Present:
Chris Allgower,
Andy Bacher,
Chris Lavelle,
Hermann Nann,
John Olmsted,
Tom Rinckel, and
Ed Stephenson
SUMMARY OF THE RUN:
There were two major objectives:
Operating parameters for the channel magnets were selected by scaling from the 3-He values. The momenta used were obtained from an energy loss model of transport through the channel. The upscaling of the quadrpole values led to currents several percent beyond the capacity of one of the supplies. So all three quadrupoles were scaled down far enough to permit operation.
Early on, it was clear that a trigger with just a coincidence among delta-E1, delta-E2 and E detectors would produce very large rates. By including the two veto detectors, these rates became easily manageable. We ran in this condition for the rest of the time. We considered, but did not use, a trigger that included hits in the Pb-glass, preferring to wait until the performance of these detectors is better understood.
Before the run, Bill Lozowski spent about two weeks developing and making thinner stripper foils for the beam injection into the CIS ring. He reduced the thickness from 3.6 micro-g/cm^2 to about 2.5 micro-g/cm^2. Two such foils were tried. The improvement in beam intensity seems to go as the reciprocal of thickness. [Since, Bill had produced a 1.7 micro-g/cm^2 foil.] CIS was also retuned to transfer beam to the Cooler at 110 MeV, a rigidity closer to that of the 200-MeV proton beam. Past tests suggest that this leads to an improvement of about 40%. No such comparison was made here as one needs Cooler tunes at both transfer energies. The debuncher was tried, but no improvement to the intensity found. No large attempt was made to optimize the operation of CIS.
Considerable tune-up time was spent at the start of the run with Terry Sloan and Bob Pollock working to optimize the tune of the Cooler. They improved the injection efficiency, localized the effects of the flattop combos at the top of the ramp, opened up the ring aperture, and got good efficiency on the ramp. This led to typical starting Cooler currents of 800-900 micro-Amps. Their tune required 15 shots to fill the ring. With pileup an issue, we looked into other ways to operate the Cooler. By turning off the RF, the beam will eventually distribute itself evenly about the ring. This led to short lifetimes because particles that were slowed down due to interactions with gas or the target could not be recaptured with the electron cooling alone. So we chose an option to rebunch on harmonic 6. The benefit of this needs to be evaluated.
Position scans were made using rates in the channel to locate good overlap with the target. At large pushing pressures, the lifetime of the beam becomes a good indicator of overlap. It was found that the gas jet target is about 6 mm to the outside of the ring from the best position of the beam following the initial tuning.
The initial setup had the Pb-glass detectors rolled back so that the full complement of scintillators needed for the luminosity calibration could be installed. So the running time was divided cleanly between the two major goals.
There was sufficient energy resolution from the new 45-degree scintillators that it was possible to utilize all of the features that were designed into the system. Each side had two tapered detectors with the tapers running in opposite directions horizontally. The sum of the pulse heights gave a good separation between protons and deuterons, although there was a large sea of signals of other energies. By selecting on the deuteron bands for the left and right systems and taking linear combinations of the position readouts, one could plot scattering angle in this sytem and position along the beam axis. The latter showed a clear peak associated with the position of the target, although the feeling was that its width was dominated by the energy resolution of the detectors.
When another detector is added at 25 degrees, it is possible to see bands associated with d+p scattering. For a deuteron beam on a deuteron target, this band is intense, suggesting that the rate is dominated by quasi-elastic scattering that breaks up the target. Since this is the calibration process, we will need to be able to subtract this contribution when HD gas is run. So the calibration will have to also include a comparison to d+d alone.
Much effort went into checking geometry and other things to determine why the rates for the left and right 25-degree systems differed by some 40%. Eventually, it was found that the ion source had malfunctioned, and both tensor polarized states had acquired a large vector polarization of the same sign. When this was eliminated, the differences were under 10%. In the process, it was realized that small geometry problems abound with these scintillators, and that some additional control is needed over the positions of these detectors.
This time the forward luminosity system was repositioned at an aximuthal angle of 24 degrees out of the horizontal plane. The silicon detector was moved, and the mount for the two forward scintillators redesigned. When the forward detectors were mounted at an angle of at least 4 degrees, there was enough recoil energy to trigger the silicon counter. All spectra in this sytem were clean. Good position spectra were obtained from the silicon detector. In this system, the phi acceptance is governed by the slits over the silicon detector. Theta is controlled by the position of the forward counters. Absorbers were used so that the deuterons would stop in the second detector, leaving a clear peak. Without absorbers, breakup protons have the same velocity as deuterons and are indistinguishable in "delta-E" scintillation detectors.
A number of times we switched to hydrogen gas to see whether the signals from d+p scattering would be distinguishable from whatever we see with d+d. For the forward system, the change to protons gives a factor of two increase in the pulse height in the silicon detector. While a clean signal is seen for the 25-45 degree scintillator system, the problem with quasi-elastic remains.
Using published d+p elastic cross sections from 146 MeV (laboratory proton energy), we attempted to calculate luminosities during the run. The 25-45 detectors gave values of 1.5 x 10^31 /cm^2/s, about half of what we had included as our goal in the proposal. Since the beam current was also more than a factor of two less, we concluded that the target thickness was at least 2.5 x 10^15 /cm^2. The forward luminosity system gave a value of 0.5 x 10^31 /cm^2/s, a factor of three less, even after a crude correction was applied for reaction losses in the copper absorbers. Some geometry checks were made to make sure that this system was reasonably well aligned. With much more sensitive alignment issues, and losses from reactions and multiple scattering, it is more difficult to see how we might get a good normalization from this system.
For the last few days of the run, we optimized the data acquisition rate as a function of the beam lifetime (controlled through the target thickness) and the time on flattop. At this optimal point, about half of the beam is used during each cycle. One problem that was found and fixed was the loss of timing signals from delta-E2(B) and E(A). These are neighboring channels on the relevant TFC-FERA system. The delay cable was replaced, which helped, and offending channels moved to spare inputs.
It was noted that the Pb-glass that were closest to the 6-degree magnet were suffering gain losses compared to their condition before the run. These performances are monitored online with a trigger on cosmic rays. Several attempts were made during the run to add sheets of mu-metal in various places. None worked. For this run, we increased the voltages so that, in most cases, useful signals could be obtained anyway. Such gain reductions were found also for the 25- degree detectors used in the luminosity system.
The missing mass algorithm that was developed for the p+d -> 3He+pi0 data was moved intact to the online situation, once it was realized that the absolute time differences were unchanged despite a reworking of the trigger system. In this algorithm, the change from one reaction to another is determined by changes to the masses of the particles involved and the energy loss table used to reconstruct the alpha particle momentum from time of flight. An attempt was also made to select on that portion of the channel scintillator spectra where 4-He should appear. This was not entirely successful, as noted below.
At the end of the run we switched from harmonic 6 to 1 and noted that the lifetime of the beam increased. This means that we should reoptimize against beam lifetime and target thickness if we decide to operate in this mode. It clearly has greater cooling capacity using the RF to capture longitudinal momentum variations. Moving from harmonic 6 to 1 may increase problems with pileup.
During the run we had two instances in which the shutdown of a helium compressor led to a loss of good vacuum. For one of the these, the problem was traced to a leak in one of the helium cooling lines. The line was replaced. Once operation for a data sample was underway, we found that we needed to regenerate the stage-2 cryopumps roughly once per day. These were partial warm-ups, intended to drive off hydrogen and air only. Stage-3 pumps needed to be cycled every two days.
EARLY REPLAY RESULTS
Ed Stephenson started the meeting by reviewing the outcome of the data taking phase of the run. He used all the runs since the repair of the TDC inputs for delta-E2(B) and E(A). This gave three days of runs (287 through 307).
By gating on missing mass above 130 MeV and Pb-glass multiplicity of 2 or larger, the scintillator spectra showed a clear group that could be associated with 4-He in the channel. This group was located at higher delta-E1 and lower delta-E2 pulse height that had been expected during the run. Windows set during the run collected few, if any, 4-He nuclei. The rates are low, about 70/hour. The missing mass shape follows the trend expected for uniform illumination of the channel acceptance in position and momentum (roughly the same shape as expected for double radiative capture). In the final replay, we will need to select that subset of events associated with either one or two 68-MeV gammas in the Pb-glass array. For now, this tool doesn't exist, so cuts were made instead on Pb-glass multiplicity. Cuts on higher multiplicity result in fewer events but no change in missing mass shape. If there are any alpha-pi0 candidates, there number is likely to be less than about 10 and in the present analysis indistinguishable from double radiative capture events.
The low alpha-pi0 event rate suggests that in future running we plan on long (up to two months) of production time and, if it gives greater luminosity, running with unpolarized beam.
One feature of the XY spectra associated with 4-He is that they are off-center horizontally at the end of the channel, a condition that can be corrected by making adjustments to the septum magnet current.
John Olmsted has been checking on a number of data issues.
The delays were set in the electronics so that the E(B) detector would be the start time for all time measurements. For some events in delta-E1(B), that detector was the start time. This becomes evident as a corner in a time (delta-E1) versus time (E) plot. The events whose time of flight lies in the correct range come from the portion of the data where the start time is E(B), as planned.
It is possible to reconstruct x-position from the fall-off of pulse heights as the source of the signals in the E-detector moves away from one of the other PMT. When compared against X3, the MWPC directly in front, a correlation plot shows a kink in the middle. Delta-E2 shows no such kink. Checking on the source of events when compared to other displays gave no clue on the origin of this feature.
Gating on proton and deuteron events in the E by delta-E2 display led to the identification of those groups in delta-E1 spectra and position spectra. Deuterons are concentrated around the outside edge of the first XY display, indicating that they are present in the channel only after having first lost energy by passing through the exit flange on the 6-degree magnet. Protons are more generally distributed.
It appears that during the last part of the run, the channel TFC began to lose events from delta-E1(A). This starts in run 295 and almost completely eliminates that detector by run 303. This problem is similar to one encountered during the run on the channels that carry delta-E2(B) and E(A). The TFC sits at the end of a 150-foot ribbon cable, so it is possible that signal losses make things marginal. Unfortunately, this is the TFC that has been calibrated to yield missing mass with the time differences.
SHUTDOWN TASKS
SOFTWARE TASKS:
For the longer term, there are a number of software tasks that will be needed to provide the tools to extract alpha+pi0 events. In the analysis described by Ed, there are several important things that could be done to enhance the quality of the result:
So we have the following list, in more or less priority order:
Compare the amount of pileup for the harmonic 1 and 6 runs. This way we can decide whether the gains are worth the loss in beam to run on harmonic 6.
We eventually need to make reliable cuts on the presence of one photon, or two photons on the two sides. The first task is photon identification and characterization.
This consists of dealing with several issues:
Check the parametrization of missing mass to see whether any of the input coefficients need to be scanned or changed. Some work is needed to determine whether other particle groups such as protons can be used as markers of the correct settings. Otherwise controlled scans may be needed to see if an alpha-pi0 peak emerges.
There are some questions here:
Eventually we need to be able to assess the number of events lost due to various mechanisms. First, we need to study the distribution of alpha events in the system and see whether edges of delta-E2, for example, are an issue. There need to be estimates of other loss mechanisms, including nuclear reactions with the materials in the channel, pileup in pulse height signals, pileup in the veto leading to lost triggers, multiple scattering, etc.
There are also some problems that appeared during the run. John's earlier report referred to some of these. There were many questions about which features in the pulse height, position, and time spectra were associated with each other. It also appears in the energy by delta-E2 plot that some fraction of the deuterons that come through the system do not trigger the veto detectors, even though these are supposed to be larger in area than the delta-E and E. This could be related to some effect associated with their extreme geometry in the system, originating with their passing through the exit flange from the 6-degree magnet.