Background Subtraction

Ideally we would want to measure only the absorption or scattering from the sample in any neutron experiment. However, in reality we also measure neutrons that are scattered from the sample environment, sample holder, and perhaps even stray neutrons or other particles that are not from the experiment. All of these contribute to what is called the background scattering. Some of these can be large contributions and some are small, but they are all unwanted in the experiment especially if the signal that we are trying to measure is subtle. If you look at the data below you can see a diffraction pattern with a smooth continuous background and some diffraction lines from the materials that are from the sample environment, in this case a cryostat. One way to deal with the this is to perform a background subtraction.

Two plots showing the scattering of neutrons in an environment. On the x-axis is the angle and on the y-axis is the intensity. The first plot shows both the raw data from the background and the sample, but also the data from only the background. The second plot shows the data from when the background is subtracted from the data consisting of both background and sample. This plot has therefore peaks with less intensity then the first graph, as some of the data has been subtracted.

Figure. A graph example of background subtraction.

Background subtraction is part of the data analysis which is performed after the measurement. What we need to do is measure the sample environment before or after we measure the sample in it. This enables us to measure the scattering from the sample environment and any other contribution to the background that does not come from the sample itself. We can then subtract this measurement from the sample measurement and obtain a cleaner data set of the sample as seen on the figure. This does not remove all unwanted neutrons in our database sample environment, for example there could be double scattering events between the sample and the sample environment, where a neutron scattered by the sample in a direction is then scattered again by the sample environment. However, this is a very rare event and would only influence very subtle measurements. Also the incoherent scattering from the sample is not removed in this way. Incoherent scattering is when neutrons are scattered by the sample but not from any internal structure, so that is has equal chance of being scattered into all angles. This give a flat background on the measurement. The amount of incoherent scattering from a sample depends on the incoherent scattering cross-sections the isotopes in the sample. For instance, hydrogen has a very high incoherent scattering cross-section, so if the sample contains hydrogen there will be a large background from the sample. If this background is too high we cannot obtain useful structural data from the sample.