*3.1. Cocrystal Screening*

A large cocrystal screen has been performed on Nefiracetam using liquid-assisted grinding (LAG) and 133 di fferent coformers. A full list is shown in the Supplementary Materials. Coformers have been selected based on crystal engineering principles, with all presenting synthons prone to cocrystal formation [30,31]. Seventeen cocrystals have been suspected based on XRPD data only, corresponding respectively to a success rate of 13%, which is in alignment with cocrystal success rates as reported in the literature [32]. Here, we report only those cases for which single crystals were obtained. This was the case for the following 13 coformers: 5-hydroxyisophthalic

acid, 5-nitroisophthalic acid, 5-cyano-1,3-benzenedicarboxylic acid, 5-bromoisophthalic acid, 2-benzoylbenzoic acid, parahydroxybenzoic acid, (RS)-phenylsuccinic acid, (RS)-2-phenylbutyric acid, (RS)-3-phenyllactic acid, gallic acid, citric acid, oxalic acid, and zinc chloride (Figure 2). Cocrystals are also suspected by XRPD and DSC for the positional coformers 2,4-dihydroxybenzoic acid (β-resorcylic acid), 2,5-hydroxybenzoic acid (gentisic acid) and 3,4-hydroxybenzoic acid (protocatechic acid), and dipiconilic acid but no single crystal confirmation has been obtained so far.

**Figure 2.** (**1**–**4**) Coformers for which a new Nefiracetam cocrystal is suspected and (**<sup>7</sup>**–**14**) coformers for which a single crystal cocrystal confirmation has been achieved.

In a more general manner, successful coformers usually tend to have at least one carboxylic acid group and a phenyl ring present. A similar trend has been observed for the cocrystallization of other racetam compounds [33,34]. Recently, we have also shown inorganic salts to be good cocrystal formers for racetam compounds [35,36], which is also confirmed here. Table 1 summarizes the main results for the 13 coformers leading to confirmed cocrystals. When the ground pattern does not match the simulated pattern from the single crystal, cocrystal polymorphs, solvates, or stoichiometrically diverse cocrystals are suspected [37,38]. This table also summarizes melting/dehydration temperatures of the Nefiracetam forms as well as the obtained cocrystals. Interestingly, we show that a very wide range of properties may be obtained using cocrystallization as a tool. Melting points vary from 68 to 243 ◦C merely by playing on the nature of the coformer. This shows the potential for variability one can achieve using cocrystallization as a crystal engineering tool. Furthermore, Table 1 also highlights the importance of screening for the various solid-state forms of a given API/coformer system, as cocrystal polymorphs or cocrystal hydrates can be readily encountered.

**Table 1.** Cocrystal forms identified, together with the melting point. Pharmaceutical systems of interest are highlighted in bold. RT: Room temperature; HT: High temperature.


The cocrystals of citric acid, oxalic acid, and zinc chloride are described in more detail as these cocrystals are pharmaceutically acceptable (Figure 3). For the analysis of the others, we refer to the Supplementary Materials. For citric acid (NCA), oxalic acid (NOA), and zinc chloride (*NZC*), LAG experiments all yielded the same outcome, irrespective of the solvent used. For these systems, a more profound form screening was performed using evaporative, slurrying, and cooling crystallization when applicable. Twelve solvents were used for a total of 125 experiments with results shown in the Supplementary Materials. These results led to no additional crystalline form for the NOA cocrystal but did lead to the identification of a second 2:1 cocrystal form for the NCA system and a cocrystal hydrate for the NZC system, highlighting the importance of screening for multiple cocrystal forms once a positive coformer has been identified.

**Figure 3.** Biocompatible Nefiracetam cocrystals identified and studied in detail in this work. Relevant intramolecular interactions between Nefiracetam and the coformers are highlighted by yellow stick contacts.

### *3.2. The 2:1 Nefiracetam-Oxalic Acid Cocrystal (NOA)*

As mentioned above, only one form of the NOA cocrystal has been identified, which is easily identifiable through its 2θ peaks at 11.3, 13.6, and 15.9◦. The bulk product pattern is furthermore shown to match the one simulated from the powder resolution analysis (Figure 4a), which shows NOA to crystallize in the monoclinic *P*21 space group. Figure 5a shows a main hydrogen bond *C*11 (4) chain pattern according to Etter's graph-set notation [39,40] built through N-H···O (2.86 (1) Å, 152.0◦ as a hydrogen bond distance N···O and NHO angle) hydrogen bonds between Nefiracetam amide moieties. Nefiracetam is linked to the oxalic acid through a hydrogen bonding pattern between the hydroxyl group from both the carboxylic acid moieties and the oxygen of the γ-lactam moiety of two different Nefiracetam molecules. These interactions (2.54 (1) Å, 167 and 171.5◦) result in a *D*11 (2) hydrogen-bond pattern. This assembly creates an overall pattern with Nefiracetam chains along the *a*-axis and oxalic acid acting as a linker between chains (zig-zag feature in Figure 5a represented by a full black line). The main crystallographic parameters related to each form and the refinement parameters are summarized in Table S2 in the Supplementary Materials. Upon heating, NOA shows a sharp melting endotherm at 162 ◦C (Figure 6c) and right upon melting, a weight loss of about 15% is observed (Figure 6a), corresponding to oxalic acid sublimation [41,42]. After 200 ◦C, degradation of Nefiracetam is observed. After a prolonged (one-month) exposure under a humidity saturated atmosphere (100% RH), no transformation of NOA into another form or deliquescence were observed by XRPD and DSC measurements (Supplementary Materials).

**Figure 4.** Simulated diffraction patterns (dashed line) and experimental powders (full line) of (**a**) Nefiracetam-oxalic acid cocrystal (NOA) (blue), (**b**) Nefiracetam-citric acid cocrystal (NCA) (blue) and NCA1 (red), and (**c**) Nefiracetam-zinc chloride ionic cocrystal (NZCW) (blue) and NZC (red). Patterns are compared to the Nefiracetam one (black).

**Figure 5.** Main intermolecular interactions and crystal packing along the a- and b-axis in the crystal structure of NCA (**a**), NCA1 (**b**), NOA (**c**), NZC (**d**), and NZCW (**e**). The tetragonal geometry of the Zn2<sup>+</sup> based-complex and the water molecules are respectively highlighted using the polyhedral and spacefill representation. Intermolecular contacts are shown using yellow dashed lines. The red dashed line is used to show the centroid-centroid distance in the NZC crystal structure.

**Figure 6.** TGA curves (**<sup>a</sup>**,**b**) and DSC curves (**<sup>c</sup>**,**d**) of Nefiracetam (black), NCA (blue), NOA (red), NZC (yellow), and NZCW (grey).

### *3.3. The 2:1 Nefiracetam-Citric Acid Cocrystal (NCA and NCA1)*

Two different polymorphs of the 2:1 Nefiracetam-citric acid cocrystal have been obtained. All screening experiments led to the same 2:1 cocrystal form (NCA). Single crystals of this form were obtained from a slow evaporation from ethyl acetate. Unexpectedly, the SCXRD diffraction experiment of NCA revealed also the presence of a second 2:1 form NCA1, in the same crystal. This latter is likely due to a phenomena of cross-nucleation, i.e., the heterogeneous nucleation of Form II (NCA1, daughter form) on the surface of the other (NCA, parent form) [43,44]. This Form II was never observed in sufficient quantity in any of the screening or bulk material preparation experiments, and is therefore likely a kinetic form, crystallizing under high supersaturated conditions, which occurs at the end of the solvent evaporation process. As during upscaling, the crystallization solvent is never fully evaporated, and the material is washed, this polymorph is very likely not present in the bulk material that fits the NCA simulated powder pattern in Figure 4b. NCA crystallizes in the triclinic *P*-1 space group. As for the anhydrous forms of Nefiracetam [3], the Nefiracetam molecules are stacked in a chain, by intermolecular amide-amide hydrogen bonds (N-H···O, 2.773 (5) Å, 168.2◦) leading to a *C*11 (4) hydrogen-bond pattern. In Figure 5b, the recognition between Nefiracetam and citric acid is due to two extra *D*11 (2) intermolecular hydrogen bonds (O-H···O, 3.18 (1) Å, 146.64◦ and O-H···O, 2.596 (5) Å, 168.01◦) which link one citric acid molecule to two Nefiracetam molecules through the oxygen of the γ-lactam moiety (Nefiracetam) and the hydroxyl of a carboxylic acid (citric acid). Moreover, a ring feature (*R*22 (8) with 2.654 (6) Å and 152.45◦) leading to a cyclic citric acid dimer is observable between two carboxylic acids from different citric acid molecules. Finally, an intramolecular hydrogen bond O-H···O (*S*11(3), 3.176 (7) Å, 121.48◦) is present in each citric acid molecule.

Polymorph NCA1 shows disorder and crystallizes in the monoclinic *P*21/*c* space group. Intermolecular interactions between citric acid and Nefiracetam are identical to those mentioned for the first NCA form (Figure 5c). In Figure 6a, TGA data shows three respective weight losses in the case of NCA. A first weight loss occurs around 80 ◦C, and must likely correspond to the loss of water molecules trapped in the channel-like structure observed for this cocrystal, evidencing that NCA is likely a non-stoichiometric hydrate. The following waves correspond to citric acid degradation (±28%, 175 ◦C) [45] and Nefiracetam degradation (from 225 ◦C), respectively. The DSC endotherm at around

80 ◦C corresponds to a potential dehydration of a non-stoichiometric hydrate, but could potentially also correspond to a melting point. DVS shows a deliquescent (+20% in weight) behavior at RH > 80%. When exposing NCA to a saturated humidity atmosphere for a long time period (one month), deliquescence is followed by Nefiracetam monohydrate crystallization (Supplementary Materials).

### *3.4. The 1:1 Nefiracetam-Zinc Chloride Ionic Cocrystal (NZC)*

The form screening with zinc chloride led to two di fferent cocrystals. The full screening results are presented in the Supplementary Materials. Slurrying crystallization in a congruen<sup>t</sup> and dried solvent, led to a 1:1 ionic cocrystal (NZC) wherein Nefiracetam is coordinated with the Zn2+. In addition, evaporative experiments at room temperature with water traces in the solvent (ethyl acetate) led to a 1:1:1 hydrated ionic cocrystal (so called NZCW) form whose structure was solved showing a zinc chloride complex with Nefiracetam and one water molecule. Only NZC was upscaled in dried acetonitrile. We were unsuccessful at obtaining larger amounts of NZCW as all experimental conditions tried led to mixtures of the Nefiracetam monohydrate, NZCW, and anhydrous NZC. The experimental and simulated XRPD data related to each system are presented in Figure 4c. Single crystals obtained from slowly cooling a supersaturated acetonitrile solution, yielded suitable monoclinic *C*2/*c* cube-like single crystals of the anhydrate (NZC) wherein Nefiracetam is complexed to zinc chloride showing a tetragonal geometry around the Zn2<sup>+</sup> ion. Nefiracetam molecules are bound to each other by two identical (by symmetry), tetrahedral complexes as shown in Figures 2 and 5d (right). Complexation occurs between zinc chloride and the γ-lactam C=O moiety of a first molecule and the amide C=O from a second molecule and vice versa. The donor moiety (N-H) from the amide is involved in a hydrogen bond pattern (*C*<sup>1</sup> 1(4), 3.496(1) Å, 146.49◦) with the chloride from the tetragonal complex along the *a*-axis. These tetrahedral complexes form a network of Nefiracetam dimer molecules as shown with black diamonds in Figure 5d. The π-stacking interactions are present between the Nefiracetam aromatic moieties with a centroid-centroid distance of about 3.54◦. This stacking may explain the higher density (1.637 Mg/m3) comparing NZC with NZCW.

NZCW crystallizes in the monoclinic *P*21 space group. The amide-amide hydrogen bond between Nefiracetam molecules as described above remains present but does not lead to a *C*<sup>1</sup> 1 (4) chain pattern as in the previous cases. Indeed, the Nefiracetam chains are interrupted by water molecules and Nefiracetam amide moieties are now involved in three *D*<sup>1</sup> 1 (2) hydrogen-bond patterns with respectively a water hydrogen (O-H···H, 2.734 (2) Å, 174.33◦), a second Nefiracetam amide moiety (N-H···O, 3.019 (3) Å, 164.80◦), and with a chloride ion (N-H···Cl, 3.331 (2) Å, 169.63◦). An infinite chain (O-H···Cl, *C*<sup>1</sup> 1 (4), 3.154 (2) Å, 169.19◦) is also observed involving the second hydrogen from the water molecule and a chloride ion. In the presence of zinc chloride, the carbonyl not belonging to the pyrrolidone group is coordinated to the Zn2<sup>+</sup> cation as well as to the water oxygen atom leading to a tetrahedral complex. This complex is highlighted in Figure 5e and forms "wave" chains stacking Nefiracetam densely (1.514 Mg/m3) along the *b*-axis. NZC shows a single melting endotherm at 243 ◦C. On the other hand, NZCW shows a continuous dehydration behavior (coordinated-water loss of ±4.5% corresponds to 1 equivalent of water) up until 175 ◦C. The DSC analysis then shows an endothermic peak at 200 ◦C which potentially corresponds to the melting of another polymorphic form of the ionic cocrystal or is a mere e ffect of the water still being present in the DSC capsule. To observe the potential transformation of NZC into NZCW, NZC was stored for one month in a saturated humidity chamber at room temperature. Even though, the DSC analysis indicates traces of the hydrated form (small endotherm around 210 ◦C), XRPD shows no such transformation. If NZCW is present in the bulk, it would only be in trace amount (Supplementary Materials). Therefore, NZC could be a potentially interesting form to market.

### *3.5. Solubility and Dissolution Profile of Nefiracetam Cocrystals*

Ethanol and acetonitrile were selected to study the dissolution profile at 18 ◦C (100 rpm) of NCA, NOA, and NZC in comparison to the API (Nefiracetam FI). As the cocrystals considered here behave

incongruently in water, with Nefiracetam monohydrate crystallizing out rapidly, we were unable to determine the apparent cocrystal solubility in this solvent. Therefore, we decided to perform the dissolution studies in EtOH and MeCN as we aim at underlining the potential impact cocrystals can have on physicochemical parameters. NCA, NOA, and NZC are incongruent in EtOH leading to a final slurry of Nefiracetam, whereas a congruen<sup>t</sup> dissolution is observed in MeCN. In the case of an incongruent system in EtOH, a "spring-parachute" [46,47] behavior (Figure 7a) is expected meaning that the concentration reaches a maximum ("apparent solubility" [48,49]) before dropping to the solubility of the drug form crystallizing out (here Nefiracetam FI). The final solubility of Nefiracetam might, however, still be different as solution interactions will occur. In the present case, the cocrystal dissolution kinetics combined to the crystallization kinetics of Nefiracetam FI occur too rapidly for the "spring-parachute" behavior to be observable. Nevertheless, one notices a clear impact of the coformer on the solubility of Nefiracetam (Table 2) with the overall solubility at 18 ◦C observed in both cases around 750 mM (+40%) instead of 519 mM (Figure 7c) in EtOH. The presence of ZnCl2 on the other hand, reduces solubility to about 199 mM.

**Table 2.** Dissolution rates and solubility at 18 ◦C in EtOH in the case of Nefiracetam FI, NCA, NOA, and NZC.


\* Dissolution rates correspond to the rate calculated when half the solubility is reached, with the corresponding time also given.

Table 3 also shows a true cocrystal solubility in MeCN where all cocrystals behave congruently. Here, one also clearly sees the importance of the coformer nature on the overall solubility, with citric acid slightly increasing the amount of Nefiracetam dissolved, whereas oxalic acid reduces this amount by about 30% in comparison to Nefiracetam FI. A striking drop in Nefiracetam present in the solution of about 90% is observed using the NZC cocrystal.

**Table 3.** Dissolution rates and solubility at 18 ◦C in MeCN in the case of Nefiracetam FI, NCA, NOA, and NZC.


\* Dissolution rates correspond to the rate calculated when half the solubility is reached, with the corresponding time also given.

The changes in solubility also reflect in the dissolution rate. However, they are all less important in comparison with the dissolution rate of the parent compound. It should nevertheless be noticed that all solid forms dissolve very rapidly, with half of the maximum solubility reached in the first minute and full solubility reached in the first 5 min of adding powder to the reactor. Therefore, we expect the bioavailability to be strongly impacted by the type of solid form used, not for the impact on the time required to reach solubility, but rather by the impact on the solubility value.

**Figure 7.** (**a**) Theoretical dissolution curve in the case of a congruen<sup>t</sup> and an incongruent system (with spring-parachute behavior). Experimental dissolution curves of Nefiracetam FI (black -), NCA (blue ), NOA (red ), and NZC (yellow •) in (**b**) MeCN at 18 ◦C with a 100-rpm stirring and (**c**) EtOH at 18 ◦C with a 100-rpm stirring.
