The two polymorphs of L-Phenylalanine reported – LHS shows Form I (established as Z’ = 4 in P21) and RHS Form IV
In my opinion, this paper highlights the chemical and crystallographic expertise that is required to understand crystal structures fully, even now when data collection and structure solution are routine procedures. Thankfully, not all of the tens of thousands of crystal structures added to the CSD every year are this problematic, but the paper does underline the need for us at the Cambridge Crystallographic Data Centre (CCDC) to expertly maintain the CSD and provide tools to help our users evaluate the data we receive. The authors of this paper state there are now four well-characterized polymorphs of L-Phenylalanine and call into question the structure of DL-Phenylalanine as a true racemate (suggesting instead a twinned conglomerate).
The CSD categorises polymorphs in several ways. The most familiar to many may be the refcode family; all determinations of a compound are grouped together, including those at different temperatures, pressures, polymorphs etc. (L-Phenylalanine can be found in QQQAUJ). In such cases, we rely on the interpretation of the scientists who report the compounds, and later developments may require us to modify these decisions. The CCDC’s ConQuest also provides ‘best representative’ lists, to analyse structures programmatically to determine a single best example of each unique polymorph (for further information on the best representative lists see doi: 10.1107/S0108768106019677). Researchers at the CCDC are also providing tools to help scientists evaluate the stability of their compounds and the likelihood of other polymorphic forms being present.
One of the structures from the CCDC’s fifth Blind Test of Organic Crystal Structure Prediction Method in 2010.
On a related note, determining the crystal structure of Phenylalanine acts to highlight the even greater challenge of trying to predict the solid form a molecule may adopt. The CCDC’s sixth blind test
of organic crystal-structure prediction methods, launched on the 12th September, will be an interesting indicator of the current state of the art. I wish all the groups taking part the best of luck; this recent paper illustrates just how much they may need it!