Some comments on Chapter 4 -Photon Beam- from Version 2 design (August 19/99)
Normally bremsstrahlung taggers are built such, that the tagging efficiency can be chosen to be large, i.e. that only moderate collimation to angles even bigger than the characteristic angle is needed. For this case it is reasonable to build the electron detectors long enough, so they see the electrons efficiently.
In the case of the tagger proposed for HallD this is different. There you will have to use narrow collimation for the photon beam. Ordinary bremsstrahlung roughly follows two body kinematics. Thus, when collimating the photons the electrons are collimated virtually as well, which is even more valid for coherent bremsstrahlung. If one narrowly collimates the photons the corresponding electrons are also emitted at small angles. The optics of the planned tagger is focusing in the bend plane and defocusing perpendicular to the bend plane. Then electrons that come at large angles perpendicular to the bend plane have no correlated photons in the collimated beam, so they can be suppressed by using short detectors.
It is planned to run the tagger at very high rates, 10 MHz per detector, which can be lowered by making the detectors shorter. There is no need to consider an efficiency for un-collimated bremsstrahlung. As mentioned above these considerations hold even better for coherent bremsstrahlung. To make the detectors shorter corresponds to a horizontal slit collimator in the photon beam.
This is a very good idea. I first proposed this idea to the collaboration about two years ago, not only as a means of improving the peak tagging efficiency (very important at high rates) but also to increase the linear polarization of the beam (see below. I was not able to generate much interest in pursuing this because people thought it looked difficult to do with scintillators. Nevertheless we (Dan Sober and myself) went ahead and prepared a whole section to go into the DRv3 on how to do this, but the collaboration decided to remove it until it could be studied further. The problem can be seen from Table 4.6 in [2]. Look at the part of the table with the quadrapole, since the quad was put in to give sufficient resolution at the focal plane of vertical scattering angle. For more details on the difference it makes to have the quad there, see Fig. 1.
The last two columns show the vertical resolution and angular dispersion at the focal plane. One characteristic angle is only about 1mm wide, so the image of the photon collimator casts a stripe down the electron focal plane that is ±0.35mm r.m.s. That is a thin piece of scintillator and looks like it might not be easy to keep aligned with the beam. The idea we were working on was not a single "short" scintillator, but a whole stack of them, maybe 5 per energy channel of 0.5mm thickness each, staggered back in the direction of the flight of the electron. That way one could "find" the collimator stripe by looking at coincidence rates between each focal-plane tube and the post-collimator photon counter. Our collaborators were not convinced that this is yet ready for the Design Report, so it has been put on the back burner, pending a study. It is on my list.
There is no need for a thin exit window at the tagger for the electrons. The lowest energy electrons will have roughly 500 MeV. An aluminium window of 2 mm thickness would result in a multiple scattering angle of roughly 6 mrad, which if the detectors axe mounted as far as 10 cm downstream from the window would result in a smearing of the electron position by less than 1 mm, which certainly can be tolerated compared with the 13 mm detector width.
Yes, Dan Sober agrees with you. We do not have to use a window as thin as was used in Hall B. If we do use the "short counter" option discussed above, however, 1mm from scattering in the exit window is probably too much. Anyway, right now our design does not use this short-counter option, so we modified the design report to reflect your comment.
In subsection 4.2.2. it says that most of the intensity of the photon beam is confined to forward angles within m/E. This is not true, there is less than 50% of the photons emitted into this forward domain. But the angle is small and the photons are thus strongly forward peaked.
Yes, most does mean more than 50% doesn't it. The correct number is less than 50%, but not very much less. Next time I will reword it to say that "most of the photons are emitted in a forward peak that has a characteristic width of m/E" or something like that.
The design gives a lot of space for the photon collimation. You propose three colli- mators and two sweeping magnets in-between. I think this is a very good design. I want to propose to consider additional measures. One should install photon beam monitors along the 80 m flight path (which should of course be in vacuum - rough vacuum would be good enough). So, most of the intensity would be pealed off the beam far away from the main detector. A simple beam position monitor which detects the part of the photon beam to be absorbed in the collimators might also be of help (e.g. ionization chambers at either side of the beam, which are run with opposite voltages and their signal simply added).
My primary concern with pre-collimation is the effect it will have on the position feedback system. Because the collimator hole is 3.4mm at 80m, keeping the beam centered on that hole will require an active feedback system that runs continuously when the beam is on. This requires that we have a segmented shower counter exactly aligned with the primary (i.e. defining collimator). The easiest way to do that is to have the shower counter built into the front portion of the collimator, that is, an active collimator. What you want is that the segments of the shower counter see an unshadowed image of the photon beam, except for the part that goes down the hole. That way count differences tell you which way to steer the beam. But if you have a pre-collimator that bites into the tails of the beam at all then the shower counter will see is the shadow of the precol. Of course you could make the precol diameter very large so that it doesn't matter, but then what is the point? This is why we have designed the collimator system around the rule, keep the alignment simple: defining collimator is active, and primary collimator is first.
From our experience with coherent bremsstrahlung it is essential that the collimated photon spectrum is known. This will be needed to determine the degree of linear polarization. Therefore, I would advise you to use the second sweeping magnet as a simple pair spectrometer. All it needs to do is detect positrons and electrons in coincidence. That is then an indicator for a photon, the energy of which is given by the tagger. The efficiency of such an instrument needs to be low of course, that will be done by the choice of pair target thickness. The leptons produced from this target would be swept out of the beam by the spectrometer.
This sounds like an interesting idea. I had thought that we would want to have the pair spectrometer at the back of the Hall, downstream of the detector. That was just because the collimator hut will be very full of background neutrons and low-energy photons. This is where your experience would be valuable to give an idea of what you kind of an environment it would be in an area a few meters downstream from what amounts to a beam dump for a 12GeV bremsstrahlung beam.
In my last comment I have proposed to collimate the electrons in the tagger by using 'short' scintillators. This will, however, give a small component of linear polariza- tion to the incoherent bremsstrahlung photons. In the geometry you design your tagger this polarization would be horizontal, so it might add to or subtract from the coherent component. If I believe a calculation that I once made (1982), then for the case k/Eo=0.5 (fractional photon energy) one might get 20% of linear polarization if the electrons are collimated to two characteristic angles (which is very narrow). I think this needs to be investigated.
Yes, in fact this was the major reason I got interested in it a while back. I noticed that CB has an azimuthal dependence to its polarization, and thought that putting the beam off-center on the collimator could enhance our polarization. That would kill our tagging efficiency, unless we could recover it using out-of-plane tagging. So that is the idea. As I said above, it needs a serious study and probably some R&D. But it sounds to me like you think this is an interesting idea. Would you be interested in discussing this further with me?
We can communicate about off-axis-bremsstrahlung, but certainly I do not advocate for using that technique. But when the electrons get collimated, even if the collimator is symmetric to the beam axis, linear polarization can result. I once made calculations in this context, old stuff, if I don't forget I might carry a copy (in the old days formulas were hand written) to JLab and leave it there for you.
OK, maybe I haven't understood you. For the coherent part, there is just one value for q (nuclear recoil), which means the photon-electron system obey 2-body kinematics. Measuring the angle of one measures the angle of the other. These are not two independent variables.
You have a choice where you do the collimation. If you leave the photon collimator centered, but it is large enough, then you get effectively off-axis collimation by selecting up-going or down-going electrons. But if the photon collimator is very small (say 0.5m/E) then there are no coincidence electrons up-going or down-going because you collimated them away at the photon collimator. I assume we are discussing the coherent component.
We agree about coherent bremsstrahlung. Strictly speaking there we have three body kinematics as well: the photon, the electron, and the virtual photon. But the crystal gives us complete information about the latter, then the electron and the photon are strictly kinematically correlated.
In bremsstrahlung we have a minimum momentum transfer, close to which the cross section is maximum. Now we look at incoherent (ordinary) bremsstrahlung. We take the case of narrow on axis collimation. i.e. in the extreme the photon goes in forward direction (theta=0). Where are then the electrons under the condition of minimal momentum transfer (the value of which is fixed, but not its direction)? They are not in forward direction but at a small angle with respect to the photon (equally distributed in the azimuth), i.e. the minimum momentum transfer has a component perpendicular to the photon. If now we collimate the electrons, i.e. by cutting electrons that go up and down off, then the electrons that go left and right remain (they are focused by the tagger) and thus we have enriched the momentum transfers that go left and right. A linear polarization in this direction will result.
It needs to be calculated how big that effect is, but in principle it is there. It will always produce the same degree and direction of polarization independent of crystal orientation. If someone plans for real off axis collimation, then both the electron and the photon need to be collimated at opposite directions in order to get the biggest degree of polarization. But this is not what I was talking about.
I brought up the topic in order to give a warning: If really we collimate the electrons by using scintillators of low height, we might get as a side effect linear polarization in the incoherent part of bremsstrahlung. You want to use a detector which covers the full azimuth. Then you would get different total degrees of polarization in the vertical and the horizontal polarization, which you make vertical and horizontal by changing the crystal alignment.
OK, understood. We agree. Whatever final design we adopt for the focal plane microscope, we will have to take into account any extra linear polarization in the peak coming from the incoherent component due to vertical misalignment.
I do not understand why you want to build the tagger such, that you can only tag down to fractional photon energies of 0.5 . From what is written in chapter 3 there are objects of interest predicted for energies above 1.5 GeV. You could get substantially higher degrees of linear polarization if you would tag photons at lower energies. Why has this k/Eo=0.5 been chosen as a lower limit? Of course money might be a reason, but then you should a least make precautions that you can use the tagger also for smaller tagged photon energies.
Hall D is being proposed as a single (ambitious) experiment. It will not be a facility, like the other JLab halls, where different groups propose experiments and get allocated beam time for various physics. The model is that of a RHIC or HEP experiment, where the running is determined by a single physics objective. Whatever other physics can be extracted from the data, that is completely open for all of us to explore, but not the running conditions. To do this physics you want to chose a single photon energy, go as high up as possible without quenching the polarization, and then run under fixed conditions. The way that PWA works is that you have to analyze the whole spectrum before you can find anything. This is not like bump hunting. The signals we are looking for will probably be hidden underneath broad and overlapping structures from known objects. You have to cover enough of the mass range to know you have a complete description of the known objects. Then you look for something left over. That leftover stuff has to possess a phase shift that you measure from well below the resonance to well above the resonance. Ideally to look for a state at 1.5GeV you would like data all the way up to 2GeV and beyond.
To produce final states up to 2GeV it is not sufficient to have photon energies over the 2GeV threshold. These cross sections turn on slowly, and at threshold the cross section is zero. A more meaningful measure of the photon energy required to produce a meson of 2GeV is to ask the question, what photon energy is sufficient to produce a 2GeV meson AND give no more than 200MeV recoil to the proton? This is more formally expressed through the parameter tmin that is discussed in the Design Report.
Presently you want to tag photons always in a limited range of energies, just where the coherent peak sits. If then you build a tagger with constant dispersion (which your tagger practically has for high energy electrons) you can use a relatively short ladder of electron detectors (at least to begin with, as a cheap solution) which you position where you need it.
Yes, we never intended to cover the whole focal plane at the resolution required for the coherent peak. I question whether it would need to be movable, but some of our colleagues who are viewing this more as a general facility want to build in more flexibility. I am certainly not opposed provided that it doesn't introduce to much cost or complexity. I think that in making the microscope movable, neither one of these is a concern.
It is important that you always maintain the option of using also circularly polarized photons and double polarization experiments. I think in the end you will be happy about every single observable that you can use. The whole proposal points at finding glueballs, exotics etc. which up to now have no experimental basis of existence, but have been predicted by LGT. If, as a result of your efforts, you will not find any of these objects - wouldn't that be an extremely important result?
This is a very important question. I agree with you that a high-statistics experiment that can explain the entire spectrum in terms of quark model states can make a very strong physics statement. There are difficulties however in convincing people that we can "explain the entire spectrum" in this way. Usually in a PWA you have to put in some kind of continuum to account for backgrounds, in addition to the resonances. The question then arises whether the searched-for states are hidden in that background or are really absent. Controlling these backgrounds is strongly related to three things: hermetic detector (large, known acceptance), large statistics so you can decompose the large number of smaller channels and not have to lump them together, and the full mass coverage that I spoke of in the preceding section.
Crystal Barrel, some might say for the first time, proved that such a background-free analysis is possible. However, they were limited on the last point by the total energy available in pbar/p at rest. The point here is that one is always obliged to guess at whatever the tails of what higher in mass are doing to the high-mass end of the spectrum. Having very large statistics at a single fixed photon energy that allows full mass coverage up to the charm threshold will allow a comprehensive analysis that will have the best chance possible of a background-free analysis.