Simple and User-Friendly Methodology for Crystal Water Determination by Quantitative Proton NMR Spectroscopy in Deuterium Oxide

In drug research and development, knowledge of the precise structure of an active ingredient is crucial. However, it is equally important to know the water content of the drug molecule, particularly the number of crystal waters present in its structure. Such knowledge ensures the avoidance of drug dosage and formulation errors since the number of water molecules affects the physicochemical and pharmaceutical properties of the molecule. Several methods have been used for crystal water measurements of organic compounds, of which thermogravimetry and crystallography may be the most common ones. To the best of our knowledge, solution-state NMR spectroscopy has not been used for crystal water determination in deuterium oxide. Quantitative NMR (qNMR) method will be presented in the paper with a comparison of single-crystal X-ray diffraction and thermogravimetric analysis results. The qNMR method for water content measurement is straightforward, reproducible, and accurate, including measurement of 1H NMR spectrum before and after the addition of the analyte compound, and the result can be calculated after integration of the reference compound, analyte, and HDO signals using the given equation. In practical terms, there is no need for weighing the samples under study, which makes it simple and is a clear advantage to the current determination methods. In addition, the crystal structures of two model bisphosphonates used herein are reported: that of monopotassium etidronate dihydrate and monosodium zoledronate trihydrate.

The term "crystal water" refers to water molecules present in a compound's crystal structure.Water is commonly incorporated into the structure upon crystallization from aqueous solutions, and compounds incorporating crystal water molecules are called hydrates.It is often important to know the crystal water content of the compound under study, especially in the case of potential drug candidates entering in vitro and in vivo studies.Interactions with water can affect the functionality of the compounds, and the hydrate forms commonly have different physicochemical and pharmaceutical properties than the corresponding anhydrous forms. 1 Water in the crystal lattice increases plasticity of the crystals and enhances tabletability. 2,3Additionally, the crystal water content affects the total molecular weight of the compound and is needed for the calculation of the drug dosage.For example, one of the early generation bisphosphonate (BP) drugs, (1-hydroxyethan-1,1-diyl)bis(phosphonic acid) a.k.a.etidronic acid, is crystallized as a monohydrate.The molecular weights of the monohydrate and anhydrous forms are 224.05 and 206.03 g/mol, respectively.If the drug dose is calculated ignoring the water molecule, then there will be a ca.8% error in the drug dosage.The error increases when a larger number of crystal water molecules are present in the solid form of the compound, highlighting the importance of precisely knowing the number of crystal water molecules in the drug's molecular unit.
A plethora of methods have been used for water content measurements of organic compounds including thermogravimetric analysis (TGA), dynamic vapor sorption (DVS), Karl Fischer (KF) titration or Karl Fischer oven (KF-oven), singlecrystal X-ray diffraction (SCXRD), powder X-ray diffraction (PXRD), solid-state NMR spectroscopy (ssNMR), Fouriertransform infrared (FT-IR), and Raman spectroscopy. 4,5TGA and DVS give information about the mass change (loss) of the sample upon controlled heating, and the water content of the analyte can be subsequently calculated. 4Thermogravimetry can often distinguish between free and bound water. 6However, in some metal-phosphonate systems, crystal and metal-bound water molecules are removed in overlapping steps; hence, the separation is difficult.Thermogravimetry can also be used in combination with mass spectrometry 6 or with differential scanning calorimetry (DSC) 5 to identify the volatile matter during the heating process.The KF titration method is based on the reaction of iodine and sulfur dioxide in the presence of water using methanol or diethylene glycol monoethyl ether as solvents. 5The KF titration is not suitable for substances that release their water very slowly or only at high temperatures, are insoluble in the solvents used in the method, or contain functional groups (especially aldehydes and ketones) that can react with the reagents used in the KF test. 5Some of these problems can be overcome by using the KF-oven method, in which the water is first removed in an oven and the released moisture is directed into a titration cell, where the water determination is performed using KF titration. 5However, the oven can be heated only up to 250 °C, causing underestimated findings in the case of water of crystallization for some substances.By using SCXRD, it is possible to determine accurate atom positions and the method gives information about both the number and location of water molecules within the crystal structure. 7ssNMR methodologies utilize samples crystallized under deuterium oxide vapor and are often combined with quantum chemistry calculations. 8The combination of ssNMR and computational methods is called "NMR crystallography", and the method is primarily used to solve molecular structures in the solid state, but it is also capable of differentiating hydrate and anhydrous forms of a crystal structure. 4,9FT-IR and Raman spectroscopies are complementary methods, and they can identify hydrates from the signal caused by the hydroxyl group of H 2 O. 4,10 Most of these methods necessitate accurate weighing, are time-consuming, require special expertise of the method/equipment in question, and/or destroy the sample material.
Liquid-state NMR spectroscopy is a versatile method suitable for various types of applications, e.g., structure determination, in situ monitoring, quantification of impurities, or quantitative analysis of mixtures. 11−14 NMR spectroscopy is also increasingly being applied in pharmaceutical research and industry. 15,16mportantly, most modern drug synthesis laboratories have NMR equipment for structure and purity analysis, and the same instrumentation can also be used for other applications.Proton ( 1 H) NMR spectroscopy is well-suited for water determination due to its quantitative nature: 1 H NMR signal areas are directly proportional to the number of nuclei giving rise to the signal. 12,17mportant criteria for the proper application of quantitative NMR (qNMR) spectroscopy include the high solubility of the analyte in the solvent in question, and the presence of some detectable nucleus, which is often hydrogen. 18Additionally, several acquisition parameters need to be optimized to obtain quantitative results. 11,17,19For example, the repetition time (also called recycling time) between pulses needs to be ≥ 5 × T 1 of the slowest relaxing nucleus, and the number of scans must be suitable to reach adequate signal-to-noise ratio for reliable quantification. 12,20A 90-degree pulse is the best choice for qNMR but also shorter pulses can be used if the available measurement time is limited. 19n the present study, we introduce a novel, simple, and userfriendly method for crystal water determination of water-soluble compounds by liquid-state 1 H NMR spectroscopy.The method is applied to several commercial or synthesized "model" compounds, and the NMR results are compared to those obtained with TGA and/or SCXRD.Several of the tested compounds are BPs that are well-known drugs used for bonerelated diseases, such as osteoporosis. 21BPs are often soluble only in water, and they can contain a variable number of crystal waters in their structure.
NMR Sample Preparation.NMR samples were prepared by pipetting ca.500 μL of a reference solution into a 5 mm NMR tube.After the 1 H NMR measurement of the reference solution, a small amount of the compound under study was placed in the same NMR tube using an appropriate spatula.The solution was gently mixed by inverting the NMR tube several times until all solids were dissolved, and the 1 H NMR measurement was repeated using the same parameters as for the reference sample.
The repeatability of the method was tested by preparing some NMR samples in duplicates, and the relative standard deviation was calculated for the duplicate results.
1 H NMR Spectroscopy.The samples were measured on a Bruker AVANCE III HD (Bruker BioSpin GmbH, Karlsruhe, Germany) 600 MHz NMR spectrometer equipped with a cryogenically cooled TCI cryoprobe using a zg pulse sequence containing a 90°excitation pulse.The acquisition parameters included 8 scans (NS), 84132 data points (TD), spectral width (SW) 20.0246 ppm, receiver gain (RG) 2, acquisition time (AQ) 3.5 s, and relaxation delay (D1) 60 s.Some of the spectra were measured also with AQ 8 s and/or D1 200 s.The spectra were measured at a 295 K temperature.
The free induction decays were multiplied with an exponential line-broadening function (LB = 0.5 Hz) before Fourier transformation.Phase correction and integration of signals were performed manually by the Topspin 3.6.2version (Bruker BioSpin GmbH, Karlsruhe, Germany).The integral of the reference compound (either TSP or formic acid) was always set to 1.0.Water content of the samples was calculated using eq 1.
where I W2 = integral of the HDO signal in the spectrum containing the compound X.I W1 = integral of the HDO signal in the spectrum containing only the reference solution.N ex = number of exchangeable protons in the structure of the compound X.I X = integral of the compound X signal.N X = number of hydrogens yielding the 1 H NMR signal of the compound X.Estimates of T 1 times of formic acid and HDO were determined by measurements using the t1ir1d pulse program.The first spectrum was measured using a very short delay (d7 = 0.000003 s), and the phases were manually corrected to get all the NMR peaks negative.The measurement parameters included NS 1, TD 81920, SW 21.036 ppm, RG 57, AQ 3.24 s, and D1 150 s.Several other spectra were measured with otherwise the same parameters except that the d7 delay was gradually increased until the peak of interest was nulled.Below the correct d7 the signal is negative, whereas above the correct d7, the signal is positive.The T 1 time for the signal of interest is d7 for the null divided by the natural logarithm of 2.
Thermogravimetric Analysis.The samples were measured with a NETZSCH TG 209 F1 Libra thermogravimetric analyzer in an open Al 2 O 3 pan to determine the content of the crystal water in the compounds.The samples were first heated isothermally at 70 °C for 30−120 min to remove the "free" surface water and further heated with a heating rate of 10 °C/ min up to 300 °C.The mass loss between the end of the isothermal step and 200 °C was considered as crystal water removed from the compound.All measurements were carried out under a 20 mL/min N 2 gas flow with three replicates.
Single-Crystal X-ray Diffraction.Measured crystals were prepared under inert conditions and immersed in perfluoropolyether as a protecting oil for manipulation.Suitable crystals were mounted on MiTeGen Micromounts, and these samples were used for data collection.Data for the compounds monopotassium etidronate dihydrate and monosodium zoledronate trihydrate were collected with a Bruker D8 Venture diffractometer with graphite monochromated Cu Kα (λ = 1.54178Å).The data were processed with the APEX3 suite [Bruker APEX3.APEX3 V2019.1;Bruker-AXS: Madison, WI, USA, 2019.].The structures were solved by intrinsic phasing using the ShelXT program, 24 which revealed the position of all non-hydrogen atoms.These atoms were refined on F 2 by a fullmatrix least-squares procedure, using the anisotropic displacement parameter. 25All hydrogen atoms were located in difference Fourier maps and included as fixed contributions riding on attached atoms with isotropic thermal displacement parameters 1.2 or 1.5 times those of the respective atom.The Olex2 software was used as a graphical interface. 26Molecular graphics were generated using Mercury. 27The crystallographic data for the reported structures were deposited with the Cambridge Crystallographic Data Center.(e.g., −OH and/or −NH x ) or ionic (e.g., phosphonate) groups and are water-soluble.It is important to note that deuterium oxide always contains a small amount of HDO and will give a residual signal at about 4.7 ppm.This residual HDO content of the solvent must be considered in the crystal water calculations.Our proposed approach is to perform the measurements in a deuterium oxide solution containing a reference compound.The reference compound can be arbitrarily selected but should give a signal that does not overlap with the HDO or the analyte signals to enable accurate integration.We tested the performance of two reference compounds, TSP and formic acid.TSP is commonly used both as a chemical shift reference and a concentration reference and is commonly available in NMR laboratories.TSP gives only one singlet signal at 0 ppm that does not typically overlap with the other signals.Formic acid is a simple and commonly available compound that gives one singlet signal at approximately 8.46 ppm.The use of formic acid is recommended particularly if the analyte compound must be recovered after the water content measurement because formic acid can be easily evaporated, in contrast to TSP, which will remain with the sample.
The workflow of the protocol for crystal water determination is shown in Scheme 1.The method is designed to be simple and easy to perform and does not require the expertise of an NMR specialist or even weighing the compound under analysis.The protocol requires 1 H NMR spectra to be recorded before and after the addition of the analyte compound, and the result can be calculated after integration of the reference compound, analyte, and HDO signals using eq 1.As an example, the 1 H NMR spectra of a reference solution and a reference solution after addition of etidronic acid monohydrate together with the crystal water calculation are shown in Figure S1.Commonly, it is sufficient to integrate only one of the analyte signals.In certain cases, some of the analyte signals overlap with the HDO signal.This can be taken into account in the calculation by subtracting the integral value corresponding to the number of protons overlapping the water signal, as demonstrated for monosodium zoledronate trihydrate in Figure S2.Alternatively, it would be possible to use deconvolution for signal area determination instead of integration. 28,29obustness of 1 H NMR Measurement Parameters.Small molecules have long longitudinal relaxation times (T 1 ).
The relaxation times of HDO and formic acid protons were determined to be ∼9 and 11 s, respectively.For quantitative measurements, the repetition time (acquisition time + relaxation delay) between excitation pulses is advised to be ≥5 times the longest T 1 to allow ≥99.3% relaxation. 12Thus, the repetition time must be ≥45 s to allow HDO protons to be adequately relaxed.It is also important to use a long-enough acquisition time to collect the fully decayed free induction decay (FID). 30owever, acquisition time that is too long collects only noise.Commonly, the acquisition time is set to 5−8 s for quantitative measurements. 30he robustness of the measurement parameters was tested by changing the acquisition time and the relaxation delay resulting in repetition times of 63.5, 68, 203.5, and 208 s.Crystal water determinations for citric acid monohydrate and sodium acetate trihydrate with two different acquisition times and relaxation delays, as well as using either TSP or formic acid as reference compounds are presented in Table 1.The difference between results obtained with acquisition times of 3.5 and 8 s is negligible, and thus, the acquisition time of 3.5 s is already sufficient.According to our results (Table 1), the relaxation delay of 60 s is adequate also when using formic acid as the reference compound even though the T 1 time is a bit longer for formic acid than for HDO.This is because the reference compound signal does not need to be fully relaxed, as the same measurement parameters are used in both measurements, and the reference compound signal is always set to a constant value.The most important thing is to get the HDO and analyte compound protons relaxed because their integrals are used for the water content calculation.The measurement time for one spectrum using an acquisition time of 3.5 s and a relaxation delay of 60 s is ca.13 min.
Required Analyte Concentration.NMR samples from several compounds including a variable number of exchangeable protons were prepared with varying concentrations in order to evaluate any concentration effects.Variable-concentration experiments were performed for commercial compounds (Table 2), and the samples were prepared without exact weighing by adding low, average, and high amounts of analyte compounds to the NMR tubes to keep the method as simple as possible during all stages of the protocol.Surprisingly, all lowconcentration samples gave results that were systematically ∼10% higher (per crystal water molecule) than the actual crystal water in the compounds.For the higher-concentration samples, the results were more accurate (Table 2).Concentrations in mmol/L and mg/mL units were calculated from the 1 H NMR spectra, and according to the results obtained, a reasonable concentration for accurate water content measurements is around 100 mmol/L or ca.10−20 mg/mL (5−10 mg in 0.5 mL reference solution in a 5 mm NMR tube); however, this somewhat depends on the individual compound.The number of exchangeable protons does not seem to influence the method's accuracy.Notably, ATP (and analogues) can exist in various hydrate forms, and it is pivotal to check the crystal water content prior to preparing ATP solutions with certain concentrations.
Comparison with Other Methods and Reproducibility.In order to compare our qNMR methodology with those of two commonly used water determination methods, a set of compounds were also measured by TGA and SCXRD (Table 3).The compounds included two commercially available citrates and four BPs.The TGA data for all six compounds are shown in Figure S3.According to the TGA and qNMR results, both citrates contain two crystal waters.Trisodium citrate 5.5hydrate had been stored unopened for at least 15 years before the qNMR and TGA measurements, and thus, the crystal water content of the compound stabilized to two per citrate molecule.
Structural elucidation for all four BPs included in this work was performed by SCXRD.Notably, the structures of the new compounds monopotassium etidronate and monosodium zoledronate are reported here for the first time.The structure of etidronic acid monohydrate has been published before. 31Its asymmetric unit is shown in Figure 1A.There is one crystal water.
The etidronate anion in the structure of monopotassium etidronate is found in its monodeprotonated state, and it is charge-balanced by the K + cation (Figure 1B).The K + center is 7-coordinated in a capped trigonal prismatic geometry (Figure 1D), surrounded by three types of etidronate anions.The first forms a tris-chelating arrangement (via the C−OH group and two protonated P−OH moieties, one from each phosphonate).The second type of etidronate binding is terminal, with two P− OH groups from two different etidronate ligands binding to the K + center.Lastly, one etidronate molecule binds to the K + center in a bidentate fashion (via the C−OH group and a protonated P−OH group).The K−O bond distances are in the range of 2.749−3.080Å, within the expected range observed in other Kphosphonate compounds. 7,32There are two crystal waters in the structure (Figure 1B), which are hydrogen-bonded to neighboring phosphonate oxygens.
The structure of monosodium zoledronate (Figure 1C) exhibits additional complexity, as there are two crystallographically distinct Na + centers (Figure 1E,F).The first (Na1) is found in a distorted octahedral coordination environment bound by four phosphonate oxygens and two −OH groups (in a trans arrangement).The second (Na2) has a peculiar distorted trigonal pyramidal geometry, and it is bound by two phosphonate oxygens in a bidentate fashion, as well as by three water molecules in a meridional arrangement.There are  three crystal waters in its structure.They form multiple hydrogen bonds with phosphonate oxygens and other water molecules.
In the case of etidronic acid, the TGA showed that the crystal water is loosely bound because it was removed already during the 70 °C isothermal phase, which made it difficult to observe but was not a problem with the qNMR method.The results for monopotassium etidronate dihydrate and monorubidium etidronate dihydrate were well in line between all three methods, giving two crystal waters for both compounds.Regarding monosodium zoledronate trihydrate, the SCXRD analysis was performed a couple of years prior to the present qNMR and TGA measurements and showed three crystal waters per zoledronate molecule.However, according to the TGA and qNMR results, the compound currently contains only two crystal waters per zoledronate molecule.Similarly, as with trisodium citrate 5.5-hydrate, all the crystal waters in monosodium zoledronate trihydrate are not stable.Both trisodium citrate 5.5-hydrate and monosodium zoledronate trihydrate have lost some of the crystal water during long storage and/or handling.Hence, for commercially available compounds, even though the water content of the compound is usually described on the container label, the only way to be confident about the water content of the analyte is to measure the water content before its use, and the qNMR method described here allows for fast and simple determination.
It is worth mentioning that the BP samples were sent from Heraklion (Crete, Greece) to Kuopio (Finland) during the month of February.The first 1 H NMR measurements of monosodium zoledronate trihydrate were performed a few days after the sample's arrival.The measurements showed the presence of three crystal waters (specifically, 2.99 and 3.10 in two different measurements with varying concentrations).The next qNMR measurement was performed at least three months later together with TGA.Both methods showed the presence of two crystal waters.The air moisture levels in Kuopio (Finland) and in Heraklion (Crete, Greece) at the end of February are different, being much lower in Finland than in Crete where the monosodium zoledronate trihydrate was crystallized and measured by SCXRD.It can be speculated that the storage of the monosodium zoledronate trihydrate at ambient temperature in lower moisture conditions may have induced the loss of one water molecule.
The majority of the qNMR water content determinations for the compounds shown in Table 3 were performed in duplicates, and the relative standard deviations were within 0.19−3.24%.The corresponding relative standard deviations for the TGA data were 0.05−4.45%,indicating that the repeatabilities of the NMR and TGA methods are on the same level.Higher RSD % for the monorubidium etidronate is caused by low concentrations of the samples since the lower analyte concentrations tend to overestimate the results as discussed above.When the concentrations of the replicates are at least ca. 100 mmol/L, the relative standard deviations are well below 2%.
Strengths and Limitations.The qNMR methodology described in this paper is novel, fast, nondestructive, and accurate.The sample preparation is simple because (a) no weighing of the sample is required (adding ∼5−15 mg of the analyte compound using an appropriate spatula to the NMR tube), and (b) the volume of the reference solution does not have to be precise (just pipetting around 0.5 mL of the reference solution using Pasteur pipet to the NMR tube).If the concentration of the analyte appears to be too low, it is possible to simply add more solids to the same NMR tube and repeat the measurement.Furthermore, NMR spectroscopy enables the identification of possible organic impurities present in the sample with the same measurement.Additionally, the availability of basic NMR equipment in modern drug synthesis laboratories enables an easy adaptation of the qNMR methodology for water determinations.
Interestingly, recent technical developments on benchtop NMR spectrometers have made NMR spectrometers more accessible since the cost of such instrumentation is considerably lower than that of a high-field spectrometer. 35Benchtop NMR spectrometers have lower sensitivity and resolution than their higher-field counterparts, but they can be utilized for various purposes for which high sensitivity and resolution are not required.Detection and quantification limits for benchtop NMR spectroscopy are at a millimolar range that would be adequate for crystal water determination because the preferred analyte concentration is ca. 100 mM.In the future, the applicability of benchtop NMR for crystal water determination will be explored in practice.
It must be acknowledged that the qNMR method cannot distinguish between crystal water and surface water.If the analyte compound is prone to adsorb surface water, then it is advised to dry the analyte compound under high vacuum to remove surface water prior to the qNMR analysis.This handling should not remove the crystal water if the sample is not heated, unless the crystal water is exceptionally loosely bound.This was tested to the etidronic acid monohydrate, and crystal water remained in the structure.Also, the analyte compound must have at least one nonexchangeable proton and produce a signal that does not overlap with the HDO or the reference compound signals.

■ CONCLUSIONS
Accounting for the simplicity, repeatability, accuracy, nondestructiveness, and speed, the qNMR method described herein appears to be a convenient and user-friendly alternative to currently existing methods for crystal water determination.It is anticipated that this method will be broadly adopted in the fields of structural and medicinal/pharmaceutical chemistry, where water content measurements are crucial.

■ ASSOCIATED CONTENT
Scheme 1. Workflow of Crystal Water Determination a

Figure 1 .
Figure 1.(A) Asymmetric unit of etidronic acid monohydrate.(B) Asymmetric unit of monopotassium etidronate.(C) Asymmetric unit of monosodium zoledronate.(D) Coordination environment of the K + center in the structure of monopotassium etidronate.(E,F) Coordination environment of the two different Na + centers in the structure of monosodium zoledronate.Color codes: metal centers (K + or Na + ), magenta; O, red; P, orange; C, black; H white. Crystal waters are shown in green.

Table 1 .
Crystal Water Determination Data for Commercially Available Compounds Obtained Using Variable Measurement Parameters and Reference Compounds

Table 2 .
Crystal Water Determination Data for Commercially Available Compounds Using Various Analyte Concentrations Number of exchangeable protons in the structure.b Two crystal water molecules per acetate molecule. a

Table 3 .
Comparative Crystal Water Contents of Selected Compounds Determined by qNMR, TGA, and SCXRD a Number of exchangeable protons in the structure.b Relative standard deviation (RSD %) was calculated for the replicates using the equation RSD % = 100 × SD/μ, where SD is the sample standard deviation and μ is the sample mean.c Concentrations of duplicates reported separately.n.d.= not determined.