Greetings to all,
The Compton Experiment ended an extremely successful running period
with a record peak brightness beam. This experiment was designed,
constructed and carried out in a record time of 15 months, thanks
to the leadership of Igor Pogorelsky, the outstanding engineering
of John Skaritka, the excellent electron beam provided by Vitaly Yakimenko
and the dedication of rest of the local team, including Ping He, Shigeru
Kashiwagi, Karl Kusche, Akira Tsunemi and others. Thanks also for
our Japanese collaborators, headed by Professor Hirose (TMU), including
Dr. Tsunemi (SHI), Professor Washio (Waseda), Prof. Urakawa (KEK)
and otherws. Also thanks for funding provided by the US-Japan Collaboration
in High Energy Physics.
With the improvements in the ATF CO2 laser and the bunch compressor,
all to be finished next year, this will be a truly remarkable source
of femtosecond hard x-rays with an amazing peak brightness, just right
to start studying x-ray science with an LCLS class pulse length.
The data analysis of the Inverse Cerenkov Accelerator stage of the
STELLA experiment shows very good agreement with the model predictions.
Ilan Ben-Zvi.
Back to Top
Compton
Experiment (Reported by Igor Pogorelsky) |
In the previous report, we formulated the following goals for
the next week runs:
- Precise measurements of the electron and laser beam sizes in the
interaction region in order to make comparisons with the theoretical
predictions on the produced x-ray photons. Any improvement in the
electron and laser spot matching may lead to further enhancement
of the signal.
That is exactly what was accomplished during the last two weeks of
runs. In particular, by improving matching between the electron and
laser beams, we demonstrated a peak flux of 2x10**18 photons/second
at an x-ray energy of 6.5 keV. Below, we chronologically describe
the last Compton run in more details:
Week of September 20
The electron beam was reasonably well characterized using conventional
techniques: phosphor screen, Cerenkov radiation, transverse wire scans
with measurement of the x-ray signal. The confirmed cross section
of the e-beam is sigma_horiz=60 microns, sigma_vert=120 microns. Due
to low emittance, the beam can be considered as "filament"
(constant cross-section) over the interaction region.
Due to non-conventional nature of the laser focus ("donut"
beam focused by parabolic mirror) and lack of a reliable diagnostics,
its characterization required more attention. We used several methods
to study the laser focal region:
- magnified imaging of the focal area with the pyroelectric camera;
- transverse scan with the 3 mil wire measuring the energy transmission
behind the wire with a joulemeter;
- observation of burn patterns on the thermal paper at different
positions at the focus area;
- measurement of the energy transmission through a pinhole.
Based on these observations, the length of the interaction region
is about 5 mm. At ± 4 mm from the focal point we observe a well defined
"hole" at the axis in the radial intensity distribution.
The FWHM at the waist is 160 um or less. Higher resolution measurements
of the laser focus are not very conclusive because of possible aberration
in the imaging system, plasma formation on the pinhole and wire, and
a narrow dynamic range of the thermal paper. Note that the ~5 mm length
of interaction region implies ~70 micron sigma according to the formula
for Gaussian beams.
After obtaining data about the laser and electron focus, we were
ready to verify prediction for the number of generated Compton x-rays.
To calculate the photon number we use analytical formula
Nx~6.7´ 1011 x E[J] x Q[nC] x f x
Lambda[microns] / r**2[microns]
, where E is a portion of the laser pulse that overlaps with
the e-beam within the interaction region, f is the spatial
overlap factor between the laser and electron beam. This formula is
similar to one used by P. Sprangle, E. Esarey, et al in the NRL Compton
experiment. For Gaussian beams, this formula agrees within 10% accuracy
with simulations done by our collaborators at KEK using code CAIN.
This code does not permit to calculate non-Gaussian or aberrated beams
where the relation between the transverse and longitudinal dimensions
of the laser focus is not well defined. In the formula, we can assess
these parameters independently. Thus, we take 5 mm for the interaction
length and leave a radius in the laser focus open for discussion.
5 mm interaction length implies that just a 30 ps slice of the laser
pulse participates in the interaction. By plotting the Compton signal
against the optical delay scan and normalizing the obtained curve
to 200 mJ (typical laser energy during the run) we obtain the peak
power ~500 MW. That means that just 15 mJ of the laser energy participates
in the interaction. Substituting this and all other known data into
the formula for Nx we obtain. Nx ~ 1011 x
f / r**2. Based on observations, the right number for
the laser focal size is somewhere between 100-50 microns. It makes
a factor of 4 difference in the quadratic dependence. However it does
not make much difference in the calculated Nx number. Indeed,
if we assume r = 100 microns, then the laser spot covers essentially
the whole e-beam cross section, and we can take f=1. When r
= 50 microns, the laser beam covers only about 30% of the e-beam,
and the calculated photon number is close to the previous case. Thus,
we can calculate the expected x-ray number Nx=10**7 with a
reasonable confidence.
This study also indicated a possibility of improvement of the results
by better focusing of the electron beam. This suggestion to focus
the e-beam to sigma 40 microns by using an additional quad close to
the Compton cell made by Vitaly was implemented in a matter of a few
days thanks to John’s support.
Week of September 27
The first day was quad installation into the beamline. In the second,
the last day of the Compton run we demonstrated that the e-beam is indeed
sigma 40 microns. This immediately resulted in a significant increase
of the Compton signal to 2 V (from 0.8 V the last week). Note also,
that this week result is obtained at the twice less charge per bunch
and at the reduced bunch length (3.5 ps). Scan of the interaction region
by transverse steering of the e-beam confirms that the laser waist radius
is about 50 microns (at FWHM). For these new conditions, the recalculated
expected Nx=2x10**7
Let us now estimate the generated photon number based on the empirical
results.
- Using the detector calibration made by Peter Siddons, the 2 V signal
corresponds to 4x106 photons in the spectral range 5-6.5
keV (see below regarding the spectral range). This is two times higher
then a cross-calibration of the detector done with a 7 keV CW signal
in the NSLS x-ray beamline. Application of the CW calibration to the
dynamic response for the pulsed regime is made assuming the detector
time constant that is not very accurately defined. Using the average
between these two calibrations, we get 3x10**6 photons on the detector.
- Using foil filters we demonstrate that the detected photons are
in the spectral range of 5-6.5 keV. This agrees with the calculated
cut-off 4.6 keV based on the acceptance angle of the detector.
- The air (20 cm) and Be window (250 um) transmission within this
spectral range is ~50% (averaged). Taking into account this attenuation,
the detected photon number, is 6x10**6 (1.7x10**18 photons/sec).
- We do not have yet a uniform understanding what is the proportion
of x-rays in the spectral range of 5-6.5 keV to the total Compton
spectrum. According to one analysis we are missing about a factor
of two to compare with predictions, another analysis gives an agreement
with the analytical prediction. Work is in progress to resolve this
issue before the Symposium on Laser-Electron Interactions, Tokyo,
October 11-15, 1999.
Concluding remarks:
- Let us compare our results with the published LBNL result. LBNL
used a 100 fs terawatt solid state laser in the 90 degrees interaction
geometry and obtained 5x10**4 photons at 30 keV. The x-ray pulse duration
defined by the e-beam transverse transition time was 300 fs. Thus,
the corresponding intensity is 1.7x10**17 photons/second. The higher
yield demonstrated at the ATF are due to several factors:
- the ATF high brightness electron beam
- using a CO2 laser that carries 10 times more photons
then the 1 micron laser pulse of the same energy.
- using 180 degrees geometry that permits interaction over
the Rayleigh distance which is r / Lambda longer then
r-the 90 degrees interaction distance. That permits the electrons
and photons to stay in interaction longer time producing more scattering
events
- note that the 180 degrees geometry allows potentially
the shortest x-ray pulse duration defined by the electron bunch length.
With the forthcoming bunch compression at the ATF, bunch length under
100 fs should be possible. In the 900 geometry, the x-ray
pulse is defined by a square root average of the laser pulse and the
electron beam transit time (min 300 ps for the 100 microns electron
beam focus).
- Based on our result, we can make some extrapolation. For example
if we produce and bring to interaction 30 J, 30 ps laser pulse we
can expect 1000 times stronger signal (just compare with the current
15 mJ). If we further improve the focus and use 30 fs bunches with
1 kA current, we can achieve 1022-1023 photons/second,
etc. These goals will be subjects of future Compton runs.
- The Compton run 1999 is finished. The results are still
being processed and will be presented in later reports. Thanks everybody!
- Back to Top
STELLA
Experiment (Reported by Wayne Kimura) |
Data reduction and comparison with the model of the ICA results
from the Aug-99 run continues. The results are very good and demonstrate
the best agreement between the data and model that has ever been
observed with the ICA system.
Attached are two figures showing a comparison between the data and
model of the electron energy spectrum with the laser off,
October_1_STELLA_Laser_off
and the laser on,
October_1_STELLA_Laser_on.
The solid line is the data and the histogram is the model predictions
for 10,000 electrons. For the laser-off plots both the data and model
have been normalized to unity. For the laser-on plots, the decrease
in the magnitude of the plots has been preserved relative to the laser-off
cases. This allows both the overall shape and magnitude of the laser-on
data to be compared with the model predictions.
The experimental conditions during the run were as follows: E = 45
MeV, normalized rms emittance = 0.8 mm-mrad, intrinsic energy spread
(1sigma) = 0.28%, e-beam average focus radius in the gas cell = 65 micron,
and hydrogen gas pressure = 1597 Torr. The model used the average laser
beam size that was measured coming into the gas cell and also included
all scattering effects from the gas cell windows and gas.
The histogram shown in the laser-on figure is for a laser peak power
of 45 MW. This is 2-3 times smaller than what we measured entering the
gas cell. (Analysis is continuing to determine a better estimate for
the laser power entering the cell.) This seems to indicate that not
all the laser power is being effectively used to accelerate/decelerate
the electrons. We believe a likely reason for this is because the radially
polarized laser beam quality needs to be improved. For example, the
laser beam may be substantially less than 100% radially polarized, e.g.,
perhaps ~50%.
On the positive side, the very good agreement in overall shape and
magnitude between the laser-on data and model also implies that the
electrons are being accelerated/decelerated well with the available
useful laser power. Earlier ICA data displayed Gaussian-shaped spectra,
which indicated that many of the electrons were not being effectively
accelerated/decelerated. This improvement in the shape of the spectra
we believe is due to the small e-beam focus inside the ICA interaction
region made possible by the very low emittance of the e-beam and the
upgraded beamline optics. The 65-micron (1sigma) e-beam focus now fits
well within the 200-micron (FWHM) laser beam focus. (In the earlier
ICA runs, the e-beam focus was 500-micron diameter.)
This new spectrometer data is also noteworthy because the entire laser-on
spectrum was obtained in a single shot. This greatly facilitated the
process of comparing the data with the model. Previously, with the old
spectrometer it was necessary to splice together multiple shots, which
introduced a great deal of uncertainty caused by shot-to-shot fluctuations
in the e-beam and laser beam characteristics.
A paper that reports our latest ICA results has been submitted for
publication in the IEEE Transitions on Plasma Science Special Issue
on Second Generation Plasma and Laser Accelerators.
In other progress, the new permanent magnet wiggler for the IFEL has
been assembled and is undergoing magnetic tuning. Most of the parts
for the lifting system, which will allow the wiggler to be extracted
away from the beamline when not in use, have been fabricated or delivered.
Once the wiggler has been tuned, the entire system will be assembled
at STI. It will then be sent to BNL for surveying and installation on
Beamline #1. Assuming no unforeseen problems, this should occur during
the second week of October.
Back to Top
VISA Diagnostics (reported by Alex Murokh):
Last Monday, Alex and Vitaly conducted an experiment to test a YAG-crystal
applicability as a VISA beam profile monitor. For this purpose they
constructed a target, which included phosphor screen, mirror for OTR,
thin wire and two YAGs of different thickness. The target was positioned
at the location of a Compton experiment at the ATF, and they were able
to use the existing optical set-up to measure the beam spot size with
the various diagnostics.
The results are rather disappointing: YAGs do not perform as well as
hoped. For example, the following data have been collected with virtually
the same beam. See the data in the table and images on the file
October_1_VISA_YAG_test
| Diagnostic |
OTR |
Wire scan |
Phosphor |
1/4-mm YAG |
1/2-mm YAG |
| Beam HWHM [microns] |
84 |
102 |
166 |
186 |
207 |
The nature of image blooming on the YAGs is not obvious. We have considered
three possible explanations:
- Physical saturation of the photon generation mechanism;
- Multiple reflections and light scattering over imperfections;
(The last two may not be a problem for VISA, where the acceptance angle
is smaller by the order of magnitude)
In the nearest future we would like to run another round of experiment,
in order to establish the true cause of the observed
phenomena.
VISA Wiggler (Reported by George Rakowsky and Lowell Klaisner):
The two VISA sections that had been sent back to SLAC for straightening
(#4 and #2) returned to BNL. Upon inspection, the straightened Section
#4 showed a bow again, not as bad as before, but still unacceptable.
Ben Poling assured us that he had straightened it to <0.0035".
Section #2 looks OK. We conclude that there must be a built-in strain
in the strongback or the spacer bar. The sandwich is bolted together
and maintains its shape by friction. Shipping it cross-country by FedEx
exposes it to shocks from rough handling, as well as possibly large
temperature swings in the cargo hold. (The magnets are well insulated
in many layers of bubble-wrap and peanuts, so that is less likely to
cause creep.) Section #4 was shipped back to SLAC at Ben Poling’s and
Max’s suggestion. Ben will try straightening it again. In addition,
both wire finders (shipped at the same time) developed a severe wobble
in their precision translation stages, with up to 0.025" motion.
It is impossible to get a meaningful calibration now. The stages are
no longer parallel and rock in their bearings. It is very likely that
these devices were also damaged in shipment. The large mass of the laser/detector
assembly with its long moment arm could place a large torque on the
bearings under high-g loading. There were more calls to Ben and Brendan
at SLAC. Brendan has a spare precision stage which he will check and
send to us. New units were ordered.
Robert Ruland was here last week, attending the International magnetic
Measurement Workshop (IMMW-11). On Thursday, Sept. 24, Rakowsky and
Ruland gave a two-part talk on "Trajectory Straightening, Fiducialization
and Alignment of the Strong-Focusing VISA Undulator Using Pulsed Wire
and Interferometric Techniques". In the afternoon we hosted a lab
tour, with stops in the Magnetic Measurement Laboratory and the VISA
pre-assembly area. There were approximately 45 attendees and all visited
our show-and-tell. We had set up Section #1 in the chamber, with both
interferometers in place. We also set up Sections 2 and 3 on the pulsed
wire bench and gave a "live" trajectory tuning demo. We got
very positive feedback. This is the good news.
This week, the VISA undulator section #4 was measured and was bowed
with a 1 mm sagitta. Ben Poling loosened the bolts holding the strong
backs to the magnet support rails. The unit relaxed to straight with
a 35 micron straightness as measured on the CMM. This eliminates some
theories of magnetic forces creating a conditionally stable mechanical
structure. There is no evidence of "oil canning", i.e. some
sort of non-linear force that wants to force the structure into a curve.
It appears that the unit was bent (by an unknown mechanism - probably
dropped in transit) and was held in that bent condition by the static
friction between the strong backs and the magnet support rails. When
that static friction was released the unit returned to its original
condition. The individual magnets may have moved during the impact or
the bending but we will have to wait for magnetic measurements to confirm
that. We plan to store this in a "clean" room with an air
conditioner and see if there are any deformations with temperature.
None are expected. I plan to measure the spring constant of the structure
to estimate the force required to cause the bending that has been observed.
We will contact a local company that specializes in shipping delicate
equipment and arrange shipment early next week. No reason why the other
units should deform if handled carefully. Once in the vacuum chamber
they should be fine.
Back to Top
Computer
and Control System (Reported by Bob Malone) |
As part of ATF’s control system upgrade, we have been extensively
testing Vista’s Vsystem software tools operating under Linux on
Intel PC hardware. Over the past several weeks, a number of small
test applications have been developed to exercise both database
generation/access and the Vsystem applications program interface
(API). These test programs have also helped to uncover potential
problems which will have to be addressed during final porting of
our present VAX-based system. Thus far, the results have been very
positive:
1) Virtually all of the present ATF databases should port to the
new system without problems. Most database porting issues can be
resolved at the VAX end, generating a new database which is ready
for Linux. Tests show that once these minor changes are made on
the VAX, the resulting database can be processed under Linux with
no additional modifications. The time needed to generate a database
under Vsystem on VMS was ~ O(n^2) but has been improved under the
Linux release to ~ O(n). (n=number of channels in database)
2) Most all parameters to the API calls remain the same when moving
from VMS to Linux. Some problems do occur, for example, with character
strings which will now have to be passed by descriptor. Details
such as this can be hidden by rewriting parts of the present VMS
"intercept" library which traps all API calls and translates
arguments. The intercept library has always been part of the ATF
design, having been implemented almost 10 years ago with with express
purpose of easing system migration. Design of a new intercept library
is underway.
3) Stability: As of today (October 1, 1999) the test system has
been operating non-stop for 41 days without any errors. While we
have been testing only software operations (no hardware data acquisition)
the Linux/Vsystem combination appears to be quite stable and reliable.
Final conclusions, however, can’t be made until we purchase new
Ethernet crate controllers and subject the system to intense loading
by hardware data acquisition and software applications. Also, hardware
interrupt latencies still need to be measured and evaluated.
Back to Top
Technical
Operations (Reported by Bill Cahill) |
The FD5 magnet transfer switch installed two weeks ago suffered
dearly from rf modulator noise. After many days, Marc was able to
isolate the system and it now appears to be working. The gun modulator
1 kw amplifier was retuned and tested and should be putting out
more power. Unfortunately, when the system was brought up on Friday,
the output was down considerably. It appears that we lost a tube
in the drive stage. The linac modulator Spellman high voltage power
supply tripped on over-temperature. The cooling fan had failed.
The fan was replaced and the system in now operational. Four target
assemblies, a pinhole target assembly were completed for the Compton
experiment. The lens and cube assembly was completed for VISA and
is now in pre-survey. Ongoing work in progress by the NSLS interlocks
group (Scott Buda) for the interlock system of the terawatt laser.
A problem in the rf gun modulator 1 kw power supply that has eluded
us for a while has been found. An intermittence in the 3rd
stage filament power supply was causing the output power to drop off
radically, causing a loss of beam. Extremely difficult to find. The
unit was repaired, retuned and is now running to specifications. A capacitor
in the gun modulator had failed. This capacitor is of the "new"
batch from Maxwell Corp. that had only a few months of service. An investigation
of another capacitor company is in progress to find a more reliable
product. The 3 YAG crystal replacement foils requested by X.J. Wang
for beamline 3 are complete.
Back to Top
