Reply and Comments by SiD
Q1(B) | Q2(geo) | Q3(lowQ) | Q4(max.bkg) | Q5(bkg.prms) | Q6(L*) | Q7(VTX_R) | Q8(2&20mr) | Q9(headon) | Q10(mini-veto) | Q11(bkg@2,20mr) | Q12(DID) | Q13(anti-solenoid) | Q14(up/downstream Pol) | Q15(Z-pole ) | Q16(e-e-) | Q17(detector assembly) | Q18(exp.hall)
- What factors determine the strength and shape of the magnetic field in
your detector? Give a map of the field, at least on axis, covering
the region up to +-20 m from the IP. What flexibility do you have to
vary the features of this field map?
- We have 2-d and 3-d models that are reasonably easy to make modest modifications or extensions to; the 3-d model is very cpu-intensive to rerun so we prefer to iterate in the first instance on the 2-d model. We have provided an "on-axis" fieldmap from the 2-d model, and the same for the inner fieldmap from the 3-d model.
- The field is determined by a uniform solenoidal winding in an octagonal iron box, with sufficient linear current density to yield 5T at (0,0,0). No extra windings are added to enhance uniformity in the tracking volume. The field along the beam through the end iron may show the influence of the many (23) rather thin (10 cm) iron sheets and gaps (5 cm) we used. We assume the addition of "anti-solenoids" along the beam might require us to decrease/smooth the main solenoid field there by e.g. the addition of steel washers in the gaps of the end plates.
- For the main field in the detector volume, extra windings can be added, e.g. at the coil ends, to improve the uniformity of the tracking field, but we guess that unless a TPC is used, such refinement isn't needed.
- We have a 3-d model fieldmap also. One quarter of the model incorporates all rotational asymmetry of the iron yoke. (See Rich Smith ILCWS05 talk).
- We are working on incorporating a dipole ("DID") on the outside of the solenoid windings to provide "compensation" for a large crossing angle per Seryi and Parker. We hope to have this model well underway by Snowmass. We also hope to have some cost parameters for our base-line model.
- We have not yet looked at anti-solenoids (LCC note 142). We would guess it will be straghtforward to support the main solenoid against any forces they generate on it.
- Provide a GEANT (or equivalent) geometry description of the detector
components within 10 meters in z of the IP and within a radial
distance of 50 cm from the beamline.
- We have both GEANT3 and GEANT4 models and can provide them in any convenient requested format.
- Would you mind if the baseline bunch-spacing goes to ~150 ns instead
of ~300ns; with ~1/2 the standard luminosity per crossing and twice as
- The SiD detector technology that we have considered so far is all intrinsically fast on the scale of 150 ns, so that the issue of the 150 ns spacing really is an issue for the electronics. (Note that this distinction is ill defined for the vertex detector) The SiD electronics concept (so far) for non-very-forward systems involves measurement of the amplitude and time of signals as they occur, buffered up to four measurements per train. When the issue of 150 ns first came up, we changed the clock (and ADC) architecture to 13 bits, so we think that, unless the background per train were to go up by a large factor, we would not be concerned about the difference between 300 and 150ns bunch spacing.
- The very-forward detectors would measure every pulse. Again, given the primitive state of thinking, we donft believe we mind whether there are 3000 or 6000 buffers. Note that this design may have some relevance for the machine instrumentation.
- The vertex detector is most likely going to evolve from some CMOS like structure that does not involve shifting charge as in a CCD. Since the number of hit pixels per train would not change significantly, and 150 ns is slow compared to the logic times involved in these structures, it should not matter. Note that this conclusion is based on the rather minimal R&D that has been accomplished to date.
- For each of your critical sub-detectors, what is the upper limit you
can tolerate on the background hit rate per unit area per unit
time (or per bunch)? Which kind of background is worst for each of
these sub-detectors (SR, pairs, neutrons, muons, hadrons)?
- We would like to distinguish between backgrounds in normal operating conditions and machine eaccidentsf; we feel we may not have enough information on the latter at this stage.
- a) Silicon outer tracker
For silicon strip detectors, their operation in high radiation environments has been carefully studied. We don't anticipate that radiation damage will on its own be a problem for the outer tracker.
The real issues for backgrounds are in the areas of pileup and pattern recognition.
Pattern recognition is hard to do right, but it's probably a safe assumption that background hits won't pose a significant problem for the outer tracker unless the occupancy exceeds a few percent. For 1% occupancy in a 50um x 10cm detector with hit timing, we would require the flux to be < 0.2 particles / cm2 / bunch crossing. If fluxes are higher than that near the beam pipe, we might want to consider pixel detectors that have much smaller active areas.
The bigger issue is probably pileup. Given the triggerless design for SiD, each hit for an entire train must be recorded. The SLAC/Oregon groups are designing a readout chip with four analog buffers per channel. If we use something like this, we would probably need to keep background rates below 1 / channel / bunch train or 20 particles / cm2 / bunch train. Note that this is per bunch train, where the above number was per bunch. For 3000 bunches this would require a flux less than 0.01 particles / cm2 / beam crossing.
If the flux is much higher than this it may be necessary to explore different readout schemes. For example, one might digitize hits and use digital rather than analog storage. If a hit strip requires 20 bits for charge and time tagging, and you provide storage for 20 hits per bunch train for 2000 strips, the memory required would be 100 kB, which doesn't sound too hard. Thus, it seems like a digital storage architecture could eliminate pileup if that's a problem.
- b) EM calorimeter
For the SiD ECal, there is no damage/background issue from collisions as far as we can tell so far. We have been looking at the issue of damage due to an accelerator accident, and may be able to provide information for this soon.
- c) Hadronic calorimeter based on RPCs
Machine accidents should not be a problem for the RPCs themselves. If necessary we would need to consider radiation-hard electronics on the detector.
RPCs need 1 ms to recharge the ~ 1 cm2 area around the avalanche. So rates of the order of 100Hz/cm2 would result in unmanageable dead times. Based on preliminary PYTHIA simulations to estimate the effect we conclude that RPCs should be ok even at forward angles, but there are uncertainties on the cross sections.
- - SR: most likely no problem, since the HCAL is 'shielded' by the ECAL
- - One-photon interactions: easy
- - Two-photon interactions: more difficult
- - electrons: not a problem since they are (mostly) absorbed by the ECAL
- - muons: not a problem, the rate is manageable
- - hadrons: the rate in the very forward region is a bit of a concern.
- Can the detector tolerate the background conditions for the ILC
parameter sets described in the Feb. 28, 2005 document at
Please answer for both 2-mrad and 20-mrad crossing angle geometries.
If the high luminosity parameter set poses difficulties, can the
detector design be modified so that the gain in luminosity offsets the
reduction in detector precision?
- We are evaluating the beam-beam EM background fluxes for these cases and should have them by Snowmass. (see also Q. 11)
- What is your preferred L*? Can you work with 3.5m < L* < 4.5m?
Please explain your answer.
- The L* preferred for SiD is that which is most likely to produce the most luminosity with the least background, while not interfering with the acceptance of SiD. So it is difficult to answer, as it appears to be coupled to questions of crossing angle and required stability for the final quads. (We are not interested in a tube stabilizing the quads that goes through the middle of the detector). The range 3.5m < L* < 4.5m seems generally acceptable to us.
- What are your preferred values for the microvertex inner radius and
length? If predicted backgrounds were to become lower, would you
consider a lower radius, or a longer inner layer? If predicted
backgrounds became higher, what would be lost by going to a larger
radius, shorter length?
- This is a very detailed question that needs a lot of simulation work. We understand that Sonja Hillert (from the UK LCFI Collaboration) will present important new results on this topic at Snowmass, in the context of the CCD-based VXD design.
- Our present geometry is given in lcsim.org. The SiD approach for now is to go with a shortened barrel plus four layer endcap, but this approach needs some engineering to know just how thin the endcap can be, and additional study of how it performs given a realistic material budget. This will not be settled by Snowmass. We find detailed optimizations for this question slightly premature. We see it evolving for years as more realistic luminosity requirements and backgrounds are evaluated and the physics requirements better understood.
- Are you happy that only 20mr and 2mr crossing angles are being studied
seriously at the moment? Are you willing to treat them equally as
possibilities for your detector concept.
- We think the present strategy of studying 2mr and 20mr as eextremef cases is acceptable. However, SiD would be interested in the smallest crossing angle that does not compromise downstream E and P measurement, does not increase backgrounds, does not significantly increase the risk of backgrounds, and does not reduce the reliability of the machine (e.g. thermal load on FF superconducting quad). This may well be more than 2 and less than 20 mrad. If 2mrad will not allow downstream monitoring of polarization and energy, we would like to see study of a "smallest possible crossing angle" solution which does.
- Is a 2mr crossing angle sufficiently small that it does not
significantly degrade you ability to do physics analysis, when
compared with head-on collisions?
- It is probably acceptable, but we would wish to do a more detailed physics study of electron identification in the forward regions. See also our response to Q. 8.
- What minimum veto and/or electron-tagging angle do you expect to use
for high energy electrons? How would that choice be affected by the
crossing angle? How does the efficiency vary with polar angle in each
- These are detailed questions that we would need to simulate in some detail.
- What do you anticipate the difference will be in the background
rates at your detector for 20mr and for 2 mr crossing angle? Give
your estimated rates in each case.
- We are simulating pair backgrounds from beam-beam interactions and will have results for Snowmass. Time permitting, we hope to have results for radiative bhabhas and SR photons also.
- What is your preliminary evaluation of the impact of local solenoid
compensation (see LCC note 143) inside the detector volume, as needed
with 20mr crossing angle, on the performance of tracking detectors
(silicon, and/or TPC, etc.)
- Our gut feeling is that this will not be an issue provided the field is known reasonably well. We need to quantify this statement. However both the VXD and central tracker will have Lorentz effects for the ionization electron paths in the silicon at some level. This will need study.
- Similarly, what is you preliminary evaluation of the impact of
compensation by anti-solenoids (LCC note 142) mounted close to the
- Regarding the magnetic field influence on the detector we are not concerned as long as the field is known well (see previous answer). The mechanical issues of forces and supports for the solenoids will require a detailed study, but we do not think these will be insurmountable.
- Do you anticipate a need for both upstream and downstream polarimety
and spectrometry? What should be their precision, and what will the
effect of 2 or 20 mr crossing angle be upon their performance.
- It is desirable to have both upstream and downstream polarimetry. Similar conclusions can be drawn for energy measurements. The downstream energy spectrometer and polarimeter can be closer to the IP with less extrapolation error from their measurements to the relevant beam quantities at the IP. Given the high precision desired for both energy and polarization measurements, it is also very desirable to have redundant measurements of these quantities by independent techniques. In addition, the extraction line more easily accommodates a back-scattered Compton gamma measurement to complement the back-scattered Compton electron measurement. Beam-beam collision effects can be directly measured with extraction line diagnostics by comparing measurements with and without collisions.
- Reasons for both, assuming identical performance capabilities:
- - Independent measurements/cross checks
- - reduced systematics
- - better ability to estimate systematics
- - some independent systematics
- - precision measurements are a key aspect of the ILC
- - having both upstream and downstream polarimeters help spin alignment procedure and estimate of errors also. We are still exploring the design issues for both.
- Upstream energy spectrometer issues:
- - chicane magnets need to be ramped frequently for calibrations (every 10 minutes?);
- - how to get reliable energy measurements during ramps?
- - can luminosity be delivered when chicane magnets are ramping?
- - chicane magnet fields are low (~1 kG at 250 GeV; 0.2kG at 50 GeV); will precision be adequate for Z-pole running?
- - dispersion is only 5mm and position differences due to 10-4 energy diff will be 500nm; this is small compared to betatron spot size -- is this a problem?
- Upstream polarimeter issues:
- - laser system design complexity
- Downstream polarimeter and energy spectrometer issues:
- - backgrounds
- - is 2mrad feasible at all? Larger beam losses and larger beam spotsize at Compton IP give either worse signal:background or require larger laser power -- needs further optimization
- - effect of synchrotron energy losses
- Is Z-pole calibration data needed? If so, how frequently and how
much? What solenoid field would be used for Z-pole calibration? Are
beam energy or polarization measurements needed for Z-pole
- We have not yet given this issue real study, but expect to need some runs at the Z to get enough tracks to align the tracking detectors and perhaps to cross calibrate the calorimeters. Ideally these would be at full field. Experience from SLD shows that of order 500k Zs was just about sufficient to align a system of 96 CCDs including non-planar shape corrections for the sensors in the vertex detector. We think that the trackers need to be designed with an alignment friendly awareness - nice overlap regions and lever-arms and preferably a high degree of symmetry.
- We have not thought much about aligning the endcap yet. That could require more data.
- If the central tracker alignment were based on the SLD VXD alignment strategy, the statistics required may well be higher given the larger volume and many more overlapping regions to deal with.
- We would expect to have to (re-)align after each major detector access. In principle this ought to be no more than a few times per year.
- Would you like the e-e- option to be included in the baseline, and
if so what minimum integrated luminosity would you want?
- Not for now, unless SUSY is discovered.
- What will be your detector assembly procedure.
- What size is required for the detector hall?
- We think these questions relate to the footprint of the detector hall and its transverse and longitudinal placement w.r.t. the other hall, beam dumps etc. We can try to provide some anwers at Snowmass.