I have looked at several HCAL options, comparing both absorber and active media types.

I first compared HCAL versions with identical scintillator readout, one version had 0.7 cm W absorber per layer and one had 2 cm SS as the absorber. In both versions, the HCAL was 4 nuclear interaction lengths thick, 55 layers of W/Scin compared to 34 layers SS/Scin. My specific finding was that the W/Scin HCAL performed better than the SS/Scin HCAL both for single particle energy resolution and for PFA results - both perfect PFA and the actual algorithm. Showers in W were more compact than showers in SS - confirming results I had seen from H. Videau earlier. My main motivation for this study was to see if a more compact HCAL could be built using a dense absorber, thus saving R**2 which presumably contributes to the cost of the magnet. My general conclusion was that not only was this goal achieved, but that the W/Scin HCAL even performed better (PFA and single particles) than the SS version.

Next I looked at 2 version of HCAL with W absorber, one with scintillator and one with RPC as active media. Both these were analyzed as digital calorimeters. My specific findings here were that the W/Scin digital HCAL and the W/RPC digital HCAL had similar performance as determined by calculating perfect PFA. The scintillator version had slightly better PFA resolution, presumably because of the higher number of hits per GeV for neutrals which, in digital mode, translates directly into better resolution.

Despite this, the SiD detector concept chose as its HCAL 2 cm SS absorber with RPC readout. I then looked at the perfect PFA performance of this detector and found that it performed worse than both the W/Scintillator and the W/RPC HCALs. In fact, the SiD combines the absorber with worse properties with the active media with fewer hits, so it was no surprise that the perfect PFA performance was so poor. In fact, it is impossible to obtain 30%/sqrt(E) resolution for the SiD detector with this option.

I then made suggestions as to how the performance of the SS/RPC HCAL could be improved based on all of my observations and found that these improvements led to a larger volume for the HCAL. I then suggested that maybe the optimal use of RPCs (generally gas HCAL) would be found in a larger volume, lower B-field detector concept like the LDC, rather than the compact, higher B-field SiD. It seems to me, supported by the simulated detectors that I analyzed, that the optimal HCAL configuration for a
compact, high B-field detector should have a dense absorber combined with a solid (or maybe liquid) active media. This optimizes (means minimizes) the outer radius of the HCAL which directly saves magnet costs as mentioned above while maintaining good resolution for the neutral component of jets. Of course, things like transverse segmentation and the minimum calorimeter radius affect the final PFA performance of the detector and are used to ultimately determine if a particular concept is viable - but, as I showed, the best perfect PFA performance I got for a compact, high B-field detector was with a W/Scin HCAL. It also seems to me that there might be a different optimal HCAL for the compact detector than for a large, low B-field detector. I wouldn't be surprised if it turned out that a SS/RPC HCAL would be a good choice for the LDC detector and that a W/Scintillator HCAL would be better for the SiD. I would recommend that the LDC consider a 1 cm SS absorber/1.2 mm RPC per layer HCAL. A 4 lambda deep HCAL of this construction would have 67 of these layers and would be ~120 cm from IR to OR, an increase of ~25% compared to the 2 cm SS version. By thinning the absorber, I think the resulting neutral particle resolution obtained in a PFA would allow the 30%/sqrt(E) goal to be obtained.

Hi Dean,
Yes, I think it does - there are more samples in the W version, keeping the total thickness of the HCAL to 4 interaction lengths. For some reason, the original SS design had absorbers of thickness 1 radiation length per layer (2 cm SS). I suspected that for neutral showers in an HCAL, its not radiation length that's important, but nuclear interaction length. I wanted to see if you could get better performance out of a calorimeter with optimal interaction length sampling while reducing the overall volume of the calorimeter to save on magnet cost. The original 2 cm SS absorbers represent 12% sampling in interaction length. I chose .7 cm W absorbers for the other option which is ~7% sampling in interaction length - 2 radiation lengths. As I suspected, the finer sampling in the W performed better and resulted in a smaller volume HCAL. I think to make the SS version perform better, you should use 1 cm thick layers, which is 6% interaction lengths or 1/2 radiation length. However, this makes the overall size of the HCAL much bigger. (All of this assumes a 4 interaction length total thickness of the HCAL).
I only had simulated data with 2 different interaction length layers and 2 different radiation length layers, so I could use at least one more point for each to really understand the dependence. We will study this before Smass.