Proposed Profile Monitor Designs for the
Advanced Hydrodynamic Facility (AHF)1
William. C. Sellyey, James F. OHara
Los Alamos National Laboratory, MS H81 7, Los Alamos, NM
Abstract. The AHF consists of a LINAC, a booster ring, a 50 GeV Synchrotron and an elaborate
set of beam-lines for simultaneously delivering a string of 24 proton bunches from 12 different
directions to a firing site (fs) chamber in which an explosion is in progress. This paper will
discuss profile instrumentation being considered for the fs beam-lines and the Synchrotron. In
the beam-lines most profiling devices will probably be harps but some fluorescent screen/camera
systems may be used. For the Synchrotron RGIPM's may be used for observing individual
bunches. The MCP usually placed in the vacuum for such devices might be replaced by a
scintillator viewed by a lens plus multianode PMT. In order to observe individual bunches, it
may be necessary to increase the local vacuum pressure using a gas jet, molecular beam or a gas
puff. Another option may be to view gas fluorescence with the same optical arrangement as used
for the RGIPM. A carbon wire moving with velocities of 1 to 5 m/s is being considered as an
intercepting device to observe stored beams consisting of one or two bunches. A quadrupole
moment measuring system for determining transverse emittance is being investigated.
Harps in the firing site beamlines
Harps can be used anywhere in the beam transport to obtain projections of
beam profiles in the direction perpendicular to the secondary emission (SE) wires. In
this section a harp based profile measurement for observing profile projections of
beam pulses separated by 200 ns will be described.
All beam bunches in all (six) dimensions will be assumed Gaussian. One way
to process the signal from a wire is to send it through a Gaussian filter with rms time
width <JF=50 ns. This is a factor of 5 times longer than the bunch length in the beam
lines and thus <JF dominates. It can be shown that the fractional error in the rms spatial
beam width is
s
.
m172 eQoAxR
AV is the rms amplifier noise, m is the number of wires per mm, 8 is the secondary
emission coefficient, Q0 is the charge in one bunch, Ax is the width of one wire and R
is the signal cable impedance. To get a specific error estimate let 8=0.02, m=3/mm,
Ax=0.1 mm, R=50 Q, Q0=10n particles = 1.6*10~8 C. Take s=l mm, a typical beam
size in much of the transport. AV is obtained from a stretcher amplifier designed and
constructed for amplifying stripline signals [1]. This amplifier has a gain of 31.5 and a
1
This work was sponsored by the US Dept. of Energy under contract W7405-ENG-36.
CP648, Beam Instrumentation Workshop 2002: Tenth Workshop, edited by G. A. Smith and T. Russo
© 2002 American Institute of Physics 0-7354-0103-9/02/$19.00
275
36 MHz rms width. The filter considered here has a 20 MHz rms width, and thus will
pass less noise than the 30 MHz amplifier. The noise referred to the input of the 36
MHz unit was 25 jiV and this is used here as an overestimate of the noise. Inserting
these numbers into the above formula gives As/s=.006 and this is clearly an adequate
result. Temperature estimates for a 100 jim carbon wire indicates that it will never go
above 465 °C anywhere in the beam lines.
Viewscreens in the firing site beam lines
Many types of imaging systems have been designed for the beam transport
lines from the Synchrotron to the firing sights. The screen materials considered
include Cromox-6 (Al2O3(Cr)), LSO and shiny materials for OTR screens. Cameras
investigated include the non-rad hard CCD cameras like the Cohu 2622 and rad-hard
cameras manufactured by CIDTEC. Lenses can be off the shelf compound devices,
large diameter compound achromats or multiple fused silica. Only one system will be
described here.
This system would be used in circumstances where the beam intensity is high
at 3*1013 particles. It will use an OTR viewscreen that may be made out of
Molybdenum coated with a shiny material like Ag or Al. If the surface is smooth, the
peak backward OTR intensity will be generated in a cone with an opening angle of
0.02 radians. This is an inconvenient situation because it can lead to position
dependent imaging on the camera faceplate. However it has been shown that if the
surface is rough, the OTR light does not go into a cone [2]. It is assumed here that it
can be emitted into 2n (this needs to be demonstrated). The light intensity can be
calculated [3]. The result is that there are about 2 photons per 100 nm emitted at the
OTR surface for each 1000 protons passing through the surface. A 150 mm diameter
achromatic doublet with a i m focal length could be used as the imaging lens. The lens
would be placed l l m from the beamline, and the camera would be 12.1 m from the
beam line. This results in a magnification of 0.1. The camera is a Cohu 2622 CCD
camera. If 100 nm of light is imaged, than 3*1013 protons will produce Nph=6*1010.
The number of photons reaching the camera is N=1.39*106. It can be shown that the
fractional error in the rms width of a Gaussian shaped beam is
=
7i1/2Z
X ~ (AxAy J
8N
/2
N
Ax and Ay are the camera pixel sizes SN is the photon number equivalent of the
camera (read) noise. SN =330 at 700 nm, Ax=8.4 jim, Ay=9.8 jim and in the beamlines
typically rms beam sizes are 1 mm. The error calculation refers to the camera faceplate
and thus E=0.1 mm. This results in AE/Z=0.05. By using a larger light bandwidth,
losses in the lens system can be compensated.
The temperature rise in almost any material used for an OTR screen may be
significant but not serious because conduction carries the heat away from the beam
impact area during the 25 seconds between beam pulses.
Replacing the CCD camera by a gated, intensified camera will enable the
measurement of individual transverse bunch shapes for normal intensity and low
intensity bunches.
276
Gas Fluorescence in the Synchrotron
In a typical gas fluorescence imaging system, protons (or other charged
particles) impart energy to a residual gas molecule and some of the molecules will
emit this energy as photons. Lenses are used to produce an image of the gas
fluorescence. If the gas fluorescence has a sufficiently short decay constant, the
fluorescent image is expected to be an accurate representation of the transverse beam
profile. Gas fluorescence of the 391.4 nm NZ+ line [4] has been used extensively to
observe transverse beam profiles [5] at low beam energies. Recently it has been shown
that this can also be done at 10's of GeV [6] and therefore it will be considered here.
The approximate cross section at 50 GeV for producing a 391A nm photon is found to
beojt=3.3*l(r20cm2.
Two types of systems will be describe here, both of which will be designed to
operate in a high radiation environment, but one will depend on shielding to protect
conventional cameras and optics. The other will be made of radiation hard
components. This second system will consist of a fused silica view port to let out the
beam produced fluorescence light, a radiation hard lens system and a radiation hard,
position sensitive, high gain photon detecting device. One possibility for the photon
detector is the Hamamatsu R5900U-07-L16 with a fused silica window. It has 16 strip
photocathodes and corresponding anodes. Each strip is 0.8 mm by 16 mm with 0.2
mm space between strips. Typical electron gain is 106 for each channel. There is at
least one manufacturer who makes a microchannel plate based position sensitive
photon detector (Photec Inc.) that might be usable here. They are also capable of
manufacturing devices to our specifications. In the shielded system, a gated intensified
camera with a lens will be placed in a shielded environment some distance from the
beam.
A formula for calculating the statistical error of a measured rms width of a
Gaussian charge distribution was derived. It assumes that there is no background
signal or other noise sources other than what results from counting statistics and thus
gives the lowest error that can be achieved. The fractional rms error is given by
- - >1/2
1+
AG
(3)
Here Nb is the number of photons produced my the beam in the length of the beam that
is imaged, g is the fraction of the photons collected by the lens system and converted
to photoelectrons and G is the electron multiplication factor of the photon detector.
The rms fractional fluctuation AG/G is typically of the order of 0.5 and thus the
coefficient of 1/g is taken as 1.25. A "low" aberration fused silica lens system will be
described below. To calculate signal levels, we only need to know that the aperture of
the system is 45 mm and that the lens focal length is 500 mm. An inverting, unity
magnification system is used. To be specific, it is assumed that an R5900U-07-L16 is
used. The fraction of light from the beam entering the aperture is 1.3*10~4, the product
of quantum efficiency (0.2) and fractional strip width (0.8 ) is 0.16 which results in
g=2.03*10~5. Equation 3 can be solved for Nb and if a 5 % measurement of the rms
width is to be made, Nb =1.8*107 photons will need to be generated in 1.6 cm length
277
of the beam. The number of photons produced in L=1.6 cm of beam per turn is Nbo=
CT^PN NsL=5.61*103 where pN=3.54*109/cm3 is the density of gas molecules at 10~7
Torr. The number of turns for a 5 % measurements will be Nb/ Nbo= 3.2* 103 and this
will take 16.6 ms. Shorter measurement times could be obtained if the local gas
pressure can be increased. To observe individual bunches, the local gas pressure
would need to be increased to 10~2 Torr and it is unlikely that this can be done without
adversely affecting the circulating beam.
The measurement error in the rms beam width is also affected by the 1 mm width of
the detector photocathodes as well as optical aberrations. The rms width of a 1 mm
strip is WmiS=282 jim. It is straightforward to show that for a circular lens of radius r
the rms width of the image of a line source because of chromatic aberration is given
by
1 I An
I is the image distance, f is the focal length, n is the average index of refraction and
Anrms is the rms deviation of the index of refraction. The FWHM line width is
estimated as less than 7 nm [4]. Assuming a Gaussian distribution and using a table of
refractive index for fused silica [7] AnnnS=3.57*10~4 mm, n=1.471. Using I/f=2, and
r=22.5 mm gives Orms=17 |im.
To minimize aberrations, two planoconvex lenses will be use with the convex
faces nearly touching. Ray trace equations were derived for this lense arrangement and
used to write ray tracing software. The rms image widths of a line source were
calculated for various transverse and longitudinal object positions for a 50 mm lens
aperture and a focal length of 1 m for each lens. The image position was kept at 1 m.
The result is that a 50 mm wide detector would have a 50 mm field of view and a 52
mm depth of field while keeping the line image rms spread below H=0.33 mm. This
depth and width of field of view should be adequate for imaging both the horizontal
and vertical beams. If both horizontal and vertical profiles are imaged simultaneously
and at the same place, the position information from one can be used to correct for
spot smearing in the other.
The total rms blurring caused by the three effects discussed above is
[Wrms2+ocrms2+H2]1/2=434 jim. The rms width of a typical beam will be observed as
[4342+19002]=1949 |im and this 49 |im overestimate is less than the 5 % statistics
error (1900*. 05=95 jim). The most important contribution to the broadening comes
from the depth of field and strip width, and both of these can be corrected for.
The second type of system depends on shielding to protect a lens and camera
from radiation. To be specific, a Xibion ISG-750 intensified camera could be used as
the imaging device, combined with a single 150 mm diameter achromatic doublet with
a i m focal length (Melles Griot 01 LAO 367). The lens would be placed l l m from
the beamline, and the camera would be 12.1 m from the beam line. This results in a
magnification of 0.1. The 12.1 X 9.2 mm2 intensifier active area will translate into a
121 X 92 mm2 field of view at the beamline. Equation 3 can again be used to
calculate how long it will take to record enough data to make a 5 % measurement in
the rms width of a beam. With a 19 % quantum efficiency and an aperture of 135 mm,
g=1.79*10'6. This gives Nb=2.1*108 and with L=12.1 cm, Nbo=a?tpN NSL=4.2*104 at
278
10~7 Torr. Thus it will take 5000 turns or 26 ms to make a 5 % measurement of the rms
width. The ISG-750 output signal is 30 frames per second interline transfer. Thus it is
limited to making at most 30 measurements per second. It is a gateable camera, so the
signal from a single bunch could be integrated over many turns, especially if the local
gas pressure is increased.
The rms resolution of the ISG-250 is about 16 microns. Aberration and field of
view rms image smearing is less than 1 jim over an 80 mm field of view and an 80
mm depth of field. When added in quadrature to a 1.9 mm beam rms width they have a
less than 1 micron effect on the width. The lens focal length could be reduced by a
factor of three (say by adding lenses after the primary lens) and this would decrease
the observation time to get a 5 % rms by about a factor of 9 if a sufficiently large
viewport is available.
In the above discussion, optical losses have been ignored although they may be
as large as 50 %. Thus, measurement times may be underestimated by a factor of two.
Residual Gas lonization profile monitor (RGIPM) in the Synchrotron
For an RGIPM system approximately uniform and parallel electric and
magnetic fields will need to be established in the region of the measurement. Many
authors have described this type of system and the details of producing the fields will
not be discussed here. The electron detection system will be designed so that a 5%
measurement can be made in the rms width of a single bunch for each bunch on every
turn in the Synchrotron. Additionally, the only detector component in the vacuum will
be an LSO scintillator and this will be made as thin as possible to reduce unwanted
background signals. The beam transverse distribution will be reproduced on the
surface of the LSO by scintillation cased by 5 keV electrons accelerated and guided to
the LSO by the E and B fields. This distribution will be viewed by an imaging system
outside the vacuum.
Most RGIPM systems presently in operation use microchannel plates (MCP) in
the vacuum chamber as a signal amplifier. A problem with this is that the MCP's are
limited in total charge they can produce per unit area. In time this leads to uneven gain
over the MCP surface because, on average, transverse beam charge distributions are
not uniform. The LSO scintillator will be used here in an attempt to circumvent this
problem.
The light attenuation of LSO caused by Co60 y's has been measured [8] and is
quoted as 7% /(cm*108Rad) at the LSO emission wavelength. Assuming an average
beam size of 1.9 mm rms, and that 7000 electrons are produced per cm which are then
accelerated through a 5 kV potential (see below) it will take about ten days of
continuous operation for 8 hr/day to reduce the LSO light transmission by 14 %. It has
also been established that heating the LSO to 300 °C for about one day will restore it
to essentially its original light transmission characteristic [8]. Thus by providing a
means of heating the LSO it may be possible to use the same scintillator indefinitely.
If permanent darkening is expected to develop after say 100 heating cycles it would
probably not be difficult to design the RGIPM such that the LSO could be easily
replaced perhaps once every one or two years. It has been demonstrated that fused
silica radiation damage can largely be reversed by heat [9]. Thus the quartz viewport
279
and lenses of this system might also be kept free of radiation darkening by making
provisions for heating.
Each 5 keV electron hitting the LSO will produce M=150 photons [11]. These
will be observed by the same type of dual silica lens arrangement as discussed earlier
with a lens focal length of 150 mm, diameter of 25 mm and the magnification of -1. It
can be shown that the fractional error in the rms width of a Gaussian beam is given by
,1/2
1+
g
AG
G
.0705 Ax
^r^
M
Ax is the width of the beam being sampled, Nbe is the number of electrons produced by
the beam and the rest of the symbols have been defined previously. It will again be
assumed that the photo detector is an R5900U-07-L16 and that Ax is 1 mm. The value
of g is 2.25*10~4 and is again determined by the aperture and properties of the tube.
Assume E is 1.9 mm and solve for Nbe with a 5 % fractional error in E gives
Nbe=1.235*104. Assume that Xenon at a partial pressure of 2*10~7 Torr is introduced
into the beam tube at the measurement location. For Xenon a proton produces 44
(primary) electrons per cm at one atmosphere pressure [10]. If each bunch contains
1.25*1012 protons, 14000 electrons per centimeter will be produced by each bunch.
The length of beam that is imaged is 1.6 cm and thus the total number of electrons is
22400. This is enough to result in a 5 % measurement and also account for a 45 % loss
in photons through the optical system.
Using the spectrum of LSO [11] the rms chromatic aberration for this lens
system is 0^=63 jim. Restricting the field of view to 40 mm keeps the rms width of a
line image below 330 jim. Combining these in quadrature with the rms width of a
detector strip results in 439 jim. Finally combining this with the 1900 jim nominal
beam size results in 1950 jim.
In order to cover a 40 mm field of view, one can use three lenses each with an
R5900U-07-L16 tube for imaging. One would be in the center with its axis
perpendicular to the LSO sheet. The other two would have to be at an angle and for
these the resolution analysis will be more complicated. Alternatively, it may be
possible to have a photon counting MCP device made with 40, 1 mm detection strips.
Quadrupole moments in the Synchrotron
Some very innovative work has been done at CERN for measuring quadrupole
moments of beam bunches in a Synchrotron [12] and the CERN design may be
adapted for use in the AHF system. However it is worth investigating the possibility of
using a conventional stripline because in AHF the quadrupole signal will be relatively
large.
Consider a conventional stripline position monitor in a circular beam tube with
signal outputs T(top), B(bottom), L(left) and R(right). If a Gaussian shaped beam is
centered and the strips are narrow, the quadrupole signal from such a device will be
(T + B)-(L + R)
T+B+L+R
280
(6)
Here the a's are the rms beam widths and b is the beam tube radius. Present models of
the beam indicates that the largest a2 differences will be (2.68 mm)2-(1.12 mm)2 in
nondispersive regions at 50 GeV. The vertical beam tube diameter is limited to about
46 mm by the bending magnets. In dispersion free regions, it may be possible to use a
circular cross section beam tube with a radius of 23 mm. With these numbers, the
quadrupole signal will be 0.026. Figure 1 shows the signal one would get from one
bunch at 50 GeV from an electrode of a 0.5 ns long stripline after it is passed through
a 10 MHz rms low pass Gaussian filter. If the strip is assumed to intercept 10 % of the
wall current the quadrupole voltage signal will be of the order of 2.8*0.4 V*0.026 =29
mV. It is easy to build amplifiers with 10 MHz bandwith that have an input noise of
25 jiV and thus signal to noise will be about 800. At injection S/N will be 320, thus
amplifier noise will not be a problem.
The proposed processing electronics is shown in figure 1. The most
troublesome part of the signal is the monopole part. It is (largely) eliminated by taking
a difference between T and R signals and B and L signals with 180° hybrids (MACOM model HH-108-SMA). The output from the hybrids would go through Gaussian
filters, amplifiers and than the signals would be digitized. Current, position and
quadrupole information would all be computed digitally.
The striplines would be terminated externally at both ends to minimize
reflections and so calibration signals can be injected to measure how well the
difference is being taken by the hybrids. It may be useful to add adjustable attenuators
and phase shifters before the hybrids so corrections can be made to assure precise
cancellation of the monopole signals.
Figure 1. At left, calculated signal from one strip after passing through a Gaussian filter. Right,
processing electronics.
Flying wire in the Synchrotron
In one design that has been investigated, a rotating wheel of 30 cm diameter
would cause a C wire on a fork to reciprocate through the beam. An 8 jim wire would
have a maximum acceleration of am=0.38 g (g=9.8 m/s2), would pass through the
beam f=1.6/s and reach a maximum temperature of 1750 °K if the wheels angular
velocity was co=5 rad/s. The effect on the beam with the 8 jim wire will be about the
same as if Nitrogen were present in the beam tube at 4*10~7 Torr. Using a formula for
emittance growth [13], the 8 jim wire would cause about a 10 % emittance growth in
the 20 s of an acceleration cycle. Thus it would be useful for calibrating the non-
281
intercepting devices under conditions almost identical to an unperturbed beam. With
co=15 rad/s a higher sampling rate of 4.8/s could be achieved with no increase in
emittance growth. The peak temperature of the C wire will be 1100 K and am=3.44 g.
There are two ways the relative charge density being intercepted by the wire
can be measured. One is to measure secondary emission and the same electronics may
be used as described for the harps. Equation 1 can be used to estimate the
measurement error due to amplifier noise. m=183/mm, the number of samples per
mm, is calculated using the wire velocity of 1.05 m/s and the revolution time of T=5.2
jis. Using s=1.5 mm, x=35 jim, Q0=7*10~7 C and the other quantities the same as for
the harps one gets that the S/N=5000.
The second method is to detect radiation by attaching a scintillator like LSO to
a photomultiplier. It has been calculated that, on average, each proton from a
circulating bunch will deposit 10 jieV of energy per cc of LSO when a 35 jim C wire
is in the center of the beam, and the detector is 100 m downstream of the wire and 20
cm from the (straight) beamline center [14]. For a beam bunch of 1.25*1012 protons,
this becomes 125 MeV/cm3. For an 8 jim wire this results in 6.5 MeV/cm3. One
photon is produced per 340 eV of deposited energy [11], and thus 1.9*104/cm3 photons
are produced. Using the sampling rate m=183/mm, one gets that a total of 1.24*108
photons/cc are produced in one sweep through the beam. Equating this to Nb in
equation 3 and using a detection efficiency of g=.05 results in S/N=835 if the LSO
volume is 1 cm3. Energy deposition results for lower beam energies are not yet
available, but it is likely that several cm3 of LSO will be needed. In the actual system
the noise will probably be dominated by beam loss.
References
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
13)
14)
W. C. Sellyey and R. W. Kruse, "Operational Amplifier Based Stretcher for Stripline Beam Position
Monitors", 1991 IEEE Particle PAC, San Francisco, California, pp. 1145-1147, May 1991
S. Reiche, J. B. Rosenweig, "Transition Radiation for Uneven, Limited Surfaces", Proceedings of the
2001 PAC, Chicago, P-1282
D. W. Rule, R. B. Fiorito, "The Use Of Transition Radiation As A Diagnostic For Intense Beams",
NSWC 84-134, Naval Surface Weapons Center, Dahlgren, Virginia 22448, Silver Springs, Maryland
20910
R. H. Hughes, J. L. Philpot and C. Y. Fan, "Spectra Induced by 200-kev Proton Impact on Nitrogen",
Phys. Rev. V 123, Number 6, Sep. 15, 1961 P-2084.
J. H. Kampershor, et. el., "Initial Operation of the LEDA Beam-Induced Fluorescence Diagnostic" BIW
2000, AIP Conference Proceedings 546, P-456
Burton G., et. el., "THE LUMINESCENCE PROFILE MONITOR OF THE CERN SPS", CERN-SL
DIVISION, CERN-SL-031 BI, presented at EPAC 2000-Vienna-Austria, 26-30 June 2000
J. A. Melles, editor, Melles Griot Inc., "Optics Guide 5", 1990
Masaaki Kobayashi, Mitsuru Ishii, Charles L. Melcher, "Radiation damage of cerium-doped
oxyorthosilicate single crystal", NIM in Phys. Res. A 335(1993) 509-512
W. Tighe et. el., "Proposed experiment to investigate use of heated optical fibers for tokamak diagnostics
during D-T discharges", RSI 66(1), Jan 1995, P-907
F. Sauli, "PRINCIPLES OF OPERATION OF MULUWIRE PROPORTIONAL AND DRIFT
CHAMBERS", CERN 77-09, 3 May 1977
V. V. Avdeichikof et. el., "Light output and energy resolution of Csl, YAG, GSO, BGO and LSO
scintillators for light ions", Nuclear Instruments and Methods in Physics Research A 349 (1994) 216-224
A. Jansson, D.J. Williams, "A new optimized quadrupole pick-up design using magnetic coupling", NIM
A 479 (2002) 233-242
D. A. Edwards, M. j. Syphers, "An Introduction to the Physics of High Energy Accelerators", John Wiley
& Sons, Inc.
W.P. Lysenko, LANL, private communication
282