2.1 Materials and Methods

All reagents for synthesis and analysis were of analytical grade and were used as received without further purification. MET·HCl 1 was purchased from Tokyo Chemical Industry Co. All other reagents were obtained from Sigma Aldrich Chemical Co. and the solvents were purchased from JT Baker. The disodium succinate salt was obtained by vigorous grinding (agate mortar and pestle) of succinic acid (H₂SUC) in the presence of NaOH (1:2 molar ratio). The reaction is exothermic and water is produced during the process affording a white paste. This paste was further grinded for 10 min. affording a white powder that were desiccated in an oven at 110º C for 4 h.

Na₂[SUC]. IR-ATR (cm^-1). 3361 (w), 3250 (w), 1669 (m), 1544 (s), 1401 (s), 1237 (w), 1157 (w), 816 (m), 729 (s), 663 (s), 552 (s). ssCPMAS ¹³C NMR(ppm). Na+-OOC-*CH₂-*CH₂-COO-Na+ (δ = 34.74), Na+-OOC*-CH₂-CH₂-C*OO-Na+ (δ= 181.60).

2.2 Synthesis by Melt Mixing

MET·HCl, 1 (165.6 mg, 1.0 mmol) and dehydrated disodium succinate (81 mg, 0.5 mmol) were melted at 220 °C for 5 min. and the resulting white powder was ground for 10 min at room temperature. Suitable single crystals of 2 for X-Ray diffraction were obtained by slow evaporation of saturated methanolic of 1 at room temperature. Figures of the crystalline structures of 2 were generated by the programs X-Seed ^[13], and Olex 2 ^[14].

2.3 Spectroscopic characterization and thermal analyses

Infrared spectra were recorded on a Bruker Tensor 27 ATR spectrophotometer and measured in the range of 4000-400 cm⁻¹. Melting points of 1 and 2 were determined using a Fisher Johns apparatus and are reported without correction. Thermogravimetric analysis and differential scanning calorimetry (TGA-DSC) were performed on a NETZSCH STA 449F3 apparatus. The analyses were performed at a heating rate of 10 ºC/min within the temperature range of 30-400 ºC using a current of nitrogen as inert gas purge (10 mL/min). Solid-state ¹³C NMR spectra of 1 and 2 were recorded at 298 K on a Varian 300 spectrometer operating at Larmor frequencies of 75.4, 79.3. The ¹³C experiments were performed with cross-polarization (CP), high power decoupling, and magic angle spinning (MAS) and contact time t_CP 1500, using a Bruker double-bearing probe head, spinning speed 4.5 KHz.

2.4 Crystallographic investigations

Powder X ray diffraction experiments were performed in a Bruker D500 diffractometer operated at 40 kV and 35 mA with Cu Kα radiation (1.5406 Å), radius of 217 mm, divergence slit 0.6 mm and anti-scattering slit 0.1 mm, with soller of 2.6º, graphite monocromator and scintillation detector. The range of measurement was from 5-75º.

The relevant details of 2, data collection and structure refinement can be found in Fig. 2. Data collection for 2 was performed at 298 ºC using MoKα radiation on a Bruker APEX II CCD instrument. Semi-empirical absorption correction was applied to all data sets ^[15]. Structure solutions, refinements, and geometrical calculations were carried out by SHELXTL ^[16]. The structure was refined using full-matrix least squares methods on F2 with all non-H atoms anisotropically defined. The hydrogen atoms were placed in calculated positions using the “riding model” with Uiso = aUeq (where a is 1.5 for -CH₃ and -OH moieties and 1.2 for others).

Fig. 1 DSC thermograms of 1 and 2. The peaks in the figure are related to the first thermal event (left). TGA profiles of 1 and 2 (right). 

Fig. 2 Unit cell of 2. Ellipsoids were drawn at 50% of probability. Crystal data for (2): formula C₆H₁₄N₅O₂, F.W. 188.22, Monoclinic, P2₁/n, a 9.5551(3) Å, b 8.8279(3) Å and c 10.8487(3) Å, α 90º, β 92.4900(10)º and γ 90º, V 897.03(5) ų, Z = 4, ρcalc 1.367 (g cm^-3), μ(MoKα) 0.105 mm^-1, F(000) 404, crystal size 0.374 Å x 0.22 Å x 0.18 Å mm³, 2Ф range 5.562-50.576º, T 298(2) K, Reflections collected 7485, Independent reflections 1657 [Rint= 0.0282, Rσ= 0.0211], data/restrains/parameters 1657 / 6 / 139, GOF on F² 1.073, Final R indexes [I>= 2σ (I)] R1= 0.0345, wR2= 0.0909, Final R indexes [all data] R1 = 0.0408, wR2 = 0.0971, Largest diff. peak / hole / eÅ^-3 0.29/-0.21. 

3.1 DSC-TGA studies

Evidence for the formation of the salt 2 was initially obtained from the DSC and TGA studies. DSC experiments were used to follow the progress of the reaction. The DSC thermograms of 1 and 2 are shown in Fig. 1. The peak of 2 at 243 °C does not correspond to the melting point of either of the starting materials ^[17-18], indicating the formation of a new solid phase. The relevant thermal data are collected in Table 1.

Thus the results obtained for 2 clearly show the formation of a new salt with higher thermal stability, properly fitting with those reported for other co-crystals/salts containing carboxylic acid functional moieties ^[7e]. The DSC profile of 2 exhibits one thermal event at 210-250 ºC due to the melting of the salt. The second process, in the range 250-290 ºC, can be attributed to the decomposition of two H-MET⁺ cations followed by the decomposition of SUC^2-. This fact is supported by simultaneous TGA measurements.

3.2 Crystal structure

The salt 2 crystallizes in the monoclinic P2₁/c space group (Fig. 2). The asymmetric unit includes one H-MET⁺ and half molecule of SUC^2-, in agreement with the stoichiometry ratio found by DSC-TG analysis (Fig. 2).

So far, there are three tautomeric forms reported for the cation H-MET⁺ in the solid state (Scheme 2). In our case, the H-MET⁺ in 2 corresponds to tautomer (c). Thus, the crystal structure shows the C-N bond distances in H-MET⁺ to be in the range of 1.330(2)-1.453(2) Å, these values are intermediate between a single and a triple bond ^[19-20], indicating a delocalized π electron system across the molecule. The H-MET⁺ cation is non-planar with a torsion angle C3-N2-C4-N5 of 144.55º.

Scheme 2 Tautomers of H-MET⁺ 

The H-MET⁺ is linked to SUC^2- via different hydrogen bonds with different motifs (Fig. 3). Table 2 contains the geometric parameters for selected hydrogen bonds. The most relevant motifs identified (based on graph set theory ^[21]) are guanidinium···guanidinium [dimer R² ₂(8) motif I], charged-assisted hydrogen bonds guanidinium···carboxylate [dimer R² ₂(8) motif II] and N-H···O^- (carboxylate) [tetramer R² ₄(8) motif III] (Fig. 3, Table 2).

Fig. 3 Hydrogen bonding interactions in crystal structure of 2. 

Motifs I-III produce infinite layers along the a-axis (Fig. 3). In this case, motif I is perfectly aligned between the planes N2-C3-N3-3HA and N2-C3-N3-3HA. However, motif II exhibits a deviation of 25.93º between the plane O2-C1-O1 (carboxylate) and the plane H1B-N1-C3-N3-H3B. A comparison of this value with similar structures deposited at CSD (6.5º aminoguanidinium hydrogensuccinate-GOLXUB ^[22], 12.45º biguanidium hydrogen succinate-PUNPUK ^[23], 9.27º guanidinium hydrogen succinate-SEWXAU ^[24], 6.46º 2-phenylbiguanidinium hydrogen succinate methanol solvate-WADTEC ^[25]) reveal a significant deviation of motif II in 2. This probably being due to the juxtaposition of motif II with motif III in the crystal lattice (Fig. 3). Thus constraining the spatial approaching of the carboxylate and guanidinium moieties; moreover, in motif III the tetramer R² ₄(8) is completely planar.

Other identifiable motifs in the structure, such as N-H···O^- (carboxylate) [R² ₄(12) formed by N3-H3B···O1 2.051(13) Å, N4-H4A...O1 2.001(12) Å and N4-H4B···O1 2.154(13) Å] and N-H···O^- (carboxylate) [R² ₂(13) formed by N1-H1A···O2 2.045(15) Å and N4-H4A···O1 2.001(12) Å] allow the extend of the 3D arrangement along the c-axis.

3.3 Spectroscopic analysis

Studying the infrared spectra in two sub-regions allows the identification of hydrogen bonding in the solid state: 3550-3250 cm^-1 for N-H stretching vibrations and 1650-1581 cm^-1 for N-H deformation vibrations ^{[17, 26]}. It has been reported that when hydrogen bonding (X-H···Y) occurs in solid state, the X-H stretching vibrations decrease in frequency and conversely the X-H deformation vibrations shift to higher frequencies ^[27]. Additionally, hydrogen bonding can be associated with the broadening and enhance of intensity of X-H stretching vibrations ^[27]. Thus, in order to provide further evidence for the possible presence of motif II, we analyzed comparatively these regions and vibrational bands in the IR spectra of 1 and 2 (Fig. 4, Table 3).

Fig. 4 Infrared spectra of 1 and 2. 

N-H stretching vibrations, although special attention was placed in the bands at 3368 cm^-1 N-H stretching, 3292 cm^-1 asymmetric N-H stretching and 3149 cm^-1 symmetric N-H stretching of 1 (Table 3) ^[26], a straightforward comparison with 2 was not possible due to the broadness of the bands found. In the case of the N-H deformation vibrations, for 2 they shift to higher frequencies (Table 4) compared with those of 1 at 1623 and 1555 cm^{-1 [(26]}. This information, in combination with the band at 1650 cm^-1 due to large proton polarizability and that due to the unbound -COO^- at 1595 cm^{-1 [(28]}, may suggest the predominant presence of hydrogen bonding (motif II) in 2 (Table 3).

3.4 Powder X-Ray difraction (PXRD)

The collection of the PXRD measurements of 1-2 is shown in Fig. 5, Na₂[SUC] and simulated of 2. The experimental and simulated PXRD measurements of 2 are practically identical, and different when compared to 1 and Na₂[SUC] salt.

Fig. 5 X-Ray powder diffractions of starting salts (1 and Na₂[SUC]) and experimental and simulated diffraction of product 2. 

3.5 ssCPMAS ¹³C NMR analysis

The ssCPMAS ¹³C NMR spectra of 1, 2 and Na₂[SUC] were determined (Fig. 6, see Scheme 3 for the atom numbering of structure). The spectrum of 1 shows three signals, at δ 38.5, 40.5 and a broad one at δ 157.6 and 160.9 ppm. The up field signals (δ 38.5 and 40.5 ppm) correlate well with the solution NMR data (δ 38.4 ppm in D₂O)¹⁷. The appearance of the two signals in the solid-state experiment allow to distinguish two different -CH₃ groups. This observation agrees well with the results observed in the X-ray crystal structure of Metformin hydrochloride [CSD refcode JAMRIY01, polymorph A²⁰], where one methyl group has the interaction C3-H11···Cl1 [2.897(16) Å and 159.53º] while the other shows short contacts C4-H8···C1 [2.764(17) Å and 143.95º]. The broad downfield signal corresponds to the quaternary carbons: C2 (δ 160.9 ppm) and C3 (δ 157.6 ppm).

Fig. 6 ssCPMAS ¹³C NMR spectra of 1, Na2[SUC] and 2. 

Scheme 3 Numbering of carbon atoms of 1 and Na₂[SUC] for the assignation of chemical shifts. 

For the case of Na₂[SUC], assignation of the signals in the CPMAS ¹³C NMR spectra was straightforward based on comparison with the reported data ^[29]. The CPMAS ¹³C spectrum of 2 exhibits five signals, two downfield at δ 34.30, 38.45 and 40.47, which by comparison with the spectra of the precursors, the signal at δ 34.30 ppm corresponds to C4 of succinate anion and the other two intense signals are due to C1. According with the X-ray structure of 2, two distinguishable -CH₃ are noted due to short contact C6-H6A···O1 [2.702 Å and 114.10º] and C5-H5B···N1 [2.736 Å and 154.82º]. In addition, at up field two more signals are observed; the broad one, again, composed of two overlapped signals due to C2 and C3. Whilst the signal at δ 181.5 ppm is due to C5 of the succinate anion. No significant ∆δ (-0.07 ppm) is observed in this signal compared with Na₂[SUC] (Table 4).

3.6 Dissolution rates profiles

In order to explore the potential of 2, for comparative purposes, we determined the intrinsic dissolution rate constants of 2 and 1 under physiological conditions. The experiments were performed using tablets of 1 and 2 (crushed single crystals were employed), prepared with a hydraulic press at a total force of 220 kg/cm². The dissolution rates were determined with a Wood’s apparatus with phosphate buffer at pH 6.8. Experiments were carried out by triplicated at 37 ºC under constant stirring (50 rpm) in a constant volume of 900 mL. Determinations were done using a UV-vis spectrometer (Thermo Scientific Evolution 60) monitoring the absorption band at 233 nm using a validated method for MET·HCl ^[30], permitting quantify the amount of drug dissolved in a giving period of time to establish the dissolution rate profile.

The comparison of intrinsic dissolution profiles of 1 and 2 is shown in Fig. 7. One can easily observe that 1 (K_int = 2.94 mg/min*cm²) shows an initial rapid dissolution and minimal variations of this behavior during the period of time of the determinations, conversely for 2 (K_int = 0.885 mg/min*cm²) the dissolution rate increases constantly, allowing us to conclude that indeed 2 exhibits a sustained-release drug delivery behavior compared with 1. This behavior can be attributed to the markedly difference between the values of hydration energy of chloride anion (ΔG°= −340 kJ mol⁻¹) ^[30] and succinate (ΔG° = −118 kJ mol⁻¹) ^[31].

Fig. 7 Intrinsic dissolution profiles of 1 and 2.