Beam characteristics

Beam intensity should normally be fixed to 1.0, leading to a total reflectivity of 1.0 at low angles. If for some reason this is not the case (e.g., because of an incident medium absorption coefficient not accounted for during reduction) then you can adjust the beam intensity here. Beam intensity is also a fit parameter, BI.

Background is the expected background count rate, below which all data is noise. With sensitivity down to around 1e-7 for the reflectivity measurements, the default background level of 1e-10 is typical and is essentially zero. Background is also a fit parameter, BK.

Wavelength in Angstroms is the incident wavelength of the beam. The beam is assumed to be monochromatic. At the time of writing, the following defaults are assumed for the instruments:

	XRAY 1.5416 Angstroms
	NG-1 4.75 Angstroms
	NG-7 4.768 Angstroms
	CG-1 5.0 Angstroms

Wavelength is available in the header of the reflred data file and eventually reflfit will be clever enough to read it, but for now you must enter it yourself.

Wavelength divergence in Angstroms is the spread of neutron velocities. It does not vary with Q. Typical values are characteristic of the monochromator at the instrument:

	XRAY 0.005 Angstroms
	NG-1 0.05 Angstroms
	NG-7 0.05 Angstroms
	CG-1 0.05 Angstroms

Angular divergence in radians is the spread of incident neutron angles. Since angular divergence depends on the collimation, and slits are typically opened with increasing Q, there is an implicit Q dependence in the value used. We currently do not have an algorithm for determining angular divergence directly from the slit settings.

After the ideal reflectivity is computed for the layers described in the layer table, a convolution is performed to account for resolution of the instrument. The convolution uses a gaussian whose full-width at half maximum (w) is computed from the wavelength (L), the wavelength divergence (dL) and the angular divergence (dtheta). At each point Q in your data set, w is set to

w = Q * (dL/L + dtheta/theta),

where theta is computed from the linear approximation

theta = L*Q/4/pi.

The convolution at Qj is computed by the normalized sum over i of

Ri * exp(-(Qj - Qi)**2 / (2 (w / sqrt(8 ln 2))**2)

as long as the contribution to the convolution at your measured datapoint Qi is greater than 0.1%. To handle the boundary points, the data set Q values are extended towards positive infinity and towards negative infinity using the Q-spacing at the appropriate boundary of the data set.

2003-09-22


Browse Index