35 Cl – 1 H Heteronuclear correlation magic-angle spinning nuclear magnetic resonance experiments for probing pharmaceutical salts

Heteronuclear multiple-quantum coherence (HMQC) pulse sequences for esta-blishing heteronuclear correlation in solid-state nuclear magnetic resonance (NMR) between 35 Cl and 1 H nuclei in chloride salts under fast (60 kHz) magic-angle spinning (MAS) and at high magnetic field (a 1 H Larmor frequency of 850 MHz) are investigated. Specifically, recoupling of the 35 Cl – 1 H dipolar interaction using rotary resonance recoupling with phase inversion every rotor period or the symmetry-based SR4 21 pulse sequences are compared. In our implementation of the population transfer (PT) dipolar (D) HMQC experiment, the satellite transitions of the 35 Cl nuclei are saturated with an off-resonance WURST sweep, at a low nutation frequency, over the second spinning sideband, whereby the WURST pulse must be of the same duration as the recoupling time. Numerical simulations of the 35 Cl – 1 H MAS D-HMQC experiment performed separately for each crystallite orientation in a powder provide insight into the orientation dependence of changes in the second-order quadrupolar-broadened 35 Cl MAS NMR lineshape under the application of dipolar recoupling. Two-dimensional 35 Cl – 1 H PT-D-HMQC MAS NMR spectra are presented for the amino acids glycine

dipolar interaction using rotary resonance recoupling with phase inversion every rotor period or the symmetry-based SR4 2 1 pulse sequences are compared. In our implementation of the population transfer (PT) dipolar (D) HMQC experiment, the satellite transitions of the 35 Cl nuclei are saturated with an offresonance WURST sweep, at a low nutation frequency, over the second spinning sideband, whereby the WURST pulse must be of the same duration as the recoupling time. Numerical simulations of the 35 Cl-1 H MAS D-HMQC experiment performed separately for each crystallite orientation in a powder provide insight into the orientation dependence of changes in the second-order quadrupolar-broadened 35 Cl MAS NMR lineshape under the application of dipolar recoupling. Two-dimensional 35 Cl-1 H PT-D-HMQC MAS NMR spectra are presented for the amino acids glycineÁHCl and L-tyrosineÁHCl and the pharmaceuticals cimetidineÁHCl, amitriptylineÁHCl and lidocaineÁHClÁH 2 O. Experimentally observed 35 Cl lineshapes are compared with those simulated for 35 Cl chemical shift and quadrupolar parameters as calculated using the gauge-including projector-augmented wave (GIPAW) method: the calculated quadrupolar product (P Q ) values exceed those measured experimentally by a factor of between 1.3 and 1.9.

| INTRODUCTION
Chloride salts have wide application in the isolation and purification of active pharmaceutical ingredients (APIs). The use of salts provides higher concentration in solution, and they readily undergo crystallisation. Another advantage for using salts is that they provide more stability as preservatives, acting as an antimicrobial component designed to destroy or inhibit the growth of bacteria, yeast or moulds and extend the shelf life of medicine. [1] Last but not least the use of salts increases the bioavailability of the APIs by increasing their solubility and/or enhancing their permeability across membranes. [2] The relevance of chlorine when preparing pharmaceuticals is evident considering that a large proportion (up to about 50%) of APIs are formulated as HCl salts. [3] Introduction of a chloride ion is associated with protonation of a specific site in the API so as to achieve charge balance; this usually affects the position of the hydrogen atoms, affecting the intermolecular interactions, the supramolecular assembly and thus the solubility, stability, bioavailability and biological activity [4] ; therefore, high-resolution structural characterisation of the API and polymorph identification [5] is required.
As well as complementary 1 H, 13 C, 14/15 N solid-state NMR experiments, [6] it is clearly beneficial to directly probe the chlorine nuclei. The 35 Cl nucleus has 75% natural abundance and is a quadrupolar nucleus, with a nuclear spin I = 3/2. The quadrupolar interaction provides information about symmetry and the local environment of such nuclei, therefore contributing to the structural characterization endeavour with complementary information, hence the continuous effort to develop NMR methodology to obtain structural information from 35 Cl NMR. [7][8][9][10][11] In many organic compounds where it is covalently bound, the 35 Cl nucleus experiences a very large quadrupolar interaction not amenable to MAS experiments. [12] However, in hydrochloride salts of, for example, amino acids and pharmaceuticals, the chloride ions are ionically bonded; therefore, the quadrupolar interaction is reduced due to increased symmetry, and the 35 Cl MAS NMR signal can be measured at high magnetic fields and/or under very fast MAS. [5,[13][14][15][16][17][18] When dealing with half-integer quadrupolar nuclei, it is usually beneficial to saturate the satellite transitions and thus increase the polarisation of the central transition before the 90 excitation pulse. [19][20][21][22][23] This can be achieved using a fast 180 phase alternating pulse train that induces rotor-assisted population transfer (RAPT), [19] or a double frequency sweep (DFS) achieved by a time dependent amplitude modulated pulse that sweeps over the satellite transitions, [20] or a hyperbolic secant π inversion pulse, [21,24] or a WURST shaped pulse, [22] or repetitive sideband-selective DFS, [25,26] or using quadruple frequency sweeps. [27] An analysis of all orientations in a powder shows that such saturation can be uniformly achieved over all different orientations for the quadrupolar interaction in a powder, and therefore, no significant lineshape distortion is expected. [20] In this work, we are using an off-resonance WURST pulse sweeping over a spinning sideband, which provides a good saturation of the satellite transition and a significant signal enhancement in the 35 Cl MAS NMR spectra.
Accurate localization of chloride ions in pharmaceutical salts is important for structural characterization. In this context, it is possible to measure internuclear distances between 35 Cl and 13 C nuclei using a REAPDOR experiment as was demonstrated for 10% 13 C-enriched tyrosineÁHCl and natural abundance glycineÁHCl. [28] Proximities between 13 C and 35 Cl in histidine HCl monohydrate have also been probed using a DNP-enhanced D-HMQC experiment. [7] Further structural information can be obtained by investigating proximities between Cl and H atoms. With advances in fast MAS technology, involving the 1 H nucleus in solid-state NMR experiments becomes advantageous. Under fast MAS, not only the resolution in the 1 H MAS NMR spectra is enhanced, but also the spin-echo lifetimes become considerably longer opening up the possibility to investigate 1 H environments by recoupling the heteronuclear dipolar coupling between protons and another adjacent NMR active nucleus such as 14 N, 17 O or 35 Cl. [29][30][31][32] Here, we consider rotary resonance recoupling [33] with the RF phase inverted every rotor period [34] and symmetry-based recoupling [35][36][37] of the 35 Cl-1 H dipolar coupling.
For two-dimensional (2D) heteronuclear experiments considered in this paper, the nuclei are specified in the order of, first, the indirect and, second, the direct dimension, that is, F 1 -F 2 . The 14 N-1 H dipolar HMQC (D-HMQC) MAS experiment [38][39][40] has been successfully used to detect nitrogen-hydrogen proximities, [29] for example, to identify specific intermolecular hydrogen bonding interactions in an indomethacin-nicotinamide co-crystal, [41] a nicotinamide palmitic acid co-crystal and an acetaminophen-polyvinylpyrrolidone solid dispersion, [42] or to probe nitrogen protonation in cimetidine. [43] The spin dynamics of quadrupolar nuclei under the HMQC experiment has been recently described. [44,45] Specifically, Wang et al. [44] introduced the population transfer (PT)-D-HMQC experiment to probe 27 Al-31 P correlations where the satellite transitions of the 27 Al are saturated with a WURST irradiation during symmetry-based recoupling. In addition, frequency selective HMQC or RESPDOR [46,47] experiments that have been implemented for 14 N-1 H can also be applied in 35 Cl-1 H applications. Venkatesh et al. have presented a 35 Cl-1 H D-HMQC experiment that incorporates saturation of the satellite transitions using RAPT during the mixing time. [31] Venkatesh et al. have also investigated an alternative experiment, D-RINEPT, to obtain 35 Cl-1 H correlations. In this paper, we explore the applicability of the 35 Cl-1 H PT-D-HMQC experiment to probe proximities between hydrogen and chlorine atoms in the amino acids glycineÁHCl and L-tyrosineÁHCl and the pharmaceuticals cimetidineÁHCl, amitriptylineÁHCl and lidocaineÁHClÁH 2 O.

| EXPERIMENTAL AND COMPUTATIONAL DETAILS
L-TyrosineÁHCl, glycineÁHCl, lidocaineÁHClÁH 2 O, cimetidineÁHCl and amitriptylineÁHCl were purchased from Sigma-Aldrich and used without further purification. Experiments were performed on a Bruker Avance Neo spectrometer with a Larmor frequency of 850.2 and 83.3 MHz for 1 H and 35 Cl, respectively, using a 1.3-mm Bruker triple-resonance HXY probe operating in doubleresonance mode at 60 kHz MAS. The 35 Cl NMR spectra were referenced to a NaCl 0.1-M aqueous solution with the 35 Cl peak set to 0 ppm. [13] To convert to a chemical shift scale where the 35 Cl of solid NaCl is set to 0 ppm, it is necessary to add 46.34 ppm (Eq. 6 in Bryce et al. [13] ). The 1 H NMR spectra were referenced to the 1 H peak of adamantane set to 1.8 ppm. [32,48,49] A 1 H nutation frequency of 140 kHz was used for 1 H 90 and 180 pulses. 1 H one-pulse MAS NMR spectra were measured using the standard Bruker background suppression pulse sequence consisting of a 180 pulse followed by two 90 pulses. [50] The same recycle delay of 1 s was used for each sample, though it may be possible to optimise sensitivity by optimising the recycle delay to the sample's specific 1 H T 1 value.
Phase-inverted (i.e., x-x phase inversion each rotor period) R 3 [33,34,51,52] or SR4 2 1 [35][36][37]40] heteronuclear recoupling was applied on the 1 H channel at a nutation frequency of 120 kHz. WURST [22,23,53,54] saturation pulses at 5 kHz (central transition) nutation frequency were applied on the 35 Cl channel during the recoupling. In our implementation for the 35 Cl experiments in this paper, the WURST pulses are centred at an offset of 120 kHz, that is, the second spinning sideband and sweep over 40 kHz from low to high ppm. We observe that the success of the 35 Cl-1 H experiment requires the duration of the WURST saturation pulses to be the same as the duration of the SR4 2 1 recoupling as has previously been applied by Wang et al. [44] in a PT-D-HMQC 27 Al-31 P experiment. [45] The 35 Cl central transition was excited with a 2.1-μs pulse at a central transition nutation frequency of 120 kHz. The 2D D-HMQC spectra were acquired using States phase incrementation in the t 1 dimension. [55] 35 Cl MAS spectra were measured using the standard Bruker spin-echo pulse sequence preceded by a WURST saturation.
First-principle calculations were performed using the academic release version 16.1 of CASTEP, [56] using the Perdew-Burke-Ernzerhof (PBE) exchange correlation functional, [57] a plane-wave basis set with ultrasoft pseudopotentials and a plane-wave cut-off energy of 700 eV. Geometry optimization was performed starting with the respective X-ray structure as deposited at the Cambridge Structural Database. NMR parameters were calculated using the gauge-including projectoraugmented wave (GIPAW) [58] method with a reference chemical shielding of 30 and 962 ppm for 1 H and 35 Cl, respectively.

| Evaluation of 35 Cl-1 H D-HMQC MAS NMR pulse sequences
The 35 Cl-1 H D-HMQC solid-state NMR pulse sequences employed in this work are shown in Figure 1. SR4 2 1 [35][36][37]40] or phase-inverted R 3 [33,34,51,52] recoupling is applied on the 1 H channel to recouple the 35 Cl-1 H heteronuclear dipolar coupling. Note that, in both cases, the nutation frequency is set to be twice the spinning frequency such that the 35 Cl-1 H heteronuclear dipolar interaction is reintroduced, while removing the 1 H-1 H homonuclear dipolar interaction. On the 35 Cl channel, a WURST saturation pulse [22,23,53,54] sweeping over the F I G U R E 1 Pulse sequences for a 35 Cl-1 H dipolar heteronuclear multiple-quantum coherence (D-HMQC) magicangle spinning (MAS) nuclear magnetic resonance (NMR) experiment (a) without or (b) with a WURST saturation pulse applied to 35 Cl. Phase-inverted R 3 or SR4 2 1 is employed to recouple the 35 Cl-1 H dipolar interaction. The simultaneous application of satellite transition saturation, here using WURST, and heteronuclear recoupling is referred to as a population transfer (PT)-D-HMQC experiment. [44] The delays before and after the application of the first recoupling block and after the application of the second recoupling block are kept as short as possible second spinning sideband (with MAS at 60 kHz, the WURST offset is set to 120 kHz) saturates the satellite transition of the 35 Cl and increases the polarisation of the central transition. The pulse sequence in Figure 1a with SR4 2 1 recoupling has been demonstrated by Pandey et al. on L-tyrosineÁHCl, L-histidineÁHClÁH 2 O, procainamideÁHCl and aminoguanidineÁHCl. [30] Venkatesh et al. use a PT-D-HMQC pulse sequence similar to that shown in Figure 1b except that for saturation of the satellite transitions, they use RAPT, [19] whereas we found that at fast spinning frequency, a low-nutation frequency WURST shape pulse, sweeping a 40 kHz range over the second spinning side band, saturates the satellites transitions efficiently. Figure 2 presents 2D 35 Cl-1 H D-HMQC MAS NMR spectra of L-tyrosineÁHCl with SR4 2 1 heteronuclear recoupling recorded (a) without or (b) with the application of WURST saturation pulses to 35 [30] the spectrum obtained with the WURST saturation pulses applied to 35 Cl only shows 1 H peaks with the 35 Cl central transition.
Venkatesh et al. have previously observed such a suppression of the satellite transition peak using RAPT. [31] In the case that the 90 pulses on 35 Cl are selective on the 35 Cl central transition, the signal from the 35 Cl satellite transitions would also be reduced. [31,59] Figure 3 presents comparisons of summed rows from the 2D 35 Cl-1 H D-HMQC MAS NMR spectra of L-tyrosineÁHCl. Specifically, Figure 3a corresponds to a sum over only the 35 Cl central transition peaks. The application of SR4 2 1 recoupling with 35 Cl WURST satellite transition saturation gives 1.8 and 2.2 times more signal for the OH and NH 3 + peak, respectively. Similar gains have been reported for 35 Cl-1 H PT-D-HMQC MAS NMR spectra of L-histidine HCl. [31] This signal gain is due to transfer from the satellite transitions into the central transition: in Figure 3b, which corresponds to the sum over the satellite and central transitions, the signal intensity is very similar for SR4 2 1 recoupling with and without WURST satellite transition saturation. Note that for the same recoupling time, we observe greater signal intensity with SR4 2 1 as compared with phase-inverted R 3 recoupling. All subsequent 35 Cl-1 H D-HMQC MAS NMR spectra presented in the rest of this paper were recorded using the PT-D-HMQC pulse sequence in Figure 1b, that is, with SR4 2 1 recoupling and WURST saturation of the 35 Cl satellite transitions.

| Dependence on recoupling time and on quadrupolar orientation
This section explores the extent to which performing the PT-D-HMQC experiment with different recoupling times reveals quantitative information about 35 Cl-1 H distances. Figure 4a compares the signal intensity for the L-tyrosineÁHCl 1 H resonances (the displayed 1 H spectral width is from 1.0 to 11.2 ppm) for different SR4 2 1 recoupling times at 60 kHz MAS. For this experiment, both F I G U R E 2 (a) 35 Cl-1 H (850 MHz) dipolar heteronuclear multiple-quantum coherence (D-HMQC) magic-angle spinning (MAS) (60 kHz) nuclear magnetic resonance (NMR) spectra with skyline projections of L-tyrosineÁHCl recorded with τ 1 RCPL = τ 2 RCPL = 400 μs of SR4 2 1 heteronuclear recoupling (a) without or (b) with WURST saturation of the 35 Cl satellite transitions (see pulse sequence diagrams in Figure 1). One hundred twenty-eight transients were co-added for each of 100 rotor-synchronised t 1 free-induction decays (FIDs) with a 1 s recycle delay, corresponding to a total experimental time of 4 h. Positive and negative contours are shown in blue and green, respectively, with the base contour level at (a) 15% and (b) 12% of the maximum peak intensity. The furthest left vertical projection shows a spin-echo 35 Cl MAS NMR spectrum. A two-dimensional (2D) spectrum obtained with phase-inverted R 3 recoupling is shown in Figure S1 the τ 1 RCPL and τ 2 RCPL recoupling durations are increased from 100 μs to 1 ms. For both the NH 3 + 1 H resonance at 7.3 ppm and the OH 1 H resonance at 9.7 ppm, a clear maximum is observed at 400 μs (24 rotor periods). The experimental results when either τ 1 RCPL or τ 2 RCPL is increased independently are shown in Figure S2. In this F I G U R E 3 A comparison of summed rows from 35 Cl-1 H (850 MHz) dipolar heteronuclear multiple-quantum coherence (D-HMQC) (τ 1 RCPL = τ 2 RCPL = 400 μs) magic-angle spinning (MAS) (60 kHz) nuclear magnetic resonance (NMR) spectra of L-tyrosineÁHCl recorded with phase-inverted R 3 (green) or SR4 2 1 recoupling (pulse sequence in Figure 1a) (red) and with SR4 2 1 recoupling and WURST saturation of the 35 Cl satellite transitions (pulse sequence in Figure 1b) (blue). The sum of the F 2 1 H spectra over the 35 Cl  35 Cl (with C Q = 2.4 MHz and η = 0.72) and one 1 H, starting from I x coherence and monitoring evolution of the I y S z term, for a dipolar coupling (collinear principal axes systems, PASs) of 1.2 (circles), 1.0 (triangles) and 0.8 kHz (squares), corresponding to a Cl-H distance of 2.14, 2.27 and 2.45 Å, respectively case, evolution under the 1 H CSA distorts the recoupling build-up profile.
It is informative to compare the experimental results to simulations for a 35 Cl-1 H 2-spin system undergoing SR4 2 1 recoupling using the SIMPSON simulation package. [60] The PT-D-HMQC experiment aims to excite and detect the double quantum (DQ) coherence term between 1 H and 35 Cl spins; therefore, the spin dynamics of the PT-D-HMQC experiment can be described using product operator formalism in a similar way as presented by Mandal et al. [61] Therefore, we used the SIMPSON programme (code shown in the Supporting Information, Section S6) to calculate the evolution of the I y S z term under the R4 2 1 recoupling sequence when starting with I x . Simulated build-up for a dipolar coupling of 1.2, 1, and 0.8 kHz (corresponding to a Cl-H distance of 2.14, 2.27 and 2.45 Å, respectively) are shown in Figure 4b. Comparing the shape of the experimental build-up in Figure 4a for the L-tyrosine HCl OH 1 H resonance to the simulated build-up curves, there is good agreement for the initial build-up of the 1.2 kHz (2.1 Å) simulated data with the experimental data for the OH peak, whereas the experimental data for the NH 3 peak have maximum intensity that best matches the simulated data for 1.0 kHz. In this respect, for the crystal structure (CSD code LTYRHC10 [62] ), after density functional theory (DFT) (CASTEP) geometry optimization, the closest Cl-H (OH group) distance is 2.08 Å. For the NH 3 + group, the Cl-H distances are 2.3, 2.4 and 2.5 Å; considering also the NH 3 + group rotation, it is informative that there is a good match between the experimental data in Figure 4 and the simulated two-spin build-up curve for a distance of 2.27 Å. When τ 1 RCPL is equal to τ 2 RCPL , the 1 H CSA is refocused by the 180 pulse. However, when this is not the case, the 1 H CSA has a strong influence on the recoupling sequence as shown experimentally (compare Figures 4a and S2) and in the simulations in Figure S3.
The simulations in Figure 4b are for a powder average. However, it must be remembered that different orientations in the powder, corresponding (when averaged over a rotor period for MAS) to different parts of the second-order quadrupolar-broadened lineshape, experience different quadrupolar interaction magnitudes and have a different relative orientation of the quadrupolar tensor with respect to the 35 Cl-1 H internuclear vector (corresponding to the principal axis of the dipolar coupling) so that they will be affected differently by the recoupling sequence. Therefore, we performed numerical simulation of the intensity of 35 Cl-1 H D-HMQC experiment individually for each orientation of the 35 Cl quadrupolar tensor in a powder distribution. This reveals a strong dependence of the heteronuclear dipolar recoupling on the orientation of the quadrupolar tensor as is shown in Figure 5c. Similar symmetry of the scaling factor has been obtained for phase-inverted R 3 recoupling (see Figure S5a). Such an investigation into the contributions from different orientations is useful in providing an insight into the experimental observations, specifically when looking at how different regions of the quadrupolar lineshape are affected by the recoupling . Figures 5a,b considers the second-order quadrupolar lineshape for a powder sample with C Q = 2.4 MHz and η = 0.72. Different colours are used to identify the orientations that contribute to different frequency intervals of the lineshape. Figure 5c shows how the SR4 2 1 recoupling is distributed over all the orientations in a powder with regions displayed in black being preferentially recoupled, whereas the orientations displayed in yellow experience a weak recoupling to a 1 H nucleus. Quantifying the change in intensity upon recoupling for each orientation that contributes to the second-order quadrupolar lineshape results in a distorted spectrum as shown in Figure 5d for 266.67 μs of SR4 2 1 recoupling. Figure S4 shows corresponding plots for 33.3 or 1,000 μs of SR4 2 1 recoupling, whereas Figure S5 considers different relative orientations of the dipolar and quadrupolar tensors.

| 1 H and 35 Cl MAS NMR onedimensional spectra and GIPAW calculations for amino acids and pharmaceuticals
This and the following section present MAS NMR spectra for two amino acids, glycineÁHCl as well as L-tyrosine HCl, and three pharmaceuticals, amitriptylineÁHCl, F I G U R E 6 1 H (850 MHz) magic-angle spinning (MAS) (60 kHz) nuclear magnetic resonance (NMR) single-pulse spectra with background suppression [50] (top, in blue) together with stick-spectra (bottom, in red) representations of the gauge-including projectoraugmented wave (GIPAW)-  Figures S7-S11). Specifically, this section presents one-dimensional 1 H and 35 Cl MAS NMR spectra in Figures 6 and 7, respectively. GIPAW calcula-tions of the 1 H and 35 Cl chemical shift and 35 Cl quadrupolar parameters (the quadrupolar coupling constant, C Q , the asymmetry parameter, η Q , and, hence, the quadrupolar product, P Q ) have been performed in  CASTEP for geometry-optimised crystal structures (see Table 1). For 1 H, the GIPAW-calculated chemical shifts are presented as stick spectra in Figure 6, with the experimental and GIPAW-calculated values tabulated in Tables S1-S5. GIPAW calculations for L-tyrosineÁHCl and glycineÁHCl and on cimetidineÁHCl have been previously reported. [15,63,64] 1 H MAS NMR spectra of cimetidineÁHCl have been previously presented in Maruyoshi et al. [65] When considering 35 Cl chemical shifts, it is important to consider the choice of reference. In this work we follow Bryce et al. [13] and use NaCl 0.1-M aqueous solution. To convert to a chemical shift scale where the 35 Cl chemical shift of solid NaCl is set to 0 ppm, as for example used by Gervais et al., [15] it is necessary to add 46 ppm (Eq. 6 in Bryce et al. [13] ). It is observed from Figure 7 and Table 1 that the GIPAW-calculated 35 Cl C Q values are a factor of 1.3-1.9 times larger than that determined experimentally. Such a discrepancy has been noted previously [15,64,66,67] ; in this context, Socha et al. have shown that better agreement is found for calculations performed using hybrid and meta-generalised gradient approximation (GGA) functionals. [64] Moreover, Holmes et al. [68,69] have used a large training set of organic solids for the parameterization of a DFT dispersion correction so as to achieve better agreement between experimental and calculated electric field gradient tensors.
3.4 | 35 Cl-1 H PT-D-HMQC MAS NMR spectra for amino acids and pharmaceuticals  (Figure 8a), cross peaks are observed with the amine and the carboxylic acid protons, at 7.7 and 11.8 ppm, respectively, for which the closest H-Cl distances (see Figure S8 and Table S2) to the chloride ion are at 1.9 and 2.1 Å, respectively. Note that the quadrupolar product, P Q , is biggest of all the samples studied in this paper (6.6 MHz, see Table 1), as evident from the greater extent of second-order quadrupolar broadening of the 35 Cl lineshape.
For amytriptilineÁHCl (Figure 8b), the crystal structure (see Figure S9 and Table S3) indicates that a NH proton H93 at 11.3 ppm, a CH 2 proton H65 at 2.6 ppm and a CH 3 proton H89 at 2.8 ppm are closest to the chloride ion at 1.9, 2.7 and 2.7 Å, respectively. For cimetidineÁHCl (Figure 8c), the crystal structure (see Figure S10 and Table S4) shows that the NH protons H1 at 15.0 ppm and H9 at 15.1 ppm, the imidazole H5 at 9.1 ppm and the CH 2 H33 at 4.0 ppm are at 2.0, 2.0, 2.6 and 2.8 Å away from the chloride ion, respectively. For lidocaineÁHClÁH 2 O (Figure 8d), the crystal structure (see Figure S11 and Table S5) indicates the NH proton H89 at 11.9 ppm, the H 2 O H97 at 4.3 ppm and the CH 3 H81 at 1.3 ppm are at 2.0, 2.2 and 2.9 Å, respectively away from 35 Cl. In each of Figure 8b-d, while Cl-H cross peaks are observed at lower ppm 1 H resonances, corresponding to the most intense alkyl 1 H peaks in the one-pulse spectra in Figure 6, disappointingly no peaks are evident for the low-intensity high-ppm NH resonances. Especially for Figure 6b-d where the 1 H MAS intensity of the OH and NH moieties is considerably lower compared with the aliphatic 1 H signal, a considerable boost in sensitivity of the PT-D-HMQC is required for the 35 Cl-1 H correlation for the OH and NH sites to become visible.

| CONCLUSIONS AND OUTLOOK
This paper presents applications of a 35 Cl-1 H D-HMQC MAS NMR experiment incorporating WURST saturation of the 35 Cl satellite transitions in a PT-D-HMQC approach to probe and identify proximity between chloride ions and protons in amino acids and pharmaceutical HCl salts. Density-matrix numerical simulation (using SIMPSON) of the effect of orientation of the quadrupolar tensor on the recoupling performance shows a strong anisotropic dependence that distorts the quadrupolar lineshape. For L-tyrosineÁHCl, the distance between 1 H and 35 Cl atoms can be quantified by recording D-HMQC build-up curves, as recorded by incrementing the recoupling time. DFT calculations using the CASTEP GIPAW code gives calculated 35 Cl quadrupolar coupling constants larger than those determined by experiment as observed by others.
The PT-D-HMQC MAS NMR spectra presented in this paper often suffer from marked t 1 noise due to the strong dependence of the recoupling Hamiltonian rotor phase on small fluctuations in the MAS frequency. In this respect, we note that various approaches to reduce t 1 noise, for example, based on using pi pulses to remove the CSA evolution during recoupling, have been introduced. [59,72,73] A 35 Cl-1 H DQ HMQC experiment, recently introduced by Hung et al., [74] provides higher resolution for 35 Cl by reducing the second-order quadrupolar broadening for a DQ coherence corresponding to the sum of the central transition and the satellite transition for the spin-3/2 35 Cl nucleus. Also, it may be convenient to exploit the shorter T 1 of the quadrupolar nucleus and perform X detected X-1 H D-INEPT experiments as has been demonstrated by Venkatesh et al. [31] for 35 Cl, 71 Ga and 27 Al. There is thus much potential for the further development of 35 Cl-1 H heteronuclear correlation MAS NMR experiments.