7. Answers to a few questions
7.1 Can the code handle data in LEED format?
No, it cannot and it never will. You have to convert your data to standard crystallographic units including the symmetry.
7.2 What if I do not know the symmetry?
You should know the Patterson symmetry (symmetry of the diffraction pattern), and from this you can work out what possible symmetries you might have. For a surface, in some cases you can use STM images to eliminate possibilities. For a bulk TED case, you can used CBED. In the worst case, you will have to run calculations for different symmetries and add the set of symmetries to your feasible set. (This is not uncommon for surface data.)
7.3 What if I do not know the number of atoms?
You can run the program for different assumptions about the contents. The "guess" method where you only set 1 atom is, very empirically, not bad. You should consider the number of atoms as an element of your feasible set. (In general, the precise number of atoms does not matter.)
7.4 What do I do about the Temperature Factors for a surface?
While there are literature methods for estimating these, I have never seen them work for surface data. If you do not specify them, the program will use a default value. A reasonable guess is the bulk value. It is generally wise to underestimate these, but in practice they are not so important.
7.5 What do I do about the Temperature Factors for bulk TED data?
This has a very complicated answer. Because of dynamical effects, rather than simply have peaks at the atomic sites you can have larger features. To be more accurate, you can have hydrogrenic states similar to 1s, 2s and so forth in the two dimensions normal to the beam. Light atoms will be primarily 1s in character, heavier will have more 2s. As a consequence, heavier atoms have in effect much larger temperature factors. It is reasonable to use 0.3 for a light atom, and 3.0 for a heavy atom (e.g. Br). Some more work needs to be done here, and the numbers are thickness dependent. (Remember, you do not have Kinematical diffraction, you are finding feasible solutions only.)
7.6 What if I have twins in my data?
Twinning, particularly for surface data, is a severe problem. This is particularly the case if there is double-positioning, i.e. two different domains which average to a higher symmetry. You can run in the higher symmetry, and in some cases you will find partially occupied sites. You will then have to refine either with partial occupancies of the two (or more) or as an incoherent summation. Which you do depends upon the coherence of the source in your experiment, and there is no simple answer (i.e. try both). In some cases you may have three different domains where all the spots overlap giving an apparent p6mm Patterson symmetry although the true structure may only be p2mm. At present, I know no way to solve this, although there are possibilities. (The peaks2D code does not at present include twinning, and this will be added to it at some future date.)
7.7 Can the program handle measurement errors?
Not at the moment. In principle, these can be included but this requires more work. They will be included in peaks2D at some later date.
7.8 Why does the program not do a full solution, as 3D codes do?
In two dimensions atoms can overlap, so it is hard to automate the last step. The program peaks2D finds most of the peaks, but in many cases you will have to do this yourself. (If you have SEMPER, a set of programs is available.)
7.9 Does the program work in 3D?
There is a version for 3D surface structures, but it will not be released until some time in 1999 at the earliest.
7.10 The program gives rubbish for my bulk TED structure
If your thickness is more than 10-20nm and/or it does not contain well separated columns of atoms, it is likely to give wrong results. More work is needed for the general case with bulk TED data. If it is reasonably thin and has well-separated atomic columns, check that you are only including well determined phases (from HREM).