Discussion List Archives

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Phase Identifiers 2

Dear Colleagues

     Now that summer is over, I would like to resume our
discussion of phase identifiers.  Before doing so, I would like
to welcome two new members to the group: John Westbrook from the
Protein Data Bank and Pierre Villars of the Pauling File.  Pierre
has had considerable experience in looking at the properties of
different phases, particularly those of binary compounds.

     The previous contributions to this discussion can be
reviewed at:

     http://agate.iucr.org/iucr-top/lists/phase-identifiers

but a short summary will provide a context for our continuing
discussions.

1. Our task is to produce an identifier suitable for computer use
that would uniquely and unambiguously identify a particular
material phase.  Such an identifier would be used, e.g.,  for
connecting information on the same phase stored in different
databases.

2. We decided that the identifier may be composed of several
components.  For convenience in discussing what these components
might be, we decided to express the identifier in terms of CIF-
like items, each item corresponding to one of the components.
The final format or formats of the identifier are to be
determined later once the content has been settled.

3. We distinguished between internal and external identifiers,
the internal identifiers being derived form the properties of the
phase itself, the external identifiers being a code arbitrarily
assigned by an external agency (e.g., CAS numbers, CSD-REFCODES).
We agreed that our identifier may be compose of a combination of
internal and external components.

4. We identified enough difficulties with characterizing a given
phase using internal components that it is unlikely that we can
satisfy all the conditions specified in Section 1 above and we
may have to be content with an identifier that, while matching
the target phase, may also match a small number of other phases
that are not being targeted.  To this end we accept that at a
given stage in characterizing a phase, there may not be enough
information to fully identify the phase, and as a result not all
of the components of the identifier may be present.  The more
complete the identifier, the more likely it is that only the
target phase will be retrieved.

5. Insofar as possible, we should avoid the use of numbers that
are subject to experimental uncertainty (e.g. density) as such
numbers are not unique.  Integers, e.g., space group numbers, may
be used since they are both unique and unambiguous.

6. A list of components that have so far been proposed are:

     a) composition
     b) phase type
     c) crystal system
     d) space group number
     e) atom count in unit cell
     f) CAS number

a) The suggested way of giving the composition is to use the sum
formula normalized so that the multiplicities of the different
elements all lie between 0 and 1.0 (the element with the highest
multiplicity has its multiplicity set to 1.0).  Other
multiplicities are given as fractions (not decimal numbers) for
stoichiometric substances.  Where the multiplicity is irrational
or not well defined, a range of values may be given in a loop,
e.g. for Pb(Ti,Zr)O3 we would give:

_composition_formula  'O1 Pb1/3 Ti*1/3 Zr*1/3'
loop_
     _composition_range_element
     _composition_range_low
     _composition_range_high
          Ti   0.5  0.6
          Zr   0.4  0.5

Note that specifying Ti* as having a multiplicity of 1/3 means
its range of composition values may meaningfully lie between 0
and 1.0, since the limiting compositions are PbTiO3 and PbZrO3.

The ranges are not unique and may have different values depending
on the information available.  However, a match would be found
provided the range specified in the query overlapped the range
given in the target symbol.  However, in evaluating the
composition range, both the rational multiplicity given in the
formula and the decimal value given in the range loop must be
taken into account, i.e., for Ti the composition would range
between 0.5/3 and 0.6/3.

The formulae could be simplified by omitting 1 wherever it occurs
as an integer, e.g., 'O Pb/3 Ti*/3 Zr*/3'.   Al2SiO5 would then
appear as 'Al2/5 O Si/5'.

A material such as Fe S(1+x) would be given as:  'Fe* S1' or more
simply 'Fe* S'  with Fe having a composition range from say 0.90
to 0.95.  Ba2Cu3YO(6+x), 0<x<1, in which the major component is
variable could be rendered: 'Ba2/7 Cu3/7 O* Y/7' with O having
the range 0.86 to 1.0 (i.e., 6/7 to 7/7).  Note that in this case
O would be identified as the major component (or at least
potentially the major component) by the absence of a fractional
multiplier.  Compare with 'O Pb/3 Ti*/3 Zr*/3'.

Assuming that we adopt this convention we need to address the
following questions:

i) Does this description lead to a unique description while at
the same time cover all the possibilities?

ii) Should there be some lower limit to the multiplicities that
are included in the formula so that a long list of minor
components such as is found, e.g., in many minerals, is not
given, or can software be designed to ignore components that are
present below a prescribed cut-off?  If the formula should omit
the minor components, what should the cut-off be and should the
minor components be listed separately, i.e., might they be
important in some aspects of phase identification?

iii) Is there a better way of expressing the composition?

b) phase type
     This would be a flag to indicate the type of phase that was
being described.  Possible values of this flag are:
     liq  liquid
     xtl  crystal
     lxt  liquid crystal
     qxt  quasi-crystal
     gls  glass
     amp  amorphous
     inc  incommensurate crystal (strictly a subgroup of xtl)
Other values suggested by Sidney Abrahams (and perhaps represent
a different component of the identifier) are:
               com  composition-change morphotropic phase (i.e., an
          inhomogeneous crystal containing regions of different
          composition and structure).
     pol  polytype phase
     tra  transient-structure phase
               non  non-crystalline phase (but it might be better to use
          some of the values given above)
It would be necessary to develop some tight definitions for these
flags to allow their unambiguous assignment, indicating which
might be subgroups (e.g, amp and gls as subgroups of non) so that
a search program would recognize amp and non (for example) as a
match.  If two of these flags are needed (e.g., xtl and pol) they
should be assigned to different components.  Each component
should contain only one flag.

Items c to e below would only be needed for a crystalline phase.

c) crystal system
     This would be one of the standard symbols currently used in
the Pearson symbol:
a, m, o, t, h and c.

d) space group number
     The allowed range is from 1 to 230, but enantiomorphic space
groups would have to be arbitrarily assigned to the lower space
group number, e.g., 144 should be used to represent both 144
(P31) and 145 (P32), that is, 145 would not appear in the
enumeration list for this component.  Chirality, if known and
important, should appear as a separate component (any suggestions
how this should be expressed?)   The space group may not always
be known, in which case the crystal system would provide partial
information.  Other possible partial symmetry components would be
the crystal point group and the lattice centring (P, E (one face
centred), F, I, C).  If all trigonal, hexagonal and rhombohedral
space groups are represented by the crystal system component 'h',
should we include R centring to identify rhombohedral space
groups?  All symbols used to describe the symmetry must be
setting-independent.

e) atom count in unit cell
     This is used in the Pearson symbol and, providing the unit
cell is clearly identified, it should be unambiguous except where
partial occupancy occurs.  It might be possible to normalize this
to the atom sites indicated in the chemical formula, e.g., in
'Fe* S' discussed above, the atom count would be 24 for space
group 190 (P-62c).  That is there are nominally 2 atoms in the
formula unit and these are repeated 12 times by symmetry in the
unit cell. Since this number must be an integer, how should it be
rounded in the cases where the number of atoms in the cell is
non-integral?

f) CAS number
     If this were used it would have to be a number that
corresponds to the actual composition of the phase.  Different
CAS numbers may be assigned for a molecule and for the same
molecule with solvent of crystallization, e.g., CuSO4 and CuSO4.
5H2O have different CAS numbers.  Other possible external
components such as the CSD REFCODE are usually specific to a
substance but not to the phase.

I welcome discussion on these proposals and suggestions for other
items that might be included in the phase identifier.  Please
respond by replying to this message.  That will ensure that your
reply is linked to this message when browsing the discussion list
by thread.

                    Best wishes

                         David


*****************************************************
Dr.I.David Brown,  Professor Emeritus
Brockhouse Institute for Materials Research,
McMaster University, Hamilton, Ontario, Canada
Tel: 1-(905)-525-9140 ext 24710
Fax: 1-(905)-521-2773
idbrown@mcmaster.ca
*****************************************************

Reply to: [list | sender only]
International Union of Crystallography

Scientific Union Member of the International Science Council (admitted 1947). Member of CODATA, the ISC Committee on Data. Partner with UNESCO, the United Nations Educational, Scientific and Cultural Organization in the International Year of Crystallography 2014.

International Science Council Scientific Freedom Policy

The IUCr observes the basic policy of non-discrimination and affirms the right and freedom of scientists to associate in international scientific activity without regard to such factors as ethnic origin, religion, citizenship, language, political stance, gender, sex or age, in accordance with the Statutes of the International Council for Science.

Useful links