Online Database of EPR and ENDOR Spectra

Author: Burkhard Kirste, Freie Universität Berlin, Institut für Chemie und Biochemie - Organische Chemie


In the active time of our EPR/ENDOR group (Prof. Kurreck), quite a few EPR and ENDOR spectra of small organic radicals, multispin systems and bioorganic radicals have been measured. Since 1984, digital spectra from computer-interfaced spectrometers are available. However, these spectra are stored locally in some kind of home-brewn binary format. In order to make these spectra publicly available, several problems have to be attacked. Thus, data should be converted on request to some kind of useful interchange format (e.g. JCAMP or ASCII) by a cgi script, possibly also to a displayable picture. Second, search facilities are required, offering different types of access: by name, by CAS registry number, by molecular formula, by structure. For the latter, available Java based utilities might be used (e.g. the JME Molecular Editor written by Peter Ertl, Novartis Pharma AG).


Although there are quite a few spectral databases available, commercial or freely available in the Internet, to my knowledge only one of them is offering a few EPR spectra: the Spectral Database for Organic Compounds, SDBS (from AIST, Japan).

Regarding my own collection of EPR, ENDOR and TRIPLE spectra, I have to admit that all original spectra recorded on paper or on punched tape, before 1984, are lost meanwhile. However, all spectra digitally recorded, from the period 1984 - 1993, are still available. The first of them, originally stored on magnetic tape or on an HP 1000 minicomputer, had to be transferred via serial line to a more compatible personal computer.

This personal collection consists of more than 1200 spectra, but only some of them are of general interest; most of them are series of measurements in liquid crystals or in reversed micelles. The data are stored in a home-brewn binary format and have to be converted to, e.g., the JCAMP-DX format for public use.

Setting of Problem

  1. Conversion of spectra to the JCAMP-DX format. This is fairly easy for ENDOR spectra, because our data sets contain most experimental parameters, and for the data points y values (intensities) have been recorded, and the corresponding x values (frequencies in MHz) can be assigned unambiguously. Regarding EPR spectra, only y values (intensities) have been recorded; the field sweep was linear and has to be reconstructed from the scan width (in Gauss) and the central field setting. Hence, they do not allow reliable conclusions regarding g values.
  2. Description of the sample (identity of the radical or paramagnetic species, solvent, temperature, perhaps concentration etc.)
  3. Substance identity of the radical (name, synonyms, structure, CAS registry number (RN))
  4. References, if applicable
  5. Preview of spectra
  6. Search facilities: by name, structure, molecular formula or CAS RN
  7. Reverse lookup: from data (HFC) to radical

Results and Discussion

Search facility: The lookup of compounds by name or by CAS registry number does not present major problems, it can fairly easily be implemented by means of a CGI script. More ambitious is the problem of structure or substructure search. For this purpose, the Java applet JME Molecular Editor written by Peter Ertl, Novartis Pharma AG, is used. Our (tiny) Search page for molecules and free radicals is based on a CGI Perl script (OCDB: Organic Chemistry DataBase). It allows text searches, including the possibility of using SMILES strings generated by the JME Molecular Editor. In the current implementation, only the first hit (if any) is displayed; this requires some thought in the ordering of the database entries (e.g., tetrahydrofuran should be placed before 2-methyltetrahydrofuran).

Database entries show the substance identification in the first part: The structure of the compound is displayed in the first place (generated on the fly from the JME string by the Java applet), followed by the SMILES string, the CAS registry number, the molecular formula, the molecular weight, and the name (German, English, synonyms, CAS name). Unfortunately, there is one problem with the JME representation: there is no designation for radicals; e.g., the structure of galvinol (with OH) is displayed for the galvinoxyl entry (Coppinger's radical). Radical anions (e.g. of anthracene) are included in the entry of the neutral (diamagnetic) compound. In the second part, links to relevant information are provided, such as (MDL) mol files, spectra, and references.

Info pages: For the more interesting radicals, individual web pages have been written, for example for Yang's bradical. They provide the name(s) and an image of the structural formula, links to the EPR/ENDOR spectra (allowing download in JCAMP-DX format), simulations of EPR spectra, tables of HFC (hyperfine coupling constants in MHz), references etc. Previews of the spectra have been generated by means of the Java tool JSpecView. An index is providing a listing of the available info pages: EPR/ENDOR index at - info pages for a selection of free radicals.

Reverse lookup: as for (13C) NMR and mass spectra, it might be interesting to be able to retrieve the structure of the compound, given the spectrum. This problem has been tackled here in a rudimentary way be providing a sorted listing of HFC (in MHz); HFC are ordered by decreasing absolute values. Although this table might be searched with the numerical values of HFC, this is probably of limited use because of numeric inaccuracies caused by experimental errors (typically ± 0.01 MHz) and the dependence of HFC on solvent and temperature.

Conclusions and Outlook

As an alternative to a database of digital (or digitized) EPR/ENDOR spectra, the corresponding data might be integrated directly into online publications by using CML (Chemical Markup Language) or XML. It might be worth while to create a big database of all available EPR and ENDOR spectra, maybe in analogy to the NMR Shift database (