Acetylsalicylic acid (ASA or aspirin) powder with a mass fraction purity of >99% was obtained from Sigma (Sigma-Aldrich, Gillingham, UK). Absolute ethanol was purchased from Fisher Scientific (Loughborough, UK).
Solubility Studies
Eleven combinations of binary solvent mixtures were prepared using suitable volume fractions of water-ethanol increasing consecutively from 0.0 to 1.0 ethanol: water, the neat solvents. A shake-flask technique with spectrophotometry was used to determine drug solubility in these solvents. Excess amounts of drug were added to the mixtures to allow saturation concentration to be reached. The eleven samples were then incubated at 37°C and 25°C in a thermostatic water bath shaker (Cambridge, Crafton) at atmospheric pressure while constantly shaking at a speed of 200 rpm, for a minimum of 72 h (the preliminary results showed that 72 h was sufficient to reach equilibrium condition).
The suspensions were allowed to settle for 1 h, and then the supernatants were filtered to remove the excess solid using a syringe-driven filter unit (pore size 0.20 μm). A volume of either 0.2 mL or 1 mL (depending on the concentration of drug in the saturated sample) of the filtrate from each sample was immediately diluted quantitatively using an appropriate amount of the same ethanol: water solvent mixture. Dilutions in the range of 10–10,000 times were made depending on the concentration of the drug in the filtered solutions. A Knauer HPLC instrument (Berlin, Germany) composed of a K-1001 HPLC pump, a BioTech. degasser, a sample loop (20 μL), and a K-2600 ultraviolet detector was used for the determination of acetylsalicylic acid and salicylic acid. The chromatographic data processing was performed by employing the Chromgate software (version 3.1). The separation was performed by a stationary phase of C18 XBridge analytical column (5 μm × 250 mm × 4.6 mm) from Waters Co. (Ireland), and the mobile phase consisted of phosphoric acid:acetonitrile:water (2:400:600 v/v). The freshly prepared mobile phase was filtered using a vacuum filter system equipped with a 0.45-μm membrane filter (Millipore Corp., Billerica, Massachusetts) and degassed by ultrasonic for 15 min. Chromatography was run at 25°C by pumping the mobile phase at a flow rate of 1.5 mL/min. The UV detector recorded the column effluent at 254 nm (36). The calibration curve between the peak area and the concentration in the range of 10–1000 mg.L−1 for acetylsalicylic acid is Y = 3336.8 CASA + 181,813 and in the range of 0.5–20 mg.L−1 for salicylic acid is Y = 39,158 CSA + 14,652. The solubility data are obtained from the interpolation of these plots.
Density Measurement of the Solutions
In order to measure the true density of each saturated solution, 1 mL of each filtrate was weighed accurately using an accurate balance with a precision of 0.1 mg (Sartorius, Ireland). The true density values are needed to convert molarity and mole fraction. Three repeats of the procedure were completed for all seven drugs at 37°C and 25°C, and the results averaged. It should be noted that the density of the saturated solution of acetylsalicylic acid is overestimated since there is a considerable amount of salicylic acid which increases the density value.
Differential Scanning Calorimetry
A differential scanning calorimeter (DSC7, Mettler Toledo, Switzerland) was employed to investigate the thermal behavior (enthalpy and melting point) of acetylsalicylic acid before and after the solubility test. This information allows us to identify whether a different polymorphic form is produced during the solubility test. The samples studied through the DSC machine were the pure drug sample and the samples obtained after equilibration with 0, 0.5, and 1 ethanol fractions at both 25 and 37°C temperatures. The acetylsalicylic acid particles left in the solubility test were collected and dried. The dried samples were placed in DSC pans and heated between 25 and 300°C at a scanning rate of 10°C/min under nitrogen gas (50 mL/min). After obtaining the DSC traces for each sample, the melting points and enthalpies of fusion were calculated by the software provided.
Fourier Transform Infrared Spectroscopy
In order to explore any changes in the structure of the solid extracted from the solubility test as a result of possible hydrolysis of acetylsalicylic acid and precipitation of any by-products, FT-IR was employed (Perkin Elmer’s Spectrum One, Shelton, CT, USA). Briefly, methanol was used to clean the instrument to remove any residual matter left on the apparatus, followed by placing a few milligrams of each of the separated solid particles after the solubility test. The sample was pressed with a pressure of 100 bar followed by scanning the sample three times over a range of 4000 cm−1 to 500 cm−1 to obtain spectra with appropriate resolution.
X-ray Powder Diffraction
The XRPD patterns were obtained for all samples including original acetylsalicylic acid using a D2 Phaser diffractometer (Bruker AXS GmbH, Karlsruhe, Germany). All samples produced were scanned in Bragg-Brentano geometry, over a scattering (Bragg, 2θ) angle range from 5 to 50°, in 0.02° steps at 1.5° min−1(37). Microsoft Excel was used to analyze and plot the collected XRPD patterns. The crystallinity of the samples was also determined to elucidate the effect of the type of solvent (water and ethanol) on the crystallinity of the recovered acetylsalicylic acid samples. The area under the curve for the “distinctive crystalline peaks” at 7.8 and 15.6 2θ angles was measured for each XRPD diffractogram and used in the determination of crystallinity (%) using the equation 1(38).
$$\mathrm{Crystallinity}=\frac{\mathrm{Area}\ \mathrm{of}\ \mathrm{cryatlline}\ \mathrm{peaks}}{\mathrm{Area}\ \mathrm{of}\ \mathrm{all}\ \mathrm{peaks}\ \left(\mathrm{crystalline}+\mathrm{amorphous}\right)}\times 100$$
(1)
Mathematical Modeling of the Solubility Data
The solubility values determined for acetylsalicylic acid in ethanol + water mixtures are correlated and back-calculated utilizing the mathematical cosolvency models such as the Yalkowsky, Jouyban-Acree, modified Wilson, and PC-SAFT models, and details for each studied model are discussed below.
Yalkowsky Model
The Yalkowsky model was employed to express the natural logarithm of solubility in a mixture of solvent + cosolvent (39).
$$\ln {C}_m={f}_1\ln {C}_1+{f}_2\ln {C}_2$$
(2)
where C1 and C2 are solubility data in mono-solvents 1 and 2 in molar fraction unit, xm is the solubility of the drug in the solvent mixture, and f1 and f2 are volume fractions of solvents 1 and 2 in the absence of the drug. After modification of Eq. 2 (i.e., substitution of f2 with (1−f1) and subsequent rearrangements), Eq. 3 can be obtained as (40):
$$\ln {C}_m=\ln {C}_2+\left(\ln \frac{C_1}{C_2}\right){f}_1=\ln {C}_2+\sigma .{f}_1$$
(3)
the σ is the model constant. A linear relationship has been shown between the logarithm of octanol-to-water partition coefficient of the solute (log P) and σ as below (41):
$$\sigma =M.\log P+N$$
(4)
where M and N are the cosolvent constants. After replacing Eq. 4 in Eq. 3, a predictive mathematical equation is attained (41).
$$\ln {C}_m=\ln {C}_2+{f}_1\left(M.\log P+N\right)$$
(5)
By employing M and N values obtained from the literature for used cosolvent (ethanol) and log P of a drug, the solubility of the drug in the solvent mixture can be computed only using solubility data in water.
Jouyban-Acree Model
The Jouyban-Acree model as a simple linear cosolvency model used for binary mixtures of solvents at various temperatures can be presented by Eq. 6(17):
$$\ln {C}_{m,T}={f}_1.\ln {C}_{1,T}+{f}_2.\ln {C}_{2,T}+\frac{f_1.{f}_2}{T}\sum_{i=0}^2{J}_i.{\left({f}_1-{f}_2\right)}^i$$
(6)
Ji is the model parameter calculated using linear regression of (lnCm, T − f1. ln C1, T − f2. ln C2, T) vs\(\frac{f_1.{f}_2}{T}\), \(\frac{f_1.{f}_2\left({f}_1-{f}_2\right)}{T}\), and \(\frac{f_1.{f}_2{\left({f}_1-{f}_2\right)}^2}{T}\) and other model parameters have the same meanings as those of the above model.
The Modified Wilson Model
In addition to linear models employed for fitting and prediction of solubility values, the non-linear model of modified Wilson is also utilized for modeling the solubility data in the solvent mixtures at isothermal conditions. The equation is as (42):
$$-\ln {C}_m=1-\frac{f_1\left[1+\ln {x}_1\right]}{f_1+{f}_2{\lambda}_{12}}-\frac{f_2\left[1+\ln {x}_2\right]}{f_1{\lambda}_{21}+{f}_2}$$
(7)
λ12 and λ21 are the model constants computing using nonlinear analysis.
PC-SAFT Model
The perturbed chain SAFT equation of state (EOS) or PC-SAFT was first proposed and developed by Gross and Sadowski in 2001 (43) as an alternative to the original version of SAFT derived by Chapman et al.(44). The residual molar Helmholtz energy of the PC-SAFT (ares) obtained by the Helmholtz energy contributions from the reference system hard chain (ahc), dispersion force (adisp), and hydrogen bonding (aassoc) is obtained as follows:
$${a}^{\mathrm{res}}=a-{a}^{\boldsymbol{ideal}}={a}^{\mathrm{hc}}+{a}^{\boldsymbol{disp}}+{a}^{\boldsymbol{assoc}}$$
(8)
In PC-SAFT, pure components can be described using five pure-component parameters: (i)m, number of segments per chain; (ii)σ, diameter of each segment in Angstrom (Å); (iii)ε, energy parameter for each segment in Joules (J); (iv) κAiBi, effective volume of the association (Å3); (v)εAiBi, energy parameter of the association (bar.l/mol); (vi)NumAss, number of association sites (\({N}_i^{\boldsymbol{assoc}}\)). The parameters of each component are reported in Table I(43, 45). The interaction parameters for binary systems (ethanol + water), (acetylsalicylic acid + water) and (acetylsalicylic acid + ethanol), kij, for the purely predictive model were set to zero.
Table I Pure Component Parameters for the Substances The fugacity coefficient for component k (ϕk) and compressibility factor (z) using the PC-SAFT EOS are computed as follows:
$$\mathit{\ln}{\phi}_k={a}^{\mathrm{res}}+\left(z-1\right)+{\left(\frac{\partial {a}^{\mathrm{res}}}{\partial {x}_k}\right)}_{T,V,{X}_{i\ne k}}-\sum_{j=1}^N\left[{X}_j{\left(\frac{\partial {a}^{\mathrm{res}}}{\partial {X}_k}\right)}_{T,V,{X}_{i\ne j}}\right]- lnz$$
(9)
$$z=1+\rho {\left(\frac{\partial {a}^{\mathrm{res}}}{\partial \rho}\right)}_{T,{X}_i}$$
(10)
where ρ is the molar density. Using PC-SAFT, the activity coefficients are calculated from the fugacity coefficients through Eq. 11:
$${\gamma}_i=\frac{\phi_i}{\phi_i^0}$$
(11)
where ϕi and \({\phi}_i^0\) are the fugacity coefficients of component i in the mixture and that of the pure compound, respectively. In solid-liquid equilibria, the solid solubility in the liquid phase is calculated according to the following expression (46):
$$\mathit{\ln}{x}_i=\frac{\Delta {H}_m}{R}\left(\frac{1}{T_m}-\frac{1}{T}\right)-\mathit{\ln}{\gamma}_i$$
(12)
where xi and γi represent the solubility and activity coefficient of compound i. In this study, the activity coefficient of compound i (γi) was determined via Eq. 11. Since the activity coefficient depends on solubility in mole fraction (xi), solubility must be determined from the iterations with Eq. 12. In the equation above, ΔHm and Tm represent fusion enthalpy and melting point temperature, respectively, and their values are presented in Table II.
Table II Melting Points and Enthalpy of Fusion for Drug Crystals Obtained From Different Concentrations of Ethanol + Water at 25 and 37°C All explained models are correlated to the measured solubility values of acetylsalicylic acid, and the mean relative deviation (MRD%) (Eq. 13) is used to obtain the model’s accuracy.
$$MRD\%=\frac{100}{N}\sum \left(\frac{\left|\boldsymbol{Calculatedvalue}-\boldsymbol{Observedvalue}\right|}{\boldsymbol{Observedvalue}}\right)$$
(13)
N is the number of data points in each set.