Some important characteristics are described of one- and two-dimensional gas proportional detectors with delay line position readout, which are currently in use at beam-line X12B of the National Synchrotron Light Source. The importance of careful choice of geometric design parameters is emphasized with respect to position linearity and electrical stability; position resolution of 100 µm FWHM and counting rates up to several times 105 s-1 are obtained in calibration studies. Results from some recent beam-line experiments further illustrate the excellent position resolution and linearity, and demonstrate the usefulness of these detector systems for small angle scattering, particularly in dynamic studies.
*This research was supported by the U.S. Department of Energy:
Contract No. DE-AC02-76CH00016
I. INTRODUCTION
Beam-line X12B at Brookhaven's National Synchrotron Light Source has been
designed to carry out time-resolved and static X-ray diffraction studies of
macromolecular systems. The general beam-line arrangement is intended to provide
a widely accessible and user-friendly facility for the acquisition and on-line
analysis of diffraction data from ordered and disordered phases1. For
the last few years experiments have been performed using, first, a
one-dimensional delay line detector2, and more recently, over the
last two years, a two-dimensional delay line detector3.
We describe here some features of the detector systems, and results from some
recent experiments that have been performed at the beam-line. The goal of this
activity is to provide detectors with position resolution in the range 100 to
500 µm, count rate capabilities of several hundred thousand to tens of million
per second, good detection efficiency in the X-ray energy range 3 to 15 keV,
absolute position stability of the order 50 µm and capabilities of dynamic
studies.
II. DETECTORS
Position sensing in gas proportional detectors by delay line is well
understood and utilized2,3,4,5,6,7; for single photon detection,
encoding by delay line provides an optimum combination of excellent position
resolution and high counting rate capability compared with equivalent encoding
techniques using alternate forms of charge division such as a resistive line.
Furthermore, at very high counting rates, the timing information from a cathode
delay line, combined with a prompt anode signal, can be used to reduce pile-up
distortion to negligible levels. Delay lines can be realized in various physical
forms, from those which are self-contained in their own physical enclosure to
those that are part of the cathode design of the detector itself. The approach
used with our detector systems is to fabricate the delay line in its own
enclosure; it is interfaced to the detector cathodes through appropriate
connecting ribbons. In this way the same detector can be used with delay lines
of differing transit times, or detectors of different size can be used with the
same line, offering flexibility with different counting rate applications. Each
delay line, specially designed for these detectors, has 40 nodes and contains
inductors which are wound on soft iron cores. Its enclosure has dimensions 20cm
x2.5cm x2.5cm, which also shields magnetic fields up to 0.1 T.
The basic design of the two-dimensional detector has been described in reference 3.
X-rays enter the front of the detector through a 10 cm x10 cm beryllium window,
whose thickness is 250 µm. Behind the window is a gas volume of depth 6.35 mm;
this comprises, first, an absorption and drift depth of 3.85 mm, then the upper
(wire) cathode, the anode wire plane, and finally the lower (printed circuit)
cathode. The anode-cathode spacing, d, is 1.25 mm, and the anode wire
spacing, s, is 1.1 mm. The wires of the upper cathode (Y-axis) are
parallel to the anode wires, while the strips on the lower cathode (X-axis) are
perpendicular to the anode wires. The actual electrode spacings are determined
in the following manner. In order to obtain a fine position resolution across
the anode wires, it is desirable to make s as small as possible; the
fraction of anode charge induced on the cathodes, at most 0.5 on each cathode,
is a function of the ratio s/d, and begins to drop significantly8
when s/d falls below unity. Thus, as s is made smaller, so
also should d; since, however, the number of cathode nodes cannot exceed
those on the delay line, there is a minimum value of d that can be used
while still achieving good position linearity and resolution with the cathode.
Therefore an acceptable minimum value for d is first determined, then s
should be approximately the same value. The relatively small spacings used in
the present design, d = 1.25 mm and s = 1.1 mm, also require
special attention in the fabrication of the cathodes and anode, in particular to
the edges of the wire chamber proper, where discharge or surface currents can
occur unless appropriate design criteria are adopted. As will be discussed in
section IV, this particular two-dimensional detector has performed in a very
stable manner for over two years.
The position encoding principle within the proportional chamber is based on
interpolating cathode strips9,10,11, in which the center of gravity
of the cathode induced charge from an anode avalanche is found by the weighted
average of charge collected by an array of strips or wires. With a total of 40
nodes on the delay line, and a 10 cm sensing length. the cathode node spacing is
2.5 mm (in practice it is slightly larger since the cathodes should extend
beyond the detector sensitive area). With d = 1.25 mm the "single
strip" per node usually employed in detectors of this type would result in
a highly non-linear readout, because the "footprint" of the cathode
induced charge, a quasi-Lorentzian profile with FWHM approximately 1.5 d,
is smaller than the strip width. Recently we have investigated a variety of
cathode encoding patterns which optimize both position resolution and position
linearity with the small electrode spacings used here. A near optimum cathode
for the two-dimensional detector is the two-intermediate strip cathode3,12;
both cathodes of the two-dimensional detector are constructed in this form. A
photograph of the upper (wire) cathode is shown in figure 1, the 40 read-out
nodes showing clearly on the circuit board at the right hand side.
2D Multi-wire Anode, Y-cathode Frame
2D Multi-wire Proportional Chamber, Delay Lines, Preamps
Figure 1 Photograph of main G10 frame of two-dimensional
proportional detector; the outside dimensions are 15.2cm x15.2cm. This shows the
wires of the upper cathode, epoxy bonded at left and right side of frame, and
connected to their printed circuit board at right hand edge (which then
interfaces to a 40 node delay line); the cathode wires are 50 µm diameter gold
plated tungsten. Anode wires (not attached in this picture) are bonded to the
same frame in same direction as the cathode wires.
On this electrode there are 240 wires (50µm gold plated tungsten), connected
together in pairs; the central pair of every group of three adjacent pairs of
wires forms the nodes to the delay line. The pair of wires at each side of the
node is capacitively coupled to the node (through the natural capacitance, C1,
between each pair of wires), and they are referred to as intermediate groups. A
detail of the wiring arrangement between two nodes is shown in figure 2; the
intermediate groups are held at the correct DC potential through 1MW
chip resistors (located on the opposite side of the printed circuit board and
connected via plated through holes). Typically C1 is just a few pF;
chip capacitors, Cc, several times the value of C1, couple
intermediate groups to improve position linearity12. This type of
cathode not only yields excellent linearity but is well suited for delay line
use because of the very low inter-node capacitance which arises from having four
adjacent groups of wires in series between nodes.
Delay Line - Cathode Wire Coupling Scheme
Figure 2 Schematic arrangement of cathode connection to delay
line, showing natural capacitance C1 between wire groups (or strips);
chip capacitors Cc and 1MW
chip resistors are added to the cathode circuit board.
The lower printed circuit cathode is fabricated also with two intermediate
strips between nodes. In place of the pairs of wires of the upper cathode there
are narrow conducting strips, normally consisting of copper since the board is
manufactured by standard printed circuit techniques. However, these produce
unwanted fluorescence when the detector is used with X-rays of energy greater
than 8 keV, and the fabrication process is taken one step further to replace the
copper strips with aluminum strips.
III. POSITION RESOLUTION AND COUNTING RATE
A. Position Resolution
It is important that the required position information be obtained at a
small anode avalanche size, so that at high counting rates the undesirable
consequences of space charge saturation and anode deposit formation that can
result from large avalanche size are avoided. References 2 and 4 describe in
detail the conditions under which minimum noise is achieved from the
preamplifiers terminating each end of the delay line; position resolution varies
as tw1.5/(Zo
), where tw is a parameter describing the time width of
the signal current from the delay line into the preamplifiers, Zo
is the characteristic impedance of the line, and its transit time. For the set
of one- and two- dimensional detectors used until about one year ago, delay
lines with = 1 µs and Zo = 500 W
were used. In the axis along the anode wires, the X-axis for the two-dimensional
detector, the relationship between position resolution and anode charge is shown
by the circles in figure 3, for 5.4 keV X-rays in Ar/20%CO2. A
minimum resolution of about 100 µm FWHM is achieved at an anode charge of
approximately 0.7 pC; with avalanche sizes larger than this, position resolution
begins to deteriorate, because of spreading of the avalanche along the anode
wire13. The dashed line is the electronic noise contribution. The 100
µm resolution limit is due primarily to the range of photo- and Auger electrons13.
In order to increase the counting rate capability, we have recently installed
0.5 µs delay lines on each axis of the two-dimensional detector; these have a
characteristic impedance of 300 W. The resulting
position resolution vs anode charge is shown by the squares in figure 3. To a
first order the resolution is a factor two worse with the 0.5 µs line, at
a given anode charge, primarily due to the reduction in delay line transit time.
Relationship between Anode Charge and Resolution
Figure 3 Position resolution vs anode charge (measured in 1 s)
for the two-dimensional detector, for delay lines with 1 µs and 0.5 µs transit
times. Detector area is 10 cm x10 cm; position resolution of approximately
1 in 103 is achieved with 1 µs line.
An important feature of the resolution curve for the 0.5 µs line is the fact
that the minimum value achieved, 150 µm FWHM, is larger than for the 1 µs
line. This is due to the combined effects of shorter transit time delay line and
anode avalanche spreading. A larger avalanche size is required to achieve a
given resolution limit due to electronic noise, but the avalanche size at which
resolution deterioration begins is unchanged since this is purely a gas effect.
Thus, the higher counting rate capability is achieved with some compromise to
resolution. We intend in the near future to examine the performance of the
detector with delay lines whose transit times are smaller yet, in the region of
0.2 µs.
Position resolution, and linearity, across the anode wires (Y-axis) of the
two-dimensional detector is more complex to analyze than along the anode wires.
The discrete locations of the anode wires tends to bias recorded positions at or
close to the wire positions; this yields a position modulation in the Y-axis of
the recorded image with a period equal to the anode wire spacing; the amplitude
of the modulation is dependent upon various factors, anode wire spacing and
lateral electron diffusion in the drift region being critical parameters. With
an argon based gas filling, the anode modulation from the two-dimensional
detector is significant, as will be seen in section IV. A compensation matrix,
generated from uniform illumination, is used to smooth the data on a pixel by
pixel basis, essentially removing the modulation, but it is important to
appreciate that this does not enhance the Y-axis position resolution.
B. Counting Rate
Each position axis of the detector generates two timing signals, one from
each end of its corresponding delay line, stop T1, and stop T2.
A prompt signal from the anode, TA, is also generated. A TDC
capable of accurately analyzing these signals, and with an intrinsic
differential non-linearity of only 0.1%, has been designed and fabricated at
this laboratory14. While an X-ray event's position is determined from
(T1-T2), the extra information from TA
allows two important veto capabilities to be built into the TDC, which prevent
mis-recording of position at very high counting rates. These are i) logic
which ensures that a pre-determined minimum time occurs between each anode
signal (the minimum inter-arrival time discriminator) and ii) logic to ensure
that the sum of the time difference between each stop signal and the anode
signal corresponds to the delay line transit time (arithmetic sum
discriminator). The solid curve in figure 4 shows the relationship of TDC
conversion rate vs incident photon rate from one position axis when the detector
is uniformly irradiated with X-rays. A 1 µs delay line was used and both of the
above mentioned discriminators were active; the curve is taken from the TDC
evaluation measurements in reference 14. We have now developed a simple computer
simulation of the TDC operation; this has allowed us to predict the TDC
throughput as a function of delay line transit time for both uniform and
non-uniform irradiation.
Full details of the simulation will be described in a later report; an array of
anode signals (TA) is generated whose time intervals follow an
exponential distribution with a mean interval equal to the inverse of the
incident photon rate. Two additional arrays (T1,T2),
representing the timing of signals at each end of the delay line, are then
calculated with the probability of the event position and the delay line transit
time being taken into account. The simulation includes the mathematical
equivalent of the minimum inter-arrival discriminator and the arithmetic sum
discriminator, and includes the effect of the approximately 70 ns dead-time of
the TDC for one-axis. The signals, TA, can correspond to
photons with uniform spatial distribution across the entire detector length, or
to distributions with localized diffraction maxima.
The open curve (a) in figure 4 illustrates the TDC conversion rate vs incident
photon rate that is predicted by this simulation for uniform irradiation of
one-axis with a 1 µs line; it follows very closely the experimental
measurement, a broad maximum throughput of just over 250 kHz occurring for an
incident photon rate of nearly 1 MHz. The open curve (b) in figure 4 illustrates
the equivalent prediction for one-axis with a 0.5 µs line; as expected the
maximum throughput and the incident flux at which it occurs, are higher by about
a factor two compared with the 1 s line. We have also used the simulation to
determine the system throughput vs incident photon rate for non-uniform
irradiation. In general TDC conversion rate vs incident photon rate follows the
trend indicated in figure 4, with slightly higher maximum conversion rates
achieved when events are clustered around the center of the detector, and
conversely a lower conversion rate when events are clustered near the edges of
the detector. For two peaks whose intensity ratio is 103:1, the intensity ratio
of encoded events remains within a few percent of its correct value even when
the TDC conversion rate is at a maximum.
TDC Efficiency vs Incident Rate
Figure 4 TDC conversion rate vs incident photon rate for
uniform irradiation. Solid curve (from reference 11) is experimental data for 1
µs delay line, open curves are results from computer simulation for 1 µs and
0.5 µs delay lines.
IV. BEAM-LINE PERFORMANCE
The detectors are operated with either Ar/20%CO2 or Xe/10%CO2;
generally the X-ray energy range 3 to 9 keV can be covered with good resolution
and reasonable efficiency by the argon mixture, and up to 15 keV can be covered
by the xenon mixture. A two-dimensional detector system has been in almost daily
operation at the X12B beam-line for over two years, regularly experiencing
incident photon rates of 105 to 106 s-1; during
this time it has been removed for major servicing just once, to reconnect one
anode wire which became isolated from the high voltage. It has otherwise
functioned in a very stable and reliable mode; this successful operation results
from the careful construction, correct choice of detector geometric parameters,
use of low-outgassing materials in construction and appropriate choice of very
high purity gas mixture. Background counts are due to only cosmic rays and local
radioactive emissions, corresponding to about 5 s-1 for the 10 cm x10
cm area, and therefore have negligible contribution to diffraction images.
A resolution and global linearity demonstration is shown in figure 5, which is a
log-transformed image of a data frame acquired from a calibration sample
consisting of silver behenate; it is taken with 11.6 keV X-rays and a gas
filling of Xe/10%CO2. Anode wire modulation in the Y-axis is clearly
evident in the raw data of the left-hand half of the image. The data in the
right-hand half have been compensated by the pixel-based analysis process
referred to in the last section. The Debye-Scherrer rings resulting from the
lamellar spacing (58.39 Å) of silver behenate occur near the center of the
frame, while diffraction features from the crystalline lattice occur near the
periphery.
Partially Compensated Data Frame
Figure 5 Debye-Scherrer rings from sample of silver behenate,
illustrating excellent global linearity of the delay line readout. The left-hand
half of the image is raw data, and shows modulation caused by the discrete
positions of anode wires (whose pitch is 1.1 mm). In the right hand half of the
image, the raw data has been smoothed by analysis with a compensation matrix.
An example of dynamical data acquisition is shown in figure 6. This
illustrates measurements of the apparent rate constants of thermal solid-phase
transitions in a powder of the triglyceride trilaurin, using 9 keV X-rays and Ar/20%CO2.
A crystalline powder of trilaurin was melted in a temperature-jump device and
flash-cooled, forming the alpha phase of the system. Diffraction data was then
acquired over 500 ms increments while the sample was subjected to a rapid
thermal ramp (80 C/min), bracketing two crystalline phase transitions: alpha
beta', then beta' beta. The resultant data frames were compensated and radially
averaged to generate intensity vs. Q profiles. Figure 6 presents a time series
of radial profiles from a single pass of the thermal cycle. Diffracted intensity
is encoded as a gray scale (white black representing increasing intensity, using
a linear scale), with the horizontal axis representing scattering angle; the
alpha phase, indicated by the single intense peak occurring during the first 20
s, is centered on Q=1.4 Å-1.
Thermal Phase transitions in a Triglycerid Powder
Figure 6 Example of kinetics of solid-solid phase transitions
in triglycerides. Each vertical time bin is 500 ms.
Figure 7 is an example of crystallographic data acquired with a small crystal
of whale myoglobin, using 11.6 keV and Xe/10%CO2. Figure 7(a) and (b) are two
oscillation (1.5º) data frames scaled to maximize the apparent contrast of weak
reflections. Although the strong peaks appear saturated in the image, the actual
data from the detector is histogrammed into a 32 bit deep frame store. Thus,
there is no saturation in the raw data, and the dynamic range available in
enormous. Furthermore, there are no serial readout delays between frames. Figure
7(c) and (d) are the log-transforms of (a) and (b).
Rotation Diffraction Patterns from a Myoglobin Crystal
Figure 7 Crystallographic images from a small crystal of whale
myoglobin. (a) and (b) are two frames 1.5 apart with intensity on a linear scale
(but scaled to emphasize weak reflections), while (c) and (d) are the same two
frames with a logarithmic intensity scale.
Diffraction Pattern of Frog Psoas Muscle
The above figure is a false-color image of a wide-angle exposure
(30 sec, sdd: 50 cm) of a small frog psoas muscle fiber bundle. The image was
compensated for detector non-uniformity and log transformed to emphasis weak
features of the pattern. Note that the borders of the image were not corrected
for detector nonuniformity (ie. not compensated), demonstrating the effect of
anode response modulation upon the raw data.
V. CONCLUDING REMARKS
We have described the performance of a relatively new gas proportional
chamber system which has been shown to provide performance ideally suited for
the requirements of the diffraction studies at beam-line X12B of the NSLS. The
delay line detectors in routine daily use are extremely stable and reliable. The
beam-line now has a suite of detectors which includes not only the one- and
two-dimensional delay line detectors described here, but also a very high rate
one-dimensional wire chamber with a fast low noise shaping amplifier on each of
the 100 anode wires (106 photons s-1 per wire). This is
particularly well suited to time resolved studies of certain biological samples
and will be brought into operation in the near future. Two-dimensional detectors
based on cathode pad readout from many adjacent rows of pads are presently under
design to increase counting rate capability for area detectors to over 108
photons s-1.
ACKNOWLEDGEMENTS
We are grateful to Veljko Radeka and Joachim Fischer for their invaluable
advice and continued support in this work. The skill of Gene Von Achen in the
assembly of most of the detector systems is greatly appreciated. The assistance
of Lee Rogers and Joe Harder in preparing the electronics is acknowledged.
1 M. Capel, Synch. Rad. News 6 23 (1993)
2 R.A. Boie, J. Fischer, Y. Inagaki, F.C. Merritt, V. Radeka, L.C.
Rogers and D.M. Xi, Nucl. Instrum. & Meth. 201 93 (1982)
3 G.C. Smith, B. Yu, J. Fischer, V. Radeka and J. A. Harder, Nucl.
Instrum. & Meth. A323 78 (1992)
4 V. Radeka, IEEE trans. Nucl. Sci. NS-21 51 (1974)
5 V. Perez-Mendez, M. Greenstein and D. Ortendahl, IEEE Trans. nucl.
Sci. NS-24 209 (1977)
6 A. Gabriel and M.H.J. Koch, Nucl. Instrum. & Meth. A313
549 (1992)
7 R.A. Lewis, W. Helsby, A. Jones, F. d'Annunzio, C. Hall, B. Parker,
I. Sumner and J. Worgan, Proceedings of the European Workshop on X-ray Detectors
for Synchrotron Radiation (ed. A.H.Walenta, Univ. of Siegen) 257 (1992)
8 E. Gatti, A. Longoni, H. Okuno and P. Semenza, Nucl. Instrum. &
Meth. 163 83 (1979)
9 I. Endo, T. Kawamoto, Y. Mizuno, T. Ohsugi, T. Taniguchi and T.
Takeshita, Nucl. Instrum. & Meth. 188 51 (1981)
10 F. Piuz, R. Rosen and J. Timmermans, Nucl. Instrum. & Meth. 196
451 (1982)
11 J.S. Gordon and E. Mathieson, Nucl. Instrum. & Meth. 206
267 (1984)
12 E. Mathieson and G.C. Smith, IEEE Trans. Nucl. Sci. 36 305
(1989)
13 J. Fischer, V. Radeka and G.C. Smith, Nucl. Instrum. & Meth. A252
239 (1986)
14 J.A. Harder, Nucl. Instrum. & Meth. A265 500 (1988)