kmno4 - 1-3-2014 at 06:33
It is commonly known that metallic elements exist in simple compounds as cations (sodium , potassium... etc).
So, experimentally it can be found ionic (cationic) radius in such ionic compounds (NaCl, KF.... etc). Of course, the radius changes in some limits,
depending on compound.
Available tables (see http://en.wikipedia.org/wiki/Ionic_radius for example) give also ionic radius for non-metals: boron, carbon, silicon, sulfur, nitrogen... etc.
However it is know that such cations do not exist in any chemical compound (C(4+), B(3+), N(5+).....etc.)
Can someone enlighten me - what is the purpose of giving radii for non-existent ions ?
Another question - how - experimentally - are these values determined ?
turd - 1-3-2014 at 10:31
They are determined from the crystal structures of carbonates, borates and nitrates and corrected for shortening of the bonds by covalency. See the
original references by Shannon which give the methods and the employed crystal structures.
For a slightly more involved empirical estimation of expected interatomic distances, see the bond valence method of I. D. Brown (https://en.wikipedia.org/wiki/Bond_valence_method and references therein).
Edit: corrected typo.
[Edited on 1-3-2014 by turd]
leu - 1-3-2014 at 13:57
The attached article:
THE PRINCIPLES DETERMINING THE STRUCTURE OF COMPLEX IONIC CRYSTALS
Linus Pauling
J. Am. Chem. Soc., 1929, 51 (4), pp 1010–1026
explains the reasons such crystals exist 
Attachment: ja01379a006.pdf (578kB)
This file has been downloaded 528 times
kmno4 - 24-3-2014 at 04:01
Unfortunately, the lecture of Shannon's papers (thanks turd) gave me nothing. Because of unknown reasons, Shannon sees "ions" in
every single compound mentioned in the papers, e.g. he makes no difference between Na, C and O in Na2CO3: this compound contains Na(+), C(+4) and
O(-2) "ions". It looks to me like some kind of chemical manner.
In some cases, ionic radii have close to zero or even negative values (of course in Shannon's sense).
Citation (DOI:10.1107/S0567740869003220, paper attached):
Furthermore, with smaller anions one is not forced to
assume negative radii for C(4+) in the carbonate ion
and N(+5) in the nitrate ion. Although there is intrin-
sically no objection to negative radii so long as one does
not assume a hard sphere model, it is esthetically more
pleasing to have positive radii. However, even with
Fumi & Tosi's crystal radii it is not possible to achieve
a complete set of positive radii. (...)
It sounds for me very curiously 
The reason of "negative radii" is obvious: wrong assumption that oxygen in e.g. carbonate ion has similar volume/radius like in e.g. sodium oxide.
Shannon mentions this "covalent issue", but explains it in some twisted way........
I do not like his concepts and "ions" at all.
In attachment
Acta Cryst. (1969). B25, 925-946
Effective ionic radii in oxides and fluorides by R. D. Shannon and C. T. Prewitt
Attachment: a06734.pdf (845kB)
This file has been downloaded 586 times