The effect of tailor-made additives on crystal growth of methyl paraben: Experiments and modelling

Author’s Accepted Manuscript The Effect of Tailor-made Additives on Crystal Growth of Methyl paraben: Experiments and Modelling Zhihui Cai, Yong Liu, Yang Song, Guoqiang Guan, Yanbin Jiang www.elsevier.com/locate/jcrysgro

PII: DOI: Reference:

S0022-0248(16)30971-X http://dx.doi.org/10.1016/j.jcrysgro.2016.12.103 CRYS23962

To appear in: Journal of Crystal Growth Received date: 7 July 2016 Revised date: 20 December 2016 Accepted date: 27 December 2016 Cite this article as: Zhihui Cai, Yong Liu, Yang Song, Guoqiang Guan and Yanbin Jiang, The Effect of Tailor-made Additives on Crystal Growth of Methyl paraben: Experiments and Modelling, Journal of Crystal Growth, http://dx.doi.org/10.1016/j.jcrysgro.2016.12.103 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The Effect of Tailor-made Additives on Crystal Growth of Methyl paraben: Experiments and Modelling

Zhihui Cai, Yong Liu, Yang Song, Guoqiang Guan, Yanbin Jiang*

School of Chemistry and Chemical Engineering, South China University of Technology, Guangzhou 510640, China *

Corresponding

author.

Yanbin

JiangTel.:+86-20-8711-2051;

[email protected] (Y. Jiang)

ABSTRACT In this study, methyl paraben (MP) was selected as the model component, and acetaminophen (APAP), p-methyl acetanilide (PMAA) and acetanilide (ACET), which share the similar molecular structure as MP, were selected as the three tailor-made additives to study the effect of tailor-made additives on the crystal growth of MP. HPLC results indicated that the MP crystals induced by the three additives contained MP only. Photographs of the single crystals prepared indicated that the morphology of the MP crystals was greatly changed by the additives, but PXRD and single crystal diffraction results illustrated that the MP crystals were the same polymorph only with different crystal habits, and no new crystal form was found compared with other references. To investigate the effect of the additives on the crystal growth, the interaction between additives and facets was discussed in detail using the DFT methods and MD simulations. The results showed that APAP, PMAA and ACET would be selectively adsorbed on the growth surfaces of the crystal facets, which induced the change in MP crystal habits.

Keywords: A1. Adsorption; A1. Computer simulation; A1. Impurities; A1. Single crystal growth; B1. Aromatic compounds

1. Introduction 1

Solution crystallization is widely used in agriculture, food, pharmaceutical and special chemical separation and purification [1]. Today, more attention is paid to the quality control of solution crystallization processes which alter solvents, crystallization temperature, degree of supersaturation and additives, etc., which would have great impact on the properties of crystal products including polymorph and morphology [2]. In solution crystallization, the interaction of solvent with solute in bulk solution and crystal surface has great impact on crystal growth. On one hand, the ordering of solvent near the solvent-crystal interface could contribute to a rough crystal surface and thus enhance the rate of crystal growth [3]. On the other hand, the competitive adsorption of solvent and solute on the crystal growth sites could lead to retarded crystal growth [4]. Thus, the morphology of the growth crystal would be significantly affected by the behavior of the solvent. However, the control of crystal growth on specific crystal facets with solvent is still a great challenge. In many cases, with just a small dose of additives incorporated into the solution, significant changes of specific crystal facets could be induced, which indicates that the control of crystal growth of specific facets is more feasible with additives. Tailor-made additives, which are molecules that share the same molecular structure as a parent molecule with only slight structural variations, have previously been demonstrated as a useful means to control crystallization dynamics in a solution. For example, tailor-made additives were added to solutions of a crystallizing parent molecule to alter the crystal growth rate, size, shape, or even control the transformation of crystal polymorphism, etc [5, 6]. Controlling crystal growth with tailor-made additives is attracting more and more interest. In most cases, additives have a negative effect on crystal growth, which can contribute to the retarded growth on specific crystal facets. At the molecular level, the negative effect of additives on crystal growth can be well described by the adsorption of additives on the growth site on the crystal facets, which could inhibit the adsorption of the substrate molecules [7, 8]. However, the mechanism of additives is usually poorly understood on the molecular level, as it is difficult to explore by experimental characterization. Recently, a method of molecular simulation coupled with experimental and diffraction data became popular for crystallization analysis. For example, molecular dynamics (MD) simulation is an effective technology for studying 2

crystal growth, especially in the presence of impurities or additives. [9, 10] Yang et al. studied the mechanism by which L-valine affects the growth of L-alanine on (0 1 1) and (1 2 0) facets and suggested that L-valine tended to be adsorbed on the step on (1 2 0) and on the terrace of (0 1 1), which could contribute to significant inhibition effects of the growth on (1 2 0) and less effect of the growth on (0 1 1), respectively [11]. Yani et al. performed MD simulation to investigate the effect of PVP, HPMC and lecithin on crystal growth of salbutamol sulfate, and it was found that PVP was the most effective growth inhibitor among the three additives, which was consistent with the experimental study of Xie, et al [12-13]. Density functional theory (DFT) also provided reliable methods for describing the adsorption behaviour of various systems. Although DFT is thought to have high accuracy and a wider use in the adsorption systems with strong interaction [14-17], it is a relatively acceptable method with dispersion correction to deal with the systems with weak interactions. Ding et al. employed DFT methods and MD simulation to analyze the mechanism and thermodynamics of the adsorption behaviour of polybrominated diphenyl ethers on grapheme, and suggested that the interaction between the pollutant molecules and graphene was physisorption [18]. Furthermore, there are various reported mechanistic methods, which are also important in modelling the crystal growth with additives and predicting the effect of additives. For example, Sizemore et al. developed a mechanistic-based model for describing the effect of imposters on growing crystal surface. This model was successfully applied to the crystal growth of α-glycine in the presence of L-alanine and well consistent with experimental data [19]. In this study, methyl paraben (MP) was selected as the model component to investigate how the three tailor-made additives, i.e. acetaminophen (APAP), p-methyl acetanilide (PMAA) and acetanilide (ACET), affect the growth of MP crystal. The molecular structure of MP, APAP, PMAA and ACET are similar, as shown in Fig. 1. Particular effort was given to describe the mechanism of additives, where DFT methods were employed to calculate the adsorption energy between the MP crystal facets and the additives, the electron cloud distribution and the electrostatic EPSs charge gradient, while MD simulation was performed to describe the dynamic performance of additives on the solvated interfaces.

2. Experimental section 3

2.1. Materials MP was purchased from Tianjin Kermel Chemical Reagent Co. Ltd., APAP and ACET purchased from Shanghai Crystal Pure Reagent Co., Ltd., PMAA purchased from J&K Chemicals, ethanol purchased from Sinopharm Chemical Reagent Co. Ltd., were analytical reagents. Chromatographically pure methanol was purchased from Shanghai Ling Feng Chemical Reagent Co., Ltd. Water was distilled and deionized. 2.2. Selection of suitable mole ratio for additives For selecting a suitable mole ratio of additives, the solubility of MP, APAP, PMAA and ACET in ethanol at 25 °C was measured using the gravimetric method, which is similar to that in our previous work [20-21]. The results showed that MP has the highest mole ratio solubility of 14.71 mol%, while APAP has the lowest of 5.82 mol%, and ACET and PMAA have similar and medium solubility of 10.36 mol% and 9.36 mol%, respectively. According to these solubility differences, proper dosages of tailor-made additives, which would avoid the spontaneous nucleation of the additives, can be determined to induce the crystallization of MP. In this study, the mole ratio between MP and additives is fixed at 1:0.1 and 1:0.01. 2.3. Slow solvent evaporation Single crystal preparation was carried out as follows: 4 g MP (2.632×10-2 mol) and additive with 0, 2.632 × 10-4 mol or 2.632 × 10-3 mol, corresponding to the designed mole ratio, i.e. 0, 0.01 and 0.1 respectively, was put into a 50 mL glass beaker containing 10 mL ethanol, stirred for at least 1 h and saturated at room temperature, then the solution was filtered by a 0.22 m membrane filter. 2 mL of the filtered solution was transferred into a glass vial and left undisturbed for one week in an athermostat container at 25 °C. The rate of evaporation was adjusted by covering the solution with plastic film with 5 small holes with a diameter of about 1 mm. When there were crystals that began to precipitate in the solution, the tubes were shaken slightly to make the crystal growth symmetrically perfect. The prepared single crystals of certain size and perfect morphology were carefully washed with pure saturated solution and ethanol to reduce the imperfection of the crystal and the impurity of the crystal surfaces [11].

2.4. Analysis of crystal purity 4

To analyze the purity of the prepared crystals, a known amount of the crystals was dissolved in a known amount of mobile phase. Then the purity of the resulting solution was assessed by high performance liquid chromatography (HPLC) (Agilent 1200, Agilent Technologies, Ltd., USA) at 25 oC with a sample injection of 5 μL, which used a non-polar chromatography column (Agilent HC-C18, 250mm × 4.6mm i.d., 5 μm, USA), a UV wavelength detector followed at 254 nm, and a mobile phase consisting of methanol : water = 85 : 15 (v/v). 2.5. Single crystal X-ray structure analysis An X-ray single crystal diffractometer (Smart Apex II, Bruker-axs, Germany) was used for analyzing the structure of the crystals. The data were corrected by means of ψ-ω scanning, and 9867 diffraction data were collected using the Smart program with independent diffraction data of 2586 (Rint = 4.9245). Then the structure and parameters of the crystal were analyzed using SHELXL software. 2.6. Powder X-ray diffraction An X-ray diffractometer (D8 ADVANCE, Bruker AXS, Germany) with Cu-ka radiation generated at 40 mA and 40 kV, was used for attaining the powder X-ray diffraction (PXRD) patterns of the crystals. All samples were scanned between 5

o

and 45 o (2θ) with a scanning step of 0.2. The data of the PXRD was indexed to identify the morphology of each induced crystal. 2.7. Computational detail and methods Materials Studio 4.3 (Accelrys Software Inc., CA, 2008) was selected to perform DFT methods and MD simulations. The structure and parameters of a single crystal of MP was obtained using SHELXL software to analyze the experimental structure, and the unit cell of MP crystal was obtained for the following simulations. The prediction of powder diffraction patterns was simulated using a Reflex module, which was verified by comparison with the experimental PXRD patterns. In this study, DFT methods were employed to calculate the adsorption energy, electron cloud distribution and the electrostatic EPSs charge gradient of additive molecules on the non-solvated crystal surface. Surfaces were cleaved, in order to gain three-dimensional surface structures, while a vacuum thickness of 30.00 Å was initialized and all atoms in the built cell box were fixed. Then, a molecule of additive was added to the middle of the vacuum layer and relaxed. Based on preliminarily optimizing the initial structures to find the structure with optimal 5

energy, several orientations of additive were tested to find the optimal orientation. Generalized Gradient Approximation (GGA) functional was selected to describe the adsorption of additives, and electron density and population analysis with ESP charges were also calculated to analyze the electron cloud and electrostatic field. The adsorption energy of additive (Eadsorption) on the crystal facet can be calculated as follows [22]. Eadsorption  Emolecule surface  ( Emolecule  Esurface)

(1)

where Emolecule+surface is the total energy of the adsorption system, Emolecule and Esurface are the energies of the additive molecule and the assigned crystal surface of MP, respectively. To study the dynamic performance of additives on the specific crystal surfaces, MD simulations with canonical ensemble (NVT, constant particle number, volume and temperature) were performed. Non-bond energy was determined using the Ewald summation method. To find the optimal force field for the simulation, COMPASS, Dreiding, Universal, CVFF and PCFF force fields with use current force field assigned, charge using QEq and charge using Gasteiger charge rules were selected to optimize the unit cell of MP and a single molecule. When the original unit cell was optimized using a PCFF force field with charge using the Gasteiger charge rule, both the optimized unit cell and MP molecule had the least deviation in the parameters and configurations. Therefore, PCFF and charge using Gasteiger was the optimal force field and charge rule respectively, and was used in all MD simulations. The crystal facet of MP was cleaved and extended to a 3 × 3 supercell with a vacuum thickness of 50.00 Å. A solvated interface was built by inserting a solvent layer with at least 400 ethanol molecules. The solvent layer and the top layer of the crystal facet were relaxed. 100 ps was given for the solvated interface to reach equilibrium with a time step of 1fs at 298 K. Then an additive molecule was put in the middle of the solvent layer. Anneal task was implemented to optimize the structure of the solvated interface. All MD simulations were calculated for 200 ps with a time step of 1 fs at 298 K. The mean square displacements (MSDs) and radial distribution functions (RDFs) of the additive molecule were selected to further study the mobility of additives and intermolecular interaction between additives and substrates, respectively.

6

3. Results and discussion 3.1. Purity of the prepared single crystals HPLC was employed to analyze the purity of the single crystals prepared. The different separation capabilities of MP, APAP, PMAA and ACET are shown in Fig. 2, which indicates that the specific retention time of MP, APAP, PMAA and ACET was 3.5 min, 2.1 min, 3.7 min and 3.1 min, respectively. Fig. 3 displays the retention time of the prepared single crystals induced by the three additives, it indicates that only one peak with a similar retention time from 3.491 to 3.495, could be found. Therefore, it suggests that only pure MP crystal was obtained under the induction of the tailor-made additives, and no additives incorporated into the bulk crystal of MP. 3.2. PXRD results of MP single crystals Fig. 4 show PXRD patterns of the MP single crystals obtained under different concentrations of different additives, characteristic peaks were observed at 2θ values of 9.0, 9.7, 13.8, 15.3, 17.0, 18.5, 20.0, 21.5, 22.8, 25.2, 26.2 and 27.6 in both figures, which suggests that only form I crystal reported in the literature [23] was obtained. However, intensity variation of the characteristic peaks, which related to the habit of crystals, was observed. Hence, it can be concluded that the addition of APAP, PMAA and ACET only changes the morphology (or habit) of MP crystals rather than the crystal form. 3.3. Results of single crystal X-ray structure analysis Fig. 5 shows the structure and molecular packing of MP single crystals. MP molecules are interlinked with each other by O⋯HO hydrogen bonds to build a three-dimensional hydrogen-bonded network in the crystal, as shown in Fig. 5b. The cell parameters for the crystal obtained of Form 1-I are as follows: α 90o, β 120.7o, γ 90o, a 13.568Å, b 16.958Å, c 11.021Å, with space group of Cc and 12 MP molecules in a unit cell. As shown in Fig. S1, the simulated PXRD pattern of the calculated crystal structure shares the same characteristic peaks with the experimental PXRD pattern, which suggests that the calculated structure is consistent with the experimental structure, thus the calculated structure would be employed as the original structure for the following simulation and analysis. Crystallographic data of the crystal obtained and those from the literature [23-25] are summarized in Table 1. The crystal structure obtained in this work is similar to the structure reported in the literature [23-25], especially the structure reported by 7

Lin et al [24]. with cell parameters of α 90o, β 130.10o, γ 90o, a 13.568Å, b 16.958Å, c 12.458Å, which had a similar simulated PXRD pattern as that obtained in this work and shown in Fig. S2 of the Supporting information. However, there are slight differences in value of β and c, as shown in Table 1, which could be due to the effect of additives. 3.4. Crystal growth induced by three additives Fig. 6 shows the habits of MP crystals without additives and induced by three additives. When no additives were added to the solution, no steady morphology of MP single crystals could be formed, as shown in Fig. 6a. However, it is interesting that three steady morphologies can be formed under the induction of the three additives in a number of parallel experiments, as shown in Fig. 6b to 6d. Fig. 6b indicates that the MP crystal morphology induced by ACET is an octahedron with four crystal facets (h k l), i.e. (1 1 0), (-1 1 0), (1 1 -1) and (-1 1 1) exposed. However, there are some differences in the surface area of the facets, which might be due to the impact of different ACET mole ratios, as shown in Fig. 6e and 6h. When the mole ratio of ACET increases from 0.01 to 0.1, surface (-1 1 0) grows bigger whereas (1 1 -1) becomes smaller, it suggests that the presence of ACET retards the growth of surfaces (-1 1 0) and has less impact on the growth of (1 1 -1). Fig. 6c, 6f and 6i indicate that the main exposed crystal facets induced by APAP include (1 1 0), (1 -1 1), (1 1 -1), (-1 1 1), (0 2 1), (0 2 -1), (0 2 0). When comparing crystal habits in Fig. 6i with the one in Fig. 6f, it could be found that there was a significant increase in surface area of (0 2 1) and less significant change in (-1 1 1). The results suggest that the presence of APAP prohibits the growth on (0 2 1) and has less effect on facet (-1 1 1). Fig. 6d, 6g and 6j show that for MP crystal habits induced by PMAA, (1 1 0), (1 1 -1), (-1 1 0), (-1 1 1) and (0 2 0) are the main exposed facets. With the increasing of the PMAA mole ratio, facet (0 2 0) becomes larger, while facet (1 1 -1) shrinks significantly, indicating that PMAA retards the growth on (0 2 0) effectively and shows less effect on the growth on (1 1 -1). The results above indicate that the three additives have different effects on growth on the specific crystal facets, which lead to the change in crystal habit. However, the mechanism of morphology changes induced by additives is not clear yet, thus, further discussion on the mechanism of the tailor-made additives on the 8

specific crystal facets of MP was conducted by analyzing the interaction between additives and crystal facets. 3.5. The effect of additives on MP crystal surfaces using DFT methods Eadsorption, the electron cloud distribution and the electrostatic EPSs charge gradient was calculated to study the interaction between additives and the MP crystal facets. As solvent was absent in the structure, the calculated adsorption energy of additives could only serve as a qualitative indicator of the relative effect of additive on the crystal growth of MP. Table 2 lists the calculated results of Eadsorption on the main facets for three additives, it suggests that the Eadsorption on the main facets has quite different distributions for different additives. From Table 2, it can be found that all values of Eadsorption of tailor-made additives are negative, which suggests that a more stable interface could be formed through the adsorption of tailor-made additives on the crystal facets, thus the adsorption of additives on the facets is favorable. More importantly, more negative Eadsorption means more stable adsorption and stronger interaction, i.e., stronger interaction between additive and specific crystal facet could be formed. It indicates that additives might be more favourable to occupy the growth site and disturb the adsorption of solute molecule to this facet. Furthermore, when the concentration of additives increases, more growth site could be occupied by additives, and thus the growth on this facet would be significantly inhibited. On the contrary, less negative Eadsorption means additives are less favourable to occupy the growth site and disturb the adsorption of solute molecule less. Therefore, when the concentration of additives increases, the effect of additives on these facets would be not significant. As shown in Table 2, the most negative Eadsorption (-14.180 kcal·mol-1 for ACET, -24.491 kcal·mol-1 for APAP and -14.397 kcal·mol-1 for PMAA) was obtained on facet (-1 1 0) for AECT, (0 2 1) for APAP and (0 2 0) for PMAA respectively, where the growth was significantly retarded, when increasing the molar ratio of additives from 0.01 to 0.1. The least negative Eadsorption (-8.793 kcal·mol-1 for ACET, -9.912 kcal·mol-1 for APAP and -8.230 kcal·mol-1 for PMAA) was obtained on facet (1 1 -1) for ACET, (-1 1 1) for APAP and (1 1 -1) for PMAA respectively, where facet growth was not effectively retarded. Therefore, the effect of different additives on the growth of the specific MP crystal facets could be well predicted by the calculation of Eadsorption on facets. 9

Fig. 7 shows the electron cloud distribution and the electrostatic EPSs charge gradient of ACET molecule on facet (-1 1 0) and (1 1 -1). As shown in Fig. 7a and 7b, more overlap of electron cloud between the ACET molecule and MP molecules on (-1 1 0) was observed, which suggests that the interaction of the ACET molecule with MP molecules on (-1 1 0) was stronger. Fig. 7c and 7d further verify that there are more like-charges distributing around the ACET molecule on (1 1 -1), which would lead to the compression of the electron cloud of the ACET molecule, suggesting that the interaction on (1 1 -1) was weaker. These results agree with and verify the calculated result of adsorption energy listed in Table 2. Hence, ACET molecules are more likely to be absorbed to facet (-1 1 0) and inhibit the adsorption of MP molecules. Therefore, ACET has a negative effect on the growth rate of facet (-1 1 0). On the contrary, ACET has less effect on the growth of facet (1 1 -1). The effect of APAP on the growth of (0 2 1) and (-1 1 1), as well as PMAA on the growth of (0 2 0) and (1 1 -1), is similar to the effect of ACET on facet (-1 1 0) and (1 1 -1), as shown in Fig. S3 and Fig. S4 of the Supporting information. From the view point of energy and electronics, these results indicate that the calculation using DFT methods is reliable to predict the effect of the three additives on specific crystal facets of MP. To better understand the effect of the three additives, dynamic performance of the additives was investigated as follow. 3.6. Interaction between crystal facets and additives Surface-docking model, which was proposed by Lu et al [26] and Myerson et al [27] and was successfully applied to predict the impurity effect by Yang et al [11, 28], was employed to perform MD simulation for analyzing the mobility of additives, as well as the interaction between additives and MP molecules on the crystal facets. Because solvent has significant effect on the dynamic performance of additives, all the investigated interfaces were solvated. During the MD simulation process, it can be found that ACET, APAP or PMAA molecules are gradually adsorbed to the crystal facets of (-1 1 0), (0 2 1) or (0 2 0) respectively, but is difficult to be adsorbed to the crystal surface of (1 1 -1), (-1 1 1) or (1 1 -1) in the given simulation time. RDFs analysis was carried out to analyze the interaction between the bonding sites of the additive molecule and the possible bonding sites of the MP molecules, which surround the adsorbed additive molecule in the following 100 ps. RDFs is defined as the ratio (g(r)) of the 10

probability density that a particle distributes around a given particle in a distance of r (ρ(r)) and the probability of random distribution ρ: g (r ) 

 (r ) 

(2)

Thus, RDFs is an indicator which describes the probability that other particles distribute around a given particle. MSDs analysis of additives was carried out to investigate the mobility of different additives on different interfaces, with the simulation time dependence of MSDs, the diffusivities of additive can be calculated by the following equation:

D

1 d N 2 lim i 1  ri (t )  ri (0) 6 N t  dt

(3)

where N is the number of investigated particles, ri(t) and ri(0) are the position vectors of particle i at the simulation time of t and 0 respectively. Fig. 8 shows the RDFs results of the additive molecules on the different solvated interfaces. As shown in Fig. 8a, MP and three tailor-made additive molecules have various possible bonding sites to form a hydrogen bond with other molecules, which serve as both hydrogen-bond donor and receiver. All three tailor-made additives have acetamido group, where H, N and O atoms are possible bonding sites. Furthermore, there is an additional hydroxyl group in the APAP molecule, where the H and O atoms could serve as possible bonding sites. For MP, the hydroxyl group and ester group may also offer various bonding sites for hydrogen bonding. As shown in Fig. 8b, at distances of less than 2.5 Å, which could serve as the critical distance of hydrogen bond, peaks for all the bonding sites of APAP are observed, which indicates that all bonding sites of APAP molecule could form a hydrogen bond with a solute molecule. Distinct and high peaks of the H1, H2 and O2 curves are observed in this region, while only small peaks of the N1 and O1 curves are observed, which indicates that the APAP molecule binds with surface MP molecules mainly with H1, H2 and O2, and less with N1 and O1. Similarly, the RDFs results shown in Fig. 8c and 8d indicate that ACET molecule binds with the surface MP molecules on (-1 1 0) mainly with O1 and H1, while PMAA binds with surface MP molecules on (0 2 0) mainly with H1. Fig. 9 shows the crystal surface structure of facet (0 2 1) and snapshots of the various hydrogen bonding to verify the RDFs results, where hydrogen bonds 11

binding by APAP and solute molecules are marked by an orange oval. As shown in Fig. 9a, there are unbonded phenolic hydroxyl groups (marked as a blue oval) and ketonic groups (marked in the black oval) on (0 2 1), which can serve as bonding sites for hydrogen bonding. It can be deduced that the molecular orientation of MP on (0 2 1) makes crystal grows mainly by the formation of new O⋯HO hydrogen bonds with the unbonded groups of the surface MP molecule and the MP molecules which migrate from the solution. As APAP molecules were adsorbed to the surface of (0 2 1), solvent molecules would be desorbed from the crystal surface. APAP would occupy the growth sites of solute and form various hydrogen bonds with the solute molecules. As shown in Fig. 9b to 9f, hydroxyl groups of MP were the main bonding sites to bind with APAP molecules. The binding of the surface molecules with the MP molecules from the solution was disturbed. As the strength of hydrogen bonding is much larger than close packing, the disturbance effect of APAP on the growth on (0 2 1) should be very significant. Therefore, the growth on (0 2 1) was significantly retarded as a result of the adsorption of APAP molecules on the crystal surface. Also, the mobility of APAP on the interface was limited. Fig. 10 shows the crystal surface structure of facet (-1 1 1) and snapshots of the simulated solvated interface structure of (-1 1 1) at 0 ps and 200 ps. As a result of the molecular orientation of MP on (-1 1 1), Fig. 10a indicates that there are less unbonded sites exposed, and crystal grows mainly by close packing, rather than the formation of new hydrogen bonds. As shown in Fig. 10b and 10c, APAP molecules are difficult to be adsorbed to (-1 1 1), and forms no hydrogen bond with the solute molecule in the given simulation time. Therefore, APAP has less effect on the growth of (-1 1 1) and would have a higher mobility on (-1 1 1). The MSD results are shown in Fig. 11, where the corresponding diffusivities calculated by Eq. 3 are marked in round brackets. Fig. 11a indicates that APAP has significantly larger diffusivities on (-1 1 1), thus APAP has higher mobility on (-1 1 1), which verifies the results in Fig. 9 and 10. Similar results can be found for the effect of ACET on facet (-1 1 0) and (1 1 -1), and PMAA on facet (0 2 0) and (1 1 -1). Hydrogen bonding was also observed between ACET and facet (-1 1 0), as well as PMAA and facet (0 2 0), as shown in Fig. S5 and S6, respectively. No adsorption of ACET on (1 1 -1) and PMAA on (1 1 -1) or hydrogen bonding was observed in the given simulation time, as shown in 12

Fig. S7. Furthermore, Fig. 11b indicates that ACET has higher mobility on (1 1 -1), and Fig. 11c indicates that PMAA has higher mobility on (1 1 -1), which verifies the results in Fig. S5, S6 and S7. The DFT calculation and MD simulation results well explain the experimental phenomena of the retarded growth and have less effect on specific crystal facets of MP under the induction of different additives, which provides new insights into the mechanism of the tailor-made additives on the crystal growth.

4. Conclusions The effect of APAP, PMAA and ACET, which were selected as tailor-made additives for MP, on the crystal habits of MP was studied by molecular simulation combining with experiment and characterization. Different steady morphologies of MP crystal were found under the induction of different additives, and the growth changes of specific crystal facets increased with increasing the additives concentration. The results of HPLC and PXRD suggest that the presence of ACET, APAP and PMAA have influence on the crystal habit of MP, rather than its crystal structure, and the results of single crystal X-ray indicated that no new crystal form was found. The DFT calculations and MD simulations were performed to provide evidence for the mechanism of morphology changes of the induced MP crystal, which was based on the study of the interaction between additives and crystal facets of MP. The simulation results show that additives were selectively adsorbed to the crystal facets, which could inhibit the crystal growth and contribute to the changes on specific crystal facets. The results provide new insights into the mechanism of tailor-made additives on the crystal growth.

Appendix A. Supporting information Supplementary data associated with this article can be found in the document namely Supporting information.

Acknowledgments Financial support from the National Natural Science Foundation of China (Nos. 21276091, 91434126) is greatly appreciated. 13

References [1] R. Davey, J. Garside, From Molecules to Crystallizers; Oxford University: Oxford, UK, 2000. [2] M. Kitamura, Strategy for control of crystallization of polymorphs, CrystEngComm 11 (2009) 949-964. [3] J. R. Bourne, R. J. Davey, The role of solvent-solute interactions in determining crystal growth mechanisms from solution: II. The growth kinetics of hexamethylene tetramine, J. Cryst. Growth 36 (1976) 278–296. [4] I. Weissbuch, R. Popoviti-Biro, M. Lahav, L. Leiserowitz, Understanding and control of nucleation, growth, habit, dissolution and structure of two- and three–dimension crystal using ‘Tailor–made’ auxiliaries, Acta Crystallogr., Sect. B: Struct. Sci. 51 (1995) 115–148. [5] R. Q. Song, H. Coelfen, Additive controlled crystallization, CrystEngComm 13 (2011) 1249-1276. [6] M. Lahav, L. Leiserowitz, Tailor-made auxiliaries for the control of nucleation, growth and dissolution of two- and three-dimensional crystals, J. Phys D Appl. Phys. 26 (1993) B22-31. [7] Z. Berkovitch-Yellin, J. V. Mil, L. Addadi, M. Idelson, M. Lahav, L. Leiserowitz, Crystal morphology engineering by “Tailor-made” inhibitors: a new probe to fine intermolecular interactions, J. Am. Chem. Soc. 107 (1985) 3111-3122. [8] X. L. Mao, X. F. Song, G. M. Lu, Y. Z. Sun, Y. X. Xu, J. G. Yu, Control of crystal morphology and size of calcium sulfate whiskers in aqueous HCl solutions by additives: experimental and molecular dynamics simulation studies, Ind. Eng. Chem. Res. 54 (2015) 4781-4787. [9] D. Kumar, R. Thipparaboina, S. R. Modi, A. K. Bansal, N. R. Shastri, Effect of surfactant concentration on nifedipine crystal habit and its related pharmaceutical properties, J. Cryst. Growth 422 (2015) 44-51. [10] U. Aschauer, F. Jones, W. R. Richmond, P. Bowen, A. L. Rohl, G. M. Parkinson, H. Hofmann, Growth modification of hematite by phosphonate additives, J. Cryst. Growth 310 (2008) 688-698. [11] X. Y. Yang, G. Qian, X. Z. Duan, X. G. Zhou, Effect of impurity on the lateral crystal growth of L-alanine: a combined simulation and experimental study, Ind. Eng. Chem. Res. 51 (2012) 14845-14849. 14

[12] Y. Yani, P. S. Chow, R. B. H. Tan, Molecular simulation study of the effect of various additives on salbutamol sulfate crystal habit, Mol. Phamaceutics 8 (2011) 1910-1918. [13] S. Y. Xie, S. K. Poornachary, P. S. Chow, R. B. H. Tan, Direct precipitation of micron-size salbutamol sulfate: new insights into the action of surfactants and polymeric additives, Cryst. Growth Des. 10 (2010) 3363-3371. [14] R. Gonzalez-Hernandez, A. Gonzalez-Garcia, W. Lopez-Perez, Density functional theory study of the adsorption and incorporation of Sc and Y on the AlN(0001) surface, J. Cryst. Growth 443 (2016) 1-7. [15] P. Kempisty, S. Krukowski, Crystal growth of GaN on (0 0 0 1) face by HVPE-atomistic scale simulation, J. Cryst. Growth 303 (2007) 37-43. [16] P. Kempisty, P. Strak, K. Sakowski, S. Krukowski, General aspects of the vapor growth of semiconductor crystals–A study based on DFT simulations of the NH3/NH2 covered GaN(0001) surface in hydrogen ambient, J. Cryst. Growth 390 (2014) 71-79. [17] K. Itoh, A. Iwa, Y. Uriu, K. Kadokura, Infrared absorption spectroscopic and DFT calculation studies on the adsorption structures of nitromethane on the single crystals of Cu and Ag, Surf. Sci. 602 (2008) 2148-2156. [18] N. Ding, X. F. Chen, C. M. L. Wu, Interactions between polybrominated diphenyl ethers and graphene surface: a DFT and MD investigation, Environ. Sci.: Nano 1 (2014) 55-63. [19] J. P. Sizemore, M. F. Doherty, A new model for the effect of molecular imposters on the shape of faceted molecular crystals, Cryst. Growth Des. 9 (2009) 2637-2645 [20] R. Zhang, J. Ma, J. Li, Y. B. Jiang, M. Y. Zheng, Effect of pH, temperature and solvent mole ratio on solubility of disodium 5′-guanylate in water + ethanol system, Fluid Phase Equilib. 303 (2011) 35-39. [21] Y. Zhang, Y. B. Jiang, K. Li, Y. Qian, Solubility of β-artemether in methanol + water and ethanol + water from (288.85 to 331.95) K, J. Chem. Eng. Data 54 (2009) 1340-1342. [22] D. S. Coombes, C. R. A. Catlow, J. D. Gale, M. J. Hardy, M. R. Saunders, Theoretical and experimental investigations on the morphology of pharmaceutical crystals, J. Pharm. 91 (2002) 1652-1658. [23] D. Vujovic, L. R. Nassimbeni, Methyl paraben - A new polymorph?, Cryst. 15

Growth Des. 6 (2006) 1595-1957. [24] X. T. Lin, Studies on the crystal structure of p-substituted benzoates: I. the crystal structure of methyl p-hydroxybenzoate, Chin. J. Struct. Chem. 2 (1983) 213-217. [25] H. K. Fun, S. R. Jebas, A second monoclinic polymorph of methyl 4-hydroxybenzoate, Acta Crystallogr. Sect. E: Struct. Rep. Online 64 (2008) o1255-u1833. [26] J. J. Lu, J. Ulrich, An improved prediction model of morphological modifications of organic crystals induced by additives, Cryst. Res. Technol. 38 (2003) 63-73. [27] A. S. Myerson, S. M. Jang, A comparison of binding energy and metastable zone width for adipic acid with various additives, J. Cryst. Growth 156 (1995) 459-466. [28] X. Y. Yang, G. Qian, X. Z. Duan, X. G. Zhou, Impurity effect of l-valine on l-alanine crystal growth, Cryst. Growth Des. 13 (2013) 1295-1300.

MP

ACET

APAP

PMAA

Fig. 1. Molecular structure of MP, APAP, PMAA and ACET, where H atoms are marked in white, O atoms are marked in red, N atoms are marked in blue and C atoms are marked in grey.

16

mAU

APAP

MP

PMAA

1400 1200

1000

ACET

800

600 400 200 0 0

1

3

2

5

4

6 min

Fig. 2. HPLC chromatograms of MP, APAP, PMAA and ACET. a. APAP

3.491

b. PMAA

3.494

3.495

c. ACET

2

3

4

5

6 min

Fig. 3. HPLC chromatograms of MP crystals grown with different additives.

17

Relative Intensity

a. With 0.01 (mole ratio) additives

ACET APAP PMAA No additive

5

10

15

20

25 30 2θ (°)

35

40

45

Relative Intensity

b. With 0.1 (mole ratio) additives

ACET

APAP PMAA No additive

5

10

15

20

30 25 2θ (°)

35

40

45

Fig. 4. The XRD spectra of MP crystal grown with no additives, 0.01 and 0.1 (mole ratio) additives.

18

a. Crystal Structure

b. Molecule Packing

Fig. 5. Crystal structure and molecular packing of MP, where H atoms are marked in blue, O atoms are marked in red and C atoms are marked in grey.

b

a

c. Morphology induced by APAP

b. Morphology induced by ACET

(-1 1 0)

(0 2 -1) (1 1 -1)

(1 1 -1)

(-1 1 1)

c

d. Morphology induced by PMAA

(0 2 0) (1 1 0)

(-1 1 0)

(1 1 0)

(0 2 0) (1 1 0) (-1 1 1)

e. MP:ACET=1:0.01

(1 1 -1 ) (1 -1 -1)

(0 2 1)

f. MP:APAP=1:0.01

g. MP:PMAA=1:0.01

(1 1 -1)

a. MP: No additive

(0 2 -1)

(-1 1 0)

(0 2 0) (-1 1 1)

(1 1 -1)

(-1 1 0) (0 2 0) (1 1 0)

(1 1 0)

(-1 1 1)

h. MP:ACET=1:0.1

(0 2 1)

i. MP:APAP=1:0.1

(1 1 0) (-1 1 0) (1 1 -1 )

j. MP:PMAA=1:0.1

Fig. 6. MP crystal morphologies and experimental habits induced by additives.

19

Fig. 7. Electron cloud distribution and electrostatic field of ACET molecule in facets (-1 1 0) and (1 1 -1). Additive molecule is marked in blue, while substrate molecules are marked in red. The electron cloud is shown by green dots. The strongest charge is marked in red while the weakest is marked in blue.

20

16

b. APAP on (0 2 1)

a. Bonding sites O2

O1

H1

12

N1 H2

N1 10

H1

APAP

g(r)

O1

ACET O1

H1'

8 6

O3'

4

N1

O1'

H1

O2'

PMAA

2

MP

0 0

22

2

3

4

5 6 r (Å)

7

8

9

d. PMAA on (0 2 0)

H1 N1 O1

18

10

H1 N1 O1

14

16

12

14

10

12

g (r)

g(r)

1

16

c. ACET on (-1 1 0)

20

H1 H2 N1 O1 O2

14

10

8

8

6

6

4

4

2

2

0

0 0

1

2

3

4

5 6 r (Å)

7

8

9

10

0

1

2

3

4

5 r (Å)

6

7

8

9

10

Fig. 8. RDFs results of the additive molecules on the different solvated interfaces

a. Surface structure of (0 2 1)

b.O2⋯H1' at 128 ps

c. H2⋯O1' at 116 ps

O2 H1' O1' d. H1⋯O1' at 140 ps

e. N1⋯H1' at 155 ps N1

H2

f. H2⋯O2' and O2⋯H1' at 156 ps H2 O2

H1'

H1

O2'

O1'

H1'

Fig. 9. Crystal surface structure of facet (0 2 1) and snapshots of different hydrogen bonding diagram between bonding sites of APAP molecule and solute 21

molecules. a. Solute molecules: colourful stick; b to f. Solute molecules: colourful line; Ethanol: blue line; APAP molecule: colourful ball and stick. The same legend is used in Fig. 10.

Fig. 10. Crystal surface structure of facet (-1 1 1) and snapshots of the simulated solvated interface structure of (-1 1 1) at 0 ps and 200 ps.

22

Mean Square Displacement (Å2)

1000 a. APAP

800 600 400 200 (0 2 1) (D = 2.885×10-9 m2/s)

0

Mean Square Displacement (Å2)

(-1 1 1) (D = 8.373×10-9 m2/s)

0

20 40 60 80 100 120 140 160 180 200 t (ps)

700 b. ACET

600 (1 1 -1) (D = 5.144×10-9 m2/s)

500 400 300

(-1 1 0) (D = 1.380×10-9 m2/s)

200 100

0 0

20 40 60 80 100 120 140 160 180 200 t (ps)

Mean Square Displacement (Å2)

300 c. PMAA

250 200 (1 1 -1) (D = 2.118×10-9 m2/s)

150 100 50 (0 2 0) (D = 1.563×10-10 m2/s)

0

0

20 40 60 80 100 120 140 160 180 200 t (ps)

Fig. 11. Mean square displacement of three additives on the different solvated interface.

23

Table 1 Comparison of single crystal analytical results obtained by experiment with the reported crystal structure. Crystal

Experiment

Reference24

Reference23

Reference25

chemical formula formula mass crystal system space group a/ Å b/ Å c/ Å α/° β/° γ/° unit cell formula units, Z unit cell volume/ Å3 temperature/K unique reflections observed reflections R1[ I > 2σ(I)] wR(F2)[ I > 2σ(I)] R1[all] wR(F2)[all] Rint

C8H8O3 152.14 monoclinic Cc 13.568(3) 16.958(3) 11.021(2) 90.00 120.07(3) 90.00 12 2194.6 298 4895 4257 0.0415 0.1254 0.0461 0.1357 0.0438

C8H8O3 152.14 monoclinic Cc 13.568(5) 16.958(7) 12.458(6) 90.00 130.10(3) 90.00 12 2192.7 298

C8H8O3 152.14 monoclinic Cc 13.006(3) 17.261(4) 12.209(2) 90.00 129.12(3) 90.00 12 2126.6 113 4338 3800 0.0330 0.0857 0.0402 0.0902 0.0459

C8H8O3 152.14 monoclinic Cc 12.971(4) 17.249(7) 10.8428(3) 90.00 119.260(1) 90.00 12 2116.32 100 3278 2705 0.0425 0.1098 0.0542 0.1163 0.0474

0.054

Table 2 Adsorption energy of additives on different MP crystal surfaces. Additives ACET APAP PMAA

Eadsorption/kcal·mol-1 110

-1 1 0

1 1 -1

-1 1 1

021

0 2 -1

020

-13.925 -15.183 -13.727

-14.180 -12.947 -12.234

-8.793 -11.910 -8.230

-9.052 -9.912 -9.272

/ -24.491

/ -14.562

/

/

/ -12.035 -14.397

24

Highlights 

The effect of tailor-made additives on the crystallization of MP was investigated.



The crystal habit of MP crystals was greatly changed by the additives.



The change of crystal habits was induced by selective adsorption of additives.

Graphical Abstract MP-Ethanol-Solution + Additives Single Crystal Preparation H1

Analyze the Effect of Additives

(0 2 -1) (1 1 -1) (-1 1 0) (0 2 0)(1 1 0)

(-1 1 0) (0 2 0) (1 1 0) (-1 1 1) (0 2 1)

(-1 1 1)

MP:APAP=1:0.01

MP:APAP=1:0.1

(0 2 1)

Effect of APAP on facet (0 2 1)

Growth with Less Effect Growth with retardation

Effect of APAP on facet (-1 1 1) Relative Intensity

(0 2 -1) (1 1 -1)

With 0.1 APAP With 0.01 APAP With no additive

5

10

15

20

25 30 2θ (°)

25

35

40

45