spectroscopy
Universit
29
The driving force for these wide-area devices is monitored in real time. Detector advances also of-
fer considerable improvement in the speed of data
collection for related techniques such as angle-re-
solved PES and X-ray standing wave spectroscopy.
Few of the detector systems under development
reserved.
*
Corresponding author. Tel.: +44 1970 622803; fax: +44
1970 622826.
E-mail address: a.evans@aber.ac.uk (D.A. Evans).
Nuclear Instruments and Methods in Physics
0168-583X/$ - see front matter C211 2005 Elsevier B.V. All rights
1. Introduction
Often the main limitation in spectroscopic tech-
niques is the e?cient detection of radiation (both
electromagnetic and particulate) and improve-
ments in detection technology lead inevitably to
significant advances in scientific knowledge. One
example is the transformation of optical spectros-
copy by the appearance of CCD detectors.
the enormous increase in collection e?ciency.
In electron spectroscopy, detection is still largely
dominated by discrete channeltron systems,
although there is a considerable e?ort worldwide
to develop the electron equivalent of CCD detec-
tors. The potential applications for such detectors
are many ? in photoelectron spectroscopy (PES)
for example, such detection systems enable pro-
cesses such as catalysis and thin film growth to be
Abstract
In electron spectroscopy, multi-channel detection combined with intense radiation sources provides the optimum
experimental configuration. Building on the 5 mm, 192-channel ion detector developed at Aberystwyth, longer arrays
have been fabricated for the detection of electrons in a commercial hemispherical analyser. The performance and reli-
ability of a 10 mm, 384-detector array is discussed and the first array-detected photoelectron spectroscopy data for sin-
gle-crystal diamond are presented. In scanning mode, the detector shows a large improvement compared to single
channeltron detection and this improvement allows data to be collected in snapshot mode (1 s per spectrum) to enable
real-time measurements.
C211 2005 Elsevier B.V. All rights reserved.
Keywords: Real time XPS; Electron detector; Multianode; Array detection
A fully integrated multi-channel
for electron
D.P. Langsta?, A. Bushell,
Institute of Mathematical and Physical Sciences,
Available online
doi:10.1016/j.nimb.2005.06.187
detector
T. Chase, D.A. Evans
*
y of Wales, Aberystwyth, SY23 3BZ, UK
August 2005
Research B 238 (2005) 219?223
www.elsevier.com/locate/nimb
havehoweverbeenabletodemonstratetherequired
electricalstabilityneededincontinuousoperationin
commercialsystems.In2001,aprojectwasinitiated
to develop a new application in electron detection
for the 5 mm, 192-channel ion detector originally
developed at Aberystwyth over many years [1,2].
This has resulted in the design and fabrication of a
10 mm, 384-detector array electron detector and a
more advanced 20 mm, 768-detector array.
Other designs of multi-channel detector have
been fabricated using discrete devices for amplifi-
cation/discrimination and counting pulses [3,4],
but this design is unique in that all the front end
electronics are integrated on a single chip, together
prises a custom silicon integrated circuit mounted
220 D.P. Langsta? et al. / Nucl. Instr. and Meth.
on a ceramic substrate in proximity to the rear face
of a multi-channel plate (MCP). The detector
operation is illustrated schematically in Fig. 1.
An electron incident on the MCP is amplified
and the electron pulse is detected by the collector
anodes of the array.
Fig. 1. The principle of operation of the array detector is based
on the direct detection of electrons by an array of collection
anodes placed beneath a multi-channel plate electron multiplier.
Each anode is in direct contact with its pre-amplifier and
with the collection anodes. This level of integration
results in greatly simplified assembly onto the cera-
mic module, as only 25 bond wires need to be con-
nected to the circuit for power, data and control
signals to and from the external controller via
the vacuum feedthrough.
2. Device specifications and performance
The UWA charge sensitive detector array com-
discriminator circuitry.
The detector chip contains a number of chan-
nels, or pixels, on a pitch of 25 lm. The results
presented here are taken from a device containing
384 channels with a total active area of 9.6 mm ?
2 mm. A more advanced device containing 768
channels with a total active area of 19.2 mm ?
3 mm is currently being characterised. Each chan-
nel has a metal anode to collect the electrons as
they emerge from the MCP, a charge sensitive
amplifier to produce a digital signal in response
to the electron pulse, a 16-bit counter to accumu-
late the counts as they arrive and circuitry to read
out the data sequentially from all channels in the
array.
The detector chip is operated at around 2 kV
relative to the analyser, and in order to provide
this level of isolation, communication with the
array is achieved via a floating controller. This
controller, developed in collaboration with the
CLRC Daresbury laboratory, provides timing,
control and readout of data from the detector
head as well as control of the MCP power supply.
For application in electron spectroscopy, the
detector device is bonded to a ceramic substrate
that is mounted as a modular unit in a hemispher-
ical electron analyser (VG CLAM4) such that the
array detector lies at the focal plane. All materials
used are ultrahigh vacuum (UHV) compatible
such that, following incorporation of the detector
module, the spectroscopy chamber achieved its
routine base pressure of below 10
C010
mbar.
Preliminary testing of the array detector
included measurement of the point spread function
for the detector (using a technique of speckle imag-
ing [5,6]), the determination of the noise floor,
measurement of the array uniformity and determi-
nation of readout speed and maximum count rate.
To determine the point spread function, the array
was uniformly illuminated with electrons and a
short integration time chosen such that no counter
received more than one event per integration peri-
od. The mean size of the point spread was found to
be 5.5 channels, corresponding to a spatial resolu-
tion of 137 lm. This compares favourably with the
figure of 150 lm obtained for the LBNL detector
[7] and 470 lm for the Trieste device [4].
The noise floor of the detector was determined
in Phys. Res. B 238 (2005) 219?223
by operating the array for an extended period in
mode required a total collection time of 200 s, and
Meth.
the absence of any ionising radiation or charged
particles. The noise value of 1.02 ? 10
C04
events
s
C01
pixel
C01
is solely due to the MCP in the system,
and is in good agreement with the figure quoted by
the manufacturers [8]. This noise floor is well
below the noise inherent in the XPS process.
The uniformity of response was determined by
recording a large number of spectra at a series of
electron energies such that every pixel on the array
was exposed to the whole spectrum and thus all
received equal numbers of electrons. The unifor-
mity of the detector, given by the standard devia-
tion of the response divided by the mean was
found to be 7.9%. Although this is higher than
the figure of 4% quoted for the LBNL detector
[7] , most of this non-uniformity can be removed
by correction techniques [9].
The readout speed of the detector is currently
limited by the cabling between the vacuum feed-
through and the control electronics. At present
the readout speed is set to 2 lS/pixel, giving an
overall readout period of 768 lS across the entire
384 pixel array. It is hoped to improve upon this
by mounting the control electronics directly on
the vacuum flange rather than at the end of
1.8 m of multiway cable as at present. The LBNL
detector uses double bu?ered counters, so the
readout time is e?ectively zero [7], whereas the
Trieste device has a quoted speed of 0.2 mS to read
out 96 detectors [4].
At present it has not been possible to directly
determine the maximum count rate as the labora-
tory X-ray source used does not generate a su?-
cient photo-current to saturate the device, even
at the low-energy secondary electron tail. Earlier
experiments with a similar MCP based device
[10] yielded a maximum count rate in excess of
10
4
counts per second per pixel, leading to an over-
all count rate of approximately 4 MHz across the
array. The exact figure for the 384- and 768-chan-
nel devices will be determined in a series of exper-
iments planned for the synchrotron radiation
source. It should be noted that the count rate lim-
itation is due to the MCP characteristics and that
higher figures could be obtained by using C212hotC213
MCPs, which may themselves bring other require-
ments such as the need for cooling to be built into
D.P. Langsta? et al. / Nucl. Instr. and
the detector head [4].
so it is clear that the array provides a significant
increase in total counts. It should be noted that
the sample preparation was not identical in both
cases, resulting in di?erent intrinsic peak widths
and so no comparison can be made regarding the
energy resolution di?erence between the two modes.
Similarly, it has not been possible to directly
determine the linearity of the present device using
the laboratory X-ray source, however the previous
device [10] exhibited good linearity over at least 6
orders of magnitude.
3. Application in electron spectroscopy
The UWA detector, incorporated in a VG
CLAM4 electron analyser, has been used to detect
photoelectrons excited by a range of radiation
sources. Most of the initial experiments have been
performed using a Mg Ka flood X-ray source,
using several metal and semiconductor test sam-
ples. The 384-channel array has been continuously
operational for over four months and this confirms
that it is physically and electrically robust.
The detector can operate in two modes: energy
scanning and energy fixed. The first mode takes
advantage of the increased collection area across
the focal plane and is used for acquiring high-res-
olution data over a wide energy range. In the sec-
ond mode of operation, the analyser energy is fixed
and the spectral information is obtained by the
intensity image across the 384 pixels. The two
modes are compared with a single channel detector
under similar experimental conditions in Fig. 2.A
polished (111) diamond surface has been selected
as the test sample since this provides a narrow C
1 s emission peak.
The data obtained with a single channeltron in
scanning mode (Fig. 2(a)) were collected in 250 s
by changing the retard potential of the electron
analyser in increments of 0.1 V. The data obtained
with the array in scanning mode (Fig. 2(b)) were
obtained using the same analyser parameters (pass
energy = 20 eV; entrance slit = 1 mm), and the
spectrum shown is the result of weaving together
400 array images. The data acquired in this woven
in Phys. Res. B 238 (2005) 219?223 221
However, the value obtained for the full-width half
detectors is the non-uniformity that is always pres-
Meth.222 D.P. Langsta? et al. / Nucl. Instr. and
maximum (FWHM) of the woven data is compa-
rable to values reported for diamond surfaces
under similar experimental conditions [11].
To enable real-time measurements, data must
be recorded with a collection time smaller than
the timescale of the process under investigation.
This is possible with the UWA detector array in
snapshot mode as shown for a diamond (111) sur-
face in Fig. 2(c). This spectrum was recorded with
data collection time of 1 s. In order to image the
C1 s peak across the 384-channel array, a pass
energy of 100 eV was required. This increases the
count rate compared to spectra (Fig. 2(a) and
(b)), and also introduces an asymmetry in the peak
has been successfully constructed and incorporated
Fig. 2. A comparison of the performance of a (a) channeltron
detector with a 384-channel array detector in (b) weaving
and (c) snapshot modes. The data were recorded with a Mg Ka
X-ray source for a diamond (111) surface. Spectra (a) and (b)
were recorded with the same analyser entrance slit (1 mm) and
pass energy (20 eV) and the data collection time is similar.
Spectrum (c) is corrected for array non-uniformity and data
were collected in 1 s.
into a hemispherical electron energy analyser for
photoelectron spectroscopy. Test results confirm
that the detector is reliable and robust and is able
toout-performexisting single-channeldetectorsys-
tems. By operating in snapshot mode, it has been
shown that data can be recorded at the 1 s time-
scale, enabling real-time experiments to be per-
formed. Further advances are likely by using
high-flux synchrotron radiation sources where the
UWA detector is particularly well suited since the
small source size means that there is no redundant
detectorarea.UsingfloodX-raysources,thedetec-
tor operates at less than 1% of its maximum count
rateandthisislimitedbycurrent MCPtechnology.
The detector is therefore ideal for use with the lat-
est high-brightness X-ray sources.
Acknowledgements
This work was funded through the EPSRC
ent due to manufacturing tolerances and can have
contributions from the MCP, the integrated array,
or the mounting of the MCP on the array. This has
been a key problem since the advent of the MCP in
the late 1960s, but this has been largely resolved
using matrix deconvolution methods [12]. Where
the spectral features are slow-varying, it is also
possible to correct for the non-uniformity using a
spectral region that is featureless. The data of
Fig. 2(c) have been corrected in this way, and it
is clear that the data quality of the 1 s snapshot
is su?cient for this mode to be used in real-time
experiments. These studies are currently underway
and further improvements are planned using more
intense synchrotron radiation sources and a larger
768-channel array.
4. Conclusions
A fully integrated 384-channel electron detector
that is due to analyser aberration at this pass en-
ergy (the detector response is una?ected).
One issue that must be addressed for array
in Phys. Res. B 238 (2005) 219?223
Real-time Electron Energy Spectroscopy (REES)
programme (GR/R08575, GR/S74126). Jon
Headspith, Richard Farrow and Dave Teehan at
the CLRC Daresbury laboratory are thanked for
providing electronics and detector flange design,
and Keith Birkinshaw and Stephen Evans are
thanked for many invaluable discussions.
References
[1] K. Birkinshaw, Int. Rev. Mass Spectrom. 15 (1996) 13.
[2] D.P. Langsta?, M.W. Lawton, T.M. McGinnity, D.M.
Forbes, K. Birkinshaw, Meas. Sci. Technol. 5 (1994) 389.
[3] A. Nambu, J.-M. Bussat, M. West, B.C. Sell, M. Watan-
abe, A.W. Kay, N. Mannella, B.A. Ludewigt, M. Press, B.
Turko, G. Meddeler, G. Zizka, H. Spieler, H. van der
Lippe, P. Denes, T. Ohta, Z. Hussain, C.S. Fadley, J.
Electron Spectrosc. Rel. Phenom. 137?140 (2004) 691.
[4] L. Gori, R. Tommasini, G. Cautero, M. Barnaba, A.
Accardo, S. Carrato, G. Paolucci, Nucl. Instr. and Meth. A
431 (1999) 338.
[5] M. Sinha, D. Langsta?, D. Narayan, K. Birkinshaw, Int. J.
Mass Spectrom. Ion Phys. 176 (1998) 99.
[6] D. Narayan, D. Langsta?, M. Sinha, K. Birkinshaw, Int. J.
Mass Spectrom. Ion Phys. 176 (1998) 161.
[7] J.M. Bussat, C.S. Fadley, B.A. Ludewigt, G.J. Meddeler,
A. Nambu, M. Press, H. Spieler, B. Turko, M. West, G.J.
Zizka, IEEE Trans. Nucl. Sci. 51 (2004) 2341.
[8] Hamamatsu Photonics K.K., MCP Assembly, Shizuoka-
ken, Japan, Hamamatsu Photonics K.K. Electron Tube
Centre, 1991.
[9] K. Birkinshaw, Int. Rev. Mass Spectrom. 215 (2002)
195.
[10] D.P. Langsta?, Int. J. Mass Spectrom. Ion Phys. 215
(2002) 1.
[11] S. Evans, The Properties of Natural and Synthetic
Diamond, London, Academic Press, 1992.
[12] K. Birkinshaw, Int. Rev. Mass Spectrom. 181 (1998) 159.
D.P. Langsta? et al. / Nucl. Instr. and Meth. in Phys. Res. B 238 (2005) 219?223 223