Synchrotron radiation induced TXRF†
C. Streli,
*
a
P. Wobrauschek,
a
F. Meirer
a
and G. Pepponi
b
Received 18th December 2007, Accepted 10th April 2008
First published as an Advance Article on the web 6th May 2008
DOI: 10.1039/b719508g
The use of synchrotron radiation (SR) as an excitation source for total reflection X-ray fluorescence
analysis (TXRF) offers several advantages over X-ray tube excitation. Detection limits in the fg range
can be achieved with efficient excitation for low Z as well as high Z elements due to the features of
synchrotron radiation and in particular the high brilliance in a wide spectral range and the linear
polarization in the orbital plane. SR-TXRF is especially interesting for samples where only small
sample masses are available. Lowest detection limits are typically achieved using multilayer
monochromators since they exhibit a bandwidth of about 0.01 DE/E. Monochromators with smaller
bandwidth like perfect crystals, reduce the intensity, but allow X-ray absorption spectroscopy (XAS)
measurements in fluorescence mode for speciation and chemical characterisation at trace levels.
SR-TXRF is performed at various synchrotron radiation facilities. An historical overview is presented
and recent setups and applications as well as some critical aspects are reviewed.
Introduction
Total reflection X-ray fluorescence analysis is a well established
analytical technique for the detection of major, minor and trace
elements
1–3
especially suited for samples, where only small
specimen mass is available. Moreover TXRF can be used to
investigate wafer surface contaminations or determine depth
profiles in the near surface region and determine the implanta-
tion dose when using the angular dependence of the fluorescence
signal.
4
Detection limits achieved with X-ray tube excitation are
in the range of 1 pg.
5
If the detection limits are to be further improved it is helpful to
look at the definition of the limits of detection (LD):
LD ¼
3
ffiffiffiffiffiffiffi
N
B
p
N
N
m
sample
or LD ¼
3
ffiffiffiffiffi
I
B
p
S
1
ffiffi
t
p
(1)
where, N
N
are the net counts, N
B
the background counts, m
sample
is the sample mass, I
B
is the background intensity, S is the
sensitivity (net intensity/m
sample
) and t is the measuring time.
One can easily see from eqn (1) that there are different ways of
improving the detection limits, namely increasing the sensitivity
S (I
N
/m), reducing the background, and increasing the measuring
time, which, however, is limited for practical reasons.
Besides using total reflection geometry for reducing the spectral
background and doubling the fluorescence signal from the
sample, a further possibility to reduce scatter contributions from
the sample itself is the use of linearly polarized primary
radiation.
6,7
Due to the anisotropic emission characteristics of the
scattered radiation based on the classical dipole oscillator
emission it is advantageous to place a detector in such a position
that only the isotropic emission of the fluorescence signal is
detected. Hence the combination of TXRF with polarized
radiation leads to a lower background. Moreover, the use of
monochromatic primary radiation improves the background
conditions because only photons of one energy can be scattered.
An increase in sensitivity can be attained by using a tunable intense
excitation source, enabling the exciting energy to be adjusted
to just above the absorption edge of the element of interest.
Christina Streli (far left) is Professor at the Atominstitut of the
Vienna University of Technology, Austria. Her research interest is
the development of special X-ray methods for interdisciplinary
applications with the main focus using synchrotron radiation as the
excitation source. Peter Wobrauschek (far right) is retired
Professor at the Atominstitut working in the same group as
Christina Streli. Florian Meirer (second from left) is undertaking
his PhD in this group. Giancarlo Pepponi (second from right) is
senior scientist at the Centre for Materials and Microsystems of
the Fondazione Bruno Kessler—IRST in Trento, Italy. Since the
attainment of his PhD in the group of Christina Streli in 2003,
there has been a long term cooperation on SR-TXRF between the
two groups.
a
Atominstitut, Vienna University of Technology, Vienna, Austria. E-mail:
streli@ati.ac.at
b
Centre for Materials and Microsystems of the Fondazione Bruno
Kessler—IRST, Povo, Trento, Italy
† This paper is part of a JAAS themed issue on Synchrotron Radiation,
with guest editors Alex von Bohlen and Metin Tolan.
792 | J. Anal. At. Spectrom., 2008, 23, 792–798 This journal is ª The Royal Society of Chemistry 2008
CRITICAL REVIEW www.rsc.org/jaas | Journal of Analytical Atomic Spectrometry
Synchrotron radiation with its outstanding properties has
offered new possibilities for improving the power of TXRF. The
intense beam with a continuous spectral distribution from
photon energies in the infrared region to high energy photons as
well as the linear polarization in the orbit plane and its natural
collimation are features best suited for excitation in total
reflection geometry.
For optimal excitation conditions the spectral distribution can
be modified by elements like cut-off mirrors, monochromators
and filters. Details can be found in ref. 8.
Multilayer monochromators are best suited for XRF
analysis.
9,10
In comparison to crystal monochromators they offer
a larger bandwidth (DE/E z 0.01), which leads to a much larger
photon flux on the sample. Another advantage is the possibility
of selecting the excitation energy just below a matrix element
with high concentration and just above the absorption edge of
the element of interest (‘‘selective excitation’’), with the possible
drawback, however, of an increased background due to Raman
scattering.
11
If the experiments are performed in air, scattering of the
exciting radiation contributes to the background. Therefore all
measurements should be done in a vacuum chamber.
The combination of TXRF with synchrotron radiation may be
performed with various geometrical arrangements for reflector
and detector. Fig. 1 shows three possibilities.
For geometry A the polarization effect is fully utilized by
positioning the detector axis in the plane of the orbit. Scattered
radiation is not emitted in that direction. The sample is excited
efficiently, and full homogenous illumination of the sample by
the width of the beam in the horizontal plane is given. There are
hardly any losses due to the collimators because the beam is
naturally collimated in the vertical plane (0.1 to 0.2 mrad,
depending on the energy). The detection of the fluorescence
signal is not optimal because the detector must be sidelooking to
use the polarization effect. The fluorescent radiation has a long
path in the sample to reach the detector giving rise to absorption
of the fluorescence in the sample and possible quantification
errors.
The excitation conditions for the arrangement as displayed in
Fig. 1B are poor. Most of the photons in the horizontal plane are
absorbed in the collimation system. The intensity distribution
and the degree of linear polarisation drop along the vertical axis
when moving from the orbital plane and therefore the fluores-
cence intensity also drops with the deviation of sample regions
from the plane of reference. Thus a restriction to 2 mm sample
diameter is advisable due to the intensity and polarization
distribution in the vertical plane. However the detection
efficiency is high because of the large solid angle obtained due to
the small distance between reflector and detector.
Excellent excitation and detection will be achieved with
arrangement C. This combination of sample–detector position
though results in a complete loss of the use of the polarization
effect. If the sample is small, which is the case in ultra trace
analysis, the scattering contribution from the sample itself is
negligible. Scattering from the substrate is reduced by total
reflection.
Historical review
The first experiments were published by Iida et al.
12
in 1986
performed at the Photon factory in Japan with a vertical reflector
and sidelooking detector. As the used monochromator was Si
111, the obtained detection limits were in the pg range, compa-
rable with tube excitation. In 1988 Pella and Dobbyn
13
published
results from their experiments at NSLS using a horizontal
reflector and a sidelooking detector.
The first experiments of the Atominstitut group were
performed at SSRL, beamline 10-2 together with F. Hegedu
¨
sin
1991. The determination of transmutational elements like Ni in
a Cu matrix after p + bombardement was the task.
14
A crystal
monochromator and a sidelooking detector geometry was used.
The reflector was mounted vertically.
In 1994 Brennan et al. performed experiments at SSRL at
a wiggler beamline—beamline 6-2.
15
The main goal was to
optimize the setup for Si wafer surface characterization. LD’s of
1 10
8
atoms cm
2
have been achieved using a double
ML-monochromator and a Teflon-filter
16,17
to prevent saturation
of the detector.
The first experiments at HASYLAB, DESY Hamburg,
Beamline L, which is a bending magnet beamline, were
performed by the Atominstitut group in 1994. The arrangement
was a vertical reflector and sidelooking detector. Detailed
information concerning the experiments are given in ref. 18 and
19. Detection limits for Ni with an excitation energy of 10 keV
Fig. 1 Three possibilities of arranging wafer and detector for SRTXRF.
This journal is ª The Royal Society of Chemistry 2008 J. Anal. At. Spectrom., 2008, 23, 792–798 | 793
were found to be 13 fg or 1.3 10
8
atoms cm
2
for an inspected
area of 1 cm
2
. A multilayer was used to monochromatize the
beam and an Al filter was applied to prevent the detector being
saturated by the Si signal. Also, comparison of the various
geometries has been performed by the Atominstitut group.
20
Pepponi et al.
21
compared the vertical and horizontal geometry
for different kinds of samples and found that for samples with
a strong scattering matrix the horizontal geometry provided
better results; for samples with no matrix the vertical geometry
provided better results. He also compared SR-TXRF with
microanalysis on ultra-thin foils and found that TXRF obtained
better detection limits.
22
At the ESRF, the European Synchrotron Radiation Facility in
Grenoble, a feasibility test in 1996 showed that third generation
machines, as the fully dedicated and application oriented
installations like ESRF are called nowadays, offer new possi-
bilities for mapping the distribution of metallic contaminants
over the wafer surface with a LD in the 10
8
atoms cm
2
range.
23
A
dedicated setup for a 300 mm wafer analysis was built—details
can be found in ref. 4—but the setup was not accepted by the
semiconductor industry.
For the low Z elements the ideal source for efficient excitation
is definitely synchrotron radiation due to its high intensity also in
the low energy region. Experiments have been started with the
work of Madden et al. in 1993
24
at SSRL, Beamline III-4 with
filtered white radiation. The Atominstitut group started their
experiments also in 1993 at Beamline III-4 and investigated in the
following years the spectral distribution and geometries.
25–27
The
best suitable beamline at SSRL turned out to be the bending
magnet beamline BL 3-3 in combination with a multilayer
monochromator. Details of the setup can be found in ref. 28.
Droplet samples were compared with spin coated wafer
samples.
29
Detection limits of about 60 fg for Na were achieved.
Angle scans showed different adsorption behavior for the
elements Al, Mg and Na in a multielement sample.
In 2000 Baur et al.
11,30
performed experiments at Beamline
III-4 on low Z elements investigating the resonant Raman effect
on spectra from Si wafers excited with an energy just below the
absorption edge of Si. The authors obtained 2.8 10
10
atoms
cm
2
detection limits for Al on Si wafers.
In 1999 the ATI group started experiments at the plane grating
monochromator beamline for undulator radiation of the
Physikalisch-Technische Bundesanstalt (PTB) at BESSY2,
Berlin, Germany. The beamline could also be operated in wiggler
mode, so excitation of Na, Mg and Al was possible, the detection
limits obtained were in the low pg range.
31
As this beamline is
best suited for delivering a low energy (< 1 keV) beam with
extremely high spectral purity, detection limits for C and N could
be determined to be below 1 pg. Also, a comparison between the
possible geometries has been performed, the vertical arrange-
ment of the wafer and sidelooking detector turned out to provide
better results. Details can be found in ref. 32. Low Z elements on
Si wafers were measured,
33
droplet samples containing B were
analyzed and detection limits of 3 ng were achieved. In addition,
samples of a 5 nm carbon monolayer and carbon–nickel–carbon
multilayer on silicon wafers were measured simultaneously for
thickness and density. Further studies on resonant Raman
scattering were performed, especially with regard to the influence
of the excitation energy, set below the absorption edge of
silicon, on the spectral background affecting the low Z element
(see ref. 33).
Recent activities
A vacuum SR-TXRF setup has been available since 2004 at
HASYLAB, Beamline L, Hamburg, Germany.
34
The TXRF
vacuum chamber is exchangeable with the instrumentation of the
microfocus setup at Beamline L, a bending magnet beamline.
The translation and rotation stages of the microfocus setup can
be directly used for adjustment. A sample loader for 30 mm
circular sample reflector—arranged in the vertical—is available
as well as a sample holder for Si wafers up to 100 mm diameter.
For the 30 mm reflectors an 8 stage sample changer has been
installed recently.
35
Fig. 2 shows a spectrum obtained from a Ni
sample—from this, detection limits of 8 fg have been deduced.
Details can be found in ref. 36. Using the Si 111 crystal mono-
chromator available at beamline L, X-ray absorption near edge
structure (XANES) investigations in total reflection geometry on
trace elements are also feasible. The setup—now equipped with
a50mm
2
SDD—is available for all users.
The SR-TXRF setup at NSLS, Campinas, Brazil is also
operated in a vacuum chamber; white beam excitation of
a bending magnet beamline as well as monochromatic radiation
from Si 111 is used.
37
The sample reflector is located vertically.
Detection limits of 0.04 mgml
1
are reported. Mainly environ-
mental samples are analyzed.
The SR-TXRF setup at Bejing Synchrotron, Bejing, China, is
operated in air using white beam excitation of a bending magnet
beamline.
38
A SR-TXRF setup for wafer surface analysis at the PTB
Beamline at BESSY2, Berlin, Germany, has been designed by the
Physikalisch-Technische Bundesanstalt.
39
This beamline
provides low energy radiation (0.1 keV and 1.9 keV) from an
undulator monochromatized by a plane grating monochromator
with high energy resolution. The instrument can handle up to
a 300 mm wafer and is also suited for EDXRF analysis of thin
structures deposited in silicon wafers. The most prominent
features are a high vacuum load-lock combined with an
equipment front end module and an UHV irradiation chamber
with an electrostatic chuck mounted on an 8-axis manipulator.
Fig. 2 Spectrum of a sample containing 100 pg of Ni on a Si wafer as
sample reflector excited with 17 keV using the multilayer mono-
chromator, lifetime: 60 s, the detection was found to be 8 fg (1000 s), with
a sensitivity of 340 cps ng
1
mA
1
.
794 | J. Anal. At. Spectrom., 2008, 23, 792–798 This journal is ª The Royal Society of Chemistry 2008
The whole wafer surface of a 200 and a 300 mm wafer can be
scanned.
40
Recent activites report on detection limits below
100 fg for Al and below 40 fg for Na. Also reference free
quantification was performed successfully.
41
A SR-TXRF setup for wafer analysis is also available at
SSRL, Stanford, California.
42
A vacuum chamber in a clean
room environment has been installed at Beamline 6-2, a wiggler
beamline suitable to analyze wafers up to 250 mm wafers. The
beam is monochromatized by a double multilayer mono-
chromator. The wafer is held by an electrostatic chuck mounted
vertically, the detector is sidelooking and specially adapted to not
get spurious peaks from the detector collimator. Detection limits
of 1 10
8
atoms cm
2
for Ni are reported. The beamline is also
equipped with a Si 111 double crystal monochromator, so also
absorption spectroscopy can be performed. No automatic wafer
load lock is installed up to now, but cleanroom environment class
100 is established around the vacuum chamber. Recently the
successful measurement of the elemental composition of material
from the NASA Gemini mission to get information about the
solar elemental abundances
43
has been reported. Wafers were
exposed to solar winds, the ions implanted in the wafer surface.
Using angle dependence measurements the elemental composi-
tion could be determined.
SR-TXRF experiments with a wavelength dispersive (WD)
spectrometer at SPring8, Harima, Japan, were described by
Sakurai et al.
44
At beamline 40XU, a helical undulator source
with a Kirkpatrick-Baez (K-B) optic for focusing and high
harmonic suppression is available. This quasi monochromatic
radiation was used for TXRF followed by a Johansson-type
spectrometer equipped with a Ge (220) curved analyzing crystal
and with a YAP:Ce detector. The absolute and relative detection
limit for nickel are 3.1 10
16
g and 3.1 ppt (pg g
1
) for a 0.1 mL
droplet of pure water, respectively, which is nearly 50 times better
than the current best data achieved by conventional energy-
dispersive TXRF using a Si(Li) detector system. Awaji et al.
described a further WD setup at SPring8.
45
At beamline 16XU,
an undulator beamline, a Si 111 double crystal monochromator
was used in combination with a Rh-coated focusing mirror to
suppress the higher harmonics. The sample stage for wafer
handling was taken from a Rigaku TXRF3000 spectrometer,
from a load lock chamber. Kurunczi and Sakurai described
a special method of sample preparation of water samples for
WD-SRTXRF; they used HF etching to obtain a hydrophobic
silicon surface and were able to produce a microdroplet of
80 mm.
46
XAS measurements in TXRF geometry
SR-TXRF was used for XAFS (X-ray absorption fine structure)
measurements (i.e. NEXAFS (near edge X-ray absorption fine
structure), XANES (X-ray absorption near edge structure) and
EXAFS (extended X-ray absorption fine structure)) in fluores-
cence mode at various SR facilities which offers new possibilities
of speciation on tiny sample amounts and low concentrations at
trace levels. At Hamburg’s HASYLAB, Beamline L, XAFS in
fluorescence mode using total reflection geometry was compared
with standard 45
geometry and verified that the sensitivity
in total reflection geometry was increased thereby lowering
the accessible concentration range.
47
Pepponi et al.
48
reported
NEXAFS measurements of organic contaminants on silicon
wafers that were performed at the PTB plane grating mono-
chromator beamline for undulator radiation at the electron
storage ring BESSYII. The K edges of C, N and O were
examined and speciation was performed. The same setup was
used by To
¨
ro
¨
k et al.
49
for NEXAFS measurements on nitrogen
compounds in aerosol samples collected on silicon wafer
surfaces. The detection limits in the low pg range for nitrogen
allow analysis of samples collected in only 10 min with acceptable
accuracy. Osa
´
n et al.
50
used the PTB laboratory at BESSY, to
apply the SR-TXRF-XANES technique to low Z elements. They
reported the ammonium to nitrate ratio in Antarctic fine aerosols
collected from less than 2 m
3
of air. For Antarctic fine aerosols in
the size range of 0.25–0.5 mm, nitrogen was observed to be
present almost entirely as the ammonium species. When the size
of aerosol particles increased in the range of 0.25–2 mm, the
content of ammonium decreased and that of nitrate increased.
Streli et al.
36
used this combination of techniques for arsenic
speciation (As
III
and As
V
) in xylem sap of cucumber plants and
achieved very good detection limits of As of 170 ng l
1
. Recently
the results on As speciation in xylem sap of cucumber plants have
been published by Meirer et al.
51
Fig. 3 shows some XANES
spectra.
Giubertoni et al.
52
and d’Acapito et al.
53
recently reported
EXAFS measurements at the GILDA beamline under grazing
incidence conditions for the characterization of dopants (local
structure related to electrical activation) in Si wafers for the
production of Ultra Shallow Junctions.
XANES studies of Cu on Si wafer surface have been
performed by Singh et al.
54
at SSRL Beamline 6-2.
Applications
The new applications are annually reviewed in the Atomic
Spectrometry Updates: X-ray fluorescence spectroscopy
55–57
Environmental
An interesting application is the analysis of aerosol samples
collected in a Berner-impactor (12 stages) directly on Si wafers,
Fig. 3 XANES spectra of As
V
and As
III
standard solutions, a nutrient
solution containing As
V
and of xylem sap extracted from a cucumber plant.
This journal is ª The Royal Society of Chemistry 2008 J. Anal. At. Spectrom., 2008, 23, 792–798 | 795
showing the advantage of the very low sample mass required for
SR-TXRF. Experiments at Hasylab, Beamline L showed that 1 h
sampling time was sufficient to get a reasonable signal—this will
lead to detailed studies of air pollution. Fittschen et al.
58
published
a new technique for the deposition of standard solutions by inkjet
printers and its applicability for aerosol analysis. Droplet sizes of
50–200 mm were achieved (which are smaller than the available
synchrotron beam) and the authors reported an absolute
calibration of the number of droplets versus measured Co inten-
sity. These results lead to the conclusion that this technique is very
promising for the quantification of aerosol samples collected by
those impactors that produce a pattern and not a single spot.
Groma et al. also performed measurements of aerosols, sampled
with a May impactor, which collects the aerosols in the shape of
a strip on Si wafers.
59
The quantification was performed using
a Cr strip as external standard. Detection limits of about 1 pg m
3
have been obtained for 20 min sampling time. Both groups drew
attention to the considerable advantage of SR-TXRF for aerosol
analysis in that a short collection time was sufficient to gather
enough sample mass for analysis, to provide a means for
monitoring of rapid changes in aerosol composition.
The SR-TXRF setup at the NSLS SR facility in Brazil was
used for various applications: The determination of various
elements in Brazilian wines,
60
in several mineral waters
commonly available in Brazil
61
and in sea water after salt matrix
removal with Pyrolidine-dithiocarbamate.
62
Moreira et al.
studied the metal absorption in culture corn irrigated with
domestic sewage and found higher metal concentration in the
plant irrigated with sewage than in the plant irrigated with water.
Salvador et al.
63
determined trace elements in various plants to
perform environmental pollution control successfully. De Vives
et al.
64
analysed tree rings from samples of wood collected from
a specific species, Caesalinia peotophoroides, which is common in
Brazil in both urban and country areas. The samples were
digested prior to measurement by TXRF and the authors found
a decrease in the K/Ca, K/P and Pb/Ca ratios towards the bark.
The same group reported the use of fish samples as environ-
mental monitors and discussed the risks to human health by the
ingestion of fish contaminated by metals and other toxic
elements.
65
The ability of Tillandsia (an epiphyte widely used as
an atmospheric monitor) to accumulate heavy metals was
successfully tested by Wannaz et al.
66
to provenance atmospheric
emission sources in Argentina. Espinoza-Quinones et al.
67
compared PIXE with SR-TXRF analysing river water and found
As, Cr, Cu and Zn to be above the limits recommended by
environmental legislation.
Industrial/biochemical
The SR-TXRF setup for wafer analysis at SSRL was used to
determined Fe concentrations on silicon wafer surface after wet
cleaning treatment by SR-TXRF, TOF-SIMS, lifetime and deep
level transient spectroscopy using SR-TXRF as reference.
68
Good correlations were found.
SR-TXRF at the Taiwan National Synchrotron radiation centre
was applied by Chang et al.
69
to study SiO
2
/Ta
2
O
5
multilayer
films on glass substrates.
Several topics have been investigated at the Brazilian Nation
Synchrotron Light source in Campinas: Trace elements in
different pharmaceutical forms of diclofenac sodium have been
determined by Zucchi et al.
70
—differences in trace element
content could be verified. Perspex, Kimfoil and Mylar were
compared as substrates for the analysis of liquid samples by Poli
et al.
71
They found that thin polymer foils produced less back-
ground, but unfortunately they did not compare their substrates
with the commonly used quartz or Silicon reflectors. Novikova
et al.
72
investigated the protective effect of xydiphone on
membrane-bound enzyme damaged by lead ions. They deter-
mined the position of Pb ions within the molecular film by TXRF
before and after the xydiphone treatment and found that
xydiphone effectively eliminated the Pb ions incorporated in the
Ca-ATPase molecules. Al
2
O
3
ceramic powders were analysed by
Peschel et al.
73
using the SR-TXRF facility at Hasylab.
SR-TXRF was used to check the homogeneity of elemental
distribution, which limited the precision of the measurements
Clinical applications
Canellas et al.
74
determined trace and major elements in serum of
patients with chronic myelogenous leukaemia (CML). They
found that the concentrations of Ca, Cr, Cu, Fe, Mn, P, Rb and S
differed significantly between groups of healthy and CML
patients. Serpa et al.
75
studied cognitive impairment related to
changes in elemental concentrations in the brain of old rats.
Higher Br and Cu values were found in certain brain regions of
the cognitively impaired group in comparison with the control
group.
Critical aspects
Though the advantages of SR-TXRF dominate, there are some
aspects which have to be considered critically: first of all there is
a limitation of Synchrotron radiation facilities and only few of
them offer a TXRF set-up. Beamtime has to be paid or
a proposal has to be submitted, this has to be reviewed and
beamtime has to be assigned, if the proposal is approved. So
a detailed project planning in time and tasks has to be made in
advance. A further critical aspect is requirements on the sample
reflectors, they have to be flat and smooth, l/20, with l of visible
light is a good measure for flatness, the roughness should not
exceed 2 nm. Specially polished quartz reflectors or silicon wafers
are best suited. As the technique is very sensitive it will also detect
contaminations of the sample reflectors easily and the prepara-
tion of a clean blank reflector is sometimes tricky and needs the
expertise of chemists. This is especially an issue if the samples are
collected on reflectors in advance and no checking of the clean-
ness with this sensitivity is possible, e.g. aerosol collection. A last
drawback one has to consider is that absolute quantification is
normally not possible and an internal standard is required.
Conclusions
SR-TXRF combines successfully the use of synchrotron
radiation with the intrinsic advantages of TXRF offering low
detection limits (fg) for small amounts of samples if the exciting
synchrotron radiation is selected by means of a multilayer
monochromator. Using crystal or plane grating mono-
chromators also XAS measurements can be performed to obtain
chemical information (oxidation state, compound, bonds, local
796 | J. Anal. At. Spectrom., 2008, 23, 792–798 This journal is ª The Royal Society of Chemistry 2008
