research papers
Calibrating SANS data for instrument geometry and pixel sensitivity effects: access to an extended Q range
aHeinz Maier-Leibnitz Zentrum (MLZ), Technische Universität München, Lichtenbergstrasse 1, 85747 Garching bei München, Germany, and bGerman Engineering Materials Science Centre (GEMS) at Heinz Maier-Leibnitz Zentrum (MLZ), Helmholtz-Zentrum Geesthacht, Lichtenbergstrasse 1, 85747 Garching bei München, Germany
*Correspondence e-mail: sebastian.busch@hzg.de
An improved data-reduction procedure is proposed and demonstrated for small-angle neutron scattering (SANS) measurements. Its main feature is the correction of geometry- and wavelength-dependent intensity variations on the detector in a separate step from the different pixel sensitivities: the geometric and wavelength effects can be corrected analytically, while pixel sensitivities have to be calibrated to a reference measurement. The geometric effects are treated for position-sensitive 3He tubes, where they are anisotropic owing to the cylindrical geometry of the gas tubes. For the calibration of pixel sensitivities, a procedure is developed that is valid for isotropic and anisotropic signals. The proposed procedure can save a significant amount of beamtime which has hitherto been used for calibration measurements.
Keywords: small-angle neutron scattering; data reduction; calibration; corrections.
1. Introduction
Small-angle neutron scattering (SANS) is a commonly used technique to determine the size, shape and arrangement of structures on a mesoscopic length scale in physics, chemistry, biology and materials science (Lindner et al., 2000; Wang et al., 2004; Petoukhov & Svergun, 2005; Davies et al., 2006; Schmidt-Rohr & Chen, 2008; Fratzl, 2003; Das et al., 2012; Martin et al., 2013; Milde et al., 2013; Gilles et al., 2014; Jaksch et al., 2014; Tokunaga et al., 2015; Lang et al., 2016).
Of course, the measured data have to be reduced in order to remove instrumental influences and obtain pure information about the sample. There are a wide range of publications on the subject of correcting SANS data. The goal is to obtain a smeared sample-dependent scattering ; Strunz et al., 2000; Keiderling, 2002; Ghosh et al., 2000; Zemb & Lindner, 2002; Brûlet et al., 2007; Zhang et al., 2010; Barker & Mildner, 2015). All these methods rely on a common step, namely the calibration of the measured sample scattering signal by the signal of a well characterized isotropic scatterer, such as water, Plexiglas or vanadium, measured at the same instrumental settings as the sample.
(Wignall & Bates, 1987These calibration measurements serve three purposes: (i) correction of geometric effects of the instrumental setup; (ii) correction of different detection sensitivities of different detector pixels; and (iii) calibration of the measured signal to absolute units. Unfortunately, several problems arise when performing such a calibration measurement. It is basically impossible to find a sample that scatters completely isotropically, even if incoherent, owing to inelastic (May et al., 1982; Lindner et al., 2000) or multiple (Copley, 1988; Shibayama et al., 2005) scattering. Additionally, large detector distances require unfeasibly long measurements to achieve reasonable statistics. Therefore, the measurements are usually performed at short and medium detector distances and used for large detector distances as well, or the measurements at large detector distances are not corrected at all.
It is therefore advantageous to create a detector efficiency file at an optimal geometry for the respective standard and to transform it to the actually used geometry afterwards. This results in a significant reduction in the beamtime that has to be spent on calibration measurements. To do so, an accurate solid-angle correction has to be performed and possible anisotropic effects of the detector system must be accounted for. In this contribution, several non-trivial effects are considered that are important for detectors consisting of an array of position-sensitive 3He tubes. This detector type is used in many instruments, such as SANS-1 at the Heinz Maier-Leibnitz Zentrum (MLZ) in Garching near Munich, Germany (Mühlbauer et al., 2016; Heinz Maier-Leibnitz Zentrum, 2015), KWS-1 at MLZ (Feoktystov et al., 2015), D22 at ILL, Grenoble, France (ILL, 2016), LOQ/SANS2D at ISIS, Oxfordshire, UK (Heenan et al., 1997; Duxbury et al., 2014), EQ-SANS at SNS, Tennessee, USA (Boukhalfa et al., 2013), V4 at Helmholtz-Zentrum Berlin, Germany (Keiderling & Jafta, 2016), and QUOKKA at OPAL, New South Wales, Australia (Gilbert et al., 2006).
The solid-angle correction for these instruments differs from that for multi-wire area detectors since the cylindrical detector tubes have an anisotropic geometry. In addition, shadowing effects arise at large angles, since multiple detection volumes intersect the trajectory of a scattered neutron. The neutron might even pass a first detector tube and be recorded in a neighbouring one, creating a parallax effect. Since the transmission of 3He depends on the neutron wavelength, this effect has to be understood for the whole neutron spectrum of the source. Recent SANS instruments support even more extreme geometries, such as a sideways movable detector to reach scattering angles 2θ of ∼45° (SANS-1 at MLZ). Additional corrections have to be taken into account if the detector is rotated to compensate partially for the different incident angles on the detector plane.
In the following, a procedure is proposed which solves several of the above-mentioned problems for position-sensitive 3He tubes. The tools for transforming a calibration measurement to arbitrary SANS geometries are provided. This includes all available wavelengths from a cold neutron source and large scattering angles and is tested up to ∼40°. In addition, a novel procedure to correct the pixel sensitivity is introduced. Rather than relying on a completely isotropic scattering pattern, it works with any scattering pattern that is measured twice with a small lateral detector movement in between. This is advantageous, as basically no sample scatters isotropically, as mentioned above.
Several effects will be discussed that distort the measured intensity: the varying solid-angle coverage of the pixels (a change in the intensity of the order of 50%), shadowing (of the order of 25%), pixel sensitivity (of the order of 10%), scattering angle-dependent sample transmission (of the order of 3%) and finally the parallax effect (of the order of 1%).
This contribution is organized as follows. First, the geometry of the instrument will be defined. Then, four factors that influence the detected intensity will be discussed in turn: (i) the solid angle covered by the pixels; (ii) an idealized detector efficiency including shadowing and parallax effects; (iii) different pixel sensitivities; and (iv) the sample transmission for scattering at large angles. Finally, the combined influence of these factors will be demonstrated on a measured scattering pattern of vanadium.
2. Geometry of the instrument
SANS data are usually given by an intensity matrix , where a × b is the detector image resolution (for SANS-1 at MLZ, a = b = 128). Coordinates in this detector-based coordinate system will be denoted by uppercase vector R = (X, Y, Z) (see Fig. 1a). The origin is set to the top-left pixel as seen from the sample position. XY is the detector plane and Z is the direction of the detector normal vector, all measured in pixels (px) and enumerated by i,j with and . At SANS-1 at MLZ, the detector consists of tubes with an active length of slightly over 1 m, an inner diameter of 7.324 mm and a pitch of 8 mm. Each pixel has a size of px = py = 8 mm.
The coordinates of all pixels can also be given in a sample-based coordinate system, denoted by lowercase vector r = (x, y, z) (Figs. 1b and 1c). The origin is the centre of the illuminated sample volume, z is the direction of the incident neutron beam, and x and y are in the horizontal and vertical directions, respectively. The units of this system are metres. If the detector is not rotated (ω = 0), the two coordinate systems have parallel directions of their normal vectors, , i = x, y, z.
It is customary for the detector to be able to move along the z direction in order to resolve a wide range of neutron scattering angles 2θ. In more flexible setups, such as that of SANS-1 at MLZ, the detector can also be moved sideways by a lateral displacement xpiv and rotated by an angle ω. For this purpose, it is mounted on a hinge that has a rotation axis parallel to y. The pivot point is located in the detector system at Xpiv = 63.5 px and Zpiv = 9.14 px 73.1 mm.
The affine transformation between the two coordinate systems is given by
with the dimension matrix
the rotation matrix around the y axis
and translation vectors
where (Cx, Cy) is the position of the direct beam centre on the detector. Note that gravity effects can be taken into account in the vector T2 through translation of the system's Y coordinate by the corresponding offset.
The set of pixels in the sample system is defined by their centre coordinates = (xi,j, yi,j, zi,j) and the set of pixels in the detector system by their number = (Xi,j, Yi,j, 0) , , . With the above transformations, the pixels can be converted from one system into the other.
Pixel pi,j has a distance
from the sample and the shortest distance between the sample and the corresponding detector tube is at yi,j = 0, i.e.
A neutron that is detected in pixel pi,j has been scattered by the sample under a scattering angle
The incident angle with which the neutrons impinge on the detector can be calculated using the detector normal nDet = [−sin(ω), 0, cos(ω)] as
Owing to the geometry of the problem, it is beneficial to use spherical coordinates. The coordinate transformation for vector r is given by (Fig. 1c)
In this coordinate system, pixel pi,j has an azimuthal angle of
and a polar angle
The signum function is used to orient the angles in a positive or negative direction.
If gravitational effects are taken into account, the polar angle ϑ can easily be adjusted for the loss of height of the neutrons after longer flight distances by an offset .
The instrument sample-to-detector distance d0 conventionally assumes that the detector is perpendicular to the direct beam direction. For detectors that can be moved laterally and rotated by an angle ω around a pivot point, which is not necessarily in the detector plane, a different definition is needed. The distance for the primary beam detected in pixel C is given by [as calculated from equation (1)]
and the shortest distance from the sample-to-detector plane is
3. Solid-angle corrections for a flat two-dimensional detector
It is instructive to look first at the detector as if it were a completely flat two-dimensional plane. In this case, the only effect that has to be accounted for is the variation in the solid-angle element of the different pixels. This yields the well known solid-angle correction (Grillo, 2008; He et al., 2015)
where is the distance from the sample to the detector and px,py are the pixel sizes in the x and y directions, respectively. The first term represents a scaling factor taking into account that a pixel of a given size covers a smaller solid-angle element when the detector is further away from the sample. The second term, , takes into account that the detector is not a spherical shell and therefore pixels with are even further away from the sample. The third term, , accounts for the decrease in solid-angle coverage because of the angle between the pixel surface and the scattered neutron beam.
For a detector consisting of cylindrical tubes, the correction has to be adjusted because of the broken symmetry. While the cylinder surface tilts away from the sample as the scattering angle increases along the cylinder axis, just as it does for a flat detector, this is not the case for sideways scattered neutrons that reach a different detector tube. He et al. (2015) suggest a correction for the tube geometry according to
where is the y component (i.e. the polar angle) of the scattering angle 2θ, i.e. = .
This formula can be adjusted to a detector system rotated by an angle ω by introducing the incident detector angle from equations (8) and (11) (Fig. 1). Therewith, is replaced by , cos (2θ) by cos(α) and by the polar angle
This correction is shown in Fig. 2 and is indeed the most important term of all the corrections discussed in this contribution. However, for d below ∼2.5 m, it becomes apparent that this treatment of a tube detector as a flat two-dimensional structure is not enough and some effects caused by the three-dimensional nature of the tubes are not accounted for. These additional corrections will be discussed below.
4. Idealized detector efficiency and shadowing effects in a three-dimensional detector
At large scattering angles the detection volumes partly shadow each other, effectively reducing the illumination intensity. In the x direction there is a certain critical angle at which the shadowing effect starts. In the y direction shadowing is present for any . To correct for these effects, the idealized efficiency of pixel pi,j (not taking into account manufacturing differences or effects like dead time) is calculated by a numerical integration of a function of the form
where E and T are, respectively, the absorption and transmission probability functions of a neutron with a flight trajectory through the infinitesimal detecting element dΩ. The function E describes the probability that a neutron is absorbed within a given pixel and T the probability that neighbouring pixels have not absorbed the neutron on its way. denotes the solid-angle element covered by pixel pi,j. arises from the absorption function of the tube wall material, where n is the number of walls that the neutron intersects between the sample and 3He. Normally, n = 1, but if several tubes intersect, i.e. if shadowing occurs.
In order to calculate E and T, the functions are parametrized in spherical coordinates (Fig. 1c):
and
Note that d and are essentially constant over a single solid-angle element , i.e. it is possible to write for the Jacobian determinant in the integral. The use of a cosine instead of the usual sine function is due to the definition of the coordinate system.
The transmission of 3He depends on the gas pressure p, the neutron wavelength λ and the flight path length s that the neutron has to travel within the gas (Masalovich, 2007):
The absorption accounts for the neutrons which were not transmitted:
The distance s is a function of the exact position and angle of the incoming neutron and will be parameterized below.
In the following, the influences of the absorption probability E and of the shadowing and parallax effect T will be discussed in four stages: first, only the absorption probability without shadowing effects will be discussed at the same height as the sample, i.e. y = 0, before shadowing and parallax are included. Then, again without shadowing, the absorption probability will be described in the x and y directions, and finally the shadowing and parallax effects are included in these directions.
4.1. Absorption probability E in the x direction (y = 0, no shadowing)
At this point let (y = 0, ϑ = 0), i.e. the neutron flight trajectory is within the horizontal plane from the sample position. Hence, neutrons within this plane `see' a horizontal cut through the cylindrical tube, which is a circle of radius r. The azimuthal opening angle ψ from the sample position to this circle is
With being the distance from the centre of the tube (r is the tube radius), the path length function s of the neutron depends on h according to (see Fig. 3)
The distance h further depends on the azimuthal opening angle ψ and the pixel distance dxzi,j from the sample:
By definition of the sample-to-pixel distance di,jxz in equation (6), these opening angles are also valid for a detector system rotated by an angle ω. For pixel pi,j, the azimuthal angle φ of the spherical coordinate system is related to the azimuthal opening angle ψ by
Using this substitution, the integrand of function (19) for non-shadowed pixels along the y = 0 plane is a function of the azimuthal opening angle ψ of pixel pi,j:
The effect is shown in Fig. 4(a) for different neutron wavelengths. It can be seen that the absorption probability decreases close to the left and right edges of the tube because the neutron travels through a smaller volume of the 3He gas.
4.2. Including the shadowing and parallax effect T in the x direction (y = 0)
Since the neutrons can hit the detector at an angle, they can cross other detector tubes before they are recorded. As a result, neutrons on the same flight trajectory can be recorded in two neighbouring pixels and a parallax effect emerges. From Fig. 5, it can be seen that the corresponding shadowing effect in the x direction occurs for
For the detector on the SANS-1 beamline at MLZ, φcrit ≃ 21° + ω.
In order to take this effect into account for pixel pi,j at azimuthal angle , the transmission function T in the integrand of equation (19) is determined by the neutron flight path length through the neighbouring tubes that cover the same solid-angle element (Fig. 5). For y = 0 and ϑ = 0, the path length through the neighbouring tubes can be determined as a function of the azimuthal angle φ shifted by the offset of the neighbouring tube's azimuthal angle = − . Here, a plus sign + denotes the case when the tube is shadowed on the right-hand side and a minus sign − the case on the left-hand side:
where the substitution of equation (25) was used for the azimuthal angle φ in the second step. An additional factor of the tube material transmission is required, since the neutrons cross the material again when leaving the shadowing tube. The neutron damped by the neighbouring tubes is depicted in Fig. 4(b).
4.3. Absorption probability E in the x and y directions (no shadowing)
The intersections of the horizontal plane y = 0 with the detector tubes are circles of radius r. The situation changes if the neutrons are scattered up- or downwards and impinge on the detector at a nonzero angle ϑ in the y direction: the mean path length is increased because the neutrons `see' a skewed cut through the cylinder, which is in fact an ellipse (Fig. 6) with increasing eccentricity (Fig. 7). Hence, there is a higher probability of them being detected. For pixel pi,j, the circular is deformed to an ellipse with minor axis r and major axis (Fig. 6) and hence s is increased by a factor of in the integrand of equation (26):
Note that is essentially constant over the solid-angle element and it is therefore possible to use .
4.4. Including the shadowing and parallax effect T in the x and y directions
As can be readily seen in Fig. 7, the shadowing in the x direction occurs for at exactly the same positions as for y = 0, i.e. the considerations from above are also valid in this case. In the y direction, shadowing and parallax occur for any polar angle . The path length function s has to be considered within each respective detector voxel separately. Its intersection with the upper and lower end planes of the voxel (Fig. 8) has to be calculated.
Without neglecting useful cases, it can be assumed that the neutron trajectory intersects not more than one of the confining horizontal planes. If the neutron intersects the upper or lower plane, the path length function is subdivided into s = sin + sout, with the path length within the voxel sin and that outside of it sout (Fig. 8). This subdivision can be calculated analytically by geometric considerations (Appendix A). Subsequently, similar to the x direction, the transmission function T for pixel pi,j from equation (28) is further extended by the transmission function of the neutron flight path sout in the voxel above () or below ():
The idealized efficiency is therefore determined by an integration over the azimuthal opening angle ψ and the polar angle ϑ. The integration limits of the polar angles are determined by the confining planes A and E according to
Finally, the idealized efficiency of pixel pi,j is calculated by
The integration can be performed numerically, for example by Gauss–Tschebyschow quadrature (Abramowitz & Stegun, 1965), with reasonable speed and very good accuracy.
The effect of a shadowing neighbouring tube on the right for a single pixel is shown in Figs. 9(c) and 9(f). The effect of a shadowing neighbouring voxel below in the same tube can be seen in Fig. 9(f).
Figs. 10(a) and 10(b) depict the idealized efficiency function due to the shadowing effects and changed neutron path length s in the tube for neutron wavelengths λ = 4.9 and 12 Å without any solid-angle effects. The detector is positioned at d0 = 1.111 m and xpiv = 0.5 m. The shadowing effect is strongly visible, as well as a weak efficiency increase in the Y direction due to the increase in the neutron flight path s within the tube. Fig. 10(c) shows the same situation but with the detector system rotated by ω = 20°. The visible effect is purely the efficiency change due to increased s. With this geometry the shadowing is not visible in the two-dimensional image.
5. Pixel sensitivity
So far, only geometric effects have been dealt with – it has been implicitly assumed that every neutron absorbed by 3He corresponds to a detection event. This is not the case in practice and every pixel has an individual sensitivity ∊. The task of determining these sensitivities is often performed using an incoherent scatterer (vanadium, Plexiglas or water) which is assumed to scatter isotropically. As mentioned above, even a purely incoherent scatterer cannot be expected to result in a flat intensity distribution on the detector after all geometric corrections. In the following, a procedure is presented which does not require the sample to scatter isotropically, but instead uses two measurements of the same arbitrary sample with a displaced detector to determine the pixel sensitivity, inspired by procedures used for X-rays (Hagemann et al., 2017).
For this procedure to work, it is essential to have measurements in which the intensity fluctuations due to counting statistics are much smaller than the systematic effects that are to be corrected. In the current case of sensitivity differences of the order of 10−2, one would aim for a error of the order of 10−3 or better, corresponding to at least 106 neutrons per pixel. A strongly scattering sample is therefore beneficial; in the present work, glassy carbon was used.
Two measurements of the scattering of glassy carbon were performed, one with the detector further left (denoted l) and one with the detector shifted one pixel distance of 8 mm to the right (denoted r). The pixel number i in a given row increases from left to right, while the row number is not given for clarity. When the detector is in position r, pixel number i is illuminated with a true intensity Ii and will register ci counts owing to its sensitivity ∊i, with statistical error (ci)1/2. After translating the detector left into position l, the true intensity Ii is now recorded in pixel number i + 1:
which can be combined and rearranged to
where ri is a shorthand notation for the ratio of measured counts and σ is the propagated statistical error of the counts. In reality, the ratio of the sensitivity and the ratio of the measured counts are not identical for all pixels owing to statistical fluctuations. Instead of being equal, they should be as similar as possible, i.e. the aim is to minimize the quantity
The whole set of pixel efficiencies is the solution to
The gradient vector of this function can be calculated analytically, which speeds up the fitting process considerably.
The advantage of this procedure, using the ratios of neighbouring efficiencies, is that it works with arbitrary scattering patterns. Drawbacks are the necessity for very good statistics and the cancellation of a factor that is constant over the whole detector, i.e. an overall detector sensitivity prefactor. It follows that, with this procedure alone, the data cannot be put on an absolute scale; an additional constraint – for example the comparison of a glassy carbon measurement with the result of an X-ray measurement of the same sample – is required to fix this prefactor and obtain absolute units.
The result of the procedure presented in this section is shown in the results, using the measurements of a glassy carbon sample.
6. Sample effects
There are some effects of the sample which influence the detected intensity so that it is not proportional to the scattering function. They have been described in detail elsewhere and are summarized in the following.
6.1. Transmission of the sample
When neutrons are scattered to large angles (i.e. 2θ 5°), the sample transmission is a function of the scattering angle 2θ and an angle-dependent transmission has to be applied. This effect can be neglected at very small angles. However, it has to be taken into account when larger scattering angles are considered.
The neutron absorption from a plate-like sample with thickness l is a function of the sample thickness and a material constant μ. Experimentally, the constant μ is usually approximated by a measurement of the sample transmission Ts at 2θ = 0 (Calmettes, 1999):
This attenuation of the transmitted beam is, on the one hand, due to simple absorption and, on the other hand, due to scattering (both coherent and incoherent) to large angles where the neutrons no longer impinge on the detector and are effectively lost.
An additional effect that has to be taken into account is that the distance the scattered neutrons travel through a sample also depends on the scattering angle (Fig. 11). As the scattering angle increases, the flight distance increases. A neutron that is scattered after a distance x after entering the sample by a scattering angle 2θ traverses a distance of (Guinier, 1964)
For scattering events that occur after passing a distance x within the sample, the neutron amplitude is attenuated by a factor of with the material-dependent value μ. The neutron amplitude that is scattered at x to an angle of 2θ is attenuated by an additional factor . Then, the angle-dependent transmission is calculated analytically according to (Brûlet et al., 2007; Grillo, 2008)
where
with f(2θ) = −1 + 1/cos (2θ). The effect is depicted in Fig. 12.
6.2. Inelastic scattering
An Q is calculated wrongly, and second, the detected intensity depends on the energy-dependent detection probability of the detector and a normalization factor. This affects both the total number of neutrons recorded by the detector and the Q dependence of the scattering signal, since the amount of increases with Q. On spectrometers, the corresponding Debye–Waller factor can be accounted for; there are also correction procedures for diffractometers (Placzek, 1952) but they are less well defined.
between the neutron and the sample causes two problems. First, the momentum transferIt has been shown (Ghosh & Rennie, 1999; Ghosh et al., 2000) that an between the sample and the neutron is noticeable for many standards normally used in neutron scattering (in descending order water, Plexiglas, vanadium, glassy carbon) and has the systematic effect of increasing the neutron energy towards thermal wavelengths of ∼2 Å.
An ), which makes it particularly suitable for calibration purposes. Its only drawback is that the scattered intensity is markedly anisotropic, while the other, incoherent, standards scatter isotropically to a zeroth approximation. However, this does not pose a problem for the determination of the absolute scattering intensity, and the voxel sensitivities can also be calibrated with such an anisotropic scatterer, as shown in §5.
from the sample to the neutron requires both thermal motion in the sample and a scattering process that can transfer its energy to the neutron. probes single-particle motions and is therefore particularly susceptible to such an The coherent scatterer glassy carbon causes orders of magnitude less than the other standards mentioned above (Barker & Mildner, 20156.3. Multiple scattering
Multiple scattering distorts the scattered intensity (Schelten & Schmatz, 1980). The calibration methods presented in this contribution are valid independent of the amount of multiple scattering in the calibration standard but do not correct for multiple scattering in the sample.
7. Results and discussion
All measurements were carried out on the SANS-1 instrument at the Heinz Maier-Leibnitz Zentrum (MLZ) in Garching near Munich, Germany (Mühlbauer et al., 2016; Heinz Maier-Leibnitz Zentrum, 2015). The instrument is equipped with an array of 128 position-sensitive 3He detectors of type Reuter-Stokes P4-0341–201 installed with a pitch of 8 mm. They have an active length of slightly more than 1 m, an inner diameter of 7.324 mm and a gas pressure pHe of 15 bar (1 bar = 100 kPa). The detector system can be moved sideways by up to 570 mm and rotated by angles of up to 21°.
The determination of pixel sensitivities ∊ via equation (37) was performed with an established reference sample of glassy carbon (Zhang et al., 2010). Its fabrication procedure is described by Moreno-Castilla et al. (1980). The sample had a thickness of 1 mm and was measured with a neutron wavelength of λ = 4.9 Å, a collimation of 4 m and a detector distance of d0 = 4.0 m twice: once with a lateral displacement of the detector of xpiv = 4 mm and once with xpiv = 12 mm. In order to achieve the necessary pixel statistics, the sample was measured for 10 h at both positions. The resulting sensitivity matrix is shown in Fig. 13, by itself and in combination with the idealized detector efficiency and shadowing of equation (33).
It can be seen that the procedure works very well for local sensitivity fluctuations without distorting or smearing the general shape. In particular, there is no drift across the detector from top to bottom, which might have happened since the detector can only be moved sideways and not lifted. Some features that are most probably not in the true scattering pattern have not been removed completely, in particular a vertical line of lower sensitivity around X = 30 and a slight distortion of the otherwise azimuthally symmetric pattern around the beamstop. However, it was decided not to enforce azimuthal symmetry and not to correct for these small effects, in order not to introduce artefacts.
Such a pixel sensitivity fluctuation map can be determined once – here at a neutron wavelength of 4.9 Å – and then used at all instrumental geometries and neutron wavelengths. The overall wavelength dependence of the detection probability is included in equation (33). The fluctuations in the pixel sensitivities are independent of wavelength (Appendix B).
Using this sensitivity matrix, the rest of the procedure is demonstrated on the scattering of a vanadium sample which is a nearly perfect incoherent scatterer (Sears, 1992). For the vanadium measurements, a neutron wavelength of λ = 4.9 Å, a collimation of 6 m and a detector distance of d0 = 1.111 m were used. Additionally, offsets of xpiv = 500 mm with two distinct rotation angles ω of 0 and 20° were chosen to obtain different tube shadowing conditions. These settings require the strongest corrections of the commonly used SANS settings. The measurement time was always 30 min.
The efficiency maps for these measurement conditions were calculated by multiplying the effects from the solid-angle corrections of equation (16), the idealized detector efficiency and shadowing η of equation (33), the non-ideal detection probability ∊ of equation (37), and the sample transmission correction Ts of equation (41). Fig. 14 shows the efficiency maps for two detector angles ω. Comparison with Fig. 2 shows that the solid-angle effect is the leading order correction, but the other effects introduced in this contribution have a clearly visible influence.
These efficiency maps were applied to the scattering of a vanadium sample. Figs. 15(a) and 15(c) show the results of vanadium raw data for ω = 0 and ω = 20°. Around 2θ = 0 (marked by a cross), the area of the beamstop is masked. In addition, the detector is partly overshadowed from the nose system towards the detector on the right-hand side of the pattern, which was masked as well.
With the detector orthogonal to the incoming neutron beam, ω = 0, the scattering intensity decreases by ∼50% towards the largest scattering angles owing to the solid-angle and detector effects. The rotated detector has its shortest distance to the sample nearly at its centre. For that reason, the scattering intensity decreases around that point by 10% towards the sides of the detector. Fig. 15 shows the respective scattering patterns divided by the predicted efficiencies from Fig. 14. The corrections reproduce constant scattering patterns, superimposed with statistical noise, as expected from a vanadium sample. The measurement time of the sample was not enough to resolve the sensitivity fluctuations (Fig. 13) between the pixels, but it is clear that there are no distinct systematic features visible on the two-dimensional image after the correction.
The azimuthally averaged scattering cross sections are depicted in Fig. 15. Even at the largest 2θ values of ∼40°, the procedure reproduces basically a constant. The intensity shows a very small decrease at very large 2θ values, which might be induced by effects from vanadium. At ω = 20°, the averaged raw data show a kink at 2θ ≃ 23° which originates from superimposed effects in the vertical and horizontal directions. As soon as the scattering angle 2θ reaches the upper or lower boundary of the detector, the effects in the horizontal direction become suddenly more prominent, which shows the necessity for the applied anisotropic correction procedure.
8. Conclusions
This study has presented a comprehensive procedure for correcting the geometric distorting effects of 3He counter tubes. The procedure makes it possible to create detector efficiency maps for arbitrary SANS geometries from a single measurement. It has been demonstrated that the procedure works even for extreme SANS geometries such as scattering angles of 2θ = 40° and rotated detector systems.
Additionally, a procedure for correcting the individual pixel efficiencies by taking two measurements of glassy carbon with a laterally offset detector has been demonstrated.
The procedure developed in this contribution can be easily generalized to include instrumental resolution effects. The idealized detector efficiency η is a function of λ and can therefore be smeared with a known Δλ/λ distribution function.
Performing these calibration measurements with various wavelengths at the beginning of a reactor cycle allows the calibration of all sample measurements in a very effective way, to save beam time and improve the quality of the detected data.
APPENDIX A
Voxel limits
The intersection of a voxel's upper or lower confining plane with the trajectory of a neutron can be calculated analytically. In the following, the geometric considerations are presented (Fig. 16). Using equations (23) and (24), the function s for different ψ reads
The condition that plane D from Fig. 8 cuts plane B or C at the boundaries of a voxel a distance d inside the tube and a distance dxz from the sample position is fulfilled if
with
where + is the condition for plane B and − the condition for plane D.
The split = + is performed by the following set of equations for the case and (Fig. 16):
The other cases work analogously. sout/in is decided by the angle cases or and whether one looks at plane B or C.
APPENDIX B
Wavelength dependence of pixel sensitivity
The key idea of this contribution is to separate the influence of different instrumental settings (which can be calculated) from the varying pixel sensitivities across the detector (which have to be determined experimentally). The aim is to measure the pixel sensitivities at only one instrumental setting and apply them to all measurements. Since these will most probably include measurements at other neutron wavelengths, the features in the sensitivity matrix must therefore be wavelength independent for this procedure to work.
Such an independence of neutron wavelength is to be expected, as the variations in pixel sensitivity are caused after the conversion of neutrons to charged particles in the nuclear reaction with 3He. The energy of these particles is independent of the neutron wavelength, which therefore does not influence the result.
This expectation was tested experimentally by measuring a sample, in this case a 1 mm thick water layer, in one instrumental geometry with two neutron wavelengths, 4.9 and 6.0 Å. While features of the pixel sensitivity are clearly visible in both measurements (Fig. 17), they cancel each other perfectly when dividing one by the other. The pixel sensitivity fluctuations are therefore not dependent on the neutron wavelength in this wavelength range.
Acknowledgements
We would like to thank Jean-François Moulin (German Engineering Materials Science Center at Heinz Maier-Leibnitz Zentrum, Helmholtz-Zentrum Geesthacht) for fruitful exchanges and critical reading of the manuscript, Sebastian Mühlbauer (Technische Universität München) for stimulating discussions, Karl Zeitelhack (Technische Universität München) for valuable input regarding the inner workings of 3He detectors, and Jan Ilavsky (Argonne National Laboratory, USA) for providing the glassy carbon sample. This work was based on experiments performed at the SANS-1 instrument operated by HZG and FRM II at the Heinz Maier-Leibnitz Zentrum (MLZ), Garching, Germany.
Funding information
The following funding is acknowledged: Deutsche Forschungsgemeinschaft (grant No. GI 242/4-1).
References
Abramowitz, M. & Stegun, I. A. (1965). Editors. Handbook of Mathematical Functions: With Formulas, Graphs, and Mathematical Tables. New York: Dover Publications. Google Scholar
Barker, J. G. & Mildner, D. F. R. (2015). J. Appl. Cryst. 48, 1055–1071. Web of Science CrossRef CAS IUCr Journals Google Scholar
Boukhalfa, S., He, L., Melnichenko, Y. B. & Yushin, G. (2013). Angew. Chem. Int. Ed. 52, 4618–4622. Web of Science CrossRef CAS Google Scholar
Brûlet, A., Lairez, D., Lapp, A. & Cotton, J.-P. (2007). J. Appl. Cryst. 40, 165–177. Web of Science CrossRef IUCr Journals Google Scholar
Calmettes, P. (1999). J. Phys. IV, 9, 83–93. CAS Google Scholar
Copley, J. R. D. (1988). J. Appl. Cryst. 21, 639–644. CrossRef CAS Web of Science IUCr Journals Google Scholar
Das, P., Rastovski, C., O'Brien, T. R., Schlesinger, K. J., Dewhurst, C. D., DeBeer-Schmitt, L., Zhigadlo, N. D., Karpinski, J. & Eskildsen, M. R. (2012). Phys. Rev. Lett. 108, 167001. Web of Science CrossRef PubMed Google Scholar
Davies, T. S., Ketner, A. M. & Raghavan, S. R. (2006). J. Am. Chem. Soc. 128, 6669–6675. Web of Science CrossRef PubMed CAS Google Scholar
Duxbury, D., Heenan, R., McPhail, D., Raspino, D., Rhodes, N., Rogers, S., Schooneveld, E., Spill, E. & Terry, A. (2014). J. Instrum. 9, C12051. Web of Science CrossRef Google Scholar
Feoktystov, A. V., Frielinghaus, H., Di, Z., Jaksch, S., Pipich, V., Appavou, M.-S., Babcock, E., Hanslik, R., Engels, R., Kemmerling, G., Kleines, H., Ioffe, A., Richter, D. & Brückel, T. (2015). J. Appl. Cryst. 48, 61–70. Web of Science CrossRef CAS IUCr Journals Google Scholar
Fratzl, P. (2003). J. Appl. Cryst. 36, 397–404. Web of Science CrossRef CAS IUCr Journals Google Scholar
Ghosh, R. E., Egelhaaf, S. U. & Rennie, A. R. (2000). A Computing Guide for Small-Angle Scattering Experiments. Institut Laue–Langevin, Grenoble, France. Google Scholar
Ghosh, R. E. & Rennie, A. R. (1999). J. Appl. Cryst. 32, 1157–1163. Web of Science CrossRef CAS IUCr Journals Google Scholar
Gilbert, E. P., Schulz, J. C. & Noakes, T. J. (2006). Phys. B Condens. Matter, 385–386, 1180–1182. Web of Science CrossRef CAS Google Scholar
Gilles, R., Mukherji, D., Eckerlebe, H., Karge, L., Staron, P., Strunz, P. & Lippmann, T. (2014). J. Alloys Compd. 612, 90–97. Web of Science CrossRef CAS Google Scholar
Grillo, I. (2008). Small-Angle Neutron Scattering and Applications in Soft Condensed Matter, pp. 723–782. Heidelberg: Springer. Google Scholar
Guinier, A. J. (1964). Théorie et Technique de la Radiocristallographie. Paris: Dunod. Google Scholar
Hagemann, J., Robisch, A.-L., Osterhoff, M. & Salditt, T. (2017). J. Synchrotron Rad. 24, 498–505. Web of Science CrossRef IUCr Journals Google Scholar
He, L., Do, C., Qian, S., Wignall, G. D., Heller, W. T., Littrell, K. C. & Smith, G. S. (2015). Nucl. Instrum. Methods Phys. Res. Sect. A, 775, 63–70. Web of Science CrossRef CAS Google Scholar
Heenan, R. K., Penfold, J. & King, S. M. (1997). J. Appl. Cryst. 30, 1140–1147. Web of Science CrossRef CAS IUCr Journals Google Scholar
Heinz Maier-Leibnitz Zentrum (2015). J. Large-Scale Res. Facil. 1, A10. Google Scholar
ILL (2016). D22 – A Small-Angle Neutron Scattering Diffractometer, https://www.ill.eu/instruments-support/instruments-groups/instruments/d22/description/instrument-layout/. Google Scholar
Jaksch, S., Schulz, A., Kyriakos, K., Zhang, J., Grillo, I., Pipich, V., Jordan, R. & Papadakis, C. M. (2014). Colloid Polym. Sci. 292, 2413–2425. Web of Science CrossRef CAS Google Scholar
Keiderling, U. (2002). Appl. Phys. A, 74, s1455–s1457. Web of Science CrossRef CAS Google Scholar
Keiderling, U. & Jafta, C. (2016). J. Large-Scale Res. Facil. 2, A97. Google Scholar
Lang, D., Karge, L., Schatte, J., Gilles, R., Dallinger, R., Weissensteiner, I., Staron, P., Knabl, W., Primig, S. & Clemens, H. (2016). J. Alloys Compd. 688, 619–631. Web of Science CrossRef CAS Google Scholar
Lindner, P., Leclercq, F. & Damay, P. (2000). Phys. B Condens. Matter, 291, 152–158. Web of Science CrossRef CAS Google Scholar
Martin, S. L., He, L., Meilleur, F., Guenther, R. H., Sit, T. L., Lommel, S. A. & Heller, W. T. (2013). Arch. Virol. 158, 1661–1669. Web of Science CrossRef CAS PubMed Google Scholar
Masalovich, S. (2007). Nucl. Instrum. Methods Phys. Res. Sect. A, 581, 791–798. Web of Science CrossRef CAS Google Scholar
May, R. P., Ibel, K. & Haas, J. (1982). J. Appl. Cryst. 15, 15–19. CrossRef CAS Web of Science IUCr Journals Google Scholar
Milde, P., Köhler, D., Seidel, J., Eng, L., Bauer, A., Chacon, A., Kindervater, J., Mühlbauer, S., Pfleiderer, C., Buhrandt, S., Schütte, C. & Rosch, A. (2013). Science, 340, 1076–1080. Web of Science CrossRef CAS PubMed Google Scholar
Moreno-Castilla, C., Mahajan, O., Walker, P., Jung, H. & Vannice, M. (1980). Carbon, 18, 271–276. CAS Google Scholar
Mühlbauer, S., Heinemann, A., Wilhelm, A., Karge, L., Ostermann, A., Defendi, I., Schreyer, A., Petry, W. & Gilles, R. (2016). Nucl. Instrum. Methods Phys. Res. Sect. A, 832, 297–305. Google Scholar
Petoukhov, M. V. & Svergun, D. I. (2005). Biophys. J. 89, 1237–1250. Web of Science CrossRef PubMed CAS Google Scholar
Placzek, G. (1952). Phys. Rev. 86, 377–388. CrossRef CAS Web of Science Google Scholar
Schelten, J. & Schmatz, W. (1980). J. Appl. Cryst. 13, 385–390. CrossRef CAS IUCr Journals Web of Science Google Scholar
Schmidt-Rohr, K. & Chen, Q. (2008). Nat. Mater. 7, 75–83. Web of Science PubMed CAS Google Scholar
Sears, V. F. (1992). Neutron News, 3(3), 26–37. CrossRef Google Scholar
Shibayama, M., Nagao, M., Okabe, S. & Karino, T. (2005). J. Phys. Soc. Jpn, 74, 2728–2736. Web of Science CrossRef CAS Google Scholar
Strunz, P., Šaroun, J., Keiderling, U., Wiedenmann, A. & Przenioslo, R. (2000). J. Appl. Cryst. 33, 829–833. Web of Science CrossRef CAS IUCr Journals Google Scholar
Tokunaga, Y., Yu, X., White, J., Réçénnow, H. M., Morikawa, D., Taguchi, Y. & Tokura, Y. (2015). Nat. Commun. 6, 7638. Web of Science CrossRef PubMed Google Scholar
Wang, H., Zhou, W., Ho, D. L., Winey, K. I., Fischer, J. E., Glinka, C. J. & Hobbie, E. K. (2004). Nano Lett. 4, 1789–1793. Web of Science CrossRef CAS Google Scholar
Wignall, G. D. & Bates, F. S. (1987). J. Appl. Cryst. 20, 28–40. CrossRef CAS Web of Science IUCr Journals Google Scholar
Zemb, T. & Lindner, P. (2002). Neutrons, X-rays and Light: Scattering Methods Applied to Soft Condensed Matter. Amsterdam: North-Holland. Google Scholar
Zhang, F., Ilavsky, J., Long, G. G., Quintana, J. P. G., Allen, A. J. & Jemian, P. R. (2010). Metall. Mater. Trans. A, 41, 1151–1158. Web of Science CrossRef Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.