Professional Documents
Culture Documents
cm
__ NUCLEAR
INSTRUMENTS
l!B
ELSEVIER Nuclear Instruments and Methods in Physics Research A 401 ( 1997) 442462
8 METHODS
IN PHYSICS
RESEARCH
Sectlon
A
Abstract
Bremsstrahlung is produced in electron storage rings when the particle beam interacts with the storage ring components or
with the residual gas molecules in the storage ring vacuum. The interaction of the particle beam with the gas molecules occurs
continuously during storage ring operation. At straight sections of a storage ring, the gas bremsstrahlung contribution from
each interaction adds up to produce a narrow monodirectional photon beam that travels down the beamlines. At the Advanced
Photon Source (APS), where the stored electron beam energy is 7 GeV, gas bremsstrahlung generated in the straight sections of
the insertion device beamlines, which is a total of 15.38 m in effective length, can be significant. This paper presents the results
of the bremsstrahlung spectrum and energy measurements from the insertion device beamlines of the APS using a hermetic
lead glass calorimeter. The calorimeter consists of 25 lead glass blocks. each 6.3 cm x 6.3 cm x 35 cm, stacked into a 5 x 5
array. Phototubes connected to the lead glass blocks collected the signals. The calorimeter is calibrated using the maximum
energy of the observed bremsstrahlung spectrum. The total energy radiated and the bremsstrahlung spectrum were measured
as a function of the storage ring beam current for both electrons and positrons of 7GeV energy. Vacuum in the straight
section was continuously monitored, and measurements to determine the nature of the residual gas inside the storage ring
were performed. The measured bremsstrahlung spectra are fitted to a function that shows an approximate l/E behavior. The
gas bremsstrahlung rate from the APS undulator beamline straight path of 15.38 m is measured as 60.0 I!I 2.0 GeV/s/nT/mA.
Significant contribution of bremsstrahlung from sources other than the residual gas molecules is observed. The maximum
bremsstrahlung energy measured in a beamline during the experiment is 369.0 ?C22.0 GeV/s/nT/mA.
the inverse square of the mass of the particle. How- produce a narrow monodirectional bremsstrahlung
ever, as the velocity of the particle approaches rela- beam. This beam travels down the beamline, along
tivistic values, its mass is of decreasing importance. with the synchrotron radiation, with a characteristic
Thus, bremsstrahlung generation by charged particles emission angle [7] (73 prad at the APS). Because
other than electrons is not easily detected unless they the storage ring beam straight paths at the APS are
have relativistic velocities. 15.38 m long (in the line of sight of the beamlines),
From the experimental standpoint, electrons have gas bremsstrahlung production in the beamlines is
been the most important source of bremsstrahlung. significant. Bremsstrahlung can develop into an elec-
The differential cross section do for the emission of tromagnetic shower within any beamline component
a photon in the energy range between E;, and E, + dE,., it encounters. In addition, it can also generate pho-
by an incident electron of kinetic energy E, in a ma- toneutrons from targets of high-Z materials [8] like
terial of nuclei charge Ze can be expressed as [l, 21 photon beam stops and collimators.
Gas bremsstrahlung measurements have been per-
,(E+moc2)dE, formed earlier in the storage rings at KEK [9] in
do- = aoB(Z, E,/E)Z E
Ej Japan and Frascati [ 1O-l 21 in Italy. These measure-
(cm’/nucleus), ments were done at lower stored beam energies com-
(1)
pared to the 7 GeV particle beam energy of the APS
where (~0= &e4/mic4, mOc2 being the rest mass of storage ring. Monte-Carlo electromagnetic shower
the electron, and B is a coefficient that is a slowly codes [ 131 have also been used to quantitatively es-
varying function of Z and E,/E. The numerical evalu- timate gas bremsstrahlung from storage rings. These
ation of B as a function of Z and E,/E has been done codes assume air at atmospheric pressure as the parti-
by Heitler [2] and has been verified for particle ener- cle’s traveling medium. The results were then linearly
gies up to 500MeV. Although a comprehensive the- extrapolated to typical storage ring vacuum pressure.
ory of bremsstrahlung at all particle energies does not But, in reality, considering a single vacuum value will
exist as of yet, several equations that could be used not suffice as there exists a pressure profile along the
for different electron energies and various materials length of the straight paths in a storage ring. Also, the
are available [3, 41. chemical composition of the residual gas molecules
The Advanced Photon Source (APS) at Argonne along the particle trajectory in the storage ring vacuum
National Laboratory consists of a 200MeV elec- chamber is different from that of air and may even
tron linac, a positron converter target, a 450MeV change during the course of storage ring operation.
positron linac, a positron accumulator ring, a 7GeV In previous studies [ 141, the discrepancy between the
booster synchrotron, and a 7 GeV electron/positron reported measurements and the Monte-Carlo simula-
storage ring along with a number of experimental tions is at least a factor of 3. An existing empirical
photon beamlines utilizing the synchrotron radia- formula [ 151 has been used to roughly estimate the
tion [5]. Bremsstrahlung is produced in the APS gas bremsstrahlung dose rate. But, from the point of
storage ring when the circulating beam scatters from view of the radiation protection of the personnel, an
the storage ring components and/or from the resid- actual experimental measurement of this dose rate is
ual gas molecules in the vacuum [6]. Interactions of imperative.
the beam with the storage ring components gener-
ally occur when the beam deviates from the ideal
closed orbit or during injection and beam dumps. 2. The experimental setup
Scattering from the residual gas molecules in the
storage ring vacuum occurs continuously during op- Photons entering a medium transfer their energy by
eration, which generates the gas bremsstrahlung. Gas means of repeated radiative and collision interactions,
bremsstrahlung becomes very important for large generating a multiplicative electromagnetic shower.
storage rings with long straight beam paths on either An accurate measurement of the gas bremsstrahlung
side of the insertion device (ID) straight sections, spectrum can therefore be conducted by means
as the contribution from each interaction adds up to of a sensitive electromagnetic calorimeter that
444 M. Pishurody et al. J Nucl. Instr. and Meth. in Phys. Rex A 401 (1997) 442-462
Bremsstrahhmg Photon
Fig. 1. The lead glass electromagnetic calorimeter. Details of a single lead glass block are also shown.
possesses good resolution at energies around The electromagnetic calorimeter used to measure
a few GeV. Approximately, 1.2 x lo5 gas bremsstrah- the bremsstrahlung energy is shown in Fig. 1. It con-
lung photons are estimated to be produced per second sists of 25 lead glass blocks tightly stacked into a 5 x 5
in a 15.38m-long typical straight path of the APS array. Each lead glass block used has dimensions of
storage ring. Thus, the calorimeter should also have 6.3 cm x 6.3 cm x 35 cm (13.5X0). The dimension of
a fast time response to cope with such high photon the entire bulk of the detector is sufficient to contain
rates. the electromagnetic shower in the transverse direc-
tion. The 13.5X0 depth will contain about 96% of
2.1. Lead glass calorimeter the shower longitudinally. Each block is enclosed in
a metal light-tight can with a photomultiplier tube
Lead glass, which has a radiation length (X0) of (PMT) and base fastened to the end. An LED can be
about 2.6cm, is an ideal material for the desired attached to the center of the front of each detector as
calorimeter. It is composed of 50.8% PbO, 41% SiOz, a source for gain and stability studies. During data
and 7% K20 by weight. The energy resolution of this acquisition, the incident photon beam is aligned on
material is typically about S-10%/&. The charac- the center of the central lead glass block of the 5 x 5
teristic response time of lead glass is considered fast matrix (see Fig. 1).
and is typically about 20ns. Approximately, 20X0 of The first optical enclosures (FOE) or the white
material, and therefore a 52 cm long lead glass block, beam experimental stations of the insertion device
will contain practically 100% of the electromagnetic (ID) beamlines at the APS, which are adequately
shower longitudinally. The transverse development of shielded, were found to be suitable for conducting
electromagnetic showers in different materials scales the measurements of gas bremsstrahlung characteris-
fairly accurately with the Moliere radius. The Moliere tics. The ID was kept magnetically open to the full
radius of lead glass is about 3 cm, which means that extent in order to minimize the synchrotron radiation
an electromagnetic cascade can be completely con- background. However, there was still a small contri-
tained transversely in a 30cm diameter block of this bution from the synchrotron radiation produced by the
material. upstream and downstream bending magnets, which
M. Pisharody et al. / Nucl. Instr. and Meth. in Phvs. Rex A 401 (1997) 442-462 445
travels along the straight path of the particle, arriving estimated to be generated per GeV photon, the upper
at the calorimeter location. This bending magnet syn- end of the spectrum will easily overflow unless the
chrotron energy spectrum ranges from 0 to SOOkeV PMT outputs are attenuated. The attenuation factor
with a peak energy of -20 keV. A thin copper shield has been chosen so as to make use of the full dynamic
of 0.5X0 (6mm) was placed on the calorimeter front range of the 11 bit ADCs with as little overflow as
face to eliminate this remaining residual synchrotron possible. The attenuation on each PMT channel has
radiation. been set to approximately 25 dB, except for the central
PMT which splits half of its current to the trigger
circuit as shown in Fig. 2. To compensate for the split,
the attenuation on this PMT channel has been adjusted
3. Data acquisition system
by a factor of two less than the remaining channels.
The trigger is used both to generate a gate to the
A schematic of the data acquisition system (DAQ) ADCs upon a real energy deposit, as well as to ex-
is shown in Fig. 2. Signals from individual lead glass clude any residual low energy synchrotron radiation.
detector photomultiplier tubes of the calorimeter are Using a discriminator, a threshold of 150 MeV is
read out into two 16-input fast encoding and read- set to eliminate events from the low-energy region.
out ADCs (FERA) through a passive attenuator re- Bremsstrahlung energy lost in this cutoff region is
sistor array. The ADCs are each of 11 bit dynamic then corrected for in the final analysis using a fit and
range and have a constant short conversion time of extrapolation procedure, Also, a delay has been intro-
8.5 us. The attenuators prevent the ADCs from over- duced to signals from all the PMT channels relative
flowing. As will be discussed in the next section, the to the trigger channel split from the central detector.
gain of the PMTs at their selected operating voltages Because the transfer of acquired data to a computer
is roughly 106. With about 400 photoelectrons typically takes many milliseconds per event, the
Disk
T
CAMAC
AttenuatorBox
n
PMTI
F D M
R E
E I M
PMT13(central) V 0 MicroVAX
R E R
R Y
A
PMT25
GateGenerator Discriminator
digitized data from the ADC is stored into a 16 K mem- electrons Npe is approximated as
ory unit for subsequent slow read out by the computer.
Transfer of data from the ADCs to the memory oc- N = WW - PADC 2
PC (3)
curs after each gate and proceeds at 100 ns per word. ( OADC b
This eliminates significant dead-time corrections that
where PADCis the mean ADC pedestal. The gain is
will have to be applied otherwise. Once the memory
then determined from the relation
unit buffer is full, an inhibit is generated that prevents
further readouts of the ADCs by the driver. The crate G- (ADC) X 0.25 pC
signals the computer that the buffer is to be read out. ADC channel
NPe
After reading out the full memory unit, the computer
x 6.242 x lo6 electrons
resets the memory address of the unit and clears the (4)
ADC inhibit for the next batch of events. A MicroVax 1pc
computer is used to control the CAMAC modules and Fig. 4 shows a typical variation of the photoelectron
to store the data. yield and gain as a function of the high voltage applied
The digitization time of the ADC is about 8.5 ps, to the lead glass detector given in Fig, 3.
and readout time by the driver into the memory module The operating voltage of the phototubes depends on
is about 2 us. This implies an approximate dead time of both the number of photoelectrons expected at 7 GeV
11 ps between accepted triggers. Thus, the ADCs will and the region of linear operation. We estimated the
convert and the memory will store all events triggered expected photoelectron yield at 7 GeV by using the
up to a rate of about 90 kHz (l/l 1~s). Any trigger endpoint of the bremsstrahlung spectrum, which was
rate above this will have to be dead-time corrected obtained in an earlier measurement made with a single
in the analysis. By introducing a known delay time lead glass block. By matching the pulse height for the
7 between ADC gates, the measured rate NM can be end point of the bremsstrahlung with the mean pulse
corrected for dead time by using the relation height from an LED at identical operating conditions,
the number of photoelectrons at 7 GeV was found to be
NM approximately 2700. The initial high voltages for the
NR = ___
1 - NMz’
(2)
PMTs were then chosen in such a way that each of the
tubes operate in their linear region with approximately
where NR is the dead-time-corrected actual trigger
similar gains.
rate. The delay r is set at 15 us, close to the ADC dead
The calorimeter is mounted on a platform that slides
time of 11 us.
on a pair of rails. A step motor is used to power the
Apart from the ADC data, three scalers were also
sliding mechanism. The detector volume can thus be
written out into the data stream as shown in Fig. 2.
moved beam-left or beam-right using this arrange-
They include a pulser running at a fixed frequency
ment. This sliding surface with the detector attached
(Sl ) that is used to establish the rate of triggered
is then placed on a table that can be manually raised
events and two numbers representing dead-timeless
and lowered. These lateral and vertical movements
triggered events (S2) and accepted triggered events
thus enable us to scan each lead glass block with the
(S3) within the dead-time limit.
bremsstrahlung beam.
The ADC distribution of the pulsed LED source As discussed in previous section, the PMT gains
follows Poisson statistics, the mean of which ((ADC)) have so far been set only by a pulsed LED source.
changes with applied high voltage. An example of The bremsstrahlung beam itself is used to fine-tune
such a distribution for a typical lead glass detector the gains more accurately prior to the data acqui-
is given in Fig. 3. Using the mean value (ADC) and sition. The procedure involves scanning each lead
width aAoc of the distribution, the number of photo- glass block of the calorimeter by the bremsstrahlung
M. Pisharody er al. 1 Nucl. Instr. and Meth. in Phys. Res. A 401 (1997) 442462 441
v-r-
22250
= 660.5f0.2
1
j
oas= 16.6fO.l
1750
1
185OV
1500
t
1250 L
1000
1
250
E
AOC Channel
Fig. 3. Pedestal-subtracted ADC distribution for various PMT high voltages, using a pulsed LED source attached to the front of a
typical lead glass detector. As an example, the inset shows details such as (ADC) and crA~c of the hatched distribution for 2100 V. The
photoelectron yield in this case is approximately 1220.
beam. The beam is made to fall at the center of each out during the scanning procedure for approximately
lead glass block using the mechanism that moves identical beam currents.
the calorimeter. Individual outputs are then displayed
through a qVt multichannel analyzer module. The
observed bremsstrahlung distribution drops at the 6. Data collection
maximum particle energy of 7 GeV, thus defining an
endpoint for the spectrum. This well-defined endpoint The Advanced Photon Source (APS) stored elec-
is used to fine-tune the PMT gains. By scanning the trons in the storage ring from the time it became oper-
beam from block to block, the individual PMT high ational until July of 1996. Positrons have been stored
voltages are adjusted within their respective operating from then on, starting from August of 1996. A total of
ranges, so as to fix the endpoint of the bremsstrahlung four different measurements of the gas bremsstrahlung
distribution from each detector approximately at the intensity were conducted at various insertion device
same qVt channel. Data for the purpose of absolute (ID) beamlines during a time interval of five months.
energy calibration of the calorimeter were written Run #l and Run #2 were performed at ID Sector 12
448 M. Pisharody et al. / Nucl. Instr. and Meth. in Phys. Res. A 401 (1997) 442-462
,.%102pl / I I II, 1 1 , 1 , 11 1
k?
,s 9000
S .
6’) ..’
8000
1
_I
7000 .
1 4
6000
.
5000. -
r, ,,,,1,,,,1,,~,1~~~~1~~~~~~“1~
1800 1850 1900 1950 2000 2050 2100
PMT High Voltage (Volts)
Fig. 4. (a) Calculated photoelectron yields (Npe) for various PMT high voltages. obtained using Eq. (3). (b) Calculated gains (C) as a
function of PMT high voltages as determined by Eq. (4)
with electron beam in the storage ring. ID Sector 10 used to pedestal-subtract the experimental data writ-
was used for Run #3 and ID Sector 13 for Run #4, ten out. During a pedestal run, the station shutter is
both with stored positron beam. The data analyzed in kept closed to prevent the photon beam from striking
this paper come from all these four runs. the calorimeter, although it was found that the mean
The lead glass calorimeter setup was placed inside pedestal value was unaffected by the beam.
the experimental station in such a way that the central
lead glass block of the calorimeter is aligned along 6.1. Storage ring vacuum analysis
the photon beam path [16]. When the beam shutter
opens, those events satisfying the preset discrimina- The intensity of gas bremsstrahlung is a func-
tor threshold trigger the gate and the corresponding tion of the stored beam current, atomic number of
ADC digitized outputs are written out to the computer the residual gas, and the residual gas pressure of
hard disk. The scalers along with 25 ADC channel the vacuum in which it has been generated [3, lo].
signals are stored on an event-per-event basis. Dif- Determination of the correct value of the residual gas
ferent runs are performed corresponding to different vacuum and its chemical composition is thus impera-
storage ring beam currents. Vacuum values from rel- tive for subsequent intensity and energy calculations.
evant ion pumps (IPs) and ion gauges (IGs) in the Bremsstrahlung photon(s) could be generated any-
straight section are also recorded for the time period where in a 15.38 m-long effective straight path that is
of each run. The ADC pedestal runs are also con- in the line of sight of an ID beamline. These straight
ducted from time to time using an external gate from paths are equipped with ion pumps and gauges that
a random puber to trigger. This information is later provide information about the vacuum along the ID
M. Pisharody et al. / Nucl. Instr. and Meth. in Phys. Res. A 401 (1997) 442-462 449
sector [ 161. A typical vacuum profile of the ion gauge The vacuum values, averaged over the data acqui-
IG of Sector 12-ID is shown in Fig. 5 for a duration sition run time, are calculated from all the ion pump
of 24 h. The sudden drops in the profile correspond and gauge readings. The straight path segment-length
to a zero beam current state (due to beam loss) of weighted values of these time-averaged individual
the storage ring. Maximum vacuum is attained un- vacuum readings are then used for normalizing vari-
der such zero beam current conditions. For non-zero ous results. Although vacuum fluctuations are present
beam currents, the average vacuum is typically of the on a longer time scale as shown in Fig. 5, no signifi-
order of a nTorr. cant variations (> 10%) are observed over the typical
data collection time period of 16 min.
Table 1
Summary of the RGA studies
5
cn IO
s
8 IO'
=
p 103
102
10
1 1
0 1000 2000 0 1000 2000 0 1000 2000
ADC Channel ADCChannel ALXChannel
IO 10 10
1 1 1
0 IWO 2000 0 1000 2000 0 1000 2000
ADCChonnel ADCChonnrl AlXChannel
10 10 10
1 1 1
0 1000 2000 0 1000 2000 0 1000 2000
ADCChonnel ADCChonnel AXChannel
#7 #8 #9
Fig. 6. Typical uncorrected raw ADC pedestal-subtracted spectra from the central 3x3 lead glass blocks of the calorimeter. The cente
block (#I 3) contains ~92% of the energy deposited. Lateral energy spread is visible in all the surrounding blocks.
M. Pisharody et al. / Nucl. huir. and Meth. in Phvs. Rex A 401 (1997) 442462 451
be distributed over many ADC channels instead of 6.1 GeV instead of the expected cutoff value of 7 GeV.
sharply dropping at the cutoff limit. The functional The calibration constants C,, and the resolution values
form of the ADC distribution can be assumed as [ 171 cn obtained through the fits are given in Table 2. The
complete bremsstrahlung energy spectrum can then be
dW’L C,co1 obtained using the relation
d(ADC)
E; 1
InaY
-(Ey - ADC/C)* dE
=N
J $5 exp
E: In,” i’ 2o,2E, Y)
Channel
ADC Total
Fig. 7. The summed ADC spectrum of all nine lead glass block detectors of the calorimeter. Additional entries after the cutoff are from
multi-bremsstrahlung events as discussed in the text. The inset shows the fit to the end-point region of the spectrum using Eq. (5), which
yields values for the calibration constant and resolution of the calorimeter.
As an example to demonstrate our calibration ability for this to occur at the storage ring vacuum
procedure and results, the summed ADC spectrum has been estimated to be less than 1%. The dou-
of all the nine lead glass blocks from a data set ble bremsstrahlung production will roughly follow a
acquired at an electron beam current of 6OmA is l/E2 distribution. These events are embedded within
shown in Fig. 7. The inset shows the cutoff region the l/E bremsstrahlung profile and are impossible to
of the spectrum with the fit as discussed above. The distinguish, except beyond the 7 GeV cutoff. Because
fit unfolds the effective calibration constant Ce~ and the experiment cannot resolve these simultaneously
resolution 00 of the calorimeter. These values are generated photons, only their added total energy is
listed in the inset as fit results. Fig. 8 shows the cor- detected by the calorimeter. This results in the pres-
responding energy spectrum for the data in Fig. 7. ence of what appears to be extra pileup events after
The ADC pedestals have been corrected for all the the cutoff of the l/E bremsstrahlung spectrum. The
nine channels used in the analysis. The additional effect does not seem to shift the knee of the spec-
entries beyond the bremsstrahlung cutoff in Fig. 8 are trum. Thus, the end point of the spectrum can still
from real multiple bremsstrahlung events. The prob- be fit as accurately as described in the calibration
M. Pisharody et al. / Nuci. Instr. and Meth. in Phys. Res. A 4001 (1997) K-462 453
GeV
Bremsstrohlung Energy
Fig. 8. A typical uncorrected bremsstrahhmg energy spectrum. Calibration constant Ccc obtained from the fit procedure has been used to
convert the ADC distribution shown in Fig, 7. The cutoff, smeared by the finite resolution of the calorimeter, can be seen at around 6.77 GeV.
procedure section, even in the presence of such ex- factor W’, which is given by
tra events near the cutoff region. Fits similar to that
in Fig. 7, conducted on data sets from various runs, (8)
yield a mean calibration constant (( &)) value of
250.0flO.O ADC/GeV and a mean resolution value of
where n is the total number of events in the ith buffer.
(lO.Of0.7)%/~.
The sum of these weighted individual buffer rate spec-
tra, averaged over the total number of buffers b used
7.3. The measured bremsstrahlung energy spectrum for the analysis, yields the spectrum for the average
photon production rate (N):
._
0 1 2 3 4 5 6 7 8 910 0 1 2 3 4 5 6 7 6 9 10
GeV GeV
Photon Energy (EJ Photon Energy (5)
2 3 4 5 0 7 6 9 10
GA’ GA’
Photon Energy (E,) Photon Energy (E,)
Fig. 9. The bremsstrahlung for dead-time losses. for typical beam currents from the electron beam Runs #l and
energy spectra, corrected
#2 in Sector 12-ID ((a) and (b)) and from the positron
beam Runs #3 in Sector lo-ID and #4 in Sector 13-ID ((c) and (d)). The y-axis
represents the photon production rate (N) normalized to corresponding vacuum. The lowest and highest beam currents are marked near
the respective spectra. The beam currents for the intermediate plots are shown on the side in that order.
production rate (N) determined by Eqs. (8) and where AN is the photon production rate at a given
(9), normalized to the estimated vacuum, of photons energy. The total bremsstrahlung energy radiated in
having corresponding energies Ey shown on the x- units of (GeV) (s)-‘(nT)-’ is also calculated for
axis. The results shown in Fig. 9 are corrected for the corresponding beam currents by using the equation
data acquisition system dead time and ADC pedestals.
The total number of bremsstrahlung photons emitted
(11)
in units of (s)-‘(nT)-’ are now determined for the
corresponding beam currents using the relation
The finite sum values calculated from Equation (10)
and Equation (11) include both single and multiple
N:o’ = ‘2 AN, (10)
E ml” photon events, as well as their energies, The detailed
M. Pisharody et al. / Nucl. Instr. and Meth. in Phys. Res. A 401 (1997) 442-462 455
Table 3
Observed dead-time-corrected bremsstrahlung energy originating from the 15.38m particle beam straight path (not corrected for other
systematic errors)
results obtained along with the respective vacuum val- along the y-axis in order to be able to clearly distin-
ues are tabulated in Table 3. guish between the unhatched (corrected) and hatched
(uncorrected) distributions. Evidently, as can be seen
from Eq. (2), the correction is relevant only at high
8. Systematic corrections rates for the r value that has been selected. This can
be seen at the low energy ends of the spectra in both
8.1. Dead-time correction Figs. 10(a) and (b), where the rate of photon produc-
tion is high. The effect of the dead-time correction
As mentioned earlier, the energy spectra in becomes totally insignificant as we go to the high en-
Fig. 9 are already corrected for dead-time effects of ergy end of the spectrum. On an average, a maximum
the data acquisition system. Eq. (2) is used to cal- of 3.0% correction is observed for typical data corre-
culate the required correction. The value for r in the sponding to the largest beam current.
equation is taken as 15 us, which has been preset in
the data acquisition system through a gate and delay
module. 8.2. Longitudinal shower leakage correction
To demonstrate the effect of dead-time correction
using Eq. (2), two typical corrected spectra, super- In order to completely contain an electromagnetic
imposed on the corresponding uncorrected ones, are shower in the longitudinal direction, a minimum of
shown in Fig. 10. The figure is plotted in linear scale 20x0 material of the shower medium is required.
456 M. Pisharody et al. J Nucl. Ins@. and Meth. in Phvs. Rex A 401 (1997) 442-462
1 Deed-time Corrected
/ Deod-time Corrected
Oeod-time Uncorrected
GeV GeV
Photon Energy (E,) Photon Energy (E,)
Fig. 10. Typical energy spectra, uncorrected and corrected for dead-time, from Run #2 in Sector 12-ID for two different beam currents.
Shower leakage through the back side of our 13.5X0 leakage. The average leakage energy in the lateral
calorimeter is therefore possible, requiring necessary direction is found to be zero.
correction to the measured total bremsstrahlung en-
ergy values in Table 3. 8.3. Low energy or threshold correction
An EGS4 [ 181 Monte-Carlo simulation has been
run to understand the correlation between the incom- The discriminator threshold trigger discussed ear-
ing bremsstrahlung energy and the fraction of it leak- lier cuts off events having energy approximately below
ing out through the back of the lead glass calorimeter. 150MeV. Along with the synchrotron events, some
The geometry selected consists of lead glass blocks, low energy bremsstrahlung will also be lost. The re-
each 35 cm long, in a 3 x 3 array. Fig. 11 shows the sults shown in Table 3 need to be corrected for the
correlation between the incoming bremsstrahlung en- lost bremsstrahlung energy.
ergy spectrum and the leakage energy spectrum at the To estimate the missing bremsstrahlung energy be-
back side of the calorimeter. From Fig. 11, it is clear low the threshold, fits to the spectra in Fig. 9 were per-
that leakage at the back side is zero at low incident formed. It is convenient to convert the (N)cc (l/E,)
bremsstrahlung energies (below 1 GeV). Thus, elec- distributions in Fig. 9 to a constant flat spectrum
tromagnetic showers generated by incoming photons using the following relation:
below 1 GeV are fully contained by the calorimeter.
1
Above 1 GeV however, Fig. 11 shows leakage from Ey x (N)cc E, x - = Constant. (12)
the rear side of the calorimeter. On average, approxi- E?
mately 4.3&O. 1% of the incoming energy is estimated Eq. (12) will yield a more or less flat distribution,
to be leaking out through the back end of the calorime- which can then be easily fitted to a straight line. Such
ter at energies above 1 GeV. This average percent a straight line fit to one of the typical distributions,
fraction is used to account for the longitudinal shower along with the resulting fit parameters, is shown in
M. Pisharody et al. / Nucl. Instr. and Meth. in Phys. Rex A 401 (1997) 442462 451
0
0 1000 2000 3000 4000 XaO 8000
BrZstrah:iYg
(Mev)
Bremsstrahlung Energy vs Bock Leakage Energy
Fig. 11. Results from EGS4 Monte-Carlo run. Scatter plot between incoming bremsstrahlung energy spectrum and shower leakage energy
spectrum at the back side of the calorimeter, showing the correlation
Fig. 12. Integrating the fit function along the entire rent), longitudinal leakage (4.3fO.l%), and threshold
energy range of the spectrum will give us the total cutoff effects (2.8fO. 1%).
energy radiated. The fraction of the energy lost due to The normalized bremsstrahlung energies show
the threshold cutoff is found to be a constant for all the consistent values within each individual run, but they
beam currents and runs. This fraction, calculated for vary from run to run and sector to sector. The re-
several beam currents in each of the runs, is tabulated sults from Table 5 are also shown plotted in Figs. 13
in Table 4. It can be seen from the table that, irrespec- and 14. Both these figures show the bremsstrahlung
tive of the beam current, approximately 2.8fO.l% energy E, normalized to the beam current, as a
of energy is being lost below the discriminator function of the storage ring vacuum P for different
threshold. experimental runs. Fig. 13 represents data from Run
numbers 1 (12-ID), 2 (12-ID), and 3 (lo-ID), show-
ing straight line fits of the form E =A + BP. The
9. Results errors shown are the measure of the deviation of each
data point from the mean value as obtained from a
Table 5 gives the corrected bremsstrahlung energy least-square fit. The y-intercepts A of these straight
results, normalized to vacuum and storage ring beam lines give the bremsstrahlung energy rates from
current. The results are corrected for the dead-time the storage ring straight path of 15.38 m, when the
losses (a maximum of 3.0%, for the largest beam cur- vacuum of the storage ring is zero nTorr. This is the
458 M. Pisharodv et al. / Nucl. Instr. and Met/t. in Phys. Res. A 401 (1997) 442-462
0 1 2 3 4 5 6 7 8 9 10
GeV
Photon Energy (5)
Fig. 12. Dead-time corrected bremsstrahlung spectrum with the y-axis in units of E, x Rate (A’), fitted with a straight line yielding the
fit parameters A and B (inset).
Table 4
Low-energy correction evaluated for a few typical beam currents
Table 5
Corrected bremsstrahlung energy results from the 15.38 m particle beam straight path
contribution to the total bremsstrahlung energy rate correspondingly enhances the gas bremsstrahlung
by non-gas interactions in the straight path. This can production rate. It is also observed that, in the beam
differ from sector to sector. The slope B of these fitted current range for which the measurements were con-
straight lines gives the gas bremsstrahlung energy rate ducted, the bremsstrahlung energy rate for a given vac-
from the 15.38 m straight path, normalized to both the uum is proportional to the current. Thus, the variations
vacuum and beam current. The gas bremsstrahlung in the normalized bremsstrahlung energy rate seen in
energy rate is calculated as 60.0120 GeV/s/nT/mA. Table 5 are attributed to the differences in the non-gas
This value is consistent for the data from Runs #l bremsstrahlung contribution from sector to sector and
and #2, both conducted at 12-ID, and also for the data to the enhanced gas bremsstrahlung production rate
from Run #3 conducted at lo-ID. However, as shown due to the presence of vacuum leaks.
in Fig. 14, the data from Run #4 obtained from 13-ID
show a corresponding gas bremsstrahlung energy rate 9.1. Bremsstrahlung dose estimations
that is approximately a factor of four higher than the
other three data sets. This observation is consistent The bremsstrahlung rate profiles in Fig. 9 are
with a vacuum leak detected in this sector during the converted into their respective fluence spectra by
data collection. As mentioned earlier, the presence of dividing them by a cross-sectional area that corre-
vacuum leaks in a sector can cause a change in the ef- sponds to the average bremsstrahlung beam size at
fective Z ofthe residual gas in the straight path, giving the detector location. This cross-sectional area is
rise to three to four times higher Z2 values than what determined from the average bremsstrahlung emis-
is expected under normal operating conditions. This sion angle [7] at the source point, 26m upstream.
460 Pisharody et al. 1 Nucl. Instr. and Meth. in Phys. Rex A 401 (19971 442-462
~ .t,. . i:-.-_:;..,...‘,...
y.-:: /............
.‘-
.-
I’
o”ll’ll~‘~““~~l~~‘I’~,
0 2 4 6 8 10 12
Fig. 13. Bremsstrahlung energy, normalized to the beam current, as a function of the storage ring vacuum for the experimental runs
conducted at Sectors IO-ID and 12-1D. The slopes of the fitted straight lines give the gas bremsstrahlung energy rate, the value of which
is given at the top. The y-intercepts represent the non-gas bremsstrahlung energy rate from these beamlines.
At 7 GeV APS electron energy, an average emission all practical purposes. The maximum dose-equivalent
angle value of 73 prad can be estimated. The aver- values corresponding to a few typical results in
age bremsstrahlung beam cross sectional area, 26m Table 5, calculated by summing the events un-
downstream with this characteristic angle of emission, der the dose-equivalent spectra along the entire
is then approximately 0.113 cm*. A set of already ex- energy range, are given in Table 6 in units of
isting fluence-to-dose-equivalent conversion factors Sv/h/nT/mA. From this table, it can be seen that the
(F,,,v in units of Svcm*) [19] is used to convert dose-equivalent rate that closely represents the gas
the measured fluence spectra to corresponding dose bremsstrahlung is 4.2 x 10e4 Sv/h/nT/mA, which
equivalent spectra. The correction for the dose from corresponds to 0.13 Sv/h/nT for a beam current of
the events missed due to the threshold setup has also 300mA. The maximum bremsstrahlung energy rate
been investigated. Earlier, it was shown that approxi- measured in this experiment at an ID beamline is
mately 2.8% of the total energy is lost in the threshold 369.0 f 22.0GeV/s/nT/mA as can be seen from
cutoff region of O-l 50 MeV. Since dose is defined as Table 5. This corresponds to a dose-equivalent rate
the energy absorbed per unit mass, its contribution in of 2.18 x 10e3 Sv/h/nT/mA. At this rate, the dose-
the threshold cutoff region will also be 2.8% of the equivalent value is 0.65 Sv/h/nT for a beam current
total dose which is small enough to be ignored for of 300 mA.
M. Pisharody et al. 1 Nucl. Instr. and Meth. in Phys. Rex A 401 (1997) 442462 461
_,_.
* ....‘“’
#2, Sector 12
_,ID,
_ e- ,,,,&........*,_....
.-.-.,;
.
_.-. ........‘.
_._ _,_.A+-.:,:.: .........,.,,
-.-.i: ,,.......
...‘““‘.‘#I( sector, 2 ,lJ,,?-
,.._.-‘-‘r;
..‘. ._,.......
,_._.- ,_.-’ ,.,....
.,......” ___---_.-’
,,,..,....
,..... l_._---
.., __.---
-d-e
___---
___---__-- 13, Sector 10 ID,Cf
Fig. 14. Data from the run conducted at Sector 13-ID along with the other three data sets shown in Fig. 13. The gas bremsstrahlung energy
rate from 13-ID is approximately a factor of four higher compared to those from Sectors lo-ID and 12-ID, consistent with a variation in
the effective Z of the storage ring residual gas due to vacuum leak.
Table 6
Normalized maximum dose-equivalent values corresponding to the bremsstrahlung energy radiated from the 15.38 m of particle beam
straight path. Only one typical value per run is shown
10. Conclusions authors are grateful to the members of the APS Col-
laborative Access Teams (CATS) belonging to Sec-
The lead glass calorimeter, by virtue of its sen- tors 10 (MR CAT), 12 (BESSRC CAT). and 13 (GSE
sitivity and fast response, is found to be a very CARS CAT) for providing the beamlines in order to
effective detector for the accurate measurement of conduct the experiment.
bremsstrahlung from electron storage rings. This This work is supported by U.S. Department of En-
calorimeter can be calibrated using the maximum en- ergy, BES-Material Sciences, under contract no. W-
ergy of the observed bremsstrahlung spectrum, which 31-109-ENG-38.
is equal to the energy of the particle beam in the stor-
age ring. The gas bremsstrahlung rate from the APS
undulator beamline straight path of 15.38m is mea- References
sured as 60.0f2.0 GeV/s/nT/mA. This corresponds
to a maximum dose-equivalent rate of approximately [II H.A. Bethe, W. Heitler, Proc. Roy. Sot. A 146 (1934) 84.
121W. Heitler, The Quantum Theory of Radiation, Oxford
4.2 x 1O-4 Sv/h/nT/mA. The analysis shows that University Press. London, 1944.
there is a significant bremsstrahlung contribution [31 H.W. Koch, J.W. Motz, Rev. Mod. Phys. 31 (1959) 920.
from sources other than the residual gas molecules in [41 B. Rossi. High Energy Particles, Prentice-Hall, Englewood
Cliffs, NJ. 1952.
the beam straight path of the storage ring. The maxi-
[51 G.K. Shenoy. P.J. Viccaro, D.M. Mills, Argonne National
mum bremsstrahlung energy rate measured in a sector Laboratory Report. ANL-88-9, Argonne, IL, 1988.
during this experiment is 369.0*22.0 GeV/s/nT/mA, [61 H.J. Moe, National Laboratory Report, APS-LS-141,
which corresponds to a maximum dose-equivalent Argonne, IL, 1991.
value of approximately 2.18 x 1O-3 Sv/h/nT/mA. [71 L.I. Schiff, Phys. Rev. 70, (1946).
In the beam current range for which the measurements PI G. Bathow, E. Freytag, K. Tesch, Nucl. Phys. B 2 (1967)
669.
were conducted, the bremsstrahlung energy rate for
[91 S. Ban, H. Hirayama, S. Miura, Hlth Phys. 57 (1989) 407.
a given storage ring vacuum is found to be propor- 1101 A. Rindi, Hlth Phys. 42 (1982) 187.
tional to the current. A measurement of the residual [Ill A. Esposito, M. Pelliccioni, Laboratori Nazionali di Frascati,
gas composition inside the storage ring straight path LNF-86.23NT Italy, 1986.
shows CO, CO?, and Hz to be the major constituting UT A. Esposito. M. Pelliccioni, Proc. Health Physics Society 20th
Midyear topical Symp., Reno. NV, USA, CONF-8602106,
components. 1987, p. 495.
[I31 G. Tromba, A. Rindi, Nucl. Instr. and Meth. A 292 (1990)
700.
Acknowledgements [I41 A. Ferrari, M. Pelliccioni, P.R. Sala, Nucl. Instr. and Meth.
B 83 (1993) 518.
The authors wish to thank Mark Keeffe for his as- u51 J.C. Frank, Laboratoire pour I’Utilisation du Rayonnement
Electromagnetique, LURE EP 88-01, Orsay, France, 1988.
sistance in designing the hardware for the lead glass
[I61 M. Pisharody, PK. Job, S. Magill, J. Proudfoot. R. Stanek,
calorimeter mount, Bill McHargue for providing the Argonne National Laboratory Report, ANL/APS/LS-260,
attenuator circuit module and associated electronics, Argonne, IL. 1997.
and Brian Rodricks for his help with the data acqui- [I71 J. Andruszkow et al., Calibration Procedure of the
Calorimeters of the ZEUS Luminosity Monitor, Deutsches
sition (DAQ) components and software. We are also
Elektronen-Synchrotron (DESY), HERA ZEUS-94-071,
thankful to Ed Semones for his help in the residual Hamburg, Germany, 1994.
gas analysis (RGA) effort, and Don Jankowski for [I81 W.R. Nelson, H. Hirayama, D.W.O. Rogers. EGS4 Code
his technical assistance. Special acknowledgements to System, Stanford Linear Accelerator Center, SLAC-265,
Harold Moe for his valuable suggestions during the Stanford, CA, 1985.
course of the experiment and analysis. Finally, the [I91 D.W.O. Rogers, Hlth Phys. 46 (1984) 891.