Having tried SHELX(S),SIR2004,(SUPER)FLIPPER(WinGX) etc. to no avail and before badmouthing the chemist for really not trying,reexamine:

  • Data: Weak(I/sigma(I) < 1 is going to be a challenge)
  • Composition: A badly incorrect formula can throw out many direct methods programs; some charge flipping programs are more resistant as they use Fs and not Es,
  • Cell: Danger signs include:
    • Large number of reflections not 'explained'.
    • Easily changable cell.
    • Unexpectedly large cell.
    • Uncommon crystal system: most common S.G.s:P21/c, P-1, P212121, C2/c, P21, Pbca, Pna21, Cc, Pbcn and C2
    Cell_now may be of assistance, even when nonmerohedral twinning, for which it was designed to find, is minimal or absent.
    cell_now [1.p4p](loaded with selected rlps)
    From the selection of cells produced choose one(usually the first, but may prefer e.g. 'higher symmetry' other). This is used to index the rlps in various orientations*; a (multiple) p4p file is written after each and RLATT(current .p4p file) can be used to view colour coded rl:white rlps indexed by first component,green by current component and red are unindexed; an analysis shows how many are unique to each orientation, how many composite and how many left.


  • Disorder: The implicit model should be chemically plausable;e.g. bent phenyl rings aren't. Working on the principle that 'Nothing succeeds like excess' W(hole) M(olecule) D(isorder) may be allowable. This illustates the use of the FRAG facility in SHELXL taken to this extreme.
    Have the right bits in each PART; swapping may help(see s2512)
  • Twinning:If not apparent at the indexing stage, the |E**2 - 1| statistic should give a clue as does the presense of METRICALLY higher symmetry cells , both which may be found in XPREP. Pseudo-translation CAN be indicative but may be just indicating centring etc.
    • RTFM SHELX and see if any of the 'receipe' matrices apply
    • Use WinGx to run one of the twinning programs to generate matrices and HKLF5 data
    • Heroically do it by hand:E.g. if metric symmetry is higher than the actual as in Orthorhombic instead of Monoclinic,to see if a non-monoclinic operation(nm-op) is causing the problem:
      Mathematica may be found useful
    • Collecting data at a different temperature may help, i.e. closer to the preparation temperature to minimise the probability of any phase change.
▲ Up to the top