Dario Braga, Laura Chelazzi, Iacopo Ciabatti and Fabrizia Grepioni*
Dipartimento di Chimica G. Ciamician, Università degli Studi di Bologna, Via Selmi 2, 40126 Bologna, Italy. E-mail: fabrizia.grepioni@unibo.it; Fax: +39 051 2099456
First published on 2nd August 2012
L-Serine and DL-serine have been treated with oxalic acid under different experimental conditions, e.g. crystallization from solution and slurry, kneading and dry mixing, yielding the molecular salts [L-serH]2[ox]·2H2O forms I and II, [L-serH][Hox] and [DL-serH]2[ox]·2H2O, all fully characterized by powder or single crystal X-ray diffractions; the relationship between the different crystal forms, relative thermal stability, dependence on relative humidity and possible interconversions have all been explored.
A recent development of these strategies has arisen from the possibility of using co-crystallization to tackle enantiomer–racemate relationships when chiral molecules are involved.5 In a previous study we have reported the mechanochemical preparation of co-crystals of dextro- (R,R), levo- (S,S), meso- (R,S) and racemic (R,R–S,S) tartaric acid, C4O6H6, with pyrazine, C4N2H4. Pyrazine was chosen not only because it would easily bind to tartaric acid via N⋯H–O hydrogen bonds, but also because any unreacted pyrazine can easily sublimate at room temperature as a pure component,6 thus simplifying analysis and characterization of the reaction product. In a related experiment Friščić et al. had previously investigated the reaction of chiral co-crystals of caffeine·D-tartaric acid with co-crystals of caffeine·L-tartaric acid in the presence of a few drops of nitrobenzene.7
In this paper we report the result of co-assembly processes between L-serine [2(S)-amino-3-hydroxypropanoic acid] and DL-serine [2(RS)-amino-3-hydroxypropanoic acid] with oxalic acid (Scheme 1). Depending on stoichiometric ratio and/or preparation conditions, two polymorphic modifications of the dihydrated 2:
1 salt [L-serH]2[ox]·2H2O (form I and form II), one crystal form of the anhydrous 1
:
1 molecular salt [L-serH][Hox] obtained by direct mixing (grinding and kneading) and the dihydrated 2
:
1 salt of racemic DL-serine, namely [DL-serH]2[ox]·2H2O, have been obtained. Structural characterization has been possible, for the last salt and polymorph II of [L-serH]2[ox]·2H2O, via powder data.
![]() | ||
Scheme 1 Serine molecule (left) and serinium cation (right). |
It is important to stress here that distinction between molecular salts and co-crystals is usually made on the basis of H-atoms positions along the O–H⋯O or N–H⋯O hydrogen bonds. When COOH groups are involved, as in the case of oxalic acid as the co-former of choice, an analysis of C–O distances helps in distinguishing between COOH and COO− groups; chemical sense and a careful study of hydrogen bonding patterns are also needed before a sound conclusion can be reached. Although, in our experience, molecular salts and co-crystals can show extremely similar behaviour,8 the implications, especially in the pharmaceutical field, can be of outmost relevance.§ An adequate choice of the “co-former” may strongly depend on the task at-hand.
All solid-state materials have also been investigated by differential scanning calorimetry and variable temperature X-ray powder diffraction, as well as by a series of experiments in solution, in slurry and under different humidity conditions, in order to ascertain relative stability and possible interconversion pathways.
Reagents ratio | Method | Product |
---|---|---|
L-ser![]() ![]() ![]() ![]() | c, k, s | [L-serH]2[ox]·2H2O I |
L-ser![]() ![]() ![]() ![]() | g | [L-serH]2[ox]·2H2O II |
L-ser![]() ![]() ![]() ![]() | g, k | [L-serH][Hox] |
DL-ser![]() ![]() ![]() ![]() | c, g, k, s | [DL-serH]2[ox]·2H2O |
DL-ser![]() ![]() ![]() ![]() | c, g, k, s | [DL-serH]2[ox]·2H2O + DL-ser |
In order to establish a sequence of relative stability for the molecular salts obtained with L-serine, slurry experiments were performed. The results are summarized in Table 2, and show that [L-serH]2[ox]·2H2O form I is the most stable form under ambient conditions.
Starting material | Product |
---|---|
[L-serH]2[ox]·2H2O I | [L-serH]2[ox]·2H2O I |
[L-serH]2[ox]·2H2O II | [L-serH]2[ox]·2H2O I |
[L-serH][Hox] + L-ser | [L-serH]2[ox]·2H2O I |
Direct mixing of L-serine and oxalic acid by grinding or kneading in molar ratio 1:
1 invariably leads to the formation of the anhydrous species [L-serH][Hox]. In the slurry tests, crystallization from solution yields instead the 2
:
1 dihydrated adduct [L-serH]2[ox]·2H2O form I, while the excess of oxalic acid is recovered as H2ox·2H2O. The structure of the anhydrous 1
:
1 compound has been determined directly from powder diffraction data (see below).
We also performed direct reactions in the solid state in order to see whether the forms obtained could be further converted. The results are summarized in Table 3.
Preparative attempts with DL-serine invariably yielded the molecular salt [DL-serH]2[ox]·2H2O; even the direct reaction of L-serine, D-serine and oxalic acid in a 1:
1
:
1 molar ratio resulted in formation of the racemic crystal [DL-serH]2[ox]·2H2O plus DL-serine, instead of the alternative products constituted by a racemic conglomerate of [L-serH]2[ox]·2H2O and [D-serH]2[ox]·2H2O (Fig. 1).
![]() | ||
Fig. 1 Schematic representation of the relationship between the different crystal forms obtained by reacting L-serine with oxalic acid. |
A further point of interest is represented by the results of the dehydration processes on forms I and II of [L-serH]2[ox]·2H2O. Dehydration of [L-serH]2[ox]·2H2O does not result in formation of the anhydrous phase [L-serH]2[ox], either under thermal treatment or in the presence of P2O5. As shown by variable temperature powder diffraction, loss of water results in phase transition to the 1:
1 molecular salt [L-serH][Hox], accompanied by extrusion of L-serine from the crystalline material (see ESI‡); more accurate data on the transitions have been obtained from DSC (see Table 4).
Variable temperature X-ray powder diffraction | ||
---|---|---|
Phase trans.a | Product | |
[L-serH]2[ox]·2H2O I | 80–85 °C | [L-serH][Hox] + L-ser |
[L-serH]2[ox]·2H2O II | 70–75 °C | [L-serH][Hox] + L-ser |
Thermal behaviour of the molecular salts was examined by DSC measurements (see ESI‡), both in open and sealed pans. Measurements in sealed pans allow us to determine congruent melting of the hydrated phases, while in open pans loss of water is observed upon heating the hydrated samples, and the 1:
1 anhydrous phase is formed, accompanied by extrusion of L-ser. DSC measurements in sealed pans also allow us to characterize the dimorphic system as a monotropic system, because no solid–solid phase transition is observed before melting, which occurs at 87 °C for form I and at 77 °C for form II (onset values). Thermal stability of the anhydrous molecular salt [L-serH][Hox] is much higher, as melting point is observed at ca. 125 °C (onset), followed by decomposition (see ESI‡).
The behaviour of [L-serH]2[ox]·2H2O forms I and II under controlled humidity at RT was also investigated (see Table 5). If form II is exposed to high humidity (RH 84% in the presence of a supersaturated solution of KCl), it converts into the more stable form I, while no transition is observed under low humidity conditions (RH 11% with a supersaturated solution of LiCl). Exposure to a dry atmosphere, i.e. to P2O5, does not result in dehydration; this fact is better understood if the packing features of forms I and II are investigated (see below).
Reagents | RH% = 11a | RH% = 84b | P2O5c |
---|---|---|---|
a Obtained with LiCl.b Obtained with KCl.c All attempts to solve the structure from powder diffraction data have been thus far unsuccessful. | |||
[L-serH]2[ox]·2H2O I | No change | No change | No change |
[L-serH]2[ox]·2H2O II | No change | Form I | No change |
[L-serH][Hox] | No change | New phasec | No change |
[DL-serH]2[ox]·2H2O | [L-serH]2[ox]·2H2O form I | [L-serH]2[ox]·2H2O form II (powder data) | [L-serH][Hox] (powder data) | |
---|---|---|---|---|
Chemical formula | C8H16N2O10·2H2O | C8H16N2O10·2H2O | C8H16N2O10·2H2O | C5H9NO7 |
Mr | 336.26 | 336.26 | 336.24 | 195.13 |
T/K | 293 | 293 | 293 | 293 |
λ (Å) | 0.71073 | 0.71073 | 1.54056 | 1.54056 |
Crystal system | Monoclinic | Monoclinic | Monoclinic | Monoclinic |
Space group | P21/C | P21 | P21 | P21 |
a/Å | 4.8685(3) | 4.8832(6) | 12.5711(6) | 14.3723(8) |
b/Å | 17.2207(12) | 11.9276(14) | 11.2144(5) | 6.2614(4) |
c/Å | 17.2062(15) | 12.4798(15) | 5.2079(2) | 9.1815(5) |
α/degrees | 90 | 90 | 90 | 90 |
β/degrees | 91.668(7) | 92.018(10) | 100.529(3) | 92.112(4) |
γ/degrees | 90 | 90 | 90 | 90 |
V/Å3 | 1441.94(18) | 726.43(15) | 721.84(3) | 825.69(8) |
Z | 4 | 2 | 2 | 4 |
dcalc/mg cm−3 | 1.549 | 1.528 | — | — |
μ/mm−1 | 0.148 | 0.147 | — | — |
Measd reflns | 10902 | 5125 | ||
Unique reflns | 3502 | 2878 | — | — |
Rint | 0.051 | 0.073 | — | — |
GoF | 1.255 | 1.149 | — | — |
R1 (I>2σ(I) | 0.1065 | 0.0963 | — | — |
wR2 (all) | 0.2465 | 0.1693 | — | — |
Rwp | — | — | 9.31 | 10.28 |
χ2 | — | — | 3.9 | 5.95 |
The formation of all molecular salts is the result of a one- or two-proton transfer process from oxalic acid to the amino acid serine, as evidenced by an analysis of all C–O distances. On the basis of pKa values for the second deprotonation of oxalic acid (pK = 4.14) and deprotonation of the COOH group on serine (pK = 2.19), it is clear that complete deprotonation of oxalic acid is possible only in the presence of an excess of serine, and the resulting salts show indeed a 2:
1 stoichiometric ratio of serine
:
oxalic acid. Monodeprotonation of the oxalic acid is observed only in anhydrous [L-serH][Hox], in keeping with a stoichiometric ratio of 1
:
1.
Relevant crystal packing features for the four molecular salts will now be discussed and compared, starting with [DL-serH]2[ox]·2H2O and the two polymorphic forms of [L-serH]2[ox]·2H2O, which will be described together due to their packing similarities; this will be followed by a description of anhydrous [L-serH][Hox].
![]() | ||
Fig. 2 (a) Chessboard disposition of channels, formed by piling serinium cations parallel to the a-axis direction in crystalline [DL-serH]2[ox]·2H2O. (b) The serinium cations along the piles interact with each others only via CH⋯O interactions. |
The channels evidenced in light-blue are filled with water of crystallization (blue spheres in Fig. 3a); the water molecules along the channels are linked via hydrogen bonds in infinite zig–zag chains, as shown in Fig. 3b [Ow⋯Ow 2.750(5) and 2.838(5) Å]. The role of water in this structure is not limited to its filler capacity: the water act as a hydrogen bonding “glue” towards the serinium cations, as it can be seen in Fig. 4, thus stabilizing the cationic 3D-network. For this task each serinium cation makes use of both the –OH and the –NH3+ groups, one of which is bound by water in one channel, while the second is bound to water molecules in the adjacent channel; in this way each cation bridges two water-filled channels [Ow⋯OOH 2.751(6) and 2.791(5) Å; Ow⋯NNH3+ 2.826(6) and 2.838(5) Å].
![]() | ||
Fig. 3 (a) Filling of light-blue channels from Fig. 2a with water of crystallization (blue spheres); (b) the water molecules along the channels are linked to each other via hydrogen bonds in infinite zig–zag chains, and also interacts with the serinium cations. |
![]() | ||
Fig. 4 Water-filled channels in crystalline [DL-serH]2[ox]·2H2O: the water molecules act as a hydrogen bonding “glue” towards the serinium cations, thus stabilizing the cationic 3D-network. |
The empty channels are then filled with the oxalate dianions (Fig. 5a); the carboxylate oxygen atoms on each dianion interact via charge-assisted hydrogen bonds with six serinium cations, as shown in Fig. 5b [OCOO−⋯OCOOH 2.518(5) and 2.536(5) Å, OCOO−⋯OOH 2.777(5) and 2.779(5) Å, OCOO−⋯NNH3+ in the range 2.798(5)–2.947(5) Å].
![]() | ||
Fig. 5 (a) View down the a-axis of the cationic framework, constituted of serinium cations and water molecules, with oxalate dianions (in orange) filling the channels. (b) Each oxalate dianion interacts via charge-assisted hydrogen bonds with six serinium cations. |
The same description of the packing features in crystalline [DL-serH]2[ox]·2H2O also applies to the two polymorphic modifications of crystalline [L-serH]2[ox]·2H2O. Both form I and form II are characterized by the presence of channels extending along the a- and c-axis direction, respectively, which are filled with water molecules and oxalate dianions; a schematic comparison of the three crystal packings is presented in Fig. 6.
![]() | ||
Fig. 6 Comparison of crystal packings, all viewed down the channels direction, for (top) [(DL)-serH]2[ox]·2H2O, (bottom left) [L-serH]2[ox]·2H2O form I and (bottom right) [L-serH]2[ox]·2H2O form II. |
As observed in [(DL)-serH]2[ox]·2H2O, the water molecules accommodated within the channels form infinite hydrogen bonded chains [Ow⋯Ow 2.866(9), 2.775(8) Å; 2.84(1), 2.70(1) Å in forms I and II, respectively], which are then bound to the serinium cations [Ow⋯OOH 2.721(8), 2.851(9) and 2.84(1), 2.70(1) Å; Ow⋯OCO 2.928(9) and 3.19(1) Å; Ow⋯NNH3+ 2.855(9), 2.828(9) and 2.771(9) Å in forms I and II, respectively], as it is shown in Fig. 7. The most notable difference with respect to [DL-serH]2[ox]·2H2O is that in [L-serH]2[ox]·2H2O forms I and II the cationic frameworks are now chiral.
![]() | ||
Fig. 7 Water-filled channels in [L-serH]2[ox]·2H2O form I (top left) and form II (top right); side-view of the channels, showing the zig–zag water chains in forms I (bottom left) and II (bottom right) (H atoms in form II not observed, as these are powder data). |
The oxalate dianions interact with the cationic framework via charge-assisted hydrogen bonds; as observed in [(DL)-serH]2[ox]·2H2O, each dianion is bound to six serinium cations, but the conformation and the orientation of the cations around the oxalate dianion are markedly different within the polymorphic pair, as it can be observed in Fig. 8 [OCOO−⋯OOH/COOH 2.502–2.701 and 2.50(1)–2.95(1) Å; OCOO−⋯NNH3+ 2.773(8)–2.909(8) and 2.68(1)–2.988(9) Å for forms I and II, respectively]. Fig. 9 shows how forms I and II can be easily differentiated on the basis of their X-ray powder patterns.
![]() | ||
Fig. 8 Hydrogen bond interactions between oxalate dianions and the surrounding serinium cations in [L-serH]2[ox]·2H2O forms I (left) and II (right); H atoms not observed in form II (powder data, see Experimental section). |
![]() | ||
Fig. 9 Comparison of calculated XRPD patterns for [L-serH]2[ox]·2H2O form I (bottom) and form II (top). |
![]() | ||
Fig. 10 (a) View down the b-axis: the structure can be described as a sequence of ABCB′ layers formed alternately by serinium cations and oxalate anions bonded together by hydrogen bonds. (b) View down the a-axis – left: serinium cations belonging to layer A; right: serinium cations belonging to layer B. N⋯O and O⋯O hydrogen bonds between adjacent layers not shown for clarity. |
It is not surprising that [L-serH]2[ox]·2H2O, both as form I and form II, converts into [L-serH][Hox] upon water loss in a heating process: if water is removed from the channels, the cationic framework becomes unstable, and the crystal “solves” the problem by eliminating also one cation as a serine molecule; the extra H-atom on the hydrogen oxalate can thus be employed to form infinite chains of anions placed side by side as to form layers, which in turn are stabilized by close interactions with the remaining serinium cations layers.
The solid–solid reactions with co-crystals or between co-crystals can be used not only to produce new crystal forms with respect to conventional reactions in solution, but also to interconvert crystal forms, in a sort of supramolecular metathesis. The combined experiments suggest a scale of solid state stability, form I > form II > anhydrous DL-serine. The structures of form I of [L-serH]2[ox]·2H2O and that of [DL-serH]2[ox]·2H2O have been determined by single-crystal X-ray diffraction while those of [L-serH]2[ox]·2H2O form II and that of [L-serH][Hox] have been determined from powder diffraction data alone.
Undoubtedly the increasing capacity in determining crystal structures from good quality powder diffraction data collected on laboratory equipment is of paramount importance in tackling structural problems related to the mechanical mixing of reactants, which is, in the case of co-crystals and very recently also in the cases of ionic co-crystals,15 the most efficient way to prepare new materials with active pharmaceutical ingredients.
Footnotes |
† This article is included in the All Aboard 2013 themed issue. |
‡ Electronic supplementary information (ESI) available: Comparisons between experimental and calculated XRPD diffractograms, Rietveld refinement plots, DSC measurements, variable temperature XRPD diffractograms for [L-serH]2[ox]·2H2O forms I and II. CCDC 881344–881347. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c2nj40379j |
§ Co-crystals of an active pharmaceutical ingredient (API) are seen as pure “API-excipient” associations, and “an API that has been processed with a co-crystallizing excipient to generate an API-excipient co-crystal may be treated as a drug product intermediate”; in contrast, “per current regulatory scheme, different salt forms of the same active moiety are considered different active ingredients” http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM281764.pdf |
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2013 |