Polymorphism and conformations of mefenamic acid in supercritical carbon dioxide

J. of Supercritical Fluids 152 (2019) 104547

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu

Polymorphism and conformations of mefenamic acid in supercritical carbon dioxide Roman D. Oparin a,∗ , Yevhenii A. Vaksler b,c , Michael A. Krestyaninov a , Abdenacer Idrissi b , Svitlana V. Shishkina c,d , Michael G. Kiselev a a

G.A. Krestov Institute of Solution Chemistry of the Russian Academy of Sciences (RAS), Akademicheskaya str. 1, Ivanovo, 153045, Russia Laboratoire de Spectrochimie Infrarouge et Raman (UMR CNRS A8516), Université Lille, 59655, Villeneuve d’Ascq Cedex, France c V.N. Karazin Kharkiv National University, 4 Svobody Sq., Kharkiv, 61022, Ukraine d SSI “Institute for Single Crystals” NAS of Ukraine, 60 Nauky ave., Kharkiv, 61001, Ukraine b

h i g h l i g h t s

g r a p h i c a l

a b s t r a c t

• Combination of IR and Raman spectroscopy and quantum chemical calculations. • Correlation of mefenamic acid polymorphs and conformers in supercritical solution. • Possibility monitoring the polymorphic transformation of mefenamic acid. • Monitoring of MA conformers in a supercritical fluid solution.

a r t i c l e

i n f o

Article history: Received 18 April 2019 Received in revised form 23 May 2019 Accepted 23 May 2019 Available online 29 May 2019 Keywords: Mefenamic acid Polymorphism Conformational manifold Supercritical fluid technologies IR spectroscopy Raman spectroscopy

a b s t r a c t In this paper we have studied the conformations of mefenamic acid molecules in its saturated solution in supercritical carbon dioxide being in contact with bottom phase containing different polymorphs of mefenamic acid under isochoric heating conditions in the temperature range of 80–220 ◦ C. For this purpose, we used in situ IR and Raman spectroscopy as well as micro IR and micro Raman spectroscopy and quantum chemical calculations. We have found that there is a correlation between the polymorphic modifications of the crystalline form of mefenamic acid and its conformers in CO2 fluid phase. The results of this work have shown that it is possible to monitor the polymorphic transformation of the crystalline phase of mefenamic acid by screening its molecular conformations in a fluid solution. This finding opens up the possibilities to manage the obtaining of certain polymorph of mefenamic acid with high purity degree from its fluid solution by means of variation of thermodynamic parameters of this solution. © 2019 Elsevier B.V. All rights reserved.

1. Introduction

Abbreviations: MA, mefenamic acid; API, active pharmaceutical ingredient; IR, infrared; HTHP, high pressure high temperature; scCO2 , supercritical carbon dioxide. ∗ Corresponding author. E-mail address: [email protected] (R.D. Oparin). https://doi.org/10.1016/j.supflu.2019.104547 0896-8446/© 2019 Elsevier B.V. All rights reserved.

According to the Biopharmaceutics Classification System from the World Health Organization and the European Medicines Agency [1,2] all the medications can be characterized by individual solubility and permeability values, which are two main attributes affecting

2

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 1. (a) General view of MA molecular structure; (b) Crystal structure of the MA polymorph I [8], (b,c) Crystal structures of the MA polymorph II [9]. For form II two crystal structures exist for which the values of dihedral angle related to rotation of two-methyl substituted benzene ring differ. These crystal structures were built in the MERCURY program [20].

the bioavailability of drugs. Most active pharmaceutical ingredients (APIs) have low aqueous solubility [3,4]. All this has led to the attempts to improve their solubility by applying various techniques such as micronization or loading of APIs into the porous matrix. The control of the polymorphic form of the API was considered as an efficient way to improve the aqueous solubility [5,6]. Indeed, adequate selection of a polymorph form makes it possible to choose the most appropriate dissolution profile available for the drug and thus to control its period of therapeutic effect and release rate of API. Mefenamic acid (MA) is a well-known analgesic-antipyretic agent belonging to the fenamates class (anthranilic acid derivatives) of nonsteroidal anti-inflammatory drugs (NSAIDs) and is used to treat mild to moderate pain. The general view of MA molecular structure is presented in Fig. 1a. Three polymorphs of MA have been discovered [7–9] (see Fig. 1b–d) and the properties of the first two have been characterized. The third form has very low stability and moreover it cannot be obtained by direct crystallization [9]. It has been shown that the second form has a higher dissolution rate [7,10]. However, the ratio of polymorphs in commercial forms of MA is not usually specified and this influences significantly the measured values of the relative concentration of MA in blood plasma. This is illustrated by the enormous scattering of the reported data which range between 28 and 86% [11]. It should be mentioned that this ratio also is not available when quantifying the solubility of MA [10,12–19] and the solubility of the commercial form of MA in water at normal conditions has an extremely low value (the molar fraction is lower than 10−4 ) [12]. In order to control the formation of specific polymorphic form, thermal conversion upon heating above 155 ◦ C at ambient pressure [15] and recrystallization from polar aprotic solvents [7,21,22], among which DMF is the most popular [13,23], have been proposed. However, there are three issues, which significantly limit the use of these methods: high speed of sublimation, which is faster than the conversion between crystalline forms [15], toxicity of DMF and incomplete conversion from the first polymorph into the second one [13,24] and finally, the second polymorph form is not stable in

the wet state, which is a drawback during both its synthesis and long-term storage [25,26]. Nevertheless, it was shown that when MA is kept in dry state, the back-transformation of the second form to the first one does not occur even within 90 days [26,27]. Based on these results, we propose in this paper to use supercritical CO2 to prepare specific polymorphic form of MA with the advantage of a complete control the environment (in particular of the humidity) [28,29]. Furthermore, the process of synthesizing new polymorphs has been so far assessed by analyzing the final form by off-line analytical methods, which often are empiric, however, the changes occurring during their formation have not been quantified [30,31]. As a consequence, we propose in this paper to overcome this limitation and to apply in situ IR spectroscopy which has the advantage of monitoring the processes occurring “in place” or “in work”. It gives precise information on the recrystallization dynamics and product formation and points out to the factors that influence the transformation between the various polymorphs [32,33]. The choice of MA as a subject of inquiry in the present investigation is dictated by the fact that MA molecule contains two aromatic fragments conjugated via NH group and one of them is conjugated with carboxylic group. An accurate attribution of the spectral signatures of the MA polymorphs is available and associated with the arrangement of the NH and CO moieties in MA molecular structure [34]. In its turn the quantum chemical calculations allow the spectral assignment of the MA conformers in vacuum. As we have shown in our previous work [35–37], this assignment is mandatory in order to identify, using IR spectroscopy, the conformations of MA molecules that are dissolved in CO2 phase. We then showed the correlation between the polymorphic forms of MA and the conformations of its molecules occurring in the supercritical CO2 . In addition to above mentioned advantages of using in situ vibration spectroscopy, these methods also allows to characterize the conformations of MA molecules that are dissolved in scCO2 phase being in contact with the crystalline form of MA. The experimental conditions (temperature, pressure, and the control of the environment content) was strictly controlled and the

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

problem of reproducibility of polymorphs synthesis encountered using other methods was resolved [38,39]. In addition, the proposed method of synthesizing a specific polymorph form of MA can be considered as environmentally friendly because of the low toxicity of CO2 . Before going further, we want to clarify what we mean by the correlation between the conformation of MA that are defined in vacuum (basing on results of quantum chemical calculations) and then are considered as equivalent to those that would define in CO2 phase (basing on results of IR spectroscopy) and those that are present in the different solid forms of MA. The similarity between the molecular conformations in vacuum and in those in the crystalline form was suggested [40]. However, this was challenged by X-ray data. [41,42]. However, the exact path way involved in the formation of these polymorphic forms is still a subject of intensive researches [43–47]. In our work, we show that the conformational features of MA in CO2 phase and in its polymorphic forms are interdependent. The paper is organized as follow: in the first part we will introduce the materials and the methods we used in our project, the second part will present the results and their discussion and finally a conclusion will be made in the last part.

3

2. Materials and methods 2.1. Materials The mefenamic acid (2-[(2,3-Dimethylphenyl)amino]benzoic acid, N-(2,3-Xylyl)anthranilic acid), was purchased from “SigmaAldrich” (Ref. N. M4267-50 G). The CO2 gas (99.99% purity) was purchased from “The Linde Group. Gas Industriels France”. 2.2. Methods 2.2.1. In situ IR spectroscopy In order to characterize the conformational equilibriums of MA in the scCO2 phase which is in contact with the excess of MA crystalline form (with respect to the expected solubility of MA in scCO2 ) under isochoric heating conditions we used the same experimental setup as described in detail in our previous work [48]. However, in the present work we used an improved version of the high-pressurehigh-temperature (HPHT) optical cell with a variable optical path length (see Ref. [37]). This cell allows reaching the equilibrium concentration of the solute in scCO2 relatively fast and to record spectra with a high resolution as well as with a high signal-to-noise ratio

Fig. 2. (a) Schematic representation of the experimental HPHT optical cell for IR spectroscopy with a variable optical path length; (c) Schematic representation of the experimental HPHT optical cell for Raman spectroscopy; (b,d) Cells sectional views showing their internal geometry and position of the IR beam or excitation red laser beam relative to the bottom part of the cell containing the sample solid phase.

4

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

in a wide range of thermodynamic parameters of state. The general and sectional views of this cell are shown in Fig. 2a,b. In our cell we used the optical windows with an effective working diameter of 8 mm and thickness of 9 mm to stand the high pressure up to 1000 bar. The windows made of silicon allowed us to measure the IR spectra in the wavenumber range of 1000–7000 cm−1 . The sample of initial (commercial) crystalline form of MA with a diameter of 9 mm and thickness of 7 mm was placed into a sample holder in the bottom part of the cell (see Fig. 2a). In order to avoid the contamination of the initial sample by the residuals of atmospheric components (in particular, water and oxygen), the HPHT cell was pumped out to a residual pressure of 0.1 mbar. After that, the cell was filled with dry CO2 through a stainless steel capillary connected to a high pressure setup allowing the adjustment of the pressure with an accuracy of ±0.5 bar. The cell was heated by means of four cartridge heaters disposed in its body corners. Three thermocouples were used for temperature control. One of them was located in the vicinity of one of the four cartridge heaters and was connected to a proportional-integral-derivative (PID) controller allowing controlling the temperature with an accuracy of about ±1 ◦ C. The second thermocouple was located in the upper part of the cell body close to the fluid solution phase and the third one was located in the bottom part of the cell containing the crystalline form of MA. By choosing the thermocouple positions it is possible to control the temperature gradient between the bottom and the fluid phases. In our experiment, this gradient did not exceed 3 ◦ C even at the highest temperature studied. The measured mid IR spectra corresponded to the CO2 phase of a binary [MA–scCO2 ] mixture (see the sectional view in Fig. 2b). These spectra were obtained using a FTIR spectrometer Bruker Vertex 80 equipped with a DTGS detector in the wavenumber range of 1000–4000 cm−1 for isochoric conditions in the temperature range of 80–220 ◦ C with a step of 10 ◦ C at a density of CO2 corresponding to * = 1.1cr (CO2 ), where cr (CO2 ) = 10.625 mol·L−1 is the critical density of carbon dioxide. For this isochore, the pressure was in the range of 172–459 bar. At each temperature the initial spectrum was measured immediately when the temperature reached the target value. Since it takes some time to achieve the equilibrium concentration of MA in the scCO2 phase (when the intensity of the spectra remains constant), we have then recorded the dynamic spectra every 30 min to get information on the dissolution of MA in the scCO2 phase at a given temperature during the equilibration process. The time it took to reach this equilibrium was 6 h for the temperatures in the range of 80–170 ◦ C, 24 h at T = 180 ◦ C, 48 h at T = 190 ◦ C, and 12 h for the temperatures in the range of 200–220 ◦ C. Since we used a single-beam spectrometer in our experiment, for each measurement in order to exclude the influence of the atmosphere components (in particular, water vapor) on the sample spectra, we continuously pumped the dry air through the sample spectrometer chamber where the optical cell was positioned. For each temperature, we recorded the spectra of the empty cell (the silicon windows spectra) and the spectra of the cell filled only with CO2 . Finally, we recorded the spectra of the binary mixture where MA was placed in the bottom part of the cell (see Fig. 2b) that was filled with CO2 . All these spectra were measured with a resolution of 1 cm−1 . For each spectrum, 128 interferograms were recorded and averaged out to increase the signal-to-noise ratio. The resulting spectra of MA diluted in scCO2 were calculated by a direct subtraction of the spectra measured for the cell filled only with CO2 from the spectra of the binary mixture. The final spectra were corrected by baseline subtraction. Because of the increase in the MA solubility in the CO2 phase with temperature, we used two different optical path lengths. In the temperature range of 80–160 ◦ C when the solubility of MA in scCO2 is low, the optical path length was 1.6 mm, while in the temperature range of 150–220 ◦ C the path length was decreased to 0.137 mm. The values of these path lengths were opti-

mized to avoid the oversaturation of the analytical spectral bands of MA in scCO2 . 2.2.2. In situ Raman spectroscopy In order to follow up the changes occurring on the surface of the MA crystalline sample that was placed in the sample holder and was in contact with the scCO2 phase, we used in situ Raman spectroscopy. The MA sample was 14 mm in diameter and 3 mm thick (see the sectional view in Fig. 2d). Using a sapphire window also allowed us to visually observe the changes of the crystalline sample in the holder in real time in the temperature range of 80–190 ◦ C. A He-Ne laser (␭ =632.81 nm) with an output power of about 18 mW was connected to an optical fiber directing the laser beam into the specific module “Power Head” [49]. This module was used to ensure remote collection of the Raman scattering with optimal sensitivity. The laser beam was focused on the sample by a lens with a focal length of 40 mm. The laser beam power at the focal point was about 10 mW. The Raman light was collected in the backscattering mode by the same lens and then was focused inside the module at the entrance of the collecting optical fiber connected to a visible Raman spectrometer LabRAM HR Evolution (produced by HORIBA Jobin Yvon, France). The spectra were registered in the wavenumber range of 50–4000 cm−1 . The diffraction grating of 600 grooves/mm was used. For each measurement we used different acquisition time and number of accumulations. These values were optimized to get spectra with a good signal-to-noise ratio. 2.2.3. Micro IR spectroscopy The micro IR reflectance spectra of the MA crystalline forms were recorded using a Vertex 70 spectrometer equipped with a liquid nitrogen cooled MCT detector. This spectrometer was coupled with a Hyperion 3000 FT-IR microscope (produced by Bruker Optik GmbH) equipped with a 15x magnifying objective and was operated in the reflectance mode using an aperture size of 160 x 160 ␮m. In order to obtain the spectra with a good signal-to-noise ratio, 256 interferograms were recorded and averaged out for each measurement. Then the resulting absorption spectra in the wavenumber range of 800–3600 cm−1 with a 1 cm−1 spectral resolution were derived from the recorded reflectance spectra by applying the Kramers–Kronig transform algorithm. 2.2.4. Micro Raman spectroscopy Raman spectra of the MA crystalline forms were obtained using the same visible Raman spectrometer (see the in situ Raman spectroscopy part), combined with a confocal microscope (10–100× magnifying objectives were used), in the back-scattering geometry in the spectral range of 50–3600 cm−1 and with an add-on Linkam THMS600, which allows studying the solid phase under heating and cooling in the temperature range from −196 ◦ C to 600 ◦ C with an accuracy of 0.1 ◦ C. A He-Ne laser (␭ =632.81 nm) with the energy of 15 mW at the sample was applied for excitation. The Raman signal was collected with a CCD–detector (1024 × 256 pixels) placed after a diffraction grating (1800 grooves/mm) giving the final spectral resolution of about 0.3 cm−1 . The spectra were accumulated in each single scan with the exposure time of 300 s per each orientation of the grating that assured the spectra with a good signal-to-noise ratio. 2.2.5. Quantum chemical calculations The electronic structure calculations of the MA molecule were performed using the GAUSSIAN 09 suite of programs [50]. The optimization of the geometry and calculations of vibration frequencies were performed using the DFT hybrid Becke three-parameter functional B3LYP [50,51] with 6-311++g(2d,p) basis sets. The conformational search was based on the analysis of the potential energy surface scans. The potential energy profiles were calculated

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 3. IR spectra of three samples of crystalline mefenamic acid commercial form. The area limited by a rectangle shows the N–H stretching vibration region and corresponds to the insert.

along the three dihedral angles, which determine the rotations of the carboxylic group and two aromatic fragments of the molecule, respectively. The preliminary scanning of the hyper-surface of potential energy was performed by semi empirical method PM3 and the obtained set of conformers was further optimized using the B3LYP functional with the 6-311++g(2d,p) basis set. The structure of each conformer corresponded to the local minimum of potential energy, which was confirmed by frequency calculations. 16 MA conformers have been identified, but only 4 most stable conformers (Ia, Ib, IIa, IIb) were further taken into account. These four conformers are determined by the values of two dihedral angles, which will be discussed further in the text. For these conformers the parameters of intramolecular hydrogen bonds were calculated along with the transition energy barriers between the four conformational states that were determined by scanning the potential energy surface. 3. Results and discussion 3.1. Analysis of initial crystalline form of mefenamic acid In order to characterize the MA initial form used in our experiment, we measured the IR spectra of three MA samples in KBr tablets. The analysis of these spectra presented in Fig. 3 in the wavenumber range corresponding to the N–H stretching vibrations region (3250–3400 cm−1 ) has revealed the presence of a small spectral contribution related to polymorph II. It means that the initial commercial form of MA used in our study is not pure polymorph I (named «PI»), and it contains a small amount of polymorph II (named «PII»). This is in good agreement with results of Ref. [15,16]. Moreover, a quantitative analysis of these three spectra in the N–H stretching vibrations region has shown that the average percentage of polymorph II is 13.6% that is commensurable with the value of 10% given in Ref. [15]. 3.2. Thermal conversion of crystalline form of mefenamic acid In this work, we have also followed the thermal polymorphic conversion of MA at different temperatures. To do this we continuously measured the Raman spectra of single crystals of MA that were placed in temperature controlled plate of Linkam system. These measurements were fulfilled in fast mode when acquisition time was 5 s and only one orientation of the grating was used. The wavenumber range of 1300–1360 cm−1 was used as analytical. We chose this spectral domain because it can be considered as indicator of different polymorphs. Indeed in the case of polymorph

5

I it contains one spectral contribution whereas for polymorph II two spectral contributions are observed (see Fig. 4).This figure also shows the differences between Raman spectra of a single crystal of the MA initial form (PI) and a single crystal of the obtained MA polymorph II (PII) in two spectral domains (50–1800 cm−1 and 2700–3600 cm−1 ). We have found that the transition of polymorph I into polymorph II begins at the temperature slightly above 160 ◦ C. However, the rate of this process is slow and depends on size of crystal. Though the polymorphic transition of small crystals occurs faster, nevertheless it may take the hours. Finally we have found that fast phase transition occurs in the temperature range of 180–190 ◦ C. In this temperature range irrespective of crystal size the transition between polymorph I and II takes minutes. This fact is in good agreement with the results of Ref. [15], where the authors showed that starting from the temperature of 150 ◦ C, the polymorphic transformation of MA from polymorph I into polymorph II is observed, but the dynamics of this process is extremely slow. Nevertheless, these authors did not succeed in achieving the full polymorphic conversion at this temperature even after 140 h, and the full polymorphic conversion was reached only at 160 ◦ C, after 35 h.

3.3. Thermal conversion of mefenamic acid crystalline form being in contact with scCO2 phase: in situ Raman spectroscopy We have also studied the thermal conversion of the MA crystalline form being in contact with a scCO2 phase under isochoric heating. We have measured a set of Raman spectra of the crystalline MA sample surface in the bottom part of the cell at different temperatures. Fig. 5a,b shows the temperature evolution of these spectra. Here we monitored the polymorphic conversion on the base of observation of spectral bands change in three wavenumber ranges (50–200 cm−1 , 1300–1360 cm−1 , and 3300–3400 cm−1 ). It has been found that in the temperature range 80–140 ◦ C the obtained spectra correspond to polymorph I. Further, at the temperature of 160 ◦ C, the spectral signature reveals the transition of polymorph I to polymorph II. However, as in case of the crystalline MA transition, without the scCO2 phase, the rate of this process is slow. Finally, full polymorphic transition has been found at the temperature of 190 ◦ C. Moreover, the visual inspection of the crystalline MA surface being in contact with a scCO2 phase also allowed detecting the beginning of the polymorphic transition at 160 ◦ C. The emergence of dark domains on a bright background (see Fig. Sup. 3) may indicate the beginning of this transition. Then, the full polymorphic transition on the surface of the sample has been found at the temperature of 170 ◦ C. Nevertheless, this visual analysis has not allowed us to state the fact of a full polymorphic transition in the sample bulk and gives only a general idea about the temperature range of this polymorphic transition. It has also been discovered that after 30 h when the temperature was maintained at 190 ◦ C, the crystalline sample was melted and became transparent (see the last photo in Fig. Sup. 3). As a result, strong fluorescence appears and gives contribution into the Raman spectrum (see Fig. 5c). This fluorescence occurs because of the interaction of the laser beam with the metal alloy of the sample holder. According to various literature data (see, e.g., Ref. [23,34,52]) the MA melting point of polymorph II is expected in the temperature range of 230–233 ◦ C, however, in the case of a binary system of [MA–scCO2 ] under isochoric heating conditions, the MA melting point is reached at 190 ◦ C under pressure of 459 bar. Indeed, this is in good agreement both with our findings, namely for a binary system of [Ibuprofen–scCO2 ] [37] and with the literature data [53–55],

6

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 4. Comparison of Raman spectra of single crystals of mefenamic acid polymorph I (PI) and polymorph II (PII) in two spectral domains of 50–1800 cm−1 and 2700–3600 cm−1 . The area limited by a rectangle corresponds to the spectral band used as an indicator of thermal transformation of polymorph I to polymorph II during measurements (see text).

Fig. 5. Thermal evolution of Raman spectra in wavenumber ranges of 50–1800 cm−1 (a) and 2800–3600 cm−1 (b) of crystalline mefenamic acid being in contact with scCO2 phase under isochoric heating. Comparison of Raman spectra of crystalline mefenamic acid and its melted phase (c).

where the authors noted the depression of melting temperatures of organic solids by high-pressure CO2 . 3.4. Choice of analytical spectral domains In order to analyze the experimental IR spectra, two spectral domains were chosen as analytical. These domains are located in the wavenumber ranges of 1400–1550 cm−1 and 3250–3500 cm−1 (see Fig. 6). The first spectral domain is associated with complex vibrations that include the vibrations of the MA molecule aromatic system and rocking vibration of the N–H group; the second wavenumber range is associated with the stretching vibration of

the N–H group. This choice is linked with the fact that changing the MA conformations induces a direct change in the N–H vibrations because the amino-group can be directly involved in the formation of intramolecular hydrogen bonds with hydroxyl or carbonyl oxygen of carboxyl fragment depending on its orientation. Thus, the change in the parameters of these bonds as a consequence of conformational transition will directly affect the vibration parameters of the N–H group itself. On the other hand, conformational transition affects the vibration of atoms of aromatic fragments indirectly, namely by means of electronic density redistribution, because of their close positions to the MA functional groups (in particular, C O, N–H, O–H) that are sensitive to the conformation

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

7

Fig. 6. Analytical spectral domains in the wavenumber ranges of 1400–1550 cm−1 and 3250–3500 cm−1 of mefenamic acid IR spectrum measured at 170 ◦ C in scCO2 phase as an example.

Fig. 7. Schematic representation of four most stable conformers of mefenamic acid obtained by quantum chemical calculations.

changes. Therefore, the emergence of a new spectral contribution in this spectral domain that is induced, for instance, by the change in the thermodynamic conditions, is a signature of the change in the MA conformation. In our recent works [35,37], by analyzing paracetamol and ibuprofen as an example, we have also shown the advantage of using the spectral domain of 1400–1550 cm−1 associated with the aromatic ring for screening the conformational transitions of pharmaceutical molecules diluted in scCO2 . Another reason to choose the N–H stretching vibration mode as analytical is associated with differences in its behavior in the various polymorphs of MA. Indeed, the spectral domain of 3300–3350 cm−1 was successfully used for screening the polymorphic transition of crystalline MA as has been shown above in the text and in a number of works (see e.g., [15,16]). The authors of these works have shown that the appreciable change in the N–H stretching band position that occurs in the range of 3300–3350 cm−1 allows distinguishing between polymorphs I and II. As a result of the transition from polymorph I to polymorph II, the position of this band shifts from 3312 cm−1 to a higher frequency of 3353 cm−1 . Such behavior is consistent with the crystallographic results. For polymorph I, the proton of the N–H group is involved in a relatively strong interaction with the carbonyl oxygen (the intramolecular hydrogen bond). (see Fig. 1b) Indeed, in this case the N· · ·O distance is 2.636(2) Å [8]. However, transition to polymorph II, because of the changes in the molecular geometry (see Fig. 1 c,d), changes this distance too. Thus, for two different configurations of polymorph II,

the N· · ·O distance increases up to 2.67(1) Å and 2.72(2) Å, respectively [9] (these distances were measured in the MERCURY program [20]). Therefore, it should be expected that the N–H bond length will also decrease upon this polymorphic transition. All these changes indicate the change in the parameters of the intramolecular hydrogen bond as a consequence of the polymorphic transformation, namely the weakening of the interaction between the proton of the amino-group and the carbonyl oxygen. This shifts the amino-group vibration toward higher energies (the blue shift of the N–H stretching band). Furthermore, the IR and Raman spectroscopic studies and computational studies of the MA polymorphic forms point out to the noticeable difference in the position of the N–H stretching spectral band (see, e.g., Ref. [34,40]).

3.5. Assignment of spectral contributions In order to assign the spectral contributions of the analytical spectral domains presented in Fig. 6 to the different MA conformers, we relied on the results of our quantum chemical calculations. Thus, four most stable conformers were taken into account (see above). These conformers, named Ia, Ib, IIa, IIb are schematically presented in Fig. 7. They can be realized through the variation of two dihedral angles of ␶1 and ␶2 (see Fig. 8). The values of these angles for the given conformers are presented in Table 1. For these conformers it has been found (see Fig. 9) that the energy barriers quantifying the conformational transition (that

8

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 8. Schematic representation of the mefenamic acid molecule with atoms forming the dihedral angles of ␶1 (C4-C3-C10-O11) and ␶2 (C4-N15-C17-C19). These atoms are highlighted with the green color.

Fig. 9. Energy barriers of conformational transitions within pairs Ia→IIa→Ia and Ib→IIb→Ib that are related to the variation of dihedral angle ␶1 and within pairs Ib→Ia and IIb→IIa that are related to the variation of dihedral angle ␶2 . Table 1 Values of the dihedral angles of ␶1 and ␶2 , N–H bond length in the molecule aminogroup, and potential energy for 4 most stable conformers obtained by quantumchemical calculations. Parameter ␶1 ␶2 N–H Conformer energy

conformer Ia conformer Ib conformer IIa ◦

3.7 134.5◦ 1.017Å 0 kJ·mol−1

−2.0 76.5 1.017 Å 1.46 kJ·mol−1

180.0 135.1 1.015 Å 16.07 kJ·mol−1

conformer IIb 177.0 77.0 1.010 Å 16.65 kJ·mol−1

are related to the variation of dihedral angle ␶1 ) Ia→IIa and Ib→IIb are high (>33 kJ/mol). This definitively excludes the possibility of a spontaneous thermal transition between the conformers within these pairs even at the highest investigated temperature (RT493K = 4.1 kJ/mol). However, as shown in Fig. 9, the energy barriers of the conformational transitions Ia→Ib and IIa→IIb that are related to the variation of dihedral angle ␶2 do not exceed 1.5 kJ/mol. Thus, because of the fact that these barriers are lower than the thermal fluctuations even at room temperature (RT298K = 2.5 kJ/mol), the formation of each conformer within these pairs is equiprobable over the whole investigated temperature range. Consequently, these conformers may be grouped into two pairs, namely Ia–Ib and IIa–IIb; and each of these pairs may be considered as a single conformer. These two conformers were named as conformer I and conformer II, respectively.

In order to analyze the changes in the experimental spectra and to get information on the conformational state of the MA molecules in the scCO2 phase, we have also calculated the IR spectra for these conformers within our quantum chemical calculations. These spectra are presented in Fig. 10a. It is obvious from these spectra that the N–H band positions for conformers IIa and IIb are blue-shifted relative to those calculated for conformers Ia and Ib. Therefore, taking into account this difference in the N–H band positions, which correlates with its behavior during the transition from polymorph I to polymorph II (see, e.g., Ref. [15,16]), the difference in the potential energies for these conformers as well as the difference in the N–H bond lengths for these conformers (see Table 1), we have made a hypothesis that the conformational pair of Ia–Ib may be related to polymorph I, while that of IIa–IIb may be related to polymorph II. However, it is obvious that the MA molecules geometry for conformers IIa and IIb in vacuum differs from that of the conformers in polymorph II. Indeed, according to the analysis of the dihedral angle values presented in Table 1 for MA conformers IIa and IIb, the carbonyl oxygen is directed to the opposite side of the amino-group proton, whereas in the crystalline forms of polymorph II the carbonyl oxygen is directed toward the amino-group proton as in polymorph I (see, e.g., Ref. [9]). These discrepancies may be explained by the constrained geometry of the molecules in crystal, where it is defined by intermolecular bonds. However, the molecule transition from crystal to solution, where at low concentrations the molecule

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

9

Fig. 10. (a) IR spectra for 4 most stable conformers Ia, Ib, IIa, IIb obtained by quantum chemical calculations. (b,c) Sums of the IR spectra for pairs of conformers Ia and Ib (Conformer Ia + Ib) and conformers IIa and IIb (Conformer IIa + IIb).

exists as a monomer, is accompanied by the relaxation of the above mentioned constrained geometry, which leads the molecule to the closest local energy minimum. Because according to our statement that each of the pairs of Ia–Ib and IIa–IIb may be considered as a single conformer, we summarized the spectra of conformers within these pairs. These summarized spectra, named Ia + Ib and IIa + IIb are also presented in Fig. 10b for the wavenumber range of 1390–1610 cm−1 and in Fig. 10c for the wavenumber range of 3200–4000 cm−1 . The spectra shown in Fig. 10c demonstrate an appreciable shift in the N–H stretching band from 3480 cm−1 to a higher frequency of 3600 cm−1 at the transition from conformer I to conformer II. This change is consistent with the N–H stretching band position shift during the MA crystalline polymorphic transformation. Additionally, in the wavenumber range associated with the vibrations of the aromatic system we may also clearly distinguish different conformers, each of which gives two separated spectral contributions to the summarized spectrum (see Fig. 10b). 3.6. Analysis of mefenamic acid conformers in CO2 phase: in situ IR spectroscopy Fig. 11a,b demonstrates the evolution of experimental spectra (the two analytical spectral domains) of MA in the scCO2 phase with

a temperature increase in the range of 80–160 ◦ C. These spectra correspond to the equilibrium state at each temperature. The spectral bands intensity increase is related to the increase in the MA concentration in the scCO2 phase with the temperature increase. However, the spectral band shape is not affected. The analysis of the experimental spectra in the wavenumber ranges of 1400–1550 cm−1 and 3250–3500 cm−1 shows that in the first spectral domain two spectral contributions (at 1456 and 1515 cm−1 ) and in the second spectral domain the spectral contribution at 3346 cm−1 are predominant. According to the calculated spectra (see Fig. 10b,c) we can assign these contributions to MA conformer I. Nevertheless, even at these temperatures the experimental spectra contain very small contributions at 1475 and 3436 cm−1 , which we can attribute to MA conformer II. In our previous works [35,37], by analyzing pharmaceutical substances possessing different types of polymorphism as an example, we showed that a polymorphic transformation of crystalline form being in contact with a fluid solution (scCO2 phase) leads to a conformational transition of the molecules of these pharmaceutical substances in the fluid phase. Moreover, in the equilibrium conditions between the solid and fluid phases, the molar ratios of the polymorphs in the solid phase are correlated with the molar ratios of the conformers in the fluid solution phase. Therefore in the case of interface between the bottom phase containing the crys-

10

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 11. IR spectra of mefenamic acid diluted in scCO2 phase measured in the temperature range of 80–160 ◦ C: (a,b) in two analytical spectral domains 1400–1550 cm−1 and 3200–3500 cm−1 , respectively (the spectra were measured at the sample optical path length of 1.6 mm); and measured in the temperature range of 150–190 ◦ C: (c,d) in the same wavenumber ranges (here the sample optical path length was of 0.137 mm.

talline initial form of MA and the CO2 phase containing molecules of diluted MA, the presence of MA conformer II in the scCO2 phase is an indicator of the presence of polymorph II in the initial commercial crystalline MA as was discussed in part 3.1. This is also in good agreement with results of Ref. [56] where authors on the basis of results of NMR spectroscopy have shown that in saturated solution of commercial MA in DMSO being in equilibrium with solid phase of MA, mainly two stable conformers exist. Moreover the molar fractions of these conformers correspond to molar fractions of MA polymorphs in its solid phase, where the fraction of the first polymorph is prevailing. The analysis of the spectra presented in Fig. 11a,b also shows that the intensities associated with conformers I and II increases, while their ratio remains constant. It means that in this temperature range the percentages of these conformers do not change. Thus, we may assume that in this temperature range the polymorphic transformation from polymorph I into II in the MA crystalline phase does not occur. Fig. 11c,d shows the spectra of MA diluted in the scCO2 phase for the two analytical spectral domains in the temperature range of 150–190 ◦ C. For each temperature in the range of 170–190 ◦ C both spectra that were measured immediately after reaching the targeted temperature and those after 6 h of equilibration are presented. Examining the spectra that are presented in Fig. 11c,d, we can observe that starting from the temperature of 170 ◦ C the spectral contributions, which are attributed to MA conformer II and located at 1475, 1500 and 3436 cm−1 , increase appreciably. It has also been found that the rate of increasing these contributions is higher than the rate of increasing contributions that are related to conformer I (at 1456, 1515 and 3346 cm−1 ). We interpret this behavior as indication of the transition from conformer I to conformer II.

After that, we compared the intensity changes in these spectral contributions at the temperatures of 170 ◦ C, 180 ◦ C and 190 ◦ C over the first 6 h and found that with the temperature increase, the rate of transition of conformer I to conformer II also increase. Therefore, the highest rate of conformational transitions was found at 190 ◦ C. This behavior may be attributed to the beginning of fast polymorphic transformation of the MA crystalline phase in the bottom part of the cell. This is in good agreement with the results of Ref. [15], and our finding mentioned in parts 3.2 and 3.3 of this paper. In order to confirm this hypothesis, we examined in detail and compared the dynamics of the changes in the MA IR spectra in the phase of scCO2 at two temperatures (180 and 190 ◦ C). For the temperature of 180 ◦ C, the spectra were measured every 30 min over 24 h and for the temperature of 190 ◦ C, the spectra were also measured every 30 min over 48 h. The results of these measurements are summarized in Figs. 12 and 13 where the dynamics of the MA spectral change in the scCO2 phase as a function of time at the temperature of 180 ◦ C and 190 ◦ C are presented, respectively. It is obvious from Fig. 12 that the MA concentration in scCO2 increases. The analysis of these spectra shows that the intensities of the spectral bands located at 1475, 1500 and 3436 cm−1 attributed to MA conformer II increase faster than those located at 1456, 1515 and 3346 cm−1 attributed to conformer I. Such behavior testifies the increase in the percentage of conformer II. Nevertheless, even after 24 h, the spectral bands that were attributed to conformer I were still observed in the spectrum. It means that over 24 h the full transition of conformer I to conformer II is not achieved, though we might expect the full transition because according to our own finding (see above) polymorph conversion at this temperature should take less time. Fig. 13 shows the dynamics of MA spectral changes in the scCO2 phase as a function of time at the temperature of 190 ◦ C.

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

11

Fig. 12. Dynamics of spectral changes in spectral domains of 1400–1550 cm−1 and 3200–3500 cm−1 of IR spectra of mefenamic acid in scCO2 phase kept at the temperature of 180 ◦ C for 24 h.

It has been found that in the first 19 h, the spectral contributions intensities that are related to MA conformer II (at 1475, 1500 and 3436 cm−1 ) are higher than the rise of intensities of the spectral contributions that are attributed to conformer I (1456, 1515 and 3346 cm−1 ). This leads to an increase in the molar fraction of conformer II. In the time range of 19–21.5 h the intensity of the spectral band that is related to the N–H stretching vibration of conformer I reaches its maximum. Further, in the time range of 22–32 h all the spectral components that are related to this conformer considerably decrease, while the intensities of the spectral bands associated with conformer II continue rising. As a result, after 32 h it becomes impossible to detect the spectral signatures of conformer I. Therefore, we may hypothesize that at 190 ◦ C the full transition of conformers I to conformers II occurs. However, the spectra at 190 ◦ C behave in an interesting way in the time interval of 35–48 h. Namely, in both analytical spectral domains, we observe an avalanche-like increase in the spectral bands integral intensities. Indeed, in the wavenumber range of 1400–1550 cm−1 the spectral bands become wider and reach their oversaturation after 44.5 h. Though we might expect an exponentially-damped increase in their integral intensities with time at a constant temperature (due to the limit of the MA solubility in the scCO2 phase spectral intensity asymptotically reaches maximum value within certain time and remains constant), on the contrary we observe acceleration of the integral intensities increase. The same scheme of changes is also observed in the N–H vibration mode, where the spectral band that is associated with conformer II also broadens, and it is accompanied by simultaneous appearance of a pronounced contribution on its low-frequency slope (3380–3400 cm−1 ), whereas its integral intensity increases exponentially. This unusual change is correlated with the occurrence of critical phenomena that induce the change in the extinction coefficients of these spectral bands. Indeed, as it was shown before, the melting of MA polymorph II occurs at 190 ◦ C and this is the first-order phase transition [57] that may be traced back to the appearance of critical opalescence in the scCO2 fluid phase being in contact with the MA solid phase. This is also confirmed by the behavior of the raw spectra (without baseline correction) of MA in the scCO2 phase that demonstrate an increase in the spectral intensity accompanied by spontaneous growth in the spectral background at 190 ◦ C (see Fig. Sup. 1). Such phenomenon is due to the increasing of continuous absorption induced by a large density fluctuation of the MA molecules in the sc CO2 phase. It is precisely this fact that explains the occurrence of nonlinearity in the behavior of the extinction coefficient. Of course, it results in the non-equilibrium state of the system that is characterized by the presence of transient MA molecular clusters. This may also be confirmed by the considerable change in the

absorption of the solvent itself. At the same time, we observe a considerable decrease in the integral intensities of the CO2 combination spectral bands (2␯2 ␯3 and ␯1 ␯3 ), which are both red shifted and narrowed (see Fig. Sup. 2a). These critical phenomena are expected to disappear when the temperature is increased further (above 190 ◦ C). Indeed, the subsequent heating to 200 ◦ C and further to 210 ◦ C and 220 ◦ C leads again to an appreciable decrease in the integral intensities and narrowing of the spectral contributions in the wavenumber range of 1400–1550 cm−1 (see Fig. 14a). The spectral band associated with the N–H stretching vibration of conformer II loses its low-frequency shoulder and its total integral intensity also decreases considerably. Nevertheless, these changes do not result in the reappearance of an N–H stretching band that is related to conformer I. Finally, the spectra of MA and CO2 (combination bands) again become similar to the spectra recorded at 190 ◦ C in the time range before the avalanche-like increase in the spectral intensity occurs (see Fig. 14b and Fig. Sup. 2b). Furthermore, the similarity between the MA spectra at the temperatures of 210–220 ◦ C and those at 190 ◦ C that were measured in the time interval from 30 to 35 h allows us to state that even at such high temperatures no thermal decomposition of MA takes place. Taking into account the full transition of MA conformer I to conformer II at 190 ◦ C in scCO2 phase being in contact with the MA crystalline phase, we may suppose that, in the bottom phase, the total conversion of polymorph I into polymorph II also takes place. Therefore, it might be expected that after a fast cooling down of the cell, pure polymorph II can be obtained. To confirm this hypothesis we analyzed the recrystallized MA obtained from the binary mixture of [MA–scCO2 ]. 3.7. Analysis of crystalline MA obtained by crystallization from binary mixture of [MA–scCO2 ] We have analyzed two processed MA samples obtained after cooling down the optical cell and the subsequent evacuation of CO2 . The first sample was extracted from the bottom part of the cell that contained an excess of MA, whereas the second one was collected from the upper part of the cell, which corresponded to the phase of scCO2 (see the sectional view in Fig. 2b). This sample was obtained as a deposit on the cell walls as a result of sedimentation from a fluid solution upon its fast cooling down. The main difference between these two samples was in their particles sizes. Thus, the sample obtained from the scCO2 phase was microcrystalline. Then, we recorded full range micro Raman and micro IR spectra of these two samples and compared them with the spectra of MA of PII form (see Fig. 15). It has been found that for these samples, both the IR and Raman spectra totally correspond to the spectra

12

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

Fig. 13. Dynamics of spectral changes in spectral domains of 1400–1550 cm−1 and 3200–3500 cm−1 of IR spectra of mefenamic acid in scCO2 phase kept at the temperature of 190 ◦ C for 48 h. Different colors show the certain key time ranges.

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

13

Fig. 14. IR spectra of mefenamic acid diluted in scCO2 phase in the temperature range of 190–220 ◦ C in two spectral domains: (a) 1000–1800 cm−1 and (b) 2650–3550 cm−1 . The spectra were measured at the sample optical path length of 0.137 mm.

Fig. 15. Comparison of Raman and IR spectra of the mefenamic acid samples obtained after processing in an experimental cell (the first sample was extracted from the bottom part of the cell that always contained an excess of MA, the second sample was collected from the upper part of the cell, where it was obtained as a deposit on the cell walls as a result of sedimentation from a fluid solution upon its fast cooling down) with spectra of mefenamic acid polymorph II (PII) obtained via thermal conversion of polymorph I using the Linkam system with a temperature controlled plate.

of pure polymorph II. Indeed, the main signature of polymorph II is located in the N–H stretching region, whereas the signature of polymorph I in this region is totally absent. At the same time, there are also a number of spectral peculiarities in the wavenumber ranges of 50–1800 cm−1 for Raman spectra and 800–1800 cm−1 for IR spectra, which clearly prove that the MA samples obtained after processing in the experimental cell are pure polymorph II. Therefore, we may assert that, on the one hand, there are correlations between the conformational state of MA molecules in the scCO2 phase and the polymorphism of the crystalline phase being in contact with each other. On the other hand, we can also expect the correlations between the polymorphism of the MA form obtained by crystallization from the fluid phase and the conforma-

tional state of its molecules in the scCO2 phase, from which this form was crystallized. 4. Conclusions In this work we have studied the conformational equilibrium of mefenamic acid molecules in its saturated solution in supercritical carbon dioxide being in contact with bottom phase containing different polymorphs of mefenamic acid under isochoric heating conditions in the temperature range of 80–220 ◦ C. For this study we applied a combination of experimental (in situ IR and in situ Raman as well as micro IR and micro Raman spectroscopy) and computational (quantum chemical calculations) methods. We have found

14

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547

that in the case of solid-fluid interface there is a correlation between the polymorphic modifications of the crystalline form of mefenamic acid and its conformers in CO2 fluid phase. This phenomenon occurs as consequence of equality of chemical potentials of mefenamic acid polymorph in solid phase and mefenamic acid conformer in CO2 fluid phase, so that the transformation of polymorph leads to change of chemical potential of solid phase which in turn should result in the change of chemical potential of another phase as follows from conformational transition. The similar results were also shown in our previous works [35,37]. Thus, we have shown that it is possible to monitor the polymorphic transformation of the MA crystalline phase by screening its molecular conformations in the fluid solution. However, the question regarding the mechanism of MA molecules relaxation in CO2 phase at their transition from crystal to solution remains unanswered and requires additional study. Acknowledgements This work was supported by the Russian Foundation for Basic Research (grants No. 17-03-00459 and 18-29-06008); by the French Embassy in Russia (Metchnikov program 2018, file No. 915125G); by Ministry of Education and Science of the Russian Federation (contract No. 01201260481). The IR spectroscopy experiment was performed using the molecular fluid spectroscopy facility (http://www.ckp-rf.ru/usu/ 503933/) of G.A. Krestov Institute of Solution Chemistry of the Russian Academy of Sciences (ISC RAS) (Russia), Raman experiment was performed in the Laboratory of IR and Raman Spectral Chemistry of the University Lille 1 (LASIR) (France). Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.supflu.2019. 104547. References [1] CPMP/EWP/QWP/1401/98 Rev. 1/ Corr – Guideline on the Investigation of Bioequivalence, European Medical Agency, 2010, Retrieved from http://www. ema.europa.eu/docs/en GB/document library/Scientific guideline/2010/01/ WC500070039.pdf. [2] General Notes on Biopharmaceutics Classification System (BCS)-Based Biowaiver Applications, World Health Organization, 2014, Retrieved from http://apps.who.int/medicinedocs/documents/s21674en/s21674en.pdf. [3] K.T. Savjani, A.K. Gajjar, J.K. Savjani, Drug solubility: importance and enhancement techniques, ISRN Pharm. 2012 (2012) 1–10. [4] K. Prakash, R. Jieun, K. Hyeongmin, K. Iksoo, K. Jeong Tae, K. Hyunil, C. Jae Min, Y. Gyiae, L. Jaehwi, Pharmaceutical particle technologies: an approach to improve drug solubility, dissolution and bioavailability, Asian J. Pharm. Sci. 9 (2014) 304–316. [5] A. Sharma, R. Mishra, P. Tandon, Polymorphism in Pharmaceutical Compounds, Advancements and Futuristic Trends in Material Science, 2011 https://www.academia.edu/24152922/Polymorphism in Pharmaceutical Compounds. [6] M. Pudipeddi, A.T.M. Serajuddin, Trends in solubility of polymorphs, J. Pharm. Sci. 94 (2005) 929–939. [7] A.J. Aguiar, J.E. Zelmer, Dissolution behavior of polymorphs of chloramphenicol palmitate and mefenamic acid, J. Pharm. Sci. 58 (1969) 983–987. [8] J. McConnell, F. Company, N-(2, 3-Xylyl) anthranilic acid, C15H15NO2, mefenamic acid, Cryst. Struct. Commun. 5 (1976) 861–864. [9] S. SeethaLekshmi, T.N. Guru Row, Conformational polymorphism in a non-steroidal anti-inflammatory drug, mefenamic acid, Cryst. Growth Des. 12 (2012) 4283–4289. [10] S. Romero, B. Escalera, P. Bustamante, Solubility behavior of polymorphs I and II of mefenamic acid in solvent mixtures, Int. J. Pharm. 178 (1999) 193–202. [11] D. Shinkuma, T. Hamaguchi, Y. Yamanaka, N. Mizuno, Correlation between dissolution rate and bioavailability of different commercial mefenamic acid capsules, Int. J. Pharm. 21 (1984) 187–200. [12] S.K. Abdul Mudalip, M.R. Abu Bakar, P. Jamal, F. Adam, Solubility and dissolution thermodynamic data of mefenamic acid crystals in different classes of organic solvents, J. Chem. Eng. Data 58 (2013) 3447–3452.

[13] C.E. Nicholson, S.J. Cooper, Crystallization of mefenamic acid from dimethylformamide microemulsions: obtaining thermodynamic control through 3D nanoconfinement, Crystals 1 (2011) 195–205. [14] R.K. Gilpin, W. Zhou, Infrared studies of the polymorphic states of the fenamates, J. Pharm. Biomed. Anal. 37 (2005) 509–515. [15] R.K. Gilpin, W. Zhou, Infrared studies of the thermal conversion of mefenamic acid between polymorphic states, Vib. Spectrosc. 37 (2005) 53–59. [16] N. Abbas, I.D.H. Oswald, C.R. Pulham, Accessing mefenamic acid form II through high-pressure recrystallisation, Pharmaceutics 9 (2017) 16–26. [17] A. Adam, L. Schrimpl, P.C. Schmidt, Some physicochemical properties of mefenamic acid, Drug Dev. Ind. Pharm. 26 (2000) 477–487. [18] N.M. Rawashdeh, N.M. Najib, I.M. Jalal, Comparative bioavailability of two capsule formulations of mefenamic acid, Int. J. Clin. Pharmacol. Ther. 35 (1997) 329–333. [19] T. Bapurao, Y. Ashokraj, R. Panchagnula, P. Pillai, Mefenamic acid: new polymorph or crystal defect? Pharm. Technol. Europe 18 (2006) http://www. pharmtech.com/mefenamic-acid-new-polymorph-or-crystal-defect. [20] C.F. Macrae, P.R. Edgington, P. McCabe, E. Pidcock, G.P. Shields, R. Taylor, M. Towler, J. van de Streek, Mercury: visualization and analysis of crystal structures, J. Appl. Crystallogr. 39 (2006) 453–457. [21] R. Panchagnula, P. Sundaramurthy, O. Pillai, S. Agrawal, Y.A. Raj, Solid-state characterization of mefenamic acid, J. Pharm. Sci. 93 (2004) 1019–1029. [22] A.M. Siti Kholijah, A.B. Mohd Rushdi, J. Parveen, A. Fatmawati, C.M. Rohaida, S. Siti Zubaidah, M.A. Zatul Iffah, M.S. Shalyda, Effects of solvents on polymorphism and shape of mefenamic acid crystals, Matec Web Conf. 150 (2018) 02004. [23] S. Cesur, S. Gokbel, Crystallization of mefenamic acid and polymorphs, Cryst. Res. Technol. 43 (2008) 720–728. [24] K. Bodnár, S.P. Hudson, Å.C. Rasmuson, Stepwise use of additives for improved control over formation and stability of mefenamic acid nanocrystals produced by antisolvent precipitation, Cryst. Growth Des. 17 (2017) 454–466. [25] M. Otsuka, F. Kato, Y. Matsuda, Effect of temperature and kneading solution on polymorphic transformation of mefenamic acid during granulation, Solid State Ion 172 (2004) 451–453. [26] F. Kato, M. Otsuka, Y. Matsuda, Kinetic study of the transformation of mefenamic acid polymorphs in various solvents and under high humidity conditions, Int. J. Pharm. 321 (2006) 18–26. [27] M. Dixit, A.G. Kini, P. Kulkarni, Enhancing the dissolution of polymorphs I and II of mefenamic acid by spray drying, Turk. J. Pharm. Sci. 9 (2012) 13–26. [28] M. Tuerk, D. Bolten, Polymorphic properties of micronized mefenamic acid, nabumetone, paracetamol and tolbutamide produced by rapid expansion of supercritical solutions (RESS), J. Supercrit. Fluids 116 (2016) 239–250. [29] A.Z. Hezave, F. Esmaeilzadeh, Micronization of drug particles via RESS process, J. Supercrit. Fluids 52 (2010) 84–98. [30] L.X. Yu, M.S. Furness, A. Raw, K.P.W. Outlaw, N.E. Nashed, E. Ramos, S.P.F. Miller, R.C. Adams, F. Fang, R.M. Patel, F.O. Holcombe, Y.-y. Chiu, A.S. Hussain, Scientific considerations of pharmaceutical solid polymorphism in abbreviated new drug applications, Pharm. Res. 20 (2003) 531–536. [31] M. Antonio, R.M. Maggio, Assessment of mefenamic acid polymorphs in commercial tablets using chemometric coupled to MIR and NIR spectroscopies. Prediction of dissolution performance, J. Pharm. Biomed. Anal. 149 (2018) 603–611. [32] N. Tumanova, N. Tumanov, K. Robeyns, F. Fischer, L. Fusaro, F. Morelle, V. Ban, G. Hautier, Y. Filinchuk, J. Wouters, T. Leyssens, F. Emmerling, Opening Pandora’s Box: chirality, polymorphism, and stoichiometric diversity in Flurbiprofen/Proline cocrystals, Cryst. Growth Des. 18 (2018) 954–961. [33] H. Kulla, S. Greiser, S. Benemann, K. Rademann, F. Emmerling, In situ investigation of a self-accelerated cocrystal formation by grinding pyrazinamide with oxalic acid, Molecules 21 (2016) 917–925. [34] V.R.R. Cunha, C.M.S. Izumi, P.A.D. Petersen, A. Magalhães, M.L.A. Temperini, H.M. Petrilli, V.R.L. Constantino, Mefenamic acid anti-inflammatory drug: probing its polymorphs by vibrational (IR and Raman) and solid-state NMR spectroscopies, J. Phys. Chem. B 118 (2014) 4333–4344. [35] R.D. Oparin, M. Moreau, I. De Walle, M. Paolantoni, A. Idrissi, M.G. Kiselev, The interplay between the paracetamol polymorphism and its molecular structures dissolved in supercritical CO2 in contact with the solid phase: In situ vibration spectroscopy and molecular dynamics simulation analysis, Eur. J. Pharm. Sci. 77 (2015) 48–59. [36] R.D. Oparin, E.A. Vorobyev, M.G. Kiselev, A new method for measuring the solubility of slightly soluble substances in supercritical carbon dioxide, Russ. J. Phys. Chem. B 10 (2016) 1108–1115. [37] R.D. Oparin, D.V. Ivlev, A.M. Vorobei, A. Idrissi, M.G. Kiselev, Screening of conformational polymorphism of ibuprofen in supercritical CO2, J. Mol. Liq. 239 (2017) 49–60. [38] J.D. Dunitz, J. Bernstein, Disappearing polymorphs, Acc. Chem. Res. 28 (1995) 193–200. [39] B. Dejan-Kreˇsimir, R.W. Lancaster, B. Joel, Disappearing polymorphs revisited, Angew. Chemie Int. Ed. 54 (2015) 6972–6993. [40] S. Jabeen, T.J. Dines, S.A. Leharne, B.Z. Chowdhry, Raman and IR spectroscopic studies of fenamates – conformational differences in polymorphs of flufenamic acid, mefenamic acid and tolfenamic acid, Spectrochim. Acta A. Mol. Biomol. Spectrosc. 96 (2012) 972–985. [41] E.H. Lee, S.R. Byrn, M.T. Carvajal, Additive-induced metastable single crystal of mefenamic acid, Pharm. Res. 23 (2006) 2375–2380.

R.D. Oparin et al. / J. of Supercritical Fluids 152 (2019) 104547 [42] X. Yang, B. Sarma, A.S. Myerson, Polymorph control of micro/nano-sized mefenamic acid crystals on patterned self-assembled monolayer islands, Cryst. Growth Des. 12 (2012) 5521–5528. [43] V. Dhanaraj, M. Vijayan, Structural studies of analgesics and their interactions. XII. Structure and interactions of anti-inflammatory fenamates. A concerted crystallographic and theoretical conformational study, Acta Crystallogr. Sect. B 44 (1988) 406–412. [44] J.J. Lozano, R. Pouplana, M. López, J. Ruiz, Conformational analysis of the antiinflammatory fenamates: a molecular mechanics and semiempirical molecular orbital study, J. Mol. Struct. Theochem 335 (1995) 215–227. [45] O.G. Uzoh, P.T.A. Galek, S.L. Price, Analysis of the conformational profiles of fenamates shows route towards novel, higher accuracy, force-fields for pharmaceuticals, J. Chem. Soc. Faraday Trans. 17 (2015) 7936–7948. [46] W. Du, A.J. Cruz-Cabeza, S. Woutersen, R.J. Davey, Q. Yin, Can the study of self-assembly in solution lead to a good model for the nucleation pathway? The case of tolfenamic acid, Chem. Sci. 6 (2015) 3515–3524. [47] R.R. Ramos, C.C. da Silva, F.F. Guimarães, F.T. Martins, Polymorphism and conformerism in chalcones, CrystEngComm 18 (2016) 2144–2154. [48] R.D. Oparin, A. Idrissi, M.V. Fedorov, M.G. Kiselev, Dynamic and static characteristics of drug dissolution in supercritical CO2 by infrared spectroscopy: measurements of acetaminophen solubility in a wide range of state parameters, J. Chem. Eng. Data 59 (2014) 3517–3523. [49] B. Grignard, B. Gilbert, C. Malherbe, C. Jérôme, C. Detrembleur, Online monitoring of heterogeneous polymerizations in supercritical carbon dioxide by Raman spectroscopy, ChemPhysChem 13 (2012) 2666–2670. [50] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr, J.E.

[51] [52]

[53]

[54]

[55] [56]

[57]

15

Peralta, F. Ogliaro, M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. Farkas, J.B. Foresman, J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian 09, Revision D.01, Gaussian, Inc., Wallingford CT, 2013. A.D. Becke, Density-functional exchange-energy approximation with correct asymptotic behavior, Phys. Rev. A 38 (1988) 3098–3100. A. Zeinolabedini Hezave, M.H. Khademi, F. Esmaeilzadeh, Measurement and modeling of mefenamic acid solubility in supercritical carbon dioxide, Fluid Phase Equilib. 313 (2012) 140–147. Y. Takebayashi, K. Sue, T. Furuya, Y. Hakuta, S. Yoda, Near-infrared spectroscopic solubility measurement for thermodynamic analysis of melting point depressions of biphenyl and naphthalene under high-pressure CO2, J. Supercrit. Fluids 86 (2014) 91–99. H. Uchida, M. Yoshida, Y. Kojima, Y. Yamazoe, M. Matsuoka, Measurement and correlation of the solid−liquid−gas equilibria for the carbon dioxide + S-(+)-ibuprofen and carbon dioxide + RS-(±)-ibuprofen systems, J. Chem. Eng. Data 50 (2005) 11–15. R.M. Lemert, K.P. Johnston, Solid-liquid-gas equilibria in multicomponent supercritical fluid systems, Fluid Phase Equilib. 45 (1989) 265–286. I.A. Khodov, K.V. Belov, S.V. Efimov, L.A.E.Bd. Carvalho, Determination of preferred conformations of mefenamic acid in DMSO by NMR spectroscopy and GIAO calculation, in: AIP Conference Proceedings, 2019, pp. 2063, 040007. J.W. Forbes, First-order polymorphic and melting phase transitions under shock loading, in: Shock Wave Compression of Condensed Matter. Shock Wave and High Pressure Phenomena, Springer, Berlin, Heidelberg, 2012, pp. 201–241.