- Email: [email protected]
Perfusion calorimetry in the characterization of solvates forming isomorphic desolvates
European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Contents lists available at ScienceDirect
European Journal of Pharmaceutical Sciences journal homepage: www.elsevier.com/locate/ejps
Perfusion calorimetry in the characterization of solvates forming isomorphic desolvates Julia Baronsky a,b, Martina Preu b, Michael Traeubel b, Nora Anne Urbanetz c,⇑ a
Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine-University, Universitaetsstrasse 1, D-40225 Duesseldorf, Germany BAH-RD-A, Bayer Animal Health GmbH, Alfred-Nobel-Strasse 50, D-40789 Monheim, Germany c Research Center Pharmaceutical Engineering GmbH, Inffeldgasse 21a/II, A-8010 Graz, Austria b
a r t i c l e
i n f o
Article history: Received 3 February 2011 Received in revised form 9 June 2011 Accepted 17 June 2011 Available online 25 June 2011 Keywords: Perfusion calorimetry Microcalorimetry Non-stoichiometric solvate Isomorphic desolvate
a b s t r a c t In this study, the potential of perfusion calorimetry in the characterization of solvates forming isomorphic desolvates was investigated. Perfusion calorimetry was used to expose different hydrates forming isomorphic desolvates (emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate) to stepwise increasing relative vapour pressures (RVP) of water and methanol, respectively, while measuring thermal activity. Furthermore, the suitability of perfusion calorimetry to distinguish the transformation of a desolvate into an isomorphic solvate from the adsorption of solvent molecules to crystal surfaces as well as from solvate formation that is accompanied by structural rearrangement was investigated. Changes in the samples were confirmed using FT-Raman and FT-IR spectroscopy. Perfusion calorimetry indicates the transformation of a desolvate into an isomorphic solvate by a substantial exothermic, peak-shaped heat flow curve at low RVP which reflects the rapid incorporation of solvent molecules by the desolvate to fill the structural voids in the lattice. In contrast, adsorption of solvent molecules to crystal surfaces is associated with distinctly smaller heat changes whereas solvate formation accompanied by structural changes is characterized by an elongated heat flow. Hence, perfusion calorimetry is a valuable tool in the characterization of solvates forming isomorphic desolvates which represents a new field of application for the method. Ó 2011 Published by Elsevier B.V.
1. Introduction Active pharmaceutical ingredients as well as excipients may exist in different solid forms which is of great importance for pharmaceutical development (Burger et al., 1999; Newman and Byrn, 2003; Othman et al., 2007). One important type of solid forms are solvates. A solvate may be defined as a crystalline multi-component system in which a solvent (or several solvents) is coordinated in or accommodated by the crystal structure (Griesser, 2006). In case the solvent is water, the solvate is called hydrate. The omnipresence of water makes hydrate formation an important issue in pharmaceutical industry (Giron et al., 2002; Redman-Furey et al., 2005). The formation of organic solvates plays a significant role also, since pharmaceutical substances are often exposed to organic solvents during production and processing, for example. Solvates may be divided into stoichiometric and non-stoichiometric solvates (Griesser, 2006). A stoichiometric solvate is a solvate
⇑ Corresponding author. Tel.: +43 316 873 9720; fax: +43 316 873 109720. E-mail address: [email protected] (N.A. Urbanetz). 0928-0987/$ - see front matter Ó 2011 Published by Elsevier B.V. doi:10.1016/j.ejps.2011.06.008
that, within the range of solvent activity its crystal structure exists in (at given pressure and temperature), has a fixed number of solvent molecules inside the crystal lattice, while, for a non-stoichiometric solvate, the number of solvent molecules inside the lattice is dependent on solvent activity. Non-stoichiometric solvates may desolvate retaining their crystal structures (Stephenson et al., 1998). The isostructural unsolvated form generated by desolvation is called isomorphic desolvate or desolvated solvate. Due to the fact that desolvation causes only slight changes to the crystal lattice, the identification of solvates forming isomorphic desolvates might not be straightforward (Griesser, 2006). Methods that have been shown to be suitable to identify this type of solvates are single crystal structure analysis (Ruth et al., 2003), X-ray powder diffraction (XRPD) (Stephenson et al., 1998) and spectroscopic techniques (Stephenson et al., 1997). XRPD detects any shrinking of the lattice which may occur upon desolvation (Rollinger and Burger, 2002). Spectroscopic techniques, such as solid-state nuclear magnetic resonance spectroscopy and Raman spectroscopy, are sensitive to the changes in the local environment of molecular groups inside the lattice induced by the departure of the solvent molecules (Miroshnyk et al., 2006; Ruth et al., 2003). However, as some of the methods might not be available or in
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
some cases, a method might fail to detect whether a solvate forming an isomorphic desolvate is present (Stephenson et al., 1998), further methods suitable to characterize this type of solvates are needed. The aim of this study is to elucidate the potential of perfusion calorimetry in the characterization of solvates converting into isomorphic desolvates. In perfusion calorimetry, a sample is exposed to a controlled relative vapour pressure (RVP) of solvent while measuring thermal activity (Gaisford, 2007). The method has been used to study solvent-solid interactions, e.g. the adsorption of solvent molecules to crystal surfaces and the sorption of solvent molecules by amorphous substances (Carvajal and Staniforth, 2006; Danforth and David, 2006; Timmermann et al., 2006). However, its potential in the investigation of solvate formation has been barely used (Lehto and Laine, 1998; Markova et al., 2001). In this study, different substances that are known to be hydrates forming isomorphic desolvates were investigated using perfusion calorimetry. The aim was to assess whether perfusion calorimetry is suitable to detect solvates changing into isomorphic desolvates which would represent a new field of application for the method. The substances studied are the non-stoichiometric hydrates II–IV of emodepside (veterinary drug substance; Baronsky et al., 2009b), erythromycin A dihydrate and spirapril hydrochloride monohydrate (Stephenson et al., 1998). The hydrates were first desolvated in situ inside the sample vessel using dry nitrogen. The isomorphic desolvates generated were then exposed to stepwise increasing RVP values of water (relative humidities, RH) to study the heat changes associated with hydrate formation. In addition, the isomorphic desolvates were exposed to stepwise increasing RVP values of methanol (RVPMeOH) to investigate whether methanol molecules enter the structural voids inside the crystal lattices previously occupied by water molecules. Furthermore, emodepside I (non-solvated crystalline form of emodepside; Baronsky et al., 2009b) and sulfaguanidine anhydrate were used as model substances to test the ability of perfusion calorimetry to distinguish isomorphic solvate formation from solvent adsorption to crystal surfaces and from solvate formation that is accompanied by structural rearrangement.
2. Materials and methods 2.1. Materials Emodepside I was purchased from KPV, Kiel, Germany. The hydrates II–IV of emodepside were prepared using methods described in literature (Baronsky et al., 2009b). The hydrates II and IV were formed by cooling a hot ethanol 96% (V/V) solution (hydrate II) and methanol solution (hydrate IV), respectively, at room temperature. Hydrate III was produced by slurry phase conversion of emodepside II hydrate at room temperature using acetone. Erythromycin A dihydrate, analytical standard, was obtained from Riedel de Haen, Seelze, Germany. The desolvate of erythromycin A dihydrate was obtained by storing the dihydrate over P205 for 4 days. Spirapril hydrochloride monohydrate CRS was purchased from EDQM (European Directorate for the Quality of Medicines). Regarding sulfaguanidine, the monohydrate was supplied by Fluka, Buchs, Switzerland. The anhydrate used for experiments was prepared as described in literature (Gift and Taylor, 2007). The organic solvents used in this study (for the preparation of the hydrates II–IV of emodepside as well as for perfusion experiments) were of analytical grade and purchased from Merck, Darmstadt, Germany. The water used in the perfusion experiments was Milli-Q water obtained by a Milli-Q Gradient System (Millipore, Schalbach, Germany).
75
2.2. Methods 2.2.1. Perfusion calorimetry Experiments were performed using a 2277 Thermal Activity Monitor (TA Instruments, New Castle, Delaware). To expose the sample to a controlled RVP of solvent during measurement, vapour pressure control devices based on two mass flow controllers were used. They are composed of the 2250010 4 ml RH Perfusion Ampoule, two mass flow controllers (Bronkhorst HighTech, AK Ruurlo, Netherlands) and the 3811 Flow Control Module (TA Instruments, New Castle, Delaware). The performances of the control devices were tested using the method described by Baronsky et al., 2009a. The results from the testing procedure were used to adapt RVP settings in order to create the desired RVP values inside the reaction vessel (RHin values and RVPin values of methanol, respectively). All experiments were carried out at 25 °C. The total flow rate was set to 80 ml h 1 (0 °C, 1.01325 bar). The amount of solvent (water or methanol) placed into the first and second solvent reservoir of the perfusion ampoule was 1 ml and 0.5 ml, respectively. The heater of the perfusion ampoule was set to 45 °C. The solid material under investigation was accurately weighted and transferred into the reaction vessel. The weight of sample ranged between 1 mg and 10 mg. The loaded reaction vessel was connected to the perfusion ampoule which was then equilibrated in the different equilibration positions for 20 min before starting the measurement. At the beginning of measurement, the sample was exposed to 0% RVP (drying period). After the heat flow curve of the sample had returned to the blank curve (attainment of equilibrium of the sample), the RVP inside the vessel (RHin and RVPin of methanol, respectively) was increased to 10%, 20%, 30%, 50%, 70% and 90%. A time delay of 4 h per step was sufficient for the solvates forming isomorphic desolvates to attain equilibrium at each step. Only in case of emodepside II subjected to methanol vapours, it was not sure whether the heat flow curve of the sample had returned to the blank curve at each step after 4 h. Thus, in this case, RVP steps were run for 6 h instead of 4 h. For each type of experiment, a blank curve was recorded by repeating the experiment using an empty sample vessel. Each experiment was performed having an empty, closed stainless steel ampoule in the reference position. Before starting a series of measurements, the calorimeter was calibrated using a static calibration. The calibration heat flow generated by the internal calibration heater resistors was 1000 lW. The calibration was performed having the perfusion ampoule inside the measuring position and the reference ampoule inside the reference position. During calibration, the solvent reservoirs and the reaction vessel of the perfusion ampoule were empty and the gas flow was turned off. 2.2.2. Spectroscopic analysis After having finished a perfusion experiment, the sample was subsequently analysed using both FT-Raman spectroscopy and FT-IR analysis. The sample was transferred into the sample holder (FT-Raman spectroscopy) and the ATR unit (FT-IR spectroscopy), respectively, as quickly as possible in order to minimize the time prior to measurement (time in which the sample was exposed to ambient conditions). 2.2.3. FT-Raman spectroscopy Raman spectra were acquired with a Bruker RFS 100S spectrometer (Bruker Optik GmbH, Ettlingen, Germany) using a Nd:YAG laser (1064 nm) as excitation source and a high sensitivity Gedetector (cooled with liquid nitrogen). The spectra were recorded from 3500 cm 1 to 20 cm 1 at a resolution of 1 cm 1 (128 scans) using a laser output of 750 mW.
76
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
2.2.4. FT-IR spectroscopy Infrared analysis was carried out with a FT-IR Bruker Tensor 37 spectrometer (Bruker Optik GmbH, Ettlingen, Germany) using a Heated Diamond Top-plate as sampling system (ATR method). Spectra were recorded from 4000 cm 1 to 550 cm 1 at a resolution of 2 cm 2 (64 scans). 2.2.5. Data treatment Some of the data were treated using the standard normal variate (SNV) method to correct multiplicative effects. Regarding FT-IR spectra, the entire measuring range was used for SNV. For FTRaman spectra, SNV was performed over the range of 3200– 570 cm 1 (emodepside samples), 3200–550 cm 1 (spirapril hydrochloride samples) and 3200–20 cm 1 (erythromycin A samples), respectively. 3. Results 3.1. Thermal activity at different RH Figs. 1 and 2 show the heat flow curves of the different samples exposed to increasing RH in comparison to the corresponding blank curves. At the beginning of the experiment (0% RH), the nonstoichiometric hydrates II–IV of emodepside, erythromycin A dihydrate and spirapril hydrochloride monohydrate show considerable endothermic heat flows corresponding to dehydration. For erythromycin A dihydrate, the shape of the endothermic heat flow curve differs from the shape of the curves of the other hydrates. This might be explained by a strong hydrogen bonding of the water molecules in case of erythromycin A dihydrate (Stephenson et al., 1998) which might hinder the dehydration process. After a few hours, the heat flow curves have returned to the blank curve revealing the end of the dehydration processes. When increasing
the RH to 10%, the isomorphic desolvates (emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate) obtained by dehydration all show distinct exothermic peaks compared to the blank curve caused by the incorporation of water molecules into the crystal lattices. The heat flow curves have returned to the blank curve after about 1–1.5 h indicating the attainment of equilibrium. For the desolvates II–IV of emodepside, gravimetric analysis of the water sorption that occurs when increasing RH from 0% to 10% was performed. Measurements reveal that the uptake of water molecules is practically finished after 20 min (desolvate II), 30 min (desolvate III) and 40 min (desolvate IV), respectively (data not shown). Hence, the equilibration times of the desolvates II–IV obtained from the perfusion experiments are longer compared to the corresponding gravimetric equilibration times. Possible explanations for the discrepancies between the equilibration times gained from calorimetric and gravimetric analysis will be discussed below. Also at higher RH values, the non-stoichiometric hydrates II–IV of emodepside show distinct exothermic peaks that correspond to the incorporation of further water molecules into the crystal lattices (Fig. 1). By contrast, the sample curves of erythromycin A and spirapril hydrochloride more or less correspond to the blank curves at higher RH values (Fig. 2). There are slight differences visible between the sample curves of erythromycin A and spirapril hydrochloride and the blank curves (sample curves fall below the blank curves at higher RH). The reason for this is that the heat changes associated with the perfusion of the ampoule slightly differ between sample curves and blank curves. Regarding emodepside I and sulfaguanidine anhydrate, water molecules adsorbed to crystal surfaces desorb at 0% RH resulting in slight endothermic heat flows at the beginning of the experiment (hardly visible for emodepside I) (Figs. 1 and 2). At 10% RH, emodepside I and sulfaguanidine anhydrate show minor exothermic heat
Emodepside II
Emodepside I
100
200
100
200
60 0
40
-100
Emodepside I, 9.34 mg blank RHin
5
10
15
20
25
60 0
40
-100
0 0
30
5
10
Emodepside III
25
30
200
100 ^exo
60 0
40
-100
Emodepside III, 3.23 mg blank RHin
10
15
Time / h
20
25
30
20
RHin / %
80
100
Heat flow / µW
^exo
Heat flow / µW
20
Emodepside IV 100
200
5
15
0
Time / h
Time / h
0
20
Emodepside II, 7.25 mg blank RHin
80
100
60 0
40
-100
Emodepside IV, 3.19 mg blank RHin
0 0
5
10
15
20
25
RHin / %
0
20
80
100
RHin / %
80
100
Heat flow / µW
^exo
RHin / %
Heat flow / µW
^exo
20 0
30
Time / h
Fig. 1. Heat flow curves of modification I and hydrates II–IV of emodepside being first dried at 0% RH and then exposed to increasing RH (25 °C). Sample curves are presented in comparison to the blank curve. RHin is the RH value generated inside the sample vessel.
77
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
100
200
80
150
60
^exo
50 40 0 20
-50
Heat flow / µW
Heat flow / µW
150
100
RHin / %
Erythromycin A, 1.26 mg blank RHin
Spirapril hydrochloride
100
Spirapril hydrochloride, 3.47 mg blank RHin
80
100
60
^exo
50
40
0 20 -50 0
0 0
10
20
RHin / %
Erythromycin A 200
0
30
10
20
30
Time / h
Time / h Sulfaguanidine
200
100
^exo
80
100
60
50
40
0 Sulfaguanidine, 7.67 mg blank RHin
-50 0
5
10
15
20
25
RHin / %
Heat flow / µW
150
20 0
30
Time / h
6
250
200
150
-1
I kJ*mol-1 II kJ*mol-1 III kJ*mol-1 IV kJ*mol-1 I mol*mol-1 II mol*mol-1 III mol*mol-1 IV mol*mol-1
5 4 3
100 2 50
1
0 0
20
40
60
80
Total water sorbed / mol*mol
Cumulative heat of sorption / kJ*mol
flows which stands in opposition to the distinct exothermic peaks observed for the desolvates. The minor exothermic peaks correspond to the adsorption of water molecules to the crystal surfaces. Also at higher RH, the thermal activity of emodepside I exceeds the activity of the blank curve only slightly indicating the adsorption of further water molecules. For sulfaguanidine anhydrate that is known to transform into the monohydrate at higher RH (anhydrate and hydrate have different crystal structures), the heat flow is found to slowly increase after about 1 h at 70% RH suggesting the start of hydrate formation. At 90% RH, the slope of the heat flow curve becomes steeper indicating an increase in the rate of the transition into the hydrate. The heat flow reaches a maximum after about 2 h before it starts to return to the baseline. FT-Raman analysis of the sample after the experiment confirms the transformation into sulfaguanidine monohydrate. The RH step at which hydration starts and the observed dependence of the rate of transition on RH are consistent with literature (Gift and Taylor, 2007). Except for sulfaguanidine that has not attained equilibrium at high RH steps, the heats of sorption at the individual RH steps were determined for the different samples measured. To obtain the heats of sorption, the corresponding blank curve was subtracted from the sample curve and the areas under the peaks were calculated. The heats of sorption were summed up to the cumulative heat of sorption as a function of RH (calorimetric sorption isotherm). For the different solid forms of emodepside, the perfusion experiment was repeated twice and standard errors for the cumulative heats of sorption at the different RH steps were calculated. Fig. 3 shows the cumulative heats of sorption (per mole anhydrous emodepside) including standard errors for emodepside I–IV as a function of RH. The small error bars reveal a good reproducibility of the heat flow measurements. Fig. 3 compares the calorimetric sorption isotherms of emodepside I–IV to the corresponding isotherms determined from gravimetric studies which were already published (Baronsky et al., 2009b). For emodepside I–IV, the shapes
-1
Fig. 2. Heat flow curves of erythromycin A dihydrate, spirapril hydrochloride monohydrate and sulfaguanidine anhydrate being first dried at 0% RH and then exposed to increasing RH (25 °C). Sample curves are presented in comparison to the blank curve. RHin is the RH value generated inside the sample vessel.
0 100
RH / % Fig. 3. Comparison of the calorimetric water vapour sorption isotherms of emodepside I–IV (obtained at 25 °C) versus the corresponding gravimetric sorption isotherms recorded at 22 °C (Baronsky et al., 2009b).
of both isotherms are in good agreement which shows that the heat of sorption is proportional to the amount of water sorbed. Slight differences found between the calorimetric and the gravimetric isotherm for emodepside II–IV may be explained by the fact that the heat of sorption is not only a function of the number of water molecules taken up by the solid, but it is also dependent on the strength of water bonding (Sheridan et al., 1995). The heat of sorption further depends on the accommodations of the crystal lattice that are induced by the incorporation of the water molecules. These accommodations include changes in the local environment inside the lattice and possible adaptations of the entire lattice (lattice expansion). Fig. 4 shows the calorimetric isotherms obtained for erythromycin A and spirapril hydrochloride. The
78
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
3.2. Thermal activity at different RVP of methanol (RVPMeOH)
120
The measured heat flow curves and the corresponding blank curves are given in Figs. 5 and 6. Similar to the perfusion experiments performed with water, the desolvates II–IV of emodepside and the desolvate of erythromycin A dihydrate formed during the initial drying period exhibit substantial exothermic peaks at 10% RVPMeOH. This finding indicates that the desolvates incorporate methanol molecules without changing their three-dimensional structures. Fig. 5 shows that at 10% RVPMeOH, the heat flow of desolvate II decreases less rapidly compared to the heat flows of the desolvates III and IV. The resulting longer equilibration time of desolvate II might be related to slight accommodations of the crystal lattice induced by the incorporation of the methanol molecules. At higher RVPMeOH values, emodepside II–IV show further exothermic peaks. However, the heat changes occurring at higher RVPMeOH are considerably smaller compared to the first RVPMeOH step. In case of emodepside II, the baselines of the sample curve and the blank curve are offset at higher RVPMeOH. The reason for this offset is that the heat changes associated with the perfusion of the ampoule slightly differ between sample curve and blank curve. For erythromycin A, remarkable enthalpy changes are recorded at 70% and 90% RVPMeOH. Visual examination of the sample after the measurement revealed that the sample has changed into a saturated methanol solution. To verify the incorporation of methanol molecules into the crystal lattices of the desolvates II–IV, the samples were analysed by FT-Raman and FT-IR spectroscopy after having finished the measurements. The obtained spectra were compared to the spectra of the desolvates II–IV and to the spectra of the non-stoichiometric hydrates II–IV recorded at ambient conditions. Both the FT-Raman and FT-IR spectra of the samples exposed to methanol vapours slightly differ from the spectra of the desolvates and hydrates. The fact that only small differences are found between the spectra
100 80 60 40
Erythromycin A Spirapril hydrochloride
0 0
20
40
60
80
100
RHin / % Fig. 4. Calorimetric water vapour sorption isotherms of erythromycin A and spirapril hydrochloride (25 °C).
shapes of the isotherms agree well with the shape of the corresponding gravimetric isotherms reported in literature (Stephenson et al., 1998). However, what differs from the gravimetric results is the relative position of the calorimetric isotherms. The cumulative heat of sorption of erythromycin A is about 5 times the cumulative heat of sorption of spirapril hydrochloride (Fig. 4). In opposition to that, the total amount of water sorbed by erythromycin A (mol mol 1) is twice the amount incorporated by spirapril hydrochloride (Stephenson et al., 1998). The higher heat of sorption (heat of hydration) found for erythromycin A indicates a stronger decrease in energy induced by the presence of the water molecules relative to the decrease in energy of spirapril hydrochloride.
Emodepside I
Emodepside II 100
200
60 40
0
20
Emodepside I, 8.84 mg blank RVPin
-200 0
5
10
15
20
25
Heat flow / µW
80
^exo
400
RVPin MeOH / %
Heat flow / µW
100
^exo
400
80
200
60 40
0
0
30
20
Emodepside II, 8.37 mg blank RVPin
-200 0
10
20
Time / h
30
0 40
Time / h Emodepside IV
Emodepside III
100
100 ^exo
200
60 40
0 Emodepside III, 9.38 mg blank RVPin
-200
5
10
15
Time / h
20
25
30
20
Heat flow / µW
80
0
^exo
400
RVPin MeOH / %
Heat flow / µW
400
RVPin MeOH / %
20
80
200
60 40
0 Emodepside IV, 9.36 mg blank RVPin
-200 0
0
5
10
15
20
25
20
RVPin MeOH / %
Cumulative heat of sorption / kJ*mol
-1
140
0
30
Time / h
Fig. 5. Heat flow curves of modification I and hydrates II–IV of emodepside being first dried at 0% RVP and then exposed to increasing RVPin values of methanol (25 °C). Sample curves are presented in comparison to the blank curve. RVPin is the RVP value of methanol generated inside the sample vessel.
79
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Erythromycin A
Spirapril hydrochloride 600
400
80
^exo
60
200 40 0
20
100 Spirapril hydrochloride, 8.02 mg blank RVPin
400
80
^exo
60
200 40 0
20 0
0 0
10
20
30
RVPin MeOH / %
Erythromycin A, 4.37 mg blank RVPin
Heat flow / µW
100
RVPin MeOH / %
Heat flow / µW
600
0
10
20
30
Time / h
Time / h
Fig. 6. Heat flow curves of erythromycin A dihydrate and spirapril hydrochloride monohydrate being first dried at 0% RVP and then exposed to increasing RVPin values of methanol (25 °C). Sample curves are presented in comparison to the blank curve. RVPin is the RVP value of methanol generated inside the sample vessel.
Raman Intensity (SNV corrected)
reveals that methanol molecules entered the crystal lattices of the desolvates II–IV without changing the crystal structures. Consequently, isomorphic methanolates have formed named methanolate II–IV. Fig. 7 presents the Raman spectra of the methanolates in comparison to the spectra of both desolvates and hydrates (Baronsky et al., 2009b). The CAH and the C@O stretching region are presented since changes are most distinct in these regions. The spectrum of methanolate II differs from the spectra of desolvate II and hydrate II in the relative intensities of the bands present in the CAH stretching region. Differences are also found in the C@O stretching region, e.g. the band present at 1729 cm 1 in the spectra of hydrate II and desolvate II is shifted to 1724 cm 1 in the spectrum of methanolate II. Regarding the CAH stretching region of methanolate III, the band maximum at 2941 cm 1 found in the spectra of desolvate III and hydrate III is absent. In the C@O stretching region, the spectrum of methanolate III more or less corresponds to the spectrum of hydrate III. In case of methanolate IV, the interaction of methanol molecules with C@O groups of the lattice causes the shoulder of the band at 1654 cm 1 present in the spectrum of the desolvate to shift to 1633 cm 1 in the spectrum of methanolate IV. The described differences between the spectra of the methanolates and the spectra of the corresponding hydrates and desolvates represent the changes in the local environment of
0% RH 45% ± 15% RH 90% RVPMeOH II
III
IV 3100
3000
2900
2800
1750
1650
Wavenumber / cm-1 Fig. 7. FT-Raman spectra of the desolvates II–IV of emodepside exposed to methanol vapours compared to the spectra of the desolvates II–IV (0% RH) and the hydrates II–IV (45% ± 15% RH) (Baronsky et al., 2009b).
the CAH and C@O groups caused by the presence of methanol molecules. Similar to the water molecules present in the hydrates, methanol molecules interact with C@O groups of the lattices leading to a shift of C@O stretches towards lower wavenumbers (in comparison to the C@O stretches present in spectra of the corresponding desolvates) (Baronsky et al., 2009b). Regarding methanolate II and IV, the shift of the C@O vibrations is more pronounced than the shift of the C@O vibrations in the spectra of the hydrates II and IV indicating hydrogen bonding to be stronger in the methanolates. For the spectrum of methanolate II, the differences to the spectrum of desolvate II are more pronounced than the differences between the spectra of hydrate II and desolvate II which suggests that the incorporation of the methanol molecules induces more changes in the local environment inside the crystal lattice than the uptake of water molecules. This suggestion is in agreement with the elongated equilibration time found for methanolate II (Fig. 5). To confirm that methanol molecules have entered the lattice of the desolvate of erythromycin A at 10% RVPMeOH, the experiment was stopped at the end of the 10% step and the sample was examined using FT-Raman and FT-IR spectroscopy. The FT-Raman spectrum of the sample more or less corresponds to the spectrum of the desolvate of erythromycin A (data not shown). There might be slight differences between both spectra, however, due to a low signal-to-noise ratio, it is not possible to decide whether differences are significant. In opposition to that, the FT-IR spectrum of the sample shows slight but significant differences in comparison to the spectra of the desolvate and the dihydrate (Fig. 8). For example, differences are readily visible in the C@O stretching region. While all solid forms have a strong band at 1713 cm 1, the sample obtained from the perfusion experiment additionally exhibits an absorption maximum at 1732 cm 1 whereas in case of the desolvate, two bands overlap with the band at 1713 cm 1 being visible as shoulders. Summarising, spectroscopic analysis confirms the transformation of the desolvate into an isomorphic methanolate for emodepside II–IV and erythromycin A which was suggested from perfusion calorimetry. In opposition to the desolvates of emodepside II–IV and erythromycin A, the heat flow curve of the desolvate of spirapril hydrochloride is found to correspond to the blank curve throughout the RVPMeOH range (Fig. 6). This finding suggests that the formation of an isomorphic methanolate does not occur since the structural voids present in the lattice of the desolvate are not sufficient in size to accommodate methanol molecules. After having finished the measurement, the sample was analysed using FT-Raman and FTIR spectroscopy. In both cases, the spectrum of the monohydrate was obtained (data not shown). A possible explanation for this finding is that the desolvate is still present at the end of the experiment and, due to its high hygroscopicity, re-hydrates during the
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Absorbance Units
80
0% RH
45% ± 15% RH
10% RVPMeOH 3500
3000
2500
2000
1500
1000
Wavenumber / cm-1 Fig. 8. FT-IR spectrum of the desolvate of erythromycin A exposed to 10% RVP of methanol compared to the spectra of the desolvate (0% RH) and the hydrate of erythromycin A (45% ± 15% RH).
transfer of the sample into the sample holder (FT-Raman spectroscopy) and the ATR unit (FT-IR spectroscopy), respectively, so that the spectrum of the monohydrate is obtained. This explanation supports the supposition from perfusion calorimetry that methanol molecules do not enter the lattice of the desolvate of spirapril hydrochloride. Regarding emodepside I, the sample curve is similar to the blank curve up to 70% RVPMeOH (Fig. 5). Only minor exothermic peaks are measured compared to the blank (hardly visible) suggesting the adsorption of methanol molecules to the crystal surface. At 90% RVPMeOH, a sharp initial peak occurs followed by a prolonged exothermic event. Analysing the sample after the measurement using FT-Raman and FT-IR spectroscopy reveals that emodepside I has transformed into methanolate IV of emodepside. For emodepside II–IV, the perfusion experiment was repeated twice. The measurements were used to calculate the cumulative heats of sorption ± standard errors as a function of RVPMeOH (Fig. 9). The standard errors displayed in Fig. 9 are small (most of the times hidden behind the symbols used) which reveals that a good reproducibility of the heat flow measurements is not only obtained for water (Fig. 3) but also for methanol as solvent. The calorimetric sorption isotherm of emodepside I shown in Fig. 9 was
Fig. 9. Calorimetric methanol vapour sorption isotherms of emodepside I–IV (25 °C).
determined from the curve presented in Fig. 5 neglecting the fact that equilibrium is not attained at the end of the 90% RVPMeOH step. The inaccuracy of the isotherm at 90% RVPMeOH is accepted, since the isotherm is only used to reveal the differences between its shape and the shape of the isotherms of emodepside II–IV. Fig. 9 shows that the isotherm of emodepside I considerably differs from the isotherms of the non-stoichiometric methanolates II–IV forming isomorphic desolvates. While the isotherm of emodepside I is typical for a solvate formation that is accompanied by structural rearrangement, the isotherms of emodepside II–IV resemble type I isotherms. This type of isotherms is frequently found in non-stoichiometric solvates forming isomorphic desolvates and therefore indicates the presence of this type of solvates (Authelin, 2005).
4. Discussion In each case studied, the transformation of the desolvate into the isomorphic solvate (hydrate or methanolate) is characterized by a substantial exothermic event at low RVP. This thermal activity results in a notable increase of the calorimetric isotherm in the region of low RVP. The considerable heat production reflects the uptake of a substantial number of solvent molecules at low RVP which is characteristic for desolvates. The driving force for solvent incorporation is to reduce the molecular vacuum present in the lattice of the desolvate (Stephenson et al., 1998). The transformation of the desolvate into the isomorphic solvate is not only indicated by the shape of the calorimetric isotherm, but also by the shape of the heat flow curve associated with the solvate formation. In each case studied, the heat flow curve is peak-shaped, i.e. the heat flow rapidly increases and decreases, which reflects the fast uptake of solvent molecules and the resulting short equilibration time which are typical for desolvates forming isomorphic solvates (Stephenson et al., 1998). The rapidness of solvation is due to several factors, one of them being the low activation energy of the process caused by the fact that structural changes are not required. As described in the section on results, the calorimetric equilibration times of the desolvates II–IV of emodepside at 10% RH differ from the corresponding gravimetric equilibration times. One possibility is that the longer equilibration times observed in the calorimetric experiments may be due to thermal activity associated with slight accommodations of molecular groups inside the lattices induced by the incorporation of water molecules. However,
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
it might also be possible that the equilibration process inside the sample vessel is limited by the delivery of water molecules via the wet line. Whether the water supply is the rate limiting step depends on the ratio of the amount of water delivered by the wet line to the amount of water sorbed by the sample. In order for the water supply to not limit the equilibration process, an excess of water has to be supplied relative to the amount of water sorbed (Danforth and David, 2006). Since the amount of water taken up by the sample is proportional to the weight of sample, the weight of sample used for measurement is important. In case of emodepside I, for which the water supply was not expected to limit the equilibration process since the solid form adsorbs only minor amounts of water, a weight of sample that just covers the bottom of the vessel was used. Although a higher weight of sample would be preferable in terms of the signal-to-noise ratio, it would lead to a layer formed by the sample. As the diffusion of the water molecules through the layer might be a limiting factor in the attainment of equilibrium, the amount of sample just covering the bottom of the vessel was used (about 9 mg). For emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate, the weights of sample used were below the amounts needed to cover the bottom of the vessel in order to prevent the amount of water entering the vessel to limit the equilibration process. The weight of sample used was inversely proportional to the amount of water sorbed by the desolvate at 10% RH (known from gravimetric analysis). Despite of the reduced weight of sample used, it is not sure whether the number of water molecules delivered by the wet line still limits the equilibration of the desolvates in the vessel. Possibilities to investigate whether the water supply is rate limiting are to further reduce sample weights or to increase the total flow rate in order to increase the amount of water being delivered by the wet line. If the process was limited by the amount of water entering the vessel, shorter equilibration times would be observed in both cases but the heat flow curves would still be peak-shaped. However, the main focus of this study was not to determine the exact equilibration times but to obtain the shape of the curves which provides indication about the kinetics of the process observed. Furthermore, both the reduction of the weight of sample and the increase of the total flow rate would result in a lower signal-tonoise ratio. Therefore, the experimental setup was not changed. The investigation of emodepside I and sulfaguanidine anhydrate has shown that perfusion calorimetry is suitable to distinguish the transformation of a desolvate into an isomorphic solvate from both adsorption of solvent molecules and solvate formation that is accompanied by structural changes. Although the adsorption of water molecules to crystal surfaces of sulfaguanidine anhydrate (Fig. 2) as well as the adsorption of water molecules and methanol molecules to the crystal surfaces of emodepside I (Figs. 1 and 5) exhibit sorption kinetics similar to the kinetics found for desolvateisomorphic solvate formation, the heat changes associated are distinctly smaller leading to different calorimetric isotherms. The transformations of sulfaguanidine anhydrate into the monohydrate (Fig. 2) and of emodepside I into methanolate IV (Fig. 5), both representing solvate formation that is accompanied by changes in the crystal structure, differ from the transformation of the various desolvates into isomorphic solvates by the kinetics of the processes. Compared to the rapid increase and decrease of the heat flow observed for desolvate-isomorphic solvate formation, the heat flow associated with the transition of sulfaguanidine anhydrate and emodepside I, respectively, changes less rapidly resulting in an elongated equilibration time. The elongated equilibration time is caused by the structural changes that are not required in case of the transformation of a desolvate into an isomorphic solvate. However, it has to be taken into consideration that the energy barriers to structural rearrangement may (partly) be overcome by milling the sample, for example (generation of defective sites that facili-
81
tate solid-solid transition). In this case, a peak-shaped heat flow curve may be obtained. Consequently, the history of the sample has to be known.
5. Conclusions The investigation of several non-stoichiometric solvates forming isomorphic desolvates has shown that perfusion calorimetry is a valuable tool to indicate the transformation of a desolvate into an isomorphic solvate. This represents a new field of application for the method. The process of the desolvate transforming into an isomorphic solvate is characterized by a notable exothermic, peak-shaped heat flow at low RVP which reflects the rapid uptake of solvent molecules by the desolvate to reduce the molecular vacuum present in the lattice. The method allows the user to both generate the desolvate and expose it to different RVP of solvent in a single experiment. Furthermore, perfusion calorimetry does not require any sample preparation and it can be used for different solvents. Since the presence of a desolvate/isomorphic solvate system is, among other things, deduced from the shape of the heat flow curve, it has to be taken into consideration that the amount of solvent supplied by the wet line may delay the attainment of equilibrium and thus may affect the shape of the curve. Possibilities to prevent the amount of solvent delivered to limit equilibration are the use of a small weight of sample and the increase of the total flow rate used. Although perfusion calorimetry, due to its unspecific nature, is not able to prove the existence of a desolvate transforming into an isomorphic solvate, it is suitable to provide indication within a short period of time.
References Authelin, J.-R., 2005. Thermodynamics of non-stoichiometric pharmaceutical hydrates. Int. J. Pharm. 303, 37–53. Baronsky, J., Bongaerts, S., Traeubel, M., Urbanetz, N., 2009a. Isothermal microcalorimetry perfusion experiments: a method to verify the relative vapour pressure of organic liquids inside the sample vessel. Thermochim. Acta 493, 37–41. Baronsky, J., Bongaerts, S., Traeubel, M., Weiss, H.-C., Urbanetz, N., 2009b. The study of different solid forms of emodepside. Eur. J. Pharm. Biopharm. 71, 88–99. Burger, A., Henck, J.-O., Hetz, S., Rollinger, J.M., Weissnicht, A.A., Stöttner, H., 1999. Energy/temperature diagram and compression behavior of the polymorphs of D-mannitol. J. Pharm. Sci. 89, 457–468. Carvajal, M.T., Staniforth, J.N., 2006. Interactions of water with the surfaces of crystal polymorphs. Int. J. Pharm. 307, 216–224. Danforth, P.M., David, L., 2006. Rapid assessment of the structural relaxation behavior of amorphous pharmaceutical solids: effect of residual water on molecular mobility. Pharm. Res. 23, 2291–2305. Gaisford, S., 2007. Pharmaceutical Isothermal Calorimetry. Informa Healthcare Inc., New York, pp. 33–37. Gift, A.D., Taylor, L.S., 2007. Hyphenation of Raman spectroscopy with gravimetric analysis to interrogate water–solid interactions in pharmaceutical systems. J. Pharm. Biomed. Anal. 43, 14–23. Giron, D., Goldbronn, Ch., Mutz, M., Pfeffer, S., Piechon, Ph., Schwab, Ph., 2002. Solid state characterizations of pharmaceutical hydrates. J. Therm. Anal. Calorim. 68, 453–465. Griesser, U.J., 2006. The importance of solvates. In: Hilfiker, R. (Ed.), Polymorphism in the Pharmaceutical Industry. VILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, pp. 211–233. Lehto, V.-P., Laine, E., 1998. Assessment of physical stability of different forms of cefadroxil at high humidities. Int. J. Pharm. 163, 49–62. Markova, N., Sparr, E., Wadso, L., 2001. On application of an isothermal sorption microcalorimeter. Thermochim. Acta 374, 93–104. Miroshnyk, I., Khriachtchev, L., Mirza, S., Rantanen, J., Heinämäki, J., Yliruusi, J., 2006. Insight into thermally induced phase transformations of erythromycin a dihydrate. Cryst. Growth Des. 6, 369–374. Newman, A.W., Byrn, S.R., 2003. Solid-state analysis of the active pharmaceutical ingredient in drug products. Drug Discov. Today 8, 898–905. Othman, A., Evans, J.S.O., Evans, I.R., Harris, R.K., Hodgkinson, P., 2007. Structural study of polymorphs and solvates of finasteride. J. Pharm. Sci. 96, 1380–1397. Redman-Furey, N., Dicks, M., Bigalow-Kern, A., Cambron, R.T., Lubey, G., Lester, C., Vaughn, D., 2005. Structural and analytical characterization of three hydrates and an anhydrate form of risedronate. J. Pharm. Sci. 94, 893–911.
82
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Rollinger, J.M., Burger, A., 2002. Physico-chemical characterization of hydrated and anhydrous crystal forms of amlodipine besylate. J. Therm. Anal. Calorim. 68, 361–372. Ruth, L.T., Griesser, U.J., Morris, K.R., Byrn, S.R., Stowell, J.G., 2003. X-ray diffraction and solid-state NMR investigation of the single-crystal to single-crystal dehydration of thiamine hydrochloride monohydrate. Cryst. Growth Des. 3, 997–1004. Sheridan, P.L., Graham, B., Storey, D., 1995. Development of a flow microcalorimetry method for the assessment of surface properties of powders. Pharm. Res. 12, 1025–1030.
Stephenson, G.A., Groleau, E.G., Kleemann, R.L., Xu, W., Rigsbee, D.R., 1998. Formation of isomorphic desolvates: creating a molecular vacuum. J. Pharm. Sci. 87, 536–542. Stephenson, G.A., Stowell, J.G., Toma, P.H., Pfeiffer, R.R., Byrn, S.R., 1997. Solid-state investigations of erythromycin a dihydrate: structure, NMR spectroscopy, and hygroscopicity. J. Pharm. Sci. 86, 1239–1244. Timmermann, I.L., Steckel, H., Trunk, M., 2006. Assessing the re-crystallization behaviour of amorphous lactose using the RH-perfusion cell. Eur. J. Pharm. Biopharm. 64, 107–114.
Contents lists available at ScienceDirect
European Journal of Pharmaceutical Sciences journal homepage: www.elsevier.com/locate/ejps
Perfusion calorimetry in the characterization of solvates forming isomorphic desolvates Julia Baronsky a,b, Martina Preu b, Michael Traeubel b, Nora Anne Urbanetz c,⇑ a
Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine-University, Universitaetsstrasse 1, D-40225 Duesseldorf, Germany BAH-RD-A, Bayer Animal Health GmbH, Alfred-Nobel-Strasse 50, D-40789 Monheim, Germany c Research Center Pharmaceutical Engineering GmbH, Inffeldgasse 21a/II, A-8010 Graz, Austria b
a r t i c l e
i n f o
Article history: Received 3 February 2011 Received in revised form 9 June 2011 Accepted 17 June 2011 Available online 25 June 2011 Keywords: Perfusion calorimetry Microcalorimetry Non-stoichiometric solvate Isomorphic desolvate
a b s t r a c t In this study, the potential of perfusion calorimetry in the characterization of solvates forming isomorphic desolvates was investigated. Perfusion calorimetry was used to expose different hydrates forming isomorphic desolvates (emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate) to stepwise increasing relative vapour pressures (RVP) of water and methanol, respectively, while measuring thermal activity. Furthermore, the suitability of perfusion calorimetry to distinguish the transformation of a desolvate into an isomorphic solvate from the adsorption of solvent molecules to crystal surfaces as well as from solvate formation that is accompanied by structural rearrangement was investigated. Changes in the samples were confirmed using FT-Raman and FT-IR spectroscopy. Perfusion calorimetry indicates the transformation of a desolvate into an isomorphic solvate by a substantial exothermic, peak-shaped heat flow curve at low RVP which reflects the rapid incorporation of solvent molecules by the desolvate to fill the structural voids in the lattice. In contrast, adsorption of solvent molecules to crystal surfaces is associated with distinctly smaller heat changes whereas solvate formation accompanied by structural changes is characterized by an elongated heat flow. Hence, perfusion calorimetry is a valuable tool in the characterization of solvates forming isomorphic desolvates which represents a new field of application for the method. Ó 2011 Published by Elsevier B.V.
1. Introduction Active pharmaceutical ingredients as well as excipients may exist in different solid forms which is of great importance for pharmaceutical development (Burger et al., 1999; Newman and Byrn, 2003; Othman et al., 2007). One important type of solid forms are solvates. A solvate may be defined as a crystalline multi-component system in which a solvent (or several solvents) is coordinated in or accommodated by the crystal structure (Griesser, 2006). In case the solvent is water, the solvate is called hydrate. The omnipresence of water makes hydrate formation an important issue in pharmaceutical industry (Giron et al., 2002; Redman-Furey et al., 2005). The formation of organic solvates plays a significant role also, since pharmaceutical substances are often exposed to organic solvents during production and processing, for example. Solvates may be divided into stoichiometric and non-stoichiometric solvates (Griesser, 2006). A stoichiometric solvate is a solvate
⇑ Corresponding author. Tel.: +43 316 873 9720; fax: +43 316 873 109720. E-mail address: [email protected] (N.A. Urbanetz). 0928-0987/$ - see front matter Ó 2011 Published by Elsevier B.V. doi:10.1016/j.ejps.2011.06.008
that, within the range of solvent activity its crystal structure exists in (at given pressure and temperature), has a fixed number of solvent molecules inside the crystal lattice, while, for a non-stoichiometric solvate, the number of solvent molecules inside the lattice is dependent on solvent activity. Non-stoichiometric solvates may desolvate retaining their crystal structures (Stephenson et al., 1998). The isostructural unsolvated form generated by desolvation is called isomorphic desolvate or desolvated solvate. Due to the fact that desolvation causes only slight changes to the crystal lattice, the identification of solvates forming isomorphic desolvates might not be straightforward (Griesser, 2006). Methods that have been shown to be suitable to identify this type of solvates are single crystal structure analysis (Ruth et al., 2003), X-ray powder diffraction (XRPD) (Stephenson et al., 1998) and spectroscopic techniques (Stephenson et al., 1997). XRPD detects any shrinking of the lattice which may occur upon desolvation (Rollinger and Burger, 2002). Spectroscopic techniques, such as solid-state nuclear magnetic resonance spectroscopy and Raman spectroscopy, are sensitive to the changes in the local environment of molecular groups inside the lattice induced by the departure of the solvent molecules (Miroshnyk et al., 2006; Ruth et al., 2003). However, as some of the methods might not be available or in
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
some cases, a method might fail to detect whether a solvate forming an isomorphic desolvate is present (Stephenson et al., 1998), further methods suitable to characterize this type of solvates are needed. The aim of this study is to elucidate the potential of perfusion calorimetry in the characterization of solvates converting into isomorphic desolvates. In perfusion calorimetry, a sample is exposed to a controlled relative vapour pressure (RVP) of solvent while measuring thermal activity (Gaisford, 2007). The method has been used to study solvent-solid interactions, e.g. the adsorption of solvent molecules to crystal surfaces and the sorption of solvent molecules by amorphous substances (Carvajal and Staniforth, 2006; Danforth and David, 2006; Timmermann et al., 2006). However, its potential in the investigation of solvate formation has been barely used (Lehto and Laine, 1998; Markova et al., 2001). In this study, different substances that are known to be hydrates forming isomorphic desolvates were investigated using perfusion calorimetry. The aim was to assess whether perfusion calorimetry is suitable to detect solvates changing into isomorphic desolvates which would represent a new field of application for the method. The substances studied are the non-stoichiometric hydrates II–IV of emodepside (veterinary drug substance; Baronsky et al., 2009b), erythromycin A dihydrate and spirapril hydrochloride monohydrate (Stephenson et al., 1998). The hydrates were first desolvated in situ inside the sample vessel using dry nitrogen. The isomorphic desolvates generated were then exposed to stepwise increasing RVP values of water (relative humidities, RH) to study the heat changes associated with hydrate formation. In addition, the isomorphic desolvates were exposed to stepwise increasing RVP values of methanol (RVPMeOH) to investigate whether methanol molecules enter the structural voids inside the crystal lattices previously occupied by water molecules. Furthermore, emodepside I (non-solvated crystalline form of emodepside; Baronsky et al., 2009b) and sulfaguanidine anhydrate were used as model substances to test the ability of perfusion calorimetry to distinguish isomorphic solvate formation from solvent adsorption to crystal surfaces and from solvate formation that is accompanied by structural rearrangement.
2. Materials and methods 2.1. Materials Emodepside I was purchased from KPV, Kiel, Germany. The hydrates II–IV of emodepside were prepared using methods described in literature (Baronsky et al., 2009b). The hydrates II and IV were formed by cooling a hot ethanol 96% (V/V) solution (hydrate II) and methanol solution (hydrate IV), respectively, at room temperature. Hydrate III was produced by slurry phase conversion of emodepside II hydrate at room temperature using acetone. Erythromycin A dihydrate, analytical standard, was obtained from Riedel de Haen, Seelze, Germany. The desolvate of erythromycin A dihydrate was obtained by storing the dihydrate over P205 for 4 days. Spirapril hydrochloride monohydrate CRS was purchased from EDQM (European Directorate for the Quality of Medicines). Regarding sulfaguanidine, the monohydrate was supplied by Fluka, Buchs, Switzerland. The anhydrate used for experiments was prepared as described in literature (Gift and Taylor, 2007). The organic solvents used in this study (for the preparation of the hydrates II–IV of emodepside as well as for perfusion experiments) were of analytical grade and purchased from Merck, Darmstadt, Germany. The water used in the perfusion experiments was Milli-Q water obtained by a Milli-Q Gradient System (Millipore, Schalbach, Germany).
75
2.2. Methods 2.2.1. Perfusion calorimetry Experiments were performed using a 2277 Thermal Activity Monitor (TA Instruments, New Castle, Delaware). To expose the sample to a controlled RVP of solvent during measurement, vapour pressure control devices based on two mass flow controllers were used. They are composed of the 2250010 4 ml RH Perfusion Ampoule, two mass flow controllers (Bronkhorst HighTech, AK Ruurlo, Netherlands) and the 3811 Flow Control Module (TA Instruments, New Castle, Delaware). The performances of the control devices were tested using the method described by Baronsky et al., 2009a. The results from the testing procedure were used to adapt RVP settings in order to create the desired RVP values inside the reaction vessel (RHin values and RVPin values of methanol, respectively). All experiments were carried out at 25 °C. The total flow rate was set to 80 ml h 1 (0 °C, 1.01325 bar). The amount of solvent (water or methanol) placed into the first and second solvent reservoir of the perfusion ampoule was 1 ml and 0.5 ml, respectively. The heater of the perfusion ampoule was set to 45 °C. The solid material under investigation was accurately weighted and transferred into the reaction vessel. The weight of sample ranged between 1 mg and 10 mg. The loaded reaction vessel was connected to the perfusion ampoule which was then equilibrated in the different equilibration positions for 20 min before starting the measurement. At the beginning of measurement, the sample was exposed to 0% RVP (drying period). After the heat flow curve of the sample had returned to the blank curve (attainment of equilibrium of the sample), the RVP inside the vessel (RHin and RVPin of methanol, respectively) was increased to 10%, 20%, 30%, 50%, 70% and 90%. A time delay of 4 h per step was sufficient for the solvates forming isomorphic desolvates to attain equilibrium at each step. Only in case of emodepside II subjected to methanol vapours, it was not sure whether the heat flow curve of the sample had returned to the blank curve at each step after 4 h. Thus, in this case, RVP steps were run for 6 h instead of 4 h. For each type of experiment, a blank curve was recorded by repeating the experiment using an empty sample vessel. Each experiment was performed having an empty, closed stainless steel ampoule in the reference position. Before starting a series of measurements, the calorimeter was calibrated using a static calibration. The calibration heat flow generated by the internal calibration heater resistors was 1000 lW. The calibration was performed having the perfusion ampoule inside the measuring position and the reference ampoule inside the reference position. During calibration, the solvent reservoirs and the reaction vessel of the perfusion ampoule were empty and the gas flow was turned off. 2.2.2. Spectroscopic analysis After having finished a perfusion experiment, the sample was subsequently analysed using both FT-Raman spectroscopy and FT-IR analysis. The sample was transferred into the sample holder (FT-Raman spectroscopy) and the ATR unit (FT-IR spectroscopy), respectively, as quickly as possible in order to minimize the time prior to measurement (time in which the sample was exposed to ambient conditions). 2.2.3. FT-Raman spectroscopy Raman spectra were acquired with a Bruker RFS 100S spectrometer (Bruker Optik GmbH, Ettlingen, Germany) using a Nd:YAG laser (1064 nm) as excitation source and a high sensitivity Gedetector (cooled with liquid nitrogen). The spectra were recorded from 3500 cm 1 to 20 cm 1 at a resolution of 1 cm 1 (128 scans) using a laser output of 750 mW.
76
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
2.2.4. FT-IR spectroscopy Infrared analysis was carried out with a FT-IR Bruker Tensor 37 spectrometer (Bruker Optik GmbH, Ettlingen, Germany) using a Heated Diamond Top-plate as sampling system (ATR method). Spectra were recorded from 4000 cm 1 to 550 cm 1 at a resolution of 2 cm 2 (64 scans). 2.2.5. Data treatment Some of the data were treated using the standard normal variate (SNV) method to correct multiplicative effects. Regarding FT-IR spectra, the entire measuring range was used for SNV. For FTRaman spectra, SNV was performed over the range of 3200– 570 cm 1 (emodepside samples), 3200–550 cm 1 (spirapril hydrochloride samples) and 3200–20 cm 1 (erythromycin A samples), respectively. 3. Results 3.1. Thermal activity at different RH Figs. 1 and 2 show the heat flow curves of the different samples exposed to increasing RH in comparison to the corresponding blank curves. At the beginning of the experiment (0% RH), the nonstoichiometric hydrates II–IV of emodepside, erythromycin A dihydrate and spirapril hydrochloride monohydrate show considerable endothermic heat flows corresponding to dehydration. For erythromycin A dihydrate, the shape of the endothermic heat flow curve differs from the shape of the curves of the other hydrates. This might be explained by a strong hydrogen bonding of the water molecules in case of erythromycin A dihydrate (Stephenson et al., 1998) which might hinder the dehydration process. After a few hours, the heat flow curves have returned to the blank curve revealing the end of the dehydration processes. When increasing
the RH to 10%, the isomorphic desolvates (emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate) obtained by dehydration all show distinct exothermic peaks compared to the blank curve caused by the incorporation of water molecules into the crystal lattices. The heat flow curves have returned to the blank curve after about 1–1.5 h indicating the attainment of equilibrium. For the desolvates II–IV of emodepside, gravimetric analysis of the water sorption that occurs when increasing RH from 0% to 10% was performed. Measurements reveal that the uptake of water molecules is practically finished after 20 min (desolvate II), 30 min (desolvate III) and 40 min (desolvate IV), respectively (data not shown). Hence, the equilibration times of the desolvates II–IV obtained from the perfusion experiments are longer compared to the corresponding gravimetric equilibration times. Possible explanations for the discrepancies between the equilibration times gained from calorimetric and gravimetric analysis will be discussed below. Also at higher RH values, the non-stoichiometric hydrates II–IV of emodepside show distinct exothermic peaks that correspond to the incorporation of further water molecules into the crystal lattices (Fig. 1). By contrast, the sample curves of erythromycin A and spirapril hydrochloride more or less correspond to the blank curves at higher RH values (Fig. 2). There are slight differences visible between the sample curves of erythromycin A and spirapril hydrochloride and the blank curves (sample curves fall below the blank curves at higher RH). The reason for this is that the heat changes associated with the perfusion of the ampoule slightly differ between sample curves and blank curves. Regarding emodepside I and sulfaguanidine anhydrate, water molecules adsorbed to crystal surfaces desorb at 0% RH resulting in slight endothermic heat flows at the beginning of the experiment (hardly visible for emodepside I) (Figs. 1 and 2). At 10% RH, emodepside I and sulfaguanidine anhydrate show minor exothermic heat
Emodepside II
Emodepside I
100
200
100
200
60 0
40
-100
Emodepside I, 9.34 mg blank RHin
5
10
15
20
25
60 0
40
-100
0 0
30
5
10
Emodepside III
25
30
200
100 ^exo
60 0
40
-100
Emodepside III, 3.23 mg blank RHin
10
15
Time / h
20
25
30
20
RHin / %
80
100
Heat flow / µW
^exo
Heat flow / µW
20
Emodepside IV 100
200
5
15
0
Time / h
Time / h
0
20
Emodepside II, 7.25 mg blank RHin
80
100
60 0
40
-100
Emodepside IV, 3.19 mg blank RHin
0 0
5
10
15
20
25
RHin / %
0
20
80
100
RHin / %
80
100
Heat flow / µW
^exo
RHin / %
Heat flow / µW
^exo
20 0
30
Time / h
Fig. 1. Heat flow curves of modification I and hydrates II–IV of emodepside being first dried at 0% RH and then exposed to increasing RH (25 °C). Sample curves are presented in comparison to the blank curve. RHin is the RH value generated inside the sample vessel.
77
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
100
200
80
150
60
^exo
50 40 0 20
-50
Heat flow / µW
Heat flow / µW
150
100
RHin / %
Erythromycin A, 1.26 mg blank RHin
Spirapril hydrochloride
100
Spirapril hydrochloride, 3.47 mg blank RHin
80
100
60
^exo
50
40
0 20 -50 0
0 0
10
20
RHin / %
Erythromycin A 200
0
30
10
20
30
Time / h
Time / h Sulfaguanidine
200
100
^exo
80
100
60
50
40
0 Sulfaguanidine, 7.67 mg blank RHin
-50 0
5
10
15
20
25
RHin / %
Heat flow / µW
150
20 0
30
Time / h
6
250
200
150
-1
I kJ*mol-1 II kJ*mol-1 III kJ*mol-1 IV kJ*mol-1 I mol*mol-1 II mol*mol-1 III mol*mol-1 IV mol*mol-1
5 4 3
100 2 50
1
0 0
20
40
60
80
Total water sorbed / mol*mol
Cumulative heat of sorption / kJ*mol
flows which stands in opposition to the distinct exothermic peaks observed for the desolvates. The minor exothermic peaks correspond to the adsorption of water molecules to the crystal surfaces. Also at higher RH, the thermal activity of emodepside I exceeds the activity of the blank curve only slightly indicating the adsorption of further water molecules. For sulfaguanidine anhydrate that is known to transform into the monohydrate at higher RH (anhydrate and hydrate have different crystal structures), the heat flow is found to slowly increase after about 1 h at 70% RH suggesting the start of hydrate formation. At 90% RH, the slope of the heat flow curve becomes steeper indicating an increase in the rate of the transition into the hydrate. The heat flow reaches a maximum after about 2 h before it starts to return to the baseline. FT-Raman analysis of the sample after the experiment confirms the transformation into sulfaguanidine monohydrate. The RH step at which hydration starts and the observed dependence of the rate of transition on RH are consistent with literature (Gift and Taylor, 2007). Except for sulfaguanidine that has not attained equilibrium at high RH steps, the heats of sorption at the individual RH steps were determined for the different samples measured. To obtain the heats of sorption, the corresponding blank curve was subtracted from the sample curve and the areas under the peaks were calculated. The heats of sorption were summed up to the cumulative heat of sorption as a function of RH (calorimetric sorption isotherm). For the different solid forms of emodepside, the perfusion experiment was repeated twice and standard errors for the cumulative heats of sorption at the different RH steps were calculated. Fig. 3 shows the cumulative heats of sorption (per mole anhydrous emodepside) including standard errors for emodepside I–IV as a function of RH. The small error bars reveal a good reproducibility of the heat flow measurements. Fig. 3 compares the calorimetric sorption isotherms of emodepside I–IV to the corresponding isotherms determined from gravimetric studies which were already published (Baronsky et al., 2009b). For emodepside I–IV, the shapes
-1
Fig. 2. Heat flow curves of erythromycin A dihydrate, spirapril hydrochloride monohydrate and sulfaguanidine anhydrate being first dried at 0% RH and then exposed to increasing RH (25 °C). Sample curves are presented in comparison to the blank curve. RHin is the RH value generated inside the sample vessel.
0 100
RH / % Fig. 3. Comparison of the calorimetric water vapour sorption isotherms of emodepside I–IV (obtained at 25 °C) versus the corresponding gravimetric sorption isotherms recorded at 22 °C (Baronsky et al., 2009b).
of both isotherms are in good agreement which shows that the heat of sorption is proportional to the amount of water sorbed. Slight differences found between the calorimetric and the gravimetric isotherm for emodepside II–IV may be explained by the fact that the heat of sorption is not only a function of the number of water molecules taken up by the solid, but it is also dependent on the strength of water bonding (Sheridan et al., 1995). The heat of sorption further depends on the accommodations of the crystal lattice that are induced by the incorporation of the water molecules. These accommodations include changes in the local environment inside the lattice and possible adaptations of the entire lattice (lattice expansion). Fig. 4 shows the calorimetric isotherms obtained for erythromycin A and spirapril hydrochloride. The
78
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
3.2. Thermal activity at different RVP of methanol (RVPMeOH)
120
The measured heat flow curves and the corresponding blank curves are given in Figs. 5 and 6. Similar to the perfusion experiments performed with water, the desolvates II–IV of emodepside and the desolvate of erythromycin A dihydrate formed during the initial drying period exhibit substantial exothermic peaks at 10% RVPMeOH. This finding indicates that the desolvates incorporate methanol molecules without changing their three-dimensional structures. Fig. 5 shows that at 10% RVPMeOH, the heat flow of desolvate II decreases less rapidly compared to the heat flows of the desolvates III and IV. The resulting longer equilibration time of desolvate II might be related to slight accommodations of the crystal lattice induced by the incorporation of the methanol molecules. At higher RVPMeOH values, emodepside II–IV show further exothermic peaks. However, the heat changes occurring at higher RVPMeOH are considerably smaller compared to the first RVPMeOH step. In case of emodepside II, the baselines of the sample curve and the blank curve are offset at higher RVPMeOH. The reason for this offset is that the heat changes associated with the perfusion of the ampoule slightly differ between sample curve and blank curve. For erythromycin A, remarkable enthalpy changes are recorded at 70% and 90% RVPMeOH. Visual examination of the sample after the measurement revealed that the sample has changed into a saturated methanol solution. To verify the incorporation of methanol molecules into the crystal lattices of the desolvates II–IV, the samples were analysed by FT-Raman and FT-IR spectroscopy after having finished the measurements. The obtained spectra were compared to the spectra of the desolvates II–IV and to the spectra of the non-stoichiometric hydrates II–IV recorded at ambient conditions. Both the FT-Raman and FT-IR spectra of the samples exposed to methanol vapours slightly differ from the spectra of the desolvates and hydrates. The fact that only small differences are found between the spectra
100 80 60 40
Erythromycin A Spirapril hydrochloride
0 0
20
40
60
80
100
RHin / % Fig. 4. Calorimetric water vapour sorption isotherms of erythromycin A and spirapril hydrochloride (25 °C).
shapes of the isotherms agree well with the shape of the corresponding gravimetric isotherms reported in literature (Stephenson et al., 1998). However, what differs from the gravimetric results is the relative position of the calorimetric isotherms. The cumulative heat of sorption of erythromycin A is about 5 times the cumulative heat of sorption of spirapril hydrochloride (Fig. 4). In opposition to that, the total amount of water sorbed by erythromycin A (mol mol 1) is twice the amount incorporated by spirapril hydrochloride (Stephenson et al., 1998). The higher heat of sorption (heat of hydration) found for erythromycin A indicates a stronger decrease in energy induced by the presence of the water molecules relative to the decrease in energy of spirapril hydrochloride.
Emodepside I
Emodepside II 100
200
60 40
0
20
Emodepside I, 8.84 mg blank RVPin
-200 0
5
10
15
20
25
Heat flow / µW
80
^exo
400
RVPin MeOH / %
Heat flow / µW
100
^exo
400
80
200
60 40
0
0
30
20
Emodepside II, 8.37 mg blank RVPin
-200 0
10
20
Time / h
30
0 40
Time / h Emodepside IV
Emodepside III
100
100 ^exo
200
60 40
0 Emodepside III, 9.38 mg blank RVPin
-200
5
10
15
Time / h
20
25
30
20
Heat flow / µW
80
0
^exo
400
RVPin MeOH / %
Heat flow / µW
400
RVPin MeOH / %
20
80
200
60 40
0 Emodepside IV, 9.36 mg blank RVPin
-200 0
0
5
10
15
20
25
20
RVPin MeOH / %
Cumulative heat of sorption / kJ*mol
-1
140
0
30
Time / h
Fig. 5. Heat flow curves of modification I and hydrates II–IV of emodepside being first dried at 0% RVP and then exposed to increasing RVPin values of methanol (25 °C). Sample curves are presented in comparison to the blank curve. RVPin is the RVP value of methanol generated inside the sample vessel.
79
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Erythromycin A
Spirapril hydrochloride 600
400
80
^exo
60
200 40 0
20
100 Spirapril hydrochloride, 8.02 mg blank RVPin
400
80
^exo
60
200 40 0
20 0
0 0
10
20
30
RVPin MeOH / %
Erythromycin A, 4.37 mg blank RVPin
Heat flow / µW
100
RVPin MeOH / %
Heat flow / µW
600
0
10
20
30
Time / h
Time / h
Fig. 6. Heat flow curves of erythromycin A dihydrate and spirapril hydrochloride monohydrate being first dried at 0% RVP and then exposed to increasing RVPin values of methanol (25 °C). Sample curves are presented in comparison to the blank curve. RVPin is the RVP value of methanol generated inside the sample vessel.
Raman Intensity (SNV corrected)
reveals that methanol molecules entered the crystal lattices of the desolvates II–IV without changing the crystal structures. Consequently, isomorphic methanolates have formed named methanolate II–IV. Fig. 7 presents the Raman spectra of the methanolates in comparison to the spectra of both desolvates and hydrates (Baronsky et al., 2009b). The CAH and the C@O stretching region are presented since changes are most distinct in these regions. The spectrum of methanolate II differs from the spectra of desolvate II and hydrate II in the relative intensities of the bands present in the CAH stretching region. Differences are also found in the C@O stretching region, e.g. the band present at 1729 cm 1 in the spectra of hydrate II and desolvate II is shifted to 1724 cm 1 in the spectrum of methanolate II. Regarding the CAH stretching region of methanolate III, the band maximum at 2941 cm 1 found in the spectra of desolvate III and hydrate III is absent. In the C@O stretching region, the spectrum of methanolate III more or less corresponds to the spectrum of hydrate III. In case of methanolate IV, the interaction of methanol molecules with C@O groups of the lattice causes the shoulder of the band at 1654 cm 1 present in the spectrum of the desolvate to shift to 1633 cm 1 in the spectrum of methanolate IV. The described differences between the spectra of the methanolates and the spectra of the corresponding hydrates and desolvates represent the changes in the local environment of
0% RH 45% ± 15% RH 90% RVPMeOH II
III
IV 3100
3000
2900
2800
1750
1650
Wavenumber / cm-1 Fig. 7. FT-Raman spectra of the desolvates II–IV of emodepside exposed to methanol vapours compared to the spectra of the desolvates II–IV (0% RH) and the hydrates II–IV (45% ± 15% RH) (Baronsky et al., 2009b).
the CAH and C@O groups caused by the presence of methanol molecules. Similar to the water molecules present in the hydrates, methanol molecules interact with C@O groups of the lattices leading to a shift of C@O stretches towards lower wavenumbers (in comparison to the C@O stretches present in spectra of the corresponding desolvates) (Baronsky et al., 2009b). Regarding methanolate II and IV, the shift of the C@O vibrations is more pronounced than the shift of the C@O vibrations in the spectra of the hydrates II and IV indicating hydrogen bonding to be stronger in the methanolates. For the spectrum of methanolate II, the differences to the spectrum of desolvate II are more pronounced than the differences between the spectra of hydrate II and desolvate II which suggests that the incorporation of the methanol molecules induces more changes in the local environment inside the crystal lattice than the uptake of water molecules. This suggestion is in agreement with the elongated equilibration time found for methanolate II (Fig. 5). To confirm that methanol molecules have entered the lattice of the desolvate of erythromycin A at 10% RVPMeOH, the experiment was stopped at the end of the 10% step and the sample was examined using FT-Raman and FT-IR spectroscopy. The FT-Raman spectrum of the sample more or less corresponds to the spectrum of the desolvate of erythromycin A (data not shown). There might be slight differences between both spectra, however, due to a low signal-to-noise ratio, it is not possible to decide whether differences are significant. In opposition to that, the FT-IR spectrum of the sample shows slight but significant differences in comparison to the spectra of the desolvate and the dihydrate (Fig. 8). For example, differences are readily visible in the C@O stretching region. While all solid forms have a strong band at 1713 cm 1, the sample obtained from the perfusion experiment additionally exhibits an absorption maximum at 1732 cm 1 whereas in case of the desolvate, two bands overlap with the band at 1713 cm 1 being visible as shoulders. Summarising, spectroscopic analysis confirms the transformation of the desolvate into an isomorphic methanolate for emodepside II–IV and erythromycin A which was suggested from perfusion calorimetry. In opposition to the desolvates of emodepside II–IV and erythromycin A, the heat flow curve of the desolvate of spirapril hydrochloride is found to correspond to the blank curve throughout the RVPMeOH range (Fig. 6). This finding suggests that the formation of an isomorphic methanolate does not occur since the structural voids present in the lattice of the desolvate are not sufficient in size to accommodate methanol molecules. After having finished the measurement, the sample was analysed using FT-Raman and FTIR spectroscopy. In both cases, the spectrum of the monohydrate was obtained (data not shown). A possible explanation for this finding is that the desolvate is still present at the end of the experiment and, due to its high hygroscopicity, re-hydrates during the
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Absorbance Units
80
0% RH
45% ± 15% RH
10% RVPMeOH 3500
3000
2500
2000
1500
1000
Wavenumber / cm-1 Fig. 8. FT-IR spectrum of the desolvate of erythromycin A exposed to 10% RVP of methanol compared to the spectra of the desolvate (0% RH) and the hydrate of erythromycin A (45% ± 15% RH).
transfer of the sample into the sample holder (FT-Raman spectroscopy) and the ATR unit (FT-IR spectroscopy), respectively, so that the spectrum of the monohydrate is obtained. This explanation supports the supposition from perfusion calorimetry that methanol molecules do not enter the lattice of the desolvate of spirapril hydrochloride. Regarding emodepside I, the sample curve is similar to the blank curve up to 70% RVPMeOH (Fig. 5). Only minor exothermic peaks are measured compared to the blank (hardly visible) suggesting the adsorption of methanol molecules to the crystal surface. At 90% RVPMeOH, a sharp initial peak occurs followed by a prolonged exothermic event. Analysing the sample after the measurement using FT-Raman and FT-IR spectroscopy reveals that emodepside I has transformed into methanolate IV of emodepside. For emodepside II–IV, the perfusion experiment was repeated twice. The measurements were used to calculate the cumulative heats of sorption ± standard errors as a function of RVPMeOH (Fig. 9). The standard errors displayed in Fig. 9 are small (most of the times hidden behind the symbols used) which reveals that a good reproducibility of the heat flow measurements is not only obtained for water (Fig. 3) but also for methanol as solvent. The calorimetric sorption isotherm of emodepside I shown in Fig. 9 was
Fig. 9. Calorimetric methanol vapour sorption isotherms of emodepside I–IV (25 °C).
determined from the curve presented in Fig. 5 neglecting the fact that equilibrium is not attained at the end of the 90% RVPMeOH step. The inaccuracy of the isotherm at 90% RVPMeOH is accepted, since the isotherm is only used to reveal the differences between its shape and the shape of the isotherms of emodepside II–IV. Fig. 9 shows that the isotherm of emodepside I considerably differs from the isotherms of the non-stoichiometric methanolates II–IV forming isomorphic desolvates. While the isotherm of emodepside I is typical for a solvate formation that is accompanied by structural rearrangement, the isotherms of emodepside II–IV resemble type I isotherms. This type of isotherms is frequently found in non-stoichiometric solvates forming isomorphic desolvates and therefore indicates the presence of this type of solvates (Authelin, 2005).
4. Discussion In each case studied, the transformation of the desolvate into the isomorphic solvate (hydrate or methanolate) is characterized by a substantial exothermic event at low RVP. This thermal activity results in a notable increase of the calorimetric isotherm in the region of low RVP. The considerable heat production reflects the uptake of a substantial number of solvent molecules at low RVP which is characteristic for desolvates. The driving force for solvent incorporation is to reduce the molecular vacuum present in the lattice of the desolvate (Stephenson et al., 1998). The transformation of the desolvate into the isomorphic solvate is not only indicated by the shape of the calorimetric isotherm, but also by the shape of the heat flow curve associated with the solvate formation. In each case studied, the heat flow curve is peak-shaped, i.e. the heat flow rapidly increases and decreases, which reflects the fast uptake of solvent molecules and the resulting short equilibration time which are typical for desolvates forming isomorphic solvates (Stephenson et al., 1998). The rapidness of solvation is due to several factors, one of them being the low activation energy of the process caused by the fact that structural changes are not required. As described in the section on results, the calorimetric equilibration times of the desolvates II–IV of emodepside at 10% RH differ from the corresponding gravimetric equilibration times. One possibility is that the longer equilibration times observed in the calorimetric experiments may be due to thermal activity associated with slight accommodations of molecular groups inside the lattices induced by the incorporation of water molecules. However,
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
it might also be possible that the equilibration process inside the sample vessel is limited by the delivery of water molecules via the wet line. Whether the water supply is the rate limiting step depends on the ratio of the amount of water delivered by the wet line to the amount of water sorbed by the sample. In order for the water supply to not limit the equilibration process, an excess of water has to be supplied relative to the amount of water sorbed (Danforth and David, 2006). Since the amount of water taken up by the sample is proportional to the weight of sample, the weight of sample used for measurement is important. In case of emodepside I, for which the water supply was not expected to limit the equilibration process since the solid form adsorbs only minor amounts of water, a weight of sample that just covers the bottom of the vessel was used. Although a higher weight of sample would be preferable in terms of the signal-to-noise ratio, it would lead to a layer formed by the sample. As the diffusion of the water molecules through the layer might be a limiting factor in the attainment of equilibrium, the amount of sample just covering the bottom of the vessel was used (about 9 mg). For emodepside hydrates II–IV, erythromycin A dihydrate and spirapril hydrochloride monohydrate, the weights of sample used were below the amounts needed to cover the bottom of the vessel in order to prevent the amount of water entering the vessel to limit the equilibration process. The weight of sample used was inversely proportional to the amount of water sorbed by the desolvate at 10% RH (known from gravimetric analysis). Despite of the reduced weight of sample used, it is not sure whether the number of water molecules delivered by the wet line still limits the equilibration of the desolvates in the vessel. Possibilities to investigate whether the water supply is rate limiting are to further reduce sample weights or to increase the total flow rate in order to increase the amount of water being delivered by the wet line. If the process was limited by the amount of water entering the vessel, shorter equilibration times would be observed in both cases but the heat flow curves would still be peak-shaped. However, the main focus of this study was not to determine the exact equilibration times but to obtain the shape of the curves which provides indication about the kinetics of the process observed. Furthermore, both the reduction of the weight of sample and the increase of the total flow rate would result in a lower signal-tonoise ratio. Therefore, the experimental setup was not changed. The investigation of emodepside I and sulfaguanidine anhydrate has shown that perfusion calorimetry is suitable to distinguish the transformation of a desolvate into an isomorphic solvate from both adsorption of solvent molecules and solvate formation that is accompanied by structural changes. Although the adsorption of water molecules to crystal surfaces of sulfaguanidine anhydrate (Fig. 2) as well as the adsorption of water molecules and methanol molecules to the crystal surfaces of emodepside I (Figs. 1 and 5) exhibit sorption kinetics similar to the kinetics found for desolvateisomorphic solvate formation, the heat changes associated are distinctly smaller leading to different calorimetric isotherms. The transformations of sulfaguanidine anhydrate into the monohydrate (Fig. 2) and of emodepside I into methanolate IV (Fig. 5), both representing solvate formation that is accompanied by changes in the crystal structure, differ from the transformation of the various desolvates into isomorphic solvates by the kinetics of the processes. Compared to the rapid increase and decrease of the heat flow observed for desolvate-isomorphic solvate formation, the heat flow associated with the transition of sulfaguanidine anhydrate and emodepside I, respectively, changes less rapidly resulting in an elongated equilibration time. The elongated equilibration time is caused by the structural changes that are not required in case of the transformation of a desolvate into an isomorphic solvate. However, it has to be taken into consideration that the energy barriers to structural rearrangement may (partly) be overcome by milling the sample, for example (generation of defective sites that facili-
81
tate solid-solid transition). In this case, a peak-shaped heat flow curve may be obtained. Consequently, the history of the sample has to be known.
5. Conclusions The investigation of several non-stoichiometric solvates forming isomorphic desolvates has shown that perfusion calorimetry is a valuable tool to indicate the transformation of a desolvate into an isomorphic solvate. This represents a new field of application for the method. The process of the desolvate transforming into an isomorphic solvate is characterized by a notable exothermic, peak-shaped heat flow at low RVP which reflects the rapid uptake of solvent molecules by the desolvate to reduce the molecular vacuum present in the lattice. The method allows the user to both generate the desolvate and expose it to different RVP of solvent in a single experiment. Furthermore, perfusion calorimetry does not require any sample preparation and it can be used for different solvents. Since the presence of a desolvate/isomorphic solvate system is, among other things, deduced from the shape of the heat flow curve, it has to be taken into consideration that the amount of solvent supplied by the wet line may delay the attainment of equilibrium and thus may affect the shape of the curve. Possibilities to prevent the amount of solvent delivered to limit equilibration are the use of a small weight of sample and the increase of the total flow rate used. Although perfusion calorimetry, due to its unspecific nature, is not able to prove the existence of a desolvate transforming into an isomorphic solvate, it is suitable to provide indication within a short period of time.
References Authelin, J.-R., 2005. Thermodynamics of non-stoichiometric pharmaceutical hydrates. Int. J. Pharm. 303, 37–53. Baronsky, J., Bongaerts, S., Traeubel, M., Urbanetz, N., 2009a. Isothermal microcalorimetry perfusion experiments: a method to verify the relative vapour pressure of organic liquids inside the sample vessel. Thermochim. Acta 493, 37–41. Baronsky, J., Bongaerts, S., Traeubel, M., Weiss, H.-C., Urbanetz, N., 2009b. The study of different solid forms of emodepside. Eur. J. Pharm. Biopharm. 71, 88–99. Burger, A., Henck, J.-O., Hetz, S., Rollinger, J.M., Weissnicht, A.A., Stöttner, H., 1999. Energy/temperature diagram and compression behavior of the polymorphs of D-mannitol. J. Pharm. Sci. 89, 457–468. Carvajal, M.T., Staniforth, J.N., 2006. Interactions of water with the surfaces of crystal polymorphs. Int. J. Pharm. 307, 216–224. Danforth, P.M., David, L., 2006. Rapid assessment of the structural relaxation behavior of amorphous pharmaceutical solids: effect of residual water on molecular mobility. Pharm. Res. 23, 2291–2305. Gaisford, S., 2007. Pharmaceutical Isothermal Calorimetry. Informa Healthcare Inc., New York, pp. 33–37. Gift, A.D., Taylor, L.S., 2007. Hyphenation of Raman spectroscopy with gravimetric analysis to interrogate water–solid interactions in pharmaceutical systems. J. Pharm. Biomed. Anal. 43, 14–23. Giron, D., Goldbronn, Ch., Mutz, M., Pfeffer, S., Piechon, Ph., Schwab, Ph., 2002. Solid state characterizations of pharmaceutical hydrates. J. Therm. Anal. Calorim. 68, 453–465. Griesser, U.J., 2006. The importance of solvates. In: Hilfiker, R. (Ed.), Polymorphism in the Pharmaceutical Industry. VILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, pp. 211–233. Lehto, V.-P., Laine, E., 1998. Assessment of physical stability of different forms of cefadroxil at high humidities. Int. J. Pharm. 163, 49–62. Markova, N., Sparr, E., Wadso, L., 2001. On application of an isothermal sorption microcalorimeter. Thermochim. Acta 374, 93–104. Miroshnyk, I., Khriachtchev, L., Mirza, S., Rantanen, J., Heinämäki, J., Yliruusi, J., 2006. Insight into thermally induced phase transformations of erythromycin a dihydrate. Cryst. Growth Des. 6, 369–374. Newman, A.W., Byrn, S.R., 2003. Solid-state analysis of the active pharmaceutical ingredient in drug products. Drug Discov. Today 8, 898–905. Othman, A., Evans, J.S.O., Evans, I.R., Harris, R.K., Hodgkinson, P., 2007. Structural study of polymorphs and solvates of finasteride. J. Pharm. Sci. 96, 1380–1397. Redman-Furey, N., Dicks, M., Bigalow-Kern, A., Cambron, R.T., Lubey, G., Lester, C., Vaughn, D., 2005. Structural and analytical characterization of three hydrates and an anhydrate form of risedronate. J. Pharm. Sci. 94, 893–911.
82
J. Baronsky et al. / European Journal of Pharmaceutical Sciences 44 (2011) 74–82
Rollinger, J.M., Burger, A., 2002. Physico-chemical characterization of hydrated and anhydrous crystal forms of amlodipine besylate. J. Therm. Anal. Calorim. 68, 361–372. Ruth, L.T., Griesser, U.J., Morris, K.R., Byrn, S.R., Stowell, J.G., 2003. X-ray diffraction and solid-state NMR investigation of the single-crystal to single-crystal dehydration of thiamine hydrochloride monohydrate. Cryst. Growth Des. 3, 997–1004. Sheridan, P.L., Graham, B., Storey, D., 1995. Development of a flow microcalorimetry method for the assessment of surface properties of powders. Pharm. Res. 12, 1025–1030.
Stephenson, G.A., Groleau, E.G., Kleemann, R.L., Xu, W., Rigsbee, D.R., 1998. Formation of isomorphic desolvates: creating a molecular vacuum. J. Pharm. Sci. 87, 536–542. Stephenson, G.A., Stowell, J.G., Toma, P.H., Pfeiffer, R.R., Byrn, S.R., 1997. Solid-state investigations of erythromycin a dihydrate: structure, NMR spectroscopy, and hygroscopicity. J. Pharm. Sci. 86, 1239–1244. Timmermann, I.L., Steckel, H., Trunk, M., 2006. Assessing the re-crystallization behaviour of amorphous lactose using the RH-perfusion cell. Eur. J. Pharm. Biopharm. 64, 107–114.