(344e) Process Parameters in the Purification of Curcumin By Cooling Crystallization

Structurally related impurities (often by-products) are of great importance in the pharmaceutical industry. Such impurities are hard to remove completely by crystallization, because of their structural similarity to the solute molecule. Impurities may become incorporated into structures and influence nucleation^1,2 and growth², and thus polymorphic outcome, crystal shape and size.^2-4 Often multi-step successive crystallizations are needed to reach a desired purity, significantly lowering the product yield.

Curcumin, a major active component of the Indian spice Turmeric (Curcuma Longa), has a wide range of bioactivities (antimicrobial, anti-inflammatory and anticancerogenic properties).Extract obtained from Turmeric comprises a mixture of curcumin, demethoxycurcumin (DMC) and bisdemethoxycurcumin (BDMC), known as curcuminoids. Commercially available crude curcumin contains significant amounts of the other two curcuminoids, typically about 17% DMC and 3% BDMC. Separation and purification of curcumin is quite challenging, since curcuminoids are structurally very similar, with differences only in the number of methoxy groups.

In this work, purification of crude curcumin by successive cooling crystallizations has been investigated.⁵ Statistical experimental design techniques were used to identify the effect of cooling rate, seeding, seed polymorph and agitation on the yield, crystal purity, particle size, shape and structure as well as determining the minimum number of crystallization steps needed to arrive at a particular product specification. Isopropanol was selected as the solvent for the curcumin crystallization based on determined solubilities of the individual curcuminoids. The composition of solid curcumin samples and concentration of individual curcuminoids dissolved in the mother liquor were quantified by HPLC. The polymorphic composition of the crystals was determined by Raman spectroscopy and by X-ray powder diffraction. Particle size and shape distribution was determined by light microscopy and image-processing analysis, using Feretâ??s diameter and aspect ratio as particle characteristics. Molecular dynamics simulations were performed to calculate curcumin lattice energy as a function of curcuminoid impurity concentration (GAFF 600 molecules, Gromacs).

The results showed that cooling crystallization of curcumin is sensitive to process parameters including seed polymorph, stirring method, cooling rate and the number of crystallization steps, the latter being closely related to impurity concentrations. Manipulation of these process parameters allows for polymorph selection, particle habit and size control, purity and yield control. It was found that removal efficiency of DMC, the major impurity present in crude curcumin, exponentially decreases as a function of the number of successive crystallizations. Crystallization operated in seeding mode and slow cooling rate increases the purity of the obtained curcumin crystals. In particular, using seeds of the metastable Form II improves the purification of curcumin significantly. The computational results showed that incorporated impurity decreases the lattice energy, and that this decrease is more pronounced for the metastable Form II, indicating a weaker affinity of the impurity molecules for the metastable Form II curcumin. At substantially optimal crystallization conditions high purity curcumin (>99.1%) can be obtained by a reduced number of successive crystallization steps. Such a decrease in the total number of successive crystallizations increase the overall curcumin yield from 28% to 50%.

1.  Pino-Garcia, O.; Rasmuson, A. Cryst. Growth Des. 2004, 4, 1025â??1037.
2.  Shtukenberg, A.; Lee, S.; Kahr, B.; Ward, M. Annu. Rev.Chem. Biomol. Eng. 2014, 5, 77â??96.
3.  Cashell, C.; Corcoran, D.; Hodnett, B. K. Cryst. Growth Des. 2005, 5, 593â??597.
4.  Liu, J.; Svärd, M.; Hippen, P.; Rasmuson, A. C. J. Pharm. Sci. 2015, 104, 2183â??2189.
5.  Ukrainczyk, M.; Hodnett, K. B.; Rasmuson, A. C. submitted to Org. Process Res. Dev. (2016).

This work has been supported by Science Foundation Ireland, Grant number: 12/RC/2275.