Crystalline mesophases: Structure, mobility, and pharmaceutical properties

Advanced Drug Delivery Reviews 100 (2016) 194–211

Contents lists available at ScienceDirect

Advanced Drug Delivery Reviews journal homepage: www.elsevier.com/locate/addr

Crystalline mesophases: Structure, mobility, and pharmaceutical properties☆ Evgenyi Shalaev a, Ke Wu a, Sheri Shamblin b, Joseph F. Krzyzaniak b, Marc Descamps c a b c

Allergan plc, 2525 Dupont Drive, Irvine, CA 92612, USA Pfizer, Inc., Groton, CT 06340, USA University of Lille, UMET (Unité Matériaux Et Transformations), 59655 Villeneuve d'Ascq CEDEX, France

a r t i c l e

i n f o

Article history: Received 2 November 2015 Received in revised form 3 April 2016 Accepted 5 April 2016 Available online 8 April 2016 Keywords: Disorder Condensed phase Amorphous Liquid crystals Plastic crystals Phase diagram Glass transition Molecular mobility

a b s t r a c t Crystalline mesophases, which are commonly classified according to their translational, orientational, and conformational order as liquid crystals, plastic crystals, and conformationally disordered crystals, represent a common state of condensed matter. As an intermediate state between crystalline and amorphous materials, crystalline mesophases resemble amorphous materials in relation to their molecular mobility, with the glass transition being their common property, and at the same time possessing a certain degree of translational periodicity (with the exception of nematic phase), with corresponding narrow peaks in X-ray diffraction patterns. For example, plastic crystals, which can be formed both by near-spherical molecules and molecules of lower symmetry, such as planar or chain molecules, can have both extremely sharp X-ray diffraction lines and exhibit glass transition. Fundamentals of structural arrangements in mesophases are compared with several types of disorder in crystalline materials, as well as with short-range ordering in amorphous solids. Main features of the molecular mobility in crystalline mesophases are found to be generally similar to amorphous materials, although some important differences do exist, depending on a particular type of mobility modes involved in relaxation processes. In several case studies reviewed, chemical stability appears to follow the extent of disorder, with the stability of crystalline mesophase found to be intermediate between amorphous (least stable) and crystalline (most stable) materials. Finally, detection of crystalline mesophases during manufacturing of two different types of dosage forms is discussed. © 2016 Elsevier B.V. All rights reserved.

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (Dis)order in condensed phases and structure of crystalline mesophases . . . 2.1. Introduction to disorder . . . . . . . . . . . . . . . . . . . . . 2.2. Crystalline mesophases — orientationally disordered crystals . . . . . 3. Molecular mobility in crystal mesophases . . . . . . . . . . . . . . . . . 4. Thermodynamic aspects of crystalline mesophases . . . . . . . . . . . . . 5. Pharmaceutical properties of crystalline mesophases . . . . . . . . . . . . 5.1. Chemical stability . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Physical stability . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Solubility and dissolution rate . . . . . . . . . . . . . . . . . . . 5.4. Crystalline mesophases encountered in pharmaceutical manufacturing 6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

☆ This review is part of the Advanced Drug Delivery Reviews theme issue on “Amorphous pharmaceutical solids”.

http://dx.doi.org/10.1016/j.addr.2016.04.002 0169-409X/© 2016 Elsevier B.V. All rights reserved.

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

195 195 196 197 201 204 206 206 206 208 208 209 210 210

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

1. Introduction Crystalline mesophases [1] represent a common state of condensed matter, occupying space between amorphous materials with shortrange order and long-range disorder, from one side, and crystalline materials possessing long-range order on the lengthscale of tens of nms, from the other side. In this review, we adopt a broad definition of crystalline mesophases, which includes liquid crystalline structures, conformationally disordered crystals, and plastic crystals. Disordered solids, which in pharmaceutical applications are generally associated with amorphous materials, have attracted major attention of pharmaceutical scientists for two main reasons. First of all, amorphous state can provide significant practical benefits, such as greater apparent solubility for small molecules, or better stability for protein drugs achieved by mixing protein and lyoprotector molecules on molecular level, provided that lyoprotector remains in the amorphous state. On the other hand, amorphous materials, being thermodynamically less stable than corresponding crystalline state, can undergo physical and chemical transformations, thus limiting their shelf life. Rates of chemical reactivity in the amorphous vs. crystalline materials are documented for many systems, with amorphous materials (including their glassy state, i.e., below the Tg) demonstrating significantly higher degradation rates [8–10]. Correspondingly, major efforts have been devoted to investigation of properties and stability of amorphous state. Dynamics of amorphous materials, which is associated with both the glass transition (Tg) and various sub-Tg mobility modes, such as Johari–Goldstein relaxation, has been investigated in many details [11–13]. Furthermore, structural aspects of amorphous pharmaceutical materials have attracted increasing attention recently [14,15]. Structural and dynamic properties of crystalline mesophases, from the other hand, have not been discussed at the same level of details in pharmaceutical literature, possibly due to some practical challenges associated with their detection and recognition. In many cases, it is not straightforward to distinguish crystalline mesophases from either crystalline or amorphous state, especially if experimental characterization is limited to a single method. E.g., X-ray powder diffraction (XRPD), which is a main solidstate characterization tool, would not allow distinguishing between a regular crystal and plastic crystal, because plastic crystals would also have sharp and strong diffraction lines. Furthermore, observation of the glass transition by DSC would not necessary be a unique signature of an amorphous material, because it is an inherent property of crystal mesophases as well. The general relevance of mesophases to pharmaceutical systems has been recognized relatively recently [2,3], although the importance of liquid crystalline materials has long been known for some specific cases, such as topical formulations [4] and some drug delivery systems, e.g., liposomes [5]. In two earlier comprehensive reviews on crystalline mesophases in pharmaceuticals, both published in 2005 [3,16], the importance of crystalline mesophases for pharmaceutical R&D has been convincingly presented. In particular, numerous examples of active pharmaceutical ingredients were provided, including both small molecular weight compounds and macromolecules, and covering a range of therapeutic classes. It was also proposed [2], based on the original idea from [17], that milling-induced amorphization of crystalline materials can go through transient crystalline mesophase structures. If this is indeed the case, crystalline mesophases can be even more ubiquitous considering that milling (grinding) is a common industrial process. Furthermore, potential connections between crystalline mesomorphism and polyamorphicity – with the latter one also getting attention in pharmaceutical science – have been suggested. For example, a famous case of a polyamorphic form of triphenyl phosphite (so called “glacial phase”) could indeed represent a plastic crystal composed of nanocrystals rather than a true polyamorphous state [18]. Similar to amorphous materials, mesophases can be converted to low temperature (and thermodynamically stable) crystal phase, although such transition can be kinetically hindered under typical

195

experimental conditions. Upon deep-enough undercooling (also known as supercooling), mesophases exhibit a “freezing” of molecular motions similar to the liquid-to-glass transition. Correspondingly, they can be in a state with either dynamic disorder (above the calorimetric glass transition temperature, Tg) or static-frozen disorder (below the Tg). For example, a continuous cooling of a metastable disordered crystalline phase of ethanol, which represents a monotropic situation with the ordered stable phase, results in a freezing of disorder at temperatures below the Tg [6]. An important fundamental feature of mesophases, which separates them from amorphous glasses, is that the glass transition can take place in a thermodynamically stable crystalline state as indeed observed experimentally in orientationally disordered SnCl2·2H2O and TlNO2 crystals [7]. In this review, fundamental properties of crystalline mesophases are summarized, covering both structural and molecular mobility aspects. Particular attention is paid to establishing a comparison with the nature of disorder existing in both conventional crystals and amorphous compounds, whereas a specific case of mesophases, i.e., orientationally disordered crystals, is considered in more detail. The structural overview is followed by the discussion of dynamic properties of crystalline mesophases, in which molecular mobility in mesophase is compared with the mobility in an amorphous state of the same molecule. Furthermore, general principles which determine relative thermodynamic stability of crystalline mesophases are presented, with the help of relevant phase diagrams. Finally, examples of instability of crystalline mesophases vs. amorphous and crystalline states of the same material are discussed. Several of the examples considered represent real situations encountered during development of novel drug candidates, covering both stability and formation of crystalline mesophases and change in their properties during pharmaceutical processing. The examples show that the chemical stability of crystalline mesophases is indeed intermediate between amorphous (least thermodynamically stable) and crystalline (most thermodynamically stable) forms. This conclusion is consistent with an early report which compared chemical stability of a crystalline mesophase with the same molecule in the crystalline state [19]. On the physical stability subject, we note that crystalline mesophases can be expected to crystallize much more easily above their Tg — provided that the mesophase is thermodynamically metastable in respect to the crystalline phase at a particular condition (e.g., temperature, pressure, solvent). We conclude that, while there is a growing awareness of pharmaceutical scientists about importance of the mesophases in pharmaceutical R&D and manufacture, potential pharmaceutical advantages of crystalline mesophases, such as their better stability than that of amorphous materials and higher apparent solubility/faster dissolution than crystalline structures, are probably under-explored. 2. (Dis)order in condensed phases and structure of crystalline mesophases Structure and order hierarchy of different states of condensed matter is presented schematically in Fig. 1. Mesophases refer to those phases intermediate to crystalline and amorphous materials in terms of structural (dis)orders [20]. A number of mesophases have been discovered with various degrees of translational, rotational, and conformational disorder, and they are commonly classified as belonging to one of the three categories, that is, liquid crystal, plastic crystal, and condis phase (conformational disordered crystals) ([1,20]). Condis crystals, which possess translational and rotational order, but also have partial or complete conformational disorder [21], are the closest structurally to common crystalline materials. In condis crystals, different conformers are distributed randomly throughout of crystalline lattice, as, e.g., in pseudo racemic crystals (sometimes named pseudoracemate). Condis crystals are typically formed by macromolecules such as synthetic polymers with two or more conformers of similar overall molecular shapes, although conformational disorder can also exist in small molecules. A

196

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

Fig. 1. Order hierarchy of different states of condensed matter.

condis phase can be potentially converted to either plastic crystal or a liquid crystal depending on the specific type of order it loses during the conversion. Liquid crystals are the most widely recognized types of crystalline mesophases. They have orientational order but conformational and three-dimensional translational disorder [22,23]. Liquid crystals are usually formed by either rod- or disc-shaped molecules, and such systems can commonly adapt different mesostructures via either temperature (thermotropic transitions) or solvent content (lyotropic transitions) changes. Depending on the shape and flexibility of a particular molecule, liquid crystals can form nematic, smectic, or hexagonal structures. Nematic liquid crystals are the most similar phases to liquids with only orientational order. On the other hand, smectic or hexagonal phases have both orientational order and partial positional order in one or two dimensions. They can be classified as thermotropic or lyotropic according to their method of preparation. In addition, liquid crystals can often be differentiated based on molecular geometry (calamitic vs. discotic) or type of bond (molecular vs. ionic). Plastic crystals have 3-dimensional translational order but orientational disorder of the molecules around crystallographic position. Therefore, they are also called orientationally disordered crystals (ODIC phases). Plastic crystals are usually formed by near-spherical molecules, although plastic crystals of less symmetrical molecules, such as planar or chain molecules, are also known. They commonly have a simple crystal structure with a high symmetry, e.g., facecentered cubic or hexagonal. Molecular plastic crystals can be easily deformed by a modest mechanical force, whereas ionic plastic crystals could be more rigid. Note also that some systems can belong to two categories at the same time, e.g., the common form of crystalline ice, hexagonal ice, or the plastic crystal form of succinonitrile, can be classified as both conformationally disordered crystal and orientationally disordered crystal. In order to provide a background on the nature of disorder, which is essential in understanding partially disordered mesophases, a brief subon “Introduction to disorder” is included below, with a discussion of

three main types of disorder in crystalline materials as well as an order in amorphous materials. 2.1. Introduction to disorder The translational periodicity determines the long-range order (LRO) of crystals. In an ideal crystal, all the atoms are located in exact positions, and each particular atom would have its neighbors at exactly the same distances and angles. As a result, an X-ray diffraction pattern of an infinite crystal should consist of very narrow diffraction peaks centered at discrete diffraction angles (wave vectors Q at nodes of the reciprocal lattice). The positions of the Bragg peaks are determined by the parameters of the unit cell of the crystal lattice and given by the Bragg law. The intensity of the peaks reflects the distribution of positions of molecules within the unit cell relative to the lattice points. The calculation of the diffracted intensity at a specific Q position, I(Q), requires calculating the structure factor F(Q), which reflects the distribution of the positions of the molecules within the unit cell: IðQ Þ ∝ j FðQ Þj2

with

FðQ Þ ¼

X

f j j

  exp i 2π Q :r j

ð1Þ

The summation is taken on the atoms j of the unit cell only. fj is the atomic scattering factor of the j-th atom situated at rj from an origin. All real crystals have some extent of disorder with corresponding modifications of X-ray diffraction peak intensity or/and broadening. Furthermore, such disorder or randomness is not necessarily incompatible with the crystalline nature. Three main types of crystal disorder are discussed below. Crystal size effect: a borderline situation of “disorder” (so to speak) is linked to the finite size of single crystals which otherwise do not show any perturbation of the translational periodicity. Crystal size reduction leads to a specific type of X-ray peak broadening. This broadening can only be observed for crystallite sizes lower than 0.1–1 μm. The width of a diffraction peak, δQ, in reciprocal lattice unit Q, is inversely proportional to the number of unit cells along a given crystallographic

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

197

direction. The particularity of this broadening is that δQ does not change with the order of the line (i.e. with Q). If the line profiles are plotted on the 2θ scale, the equation relating the crystal size L, to the broadening δ(2θ) of the line observed at Bragg angle θ0, has a specific cosine dependence given by the Scherrer equation:

factors, fj in Eq. (1), have to be replaced by the temperature dependent expression:

δð2θÞ ≅ 0:9λ=L cos θ0

where K is a constant, uj is an average displacement of atom, and M is usually called the Debye–Waller factor. As a result, |F(Q)| and the Bragg scattering intensity are reduced. The reduction is more pronounced at high Q values (~ high Bragg angles) and for high temperatures where u2j is large. This reduction at high Bragg angles adds to the regular decrease of X-ray diffraction intensity linked to the nominal decrease of atomic form factors. Another example of imperfection of the first kind is that of plastic crystals as will be shown later. Imperfections of the second kind are such that the fluctuations of the inter-motifs distances increase with the distance between motifs. In that case there is neither more strict long range crystalline periodicity nor average motif. When the fluctuations only increase very slightly, the molecular positioning is still rather well defined, while not perfect. The solid is then often designated as “badly-crystallized.” Sometimes a “paracrystalline” description [24–26] can be adopted. The X-ray diffraction by such a disordered system can still show sort of diffraction peaks which are, however, broadened. Unlike size broadening, the peak width δQ increases with the order of the line (i.e. with Q). The peak intensity also decreases rather fast as |Q| increases. Fig. 3a gives a one-dimensional illustration of an imperfection of the second kind [24,25]. Two kinds of atoms, with slightly different diameters, are stacked randomly. It is clear that, in this case, the fluctuation of the inter-atomic distance increases with the distance: Dn − dn N D1 − d1 (n N 1). Fig. 3b shows the X-ray pattern corresponding to this system (in Q units). Pseudo diffraction peaks clearly appear. However, their width increases very rapidly with the order of the peak (as the square of the scattering vector Q2). At the same time, the maximum intensity decreases as 1/Q2 so that only the first orders are observable. If the atomic scattering factors of the two types of atoms are different, the peaks are dissymmetric and their maxima are displaced from ideal reciprocal lattice nodes. By comparison the peak width due to crystal size is independent of the peak index. In extreme cases of imperfections of the second kind, long-term periodicity is lost completely, as in amorphous materials. In this frame, the structure is specified by the space variation of the radial pair distribution function (PDF) g(r) which gives the probability to find an atom at the distance r from another one. Experimental determination of g(r) allows to describe the short range order (SRO) which develops between molecules. Fig. 4 illustrates the meaning of the PDF g(r), in the case of a very simple mono-atomic amorphous system. The location of the first peak corresponds to the most probably distance for the first coordination shell, the second peak corresponds to second shell, etc. The peak area can be used to calculate coordination number. The PDF analysis is used later in this section to compare local order of the same molecule in three different states of condensed matter [27].

ð2Þ

where λ is radiation wavelength. In the case of a collection of very small nanocrystals, such a broadening can become very important and leads to a considerable overlapping of the peak wings. It results in an apparent diffuse background and a diffractogram which has similarities with that of the amorphous form. Care must be taken not to confuse these situations. In this case, the presence or not of a glass transition in DSC can help in interpreting the microstructure. An ongoing difficulty in analyzing the structures is that a size effect can affect any type of phase whether it is molecularly ordered or not. Another type of disorder can arise from a local displacement of the structural elements (such as atoms, monomer units, and motif) or a chemical substitution. Disorder by chemical substitution is commonly introduced during solid-state chemical degradation of organic crystals, e.g., crystalline active pharmaceutical ingredient. A disorder induces fluctuations of the distance between homologous atoms. The characteristics of these fluctuations allow cataloguing crystalline imperfections and clarifying the limit of the crystallinity. With this respect, it is necessary to distinguish two kinds of crystalline imperfections [24–26]. Imperfections of the first kind are such that the fluctuations of the inter-atomic (inter-motifs) distances do not increase with the distance between motifs. Such imperfections preserve the long range positional order of the lattice on an average. The motif that repeats translationally is an average one. This average is either taken over time or space. The persistent translational periodicity still imposes the existence of Bragg peaks. Imperfections of the first kind do not induce broadening of the Bragg peaks. The structure factor is now an effective one: b F(Q) N. It corresponds to the average motif and takes into account all molecular positions inside the unit cell with an appropriate statistical weight. This type of disorder only produces a reduction in intensity of the diffraction peaks at high Q values. The diffracted intensity, which is removed from the Bragg peaks, is spread throughout the reciprocal space in the form of a low-intensity diffuse scattering. A typical example of imperfection of the first kind is the thermal agitation of atoms and molecules which exist in every crystal. The regular positions of molecules in a crystal are only average positions around which they are continuously vibrating and librating. The centers of mass of vibrating molecules are perfectly ordered because the vibrations of the different molecules are statistically similar. Fig. 2 outlines a simplified example of a crystalline chain of atoms. It simply demonstrates how the fluctuations induced by vibrations are independent of the distance between atoms. The amplitude of the fluctuation can be as large as 1/10 of the lattice parameter. It is highly temperature dependant. For the calculation of the diffracted intensity I(Q), the atomic form

  f j ðTÞ ¼ f j exp − KQ 2 u j 2 ¼ f j expð−MÞ

ð3Þ

2.2. Crystalline mesophases — orientationally disordered crystals

Fig. 2. One-dimensional monoatomic model of imperfection of the first kind: thermal agitation. The grey zones show the spatial extension of atomic vibrations around the average position (black point). The fluctuation of the distances is independent of the interatomic distance: D1 − d1 = Dn − dn ∀n

While structure of liquid crystals, including pharmaceuticals, is covered in detail in a number of publications [3,16,22,23], plastic crystals represent often overlooked, but nevertheless important and pharmaceutically relevant type of crystalline mesophases. Therefore, this section is focused on structural features of ODIC crystals, with the help of specific examples which include ethanol, caffeine, and succinonitrile. The Orientational Disordered Crystals (ODIC) are usually soft and may even flow under their own weight (for this reason sometimes referred to as plastic crystals [28,29]). Molecular rotation allows accommodating the generally lower molecular symmetry to the high symmetry of the

198

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

Fig. 3. (a). one-dimensional illustration of an imperfection of the second kind. Two types of atoms, with slightly different diameters, are stacked randomly. The fluctuation of the interatomic distance increases with the atomic separation. (b) Corresponding calculated X-ray diagram.

site. A majority of plastic crystals are formed by molecules with a globular shape. It is the case of a large variety of tetrahedrally symmetric molecules such as CH4, CBr4, and adamantane. However, ODIC (or rotator phase) can also be found with much less symmetric molecules if intermolecular interactions are weak enough. They can be rather flat molecules, such as caffeine, which may rotate in their plane, or chain molecules like alkanes [30] which may perform axial rotations. In

Fig. 4. Radial pair distribution function (PDF) g(r) for a simple atomic liquid (atomic diameter σ). Right: the corresponding diffracted intensity bI(Q)N A sample structure is also depicted where the solvation shells are indicated by the dotted lines. The exclusion radius can be seen in the absence of amplitude of g(r) for σ b r b 2σ.

these two latter cases, the ODIC phase is more often hexagonal with a rotational disorder which takes place around the hexagonal c axes. It should be added that the orientational freedom of molecules in ODIC can be restricted. Some short range intermolecular correlations may exist. The correlations which subsist in the disordered phases of these crystals can be of very different kinds according to the intermolecular potential and the molecular and crystal structure. Intermolecular correlation contributes to the decrease in the value of the configurational entropy associated with the orientational disorder. Therefore, the value of the transition enthalpy with the low-temperature brittle phase is decreased while that of melting can be increased. Three examples of ODIC systems are considered below, including ethanol, caffeine, and succinonitrile, to illustrate structural arrangements and corresponding XRPD which are typical for orientational disorder. Fig. 5 shows the X-ray diffractograms of three phases of ethanol [31]. The diffractogram of the plastic phase (middle) is very typical of this type of mesophase. It shows only two Bragg peaks identified as the (110) and (200) of a highly symmetric body-centered-cubic lattice. The peaks are very sharp. The intensity decreases extremely rapidly with the Bragg angle. The situation is fully illustrative of the case of a crystal imperfection of the first kind taken to the extreme. Confrontation with the two other diffractograms is instructive. The multitude of Bragg peaks of the stable phase (bottom) is a feature of the low symmetry monoclinic ordered phase. When comparing with the diffractogram of the amorphous phase, we may notice that the wave vector Q characteristic of the peak maximum in the amorphous phase is closely located with the (110) line which means that intermolecular distances are

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

199

Fig. 6. X-ray powder diagram of the high temperature hexagonal disordered phase of caffeine. We may notice the considerable decrease of peaks intensities at high Bragg angle which is due to the disorder. However, Bragg peaks remain extremely sharp, which shows that the disorder is fully compatible with the existence of a perfect (average) crystalline periodicity. Reproduced from [P. Derollez, N. T. Correia, F. Danède, F. Capet, F. Affouard, J. Lefebvre and M. Descamps, Ab Initio structure determination of the high-temperature phase of anhydrous caffeine by X-ray powder diffraction, Acta Crystallogr. Sect. B, 61 (2005) 329–334], with permission of the International Union of Crystallography.

Fig. 5. X-ray diffraction patterns of ethanol. From top to bottom: The conventional glass at 80 K, the supercooled plastic crystal phase (cubic) at 80 K, the stable monoclinic crystal at 120 K. Reprinted figure with permission from [A. Srinivasan, F. J. Bermejo, A. de Andrés, J. Dawidowski, J. Zúñiga, and Criado, Phys. Rev. B, 53, 8172–8175 (1996)]. Copyright 1996 by the American Physical Society.

similar in both phases. However, the considerable width of the amorphous halo demonstrates the lack of long range order. As for the plastic phase succinonitrile (which is discussed later in this section), X-ray diffraction pattern of plastic phase of ethanol (Fig. 5 middle) shows the presence of a noticeable diffuse scattering. The diffuse scattering intensity corresponds to the intensity removed from Bragg peaks due to the rotational disorder of the molecules. Fig. 6 shows the x-ray powder diffraction by the high temperature disordered crystalline phase of anhydrous caffeine [32,33]. In this case, the lattice is hexagonal. The high crystalline symmetry can only be understood if the caffeine molecules – which have a low symmetry – are statistically disordered around their center of mass. Bragg peaks are more numerous than in the case of ethanol, but their intensity is also decreasing fast with the Bragg angle. It is common that plastic phases give rise to extremely sharp Bragg peaks like that [34], although the limited number of Bragg reflections means that a detailed analysis of the disordered structure is challenging. Nevertheless, the structure refinements of the disordered phases is still possible, as demonstrated for caffeine based on both powder diffraction pattern [33] and single crystal [35] X-ray diffraction. Results are in many respects similar and confirm the existence of a statistical disorder of the molecules which dynamics

was revealed by dielectric relaxation spectroscopy [32,36]. Ring planes of the molecules are parallel and stacked along the c hexagonal axis. Proposed structures only differ by the number of in-plane orientations that each molecule randomly takes on its site around the hexagonal axis: either three [33] or six [35] orientations. The orientational disorder restores a site symmetry compatible with the Bravais lattice. Fig. 7 shows an instantaneous schematic view from above of the molecules in the course of their rotation around the hexagonal axis c. Crystallography only provides a static averaged view of the disordered molecular organizations and cannot give information about the dynamics of the process. Further details regarding the structure can often be provided from additional independent measurements and especially from vibrational and relaxational spectroscopies. Comparison of Raman Spectra [37] of the different phases is often useful especially

Fig. 7. The high temperature hexagonal crystalline phase of caffeine. Schematic view of a possible instantaneous configuration. Molecule are rotating around the C axis. The structure is perfectly crystalline but only on an average. Reproduced from [P. Derollez, N. T. Correia, F. Danède, F. Capet, F. Affouard, J. Lefebvre and M. Descamps, Ab Initio structure determination of the high-temperature phase of anhydrous caffeine by X-ray powder diffraction, Acta Crystallogr. Sect. B, 61 (2005) 329–334], with permission of the International Union of Crystallography.

200

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

in the low frequency domain (up to ≈200–300 cm−1) which is sensitive to intermolecular interactions. Since the disorder can be dynamic, dielectric investigations are very helpful, when molecules are not of high symmetry and bring a permanent dipole moment. Monte Carlo and molecular dynamic simulations are often very useful in revealing some specificities of the molecular disorder. In the case of caffeine, recent Monte Carlo (MC) simulations [38] and new dielectric data [36] suggest the existence of possible out-of-plane motions between molecular plane positions which are tilted by about 10° on both sides of the crystallographic (ab) plane. Analysis of the rather pronounced X-ray diffuse scattering intensity can reveal remarkable information about the interactions between molecules and their short range orientational ordering as was shown for cyanoadamantane [34]. Succinonitrile represents an interesting example of crystalline mesophase which combines isomerization (which can be considered as an example of conformational disorder) and molecular rotation (orientational disorder). Moreover, the orientational disorder is incomplete, as its ODIC phase shows orientational correlations induced by steric hindrance as can be observed via the X-ray (or neutron) 3-dimensional diffuse clouds [39]. Succinonitrile (N`C\\CH2\\CH2\\C`N) is an aliphatic dinitrile which is employed as antidepressive agent in some European countries [40,41]. Succinonitrile has recently attracted considerable attention in the technological area of energy storage [42]. At low temperature, succinonitrile is in an ordered monoclinic crystal phase [43]. As in the examples presented above, there are very few, but sharp, X-ray Bragg peaks for the ODIC phase (Fig. 8). The high degree of rotational disorder is also visible as a strongly increased bump in the background compared to the low temperature phase. The molecule exists in an equilibrium mixture of Trans and Gauche (G1 and G2) isomers (Fig. 9a). In the gaseous state, the Trans form is slightly lower in energy than the Gauche ones. As succinonitrile becomes more and more dense the proportions are reversed. In the ODIC phase, the respective proportions of Gauche (NG) and Trans (NT) isomers have been calculated from measuring the intensity ratio of two Raman components [44]. (NT/NG) decreases with temperature from 0.29 at 297 K to 0.22 at 220 K (in the slightly undercooled ODIC phase). The ratio accelerates its decay at the approach of the transition to the low temperature monoclinic phase where it abruptly cancels. Only the gauche isomers subsist in this latter phase. In the ODIC phase, succinonitrile is body centered cubic (m3m). The structure has been examined and re-examined several times by X-ray analysis during the last 50 years [45–47]. The analysis of the polarization

effect on the Raman spectra of single crystals of the ODIC phase helped to identify the twelve equilibrium positions that each of the three isomers may occupy in its lattice site [45]. The middle bond C\\C of the molecule may direct itself along the four 3-fold diagonal axis of the cubic cell. Around the 3-fold axes bearing the middle bond each CH2\\C`N group may occupy three equilibrium positions shifted from one another by a 2π/3 rotation around this axis. We can easily notice that whatever the equilibrium position of the molecule on its site, nitrogen atoms of the CH2\\C`N groups are localized in the middle of one of the sides of the cubic cell (Fig. 9b). The global disorder of a molecule around its crystallographic site can be understood as resulting from a mechanism which combines: 1) an isomerisation reaction G–NT around the 3-fold axis along which is aligned the C\\C bond; 2) a 4-fold rotation of the T isomer by which the C\\C bond changes its orientation from the initial 3-fold axis to another one; 3) an isomerization T–NG1 (or G2) around the new 3-fold axis. Each isomer may thus take twelve equilibrium positions which determine a mean molecularconfiguration compatible with the symmetry of the crystal. Incoherent quasi-elastic neutron scattering (IQNS]) experiments [47] have proven the validity of this model [48] allowing for both motions: isomerization and molecular rotation. On its site, the molecule can thus take 36 different conformations/orientations which lead to the high value of the transition entropy. Succinonitrile is a highly polar molecule in the Gauche configuration. The coupled mechanism isomerization/rotation causes an effective reorientation of the molecular dipole and explains the high value of the static dielectric constant (εS N 55) of ODIC succinonitrile. For two-second neighbor molecules, there is complete steric hindrance between configurations of neighboring molecules that would bring for both of them, a nitrogen atom into the middle of the same cubic axes. This correlation of steric origin induces a short-range ferroelectric dipolar order which tends to increase the value of the static dielectric constant [49]. It has also been shown that these correlations are responsible of the specific 3-dimensional modulation of the X-ray diffuse scattering shown in Fig. 10. Succinonitrile is an example of rotational disorder occurring in the crystal state that can be achieved by deformable molecules. It also illustrates the need to cross the experimental methods in order to gain information on the description of the disorder. To compare local structure in crystalline mesophase with that in liquid and crystalline state, the PDF analysis is particularly useful. As an example, carbon–carbon (C\\C) and bromine–bromine (Br\\Br) PDFs for three different phases of CBr4 (liquid, plastic, and ordered crystalline) are shown in Fig. 11 [27]. The carbon–carbon PDFs for both plastic and

Fig. 8. X-ray diffractogram of Succinonitrile in the low temperature monoclinic phase composed of only Gauche molecules (left). ODIC phase (right) note the few number of peaks and important diffuse background. Reproduced from [S. Hore, R. Dinnebier, W. Wen, J. Hanson, and J. Maier. Structure of plastic crystalline succinonitrile: high-resolution in situ powder diffraction, Z. Anorg. Allg. Chem., 2009, 635, 88–93]. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

201

Fig. 10. A picture of the diffuse X-ray scattering of the ODIC phase of succinonitrile showing the 3-dimentional modulation induced by intermolecular correlations.

3. Molecular mobility in crystal mesophases In this section, different types of molecular mobility observed in crystalline mesophases are considered, and compared with dynamics properties of amorphous structures. Molecular mobility in amorphous pharmaceuticals is a subject of numerous publications. In earlier investigations, the major attention was on the glass transition behavior, including studies on the depression of the glass transition temperature by water (water plasticization) [50] and the structural (global) relaxation as relevant to the Tg [51,52]. More recently, the focus has been shifted to lower-temperature (and often non-cooperative) mobility

Fig. 9. (a). The Gauche and Trans configurations of Succinonitrile. (b) The ODIC BCC structure of succinonitrile. The positions occupied by nitrogen atoms in the course of the different reorientations are shown. Are also indicated example of 1) the isomerism reaction involving Gauche (G1 and G2) and Trans isomers by rotation of a CH2\ \C`N group along one 3-fold axes and of 2) the 4-fold reorientation of the Trans isomer. (c). Example of steric exclusion of two second neighbors Succinonitrile Trans molecules.

ordered crystalline phases are much more structured than that for the liquid phase, as evidenced by the stronger first peak, the deeper minimum after the first peak, and multiple peaks and minimums expending at least up to 25 Å. On the other hand, Br\\Br PDFs deliver quite different message, with both liquid and plastic phases having very similar (and disordered) PDFs with only two definite minima, the last one at approx. 8 Å. The CBr4 case provides a representative illustration of structural relationships between three condensed phases, showing a similar extent of the translational order (represented by C-C RDFs) in ordered crystalline and plastic phases, and a striking similarity in the rotational disorder (from Br\\Br RDFs) between the plastic and liquid phases.

Fig. 11. Intermolecular partial radial distribution histograms of liquid (solid lines), plastic (grey tone lines), and ordered crystalline phase (dashed lines) of CBr4. Upper panel: carbon–carbon (CC), lower panel: bromine–bromine (BrBr). Reprinted figure with permission from [L. Temleitner and L. Pusztai, Local order and orientational correlations in liquid and crystalline phases of carbon tetrabromide from neutron powder diffraction measurements, Phys. Rev. B, 81 (2010) 134101–134108]. Copyright 2010 by the American Physical Society.

202

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

modes such as Johari–Goldstein relaxation, and their relationships with the physical stability of amorphous pharmaceuticals [53,54]. Similar to amorphous solids, crystalline mesophases are also expected to have both main and secondary relaxation events, as summarized in [55]. Moreover, it was stressed that the existence of a glass transition is a key identifier for the mesomorphic state [56]. Indeed, glass transition has been reported in different types of crystalline mesophases formed by both small molecules and polymers. For example, in hexagonal ice, which represents orientationally disordered crystal mesophase (can also be classified as CONDIs crystal), a stepwise increase in the heat capacity during heating was observed at approximately 100 K [7]. The Cp change corresponds to the glass transition and is associated with the “unfreezing” of proton disorder. Similarly, relaxation of frozen-in orientation polarization of water molecules in ice was observed at 110 K by thermally stimulated depolarization current (TSDC) method [57]. As an example of a pharmaceutical material, the glass transition in a liquid crystal form of cyclosporine was observed using DSC and dielectric spectroscopy, with both methods detecting a Tg event at approx. 125 °C [58]. Tg was also observed in a condis mesophase of a small molecular weight drug candidate, GNE068, at 135–145 °C using modulated differential scanning calorimetry (MDSC) [59]. Similar to isotropic (i.e., amorphous) glasses, Tg in mesophasic glasses depends on both chemical nature of the molecule and the presence and concentration of water. The shape of the molecule itself (e.g., rod like, banana type, spherical) has a strong impact on the interactions and consequently on the dynamics of molecules. In addition, if a particular molecule can adopt different mesostructures (e.g., inverted hexagonal vs. lamellar), the Tg can also be different depending on a specific structural organization. Impact of these three factors, e.g., molecular structure, type of crystal mesophase, and water content, on the Tg in crystalline mesophases was studied using synthetic phospholipids [60]. Phospholipids represent a relatively common class of pharmaceutical excipients, which are used in controlled release drug delivery systems. The summary of the results is presented in Fig. 12, which describes the Tg in several species of two main classes of phospholipids, phosphatidylethanolamine (PE) and phosphatidylcholine (PC) species, as a function of water content. The figure shows that water has a significant plasticizing impact, although the extent of the plasticization depends on both chemical nature of the lipid and the mesophase structure. In particular, the Tg of the non-lamellar phases (e.g., inverted hexagonal and ribbon phases of one of the PE species—1,2-dioleoylphosphatidylethanolamine, DOPE) appears to be

much less sensitive to water than the gel phase formed by the same molecule. Furthermore, the difference in the headgroup structure between PC (quaternary ammonium cation, Fig. 12) and PE (primary amine) was shown to have a significant impact on the Tg, with PC species having higher Tg values than the PE species with the same hydrocarbon tail (DOPE vs. 1,2-dioleoyl-sn-glycero-3-phosphocholine, DOPC). The results indicated also that, while the Tg appears to be relatively insensitive to the length of the hydrocarbon chain (essentially identical Tg in 1,2-dimyristoyl-sn-glycerol-3-phosphocholine (DMPC) vs. DPPC, the chain length of 14 vs. 16 carbons), introduction of the double bond into the tail could result in lowering the glass transition temperature, as the Tg for DOPC (having two unsaturated bonds) is lower than that of DPPC (saturated chain). Finally, a more ordered gel lamellar phase (with 2-dimensional order) had a higher Tg than the same molecule in a non-lamellar (inverted hexagonal for DOPE and either cubic or hexagonal for DOPC) – and less ordered – phase. A mechanistic understanding of molecular mobility requires determination of relaxation times and their temperature dependence. A comprehensive evaluation of dynamics of a crystalline mesophase, i.e., plastic crystal of caffeine, has been performed using dielectric spectroscopy [32,36] both in the equilibrium (TN 130 °C) and undercooled ODIC states. Fig. 7 shows the hexagonal average array of caffeine molecules in the crystalline lattice. The high crystalline symmetry can only be understood if the caffeine molecules – which have a low symmetry – can rotate around their center of mass. Dielectric results that prove the dynamic aspect of the rotational disorder [32,36], are presented in Fig. 13. Fig. 13 shows that both the amplitude and the corresponding frequency of the dielectric relaxation peak observed in the ODIC phase of caffeine decrease systematically with temperature. This is also supported by the decrease of εs as shown in the inset of the figure. This indicates that an antiferroelectric organization of molecular dipoles sets in when temperature decreases (decrease of the Kirkwood factor). From the dielectric spectra obtained at different temperatures, relaxation times (τ) can be obtained by fitting the experimental data with the Havriliak–Negami (H.N.) expression [61]:  α γ ε  ðωÞ ¼ ε∞ þ Δε= 1 þ ðiωτHN Þ

ð4Þ

to which a term iσ/ε0ω accounting for the conductivity is added. The parameters 0 ≤ α ≤ 1 and 0 ≤ γ ≤ 1 are respectively related to the symmetric (α) and asymmetric (γ) broadening. Fig. 14 represents temperature dependence of relaxation times in two phases (phase II and plastic crystal I) of caffeine. Two relaxation

Fig. 12. Left: Glass transition temperature of several PE and PC species as a function of water content. DOPC dried was in a non-lamellar phase, and other lipids in the gel phase, unless specified in the Figure. Reprinted with permission from [E. Y. Shalaev, G. Zografi, and P. L. Steponkus, Occurrence of glass transitions in long-chain phosphatidylcholine mesophases, J. Phys. Chem. B, 114 (2010) 3526–3533]. Copyright 2010 American Chemical Society. Right: molecular structures of DOPE (top) and DOPC (bottom).

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

Fig. 13. Dielectric loss spectra of caffeine recorded between 423 K and 283 K (step of 10 K). The inset shows the corresponding temperature evolution of static permittivity increment. Reprinted with permission from [M. Descamps, N. T. Correia, P. Derollez, F. Danede, and F. D. R. Capet, Plastic and glassy crystal states of caffeine, J. Phys. Chem. B, 109 (2005) 16092–16098]. Copyright 2005 American Chemical Society.

203

strongly non-Arrhenius [63]. It should be noted, however, that, while Arrhenius-like temperature dependence of relaxation times in crystalline mesophases appears to be much more typical, there are also cases of non-Arrhenius behavior as observed for a binary succinonitrile– glutaronitrile system and conformationally disordered Freon 112 [64, 65], as well as BB-3(1-Me) polyester [66]. Temperature dependence of the relaxation times of the BB-3(1-Me) polyester in two forms, LC and isotropic, is shown in Fig. 15. In both phases, the temperature dependence is not Arrhenius; however, a major difference in the relaxation times between the isotropic and LC glasses was observed. For the isotropic glass, the temperature dependence had two parts, which were interpreted as the evidence of two relaxation modes due to uncoupling between orientational and translational motions. An important property of any glass is its relaxation behavior when annealed below the Tg. Relaxation of glasses has a practical importance, as annealing was shown to improve stability of amorphous pharmaceuticals [67,68]. A common way to study relaxation in glasses is by monitoring enthalpy relaxation (also known as enthalpy recovery and thermal overshot) peak on DSC curves after annealing a sample below its Tg. The enthalpy relaxation is usually described using stretchedexponential function [69]

processes were observed in phase I (plastic), with the faster/lowtemperature AI process corresponding to in-plane reorientation motions of molecular dipoles, whereas the slow BI process probably relates to the out-of-plane motions, with the molecular rotation plane tilted in respect to the crystallographic (ab) plane. In the RT phase (phase II), a single relaxation process was observed due to the inplane reorientational molecular motion. The extrapolation of the τ values to 100 s provides a conventional definition of the glass transition temperature, at which a specific type of molecular mobility is “frozen” on a typical experimental time scale. For both crystalline phases, the Tg corresponding to the in-plane rotation (process A) is approximately similar at −10 °C for phase I and −13 °C for phase II, whereas the value of Tg expected for the out-of plane rotation (phase I only) would be much higher (at ≅136 °C). Fig. 14 also shows near-Arrhenius behavior of both types of relaxation processes. Such Arrhenius behavior of the relaxation time is common for plastic crystals. In other types of crystalline mesophases, including nematic and smectic liquid crystals, Arrhenius dependence of the relaxation time for rotational mobility along the short axis has also been reported [62]. Likewise, beta-relaxation in conventional glasses, which is linked to rotational mobility, is characterized by Arrhenius behavior whereas alpha-relaxation in the same system can be

where ΔHt and ΔH∞ are the measured and maximal enthalpy recoveries, respectively, t is the annealing time, τ is the mean relaxation time, and β is a parameter which reflects distribution of the relaxation times. Relaxation behavior of mesophase glasses during annealing is similar to that of amorphous glasses and is characterized by a broad distribution of relaxation times, with β values b0.5 in many cases (β = 1 would correspond to a single relaxation time). For example, β was determined to be from 0.3 to 0.5 for two phospholipid mesophases (the inverted hexagonal, HII, phase for DOPE [70] and Lβ (gel) phase for POPE [71]), 0.36 for α-relaxation process of liquid crystalline polysiloxane [72] and 0.5 for BB-3(1-Me) polyester [66]. Annealing can also result in some structural changes, as was shown for glassy plastic crystal of cyanoadamantane, in which annealing gave rise to an increase (although very slow) in orientational order [73] and a thermal overshot upon heating in DSC. This overshoot, which takes place above Tg, is probably related to the reversion of this orientational order.

Fig. 14. Relaxation times of two phases of caffeine, corresponding to the peak of the imaginary parts for three processes AI, AII, and BI. Solid lines show Arrhenius fits of the experimental data. From [M. Descamps, A.A. Decroix, Polymorphism and disorder in caffeine: dielectric investigation of molecular mobilities, Journal of Molecular Structure 1078 (2014) 165–173].

Fig. 15. Temperature dependence of the enthalpy relaxation times in isotropic liquid (circles) and the smectic liquid crystal (squares) of BB-3(1-Me) polyester. The solid lines represent VFT behavior. Reprinted with permission from M. Tokita, S. i. Funaoka, and J. Watanabe, Study on smectic liquid crystal glass and isotropic liquid glass formed by thermotropic main-chain liquid crystal polyester, Macromol., 37 (2004) 9916–9921. Copyright 2004 American Chemical Society.

 h  ΔHt ¼ ΔH∞ 1− exp −ðt=τÞβ

ð5Þ

204

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

It is of interest to compare the Tg in an amorphous glass with the Tg of the same system in a crystalline mesophase glass. Intuitively, the free volume in the more disordered (amorphous) glass could be expected to be higher, and the Tg lower, than that in more ordered, liquid crystalline, glass of the same material. Indeed, for polyacryloxybenzoic acid, for example, the Tg in an amorphous (more disorder) state was significantly (by approx. 50 °C) lower than the Tg of liquid crystalline (more order) glass [1]. However, the situation appears to be more complex, and the Tg of an amorphous state can be similar or even higher than for a mesophase system of an identical chemical composition. For example, the Tg values were similar for ethanol in both an amorphous and plastic crystalline states of ethanol [74]. A detailed comparison of the glass transition in series of binary co-polyesters was reported in [75]. In that system, the Tg of isotropic glass was higher than of the corresponding LC glass by approximately 30–40 °C. The lower Tg of the LC glasses was explained as because of different types of the molecular motions responsible for the Tg in isotropic glass (segmental motion leading to changes in the conformations of the chains) vs. LC (rotation of the elongated chains around their axis, translational motions of the chains, and motion of the small atomic groups). It was suggested further that the latter kinds of motion require smaller free volume than that of the former, which explains lower Tg for the LC glass. Similar results were obtained for BB-3(1-Me) polyester, with the Tg of smectic glass significantly (12 °C) lower than in isotropic liquid glass [66]. Even more complex pattern of the molecular mobility in crystalline mesophase vs. amorphous state was observed for calcium ketoprofen, for which mobility was evaluated by measuring temperature dependence of T1 relaxation times by solid state NMR [76]. It was found that the amorphous form generally had higher mobility than its liquid crystalline counterpart at 0, 20, 60, and 80 °C, as indicated by the shorter T1 relaxation time of the former. However, the T1 relaxation time of the mesomorphous phase seemed to reach a minimum at 40 °C and was lower than that of the amorphous phase, indicative of a greater mobility in the crystalline mesophase at 40 °C. In some systems, two glass transition events were reported, possibly due to the freezing of different types of molecular motions corresponding to different timescales. Two Tg events have been suspected in a plastic phase (phase I) of caffeine as shown in Fig. 14 [36]. They correspond to in-plane (Tg ~ −10 °C) and out-of plane (Tg ~ 136 °C) rotations, as discussed above. In another example, two Tgs have been observed for a cubic mesophase of a phospholipid (DOPC) using DSC [60]. Furthermore, a smectic glass of one of the compounds in homologous series of butyl(methoxybenzylidene)aniline) (HEXOBUTA) was reported to have two Tgs, and it was proposed that this could be an intrinsic nature of smectic structures [77]. It should be noted, however, that phospholipids in smectic (gel) phase showed a single Tg event [60], indicating that the double Tg observed in HEXOBUTA is not necessary a common property of smectic phases, but rather a specific property of a particular system. An insight into molecular nature of the molecular mobility in various disordered states can be also obtained from measuring either the strength of the dielectric relaxation event or the heat capacity change. For secondary relaxation of ethanol, for example, the strength of both main and secondary relaxation peaks was very similar between the molecular glass (orientational and translational disorder) and the glassy crystals (orientational disorder) [78]. The authors concluded that both the reorientational motions of the ethanol molecules and the local environment are similar amorphous supercooled liquid state and in plastic crystals. However, temperature dependence of the relaxation times in plastic crystal and amorphous glass was different, with more fragile amorphous state (Angell's fragility parameter m = 60) than plastic crystal (m = 40). In addition, calorimetric study of the same system (i.e., amorphous glass and plastic crystal of ethanol) revealed that the heat-capacity change at the Tg was larger for an amorphous glass of ethanol than for its plastic crystal [7], probably due to the additional contribution from unfreezing of translational mobility in the amorphous glass above the Tg.

An interesting (and pharmaceutically relevant) property of crystal mesophases is that they could change physical properties of molecules which are sequestered (confined) within its structure. One such example was reported in [79], where the impact of a confinement of sucrose by an inverted hexagonal phase, HII, of a phospholipid was observed. In that study, two Tgs of amorphous sucrose were observed under certain conditions, with the Tg2 (higher-temperature) event corresponding to the Tg of “bulk” amorphous sucrose, whereas Tg1 (lower-temperature) was assigned to the Tg of sucrose confined inside the cylinders of the HII phase with the diameter of approx. 2 nm. It was proposed that the confinement resulted in the depression of the Tg of amorphous sucrose, which is consistent with the well-known impact of confinement on the Tg of different amorphous materials. To conclude discussion of the molecular mobility, glass transition is an intrinsic “signature” of both crystalline mesophases and amorphous glasses. Properties of the glass transition in mesophases are similar to that in amorphous glasses, in both of which the main (global) relaxation has a strong non-exponential nature with a broad distribution of the relaxation times. Tg depends on the chemical nature of the compound, as well as concentration of a plasticizer (water) and a particular structural arrangement of the crystalline mesophase. Molecular nature of the glass transition in crystalline mesophases might reflect sequential freezing of separate modes having different timescales and therefore different glass transition temperatures, as was demonstrated for caffeine. When comparing the Tg and ΔCp of the same molecule in different mesomorphic states and amorphous form, more ordered states were shown a higher Tg and lower ΔCp, although multiple exceptions do exist. 4. Thermodynamic aspects of crystalline mesophases Thermodynamic relationships between crystalline and an amorphous states are relatively straightforward: a crystalline form is the thermodynamically stable form below its melting point (for singlecomponent solvent-free systems) or its liquidus (solubility) temperature for binary and multi-component systems with a solvent. Furthermore, thermodynamic stability of different crystalline forms has been discussed in the literature in many details. E.g., solvent-free polymorphs are generally divided into monotropic and enantiotropic systems with thermodynamics relationships commonly illustrated by the Gibbs free energy diagram, while the thermodynamics stability of different solvates can be tracked (at least in principle) using compositiontemperature phase diagrams or various versions of pressure–composition–temperature phase diagrams. Mesophase-forming systems, on the other hand, offer more diverse and complicated scenarios in many cases. Let's consider thermodynamic stability in a one-component mesophase-forming polymeric system with the help of the diagram shown in Fig. 16 [1]. The schematic diagram illustrates that thermodynamic stability increases from the right side of the diagram to its left side in parallel with the increase in the extent of ordering. At a fixed low temperature, for example, both the extent of order and the thermodynamic stability increases from the glass (field 11) to two-phase system consisting of mesophase glass + isotropic glass to a single-phase mesophase glass and finally (via three more intermediate states of increasing order) to the highly ordered crystalline state. The diagram can also describe temperature-induced changes; e.g., heating of a two-phase system of crystal + mesophase glass (field 10) above the Tg would produce crystal + mesophase, followed by formation of a single-phase mesophase system and finally isotropic melt. Note that this diagram is applicable to all three main classes of mesophases, i.e., conformationally disordered crystals, plastic crystals (ODIC), and liquid crystals. In systems with ODIC phase, the most common situation is that of a first-order phase transition from a low-temperature crystalline (brittle) phase (field 3 in Fig. 16) to a rotator mesophase (field 2) occurring during heating before melting (i.e., before formation of isotropic melt), and such phase transition is associated with “unfreezing” of a dynamic

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

Fig. 16. Schematic diagram of equilibrium and non-equilibrium phases in a onecomponent system with a single crystalline mesophase. Only the phases on the lefthand side (i.e., fields 1, 2, and 3) are in equilibrium, while all other structures are metastable. The diagram shows both single-structure regions and co-existence between two different structures. For example, field 10 corresponds to a (metastable) co-existence between two structures, crystal and mesophase glass. Upon heating, this system would pass the glass transition temperature of the mesophase and thus entering filed 9, co-existence of a crystalline phase and a liquid-like mesophase. Further heating results in melting of the crystalline phase and corresponding conversion of the two-phase system into a single-phase mesophase (field 2), and finally in the mesophase conversion to an isotropic melt (field 1). Reprinted from [B. Wunderlich, A classification of molecules, phases, and transitions as recognized by thermal analysis, Thermochim. Acta, 340–341 (1999), 37– 52], copyright 1999.

orientational disorder. The transition can also be assisted by conformational flexibility and a solid state isomerization reaction, as in succinonitrile. The transformation to the plastic phase can be perceived as an orientational melting of an average crystalline lattice. The highertemperature melting event involves the unlocking of just the translational degrees of freedom, and therefore is characterized usually by a low melting enthalpy (ice is an exception) and relatively high melting temperature. Succinonitrile is an example of such brittle-to-plastic crystal transition. Upon heating (see Fig. 17), succinonitrile undergoes a strongly first-order transition to an ODIC phase at 233.3 K. It shows a large transition entropy (ΔSt = 6.35 cal/deg·mole), an abrupt 6.7% volume expansion, and a significant increase of specific heat (Cp).

Fig. 17. Evolution of the Cp for succinonitrile showing the transition from brittle to ODIC then from ODIC to liquid. Is also reported a schematic evolution of the Gibbs function which illustrates the variations of entropy at each first-order transition. Note the Cp jumps after each transition. Reprinted (with modifications) with permission from [C. A. Wulff and E. F. Westrum Jr., Heat capacities and thermodynamic properties of globular molecules. VI. Succinonitrile, J. Phys. Chem. 67 (1963) 2376–2381]. Copyright 1963 American Chemical Society.

205

Correspondingly, the melting entropy at 331.3 K (ΔStm = 2.68 cal/ deg·mole) is relatively modest as it is for the increase of Cp [80]. It shows that the contribution of the unfreezing of rotational degrees of freedom to ΔCp is greatest than that corresponding to the translational diffusion of molecules which generates fluidity. The expansion coefficients in the ODIC and liquid phase are very alike [45]. The brittle crystal-to-ODIC phase transition sequence is obviously reversed during cooling, and the formation of highly ordered lowtemperature crystalline phase from the high-temperature melt can pass through an intermediate rotator (ODIC) phase. Furthermore, rotator phases may play an important role in the crystallization of polymers. For example, experimental work suggests that polyethylene crystallizes from the melt via nucleation of a rotator phase [81–83]. It should be stressed that the diagram (Fig. 16) does not cover all possible scenarios even for single-component systems. In a case of ethanol, as an example of a monotropic system, the brittle phase converts directly to isotropic melt upon heating, and therefore there is no reversible transition between the brittle and plastic phases. In this case, field 2 (mesophase) is not accessible from field 1 for ethanol, as well as for other monotropic systems. Nevertheless, this diagram represents a useful (although probably over-simplified) illustration of relative thermodynamic stability of a system with a mesophase as a function of temperature, and Fig. 16 will be used in the next section, where stability of the same chemical entity in different phases is compared. For systems with more than one mesophases, a usual sequence of thermotropic phase transitions during heating is crystalline–smectic– nematic–isotropic. However, “reentrant” phase transition, from smectic to nematic during heating, was also observed in several systems, such as crystal–nematic–smectic–nematic –isotropic phase transition sequence, which was reported for a small organic molecule, derivative of tolane [84]. For lyotropic systems, the common sequence of the phase transitions upon removal of solvent (e.g., water) is to transfer from smectic (lamellar liquid crystalline) phase to a non-lamellar phase, e.g., the inverted hexagonal phase HII [85]. In some systems, such a phospholipid (DOPE)-water binary system [86], reentrant lyotropic phase transition has been also observed. According to the temperature-composition phase diagram (Fig. 18), dehydration can result in the re-entrant phase transition, from the non-lamellar to lamellar liquid crystalline (Lα) and then back to the non-lamellar phase [87]. The phase diagram

Fig. 18. Phase diagram of water-DOPE system. The arrow shows a pathway for lyotropic phase transitions at 20 °C, with the sequence of lyotropic phase transitions presented in the right panel of the figure. Reprinted (with modifications) from [Shalaev, EY, PL Steponkus (1999) Phase diagram of 1,2-dioleoylphosphatidylethanolamine (DOPE):water system at subzero temperatures and at low water contents, Biochim. Biophys. Acta 1419, 229-247]. Copyright 1999, with permission from Elsevier.

206

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

also shows that phase relationships can be rather complicated, with the relevant thermodynamic stability of phases changing in a narrow composition (e.g., solvent content) and temperature range. For example, let’s consider a sequence of lyotropic phase transitions at a fixed temperature of 20 °C. Horizontal arrow (Fig. 18) marks a movement of the figurative point of the system upon dehydration at 20 °C, and the corresponding sequence of the lyotropic phase transitions is listed in the same figure. As water content is reduced from 30 wt% to 0.5 wt%, the system goes through nine different phase states, including the reentrant transition, Lα→Lα+HII→Lα. It should be noted that, while phase diagrams provide a comprehensive information about relative thermodynamic stability of different phases as a function of the chemical composition of a system and state variables such as pressure and temperature, the rates of corresponding phase transitions can vary significantly between systems and also can be very sensitive to temperature. Thermodynamically unstable phases can persists for long periods of time, especially below their corresponding Tg, as illustrated below in Section 5. It should also be noted that, while there are specific rules for phase diagrams of systems with macroscopic phases, crystalline mesophases might have additional degree of freedom due to a significant contribution from interfaces [88], and therefore the usual rules used in construction of phase diagrams (such as Gibbs phase rule) should probably be adjusted, e.g., to allow for an extra degree of freedom. 5. Pharmaceutical properties of crystalline mesophases Different solid phases have various degrees of molecular mobility, packing density (free volume), and local order, and these factors would impact the rate of various physical and chemical processes and therefore stability and the shelf life of pharmaceuticals. From the general perspectives, physical and chemical stability of different physical forms of the same material can be expected to follow the same rankorder as thermodynamic stability, which represented schematically in Fig. 16. Several examples are considered in this section to illustrate the relative chemical stability of crystalline mesophases in respect to crystalline and amorphous forms, followed by physical stability cases and finally examples of mesophase-related phase transformations during pharmaceutical processing. 5.1. Chemical stability An example of an enhanced chemical reactivity in a crystalline mesophase (as compared with a crystalline form of the same molecule) was reported in [89,90]. In that study, milling and freeze-drying treatments were used to produce disordered samples of a model peptide, tetraglycine methyl ester (TGME). The disordered samples had relatively sharp XRPD peaks but their intensity was significantly reduced as compared with the initial crystalline sample. Such decrease in the strength of the diffraction peaks, especially at higher angles, is typical for crystalline mesophases as discussed in Section 2. In addition, DSC tests of both milled and freeze-dried materials revealed the Tg thermal event at approximately 30 °C. Tellingly, the samples preserved their appearance as solid powder particles when heated up to 165 °C (i.e., well above the Tg of 30 °C). Therefore, it was concluded that the freeze-dried and milled samples of TGME represented crystalline mesophase, possibly conformationally disordered crystal. The chemical reactivity of both the initial crystalline TGME and crystalline mesophases was evaluated at various temperatures between 50 and 165 °C. The crystalline material had typical sigmoidal kinetics curves with characteristic induction period and acceleratory stages. The induction period is associated with a compact and rigid crystalline lattice, which creates a spatial restriction (and a physical barrier) for the conversion of the reactant (TGME) molecules to the reaction products, molecules of which could not be accommodated in the reactant crystal lattice without its disruption. A transition from the induction period to the acceleratory phase is due

to the accumulation of the sufficient amount of defects in the crystalline lattice. The newly created product molecules have different shape/size than the reactant thus creating increased stress in the lattice (chemical pressure [91]). In contrast to the sigmoidal kinetics curve of the crystalline TGME, the crystalline mesophases had deceleratory kinetics curves and significantly higher rate of reactivity (Fig. 19). Both the elimination of the induction period and the increased reactivity rate reflect more disordered nature of the crystalline mesophase. An important observation was that the overall reaction mechanism remained the same in both the crystalline form and the mesophases, indicative of a similar local order in both the crystalline state and the crystalline mesophase [89]. The reaction involves intermolecular transfer of the methyl from the ester of one molecule to the amino group of a neighbor molecule and therefore requires certain intermolecular distance and orientation. Only at the higher temperature of 165 °C the reaction mechanism in the crystalline mesophase changed from methyl transfer to polycondensation [90]; this change in the reaction mechanism was attributed to a greater mobility and molecular flexibility in the crystalline mesophase allowing formation of a four-centered activated complex. The TGME example demonstrates that the more thermodynamically stable form (field 3 in Fig. 19, right) is also less chemically reactive than the metastable crystalline mesophase (field 9). On the other hand, crystalline mesophase can be expected to be more stable than the corresponding amorphous material of the same composition, as was indeed observed for naficilin sodium [19]. In a more recent example, two solid forms of a novel drug, cevofecin, were produced, with characteristic XRPD patterns (Fig. 20). While both forms had a diffused halo typical of amorphous materials, one of the forms also had two weak but relatively sharp peaks in the lower-angle region of the XRPD pattern; the existence of these peaks was further confirmed using synchrotron SAXS (Fig. 20). Therefore, this material was identified as a crystalline mesophase, although its exact structure was not determined. Chemical stability of these two forms of cefovecin was compared at three different temperatures, 5, 25, and 40 °C, after 6 months storage. The two forms had a comparable water content of 2.5 wt% (2.2– 2.7 wt%) for crystalline mesophase and 2.6 wt% (2.2–3.2 wt%) for the amorphous material. The results are given in Fig. 20, providing both the potency and the level of individual degradants as determined by the reverse-phase high-performance liquid chromatography (RPHPLC). At all three storage temperatures, the crystalline mesophase has both higher potency and lower amount of essentially all individual impurities (with one exception of impurity no. 5) than the amorphous sample. Overall, the rank-order of chemical stability in a few cases considered above followed the thermodynamic stability, with crystalline mesophases demonstrating intermediate rates of chemical reactivity between the crystalline (lowest rate of reactivity/highest chemical stability) and amorphous (lowest chemical stability) states. 5.2. Physical stability For amorphous systems, their physical stability (i.e., absence of crystallization) is related to the rate of amorphous-to-crystalline transition, which slows down dramatically below the Tg. One could expect a similar behavior for crystalline mesophases, with the rates of phase transformation from a higher-temperature metastable mesophase to a thermodynamically stable crystal slowing down below the Tg. One such example was reported for a phospholipid, DOPC. In this case, a non-lamellar crystalline mesophase of DOPC, which is metastable in respect to the lamellar crystalline phase, Lc, was annealed at both above and below the Tg [60]. After annealing below the Tg, an enthalpy recovery endotherm was observed on the consecutive DSC curve, which is typical for both amorphous materials and mesophases as discussed in Section 3. When the same sample was annealed just above the Tg, 3-dimensional ordered crystalline phase, Lc, was formed, as evidenced

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

207

Fig. 19. Left: Kinetics curves for the solid-state methyl transfer reaction in three different forms of tetraglycine methyl ester. Dashed curves marked 2 and 3 represent fitting of the experimental data for a crystalline (3) and freeze-dried (2) samples. Symbols and solid curves represent reactivity of two milled samples of TGME. Reprinted from [E. Y. Shalaev, M. Shalaeva, S. R. Byrn, and G. Zografi, Effects of processing on the solid-state methyl transfer of tetraglycine methyl ester, Int. J. Pharm., 152 (1997), 75–88]. Copyright 1997, with permission from Elsevier. The diagram on the right (see Fig. 16) illustrates relative thermodynamical stability of the three different states of TGME, i.e., crystalline (3), crystalline mesophase (2), and two-phase disordered crystal + mesophase material (9).

by both the loss of the Tg and the observation of a strong melting peak on the DSC heating curve obtained after annealing. In another report [92], crystallization of geranoyl diethanolamide liquid crystal, forming a lamellar crystalline phase, was observed, but only after long-term (~year) storage of the sample at room temperature, which is well above the Tg of approximately − 70 °C. The

crystallization was accompanied by the loss of the glass transition, as expected, similar to the phospholipid example above. Caffeine is a system for which the molecular mobility plays a dramatic role in its physical stability. Caffeine has two known enantiotropically related polymorphic forms. The room temperature commercial form II is always accompanied by traces (several percent

Fig. 20. A: Potency (A1) and purity (A2, A3, A4) of two forms of cephalosporin after storage for 6 months at three temperatures. Crystalline mesophase (preparation 2) is more thermodynamically stable as illustrated in the Wunderlich's graph (C), and also it was shown to have a better chemical stability/lower rate of chemical reactivity. B: XRPD patterns (right) and synchrotron SAXS patterns (left) of bulk API (top patterns) and freeze-dried product (bottom patterns). The broad peak in the SAXS patterns is from the polymer tape which was used as the sample holder. C: Schematic diagram of thermodynamic stability of different phases in a single-component system (see Fig. 16).

208

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

depending of the manufacturer) of the high temperature metastable disordered phase I which is a source of instability. This latter phase is certainly produced during manufacturing process involving high temperature excursion, dehydration, or milling. The kinetics of conversion from I to II is extremely slow at room temperature, which is approximately 35 °C above the Tg (Tg of form I is approximately − 10 °C). By systematic time resolved investigation of this conversion, it could be shown that the activation energy of the process is close to that which controls the molecular motions in phase I [36]. Interestingly, the commercial form II is also dynamically disordered, albeit to a lesser degree. Up to now, no fully ordered crystal phase of this rather simple molecular compound has been found. The examples above demonstrate that, in order to understand the physical stability of metastable mesophases, it is essential to know their Tg, with the expectation that the rate of a corresponding phase transformation would slow down significantly below the Tg. However, the absolute rates of phase transformations above the Tg may vary significantly between different systems, from a few minutes (an example of DOPC above) to years (geranoyl diethanolamide and caffeine). In contrast to the stability of mesophases in the solid state as described above, their physical stability in aqueous dispersions has rarely been reported. A noticeable example was the solvent-mediated form transformation of liquid crystalline Ritonavir in 0.1 M hydrochloric acid (pH 1) at room temperature (22 °C). The in situ formed mesophase was initially generated from a suspension of the amorphous drug. Then this metastable mesophase gradually converted a crystalline form within 30 days [93].

5.3. Solubility and dissolution rate In terms of the Gibbs free energy, crystalline mesophase occupies an intermediate position between a crystalline form (lowest free energy) and an amorphous form of the same system (highest free energy),

thus creating an opportunity to increase of solubility and availability for crystalline forms with poor water solubility. Several examples of solubility and dissolution behavior of crystalline mesophases vs crystalline and amorphous forms are considered below. Ritonavir molecules selfassemble to form nanostructures with the mobility and order characteristic of a lyotropic liquid crystal [93]. The dissolution rates and solubilities of ritonavir were measured to be: amorphous N liquid crystal ≫ crystalline Form I N crystalline Form II. Specifically, the dissolution and solubility of lyotropic liquid crystal is slightly lower than that of the amorphous phase and about 20 times higher than that of Form II. In another example, the solubility and the dissolution rate of the supercooled thermotropic mesophase fenoprofen calcium in water was observed to be higher than that of the crystalline dihydrate [94]. In a separate study, solubility of the thermotropic reversed hexagonal phase fenoprofen calcium in water was compared to that of the crystalline dihydrate, as measured in both pure bulk drug and in a proprietary tablet formulation (Nalfon®) [95]. In this work, the solubility of the neat crystalline dihydrate and the drug in the tablet formulation was about 2.8 and 3.0 mg/ml, respectively, whereas the maximum solubility of the liquid crystal was 5.0 and 6.9 mg/ml, respectively.

5.4. Crystalline mesophases encountered in pharmaceutical manufacturing Manufacturing of pharmaceutical products involves various processes during which the material can change its chemical composition (e.g., by desolvation) and is exposed to various environmental conditions including temperatures from sub-zero (e.g., freeze-drying) to above 100 °C (e.g., during terminal sterilization) as well as a range of water activities and mechanical stresses. Under conditions of pharmaceutical manufacture, one can expect that materials can also undergo phase transitions including amorphous/crystalline transformations. For example, milling promotes crystalline-to-amorphous transformation in many cases. Such milling-induced disordering can proceed via

Fig. 21. (A) SAXS patterns of an aqueous solution of compound 1 obtained during cooling. SAXS/WAXS patterns were obtained on ID02 beamline of the European Synchrotron Radiation Facility (ESRF). (B) Overlay of SAXS patterns of compound 1 in solution at room temperature, in frozen solutions, and in dried samples at two different water contents. (C) d-spacing of the crystalline mesophase of frozen aqueous solutions compound 1 as a function of temperature at different concentrations. (D) chemical structure of compound 1.

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

intermediate mesophase. Indeed, milling can result in a loss of ordering in a certain crystallographic direction while preserving long-range order in other crystallographic planes [17]; such partially disordered material could retain overall rotational ordering and also one- or twodimensional translational order, i.e., representing a crystalline mesophase. Freeze-drying, another typical pharmaceutical processing, which involves dissolution of a (usually) crystalline API in an aqueous media followed by freezing, ice sublimation, and secondary drying, commonly produces amorphous structures. On the other hand, an exposure of an initially amorphous material to higher RH/temperature can result in amorphous-to-crystalline transformation. In this section, several examples of crystalline mesophases observed during typical manufacturing treatment are discussed. Formation of crystalline mesophases during freeze-drying might be relatively common. For example, crystalline mesophases in freezedried calcium benzoate and sodium deoxycholate were reported in [96,97], respectively. In another example, a crystalline mesophase was observed during freeze-drying of an exploratory drug candidate, compound 1, with the molecular structure presented in Fig. 21 [98]. While the initial material was highly crystalline, the XRPD pattern of the freeze-dried samples had a broad halo typical for amorphous solids. However, polarized light microscopy (PLM) test showed birefringence indicative of an anisotropic structure. In addition, small-angle X-ray scattering (SAXS) pattern of the freeze-dried material had a relatively sharp peak (Fig. 21). Therefore, the freeze-dried form of compound 1 can be classified as a crystalline mesophase. To understand at what stage of the manufacturing process the crystalline mesophase is obtained, low-temperature SAXS/wide-angle X-ray scattering (WAXS) study was performed. A lamellar phase, which is characterized by the SAXS peaks with the diffraction angle ratio 1:2, was observed in frozen aqueous solutions at −40 °C and −10 °C (Fig. 21). Comparison of the SAXS patterns of the frozen solutions with the one for the freeze-dried samples reveals that the second-order diffraction was no longer observed in the freeze-dried material. Evolution of the SAXS patterns, from aqueous solution at RT to frozen solution to dried materials with a higher and lower water contents, is summarized in Fig. 21. The results imply that the lamellar mesophase, which was observed in the frozen solution, transformed into another mesophase, which is characterized by a single SAXS peak centered at q of approx. 2.2 nm−1, during the drying segment of the freeze-drying process. In addition to the SAXS peak, a diffuse halo was observed in the dried samples, probably indicative of a two-phase system mesophase + isotropic glass. Manufacturing process can also change properties of crystalline mesophases formed by excipients. Such example of a drug product, which was formulated with compritol in order to control the rate of drug release, is described here. SAXS patterns of the drug product have at least four peaks with the position ratio of 1:2:3:4 (Fig. 22), indicative of the lamellar structure formed by compritol. During the manufacture, an annealing (aging) step was introduced to achieve consistent dissolution profiles. SAXS patterns were obtained for several lots both before and after aging step and are also shown in Fig. 22. The aged lots show relatively small but consistent shift in the peak position, as exemplified in Fig. 22 (insert) for one of the lot and summarized in Fig. 22, bottom. As shown in the Fig. 22, the d-spacing for freshly prepared materials varied from lot to lot, whereas aging resulted in consistent and reproducible values. Similar trend, i.e., variability for freshly made lots and consistent results after aging, was observed when the same lots were tested using dissolution method (dissolution data not shown). A potential (although probably not the only one) interpretation of these observations is that a portion of the active ingredient is confined in the interlamellar space of the compritrol matrix, and variations in the interlamellar distance reflect variable amount of drug in this phase, causing variability in the dissolution rates. After aging, the consistent interlamellar distance, and therefore consistent amount of the drug between lamellar layers, was achieved, resulting in improved reproducibility in the dissolution behavior between different lots.

209

Fig. 22. SAXS patterns (top) and d-spacing representing the first diffraction peak (bottom) of a controlled release drug product formulated in compritol.

6. Conclusion It is important for a pharmaceutical scientist to consider that any pure phase, including ordered crystals, amorphous solids, and isotropic melts, might exhibit mesophase behavior under conditions of temperature/solvent change and dissolution [99]. These mesophases and particularly liquid crystals can cause multiple phase transitions, as the system moves from the solid or semi-solid to liquid crystal (LC) to isotropic liquid phases and potentially back again if the energy is removed or if the chemical composition (e.g., due to water sorption/ desorption) is changed. LC systems have low temperature thresholds for these phase transitions, meaning that the energy needed to induce them can accidentally be exceeded by some manufacturing process. These phase changes have a considerable impact on the properties of a pharmaceutical formulation, and screening of the drug and excipients (such as cellulose-derived excipients and glycerides) for LC behavior should be a priority. Furthermore, crystalline mesophases could have some unique biophysical properties. It was proposed, for example, that certain lyotropic LC formers in liquid pharmaceutical formulations might cause hemolysis and opalescence [20]; if this is indeed the case, it would re-emphasize the need for a detailed structural characterization of pharmaceutical products. On the other hand, an intriguing observation has been recently reported by Fukushi et al. [100], who observed correlation between liquid crystallinity and anticancer activity of a series of small mesogenic molecules. They detected greater inhibitory effect of the cell proliferation with liquid crystal-forming benzoates than

210

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211

those of common crystal structures. This anti-proliferative activity of the liquid crystals was attributed to their ability to form a spherical molecular aggregation. It should be noted, however, that the mechanism of the effect of the liquid crystallinity at the cellular and molecular levels is not known, and that the alternative and simpler explanation, i.e., that the difference is due to chemical structure in the compounds tested (rather than their liquid crystalline character) cannot be ruled out. One of the potential benefits of mesophases in oral drug delivery may be the ability to form stable colloidal species in solution. It has been shown that drugs, particularly those which are amphiphilic in nature, can form aggregates and colloidal species when introduced to an aqueous environment that mimics the conditions of the intestine [101]. Furthermore, the design of drug into colloidal architectures has been shown to very effectively sustain drug concentrations higher than its thermodynamic solubility and to increase the absorption of poorly soluble drugs [102]. Typically, this is achieved by formation of an amorphous drug form stabilized by polymer [103]. One potential advantage of a mesophase over the amorphous form would be to reduce the driving force for transformation to the crystalline phase, either through self-assembly or stabilization from a polymer. Overall, we conclude that formation of mesophases could have a significant impact on biopharmaceutical properties, including degradation and drug release kinetics, and detailed investigation of all the aspects of disordered materials (structure, dynamics, and thermodynamics) is essential in order to understand performance and stability of pharmaceuticals. At the same time, mesophases could deliver similar benefit as amorphous materials, i.e., improved apparent solubility and higher dissolution rate than crystalline forms, at the same time reducing risk of physical instability (form conversion) and precipitation in solution as compared to amorphous forms.

Acknowledgments Joanne Lukaszewicz and Renuka Reddy for sample preparation and HPLC analysis for cefovecin, Jinyang Hong for supporting investigations of compound 1, and anonymous reviewers for helpful comments.

References [1] B. Wunderlich, A classification of molecules, phases, and transitions as recognized by thermal analysis, Thermochim. Acta 340-341 (1999) 37–52. [2] E.Y. Shalaev, G. Zografi, The concept of 'structure' in amorphous solids from the perspective of the pharmaceutical sciences, in: H. Levine (Ed.), Amorphous Food and Pharmaceutical Systems, The Royal Society of Chemistry 2002, pp. 11–30. [3] C.L. Stevenson, D.B. Bennett, D. Lechuga-Ballesteros, Pharmaceutical liquid crystals: the relevance of partially ordered systems, J. Pharm. Sci. 94 (2005) 1861–1880. [4] C.C. Müller-Goymann, Physicochemical characterization of colloidal drug delivery systems such as reverse micelles, vesicles, liquid crystals and nanoparticles for topical administration, Eur. J. Pharm. Biopharm. 58 (2004) 343–356. [5] B.S. Pattni, V.V. Chupin, V.P. Torchilin, New developments in liposomal drug delivery, Chem. Rev. 115 (2015) 10938–10966. [6] O. Haida, H. Suga, S. Seki, Calorimetric study of the glassy state XII. Plural glasstransition phenomena of ethanol, J. Chem. Thermodyn. 9 (1977) 1133–1148. [7] H. Suga, Ordering processes in some disordered systems, Pure Appl. Chem. 55 (1983) 427–436. [8] M.J. Pikal, A.L. Lukes, J.E. Lang, K. Gaines, Quantitative crystallinity determinations for beta-lactam antibiotics by solution calorimetry: correlations with stability, J. Pharm. Sci. 67 (1978) 767–773. [9] R. Lefort, V. Caron, J.F. Willart, M. Descamps, Mutarotational kinetics and glass transition of lactose, Solid State Commun. 140 (2006) 329–334. [10] N. Dujardin, E. Dudognon, J.F. Willart, A. Hédoux, Y. Guinet, L. Paccou, M. Descamps, Solid state mutarotation of glucose, J. Phys. Chem. B 115 (2011) 1698–1705. [11] C.A. Angell, Liquid fragility and the glass transition in water and aqueous solutions, Chem. Rev. 102 (2002) 2627–2650. [12] B.C. Hancock, G. Zografi, Characteristics and significance of the amorphous state in pharmaceutical systems, J. Pharm. Sci. 86 (1997) 1–12. [13] F. Franks, Scientific and technological aspects of aqueous glasses, Biophys. Chem. 105 (2003) 251–261. [14] S. Bates, G. Zografi, D. Engers, K. Morris, K. Crowley, A. Newman, Analysis of amorphous and nanocrystalline solids from their X-ray diffraction patterns, Pharm. Res. 23 (2006) 2333–2349.

[15] J.R. Authelin, A.P. MacKenzie, D.H. Rasmussen, E.Y. Shalaev, Water clusters in amorphous pharmaceuticals, J. Pharm. Sci. 103 (2014) 2663–2672. [16] H. Bunjes, T. Rades, Thermotropic liquid crystalline drugs, J. Pharm. Pharmacol. 57 (2005) 807–816. [17] T.P. Shakhtschneider, V.V. Boldyrev, Mechanochemical synthesis and mechanical activation of drugs, in: E.V. Boldyreva, V.V. Boldyrev (Eds.), Reactivity of Molecular Solids, John Wiley & Sons, Chichester, N.Y. 1999, pp. 271–311. [18] B.G. Demirjian, G. Dosseh, A. Chauty, M.L. Ferrer, D. Morineau, C. Lawrence, K. Takeda, D. Kivelson, S. Brown, Metastable solid phase at the crystallineamorphous border: the glacial phase of triphenyl phosphite, J. Phys. Chem. B 105 (2001) 2107–2116. [19] N. Milton, S.L. Nail, The physical state of nafcillin sodium in frozen aqueous solutions and freeze-dried powders, Pharm. Dev. Technol. 1 (1996) 269–277. [20] Y. Cui, A material science perspective of pharmaceutical solids, Int. J. Pharm. 339 (2007) 3–18. [21] W. Bernhard, C. Wei, The difference between liquid crystals and conformationally disordered crystals, in: A.I. Isayev, T. Kyu, S.Z.D. Cheng (Eds.), Liquid-Crystalline Polymer Systems, ACS Symposium Series, vol. 632, American Chemical Society 1996, pp. 232–248. [22] S. Singh, Liquid Crystals: Fundamentals, World Scientific Publishing, Singapore, 2002. [23] S. Kumar, Liquid Crystals: Experimental Study of Physical Properties and Phase Transitions, Cambridge University Press, Cambridge, UK, 2001. [24] A. Guinier, X-ray Diffraction in Crystals, Imperfect Crystals, and Amorphous Bodies, Dover, New York, 1994 (387 pp.). [25] A. Guinier, Théorie et technique de la radiocristallographie, Dunod, Paris, 1964. [26] R. Hosemann, Der ideale Parakristall und die von ihm gestreute kohärente Röntgenstrahlung, Z. Phys. 128 (1950) 465–492. [27] L. Temleitner, L. Pusztai, Local order and orientational correlations in liquid and crystalline phases of carbon tetrabromide from neutron powder diffraction measurements, Phys. Rev. B 81 (2010) 134101–134108. [28] J.N. Sherwood (Ed.), The Plasctically Crystalline State, John Wiley & Sons, Hoboken, New Jersey, 1978. [29] S.J.L. Billinge, M.F. Thorpe (Eds.), Local Structure from Diffraction, Springer, New York, 1998 (399 pp.). [30] T. Shimizu, S. Tanaka, N. Onoda-Yamamuro, S. Ishimaru, R. Ikeda, New rotator phase revealed in di-n-alkylammonium bromides studied by solid-state NMR, powder XRD, electrical conductivity and thermal measurements, J. Chem. Soc. Faraday Trans. 93 (1997) 321–326. [31] A. Srinivasan, F.J. Bermejo, A. de Andrés, J. Dawidowski, J. Zúñiga, Criado, Evidence for a supercooled pastic-crystal phase in solid ethanol, Phys. Rev. B 53 (1996) 8172–8175. [32] M. Descamps, N.T. Correia, P. Derollez, F. Danede, F.D.R. Capet, Plastic and glassy crystal states of caffeine, J. Phys. Chem. B 109 (2005) 16092–16098. [33] P. Derollez, N.T. Correia, F. Danède, F. Capet, F. Affouard, J. Lefebvre, M. Descamps, Ab Initio structure determination of the high-temperature phase of anhydrous caffeine by X-ray powder diffraction, Acta Crystallogr. Sect. B 61 (2005) 329–334. [34] M. Descamps, J.F. Willart, Slow non-equilibrium process in a glassy crystal, J. NonCryst. Solids 172-174 (Part 1) (1994) 510–519. [35] G.D. Enright, V.V. Terskikh, D.H. Brouwer, J.A. Ripmeester, The structure of two anhydrous polymorphs of caffeine from single-crystal diffraction and ultrahigh-field solid-state 13C NMR spectroscopy, Cryst. Growth Des. 7 (2007) 1406–1410. [36] M. Descamps, A.A. Decroix, Polymorphism and disorder in caffeine: dielectric investigation of molecular mobilities, J. Mol. Struct. 1078 (2014) 165–173. [37] T. Denicourt, A. Hédoux, Y. Guinet, J.F. Willart, M. Descamps, Raman scattering investigations of the stable and metastable phases of cyanoadamantane glassy crystal, J. Phys. Chem. B 107 (2003) 8629–8636. [38] N.A. Murugan, A. Sayeed, Thermal behavior of disordered phase of caffeine molecular crystal: insights from Monte Carlo simulation studies, J. Chem. Phys. 130 (2009) 204514-1–204514-8. [39] P. Derollez, J. Lefebvre, M. Descamps, Neutron diffuse scattering in the plastic phase of succinonitrile, J. Phys. Condens. Matter 2 (2015) 9975–9987. [40] M. Floreani, F. Carpenedo, R. Santi, A.R. Contessa, Studies on the systems involved in succinonitrile metabolism in liver, Pharmacol. Res. Commun. 12 (1980) 433–439. [41] A.R. Contessa, R. Santi, Liberation of cyanide from succinonitrile, Biochem. Pharmacol. 22 (1973) 827–832. [42] P.J. Alarco, Y. Abu-Lebdeh, A. Abouimrane, M. Armand, The plastic-crystalline phase of succinonitrile as a universal matrix for solid-state ionic conductors, Nat. Mater. 3 (2004) 476–481. [43] S. Hore, R. Dinnebier, W. Wen, J. Hanson, J. Maier, Structure of plastic crystalline succinonitrile: high-resolution in situ powder diffraction, Z. Anorg. Allg. Chem. 635 (2009) 88–93. [44] M. Descamps, Space correlations in succinonitrile observed by X-ray scattering, Solid State Commun. 14 (1974) 77–81. [45] H. Fontaine, M. Bee, Sur la structure du succinonitrile dans sa phase plastique, Bull. Soc. Fr. Minéral. Cristallogr. 95 (1972) 441–450. [46] P. Derollez, J. Lefebvre, M. Descamps, W. Press, H. Fontaine, Structure of succinonitrile in its plastic phase, J. Phys. Condens. Matter 2 (1990) 6893–6903. [47] M. Bee, R.E. Lechner, J.P. Amoureux, R. Fouret, Temperature dependence of rotational isomerisation and molecular reorientation rates in plastic succinonitrile, J. Phys. C Solid State Phys. 16 (1983) 4973–4983. [48] H. Fontaine, R. Fouret, gd'isomérisation dans la phase plastique du succinonitrile, Adv. Mol. Relax. Process. 5 (1973) 391–394. [49] M. Descamps, G. Coulon, Dielectric constant of disordered dipoles with steric hindrance: I. plastic succinonitrile, Chem. Phys. 19 (1977) 347–352.

E. Shalaev et al. / Advanced Drug Delivery Reviews 100 (2016) 194–211 [50] B.C. Hancock, G. Zografi, The relationship between the glass transition temperature and the water content of amorphous pharmacetuical solids, Pharm. Res. 11 (1994) 471–477. [51] B.C. Hancock, S. Shamblin, G. Zografi, Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures, Pharm. Res. 12 (1995) 799–806. [52] J. Liu, D.R. Rigsbee, C. Stotz, M.J. Pikal, Dynamics of pharmaceutical amorphous solids: the study of enthalpy relaxation by isothermal microcalorimetry, J. Pharm. Sci. 91 (2002) 1853–1862. [53] S. Bhattacharya, R. Suryanarayanan, Local mobility in amorphous pharmaceuticals: characterization and implications on stability, J. Pharm. Sci. 98 (2009) 2935–2953. [54] G.P. Johari, S. Kim, R.M. Shanker, Dielectric study of equimolar acetaminophenaspirin, acetaminophen-quinidine, and benzoic acid-progesterone molecular alloys in the glass and ultraviscous states and their relevance to solubility and stability, J. Pharm. Sci. 99 (2010) 1358–1374. [55] G.P. Johari, Secondary relaxations and the properties of glasses and liquids, in: T. Dorfmüller, G. Williams (Eds.), Molecular Dynamics and Relaxation Phenomena in Glasses, Springer, Berlin Heidelberg 1987, pp. 90–112. [56] B. Wunderlich, J. Grebowicz, Thermotropic mesophases and mesophase transitions of linear, flexible macromolecules, in: N.A. Platé (Ed.), Liquid Crystal Polymers II/III, Springer, Berlin Heidelberg 1984, pp. 1–59. [57] G.P. Johari, S.J. Jones, Study of the low-temperature "transition" in ice Ih by thermally stimulated depolarization measurement, J. Chem. Phys. 62 (1975) 4213–4223. [58] D. Lechuga-Ballesteros, A. Abdul-Fattah, C.L. Stevenson, D.B. Bennett, Properties and stability of a liquid crystal form of cyclosporine—the first reported naturally occurring peptide that exists as a thermotropic liquid crystal, J. Pharm. Sci. 92 (2003) 1821–1831. [59] P. Chakravarty, S. Bates, L. Thomas, Identification of a potential conformationally disordered mesophase in a small molecule: experimental and computational approaches, Mol. Pharm. 10 (2013) 2809–2822. [60] E.Y. Shalaev, G. Zografi, P.L. Steponkus, Occurrence of glass transitions in longchain phosphatidylcholine mesophases, J. Phys. Chem. B 114 (2010) 3526–3533. [61] S. Havriliak, S. Negami, A complex plane representation of dielectric and mechanical relaxation processes in some polymers, Polymer 8 (1967) 161–210. [62] S. Urban, C.M. Roland, Low frequency relaxation in liquid crystals in relation to structural relaxation in glass-formers, J. Non-Cryst. Solids 357 (2011) 740–745. [63] R. Casalini, C.M. Roland, Effect of regioisomerism on the local dynamics of polychlorostyrene, Macromolecules 47 (2014) 4087–4093. [64] Th. Bauer, M. Köhler, P. Lunkenheimer, A. Loidl, C.A. Angell, Relaxation dynamics and ionic conductivity in a fragile plastic crystal, J. Chem. Phys. 133 (2010) 144509. [65] U. Schneider, R. Brand, P. Lunkenheimer, A. Loidl, Excess wing in the dielectric loss of glass formers: a Johari–Goldstein β relaxation, Phys. Rev. Lett. 84 (2000) 5560–5563. [66] M. Tokita, S.i. Funaoka, J. Watanabe, Study on smectic liquid crystal glass and isotropic liquid glass formed by thermotropic main-chain liquid crystal polyester, Macromolecules 37 (2004) 9916–9921. [67] S. Luthra, M.J. Pikal, Stabilization of lyophilized pharmaceuticals by control of molecular mobility: impact of thermal history, in: F. Jameel, S. Hershenson (Eds.), Formulation and Process Development Strategies for Manufacturing Biopharmaceuticals, John Wiley & Sons, Inc., Hoboken, NJ 2010, pp. 521–548. [68] B. Wang, M.T. Cicerone, Y. Aso, M.J. Pikal, The impact of thermal treatment on the stability of freeze-dried amorphous pharmaceuticals: II. aggregation in an IgG1 fusion protein, J. Pharm. Sci. 99 (2010) 683–700. [69] K. Kawakami, M.J. Pikal, Calorimetric investigation of the structural relaxation of amorphous materials: evaluating validity of the methodologies, J. Pharm. Sci. 94 (2005) 948–965. [70] E.Y. Shalaev, P.L. Steponkus, Phase behavior and glass transition of 1,2dioleoylphosphatidylethanolamine (DOPE) dehydrated in the presence of sucrose, Biochim. Biophys. Acta Biomembr. 1514 (2001) 100–116. [71] E.Y. Shalaev, P.L. Steponkus, Glass transition of a synthetic phospholipid in the lamellar phase, J. Phys. Chem. B 107 (2003) 8734–8737. [72] J.F. Mano, N.T. Correia, J.J. Moura-ramos, S.R. Andrews, G. Williams, Dipolar relaxation mechanisms in the vitreous state, in the glass transition region and in the mesophase, of a side chain polysiloxane liquid crystal, Liq. Cryst. 20 (1996) 201–217. [73] M. Descamps, J.F. Willart, Glassy crystals: above and below Tg, AIP Conf. Proc. 708 (2004) 591. [74] F.J. Bermejo, H.E. Fischer, M.A. Ramos, A. de Andrés, J. Dawidowski, R. Fayos, Microscopic dynamics in glasses in relation to that shown by other complex systems, in: M. Rubi, C. Pérez-Vicente (Eds.), Complex Behaviour of Glassy Systems, Springer, Berlin Heidelberg 1997, pp. 44–61. [75] D. Chen, H.G. Zachmann, Glass transition temperature of copolyesters of PET, PEN and PHB as determined by dynamic mechanical analysis, Polymer 32 (1991) 1612–1621.

211

[76] F. Atassi, C. Mao, A.S. Masadeh, S.R. Byrn, Solid-state characterization of amorphous and mesomorphous calcium ketoprofen, J. Pharm. Sci. 99 (2010) 3684–3697. [77] M. Sorai, Heat capacities of thermotropic liquid crystals with low molecular weight, Thermochim. Acta 88 (1985) 1–16. [78] S. Benkhof, A. Kudlik, T. Blochowicz, E. Rössler, Two glass transitions in ethanol: a comparative dielectric relaxation study of the supercooled liquid and the plastic crystal, J. Phys. Condens. Matter 10 (1998) 8155–8171. [79] E.Y. Shalaev, P.L. Steponkus, Depression of the glass transition temperature of sucrose confined in a phospholipid mesophase, Langmuir 17 (2001) 5137–5140. [80] C.A. Wulff, E.F. Westrum Jr., Heat Capacities and thermodynamic properties of globular molecules. VI. Succinonitrile, J. Phys. Chem. 67 (1963) 2376–2381. [81] E.B. Sirota, Polymer crystallization: metastable mesophases and morphology, Macromolecules 40 (2007) 1043–1048. [82] H. Kraack, M. Deutsch, E.B. Sirota, n-Alkane homogeneous nucleation: crossover to polymer behavior, Macromolecules 33 (2000) 6174–6184. [83] H. Kraack, E.B. Sirota, M. Deutsch, Homogeneous crystal nucleation in short polyethylenes, Polymer 42 (2001) 8225–8233. [84] N.H. Tinh, H. Gasparoux, New series exhibiting pure enantiotropic nematic reentrant compounds at atmospheric pressure, Mol. Cryst. Liq. Cryst. 49 (1979) 287–291. [85] J.M. Seddon, Structure of the inverted hexagonal (HII) phase, and non-lamellar phase transitions of lipids, Biochim. Biophys. Acta Biomembr. 1031 (1990) 1–69. [86] M.M. Kozlov, S. Leikin, R.P. Rand, Bending, hydration and interstitial energies quantitatively account for the hexagonal-lamellar-hexagonal reentrant phase transition in dioleoylphosphatidylethanolamine, Biophys. J. 67 (1994) 1603–1611. [87] E.Y. Shalaev, P.L. Steponkus, Phase diagram of 1,2-dioleoylphosphatidylethanolamine (DOPE):water system at subzero temperatures and at low water contents, Biochim. Biophys. Acta Biomembr. 1419 (1999) 229–247. [88] B. Wunderlich, Thermodynamic description of condensed phases, J. Therm. Anal. Calorim. 102 (2010) 413–424. [89] E.Y. Shalaev, M. Shalaeva, S.R. Byrn, G. Zografi, Effects of processing on the solidstate methyl transfer of tetraglycine methyl ester, Int. J. Pharm. 152 (1997) 75–88. [90] E.Y. Shalaev, M. Shalaeva, G. Zografi, The effect of disorder on the chemical reactivity of an organic solid, tetraglycine methyl ester: change of the reaction mechanism, J. Pharm. Sci. 91 (2002) 584–593. [91] T. Luty, C.U. Eckhardt, General theoretical concepts for solid state reactions: quantitative formulation of the reaction cavity, steric compression, and reactioninduced stress using an elastic multipole representation of chemical pressure, J. Am. Chem. Soc. 117 (1995) 2441–2452. [92] S.M. Sagnella, C.E. Conn, I. Krodkiewska, C.J. Drummond, Nonionic diethanolamide amphiphiles with isoprenoid-type hydrocarbon chains: thermotropic and lyotropic liquid crystalline phase behaviour, Phys. Chem. Chem. Phys. 13 (2011) 17511–17520. [93] B. Rodriguez-Spong, A. Acciacca, D. Fleisher, N. Rodriguez-Hornedo, pH-induced nanosegregation of ritonavir to lyotropic liquid crystal of higher solubility than crystalline polymorphs, Mol. Pharm. 5 (2008) 956–967. [94] T. Rades, Y. Padmadisastra, C.C. Müller-Goymann, Thermal behaviour and solubility of fenoprofen calcium, Pharmazie 51 (1996) 846–851. [95] J. Patterson, J., A. Bary, T. Rades, Physical stability and solubility of the thermotropic mesophase of fenoprofen calcium as pure drug and in a tablet formulation, Int. J. Pharm. 247 (2002) 147–157. [96] A. Terakita, S.R. Byrn, Structure and physical stability of hydrates and thermotropic mesophase of calcium benzoate, J. Pharm. Sci. 95 (2006) 1162–1172. [97] Z. Su, The Investigation of Pharmaceutical Liquid Crystals: Formation, Stability and Phase Behavior, Purdue, 2011 (Thesis). [98] M. Casteel, E.Y. Shalaev, J. Hong, S. Ewing, S. Kumar, J. Krzyzaniak, Impact of solution concentration and temperature on API liquid crystal formation during lyophilization, AAPS meeting, 2011. [99] F. Atassi, S. Byrn, General trends in the desolvation behavior of calcium salts, Pharm. Res., 23 (2006) 2405-241. [100] Y. Fukushi, H. Yoshino, J. Ishikawa, M. Sagisaka, I. Kashiwakura, A. Yoshizawa, Effects of liquid crystallinity on anticancer activity of benzoate derivatives possessing a terminal hydroxyl group, Liq. Cryst. 41 (2014) 1873–1878. [101] A.K. Doak, H. Wille, S.B. Prusiner, B.K. Shoichet, Colloid formation by drugs in simulated intestinal fluid, J. Med. Chem. 53 (2010) 4259–4265. [102] William Curatolo, James A. Nightingale, Scott M. Herbig, Utility of hydroxypropylmethylcellulose acetate succinate (HPMCAS) for initiation and maintenance of drug supersaturation in the GI milieu, Pharm. Res. 25 (2008) 1419–1431. [103] D.T. Friesen, R. Shanker, M. Crew, D.T. Smithey, W.J. Curatolo, J.A.S. Nightingale, Hydroxypropyl methylcellulose acetate succinate-based spray-dried dispersions: an overview, Mol. Pharm. 5 (2008) 1003–1019.