research papers

High-resolution SAXS setup with tuneable

resolution in direct and reciprocal space:
ISSN 1600-5767
a new tool to study ordered nanostructures

Andrei P. Chumakov,a* Kirill S. Napolskii,b Andrei V. Petukhov,c,d Anatoly A.

Snigirev,e Irina I. Snigireva,a Ilya V. Roslyakovb and Sergey V. Grigorievf,g

Received 12 May 2019 a

European Synchrotron Radiation Facility, 71 Avenue des Martyrs/BP 220, Grenoble 38043, France, bLomonosov
Accepted 11 August 2019
Moscow State University, Moscow 119991, Russian Federation, cvan ’t Hoff laboratory for Physical and Colloid
Chemistry, Debye Institute for Nanomaterials Science, Utrecht University, Padualaan 8, Utrecht 3508, The Netherlands,
d
Laboratory of Physical Chemistry, Department of Chemical Engineering and Chemistry, Eindhoven University of
Edited by S. Boutet, SLAC National Accelerator Technology, The Netherlands, eImmanuel Kant Baltic Federal University, Nevskogo Street 14, Kaliningrad 236041,
Laboratory, Menlo Park, USA Russian Federation, fPhysics Institute, NRC Kurchatov Institute, Gatchina, Leningrad Region 188300, Russian Federation,
and gFaculty of Physics, Saint-Petersburg State University, Saint-Petersburg, 198504, Russian Federation. *Correspon-
Keywords: compound refractive lenses; small- dence e-mail: a.p.chumakov@gmail.com
angle X-ray scattering; microradian diffraction;
anodic alumina.
A novel compact small-angle X-ray scattering (SAXS) setup with tuneable
Supporting information: this article has
resolution in both direct and reciprocal space has been designed and tested for
supporting information at journals.iucr.org/j the study of nanostructured materials with a hierarchical structure. The setup
exploits a set of compound refractive lenses that focus the X-ray beam at the
detector position. Anodic alumina membranes with a self-ordered porous
structure were chosen as test samples. The setup allows patterns to be collected
with a minimum scattering vector value of 0.001 nm1 and gives the possibility
for an easy continuous switch between taking high-resolution statistically
averaged diffraction data of macroscopically large sample volumes and lower-
resolution diffraction on a small single domain of the anodic aluminium oxide
film. It is revealed that the pores are longitudinal and their ordering within each
domain tends towards the ideal hexagonal structure, whereas the in-plane
orientation of the pore arrays changes from domain to domain. The possible
advantages and disadvantages of the proposed compact SAXS scheme are
discussed.

1. Introduction
For precise structural characterization, microscopy and
diffraction methods are complementary and provide infor-
mation about the local and average structure of the material in
direct and reciprocal space, respectively. Similar to X-ray
diffraction which is usually applied for the identification and
refinement of crystal structures, small-angle X-ray scattering
(SAXS) is used for the analysis of materials that possess
ordering of their structure on the nano- or microscale. SAXS
setups with high resolution mainly utilize synchrotron radia-
tion. Recording the scattering at small angles assumes the
presence of such structural features of the SAXS installation
as (i) a long collimating section from the source of the
synchrotron beam to the sample (several tens of metres), (ii) a
fixed sample position and (iii) a considerable distance from
the sample to the detector (detector tube), which can reach
10–30 m (Narayanan et al., 2018; Smith et al., 2019; Nikitenko
et al., 2008). Alternatively, one can use a more compact
Bonse–Hart camera (Ilavsky et al., 2009), although this is not
useful for recording 2D scattering patterns. The standard
# 2019 International Union of Crystallography procedure for forming a synchrotron beam and the fixed

J. Appl. Cryst. (2019). 52, 1095–1103 https://doi.org/10.1107/S1600576719011221 1095

research papers
position of the sample stage in most cases make it impossible ence llong = 2/(2
) is sufficient to provide interference
to vary the illuminated sample area without realigning the between waves scattered at different depths in the sample
SAXS setup. The exception is the possibility of changing the (Petukhov et al., 2002; Grigoriev et al., 2010; Als-Nielsen &
illuminated area of the sample by changing the beam-defining McMorrow, 2011).
slits upstream of the sample position in primary or secondary It is more challenging to ensure sufficient beam coherence
short-focusing schemes. However, the change in beam size in in the transverse direction. Owing to the finite source size S,
this case will lead to a rapid reduction in the intensity, espe- the beam consists of coherent volumes (or patches) that are
cially if the size of the desired illuminated area is of the order incoherent with each other. For a freely propagating beam, the
of several micrometres. A large sample-to-detector distance size of a coherent patch is determined by the transverse
requires the availability of a long accessible physical space, coherence length, which increases linearly with increasing
and large financial and time costs (design, manufacture, distance (Born & Wolf, 1970; Pauw, 2013). For example, at the
complexity in the maintenance of mechanical and vacuum position of the CRL it is given by
systems, etc.). The problem of separating the weak small-angle R
scattered intensity from the intense divergent direct beam ltr;0 ¼ ; ð1Þ
2S
creates additional difficulties. Most of these problems could be
overcome by focusing the beam. where R is the distance travelled by the beam from the source
The discovery of a new method of focusing X-ray beams to the CRL and  is the wavelength. The angular resolution of
and the invention of compound refractive lenses (CRL) for a SAXS scheme determines the maximum scale at which
X-ray optics (Snigirev et al., 1996; Lengeler et al., 2005) electron-density correlations can be determined. By using
allowed a breakthrough in the development of new synchro- values typical for synchrotron facilities of R ’ 50 m,  ’
tron setups and facilities (Vaughan et al., 2011; Seiboth et al., 0.1 nm and S ’ 0.1–1 mm, one can estimate ltr ’ 2.5–25 mm.
2014). In particular, CRL can modify and simplify SAXS Since beam coherence over these distances is required to
setups (Petukhov et al., 2006, 2015) and create various facilities observe the correlations, ltr sets the ultimate limit of the
for the study of the local microstructure of matter by achievable angular resolution 2 ’ /(2ltr) = S/R. This reso-
diffraction methods (Snigireva & Snigirev, 2006; Riekel et al., lution can be achieved in the microradian SAXS scheme by
2010). The combination of standard and novel approaches in exploiting a divergent X-ray beam that is focused at the
instrument design leads to new experimental opportunities detector position by CRL installed just in front of the sample
and data-collection technologies (Bosak et al., 2010; Byelov et (Drakopoulos et al., 2005; Thijssen et al., 2006; Petukhov et al.,
al., 2013, Santoro et al., 2014). 2015). One can see that the ultimate angular resolution limit of
Here we introduce a novel high-resolution SAXS setup with this scheme is governed by the angular size S/R of the source
tuneable resolution in both direct and reciprocal space for the seen from the position of the sample and is independent of the
study of nanostructured materials. It is especially suited for the sample-to-detector distance.
characterization of nanomaterials possessing a hierarchical The same conclusion can also be obtained by the following
structure with simultaneous ordering on the nano- and direct-space consideration. Using the geometric optics
microscales. The performance capabilities of the suggested approach and neglecting the imperfection of the focusing
scheme were tested using porous anodic aluminium oxide elements, the smallest size of the direct beam at the detector
(AAO) films, the structure of which consists of micrometre- position can be achieved when the lens creates an image of the
sized domains with a 2D hexagonal array of sub-100 nm sized source at the detector. The demagnification factor of this
cylindrical channels. There is also long-range in-plane orien- scheme is governed by the ratio L/R of the lens–detector
tational order in the AAO structure extending over the distance L and the source–lens distance R, which defines an
millimetre scale (Napolskii et al., 2012). The advantages and ultimate limit for the image size s = SL/R. In angular terms,
disadvantages of this novel SAXS scheme with tuneable this corresponds to a resolution of () = s/L = S/R.
resolution in direct and reciprocal space are discussed. Thus, by using a convergent X-ray beam that is focused at
the detector position, one can readily achieve an angular
resolution of the order of 2–20 mrad at a synchrotron source.
With the currently ongoing upgrades at the ESRF (The ESRF
2. Description of the novel SAXS scheme Orange Book; Dimper et al., 2014) the source size will be
2.1. Microradian SAXS and X-ray beam coherence reduced even further, which will allow for sub-microradian
Since a scattering pattern is created as a result of the resolution. This resolution does not necessarily require long
interference of waves scattered at the sample, beam coherence beamlines but can also be achieved in a more compact scheme
is a crucial issue for high-resolution scattering experiments. of a few metres long, as long as the detector possesses suffi-
One usually defines the longitudinal coherence llong , which is cient resolution (a few micrometres in this case).
related to the final spectral width
 of the X-rays, and the
transverse coherence ltr , which depends on the final size of the 2.2. Novel compact SAXS scheme with tuneable resolution
radiation source. Under the conditions of most small-angle The main idea of the proposed novel high-resolution SAXS
X-ray scattering experiments exploiting a silicon mono- setup is the use of all the accessible space between the CRL
chromator (typically,
/ ’ 104), the longitudinal coher- and the fixed position of the 2D detector as a possible sample

1096 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution J. Appl. Cryst. (2019). 52, 1095–1103
 research papers
position, instead of moving the detector relative to a fixed between the sample and the detector Lsd in the converging
sample. A fixed detector position allows one to use a single X-ray beam. The diameter of the irradiated sample area D
detector that is optimized for the resolution of the optical during the focusing can be estimated by
scheme. There is one example of a laboratory SAXS instru- ( 2 )1=2
ment with a movable sample stage and fixed collimation slits ðh  sÞ Lsd 2
D¼ þs ; ð2Þ
and detector position, which allows one to switch easily from F
wide-angle to small-angle scattering mode (http://www.anton-
paar.com/corp-en/products/group/saxs/) but, to the best of our where h is the beam size at the outlet of the CRL, s is the beam
knowledge, such a geometry of SAXS setup with compound size at the detector and F is the focal length of the CRL. The
refractive lense collimation optics has not been demonstrated reduction in ltr will be negligible at the sample position Lsd_max
yet. close to the lenses. On the other hand, in a focused converging
The conceptual scheme of this SAXS setup is presented in beam not only its size but also the coherence length ltr reduce
Fig. 1. The X-ray beam of the required wavelength passes with increasing distance. Owing to the finite value of the
through the CRL with a short focal distance of the order of 1– coherence length, there is an upper limit on the separation of
2 m. The high-resolution 2D detector with a small pixel size is two objects if they are to give rise to the interference effect.
placed in the image plane. The sample stage with all the Under the conditions of a small-angle scattering experiment,
required auxiliary facilities for horizontal, vertical and rota- this corresponds to the period d of the studied nanostructure;
tional adjustment of the sample position is mounted on a rail hence Lsd_min will depend on the condition ltr  d and, using
to enable its movement along the beam between the CRL and equations (1) and (2) and neglecting ltr at the position of the
the detector. The detector stage and the rail with the sample detector, can be estimated approximately as
stage can be mounted on an optical table according to the dF 2SdF
Lsd min ¼ ¼ : ð3Þ
X-ray beam axis and the chosen experimental geometry (see ltr;0 R
Fig. S1 in the supporting information). Thus, it becomes
possible to vary the illuminated area on the sample by moving Often, the ordered regions of a nano-object have limited
the sample assembly between the short-focus CRL and the dimensions in the sample space and, thus, the sample can
detector along the axis of the converging beam. This allows us contain smaller separate ordered single-crystal regions or
to obtain highly averaged information over a large area with domains with various defects (point, linear or volumetric). As
high resolution and to zoom in on specific structural features the Lsd distance decreases, the size of the irradiated region on
of the sample without the need to realign the optical scheme. the sample also decreases. This makes it possible, in the course
Owing to beam convergence, a change in the sample posi- of the experiment, to obtain highly averaged information from
tion relative to the detector and a change in the illuminated a significant irradiated sample volume, and to localize on one
region of the sample inevitably lead to a change in the reso- or several coherent regions of the sample and obtain structural
lution in both direct and reciprocal space. For a change in the parameters for ordered nano-objects within a particular
transverse coherence length ltr , this will mainly affect the domain. Note that changing the spatial resolution in direct
available q range [q = (4/)sin(/2), where  is the scattering space is automatically linked to a change in the resolution in
angle and  is the wavelength of the incident radiation], reciprocal space.
depending on the ratio of the detector size and the sample-to-
detector distance Lsd, and also the angular resolution
q, 3. Experimental
depending on the ratio of the pixel size of the detector and Lsd.
As discussed above, the transverse coherence length ltr , 3.1. Experimental setup
depending on the source size S and distance from the source R, The compact high-resolution small-angle X-ray scattering
will be shrunk, together with the diameter of the irradiated setup with tuneable probe size was assembled on the Micro-
sample area D, proportionally to the reduction in the distance optics Test Bench installed on beamline ID06 at the European
Synchrotron Radiation Facility (ESRF) in Grenoble, France.
A schematic drawing of the setup is shown in Fig. 1. Micro-
radian X-ray diffraction with a beam energy of about 12 keV
( = 1 Å,
/ ’ 104) was realized using a set of Be CRL,
which consisted of 30 parabolic lenses (Snigirev et al., 1996)
with a radius of curvature of 200 mm, giving a focal length F of
1.45 m. The lenses were placed at a distance R of 56 m from
the undulator source, with the source size being 23  928 mm.
The primary slits, closest to the source (27.8 m), had a vertical
gap of 0.5 mm which did not restrict the beam. In contrast, the
Figure 1 horizontal gap of 0.05 mm was much smaller and effectively
The conceptual scheme of the compact high-resolution small-angle X-ray
defined the secondary source in the horizontal direction at the
scattering setup with tuneable probe size. The shaded area of the beam
denotes a coherent patch; its size has been exaggerated for better slit position. Thus in our case, the transverse coherence length
visibility. at the position of the CRL was about ltr = 121  28 mm in the

J. Appl. Cryst. (2019). 52, 1095–1103 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution 1097
research papers
vertical and horizontal directions, respectively. The colli- 3.2. Sample description
mating slits in front of the CRL limited the beam down to AAO is a typical self-ordered nanostructured material with
0.5 mm in the horizontal and vertical directions, thus fitting the a two-dimensional hexagonal network of pores (Masuda &
effective aperture of the CRL. The beam was focused on a 2D Fukuda, 1995). Since that pioneering work, which showed the
high-resolution Sensicam CCD detector which was combined opportunity for the formation of large-scale ordered struc-
with an X-ray imaging system. The X-ray imaging system tures, AAO has been intensively used as a platform for
consisted of a 9.9 mm thick LAG:Eu scintillator, an Olympus creating various nanostructured functional devices (Qin et al.,
UPLADO tenfold objective and a Sensicam QE camera with a 2005; McGary et al., 2006; Masuda et al., 2006; Banerjee et al.,
resolution of 1376  1040 pixels, a pixel size of 6.45  6.45 mm 2009; Kolmychek et al., 2018). The real structure of a self-
and a 12-bit dynamic range. Taking all optical elements into ordered AAO membrane consists of domains of longitudinal
account, the imaging setup enabled an effective pixel size of pores. In each domain, the pore structure seeks to be ordered.
0.645  0.645 mm, covering a field of view of 890  670 mm. The pores can branch and terminate during growth of the
We note that the conditions for best beam focusing are membrane, and these processes occur mainly at the domain
different in the vertical and horizontal directions. The boundaries (Napolskii et al., 2011). Defects are inherent in the
secondary source created in the horizontal direction at the real structure of anodic alumina and they have an influence on
primary beamline slits is located closer to the CRL and its both the diffraction patterns and the functional properties of
image will be created slightly further downstream. However, AAO membranes. Napolskii et al. (2010) and Roslyakov et al.
the difference in the positions of the images of the primary and (2016) demonstrated that microradian X-ray diffraction is an
secondary sources is small since the focal length F is much efficient tool for the structural characterization of self-ordered
smaller than the source-to-lens distance. The currently porous oxide films. In contrast with conventional microscopy
ongoing upgrade of the ESRF (Dimper et al., 2014) will lead to techniques, the SAXS technique allows one to determine
a nearly circular source with comparable dimensions in the quantitatively the 3D ordering of self-organized AAO struc-
vertical and horizontal directions. In this case, there will no tures.
longer be any need to create a secondary source in the hori- The anodic alumina membranes were prepared using the
zontal direction, thus removing the problems discussed above two-stage anodizing technique. High-purity aluminium foil
that were caused by the very high anisotropy of the source (99.999%, 0.5 mm thick, Goodfellow) with a coarse grain
dimensions at the time of the experiment. structure was used as the starting material. The lateral sizes of
The sample holder was installed on a motorized goniometer the aluminium grains were higher than 1 mm, and their
that allows one to vary the sample-to-detector distance Lsd thickness coincided with the foil thickness. Prior to anodizing,
from 246 to 16 mm. Using slits of 0.5  0.5 mm or the foil was mechanically polished using a diamond suspension
0.2  0.2 mm, this range of Lsd corresponds to illuminated (Struers) to a mirror finish and cleaned repeatedly with
area sizes from 87  87 mm to 7.5  7.5 mm and from acetone and deionized water in an ultrasonic bath. Then the
35.6  35.6 mm to 4.2  4.2 mm, respectively. As a test sample aluminium was anodized in 0.3 M oxalic acid (H2C2O4, 98%,
we used a 96 mm thick anodic alumina membrane with a self- Aldrich) at 40 V in a two-electrode cell using Pt wire as the
ordered porous structure (a pore diameter of ca 40 nm, with counter electrode. The temperature of the electrolyte was kept
the distance between the centres of neighbouring pores dint in the range of 273–274 K. These conditions lead to the
being 102 nm). The membrane was installed perpendicular to formation of oxide films with a highly ordered porous struc-
the incident beam on the translation/rotation stage with ture, with the domains possessing a nearly ideal 2D hexagonal
possible translation along the beam axis. Using the minimum pore arrangement (Roslyakov et al., 2017). In the plane of the
illuminated area size of 4.2 mm, focusing on the area of one or oxide film the domains have an irregular shape and their
a few structure domains became possible. Here we present maximum lateral size can reach several or even a few tens of
data for a slit size of 0.2  0.2 mm with varying diameters of micrometres. After the first stage of anodizing for 72 h, the
the irradiated area D = 35.6, 31.5, 24.7, 17.8, 11 and 4.2 mm at alumina film was selectively etched away in an aqueous
sample-to-detector distances Lsd = 246, 216, 166, 116, 66 and solution containing 35 ml l1 85% H3PO4 and 20 g l1 CrO3 at
16 mm, respectively. The minimum sample-to-detector 343 K. The second anodizing stage was performed under the
distance Lsd_min for our case at ltr = d, according to equation same conditions for 47 h. The thickness of the obtained AAO
(3), should be about 2.5–5 mm. film was about 96 mm.
Scanning electron microscopy (SEM) images were recorded
using an NVision 40 scanning electron microscope (Carl
Zeiss). To examine the in-plane orientational order in the
AAO structure, statistical analysis of the SEM images was 4. Results and discussion
performed using a colour-coding procedure (Hillebrand et al., Anodic alumina membranes possess a significant scattering
2008). The statistical distribution of the interpore distances for contrast, caused by the difference between the refractive
the nearest neighbours around the considered pore was indices of air in the longitudinal pores and of the material of
obtained by the Voronoi algorithm using the ImageJ program the pore walls. We used AAO films with a self-ordered porous
(Schneider et al., 2012) and in-house software (http:// structure (the distance between the centres of the pores dint =
eng.fnm.msu.ru/en/software/). 102 nm), consisting of domains (regions with 2D hexagonal

1098 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution J. Appl. Cryst. (2019). 52, 1095–1103
 research papers

Figure 2
2D X-ray diffraction patterns of porous anodic alumina obtained at different sample-to-detector distances Lsd and irradiated spot diameters D. (a) Lsd =
246 mm, D = 35.6 mm. (b) Lsd = 116 mm, D = 17.8 mm. (c) Lsd = 16 mm, D = 4.2 mm.

pore ordering) whose average size is 1–

10 mm. This range of sizes of ‘ideally’
ordered regions allows us to realize a
situation where the length of the
transverse coherence of the beam is
larger, comparable to or less than the
size of the domains.
2D diffraction patterns of the AAO
membrane, recorded at different
sample-to-detector distances Lsd = 246,
116 and 16 mm, are shown in Fig. 2. The
SAXS data show that some scattering
intensity is present in almost all
azimuthal directions, which is in good
agreement with the powder-like
domain structure of AAO revealed by
SEM [Figs. 3(b) and 3(c)]. However, a
clear sixfold modulation of the scat-
tered intensity suggests that, on a large
scale, a predominant range of orienta-
tions prevails. Earlier, we performed a
detailed statistical analysis of the SEM
data (Napolskii et al., 2012), which
confirmed that a seemingly powder-like
domain structure of a single SEM
image reveals a predominant orienta-
tion after averaging over many neigh-
bouring spots.
The sizes of the transverse coherence
will change during beam focusing. On
the basis of the parameters calculated
above for the vertical and horizontal Figure 3
transverse coherence of the beam at (a) The pair distribution function calculated from the momentum-transfer dependence of the X-ray
scattering for a narrow sector with an opening of ’ = 0.6 for the first-order diffraction peak (see
the position of the CRL and assuming inset). The diffraction pattern was recorded at a sample-to-detector distance Lsd of 216 mm. The
that the coherence length ltr reduces strong point Bragg reflections are marked by numerals (1), (2) and (3). (b) A SEM image of the
linearly with increasing distance in a bottom side of AAO (after removal of the barrier layer). (c), (d) The SEM image from panel (b),
subjected to the colour coding procedure. (c) The colour indicates the average angle to the nearest
convergent beam, ltr = ltr,0 Lsd/F, we
neighbours of the considered pore. (d) The colour shows the average interpore distance to the
calculate the following values of the nearest neighbours. The corresponding colour scales in degrees and nanometres are given in the
transverse coherence length for all upper right corners of corresponding panels.

J. Appl. Cryst. (2019). 52, 1095–1103 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution 1099
research papers
three positions of Fig. 2: (a) 20.53  0.40 mm, (b) marked as white and black points in the colour-coded SEM
9.68  0.19 mm and (c) 1.34  0.03 mm in the vertical and images. Thus, it can be concluded that the bright point
horizontal directions, respectively. One can see that the diffraction reflections arise from the ordered pore structure
number of high-order reflections increases when the sample is within the domains, whereas the defect areas of the porous
closer to the detector. Moreover, they shift closer to the film located near domain boundaries and point defects are the
position of the directly transmitted beam. It was revealed that sources of the small maxima on the overall background of the
the diffraction peaks consist of aggregates of small point diffraction peak.
diffraction reflections which are the result of X-ray scattering Defining the positional correlations of the pores as
on sets of pores with different interpore distances [see the m = q10/q10 from the radial width q10 of the lowest-order
inset in Fig. 2(a)]. Pore arrays with a longitudinal axis direction Bragg peak, we find that for the point diffraction reflections
coaxial with the X-ray beam produce brighter point diffraction the value m varies from 30 to 80 structure periods in each
reflections than other domains with disorientation of the pores domain. In the case of full-circle averaging, the m value varies
relative to the incident beam. It should be noted that the point in the range of 5–7 structure periods. Such a decrease in the
diffraction reflections are not speckles of the X-ray beam number of average structure periods in ordered regions with
because the brightest ones repeat every 60 , following the increasing irradiated area is in good agreement with the
hexagonal symmetry of the AAO structure. It was also assumption that pore branching and termination occur mainly
possible to observe the appearance and extinction of a single on the boundaries of the domains (Napolskii et al., 2011).
point diffraction reflection in a rocking curve experiment The azimuthal dependence of the first-order Bragg reflec-
(Fig. S2 in the supporting information). tion intensity I(’) calculated from diffraction patterns
The obtained 2D patterns of small-angle diffraction can be recorded at different diameters of illuminated area D from
analysed on the basis of the previously demonstrated tech- 35.6 to 4.2 mm is shown in Fig. 4. As the irradiated area
nique for determining the parameters of the AAO structure by decreases, the evolution of the point diffraction reflections and
the SAXS method (Napolskii et al., 2010). Let us examine the their contribution to the overall scattering intensity I(’) of the
momentum transfer q dependence of the scattering intensity diffraction peak can be observed. We can conclude that a
for a narrow sector with an opening of ’ = 0.6 for the first- significant part of the domains has an advantageously uni-
order diffraction peak [see the inset in Fig. 3(a)]. It can be seen directional orientation of the pore lines in the plane of the
that the first-order diffraction maximum, which is usually membrane, which is governed by the crystallographic orien-
smooth (Napolskii et al., 2010), splits into three independent tation of the aluminium substrate (Napolskii et al., 2012). It
and clearly distinguishable diffraction peaks [marked (1), (2) can be clearly seen that a decrease in the sample-to-detector
and (3)]. The dependencies I(q) were fitted by the sum of
Lorentz functions. The point-like diffraction peaks each have
unique locations with centres at (1) q10 = 0.06844 (4) nm1, (2)
q10 = 0.07052 (1) nm1 and (3) q10 = 0.07268 (1) nm1. These
values of the momentum transfer q correspond to structures
with the following interpore distances: (1) dint_1 = 106.0 nm,
(2) dint_2 = 102.9 nm and (3) dint_3 = 99.8 nm [see the pair
distribution function in Fig. 3(a)]. Note that the average
interpore distance calculated from the radial intensity profile
I(q) averaged over the full azimuthal range (’ = 360 ) gives
us a value of dint = 101.4 nm.
There are two possible reasons for the existence of separate
maxima on the pair distribution function corresponding to
areas with various interpore distances: (i) the AAO porous
structure consists of domains in which the average interpore
distance varies from domain to domain; (ii) areas with much
lower or higher distances between nearest-neighbour pores
are localized near defect regions, where the pore surroundings
are far from hexagonal. According to SEM images of the
bottom side of AAO porous films [Fig. 3(b)], the second
scenario is realized. Colour maps of the domain in-plane
orientation and interpore distance distributions are shown in
Figs. 3(c) and 3(d), respectively. One can see that the average
interpore distance inside domains with various in-plane
orientations does not vary significantly. As a consequence, the
multimodal nature of the radial SAXS profiles and of the Figure 4
The azimuthal intensity distribution I(’10) of the 10 diffraction maximum
corresponding pair distribution functions with sharp satellites (110 < ’ < 180 ) at irradiated spot diameters D from 36 to 4.2 mm (at
arises from defect areas of the porous structure, which are sample-to-detector distances Lsd from 246 to 16 mm).

1100 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution J. Appl. Cryst. (2019). 52, 1095–1103
 research papers
distance leads to a lesssening of the resolution in reciprocal intensity of a diffraction peak obtained from the Lorentzian fit
space and as a consequence the I(’) curves become smooth. of the radial distribution of the scattered intensity I(q) in the
As we have demonstrated in our previous work (Napolskii azimuthal range ’ = 14 . The FWHM of the rocking curve
et al., 2010; Grigoriev et al., 2010; Chumakov et al., 2013; shown in Fig. 5(a) is ! = 1.366 , which corresponds to a
Roslyakov et al., 2013), diffraction patterns are extremely correlation length of Lz ’ 3.7 mm. In the case of the small spot
sensitive to sample orientation. Rotation of the sample by as of irradiated area D = 4.2 mm, the rocking curve is formed by a
little as 0.1 around the axes orthogonal to the beam leads to a set of clearly visible peaks [Fig. 5(b)]. Taking into account the
significant modification of the diffraction pattern. The width of presence of just a few domains within the small irradiated
the rocking curve ! is related to the correlation length of the area, we can explain the narrow peaks as rocking curves from
porous structure along with the pores (Lz) via the equation the different point diffraction reflections. The ! values for
2 2 these peaks are about 0.05 and correspond to a correlation
Lz ¼ ¼ ; ð4Þ length Lz ’ 100 mm, which is very close to the membrane
qz q10 sin !
thickness (96 mm). Thus, we can conclude that the real long-
where qz is the width of the 10 diffraction maximum along the itudinal self-correlation length (the straight section of the
beam, q10 is the scattering vector corresponding to the 10 pores) inside the domain significantly exceeds the average
diffraction peak and ! is the FWHM of the rocking curve. We longitudinal correlation length of the membrane. The pores
can define the longitudinal self-correlation length as the length within the domains are mainly rectilinear through the whole
along the pore within which the difference between the pore membrane thickness.
and a best-fit straight cylinder is smaller than the pore Moving the sample closer to the detector results in focusing
diameter. of the X-ray beam on the local volume of the specimen, which
Fig. 5 presents the rocking curves for the intensity of the 10 allows us to study the local features of the AAO structure.
diffraction peak at sample-to-detector distances of (a) Lsd = Using an X-ray beam with a minimal beam size of D = 4.2 mm
216 mm and (b) Lsd = 16 mm, with the diameters of the irra- enables focusing on the area of a unit domain or on the
diated areas being D = 31.5 and 4.2 mm, respectively. Each boundary between two domains to choose the most appro-
point on the experimental curve represents the integral priate region of interest (Fig. 6). Using a sample-to-detector
distance of 16 mm, we performed a mapping of the structure

Figure 6
Diffraction patterns recorded at two points on the AAO sample situated
at a distance of 3 mm apart. The X-ray beam passes through (a) the centre
of the domain and (b) the region of the boundary of two domains. The
sample-to-detector distance Lsd is 16 mm (D = 4.2 mm). (c) A SEM image
Figure 5 of AAO in a randomly chosen region of the high-angle border between
Rocking curves of the first-order peak q10 with the azimuthal peak centre two neighbouring domains. (d) The SEM image from panel (c) subjected
’ = 0 and opening ’ = 14 at sample-to-detector distances (a) L = to colour coding based on the average angle to the nearest neighbours of
216 mm and (b) L = 16 mm. the considered pore [the colour scale is the same as in Fig. 3(c)].

J. Appl. Cryst. (2019). 52, 1095–1103 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution 1101
research papers
with a resolution equal to the minimum movement of the (ii) Each domain of the AAO membrane consists of recti-
stepper motors. linear through pores. The longitudinal pore axes between the
It can be clearly seen that the mosaicity of the porous domains have a different reorientation on an angle of about
structure inside the domain [Fig. 6(a)] is very low (’ ’ 15 ). 0.05–0.5 from domain to domain.
As a consequence, the average mosaicity of the AAO struc- (iii) The dispersion of point reflections into diffraction
ture is mainly given by the disorientation of neighbouring peaks testifies to the reorientation of domains in the plane of
domains rather than by the variation of the in-plane orienta- the membrane.
tion of the pore structure inside the domain. The diffraction (iv) The pore structure inside the domains is close to the
pattern recorded from the region of the domain boundary ideal 2D hexagonal structure. At the high-angle domain
[Fig. 6(b)] corresponds to the high-angle boundary: the six boundaries the misorientation of the nearest domains is close
first-order reflections from the first domain are rotated by ’ ’ to 30 , but low-angle boundaries lead to a smooth change of
30 with respect to the six reflections from the neighbouring the in-plane orientation of the porous system from one
domain. Such 30 boundaries are common for the AAO domain to another.
porous structure. As an example, a high-magnification SEM
image of a randomly chosen region of the AAO structure near
the domain boundary is shown in Fig. 6(c). The corresponding
colour-coded map [Fig. 6(d)] showing the 30 disorientation Acknowledgements
between two domains is in good agreement with the diffrac- The authors acknowledge the ESRF for hospitality and Yuriy
tion pattern with 12-fold intensity modulation [Fig. 6(b)]. Chushkin (XPCS/CXDI, ID10 beamline, ESRF) for fruitful
discussions.
5. Conclusions
We have successfully demonstrated the advantages of the Funding information
proposed novel SAXS scheme: Synthesis and SEM characterization of the anodic alumina
(i) Compact SAXS with a tuneable probe size is well suited films were performed with the financial support of the Russian
to the investigation of both the average and local structure of Science Foundation (grant No. 18-73-10151). The develop-
nanomaterials with spatially ordered structure. ment of the high-resolution small-angle X-ray scattering setup
(ii) The use of CRL allows one to reduce the length of the was supported by RSF grant No. 19-72-30009. AAS is grateful
SAXS setup to ca 1.5 m. to the Russian Academic Excellence Project at the Immanuel
(iii) Changing the slit gap behind the CRL, and the sample Kant Baltic Federal University for financial support of his
position between the CRL and the detector, allows one to travel.
change the irradiated area on the sample.
(iv) Diffraction mapping of the sample is possible at a
minimal sample-to-detector distance. References
(v) Scattering in air is reduced in a compact setup with a Als-Nielsen, J. & McMorrow, D. (2011). Elements of Modern X-ray
shorter lens-to-detector distance, so it is no longer necessary Physics. Chichester: John Wiley & Sons.
to use a vacuum detector tube for samples with sufficient Banerjee, P., Perez, I., Henn-Lecordier, L., Lee, S. B. & Rubloff, G. W.
contrast. (2009). Nat. Nanotechnol. 4, 292–296.
(vi) It would be possible to construct this easy scheme as a Born, M. & Wolf, E. (1970). Principles of Optics, 4th ed. Oxford:
Pergamon Press.
temporary assembly at different laboratories and synchrotron Bosak, A., Snigireva, I., Napolskii, K. & Snigirev, A. (2010). Adv.
beamlines. Mater. 22, 3256–3259.
The proposed SAXS scheme should also be examined for its Byelov, D. V., Meijer, J.-M., Snigireva, I., Snigirev, A., Rossi, L., van
possible drawbacks: den Pol, E., Kuijk, A., Philipse, A., Imhof, A., van Blaaderen, A.,
(i) The resolution parameters in reciprocal space begin to Vroege, G. J. & Petukhov, A. V. (2013). RSC Adv. 3, 15670–15677.
Chumakov, A. P., Roslyakov, I. V., Napolskii, K. S., Eliseev, A. A.,
deteriorate during the approach of the sample to the detector, Lukashin, A. V., Eckerlebe, H., Bouwman, W. G., Belov, D. V.,
owing to the shrinkage of reciprocal space in the detector Okorokov, A. I. & Grigoriev, S. V. (2013). Nanotechnol. Russia, 8,
plane. 631–638.
(ii) To achieve sufficient resolution in reciprocal space in the Dimper, R., Reichert, H., Raimondi, P., Sánchez Ortiz, L., Sette, F. &
compact SAXS scheme, a high-resolution detector with Susini, J. (2014). The ESRF Orange Book: ESRF Upgrade
Programme Phase II (2015–2022) Technical Design Study. ESRF,
minimal pixel size is required. Grenoble, France. http://www.esrf.eu/Apache_files/Upgrade/ESRF-
Summing up our test investigation of AAO membranes by orange-book.pdf.
compact SAXS with a tuneable probe size, we make the Drakopoulos, M., Snigirev, A., Snigireva, I. & Schilling, J. (2005).
following observations: Appl. Phys. Lett. 86, 014102.
(i) The interpore distance distribution is narrow. Areas with Grigoriev, S. V., Syromyatnikov, A. V., Chumakov, A. P., Grigoryeva,
N. A., Napolskii, K. S., Roslyakov, I. V., Eliseev, A. A., Petukhov,
much shorter or longer distances between nearest-neighbour A. V. & Eckerlebe, H. (2010). Phys. Rev. B, 81, 125405.
pores are localized near defect regions, where the pore Hillebrand, R., Müller, F., Schwirn, K., Lee, W. & Steinhart, M.
surroundings are far from hexagonal. (2008). ACS Nano, 2, 913–920.

1102 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution J. Appl. Cryst. (2019). 52, 1095–1103
 research papers
Ilavsky, J., Jemian, P. R., Allen, A. J., Zhang, F., Levine, L. E. & Long, Petukhov, A. V., Thijssen, J. H. J., ’t Hart, D. C., Imhof, A., van
G. G. (2009). J. Appl. Cryst. 42, 469–479. Blaaderen, A., Dolbnya, I. P., Snigirev, A., Moussaı̈d, A. &
Kolmychek, I. A., Pomozov, A. R., Leontiev, A. P., Napolskii, K. S. & Snigireva, I. (2006). J. Appl. Cryst. 39, 137–144.
Murzina, T. V. (2018). Opt. Lett. 43, 3917–3920. Qin, L. D., Park, S., Huang, L. & Mirkin, C. A. (2005). Science, 309,
Lengeler, B., Schroer, C. G., Kuhlmann, M., Benner, B., Günzler, T. F., 113–115.
Kurapova, O., Zontone, F., Snigirev, A. & Snigireva, I. (2005). J. Riekel, C., Burghammer, M. & Davies, R. (2010). IOP Conf. Ser.
Phys. D Appl. Phys. 38, A218–A222. Mater. Sci. Eng. 14, 012013.
Masuda, H. & Fukuda, K. (1995). Science, 268, 1466–1468. Roslyakov, I. V., Eliseev, A. A., Yakovenko, E. V., Zabelin, A. V. &
Masuda, H., Yamada, M., Matsumoto, F., Yokoyama, S., Mashiko, S., Napolskii, K. S. (2013). J. Appl. Cryst. 46, 1705–1710.
Nakao, M. & Nishio, K. (2006). Adv. Mater. 18, 213–216. Roslyakov, I. V., Gordeeva, E. O. & Napolskii, K. S. (2017).
McGary, P. D., Tan, L. W., Zou, J., Stadler, B. J. H., Downey, P. R. & Electrochim. Acta, 241, 362–369.
Flatau, A. B. J. (2006). Appl. Phys. 99, 08B310. Roslyakov, I. V., Koshkodaev, D. S., Eliseev, A. A., Hermida-Merino,
Napolskii, K., Roslyakov, I., Romanchuk, A., Kapitanova, O., D., Petukhov, A. V. & Napolskii, K. S. (2016). J. Phys. Chem. C, 120,
Mankevich, A., Lebedev, V. & Eliseev, A. (2012). J. Mater. Chem. 19698–19704.
22, 11922–11926. Santoro, G., Buffet, A., Döhrmann, R., Yu, S., Körstgens, V., Müller-
Napolskii, K. S., Roslyakov, I. V., Eliseev, A. A., Byelov, D. V., Buschbaum, P., Gedde, U., Hedenqvist, M. & Roth, S. V. (2014).
Petukhov, A. V., Grigoryeva, N. A., Bouwman, W. G., Lukashin, Rev. Sci. Instrum. 85, 043901.
A. V., Chumakov, A. P. & Grigoriev, S. V. (2011). J. Phys. Chem. C, Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. (2012). Nat.
115, 23726–23731. Methods, 9, 671–675.
Napolskii, K. S., Roslyakov, I. V., Eliseev, A. A., Petukhov, A. V., Seiboth, F., Schropp, A., Hoppe, R., Meier, V., Patommel, J., Lee, H. J.,
Byelov, D. V., Grigoryeva, N. A., Bouwman, W. G., Lukashin, A. V., Nagler, B., Galtier, E. C., Arnold, B., Zastrau, U., Hastings, J. B.,
Kvashnina, K. O., Chumakov, A. P. & Grigoriev, S. V. (2010). J. Nilsson, D., Uhlén, F., Vogt, U., Hertz, H. M. & Schroer, C. G.
Appl. Cryst. 43, 531–538. (2014). J. Phys. Conf. Ser. 499, 012004.
Narayanan, T., Sztucki, M., Van Vaerenbergh, P., Léonardon, J., Smith, A. J., Davidson, L. S., Emmins, J. H., Hiller Bardsley, J. C.,
Gorini, J., Claustre, L., Sever, F., Morse, J. & Boesecke, P. (2018). J. Holloway, P., Malfois, M., Marshall, A. R., Pizzey, C. L., Rogers,
Appl. Cryst. 51, 1511–1524. S. E., Shebanova, O., Snow, T., Williams, E. P. & Terrill, N. J. (2019).
Nikitenko, S., Beale, A. M., van der Eerden, A. M. J., Jacques, S. D. M., arXiv:1903.05405.
Leynaud, O., O’Brien, M. G., Detollenaere, D., Kaptein, R., Snigirev, A., Kohn, V., Snigireva, I. & Lengeler, B. (1996). Nature, 384,
Weckhuysen, B. M. & Bras, W. (2008). J. Synchrotron Rad. 15, 632– 49–51.
640. Snigireva, I. & Snigirev, A. (2006). J. Environ. Monit. 8, 33–42.
Pauw, B. R. (2013). J. Phys. Condens. Matter, 25, 383201. Thijssen, J. H. J., Petukhov, A. V., ’t Hart, D. C., Imhof, A., van der
Petukhov, A. V., Aarts, D. G. A. L., Dolbnya, I. P., de Hoog, E. H. A., Werf, C. H. M., Schropp, R. E. I. & van Blaaderen, A. (2006). Adv.
Kassapidou, K., Vroege, G. J., Bras, W. & Lekkerkerker, H. N. W. Mater. 18, 1662–1666.
(2002). Phys. Rev. Lett. 88, 208301. Vaughan, G. B. M., Wright, J. P., Bytchkov, A., Rossat, M., Gleyzolle,
Petukhov, A. V., Meijer, J.-M. & Vroege, G. J. (2015). Curr. Opin. H., Snigireva, I. & Snigirev, A. (2011). J. Synchrotron Rad. 18, 125–
Colloid Interface Sci. 20, 272–281. 133.

J. Appl. Cryst. (2019). 52, 1095–1103 Andrei P. Chumakov et al.  High-resolution SAXS with tuneable resolution 1103