**1. Introduction**

Crystal engineering, defined as preparation of new molecular solids with tailor-made properties by using intermolecular interactions [1], continues to draw the interest of a wide scientific community. A rational design of these solids is based on a thorough understanding of the supramolecular chemistry of functional groups, in particular those with a hydrogen bonding potential. Owing to their strength and directionality, hydrogen bonds are likely to dominate above all the other interactions. The extensive surveys of the Cambridge Structural Database (CSD) helped with the formulation of empirical guidelines concerning the design of molecular crystals [2]. A generally valid rule on hydrogen bonding states that all good proton donors and acceptors are normally engaged in interactions [3]. A new terminology has also emerged: a pair of complementary functional groups, linked via intermolecular interaction, such as a hydrogen bond, is known as a synthon [4]. A heterosynthon is composed of two different functional groups, whereas two identical groups make part of a homosynthon. A prominent example of a self-association motif is a well-known carboxylic acid dimer. Another rule concerns the synthon hierarchy: the heterosynthons are favored over the homosynthons. Recent reports agree that it is still impossible to predict the structure of the molecular solid [5,6]. In this context, a phenomenon of polymorphism is brought up. The term polymorphism describes the existence of the same compound in several crystal forms that differ in spatial arrangements of their components and some of their properties [7]. Polymorphs of the same compound

**Citation:** Podjed, N.; Modec, B. Hydrogen Bonding and Polymorphism of Amino Alcohol Salts with Quinaldinate: Structural Study. *Molecules* **2022**, *27*, 996. https://doi.org/10.3390/ molecules27030996

Academic Editor: Miroslaw Jablonski

Received: 21 December 2021 Accepted: 29 January 2022 Published: 1 February 2022

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

**Copyright:** © 2022 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/).

generally differ in lattice energies by a few kJ/mol at most [8]. As claimed by McCrone [9], the number of forms known for a given compound is proportional to the time and money spent in research on that compound. A systematic study of crystal structures of a large number of molecular solids, fueled also by the pharmaceutical industry [10,11], has revealed that at least every other molecule exhibits polymorphism [12]. It has been shown that hydrogen bonding potential only slightly increases a likelihood for the molecule to be polymorphic, whereas chiral molecules are somewhat reluctant towards crystallization in more than one crystal form [13].

Herein, the solid-state structures of salts of three amino alcohols with quinaldinic acid are presented. The structural formulae of the acid and amino alcohols are depicted in Figure 1.

**Figure 1.** Structural formulae of quinaldinic acid, 3-amino-1-propanol, 2-amino-1-butanol, and 2-amino-2-methyl-1-propanol.

The salts contained protonated amino alcohols as cations and quinaldinate ions as counter-anions. Single crystals of all were obtained inadvertently as by-products of the [Cu(quin)2(H2O)] reactions with the amino alcohol [14]. It has been observed previously that the amino alcohol OH group undergoes a spontaneous deprotonation in the presence of copper(II) complexes [15]. The resulting amino alcoholate ions coordinated to copper(II) in a chelating manner with the alkoxide oxygen serving as a bridge between two or among three metal ions. The amino alcoholate coordination probably assists in the deprotonation of amino alcohol. Some of our reaction systems provided a few more pieces of information concerning the formation of the amino alcoholate ions. The nature of the products, isolated from these reaction systems, strongly suggests a proton transfer from the OH group of the amino alcohol molecule to the NH2 group of another molecule. In the reaction below, the H2N–(CH2)*n*–OH denotes amino alcohol in general.

$$\text{H}\_2\text{H}\_2\text{N-(CH}\_2\text{)}\_n\text{-OH} \leftrightarrow \text{H}\_3\text{N}^+\text{-(CH}\_2\text{)}\_n\text{-OH} + \text{H}\_2\text{N-(CH}\_2\text{)}\_n\text{-O}^-$$

The H2N–(CH2)*n*–O<sup>−</sup> ions coordinated to copper(II), whereas the H3N+–(CH2)*n*–OH ions crystallized as salts with quinaldinate. Later, a more straightforward synthesis of these salts was sought. A reaction of quinaldinic acid with the excess of amino alcohol in methanol with no copper(II) complex involved was met with success. Two of the salts were found to be polymorphic. A detailed account of the solid-state structures follows.

## **2. Results and Discussion**

First, the common structural features of the title compounds are described. The crystal structures of all consist of NH2-protonated amino alcohol molecules as countercations and quinaldinate anions with carboxylate moiety. In all, the C–O bond lengths of the carboxylate are the same within the experimental error. Interestingly, in some structures, the quinaldinate ions deviate from planarity. For convenience, we have described this deviation as a twist angle between the carboxylate plane and the quinoline plane. Depending upon the structure, the quinaldinates can stack one upon another. Geometric parameters of the *π*···*π* stacking interactions are conventionally given by the centroid···centroid distance, dihedral angle, and shift distance [16]. Quinaldinate can participate in another interaction, a C–H···*π* interaction. All interactions involving *π* rings are given in Table 1. Both the cations and the anions possess groups that are hydrogen bond donors (NH3 <sup>+</sup> in protonated amino

alcohol) or acceptors (carboxylate and quinaldinate nitrogen) or both (OH in protonated amino alcohol). With the first two being good hydrogen bond donors/acceptors, their participation in hydrogen bonding is likely to govern the connectivity patterns in solid state. A detailed list of hydrogen bonds is given in Table 2, whereas all possible heterosynthons and their actual occurrences in the structures of the title compounds are given in Table 3.

**Table 1.** *π*···*π* stacking and C–H···*π* interactions [Å, ◦] in title compounds.


C–H···Ph[1+*x*, 1+*y*, *z*], H···*Cg* = 2.86, C–H···*Cg* = 167, C···*Cg* = 3.8079(14)

**Table 2.** Hydrogen bonds (Å) in title compounds.



**Table 3.** Heterosynthon occurrence in the structures of title compounds.

[a] Weak interaction. The N···N contact is longer than the sum of the corresponding van der Waals radii, 3.1 Å [17].

The crystal structure of **1** consists of 3a1pOHH<sup>+</sup> cations and strictly planar quinaldinate ions. All hydrogen bond donors and acceptors participate in intermolecular interactions. The quinaldinate nitrogen interacts only weakly with the NH3 <sup>+</sup> group: the corresponding N···N distance amounts to 3.108(3) Å, the value that is almost the same as the sum of the van der Waals radii for nitrogen atoms, 3.1 Å [17]. The connectivity pattern consists of two types of hydrogen bonds: the OH···−OOC and the NH3 <sup>+</sup>···−OOC hydrogen bonds. Each type occurs between the cation and the anion. The hydrogen bonding pattern produces infinite layers, which are coplanar with the *ab* plane and stack along the *c* crystallographic axis. Section of such a layer is depicted in Figure 2. The layers stack upon one another with significant *π*···*π* stacking interactions occurring between quinaldinates from adjacent layers (Figure S4). Parameters of the shortest *π*···*π* stacking interaction are Ph···Py type, *Cg*···*Cg* = 3.6571(15) Å, dihedral angle = 0.41(11)◦, shift distance = 1.346 Å.

**Figure 2.** Perpendicular view to the section of a layer of hydrogen-bonded cations and anions in the structure of **1**.

The 2-amino-1-butanol salt was found in two polymorphic forms, **2a** and **2b**. Both crystallize in a monoclinic *P* 21/*n* unit cell. The quinaldinates of **2a** are non-planar with the twist angle of 11.4(2)◦, whereas those of **2b** are nearly planar. The structures of both feature

the OH···−OOC and the NH3 <sup>+</sup>···−OOC synthons. In **2a**, a weak interaction occurs between NH3 <sup>+</sup> and OH groups. Once again, in neither of the two structures, the quinaldinate nitrogen is engaged in stronger intermolecular interactions. Its shortest contact occurs with the NH3 <sup>+</sup> group with the corresponding N···N distance being 3.1669(16) Å (**2a**) or 3.080(3) Å (**2b**). Hydrogen bonds link cations and anions into layers (polymorph **2a**, Figure 3) or into chains (polymorph **2b**, Figure 4). In **2a**, significant *π*···*π* stacking interactions occur between quinaldinates from adjacent layers (Figure S5). Parameters of the shortest *π*···*π* stacking interaction are Ph···Py type, dihedral angle = 0.39(7)◦, *Cg*···*Cg* = 3.5163(9) Å, and shift distance = 1.123 Å. The packing of chains in **2b** is such that no *π*···*π* stacking occurs.

**Figure 3.** A perpendicular view to the layer in **2a**.

**Figure 4.** Section of a chain in **2b**.

The 2-amino-2-methyl-1-propanol salt also exists in two polymorphic forms. The one that crystallizes in a monoclinic *P* 21/*n* cell was labeled **3a**, and the one that crystallizes in

a triclinic *P*−1 cell was labeled **3b**. The quinaldinates of the **3a** polymorph are non-planar with the twist angle of 25.48(10)◦. Apart from the usual synthon, the NH3 <sup>+</sup>···−OOC hydrogen bond, there is a short contact between the hydroxyl group of the 2a2m1pOHH<sup>+</sup> cation and the quinaldinate nitrogen with the O···N distance being 2.8187(17) Å. The NH3 <sup>+</sup>···−OOC and the OH···N(quin−) hydrogen bonds link ions into chains, which propagate along *a* crystallographic axis (Figure 5). The chains pack in a parallel fashion without any *π*···*π* stacking interactions.

**Figure 5.** Section of a chain in **3a**.

The **3b** polymorph also consists of infinite chains. The chains propagate along *b* crystallographic axis. Yet, the hydrogen bonding motif markedly differs from that in **3a**. Firstly, the quinaldinate nitrogen is engaged in weak interaction with the adjacent NH3 <sup>+</sup> moiety. The corresponding N···N contact is 3.1037(14) Å. In the infinite chain, the following synthons may be recognized: in addition to the usual NH3 <sup>+</sup>···−OOC and OH···−OOC hydrogen bonds, there is also the NH3 <sup>+</sup>···OH hydrogen bond that links the cations (Figure 6). Of the two polymorphs, only **3b** displays hydrogen-bonding interactions between the cations. The packing of the chains is such that it allows *π*···*π* stacking interactions between neighboring chains (Figure S6). The quinaldinates are again non-planar with the 17.26(9)◦ twist angle.

Products obtained upon a direct reaction of a specific amino alcohol and quinaldinic acid may be classified as salts. The combinations involving amines and carboxylic acids do not always produce salts. The frequently employed ΔpKa rule in predicting the nature of the product [18], ionic (a salt) or neutral (a co-crystal), can give indefinite answers. It has been stated that with the difference between the pKa of the base and the pKa of the acid in the −1 to 4 interval, the ionization of functional groups depends upon the whole crystal packing [18], and the product classification depends upon the position of the proton along a N···O hydrogen bond [19]. The combinations of amino alcohols, used in place of amines, and quinaldinic acid (quinoline-2-carboxylic acid) result in the ΔpKa values that do not fall into the −1 to 4 domain. Although the hydroxyl group lowers the pKa value relative to the "parent" alkylamine (For example, pKa of 2-aminoethanol is by 1.15 unit lower than pKa of ethylamine, 9.50 vs. 10.65 [20]), it is the quinaldinic acid that swings the balance in favor of the salt formation. The salt formation was further confirmed for all title compounds in the process of structure refinement by the location of proton in the electron difference maps.

**Figure 6.** Section of a chain in **3b**.

The components of the three salts contain the same functional groups. Similar connectivity patterns are thus expected. The following discussion shows to what extent this expectation was realized. It is to be noted that three compounds present a very limited data set. The general validity of the conclusions is thus to be treated with caution. Firstly, in the solid-state structures of title compounds, all good proton donors and acceptors are used in the intermolecular connectivity. All five structures conform to the predicted synthon hierarchy [2]: only the heterosynthons may be displayed and no homomeric ones. As shown in Table 3, all our salts feature the NH3 <sup>+</sup>···−OOC synthon. The second one in the order of occurrence is the OH···−OOC synthon, which is observed in all but **3a**. Interestingly, its formation is with no exception accompanied by a weak NH3 <sup>+</sup>···N(quin−) interaction. The **3a** salt, which lacks the OH···−OOC interaction, also lacks the NH3 <sup>+</sup>···N(quin−) interaction. The absence of the NH3 <sup>+</sup>···N(quin−) interaction in **3a** is compensated by the OH···N(quin−) hydrogen bond. The salt **3a** is the only compound that demonstrates this type of hydrogen bond; **3b**, the other (2a2m1pOHH)quin polymorph, also displays a specific feature, a NH3 <sup>+</sup>···OH interaction. The latter is of interest because it occurs between ions of the same type, i.e., the 2a2m1pOHH+ cations. The survey reveals that **1** and **2b** feature the same heterosynthons. The same observation pertains to the **2a**/**3b** pair. The **3a** polymorph differs from the other four structures. According to the literature, each pair of polymorphs, the **2a**/**2b** polymorphs and the **3a**/**3b** polymorphs, with differences in hydrogen bonding between their components may be thus classified as hydrogen bond isomers of the same solid [21]. The **2a**/**2b** polymorphs crystallized from the same reaction mixture, as opposed to the **3a**/**3b** polymorphs, which crystallized from different reaction mixtures. The **2a**/**2b** polymorphs are therefore concomitant polymorphs [22]. The structures of **2a** and **2b** reveal another important difference. Whereas **2a** features *π*···*π* stacking of quinaldinates, this type of interaction is lacking in **2b**. The same difference pertains to the **3a**/**3b** pair. On the other hand, the structures of all four share a common feature: the C–H···*π* type interactions.

The structures of **1**–**3b** have some structural features in common. The observed differences are a result of a complex interplay of short- and long-range intermolecular interactions that govern the supramolecular assembly during the crystallization procedure. Yet, each structure thus presents a specific situation and as such conforms with the current opinion in the field of crystal engineering that it is impossible to predict all molecular recognition events during the crystallization.
