NMR spectroscopy provides an element-specific, sensitive probe of the local environment, enabling detailed information to be extracted. However, in the solid state the vast majority of this information remains unexploited, owing to the challenges associated with obtaining high-resolution spectra and the ease with which these can be interpreted. Recent advances enabling accurate and efficient calculation of NMR parameters in periodic systems have revolutionized the application of such approaches in solid-state NMR spectroscopy, particularly among experimentalists. As NMR is sensitive to the atomic-scale environment, it provides a potentially useful tool for studying disordered materials, and the combination of experiment with first-principles calculations offers a particularly attractive approach. There are, however, significant experimental and computational challenges in the application of NMR spectroscopy to disordered materials. For example, there is no longer a single arrangement of atoms in a simple model structure that matches the material under study, and many different atomic arrangements may be required to gain insight into the interpretation and assignment of NMR spectra, and ultimately, into the structure of the material under study.

The crystal chemical flexibility of pyrochlore-based (A₂B₂O₇) oxide materials has resulted in a wide range of applications, including energy materials, nuclear waste encapsulation and catalysis. There is, therefore, considerable interest in understanding the structure–property relationships in these materials, i.e., investigating how cation/anion disorder and local structure vary with composition. Here we combine a range of ⁸⁹Y, ¹¹⁹Sn and ¹⁷O NMR experiments with periodic planewave calculations to explore cation disorder in ¹⁷O-enriched (Y,La)₂(Sn,Ti,Zr,Hf)₂O₇ phases. We show how a variety of NMR crystallographic approaches from cluster-based approaches, to ensemble-based modelling and the use of grand canonical ensembles can provide insight into the atomic-scale structure and disorder in these materials.[1-6]

[1] Reader, S. W., Mitchell, M. R., Johnston, K. E., Pickard, C. J., Whittle, K. R., & Ashbrook, S. E. (2009). J. Phys. Chem C 113, 18874.

[2] Mitchell, M. R., Reader, S. W., Johnston, K. E., Pickard, C. J., Whittle, K. R., & Ashbrook, S. E. (2011). Phys. Chem Chem. Phys. 13, 488.

[3] Mitchell, M. R., Carnevale, D., Orr, R., Whittle, K. R., & Ashbrook, S. E. (2012). J. Phys. Chem . 116, 427.

[4] Fernandes, A., Moran, R. F., Sneddon, S., Dawson, D. M., McKay, D., Bignami, G. P. M., Blanc, F., Whittle K. R. & Ashbrook, S. E. (2018). RSC Advances 8, 7089.

[5] Moran, R. F., McKay, D., Tornstrom, P. C., Aziz, A., Fernandes, A., Grau-Crespo R., & Ashbrook, S. E. (2020). J. Am. Chem. Soc. 141, 17838.

[6] Fernandes, A., Moran, R. F., McKay, D., Griffiths, B., Herlihy, A., Whittle K. R., Dawson, D. M. & Ashbrook, S. E. (2020). Magn. Reson. Chem. in press, DOI: 10.1002/mrc.5140.

Structure elucidation of amorphous molecular solids presents one of the key challenges in chemistry today. Knowledge of the atomic-level structure in such materials would be of great value for example to direct the optimisation of pharmaceutical formulations. However, the lack of long-range structural order prevents atomic-level characterization of these materials using methods like single crystal X-Ray diffraction. Solid state NMR on the other hand yields atomic-level information on amorphous materials and molecular dynamics (MD) can generate candidate sets of possible disordered structures. Directly relating these two techniques has been out of reach so far, due to the prohibitive cost of computing chemical shifts on large ensembles of large MD structures using DFT. Recently, a method based on machine learning, dubbed ShiftML (1), has emerged as a quick and accurate way to predict the chemical shifts of organic solids, even for large systems.

Here, using a machine learning model of chemical shifts, we determine the complete atomic-level structure of the amorphous form of a drug molecule by combining dynamic nuclear polarization-enhanced solid-state NMR experiments with predicted chemical shifts for MD simulations of large systems (2). From these amorphous structures we then identify H-bonding motifs and relate them to local intermolecular interaction energies.

1. Paruzzo, F. M.; Hofstetter, A.; Musil, F.; De, S.; Ceriotti, M.; Emsley, L., “Chemical shifts in molecular solids by machine learning.” Nat Commun 2018, 9, 4501. doi.org/10.1038/s41467-018-06972-x

2. Cordova, Balodis, Hofstetter, Paruzzo, Nilsson Lill, Eriksson, Berruyer, Simões de Almeida, Quayle, Norberg, Svensk Ankarberg, Schantz, Emsley, "Structure determination of an amorphous drug through large-scale NMR predictions." Nat Commun 2021, doi.org/10.1038/s41467-021-23208-7

Complex structures with subtle atomic-scale details are now routinely solved using complementary tools such as X-ray and/or neutron scattering combined with electron diffraction and imaging. Identifying unambiguous atomic models for oxyfluorides, needed for materials design and structure−property control, is often still a considerable challenge despite the advantageous optical responses, magnetic properties, and energy storage capability of numerous oxyfluorides. Amongst the long-stranding challenges are the lack of tools to resolve fluorine and oxygen and to characterize fluoride-ion and MF_n polyhedral dynamics. In this work, NMR crystallography is used in combination with single-crystal X-ray diffraction, X-ray absorption spectroscopy, and property measurements to provide a comprehensive structural picture of a series of new oxyfluoride materials and highlight the presence of previously unidentified selective fluorine-mediated dynamics.

This talk will focus on insights from ¹⁹F NMR across early transition metal oxyfluoride materials including newly discovered hafnium oxyfluorides, spin singlet Mo(IV) cluster compounds, and emerging hybrid organic–inorganic low-dimensional compounds. The first system has relevance to fluoride-doped HfO₂ electronic materials [1]; the second example features a rare triangular metal oxyfluoride cluster, [Mo₃O₄F₉]⁵⁻ (Fig. 1) [2]; and the third series of compounds are structurally diverse and provide fundamental insights into competition between centrosymmetric and noncentrosymmetric crystallization [3]. Identifying the anion (dis)order is central to building design rules for noncentrosymmetric crystals with technologically relevant properties. 1D and 2D solid-state ¹⁹F NMR experiments are supported by ab initio calculations to shed light on the anion sublattice and to assign the numerous distinct fluorine environments. In compounds with ⁹³Nb and ⁵¹V, coupling between the metal and fluorine nuclei can be used to further aid the interpretation. Variable-temperature measurements reveal fluorine dynamics that are strongly correlated to polyhedral degrees of freedom. The dual scattering and spectroscopy approach is used to demonstrate the sensitivity of ¹⁹F shielding to small changes in bond length, on the order of 0.01 Å, even in the presence of hydrogen bonding, metal−metal bonding, and electrostatic interactions.

[1] Flynn, S., Zhang, C., Griffith, K. J., Shen, J., Wolverton, C., Dravid, V. P., Poeppelmeier, K. R. (2021). Inorg. Chem. 60, 4463.

[2] Ding, F., Griffith, K. J., Koçer, C. P., Saballos, R., Wang, Y., Zhang, C., Nisbet, M., Morris, A. J., Rondinelli, J. M., Poeppelmeier, K. R. (2020). J. Am. Chem. Soc. 142, 12288.

[3] Nisbet, M. L., Pendleton, I. M., Nolis, G. M., Griffith, K. J., Schrier, J., Cabana, J., Norquist, A. J., Poeppelmeier, K. R. (2020). J. Am. Chem. Soc. 142, 7555.

Three-dimensional electron diffraction crystallography (microED) can solve structures of sub-micrometer crystals, which are too small for single crystal X-ray crystallography. However, R factors for the microED-based structures are generally high (15-30% for small molecules) because of dynamic scattering. Thus, R factor may not be reliable provided that kinetic analysis is used. Consequently, there remains 1) ambiguity to locate hydrogens and 2) assignment of nuclei with close atomic numbers, like carbon, nitrogen, and oxygen. On the other hand, ¹H solid-state NMR is readily available using fast MAS probes and ¹³C, ¹⁴N, ¹⁵N, and ¹⁷O are completely different nuclei for NMR observation. Thus, information from solid-state NMR and microED is complementary.

Herein, we demonstrate combined approach using solid-state NMR and microED to solve crystalline structure. First, well established NMR crystallography approach is employed. Isotropic chemical shifts are very sensitive to local environment, thus crystalline structure, however, there are no intuitive way to predict chemical shifts from structures. GIPAW procedure paves a way to estimates chemical shifts of crystalline materials in a very high accuracy. This makes isotropic chemical shift as a reliable measure for structure validation. While wrong position of ¹H and misassignment of carbon/nitrogen/hydrogen result in poor agreement between experimental and calculated chemical shifts, right structure can be easily chosen among candidates. We show that this approach, which is well established with XRD structures, equally works well with microED. The crystalline structures are determined by microED and validated by isotropic chemical shifts [1]. To further validate the structure, next, we demonstrate another measure using dipolar-based ssNMR experiments in addition to isotropic chemical shifts. In principle, dipolar couplings provide useful information for structure elucidation, as the size of coupling is inversely proportional to the cube of internuclear distances. However, spin dynamics is often complicated due to presence of multiples of intra- and intermolecular couplings for small molecules, making structure elucidation difficult. On the other hand, it is readily calculated for given structures if the spin system is simple enough (< 8 spins). Here we utilize ¹H-¹H selective recoupling of proton (SERP) experiments [2-4] and ¹H-¹⁴N phase-modulated rotational-echo saturation-pulse double-resonance (PM-RESPDOR) as dipolar-based NMR experiments [5, 6]. While the former selects a subset of ¹H-¹H spin systems, the latter simplifies the spin system by decoupling ¹H-¹H interactions. As a result, SERP and RESPDOR probe ¹H-¹H and ¹H-¹⁴N networks, respectively. The structure is solved by microED and then validated by evaluating the agreement between experimental and calculated dipolar-based NMR results [7]. As the measurements are performed on ¹H and ¹⁴N, the method can be employed for natural abundance samples. Furthermore, the whole validation procedure was conducted at 293 K unlike widely used chemical shift calculation at 0 K using the GIPAW method.

[1] C. Guzmán-Afonso, Y.-l. Hong, H. Colaux, H. Iijima, A. Saitow, T. Fukumura, Y. Aoyama, S. Motoki, T. Oikawa, T. Yamazaki, K. Yonekura, Y. Nishiyama*, Nat. Commun. 10, 3537 (2019).

[2] N.T. Duong, S. Raran-Kurussi, Y. Nishiyama*, V. Agarwal*, J. Phys. Chem. Lett. 9, 5948-5954 (2018).

[3] N.T. Duong, S. Raran-Kurussi, Y. Nishiyama*, V. Agarwal*, J. Magn. Reson. 317, 106777 (2020).

[4] L.R. Potnuru†, N.T. Duong†, S. Ahlawat, S. Raran-Kurussi, M. Ernst, Y. Nishiyama* and V. Agarwal*, J. Chem. Phys. 153, 084202 (2020).

[5] N.T. Duong, F. Rossi, M. Makrinich, A. Goldbourt, M.R. Chierotti, R. Gobetto, Y. Nishiyama*, J. Magn. Reson. 308, 106559 (2019)

[6] N.T. Duong, Z. Gan, Y. Nishiyama*, Front. Mol. Biosci. 8, 645347 (2021).

[7] N.T. Duong, Y. Aoyama, K. Kawamoto, T. Yamazaki, Y. Nishiyama, under review.

The determination of active site protonation states is critical to gaining a full mechanistic understanding of enzymatic transformations; yet proton positions are challenging to extract using the standard tools of structural biology. Here we make use of a joint solid-state NMR, X-ray crystallography, and first-principles computational approach that unlocks the investigation of enzyme catalytic mechanism at this fine level of chemical detail. Through this process, we are developing a high-resolution probe for structural biology that is keenly sensitive to proton positions – rivaling that of neutron diffraction, yet able to be applied under conditions of active catalysis to microcrystalline and non-crystalline materials. For tryptophan synthase, this allows us to peer along the reaction coordinates into and out of the α-aminoacrylate intermediate. By uniquely identifying the protonation states of ionizable sites on the cofactor, substrates, and catalytic side chains, as well as the location and orientation of structural waters in the active site, a remarkably clear picture of structure and reactivity emerges. Most incredibly, this intermediate appears to be mere tenths of angstroms away from the preceding transition state in which the β-hydroxyl of the serine substrate is lost. The position and orientation of the structural water immediately adjacent to the substrate β-carbon suggests not only the fate of the hydroxyl group, but also the pathway back to the transition state and the identity of the active site acid-base catalytic residue. Reaction of this intermediate with benzimidazole (BZI), an isostere of the natural substrate, indole, shows BZI bound in the active site and poised for, but unable to initiate, the subsequent bond formation step. When modeled into the BZI position, indole is positioned with C-3 in contact with the α-aminoacrylate Cβ and aligned for nucleophilic attack.

Structural analysis by XRD still remains a considerable challenge for materials that can’t be isolated as single crystals. In NMR crystallography structural constraints are extracted from the modern solid-state NMR techniques, and along with DFT (density functional theory) calculations.[1-4] NMR crystallography has been used to derive de novo structures and to aid the refinement of X-ray powder diffraction data.[1-4] In this work, computational integration of advanced solid-state NMR with PXRD (powder X-ray diffraction) and modelling is used to understand the structure of metal coordination polymers that are produced by the etching of metal nanoparticles in acidic solution.[6-8] Notably, these coordination polymers have some structural disorder which gives rise to broadened diffraction peaks. Solid-state NMR was applied to determine the number of molecules in the asymmetric unit and give insight into the geometry at the metal center. Then, the PXRD pattern of the coordination polymers was partially indexed to find probable unit cells. The position of heavy atoms was then optimized within the unit cell using the Free Objects for Crystallography (FOX) software. Finally, Rietveld refinement and DFT optimization was used to obtain a final structural model. The final NMR and PXRD derived structure is validated by comparing the experimental and simulated PXRD pattern and NMR parameters. This protocol was verified on scandium acetate which has a known single crystal structure from the literature. The protocol was then successfully applied to microcrystalline gallium and aluminum coordination polymers. We anticipate that this methodology could be extended to similar kind of coordination polymers with inherent heterogeneous character.

Figure 1. Schematic representation of the NMR crystallography approach used for finding the crystal structure of the coordination polymer.

[1] Ashbrook, S. E. & McKay, D. (2016). Chem. Commun. 52, 7186–7204.

[2] Bouchevreau, B., Martineau, C., Mellot-Draznieks, C., Tuel, A., Suchomel, M. R., Trébosc, J., Lafon, O., Amoureux, J.-P. & Taulelle, F. (2013). Chem. – A Eur. J. 19, 5009–5013.

[3] Thomas, B., Rombouts, J., Oostergetel, G. T., Gupta, K. B. S. S., Buda, F., Lammertsma, K., Orru, R. & de Groot, H. J. M. (2017). Chem. – A Eur. J. 23, 3280–3284.

[4] Xu, Y., Southern, S. A., Szell, P. M. J. & Bryce, D. L. (2016). CrystEngComm. 18, 5236–5252.

[6] Rossini, A. J., Hildebrand, M. P., Hazendonk, P. A. & Schurko, R. W. (2014). J. Phys. Chem. C. 118, 22649–22662.

[7]Chang, B., Martin, A., Thomas, B., Li, A., Dorn, R., Gong, J., Rossini, A. & Thuo, M. ACS Mater. Lett. 2, 1211–1217.

[8] Chang, B. S., Thomas, B., Chen, J., Tevis, I. D., Karanja, P., Çınar, S., Venkatesh, A., Rossini, A. J. & Thuo, M. M. (2019). Nanoscale. 11, 14060–14069.

For over 60 years, nanomaterials have consistently attracted the attention of the scientific community. In the field of nanomedicine, recent effort toward optimizing the therapeutic efficacy of newly discovered active compounds has resulted in the development of original supramolecular systems that execute multiple functions. However, the true potential of these systems has not been entirely utilized. Advancing these materials calls for precise structural analysis of individual elements and a description of the mutual relations between them. This is a stringent requirement, as these systems exist at the borderline between crystalline and amorphous solids, for which high-quality diffraction data are inherently unavailable. This contribution thus addresses our attempt to formulate an efficient experimental-computational strategy for obtaining deep insight into the structure of complex polycrystalline composites with micro- and nanodomain architecture. To determine the atomic-resolution structure of these systems, we apply a procedure based on ¹H NMR crystallography extended to describe the component-selective data. This strategy is based on the combined application of domain-selective solid-state NMR spectroscopy (ss-NMR), crystal structure prediction (CSP), and density functional theory (DFT)-based calculations of NMR chemical shifts. This combination of experimental and theoretical approaches enables one to determine the structural arrangements of molecules in situations which are not tractable by conventional spectroscopic techniques. Its applications should be of particular importance for systems in which phase transformations can occur, and new polymorphic forms can be spontaneously created under the influence of the matrix environment. The potential of this combined analytical approach is highlighted using the recently developed biodegradable, injectable polyanhydride microbead formulation of decitabine (5-aza-2'-deoxycytidine, DAC), an archetypal DNA methyltransferase inhibitor used as an efficient therapeutic for epigenetic cancer therapy. In this innovative drug-delivery formulation, which was developed to circumvent the problem of hydrolytic lability of the active compound, a mixture of microcrystalline domains of decitabine and nanodomains of sebacic acid (SA) is embedded in the semicrystalline matrix of poly(sebacic acid-co-1,4-cyclohexane-dicarboxylic acid) (PSA-co-PCH) carrier. The proposed method, which employs the confluence of computational data with measured NMR parameters, thus provides for a way to distinguish between alternative candidate structures exclusively existing in the composite assembles, and to select the ones that are the most compatible with available information. As the obtained results also opened a route toward the structure refinement of synthetic polymers with a limited amount of spectroscopic data available, finding a procedure for the reliable generation of a representative set of CSPs of synthetic polymers is thus of paramount importance. This contribution thus demonstrates the synergy effects of the proposed combination of several experimental and computational procedures, which considerably extends the NMR crystallography approach into the area of intricate mixtures and nanostructured composites. Potentiality of this approach will be also highlighted in de-nuovo determination of the crystal structure of chemotactic N-formyl-L-Met-L-Leu-L-Phe-OH tripeptide.

NMR crystallography uses a combination of solid-state NMR (SSNMR), X-ray diffraction (XRD), and quantum chemical calculations for the determination and/or refinement of crystal structures.^1,2 Currently, the majority of NMR crystallographic studies reply upon the measurement and calculation of isotropic chemical shifts (CS) or CS tensors and their computation by density functional theory (DFT) methods; by contrast, electric field gradient (EFG) tensors of quadrupolar nuclei are used much less extensively.^3–6 EFG tensors at the origins of quadrupolar nuclei are sensitive to their surrounding electronic environments, including longer-range interactions that do not greatly influence chemical shifts, yielding unique sets of quadrupolar parameters for each magnetically distinct environment. Furthermore, EFG tensors are less computationally demanding to compute than CS tensors. Since EFG tensors are sensitive probes of local atomic environments, we believe that it is crucial to explore and develop quadrupolar NMR-based crystal structure prediction (CSP) methods within the context of modern plane-wave DFT computational packages. In particular, such methods would be very useful for the study and characterization of organic solids, including pharmaceutical drug products, nutraceuticals, and a wide variety of multi-component crystals.⁵

Herein, we demonstrate the use of experimentally-measured and computationally-derived ³⁵Cl EFG tensor parameters in a new NMR crystallographic protocol for the refinement of crystal structures, which are developed and optimized on a training set of four organic HCl salts with known crystal structures. The stages of this protocol include: (i) selection/assignments of molecular fragments, charges, motion groups, and potential unit cells; (ii) simulated annealing using the Polymorph software package to generate tens of thousands of candidate structures, (iii) coarse geometry optimizations using DFT-D2* methods (which include dispersion effects),^7–9 and (iv) subsequent fine geometry optimizations. Between each of these stages, filters involving the unit cell dimensions, EFG tensor parameters, and static lattice energies have been optimized to select for the best candidate structures. The robustness of this new protocol is demonstrated via comparison of EFG tensors, PXRD patterns, and overlays of the known and refined crystal structures. Finally, this new protocol is demonstrated in several blind tests for the structural determination and refinement of organic HCl salts with unknown structures.

(1) Taulelle, F. Encycl. Magn. Reson. 2009, 1–14.
(2) NMR Crystallography; Harris, R., Wasylishen, R., Duer, M., Eds.; John Wiley & Sons Ltd.: Chichester, U.K., 2009.
(3) Ashbrook, S. E.; McKay, D. Chem. Commun. 2016, 52, 7186–7204.
(4) Bryce, D. L. IUCrJ 2017, 4, 350–359.
(5) Hodgkinson, P. Prog. Nucl. Magn. Reson. Spectrosc. 2020, 118–119, 10–53.
(6) Widdifield, C. M.; Farrell, J. D.; Cole, J. C.; Howard, J. A. K.; Hodgkinson, P. Chem. Sci. 2020, 11, 2987–2992.
(7) Grimme, S. J. Comput. Chem. 2006, 27, 1787–1799.
(8) Holmes, S. T.; Vojvodin, C. S.; Schurko, R. W. J. Phys. Chem. A 2020, 124, 10312–10323.
(9) Holmes, S. T.; Schurko, R. W. J. Phys. Chem. C 2018, 122, 1809–1820.