Stoichiometric and Non-Stoichiometric Hydrates of Brucine.

The complex interplay of temperature and water activity (aw) / relative humidity (RH) on the solid form stability and transformation pathways of three hydrates (HyA, HyB and HyC), an isostructural dehydrate (HyAdehy ), an anhydrate (AH) and amorphous brucine has been elucidated and the transformation enthalpies quantified. The dihydrate (HyA) shows a non-stoichimetric (de)hydration behavior at RH < 40% at 25 °C and the removal of the water molecules results in an isomorphic dehydrate structure. The metastable dehydration product converts to AH upon storage at driest conditions or to HyA if exposed to moisture. HyB is a stoichiometric tetrahydrate. The loss of the water molecules causes HyB to collapse to an amorphous phase. Amorphous brucine transforms to AH at RH < 40% RH and a mixture of hydrated phases at higher RH values. The third hyrdate (HyC) is only stable at RH ≥ 55% at 25 °C and contains 3.65 to 3.85 mole equivalent of water. Dehydration of HyC occurs in one step at RH < 55% at 25 °C or upon heating and AH is obtained. The AH is the thermodynamically most stable phase of brucine at RH < 40% at 25 °C. Depending on the conditions, temperature and aw, each of the three hydrates becomes the thermodynamically most stable form. This study demonstrates the importance of applying complimentary analytical techniques and appropriate approaches for understanding the stability ranges and transition behavior between the solid forms of compounds with multiple hydrates.

Introduction

Hydrates are the most common solvate forms identified among pharmaceuticals and other small organic molecules and are known to occur for at least of a third of organic (drug) molecules.[1−3] Solvent adducts often crystallize more easily than solvent-free forms, because water or small solvent molecules generally improve packing efficiency when incorporated in the crystal structure. Moreover, the hydrogen bonding sites of a (drug) molecule are frequently better satisfied by interactions with water than to the molecule itself,[4] resulting in a stabilization of the crystal structure.

Hydrates can be formed by crystallization from a water solution or from a solution of organic solvents containing a sufficient quantity of water. Hydrate formation may also occur when a solid substance just comes in contact with water (wet granulation, aqueous film coating) or on exposure to water vapor (storage).[5] The physical stability of crystalline hydrates is dependent on temperature, absolute pressure, as well as the partial water vapor pressure.[6,7] The presence of water molecules within the lattice of a compound affects the packing arrangement of molecules and the intermolecular interactions in a crystal structure and, hence, influences solubility, dissolution rate, stability, and bioavailability of pharmaceutical compounds.[8] The stability of hydrates can vary significantly, and their transformation behavior and interrelationship with other solid forms of the same compound (anhydrates, lower/higher hydrates, etc.) should be extensively studied[9−15,15] to avoid problems related to the material properties of the substance. Generally, hydrates are susceptible to dehydration during routine drying or storage conditions, which may lead to the formation of a hydrate with lower water content, a dehydrated hydrate, one or more anhydrous forms (polymorphs) or an amorphous material. The dehydration and rehydration processes of a hydrate forming system can be very complex,[16−25] and may involve multiple phases. Therefore, investigating the interactions of water with a substance, finding and characterizing[26−30] existing hydrate forms, and determining their hydration and dehydration characteristics,[31,32] as well as polymorphic transformations, is essential for the development of robust manufacturing processes of any fine chemical for both practical and regulatory reasons. Furthermore, the assessment of the physical and often also chemical stability of the drug substance and drug products is crucial.[4,8,33−36]

On the basis of the hydration/dehydration mechanisms, the continuity or discontinuity of the sorption/desorption behavior, and the involved structural changes, hydrates are commonly grouped into two main classes, stoichiometric and nonstoichiometric hydrates. Stoichiometric hydrates have a well-defined water at a given relative humidity (RH) content, and the crystal structure is clearly different from that of other solid form(s). The dehydration mechanism involves a considerable rearrangement of the host molecules. Hydrate structures hosting water molecules in open structural voids, such as channels, often show nonstoichiometric behavior. The solvent may fully or partly escape through these channels without significant changes in the crystal structure, except anisotropic expansion/distortion of the structure due to the accommodation or release of the water molecules in the structure.[37] A careful investigation of a hydrate over a wide range of relative humidities is mostly the key to establish its stoichiometric or nonstoichiometric behavior.

Nonstoichiometric hydrates may be generally rated as “problematic” and “difficult to handle” solid forms. The water in nonstoichiometric hydrates is often rather weakly bound (“free” water) and may interact with other components, compromising the stability and performance of formulated products. The variability of the water content is also highly relevant for weighing and dosing operations of a substance and may be critical for adjusting a dose uniformity in single unit dosage forms or may lead to substantial errors in any mass based values including the activity data of biologically active compounds. Avoiding variations in the water content in nonstoichiometric hydrates is often very difficult under processing conditions and requires special efforts such as a precise control of the environmental conditions (moisture and temperature). Apart from anisotropic lattice contraction, the loss of water typically does not appreciably affect its structure. Furthermore, removing the solvent often results in isomorphic desolvates (dehydrates) containing “empty space” in their crystal structure that can result in reduced chemical or physical stability.[12,38−40] In general, isomorphic dehydrates are “highly hygroscopic” and reuptake water readily when exposed to elevated humidity conditions. They are usually metastable, of higher energy, and may (besides the original solvent) also take up other solvents or even molecular oxygen[41] to minimize void space in the crystal. The reduced packing efficiency of the desolvated lattice results in a net decrease in lattice energy; i.e., it becomes less stable relative to the solvated structures.[42] It has also been observed that nonstoichiometric hydrates may lose crystallinity when the very last water molecules are removed.[43]

Brucine (2,3-dimethoxystrychnidin-10-one) is an alkaloid (Figure ), which is structurally strongly related to strychnine and is found in the seeds of the Stychnos nux-vomica tree.[44] The alkaloid features six asymmetric carbon atoms and no hydrogen-bonding donor group. Brucine can be used as a tool for stereospecific chemical syntheses and has been used as an enantioselective recognition agent in chiral resolution.[45−48] The compound is a neurotoxin, which acts as an antagonist at glycine receptors. It has been used for the treatment of liver cancer in the Chinese medicine, and additionally, antiproliferative effects in different cancer cells have been reported.[49−53] However, the use of brucine as a treatment for cancer is limited due to its narrow therapeutic window. Furthermore, it has been reported that brucine shows analgesic and anti-inflammatory properties.[54]

Figure 1

Molecular diagram of brucine (2,3-dimethoxystrychnidin-10-one).

The brucine solid form reported by Groth in 1919 is a tetrahydrate.[55] The Cambridge Structural Database (CSD)[56] contains the structures of different crystal forms of brucine, including 2 anhydrates (CSD Refcodes: MAJRIZ,[57] a low temperature anhydrate AH, and MAJRIZ01,[58] the commercially available AH), a dihydrate (CIKDOQ, HyA),[59] a 3.86 hydrate (YOYZIX,[58] which was published after we finished our experimental work, HyC), a tetrahydrate (ZZZPRW01, HyB),[60] a 5.25-hydrate (UCOJIG),[60] 12 solvates with organic solvents (JIFWEB,[61] JIFWIF,[61] JIFWOL,[61] JIFWUR,[61] PIGNUP,[61] PIGPAX,[61] PIGPEB,[61] PIGPIF,[61] PIGPOL,[61] PIGPUR,[61] PIGQAY,[61] and MAJROF[57]), and 4 heterosolvates (“mixed” solvates) with water and organic solvent molecules (DAFFUL,[62] MAJRUL,[57] HIDGOS,[63] and HIDGUY[63]). For more details, see Table S1 of the Supporting Information.

In the present study we develop and report for the first time a consistent thermodynamic and kinetic picture of three brucine hydrates (HyA, HyB, and HyC) and water-free forms thereof (AHI, HyA, amorphous). The polymorphic pair AH/AH (LT - low temperature) has already been investigated and reported to be enantiotropically related (AH → AH transformation upon cooling: −24 °C; AH → AH transformation: 36 °C upon heating) and was, therefore, not further investigated.[58] A broad range of analytical techniques were applied to characterize the three hydrates and their dehydration products. This included hot-stage microscopy (HSM), differential scanning calorimetry (DSC), thermogravimetric analysis (TGA), isothermal relative humidity (RH)-perfusion calorimetry, gravimetric moisture sorption/desorption analysis, environmental powder X-ray diffraction, vibrational spectroscopy (IR and Raman), and H2O/D2O exchange experiments. It was possible to get a comprehensive picture of the stability ranges and the thermo-physical characteristics of this complex hydrate/water-free system. This investigation can be seen as a model study for a pharmaceutical, highlighting that a molecular level understanding of water–solid interactions leads to more robust strategies for handling, processing, and using pharmaceutical solids.

Materials and Methods

Materials

Brucine dihydrate (Lot # 47H1544) was purchased from Sigma. The dihydrate (HyA) was prepared by stirring a suspension of brucine in water between 40 and 50 °C for 1 day. Similarly, the tetrahydrate (HyB) and 3.85-hydrate (HyC) were prepared by stirring a brucine suspension in water at 0 °C and between 20 and 30 °C for 1 day, respectively. Brucine anhydrate (AH) was produced by drying HyC at 160 °C in a drying oven for 30 min. The isostrucural HyA dehydrate (HyA) was produced by dehydration of HyA at 0% RH at 25 °C. Exposure of HyA to moisture resulted in an immediate back-transformation to HyA. Amorphous brucine was prepared by storing HyB over P2O5 at 60 °C for 48 h. All experiments were performed with carefully conditioned solid forms by storing the samples at: HyA, 53% RH (over a saturated Mg(NO3)2 solution) at 25 °C; HyB, 98% RH (over a saturated K2SO4 solution) at 8 °C; HyC, 92% RH (over a saturated KNO3 solution) at 25 °C; AH, stored at 0% RH (over P2O5) at 25 °C. Amorphous brucine and HyA were freshly prepared right before use.

Gravimetric Moisture Sorption/Desorption Experiments

Moisture sorption and desorption studies were performed with the automated multisample gravimetric moisture sorption analyzer SPS23-10μ (ProUmid, Ulm, D). The moisture sorption analyzer was calibrated with saturated salt solutions according to the suppliers recommendations. Approximately 100–200 mg of sample was used for each analysis. The HyB and HyC measurement cycles were started at 95% with an initial stepwise desorption (decreasing humidity) to 0%, followed by a sorption cycle (increasing humidity) back to 95% relative humidity (RH). RH changes were set to 5% for all cycles. For HyA, two measurement cycles were performed. The first started at 90% RH with an initial stepwise desorption to 0%, followed by a sorption cycle back to 95% RH, with RH changes set to 5%. The second measurement cycle started at 40% RH with an initial stepwise desorption to 0%, followed by a sorption cycle back to 40% RH, with RH changes set to 2%. The equilibria conditions for each step were set to a mass constancy of ± 0.001% over 60 min and a maximum time limit of 48 h for each step.

Determination of the Critical Water Activity (Slurry Method)

Excess of amorphous brucine was stirred (500 rpm) in ≥ 0.5 mL of methanol/water mixtures, each containing a different mole fraction of water corresponding to a defined water activity[64,65] (section 2 of the Supporting Information) at 10.0, 25.0, and 40.0 ± 0.1 °C for 7 days. Coulometric Karl Fischer Titration (C20 instrument, Mettler Toledo, CH) was applied to determine the water content in the mixtures. The water activity of solvent mixtures at 10 and 40 °C was calculated using the NRTL (Non-Random Two-Liquid) model as implemented in the ASPEN Properties software program.[66] Karl Fischer titrations and water activity calculations of the solvent mixtures were performed without brucine. Brucine samples were withdrawn, and the resulting phase (wet cake) was determined using PXRD (measured between two mylar foils to avoid a phase transformation during measurement).

Temperature-Dependent Slurry Experiments in Water

Suspensions of amorphous brucine were prepared in water and then stirred either at a constant temperature (2, 10, 15, 20, 25, 30, 35, 40, 50, 60 ± 1 °C) or cycled in between x and y °C (5–10, 10–15, 15–20, 20–25, 25–30, 30–35, 35–40, 10–20, 20–30, 30–40, 40–50, 50–60 ± 1 °C) for 7 days. The wet cakes were analyzed with PXRD (measured between two mylar foils to prevent solvent loss). For more details, see section 3 of the Supporting Information.

Powder X-ray Diffraction (PXRD)

PXRD patterns were obtained using an X’Pert PRO diffractometer (PANalytical, Almelo, NL) equipped with a θ/θ coupled goniometer in transmission geometry, a programmable XYZ stage with a well plate holder, Cu-Kα1,2 radiation source with a focusing mirror, a 0.5° divergence slit, a 0.02° Soller slit collimator on the incident beam side, a 2 mm antiscattering slit, a 0.02° Soller slit collimator on the diffracted beam side, and a solid state PIXcel detector. The patterns were recorded at a tube voltage of 40 kV and tube current of 40 mA, applying a step size of 2θ = 0.013° with 80 or 200 s per step in the 2θ range between 2° and 40°. For nonambient RH measurements, a VGI stage (VGI 2000M, Middlesex, U.K.) was used. Equilibration conditions for VGI measurements are given in section 4 of the Supporting Information.

The PXRD patterns, recorded at 25 °C, were indexed using the first 20 peaks with DICVOL04 and the space group, which was determined based on a statistical assessment of systematic absences[67] as implemented in the DASH structure solution package,[68] and agreed with the single crystal data ignoring temperature effects. Pawley fits[69] and Rietveld refinements[70] were performed with TOPAS Academic V5.[71] The background was modeled with Chebyshev polynomials, and the modified Thompson–Cox–Hastings pseudo-Voigt function was used for peak shape fitting. For the Rietveld refinements, the brucine and water molecules were treated as rigid bodies.

Thermal Analysis

Differential Scanning Calorimetry (DSC)

DSC thermograms were recorded on a DSC 7 (PerkinElmer Norwalk, Ct., USA) controlled by the Pyris 2.0 software. Using a UM3 ultramicrobalance (Mettler, Greifensee, CH), samples of approximately 5–15 mg were weighed into perforated or sealed aluminum pans or hermetically sealed (high pressure) capsules. The samples were heated using rates in between 1 and 20 °C min–1 with dry nitrogen as the purge gas (purge: 20 mL min–1). The instrument was calibrated for temperature with pure benzophenone (mp 48.0 °C) and caffeine (236.2 °C), and the energy calibration was performed with indium (mp 156.6 °C, heat of fusion 28.45 J g–1). The errors on the stated temperatures (extrapolated onset temperatures) and enthalpy values were calculated at the 95% confidence intervals (CI) and are based on at least five measurements.

Thermogravimetric Analysis (TGA)

TGA was carried out with a TGA7 system (PerkinElmer, Norwalk, CT, USA) using the Pyris 2.0 Software. Approximately 5–7 mg of sample was weighed into a platinum pan. Two-point calibration of the temperature was performed with ferromagnetic materials (Alumel and Ni, Curie-point standards, PerkinElmer). Heating rates of 2–10 °C min–1 were applied, and dry nitrogen was used as a purge gas (sample purge: 20 mL min–1, balance purge: 40 mL min–1).

Isothermal Calorimetry (IC)

RH perfusion calorimetry experiments were performed with the TAM III nanocalorimeter unit (TA Instruments, Eschborn, D) in a 4 mL stainless steel RH perfusion ampule. The RH was controlled with two mass flow controllers, and dry N2 was used as carrier gas at a constant flow rate of 100 mL h–1. Approximately 15 mg (AH) and 25 mg (HyA) of sample was used. The HyA ↔ HyA humidity profile (% RH vs time) was executed as follows: 40 → 0 → 40 → 0 → 40% RH in each one step. The HyC ↔ AH humidity profile was executed as follows: 95 → 5 → 60 → 95% RH. The RH perfusion cell was calibrated with saturated solutions of NaCl (75.3% RH), Mg(NO3)2 (52.8% RH), and LiCl (11.3% RH). The heat flow of the empty RH perfusion ampule (baseline runs with the same humidity steps) was subtracted from the heat flow of the sample measurement. The errors on the stated (de)hydration enthalpy values are calculated at the 95% confidence intervals (CI) based on at least three measurements.

FT-Raman Spectroscopy

Raman spectra were recorded with a Bruker RFS 100 Raman spectrometer (Bruker Analytische Messtechnik GmbH, D), equipped with a Nd:YAG Laser (1064 nm) as the excitation source and a liquid-nitrogen-cooled, high sensitivity Ge detector. The spectra (1064 scans per spectrum) were recorded in aluminum sample holders with a laser power of 400 mW and a resolution of 2 cm–1. Samples (H2O ↔ D2O exchange) were stored and measured in hygrostats (98% RH) as detailed in ref (72).

Results and Discussion

Moisture-Dependent Stability of Brucine Hydrates

Hydrate A

The water sorption/desorption behavior of brucine hydrates was investigated between 0% and 95% RH. HyA (Figure ) shows in the RH range between 0% and 40% a sorption/desorption profile typical for nonstoichiometric hydrates.[37] This is evident from the gradual mass changes in the lower RH range and the lack of a hysteresis between sorption and desorption curves. The missing hysteresis can be related to the fact that the water molecules are located in open channels (Figure c), enabling a fast water egress/ingress without significant distortion of the overall crystal structure. Thus, the RH stability relationship is readily apparent from Figure a. The isotherms show that HyA is stable above ca. 20% RH and releases its water only at very dry conditions. Within the RH range of 40–95%, the water content is constant, and between 40% and 20% RH, less than 0.05 mol equiv of water is released, whereas, in the range between 20% and 10% RH, the hydrate loses 0.25 mol equiv of water. At the lowest RH values, the water escapes almost completely within half a day. This information is crucial for storing and handling HyA. It has to be noted that <0.05 mol equiv of water was retained in HyA during the desorption experiments (≤ 48 h at 0% RH, n = 15).

Figure 2

(a) Gravimetric moisture sorption and desorption curves of brucine HyA at 25 °C. Note that measurement points from sorption and desorption cycles coincide. (b) Fractional occupancies of water molecules derived from Rietveld refinements of the PXRD patterns recorded at different RH values. (c) Void space analysis of HyA (CIKDOQ[59]), excluding the water molecules, showing the water channels along the crystallographic b axis. Water space was calculated using the Hydrate Analyzer tool in Mercury and a probe radius and approximately a grid spacing of 1.2 and 0.15 Å, respectively.

A nonstoichiometric (channel) hydrate is often easiest identified by water vapor sorption/desorption studies in combination with PXRD. Changes in X-ray diffraction patterns can determine if the lattice expands or contracts with changing RH and/or temperature. Therefore, the gravimetric moisture sorption/desorption studies (Figure a) were correlated with structural changes to HyA using variable-humidity PXRD at 25 °C (Figure a). There is hardly any change in the peak positions and packing features of HyA with varying RH (Figure ), which explains the rapid equilibration of the moisture in the lattice with the surrounding RH in the water sorption/desorption isotherms (Figure a).

Figure 3

(a) Moisture-dependent PXRD measurements of HyA. Numbers on the y axis indicate the moisture in % at which the powder pattern was recorded. (b) Packing diagrams of HyA highlighting the water oxygen positions (W1–W4) at different HyA hydration states. Fractions correspond to water occupancies and were derived from Rietveld refinements (Table S5 of the Supporting Information). For clarity, water hydrogen atoms are omitted in (b).

Changes in lattice parameters were quantified by indexation and Rietveld refinement of the HyA PXRD patterns recorded at different RH values (Table S5 of the Supporting Information). The lattice parameters a, b, and c changed by less than 0.6, 0.1, and 0.3%, respectively, and the cell volume by less than 0.3%. The PXRD patterns recorded at the lowest RH values differ primarily from the patterns recorded at RH conditions ≥ 10% in the intensities of low-angle peak positions (Figure a, Figure S3 of the Supporting Information), reflecting differences in occupancies of water molecule positions in the crystal lattice. The derived occupancy parameters for the four water molecules (Figure b), W1–W4, correlate well with the mole fractions measured in Figure a. Thus, fractional occupancies and water positions could be used to derive structural information about water egress. The occupancy parameters collectively decreased at lowest RH values, indicating water mobility along the water channels, parallel to the crystallographic b axis (Figure c). Even at 2% RH, all four water positions were partially occupied. Small positional changes could be seen for W1 and W3, but only at low RH values (≤ 6% RH, Figure b). Water mobility in HyA is further discussed in section . It was possible to produce an isomorphic dehydrate of HyA (HyA), although, upon exposure to moisture, rehydration to HyA could not be prevented. Long-time storage experiments of HyA at RH values < 40% (25 °C) resulted in a slow transformation to AH, and at the highest RH (≥ 92% RH), a slow transformation to HyC was observed.

Hydrate B

Brucine HyB (tetrahydrate) is fairly stable at 25 °C, exhibiting practically no weight loss between 90% and 10% RH (Figure a). At RH values below 10%, the dehydration is indicated by a single step and results in amorphous brucine. Upon increasing the RH, amorphous brucine shows an accelerating water uptake and crystallization of a mixed hydrate phase, consisting of HyC, HyB, and not further characterized phase(s), occurred at an RH of 85% RH (Figure a). One of the unidentified phases likely corresponded to the 5.25-hydrate (UCOJIG[60]). Long-time storage experiments (up to 6 months) of amorphous brucine at RH values < 43% resulted in AH, whereas, at the highest RH conditions, a fast crystallization to mainly HyC and HyB is observed.

Figure 4

(a) Gravimetric moisture sorption and desorption curves of HyB/amorphous brucine at 25 °C. (b) Moisture-dependent PXRD measurements starting from HyB. Numbers on the y axis indicate the moisture in % at which the powder pattern was recorded. Dotted lines in (b) indicate the presence of other not further characterized phase(s). B and C denote characteristic low-angle reflections of HyB and HyC, respectively.

Indexation and Pawley fitting of the HyB PXRD patterns recorded at different RHs (Figure , Figure S4 of the Supporting Information) showed that the lattice parameters do not change upon decreasing the RH from 90% to 10%, thus, indicating the presence of a stoichiometric tetrahydrate. For more details, see section 4.2 of the Supporting Information.

Hydrate C

The moisture desorption isotherm of brucine HyC shows that the hydrate is stable only at RH ≥ 55% (Figure a). In the RH range between 95% and 55%, the desorption isotherm of HyC shows a slight and constant decrease in water content from 3.65 to 3.85 mol of water per mol of brucine. At RH < 55%, the entire hydrate water is released in one step and the resulting phase corresponds to AH. Brucine AH is stable up to 75% RH. At higher moisture conditions (≥ 80% RH), the sample takes up water and a transformation to HyC occurs. The distinct steps and hysteresis between the sorption and the desorption isotherms are characteristic for a phase transformation, typically observed for stoichiometric hydrates. The PXRD measurement at 95% RH confirmed the presence of HyC after the sorption cycle.

Figure 5

(a) Gravimetric moisture sorption and desorption curves of brucine HyC ↔ AH at 25 °C. (b) Moisture-dependent PXRD measurements of HyC. Numbers on the y axis indicate the moisture in % at which the powder pattern was recorded. Due to different equilibration times and other parameters such as sample amount, dynamics of the atmosphere, etc., the hydration rates in the gravimetric moisture chamber (GMS) are different from kinetics in the moisture stage (VGI) used for the PXRD recordings.

Variable-humidity PXRD at 25 °C (Figure b) confirms the nonstoichiometric behavior of HyC at higher RH values (≥ 50%), indicated by slight shifts in the peak positions. The PXRD pattern was successfully indexed to a monoclinic unit cell (80% RH, 25 °C: a = 25.018 (1) Å, b = 12.381(<1) Å, c = 17.459 (<1) Å, β = 122.20(<1)°). Similar to HyA, a small change in cell volume of less than 0.3% could be measured. For more details, see Table S7 of the Supporting Information. Upon decreasing the RH to ≤ 40%, the transformation of HyC to AH occurs in the PXRD experiments. The reversible back transformation, AH to HyC, was observed at the highest RH.

Figures , 4, and 5 show the sorption/desorption behavior of phase pure brucine solid forms. Additional experiments, using binary and ternary mixtures of brucine phases, were performed to investigate the influence of solid form mixtures on the transformation kinetics. Such mixtures may result in very complex moisture sorption isotherms due to overlapping processes, which can only be interpreted with the results presented in section and complementary methods, such as PXRD. These studies are still in progress and will be presented elsewhere.

Water Diffusion in HyA Monitored Using H/D Exchange

A fast egress/ingress of the water of hydration of HyA is indicated by the moisture-dependent studies (Figure a). Therefore, HyA was exposed to D2O vapor (∼98% RH) and monitored by Raman spectroscopy at different time points to investigate the water dynamics in the hydrate. Stretching vibrations of the water molecules, ν(O–H), are located in the region from 3700 to 3100 cm–1 (IR spectra, Figure S8 of the Supporting Information) and ν(C–H) in the region between 3000 and 2800 cm–1 for HyA (Figure ). Stretching modes of D2O are seen in the range between 2600 and 2300 cm–1. The emergence of ν(O–D) bands in the HyA spectra on exposure to D2O confirms that H2O can be exchanged by D2O and that HyA water molecules are mobile, despite forming strong hydrogen bonding interactions. Water diffusion in and out of the HyA structure is rapid, as even after only an hour exposure time ν(O–D), vibrations are visible. H2O/D2O exchange is slower if structural rearrangements are required for the vapor egress/ingress; i.e., if water molecules are not located in open channels (e.g., DB7(z): HyA[21]). Thus, the variable RH experiments (Figure a) and H2O/D2O studies (Figure ) showed that the water of hydration can be released and exchanged very quickly from/within the HyA framework, even at RH > 40%.

Figure 6

Raman spectra of brucine HyA as a function of time exposure to D2O vapor (∼98% RH). Peaks due to O–D stretching vibrations emerge over the course of a few hours and are highlighted in yellow.

Determination of the Critical Water Activity (Slurry Method) and Long-Time Stability Experiments

Amorphous brucine was added to methanol/water mixtures of various compositions (section 2 of the Supporting Information) and equilibrated under stirring for 1 week at 10, 25, and 40 °C. Samples were withdrawn periodically and analyzed with PXRD. Slurry experiments performed at 10 °C resulted in three distinct solid forms (Figure ). At a water activity (aw) ≤ 0.1, the dimethanol solvate (JIFWEB[61]) was the only phase at equilibrium. In the aw range starting from 0.2 and up to ≤ 0.8, HyA was identified as the phase in equilibrium, whereas, at aw ≥ 0.9, HyB was obtained as the most stable form. Increasing the temperature to 25 °C resulted in three different solid forms. At low water activities (aw ≤ 0.3), AH is the phase at equilibrium, whereas, between aw ≥ 0.4 and ≤ 0.8, HyA is the stable form. Thus, the transition temperature between the dimethanol solvate and AH lies in between 10 and 25 °C. Finally, at aw ≥ 0.9, HyC remains the only phase at equilibrium. At 40 °C, only two solid forms were obtained in the slurry experiments. At aw ≤ 0.3, AH is formed in the slurry, and at aw ≥ 0.4, HyA is the stable form. The enantiotropically related low temperature anhydrate phase (AH) was not observed in the slurry experiments, indicating that the thermodynamic transition point between the polymorphic pair AH/AH is lower than 10 °C, in agreement with the DSC results reported by Bialonska et al.[58] To conclude, by carefully choosing aw (water/methanol mixtures) and temperature conditions, it is possible to produce three of the brucine hydrates, AH, and the dimethanol solvate as phase pure samples. Thus, at ambient temperature, three hydrated forms of brucine and the anhydrate can exist as the thermodynamically most stable forms depending on the water vapor pressure. Therefore, if crystallization conditions are not chosen carefully, it is likely that mixtures of solid forms, showing different stability ranges and transformation behaviors, are obtained.

Figure 7

Phase diagram after equilibration for 1 week showing the dependence of brucine solid forms on water activity/relative humidity at 10, 25, and 40 °C.

The slurry experiments in water/methanol mixtures (Figure ) provide the thermodynamic stability ranges and transition points of the brucine phases, whereas the gravimetric moisture sorption/desorption studies display the kinetically stabilized existence ranges of the forms. The desorption experiments, performed at 25 °C, resulted in metastable phases, namely, amorphous brucine (Figure ) and HyA (Figure ). The latter two phases were found to slowly transform to AH, if stored at RH values ≤ 31% (Table ), in agreement with Figure . Similarly, upon increasing the RH of amorphous brucine (Figure ), not only the thermodynamically stable HyC (at the highest RH conditions) at 25 °C was obtained but also HyB and other not further characterized hydrate phase(s). The latter phases and HyB very slowly transform to HyC if stored at 98% RH (Table ). The presence of seed crystals (i.e., more stable form) accelerated any of the metastable-to-stable form transformations in the solid state.

Table 1

Long-Time Storage Experiments (6 Months) of Brucine Solid Forms at Defined RH Conditions and 25 °C

starting form(s)a	RH/%	after 6 monthsa,b
amorphous	≤31	AH
HyAdehy	≤31	HyA + AH
AH	≤31	AH
AH	43	AH ≫ HyA
HyA	43	HyA
HyA + HyB	43	HyA > HyB
HyA + HyC	43	HyA + AH
HyC	43	AH ≫ HyAc
AH	52	AH + HyA
HyA	52	HyA
HyB	52	HyB + HyA
HyC	52	AH + HyAc
AH + HyA	75	HyA
amorphous	75	HyA
HyA + HyB	75	HyA
HyA + HyC	75	HyA > HyC
AH	92	HyC
amorphous	92	HyC + HyB ≫ unknown
HyA + HyB	92	HyA + HyB + HyC
HyC	92	HyC
HyA	98	HyA ≫ HyC
HyB	98	HyB ≫ Hy

AH - anhydrate; HyA - hydrate A (dihydrate); HyB - hydrate B (tetrahydrate); HyC - hydrate C (3.85-hydrate); HyAdehy - isomorphous HyA dehydrate.

Quantified using PXRD: x ≫ y - less than 5% y; x > y - less than 20% y; x + y - similar amounts or ± 20%.

Transformation to HyA via AH.

Temperature-Dependent Stability of Brucine Hydrates

The TGA curve of HyA (Figure , curve (i), shows a one-step mass loss of 8.06 ± 0.10%, corresponding to 1.92 mol equiv of water. The dehydration process starts immediately under dry conditions (N2 purge), explaining why the measured mass loss was slightly lower than the theoretical value for a dihydrate stoichiometry (calculated: 8.37% weight loss relative to wet sample).

Figure 8

Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) of HyA. The TGA curve (i) was recorded in a pan covered with a one pinhole lid at a heating rate of 5 °C min–1. The DSC curves were recorded in pans with five pinhole lids and heating rates of 3 °C min–1 (ii) and 5 °C min–1 (iii and v), respectively, or a sealed pan (iv, v) at a heating rate of 5 °C min–1. (v) DSC curve of AH.

DSC experiments using five pinhole lids show three events: (1) The dehydration of HyA to amorphous brucine, as confirmed with PXRD and Raman spectroscopy. The process starts above 80 °C and is finished below 120 °C if heating rates ≤ 5 °C min–1 are applied (curves ii and iii). (2) Above 120 °C, an exothermic event is recorded, corresponding to the crystallization of AH. (3) Upon further heating, the melting of AH is observed as an endothermic event with an onset temperature of 178.9 ± 0.1 °C (ΔfusH = 28.4 ± 0.1 kJ mol–1). By using hermetically sealed DSC pans (curve iv), the peritectic melting of HyA at 121.5 ± 1.0 °C with a heat of dissociation (ΔdissH) of 31.4 ± 0.8 kJ mol–1 can be measured.

Hydrate B

For HyB, the TGA curve reveals a one-step mass loss of 15.32 ± 0.03%, corresponding to 3.96 mol equiv of water (Figure , curve i). DSC experiments performed in open pans (no lid) show four thermal events: (1) The dehydration of HyB to amorphous brucine at temperatures < 80 °C, (2) the glass transition at about 90 °C, (3) crystallization of brucine AH at temperatures above 130 °C, and (4) the melting of AH at 178.9 ± 0.1 °C (curve iii). The peritectic transformation of HyB to HyA can be measured in sealed DSC pans (curve iv). The first endotherm corresponds to the peritectic transformation (Ttrs( = 68.9 ± 0.5 °C, ΔtrsH = 22.6 ± 0.1 kJ mol–1) and the second endotherm to the peritectic dissociation of HyA. Curve v (amorphous brucine) shows a glass transition, followed by the crystallization and melting of AH.

Figure 9

Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) of HyB. The TGA curve (i) was recorded in an open pan at a heating rate of 5 °C min–1. The DSC curves were recorded in open pans at heating rates of 2 °C min–1 (ii) and 5 °C min–1 (iii), respectively, or a sealed pan (iv, v), at a heating rate of 5 °C min–1. (v) DSC curve of amorphous brucine. Dashed ellipsoids in (ii, iii, and v) indicate the glass transition.

The dehydration process of HyC starts immediately under the dry atmospheric conditions (N2 purge) in the TGA. The measured mass loss of 13.75 ± 0.05%, corresponding to 3.49 mol equiv of water (Figure , curve i), is lower than the mass loss derived from the (de)sorption experiments (Figure a). This can be related to the fact that HyC is stable only at RH values ≥ 55%.

Figure 10

Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) of HyC. The TGA curve (i) was recorded in a pan covered with a one pinhole lid at a heating rate of 5 °C min–1. The DSC curves were recorded in pans with five (ii) or one (iii) pinhole lids and heating rates of 2 °C min–1 (ii) and 5 °C min–1 (iii and v), respectively, or a sealed pan (iv, v) at a heating rate of 5 °C min–1. (v) DSC curve of AH.

DSC thermograms recorded in pans covered with pinhole lids show the dehydration of HyC to AH (first endotherm, curves ii and iii) and the melting of AH at 178.9 ± 0.1 °C (second endotherm). The peritectic transformation of HyC to HyA can be measured in sealed DSC pans (curve iv). The first endotherm corresponds to the peritectic transformation (Ttrs( = 86.4 ± 0.9 °C, ΔtrsH = 21.8 ± 0.6 kJ mol–1) and the second to the peritectic dissociation of HyA.

Thermodynamic Stability and Heat of Transformations

The enthalpy of a hydrate to anhydrate transition can be estimated from DSC or IC (with the aid of an RH perfusion cell) experiments. The dehydration enthalpy, ΔdehyH, measured in open DSC pans (open or pinhole lid), can be subdivided (application of Hess’s law) into the enthalpy of hydrate-to-anhydrate transformation, ΔtrsH, and that of the vaporization of the expelled moles of water (ΔvapHH).[20,73−76]If we subtract the known enthalpy value for the vaporization of water at the dehydration temperature[77] from the measured ΔdehyH, a ΔtrsH of 20.0 ± 1.4 kJ mol–1 is obtained for the HyC → AH phase transformation (Table ). Dehydration of HyA and HyB results in amorphous brucine; thus, the heat of crystallization (exothermic) of amorphous brucine to AH, which corresponds to −ΔfusH, has to be added to ΔdehyH to estimate ΔdehyH. This results in ΔtrsH and ΔtrsH of 2.4 ± 1.0 kJ mol–1 and 22.9 ± 1.2 kJ mol–1, respectively (Table ). ΔtrsH could also be estimated from the difference of ΔdissH and ΔfusH to be 3.0 ± 0.8 kJ mol–1. The latter energy agrees with the open DSC dehydration results. Thus, from the DSC dehydration experiments, the 0 K stability order of the three hydrates was estimated as follows: HyB (most stable) > HyC ≫ HyA (least stable).

Table 2

Physicochemical Data for Amorphous Brucine, Anhydrate (AH), and Hydrates (HyA, HyB and HyC)

phase	Tfus/°C	ΔfusH/kJ mol–1	Ttrs/°C	ΔtrsH/kJ mol–1	phase after transformation	methodg
AH	178.9 ± 0.1	28.4 ± 0.1			melt	DSC
HyA	121.5 ± 1.0	31.4 ± 0.8		3.0 ± 0.8a	AH	DSC (closed)
HyA			ca. 110	28.3 ± 0.7	amorp. + AH (traces)	DSC (open)
amorph. + AH (traces)			ca. 135	–25.9 ± 0.3	AH	DSC (open)
HyA				2.4 ± 0.8b	AH	DSC (open)
HyA			25	5.3 ± 0.9	HyAdehy	RH-Perf
AHdehy				–2.9 ± 1.2c	AH	DSC + RH-Perf
HyB			68.9 ± 0.5	22.6 ± 0.1	HyA	DSC (closed)
HyB			ca. 60	22.9 ± 1.2d	AH	DSC (open)
HyC			86.4 ± 0.9	21.8 ± 0.6	HyA	DSC (closed)
HyC			ca. 60	20.0 ± 1.4	AH	DSC (open)
HyC			25	19.2 ± 0.4	AH	RH-Perf
HyB				0.8 ± 0.6e	HyC	DSC (closed)
HyB				3.6 ± 1.3f	HyC	DSC + RH-Perf

ΔfusH(HyA) – ΔfusH(AH).

ΔtrsH(HyA-amorphous) + ΔtrsH(amorphous-AH).

–ΔtrsH(HyA-HyAdehy) + ΔtrsH(HyA-AH).

ΔtrsH(HyB-amorphous) + ΔtrsH(amorphous-AH).

ΔtrsH(HyB-HyA) – ΔtrsH(HyC-HyA).

ΔtrsH(HyB-AH) – ΔtrsH(HyC-AH).

DSC - differential scanning calorimetry; open - open DSC pan or covered with pinhole lid; closed - closed or high pressure DSC pan; RH-Perf - isothermal calorimetry with the aid of an RH-perfusion cell.

DSC experiments in hermetically sealed pans can provide transformation temperatures and transformation enthalpies.[78] In the case of two hydrates with the same (or almost the same) stoichiometry, the transformation enthalpy corresponds to the enthalpy difference between these two phases. However, in the case of a change in stoichiometry, the measured enthalpy value also includes an unknown contribution from the enthalpy of solution of a fraction of the dehydration product in the liberated water. The low water solubility of brucine (<0.0008 g mL–1)[44] allows one to estimate the ΔtrsH (HyX = HyB, HyC) directly in a closed DSC pan, despite HyA and HyB/HyC having different hydration states. The measured ΔtrsH values of 22.6 ± 0.1 kJ mol–1 and 21.8 ± 0.6 kJ mol–1 for the HyB → HyA and HyC → HyA transformation (Table ), respectively, agree with the 0 K stability order derived in open DSC experiments; i.e., HyB is slightly more stable than HyC at absolute zero.

With IC, the enthalpy of dehydration (ΔdehyH) and hydration (ΔhyH) can be determined. Since the magnitude of the heat of condensation of water (ΔcondHH) is equal to the heat of vaporization of water, the transition energy of AH to HyX (ΔtrsH) can be estimated according to eq :Using a value of ΔvapH°H (25 °C) of 43.99 kJ mol–1 for −ΔcondH°H (25 °C)[77] gives a ΔtrsH of 19.2 ± 0.4 kJ mol–1 for the HyC → AH transformation (Table , Figure S11 of the Supporting Information). The IC derived value is in good agreement with the enthalpy measured with DSC (Table ). It was also possible to measure, for the first time, the enthalpy of a hydrate to an isomorphic dehydrate “transformation”. This allows an estimation of the energy contribution of the water···brucine interactions in HyA. Using eqs 1 and 2 gives a “transformation” energy (ΔtrsH = −ΔtrsH) of 5.3 ± 0.9 kJ mol–1.

The transformation enthalpy of HyA to AH (3.0 ± 0.8 kJ mol–1, 2.4 ± 0.8 kJ mol–1, Table ) is distinctly lower compared to measured transformation energies of stoichiometric dihydrates to the corresponding anhydrates (e.g., phloroglucinol: 19.1 kJ mol–1,[73] barbituric acid: 17.0 kJ mol–1,[74] DB7(z): 16.9 ± 0.5 kJ mol–1,[21] codeine HCl: 28.7 ± 0.5 kJ mol–1,[20] etc.). It may be expected that the dehydration of a nonstoichiometric hydrate consumes less energy than the dehydration reaction of a stoichiometric hydrate. A slightly higher energy contribution may be expected from the water molecules in HyA (a Z′ = 2 structure) as three water molecules interact with brucine via hydrogen bonds (Figure S1 of the Supporting Information). A rule of thumb range for the energy associated with a hydrogen bond is 6–30 kJ mol–1.[79−81] The fact that the water···brucine interactions contribute only 5.3 ± 0.9 kJ mol–1 to the lattice energy of HyA may explain why water egress/ingress is facile (no hysteresis in the sorption/desorption isotherms) and agrees with the water mobility derived in D/H exchange experiments.

On the basis of the calorimetric experiments (DSC and IC) and water activity determinations, neglecting the different hydrate stoichiometries, it can be concluded that the hydrate pair HyB and HyC is enantiotropically related, with HyB being the low temperature form (stable hydrate at temperatures < 15 °C and aw > 0.8). HyC is the most stable form in the temperature range of >15 °C and < 35 °C, albeit only at aw > 0.8. At lower aw (but still ≥ 0.4) or higher temperatures (< 121.5 °C), HyA becomes more stable than HyB and HyC. At lower RH/aw or above the peritectic dissociation temperature of HyA, anhydrous brucine, AH, is the thermodynamically stable form. AH and HyA were estimated to differ only by a few kJ mol–1 in lattice energy, the energy range observed for polymorphs, thus, rationalizing why HyA can be trapped as an experimental (intermediate) “phase”.

Conclusions

The alkaloid brucine exists in at least 4 hydrate forms,[59,60] 16 solvates,[57,61,62] 2 anhydrates,[57,58] and an amorphous form. The thermodynamic and kinetic stabilities and interrelation pathways of the three practically relevant hydrates (HyA, HyB, HyC) and the water-free forms, occurring at room and higher temperatures, of this compound have been unraveled. Depending on the water vapor pressure, all three investigated hydrates and the anhydrous form can become the thermodynamically most stable form at ambient conditions, which is an extraordinary property of this compound. The tetrahydrate (HyB) and 3.85-hydrate (HyC) are enthalpically stabilized by approximately 23 and 19 kJ mol–1, respectively, with regard to the anhydrate. The enthalpic stabilization of HyA is much smaller (3 kJ mol–1). The water···brucine interactions account only for 5 kJ mol–1 of the HyA lattice energy, and the isomorphic HyA dehydrate structure was estimated to be only a few kJ mol–1 less stable than AH, rationalizing the rapid water egress/ingress and why HyA can exist without water.

An overview of the possible (de)hydration and phase interrelations of the brucine solid forms observable during storage at temperatures > 10 °C is shown in Figure . The isomorphic dehydrate of HyA and amorphous brucine are intermediates observed upon dehydration. Phase impurities (other brucine solid state forms) influence the transformation kinetics significantly.

Figure 11

Flowcharts showing the dehydration (a), hydration (b), and interrelation pathways of brucine solid forms upon storage.

The investigated hydrates clearly demonstrate that the subdivision of hydrates (or solvates in general) into two main classes, stoichiometric and nonstoichiometric, may not always be straightforward, in particular if gravimetric moisture sorption/desorption data are not available. This problem is illustrated by HyA, where the water molecules show clear hydrogen bonds to the host molecule. Moreover, the hydrate shows a constant dihydrate stoichiometry in a wide humidity range (> 40% RH), which is a typical feature of stoichiometric hydrates. However, below 40% RH, HyA shows the typical (de)hydration behavior of nonstoichiometric hydrates; i.e., (a) the amount of water in the structure depends on the water vapor pressure (and temperature), (b) the process is highly reversible and lacks a clear hysteresis between sorption and desorption, and (c) the overall structure remains more or less unaffected during the release of water molecules from defined positions. In contrast, the sorption/desorption isotherms of anhydrous brucine and HyC show distinct steps and a clear hysteresis between the sorption and desorption process, which is characteristic for stoichiometric hydrates. Nevertheless, in a limited RH range (55–100% RH), HyC also shows features of a nonstoichiometric hydrate. HyB is a stoichiometric tetrahydrate, whose structure collapses to an amorphous phase upon dehydration.

This work represents a valuable case study in characterizing (pharmaceutical) hydrates and may be seen as a “recipe” for how different, but relatively common, analytical techniques in the field of solid state characterization can be used complementarily to successfully unravel the temperature- and moisture-dependent stability (order) and interconversion pathways of even complex hydrate systems. This information is fundamental for any (industrial) production and for choosing proper storage conditions. Moisture-dependent stabilities can be quickly accessed with automated gravimetric moisture sorption/desorption studies (dynamic vapor sorption studies), but these measurements provide kinetic data and not the thermodynamic stability data. The latter can be derived from long-time stability experiments but more quickly using slurry water activity measurements. Thus, moisture sorption/desorption studies alone cannot replace the long-time stability or water activity measurements and vice versa. Furthermore, only by complementing the sorption/desorption studies with environmental powder X-ray diffraction or vibrational spectroscopy (Raman, IR) is it possible to interpret at a structural level the interconversion (hydration and dehydration) pathways relating solid forms at each step in the moisture sorption/desorption isotherms. Isothermal (RH-perfusion) calorimetry or differential scanning calorimetry may be exchangeable in quantifying energy differences between solid forms, but this will strongly depend on the system and how easily a transformation can be induced. Thermal analysis (DSC, TGA, HSM) will quickly and easily provide information about temperature-dependent stability. To conclude, the number of approaches can be reduced if only preliminary information about a system is surveyed, but to achieve the level of understanding, as in this work, it is recommended that at least thermal, water activity, and moisture sorption/desorption studies are performed. The complementarity of the latter approaches allows one to unravel the interplay of temperature and water activity that again determines the stability ranges of hydrates and water-free forms. Because, only based on such information and knowing the transition conditions and pathways, one can avoid complications during processing, storage, and handling of any fine chemical forming complex hydrate phases.