Synthesis, Characterization, and Intrinsic Dissolution Studies of Drug–Drug Eutectic Solid Forms of Metformin Hydrochloride and Thiazide Diuretics

The mechanochemical synthesis of drug–drug solid forms containing metformin hydrochloride (MET·HCl) and thiazide diuretics hydrochlorothiazide (HTZ) or chlorothiazide (CTZ) is reported. Characterization of these new systems indicates formation of binary eutectic conglomerates, i.e., drug–drug eutectic solids (DDESs). Further analysis by construction of binary diagrams (DSC screening) exhibited the characteristic V-shaped form indicating formation of DDESs in both cases. These new DDESs were further characterized by different techniques, including thermal analysis (DSC), solid state NMR spectroscopy (SSNMR), powder X-ray diffraction (PXRD) and scanning electron microscopy–energy dispersive X-ray spectroscopy analysis (SEM–EDS). In addition, intrinsic dissolution rate experiments and solubility assays were performed. In the case of MET·HCl-HTZ (χMET·HCl = 0.66), we observed a slight enhancement in the dissolution properties compared with pure HTZ (1.21-fold). The same analysis for the solid forms of MET·HCl-CTZ (χMET·HCl = 0.33 and 0.5) showed an enhancement in the dissolved amount of CTZ accompanied by a slight improvement in solubility. From these dissolution profiles and saturation solubility studies and by comparing the thermodynamic parameters (ΔHfus and ΔSfus) of the pure drugs with these new solid forms, it can be observed that there was a limited modification in these properties, not modifying the free energy of the solution (ΔG) and thus not allowing an improvement in the dissolution and solubility properties of these solid forms.


Introduction
Pharmaceutical cocrystals are multicomponent crystalline entities composed of an active pharmaceutical ingredient (API) and a coformer in a defined stoichiometric ratio [1][2][3]. The API/coformer are held together by non-covalent interactions such as hydrogen bonds, π-π interactions and van der Waals forces. The benefits encountered in the preparation the formation of the coamorphous systems glicazide-HTZ and glicazide-CTZ have been reported, seeking to improve the dissolution properties of the thiazide drugs [57], however, with no noticeable difference between the dissolution rates of amorphous/crystalline HTZ and the release rate of HTZ from the coamorphous glicazide-HTZ [57]. However, similar IDR experiments showed that the coamorphous form HTZ-atelonol (0.5:0.5~1:1) has a K int 12.5-fold more than HTZ crystalline and 2.2-fold better than the PM HTZ-atelonol [58].
Taking this into account, we are interested in the preparation of drug-drug solid forms containing MET·HCl (classified as class III, exhibiting high solubility in water but low permeability to cell membranes) [59], in the presence of the thiazide diuretics HTZ or CTZ (Scheme 1), in order to tackle the poor solubility and limited dissolution properties of both drugs, and because many T2D patients medicated with antidiabetic drugs frequently present hypertensive complications. Thus, this paper reports the ball-milling synthesis, using neat grinding (NG) [60][61][62][63][64] or liquid-assisted grinding (LAG) [60][61][62][63][64] (varying the polarity of the solvent), and characterization [57], as well as IDR experiments to determine any modification in the dissolution properties of these solid forms compared with the pure APIs. At first, we thought we had prepared DDCs, however during characterization we discovered the formation of DDESs.
Pharmaceutics 2021, 13, x FOR PEER REVIEW 3 of 21 HTZ-atelonol (0.3:0.7~1 mol:2.5 mol) has been reported [24]. For this system, using intrinsic dissolution rate (IDR) experiments, it was found that the % release of HTZ in DDES HTZ-atelonol (0.3:0.7) compared with pure HTZ has a 10-fold improvement. Analogously, the formation of the coamorphous systems glicazide-HTZ and glicazide-CTZ have been reported, seeking to improve the dissolution properties of the thiazide drugs [57], however, with no noticeable difference between the dissolution rates of amorphous/crystalline HTZ and the release rate of HTZ from the coamorphous glicazide-HTZ [57]. However, similar IDR experiments showed that the coamorphous form HTZ-atelonol (0.5:0.5~1:1) has a Kint 12.5-fold more than HTZ crystalline and 2.2-fold better than the PM HTZ-atelonol [58]. Taking this into account, we are interested in the preparation of drug-drug solid forms containing MET·HCl (classified as class III, exhibiting high solubility in water but low permeability to cell membranes) [59], in the presence of the thiazide diuretics HTZ or CTZ (Scheme 1), in order to tackle the poor solubility and limited dissolution properties of both drugs, and because many T2D patients medicated with antidiabetic drugs frequently present hypertensive complications. Thus, this paper reports the ball-milling synthesis, using neat grinding (NG) [60][61][62][63][64] or liquid-assisted grinding (LAG) [60][61][62][63][64] (varying the polarity of the solvent), and characterization [57], as well as IDR experiments to determine any modification in the dissolution properties of these solid forms compared with the pure APIs. At first, we thought we had prepared DDCs, however during characterization we discovered the formation of DDESs. Scheme 1. Molecular structures of MET·HCl and thiazide diuretics.

Materials
All the pharmaceutical reagents were purchased from Tokyo Chemical Industry America TM (Portland, OR, USA) or Toronto Research Chemicals TM (North York, ON, Canada) and were used as received. The solvents were purchased from Tecsiquim TM (Toluca, México) and were used as received.

Materials
All the pharmaceutical reagents were purchased from Tokyo Chemical Industry America TM (Portland, OR, USA) or Toronto Research Chemicals TM (North York, ON, Canada) and were used as received. The solvents were purchased from Tecsiquim TM (Toluca, México) and were used as received.  (2-4 mg) in sealed non-hermetic aluminum pans and scanned at a heating rate of 10 • C/min from 30-400 • C under a dry nitrogen atmosphere.

Eutectic Binary Mixture Screening by DSC Data
The determination of the eutectic points was made by means of construction of binary phase and Tammann diagrams [31,65]. Different DSC scans for the diverse stoichiometric compositions were prepared (1:1, 1:2, 1:3, 1:4, 1:5, 2:1, 3:1, 4:1, 5:1) to determine the eutectic points in the solid forms MET·HCl-CTZ or MET·HCl-HTZ. The different stoichiometric samples were prepared by LAG (100 µL acetonitrile) using a Planetary Micro Mill Pulverisette TM 7 Fritsch at 600 rpm for 2 h. The DSC determinations proceeded at a heating rate of 10 • C/min using the thermal analyzer Netzsch STA 449 F3 Jupiter. Binary phase diagrams were constructed by plotting the melting temperatures from the different compositions (Supplementary Tables S3 and S4, SM †), considering the first endothermic event as the solidus point (T onset ), and the second endothermic event as the liquidus (T onset ) in function of the mole fraction of MET·HCl. Tammann diagrams were constructed by plotting ∆H fusion from the different stoichiometric ratios as a function of the mole fraction of MET·HCl. Some samples were occasionally analyzed at a heating rate of 2 or 5 • C/min to improve the viewing accuracy on some thermal events.

Nuclear Magnetic Resonance
Solid-state NMR (SSNMR) spectra were recorded in a Bruker Avance II 300 spectrometer (Billerica, MA, USA, operating at: 1 H 300 MHz, 13 [73]. In the 15 N-SSNMR spectra for MET·HCl we noted the presence of 4 signals as well as the reported NMR solution [72]. The δ 13 C CP-MAS assignation of CTZ [74][75][76] and HTZ [76][77][78] was on the basis of the reported data. were dissolved in d6-DMSO for the HMBC and HSQC experiments ( Supplementary Figures S2-S4, SM †). The atom assignation number of the native APIs is based on Scheme 2. For the HMBC and HSQC 15 N experiments we used NH3(l) δ = 0 ppm as internal reference and glycine (δ = 38 ppm) as secondary standard. The chemical shifts assignment in the 15 N-SSNMR experiments presented in Table 1 was made by analogy to the assignments in solution obtained from HMBC and HSQC experiments (Supplementary Figures S2-S4, SM †). In addition, the chemical shifts for the nuclei 1 H, 13 C and 15 N [71,72] in d6-DMSO and D2O for MET·HCl have been previously reported [73]. In the 15 N-SSNMR spectra for MET·HCl we noted the presence of 4 signals as well as the reported NMR solution [72]. The δ 13 C CP-MAS assignation of CTZ [74][75][76] and HTZ [76][77][78] was on the basis of the reported data.  A JEOL (Tokyo, Japan) scanning electron microscope (SEM) model JSM-6510LV was employed to examine the morphology of the solid forms (MET·HCl-CTZ 1:1 (χ MET·HCl = 0.5) and MET·HCl-HTZ 1:1 (χ MET·HCl = 0.5), Supplementary Figures S9 and S10, SM †) using the secondary electron detector. Element mapping was acquired with an energy dispersive spectrometer (EDS) QUANTAX 200 from Bruker (Supplementary Figures S11 and S12, SM †). The specimens' preparation was performed as follow: the dried samples were fixed on carbon tape over an Al-stub and finally coated with thin layer of gold using a Denton IV sputtering chamber.

Intrinsic Dissolution Studies
We determined the IDR constants under physiological conditions. The experiments were performed using tablets, prepared with a hydraulic press at a total force of 180-200 kg/cm 2 . The dissolution rates were determined with a Wood's apparatus according to the USP XLI. Dissolution profiles were made in 0.1 N HCl for each batch. Experiments were carried out in triplicate at 37 • C under constant stirring (50 rpm) in a constant volume of 900 mL. For CTZ and the different MET·HCl-CTZ compositions, IDR determinations were made using a HPLC Agilent 1100 with an automatic injector (DE116471) under the following chromatographic conditions: mobile phase of H 3 PO 4 0.025 M and acetonitrile (80:20) with a flow of 1 mL/min and using a Zorbax SB-C18 column with dimensions of 4.6 mm × 150 mm with particle size of 5 µm at a wavelength of 272 nm. For HTZ and the different MET·HCl-HTZ compositions, IDR determinations were performed using a UV-VIS spectrophotometer (Thermospectronic Helios Gamma (Waltham, MA, USA)) at the wavelength of 272 nm.

Saturation Solubility Experiments
An excess amount of powder (eutectics or pure drug) was weighed (approximately 33.3 mg) and dissolved in a vial with a fixed volume of water (1 mL 0.1 N HCl). The vial was magnetically stirred for 72 h at 37 • C. After the equilibrium time, an aliquot was passed through a 0.45 µm filter and properly diluted and quantified through HPLC (Agilent 1260 infinity II) using a calibration curve. The experiments were made in triplicate.

Powder X-ray Diffraction (PXRD)
The preparation of the solid forms was performed by ball-milling of MET·HCl with CTZ or HTZ (1:1) using neat grinding (NG) or liquid assisted grinding (LAG) [60][61][62][63][64]. In LAG, a solvent-screening, varying their polarity, was carried out using different solvents, such as hexane, acetone, acetonitrile and water, in order to explore their effect in the formation of new solid phases [79]. Then, all the products were analyzed by PXRD.
Analysis by PXRD (Figures 1 and 2) of the attained solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 exhibited most of the characteristic unaltered peaks of the parent APIs, suggesting the solids forms obtained were not cocrystals, and thus, Rietveld analysis was performed to determine whether new solid phases were produced [66,80].

NG and LAG Solvent-Screening
The preparation of the solid forms was performed by ball-milling of MET·HCl with CTZ or HTZ (1:1) using neat grinding (NG) or liquid assisted grinding (LAG) [60][61][62][63][64]. In LAG, a solvent-screening, varying their polarity, was carried out using different solvents, such as hexane, acetone, acetonitrile and water, in order to explore their effect in the formation of new solid phases [79]. Then, all the products were analyzed by PXRD.
Analysis by PXRD (Figures 1 and 2) of the attained solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 exhibited most of the characteristic unaltered peaks of the parent APIs, suggesting the solids forms obtained were not cocrystals, and thus, Rietveld analysis was performed to determine whether new solid phases were produced [66,80].  Based on the Rietveld refinements (Supplementary Tables S1 and S2, SM †) for MET·HCl-CTZ 1:1, two crystalline lattices are observed, MET·HCl (average 31.4%, P21/c) and CTZ (average 68.6%, P1), indicating the coexistence of two solid-phases as a conglomerate of separated components, both APIs keeping their own lattice, being more like a non-continuous single-phase than a eutectic solid. According to Cherukuvada and Nangia, "eutectic solids lack a distinct unique lattice arrangement from the individual components and retain the cohesive interactions in solid solutions" [19,20]. Similar results were obtained with the Rietveld analysis for MET·HCl-HTZ 1:1, observing MET·HCl (average

NG and LAG Solvent-Screening
The preparation of the solid forms was performed by ball-milling of MET·HCl with CTZ or HTZ (1:1) using neat grinding (NG) or liquid assisted grinding (LAG) [60][61][62][63][64]. In LAG, a solvent-screening, varying their polarity, was carried out using different solvents, such as hexane, acetone, acetonitrile and water, in order to explore their effect in the formation of new solid phases [79]. Then, all the products were analyzed by PXRD.
Analysis by PXRD (Figures 1 and 2) of the attained solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 exhibited most of the characteristic unaltered peaks of the parent APIs, suggesting the solids forms obtained were not cocrystals, and thus, Rietveld analysis was performed to determine whether new solid phases were produced [66,80].  Based on the Rietveld refinements (Supplementary Tables S1 and S2, SM †) for MET·HCl-CTZ 1:1, two crystalline lattices are observed, MET·HCl (average 31.4%, P21/c) and CTZ (average 68.6%, P1), indicating the coexistence of two solid-phases as a conglomerate of separated components, both APIs keeping their own lattice, being more like a non-continuous single-phase than a eutectic solid. According to Cherukuvada and Nangia, "eutectic solids lack a distinct unique lattice arrangement from the individual components and retain the cohesive interactions in solid solutions" [19,20]. Similar results were obtained with the Rietveld analysis for MET·HCl-HTZ 1:1, observing MET·HCl (average Based on the Rietveld refinements (Supplementary Tables S1 and S2, SM †) for MET·HCl-CTZ 1:1, two crystalline lattices are observed, MET·HCl (average 31.4%, P2 1 /c) and CTZ (average 68.6%, P1), indicating the coexistence of two solid-phases as a conglomerate of separated components, both APIs keeping their own lattice, being more like a noncontinuous single-phase than a eutectic solid. According to Cherukuvada and Nangia, "eutectic solids lack a distinct unique lattice arrangement from the individual components and retain the cohesive interactions in solid solutions" [19,20]. Similar results were obtained with the Rietveld analysis for MET·HCl-HTZ 1:1, observing MET·HCl (average 33.2%, P2 1 /c) and HTZ (average 68.6%, P2 1 ), both constituents retaining their own lattice. Hence, no effect of the polarity of the solvent was found for the LAG solvent-screening in both solid forms, the quantitative percentage ratios for the lattice structures between the APIs remaining constant in all cases.

Thermal Analysis
The products obtained through the LAG solvent-screening were also analyzed by DSC. PMs were prepared combining MET·HCl with HTZ or CTZ in a 1:1 stoichiometric ratio, and analyzed by the same technique. Figure 3 shows the DSC scans of the pure APIs and those of the different outcomes for MET·HCl-CTZ 1:1, where one can easily observe the appearance of a single endothermic event, showing a considerable reduction in the T fus (197.8-203.2 • C) compared with the parent APIs. In addition, for the PM the endothermic event exhibits two overlapped peaks (209.5 and 218.28 • C), clearly indicating that the different products obtained on the LAG solvent-screening are not PMs, since they only exhibit a single endothermic peak and not separated melting events for the individual components [30].

Thermal Analysis
The products obtained through the LAG solvent-screening were also analyzed by DSC. PMs were prepared combining MET·HCl with HTZ or CTZ in a 1:1 stoichiometric ratio, and analyzed by the same technique. Figure 3 shows the DSC scans of the pure APIs and those of the different outcomes for MET·HCl-CTZ 1:1, where one can easily observe the appearance of a single endothermic event, showing a considerable reduction in the Tfus (197.8-203.2 °C) compared with the parent APIs. In addition, for the PM the endothermic event exhibits two overlapped peaks (209.5 and 218.28 °C), clearly indicating that the different products obtained on the LAG solvent-screening are not PMs, since they only exhibit a single endothermic peak and not separated melting events for the individual components [30].
On the other hand, the endothermic events observed in any of the products of MET·HCl-HTZ 1:1, exhibit a reduction of the Tfus compared with the pure MET·HCl or HTZ (Figure 4). The PM exhibits a broad peak (196.4 °C) with a shoulder (207.92 °C), considerably differing from all other endothermic events from any of the products obtained. Furthermore, as it was for the previous case, no differences were observed for the LAGformed PMs.   On the other hand, the endothermic events observed in any of the products of MET·HCl-HTZ 1:1, exhibit a reduction of the T fus compared with the pure MET·HCl or HTZ (Figure 4). The PM exhibits a broad peak (196.4 • C) with a shoulder (207.92 • C), considerably differing from all other endothermic events from any of the products obtained. Furthermore, as it was for the previous case, no differences were observed for the LAG-formed PMs.
Pharmaceutics 2021, 13, x FOR PEER REVIEW 8 of 21 33.2%, P21/c) and HTZ (average 68.6%, P21), both constituents retaining their own lattice. Hence, no effect of the polarity of the solvent was found for the LAG solvent-screening in both solid forms, the quantitative percentage ratios for the lattice structures between the APIs remaining constant in all cases.

Thermal Analysis
The products obtained through the LAG solvent-screening were also analyzed by DSC. PMs were prepared combining MET·HCl with HTZ or CTZ in a 1:1 stoichiometric ratio, and analyzed by the same technique. Figure 3 shows the DSC scans of the pure APIs and those of the different outcomes for MET·HCl-CTZ 1:1, where one can easily observe the appearance of a single endothermic event, showing a considerable reduction in the Tfus (197.8-203.2 °C) compared with the parent APIs. In addition, for the PM the endothermic event exhibits two overlapped peaks (209.5 and 218.28 °C), clearly indicating that the different products obtained on the LAG solvent-screening are not PMs, since they only exhibit a single endothermic peak and not separated melting events for the individual components [30].
On the other hand, the endothermic events observed in any of the products of MET·HCl-HTZ 1:1, exhibit a reduction of the Tfus compared with the pure MET·HCl or HTZ (Figure 4). The PM exhibits a broad peak (196.4 °C) with a shoulder (207.92 °C), considerably differing from all other endothermic events from any of the products obtained. Furthermore, as it was for the previous case, no differences were observed for the LAGformed PMs.   have close similarity with their pure constituents" [19]. Thus, using 13 C or 15 N-SSNMR we can prove whether homosynthons remained intact. Thus, for the solid forms of the products prepared by LAG with acetonitrile (because the final powders were not thick) of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1, SSNMR experiments were performed. Atom numbering for MET·HCl, CTZ and HTZ for 13 C and 15 N nuclei are shown in Scheme 2.
Through 13 C-SSNMR experiments, it can be observed that there are no significant ∆δ of the pure APIs compared with the new solid forms MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 (Table 1). The minimal ∆δ 13 C observed for the MET·HCl-CTZ 1:1, proves that MET·HCl is not strong enough to replace the homosynthons in CTZ ( Figure 5). This is probably due to the fact that CTZ has stronger intermolecular interactions, reflected in its high T fus (364.4 • C) compared with MET·HCl (232.9 • C) [81]. A similar behavior is observed for MET·HCl-HTZ 1:1, where no significant ∆δ 13 C is observed compared with the original APIs ( Figure 6). Once again, MET·HCl cannot replace the homosynthons in HTZ. In this regard, Haneef et al. have reported that in the DDES of HTZ:ATL (0.3:0.7) [24], the robust sulphonamide (-SO 2 NH 2 ) catemer chain of HTZ cannot be interrupted by the amide group of ATL. In addition, for MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 apparently the non-self-complementarity (shape mismatch) of the initial components promotes the lack of formation of heterosynthons [19,21,82].

SSNMR
So far, we have seen that according to PXRD results, homosynthons predominate over the heterosynthons. In this regard, Cherukuvada and Nangia explained that "through different spectroscopical and PXRD analysis, eutectic solid phases and solid solutions have close similarity with their pure constituents" [19]. Thus, using 13 C or 15 N-SSNMR we can prove whether homosynthons remained intact. Thus, for the solid forms of the products prepared by LAG with acetonitrile (because the final powders were not thick) of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1, SSNMR experiments were performed. Atom numbering for MET·HCl, CTZ and HTZ for 13 C and 15 N nuclei are shown in Scheme 2.
Through 13 C-SSNMR experiments, it can be observed that there are no significant Δδ of the pure APIs compared with the new solid forms MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 (Table 1). The minimal Δδ 13 C observed for the MET·HCl-CTZ 1:1, proves that MET·HCl is not strong enough to replace the homosynthons in CTZ ( Figure 5). This is probably due to the fact that CTZ has stronger intermolecular interactions, reflected in its high Tfus (364.4 °C) compared with MET·HCl (232.9 °C) [81]. A similar behavior is observed for MET·HCl-HTZ 1:1, where no significant Δδ 13 C is observed compared with the original APIs ( Figure 6). Once again, MET·HCl cannot replace the homosynthons in HTZ. In this regard, Haneef et al. have reported that in the DDES of HTZ:ATL (0.3: 0.7) [24], the robust sulphonamide (-SO2NH2) catemer chain of HTZ cannot be interrupted by the amide group of ATL. In addition, for MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 apparently the non-self-complementarity (shape mismatch) of the initial components promotes the lack of formation of heterosynthons [19,21,82]. The 13 C-SSNMR spectra of pure CTZ ( Figure 5) exhibit five signals (C4, C6, C7, C8 and C9). However, Latosińska reported the appearance of only four signals due to a poorly resolved spectra (spun at 8.4 kHz) [76]. As is our case, the CP-MAS 13 C spectrum of MET·HCl-CTZ 1:1 only showed four signals (C4′, C6′, C7′ and C8′), lacking signals for C5′, C9′ and C10′ which were not observed, probably due to lack of efficient cross polarization 1 H-13 C. Regarding the 15 N-SSNMR experiments, for the APIs MET·HCl, CTZ or HTZ and the solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 (Figures 7 and 8), no significant Δδ were observed (Table 2). This fact indicates that there is no formation of heterosynthons and homosynthons are preserved intact.  The 13 C-SSNMR spectra of pure CTZ ( Figure 5) exhibit five signals (C4, C6, C7, C8 and C9). However, Latosińska reported the appearance of only four signals due to a poorly resolved spectra (spun at 8.4 kHz) [76]. As is our case, the CP-MAS 13 C spectrum of MET·HCl-CTZ 1:1 only showed four signals (C4 , C6 , C7 and C8 ), lacking signals for C5 , C9 and C10 which were not observed, probably due to lack of efficient cross polarization 1 H-13 C.
Regarding the 15 N-SSNMR experiments, for the APIs MET·HCl, CTZ or HTZ and the solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 (Figures 7 and 8), no significant ∆δ were observed (Table 2). This fact indicates that there is no formation of heterosynthons and homosynthons are preserved intact. Regarding the 15 N-SSNMR experiments, for the APIs MET·HCl, CTZ or HTZ and the solid forms of MET·HCl-CTZ 1:1 or MET·HCl-HTZ 1:1 (Figures 7 and 8), no significant Δδ were observed (Table 2). This fact indicates that there is no formation of heterosynthons and homosynthons are preserved intact.    However, in the 15 N-SSNMR spectra of MET·HCl-CTZ 1:1 and MET·HCl-HTZ 1:1, new signals were observed that were not detected in the 15 N-SSNMR spectra of the pure thiazide drugs. This is probably due to low-resolution spectra of the pure CTZ or HTZ due to their molecular rigidity, resulting in an inefficient cross polarization 1 H-15 N [81]. However, it may happen that once the solid forms are produced, the molecular rigidity of the thiazides relaxes, allowing a more efficient cross polarization 1 H-15 N, thus exhibiting signals for N6′ and N8′ in CTZ and N9′ and N11′ in HTZ that were unnoticeable before.

Characterization of the DDESs
According to the results from DSC, PXRD and SSNMR, MET·HCl-CTZ 1:1 and MET·HCl-HTZ 1:1 are DDES forms, since the spectroscopic and diffraction data are identical to those of the original APIs and a considerable reduction of their melting points is  However, in the 15 N-SSNMR spectra of MET·HCl-CTZ 1:1 and MET·HCl-HTZ 1:1, new signals were observed that were not detected in the 15 N-SSNMR spectra of the pure thiazide drugs. This is probably due to low-resolution spectra of the pure CTZ or HTZ due to their molecular rigidity, resulting in an inefficient cross polarization 1 H-15 N [81]. However, it may happen that once the solid forms are produced, the molecular rigidity of the thiazides relaxes, allowing a more efficient cross polarization 1 H-15 N, thus exhibiting signals for N6 and N8 in CTZ and N9 and N11 in HTZ that were unnoticeable before.

Characterization of the DDESs
According to the results from DSC, PXRD and SSNMR, MET·HCl-CTZ 1:1 and MET·HCl-HTZ 1:1 are DDES forms, since the spectroscopic and diffraction data are identical to those of the original APIs and a considerable reduction of their melting points is observed [20,30]. In addition, both solid forms cannot be considered cocrystals since there is no evidence for the formation of heterosynthons [20,28]. Furthermore, Rietveld analysis proved the simultaneous residence of two crystalline lattices, thus indicating that the components are unable to form a continuous single crystalline solid [32]. Furthermore, exhaustive crystallization attempts by evaporative methods to cocrystallize both solid forms were unsuccessful, confirming that APIs are immiscible in the solid state. One of the most recurrent forms to prove the formation of DDESs is based on the construction of binary phase diagrams by DSC screening [25,30,31,[83][84][85]. These binary phase diagrams can be produced from DSC scans at different compositions (1:1, 1:2, 1:3, 1:4, 1:5, 2:1, 3:1, 4:1, 5:1) [65], where some thermograms may show the appearance of two endothermic events. The first endothermic peak is considered as the solidus point (incongruent melting) and the second endothermic event as the liquidus point [65]. The appearance of these two endothermic events (solidus and liquidus points) in the DSC scans are attributed to the excess of one of the original components (non-eutectic phase) [65,83,84]. DSC scans that show no presence of the apparent liquidus point above the solidus indicate that the corresponding true eutectic composition was reached, since no evidence of unreacted or excess of reactants can be adverted. Tammann's graphs are very helpful to confirm that a genuine eutectic composition was found in the binary phase diagrams [30,31].
Thus, binary diagrams (Figures 9b and 10b) were constructed, plotting the solidus and liquidus points (T onset ) against the mole fraction of MET·HCl (χ MET·HCl ) at all examined compositions (Supplementary Tables S3 and S4, SM †). Tammann's triangle diagrams were plotted using the enthalpy of fusion (∆H fusion ) against the mole fraction of MET·HCl [30,31]. For more clarity, Figure 9a observed [20,30]. In addition, both solid forms cannot be considered cocrystals since there is no evidence for the formation of heterosynthons [20,28]. Furthermore, Rietveld analysis proved the simultaneous residence of two crystalline lattices, thus indicating that the components are unable to form a continuous single crystalline solid [32]. Furthermore, exhaustive crystallization attempts by evaporative methods to cocrystallize both solid forms were unsuccessful, confirming that APIs are immiscible in the solid state. One of the most recurrent forms to prove the formation of DDESs is based on the construction of binary phase diagrams by DSC screening [25,30,31,[83][84][85]. These binary phase diagrams can be produced from DSC scans at different compositions (1:1, 1:2, 1:3, 1:4, 1:5, 2:1, 3:1, 4:1, 5:1) [65], where some thermograms may show the appearance of two endothermic events. The first endothermic peak is considered as the solidus point (incongruent melting) and the second endothermic event as the liquidus point [65]. The appearance of these two endothermic events (solidus and liquidus points) in the DSC scans are attributed to the excess of one of the original components (non-eutectic phase) [65,83,84]. DSC scans that show no presence of the apparent liquidus point above the solidus indicate that the corresponding true eutectic composition was reached, since no evidence of unreacted or excess of reactants can be adverted. Tammann's graphs are very helpful to confirm that a genuine eutectic composition was found in the binary phase diagrams [30,31].
Thus, binary diagrams (Figures 9b and 10b) were constructed, plotting the solidus and liquidus points (Tonset) against the mole fraction of MET·HCl (χMET·HCl) at all examined compositions (Supplementary Tables S3 and S4, SM †). Tammann's triangle diagrams were plotted using the enthalpy of fusion (ΔHfusion) against the mole fraction of MET·HCl [30,31]. For more clarity, Figures 9a,b, and 10a,b can be observed in an enlarged form in the SM (Supplementary Figures S5 and S6, SM †).
The depression in the melting point, giving place to the characteristic V-shape in the binary diagram, indicates that a true eutectic composition has been found [20,64,86]. Conversely, a W-shaped graph suggests the formation of a cocrystal. [20,64,86]. W-shaped graphics exhibit two eutectic points and in the middle of them resides the zone where the cocrystal exists [20,64,86].  A typical V-shaped graphic for MET·HCl-CTZ (Figure 9b) indicates that at χMET·HCl = 0.5 (molar ratio 1:1), the formation of a true eutectic composition with a Teut = 198.78 °C occurs. Further, the Tammann's triangle confirmed that at χMET·HCl = 0.5 is a true eutectic point (ΔHfus = 163.3 J/g, solidus) (Supplementary Table S3, SM †), only exhibiting a single endotherm. However, for the composition χMET·HCl = 0.66 (molar ratio 2:1), a higher value (ΔHfus = 194.7 J/g, solidus) was observed, with this composition not a genuine eutectic point, due to the appearance of a second peak (liquidus) caused by the presence of unreacted or excess components.
In addition, PXRD experiments (Supplementary Figures S7 and S8, SM †) and Rietveld analysis (Supplementary Tables S5 and S6, SM †) were performed for all the different compositions examined for both MET·HCl-CTZ or MET·HCl-HTZ. Rietveld analysis for all different proportions showed the simultaneous presence of two crystalline lattices, proving that the constituents are physically separated.
Furthermore, both MET·HCl-CTZ and MET·HCl-HTZ at χMET·HCl = 0.5 were analyzed by scanning electron microscopy energy-dispersive X-ray spectroscopy SEM-EDS (Supplementary Figures S9 and S10, SM †). It has been reported that detection of light elements (C, N or O) in pharmaceutical samples can be difficult because they are not very sensitive to the EDS detector (poorer count statistics and low X-ray yields) [87,88]. For this reason, the analysis will be based on heavier atoms (Cl and S). The chemical composition acquired with EDS spectrometer for MET·HCl-CTZ (χMET·HCl = 0.5) was EDS Cl: 11.49 and S: 9.00 and calculated Cl: 15.37 and S: 13.90 (Supplementary Figure S9, SM †), whereas for MET·HCl- The depression in the melting point, giving place to the characteristic V-shape in the binary diagram, indicates that a true eutectic composition has been found [20,64,86]. Conversely, a W-shaped graph suggests the formation of a cocrystal [20,64,86]. W-shaped graphics exhibit two eutectic points and in the middle of them resides the zone where the cocrystal exists [20,64,86].
A typical V-shaped graphic for MET·HCl-CTZ (Figure 9b) indicates that at χ MET·HCl = 0.5 (molar ratio 1:1), the formation of a true eutectic composition with a T eut = 198.78 • C occurs. Further, the Tammann's triangle confirmed that at χ MET·HCl = 0.5 is a true eutectic point (∆H fus = 163.3 J/g, solidus) (Supplementary Table S3, SM †), only exhibiting a single endotherm. However, for the composition χ MET·HCl = 0.66 (molar ratio 2:1), a higher value (∆H fus = 194.7 J/g, solidus) was observed, with this composition not a genuine eutectic point, due to the appearance of a second peak (liquidus) caused by the presence of unreacted or excess components.
In addition, PXRD experiments (Supplementary Figures S7 and S8, SM †) and Rietveld analysis (Supplementary Tables S5 and S6, SM †) were performed for all the different compositions examined for both MET·HCl-CTZ or MET·HCl-HTZ. Rietveld analysis for all different proportions showed the simultaneous presence of two crystalline lattices, proving that the constituents are physically separated.
Furthermore, both MET·HCl-CTZ and MET·HCl-HTZ at χ MET·HCl = 0.5 were analyzed by scanning electron microscopy energy-dispersive X-ray spectroscopy SEM-EDS (Supplementary Figures S9 and S10, SM †). It has been reported that detection of light elements (C, N or O) in pharmaceutical samples can be difficult because they are not very sensitive to the EDS detector (poorer count statistics and low X-ray yields) [87,88]. For this reason, the analysis will be based on heavier atoms (Cl and S). The chemical composition acquired with EDS spectrometer for MET·HCl-CTZ (χ MET·HCl = 0.5) was EDS Cl: 11.49 and S: 9.00 and calculated Cl: 15 Figures S11 and S12, SM †), where it can be observed that in some zones the presence of O or S atoms is scarce, a fact that can be associated with the presence of MET·HCl molecules. This irregular distribution of O or S atoms indicates that both MET·HCl-CTZ (χ MET·HCl = 0.5) or MET·HCl-HTZ (χ MET·HCl = 0.5) are not a continuous single crystalline solid and the APIs are physically separated [32]. This physical separation between the components in both solid forms explain the differences between the calculated and experimental values observed in the EDS analysis. Since the intensity of the X-ray signal at any energy level is proportional to the concentration of that element in the path of the electron beam [88], when such element is reduced in concentration the maximum intensity and the peak-to-background are reduced [89].
According to the SEM micrographs, the morphology and texture observed in both solid forms are as solid phases with low crystallinity (Figures S9 and S10, SM †).

Microstructure Characterization
By itself, binary eutectic solids exhibit a microstructure-level periodicity different of the original pure crystalline components [90,91]. During the solidification, the effective entropy of both components changes, and this parameter can be used as an indicator to predict the microstructure adopted (Table 2). Hence, the nondimensional ∆S 0 fus , ∆S 0 fus = ∆H fus T fus /R (R: gas constant) [92,93] observed with the individual components could provide a panorama of the existing micromolecular array. The values of ∆S 0 fus can allow us to distinguish whether the solidification in the binary mixture is faceted or nonfaceted (regular) [84]. In our case, the ∆S 0 fus values observed between MET·HCl and CTZ and between MET·HCl and HTZ are similar but greater than 2 (i.e., ∆S 0 fus > 2 J/mol·K): MET·HCl (13.03 J/mol·K), CTZ (9.22 J/mol·K) and HTZ (7.50 J/mol·K, reported 9.61 J/mol·K) [94] (Table 2), hence both binary mixtures MET·HCl-CTZ (χ MET·HCl = 0.5) and MET·HCl-HTZ (χ MET·HCl = 0.5) solidified in a faceted manner (irregular microstructure) [84,93].
Furthermore, a comparison of the determined values of ∆H fus(excess) with the experimental ones can be used to differentiate whether a eutectic mixture or a PM has been obtained [30]. Thus, the thermodynamic parameters were calculated according to the following expressions (χ component = mole fraction of either component): Based on the above, negative ∆H fusion(excess) values were observed for both solids MET·HCl-CTZ (χ MET·HCl = 0.5) and MET·HCl-HTZ (χ MET·HCl = 0.5): −84.90 J/g and −110.05 J/g, thus indicating that DDESs were formed instead of PMs [30].

IDR Experiments
Due to the fact that CTZ and HTZ are the limited-water soluble drugs, in this study we only focused on the K int values for both thiazides. Thus, IDR experiments were carried out only for the solid phases MET·HCl-CTZ and MET·HCl-HTZ with compositions of 1:1 (χ MET·HCl = 0.5), 1:2 (χ MET·HCl = 0.33) and 2:1 (χ MET·HCl = 0.66), since these proportions are closer to the true eutectic points determined. Table 3 shows the K int values found and Figure 11 displays the plots of the IDR experiments.   The Kint value determined (0.076 mg/min·cm 2 ) is close to that reported for crystalline HTZ (0.098 mg/min·cm 2 ) [58]. In a previous work we reported the Kint value of MET·HCl [73]. Specifically, the composition of MET·HCl-HTZ χMET·HCl = 0.66 showed an increase of 1.21-fold compared with pure HTZ, however the compositions χMET·HCl = 0.33 and 0.5 exhibited a decrease in the dissolution rate (0.76-and 0.77-fold). This decrease in the release behavior of HTZ can be attributed to the great excess of HTZ compared with MET·HCl, limiting any probable enhancement during dissolution [95]. In addition, the compositions MET·HCl-CTZ χMET·HCl = 0.33 and 0.5 showed an increase in the amount of CTZ dissolved (1.31 and 1.65 -fold) compared with pure CTZ. However, in the case of MET·HCl-CTZ χMET·HCl = 0.66, it exhibited an increase of 6.06-fold. This should be taken with caution, since this Kint value presented a high standard deviation (Table 3). This is attributed to poor wettability and dispersibility of the solid form, as pronounced clumping occurred during the dissolution testing, limiting the reproducibility in triplicate determinations. Further- The K int value determined (0.076 mg/min·cm 2 ) is close to that reported for crystalline HTZ (0.098 mg/min·cm 2 ) [58]. In a previous work we reported the K int value of MET·HCl [73]. Specifically, the composition of MET·HCl-HTZ χ MET·HCl = 0.66 showed an increase of 1.21-fold compared with pure HTZ, however the compositions χ MET·HCl = 0.33 and 0.5 exhibited a decrease in the dissolution rate (0.76-and 0.77-fold). This decrease in the release behavior of HTZ can be attributed to the great excess of HTZ compared with MET·HCl, limiting any probable enhancement during dissolution [95]. In addition, the compositions MET·HCl-CTZ χ MET·HCl = 0.33 and 0.5 showed an increase in the amount of CTZ dissolved (1.31 and 1.65-fold) compared with pure CTZ. However, in the case of MET·HCl-CTZ χ MET·HCl = 0.66, it exhibited an increase of 6.06-fold. This should be taken with caution, since this K int value presented a high standard deviation (Table 3). This is attributed to poor wettability and dispersibility of the solid form, as pronounced clumping occurred during the dissolution testing, limiting the reproducibility in triplicate determinations. Furthermore, the aqueous solubility and dissolution profile of a given compound can be quantitatively related to its T fus and ∆H fus [96]. Usually substances with high values of T fus and ∆H fus have strong intermolecular interactions and possess low solubility [95,97]. Thus, if the formation of a binary eutectic solid leads to a modification in the thermodynamic functions representing an increase in the entropy of the mixture, the enthalpy of the mixture is also modified favorably, and the alteration of these factors may contribute to increase the negative value of the free energy of solution (∆G), overall improving the solubility of a substance [30]. Globally, the solid forms of MET·HCl-CTZ or MET·HCl-HTZ with the compositions χ MET·HCl = 0.33, 0.5 and 0.66 exhibited a decrease or a modest enhancement in the amount dissolved of the thiazide drugs. Considering the thermodynamic parameters (T fus , ∆H fus and ∆S fus , Table 2), in all cases there was a decrease in the T fus compared with the pure constituents. However, in both these solid forms at compositions of χ MET·HCl = 0.5 and 0.66, an increase in the ∆S fus (increase in randomization) was observed, suggesting an enhancement in the amount dissolved of the thiazides. However, there was also a slight increase in the ∆H fus compared with the original components, which affected the ∆G, leading to a limited modification in the amount dissolved, because ∆G tends to be more positive. In the case of the solid forms with composition χ MET·HCl = 0.33, both parameters remain almost unchanged, reflecting a limited modification in the amount dissolved: ∆H fus (limited modification in the intermolecular interactions) and ∆S fus (poor dispersibility). On the other hand, in the case of the DDES HTZ:atelonol (0.3:0.7), a 10-fold improvement of the % release of HTZ (phosphate buffer and pH = 7.4) was observed [24]. Haneef et al. attribute this improvement in dissolution rate to the fact that in the eutectic phase both components established weak intermolecular interactions. In our case, a similar analysis of the ∆H fus values for both solid forms MET·HCl-CTZ and MET·HCl-HTZ led us to conclude that a limited modification in the intermolecular interactions occurred after the solidification of the DDESs compared with their parent components.
According to the saturation solubility studies (Table 3), all the combinations for both solid forms exhibited an increase (the amount of drug dissolved in a saturated solution) compared with the pure components [98]. In this case, MET·HCl-CTZ χ MET·HCl = 0.33, 0.5 and 0.66 improved 1.51-, 1.48-and 1.44-fold respectively, and for MET·HCl-HTZ χ MET·HCl = 0.33, 0.5 and 0.66 increases in the solubility were 1.11-, 1.13-and 1.36-fold, respectively. Considering again the DDES HTZ:atelonol (0.3:0.7), an improvement in the solubility of HTZ by 14-fold was reported. The poor modifications in the solubility observed in the solid forms MET·HCl-CTZ χ MET·HCl = 0.33, 0.5 and 0.66 and MET·HCl-HTZ χ MET·HCl = 0.33, 0.5 and 0.66 is attributed to the limited alteration of ∆G in the solution.

Conclusions
Thus, this paper describes the mechanochemical preparation of the DDESs MET·HCl-CTZ and MET·HCl-HTZ by NG or LAG. Analyses by means of binary diagrams (DSC screening) allowed the determination of the true eutectic points at the composition of χ MET·HCl = 0.5 for both solid forms. In general, it has been reported that DDESs have a lower melting point than their original components, and that these solid forms have high free energy, greater molecular mobility and weaker intermolecular interactions [19,91,93,99]. These attributes may represent a substantial improvement in their aqueous solubility and dissolution properties. However, the solid forms described in this work exhibited a limited modification in these properties, this being due to the lack of or no significant alteration in the thermodynamic parameters ∆H fus and ∆S fus , compared with the original constituents, despite the fact that a considerable decrease in T fus values is observed.
Hence, this study represents a thorough characterization of two DDESs containing MET·HCl and CTZ or HTZ. This shows that despite an active research history describing organic solid eutectics [100], when related to pharmaceutical solid eutectics they still remain partially unexplored since the analytical methods available for full characterization and understanding of their molecular and structural integrity are almost incipient compared with cocrystals. Thus, we believe that the lack of proper understanding has hampered the potential applications of these solid forms [28]. Furthermore, we believe that this study sheds further light to better comprehend the behavior of these fascinating species.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/pharmaceutics13111926/s1, Table S1: Rietveld refinements (by NG or LAG solvent-screening) for MET·HCl-CTZ 1:1, Table S2: Rietveld refinements (by NG or LAG solvent-screening) for MET·HCl-HTZ 1:1, Table S3: Thermodynamic parameters for the construction of the of binary phase and Tammann's triangle diagram for MET·HCl-CTZ, Table S4: Thermodynamic parameters for the construction of the binary phase and Tammann's triangle diagram for MET·HCl-HTZ, Table S5: Rietveld refinements for the different compositions for the solid form MET·HCl-CTZ, Table S6: