crystals

Article , Stability and Desolvation of the Solvates of Sorafenib Tosylate

1, 1, 1 1 1 1,2 Peng Yang †, Chunlei Qin †, Shichao Du , Lina Jia , Yujia Qin , Junbo Gong and Songgu Wu 1,2,*

1 State Key Laboratory of Chemical Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China 2 Collaborative Innovation Center of Chemical Science and Engineering, Tianjin 300072, China * Correspondence: [email protected]; Tel.: +86-22-2740-5754 Author contributions: P. Y. and C. Q. contributed equally. †  Received: 31 May 2019; Accepted: 11 July 2019; Published: 17 July 2019 

Abstract: In this study, three solvates of sorafenib tosylate were obtained from methanol, ethanol and n-methyl-2-pyrrolidone (NMP) after solvate screening and the effect of solvent on the formation of solvate was analyzed. The solvents with high value of polarity/dipolarity and appropriate donor/acceptor propensity are more likely to form corresponding solvates. The crystal structures of the solvates were elucidated for the first time by using single crystal X-ray diffraction data. The analysis results indicate that methanol solvate and ethanol solvate are isostructural and hydrogen bonds could be formed between solvent molecules and sorafenib tosylate molecules. Hirshfeld surface analysis was used to research the interactions in the solvates, and the results reveal that the H H, ··· C H/H C and O H/ H O contacts play the vital role in molecular packing. In addition, three solvates ··· ··· ··· ··· were characterized by polarized light microscope, powder X-ray diffraction, thermogravimetric analysis and differential scanning . The solvates show different thermodynamic stability in methanol +NMP and ethanol +NMP mixtures. Furthermore, the desolvation of solvates was studied by hot stage microscope and discussed.

Keywords: sorafenib tosylate; solvate; formation; Hirshfeld surface analysis; stability; desolvation

1. Introduction Solvates, a common crystal form in the pharmaceutical industry [1,2], contains active pharmaceutical ingredient (API) molecules and solvent molecules within the crystal structure [3,4], and hydrates are the particular type of solvates with the solvent being water [5]. Solvates might exhibit different physicochemical properties compared with API, such as stability, solubility, dissolution rate, which could affect the bioavailability of pharmaceuticals [6–8]. Solvates typically appear during the purification and processing stages of API, such as mixing, wet granulating and spray drying [9,10]. In addition, solvates may exhibit the faster dissolution rate, higher solubility and better bioavailability than the stable crystal form of API [11–13]. More importantly, the desolvation of solvates could be the effective even the only way to obtain certain polymorphs [13,14]. With the purpose of controlling the formation of solvates and stabilizing the product quality, it is significant to study the solvates from the perspectives of crystal structure, formation mechanism, stability, desolvation behavior, and so on. Sorafenib tosylate (ST, C H ClF N O C H O S, CAS No.475207-59-1, Figure1) has been 21 16 3 4 3· 7 8 3 proved to be a potent inhibitor in inhibiting several receptor tyrosine kinases which are involved in tumor cell proliferation and angiogenesis [15,16]. As a significant anticarcinogen, ST exhibits higher bioavailability and lower side effect compared with sorafenib [17]. However, little study about the crystal structure, stability, formation mechanism, and desolvation behavior has been carried out in the

Crystals 2019, 9, 367; doi:10.3390/cryst9070367 www.mdpi.com/journal/crystals Crystals 2019, 9, x FOR PEER 2 of 14

Crystals 2019, 9, 367 2 of 14 the previous literature [18]. In this work, the crystal structures of three solvates obtained from methanol, ethanol and n-methyl-2-pyrrolidone (NMP) were determined by single crystal X-ray previousdiffraction, literature and the [ 18effect]. In of this solvent work, properties the crystal on structures the formation of three of solvatessolvate was obtained analyzed. from methanol,Hirshfeld surfaceethanol analysis and n-methyl-2-pyrrolidone was used to research (NMP) the interactions were determined in molecular by single packing. crystal Then X-ray the dipolarizedffraction, light and microscope,the effect of solventpowder properties X-ray diffraction, on the formation thermogr of solvateavimetric was analysis analyzed. and Hirshfeld differential surface scanning analysis wascalorimetry used to were research applied the to interactions fully characterize in molecular three solvates. packing. Furthermore, Then the polarized the solubility light microscope, of solvates powderin methanol X-ray +NMP diffraction, and ethanol thermogravimetric +NMP mixtures analysis was determined and differential to investigate scanning the calorimetry thermodynamic were stability.applied to In fully addition, characterize a new NMP three solvate solvates. was Furthermore, found and analyzed. the solubility More of importantly, solvates in methanol as a significant+NMP methodand ethanol to obtain+NMP certain mixtures polymorphs, was determined the desolvati to investigateon behavior the thermodynamic of three solvates stability. was researched In addition, by hota new stage NMP microscope solvate was and founddiscussed. and analyzed. More importantly, as a significant method to obtain certain polymorphs, the desolvation behavior of three solvates was researched by hot stage microscope and discussed.

Figure 1. The chemical structure of sorafenib tosylate. Figure 1. The chemical structure of sorafenib tosylate. 2. Experimental Section 2. Experimental Section 2.1. Materials 2.1. MaterialsST (>0.99 mass fraction) was provided by Huai’an Xinlicheng Chemical Co., Ltd. (Huai’an China), The purityST (>0.99 of ST mass was fraction) determined was by provided high-performance by Huai’an liquid Xinlicheng chromatography. Chemical Co., Methanol, Ltd. (Huai’an ethanol, China),n-propanol, The isopropanol,purity of ST n-butanol,was determined 2-butanol, by isobutanol,high-performance acetone, liquid acetonitrile, chromatography. ethyl acetate, Methanol, n-hexane, ethanol,cyclohexane, n-propanol, dichloromethane, isopropanol, toluene n-butanol, and NMP 2-butanol, are analytical isobutanol, reagents acetone, (>0.995 massacetonitrile, fraction) ethyl and wereacetate, purchased n-hexane, from cyclohexane, Tianjin Yuanli dichloromethane, Chemical Co. (Tianjin toluene China). and NMP These are chemicals analytical were reagents used without(>0.995 massfurther fraction) purification. and were purchased from Tianjin Yuanli Chemical Co. (Tianjin China). These chemicals were used without further purification. 2.2. Solvate Screening and Preparation

2.2. SolvateThe solvate Screening screening and Preparation of ST was investigated by selecting the most commonly used solvents. Excessive amount of ST and appropriate pure solvent were added into the EasyMax vessel (Mettler The solvate screening of ST was investigated by selecting the most commonly used solvents. Toledo, Zurich, Switzerland) with an accuracy of 0.01 C. After that, the EasyMax vessel was maintained Excessive amount of ST and appropriate pure solvent◦ were added into the EasyMax vessel (Mettler at the desired temperature under the agitation speed of 200 rpm for 24 hours. Then the obtained solids Toledo, Zurich, Switzerland) with an accuracy of 0.01 °C. After that, the EasyMax vessel was were collected by filtration and characterized. maintained at the desired temperature under the agitation speed of 200 rpm for 24 hours. Then the obtained2.3. Crystal solids Structure were collected Determination by filtration and characterized.

2.3. CrystalThe solvent Structure evaporation Determination method was employed to obtain the single crystals of methanol solvate and ethanol solvate of ST. Excessive amount of ST was added into methanol/ethanol. After that, the supernatantThe solvent wasevaporation withdrawn method by a syringewas employed with an to organic obtain filterthe single membrane crystals (0.22 of methanolµm) and injectedsolvate andinto ethanol a 10 mL solvate beaker. of Then ST. Excessive the beaker amount was sealed of ST with was plastic added film into and methanol/e placed intothanol. a vacuum After that, drying the supernatant was withdrawn by a syringe with an organic filter membrane (0.22 µm) and injected into oven at 25 ◦C. The crystals of the solvate with suitable size for single crystal X-ray diffraction (SCXRD) werea 10 mL collected beaker. after Then several the beaker days. was The sealed NMP with solvate plastic was film obtained and placed in the into 100 a vacuum mL Easymax drying vessel oven byat 25 °C. The crystals of the solvate with suitable size for single crystal X-ray diffraction (SCXRD) were cooling . Certain amount of ST and NMP was added into vessel at 40 ◦C. After the collected after several days. The NMP solvate was obtained in the 100 mL Easymax vessel by cooling dissolution of ST, the solution was cooled down to 5 ◦C at a rate of 0.1 ◦C/min, then the corresponding crystallization. Certain amount of ST and NMP was added into vessel at 40 °C. After the dissolution of solvate with appropriate size for SCXRD was obtained. The Rigaku-Rapid II diffractometer with a Mo ST, the solution was cooled down to 5 °C at a rate of 0.1 °C/min, then the corresponding solvate with Kα radiation source (λ = 0.71073 Å) was used to collect the SCXRD data of solvates.

Crystals 2019, 9, 367 3 of 14

2.4. Characterization CrystalsSeveral 2019, analytical9, x FOR PEER methods have been used to characterize the solvates of ST, including polarized3 of 14 light microscope (PLM), powder X-ray diffraction (PXRD), thermogravimetric analysis (TGA) and appropriate size for SCXRD was obtained. The Rigaku-Rapid II diffractometer with a Mo Kα radiation differential scanning calorimetry (DSC). source (λ = 0.71073 Å) was used to collect the SCXRD data of solvates. The morphologies of solvates were observed by a polarized light microscope (Olympus BX51, Tokyo, Japan) with a magnification of 100 . 2.4. Characterization × The PXRD patterns were collected by a D/MAX-2500 diffractometer (Rigaku, Tokyo, Japan) with Several analytical methods have been used to characterize the solvates of ST, including polarized Cu Kα radiation (1.54056 Å) over a diffraction angle (2θ) range of 2–40◦ at a scanning rate of 8◦/min. light microscope (PLM), powder X-ray diffraction (PXRD), thermogravimetric analysis (TGA) and Thermogravimetric analysis experiments were carried out on the TGA 1/SF (Mettler Toledo, differential scanning calorimetry (DSC). Zurich, Switzerland) from 25 C to 250 C at a heating rate of 10 C /min under nitrogen gas purge. The morphologies of solvates◦ were◦ observed by a polarized◦ light microscope (Olympus BX51, The DSC curves were acquired by the DSC 1/500 calorimeter (Mettler Toledo, Zurich, Switzerland) Tokyo, Japan) with a magnification of 100×. from 25 C to 250 C at a rate of 10 C/min under protection of nitrogen gas. The◦ PXRD patterns◦ were collected◦ by a D/MAX-2500 diffractometer (Rigaku, Tokyo, Japan) with 1 CuThe Kα radiation Fourier transform (1.54056 Å) infrared over a diffraction (FTIR) spectra angle were(2θ) range collected of 2–40° in the at a range scanning of 4000 rate toof 4008°/min. cm− by 1 using aThermogravimetric Nicolet-Magna FT-IR analysis 750 spectrometerexperiments were (Waltham, carried MA, out USA on the ) with TGA the 1/SF resolution (Mettler of Toledo, 4 cm− . Zurich, Switzerland) from 25 °C to 250 °C at a heating rate of 10 °C /min under nitrogen gas purge. 2.5. Solubility Profile of Solvates in Binary Solvent Mixtures The DSC curves were acquired by the DSC 1/500 calorimeter (Mettler Toledo, Zurich, Switzerland) fromTo 25 have °C to a suitable250 °C at understanding a rate of 10 °C/m onin theunder solution-mediated protection of nitrogen phase gas. transformation of ST in binary solventThe mixtures, Fourier thetransform solubility infrared of stable (FTIR) solvates spectra inwere methanol collected+ NMPin the andrange ethanol of 4000+ toNMP 400 cm mixtures−1 by −1 wasusing determined a Nicolet-Magna by an Agilent FT-IR 1200750 spectrometer HPLC system (Wal (Agilenttham, MA, Technologies, USA ) with Santa the resolution Clara, CA, of USA)4 cm at. 5 ◦C. The chromatographic column used in this work is an Agilent Eclipse XDB-C18 column (250 4.6 mm, × 5 µ2.5.m). Solubility The composition Profile of Solvates of the in mobile Binary Solvent phase isMixtures acetonitrile /water (82.5/17.5, v/v) with a flow rate of 1.5 mLTo /havemin fora suitable ST and understanding methanol/water on the (40 solution-mediated/60, v/v) with a flow phase rate transformation of 1.0 mL/min of ST for in NMP. binary The detectionsolvent wavelengthsmixtures, the ofsolubility ST and of NMP stable are solvates 265 nm in and methanol 210 nm, + respectively.NMP and ethanol + NMP mixtures wasExcess determined ST was by added an Agilent into 1200 binary HPLC solvent system mixtures (Agilent (about Technologies, 30 mL) inSanta a 50 Clara, mL Erlenmeyer CA, USA) at flask 5 and°C. this Theflask chromatographic was kept at column 5 ◦C by used a thermostatic in this work is shaker an Agilent (Tianjin Eclipse Ounuo XDB-C18 Instrument column Co.,(250 Tianjin× 4.6 China)mm, 5 with µm). an The accuracy composition of 0.1 of◦ theC. Aftermobile being phase stirred is acetonitrile/water for 24 hours, (82.5/17.5, the solution v/v) was with kept a flow static rate for 2 hoursof 1.5 tomL/min allow for undissolved ST and methanol/water particles to precipitate. (40/60, v/v) Afterwith a that, flow the rate supernatant of 1.0 mL/min was for withdrawn NMP. The by a syringedetection equipped wavelengths with of a 0.22ST andµm NMP filter are membrane 265 nm and and 210 injected nm, respectively. into a pre-weighted volumetric flask immediately.Excess ST An was electronic added into analytical binary solvent balance mixtures (ML204, (about Mettler 30 mL) Toledo, in a 50 Zurich mL Erlenmeyer Switzerland) flask with and an accuracythis flask of was 0.0001 kept g at was 5 °C used by a to thermostatic weigh the volumetricshaker (Tianjin flask Ounuo with saturatedInstrument solution. Co., Tianjin Then China) the solution with an accuracy of 0.1 °C. After being stirred for 24 hours, the solution was kept static for 2 hours to allow was diluted and analyzed by the HPLC method. Each experiment was repeated three times and the undissolved particles to precipitate. After that, the supernatant was withdrawn by a syringe equipped average value was adopted to calculate the mole fraction solubility. Finally, the solid remained in with a 0.22 µm filter membrane and injected into a pre-weighted volumetric flask immediately. An Erlenmeyer flask was analyzed by PXRD and TGA to determine its crystal form. electronic analytical balance (ML204, Mettler Toledo, Zurich Switzerland) with an accuracy of 0.0001 g Before measuring the concentration of stable solvate, the calibration curves of ST and NMP were was used to weigh the volumetric flask with saturated solution. Then the solution was diluted and determinedanalyzed by by the HPLC HPLC method method. and Each are experiment shown in wa Figures repeated2. The three experiment times and was the repeated average value three was times 2 toadopted ensure the to calculate repeatability the mole and frac accuracy.tion solubility. The calibration Finally, the curves solid of remained ST and NMPin Erlenmeyer are linear flask with wasR = 2 0.9999analyzed and Rby =PXRD0.9994, and respectively. TGA to determine its crystal form.

(a) (b)

FigureFigure 2. 2.Peak Peak areaarea versusversus concentration calibration calibration curves: curves: (a ()a ST;) ST; (b ()b NMP.) NMP.

Crystals 2019, 9, 367 4 of 14

2.6. Desolvation Behavior of Solvates The desolvation behavior of solvates was observed by hot stage microscope (HSM, Olympus BX-51, Tokyo, Japan) from 25 ◦C to 250 ◦C at a heating rate of 5 ◦C /min.

3. Results and Discussion

3.1. Solvate Screening The property parameters of solvents and crystal forms obtained in this work are given in Table1[ 19]. It can be seen that the solvates were obtained from methanol, ethanol and NMP, which can be labeled as SMe,SEt and SNMP, respectively. The result shows that the solvents play a significant role in the formation of solvate. To further understand the effect of solvent on the formation of solvate, and on the other hand, provide a suitable method for solvate screening, the properties of solvent including hydrogen bond donor propensity (Σα), hydrogen bond acceptor propensity (Σβ) and polarity/dipolarity (π*), were presented and discussed [20,21]. Table1 shows that it is easy to form the corresponding solvate from solvents with a high value of π*. This can be concluded from the fact that the values of Σα and Σβ of the alcohols in Table1 are close, but only methanol and ethanol with high values of π* can form corresponding solvates. This result indicates that the polarity/dipolarity of solvent has a vital effect on the formation of solvate, and that solvents with strong polarity/dipolarity can form corresponding solvates easily. However, it should be mentioned that some solvents with high values of π* are not necessarily able to form corresponding solvate, such as acetonitrile and dichloromethane. The reason for this phenomenon may be that the values of Σα and Σβ of these solvents are obviously lower compared with NMP, resulting in weak interactions with ST.

Table 1. The property parameters of solvents.

Solvent Σα Σβ π* Form methanol 0.43 0.47 0.60 YES ethanol 0.37 0.48 0.54 YES n-propanol 0.37 0.48 0.52 NO isopropanol 0.33 0.56 0.48 NO n-butanol 0.37 0.48 0.47 NO 2-butanol 0.33 0.56 0.40 NO isobutanol 0.37 0.48 0.40 NO ethyl acetate 0.00 0.48 0.55 NO acetone 0.04 0.49 0.71 NO acetonitrile 0.07 0.32 0.75 NO n-hexane 0.00 0.00 -0.04 NO cyclohexane 0.00 0.00 0.00 NO dichloromethane 0.10 0.05 0.82 NO toluene 0.00 0.14 0.54 NO NMP 0.00 0.77 0.92 YES

3.2. Crystal Structure of Solvate

The crystal structures of SMe,SEt and SNMP were determined for the first time by SCXRD, and the crystallographic data are summarized in Table2. The crystal structures are shown in Figure3. The crystallographic data show that SMe and SEt are both monoclinic systems with P21/m space group, while

SNMP belongs to triclinic system with P1 space group. Table2 illustrates that S Me and SEt have similar unit cell parameters, which are completely different from SNMP. Analysis of crystal structures indicates that SMe and SEt are isostructural. The geometrical parameters of hydrogen bonds and hydrogen bonds in three solvates are presented in Table3 and Figure4, respectively. Taking S Et as an example, hydrogen bonds could be formed between p-toluenesulfonic acid molecules and sorafenib molecules, and each ethanol molecule could form a hydrogen bond with a ST molecule in a crystal lattice. It is Crystals 2019, 9, x FOR PEER 5 of 14

Space group P21/m P21/m P1 Temperature/K 293(2) 293(2) 293(2) Wavelength/Å 0.71073 0.71073 0.71073 a/Å 11.684(2) 11.745(2) 9.1884(18) Crystals 2019, 9, 367b/Å 19.405(4) 19.412(4) 10.565(2) 5 of 14 c/Å 14.564(3) 14.650(3) 21.769(4) α/º 90.000 90.000 87.27(3) worth mentioning that the hydrogen bonds formed by methanol molecules and ST molecules cannot be β/º 111.53(3) 110.57(3) 81.17(3) displayed visually in Table3 and Figure4 for the reason that the methanol molecules in crystal lattice γ/º 90.000 90.000 82.21(3) are disordered. The existence of hydrogen bonds formed by methanol molecules and ST molecules Volume/ų 3071.7(11) 3127.1(11) 2068.2(7) can be verified by TGA/DSC curves of S , which are present and are discussed in a later section. Z 4 Me 4 2 In addition, it can be seen from the crystal structure of S that a single S molecule is formed θ range/° 3.01–25.50 NMP3.08–25.50 NMP 3.07–25.50 from an ST molecule and two NMP molecules. Furthermore, the hydrogen bond can be formed by one -14 ≤ h ≤ 14 -14 ≤ h ≤ 14 -11 ≤ h ≤ 11 NMP molecule and a sorafenib molecule. Another NMP molecule will not provide the possibility for Index ranges -23 ≤ k ≤ 23 -22 ≤ k ≤ 23 -12 ≤ k ≤ 12 the formation of a hydrogen bond located in the cavity. -16 ≤ l ≤ 17 -17 ≤ l≤ 17 -26 ≤ l ≤ 26 Density/g▪cm-3 1.447 1.451 1.341 Table 2. Crystallographic data of SMe,SEt and SNMP. Rint 0.0909 0.0657 0.0477 ParameterR1[I＞2σ(I)] 0.0567 SMe S0.0577Et SNMP0.0708

FormulawR2 C29H280.1575ClF3N4 O7SC30H30ClF0.17183N4O 7SC38H42ClF0.25843N6O8 S FormulaGoodness-o weightf-fit 669.061.024 683.091.015 835.291.017 CrystalCCDC system No. monoclinic1505023 monoclinic1505022 triclinic1505021 Space group P21/m P21/m P1 TheTemperature crystal structures/K of SMe 293(2), SEt and SNMP were determined 293(2) for the first 293(2)time by SCXRD, and the crystallographicWavelength /dataÅ are summarized0.71073 in Table 2. Th 0.71073e crystal structures are 0.71073 shown in Figure 3. The a/Å 11.684(2) 11.745(2) 9.1884(18) crystallographic data show that SMe and SEt are both monoclinic systems with P21/m space group, b/Å 19.405(4) 19.412(4) 10.565(2) 1 while SNMP cbelongs/Å to triclinic 14.564(3)system with P space 14.650(3) group. Table 2 illustrates 21.769(4) that SMe and SEt have similar unitα cell/º parameters, which 90.000 are completely different 90.000 from SNMP. Analysis 87.27(3) of crystal structures indicates thatβ/º SMe and SEt are 111.53(3)isostructural. The geometrical 110.57(3) parameters 81.17(3)of hydrogen bonds and hydrogen bondsγ/º in three solvates 90.000 are presented in Table 90.000 3 and Figure 4, respectively. 82.21(3) Taking SEt as an 3 example,Volume hydrogen/Å bonds could3071.7(11) be formed between 3127.1(11) p-toluenesulfonic acid 2068.2(7) molecules and sorafenib Z 4 4 2 molecules, and each ethanol molecule could form a hydrogen bond with a ST molecule in a crystal θ range/◦ 3.01–25.50 3.08–25.50 3.07–25.50 lattice. It is worth mentioning14 thath the14 hydrogen 14bondsh formed14 by methanol11 h 11molecules and ST − ≤ ≤ − ≤ ≤ − ≤ ≤ moleculesIndex cannot ranges be displayed23 visuallyk 23 in Table 3 22andk Figure23 4 for the 12reasonk 12that the methanol − ≤ ≤ − ≤ ≤ − ≤ ≤ 16 l 17 17 l 17 26 l 26 molecules in crystal lattice are− disordered.≤ ≤ The existence− ≤ ≤ of hydrogen bonds− ≤ formed≤ by methanol Density/g cm 3 1.447 1.451 1.341 molecules and· ST− molecules can be verified by TGA/DSC curves of SMe, which are present and are Rint 0.0909 0.0657 0.0477 discussed in a later section. In addition, it can be seen from the crystal structure of SNMP that a single R1[I>2σ(I)] 0.0567 0.0577 0.0708 NMP S moleculewR2 is formed from an0.1575 ST molecule and two 0.1718 NMP molecules. Furthermore, 0.2584 the hydrogen bond Goodness-of-fitcan be formed by one NMP 1.024 molecule and a sorafenib 1.015 molecule. Another 1.017 NMP molecule will not provideCCDC the No. possibility for the 1505023 formation of a hydrogen 1505022 bond located in 1505021 the cavity.

Crystals 2019, 9, x FOR PEER 6 of 14

(a)

(b)

FigureFigure 3. Crystal3. Crystal structures structures of solvates:of solvates: (a) ( balla) ball and and stick stick plots; plots; (b) ( packingb) packing of unitof unit cell. cell. Subscript Subscript 1, 21, 2

andand 3 represent 3 represent SMe S,SMeEt, SandEt and SNMP SNMP,, respectively. respectively.

Table 3. Hydrogen bonds of SMe, SEt and SNMP.

Solvate D−H···A d(D−H)/Å d(H···A)/Å d(D···A)/Å ∠(DHA)/°

N(1)-H(1A)···O(4) 0.86 2.06 2.895 162.3 SMea N(2)-H(2A)···O(4) 0.86 2.26 3.064 155.0

N(1)-H(1A)···O(5) 0.86 2.06 2.892 163.0 SEtb O(7)-H(7A)···O(6) 0.82 2.02 2.771 151.5

SNMPc N(1)-H(1A)···O(5) 0.86 2.07 2.823(4) 146.0 a.Symmetry transformations used to generate equivalent atoms: −x − 1, −y, −z. b.Symmetry transformations used to generate equivalent atoms: (1) x + 2, y, z + 1. (2) −x + 7/2, y − 1/2, −z + 3/2. (3) x+1/2, -y+1/2. c.Symmetry transformations used to generate equivalent atoms: (1) −x + 1, −y + 2, -z + 1. (2) x + 1, y, z + 1.

(a) (b)

(c)

Figure 4. The hydrogen bonds in SMe (a), SEt (b) and SNMP (c).

3.3. Hirshfeld Surface Analysis As a visual and efficient way of analyzing the interactions in molecular packing and comparing the differences in crystal structures [22–24], Hirshfeld surfaces were used to investigate the solvates

Crystals 2019, 9, x FOR PEER 6 of 14

(b)

Figure 3. Crystal structures of solvates: (a) ball and stick plots; (b) packing of unit cell. Subscript 1, 2

and 3 represent SMe, SEt and SNMP, respectively.

Table 3. Hydrogen bonds of SMe, SEt and SNMP.

Solvate D−H···A d(D−H)/Å d(H···A)/Å d(D···A)/Å ∠(DHA)/°

N(1)-H(1A)···O(4) 0.86 2.06 2.895 162.3 Crystals 2019, 9, 367 6 of 14 SMea N(2)-H(2A)···O(4) 0.86 2.26 3.064 155.0

N(1)-H(1A)···O(5)Table 3. Hydrogen0.86 bonds of SMe 2.06,SEt and SNMP. 2.892 163.0 SEtb O(7)-H(7A)···O(6) 0.82 2.02 2.771 151.5 Solvate D H A d(D H)/Å d(H A)/Å d(D A)/Å ∠(DHA)/◦ − ··· − ··· ··· SNMPc N(1)-H(1A)···O(5) 0.86 2.07 2.823(4) 146.0 a N(1)-H(1A) O(4) 0.86 2.06 2.895 162.3 SMe ··· a. N(2)-H(2A) O(4) 0.86 2.26 3.064 155.0b. Symmetry transformations used··· to generate equivalent atoms: −x − 1, −y, −z. Symmetry N(1)-H(1A) O(5) 0.86 2.06 2.892 163.0 transformationsS b used to generate··· equivalent atoms: (1) x + 2, y, z + 1. (2) −x + 7/2, y − 1/2, −z + 3/2. (3) Et O(7)-H(7A) O(6) 0.82 2.02 2.771 151.5 c. x+1/2, -y+1/2. Symmetryc transformations··· used to generate equivalent atoms: (1) −x + 1, −y + 2, -z + 1. SNMP N(1)-H(1A) O(5) 0.86 2.07 2.823(4) 146.0 (2) x + 1, y, z + 1. ··· a.Symmetry transformations used to generate equivalent atoms: x 1, y, z. b.Symmetry transformations used to generate equivalent atoms: (1) x + 2, y, z + 1. (2) x + 7/2, y− −1/2,− z −+ 3/2. (3) x+1/2, -y+1/2. c.Symmetry transformations used to generate equivalent atoms: (1)− x + 1, y −+ 2, -z−+ 1. (2) x + 1, y, z + 1. − −

(a) (b)

(c)

FigureFigure 4.4. The hydrogen bondsbonds inin SSMeMe ((aa),), S SEtEt (b(b) )and and S SNMPNMP (c).(c ).

3.3.3.3. Hirshfeld Hirshfeld Surface Surface Analysis Analysis AsAs a a visual visual and eefficientfficient wayway ofof analyzing analyzing the the interactions interactions in in molecular molecular packing packing and and comparing comparing the thediff differenceserences in crystal in crystal structures structures [22– [22–24],24], Hirshfeld Hirshfeld surfaces surfaces were were used used to investigate to investigate the solvates the solvates of ST. The Hirshfeld surfaces of SMe,SEt and SNMP were obtained by using CrystalExplorer 17.5 [25], and the corresponding 2D fingerprint plots are shown in Figure5. In addition, the percentage contributions of different interactions to the Hirshfeld surface are presented in Figure6. Due to the presence of N-H O ··· hydrogen bonds in S , the 2D fingerprint plot shows two sharp spikes, indicating strong O H/H O Et ··· ··· interactions. Similar situations could be found in the other two fingerprint plots. From Figure6, it can be found that the H H, C H/H C and O H/H O contacts make the greatest contribution to the ··· ··· ··· ··· ··· Hirshfeld surface because of the H atoms on the surface and the hydrogen bonds [26]. By comparing the three histograms in Figure6, it can also be found that the percentage contribution of H H in S is ··· Me much smaller than the values of SEt and SNMP. This may be caused by the disorder in SMe. In addition to the H H, C H and O H contacts, the H F/F H contact also makes a certain contribution to the ··· ··· ··· ··· ··· Hirshfeld surface. Crystals 20192019,, 99,, xx FORFOR PEERPEER 77 ofof 1414 of ST. The Hirshfeld surfaces of SMe,, SSEtEt andand SSNMP werewere obtainedobtained byby usingusing CrystalExplorerCrystalExplorer 17.517.5 [25],[25], and the corresponding 2D fingerprint plots are shown in Figure 5. In addition, the percentage contributions of different interactions to the Hirshfeld surface are presented in FigureFigure 6.6. DueDue toto thethe presence of N-H···O hydrogen bonds in SEtEt,, thethe 2D2D fingerprintfingerprint plotplot showsshows twotwo sharpsharp spikes,spikes, indicatingindicating strongstrong O···H/H···OO···H/H···O interactions.interactions. SimilarSimilar situationssituations could be found in the other two fingerprint plots. From Figure 6, it can be found that the H···H, C···H/H···C··H/H···C andand O···H/H···OO···H/H···O contactscontacts makemake thethe greatestgreatest contribution to the Hirshfeld surface because of thethe HH atomsatoms onon thethe surfacesurface andand thethe hydrogenhydrogen bondsbonds [26].[26]. ByBy comparingcomparing thethe threethree histogramshistograms inin FigureFigure 6,6, itit cancan alsoalso bebe foundfound thatthat thethe percentagepercentage contribution of H···H in SMe isis muchmuch smallersmaller thanthan thethe valuesvalues ofof SSEtEt andand SSNMP.. ThisThis maymay bebe causedcaused byby thetheCrystals disorderdisorder2019, 9, 367inin SSMe.. InIn additionaddition toto thethe H···H,H···H, C···HC···H andand O···HO···H contacts,contacts, thethe H···F/F···HH···F/F···H contactcontact7 alsoalso of 14 makes a certain contribution toto thethe HirshfeldHirshfeld surface.surface.

((a)) ((b))

((c))

Figure 5. 2D fingerprint plots of SMe (a), SEt (b) and SNMP (c). FigureFigure 5.5. 2D2D fingerprintfingerprint plotsplots ofof SSMeMe ((a),), S SEt ((bb)) and and S SNMPNMP (c().c ).

Figure 6. Percentage contribution of different interactions to the Hirshfeld surface. Figure 6. Percentage contribution of different interactions to the Hirshfeld surface. 3.4. CharacterizationFigure 6. Percentage contribution of different interactions to the Hirshfeld surface.

The PLM images of solvates are shown in Figure7, indicating that S Me,SEt and SNMP exhibit diamond-like, plate-like and rod-like morphologies, respectively. The experimental and calculated PXRD patterns of solvates are shown in Figure8. It can be seen that the experimental data show good consistency with the calculated results. It can also be found that SMe and SEt exhibit identical peaks, suggesting that SMe and SEt are isostructural, which is consistent with the analysis of the crystal structures described above. Crystals 20192019,, 99,, xx FORFOR PEERPEER 88 ofof 1414

3.4. Characterization

The PLM images of solvates are shown in Figure 7, indicating that SMe,, SSEtEt andand SSNMP exhibitexhibit diamond-like, plate-like and rod-likelike morphologies,morphologies, respectively.respectively. The experimental and calculated PXRD patterns of solvates are shown in Figure 8. It can be seen that the experimental data show good consistency with the calculated results. It can also be found that SMe andand SSEtEt exhibitexhibit identicalidentical peaks,peaks, suggestingsuggestingCrystals 2019, 9thatthat, 367 SSMe andand SSEtEt are isostructural, which is consistent withwith thethe analysisanalysis ofof thethe crystalcrystal8 of 14 structuresstructures describeddescribed above.above.

((a)) ((b)) ((c))

FigureFigure 7. 7. MicroscopeMicroscope images images of of S SMeMe ((a),), SSEtEt (((bb),),), SS SNMPNMP ((cc().).c ).

FigureFigure 8. 8. PXRDPXRD patterns patterns of SSMeMe (((aa),),), SS SEtEtEt ((b(b)) ) andand and SS SNMPNMP ((cc().).c ).SubscriptSubscript Subscript 11 1andand and 22 2 representrepresent represent experimentalexperimental experimental andand and calculatedcalculated results,results, respectively.respectively.

TheThe DSC DSC and and TGA TGA curves of three solvates are given in Figure 99.. TheThe solventsolvent stoichiometrystoichiometry isis verifiedverified by the weight loss inin TGATGA curves.curves. ThenThen SSMeMe/S/S/SEtEtEt isisis confirmedconfirmed confirmed toto to bebe be thethe the monosolvatemonosolvate monosolvate andand and thethe SSNMPNMP moleculemoleculemolecule isis formedformedformed by byby a sorafenibaa sorafenibsorafenib molecule moleculemolecule and twoandand NMPtwotwo NMP molecules,NMP molecules,molecules, suggesting suggestingsuggesting that the obtainedthatthat thethe obtainedresults are results consistent are consistent with the crystal with the structures. crystal structures. It can be seen It can from be seen Figure from9 that Figure S Me and9 that SEt SexhibitMe andand SalmostEtEt exhibit the almost same DSC the andsame TGA DSC curves. and TGA The curves. desolvation The ofdesolvation SMe/SEt occurs of SMe at/S/S temperaturesEtEt occursoccurs atat temperaturestemperatures ranging from rangingrangingabout 100 fromfrom◦C to aboutabout 150 ◦ C100100 with °C°C a toto loss 150150 of °C°C solvent. withwith aa Then lossloss theofof solvent.solvent. second endothermic ThenThen thethe secondsecond peak isendothermicendothermic considered to peakpeak be the isis consideredmelting peak to of be the the product after peak desolvation. of the product Finally, after SMe/S Etdesolvation.shows an endothermicFinally, SMe/S/S peakEtEt showsshows at 235 an◦anC endothermicin DSC curve peak with at a rapid 235 °C mass in lossDSC in curve TGA with curve a because rapid mass of the loss decomposition. in TGA curve It isbecause worth notingof the decomposition.that the desolvation It is worth temperature noting that of S Methe/S desolvationEt is much higher temperature than the of boilingSMe/S/SEtEt isis point muchmuch of higherhigher solvent, thanthan which thethe boilingis due to point the hydrogen of solvent, bonding which interactions is due to betweenthe hydrogen alcohol bonding molecules interactions and ST molecules. between Startingalcohol moleculesfrom 106 ◦ C,and SNMP ST molecules.shows an endothermic Starting from peak 106 in °C, DSC SNMP curve showsshows with anan distinct endothermicendothermic tailing, peakpeak which inincorresponds DSCDSC curvecurve withwith distinct the mass tailing, loss in which TGA curve corresponds before 200 with◦C. the Then mass NMP loss reaches in TGA boiling curve point before at 200 204 °C.◦C, Then resulting NMP in reachesreachesrapid mass boilingboiling loss pointpoint in the atat TGA 204204 curve °C,°C, resultingresulting and an endothermicinin rapirapid mass peak loss in in the the DSC TGA curve. curve Finally, and an S endothermicNMP starts to peakdecompose in the DSC at 236 curve.◦C. Finally, SNMP startsstarts toto decomposedecompose atat 236236 °C.°C.

Crystals 2019 9 Crystals 2019, 9, x 367 FOR PEER 9 9of of 14 14

(a) (b)

(c) (d)

FigureFigure 9. 9.DSC DSC and and TGA TGA curves curves of of S MeSMe( a(a),), S SEtEt ((bb),), SSNMPNMP ((cc)) and and S SNMP-MNMP-M (d().d ).

3.5. Solubility Solubility of of Solvate Solvate To investigate the thermodynamic stability of of solvates of of ST, the solubility of stable solvates in methanol + NMPNMP and and ethanol ethanol ++ NMPNMP mixtures mixtures was was determined. determined. It It is is worth worth emphasizing emphasizing that that a a new new NMP solvate (labeled as SNMP-M) )was was obtained obtained in in binary binary solvent solvent mixtures mixtures and and the the PXRD PXRD patterns patterns of of SNMP, ,SSNMP-MNMP-M andand ST STwere were compared compared in inFigure Figure 10. 10 According. According to to Figure Figure 10, 10 ,it it can can be be seen seen that that the the new new NMP solvate, SNMP-M, has, has a adifferent different PXRD PXRD pattern pattern compared compared with the SNMP andand ST. ST. Compared Compared to to the the SNMP andand ST, ST, the the PXRD PXRD pattern pattern of ofSNMP-M SNMP-M showsshows distinctly distinctly different different diffraction diffraction peaks peaks at 2θ at = 26.7°,θ = 7.6°,6.7◦, 9.2°,7.6◦, 9.211.9°,◦, 11.9 14.6°,◦, 14.6 17.1°,◦, 17.1 17.8°.◦, 17.8 In◦ .addition, In addition, DSC DSC and and TGA TGA curves curves of ofNMP NMP so solvateslvates are are presented presented in in Figure9 9.. ItIt cancan bebe seenseen that that the the starting starting temperatures temperatures of of desolvation desolvation for for S NMPSNMPand and S SNMP-MNMP-M isis 106 106 °C◦C and and 124 °C,◦C, respectively. The weight loss ofof SNMP duringduring the the desolvation desolvation is is ab aboutout 24.0%, 24.0%, which which means means that that thethe stoichiometry stoichiometry (solvent: API) API) is is 2:1, 2:1, and and this this can can be be confirmed confirmed by its crystal structure. The The weight weight lossloss of of S SNMP-M duringduring desolvation desolvation isis about about 10.2%, 10.2%, and and the the stoichiometry stoichiometry is is 1:1. 1:1. Moreover, FTIR spectroscopy was used used to to furthe furtherr confirm confirm the the differences differences of of two two NMP NMP solvates, solvates, and the FTIR patterns of two NMP solvates are show shownn in Figure 11.11. Compared Compared with with the the FTIR FTIR pattern pattern 1 1 1 of S NMP, ,it it can can be be seen seen that that S SNMP-MNMP-M showsshows additional additional peaks peaks at at1686 1686 cm cm−1, −1645, 1645 cm cm−1,1486− , 1486 cm−1 cmand− 1132and 1 cm1132−1 cmbecause− because of the of differences the differences of hydrogen of hydrogen bond bond network. network. It is It isworth worth mentioning mentioning that that two two NMP NMP solvates obtained obtained in in this this work were compared with three previously reported NMP solvates in the patent literature [27]. [27]. Based Based on on the the PXRD PXRD pattern patternss and and TGA TGA curves curves in in this this paper paper and and patent, patent, SNMP SNMP is sameis same with with one one of the of theNMP NMP solvates solvates in the in patent. the patent. However, However, SNMP-M SNMP-M showsshows different diff diffractionerent diffraction peaks atpeaks 2θ = at6.7°, 2θ 7.6°,= 6.7 12.4°,◦, 7.6◦ 17.1°,, 12.4 ◦17.8°, 17.1 in◦, the 17.8 PXRD◦ in the pattern PXRD compared pattern compared with the withreported the reportedNMP solvates. NMP Insolvates. addition, In addition,the starting the startingdesolvation desolvation temperatures temperatures of SNMP-M of S(aboutNMP-M 124(about °C) 124in DSC◦C)in curve DSC are curve quite are differentquite diff erentfrom fromthe results the results of NMP of NMP solvates solvates (106 (106 °C,114◦C,114 °C ◦andC and 142 142 °C)◦C) reported reported in in patent patent [27]. [27]. Therefore, SNMP-M isis further further confirmed confirmed as as a anew new NMP NMP solvate. solvate.

Crystals 2019, 9, 367 10 of 14

CrystalsCrystals 2019 2019, ,9 9, ,x x FOR FOR PEER PEER 1010 ofof 1414 Crystals 2019, 9, x FOR PEER 10 of 14

FigureFigureFigure 10. 10. 10.PXRD PXRD PXRD patterns patterns patterns of of of ST—Form ST—Form ST—Form II I (( a(aa),),), SS SNMP-MNMP-M ( (bb()b) and )and and S SNMPNMP S ( (cc).). ( c). Figure 10. PXRD patterns of ST—Form I (a), SNMP-MNMP-M (b) and SNMPNMP (c).

FigureFigureFigure 11. 11. 11.FTIR FTIR FTIR patterns patterns patterns of of of S S NMP-MSNMP-MNMP-M ( (a(a),a), ),S SNMP SNMPNMP ( (bb)() andb and) and ST ST ( ST(cc).). ( c). Figure 11. FTIR patterns of SNMP-M (a), SNMP (b) and ST (c).

FromFromFrom Figure Figure Figure 12 12a, 12a,a, it it it can can can be be be seen seen seen thatthat that thethe the stablestable stable solvate solvate is is S SMeMe when whenwhen x xBxB B≤ ≤ 0.1973, 0.1973,0.1973, and and and this this this changes changes changes From Figure 12a, it can be seen that the stable solvate is SMe when xB ≤ ≤0.1973, and this changes intointointo S NMP-M SSNMP-MNMP-M when when 0.2272 0.22720.2272 ≤≤ xxxBB B ≤≤ 0.3122;0.3122;0.3122; whenwhen when xxBB x≥≥B 0.4353,0.4353,0.4353, SSNMPNMP S isNMP is thetheis stablestable the stable solvate.solvate. solvate. InIn addition,addition, In addition, thethe into SNMP-M when 0.2272 ≤≤ xB ≤ ≤0.3122; when xB ≥ 0.4353,≥ SNMP is the stable solvate. In addition, the theexperimentalexperimental experimental resultsresults results inin in FigureFigure Figure 12b12b 12 indicatebindicate indicate thatthat that SSEtEt S existsEtexistsexists onlyonly only inin in purepure pure ethanol.ethanol. ethanol. InIn Inethanolethanol ethanol ++ NMPNMP+ NMP experimental results in Figure 12b indicate that SEt exists only in pure ethanol. In ethanol + NMP mixtures,mixtures,mixtures, when when when 0.0444 0.0444 0.0444 ≤ x≤ Bx xBB ≤ ≤0.2824 0.2824 0.2824and and andx x BxBB ≥ ≥ 0.3747, 0.3747, ST STST transforms transforms into into S SSNMP-MNMP-M and andand S SNMPNMP SNMP, ,respectively. respectively., respectively. mixtures, when 0.0444≤ ≤ xB≤ ≤ 0.2824 and xB≥ ≥ 0.3747, ST transforms into SNMP-M and SNMP, respectively. TheTheThe experimental experimentalexperimental results resultsresults reveal revealreveal thatthatthat thethe NMP NMP content contentcontent hashas has aa a significantsignificant significant effecteffect effect onon on thethe the formationformation formation ofof of The experimental results reveal that the NMP content has a significant effect on the formation of solvate,solvate, whichwhich isis thethe samesame asas thethe phenomenonphenomenon obobservedserved inin anhydrous/hydrateanhydrous/hydrate systemssystems inin previousprevious solvate,solvate, which which is is the the same same as as the the phenomenonphenomenon ob observedserved in in anhydrous/hydrate anhydrous/hydrate systems systems in inprevious previous studiesstudies [28,29]. [28,29]. The The interactions interactions between between ST ST molecules molecules and and NMP NMP molecules molecules increase increase with with the the increase increase studiesstudies [28 [28,29].,29]. The The interactions interactions betweenbetween ST molecules molecules and and NMP NMP molecules molecules increase increase with with the the increase increase ofof NMP NMP content content in in methanol methanol + + NMP NMP and and ethanol ethanol + + NMP NMP mixtures, mixtures, then then resulting resulting in in the the formation formation of of ofof NMP NMP content content in in methanol methanol+ + NMPNMP andand ethanol ++ NMPNMP mixtures, mixtures, then then resulting resulting in inthe the formation formation of of NMPNMP solvate, solvate, accordingly. accordingly. NMPNMP solvate, solvate, accordingly. accordingly.

FigureFigure 12. 12. Mole Mole fraction fraction solubility solubility of of solvates solvates in in ( (aa)) methanol methanol + + NMP NMP and and ( (bb)) ethanol ethanol + + NMP NMP mixtures mixtures FigureFigure 12. 12.Mole Mole fraction fraction solubility solubility ofof solvatessolvates in ( a) methanol methanol ++ NMPNMP and and (b ()b ethanol) ethanol + NMP+ NMP mixtures mixtures atat 5 5 °C: °C: ■ ■, ,S SMeMe/S/SEtEt; ;● ●, ,S SNMP-MNMP-M; ;▲ ▲, ,S SNMPNMP. . ■ Me Et NMP-M ▲ NMP atat 5 ◦5C: °C: ,S, MeS //SSEt; ●,,S S NMP-M; ; N, S,SNMP. .

Crystals 2019, 9, x FOR PEER 11 of 14

3.6. Desolvation of Solvates As shown in Figure 13, the desolvation process of solvates was studied visually using HSM. It can be seen from Figure 13a that SMe crystal showed no change before desolvation. Then solvent molecules escaped from crystals with the increase of temperature, resulting in the declining of crystal transparency. The opaque region expanded from the edge to the center of the crystals for the reason that desolvation process was prone to proceed on the edge of crystals [5]. After that, the crystals became almost opaque. Finally, crystals began to melt, and the transparency increased along with fusion. SEt and SNMP showed similar desolvation behavior to what SMe presented. A point worth Crystals 2019, 9, 367 11 of 14 emphasizing is that some breakages appeared on the crystal surface of SEt, but this phenomenon could not be observed on the crystal surface of SMe and SNMP. New cavities may be formed in SEt crystal 3.6.during Desolvation the desolvation of Solvates process, resulting in breakages on the crystal surface. As shown in Figure 14, the desolvation products of the solvates were analyzed and compared with ST polymorphs. From FigureAs 14b shown and in14c, Figure we can 13 ,see the that desolvation the PXRD process pattern of of solvates experimental was studied ST shows visually good usingconsistence HSM. Itwith can the be calculated seen from results. Figure 13Compareda that SMe withcrystal the PXRD showed pattern no change of ST (Form before I), desolvation. the desolvation Then products solvent molecules escaped from crystals with the increase of temperature, resulting in the declining of crystal of SNMP, SEt and SMe show the similar patterns. This means that the desolvation products of the SNMP, transparency. The opaque region expanded from the edge to the center of the crystals for the reason that SEt and SMe are Form I of ST, which may be because Form I is the most stable polymorph among the desolvationST polymorphs process [30,31]. was Additionally, prone to proceed the results on the edgeof thermal of crystals analysis [5]. of After solvates that, theare crystalssummarized became in almostTable 4 opaque.in order Finally,to concisely crystals study began the todesolvation melt, and process. the transparency increased along with fusion.

(a) (b)

(c)

FigureFigure 13. 13.HSM HSM images images of of S SMeMe ((aa),), SSEtEt (b) and S NMP (c().c).

SEt and SNMP showed similar desolvation behavior to what SMe presented. A point worth emphasizing is that some breakages appeared on the crystal surface of SEt, but this phenomenon could not be observed on the crystal surface of SMe and SNMP. New cavities may be formed in SEt crystal during the desolvation process, resulting in breakages on the crystal surface. As shown in Figure 14, the desolvation products of the solvates were analyzed and compared with ST polymorphs. From Figure 14b,c, we can see that the PXRD pattern of experimental ST shows good consistence with the calculated results. Compared with the PXRD pattern of ST (Form I), the desolvation products of SNMP,

SEt and SMe show the similar patterns. This means that the desolvation products of the SNMP,SEt and SMe are Form I of ST, which may be because Form I is the most stable polymorph among the ST polymorphs [30,31]. Additionally, the results of thermal analysis of solvates are summarized in Table4 in order to concisely study the desolvation process. Crystals 2019, 9, x FOR PEER 12 of 14

Table 4. Thermal analysis results of three solvates.

Solvate SMe SEt SNMP Solvent Methanol Ethanol N-methyl-2-pyrrolidone Stoichiometry 1:1 1:1 1:2 Theoretical weight loss/% 4.8 6.7 23.7 Experimental weight loss/% 4.6 7.0 24.0

Desolvation onset Ton/℃ 102 98 106 Crystals 2019, 9, 367 12 of 14 Desolvation product Form I Form I Form I

Figure 14.14. PXRDPXRD patterns patterns of ST—Formof ST—Form I: (b) I: experimental (b) experime andntal (c) and calculated; (c) calculated; and desolvation and desolvation products ofproducts SNMP of (a), SNMP SEt (a), (d) andSEt (d) SMe and (e). SMe (e).

4. Conclusions Table 4. Thermal analysis results of three solvates. In this work, threeSolvate solvates were obtained SMe from methanol,SEt ethanol SandNMP NMP by the solvate screening of ST. The effectSolvent of solvent on Methanolthe solvate formation Ethanol was N-methyl-2-pyrrolidone analyzed, indicating that the hydrogen bond donorStoichiometry propensity, hydrogen 1:1bond acceptor 1:1 propensity and 1:2polarity/dipolarity have Theoretical weight loss/% 4.8 6.7 23.7 an integrated effect on the formation of solvate. The crystal structures of the three solvates were Experimental weight loss/% 4.6 7.0 24.0 elucidated for the first time by using SCXRD data, suggesting that SMe and SEt are monoclinic systems Desolvation onset Ton/°C 102 98 106 with P21/m spaceDesolvation group, and product SNMP belongs to Form triclinic I system Form with I P1 space Formgroup. I Moreover, SMe and SEt are isostructural and solvent molecules form hydrogen bonds with ST molecules in the crystal 4.lattice Conclusions of solvates. Hirshfeld surface analysis was chosen to research the interactions in the solvates, and the results revealed that the H···H, C···H/H···C and O···H/ H···O contacts make the greatest contributionIn this work, to the threemolecular solvates packing. were obtained from methanol, ethanol and NMP by the solvate screening of ST. The effect of solvent on the solvate formation was analyzed, indicating that the SMe, SEt and SNMP were characterized by PLM, PXRD, DSC and TGA. The results of thermal hydrogenanalysis could bond be donor divided propensity, into three hydrogen procedures: bond desolvation acceptor propensity of solvate, and melting polarity of/dipolarity the product have after an integrateddesolvation eff andect ondecomposition the formation of of the solvate. product The after crystal de structuressolvation. of The the solubility three solvates of stable were solvates elucidated in formethanol the first + timeNMP by and using ethanol SCXRD + NMP data, mixtures suggesting at 5 that °C SwasMe andmeasured SEt are to monoclinic research the systems thermodynamic with P21/m P spacestability group, of solvates, and SNMP indicatingbelongs that to triclinicthe solubility system ofwith stable 1solvatesspace group. increases Moreover, with the S Meincreaseand S ofEt arethe isostructuralNMP mole fraction and solvent and the molecules NMP content form hydrogenhas a signif bondsicant witheffect ST on molecules the formation in the of crystal solvate. lattice A new of solvates.NMP solvate Hirshfeld was found surface and analysis confirmed. was HSM chosen analysis to research was used the interactionsto study the indesolvation the solvates, of solvates, and the results revealed that the H H, C H/H C and O H/ H O contacts make the greatest contribution to and the results show that S···Me, SEt··· and S···NMP could ···transform··· into Form I after desolvation. the molecular packing. AuthorSMe Contributions:,SEt and SNMP Conceptualization,were characterized Y.Q. and by PLM,S.W.; methodolo PXRD, DSCgy, P.Y., and C.Q. TGA. and The Y.Q.; results software, of thermal L.J. and analysisP.Y.; validation, could beS.W. divided and J.G..; into formal three analysis, procedures: S.D..; data desolvation curation, ofY.Q., solvate, C.Q. and melting S.D.; investigation, of the product P.Y. after and desolvationC.Q.; writing—original and decomposition draft preparatio of then, P.Y. product and C.Q..; after writing—review desolvation. The and solubility editing, P.Y., of stableS.D., L.J., solvates S.W. and in methanolJ.G.; supervision,+ NMP S.W. and and ethanol J.G.; project+ NMP administration mixtures at, 5S.W.◦C and was J.G.; measured funding to ac researchquisition, the S.W. thermodynamic and J.G. stability of solvates, indicating that the solubility of stable solvates increases with the increase of the NMP mole fraction and the NMP content has a significant effect on the formation of solvate. A new NMP solvate was found and confirmed. HSM analysis was used to study the desolvation of solvates, and the results show that SMe,SEt and SNMP could transform into Form I after desolvation.

Author Contributions: Conceptualization, Y.Q. and S.W.; methodology, P.Y., C.Q. and Y.Q.; software, L.J. and P.Y.; validation, S.W. and J.G..; formal analysis, S.D..; data curation, Y.Q., C.Q. and S.D.; investigation, P.Y. and C.Q.; writing—original draft preparation, P.Y. and C.Q..; writing—review and editing, P.Y., S.D., L.J., S.W. and J.G.; supervision, S.W. and J.G.; project administration, S.W. and J.G.; funding acquisition, S.W. and J.G. Funding: The authors are grateful to the financial support of Major National Science and Technology Projects 2017ZX07402003, Innovative Group Project 21621004 and China Postdoctoral Science Foundation 2018M641651. Crystals 2019, 9, 367 13 of 14

Conflicts of Interest: The authors declare no conflict of interest.

References

1. Aitipamula, S.; Chow, P.S.; Tan, R.B.H. Solvates and polymorphic phase transformations of 2-chloro-4-nitrobenzoic acid. CrystEngComm 2011, 13, 1037–1045. [CrossRef] 2. Fujii, K.; Aoki, M.; Uekusa, H. Solid-state hydration/dehydration of erythromycin A investigated by ab initio powder X-ray diffraction analysis: Stoichiometric and nonstoichiometric dehydrated hydrate. Cryst. Growth Des. 2013, 13, 2060–2066. [CrossRef] 3. Yuan, L.; Lorenz, H. Solvate formation of bis (demethoxy) curcumin: Screening and characterization. Crystals 2018, 8, 407. [CrossRef] 4. Bechtloff, B.; Nordhoff, S.; Ulrich, J. Pseudopolymorphs in industrial use. Cryst. Res. Technol. 2001, 36, 1315–1328. [CrossRef] 5. Li, L.; Du, R.; Du, Y.; Zhang, C.; Guan, S.; Dong, C.; Zhang, L. Febuxostat-minoxidil salt solvates: Crystal structures, characterization, interconversion and solubility performance. Crystals 2018, 8, 85. [CrossRef] 6. Ferrari, E.S.; Davey, R.J.; Cross, W.I.; Gillon, A.L.; Towler, C.S. Crystallization in polymorphic systems: The solution-mediated transformation of β to α glycine. Cryst. Growth Des. 2003, 3, 53–60. [CrossRef] 7. Zhao, J.; Wang, M.; Dong, B.; Feng, Q.; Xu, C. Monitoring the polymorphic transformation of imidacloprid using in situ FBRM and PVM. Org. Process Res. Dev. 2013, 17, 375–381. [CrossRef] 8. Zhu, M.; Wang, Y.; Li, F.; Bao, Y.; Huang, X.; Shi, X.; Hao, H. Theoretical model and experimental investigations on solution-mediated polymorphic transformation of theophylline: From polymorph I to polymorph II. Crystals 2019, 9, 260. [CrossRef]

9. Berzi¯ n, š, A.; Skarbulis, E.; Rekis, T.; Actin, š, A. On the formation of droperidol solvates: Characterization of structure and properties. Cryst. Growth Des. 2014, 14, 2654–2664. [CrossRef] 10. Chavez, K.J.; Guevara, M.; Rousseau, R.W. Characterization of solvates formed by sodium naproxen and an homologous series of alcohols. Cryst. Growth Des. 2010, 10, 3372–3377. [CrossRef] 11. Thakuria, R.; Nangia, A. Olanzapinium salts, isostructural solvates, and their physicochemical properties. Cryst. Growth Des. 2013, 13, 3672–3680. [CrossRef] 12. Perlovich, G.L.; Blokhina, S.V.; Manin, N.G.; Volkova, T.V.; Tkachev, V.V. Polymorphs and solvates of felodipine: Analysis of crystal structures and thermodynamic aspects of sublimation and solubility processes. CrystEngComm 2012, 14, 8577–8588. [CrossRef] 13. Zhang, Q.; Lu, L.; Dai, W.; Mei, X. Polymorphism and isomorphism of Huperzine A solvates: Structure, properties and form transformation. CrystEngComm 2014, 16, 1919–1926. [CrossRef] 14. Vangala, V.R.; Chow, P.S.; Tan, R.B.H. The solvates and salt of antibiotic agent, nitrofurantoin: Structural, thermochemical and desolvation studies. CrystEngComm 2013, 15, 878–889. [CrossRef] 15. Ferrara, N. VEGF and the quest for tumour angiogenesis factors. Nat. Rev. Cancer 2002, 2, 795–803. [CrossRef] 16. Sherman, J.H.; Kirzner, J.; Siu, A.; Amos, S.; Hussainic, I. Sorafenib tosylate as a radiosensitizer in malignant astrocytoma. J. Clin. Neurosci. 2014, 21, 131–136. [CrossRef][PubMed] 17. Porta, C.; Paglino, C.; Imarisio, I.; Ferraris, E. Sorafenib tosylate in advanced kidney cancer: Past, present and future. Anti-Cancer Drugs 2009, 20, 409–415. [CrossRef] 18. Ravikumar, K.; Sridhar, B.; Bhujanga, R.A.K.S.; Reddy, P.M. Sorafenib and its tosylate salt: A multikinase inhibitor for treating cancer. Acta Crystallogr. C 2011, 67, o29–o32. [CrossRef] 19. Abraham, M.H. Hydrogen bonding. 31. Construction of a scale of solute effective or summation hydrogen-bond basicity. J. Phys. Org. Chem. 1993, 6, 660–684. [CrossRef] 20. Marcus, Y. The properties of organic liquids that are relevant to their use as solvating solvents. Chem. Soc. Rev. 1993, 22, 409–416. [CrossRef] 21. Gu, C.; Li, H.; Gandi, R.B.; Raghavan, K. Grouping solvents by statistical analysis of solvent property parameters: Implication to polymorph screening. Int. J. Pharm. 2004, 283, 117–125. [CrossRef][PubMed] 22. Mckinnon, J.J.; Jayatilaka, D.; Spackman, M.A. Towards quantitative analysis of intermolecular interactions with Hirshfeld surfaces. Chem. Commun. 2007, 37, 3814–3816. [CrossRef] 23. Spackman, M.A.; Jayatilaka, D. Hirshfeld surface analysis. CrystEngComm 2009, 11, 19–32. [CrossRef] Crystals 2019, 9, 367 14 of 14

24. Hussain, E.; Kumar, R.; Choudhary, M.I.; Yousuf, S. Crystal engineering of naturally occurring seselin to obtain with enhanced anti-leishmanial activity, Hirshfeld surface analysis, and computational insight. Cryst. Growth Des. 2018, 18, 4628–4636. [CrossRef] 25. Turner, M.J.; McKinnon, J.J.; Wolff, S.K.; Grimwood, D.J.; Spackman, P.R.; Jayatilaka, D.; Spackman, M.A. CrystalExplorer model energies and energy frameworks: extension to metal coordination compounds, organic salts, solvates and open-shell systems. IUCrJ 2017, 4, 547–587. 26. Jia, L.; Yu, Y.; Wang, M.; Ouyang, H.; Gong, J. Preparation and characterization for multicomponent crystals of the antidiabetic drug gliquidone based on crystal engineering. CrystEngComm 2019, 21, 1617–1625. [CrossRef] 27. An, X.; Shen, S.; Wang, W. The preparation methods and properties of the polymorphs of sorafenib tosylate solvates. CN103896833A, 27 December 2012. 28. Qu, H.; Louhi-Kultanen, M.; Rantanen, J.; Kallas, J. Solvent-mediated phase transformation kinetics of an anhydrate/hydrate system. Cryst. Growth Des. 2006, 6, 2053–2060. [CrossRef] 29. Wang, X.; Wu, S.; Dong, W.; Gong, J. In situ monitoring of the solvent-mediated transformation of cefadroxil DMF solvate into monohydrate. Org. Process Res. Dev. 2013, 17, 1110–1116. [CrossRef] 30. Grunenberg, A.; Lenz, J. Thermodynamically stable form of a tosylate salt. PCT WO2006/034797, 6 April 2006. 31. Scarpitta, F.; Biljan, T. A process for the preparation of solid state forms of sorafenib besylate. PCT WO2012/071425, 31 May 2012.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).