Evidence of crystalline/glassy intermediates in bismuth phosphates

Abstract

³¹P and ¹⁷O NMR investigations have been achieved on bismuth oxide phosphates by a comparison between ordered and semi-ordered reference compounds; the wide chemical shift range for ¹⁷O is revealed to be a profitable source of information about partially ordered materials.

---

Bi₂O₃–MO–P₂O₅, (M = Co, Cu, Cd, Zn, Mn…) ternary systems have been well investigated, leading to the characterization of new bismuth oxide phosphates having particular structural relationships.^1–8 As intensively detailed,⁹ the rigid frameworks of these materials can be considered as an assembly by the edge sharing of O(Bi,M)₄ polyhedra, leading to infinite polycationic ribbons of variable width: one, two, three, … tetrahedra wide, surrounded by isolated PO₄groups. From XRD/ND crystal structure studies, we classified these compounds as “disordered” or “ordered”, depending on the competition (or not) between several O₄ configurations around the central P sites. Indeed, in ordered compounds such as BiM₂PO₆,^3,4,10,11 the Bi³⁺ cations strictly sit in the middle of ribbons, whereas the M²⁺ cations are located at their edges. Similarly, in disordered compounds, e.g. Bi_1.2M_1.2PO_5.5,⁷Bi³⁺ cations still occupy the middle of the ribbons, whereas the edges of the ribbons are filled by mixed site Bi³⁺/M²⁺. This statistical distribution leads to a variable orientation of the PO₄groups, depending on the local nature of its first (Bi, M) cationic shell. The disorder is all the more important because it also affects partially filled cationic channels (so-called tunnels hereafter) surrounded by PO₄groups in between pairs of ribbons. Of course, the notion of disorder is inexact because of the existence at the microscopic scale of incommensurate modulated phenomena (mainly along b*) in most of the disordered compounds.^12–14 This extra information does not survive over long range scales, e.g., it is not observed in the XRD of single crystals. Therefore, sometimes only ordered fragments of the disordered PO₄/tunnel interstitial areas can be assumed from the average crystal structure, on the basis of plausible interatomic distances.¹³ However, structural interactions between the edges of the ribbons, PO₄groups and the tunnel is far from being fully established, probably due to various phenomena, including anti-phase boundary defects within tunnels and the probable semi-ordered zones in these materials. In view of a complementary approach to these fascinating series and by an easy extension to different compounds, ³¹P and ¹⁷O NMR spectroscopy have been used as local probes. Indeed, recent technical advances in solid state NMR has led to the emergence of this technique by adapting it for use with low natural abundance nuclei having relatively small gyromagnetic ratios, such as ¹⁷O. In addition, it is worth mentioning that impedance spectroscopy measurements on all of these materials (ordered and disordered) show low ionic mobilities due to the strong P–O bonds involved for most of the oxygen ions. Therefore, only the static aspect is considered hereafter.

Thus, the present work focuses on a comparison between a typical ordered and disordered compound, with the aim of establishing the pertinence, complementarities, and limits of both ³¹P and ¹⁷O nuclei as probes with regard to the structural aspects of ordered vs. semi-ordered materials. With that aim in mind, two compounds have been selected from among the series:

(i) BiCd₂PO₆ was chosen as the archetype of ordered compounds. Its structure is isostructural to BiZn₂PO₆,^15,16 and it is interesting because it crystallizes in the Bbmm space group, while many members of the BiM₂PO₆ class (including the M = Zn term) adopt the less symmetrical Pnma space group. The coordination around its unique phosphorus position is constituted by two independent O2 (2×) and O3 (2×) atoms, while O1 is located in the two tetrahedra-wide ribbons at the center of a OBi₂Cd₂ tetrahedron (Fig. 1(a)).

Fig. 1 The structures of (a) BiCd₂PO₆ and (b) Bi_1.2Zn_1.2PO_5.5. BiCd₂PO₆ consist of [Cd₄Bi₂O₂]-ordered double ribbons surrounded by six isolated ordered phosphates. Bi_1.2Zn_1.2PO_5.5 is disordered because of (1) the presence of mixed Bi³⁺/Zn²⁺ sites at the edges of triple [(Bi_0.15Zn_0.85)₄Bi₄O₆] ribbons, (2) disordered tunnels, partially occupied by Zn²⁺, and (3) multiple PO₄ configurations around the same phosphorus.

(ii) The simplest disordered compounds have the Bi_∼1.2M_∼1.2PO_5.5 general formula (M = Mn, Co, Zn). Their structure (space groupIcma) is formed of triple ribbons with mixed Bi/M edges (for M = Zn: 15% Bi³⁺/85% Zn²⁺). The partially M-filled tunnels and surrounding disordered PO₄groups are shown in Fig. 1(b). Only one phosphorus position exists, even if it has finally been split into two close satellites, P1 (∼50%) and P2 (∼50%), in the published model. To avoid any paramagnetic perturbation, the M = Zn compound was selected. It is noteworthy that the influence of Cd²⁺ for Zn²⁺ replacement in BiM₂PO₆ on the ³¹P NMR chemical shift has already been fully quantified on the basis of the empirical z/a² parameter,¹⁵ and no additional contribution is expected between these two neighboring cations. The possibility of quantifying the local cationic environment of the PO₄groups in a Bi(M,M′)PO₆ statistical solid solution compounds has also been enhanced.

³¹ P MAS NMR : Fig. 2(a) and (b) show the ³¹P NMR spectra of BiCd₂PO₆ against Bi_1.2Zn_1.2PO_5.5, which clearly reveals the broadening of the signal for the latter due to the multitude of individual resonances in the disordered compound. It is comparable to the IR spectra of ordered vs. disordered compounds presented elsewhere.¹³ In that sense, the broad envelope does not show discrete contributions but rather a continuum. Here, in addition to the local distortion of each individual PO₄group, the influence of the nature of the neighboring Zn/Bi cationic shell has to be considered.¹⁵ Furthermore, the ³¹P double quantum MAS-NMR spectrum shows no particular privileged out-of-diagonal correlations (Fig. 2(c)) reminiscent of a glass-like state from the ³¹P NMR resolution.

Fig. 2 ³¹P MAS-NMR (9.4 T) spectra of (a) BiCd₂PO₆ and (b) Bi_1.2Zn_1.2PO_5.5. The spectra were acquired at an MAS speed of 10 kHz, with a short pulse excitation of 1.5 μs (20°) and a recycling delay of 20 s. (c) ³¹P double quantum MAS-NMR spectrum of Bi_1.2Zn_1.2PO_5.5. The spinning frequency was 10 kHz. The excitation and reconversion period was composed of back-to-back 90° pulses²¹ of 4 μs, which gave a total excitation/reconversion time of 400 μs. The repetition time was 30 s, preceded by a presaturation period. A total of 16 scans were used and 64 t1 increments were collected. The ³¹P chemical shift was referenced externally to an 85% H₃PO₄ solution at 0 ppm.

Since oxygen occupies both the polycationic regular sublattice and the disordered interstitial regions, ¹⁷O NMR analysis would be expected to give relevant information about disorder. Here, samples were enriched via the ¹⁷O enrichment method developed by Flambard et al.¹⁷ Due to the presence of water vapor, the sample was checked by ¹H NMR to ensure that all protons disappeared at the end of the enrichment. Another difficulty in obtaining ¹⁷O NMR spectra is the presence of the quadrupolar interactions of the nuclei (spin I = 5/2) that largely broaden signals. This requires suitable techniques, such as double rotation (DOR),¹⁸ multiple-quantum magic angle spinning (MQ-MAS)¹⁹ or satellite transition magic angle spinning (ST-MAS),²⁰ in order to remove the anisotropic broadenings that remain under magic angle spinning conditions.

¹⁷ O MAS NMR : Fig. 3 shows the high resolution MQ-MAS spectra of (a) BiCd₂PO₆ and (b) Bi_1.2Zn_1.2PO_5.5. The horizontal projections (top) correspond to MAS spectra still broadened by the second order quadrupolar interaction. The vertical projections reveal ¹⁷O isotropic spectra of the two compounds, where the quadrupolar broadening is removed, i.e., each maximum peak corresponds to a given oxygen environment. The resonance at 90 ppm, marked with an asterisk, corresponds to a spinning sideband of site A on the isotropic dimension. The two spectra show two groups of resonances, around 180–250 ppm and around 50–130 ppm (see the MAS projections). The region around 200 ppm (resonance A) is assigned to O(Bi,M)₄ tetrahedral sites in the polycationic ribbons. The value of the chemical shifts are close to those determined for OBi₄ tetrahedra in related compounds, 195 ppm in Bi₂O₃²² and 265 ppm in α-Bi₄V₂O₁₁.²³ Our assignment was indirectly confirmed by a ¹⁷OT₂ relaxation measurement (using a saturation recovery pulse sequence) performed on the BiCd₂PO₆ compound. The A site exhibited a short T₂ relaxation time of about 200 μs, maybe due to the presence of Bi quadrupolar nuclei in its first-neighbour cationic shell (¹¹¹Cd and ¹¹³Cd are non-quadrupolar). A similar measurement was not possible for the disordered compound due to the low efficiency of the isotopic enrichment (probably because the ¹⁷O-enriched water had already been used in previous experiments).

Fig. 3 ¹⁷O MQ-MAS NMR (18.8 T) spectra of (a) BiCd₂PO₆ and (b) Bi_1.2Zn_1.2PO_5.5. The spectra were acquired at an MAS speed of 20 kHz, with a recycling delay of 1 s, using the SPAM sequence.²⁴ The excitation and reconversion pulses were set to 3.75 μs and 1.20 μs, respectively, corresponding to an RF field strength of 80 kHz, followed by a selective 90° pulse of 11 μs (RF field of 8 kHz). For spectrum (a), each transient was accumulated with 72 scans, and 128 t1 data points were collected using the STATES method. For (b), a total of 4500 scans were needed and 30 t1 increments were collected. The ¹⁷O chemical shift was referenced externally to tap water.

Ordered compound: The second region around 100–160 ppm is typically in the chemical shift range of oxygens involved in PO₄groups.²⁵ In BiCd₂PO₆, it is composed of two resonances, B (120 ppm, T₂ = 5 ms) and C (100 ppm, T₂ = 500 μs), corresponding to O2 and O3. This assignment arises from their proximity or otherwise to quadrupolar Bi nuclei in their second cationic shell (Table 1). It was checked by a ³¹P–¹⁷O heteronuclear multiple quantum correlation (HMQC)²⁶ experiment that a correlation existed between the unique ³¹P site and the ¹⁷O–B sites (Fig. 4). However, no correlation signal was detected for the ¹⁷O–C sites due to the very short T₂ relaxation time (500 μs). It is also noteworthy that both A and C showed broad isotropic resonances compared to B. So far, this is not understood in this “ordered” compound. Note the presence of a broad signal of low intensity in the ¹⁷O dimension (Fig. 4), which is due to an impurity obtained after the process of enrichment and was not detected by XRD.

Table 1 The environment of the oxygen atoms (distances in Å) in BiCd₂PO₆ and Bi_1.2Zn_1.2PO_5.5. The first shell is given for the oxygen of the ribbons and the first two shells are presented for the oxygen of the PO₄groups.^15,16

	BiCd2PO6		Bi1.2Zn1.2PO5.5^a
	1st shell	2nd shell	1st shell
a The coordination of the disordered PO4groups is not accurately known.
O1–Bi1	2 × 2.27(2)		O2–Bi1	2 × 2.243(1)
Cd2	2 × 2.18(2)		Bi/Zn2	2 × 2.159(9)
O2–P	1 × 1.43(3)		O1–Bi1	4 × 2.299(1)
Cd2	2 × 2.22(2)	2 × 3.41(3)
Bi1		1 × 3.70(3)
O3–P	1 × 1.51(4)
Cd2	1 × 2.11(4)	2 × 3.29(2)
Bi1		2 × 3.41(2)
Bi1		1 × 3.65(4)

---

Fig. 4 The ³¹P–¹⁷O HMQC (18.8 T) spectrum of BiCd₂PO₆ was obtained at a MAS speed of 25 kHz by following the pulse sequence detailed by Massiot et al.²⁶ An echo was applied to the observed ¹⁷O nuclei with respective 90° and 180° pulses of 10 and 20 μs. Two 90° pulses of 4.5 μs were then applied on either sides of the ¹⁷O 180° pulse. The evolution delay was set to 3 ms for an evolution under J-coupling. A total of 512 scans were accumulated with a recycling delay of 1 s.

Semi-ordered compound: Next, we analyzed a semi-ordered compound, Bi_1.2Zn_1.2PO_5.5. The ¹⁷O MQ-MAS NMR spectrum is shown in Fig. 2(b). Two isolated regions are highlighted in the 2D spectrum. The broadness of the peaks seen in the isotropic projection is a signature of the high disorder present in this system, as discussed in the first part of this work. The assignment of both regions was deduced by analogy with the ¹⁷O spectrum obtained for BiCd₂PO₆. One region corresponds to oxygen atoms linked to ribbons (resonance A) between 160 and 240 ppm. Referring to the structure presented in Fig. 1(b), two oxygen sites should be distinguishable in the isotropic dimension, whether they are at the center (O1Bi₄) or at the edge (O2Bi₂Zn₂) of the ribbon. A close look at the two-dimensional contours clearly suggests the presence of more than two sites, probably due to the high sensitivity of the ¹⁷O NMR chemical shift to the cationic environment, even at a semi-local scale (second shell cationic neighbours). The presence of additional oxygen sites is easily explained by the existence of a mixed Bi/Zn cationic site at the edge of the ribbon. The assignment of each individual resonance is not yet possible due to the current absence of a large ¹⁷O NMR chemical shift database for these systems. The development of such a database would require a series of model compounds to be isotopically enriched for further ¹⁷O MAS NMR analysis. The second option available is to profit from the recent development of first principles calculations of NMR parameters using periodic boundary conditions.²⁷ The latter approach, which is beyond the scope of this Letter, is definitely more realistic at present.

The chemical shift region centred on 50–150 ppm exhibits two main resonances in the isotropic projection. These can be assigned to the oxygen atoms in the PO₄groups. First, it should be noted that the ¹⁷O resonances are spread over a large chemical shift range, especially in the isotropic dimension when the second order quadrupolar broadening is removed. This large distribution of chemical shift values with respect to the so-called ordered compound, BiCd₂PO₆, reveals the important disorder associated with the PO₄groups in these systems. Nevertheless, some discontinuities associated with discrete chemical shift values appear in the isotropic dimension. This strongly suggests that some PO₄ positions are privileged, as observed from the single crystal X-ray diffraction data. For example, two major competing PO₄ positions have been located in Bi_1.2M_1.2PO_5.5, while residual electronic density on Fourier difference maps is reminiscent of a number of extra orientations. Again, the current ¹⁷O NMR chemical shift database is not sufficient to fully interpret the spectrum in the P–¹⁷O–M region.

This study highlights the informative data provided by ¹⁷O NMR due to its broad range of chemical shifts compared to the ³¹P nucleus. In disordered bismuth phosphates, ¹⁷O NMR clearly provides evidence of preferential PO₄ orientations, in good agreement with the semi-ordering deduced from diffraction data, which shows modulated microdomains with loss of the order at long range scales. Of course, the methodology and conclusions developed here are not restricted to our particular chemical system, but can be generalized to other partially-ordered systems, solid solutions, composite structures and so on. Finally, it is worth stating that, to the best of our knowledge, a successful ³¹P–¹⁷O through-bond correlation has been presented here for the first time.

Experimental section

The different oxides reported in this Letter were prepared from stoichiometric mixtures of Bi₂O₃, MO (M = Cd, Zn) and (NH₄)₂HPO₄. To avoid the problem of volatile species, they were removed by solid state reparative methods, implying several heating–regrinding steps at temperatures from 200 to 700 °C. The purity of the samples was checked by powder X-ray diffraction using a Siemens D-5000 diffractometer with back-monochromatized Cu-K_α radiation. NMR experimental information is given in the Figure captions. ¹⁷O enrichment was achieved by heating samples at 650 °C for 8 h under ¹⁷O-enriched water vapor.¹⁵

Acknowledgements

The FEDER Région Nord Pas-de-Calais, Ministère de l’Education Nationale, de l’Enseignement Supérieur et de la Recherche, CNRS, USTL and ENSC-Lille are acknowledged for funding the NMR spectrometers. M. C. thanks the Région Nord Pas-de-Calais for financial support.

References

1. M. Huvé, M. Colmont and O. Mentré, Chem. Mater., 2004, 16, 2628.
2. M. Ketatni, O. Mentré, F. Abraham, F. Kzaiber and B. Mernari, J. Solid State Chem., 1998, 139, 274.
3. J. Huang, Q. Gu and A. W. Sleight, J. Solid State Chem., 1993, 105, 599.
4. F. Abraham, M. Ketatni, G. Mairesse and B. Mernari, Eur. J. Solid State Inorg. Chem., 1994, 31, 313.
5. M. Colmont, M. Huvé and O. Mentré, Inorg. Chem., 2006, 45(17), 6612.
6. M. Huvé, M. Colmont and O. Mentré, Inorg. Chem., 2006, 45(17), 6604.
7. F. Abraham and M. Ketatni, Eur. J. Solid State Inorg. Chem., 1995, 32, 429.
8. M. Ketatni, F. Abraham and O. Mentré, Solid State Sci., 1999, 1, 449.
9. F. Abraham, O. Cousin, O. Mentré and M. Ketatni, J. Solid State Chem., 2002, 167, 168.
10. X. Xun, S. Uma, A. Yokochi and A. W. Sleight, J. Solid State Chem., 2002, 167, 245.
11. A. Mizrahi, J. P. Wignacourt and H. Steinfink, J. Solid State Chem., 1997, 133, 516.
12. M. Ketatni, M. Huvé, F. Abraham and O. Mentré, J. Solid State Chem., 2002, 172, 327.
13. M. Colmont, M. Huvé, M. Ketatni, F. Abraham and O. Mentré, J. Solid State Chem., 2003, 176, 221.
14. M. Colmont, M. Huvé, F. Abraham and O. Mentré, J. Solid State Chem., 2004, 177, 4149.
15. M. Colmont, L. Delevoye, E. M. Ketatni, L. Montagne and O. Mentré, J. Solid State Chem., 2006, 179, 2111.
16. E. M. Ketatni, B. Mernari, F. Abraham and O. Mentré, J. Solid State Chem., 2000, 153, 48.
17. A. Flambard, L. Montagne and L. Delevoye, Chem. Commun., 2006, 28, 3426.
18. (a) A. Samoson, E. Lippmaa and A. Pines, Mol. Phys., 1998, 65, 1013; (b) A. Llor and J. Virlet, Chem. Phys. Lett., 1998, 152, 248.
19. L. Frydman and J. S. Harwood, J. Am. Chem. Soc., 1995, 117, 5367.
20. Z. Gan, J. Am. Chem. Soc., 2000, 122, 3242.
21. M. Feike, D. E. Demco, R. Graf, J. Gottwald, S. Hafner and H. W. Spiess, J. Magn. Reson., Ser. A, 1996, 122, 214.
22. S. Yang, K. D. Park and E. Oldfield, J. Am. Chem. Soc., 1989, 111, 7278.
23. N. Kim and P. Grey, Science, 2002, 297, 1317.
24. J.-P. Amoureux, L. Delevoye, S. Steuernagel, Z. Gan, S. Ganapathy and L. Montagne, J. Magn. Reson., 2005, 172, 268.
25. M. Zeyer, L. Montagne, V. Kostoj, G. Palavit, D. Prochnow and C. Jaeger, J. Non-Cryst. Solids, 2002, 311, 223.
26. D. Massiot, F. Fayon, B. Alonso, J. Trébosc and J.-P. Amoureux, J. Magn. Reson., 2003, 164, 160.
27. C. J. Pickard and F. Mauri, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 2451.

---