With the highly competitive development of chemical and pharmaceutical industries, mastering crystal growth is becoming increasingly necessary. Modern industrial manufacturers place high importance on the ability to grow crystals with a specific habit using tailored operating conditions. A detailed understanding of crystal growth is, therefore, vital for researchers in crystallography and crystallization to respond and realize this objective. Various models to predict crystal shape in the literature are reviewed here. The most commonly adopted are usually non-mechanistic and limited in their predictive power and utility, especially for products of industrial interest. Mechanistic models offer far more potential for rational crystal design, but

In order to develop a practical engineering tool for predicting the relative growth rates and morphology of solution-grown faceted crystals, a method for quickly determining solvent-modified crystal surface energies is required. The bulk interface approximation and model by van Oss, Chaudhury, and Good provides the most practical option available for small organic molecules. Applying these techniques to the mechanistic growth modeling of four centrosymmetric crystal systems provides evidence of the utility of this treatment, since both sublimation and solution growth shape predictions correspond to experimental shapes. The fact that the approach correctly predicts the changes between sublimation and solution growth shapes supports the ability of this technique to accurately account for the solvent effect.

Mechanistic modeling facilitates rational crystallization engineering and design space screening. For an accurate model, the dominant growth mechanism operating on each face must be determined, which is highly dependent on supersaturation. Considering the case of centrosymmetric growth units, we developed and connected existing mechanistic expressions for spiral and two‐dimensional‐nucleation growth regimes, through application of stationary nucleation rate theory. Our approach enables calculation of crossover supersaturations and forms a framework to model the specific mechanism operating on each face under given crystallization conditions. Increasing supersaturation can change the crystal morphology; as face families switch growth mechanisms, they may grow out of the steady‐state shape, or influence its aspect ratio. Application of the model to naphthalene, biphenyl, pentaerythritol and β‐HMX shows the ability to capture experimentally observed examples of such supersaturation‐dependent crystal habits. This methodology broadens the applicability of mechanistic crystal growth modeling to include higher‐supersaturation industrial processes. © 2017 American Institute of Chemical Engineers AIChE J, 63: 1338–1352, 2017

The central kinetic processes defining layer-bylayer crystal growth or dissolution are the attachment and detachment rates of growth units at kink sites; the net balance of these activated processes leads to either crystal growth or dissolution. Various sets of rate expressions for attachment and detachment processes have been used in the literature, in each case attempting to most appropriately capture the underlying surface chemistry. We examine these proposals with specific attention to thermodynamics and detailed balance criteria and then recommend which expressions to adopt.

Almost all molecules are non-centrosymmetric, which produces interaction anisotropy within a crystal lattice. This anisotropy generates multiple types of kink sites on each crystal step and repeating patterns of rows with different growth units from the perspective of the lattice interaction environment, even for pure molecular crystals. As a result, unstable edge rows may be generated that dissolve under conditions of crystal growth. A method to account for edge surface structures, considering such effects, is required to accurately model the step velocity, which is vital for a mechanistic description of crystal growth. We classify both thermodynamic and kinetic contributions to step row instability and develop expressions for kink densities and step velocities that capture these important non-centrosymmetric phenomena. To demonstrate the utility of our framework, we consider in depth the case of an alternating-row A–B step. Our mechanistic predictions compare favorably to kinetic Monte Carlo simulations across a wide range of interaction anisotropy.