Taming the dynamics in a pharmaceutical by cocrystallization: investigating the impact of the coformer by solid-state NMR

The anti-HIV pharmaceutical efavirenz is highly dynamic in its crystalline state, and we show that these dynamics can be tamed through the introduction of a coformer.

In Eq S4, rX···Y is the internuclear X···Y distance (C···H = 1.09x10 −10 m; C···F = 1.35x10 −10 m). In Eq S5, C is a factor given by Eq S6, n is the number of bonded hydrogen / fluorine atoms (X = H or F; n = 3 for the 13 CF3; n = 2 for 13 CH2, n = 1 for 13 CH). In Eq S6, S² is the order parameter. The order parameter (S²) for 1, 1b and 1c were calculated from a two-site jump model 5 using the torsion angles of interest from the X-ray structure (see Figure 2 of the main text and Table S37), whereas S² was approximated to 0.20 for 1a, 1d and 1e. The value of S² did not affect Ea and only had a minor influence on 0.

S4
Computational Details. Powder X-ray diffraction. Figure S1. Experimental and calculated powder X-ray diffraction of pure 1. The calculated pattern is for CSD structure 767883. 6 Figure S2. Experimental and calculated powder X-ray diffraction of 1a. The dagger denotes trace amount of 1 (see Figure S1 for the PXRD of 1). The calculated pattern of 1a is for CSD structure 768815. 6 Figure S3. Experimental and calculated powder X-ray diffraction of 1b. The calculated PXRD using the original crystal structure with disorder is shown in orange, whereas the calculated PXRD using the DFT-optimized structure without disorder is shown in black. The inset shows the low intensity of the calculated reflection at 14.1º. The calculated pattern is for CSD structure 767759. 6 Figure S4. Experimental and calculated powder X-ray diffraction of 1c. The calculated pattern is for CSD structure 909386. 7 Figure S5. Experimental and calculated powder X-ray diffraction of 1d. The calculated pattern is for CSD structure 909385. 7 Figure S6. Experimental and calculated powder X-ray diffraction of 1e. The calculated pattern is for CSD structure 1847168. 8

S8
Variable temperature 13 C solid-state NMR. Figure S7. 13 C solid-state NMR spectra of 1, pure Efavirenz, acquired at several temperatures (MAS = 11750 Hz, L( 13 C) = 128.9 MHz, contact time = 2 ms).      The 13 C solid-state NMR spectrum of 1 shows four resolved peaks for the CH2 carbons of the cyclopropyl ring (5 to 10 ppm), which arises due to the presence of two unique CH2 carbons on three crystallographically unique molecules (Z' = 3). The CH carbon of 1 (0 to −5 ppm) is split into two signals in a 2:1 ratio, and there are three peaks at 96 ppm, which have been assigned to the ethynyl carbon closest to the cyclopropyl group, further supporting the presence of three unique molecules of 1 in the structure (see Figure S19). In terms of 1a, the line shape observed for the CH resonance supports the presence of two distinct molecules, as seen in the X-ray structure (Z' = 2). Unfortunately, the CH2 resonances for the two unique cyclopropyl groups of 1a were not resolved in the 13 C spectrum due to their similar chemical shifts. While two molecules are observed in the structures of 1b, 1c, and 1d, a single peak was observed for both the CH2 and CH carbons. This may be a result of the 13 C atoms being in similar crystallographic environments in addition to the coalescence caused by dynamics. While a single molecule is in the structure of 1e (Z' = 1), two resonances have been observed for the CH2 carbons of the cyclopropyl group which has been attributed to the presence of two crystallographically distinct CH2 atoms on the cyclopropyl group. Figure S19. 13 C solid-state NMR spectra of 1 and the GIPAW simulated 13 C spectrum for both conformations of the disordered cyclopropyl group. "Conf A" denotes atom positions C16A, C17A, and C18A whereas "Conf B" denotes atom positions C16B, C17B, and C18B for the disordered cyclopropyl group. S15 Figure S20. 13 C solid-state NMR spectra of 1a and the GIPAW simulated 13 C spectrum for both conformations of the disordered cyclopropyl group.   In 1, two 19 F signals have been observed with an intensity ratio of 1:2, with the higher intensity being assigned by GIPAW calculation to two overlapping signals (see Figure S25). In the case of 1a, the two 19 F peaks observed experimentally are in excellent agreement with the GIPAW calculation (see Figure S26). While there are two molecules in the structure of 1b, 1c, and 1d, only a single peak has been observed in the experimental 19 F spectrum, with the GIPAW calculations suggesting only small differences in the chemical shifts between the two unique molecules (see Figure S27 to Figure S29). In 1e, a single 19 F peak is observed in the spectrum, supporting a Z' of 1 and the X-ray crystal structure (see Figure S30). Figure S25. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1. The experimental spectrum was acquired at 24ºC. The chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 145.6 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the crystallographic labels assigned to the resonance. S19 Figure S26. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1a. The experimental spectrum was acquired at 24ºC. The dagger and the dashed blue line denotes a trace amount of starting material (1). The chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 146.9 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the assigned crystallographic sites. S20 Figure S27. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1b. The experimental spectrum was acquired at 38ºC. The calculated chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 146.5 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the crystallographic labels assigned to the resonance. Figure S28. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1c. The experimental spectrum was acquired at 24ºC. The calculated chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 145.5 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the crystallographic labels assigned to the resonance. Figure S29. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1d. The experimental spectrum was acquired at 24ºC. The calculated chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 144.2 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the crystallographic labels assigned to the resonance. Figure S30. Experimental (black) and DFT-calculated (red) 19 F solid-state NMR spectrum of 1e. The experimental spectrum was acquired at 27ºC. The chemical shifts were averaged between the three fluorine atoms on the same CF3 group, and a ref of 145.0 ppm was used to reference the calculated chemical shifts. The labels above the calculated spectrum denote the crystallographic labels assigned to the resonance.
.20 c a Crystallographic site assigned using GIPAW calculations. b Disordered over two positions. c S 2 fixed to 0.2.

S28
Characterization of the rocking motion in 1. Figure S31. T1( 13 C) relaxation times of selected carbon atoms in 1. The T1( 13 C) of the cyclopropyl group have been assigned to disordered group in the structure (atoms C16A C17A C18A / C16B C17B C18A). Lines of best fit using Eq S7 are shown for the CH, CH2, and CF3 carbons. The fitting parameters can be found in Table S11. Figure S32. T1( 13 C) relaxation times of selected carbon atoms in 1. The T1( 13 C) of the cyclopropyl group have been assigned to ordered group in the structure (atoms C12A C13A C14A). Lines of best fit using Eq S7 are shown for the CH, CH2, and CF3 carbons. The fitting parameters can be found in Table S11.             C12B C13B C14B / C12D C13D C14D 5.3 a Performed on a reduced model. Figure S33. Starting geometry, transition state, and ending geometry of the motion of the cyclopropyl group (C11 C21 C31 / C12 C22 C32) in structure 1b. While the structure featured periodicity, only the molecule with motion is shown here for clarity. The rotation of the cyclopropyl group is not accompanied by the bending of the ethynyl group, and only minor rocking motions of the molecule is observed. The torsion angle between the ethynyl axis and the cyclopropyl group is shown in red. Figure S34. Starting geometry, transition state, and ending geometry of the motion of the cyclopropyl group (C12B C13B C14B / C12D C13D C14D) in structure 1c. While the structure featured periodicity, only the molecule with motion is shown here for clarity. he rotation of the cyclopropyl group is not accompanied by the bending of the ethynyl group, and only minor rocking motions of the molecule is observed. The torsion angle between the ethynyl axis and the cyclopropyl group is shown in red. Figure S35. DFT-calculated potential energy surface of an isolated molecule of Efavirenz. The torsion angle of interest has been highlighted in red on the molecular structure. Figure S36. DFT-calculated potential energy surface for the rotation of a cyclopropyl group in 1a (site 2), performed on a periodic model. The torsion angles from the X-ray structure (exp.) and DFT-optimized structure (calc.) are indicated by the red and black arrows, respectively. The second energy minimum at 130° may be inaccessible due to the high rotational energy barrier. 9 Section IV -X-ray Parameters, Crystallographic Sites, and Thermal Ellipsoid Plots.  Table S37. Torsion angle O-C-C-C, as shown on Figure 1 of the main text, measured in the experimental crystal structure and DFT-optimized structure. The probe atoms defining the torsion angles are given. The experimental errors were estimated using the thermal ellipsoids.
sample site probe atoms experimental O-C-C-C (°)  Table S39. Bending angle of the cyclopropyl-ethynyl axis (bend), as shown on Figure 1 of the main text, measured in the experimental crystal structure and DFT-optimized structure. The probe atoms defining the bending angles are given. The experimental errors were estimated using the thermal ellipsoids.