A Superbend X-Ray Microdiffraction Beamline at the Advanced Light Source
N. Tamura, M. Kunz, K. Chen, R.S. Celestre, A.A. MacDowell and T. Warwick
Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley CA 94720
Corresponding author: N. Tamura, email:
ntamura@lbl.gov
Abstract
Beamline 12.3.2 at the Advanced Light Source is a newly commissioned beamline
dedicated to x-ray microdiffraction. It operates in both monochromatic and
polychromatic radiation mode. The facility uses a superconducting bending magnet
source to deliver an X-ray spectrum ranging from 5 to 22 keV. The beam is focused
down to ~ 1 um size at the sample position using a pair of elliptically bent Kirkpatrick-
Baez mirrors enclosed in a vacuum box. The sample placed on high precision stages can
be raster-scanned under the microbeam while a diffraction pattern is taken at each step.
The arrays of diffraction patterns are then analyzed to derive distribution maps of phases,
strain/stress and/or plastic deformation inside the sample.
Keywords: x-ray microdiffraction, x-ray beamline, Laue diffraction, strain/stress
measurements, microprobe
Introduction
Over the years, X-ray microdiffraction (uXRD) has become one of the several standard
tools used at synchrotron facilities to identify and characterize materials at micron- and
submicron scale. X-ray beam sizes and therefore spatial resolutions range from a few
microns down to a few tens of nanometers depending on the source, photon energy range,
focusing optics and other beamline characteristics.
[1-3]
X-ray microdiffraction is often
used in complement to microspectroscopy techniques such as microfluorescence (uXRF),
micro X-ray absorption near-edge structure (uXANES) and micro- extended X-ray
absorption fine structure ( uEXAFS) spectroscopies and is mainly a phase identification
tool for minute or highly heterogeneous samples.
A typical modern microdiffraction setup delivers a focused monochromatic beam onto
the sample and is equipped with an X-ray area detector mounted on a diffractometer to
collect 2D diffraction patterns in transmission mode (requiring the use of a beam stop).
Data are then analyzed via software such as Fit2D.
[4]
A few beamlines at the Advanced
Photon Source (APS), Advanced Light Source (ALS), European Synchrotron Radiation
Facility (ESRF), National Synchrotron Light Source (NSLS), Canadian Light Source
(CLS), Swiss Light Source (SLS) and Pohang Light Source (PLS) offer a more or less
dedicated microdiffraction station with enhanced capabilities such as polychromatic
beam for strain measurement and 3-dimensional (depth-resolving) X-ray diffraction.
[5-6]
Along with groups at the APS and NSLS, the ALS has pioneered some of these
techniques and has offered a dedicated poly/monochromatic station on the bending
magnet beamline 7.3.3 open to the users community since 2001.
[6]
The user program on
ALS beamline 7.3.3 has proven to be highly successful. However, with the increasing
complexity of new scientific problems a number of potential limitations in beamline
capabilities became apparent. These limitations include limited spatial resolution (around
1 um) and strain resolution (typically 2. 10
-4
), limited energy range (5-12 keV), limited
flux in monochromatic mode, and time-consuming data collection and analysis.
Exchanging a regular ALS bending magnet for a superconducting magnet is a good way
to address most of these issues (Fig.1). For instance, the energy range can be extended
and the increased flux at high energy allows using a smaller source size and thus
obtaining a smaller focus spot size by flux trade-off. Increased flux also decreases data
collection time and improves strain resolution through better statistics provided by an
increased number of Laue spots. Higher energy photons also open the possibility of depth
profiling techniques as the penetration depth in the sample increases.
This paper briefly describes the beamline optic components and goes over some of the
improved capabilities and benefits obtained from the move of the ALS X-ray
microdiffraction program to a superconducting bending magnet source. A concluding
section will go over some of the first scientific applications that came out of that
beamline.
Beamline description
The general design of the beamline is similar to the previous end-station on ALS
beamline 7.3.3
[6]
. Only a few parts of the original beamline have been recycled to the
new one (tungsten rotary slits, X-ray CCD detector and diffractometer 2θ arm). Figure 2
shows the outline of the beamline in elevation and plan views. The beamline uses one of
the three 6 Tesla superconducting magnet sources of the Advanced Light Source which
provides a critical energy at about 12 keV, extending the photon spectrum range well into
the hard x-ray regime. The first optic encountered by the x-ray beam is a horizontally
deflecting toroidal mirror that conveys the beam to a virtual source just after the
experimental hutch entrance. The Pt (25 nm) and Rh (8 nm) coated silicon toroidal mirror
operates at a grazing angle of 3.5 mrad cutting off the energy spectrum at about 22 keV.
Water-cooled tungsten roll slits are used to adjust the virtual source size and therefore the
size of the focused beam onto the sample. Final focusing is achieved using a Kirkpatrick-
Baez (KB) mirror assembly contained in a vacuum box. The KB mirrors assembly is
Peltier-cooled to compensate for the heat load on the mirrors inside the optic box vacuum
chamber. The demagnifications to obtain a 1x1 um
2
size beam are approximately 16:1
and 8:1 for the vertical and horizontal KBs respectively. The choice of using KBs is
directed by the use of polychromatic radiation, as they are to date the most efficient
achromatic optics available. The beamline can switch between white and monochromatic
beam by way of a 2 channel-cut - 4 bounces Si(111) monochromator. This particular
design is aimed for having the possibility to illuminate the same spot on the sample with
either white or monochromatic beam independent of energy. The two channel-cuts are
mounted on roll-bearing rotation stages that are independently driven by linear motors via
a sine bar mechanism, replacing the previous 7.3.3 technology that used a tape-drive
system. The sample stages are placed on a high precision XYZ stage for accurate
positioning. A χ stage defines the incident beam angle onto the sample. Diffraction
patterns are collected using a MAR133 X-ray detector and X-ray fluorescence signal by a
Vortex Si-drift detector coupled with a XIA multi channel analyzer. The design is
flexible to allow for both reflective and transmitting sample geometries. Precise
positioning (within a few microns) of the sample into the focal point of the beam is
obtained via a Keyence laser triangulation tool.
The main differences with the previous bending magnet beamline are the extended
energy range (up to 22 keV) and increased flux allowing scaling down the exposure time
for diffraction patterns by almost an order of magnitude. Using polychromatic radiation,
strain states in the sample are obtained by fitting the angular differences between
reflections using a non-linear least square method. Strain resolution depends on the
number of reflections used in the fit. By extending the available energy range from 12
keV to 22 keV, the number of available spots significantly increases, allowing for almost
an order of magnitude better strain resolution if compared to identical material as
measured on ALS beamline 7.3.3 (from 2 10
-4
down to 5 10
-5
for a typical metal sample)
(Figure 3). Another difference with the previous system is the use of a vacuum box
instead of a He-filled box for the optics, giving rise to additional flux gain although the
main reason is more practical as no gas flow is necessary and risk of ozone formation is
minimized. Apart from the M1 mirror, all the optical components of the beamline are
inside a vacuum chamber sitting on an optical table inside the experimental hutch, and
evacuated first by a dry roughing pump and then by a turbo pump that provides a vacuum
of about 10
-5
Torr. Some extra care has been taken to enhance the mechanical coupling
between the optic box and the sample stages to limit differential vibrations as well as
temperature drifts. The hutch itself is maintained at a constant temperature of 22 +/-0.1
ºC. by an air conditioning system. The tungsten rotary slits are water-cooled, while the
monochromator crystals and the KB assembly use a water-based Peltier cooling system.
Beamline performance and capabilities
Beam spot size has been minimized using an on-line KB adjustment system. This consists
of a pair of adjustable size slits acting as a pinhole upstream of the KBs and a scintillator
(1 mm thick piece of single crystal CdWO
4
) and an optical CCD camera equipped with a
zoom lens. The system is positioned at the x-ray beam focal point. The technique uses the
Hartmann method to adjust the bends and pitch of the KB mirrors.
[7]
With this technique,
mirror figure errors of 0.2 urad can be measured and corrected for. The size of the spot
has been measured by a knife-edge method using a 150 um diameter tungsten wire. The
best focus obtained with the current KB mirrors assembly was 630 nm (h) x 500 nm (v)
(Fig. 4). This focus however has slightly degraded over time but stays stable around 1 um
for several months. Vibrations of the sample stage relative to the optics box have been
measured at ~500 nm rms and are probably one of the main factor preventing us from
obtaining a sub 500 nm spot size, although it is not the limiting factor for the current 1
um size beam. The vibration problem is in the process of being remediated by a clamping
system that will reduce the mechanical loop between the optic box and the upper sample
stages while the χ motor is unused.
Key to the success of the microdiffraction program is the ability to extract quantitative
information from both white and monochromatic beam patterns. On beamline 12.3.2, this
is achieved by using an in-house continuously evolving dedicated software called XMAS
(X-ray Microdiffraction Analysis Software).
[6]
This software’s main feature is its ability
to index thousands of patterns in an automated way and convert the information extracted
from individual patterns (such as crystal orientation and strain) into two-dimensional
maps. The indexing algorithm consists in finding angular matches of 3 non-collinear
experimental scattering vectors calculated from the spot position on the X-ray CCD
detector with those calculated from a known structure. This simple algorithm proved to
be highly robust allowing indexing up to 100 grains in a single “polycrystal” Laue pattern
or analyzing Laue patterns with fuzzy and/or highly distorted peaks due to mechanical
deformations. It tends however to be time-consuming in the cases of low symmetry
structures (triclinic, monoclinic) and large unit cell phases, but works well for most
applications in metal physics.
Applications
The beamline uses two modes of operations that can be employed in a complementary
way (Fig 5). For both modes, the sample can be raster-scanned under the X-ray
microbeam while X-ray diffraction patterns or a fluorescence signal are collected at each
step. No sample rotation is required. This avoids problems related to sphere of confusion
or non-constant diffraction volume. In polychromatic mode, polycrystalline samples
consisting of grains with sizes larger than the beam size (1 micron and more) as well as
smaller isolated single crystalline particles (such as single nanowires) can be investigated.
Each Laue pattern comes from a single crystal or a limited number of single crystals and
is readily indexed using the XMAS code. Besides grain orientation, Laue patterns can
also be used to measure the deviatoric strain as well as to identify and measure the
distribution of geometrically necessary dislocations (Fig 6).
[8]
One recent example using polychromatic microdiffraction is the study of the ordering of
nacre in a red abalone shell.
[9]
Nacre is a composite biomaterial made of 95% aragonite
and 5% various proteins, essentially chitin. It is 3000 times tougher than aragonite, a feat
that is far from being achieved by any man-made composites. In order to understand the
mechanism of nacre formation, polychromatic X-ray microdiffraction has been used to
complement X-PEEM (X-Ray photoelectron emission microspectroscopy) imaging
results. Polychromatic X-ray microdiffraction indicates that the material is made up of
aragonite tablets with a c-axis spread that starts from approximately 24º at the prismatic
calcite-nacre interface and goes down to 6º a few tens of microns away (Fig. 7). This
gradual ordering is compatible with a competitive growth mechanism where the
crystallites with a c-axis normal to the interface are kinetically favored.
A second recent application of polychromatic X-ray microdiffraction is the study of the
electromigration effect in a Sn-Cu solder joint as they are used in lead-free flip-chip
technology.
[10]
Grain orientation has been carefully monitored over a time period of over
40 hours during an accelerated electromigration test with a current density of 1.25 10
4
A/cm
2
at 75 ºC. It was observed that grains which are sitting in a current crowding region
at the anode corner of the solder ball show significant rotation of up to 0.4º over the
length of the experiment, while no other grain in the sample displays such changes in
orientation. Taking into account the anisotropy of the electrical conductivity in tin, it is
found that these observed grain rotations are driven by a tendency to decrease the
electrical resistivity in the solder bump.
Monochromatic beam mode X-ray microdiffraction is suitable for polycrystalline
samples with grain size much smaller than the beam size (nanometric sizes). In that case,
the diffraction patterns are powder patterns (Debye-Scherrer rings) for which the ring
positions can be fitted against a crystal structure database allowing for phase
identification. Slight distortions of the diffraction rings also provide information on the
macrostrain while broadening is associated to particle sizes and/or microstrain.
Monochromatic beam X-ray microdiffraction finds most of its applications in phase
identification in heterogeneous samples typical of environmental and geosciences. On
beamline 12.3.2, it has been used for instance to identify components in the bright
colored coatings (slips) of gallo-roman potteries dating from first century BC to the 2
nd
century AD.
[11]
One particular example of these tablewares is Terra Sigillata of the
marbled type, where the slip is brightly colored in a mixture of red and yellow colors. X-
ray microfluorescence and monochromatic X-ray microdiffraction have been combined
to identify the composition of the slip which consists of two layers, a 10 micron thick
layer of red slip and a 30 microns thick layer of yellow slip sequentially dip-deposited at
the surface of the pottery.
[12]
The red slip turns out to be the standard hematite-based slip
used in the conventional red Terra Sigillata, while the yellow color of the second slip was
found to be due to a titanium-rich mineral called pseudobrookite (Fe
2
TiO
5
). This yellow
color can only be obtained through specific firing conditions indicative of the high level
of mastery achieved by the ancient potters. Other phases that have been identified in the
slips are a spinel (MgAl
2
O
4
) in the yellow slip and corundum (Al
2
O
3
) in both layers,
while hematite and anorthite are ingredients of the body. Micrometric sized phases such
as quartz have also been identified in a complementary way, using the polychromatic
beam mode.
Conclusions and future prospects
Beamline 12.3.2 is a versatile X-ray microdiffraction station recently built at the
Advanced Light Source on a superconducting magnet source. The beamline is fully
operational and is now open to the user community. Some of the enhanced capabilities of
the beamline compared to the old bending magnet beamline 7.3.3 are a higher photon
flux on the sample, better strain resolution, broader energy bandpass, enhanced
computing capabilities (use of 24 node dual processor Linux cluster) and enhanced
sample positioning system. Problems currently or about to be addressed are: sub-500 nm