**Structural Elucidation of Enantiopure and Racemic 2-Bromo-3-Methylbutyric Acid** †

## **Rüdiger W. Seidel 1,\*, Nils Nöthling 2, Richard Goddard <sup>2</sup> and Christian W. Lehmann <sup>2</sup>**


Received: 27 May 2020; Accepted: 17 July 2020; Published: 30 July 2020

**Abstract:** Halogenated carboxylic acids have been important compounds in chemical synthesis and indispensable research tools in biochemical studies for decades. Nevertheless, the number of structurally characterized simple α-brominated monocarboxylic acids is still limited. We herein report the crystallization and structural elucidation of (*R*)- and *rac*-2-bromo-3-methylbutyric acid (2-bromo-3-methylbutanoic acid, **1**) to shed light on intermolecular interactions, in particular hydrogen bonding motifs, packing modes and preferred conformations in the solid-state. The crystal structures of (*R*)- and *rac*-**1** are revealed by X-ray crystallography. Both compounds crystallize in the triclinic crystal system with *Z* = 2; (*R*)-**1** exhibits two crystallographically distinct molecules. In the crystal, (*R*)-**1** forms homochiral O–H···O hydrogen-bonded carboxylic acid dimers with approximate non-crystallographic *C*<sup>2</sup> symmetry. In contrast, *rac*-**1** features centrosymmetric heterochiral dimers with the same carboxy *syn*···*syn* homosynthon. The crystal packing of centrosymmetric *rac*-**1** is denser than that of its enantiopure counterpart (*R*)-**1**. The molecules in both crystal structures adopt a virtually identical staggered conformation, despite different crystal environments, which indicates a preferred molecular structure of **1**. Intermolecular interactions apart from classical O–H···O hydrogen bonds do not appear to have a crucial bearing on the solid-state structures of (*R*)- and *rac*-**1**.

**Keywords:** 2-bromo-3-methylbutyric acid; 2-bromo-3-methylbutanoic acid; 2-bromoisovaleric acid; halogenated carboxylic acid; hydrogen bonding; chirality; absolute configuration; racemate; crystal structure; X-ray crystallography

#### **1. Introduction**

Halogenated organic compounds have received considerable research interest for decades, not only in the field of chemical synthesis [1–3] but also because of their biological properties [4,5]. In particular, a vast number of halogenated carboxylic acids have been synthesized and biochemically studied. Since mono-, di- and tricarboxylic acids are important intermediates in many biochemical pathways, their halogenated analogues have become an important research tool for the study of a wide range of biological processes owing to their ability to imitate the properties of the respective carboxylic acids or to inhibit crucial enzymes [6,7]. Despite tremendous research interest in halogenated carboxylic acids, the number of crystal structures of simple α-brominated monocarboxylic acids in the Cambridge Structural Database (CSD) is limited (13 as of June 2020) [8]. An example is bromoacetic acid, which forms a common *syn*···*syn* hydrogen-bonded carboxy dimer (Scheme 1) in the crystal (CSD refcode: BRMACA) [9]. Others are (−)-2-bromosuccinamic acid (BRSCAM) [10] and two crystal forms of 2,3-dibromo-3-phenylpropionic acid (CSD refcodes: ROFNOQ and ROFNOQ01) [11,12].

**Scheme 1.** Carboxy group *syn* and *anti* conformations [13].

2-Bromo-3-methylbutyric acid (2-bromo-3-methylbutanoic acid, **1**), commonly known as 2-bromoisovaleric or α-bromoisovaleric acid is a chiral α-halogenated monocarboxylic acid. Scheme 2 depicts the two enantiomers, (*S*)-**1** and (*R*)-**1**. Their resolution by fractional crystallization was reported almost 100 years ago [14]. Auterhoff and Lang, for example, used this approach to prepare both enantiomers of the hypnotic and sedative agent bromisoval (2-bromo-3-methylbutyrylurea or commonly bromovalerylurea) from (*S*)-**1** and (*R*)-**1** by reaction of the respective acid chlorides with urea [15]. Despite the fact that **1** has long been known and is commercially available, to the best of our knowledge and based on a WebCSD search in May 2020 [16], a crystal structure of **1** has not been reported so far.

**Scheme 2.** Chemical diagrams of the enantiomers of the title compound, (*S*)-**1** (**left**) and (*R*)-**1** (**right**) with stereodescriptors.

Chiral carboxylic acids have also attracted research interest in the fields of structural chemistry and crystal engineering, owing to phenomena such as frustration between molecular chirality and centrosymmetric hydrogen bond homosynthon formation in their crystal packing [17]. Enantiomeric mixtures can essentially crystallize as racemic crystals, racemic conglomerates (physical mixture of resolved crystals), inversion twins, disordered solid solutions [18] or, rarely, as kryptoracemates [19]. In this context, **1** attracted our attention. We have crystallized and investigated solvent-free **1** by X-ray crystallography in order to reveal preferred molecular conformations, crystal packing, intermolecular interactions and the outcome of crystallization of an enantiomeric mixture. Herein we report the crystal and molecular structures of (*R*)-**1** and *rac*-**1**.

#### **2. Materials and Methods**

(*S*)- and (*R*)-**1** were purchased from Sigma-Aldrich (purity 96%) and used as received. Solvents were of analytical grade and used without further purification. Crystals of (*R*)-**1** suitable for single-crystal X-ray diffraction were obtained from an ethanolic solution by slow evaporation of the solvent at ambient conditions. To obtain *rac*-**1**, equimolar amounts of (*S*)- and (*R*)-**1** were melted together on a Reichert hot-stage (Mikroheiztisch) mounted on a Nikon SMZ 1500 binocular microscope and cooled to room temperature [20]. The material so obtained was dissolved in ethyl acetate. Single-crystals of *rac*-**1** suitable for X-ray analysis appeared when the solvent was allowed to evaporate slowly at ambient conditions.

The X-ray intensity data were collected at 100(2) K on an Enraf–Nonius Kappa CCD for (*R*)-**1** and on a Bruker AXS Apex II for*rac*-**1**, using Mo *K*α radiation in both cases. The data were scaled and corrected for absorption effects with SADABS [21]. The crystal structures were solved with SHELXT [22] and refined with SHELXL-2018/3 [23]. The highest residual difference electron density peak each for (*R*)-**1** and *rac*-**1**

is ca. 0.7 Å from a bromine atom and can be ascribed to absorption effects. Carbon-bound hydrogen atoms were placed at geometrically calculated positions with Cmethine–H = 1.00 Å, Cmethyl–H = 0.98 Å and refined using a riding model with *U*iso(H) = 1.2 *U*eq(C) (1.5 for methyl groups). Torsion angles of the methyl groups were initially determined via difference Fourier syntheses and subsequently refined while maintaining tetrahedral angles at the carbon atoms. The carboxy hydrogen atoms in (*R*)-**1** were located in difference electron density maps. In subsequent refinements, the O–H distances were restrained to a target value of 0.84(2) Å. In *rac*-**1**, the carboxy hydrogen atom was placed in an idealized hydrogen bonding position with O–H = 0.84 Å and refined using a riding model. *U*iso(H) = 1.2 *U*eq(O) was used for all carboxy hydrogen atoms. Refined and post-refinement values of the Flack *x* parameter [24] were obtained with SHELXL using TWIN/BASF instructions and Parsons's method [25], respectively. The Hooft parameter [26–28] was calculated with PLATON [29]. Crystal data and refinement details for (*R*)-**1** and *rac*-**1** are listed in Table 1. Representations of the crystal and molecular structures were drawn with DIAMOND [30]. The structure overlay diagram and r.m.s. deviations of molecular structures from one another were obtained with MERCURY [31]. Packing indices were calculated with PLATON.


**Table 1.** Crystal data and refinement details for (*R*)-**1** and *rac*-**1**.

#### **3. Results**

Both (*R*)-**1** and *rac*-**1** were found to crystallize in the triclinic crystal system with two molecules in the unit cell. As shown in Figure 1, the molecules form O–H···O hydrogen bonded dimers through the carboxy groups in the *syn* conformation with a R<sup>2</sup> <sup>2</sup>(8) motif [32] in both crystal structures. Geometric parameters of the hydrogen bonds in both (*R*)-**1** and *rac*-**1** are given in Table 2, and selected bond lengths, bond angles and torsion angles are listed in Table 3. The encountered homochiral hydrogen-bonded dimer in (*R*)-**1** comprises two crystallographically unique molecules (*Z* = 2) and features approximate non-crystallographic *C*<sup>2</sup> symmetry with the twofold rotation axis passing through the center of the R2 <sup>2</sup>(8) hydrogen-bonded set and perpendicular to the mean plane of the two carboxy groups. In contrast, the hydrogen-bonded dimer in *rac*-**1** is heterochiral and lies across a crystallographic inversion center and, thus, is centrosymmetric. The molecule in the chosen asymmetric unit of *rac*-**1** (Z = 1) exhibits *R* configuration (see Figure 1, bottom). The two distinct molecules in (*R*)-**1** and the *R* enantiomer in *rac*-**1** adopt the same staggered conformation, as illustrated by a Newman projection in Scheme 3. A structure overlay diagram for the three molecular structures is depicted in Figure 2. The r.m.s. deviation of the respective non-hydrogen atoms in the two distinct molecules in (*R*)-**1** is 0.0322 Å. Between the respective non-hydrogen atoms and those of the *R* enantiomer in *rac*-**1**, the r.m.s. deviations are 0.0234 and 0.0243 Å.

**(***R***)-1**

*rac***-1** 

**Figure 1.** Homochiral and heterochiral hydrogen-bonded dimers respectively in (*R*)-**1** (**top**) and *rac*-**1** (**bottom**) in their crystal structures. Displacement ellipsoids are drawn at the 50% probability level. Hydrogen atoms are represented by small spheres of arbitrary radii. Dashed lines represent hydrogen bonds. Symmetry code: (a) −x + 2, −y + 1, −z + 1.


**Table 2.** Hydrogen bond geometry for (*R*)-**1** and *rac*-**1** (Å, ◦) 1.

**Table 3.** Selected bond lengths, bond angles and torsion angles (◦) for (*R*)-**1** and *rac*-**1** (Å, ◦) 1.


<sup>1</sup> Molecule 1 and molecule 2 in (*R*)-**1** are depicted respectively on the left- and right-hand side in Figure 1 (top).

**Scheme 3.** Newman-projection illustrating the staggered conformation of the *R* enantiomer encountered in the crystal structures of (*R*)-**1** and *rac*-**1**. For the corresponding torsion angles, see Figure 1 and Table 3.

**Figure 2.** Structure overlay of the two crystallographic distinct molecules in (*R*)-**1** (red and yellow) and the *R* enantiomer in *rac*-**1** (black).

The supramolecular structure of (*R*)-**1** in the crystal features short C–H···O contacts between the α-carbon atom of the carboxylic acid and the (formal) carboxy C=O moiety of an adjacent molecule (Figure 3). The hydrogen bonding motif descriptor is likewise R<sup>2</sup> <sup>2</sup>(8). In *rac*-**1**, the methine group of the α-carbon atom does not form a similar short C–H···O contact. The (formal) carboxy C=O moiety, however, is approached by a methyl hydrogen atom of the isopropyl group of an adjacent molecule (C···O = 3.58 Å, see Figure S1). The crystal packing in *rac*-**1** is denser than in (*R*)-**1**, which is evident from the volumes of the triclinic unit cells of both *Z* = 2 structures (Table 1). Thus, each molecule in *rac*-**1** occupies 5.2 Å<sup>3</sup> less space in the crystal than in (*R*)-**1**. The calculated densities (Table 1) and Kitaigorodskij packing indices [33] of 68.2% for *rac*-**1** and 66.2% for (*R*)-**1** further indicate a denser crystal packing in *rac*-**1** than in (*R*)-**1**. Short contacts (with respect to the sum of the corresponding van der Waals radii) of the Br···Br or C–H···Br type are neither observed in (*R*)-**1** nor in *rac*-**1**.

**Figure 3.** Section of the crystal structure of (*R*)-**1**, showing R2 <sup>2</sup>(8) motifs of intermolecular C–H···O and O–H···O contacts (*cf.* Figure 1). C21–H21···O22a: *d*(*D*···*A*) = 3.384(3) Å, <(*D*H*A*) = 145◦; C22–H22···O21b: *d*(*D*···*A*) = 3.519(3) Å, <(*D*H*A*) = 162◦. Symmetry codes: (a) x − 1, y − 1, z; (b) x + 1, y + 1, z.

#### **4. Discussion**

The *absolute structure* of a single-crystal of (*R*)-**1** grown from solution was established by anomalous-dispersion effects in the diffraction intensity measurements [34], thereby confirming the *absolute configuration* reported for the purchased bulk material [35,36]. Absolute structure parameters are listed in Table 1. The Flack *x* parameter estimated post-refinement based on quotients [25] and the Hooft parameter based on Bayesian statistics [26–28] are close to zero with adequately small standard uncertainties [18]. By way of comparison, the standard uncertainty of the refined Flack *x* parameter is larger than that of the former two parameters by a factor of ca. two [37].

The X-ray analysis of *rac*-**1** clearly revealed that an equimolar mixture of both enantiomers (Scheme 2) forms a racemic crystal upon crystallization from ethyl acetate and not a racemic conglomerate, which is observed in only ca. 10% of cases [38]. The higher crystallographic density of *rac*-**1** is in accord with Wallach's rule from 1895, which states that racemic crystals tend to be denser than the chiral counterparts [39]. This phenomenon can essentially be explained by the fact that enantiopure compounds can only crystallize in a Sohncke space group, devoid of inversion symmetry. To enable densest packing with *Z* = 1 in *rac*-**1**, the molecular dimer must be placed across a crystallographic inversion center, which would be impossible for homochiral *R*···*R* and *S*···*S* dimers. (*R*)-**1** crystallizes with *Z* = 2 in the space group *P*1, which is not among those available for densest packing of molecules of arbitrary shape [33]. Crystallization with *Z* > 1 is a common phenomenon for chiral carboxylic acids and has been described as frustration between chirality (referring to the whole molecule) and centrosymmetric dimer formation (referring to the hydrogen bond synthon) [17].

According to theoretical studies, the *syn* conformation of a carboxy group, as observed in (*R*)-**1** and *rac*-**1**, is energetically more stable than the *anti* conformation by ca. 21.4–28.9 kJ mol−<sup>1</sup> [13]. A *syn*···*syn* dimer (homosynthon) with a R<sup>2</sup> <sup>2</sup>(8) motif is a hydrogen bonding pattern commonly observed for carboxylic acids [40,41]. Its occurrence in the crystal structures of (*R*)-**1** and *rac*-**1** is thus as expected and also in accord with Etter's rules for hydrogen bonding, whereby all acidic hydrogen atoms and all good hydrogen bond acceptors are involved in hydrogen bonds, and the best donor and the best acceptor are hydrogen-bonded to one another [42]. The short C–H···O contacts observed in (*R*)-**1** (Figure 3) could be interpreted geometrically as weak hydrogen bonds [43]. It reasonable to assume that the hydrogen atom at the α-carbon atom here is prone to weak hydrogen bonds, since the carboxy group as well as the bromine atom should exert an electron-withdrawing effect. Since such contacts are not present in *rac*-**1**, their impact on the overall supramolecular structure in the crystal is probably minor.

The molecular conformations found in the crystal structures of (*R*)-**1** and *rac*-**1** are virtually identical, as evidenced by the calculated r.m.s. deviations of the corresponding heavy atom skeletons and visualized by a structure overlay diagram (Figure 2). It is reasonable to assume that a staggered conformation corresponds to a minimum energy structure, since not only the carbon chains but also the carboxy groups adopt the same orientation in the three molecular structures (Table 3) despite different crystal environments. This suggests that the observed conformation represents a preferred molecular structure of **1**.

#### **5. Conclusions**

We have revealed the crystal and molecular structures of (*R*)-**1** and *rac*-**1** by single-crystal X-ray analysis. The absolute configuration of (*R*)-**1** was confirmed by means of anomalous dispersion effects in the diffraction intensity measurements. Not unexpectedly, the ubiquitous carboxy *syn*···*syn* homosynthon was encountered in both structures. Clearly, O–H···O hydrogen bonds are the dominant intermolecular interaction in both structures. As compared with *rac*-**1**, the more open structure of (*R*)-**1** and the existence of two molecules in its asymmetric unit can be ascribed to frustration between chirality and centrosymmetric homosynthon formation. The observed denser crystal packing of centrosymmetric *rac*-**1** than of its enantiopure counterpart (*R*)-**1** is in accord with Wallach's rule. Short C–H···O contacts, as formed by the α-methine group in (*R*)-**1**, are not encountered in *rac*-**1**. This suggests that these weak intermolecular interactions may not have a crucial bearing on the packing of the hydrogen-bonded carboxylic acid dimers in the solid-state here, which appears to be essentially governed by close packing. A virtually identical molecular conformation in all in total three crystallographically distinct molecules in (*R*)-**1** and *rac*-**1**, despite different crystal environments, suggests that the observed geometry represents the preferred low energy structure.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2624-8549/2/3/44/s1, Figure S1: Section of the crystal structure of *rac*-**1**, showing short contacts between methyl hydrogen atoms of the isopropyl groups and the (formal) carboxy C=O moieties of adjacent molecules (C···O = 3.58 Å). CCDC 200603 [(*R*)-**1**] and 2006031 (*rac*-**1**) 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.

**Author Contributions:** Conceptualization, N.N., R.G. and R.W.S.; methodology, N.N. and R.G.; validation, R.W.S. and R.G.; formal analysis, R.W.S. and R.G.; investigation, N.N. and R.G.; resources, C.W.L.; data curation, R.W.S. and R.G.; writing—original draft preparation, R.W.S.; writing—review and editing, R.G.; visualization, R.W.S. and R.G.; supervision, C.W.L.; project administration, R.W.S.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank Alois Fürstner for providing laboratory resources for this project. R.W.S. is grateful to Peter Imming for his support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References and Note**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Absolute Configuration of In Situ Crystallized (+)-**γ**-Decalactone†**

**Michael Patzer, Nils Nöthling, Richard Goddard and Christian W. Lehmann \***

Max-Planck-Institut für Kohlenforschung, Kaiser Wilhelm Platz 1, 45470 Mülheim an der Ruhr, Germany; patzer@mpi-muelheim.mpg.de (M.P.); noethling@mpi-muelheim.mpg.de (N.N.);

goddard@mpi-muelheim.mpg.de (R.G.)

**\*** Correspondence: lehmann@mpi-muelheim.mpg.de

† Dedicated to Dr. Howard Flack (1943–2017).

**Abstract:** Knowledge about the absolute configuration of small bioactive organic molecules is essential in pharmaceutical research because enantiomers can exhibit considerably different effects on living organisms. X-ray crystallography enables chemists to determine the absolute configuration of an enantiopure compound due to anomalous dispersion. Here, we present the determination of the absolute configuration of the flavoring agent (+)-γ-decalactone, which is liquid under ambient conditions. Single crystals were grown from the liquid in a glass capillary by in situ cryo-crystallization. Diffraction data collection was performed using Cu-Kα radiation. The absolute configuration was confirmed. The molecule consists of a linear aliphatic non-polar backbone and a polar lactone head. In the solid state, layers of polar and non-polar sections of the molecule alternating along the c-axis of the unit cell are observed. In favorable cases, this method of absolute configuration determination of pure liquid (bioactive) agents or liquid products from asymmetric catalysis is a convenient alternative to conventional methods of absolute structure determination, such as optical rotatory dispersion, vibrational circular dichroism, ultraviolet-visible spectroscopy, use of chiral shift reagents in proton NMR and Coulomb explosion imaging.

**Keywords:** γ-(+)-decalactone; absolute configuration; in situ cryo-crystallization; flavoring agent; lactone; hydrogen bonding; crystal structure; X-ray crystallography

#### **1. Introduction**

Knowledge of the absolute configuration of small bioactive organic molecules is essential for understanding the significant different pharmacological effects of enantiomers on organisms [1,2]. The synthesis and characterization of chiral drug compounds is of considerable interest, especially in the pharmaceutical industry [3]. Additionally, enantiomers can differ in taste and smell perception [4,5]. This particularly applies to the natural compound discussed in this paper. γ-(+)-Decalactone [γ-**lac**], a compound that is liquid under ambient conditions, can be found in fruits, for example, strawberries or peaches and is suitable as a fruit flavoring agent [6,7]. In nature, the R-configuration of γ-**lac** is predominant. The absolute configuration of γ-**lac** has a significant influence on olfactory and taste perception. While the aroma of the *R*-enantiomer is reminiscent of peaches, the *S*-enantiomer smells of coconuts [8]. X-ray crystallography is able to determine the absolute configuration of enantiopure crystalline compounds by measuring intensity differences of *Bijvoet pairs* that are caused by anomalous dispersion [9]. One challenge is to determine the absolute configuration of organic drugs without atoms heavier than oxygen owing to the small anomalous scattering contribution of these elements [7]. For γ-**lac**, this is the case. The molecule contains solely carbon, oxygen, and hydrogen. The anomalous scattering contribution of an atom is not only dependent on the atom type but also on the wavelength of the radiation used. To get more accurate results, Cu-Kα radiation has an advantage in comparison with Ag-Kα or Mo-Kα radiation because of the enhanced anomalous scattering

**Citation:** Patzer, M.; Nöthling, N.; Goddard, R.; Lehmann, C.W. Absolute Configuration of In Situ Crystallized (+)-γ-Decalactone. *Chemistry* **2021**, *3*, 578–584. https:// doi.org/10.3390/chemistry3020040

Academic Editors: Catherine Housecroft and Katharina M. Fromm

Received: 30 March 2021 Accepted: 18 April 2021 Published: 21 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

factor at this wavelength [10]. Currently, there exist several approaches to circumvent these difficulties of absolute configuration determination. For example, co-crystallization with compounds that contain heavy atoms or co-crystallization with a chiral compound with known configuration are possible solutions to obtain the absolute configuration of the target molecule [11]. Similarly, in case of an amine, the formation of a hydrochloride can serve this purpose [9]. The object of this study was to determine the absolute configuration of an enantiopure compound that is liquid under ambient conditions. We chose to look at γ-**lac** because of its industrial relevance. In situ crystallization is a powerful tool to grow crystals direct on the diffractometer from pure liquid compounds in a glass capillary [12,13]. In this paper, the determination of the absolute configuration of a liquid natural organic compound is presented, together with a further technique for in situ crystal growth. This procedure can easily be transferred to other liquid or gaseous organic compounds.

#### **2. Materials and Methods**

(+)-γ-Decalactone was purchased from Sigma-Aldrich (purity > 97%) and used as received [α] 24 *<sup>D</sup>* = +41.6 (c 0.087, CHCl3). The liquid compound was crystallized in a capillary with a diameter of 0.3 mm directly (in situ) on the diffractometer (volume of approximately 0.5 μL). The low-temperature phase behavior was established beforehand by differential scanning calorimetry (DSC). Experiments were performed on a METTLER TOLEDO DSC 820 (Mettler-Toledo GmbH, Gießen, Germany). Two cycles with different cooling and heating rates were carried out in the range of −150–+20 ◦C. The first and second scan employed temperature gradients of 10 and 5 K per minute, respectively.

For single crystal growth, the compound was cooled below its liquid–solid phase transition temperature of 258 K (determined by LT-DSC) on the diffractometer using a stream of cold nitrogen gas delivered by an Oxford Cryosystems Cryostream 700 (Oxford Cryosystems, Long Hanborough, Oxford, UK). The crystalline powder thus obtained was used as starting material for crystal growth. By translation of the capillary through the cold nitrogen gas stream, a suitable single crystal was grown at the liquid–solid phase boundary (inverse zone melting) following a newly developed procedure [12].

For this purpose, a small attachment to a standard Huber goniometer head (model 1004) was constructed, in order to move the capillary along its axis at a controlled speed over a distance of several millimeters during a time period of several hours (Figure 1). Since there is a temperature gradient across the cold gas stream of the Cryostream 700, it is possible to slowly grow a single crystal at the solid–liquid interface and to continually transfer the crystallized portion of the compound into the colder part of the gas stream, until such time that a nearly perfect single crystal is located in the X-ray beam. The attachment comprises a bracket, which holds a miniaturized stepper motor equipped with a gear box and connected at one end to an indexer board and power supply. The motor axis is coupled to a square nut, which fits snugly over the height adjustment drive of the goniometer head. This attachment can be slid off after crystal growth has been accomplished, thus allowing subsequent free rotation of all goniometer axes.

**Figure 1.** (**a**) Sketch of the attachment showing its main components, (**b**) front view of attachment bracket with square nut fitting, (**c**) side view and size relations of attachment, (**d**) device attached to a Huber goniometer head (model 1004) mounted on the ϕ-axis of a Mach-III four circle goniometer.

Once crystal growth is completed, there are in some cases several crystals next to each other in the capillary. The procedure can be repeated and if the problem subsists, the major single crystal component can be selected, or the twinning identified in a three-dimensional representation of the reciprocal space.

X-ray intensity data were collected at 100(2) K on a Bruker-AXS Kappa Mach3 goniometer equipped with an APEX-II detector, using Cu-*Kα* radiation produced by a FR591 rotating anode X-ray source (BrukerNonius B. V., Delft, The Netherlands). Scaling and absorption correction were performed with SADABS (Bruker AXS Inc., Madison, WI, USA). The crystal structure was solved by SHELXT and refined using SHELXL-2018/3. No further constraints or restraints were applied. A summary of the crystallographic details is given in Table 1, while further structural details including bond lengths and angles can be found in the Supplementary Material. CCDC 2072278 contains the supplementary crystallographic data for this crystal structure. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/structures.

**Table 1.** Crystal data and refinement details for (+)-γ-Decalactone.



**Table 1.** *Cont.*

#### **3. Results**

The absolute configuration of γ-**lac** was determined using anomalous dispersion effects. The derived *Flack parameter* calculated from 717 quotients of *Bijvoet pairs* according to Parsons' method (SHELXL) is −0.019(60) and thus the *R*-configuration is verified (Figure 2). γ-**lac** consists of a non-polar alkyl chain and a more polar cyclic ester. The polar parts appear to interact via non-classical C-H···O hydrogen bonds involving the carbonyl group of the lactone. The supramolecular structure comprising these intermolecular hydrogen bonds between the polar groups is shown in Figure 3 (numerical values are given in Table 2). In the crystal layers consisting of the non-polar and the polar parts alternate along the c-axis of the unit cell. The structure in solid state appears to be governed by the polar ends of the molecule, since there are no short contacts between atoms in the non-polar region. The five-membered lactone ring adopts an envelope conformation with the apex at C3, which is 0.5 Å above the mean plane (r. m. s. 0.01 Å) formed by the other four atoms of the ring. The ring puckering parameter *ϕ* equals 291.8◦ (see Supplementary Material) [14]. The aliphatic alkyl chain adopts an all-trans conformation in the crystal structure (Figure 2) [15].

**Figure 2.** Molecular structure of the chiral compound (+)-γ-decalactone ((*R*)-5-hexyloxolan-2-one) determined by single crystal X-ray crystallography. The probability level of the displacement ellipsoids is 50%. Hydrogen atoms are shown with arbitrary sizes.

**Figure 3.** Crystal structure of γ-**lac**, view along the b-axis, C-H···O hydrogen bonds marked with black dashed lines.



#### **4. Discussion**

For molecules without atoms heavier than oxygen, the determination of the *absolute structure* becomes difficult due to the low anomalous scattering signal. Two strategies were pursued. First, Cu-Kα radiation was chosen because of the larger resonant scattering factors. This resulted in larger measured intensity differences of the Bijvoet pairs as compared with shorter wavelength radiation. Second, the diffracted intensities were measured with a high redundancy. This enabled outliers to be more easily identified and errors in the measured intensities to be reduced. Both techniques allowed the absolute structure parameters to be determined more accurately. The Flack parameter obtained from 718 quotients with its resultant standard deviation verifies the absolute configuration of the molecule and hence that of the structure (Table 1). The Flack parameter in absolute structure determination has to be evaluated carefully. There can be significant differences between the classical Flack parameter derived from fitting scale factors to the structure factors of both antipodes and the Flack parameter according to Parsons' method, which uses quotients [18]. The Flack parameter determined by Parsons' method usually has a smaller standard uncertainty than the classical Flack parameter and is implemented in standard crystallography programs like SHELXL [19,20]. A standard uncertainty for the Flack parameter smaller than 0.1 is necessary for accurate absolute structure determination in order to ensure that its value is zero within error and hence can be differentiated reliably from the configuration with the opposite hand, for which a value of one would be expected. Here, we report an example in which the Flack parameter is very sensitive to the measured intensity of just one outlier. By chance, we discovered that the reflection with Miller indices -5 2 14 and a d-spacing of 0.89 Å has a significant influence on the determination of the absolute configuration. The measured intensity was far too large (19 standard uncertainties from the expected value) and was subsequently attributed to a diffracted intensity of a second crystal in the capillary. If this reflection is included in the intensity data, the Flack parameter is −1.62(22)

(determined from 733 Bijvoet pairs according to the procedure described in [18]). On omitting this reflection, the Flack parameter assumes a meaningful value of 0.02(6) (see Supplementary Material). In the capillary, multiple crystals cannot always be avoided when a crystal is grown by translation perpendicular to the nitrogen gas stream and extreme caution is advised, as this example illustrates.

To the best of our knowledge, there is no report in the Cambridge Structural Database of the crystal structure determination of a pure γ-lactone that has only an aliphatic substituent with no functional groups in the chain. The bipolar structure of γ-**lac** can be compared with non-ionic surfactants, which have a polar head and a non-polar backbone. The packing of the molecules in the crystal is likely to be a consequence of the polarity of the molecular structure of γ-**lac**. A layer structure or in more general a structure in which some parts of a molecule agglomerate may be expected when the molecule has a hydrophobic chain and polar head.

One of the advantages of the goniometer head attachment described above is the simplicity of operation. The set-up is initialized by retracting the height adjustment until the stall guard functionality of the stepper motor controller signals that the end of travel has been reached. A 10-turn potentiometer connected to the analogue input of the controller allows setting the speed of the translational movement over a time range between minutes and hours, while two illuminated push buttons allow the direction of travel to be input.

#### **5. Conclusions**

We have shown that in situ cryo-crystallization allows one to determine the absolute configuration of the enantiopure liquid compound γ-**lac** by using single crystal x-ray crystallography. Especially for medicinal and pharmaceutical research, in situ cryocrystallization combined with X-ray crystallography promises to be a powerful and easy tool to determine the absolute configuration of drugs and small-molecule compounds that are liquid under ambient conditions. In addition, the technique allows intermolecular interactions to be studied in more detail. In this example, the polarity and molecular geometry of γ-**lac** leads to a layered structure. Intermolecular C-H···O hydrogen bonds between the lactone-part of γ-**lac** can be observed. This procedure can easily be transferred to other liquid organic drug compounds that crystallize at low temperature.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/chemistry3020040/s1, Figure S1: ORTEP-Plot of the molecular structure of (+)-γ-Decalactone. Figure S2: Screen shots of the capillary, face indexing and unit cell indexing. Figure S3: Recorded ambient to low temperature DSC curve of γ-**lac**. Figure S4: ATR-FT-IR spectra of γ-**lac**. Figure S5: Determination of the Flack-Parameter by Parsons' method; (a) The full data-set of 733 Bijvoet-Pairs was used for determination of the Flack-Parameter (variable p in linear regression), the quotient with highest difference from theory is marked with a red circle; (b) Linear regression of 732 Bijvoet pairs, the red marked quotient in (a) was omitted. Figure S6: Excerpt from the HKLF4 file for γ-**lac**. Symmetry-equivalent reflections are marked with colored boxes (point group 2 2 2: h k l = -h-k l = -h k -l = h -k -l). The reflection (-5 -2 -14) in the blue box strongly deviates from its Bijvoet-pairs (red box). Table S1: Crystal data and structure refinement. Table S2: Bond lengths [Å] and angles [◦]. Table S3: Anisotropic displacement parameters (Å2). Table S4: Hydrogen coordinates and isotropic displacement parameters (Å2). Table S5: Torsion angles [◦]. Table S6: Hydrogen bonds [Å and ◦]. Table S7: Asymmetry Parameters of the Five Membering Ring (Puckering Coordinates Analysis).

**Author Contributions:** Conceptualization, N.N.; methodology, M.P., N.N., and C.W.L.; writing original draft preparation, M.P.; writing—review and editing, all; visualization, M.P.; supervision, C.W.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We are grateful to M. Felderhoff for providing access to low temperature DSC and A. Fürstner for access to ATR-IR spectroscopy and optical rotation measurements.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Novel Ansa-Chain Conformation of a Semi-Synthetic Rifamycin Prepared Employing the Alder-Ene Reaction: Crystal Structure and Absolute Stereochemistry †**

**Christopher S. Frampton 1,\*, James H. Gall <sup>2</sup> and David D. MacNicol 2,\***


**Citation:** Frampton, C.S.; Gall, J.H.; MacNicol, D.D. Novel Ansa-Chain Conformation of a Semi-Synthetic Rifamycin Prepared Employing the Alder-Ene Reaction: Crystal Structure and Absolute Stereochemistry . *Chemistry* **2021**, *3*, 734–743. https:// doi.org/10.3390/chemistry3030052

Academic Editors: Catherine Housecroft and Katharina M. Fromm

Received: 26 May 2021 Accepted: 9 July 2021 Published: 11 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Abstract:** Rifamycins are an extremely important class of antibacterial agents whose action results from the inhibition of DNA-dependent RNA synthesis. A special arrangement of unsubstituted hydroxy groups at C21 and C23, with oxygen atoms at C1 and C8 is essential for activity. Moreover, it is known that the antibacterial action of rifamycin is lost if either of the two former hydroxy groups undergo substitution and are no longer free to act in enzyme inhibition. In the present work, we describe the successful use of an Alder-Ene reaction between Rifamycin O, **1** and diethyl azodicarboxylate, yielding **2**, which was a targeted introduction of a relatively bulky group close to C21 to protect its hydroxy group. Many related azo diesters were found to react analogously, giving one predominant product in each case. To determine unambiguously the stereochemistry of the Alder-Ene addition process, a crystalline zwitterionic derivative **3** of the diethyl azodicarboxylate adduct **2** was prepared by reductive amination at its spirocyclic centre C4. The adduct, as a mono chloroform solvate, crystallized in the non-centrosymmetric Sohnke orthorhombic space group, *P*212121. The unique conformation and absolute stereochemistry of **3** revealed through X-ray crystal structure analysis is described.

**Keywords:** Rifamycin O; ansamysin; antibacterial; semi-synthesis; Alder-Ene; conformation; zwitterionic; hydrogen bonding; absolute configuration; chirality; crystal structure; X-ray crystallography

#### **1. Introduction**

The rifamycins constitute an important class of ansamycin antibiotic active against mycobacteria and other bacterial pathogens, also exhibiting antiviral properties. These molecules are comprised of a substituted naphthalene or naphthoquinone core spanned by a seventeen-membered aliphatic ansa bridge. A vast number of semi-synthetic rifamycins have been produced by structural modification of the aromatic region of naturally occurring rifamycins [1]. The important bridging ansa moiety has not been so intensively studied, though recent highlights are the excellent antibacterial activity found for 24-desmethylrifampicin [2]; and the synthesis of C25 carbamate derivatives which are resistant to ADP-ribosyl transferases [3]. The present work was directed at the introduction of a bulky group close to the hydroxy group on C21 of Rifamycin O, **1**, Scheme 1, to inhibit transferase deactivation [4,5]. Attempts to carry out a Diels–Alder reaction with dimethyl acetylenedicarboxylate, hoping to exploit the *cisoid* diene arrangement of **1**, torsion angle 36◦, in the crystal [6] proved unsuccessful; however, gratifyingly, we found that diethyl azodicarboxylate and related diesters reacted readily and quantitatively giving a major product, along with a by-product in each case. The disappearance of a methyl doublet from the proton NMR spectrum with introduction of an allylic methyl singlet at lower field (see

methyl assignments in Experimental) clearly pointed to an Alder-Ene reaction [7,8] rather than a Diels–Alder reaction for which four methyl doublets would have been expected. MS confirmed a 1:1 adduct had been formed and inspection of a tactile Dreiding model of 1 revealed that azo nitrogen attack at the C18 alkene face giving an *S* configuration at C18 would, to effect hydrogen abstraction from C20, lead to a *trans* (*E*) double bond between C19 and C20. Corresponding alternative attack on the opposite alkene face would lead to an *R* configuration at the C18 stereogenic centre with a predicted *cis (Z)* double bond formed between C19 and C20. Although product **2** was obtained stereochemically pure at C4 (single AB quartet for the diastereotopic methylene protons of the spirolactone at C4), suitable single crystals for X-ray analysis could not be obtained. However, crystals were obtained from chloroform for **3**, which was derived from **2** by reductive amination as described in the Experimental section below and this resolved the stereochemical question and also revealed an unprecedented ansa-chain conformation.

**Scheme 1.** Rifamycin structures referred to in the text.

#### **2. Materials and Methods**

Typical reaction conditions for **2**: compound **1** (1g, 0.00132 mmol) and diethyl azodicarboxylate (0.69 g, 0.00396 mmol) were refluxed in toluene (40 mL) under argon for 5 h and then left at 50 ◦C for one week. The toluene was removed, and the reaction product boiled in iso-propanol and then cooled and filtered, yielding **2** as a yellow powder. Assignments for the methyl resonances of **2**: 1HNMR (400 MHz, CDCl3), *δ*H, C34, 0.16, 3H, *d*, *J* = 7 Hz; C33, 0.60, 3H, *d*, *J* = 7 Hz; C32, 1.02, 3H, *d*, *J* =7 Hz; C40 or C43, 1.22, 3H, *t*, *J* = 7 Hz; C40 or C43, 1.28, 3H, *t*, *J* = 7 Hz; C13, 1.68, 3H, s; C31, 1.77, 3H, *s*; C30,1.96, 3H, *s*; C36, 2.06,

3H, s; C14, 2.20, 3H, *s*; C37, 3.06, 3H, *s*. It may be noted that the spectrum, relevant to a future analysis of the conformational situation in solution, not considered here, shows retention of all functional groups including the unaltered (*E*) vinyl ether bridge component. Dimethyl azodicarboxylate and related, di*iso*propyl and dibenzyl esters, for example, all exhibited similar reactivity with respect to the Alder-Ene reaction with Rifamycin O **1**. The Alder-Ene reactions were quantitative, a single minor by-product being formed in each case; typical ratios being approximately 5:1. 1HNMR data were collected on a Bruker AV III 400 MHz spectrometer.

Compound **2** was converted with modest yield into **3** by employing the general reductive amination procedure of Cricchio and Tamborini as described in [9], in which, interestingly, the amine acts a reducing agent. An excess of dimethylamine methanol solution was added by syringe to compound **2** in dry degassed THF and this was then left in the dark at 50 ◦C for a week. The THF was removed and the reaction product was dissolved in ethyl acetate and shaken with 7.4 pH phosphate buffer. The acetate layer was washed with water, dried, and the solvent evaporated to give **3**. Crystallisation of **3** proved challenging. However, small colourless single crystals of a plate morphology suitable for X-ray analysis were obtained from a CHCl3 solution.

X-ray intensity data for **3** were collected at 100(1)K on a Rigaku Oxford Diffraction SuperNova Dual-flex AtlasS2 diffractometer equipped with an Oxford Cryosystems Cobra cooler using Cu *K*α radiation (λ = 1.54178 Å). The crystal structure was solved with SHELXT-2018/2 [10] and refined with SHELXL-2018/3 [11]. Hydrogen atoms bound to carbon were placed at geometrically calculated positions with Cmethine-H = 1.00 Å, Cmethylene-H = 0.99 Å, Cmethyl-H = 0.98 Å, Caromatic-H = 0.95 Å. These hydrogen atom positions were refined using a riding model with *U*iso(H) = 1.2 *U*eq (C) (1.5 *U*eq (C) for methyl groups). Methyl group torsion angles were allowed to refine whilst maintaining an idealized tetrahedral geometry. Heteroatom (N-H, O-H) hydrogen atoms were located via a difference Fourier synthesis and their positions and isotropic temperature factors were allowed to refine freely. Values of the Flack *x* parameter [12] were obtained from the final refinement cycle of SHELXL. Two values were calculated, the first using the TWIN and BASF instructions and the second using the Parsons method of Intensity Quotients [13]. The Hooft *y* parameter [14–16] was calculated through the implementation in the program PLATON [17]. Details of the sample, data collection and structure refinement are given in Table 1. Crystal packing and structural overlay figures were produced using the CCDC program Mercury [18].


**Table 1.** Sample, data collection and structure refinement for compound **3**.


**Table 1.** *Conts*.

CCDC 2045594 contains the supplementary crystallographic data for compound **3**, which can be obtained free of charge from The Cambridge Crystallographic Data Centre, see www.ccdc.cam.ac.uk/structures.

Data/restraints/parameters 10,050/0/650 GOF, (S) on *F*<sup>2</sup> 1.035 *R*<sup>1</sup> [*I* > 2σ(*I*)] 0.0422 *wR*<sup>2</sup> (all data) 0.1140 Flack *x* parameter (refined) −0.003(18) Flack *x* parameter (from 3814 quotients) −0.009(9) Hooft *y* parameter −0.008(6) Min/max residual density (e Å<sup>−</sup>3) 0.658/−0.481 CCDC deposition number 2,045,594

#### **3. Results**

Small colourless crystals of **3** exhibiting a plate morphology were obtained from slow evaporation of a chloroform solution. The asymmetric unit of the structure consists of a single fully ordered molecule of compound **3** and a single fully ordered molecule of chloroform as a solvate. The structure refined very well in the non-centrosymmetric Sohnke orthorhombic space group, *P*212121 and gave a final residual *R*-factor based on the observed data of *R*<sup>1</sup> [*I* > 2σ(*I*)] = 4.22 %. Figure 1 shows the asymmetric unit viewed obliquely from below the plane of the basal naphthenic moiety. Figure 2 shows a view of molecule of compound **3** with -CH hydrogen atoms removed for clarity and intramolecular contacts as green dashed lines; this view is obliquely down onto the plane of the basal naphthenic moiety. Selected torsion angle and intermolecular contact distances are listed in Table 2 along with comparative data for Rifamycin O, **1** and Rifamycin S, **4** (CSD codes PUTDUD [1] and PAFRAP [19]). Geometric hydrogen bond data are given in Table 3. The structure is zwitterionic, reflecting the high acidity of the OH group on C8, see for example [20–22]. The substituted 1,2-Dihydro-naphtho[2,1-*b*]furan moiety defined by atoms C1 to C12, O3 is planar with an r.m.s. deviation of the fitted atoms of 0.0586 Å, with atom C2 showing the greatest deviation from planarity, −0.124(3) Å. The single chloroform solvate molecule in the symmetric unit forms two short C-H···O interactions of [H46···O1, 2.395 Å] and [H46···O14, 2.255 Å]. There is possibly a small rotational disorder component to the solvent molecule, as evidenced by the small difference density maxima located near the chlorine atoms.


**Table 2.** Selected torsion angles (◦) and intramolecular contact distances (Å) for compound **3** and related structures \*.

\* Structural data for Rifamycin O and Rifamycin S, CSD codes PUTDUD and PAFRAP are taken from [1] and [19], respectively.

**Table 3.** Intra and intermolecular hydrogen bond data (Å,◦) \*.


\* Symmetry operations; - −*x*, *y*−1/2, −*z* + 1/2, --−*x* + 1, *y*−1/2, −*z* + 1/2.

The packing of molecules in the crystal is governed by the formation of two intermolecular hydrogen bond interactions. The first interaction is a hydroxy hydrogen, -OH, acting as a donor to a furanone carbonyl oxygen atom acting as an acceptor [O9-H9···O4, 2.750(4) Å]. The second interaction is an amide hydrogen, -NH, acting as a donor to a carbonyl oxygen atom acting as an acceptor [N3-H3···O11, 2.771(4) Å]. Both interactions use the same 21 screw axis symmetry operation along the *b*-axis, the second interaction is translated by one-unit cell along the *a*-axis, thus linking the molecules into a crosslinked infinite chain parallel to the *b*-axis of the unit cell, as shown in Figure 3. Details of the hydrogen bond interactions are given in Table 3.

**Figure 1.** A view of the asymmetric unit of the crystal structure showing the atom numbering scheme employed. In this figure, the structure is viewed obliquely from below the naphthalene ring. Anisotropic atomic displacement ellipsoids for the non-hydrogen atoms are shown at the 50% probability level. Hydrogen atoms are displayed with an arbitrary small radius.

**Figure 2.** A view of a molecule of compound **3** from the crystal structure showing the atom numbering scheme employed and the intramolecular hydrogen bonds as dashed lines. In this figure, the structure is viewed obliquely from above the naphthalene ring. C-H hydrogen atoms have been removed for clarity.

**Figure 3.** A view of part of the crystal packing of compound **3**, showing the formation of a layer of two crosslinked infinite chains parallel to the *b*-axis of the unit cell. Inter- and intramolecular hydrogen bond interactions are shown as dark and light blue dashed lines, respectively. Incomplete hydrogen bonds are shown as red dashed lines.

#### **4. Discussion**

The absolute stereochemistry of a single crystal of **3** has been determined through the anomalous dispersion effect on the diffracted beam intensities. This result was greatly enhanced by the fact that the crystal was a mono chloroform solvate since the anomalous scatting coefficients for the chlorine atoms are much larger than those for C, N and O for Cu *K*α radiation. For the structure as presented with the chiral centres C12, C18, C21, C22, C23, C24, C25, C26, C27 in the *S*, *R*, *R*, *S*, *R*, *R*, *S*, *R*, *S* configuration, respectively, the Flack parameter = −0.003(18). Determination of the absolute structure using Bayesian statistics on Bijvoet differences (Hooft method), reveals that the probability of the absolute structure as presented being correct is 1.000, while the probabilities of the structure being a racemic twin or false are both 0.000. The Flack equivalent and its uncertainty calculated through this program was *y* = −0.008(6). This calculation was based on the values of 4497 Bijvoet differences. The post refinement method based on 3814 intensity quotients (Parsons method) gave a value of *x*= −0.009(9). It can be seen that all three methods are in good agreement (with the exception of the standard uncertainty value which is approximately a factor of two greater for the refined parameter) and that the absolute stereochemistry of compound **3** is well defined. As can be seen, the molecule has an *R* configuration at C18 and the introduced double bond between C19 and C20 has a *cis* (*Z*) configuration. Since **3** came directly from the pure major product of the Alder-Ene reaction, this establishes that the major product has the structure **2** as formulated. A salient feature is the wide separation of O1-O10 and of O2-O9, these distances having increased by 3.994 and 4.663 Å from those of Rifamycin O, **1**, in the crystal. Also striking is the location of the methyl group, (C33), attached to C24; the shortest contacts to naphthalene ring atoms are 3.474(5) Å to C2 and 3.383(5) Å to C3. A comparison of the seventeen ring torsion angles for **3** [23], with those of its ultimate precursor **1** shows that the ring torsion angles close to the aromatic ring are only modestly changed; and values (those for **1** given first) for C2-N1-C15-C16, N1-C15-C16-C17, O3-C12-O5-C29, C12-O5-C29 C28 are −176.4◦, −173.3(3)◦; 63.6◦, 59.8(4)◦; −81.7◦, −85.2(3)◦; 65.9◦, 66.0(4)◦, respectively. The key 1,3-diol component of the ansa chain maintains its stereochemical integrity with values for C21-C22-C23-C24 of 63.2◦ and 55.0(4)◦, along with accompanying values for C20-C21-C22-C23 of -172.1◦ and -176.0(3)◦. For the structurally unaltered part of the ansa chain running from C21 to O5, the most dramatic change is found for torsion C25-C26-C27-C28, 56.0◦ and −171.6(3)◦, respectively. Close to C18, massive changes are consequent upon double-bond migration, and for C16- C17-C18-C19, C17-C18-C19-C20, C18-C19-C20-C21, C19-C20-C21-C22 the corresponding values are: 36.5◦, 146.4(3)◦; −178.6◦, −136.4(3)◦; 117.4◦, −1.2(5)◦; −172.5◦, 92.8(4)◦.

A calculated overlay of compound **3**, Rifamycin O, **1** and Rifamycin S, **4** (CSD codes PUTDUD and PAFRAP, respectively) is shown in Figure 4. The overlay was computed based on all ten of the naphthalene core carbon atoms and yielded a r.m.s deviation of 0.0642Å for compound **3** and Rifamycin O, **1** and 0.0630Å for compound **3** and Rifamycin S, **4** [18]. Overlay figures for compound **3** and Rifamycin O, **1** and compound **3** and Rifamycin S, **4** [active conformation] can be found in the supplementary data. See below for details.

**Figure 4.** Overlay of compound **3** (grey), Rifamycin O (red) and Rifamycin S (orange) [active conformation]. See text for details.

#### **5. Conclusions**

New synthetic access to modified ansamycins is important for combatting mutant strains of bacterial pathogens. We have shown that whilst pure Rifamycin O is totally resistant to Diels–Alder addition, it reacts smoothly in an Alder-Ene process with diethyl azodicarboxylate and related azo compounds to give a new series of semi-synthetic rifamycins. The stereochemistry of the predominant product, **2**, of the addition reaction has been defined by conversion into a zwitterionic derivative **3** whose structure has been defined by single-crystal X-ray analysis, which established the absolute stereochemistry at C18 as having an *R* configuration, and a *Z* configuration at the introduced double bond between C19 and C20. A secondary product has been observed though not yet isolated and we have provisionally assigned to it a structure, isomeric with **2**, having an *S* configuration at C18 and a *trans* (*E*) double bond between C19 and C20. The potential anti-bacterial properties of **2** (and its isomer) and related compounds as well as those of zwitterionic **3** still remain to be determined. It is interesting to note that Rifamycin O has itself recently been found to show promise as an alternative anti-*Mycobacterium abscessus* agent [24].

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/chemistry3030052/s1, Figure S1: Structure overlay for compound **3** and Rifamycin O, **1.** Figure S2: Structure overlay for compound **3** and Rifamycin S, **4.**

**Author Contributions:** Conceptualization, D.D.M. and J.H.G.; methodology, C.S.F., D.D.M. and J.H.G.; validation, C.S.F.; formal analysis, C.S.F. and D.D.M.; investigation, C.S.F., D.D.M. and J.H.G.; resources, C.S.F. and D.D.M.; data curation, C.S.F.; writing—original draft preparation, C.S.F. and D.D.M.; writing—review and editing, C.S.F. and D.D.M.; visualization, C.S.F. and D.D.M.; funding acquisition, D.D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded Financial support from the Malaysia HIR MOHE, Grant No.F000009-21001 is gratefully acknowledged.

**Data Availability Statement:** The data is available from the CCDC program. CCDC 2045594 contains the supplementary crystallographic data for compound **3**, which can be obtained free of charge from The Cambridge Crystallographic Data Centre, see www.ccdc.cam.ac.uk/structures.

**Acknowledgments:** We thank F. Johnson (Stony Brook) for kindly providing a pure sample of Rifamycin O.

**Conflicts of Interest:** The authors declare that there are no conflicts of interest.

**Sample Availability:** Samples of the compounds are unavailable from the authors.

#### **References**


## *Article* **Erdmann's Anion—An Inexpensive and Useful Species for the Crystallization of Illicit Drugs after Street Confiscations †**

**Matthew R. Wood 1, Sandra Mikhael 1, Ivan Bernal 1,2 and Roger A. Lalancette 1,\***


**Abstract:** Erdmann's anion [1,6-diammino tetranitrocobaltate(III)] is useful in the isolation and crystallization of recently confiscated street drugs needing to be identified and catalogued. The protonated form of such drugs forms excellent crystals with that anion; moreover, Erdmann's salts are considerably less expensive than the classically used AuCl4 − anion to isolate them, while preparation of high-quality crystals is equally easy in both cases. We describe the preparation and structures of the K+CoH6N6O8 − and NH4 +CoH6N7O8 −, salts of Erdmann's. In addition, herein are described the preparations of this anion's salts with cocaine (C17H28CoN7O12), with methamphetamine (C10H22CoN7O8), and with methylone (C22H34CoN8O14), whose preparation and stereochemistry had been characterized by the old AuCl4 − salts methodology. For all species in this report, the space groups and cell constants were determined at 296 and 100 K, looking for possible thermally induced polymorphism—none was found. Since the structures were essentially identical at the two temperatures studied, we discuss only the 100 K results. Complete spheres of data accessible to a Bruker ApexII diffractometer with Cu–Kα radiation, λ = 1.54178 Å, were recorded and used in the refinements. Using the refined single crystal structural data for the street drugs, we computed their X-ray powder diffraction patterns, which are beneficial as quick identification standards in law enforcement work.

**Keywords:** Flack test; Erdmann's anion; bath salts; street drugs; cocaine; methamphetamine; methylone; π–π interactions; racemic mimics; kryptoracemic crystallization

#### **1. Introduction**

Notes: (a) Erdmann's salt should not be confused with Erdmann's reagent (sulfuric acid containing dilute nitric acid), which has been used as an alkaloid color test [1]. (b) It also should not be confused with the cis–diamino (1,2-diamino) derivative that was described by Shintani, et al. [2]. (c) For the reader's convenience, the six letter acronyms used in the references provide easy access to the Cambridge Crystallographic Database [3] information and CIF documents.

In collaboration with the Ocean County Sheriff's Office Forensic Science Laboratory (NJ, USA), we have been engaged in studies of the nature of the street drugs commonly known as bath salts [4,5], the addictive principle of which are positively charged amino species, per se, or have been converted into hydrohalides (Cl− or Br−, or mixtures thereof) in order to make them water soluble. Some of the samples used were from police seizures, which in most cases are of unknown provenance. Because an effective method of isolating and identifying them has, traditionally, been to crystallize them as salts using the expensive AuCl4 − anion, we decided to find alternative, inexpensive anions, which would be simple to make even by our first-year chemistry major or nonmajor, students. Those salts should provide equally good, hopefully better, microscopic and X-ray diffraction quality crystals

**Citation:** Wood, M.R.; Mikhael, S.; Bernal, I.; Lalancette, R.A. Erdmann's Anion—An Inexpensive and Useful Species for the Crystallization of Illicit Drugs after Street Confiscations . *Chemistry* **2021**, *3*, 598–611. https:// doi.org/10.3390/chemistry3020042

Academic Editors: Katharina M. Fromm and Catherine Housecroft

Received: 9 March 2021 Accepted: 26 April 2021 Published: 30 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

with those of the traditional gold anion samples. Given that all of the street drugs are amines, it is not difficult to assume that, in cationic form, they will readily interact, via hydrogen bonding, with moieties that can act as proton donors–acceptors.

Since a number of these drugs contain oxygen moieties that can act as bases to proton donors, an ideal crystallization partner would be one that can function equally well as either an acid or a base. Such a reagent is Erdmann's anion, which is simple to prepare in multigram quantities at a very low cost and can act both as a proton acceptor and as a proton donor to various cationic drugs. Representative samples (cocaine, methamphetamine, methylone) were selected in order to demonstrate the practical use of the reagent. They were crystallized as Erdmann's salts by the addition of a 5% aqueous solution of either ammonium or potassium Erdmann's anion and a few milligrams of the target drug compound. As the potassium salt, Erdmann's salt was first described in 1866 [6]; later, Jørgensen improved the synthesis of the ammonium salt [7]. The crystal structure of K[Co(NH3)2(NO2)4)] was initially determined at room temperature by X-ray diffraction using FeK<sup>α</sup> radiation (λ = 1.937Å) in 1956 [8]. Here, we describe the crystal structures of both the potassium and ammonium salts at 100K using complete spheres of data and give a detailed description of the structures of complexes of three cationic drugs with Erdmann's anion.

#### **2. Materials and Methods**

Note: Origin of the drugs used in this study: Cocaine (**3**) and methylone (**5**) were obtained from drug seizures. Methamphetamine (**4**) was of pharmaceutical grade (enantiopure), purchased to set up a standard for the forensics laboratory. All other chemicals were of analytical reagent grade and were obtained from Sigma Aldrich (St. Louis, MO, USA), Fisher Scientific (Waltham, MA, USA), or VWR (Radnor, PA, USA) and used without purification. Any law enforcement seizures were of unknown provenance but were characterized by GC/MS analysis.

#### *2.1. Syntheses and Crystallization*

#### 2.1.1. Syntheses of (**1**) and (**2**)

In order to have a common source of this reagent (Erdmann's anion), it was prepared in a large scale as follows: The potassium salt of Erdmann's anion K[Co(NH3)2(NO2)4], complex (**1**) (MW = 316.12 g/mol) was prepared by weighing 40.0 g of CoCl2·6H2O (MW = 237.93 g/mol) (0.168 moles) dissolved in 100 mL of distilled water with stirring. In a separate beaker, 60.0 g NaNO2 (MW = 69.01 g/mol) (0.869 moles) and 35.0 g NH4Cl (MW = 53.492 g/mol) (0.654 moles) were dissolved in 288 mL of distilled water with stirring and slight heating. This second solution was filtered through a glass frit filter. To the second solution, 12 mL (0.180 moles) of "fresh" conc. NH4OH (15 M) was added with stirring. Both solutions were combined in a side-arm flask fitted with a rubber stopper and a glass tube (1 cm in diameter) to allow air to be drawn into the mixture. Air was bubbled vigorously through the mixture for 90 min. To the mixture was added 30g KCl (MW = 74.55 g/mol) (0.402 moles), after which it turned from brown to brownish red. The product was transferred to an evaporating dish, where it was left for 2–3 days. It yielded a yellow–brown precipitate and a red–orange liquid. The solid was filtered using glass frit filter, and the precipitate was dissolved in 300 mL of distilled H2O at 60 ◦C. After 2 min, the brown solution was filtered through a glass frit filter and then cooled in an ice bath. The resulting crystals were recovered using a glass frit filter, dissolved in 300 mL of hot water, and allowed to crystallize, yielding 28.4 g (53% yield based on Co). For the ammonium salt, complex (**2**), in a different preparation, 21.5 g NH4Cl were added at the end of the procedure, instead of KCl.

#### 2.1.2. Preparation of the Drug Crystals

Complex (**3**): A sample of a few milligrams of crystalline cocaine·HCl was dissolved on a glass slide in H2O, and a single drop of a 5% Erdmann's potassium salt solution in water was added. Crystals of cocaine–trans–diamino–tetranitrocobaltiate(III) began to form as yellow needles through slow evaporative condensation at room temperature. A suitable crystal was chosen for single crystal X-ray analysis.

Complex (**4**): Several milligrams of crystalline methamphetamine·HCl were reacted with a drop of the previously prepared 5% solution of potassium Erdmann's salt on a pre-cleaned microscope slide. Yellow rods precipitated from solution and were allowed to grow at room temperature until they reached a size necessary for X-ray diffraction.

Complex (**5**): The synthetic cathinone (methylone) was also crystallized using the potassium Erdmann's salt reagent. A few crystals of crystalline methylone·HCl in water were mixed on a glass slide with the Erdmann's salt test reagent, and small yellow rods quickly grew out of the solution. A sample suitable for the single crystal X-ray diffraction experiment was chosen for analysis.

The identities of all three illicit drug specimens were previously confirmed by standard gas chromatography–mass spectrometry practices at the Ocean County Sheriff's Office Forensic Science laboratory.

#### *2.2. Crystallographic Studies*

Each of the crystals (**1**–**5**) was mounted on a Cryoloop using Paratone–N and subsequently mounted on a Bruker Smart ApexII diffractometer. Complete spheres of data were recorded at 100 K using Cu–Kα radiation, λ = 1.54178 Å. Data processing, Lorentz polarization, and face-indexed numerical absorption corrections were performed using SAINT, APEX, and SADABS computer programs [9–11]. The structures were all solved by direct methods and refined by full matrix least squares methods on F<sup>2</sup> using the SHELXTL V6.14 program package. All nonhydrogen atoms were refined with anisotropic displacement parameters; all of the H atoms were found in difference electron density maps. The methylene, methine, aromatic, and amine H atoms were placed in geometrically idealized positions and constrained to ride on their parent C atoms, with C–H = 0.99, 1.00, and 0.95 Å, respectively; the H atoms of the nitrogen and oxygen atoms were refined positionally, and their thermal parameters were fixed to be 1.2UisoN and 1.5UisoO, respectively. For (**1**) and (**2**), structural and refinement parameters and the CCDC deposition numbers can be found in Table 1; for (**3**–**5**), the parameters and CCDC numbers are found in Table 2. The hydrogen bonding results are all found in Tables T1–T5 in the Supplementary Information.


**Table 1.** X-ray Experimental Details for the K+ and NH4 <sup>+</sup> Erdmann's Salts.


**Table 2.** X-ray Experimental Details for the Three Drug Complexes with Erdmann's Anion.


#### **3. Results**

*3.1. Structures* (**1**) *and* (**2**)

3.1.1. The Potassium Salt of Erdmann's Anion (**1**)

Since the potassium and ammonium salts are isomorphous and isostructural, only the packing diagram of the potassium salt (**1**) is displayed in Figure 1.

**Figure 1.** The surroundings around the potassium cation present in (**1**). To avoid cluttering, not all of the bonded interactions are shown here. In the ammonium salt, the nitrogen is located exactly at the site of the potassium shown above, but linkages between cation and anion are via NH4 <sup>+</sup> hydrogens and the –NO2 – oxygens on the anion.

Figures 2 and 3, below, display a segment of the packing of the cations and anions in the ammonium salt. As mentioned above (Figure 1's caption), stereochemical information for the ammonium and potassium salts are interchangeable. Note that Figures 2 and 3 are, respectively, *c* and *a* projections, chosen on purpose to display the packing from different angles.

**Figure 2.** The anions form rows parallel to the *b*-axis in the structure of the ammonium salt (**2**), which are linked by the ammonium cations. Identical rows above and below the one shown here continue ad infinitum. For clarity, the complexity of cationic–anionic interactions present is minimally illustrated here. Figures 2 and 3 also illustrate the amphoteric nature of Erdmann's anion, which is the origin of its usefulness as a co-crystallization agent.

**Figure 3.** In this complicated diagram, the dotted lines define the hydrogen bonds in the ammonium salt of Erdmann's anion (**2**) in which the anions are linked, not only to the ammonium cations, but also to one another.

#### 3.1.2. Structure of the Ammonium Salt of Erdmann's Anion (**2**)

In order to show what a versatile and powerful hydrogen-bonding moiety Erdmann's anion is, we present in Figure 3 the *a*-projection of the packing diagram of its ammonium salt.

In Figure 3, many hydrogen bonds were omitted, because they either (a) clutter the picture or (b) point up or down or both. Note that the anions engage into hydrogenbonded interactions while acting both as acids (via their –NH3 ligands) or bases (via –NO2 oxygens). In fact, it is just such a versatility that makes Erdmann's anion so attractive for the purification and crystallization of street drugs, which are often contaminated or adulterated for maximizing street profit. Thus, efficient precipitating counter anions serve the dual role of purifying the adulterated material and of providing high quality crystals for X-ray analysis.

#### *3.2. Structures of the Erdmann's Complexes with Various Street Drugs* (**3**), (**4**), (**5**)

The structures of cocaine, methamphetamine, and methylone, forensically important drugs, were determined by precipitation with the anion of Erdmann's salt, followed by single crystal X-ray diffraction analyses. The former two ((**3**) and (**4**)) crystallize in Sohncke space groups, *P*1 and *P*21, respectively; thus, their absolute configurations were determined via the Flack Parameter test (see Table 2 and the deposited CIF files for details). The Erdmann's derivative of methylone (**5**) crystallized as a racemate in space group *P*−1. In all three cases, the Erdmann's anion from either the potassium or ammonium salt readily replaced the original anion (either Cl− or Br−) when a few milligrams of the drug compound were reacted with one drop of 5% aqueous solution of Erdmann's anion on a microscope slide. The resulting precipitates are the salts formed by the protonated drug cation and the Erdmann's anion, in which the four NO2 groups and the two *trans*–NH3 groups act as good bases and acids.

A Caveat: It is possible, and sometimes likely, that a crystalline sample, such as methylone, prepared as described above, may give a Flack Parameter value close to 1.0 or 0.0 [12,13], suggesting a pure chiral substance is present despite the fact that this street drug is a manmade racemate. Such result would be due to (a) crystallization in a Sohncke space group as a result of packing, as in the case of NaClO3 or sodium uranyl acetate (both space group *P*213), or (b), if Z' = 2.0, and a pair of *near-racemic* molecules caused by small differences in dissymmetry of flexible fragments caused by packing forces; in that case, the space group may be Sohncke, and the molecules crystallize as kryptoracemates. (For a discussion of the concept of kryptoracemic crystallization, see [14–16]). Additionally, the crystalline material may simply be a case of conglomerate crystallization with Z' = 2—a widely known phenomenon since Pasteur's day. In all cases, additional measurements, such as CD (circular dichroism) in the solution, etc., would have to be made to correctly interpret the results.

#### 3.2.1. Erdmann's Salt of Cocaine, C17H28CoN7O12 (**3**)

The Erdmann's salt of cocaine, C17H28CoN7O12 (**3**), crystallizes in the triclinic Sohncke space group *P*1, with two cocaine cations and two Erdmann's anions in the asymmetric unit (Figure 4). That the space group is *P*1, and not *P*−1 is guaranteed by the fact that the sample is a natural product and that the Flack Parameter test (−0.003(4)) verifies such is the case (see Table 2). There is an intramolecular hydrogen bond in each cation from the quaternary N to the carbonyl O of the methoxy carbonyl moiety: N13–H13 ... O19 is 2.838(7) Å and N14–H14 ... O23 is 2.805(7) Å. One of the cations has an H bond to an O atom on a nitro group on Co1 [N14–H14 ... O8] = 3.052(7) Å. There also exists an H bond from the nearest Erdmann's anion to the ketone [O23] to the cation N6–H6 ... O23[x + 1, y, z] = 3.103(7) Å.

**Figure 4.** The interactions between the cations and anions and cations among themselves in the cocaine–Erdmann's salt (**3**). Note that the –NH3 ligand to Co1 [N6] acts as an acid toward the—C=O oxygen base of the drug [O23], while the drug's ammonium hydrogen atoms [N13] and [N14] act as acids to the –NO2 oxygen atoms [O14 and O8] of the anionic ligands.

#### 3.2.2. The Methamphetamine–Erdmann's Complex (**4**)

The Erdmann's salt of pharmaceutical grade enantiopure methamphetamine, C10H22Co N7O8**,** crystallizes in the monoclinic space group *P*21 with Z = 4 and Z' = 2. Had the sample been a chiraly mixed, manmade sample, these crystals would constitute a case of conglomerate crystallization, because C1 (from cation 1) and C11 (from cation 2) are both (S) (see Figure 5 below). This is a case in which the Flack Parameter test [13,14] would be of great value to the authorities as a warning of the presence of a meth lab—a nontrivially useful datum for enforcing institutions.

**Figure 5.** There are two independent cation–anion pairs in the asymmetric unit of the methamphetamine specimen we examined (**4**). Note that the cations and anions are linked largely by the NH2 moieties of both cations [N13 and N14], given that the drug has no oxygen atoms of its own. Thus, the hydrogen bonding network is not as robust as it was in the case of the cocaine (see Figure 4 above and Table T4 in the Supplementary Materials). Nonetheless, the fact that the entire lattice is strongly hydrogen bonded leads to crystals of very fine quality.

In Figure 5, one of the Erdmann's anions, with the central metal Co1, makes a hydrogen bond with both proton atoms on N13 of one of the methamphetamine cations in the asymmetric unit [N13–H14 ... O8 = 2.883(8)] and [N13–H14 ... O7 = 3.207(8)] Å. Moreover, there are two other hydrogen bonds from N13 to O15 and O16: N13–H13 ... O15[x + 1, y, z] = 3.070(9) and N13–H13 ... O16[x + 1, y, z] = 3.077(8) Å, and one from N14–H15 ... O9[x, y, z − 1] = 2.851(8) Å. The second anion makes similar H bonds to a symmetry−related cation N14-H16 ... O6[x − 1, y, z − 1] = 3.182(9) and N14-H16 ... O5[x − 1, y, z − 1] = 2.921(8) Å.

#### 3.2.3. The Methylone–Erdmann's Complex (**5**)

Methylone, C22H34CoN8O14 (**5**), also forms attractive crystalline lattices with Erdmann's anions, which are useful for its detection and examination; its packing is shown in Figure 6.

**Figure 6.** Our crystals contain racemic pairs of the drug methylone (**5**), indicating that it is a manmade product, a synthesized analog of cathinone, a stimulant found in *Catha edulis.* Again, note the robust hydrogen bonding network present, in which the anion is displaying its amphoteric nature by linking the equally amphoteric cations. That feature is absent in the case of the classically used AuCl4 − anions, which are also considerably more expensive.

Methylone crystallizes with Erdmann's anion in *P*−1 triclinic space group (**5**). A pair of symmetry-related anions are joined across the inversion center of the unit cell by a hydrogen bond from N2–H4 to O3[1 − x, 2 − y,−z] = 3.28(1) Å. Additionally, the O4 atoms of the symmetric pair are joined by hydrogen bonds: N1–H1 ... O4[1 − x, 1 − y,1 − z] = 3.16(1) Å. See Table T5 for bond distances and angles. Figure 6, above, shows the asymmetric unit with an additional symmetry–related anion and cation [−x, −y, z−] present to show the hydrogen bonds and the close contacts and to demonstrate the infinite propagation of anions these close contacts allow.

#### *3.3. Packing Considerations in the Three Drug Complexes*

Methylone (**5**) crystallizes in *P*−1, with Z' = 1; thus, no special comments are needed in this case, as is obvious from Figure 6 and comments above. The complexes with cocaine and methamphetamine, however, deserve considerably more careful examination, as illustrated in what follows:

3.3.1. Overlay Diagrams of the Drug Fragments for the Complexes with Cocaine (**3**) and with Methamphetamine (**4**)

Given that cocaine and methamphetamine crystallize with Z' = 2, it was interesting to inquire in what way the two independent components differ; therefore, we resorted to the MERCURY routine of CSD [3]. The resulting overlay Figures 7 and 8 were created with *DIAMOND* [17]. Cocaine (**3**) is shown below:

**Figure 7.** Cocaine–Erdmann complex (**3**). This is an overlay of the cationic cocaine molecule2 onto molecule1. The space group is *P*212121, and Z' = 2. The program MERCURY [3] was used to overlay cation 2 onto cation 1; then, the fit was optimized. The result displayed above amply justifies the need for Z' = 2, given the significant differences in torsional angles observed.

This is a simple case of a pure optically active natural product crystallizing in a Sohncke space group with Z' = 2 because the two molecules are stereochemically flexible and, upon crystallizing, they pack more densely this way. The Flack Parameter [12,13] properly recognizes this, given the fact that the value is −0.003(4). However, there is more to this packing mode, which will be elaborated upon in the section on Racemic Mimics. Next, we consider the case of methamphetamine (**4**):

Again, as in the case of cocaine, the sample of methamphetamine was known to be chiraly pure, (since it was purchased as a standard material). Therefore, the same comments concerning racemic mimics apply in this case, and relevant comments will be made next.

**Figure 8.** Methamphetamine–Erdmann complex (**4**). The overlay here is nearly perfect; the only nonhydrogen atoms that are barely separated enough to discern are shown above. As in the case of the cocaine complex, MERCURY was used to overlay cation 2 onto cation 1, and the fit was optimized. The few labels shown are for those atoms for which the fit was poor enough to allow the observer to note the presence of both atoms.

#### 3.3.2. Racemic Mimics

Historically, it appears that an awareness of the existence of this type of crystalline material was first published in papers by a) Furberg and Hassel, who studied the crystal structure of phenyl glyceric acid slowly grown from water [18]; b) Schouwstra, who studied crystals of DL–methylsuccinic acid grown by sublimation [19] and from water solution [20]; and c) Mostad, who examined o–tyrosine crystals grown from methanol containing small amounts of ammonia to increase its solubility [21]. In all those cases, crystals of the racemate and of the optically pure material crystallized with identical cell constants; this leads to values of Z' = 1 for the racemic samples and Z' = 2 for the pure enantiomorphs.

[Caveat: because some of those lattices contained racemic pairs and had Z' = 2.0, the authors of those days [16–19] labeled them racemates. In fact, the proper term today would be kryptoracemates, but because we do not want, at this stage, to branch out into that topic, a brief but suitable discussion of this issue is given in Supplementary Materials 2, below. We thank the referee for bringing this issue to our attention.]

Given that the two lattices (kryptoracemates and Sohncke space groups), Furberg and Hassel [18] asked, "why," and, "how?" In a remarkably clear and simple answer, they indicated that the pure chiral material seemed to crystallize "*as if a twin resembling in its packing that of the true racemate*": in other words, as a "racemic twin"; thus, the name *Racemic Mimics* that later evolved. They also proposed that substances containing flexible (dissymmetric) fragments whose torsional barriers were low would make ideal candidates for the existence of such a phenomenon, and they documented additional cases [18].

(The overlay diagrams shown in this document show the extent to which torsional differences are associated with the observed Z' value of 2.0). That was a remarkably advanced concept for its day and happens to conform to what we describe in our presentation, since we have two cases of racemic mimics in the cases of the cocaine derivative and of the methamphetamine derivative of Erdmann's salts. For readers interested in more extended commentary on this and related topics, we recommend Herbstein's authoritative compendium [22].

#### a. The Case of Cocaine (**3**)

Figure 9 shows the asymmetric unit for the structure of the cocaine-Erdmann's complex.

**Figure 9.** The center of mass (0.4741, 0.4173, 0.4689) of the cocaine–Erdmann lattice is located very near to <sup>1</sup> <sup>2</sup> , <sup>1</sup> <sup>2</sup> , <sup>1</sup> <sup>2</sup> , but in *P*1, the origin is totally arbitrary, which renders the issue moot for this case. Note, however, that is not the case for methamphetamine (see Figure 10, next).

b. The Case of Methamphetamine (**4**)

**Figure 10.** The pair of cations and anions observed in the case of methamphetamine. The intersection of the dotted lines is located at 0.4749, 0.5239, 0.5182, which is also very close to <sup>1</sup> <sup>2</sup> , <sup>1</sup> <sup>2</sup> , <sup>1</sup> <sup>2</sup> , as expected for a case of a racemic mimic. Overlays, above, provide a rationale for the reason why both the cocaine and amphetamine cations can function thus.

#### 3.3.3. π–π Contacts

The criterion for meaningful contacts between aromatic fragments labeled "π–π" interactions" in the report by Janiak [23] suggests that, given the experimental data available (see Figure 7 and relevant commentary in that paper), the range of 3.3–4.6 Å is reasonable. Using that as an acceptable gauge, our compounds do not have acceptable "contacts" in that range and should be ignored, because, in both cases considered here, they are closer to 6.0 Å. However, we think it is worth pointing out that that some meaningful "residual contacts" exist and depict them below in Figures 11 and 12. This caveat is in the same spirit as that in past discussions of the existence, or lack thereof, in hydrogen bonding discussions.

**Figure 11.** π–π interactions observed in the cocaine–Erdmann's complex crystals that are not separated by simple lattice translations, in which case the aromatic fragments in question are parallel to each other; thus, the latter are ideally suited for such electronic intermolecular interactions are and ignored here.

**Figure 12.** A packing diagram for the methylone complex with Erdmann's anion (**5**) is shown above. The same comments about π–π interactions in the case of cocaine apply here.

a. Cocaine Cation with Erdmann's Anion (**3**)

The distances between the central ring (C10–C15) and the closest ring (C27- –C32- ) range between 5.78 and 5.94 Å. The second ring atoms are generated through symmetry [x, y, z + 1]. The centroid-to-centroid distance = 5.868 Å [symm = x, y, z + 1], and the angle between the ring normal and the centroid-to-centroid vector is = 85.1◦ for this pair.

The other close π–π interaction is the central ring to the upper ring in the diagram, generated through symmetry [x, y, z + 1]; these distances range between 6.79 and 6.97 Å. The centroid-to-centroid distance = 6.868 Å [symm = x, y − 1, z], and the angle between the ring normal and the centroid-to-centroid vector is 51.7◦ for this one.

b. Methamphetamine Cation with Erdmann's Anion

In crystals of methamphetamine cation with Erdmann's anion (**4**), there are no close π–π interactions other than those dictated by translations; thus, we saw no need to illustrate those in this instance.

c. Methylone Cation with Erdmann's Anion (**5**)

The distances between the central ring (C10–C15) and the closest ring (C10- –C15- ) range between 5.86 and 5.92 Å. The second ring atoms are generated through symmetry [1 − x, 1 − y, 1 − z]. The centroid-to-centroid distance = 5.889 Å [symm = 1 − x, 1 − y,1 − z], and the angle between the ring normal and the centroid-to-centroid vector is = 54.7<sup>o</sup> for this one.

The other close π–π interaction is the central ring to the upper ring in the diagram, generated through symmetry [1 − x, −1 − y, −z]. These distances range between 5.58 and 5.62 Å. The centroid-to-centroid distance is 5.889 Å [symm = 1 − x, −1 − y, −z], and the angle between the ring normal and the centroid-to-centroid vector is 49.5◦ for this one.

In conclusion, it seems that, whenever aromatic bearing fragments are not sterically hindered, π–π interactions, other than those dictated by lattice translations short enough to be meaningful, are useful in forming sturdier lattices such as observed in the cases of (**3**) and (**5**).

#### **4. Summary of Experimental Results**

The adoption of, and continued investigation of, the utility of Erdmann's salt in forensic analytical testing schemes will aid the analyst by reducing sample preparation time, as well as reduced reagent cost, and will allow easy transfer of a sample to a confirmation technique, such as infrared spectroscopy or X-ray powder diffraction. Finally, testing with Erdmann's salt is essentially a nondestructive testing technique, preserving the resulting precipitate for further analysis or courtroom presentation.

The powder diffraction patterns of the precipitation products of the potassium Erdmann's salt with each of the street drugs described here (structures (**3**), (**4**), (**5**)) (Figures S1–S3) will provide the necessary analytical confirmation to any forensic lab that is using powder X-ray diffraction in conjunction with crystal tests. These powder patterns are calculated from the single crystal structures using the powder pattern generating routine in the SHELX package. X-ray diffraction is considered a "Category A" technique by the Scientific Working Group for the Analysis of Seized Drugs (SWGDrug), due to its high discrimination capabilities [24].

#### **5. Conclusions**

It seems we have justified the use of Erdmann's anion as a valuable, easily accessible, and inexpensive reagent for the purification and co-crystallization of samples of illicit drugs confiscated in the streets or police raids. Large amounts of the ammonium and/or potassium salts can be prepared by very simple methods described above; the advantage of such a procedure is that a single, purified, large source can then be used for future forensic studies with the confidence of uniformity.

In the cases of crystallographic tests (single crystal or powder), the resulting test specimens we tested have been very satisfactory, especially when the results are subjected to the Flack Parameter test.

**Supplementary Materials:** The supplementary materials are available online at https://www.mdpi. com/article/10.3390/chemistry3020042/s1.

**Author Contributions:** S.M. was an undergraduate student who prepared the ammonium and potassium salts of Erdmann's complexes under direct supervision of R.A.L. M.R.W. obtained the drug samples, and he prepared the drug complexes under direct supervision of R.A.L., I.B., R.A.L. and M.R.W. together wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** All data are deposited in CCDC.

**Acknowledgments:** We acknowledge the National Science Foundation for NSF–CRIF Grant No. 0443538 for part of the purchase of the X-ray diffractometer.

**Conflicts of Interest:** The authors have declared that no competing interests exist.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Chemistry* Editorial Office E-mail: chemistry@mdpi.com www.mdpi.com/journal/chemistry

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-1915-9