Salts and Co-Crystals of Gentisic Acid with Pyridine Derivatives: The

Oct 1, 2012 - E-mail: [email protected]. Cite this:Cryst. Growth Des. ... Nikola Bedeković , Vladimir Stilinović , and Tomislav PiteÅ¡a. Cryst...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/crystal

Salts and Co-Crystals of Gentisic Acid with Pyridine Derivatives: The Effect of Proton Transfer on the Crystal Packing (and Vice Versa) Vladimir Stilinović* and Branko Kaitner Department of Chemistry, Laboratory of General and Inorganic Chemistry, Faculty of Science, Horvatovac 102a, University of Zagreb, 10002 Zagreb, Croatia

Downloaded via KAOHSIUNG MEDICAL UNIV on July 25, 2018 at 14:30:38 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: A series of 22 salts and co-crystals of gentisic acid and pyridine derivatives were synthesized and their crystal structures were studied by X-ray diffraction. The proton transfer was found to to be absent when the difference in pKa values was below 2, and it always occurred when it was above 2.5. The hydrogen bonds between the carboxylic group and pyridine nitrogen atom were found in all structures but one. These hydrogen bonds are shorter and apparently stronger in cocrystals than in salts. In the structures of salts the most significant intermolecular interaction is hydrogen bonding between anions, usually leading to chains or dimers, whereas in co-crystals the direct bonding between neutral gentisic acid molecules is mostly absent. Using 4,4′-bipyridine, two desmotropes were obtained in which the difference in hydrogen bonding of gentisic acid molecules leads to different protonation in two structures.



INTRODUCTION Since the mid 1960s, the question of intermolecular proton transfer, i.e., whether a given combination of substances will crystallize as a salt or as a co-crystal, is among the most commonly recurring topics within the study of multicomponent solids.1 The question of salt/co-crystal formation is particularly significant in the case of multicomponent systems containing active pharmaceutical ingredients (API) since the exact control of the composition of a pharmaceutical solid is a necessary prerequisite for its production and practical application.2 A survey of 85 salts and co-crystals formed from a stoichiometric mixture of co-formers has shown that co-crystals exhibit a lower tendency for forming solvates and compounds of unexpected stoichiometries than co-crystals.3 This is attributed to the increase of the potential for hydrogen bonding in charged species, as they can form charge assisted hydrogen bonds, and are therefore more likely to form hydrogen bonds with additional neutral molecules, such as solvent molecules, additional molecules of (protonated) acid or (unprotonated) base, etc.3 Since many pharmaceuticals have carboxylic acid or pyridine functionalities, the proton transfer along hydrogen bonds between a carboxylic oxygen and a pyridine nitrogen atom has drawn considerable attention.4 In solution, an absolute measure of proton transfer between two molecules is the difference of their respective pKa values (for a given solvent and temperature).5 However, since a solution pKa value is dependent on specific interactions of both the protonated and deprotonated species with the solvent molecules (resulting in specific enthalpic and entropic effects), it provides only an approximate measurement of the absolute acidity of a given species. The assessment of (Brønsted) acid− base behavior in the crystalline state based on solution pKa © 2012 American Chemical Society

values is therefore only partly possible, especially since the behavior of hydrogen donors and acceptors in the crystalline state is greatly affected by the characteristics (steric parameters, polar and dispersive interactions) of their specific environments in the crystal structure. In spite of this, solution pKa values do somewhat reflect the proton affinities in the solid state, and a number of empirical rules for prediction of proton transfer based on the difference of pKa values of proton acceptor and donor [ΔpKa = pKa(protonated base) − pKa(acid)] have been suggested.6 The majority of investigators seem to be in agreement that ΔpKa greater than 3 leads to the formation of salts, negative ΔpKa will almost exclusively result in co-crystal formation, while for the interval 0 > ΔpKa > 3 no definite predictions of proton position (and therefore of salt/co-crystal formation) can be made.7 The width of this ΔpKa interval in which proton transfer may or may not occur (3 pKa units) demonstrates the effect of the local interactions in crystals which alter the acidity/basicity of molecules as compared to their behavior in solution. The position of the ΔpKa interval on the ΔpKa scale (its mean value is not 0 as might be expected, but rather 1.5) indicates a systematic bias towards co-crystals (i.e. solid systems in which proton transfer is absent), since apparently an acid must be considerably stronger (its dissociation constant must be larger, up to thousand times) in order to protonate a base in a crystal structure than in solution! In this study of proton transfer phenomenon we have prepared a library of 22 compounds by crystallizing gentisic Received: August 31, 2012 Revised: September 27, 2012 Published: October 1, 2012 5763

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

hydrogen donors, hydrogen acceptors), we were able to obtain a quite coherent picture of the interrelationship between hydrogen transfer, hydrogen bonding within the acid-pyridine pair and supramolecular architecture in this group of compounds.

acid with a number of pyridine derivatives. Gentisic acid (2,5dihydroxybenzoic acid; gentH) is a nonsteroidal antiinflammatory drug with antiaging properties.8 From the crystal engineering point of view, it is an interesting molecule due to its combination of functional groups which can participate in hydrogen bonding and thus influence (and be influenced by) proton transfer (Scheme 1). It possesses a carboxylic group



RESULTS AND DISCUSSION Pyridine derivatives chosen for this study are shown in Scheme 2. They span a pKa range from 2.1 (4-cyanopyridine) to 7.5 (2,4,6-trimethylpyridine)10 and can be divided into four groups: those without additional hydrogen donors or acceptors (1−6), those with a hydrogen donating group in the o-position which can form an additional hydrogen bond to the carboxylic group forming a two-point 2-aminopyridine−carboxylic acid11 (7− 11) or 2-hydroxypyridine−carboxylic acid heterosynthon, those with hydrogen donors in other positions (11, 13−16), and those with hydrogen acceptors (10−20). The latter group includes the only bipyridine (4,4′-bipyridine, 19) used. Proton Transfer along the Gentisic Acid−Pyridine Hydrogen Bond. In all structures but one (14·gentH) the hydrogen bond between the gentisic acid carboxyl group and the pyridine nitrogen is present, either as a bond between neutral (COO−H···N pyr ) or between charged species (COO−···H−N+pyr). In order to determine whether proton transfer has occurred, it is necessary to unequivocally determine the positions of protons. Unfortunately, the positions of hydrogen atoms determined from electron density maps obtained by X-ray diffraction are often somewhat uncertain, due to the weakness of radiation scattering on hydrogen atoms. The correct hydrogen placement is in particularly difficult when the crystal quality is rather poor, since this leads to uneven difference maps in which hydrogen atoms may be “lost”. It is therefore often advisible to rely on the changes of the molecular geometry which occur upon proton transfer to determine the position of the hydrogen atom. This is rather simple in the case of carboxylic acids, since a protonated carboxylic group is

Scheme 1. Gentisic Acid and Gentisinate Anion with the Enumeration of Oxygen Atoms Which Will Be Used Throughout the Paper

which can be deprotonated (atoms 1 and 2 in Scheme 1), one hydroxyl group which can form an intramolecular hydrogen bond with the carbonyl group (atom 3 in Scheme 1). This renders the carboxylic group more acidic, i.e., more prone to be deprotonated, ensuring at the same time that the resulting carboxylate is unsymmetrical as its oxygen atoms are not equivalent, as one participates in the intramolecular hydrogen bond, and the other does not. Along with these, the molecule also possesses a second hydroxyl group (atom 4 in Scheme 1) which cannot interact in any intramolecular interactions and is therefore free to bond to other molecules.9 Using a single carboxylic acid and a number of relatively similar organic bases (pyridine derivatives spanning a pKa range of ca. 5.5 units and possessing various functionalities: apolar, Scheme 2

5764

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

systems where hydrogen transfer occurred and those where it is absent, can be expected to provide useful data concerning the proton transfer in the solid state. Comparing the hydrogen bond geometry in co-crystals and salts, it is evident that the hydrogen bonds are shorter in cocrystals than in salts (Figure 2a). The difference between the mean donor−acceptor distance in co-crystals (2.69(5) Å) and salts (2.60(3) Å) is somewhat smaller than 0.1 Å, which can be considered statistically significant (almost two standard deviations). Correspondingly to hydrogen bond lengths, the angles are closer to the ideal 180° in co-crystals than in salts (Figure 2b). The mean angle in salts was found to be 169(7)°, and sometimes as low as 157° (mean), while in co-crystals the mean angle is 174(4)°, and there are no examples of hydrogen bond angles smaller than 167°. The larger deviation in angles is not unexpected, as the hydrogen bond angle depends on the less precisely determined position of the hydrogen atom. From these observations follows that the hydrogen bonds in systems where hydrogen transfer is absent are systematically stronger than those between pyridinium cations and carboxylate anions. The increased hydrogen bond strength in co-crystals might be brought into connection with the fact that the ΔpKa values for co-crystals are (average 0.9 ± 1.1) for the most part lower than those for salts (average 3.8 ± 0.8), as it has been well documented that the strongest hydrogen bonds occur when both molecules are of the same acidity, i.e., when ΔpKa approaches zero (the pKa slide rule),14 as this leads to stronger, more covalent hydrogen bonds, with deeper and wider potential energy minima.1e However, by plotting the hydrogen bond parameters vs ΔpKa (Figure 3) one may notice that in the ΔpKa region where salts and co-crystals coexist (ca. 2−2.5) that, in spite of having approximately identical ΔpKa values, cocrystals still have shorter and straighter O···H···N hydrogen bonds than salts. This seems to indicate the existence of an additional influence, other than equalization of pKa values, which contributes to the increase of hydrogen bond strength in gentisic acid co-crystals, as opposed to gentisinate salts. A probable explanation of this influence can be found in the difference between the electron density distribution on the pyridine nitrogen and that on a negatively charged carboxylate oxygen atom. The lone electron pair on the pyridine nitrogen is mostly situated in front of the nitrogen atom, while the maximal electron density on the negatively charged oxygen is distributed cyclically about the axis defined by the C−O− bond.15 Because of this in the case of hydrogen bonding to a neutral pyridine, the maximal electron density on the acceptor atom is in the direction of the N···H−O hydrogen bond, while the maximal electron density on the acceptor atom in N+− H···O− hydrogen bond is distributed about the direction of the hydrogen bond, i.e., there is less electron density between the hydrogen atom and the acceptor (oxygen). Because of this the N···H−O hydrogen bond is expectedly somewhat more covalent, and therefore somewhat stronger than the N+− H···O− hydrogen bond. An equivalent explanation can be given in terms of the HSAB16 (hard and soft acids and bases) concept − H+ (a hard Lewis acid) will preferably bond to the harder (Lewis) base − O− and the softer acid thus formed (hydroxyl group) interacts with the softer base (nitrogen). Therefore the apparent stabilization of co-crystals with respect to salts can be attributed to two factors. First, there is the above-described increase of hydrogen bond strength. Second, there is the energetically unfavorable increase of distance between the hydrogen ion and the anion. While in the

asymmetric (CO bond of ca. 1.20 Å, and C−OH of ca. 1.30 Å) while the carboxylate is symmetric, with an intermediate C− O bond length of ca. 1.25 Å.12 The difference between the two carboxylic C−O bond lengths (Δd = dC7−O1 − dC7−O2) can be used as an indicator of proton transfer. According to this parameter the gentisic acid-pyridine pairs are divided into two separate groups those with mean Δd = 0.082(16) Å, and those with mean Δd = −0.02(2) Å, the first corresponding to gentisic acid molecules and the latter to gentisinate anions. This conclusion is supported by the mean distances between carboxylic oxygen and the hydrogen located from the electron difference map (dOH), with dOH = 1.06(7) Å for the first and 1.74(13) Å for the latter group (see Figure S1 of the Supporting Information). Note that, as the Δd values are derived from more precisely determined heavy atom positions, their standard deviations are smaller than those of dOH values by a factor of ca. 7. On the basis of determined Δd values it can be shown that out of 21 structures where a hydrogen bond between gentisic acid carboxyl group and the pyridine nitrogen is present, in 16 the proton transfer has occurred (hydrogen bond is of the COO−···H−N+pyr type), while in the remaining five the proton transfer is absent (hydrogen bond of the COO−H···Npyr type). There is a definite correlation between the occurrence of proton transfer and pKa values of the pyridine derivatives used (Figure 1). In gentisic acid−pyridine pairs with ΔpKa < 2 no

Figure 1. Correlation of the difference in pKa values between gentisic acid and pyridine derivatives and the difference of C−O bond lengths in the carboxyl group of the gentisic acid molecule (an indicator of proton transfer). White circles correspond to co-crystals and black to salts.

proton transfer was observed. Also, the maximal ΔpKa at which a structure without proton transfer was found was 2.45. This indicates an existence of an interval of approximately 0.5 pKa units in which proton transfer may or may not occur, and also that the pyridine, in order to deprotonate gentisic acid in the solid state, must be at least 100 times more basic than it would be necessary to do the same in solution!13 The obvious difference between proton transfer in a (diluted) solution and in the solid is that in the first case the proton transfer is mediated by the solvent, whereas in the solid the donor and the acceptor are in direct contact. Therefore, a closer examination of this contact (i.e., the hydrogen bond between the carboxyl group and the pyridine nitrogen), in particular, a detailed comparison between hydrogen bonding in 5765

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

Figure 2. Correlation of the difference of C−O bond lengths in the carboxyl group of the gentisic acid molecule (an indicator of proton transfer) and O···H···N hydrogen bond (a) length and (b) angle. White circles correspond to co-crystals and black to salts.

Figure 3. Correlation of the difference in pKa values between gentisic acid and pyridine derivatives and O···H···N hydrogen bond (a) length and (b) angle. White circles correspond to co-crystals and black to salts. Dashed lines are drawn at ΔpKa 2 and 2.5. Only those structures where hydrogen atom was located from the electron difference map are taken into account.

O3 (0.35), and finally O4 (0.15). The N+−H···O− bond connecting the ion pair is more commonly formed with O1 (72%) than with O2, although the probability of O2 participating in this bond increases to ca. 50% in the case of 2-aminopyridinium cations as they always form two N−H···O hydrogen bonds with both O1 and O2 atoms of the carboxylate. O1 is also the favored proton acceptor from a neighboring ion pair (47%), followed by O2 (29%), O3 (18%) and O4 (6%). The situation among co-crystals is expectedly quite different − a gentisic acid molecule is on average an acceptor of only 0.56 hydrogen bonds, of which 0.28 with O2 as the acceptor and 0.14 with O3 and O4 each. Interestingly, the large increase of the average number of hydrogen bonds a gentisinate anion forms with its neighbors (other than the pyridinium counterion) does not lead to a larger diversity in supramolecular architectures. Rather the contrary, the majority of salts fall into two groups of structures according to the predominant hydrogen bonding motifs. The first group consists of structures where two gentisinate anions are bonded through two O4−H···O1 hydrogen bonds into a centrosymmetric (or pseudocentrosymmetric) dimer, with pyridinium cations bonded to O1 atoms of both anions

solution the resulting ions are often better solvated then the neutral molecule, which makes the dissociation more favorable, in the solid there is no such effect. H+ and the anion are in a direct contact in the solid, unlike in solution where they are separated by solvent molecules. All this makes it more difficult for deprotonation of the acid to occur in the solid than in solution, even though the base is in direct contact with the proton. The Effect of Proton Transfer on Hydrogen Bonding and Supramolecular Assembly. The simplest method of gaining insight into the effect deprotonation on hydrogen bonding potential of gentisic acid is to compare the average number of hydrogen bonds formed by a gentisic acid molecule (in co-crystals) to the number formed by gentisinate anion (in salts). In both cases all hydrogen donors participate in hydrogen bonding, so the molecules’ ability to act as hydrogen acceptor can be expected to be more revealing. A gentisinate anion was thus found to act on average as a hydrogen acceptor for 2.9 hydrogen bonds, of which one is bond with the pyridinium N−H and the remaining 1.9 bonds with other ion pairs. A further analysis demonstrates that the preferred acceptor atom is O1 (1.65 bonds), followed by O2 (0.75), 5766

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

Figure 4. (a) A centrosymmetric dimer in the structure of 4H·gent (b) pseudocentrosymmetric dimers connected via N2−H2n···O3 hydrogen bonds into a double chain in the structure of 10H·gent.

(Figure 4). This arrangement was found in four crystal structures: 4H·gent, 9H·gent, 10H·gent, and 18H·gent. Since 9 and 10 are 2-aminopyridines, they have an additional hydrogen bond donor capable of further interconnecting the dimers. In both cases this is achieved via N2−H2n···O3 hydrogen bonds, although in the case of 9H·gent they are somewhat longer (3.16 Å). In the second group, structures comprise chains of hydrogen bonded gentisinate anions, with pyridinium cations bonded to the O1 atom of the caboxylate group (in one structure, that of 11H·gent, to O2). Such chains can bee seen in crystal structures of 1H·gent, 2H·gent, 3H·gent, 5H·gent, 6H·gent, 7H·gent, 11H·gent, 13H·gent, 19H2·gent2, and 19H·gent·gentH. The formation of chains is achieved through hydrogen bonding of 5-hydroxyl group of a gentisinate usually to the neighboring gentisinate, most commonly to O2 of (in five cases) but also sometimes to O1 (in three) and also to hydroxylic O3 (in three). The geometry of chains is very variable and does not seem to be influenced by the choice of the acceptor atom. When the cation has additional hydrogen donors, they bond to gentisinate ions either within the same chain (13H·gent, Figure 5c), or in neighboring chains, which leads to 2D or 3D hydrogen bond networks. There are only two salts in which structural motifs other than rings and chains as described above are absent. In 6H·gent four gentisinate anions are connected via O4−H···O1 and O4− H···O4 hydrogen bonds into cyclic tetramers (Figure 7a). Such an arrangement allows for a lower steric hindrance by the omethyl groups of the pyridine, and is also stabilized by two additional N2−H3n···O1 hydrogen bonds between the neighboring ion pairs. The tetramers are further interconnected into layers by N2−H3n···O4 hydrogen bonds. The other salt with unusual hydrogen bonding is 10H·gent·0.5 H2O where the ring motif present in the anhydrous 10H·gent is disrupted by water molecules which bridge between three gentisinate anions generating assemblies comprising two cations, two anions and a water molecule. These assemblies are further connected into ca. 2.3 nm wide chains of approximately coplanar molecules (Figure 7b), with water molecules bridging to gentisinate anions of neighboring chains, interconnecting them into layers. The structures of co-crystals show a significantly greater diversity than those of salts. The only co-crystal found to have hydrogen bonding motif similar to chains in salts, is 12·gentH (Figure 8a), where the gentisic acid molecules are connected via O4−H···O3 hydrogen bonds, although these bonds are considerably longer (3.06 Å) than the equivalent bonds in

Figure 5. Examples of chain motifs in structures of pyridinium gentisinates: (a) in the structure of 4H·gent with gentisinates bonded via O4−H···O1 hydrogen bonds; (b) in the structure of 2H·gent with gentisinates bonded via O4−H···O2 hydrogen bonds; (c) in the structure of 13H·gent with gentisinates bonded via O4−H···O3 hydrogen bonds.

crystals of salts (2.77 Å in 13H·gent, 2.81 Å in 7H·gent). The molecule of 12 is bonded to a gentisic acid molecule by two hydrogen bonds, where the carbonyl group is the donor one (O1−H···O5), and the acceptor in the other (N1−H···O2). Note that this arrangement requires for the nitrogen atom of 12 to be protonated as well as its o-hydroxy to be deprotonated. In fact, this distribution of protons is in agreement with bond lengths in the molecule of 12 (C−O bond of 1.243 Å, neighboring C−C bond of 1.427 Å) which indicate that the molecule is present as 2-pyridone, rather than 2-hydroxypyridine tautomer. Unlike salts, which all have 1:1 stoichiometry, two co-crystals, (17)2·gent H and (20)2·gent H, have formed with a 2:1 stoichiometric ratio of pyridine and gentisic acid. These cocrystals were the only detected products of crystallization from a solution of gentisic acid and respective pyridine derivatives, even when a large excess of the acid was present. Both 17 and 20 have additional hydrogen acceptor groups other than the 5767

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

Figure 6. Interconnection of pyridinium-gentisinate chains into (a) 2D sheets in the structure of 7H·gent and (b) 3D network in the structure of 11H·gent; (viewed along the chains). In both representations one chain is highlighted green.

Figure 7. (a) A cyclic tetramer in the structure of 8H·gent. (b) Hydrogen bonding within quadruple chains which form layers in the structure 10H·gent·0.5 H2O. (Hydrogen bonding of water molecules to the neighboring chains has been omitted for clarity.)

Figure 8. (a) Hydrogen bonded chains in the structure of 12·gentH. (b) Molecular aggregates in the structure of (20)2·gentH.

pyridines, there is no bonding between gentisic acid molecules themselves. In the structure of 15·gentH there is also no direct bonding between the gentisic acid molecules, but rather there are nicotinamide molecules bridging between them. Each gentisic acid molecule is bonded to four nicotinamide molecules  it acts as a hydrogen donor in two short (O1−H···N1, to the pyridine nitrogen and O4−H···O5 to the amide oxygen of two nicotinamide molecules) and as an acceptor of two longer hydrogen bonds (N2−H2n···O3 and N2−H1n···O4 with amide NH2 groups of other two). The two shorter hydrogen

pyridine nitrogen (carboxymethyl group in 17 and cyano group in 20) which are not employed in hydrogen bonding. Instead, two molecules of pyridine bond to the acid via their pyridine nitrogen atoms, one to the carboxylic group (O1−H···N1), and the other to the 5-hydroxyl group (O4−H···N1). The first pyridine molecule is in both cases approximately coplanar with the gentisic acid molecule, while the other is coplanar with them in (20)2·gent H (Figure 8b) and at ca. 62° (dihedral angle) in (17)2·gent H. As both hydrogen donors of gentisic acid molecule are employed in hydrogen bonding with 5768

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

Crystallization from a solution containing gentisic acid and 4,4′-bipyridine yielded two types of crystals. They were always found concomitantly, and they had the same overall composition of two gentisic acid molecules per one 4,4′bipyridine molecule. In the structure of the first, 19H2·gent2, the 4,4′-bipyridine is diprotonated and hydrogen bonded to two gentisinate anions via N−H···O1 hydrogen bonds. In the structure of the second type of crystals the 4,4′-bipyridine is also bonded to two gentisic acid molecules in a quite similar manner as in the first (although there is a significant difference in dihedral angles between the planes of corresponding molecules in the two structures). However, in this case, both the position of the protons determined from the electron difference map and the bond lengths of the carboxylate group unequivocally show that 4,4′-bipyridine molecule is protonated on one nitrogen atom only, and that it is in fact bonded to one gentisinate anion (N1−H···O1 hydrogen bond) and to one neutral gentisic acid molecule (N′···H−O1′ hydrogen bond), i.e., its composition corresponds 19H·gent·gentH! The two crystalline forms thus cannot be properly described as polymorphs, since they do not comprise same molecular species, but rather the appellation desmotropes18 should be applied. The compound itself may be considered as a co-crystal of a salt (4-(4-pyridyl)-pyridinium gentisinate) and a neutral getisic acid molecule, which is usually referred to as a ionic cocrystal.19 The cause of the difference in the protonation of the bipyridine molecule in the two structures can be seen if one is to examine the surroundings of the gentisic acid molecules (or getisinate anions) in the two structures (Figure 11). In 19H2·gent2, both gentisinate molecules are acceptors of O4− H···O1 hydrogen bonds (which lead to the formation of chains as has been shown to be usual in the structures of salts). In the structure of 19H·gent·gentH however, only the deprotonated molecule is an acceptor of a O4−H···O1 hydrogen bond, while the other one is not, as its neighbor, although at a similar distance as in 19H2·gent2, is twisted by 85° with respect to it and acts as a hydrogen donor in a O4−H···O3 hydrogen bond to another gentisinate anion. It is therefore apparent that the presence of a hydrogen bond with the carboxylic group increases its acidity and allows for a proton transfer to occur. This is not entirely unexpected, as the carboxylate is a superior hydrogen acceptor to the carboxylic group, such a hydrogen bond will have a stabilizing effect for carboxylate and therefore render deprotonation more energetically favorable, much in the same way the intramolecular hydrogen bond in o-hydroxybenzoic acids renders them more acidic than their nonhydroxylic analogues.

bonds connect the molecules into helical chains, and the longer hydrogen bonds lead to interconnection of chains into layers (Figure 9), and of layers into a 3D network.

Figure 9. Hydrogen bonded layers in the structure of 15·gentH. Hydrogen bonds with other layers are not shown.

In the structures of co-crystals with pyridine-carboxylic acids (14·gentH and 16·gentH) the nicotinic and picolinic acid molecules are present as zwitterions with protonated pyridine nitrogen and deprotonated carboxylic groups. Because of this, although the molecules are neutral, they do possess protonated pyridine nitrogen atoms capable to act as hydrogen bond donors. In 16·gentH the acceptor of this bond is a gentisic acid carboxylic group. The same carboxylic group also acts as a hydrogen donor to a carboxylate oxygen of a second picolinic acid zwitterion, while the 5-hydroxyl group is a hydrogen bond donor to a third. This leads to bonding of molecules into 2Dsheets (Figure 10a). In 14·gentH the protonated nitrogen is hydrogen bonded to a carboxylate of a neighboring nicotinic acid molecule. This leads to a chain of nicotinic acid zwitterions17 with gentisic acid molecules acting as additional bridges between carboxylate groups of adjacent nicotinic acid zwitterions (Figure 10b). Effect of Crystal Packing on Proton Transfer − Desmotropy in 4,4′-Bipyridinium Gentisinate. The existence of a ΔpKa region in which both salts and co-crystals coexist is a definite indication of the influence local surrounding in the crystal structure may exert on the acidity of gentisic acid (i.e., basicity of a pyridine derivative). A most striking example of such an influence was found in the gentisic acid−4,4′bipyridine systems.

Figure 10. Hydrogen bonding in the structures of 16·gentH and 14·gentH: (a) a sideways view of a hydrogen bonded layer in 16·gentH; (b) a hydrogen bonded chain in 14·gentH. 5769

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

Figure 11. The surroundings of the 4,4′-bipyridine molecule in (a) 19H2·gent2 and (b) 19H·gent·gentH.

Table 1. An Overview of the Pyridine Derivatives Used, pKa Values and ΔpKa Values, and Obtained Compounds pyridine derivative

pKaa

ΔpKa

formula of the obtained compound

proton transfer

supramolecular motif

pyridine 2-methylpyridine 3-methylpyridine 4-methylpyridine 3,5-dimethylpyridine 2,4,6-trimethylpyridine 2-aminopyridine 2-amino-6-methylpyridine 2-amino-3-methypyridine 2-amino-3-hydroxypyridine 3-hydroxypyridine nicotic acid nicotinamide picolic acid methyl nicotinate methyl isonicotinate 4,4′-bipyridine

5.32 6.06 5.81 6.0 6.24 7.48 6.67 7.22 7.47 5.89 5.0 4.79 3.5 5.22 3.13 6.8 4.80, 2.94

4-cyanopyridine

2.1

2.5 3.3 3.0 3.2 3.4 4.7 3.9 4.4 4.7 3.1 2.2 2.0 0.7 2.4 0.3 4.0 2.0, 0.14 2.0, 0.14 −0.7

1H·gent 2H·gent 3H·gent 4H·gent 5H·gent 6H·gent 7H·gent 8H·gent 9H·gent 11H·gent 13H·gent 14·gentH 15·gentH 16·gentH (17)2·gentH 18H·gent 19H2·gent2 19H·gent·gentH (20)2·gentH

+ + + + + + + + + + + − − − − + ++ ± −

chain chain chain dimer chain chain chain tetramer dimer chain chain pyridine chain 3D network 2D layers discrete complex dimer chain chain discrete complex

All pKa values are determined in aqueous solutions at 20 °C. The pKa values for 2-amino-3-nitropyridine and 2-hydroxy-3-nitropyridine are not known.

a





CONCLUSION

EXPERIMENTAL SECTION

The samples were prepared by dissolving equimolar amounts of gentisic acid and the corresponding pyridine derivative in a hot mixture of ethanol and water (approximately 3:2) and left to cool and evaporate. Crystals of the products appeared in one to two days and were filtered from the solution. The crystal and molecular structures of all compounds were determined by single crystal X-ray diffraction. The diffraction data were collected at 295 K for all crystals.20 Diffraction measurements were made on an Oxford Diffraction Xcalibur Kappa CCD X-ray diffractometer with graphite-monochromated MoKα (λ = 0.71073 Å) radiation.21 The data sets were collected using the ω scan mode over the 2θ range up to 54°. All the measurements were performed at room temperature. The structures were solved by direct methods and refined using SHELXS and SHELXL programs.22 The structural refinement was performed on F2 using all data. All calculations were performed and the drawings were prepared using WINGX crystallographic suite of programs.23 The crystal data are listed in Table S1 of the Supporting Information. The hydrogen atoms not involved in hydrogen bonding were placed in calculated positions and treated as riding on their parent, while those involved in hydrogen binding were located from the electron difference map whenever data quality was sufficient to render it possible. In the case of structures of 8H·gent, 10H·gent, and (20)2·gent H, due to poor diffraction hydrogen atoms could not be located in the difference map and were placed on calculated positions according to the geometry of the molecules. Also, only very small crystals of 14·gentH, and (17)2·gent H were obtained, because of which the diffraction intensities were low, which led to an decrease of

The structures of salts and co-crystals of gentisic acid with pyridine derivatives studied here demonstrate several points on the subject of interrelationship between the proton transfer and supramolecular aggregation in the solid state (Table 1). On one hand, deprotonation of gentisic acid renders it a better hydrogen bond acceptor, which in turn affects the molecular aggregation in crystals of salts, making direct binding of gentisinate anions the main building motif in salts, while essentially absent in co-crystals. On the other hand, the hydrogen bonding of a gentisic acid molecule with neighbors affects its potential for proton transfer, most clearly demonstrated by the structures of two desmotropes obtained with 4,4′-bipyridine. Even more importantly, the N···H···O1 hydrogen bond along which the proton transfer seems to have an influence on the proton transfer, as N···H−O hydrogen bonds are apparently stronger than the N+−H···O−, therefore stabilizing the cocrystal, until a pyridine used becomes sufficiently basic and the proton transfer becomes energetically favored. Therefore, a better understanding of proton transfer in the solid state (i.e., when the acid and base are in a direct contact) can be reached if one takes into account not only the strength of acid and base (indicated by solution pKa values), but also their hardness. 5770

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

data quality observable by somewhat high R factors. In the crystal structure of 16·gentH a small maximum in the residual electron density remained positioned on the 2-fold axis which indicated the presence of a small amount of solvent in the structure disordered about the 2-fold axis and was modeled as an oxygen atom (presumably of a water molecule) with an occupancy of 0.09.



Lett. 2001, 532. (n) Shan, N.; Batchelor, E.; Jones, W. Tetrahedron Lett. 2002, 43, 8721. (o) Olenik, B.; Smolka, T.; Boese, R.; Sustmann, R. Cryst. Growth Des. 2003, 3, 183. (p) Bond, A. D. Chem. Commun. 2003, 250. (q) Varughese, S.; Pedireddi, V. R. Chem.Eur. J. 2006, 12, 1597. (r) Du, M.; Zhang, Z. H.; Zhao, X. J.; Cai, H. Cryst. Growth Des. 2006, 6, 114. (s) Grossel, C. M.; Dwyer, A. N.; Hursthouse, M. B.; Orton, J. B. CrystEngComm 2006, 8, 123. (t) Bhogala, B. R.; Nangia, A. New J. Chem. 2008, 32, 800. (u) Santra, R.; Ghosh, N.; Biradha, K. New J. Chem. 2008, 32, 1673. (v) Sarma, B.; Nath, N. K.; Bhogala, B. R.; Nangia, A. Cryst. Growth Des. 2009, 9, 1546. (5) Bell, R. P. The Proton in Chemistry; Chapman and Hall: London, 1973; Rossotti, F. J. C.; Rossotti, H. The Determination of Ability Constants; McGraw Hill: New York, 1961. (6) (a) Haynes, D. A.; Jones, W.; Motherwell, W. D. S. CrystEngComm 2006, 8, 830−840. (b) Stanton, M. K.; Bak, A. Cryst. Growth Des. 2008, 8, 3856. (c) Tong, W. Q.; Whitesell, G. Pharm. Dev. Technol. 1998, 3, 215. (d) Bowker, M. J. A Procedure for Salt Selection and Optimization. In Handbook of Pharmaceutical Salts; Stahl, P. H., Wermuth, C. G., Eds.; VHCA and Wiley-VCH: New York, 2002. (e) Johnson, S. L.; Rumon, K. A. J. Phys. Chem. 2008, 69, 74. (7) (a) Bhogala, B. R.; Basavoju, S.; Nangia, A. CrystEngComm 2005, 7, 551. (b) Molčanov, K.; Kojić-Prodić, B. CrystEngComm 2010, 12, 925. (8) (a) Lorico, A.; Masturzo, P.; Villa, S.; Salmona, M.; Semeraro, N.; Gaetano, G. D. Biochem. Pharmacol. 1986, 35, 2443. (b) Bian, S.; Doh, H.-J.; Zheng, J.; Kim, J. S.; Lee, C.-H.; Kim, D.-D. Eur. J.Pharm. Sci. 2003, 18, 141. (9) (a) Luo, T.-J. M.; Palmore, G. T. R. Cryst. Growth Des. 2002, 2, 337. (b) Cohen, D. E.; Benedict, J. B.; Morlan, B.; Chiu, D. T.; Kahr, B. Cryst. Growth Des. 2007, 7, 492. (c) Adam, M. S.; Gutmann, M. J.; Leech, C. K.; Middlemiss, D. S.; Parkin, A.; Thomas, L. H.; Wilson, C. C. New J. Chem. 2010, 34, 85. (d) Frišcǐ ć, T.; Lancaster, R. W.; Fabian, L.; Karamertzanis, P. G. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 13216. (e) Vishweshwar, P.; McMahon, J. A.; Peterson, M. L.; Hickey, M. B.; Shattock, T. R.; Zaworotko, M. J. Chem.Commun. 2005, 4601. (f) Bučar, D.-K.; Henry, R. F.; Lou, X.; Duerst, R. W.; MacGillivray, L. R.; Zhang, G. G. Z. Cryst. Growth Des. 2009, 9, 1932. (g) Aitipamula, S.; Chow, P. S.; Tan, R. B. H. Acta Crystallogr., Sect. E: Struct. Rep. Online 2010, 66, o1045. (h) McMahon, J. A.; Bis, J. A.; Vishweshwar, P.; Shattock, T. R.; McLaughlin, O. L.; Zaworotko, M. J. Z. Kristallogr. 2005, 220, 340. (i) Aakeröy, C. B.; Bahra, G. S.; Brown, C. R.; Hitchcock, P. B.; Patell, Y.; Seddon, K. R. Acta Chem. Scand. 1995, 49, 762. (j) SeethaLekshmi, S.; Row, T. N. G. CrystEngComm 2011, 13, 4886. (k) Stoimenovski, J.; Dean, P. M.; Izgorodina, E. I.; MacFarlane, D. R. Faraday Discuss. 2012, 154, 335. (l) Aitipamula, S.; Shan Chow, P.; Tan, R. B. H. CrystEngComm 2009, 11, 1823. (10) The pKa values used were measured in aqueous solutions at 25 °C and were taken from the IUPAC Stability Constants Database. (11) (a) Shan, N.; Bond, A. D.; Jones, W. Tetrahedron Lett. 2002, 43, 3101. (b) Bis, J. A.; Zaworotko, M. J. Cryst. Growth Des. 2008, 8, 1169. (c) Ebenezer, S; Muthiah, P. T. Cryst. Growth Des. 2012, 12, 3766. (12) Allen, F. H. Acta Crystallogr. 2002, B58, 380. (13) This procedes from the definition of pKa as the decadic logarithm of the acidity constant. Since in solution, when ΔpKa is zero exactly one-half of the overall number of acid molecules will be deprotonated, the ΔpKa which is necessary to achieve deprotonation in solids corresponds to the logarithm of the ratio of the acidity constants. Therefore, the observed difference of two pKa units necessary to induce proton transfer corresponds to a hundredfold difference in acidity constants. (14) (a) Gilli, P.; Pretto, L.; Gilli, G. J. Mol. Struct. 2007, 844−845, 328. (b) Gilli, P.; Pretto, L.; Bertolasi, V.; Gilli, G. Acc. Chem. Res. 2009, 42, 33. (15) Gillespie, R. J.; Popelier, P. L. A. Chemical Bonding and Molecular Geometry; Oxford University Press: Oxford, 2001. (16) (a) Pearson, R. G. J. Am. Chem. Soc. 1963, 85, 3533. (b) Pearson, R. G. J. Chem. Educ. 1968, 45, 581. (c) Pearson, R. G. J. Chem. Educ. 1968, 45, 643.

ASSOCIATED CONTENT

S Supporting Information *

(1) Table with crystallographic data for all compounds. (2) Table of hydrogen bond geometries. (3) ORTEP plots with atom labels. (4) Crystallographic information files (CIF). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: +385 1 4606371. Fax: +385 1 4606341. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by grant from the Ministry of Science, Education and Sport of the Republic of Croatia (Grant No. 199-1193079-3069).



REFERENCES

(1) (a) Johnson, S. L.; Rumon, K. A. J. Phys. Chem. 1965, 69, 74. (b) Stahly, G. P. Cryst. Growth Des. 2007, 7, 1007. (c) Childs, S. L.; Stahly, G. P.; Park, A. Mol. Pharmaceutics 2007, 4, 323. (d) Li, Z. J; Abramov, Y.; Bordner, J.; Leonard, J.; Medek, A.; Trask, A. V. J. Am. Chem. Soc. 2006, 124, 8199. (e) Mohamed, S.; Tocher, D. A.; Vickers, M.; Karamertzanis, P. G.; Price, S. L. Cryst. Growth Des. 2009, 9, 2881. (f) Mohamed, S.; Tocher, D. A..; Price, S. L. Int. J. Pharm. 2011, 418, 187. (g) Steiner, T.; Majerz, I.; Wilson, C. C. Angew. Chem., Int. Ed. 2001, 40, 2651. (h) Cowan, J. A.; Howard, J. A. K.; McIntyre, G.; Lo, S. M.-F.; Williams, I. D. Acta Crystallogr., Sect. B: Struct. Sci. 2003, B59, 794. (i) Parkin, A.; Harte, S. M.; Goeta, A. E; Wilson, C. C. New J. Chem. 2004, 28, 718. (j) Cowan, J. A.; Howard, J. A. K.; McIntyre, G. J.; Lo, S. M.-F.; Williams, I. D. Acta Crystallogr., Sect. B: Struct. Sci. 2005, B61, 724. (k) Schmidtmann, M.; Gutmann, M.; Middlemiss, D. S.; Wilson, C. C. CrystEngComm 2007, 9, 743. (2) (a) Morissette, S. L.; Almarsson, Ö .; Peterson, M. L.; Remenar, J. F.; Read, M. J.; Lemmo, A. V.; Ellis, S.; Cima, M. J.; Gardner, C. R. Adv. Drug Delivery Rev. 2004, 56, 275. (b) Banerjee, R.; Bhatt, P. M.; Ravindra, N. V.; Desiraju, G. R. Cryst. Growth Des. 2005, 5, 2299. (c) Vishweshwar, P.; McMahon, J. A.; Bis, J. A.; Zaworotko, M. J. J. Pharm. Sci. 2006, 95, 499. (3) Aakeröy, C. B.; Fasulo, M. E.; Desper, J. J. Mol. Pharm. 2007, 4, 317. (4) (a) Pedireddi, V. R.; Jones, W.; Chorlton, A. P.; Docherty, R. Chem. Commun. 1996, 997. (b) Sharma, C. V. K.; Zaworotko, M. J. Chem. Commun. 1996, 2655. (c) Arora, K. K.; Pedireddi, V. R. J. Org. Chem. 2003, 68, 9177. (d) Bhogala, B. R.; Vishweshwar, P.; Nangia, A. Cryst. Growth Des. 2002, 2, 325. (e) Vishweshwar, P.; Nangia, A.; Lynch, V. M. J. Org. Chem. 2002, 67, 556. (f) Bhogala, B. R.; Basavoju, S.; Nangia, A. Cryst. Growth Des. 2005, 5, 1683. (g) Shan, N.; Bond, A. D.; Jones, W. New J. Chem. 2003, 27, 365. (h) Walsh, R. D. B.; Bradner, M. W.; Fleischman, S.; Morales, L. A.; Moulton, B.; Rodríguez-Hornedo, N.; Zaworotko, M. J. Chem. Commun. 2003, 186. (i) Bhogala, B. R.; Nangia, A. Cryst. Growth Des. 2003, 3, 547. (j) Almarsson, Ö .; Zaworotko, M. J. Chem. Commun. 2004, 1889. (k) Aakeröy, C. B.; Salmon, D. J. CrystEngComm 2005, 7, 439. (l) Sharma, C. V. K.; Broker, G. A.; Szulczewski, G. J.; Rogers, R. D. Chem. Commun. 2000, 1023. (m) Tomura, M.; Yamashita, Y. Chem. 5771

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772

Crystal Growth & Design

Article

(17) Identical bonding can be found in several salts and cocrystals of nicotinic acid; see: (a) Kavuru, P.; Aboarayes, D.; Arora, K. K.; Clarke, H. D.; Kennedy, A.; Marshall, L.; Ong, T. T.; Perman, J.; Pujari, T.; Wojtas, L.; Zaworotko, M. J. Cryst. Growth Des. 2010, 10, 3568. (b) Stilinović, V.; Kaitner, B. Cryst. Growth Des. 2011, 11, 4110. (18) (a) Elguero, J. Cryst. Growth Des. 2011, 11, 4731. (b) Rubčić, M.; Užarević, K.; Halasz, I.; Bregović, N.; Mališ, M.; Đilović, I.; Kokan, Z.; Stein, R. S.; Dinnebier, R. E.; Tomišić, V. Chem., Eur. J. 2012, 18, 5620. (c) Foces-Foces, C.; Llamas-Saiz, A. L.; Claramunt, R. M.; Lopez, C.; Elguero, J. J. Chem. Soc. Chem. Commun. 1994, 1143. (d) Holczbauer, T.; Fabian, L.; Csomos, P.; Fodor, L.; Kalman, A. CrystEngComm 2010, 12, 1712. (e) Chierotti, M. R.; Ferrero, L.; Garino, N.; Gobetto, R.; Pellegrino, L.; Braga, D.; Grepioni, F.; Maini, L. Chem.Eur. J. 2010, 16, 4347. (f) Reviriego, F.; Alkorta, I.; Elguero, J. J. Mol. Struct. 2008, 891, 325. (g) Cruz-Cabeza, A. J.; Groom, C. R. CrystEngComm 2011, 13, 93. (19) (a) Braga, D.; Grepioni, F.; Maini, L.; Prosperi, S.; Gobetto, R.; Chierotti, M. R. Chem. Commun. 2010, 46, 7715. (b) Braga, D.; Grepioni, F.; Lampronti, G. I.; Maini, L.; Turrina, A. Cryst. Growth Des. 2011, 11, 5621. (c) Braga, D.; Grepioni, F.; Maini, L.; Lampronti, G. I.; Capucci, D.; Cuocci, C. CrystEngComm 2012, 14, 3521. (d) Braga, D.; Grepioni, F.; Maini, L.; Capucci, D.; Nanna, S.; Wouters, J.; Aerts, L.; Quéré, L. Chem. Commun. 2012, 48, 8219. (e) Umeda, Y.; Fukami, T.; Furuishi, T.; Suzuki, T.; Tanjoh, K.; Tomono, K. Drug Dev. Pharm. 2009, 35, 843. (20) Proton transfer can be influenced, and even induced, by the change of temperature. In this study we have limited ourselves to the state of protonation which is observed at room temperature. Because of this the measurements of poorly diffracting crystals were not repeated at low temperature. Even though such measurements might increase the data quality, they would also potentially provide different results than the room temperature measurement, and the results obtained at different temperatures would hardly be comparable to one another. (21) CrysAlis CCD and CrysAlis RED, Version 171.32.29; Oxford Diffraction Ltd: Wroclaw, Poland, 2003. (22) Sheldrick, G. M. Acta Crystallogr. 2008, A64, 112. (23) Farrugia, L. J. J. Appl. Cryst. 1999, 32, 837.

5772

dx.doi.org/10.1021/cg301267h | Cryst. Growth Des. 2012, 12, 5763−5772