The key technology of pharmaceutical process - What are the advances of crystallization technology?

At present, there are two main types of continuous crystallizers in common use: mixed suspension mixed discharge crystallizer (MSMPR) and continuous tube crystallizer (Figure 3). Comparing the two, it can be found that the material's residence time distribution is relatively wide and the residence time is long in MSMPR, while the residence time distribution is relatively narrow and the residence time is short in tubular crystallizer. In addition, compared with tubular crystallization, MSMPR is more suitable for the production of large-particle size crystals.

With the development of continuous crystallization research, the technology has been able to be used to produce some desired drug crystals. Compared with the intermittent process, the continuous crystallization process can provide potential economic value. However, not all processes are suitable for continuous crystallization at this stage. For example, in the production of chiral crystals, the yield of continuous crystallization is still very low. Hopefully, these bottlenecks can be resolved in the future.

04

Crystal Structure Prediction Technique (CSP)

CSP is primarily concerned with finding the most thermodynamically stable crystal structure and determining the most energetically favorable crystal arrangement in a solid without input of any experimental information. That is, the crystal structure with the lowest free energy can be found for a given chemical composition under a given pressure-temperature condition.

In general, a successful CSP calculation involves two aspects (Figure 4) : (1) generating searches for all potential low-energy structures; (2) Calculate the energy order of relative stability of a series of candidate structures.

Therefore, in order to predict crystal structure, two complementary CSP methods, global optimization and data mining, are proposed, and the global search of potential energy surface is solved.

Among them, global optimization strategy can achieve global prediction without any prior knowledge and data. Relying on powerful search algorithms, global optimization is able to generate entirely new crystal structures and compounds, including those that are not related to the original ideal net.

Data mining is a non-global optimization forecasting method based on existing knowledge and the content of crystal structure database, which can predict stable crystal structure very quickly. However, under the current level of computer technology and optimization methods, most successful cases of CSP using machine learning are limited to inorganic materials and organic molecular systems with low chemical complexity, and lack of promotion to complex molecules with higher relative molecular mass, so the prediction of drug crystal structure is limited to a certain extent, and there is still a lot of room for development.