Hydrogen contamination in the sample (often in the form of -OH or H₂O) presents serious difficulties for the accurate analysis of non-crystalline neutron diffraction data and is best avoided or minimised if at all possible. The reasons for this are firstly that hydrogen has a very large incoherent cross-section (~80 barns) leading to a large effective background, and secondly that the inelasticity effects for this background are very large for hydrogen, making it very difficult to deal with. It is thus most worthwhile to be able to assess during an experiment whether the sample is contaminated with hydrogen. Since neutron scattering has such a high sensitivity to hydrogen, it can sometimes be relatively straightforward to make such an assessment, and it is common for previously unsuspected hydrogen contaminations to be revealed, especially by pulsed neutron scattering. There are two approaches which can be used to judge whether a sample may be contaminated with hydrogen:

1. A quick analysis of the data by use of the AUTO program (see section 4.4) can be helpful in deciding whether a sample may be contaminated with hydrogen. Figure 7 shows the results obtained by running AUTO and PLAUTO for a phosphate glass sample which is highly contaminated with hydrogen. In this case the background level due to hydrogen contamination is clearly evident: Because of the Placzek inelasticity effect, the hydrogen background rises strongly at low Q , an effect which becomes more severe as the scattering angle is increased.
   
   
2. It is also possible to judge whether a sample is contaminated with hydrogen by studying its total cross-section s_tot(l ) as a function of wavelength. This may be extracted from the data measured by the LAD transmission monitor by running the TCR program described in the ATLAS manual [2], although a roughly correct value for the effective number density of the sample is necessary in order to obtain results which are sufficiently quantitative to investigate the possible presence of hydrogen. If the measured cross-section is roughly consistent with that calculated from tabulated cross-sections by the MUT0 program then the sample probably does not have a significant hydrogen contamination. However, a measured cross-section significantly in excess of that calculated from tabulated cross-sections indicates that the sample may well be contaminated with hydrogen.

Figure 8 shows the cross-section per molecule for H₂O [6] - this shows a wavelength-dependence which is dominated by the difference between the free (~20 barns) and bound (~80 barns) cross-sections of a hydrogen atom. Thus the additional cross-section due to hydrogen contamination will increase as the wavelength increases. Note, however, that the wavelength-dependence of the hydrogen cross-section can appear quite similar to a 1/v absorption cross-section and care should be taken to try and distinguish between these two possibilities when using the cross-section to investigate possible hydrogen contamination.

Last Updated 09 Nov 1998