B4: 5/+3 ppm and B3: +20/+8 ppm). The spectra show that boron ions enter first as B4 species before being replaced by trigonal groups at higher B2O3 contents. The high resolution permitted by the high magnetic field also allows distinguishing between different B4 groups (denoted as B4(1), B4(2), B4(3) and B4(4)) through different signals at 3, 0.5, 1.5 and 3 ppm. 119The Sn static NMR spectra (Fig. 3) show very broad signals due to the large chemical shift anisotropy (CSA). Similar features can be observed for the low (x = 0, 5)
Fig. 1 Glass transition temperature (open squares), linear evolution (dotted line) and deviation from linear evolution (full circles) versus the B2O3 glass content.
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Fig. 2
Paper
31
P (a) and
11
B (b) MAS-NMR spectra performed at 18.8 T on the 67SnO–xB2O3–(33
Fig. 3 119Sn static NMR spectra performed at 9.4 T on the 67SnO–xB2O3– (33 x)P2O5 composition line.
and intermediate (x = 12) samples, whereas broader signals are observed in the case of the high B2O3 glass contents (x = 25 and 33). 3.3
Correlation NMR
The investigation of the P–O–B linkages has been performed using the 11B(31P) D-HMQC sequence. The 2D maps obtained for x = 3, 7, 18 and 25 samples are reported in Fig. 4. The 2D maps are accompanied by the 1D MAS-NMR 11B and 31P spectra (denoted as i and iii) and by the horizontal and vertical 2D spectrum projections (denoted as ii and iv). The 2D spectrum
This journal is © the Owner Societies 2015
x)P2O5 composition line.
obtained for the very low B2O3 amount sample (x = 3) presents a correlation signal indicating the presence of the P–O–B4(1) linkage. The 1D 31P MAS-NMR and 2D projection (Fig. 4a(iii and iv)) show significant differences indicating that some P within the glass matrix are not connected to any B, in a good agreement with the low B2O3 amount of the sample. The 2D map obtained for the x = 7 sample displays a complex pattern composed by three correlation signals indicating the presence of P–O–B4(1), P–O–B4(2) and P–O–B3 linkages. For this sample again, some P5+ are not connected to B3+ ions, as indicated by the comparison between Fig. 4b(iii) (1D 31P MAS-NMR) and Fig. 4b(iv) (2D map 31P projection). Slight differences can be observed between the three correlation signals in the 2D maps, indicating that the phosphate species attached to each borate ion are different. However, these differences are weak (less than 2 ppm for the chemical shift) and, to avoid any over-interpretation, the three signals have been gathered into a single gaussian line (Fig. 4b(iv)) that will be considered as the signature of the P attached to the borate groups in the following. At higher B2O3 contents (x = 18 and 27), the 2D maps exhibit similar patterns. The maps present intense correlation signals involving P and B3 signals accompanied by correlation signals indicating the presence of P–O–B4(2) and P–O–B4(3) linkages. The presence and nature of B–O–B bonds have been investigated using the 11B DQ–SQ NMR technique. The 2D maps obtained for the x = 3, 7, 18 and 30 samples are reported in Fig. 5. No signal is observed in the 2D map of the x = 3 sample (Fig. 5a), indicating that the boron ions are completely dissolved into the phosphate network. Four different signals can be observed in the x = 7 sample 2D map. Three on-diagonal signals indicate auto-correlation for the B4(1), B4(2) and B3 species accompanied by two off-diagonal signals involving the B4(2) and B3 units. The two 2D maps obtained for the x = 18 and
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
PCCP
Fig. 4 2D 11B(31P) D-HMQC spectra obtained for x = 3 (a), 7 (b), 18 (c) and 25 (d) samples. The 2D maps are accompanied with 1D 11B and 31P MAS-NMR spectra (i and iii, respectively) and the 11B and 31P 2D projections (ii and iv, respectively).
30 samples display similar correlation schemes as B4(3) and B3 on- and B4(3)/B3 off-diagonal correlation signals.
4. Discussion 4.1
Tg evolution
While the three-domain evolution of the properties, depicted in Fig. 1, is a common feature in the case of B/P substitution in borophosphate glasses,7–10 the final increase of Tg in high B2O3 content glasses has not been widely reported, indicating that the system experiences an unusual structural evolution. If the MGFE is always positive (the actual Tg values are always higher than the linear values), it turns out that the maximum deviation occurs in low B2O3 glasses (region (I), x o 5) and higher B2O3 glasses exhibit lesser differences between the actual and the linear Tg values. These evolutions are governed by specific structural rearrangements that need to be highlighted in order to understand how the structure rules the macroscopic properties of these materials. New insight into the intermediate length scale organisation and the B/P mixing can be derived from the presented 1D/2D NMR investigation. 4.2
The phosphate speciation
As depicted in Fig. 2, the 1D 31P MAS-NMR spectra show a progressive evolution of the global chemical shift toward positive values, indicating the presence of depolymerised units.
Phys. Chem. Chem. Phys.
However, no clear numbering and assignment can be derived due to the broad and unresolved signals. The 2D 11B/31P correlation maps (Fig. 4) indicate clear correlations resulting from the formation of mixed P–O–B linkages. The latter involve different B4 groups and B3 units also participate in the mixed linkages at higher B2O3 contents. The NMR signature of the P connected to B units (denoted as QnmB in the following with n and m being the number of connected P and B atoms as previously used in the case of aluminophosphate glasses)19 can be easily derived from the 2D map by plotting the vertical projections. All the 31P projections are reported in Fig. 6a with representative fits allowing for a precise determination of the chemical shift and the full width at half maximum (fwhm) values of the PB units. These parameters have then been used as input data to deconvolute 1D 31P MAS-NMR using the dmfit software.20 In addition to the QnmB signals, Q1 (connected to 1 other P atom) and Q0 (isolated phosphate group) signals, that constitute the base formulation (x = 0) have been added to build a proper fit. If the presence of Q1 ( 19 ppm) is expected from the pure pyrophosphate structure, a small amount of Q0 ( 10 ppm) is also observed that could originate from a minor oxidation of Sn2+ to Sn4+ or to a slight deviation from the pyrophosphate composition. All the NMR parameters (chemical shifts, full widths at half maximum, and relative proportions) deduced from the deconvolutions are gathered in Table 1 and the representative simulations of the 1D 31P MASNMR experiments are reported in Fig. 6b.
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Paper
Fig. 5 2D 11B DQ–SQ spectra obtained for x = 3 (a), 7 (b), 18 (c) and 30 (d) samples. The 2D maps are accompanied with 1D 11B (i) and 11B 2D horizontal and vertical projections (ii and iii, respectively).
Fig. 6 31P projections of the 2D D-HMQC maps, accompanied with the fits (dotted lines) used to deconvolute the 1D MAS-NMR spectra (a), deconvolution of the representative 1D 31P MAS-NMR spectra using the parameters deduced from the 2D D-HMQC experiments (b).
As reported in Table 1, the chemical shift values of the QnmB species always lie between the Q0 and Q1 groups. Addition of a P–O–B linkage inducing a shielding effect, the QnmB species can be finally assigned to Q0mB and the presence of Q1mB that should produce a more negative chemical shift10,21–23 than the Q1 unit
This journal is © the Owner Societies 2015
can be reasonably ruled out. Moreover, we propose that the species is attached to a single B3+ ion, leading to a final assignment of Q01B, since no clear change in the chemical shift of this species can be observed when B2O3 increases. This assignment allows understanding the formation mechanism
Phys. Chem. Chem. Phys.
View Article Online
Paper
PCCP
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Table 1 31P NMR parameters obtained using the gaussian model. Chemical shifts (diso), full widths at half maximum (fwhm), relative proportions (rel. prop.) are given with errors of 0.2 ppm, 0.2 ppm, 1%, 0.05 MHz and 0.05, respectively
x
Qx
0
1
Q Q0
18.6 9.1
12.2 10.5
94.0 6.0
1
Q1 Q0 Q0B
19.4 10.7 15.6
12.5 11.4 11.1
3
Q1 Q0 Q0B
20.1 10.0 15.2
5
Q1 Q0 Q0B
7
Q1 Q0 Q0B
diso/ppm
fwhm/ppm
Rel. prop./%
x
Qx
Rel. prop./%
18.5 9.7
12.0 10.6
4.0 23.8
Q0B Q1 Q0
12.5 19.0 7.9
11.0 12.0 10.4
72.2 4.0 10.1
20
Q0B Q1 Q0
11.5 19.2 7.5
10.8 12.0 10
84.9 5.0 2.0
16.3 32.8 50.9
25
Q0B Q1 Q0
11.0 19.2 7.5
10.8 12.0 10
93.0 3.0 1.5
15.9 28.1 56
30
Q0B Q1 Q0 Q0B
10.8 19.1 7.5 10.5
10.7 12.0 10.0 11.0
95.5 2.0 2.0 96.0
10
Q Q0
64.1 22.9 13.0
15
12.2 10.8 11.2
37.0 27.9 35.1
20.4 10.0 15.1
11.3 10.5 11.4
18.9 9.2 13.6
12.1 10.8 11.0
of the P–O–B linkages: borate units directly interact with P2O74 groups by breaking a P–O–P bond and forming a P–O–B linkage. The Qn relative proportions, reported in Fig. 7, also support this formation mechanism. When B2O3 is introduced, the proportion of Q1 sites, consumed in the P–O–B formation, is significantly reduced in favor of the Q01B increase. The Q01B species experiences a non-linear evolution composed by an important and a moderate increase in the 0 r x r 5 and 5 r x r 17 compositional ranges, followed by an almost constant proportion in the 17 r x r 30 B2O3 range. This evolution has been compared to the statistical dimeric distribution derived from a pure random mixing of B and P (Fig. 7 dotted lines) forming dimers. The comparison clearly indicates that P and B elements preferentially interact at low amounts (x r 5) to create the mixed linkages. It has also to be observed that the phosphate speciation does not evolve after the B2O3
Fig. 7 Evolution of the phosphate speciation in the borophosphate mixed network. The proportion of PB calculated from the dimeric statistical distribution is reported using a dotted line.
Phys. Chem. Chem. Phys.
diso/ppm
fwhm/ppm
1
content reaches 17 mol%, where Q01B species fully dominate the phosphate network. Moderate formation of Q0 sites up to x = 5 is also observed before this site proportion decreases and completely disappears above 15 mol% of B2O3. 4.3
The borate speciation
The borate speciation can been precisely analysed owing to the highly resolved 11B NMR spectra obtained at 18.8 T. In addition to the usual progressive replacement of B4 by B3 species when the B2O3 content increases, our experiments also highlight a complex speciation for the tetra-coordinated boron though the presence of four different B4 sites. Quantification and evolution of the different borate species all along the composition line have been obtained by deconvolution of the spectra. All the 11B NMR parameters for both B4 and B3 have been gathered in Table 2 and the evolution of the relative proportions versus the B2O3 content is reported in Fig. 8. B4(1) is the main species at very low B2O3 amounts and decreases beyond 5 mol% of B2O3 when B4(2) starts to increase. At x = 10, B4(2) decreases and B3 and B4(3) start to dominate the borate network. B4(3) increases slowly up to the pure borate whereas B3 experiences a significant increase. B4(4) appears at x = 15 and increases moderately up to the pure borate composition but is always present as minor species. The information can also be obtained by the B3 speciation, in spite of the broad signal coming from the remaining second order quadrupolar effect. At low B2O3 amounts (3 r x r 7), only one signal is detected in the experiments with a chemical shift evolving from 15 to 18.7 ppm. The second B3 signal is observed in the x = 12 composition. For higher B2O3 contents, both signals are present with constant chemical shifts and with an increasing relative proportion of the second B3 signal. Due to the lack of additional information about the B3 signals, B3(1) and B3(2) were treat as a single component in Fig. 8 to discuss the evolution of the global B3 species all along the B/P substitution.
This journal is © the Owner Societies 2015
View Article Online
PCCP
Paper
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Table 2 11B NMR parameters, B4 and B3 units have been simulated using the gaussian/lorentzian and quadrupolar models available in the dmfit software.18 Chemical shifts (diso), full widths at half maximum (fwhm), relative proportions (rel. prop.), the quadrupolar constant (CQ) and asymmetry parameters (ZQ) are given with errors of 0.2 ppm, 0.2 ppm, 1%, 0.1 MHz and 0.1, respectively
x
B4
B3
1
B4(1) B4(2)
diso/ppm
CQ/MHz
ZQ
3
86.3 10.1 1.7
B3(1)
15.1
2.5
0.5
1.9
1.7 1.6 1.9
60.6 28.8 4.7
B3(1)
16.7
2.5
0.5
5.9
3.3 0.9 1.1
1.7 2.1 1.9
21.8 49.4 5.2
B3(1)
18.7
2.6
0.5
23.6
B4(1) B4(2) B4(3)
2.5 0.2 1.4
2.2 2.0 2.0
3.2 29.3 15.2
B3(1) B3(2)
18.7 17.7
2.8 2.8
0.6 0.5
27.7 24.6
15
B4(1) B4(2) B4(3) B4(4)
1.9 0.0 1.4 3.4
2.1 1.9 2.0 1.9
1.3 17.7 20.5 1.8
B3(1) B3(2)
18.8 17.9
2.8 2.8
0.6 0.5
30.5 28.2
20
B4(1) B4(2) B4(3) B4(4)
1.8 0.1 1.5 3.5
2.0 1.9 2.1 1.8
0.5 10.5 24.4 2.4
B3(1) B3(2)
18.8 17.9
2.7 2.7
0.5 0.5
34.4 27.9
25
B4(1) B4(2) B4(3) B4(4)
1.7 0.2 1.5 3.2
2.0 1.9 2.1 2.2
0.6 8.0 23.3 4.9
B3(1) B3(2)
18.7 17.9
2.8 2.8
0.5 0.6
40.8 22.4
30
B4(1) B4(2) B4(3) B4(4)
1.7 0.2 1.5 3.2
2.0 1.9 2.1 2.2
0.2 6.8 23.6 6.6
B3(1) B3(2)
18.9 18.0
2.7 2.6
0.5 0.5
31.9 30.9
diso/ppm
fwhm/ppm
Rel. prop./%
3.4 1.2
1.6 1.6
97.1 2.9
B4(1) B4(2) B4(3)
3.4 1.2 0.1
1.6 1.6 1.8
5
B4(1) B4(2) B4(3)
3.3 1.1 0.5
7
B4(1) B4(2) B4(3)
12
Fig. 8 B3 and B4(x) species evolution versus the B2O3 glass content. The amounts are expressed in terms of borate units present in the glass network.
This journal is © the Owner Societies 2015
Rel. prop./%
Beyond the proportion of the borate species and their evolution, the key information lies in the chemical environment and the B/P and B/B mixing they are involved in. This information can be derived and discussed from the correlation NMR results presented in the previous section and summarised in Fig. 9. This figure presents the 11B MAS-NMR spectra (Fig. 9(i)) accompanied by relevant 11B analyses extracted from the 11B(31P) D-HMQC (Fig. 9(ii), revealing the B involved in B–O–P linkages) and the 11B DQ–SQ (Fig. 9iii and iv, revealing the B involved in B–O–B linkages) correlation maps. The first B4 site, denoted as B4(1), presents a chemical shift of 3 ppm and is generally assigned to the B(OP)4 structural unit.21–23 While our results fully support this assignment at low B2O3 amounts (Fig. 9a(i–iii)), the presence of autocorrelation signals in the 11B DQ–SQ projection of the x = 7 sample (Fig. 9a(iv)) suggests the presence of the B4(1)–O–B4(1) linkage that cannot fit with the usual B(OP)4 assignment, but is rather indicating a B(OP)3(OB4) structure, as previously shown in lithium borophosphate by 11B(31P) REDOR experiments.10 Therefore, in
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
PCCP
Fig. 9 Comparison of the 11B MAS-NMR analysis of B4(1) (a), B4(2) (b), B4(3) (c) and B4(4) (d): (i): 1D MAS-NMR, (ii): 11B projection of D-HMQC maps, (iii) and (iv): 11 B projection of the DQ-DQ maps.
order to include both species under the same nomenclature (since they probably cannot be distinguished because of similar chemical shift values), we propose that B4(1) is a tetra-coordinated boron connected to P but also B4 units leading to a general B(OX4)4 assignment with X = P, B. The second site (B4(2): 0.5 ppm) is involved in the B–O–P bond (Fig. 9b(ii)), B–O–B4 (Fig. 9b(iii)), but also in B–O–B3 as indicated by Fig. 9b(iv) and by the off-diagonal correlation signals observed in the 2D 11B DQ–SQ map (Fig. 5b). This new species, which originates from the replacement of a tetracoordinated species (P/B4) by a trigonal and planar B3 unit, is the first species involving the combination between the three glasses forming polyhedra (P, B4 and B3) and can be assigned to the B(OX)4–n(OB3)n (X = P, B4) species. The third B4 species also appears to be connected to P, B4 and B3 units, as shown by Fig. 9c(ii–iv). Therefore, we proposed that significant changes in the chemical environment compared to B4(2) can only originate from an interaction with the Sn2+ ions, leading to the B(OX)n(OB3)m(OSn)p assignment. Finally, Fig. 9d show that the fourth B4 site is attached to B3, but is not involved in B4–O–P and B4–O–B4 linkages anymore (Fig. 9d(ii and iii)). This last site can thus be tentatively assigned to the B(OB3)m(OSn)p species. Even if two B3 species have been detected in the 1D MAS-NMR 11B experiments, no clear description can be made from our experiment owing to the broadness signal due to the remaining second order quadrupolar effect. The only information that can be derived is that both B3 units are connected to P, B4 and also B3 units (Fig. 4 and 5) and that the B3–O–B3 linkages become the most important structural feature at high B2O3 amounts. 4.4
The tin polyhedra 119
Sn static NMR spectra have been simulated using the The Haeberlen convention24 to determine the isotropic chemical
Phys. Chem. Chem. Phys.
Table 3 119Sn NMR parameters. Isotropic chemical shifts (diso), DCSA and ZCSA are defined using the Haeberlen convention.22 The values are given with errors of 2 ppm, 5 ppm and 0.1, respectively
x
diso/ppm
0 5 12 25 33
789.9 785.4 741.7 589.4 490.0
dCSA/ppm 670 674 688 728 760
ZCSA 0.32 0.38 0.35 0.37 0.33
shift (diso), the chemical shift anisotropy (dCSA) and the anisotropy parameter (ZCSA) for each sample (Table 3). The values confirm the observation and allow distinguishing between the low- and intermediate B2O3 content (x = 0, 5 and 12) and high B2O3 (x = 25 and 33) content samples. Our results indicate a modification of the chemical environment of tin between the two domains. However, the origin of this modification (evolution of the coordination state, changes in the next nearest neighbours) cannot be derived from our experiments. 4.5
The mixed glass former effect
As shown by the correlation NMR, the mixing of P and B units gives rise to a very complex structure composed of three P (Q1, Q0 and Q01B) and at least six B (4B4 and 2B3) species. The unusual Tg evolution (Fig. 1) is discussed as follows in the light of the structural evolution. At x = 0, the glass structure is formed by dimeric entities P2O74 compensated by Sn2+ ions. When B2O3 is inserted at low amounts (x o 5 mol%), B3+ ions react with dimeric Q1 to create 2 Q01B units by breaking P–O–P and forming P–O–B linkages. The mixed network is also based on B(OP)4 groups (no B4–O–B4 linkage as shown in Fig. 5a) that change the glass former topology. The combination of both
This journal is © the Owner Societies 2015
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
PCCP
Paper
Fig. 10 Dominating structural units in the glass network at low (a), intermediate (b) and high (c) B2O3 contents.
information indicates that B2O3 insertion lead to the formation of B4(OP0)4 structural units surrounded by Sn2+ ions. Evolution of the charge compensation by Sn2+ ions of this new species is certainly at the origin of the creation of the pure Q0 site observed in the 31P NMR analysis. This composition range is thus characterised by an important creation of P–O–B bonds, more important than expected from the pure statistical distribution (Fig. 7) and this structural parameter is certainly at the origin of the important increase of the Tg observed in the region of x r 5. This conclusion is also supported by the maximum values of B4(1) (Fig. 7) and the end of the important increase of Q01B (Fig. 8) units observed for the x = 5 composition. At medium B2O3 contents (7 r x r 15), new structural features appear and replace the B–O–P based structural units. As illustrated in Fig. 8, B4(2) dominates the borate speciation at the beginning of the composition range before being replaced by B3 and B4(3) when B2O3 increases. As discussed in the previous section, all these three units are characterised by the presence of B–O–B linkages involving both B4 and B3 units. The phosphate network also experiences significant changes. First, the structure only contains a few original dimeric species and the Q0, created though the reorganisation of the Sn2+ ions, also decreases and disappears. At the end of the compositional range, the phosphate speciation is mainly formed by Q01B entities. It is also noticeable that the Q01B evolution is now close to the evolution expected from the statistical distribution. The disappearance of B4(OP0)4 structural units at the expense of B–O–B based species, simplification of the phosphate speciation (from 3 to a single species) and the appearance of B3 units (supposed to create a less reticulated network) impact the Tg evolution but in a smaller extent than in the low B2O3 amount range, resulting in the lower Tg increase observed in Fig. 1. It is also noticeable that in the 5–10 mol% range (where significant amounts of Q0 sites are present), B3(1) is the main tri-coordinated borate unit (Table 2). The appearance and development of B3(2) in the x Z 10 samples (whit very low Q0 proportions) suggest that
This journal is © the Owner Societies 2015
B3(2) requires a higher charge compensation than B3(1). Such an analysis may support meta- and pyro-borate (mono- and di-anionic species) assignments for the B3(1) and B3(2) signals, respectively. At high B2O3 amounts (15 r x r 33), the borate network is dominated by B3 and B4(3) species, accompanied by B4(4) units in low amounts. We proposed that the main structural breakthrough of this compositional range has to be related to the phosphate speciation. Indeed, after reaching 100% Q01B, the extension of the P/B mixing would require the formation of Q02B. There is no trace in the 31P 1D and 11B(31P) D-HMQC NMR experiments of such signals that should present a shielded chemical shift compared to Q01B. Therefore, we proposed that after having created Q01B units, B3+ ions do not interact with P anymore but with other B species and Sn2+ ions. This compositional range is thus characterised by the formation of a tin borate network that induces the unusual Tg increase observed in Fig. 1. This scenario is also supported by the static 119Sn NMR spectra lineshape evolution (Fig. 3) between the low/intermediate B2O3 content region, where Sn is attached to P, and the high B2O3 content region, where Sn–O–B linkages are created. Our study finally showed that B/P mixing occurs all along the composition line in this high modifier content formulation. This mechanism is therefore different from the one previously observed in the 66.6Na2S–xB2S3–(33.3 x)P2S5 glass system,25 that presents only a low degree of B–S–P connectivity. This difference indicates that the anion nature has a clear and major impact on the extent and nature of mixing in the glass network.
5. Conclusions The intermediate length scale structure of the complete 67SnO– xB2O3–(33 x)P2O5 composition line has been investigated using a high field correlation NMR protocol. All along the B/P substitution, 31P and 11B(31P) D-HMQC experiments show that the phosphate network structure evolved from dimer species
Phys. Chem. Chem. Phys.
View Article Online
Published on 07 July 2015. Downloaded by Florida State University on 28/09/2015 11:25:09.
Paper
(P2O74 ) to isolated P connected to 1 B3+ ions (Q01B). The 11B MAS-NMR experiments indicate the presence of 4 tetracoordinated and 2 tri-coordinated species all along the B/P substitution. 11B(31P) D-HMQC and 11B DQ–SQ NMR experiments were used to tentatively assigned each species in terms of Bx–O–P and Bx–O–Bx bonds they are involved in. The complete set of data was then used to discuss the non-linear Tg evolution observed in the system. We determined that, at low B2O3 amounts (x r 5), the Tg increase can be explained by the preferential formation of P–O–B linkages leading to the replacement of P2O74 by Q01B and B(OP)4 species (Fig. 10a). At higher B2O3 amounts (7 r x r 15), the appearance of B4–O–B4, B4–O–B3 and B3–O–B3 linkages and the complete disappearance of dimeric species in benefit of Q01B species lead to a complex but less reticulated structure that contributes to the Tg increase but in a lower extent than at low B2O3 amounts (Fig. 10b). Finally, at high B2O3 amounts (15 r x), we showed that higher BP mixing that should produce Q02B units does not occur. Therefore, we proposed that the unusual Tg increase originate from the direct interactions between the B3+ and Sn2+ ions (Fig. 10c).
Acknowledgements G.T. would like to thank Region Nord pas de Calais, Europe (FEDER), CNRS, University of Lille and TGIR-RMN-THC FR3050 CNRS for funding and the anonymous reviewers for guidance and insightful comments.
References ¨rjesson, J. Power Sources, 1 C. Gejke, E. Zanghellini and L. Bo 2003, 119–121, 576–580. 2 A. Hayash, T. Konishi, K. Tadanaga, T. Minami and M. Tatsumisago, J. Non-Cryst. Solids, 2004, 345–346, 478–483. 3 T. Konishi, A. Hayashi, K. Tadanaga, T. Minami and M. Tatsumisago, J. Non-Cryst. Solids, 2008, 354, 380–385. 4 J. W. Lin, M. L. Schmitt, R. K. Brow and S. W. Yung, J. NonCryst. Solids, 2010, 356, 1379–1384. 5 A. Saitoh, G. Tricot, P. Rajbhandari, S. Anan and H. Takebe, Mater. Chem. Phys., 2015, 149–150, 648–656.
Phys. Chem. Chem. Phys.
PCCP
6 H. Takebe, T. Kobateke and A. Saitoh, Phys. Chem. Glasses: Eur. J. Glass Sci. Technol., Part B, 2013, 54, 182–186. 7 T. Tsuchiya and T. Moriya, J. Non-Cryst. Solids, 1980, 38–39, 323–328. 8 D. Zielnok, C. Cramer and H. Eckert, Chem. Mater., 2007, 19, 3162–3170. 9 B. Raguenet, G. Tricot, G. Silly, M. Ribes and A. Pradel, Solid State Ionics, 2012, 208, 25–30. 10 B. Raguenet, G. Tricot, G. Silly, M. Ribes and A. Pradel, J. Mater. Chem., 2011, 21, 17693–17704. 11 R. Christensen, J. Byer, G. Olson and S. W. Martin, J. NonCryst. Solids, 2012, 358, 826–831. 12 Q. Jiang, H. Zeng, Z. Liu, J. Ren, G. Chen, Z. Wang, L. Sun and D. Zhao, J. Chem. Phys., 2013, 139, 124502. 13 Z. Gan, J. Magn. Reson., 2007, 184, 39–43. 14 G. Tricot, J. Trebosc, F. Pourpoint, R. Gauvin and L. Delevoye, Annu. Rep. NMR Spectrosc., 2014, 81, 145–184. 15 Q. Wan, B. Hu, O. Lafon, J. Trebosc, F. Deng and J.-P. Amoureux, J. Magn. Reson., 2009, 200, 251–260. 16 A. Brinkmann and A. Kentgens, J. Am. Chem. Soc., 2006, 128, 14758–14759. 17 G. Tricot, B. Raguenet, G. Silly, M. Ribes, A. Pradel and H. Eckert, Chem. Commun., 2015, 51, 9284–9286. 18 A. Brinkmann and M. Eden, J. Chem. Phys., 2004, 120, 11726–11745. ¨llen, G. Tricot and S. Wegner, Solid State Nucl. 19 L. van Wu Magn. Reson., 2007, 32, 44–52. ´, 20 D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calve B. Alonso, J.-O. Durand, B. Bujoli, Z. Gan and G. Hoatson, Magn. Reson. Chem., 2002, 40, 70–76. 21 S. Elbers, W. Strojek, L. Koudelka and H. Eckert, Solid State Nucl. Magn. Reson., 2005, 27, 65–76. 22 D. Raskar, M. Rinke and H. Eckert, J. Phys. Chem. C, 2008, 112, 12530–12539. 23 D. Larink, H. Eckert, M. Reichert and S. W. Martin S, J. Phys. Chem. C, 2012, 116, 26162–26176. 24 U. Haeberlen, in Advances in Magnetic Resonance, Suppl. 1, ed. J. S. Waugh, Academic Press, New York, 1976. 25 D. Larink, H. Eckert and S. W. Martin, J. Phys. Chem. C, 2012, 116, 22698–22710.
This journal is © the Owner Societies 2015
