**2. Materials and Methods**

*Starting Materials.* Nefiracetam was purchased from Xiamen Top Health (Fujian, China). The solvents were sourced from VWR (Leuven, Belgium) and directly used without any purification steps. In the cases of ethyl acetate and acetonitrile, they were dried using a dessicant as the calcium hydride (CaH2). Nefiracetam was purified by a slurry crystallization in ethyl acetate (room temperature, ca. 300 rpm and overnight) and the solid phase was filtrated, washed, and dried. All coformers were commercially available from Alfa Aesar (Kandel, Germany), Acros Organics (Leuven, Belgium), TCI (Zwijndrecht, Belgium), and Merck (Overijse, Belgium) and used as received.

*Nefiracetam Cocrystals Screening.* A total of 133 cocrystallization agents typically used in cocrystallization efforts were selected among carboxylic acids, amides, sugars, inorganic salts and amino acids, other racetams, and profens. The cocrystal screening was performed through liquid assisted grinding (LAG) using a MM 400 Mixer Mill grinder manufactured by Retsch (Haan, Germany). The device is equipped with two grinding cells in which five 2 mL Eppendorf tubes can be set. To do so, equimolar amounts of Nefiracetam (0.2 mmol) and coformer (0.2 mmol) were weighted in an Eppendorf, and 4–5 stainless steel beads and 10 μL of solvent (methanol) were appended. Once the jars were filled, the milling runs for 90 min at 30 Hz.

For the three cocrystals studied in more detail in this work, these experiments were repeated using 12 different solvents in the LAG experiments as the nature of the solvent has been shown to impact on the outcome of such an experiment [24,25]. Using a 2:1 Nefiracetam/acid ratio for oxalic and citric acid and a 1:1 ratio for ZnCl2. The tested solvents were ethanol (EtOH), methanol, acetonitrile (MeCN), tetrahydrofuran, acetone, dichloromethane, chloroform, ethyl acetate, methyl acetate, diethylether, 2-propanol, and water, using approximately 10 μL of solvent.

*Single Crystal Growth.* Single crystals were mainly obtained from evaporative experiments. Nefiracetam and coformer were added in the molar ratio found in the cocrystal, and solids dissolved

by the addition of a su fficient amount of solvent. Solutions were then left to evaporate slowly (over periods ranging from three to seven days) at room temperature, and single crystals retrieved. A 2:1 Nefiracetam-citric acid cocrystal Form I (NCA) and 1:1:1 Nefiracetam-zinc chloride-water ionic cocrystal hydrate (NZCW) suitable crystals were obtained from ethyl acetate, while the 2:1 Nefiracetam-oxalic acid cocrystal (NOA) crystals was from ethyl acetate and tetrahydrofuran. Cooling experiments were performed by preparing supersaturated Nefiracetam and coformer solutions. An excess of Nefiracetam and coformer (equimolar) was added to a given solvent volume at room temperature, and the vials were placed at 15 ◦C below the related solvent boiling point until the full dissolution. Once the full dissolution was achieved, they were stored at −15 or 9 ◦C in the case of water. Solid phases were then retrieved and analyzed. The 1:1 Nefiracetam-zinc chloride ionic cocrystal (NZC) suitable crystals were obtained from a cooling crystallization in dried acetonitrile.

*Nefiracetam Cocrystal Bulk Material Preparation.* The bulk material for dissolution testing was prepared through crystallization from the solution. To identify an appropriate solvent, slurrying experiments were performed by placing excess amounts of Nefiracetam-coformer in suspension in di fferent solvents at 25 ◦C. Vials were sealed and the suspension was left over three days at 25 ◦C stirring at 700 rpm using a Cooling Thermomixer HLC manufactured by Ditabis (Pforzheim, Germany). Each vial was seeded with all possible solid forms (Nefiracetam and cocrystal) after 2 h of stirring. After three days, solid phases were retrieved and analyzed. The 12 aforementioned solvents were used for this section. A solvent was then selected in which the system behaves congruently (meaning a full transformation to cocrystal occurred over the three days), and an upscaled solvent-mediated cocrystal formation was performed. This upscaling was performed in an EasyMax 102 (Mettler Toledo, Zaventem, Belgium) crystallizer equipped with a hermetically closed 100 mL flask, under mechanic stirring (150 rpm) at 25 ◦C. After three days, the solids were retrieved, washed, and dried overnight at 50 ◦C. The solids were then used for dissolution measurements. NCA was upscaled to 30 g of the bulk material in ethyl acetate while NOA and NZC were upscaled in acetonitrile.

*Single-Crystal X-Ray di*ff*raction (SCXRD).* Suitable crystals of NZCW to perform the SCXRD analysis have been analyzed using a Rigaku (Neu-Isenburg, Germany) Ultra X18S rotating anode, FOX3D mirrors. The di ffracted beams were collected on a MAR345 image plate detector using MoK α (λ = 0.71073 Å). Single-crystal X-ray di ffraction data for NCA, 2:1 Nefiracetam-citric acid cocrystal Form II (NCA1), and NZC (100 K) were collected on an Oxford Di ffraction Gemini R Ultra di ffractometer (Ruby CCD detector using CuK α radiation) and crystals of NOA were measured at the SNBL beamline (Pilatus 2 M hybrid pixel detector), ESRF Grenoble. All the methodologies for the data reduction [26], resolution and refinement [27], and validation [28] were specified as in our previous work [3]. The images of the crystal structures were drawn using the software Mercury 4.1.3 [29]. CCDC 2010261-2010276 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/structures.

*Powder X-Ray Di*ff*raction (XRPD)* data were collected on a di ffractometer Bragg-Brentano manufactured by PANalytical (Eindhoven, Netherlands). The X-Ray source, a Ni-filtered CuK α (λ = 1.54179 Å), was used at 40 kV and 30 mA. The detection was carried out using a X'Celerator detector. All the powders were analyzed in a 2θ angle range from 4 to 40◦ for a total scan time of 6 min 42 s (step size = ca. 0.0167◦).

*Di*ff*erential Scanning Calorimetry (DSC)* measurements were performed from 25 to 175 ◦C at a scanning rate of 5 ◦C·min−<sup>1</sup> on a TA instrument DSC2500 (Zellik, Belgium) in the cases of NCA and NOA. The temperature range was extended to 275 ◦C for NZC and NZCW. Solid samples (6–7 mg) were placed in an aluminum crucible (40 μL) with pierced sealed lids and nitrogen was used as purge gas with a flow rate of 50 mL·min−1. Indium was used as a reference. Regarding the DSC measurements performed on the suspected and other confirmed cocrystals, the temperature range was extended to 200 and 225 ◦C.

*Thermogravimetric Analysis (TGA).* These analyses were carried out on a TGA-STDA 851e manufactured by Mettler Toledo (Columbus, OH, USA). The samples were analyzed in a temperature range from 25 to 400 ◦C (600 ◦C for NZC and NZCW). The scanning rate applied during the analysis was 10 ◦C·min−1. About 7–10 mg of solid materials were placed in an aluminum oxide crucible and a purge gas (nitrogen) was used with a flow rate of 50 mL·min−1.

*Dynamic Vapor Sorption (DVS)* analyses were performed at 25 ◦C on a Q5000 SA from TA instruments (New Castle, DE, USA). Solid samples (weight from 7 to 14 mg) were placed in a hanging platinum crucible and an empty crucible is used as the reference. These crucibles were placed in a chamber with controlled humidity and temperature. Nitrogen was used as flowing gas. Samples were first dried at 40 ◦C for 1 h before being exposed to variable relative humidity within a range of 10% to 90%. The equilibrium was considered to have been reached when the weight change was less than 0.010 mg·min−<sup>1</sup> or no weight change has occurred for 2 h.

*Moisture Exposure.* Samples (100 mg) of NCA, NOA, and NZC were stored for one month at room temperature in a sealed box containing an open water flask. The atmosphere was assumed to be saturated in water (100% RH). Both the XRPD and DSC analyses were performed on those powders after the exposure.

*High Performance Liquid Chromatography (HPLC).* The calibration line (linear equation) for the Nefiracetam dosage was drawn by interrelating the area under the curve (AUC) with the concentration prepared. To do so, Nefiracetam samples were diluted 1000 times using a 1:1 acetonitrile-Milli-Q water diluent (in volume). Nefiracetam chromatograms were then recorded according to the HPLC parameters: Device: Waters Alliance 2695 (Zellik, Belgium); Column: Waters Sunfire C18, 4.6 × 100 mm, 3.5 μm; Detector: PDA 2998 (extraction at λ = 210 nm); T◦ = 40 ◦C; injection volume: 10 μL; flow: 1.2 mL/min; mobile phase A: H2O + 0.1% H3PO4; mobile phase B: CH3CN + 0.1% H3PO4; gradient: 0 to 0.5 min at 30% B; 0.5 to 4.5 min 30% B →90% B; 4.5 to 6.5 min at 90% B. The calibration line is reported in the Supplementary Materials.

*UltraViolet Spectroscopy (UVS).* The calibration line for UV spectroscopy was drawn in order to dose the Nefiracetam by matching the absorbance at λmax (= 263 nm) and the concentration prepared with a linear equation. Acetonitrile was used as a diluent and blank. Sample UV-absorption spectra were recorded from 300 to 200 nm using a UV-1700 PharmaSpec spectrophotometer manufactured by SHIMADZU (Wemmel, Belgium).

*Dissolution Experiments.* Nefiracetam FI, NCA, NOA, and NZC were first ground using a mortar and pestle to ge<sup>t</sup> approximately the same particle size range. An excess amount of compound was added to 50 mL of solvent at 18 ◦C in a 100 mL flask under magnetic 100 rpm stirring. The 10 μL sampling occurred using a syringe equipped with a micro filter (0.02 μm). Sampling was performed every 10–15 s during the first minute up to t = 1 min, every 30 s up to t = 5 min, and every minute up to t = 15 min. A final sampling (triplicate) was performed after 24 h considering the equilibrium has been reached. Samples were diluted from 2000 to 5000 times depending on the fraction considered with a MeCN-water Milli-Q (1:1 in volume) diluent. Those fractions were then injected in HPLC and dosed as mentioned above. Concerning NZC, the 10 μL-fractions were diluted 250 times with MeCN before being dosed using a UV-1700 PharmaSpec (SHIMADZU) spectrophotometer. HPLC data as well as the UVS data related to the dissolution experiments are presented in the Supplementary Materials.
