The following changes are relative to 3.2.0, which was the last released version (May 2023). Only the most significant changes are detailed below; the full list of changes is available here.

A collaboration with the NanED group led to a number of new data names, and additions to existing definitions, that were incorporated into the core dictionary as they had general application outside ED. Among other things, the new definitions cover:

• Sample tracking
• Sample mosaicity
• Choice of diffraction theory
• Sample shape refinement

Further documentation and examples of use are available in the cif_ed repository.

New data names for flux density, total dose, and total exposure time were defined as a step towards characterisation of radiation damage.

In addition to the previously-existing Cromer-Mann and Fox parameterisations of atomic form factors, the ATOM_TYPE_SCAT category now includes data names for recording exponential polynomial, Gaussian, and inverse Mott-Bethe descriptions of form factors over user-defined ranges of sinθ/λ.

A separate category, ATOM_SCAT_VERSUS_STOL, has been added where scattering factors as a function of sinθ/λ can be tabulated.

Reflection lists used to be indexed using hkl. However, there are important situations where hkl may not identify a unique reflection: for example, lists of unmerged reflections from imaging detectors and higher-dimensional modulated structures. Therefore, use of hkl as the unique key to a reflection list has been superseded by use of opaque identifiers _diffrn_refln.id and _refln.id (typically an integer numbering). Of course, software that assumes hkl does not repeat will continue to process CIF files correctly for types of data where this assumption holds, such as merged reflection lists from untwinned, non-modulated structures.

A few data names have been added to capture permanent identifiers linked to external metadata systems. _exptl.method_id_PaNET identifies the relevant technique from the PaNET ontology. The DIFFRN_COMPONENT_PID category lists identifiers for instrumentation components such as detectors or entire lab diffractometers. Currently, DataCite DOIs and ePICs handles pointing to metadata matching the RDA PIDINST schema are supported.

The work on adding support for persistent identifiers was achieved in collaboration with the European funded OSCARS project, work-package 2.

Both AUDIT_AUTHOR and PUBL_AUTHOR categories now separate name and address. AUDIT_AUTHOR also includes email, fax (remember those?) and telephone contact details.

We now recommend that dictionaries include a list of dictionary authors. If your name is missing from the newly-added author list and should be on it, please raise an issue.

All definitions related to spreading data across multiple data blocks have been moved to the multiblock dictionary. At the same time, the auxiliary dictionaries containing tables of default enumerated values (see next item), the attribute templates, and the DDLm dictionary itself have been moved to separate repositories.

The DDLm dictionary language introduced default values that depend upon the value of another data name, for example, the default value for the atomic number of an atom could be found from the atomic symbol. These lists of default values are contained within an auxiliary dictionary file. DDLm and the core dictionary have been updated to allow more than one data name to be used when looking up the default value, so that items like the neutron scattering length (which depends on both atomic number and atomic weight) could have default values assigned. This change resulted in the replacement of single-valued _enumeration.def_index_id in certain definitions being replaced by list-valued _enumeration.def_index_ids. Dictionary-aware software which utilises default enumerations should be updated accordingly.