(136g) Enabling Rational Morphology Design and Crystal Engineering with Addict Software for Mechanistic Crystal Growth Models

P. Zhu, J. Li, C. Tilbury, Y. Sun, K. Girard, Y. A. Abramov, M. F. Doherty

Mechanistic crystal growth models offer the potential for improved engineering of crystalline products. These methods adopt an accurate description of the underlying surface growth physics& solid-state chemistry, resulting in the ability to predict effects of the growth environment on crystal shapes and enabling a systematic sweep of the design space to discover optimal growth conditions. However, this increased fidelity comes at a price of, in general, requiring significant time invested in developing expertise and then routines that implement these methods. We have developed proof-of-concept software that delivers an automated implementation of the mechanistic spiral growth model for non-centrosymmetric molecules. This software, called ADDICT (Advanced Design and Development of Industrial Crystallization Technology), can make predictions for any neutral organic molecule; developments are underway to broaden its applicability to organic salts.

ADDICT provides two significant benefits. First, it requires little expertise to use, yet yields the power of mechanistic predictions and conveys increased ability for rational design of crystal products. Only the crystallography (predictions are for a specific polymorph) and a partial charge calculation are required, then ADDICT can predict the crystal shape at specified growth conditions; over 30 solvents are included within the program. ADDICT also contains a visualization tool to indicate hypotheticalcrystal shapes for the crystallography, which can guide selection of design conditions to modify relative growth rates appropriately and realize an optimal product. Second, ADDICTenables increased testing of mechanistic methods for a variety of crystal systems. This can guide model development; one of the key advantages of adopting a mechanistic approach is the ability to systematically improve the underlying engine driving shape predictions, which has guided concurrent research.

In this presentation, we provide an overview of our automation strategy and software capabilities. We demonstrate shape predictions for three systems: two complex organic aromatics^1,2 and a Pfizer pharmaceutical molecule, and then compare to experimental morphologies.

1: 1,2,3,4,6,7-Hexahydro-5H-benzo(f)cyclopenta(c)isoquinolin-5-one

CSD: ABACAI

Hongbin Li, Hua Yang, J.L. Petersen, K.K. Wang, J. Org. Chem. (2004), 69, 4500, doi:10.1021/jo049716t

2: 3-(Acetoxymethyl)-1,4-diphenylazetidin-2-one

CSD: ABAFAL

A. Basak, S.S. Bag, P.A. Mazumdar, V. Bertolasi, A.K. Das, J. Chem. Res. (2004), 318