Professional Documents
Culture Documents
Current Perspectives
art ic l e i nf o
a b s t r a c t
Article history:
Received 23 June 2013
Received in revised form
2 September 2013
Available online 10 October 2013
Neutron imaging became important when, besides providing impressive radiographic and tomographic
images of various objects, physical, quantication of chemical, morphological or other parameters could
be derived from 2D or 3D images. The spatial resolution of approximately 50 mm (and less) yields real
space images of the bulk of specimens with more than some cm3 in volume. Thus the physics or
chemistry of structures in a sample can be compared with scattering functions obtained e.g. from
neutron scattering. The advantages of using neutrons become more pronounced when the neutron spin
comes into play. The interaction of neutrons with magnetism is unique due to their low attenuation by
matter and because their spin is sensitive to magnetic elds. Magnetic elds, domains and quantum
effects such as the Meissner effect and ux trapping can only be visualized and quantied in the bulk
of matter by imaging with polarized neutrons. This additional experimental tool is gaining more and
more importance. There is a large number of new elds that can be investigated by neutron imaging,
not only in physics, but also in geology, archeology, cultural heritage, soil culture, applied material
research, magnetism, etc. One of the top applications of polarized neutron imaging is the large eld
of superconductivity where the Meissner effect and ux pinning can be visualized and quantied. Here
we will give a short summary of the results achieved by radiography and tomography with polarized
neutrons.
& 2013 Elsevier B.V. All rights reserved.
Keywords:
Radiography and tomography
Imaging
Polarized neutrons
Meissner effect
Flux trapping
Superconductivity
Magnetic domains
1. Introduction
The applications of neutron radiography and tomography
(shortly imaging) cover a large number and quite different
research disciplines, that are all tasked with investigating the
composition of samples under special conditions such as
temperature, pressure, magnetic or electric elds, etc. and
possibly under additional boundary conditions, which inuence the measuring process. The information one obtains from
these experiments can always be distinguished into two large
classes, real space data (e.g. images,) and reciprocal space data
(scattering pattern, scattering curves). Both can be transformed into images or graphs, whereas both presume a (lot
of) theory that predicts and describes the information originally collected. Each light microscope provides real data information, i.e. images from the surface of a sample. On the other
hand, radiography, computer tomography (CT) or nuclear
magnetic resonance yield images from the bulk of a body. But
one realizes the difference between images delivered by a
n
189
2. Theoretical background
2.1. General remarks
Spin based radiography and tomography with polarized neutrons combines, absorption (attenuation) and phase based interactions, i.e. with the option of polarized neutrons enormously
enlarges the eld of applications. Additionally new techniques are
emerging that reach higher resolution and exibility such as the
Larmor Labeling neutron imaging, which uses the Larmor precession of the neutron spin in magnetic elds (see e.g. Ref. [9])
Investigations of matter by thermal and cold neutron beams
are very popular, especially if magnetism comes into play. Neutrons have some very specic properties that make them superior
to other probes such as X- or gamma rays, electrons, protons,
molecular beams, light and laser if one wants to investigate not
only surface but also the bulk of a sample. The interaction of
(thermal and cold) neutrons with matter is described by the
nuclear interaction potential, given by the special isotopedifferentiating scattering lengths and there is no dependency on
the atomic number. The low attenuation of neutrons by most
elements, with some exceptions, allows for investigating large
samples (in comparison to electrons or X-rays). The spin interaction with the magnetism of a sample together with their low
velocities (for time of ight measurements) make thermal and
cold neutrons a unique probe for condensed matter research. The
interaction of neutrons with magnetically ordered matter includes
the interaction with the nuclei (coherent scattering length bc) and
with the electronic magnetic moments of incompletely lled
3d-. 4d-, 4f- and 5f-shells (see e.g. [31]). The interaction with
magnetic elds H is described by the neutron spin (and the
Fig. 1. Layout of an instrument for imaging with polarized neutrons: neutrons from
a source are polarized and guided to the sample, which e.g. depolarizes the beam,
which is spin-analyzed and then 2D-detected.
Fig. 2. Pin hole geometry: D diameter of the aperture, L distance of D from two
points, ld distance of the points from detector (screen), and d1 and d2 blurred
images of point 1 and point 2 respectively (see text).
190
Fig. 3. For a given L/D, the distance ld of the detector screen from P1 and P2
inuences the resolution to distinguish the two points. At ld P1 and P2 can be
separated, as well as at ld x1; at ld x2 no separation is possible.
2 =2
1 fx x0 2 z z0 2 g3=2
2
here is the slope of the tangent in x0 and z0, and x0 and z0 are the
coordinates of the center of the image point. Often it is better (and
easier) to determine the edge spread function E(x) which can be
tted to the measured data (background and amplitude must be
tuned to the data),
Ex
1
1 ex x0
191
m
L t L Bt L Bds L Bs
7
h
v
where v is the velocity of the neutron, B is the magnetic eld, m is
the neutron mass, s is the path length of the neutron in the magnetic
eld B, is the neutron wavelength and h is the Planck constant. The
velocity and the wavelength are related to each other (de Broglie) as
h/p, pmv, mmass of the neutron, and vvelocity. h/
m3.956034 10 7 m2/s, for v3956 m/s, the wavelength
1 10 10 m0.1 nm1 . The neutron spin performs Larmor
precessions L also when the neutron passes through an (inhomogeneous) magnetic eld B, and thus the path integral depends on the
particular path through the eld. The angle of rotation is given by Eq.
Fig. 5. Layout of Polarized Neutron Tomography II (PONTO). A graphite monochromator reects neutron wavelength between 0.28 nm and 0.47 nm to the optical bench
which can be rotated around the C-monochromator. Length of the instrument 4 m,
items on the optical bench are displayed in an exaggerated manner.
moment m. For this idea one has to consider the magnetic moment
m, which is classical derived from a rotating electric charge e on a
circle with the radius r and the velocity v as
ev
e !
!
jl j
3
r2
2 r
2m
!
and analogous with this, in quantum mechanics the spin S is
!
connected with angular momentum l as
From Eq. (7) one also realizes that the slower neutrons, the more
Larmor precession are performed in B.
For the visualization of the magnetism of a sample the beam
must be polarized and analyzed with a spin polarizer and an
analyzer and a 2D-detector (Fig. 1). The sensitivity of the spin
direction on B is quite high: For e.g. B 1 mT and a neutron
velocity of 103 m s 1 (wavelength 0.3956 nm) the spin rotates in
!
!
l
S
!
and !
m s g s mk
m l g l mk
e
meV
mk
3:1525 10 14
2mP
T
5
k g n S t B tj j x; y; z
dt
h
j
The motion of the spin can be seen as follows. If the spin S S(t) is
given in an rectangular coordinate system, where the magnetic eld
B is parallel to the z-axis, S(t) can be written as (see also small sketch)
0
1
cos
!
B
C
St S@ sin A
0
I spin x;z
192
Fig. 6. (a) Polarized neutrons and a homogeneous magnetic eld create an image of non-equidistant fringe pattern, and (b) theory and experimental result (neutron images)
of the increasing B-eld in the copper coil.
1
9
where p is the scanning variable over the 2D function {f} and the
angle under which the sample is scanned. For investigations of
ux trapping the samples were scanned under 01 and 901 with
1
Ix; z I 0 T exp
sds 1 cos x; z
2
path
|{z}|{z}
I att x;z
I spin x;z
1
Iy; z I 0 T exp
sds 1 cos y; z
2
path
|{z}|{z}
I att y;z
10
I spin y;z
superconducting sample and the magnetism inside it. An interesting physics is the co-existence of two phases, superconducting
and normal conducting especially in superconductor type I samples, because, in contrast to type II superconductors, these should
not exhibit any ux trapping or ux pinning. Thus investigations of
the bulk of samples that are in the superconducting state unveils
size, shape and structure of these trapped or pinned magnetic
elds and give answers on how this can happen.
3.1. Imaging magnetic elds of superconducting samples
The main purpose of experiments with polarized neutrons was
to visualize expelled and trapped magnetic elds of samples in the
superconducting state. The samples were lead (superconductor
type I) and niobium (superconductor type II). The suppressed
Meissner effect and ux trapping are better known from superconductor type II; however as mentioned above they were also
observed in type I superconductors, such as in lead. To explain and
underline the precision of the experimental results, the starting
point was the image of the magnetic eld created by a Cu coil.
An electric current in the coil creates an homogeneous magnetic
eld B, as is known from fundamental electrodynamics. The
strength of B is given by the current I [Amperes] (varied from
4 A up to 6 A), the length of the coil (90 mm) and the number of
windings (205). With Eq. (7) the number of the Larmor precessions can be calculated for each path through the coil, which yields
an image as shown in Fig. 6.
The spin analyzer behind the sample is transparent for parallel
spins (spin up) and opaque for anti-parallel spins, i.e. for a spin
rotation of (spin down). The spin rotation angle (Eq. (7)) is
proportional to Bs, and s is the path length through the sample
as shown in Fig. 6. Due to different s (and given B) the spin
to [8]
"
Bc T Bc 0 K 1
T
Tc
2 #
1 N
11
1
1 a=hq
12
where a ( 6 mm) and h ( 15 mm) are the radius and the half
length of the cylinder, respectively, and q is given by as
q
4
2
13
Fig. 7. Temperature dependence of the critical magnetic eld for the used Pb
sample geometry 01. At T 5.5 K the applied external magnetic eld was
Bext 6.4 mT, which is well below Bc 25 mT (blue curve). (For interpretation of
the references to color in this gure legend, the reader is referred to the web
version of this article.)
Fig. 8. External magnetic eld orientation with respect to the sample and neutron
ight direction.
193
Fig. 9. Pb polycrystalline sample, polarized neutron images of a partial Meissner effect: rod axis 01 with respect to the neutron beam direction: (a) T 5.5 K and (b) overlap
calculated image and experiments, at T 7.4 K the fringes disappear (see text) (courtesy PRB).
194
Fig. 10. Pb polycrystalline sample: rod axis 901 with respect to the neutron ight direction: (a) T 5.5 K, (b) T 6.8 K, and (c) T 8 K. No further images processing was
applied to see these rst original images of a trapped magnetic eld which is not homogeneously distributed in the homogeneous sample (permitted reuse, by courtsey of
Phys. Rev. B).
Fig. 11. (a) Variation of the trapped eld B, FWHM is kept constant, the eld was adapted to the geometry of the lead cylinder; note the sensitivity of B on the fringe pattern,
tuning further the FWHM of the Gaussian function the calculation agree perfectly with the pattern for both, the suppressed Meissner effect (Fig. 9 and ux trapping
((Fig. 10b) [61], permitted reuse, by courtesy of Phys. Rev. B).
Fig. 12. 3D model of the trapped magnetic eld. The strength of B increases towards the rod axis and decreases to zero at the inner side of the sample surface [61] (permitted
reuse, by courtesy of Phys. Rev. B).
the amount of trapped eld. This series did not result in the best
t with respect to the FWHM, which also had to be adapted.
The Gaussian shaped eld is constant, parallel to the rod axis,
but squeezed around the rod axis, the FWHM was estimated to be
1.45(5) mm. The expelled eld was also tted by a Gaussian
function, which has its maximum at the rod axis but is zero inside
the sample. The t was compared with the experimental images as
shown in Fig. 11, a model of the calculated trapped eld is shown
in Fig. 12.
195
Fig. 13. Observation of partial Meissner effect and ux pinning in a composed Pb sample (left sketch), [62] (permitted reuse, by courtesy of Appl. Phys. Lett.).
Fig. 14. (a) Magnetic ux trapping in a Nb sample, Tc 9.25 K. The sample was not treated (chemically polished). One recognizes a temperature dependence of ux
trapping (T 5.5; T 9 K) that vanishes at T 11 K 4Tc 9.25 K. Only less than a quarter of the disc is shown (diameter 45 mm). (b) Nb sample with holder for the cryostat.
Fig. 15. Images obtained from the same sample as in Fig. 14, but the surface of the sample was this time chemical polished. This treatment removes (mosaic) surface
structures and apparently inuences the trapped ux. For T 4Tc the patterns disappear.
196
Fig. 17. Measured transmitted intensity with the grating interferometer ([25],
permitted reuse, by courtesy of Nat. Commun.).
Fig. 18. TalbotLau neutron radiographies of magnetic domains. (a) Sketch of the FeSi sample. (b)(f) Radiographic projection images at different external magnetic elds.
(c)(f) Equal eld strength in (c) and (h), (white bar 3 mm), ([25], permitted reuse, by courtesy of Nat. Commun.).
197
Fig. 19. Reconstruction from a TalbotLau neutron tomography of a FeSi wedge. (a) and (b) Sketches of the FeSi wedge ((a) front view and (b) side view). (c) Horizontal slices
at different wedge thicknesses as indicated in (b). (d)(g) Cross-sections through the TalbotLau neutron tomogram along different planes (dotted by (d)(g) in (a) and (b))
that show the high complexity of the 3D shape of magnetic domains. (h) Horizontal slice (as in (c)). (i) Enlargement of the area marked in red in (h) ([25], permitted reuse, by
courtesy of Nat. Commun.). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)
4. Summary
Imaging with polarized neutrons is widely used [23] and the eld
of applications is increasing, because of the large potential of this
method. Up to now imaging with polarized neutrons is still something exotic, it requires knowledge about radiography and
198