electronic reprint
Journal of
Synchrotron
Radiation
ISSN 0909-0495
Interferometer-controlled scanning transmission X-ray microscopes at
the Advanced Light Source
A. L. D. Kilcoyne, T. Tyliszczak, W. F. Steele, S. Fakra, P. Hitchcock, K. Franck, E.
Anderson, B. Harteneck, E. G. Rightor, G. E. Mitchell, A. P. Hitchcock, L. Yang, T.
Warwick and H. Ade
Copyright © International Union of Crystallography
Author(s) of this paper may load this reprint on their own web site provided that this cover page is retained. Republication of this article or its
storage in electronic databases or the like is not permitted without prior permission in writing from the IUCr.
J. Synchrotron Rad.
(2003). 10, 125–136 A. L. D. Kilcoyne
et al.
Microscopes at the Advanced Light Source
Interferometer-controlled scanning
transmission X-ray microscopes at the
Advanced Light Source
A. L. D. Kilcoyne,
a
T. Tyliszczak,
b,c
W. F. Steele,
c
S. Fakra,
c
P. Hitchcock,
a,b
K. Franck,
c
E. Anderson,
d
B. Harteneck,
d
E. G. Rightor,
e
G. E. Mitchell,
e
A. P. Hitchcock,
b
L. Yang,
b
T. Warwick
c
and H. Ade
a
*
a
Department of Physics, North Carolina State University,
Raleigh, NC 27895, USA,
b
Brockhouse Institute for Material
Research, McMaster University, Hamilton, ON L8S 4M1,
Canada,
c
Advanced Light Source, Lawrence Berkeley National
Laboratory, Berkeley, California 94720, USA,
d
Center for X-ray
Optics, Lawrence Berkeley National Laboratory, Berkeley,
California 94720, USA, and
e
Dow Chemical, 1897 Bldg,
Midland, MI 48667, USA. E-mail: harald_ade@ncsu.edu.
Two new soft X-ray scanning transmission microscopes located at the
Advanced Light Source (ALS) have been designed, built and
commissioned. Interferometer control implemented in both micro-
scopes allows the precise measurement of the transverse position of
the zone plate relative to the sample. Long-term positional stability
and compensation for transverse displacement during translations of
the zone plate have been achieved. The interferometer also provides
low-distortion orthogonal x, y imaging. Two different control systems
have been developed: a digital control system using standard VXI
components at beamline 7.0, and a custom feedback system based on
PC AT boards at beamline 5.3.2. Both microscopes are diffraction
limited with the resolution set by the quality of the zone plates.
Periodic features with 30 nm half period can be resolved with a zone
plate that has a 40 nm outermost zone width. One microscope is
operating at an undulator beamline (7.0), while the other is operating
at a novel dedicated bending-magnet beamline (5.3.2), which is
designed speci®cally to illuminate the microscope. The undulator
beamline provides count rates of the order of tens of MHz at high-
energy resolution with photon energies of up to about 1000 eV.
Although the brightness of a bending-magnet source is about four
orders of magnitude smaller than that of an undulator source, photon
statistics limited operation with intensities in excess of 3 MHz has
been achieved at high energy resolution and high spatial resolution.
The design and performance of these microscopes are described.
Keywords: X-rays; zone plates; scanning microscopy; NEXAFS.
1. Introduction
Structure and phenomena on a length scale from nanometres to
micrometres are the focus of much current scienti®c research. For
example, heterogeneous composition and chemical properties on this
scale are key to the behaviour of multiphasic polymeric materials,
biomaterials and composites. Furthermore, environmental and
biological processes often involve microscopically variable chemistry.
There is therefore a growing need for detailed compositional and
chemical analysis at high spatial resolution. Transmission X-ray
microscopy in both scanning (STXM) and full-®eld variants (TXM) is
becoming increasingly important on account of the relatively low
radiation damage and compositional information that is provided by
near-edge X-ray absorption ®ne-structure (NEXAFS) spectroscopy.
Various implementations of zone-plate-based scanning transmission
X-ray microscopes (STXM) have been developed during the past two
decades at several synchrotron radiation facilities (Kirz & Rarback,
1985; Kirz et al., 1995; Warwick et al., 1998; McNulty et al., 1998;
Kaulich et al., 1999). These instruments have been used to study many
different types of samples including polymeric (Ade et al., 1992, 1995;
Ade & Hsiao, 1993; Ade & Urquhart, 2002), geochemical (Botto et
al., 1994; Cody et al., 1998), environmental (Plaschke et al., 2002),
magnetic (Kim et al., 2001), extraterrestrial (Flynn et al., 1998; Keller
et al., 2000) and biological (Kirz et al., 1995) samples. Most STXM
studies use the information provided by the NEXAFS of elements
with absorption edges in the soft X-ray energy range, particularly of
carbon, nitrogen and oxygen (Sto
È
hr, 1992). The ability of STXM to
adapt to a wide range of sample environments, such as magnetic ®elds
(Kim et al., 2001), variable temperature (Wang et al., 2000) and
polymeric and environmental samples in aqueous media (Neuha
È
usler
et al., 1999; Mitchell et al., 2002), is an important aspect of the tech-
nique, as is the relatively low amount of radiation damage caused by
soft X-rays (Rightor et al., 1997; Coffey et al., 2002). In order to scan
the photon energy for NEXAFS we need to use a synchrotron
radiation facility as a tunable source for these instruments. In order to
acquire useful information in a reasonable time, high photon ¯ux in a
small spot is required. During the past decade STXMs have been
operated exclusively at high-brightness undulator sources (Kirz et al.,
1995; Warwick et al., 1998). Since the number of straight sections at
high-brightness synchrotron facilities is very limited, the use of
undulator sources greatly limits the potential for growth of STXM
technology. Furthermore, since the overall utility of NEXAFS
microscopy to a wide variety of ®elds is now without question, the use
of high-brightness bending magnets would offer increased access.
Two STXMs that signi®cantly extend prior instrumental capabil-
ities have been commissioned at the Advanced Light Source (ALS)
in Berkeley. Their design and performance are described and
discussed. One of these microscopes is the upgrade to the ®rst-
generation STXM at the ALS and is located on undulator beamline
7.0. In an important new development, the other new microscope has
been installed on a dedicated bending-magnet beamline (5.3.2) in
order to explore and, if adequate performance is found, exploit the
utility of high-brightness bending magnets for STXM illumination.
The 5.3.2 beamline itself was highly optimized for STXM use and was
described in a companion paper (Warwick et al., 2002). This bending-
magnet beamline and the STXM implemented there have been
shown to have excellent performance. The STXMs described here
will be designated by their present beamline number at the Advanced
Light Source and will be referred to as the 5.3.2 STXM and the 7.0
STXM, respectively. These new STXMs share many design features
and both use laser interferometry to accurately position the sample
relative to the X-ray optics. We will ®rst discuss the basic operating
principles of a STXM and describe the prior state-of-the-art tech-
nology to provide context. Subsequently, the design of the new ALS
STXM instruments will be presented and their performance will be
discussed with results from the 5.3.2 STXM as examples.
1.1. Basic operating principle of a zone-plate-based STXM
In a scanning X-ray microscope, a small spot of X-rays is raster
scanned relative to the sample to create an image one pixel at a time
while a suitable signal is monitored under computer control. The
most common signal is the photon ¯ux that is transmitted through the
sample (STXM). In addition, measurements in the scanning instru-
ments have also used front-face electron yield, luminescence yield
and X-ray ¯uorescence yield. Because of challenges in scanning
X-ray optics, samples are typically scanned relative to a stationary
J. Synchrotron Rad. (2003). 10, 125±136 # 2003 International Union of Crystallography
Printed in Great Britain ± all rights reserved 125
research papers
electronic reprint
research papers
126 A. L. D. Kilcoyne et al.
Microscopes at the Advanced Light Source J. Synchrotron Rad. (2003). 10, 125±136
X-ray beam in a STXM. Two-axis x, y piezoelectric stages have
generally been used for ®ne motion over a limited ®eld of view. This
x, y piezo stage itself is mounted on top of x, y, z stepping motor
stages that provide motion in excess of many millimetres for large-
scale images or coarse positioning of dispersed small samples. The
small X-ray spot is produced by a zone-plate lens. This is a variable-
line-spacing circular diffraction grating.
In order to suppress unwanted diffraction orders that would
decrease the available signal-to-background ratio, the zone plate is
fabricated with a central stop. A slightly smaller pinhole, the order
selection aperture (OSA), is placed between the zone plate and the
sample at about 0.75 times the zone-plate-to-sample distance. This
allows only the positive ®rst diffraction order to pass (see Fig. 1).
Careful transverse and longitudinal alignment of the OSA with the
zone plate is an essential aspect of tuning and operating a STXM. The
alignment determines the imaging properties of the microscopes and
the higher-order spectral contamination that may pass the zone-plate/
OSA arrangement.
The numerical aperture (NA) of a zone plate, i.e. the largest
diffraction angle for the variable zones, is determined by the size of
the outermost zone width r at a ®xed wavelength of the incident
X-rays [NA = /(2r)]. For fully coherent illumination, plane-waves
or point sources, the Raleigh resolution of a zone plate, which
corresponds to the ®rst minimum of the Airy diffraction pattern,
depends on the size of the central stop relative to the size of the zone
plate and ranges from 0.9r to 1.22r (Michette, 1986). In practice, a
®nite source is used for illumination, and the diffraction pattern is
convolved with the de-magni®ed ®nite source to obtain the actual
point-spread function of the STXM. The illumination parameter
p = d/, where d is the illuminating source size and is the full
acceptance angle of the zone plate, is a convenient way to char-
acterize the illumination. p = 0.5 corresponds roughly to the half-
Airy-disk illumination criterion, and represents a good trade-off
between ¯ux and achievable spatial resolution (Kirz et al., 1995). The
point-spread function and the modulation-transfer function of a zone
plate as a function of p have been calculated (Winn et al., 2000), and
the result shows in great detail how the imaging characteristics
depend on the illumination characteristics. Random or systematic
departures of the zones from circularity will also degrade the optical
properties of a zone plate. In order to achieve adequate performance
the zones have to be placed with an absolute accuracy of about 0.3r
(Michette, 1986). The achievable resolution depends on the
capabilities of the zone-plate-fabrication technology and on the
degree of coherence of the illuminating beam. Zone plates with
outermost zone widths as small as 20±25 nm have been fabricated,
and a spatial resolution suf®cient to resolve half-period features
25±30 nm in size has been achieved (Anderson et al., 2000; Spector et
al., 1997; Schneider et al., 1995; Denbeaux et al., 2001). Details of soft
X-ray optics for STXM have been discussed in articles by Jacobsen et
al. (1991, 1992).
Zone plates are achromatic lenses with a focal length proportional
to the photon energy. Thus, during the acquisition of NEXAFS
spectra or an image-sequence at many energy values (sometimes
referred to as `stacks'), a STXM microscope has to be refocused
synchronously with the photon energy changes (Jacobsen et al., 2000).
Depending on the focal length of the zone plate used, this requires
motions over 150±200 mm for a 30 eV-wide C 1 s NEXAFS scan or
> 1 mm for a change between different absorption edges. This range
can typically only be covered with conventional roller-bearing or
crossed-roller-bearing stages. Unfortunately, these stages exhibit
transverse motion, which is referred to as run-out, of the order of
several hundreds of nanometres during a translation. Unless
controlled, the run-out results in a relatively random misregistration
of subsequent images of the same sample area or a `blurring' of the
spot size sampled during the acquisition of a spectrum from a point
on the sample. If in-focus point spectra are acquired from small highly
contrasting sample features, the spectra will contain artifacts with
large spectral amplitudes if the photon beam moves on and off a
feature of interest because of uncorrected transverse run-out.
Overall, a STXM requires control and operational adjustment of
the z position of the zone plate, at least x and y control of the OSA, x,
y and z control over the sample, and generally x, y and z control over
the detector.
For alignment, since the X-ray beam is ®xed in space externally,
either the zone plate or the whole instrument needs to be translated
in x and y relative to the X-ray beam with about 10 mm precision. The
translation stage that provides the focusing has to be aligned in two
angles to be parallel with the externally de®ned optical axis. The
rotational degree of freedom around the optical axis, i.e. roll, is not
important.
1.2. Prior state-of-the-art technology
Pioneering efforts to construct the ®rst soft X-ray STXM were
made in the early 1980s (Rarback et al., 1984) at bending-magnet
beamline U15 at the vacuum ultraviolet ring at the National
Synchrotron Light Source (NSLS). The U15 STXM utilized a
toroidal-grating monochromator with a resolving power E/E = 200
at photon energies within the so-called water window between the
absorption edges of oxygen and carbon. An initial spatial resolution
of about 300 nm could be achieved with a zone plate that had an
outermost zone width r = 150 nm. Typical count rates were on the
order of 10 kHz.
The Stony Brook effort was ®rst relocated within the NSLS to a
temporary undulator at beamline X17 in 1986±1987 (Rarback et al.,
1988) and, subsequently, in 1988±1989 (Buckley et al., 1989), to its
present location at beamline X1A. The spatial resolution improved to
75 nm in 1988 (25±75% knife-edge test) with a 50 nm outermost
zone-width zone plate (Rarback et al., 1988). By 1991, 35 nm half-
period features could be resolved (Jacobsen et al., 1991; Kirz et al.,
1992), and, by 1997, half-period features of less than 30 nm in size
could be resolved (Spector et al., 1997). Typical count rates with the
NSLS X1A microscope for operation near the carbon edge and with
low higher-order spectral contamination are about 500 kHz. For
Figure 1
Focusing scheme of a STXM.
electronic reprint
much of the past two decades, the Stony Brook efforts have de®ned
the state-of-the-art in STXM technology.
The X17 and the initial X1A STXM used a custom-built laser
interferometer that included custom electronics to provide high-
speed control and readout of the sample position relative to a
reference prism (Rarback et al., 1988; Shu et al., 1988). These
microscopes were designed to be used at a ®xed photon energy within
the water window. The interferometer had a least-count step size, or
sensitivity, of 31.5 nm and was somewhat dif®cult to maintain. This
sensitivity was insuf®cient for the high spatial resolution achieved,
and, when integrated capacitance-controlled scanning stages with
much higher sensitivity became available, from Queensgate and later
from Physik Instrumente, two further incarnations of the Stony
Brook STXMs were developed, in 1990±1992 and 1998±2000,
respectively, which could take full advantage of the improvements
made in zone-plate technology. Neither of these microscopes was
designed to control run-out directly in hardware or ®rmware. In-
accurate registration between images is corrected with post proces-
sing based on image-correlation procedures (Jacobsen et al., 2000).
Point spectra and line spectra cannot be measured at well de®ned
locations with an accuracy near the spatial resolution limit. This
information has to be extracted from image sequences after data
acquisition. Details about the Stony Brook microscopes have been
published (Jacobsen et al., 1991; Maser et al., 2000; Feser et al., 1998).
The Stony Brook team also developed a cryo-STXM with tomo-
graphy capabilities (Wang et al., 2000).
At the ALS, a STXM was developed in the mid-1990s by a team
that included several of the present authors (Warwick et al., 1998).
The STXM was located on beamline BL7.0, an undulator beamline
that was conceived and optimized for spectroscopy. This ®rst-
generation ALS BL7.0 STXM was based on a capacitance-controlled
piezoelectric stage from Queensgate Inc. as a ®ne stage. It used
Newport PM-500 stages as sample coarse stages and for the zone-
plate z motion. The coarse and ®ne stages could be rotated together
to provide in-focus scans over the full ®eld of view when polar sample
rotation with respect to the polarization vector of the photon beam
was desired (Kim et al., 2001). The interface electronics were based
on VXI crate/modules controlled by a Unix workstation. A second
Unix workstation was dedicated for a graphical user interface (GUI).
The GUI and low-level interface software were written in Labview.
Spatial resolution of 60 nm FWHM was measured in a knife-edge test
(Warwick et al., 1999). Routine operation used a slightly incoherent
beam that resulted in a degraded spatial resolution on the order of
100 nm with improved noise characteristics. The microscope suffered
from signi®cant run-out of the zone-plate ®ne and coarse stages
(> 500 nm) and from magni®ed Abbe
Â
errors, the latter due to an off-
centre sample mount. In addition, the STXM operation has to be
shared with several other instruments on BL7.0 on a daily timeshared
basis. One of the microscopes described below is the upgrade to this
®rst-generation STXM at the ALS.
A STXM, built by a group from King's College, London (Kenney
et al., 1989), was operated for a short time at the Daresbury U6
undulator beamline. The Go
È
ttingen X-ray microscopy group devel-
oped and operated for a short period a STXM located at a bending
magnet of the BESSY-I facility. This group is now developing a new
STXM and a new full-®eld transmission X-ray microscope (TXM),
which are both located at the same undulator at BESSY-II (Gutt-
mann et al., 2001). Scanning X-ray microscopes at higher photon
energy have been implemented at the Advanced Photon Source
(McNulty et al., 1998) and at the European Synchrotron Radiation
Facility (Kaulich et al., 1999). Several scanning photoemission
microscopes (SPEM) have been built at a number of facilities. These
SPEMs could be used, but are not optimized, for transmission
experiments (Ade et al., 1990; Ko et al., 1995, 1998; Marsi et al., 1997;
Ng et al., 1994; Shin & Lee, 2001; Welnak et al., 1995).
2. Instrument description: the 5.3.2 and 7.0 STXMs at the ALS
2.1. Technical goals and requirements
The new ALS STXMs will be used almost exclusively for NEXAFS
spectromicroscopy. This requires energy scanning and continuous
refocusing. Prior experience showed that post-processing of data via
image-correlation techniques does not always eliminate the run-out
or drift that occur during data acquisition of image sequences, the
result being residual spatial degradation (Hitchcock et al. , 2003).
Some correlation algorithms can be unreliable, and tedious manual
image registration may be required. Our goal is to acquire data in a
sequence of photon energy values in registry at the spatial-resolution
limit of the microscope. Run-out has to be eliminated at the hardware
level or automatically compensated for during data acquisition. In
addition, in order to eliminate systematic image shifts as the focal
length is changed, the microscope mechanical reference system has to
be precisely colinear with the optical axis de®ned by the X-ray beam.
Low run-out and component colinearity mean that high-quality
spectroscopic information from a few speci®c locations in the sample
is immediately availability to the instrument operator, and thus the
ef®ciency and productivity of the instruments are improved.
Although the mechanical design of these two new microscopes at
the ALS differs in some ways, they share the same basic concepts and
components. The main new feature is the differential measurement of
the relative position of the zone plate and the sample, which greatly
improved the run-out of the new microscopes compared with that of
the old 7.0 STXM. The 5.3.2 STXM is described here in detail. Some
of the differences of the 7.0 STXM are pointed out.
2.2. System components
The following major system components have been incorporated:
zone-plate z stage; OSA x, y stages; sample x, y coarse stage; sample
x, y ®ne piezo stage; sample z stage; detector x, y, z stage; inter-
ferometer system; mounting platform; passive vibration isolation to
the ¯oor and vacuum/He enclosure. The arrangement and stacking of
these stages is shown schematically in Fig. 2. Fig. 3 shows three-
dimensional CAD drawings of the 5.3.2 STXM. Annotated photo-
graphs of the instrument as implemented at beamline 5.3.2 are shown
in Fig. 4.
J. Synchrotron Rad. (2003). 10, 125±136 A. L. D. Kilcoyne et al.
Microscopes at the Advanced Light Source 127
research papers
Figure 2
Schematic of the 5.3.2 STXM components.
electronic reprint
research papers
128 A. L. D. Kilcoyne et al.
Microscopes at the Advanced Light Source J. Synchrotron Rad. (2003). 10, 125±136
2.3. Mounting platforms and vibration control
Traditional air±table vibration-reduction systems are low-
frequency oscillators that isolate a heavy table from ¯oor vibrations
at frequencies above the lowest rigid-body mode, which typically
resonates at about 1 Hz. These systems are soft and can easily be
moved by external forces. Such systems will not maintain the long-
term angular alignment that is required with respect to the optical
axis.
Rather than isolate the instruments from ¯oor vibrations, we have
built these microscopes on a rigid polymer composite base ± referred
to as polymer granite or polymer concrete (Zanite, Precision Polymer
Casting) ± that avoids resonant ampli®cation at low frequencies. The
top surface of the polymer-granite block moves with the ¯oor. The
microscopes themselves are built so that only a small fraction of this
amplitude (typically 10 nm) shows up as problematic transverse
vibration of the zone-plate lens relative to the sample. A multi-tier
approach has been implemented. An inner mounting base plate for
the STXM (see Fig. 3) is supported by a high-stiffness six-strut
adjustable support system (Thur et al., 1997) mounted on the heavy
polymer-granite block. This support system allows the adjustment of
three translation and three angular degrees of freedom in order to
accurately position the whole microscope and hence the zone plate
and its translation axis with respect to the optical axis. The zone plate
and sample are mounted from this `inner' base plate with the set of
stages as depicted in Figs. 2 and 3.
A vacuum vessel is employed,
which covers the instrument without
transmitting vacuum forces to the
alignment mechanism and the
aligned components inside and
without transmitting vibrations of
the beamline components to the
instrument. The chamber seals to a
metal plate embedded into the top
of the granite, and this plate does
not move or de¯ect in any appreci-
able way as the pressure is changed.
The ALS ¯oor vibrations have a
pronounced 110 Hz component. An
attempt was thus made to keep
resonance frequencies of all
components above 130 Hz. The
polymer-granite block, which weighs
about 1200 kg, is supported on the
¯oor by four 2 inch-thick poly-
styrene-foam pads. The foam has
suf®cient strength and long-term
stability to support the massive
block yet provides for some passive
damping of vibrations. The 5.3.2
STXM six-strut support system uses
0.75 inch turnbuckle struts, with a
stiffness of 52 N mm
ÿ1
each, in a
con®guration that provides the most
stiffness in the y direction (three
struts), a medium stiffness in the x
direction (two struts) and the least
stiffness in the z direction (one
strut). This stiffness is matched to
the relative magnitude of the hori-
zontal (x) and vertical (y) vibrations
in the ¯oor, as well as the relative
tolerance of the microscope to these
vibrations: x, y has a tolerance of
10 nm (one-quarter of the transverse
resolution), while z has a tolerance
of about 100 nm (one-quarter of the
longitudinal resolution). The three
struts in the y direction that support
the 180 kg microscope result in a
resonant frequency of about 150 Hz.
The BL7.0 STXM is lighter and uses
0.5 inch turnbuckle struts in a
slightly different con®guration.
Figure 4
Photographs of the STXM at beamline 5.3.2.
Figure 3
Annotated CAD drawings of the instrument at beamline 5.3.2.
electronic reprint
