sPHENIX EMCal discussion

[Mickey Chiu <chiu AT bnl.gov>]

Hi Anne, et al,

I think

> trading the EMCal performance for a barrel time of flight detector.

is not how I would characterize John's proposal.  First, it's not at all 
clear you're losing much, if any, performance.  Secondly, the trade is 
for a tracking detector we actually want, rather than one that is forced 
upon us by financial circumstances.  I'm assuming we could save enough 
to make this statement though I haven't seen the numbers.  If this is 
true, then a more accurate statement would be "replace building a new 
EMCAL with building an entirely new tracking system."  The question is 
really between whether a new EMCAL with a whacked up tracking system is 
a better realization of the sPHENIX physics goals of jets and upsilons, 
or would an old EMCAL and new tracking system be better.  Of course if 
you don't save enough then what I say above isn't true...  The TOF stuff 
I mention only of course because I'm working on it, but I do think that 
it is only possible with outside funding or if there is enough 
contingency left over.  It's like a cherry you can put on the top if we 
can afford it, but it's not the main course.

The PbSc Moliere radius of 3 cm is listed in table 1 here:

  http://arxiv.org/pdf/nucl-ex/0202009.pdf

In the olden days when I was looking for the PbSc Moliere radius and 
noticed that it was much smaller than the segmentation, I was surprised 
and checked it with Edward and he confirmed it.  It might be a good idea 
to re-confirm it with simulation and test-beam if we decide to go this 
route.

I agree we'd need a full simulation to be sure, but I think what John 
had proposed really wouldn't be much of a negative for the Upsilons or 
jets.  The reduced measurement capabilities for the Upsilons in the 
triangles between sectors would be much less than the 40% loss we've 
already deemed acceptable for the Upsilon, and for a jet measurement 
where you just integrate in a cone I think the weird geometry there 
isn't going to be too bad.  As for projectivity in phi, we had zero 
problems in PHENIX with the 20 degree angle of incidence using the 
GAMS-derived clustering, and the PHENIX EMCAL had zero projectivity, so 
I see no major issues there.  The projectivity in eta was already a 
major issue in the baseline sPHENIX EMCAL, so that might be a wash.  I 
do think that someone would need to look into the 
clustering/reconstruction at polar angles higher than 20 degrees, since 
I think there the GAMS-derived clustering we used in PHENIX may have 
some issues.  There would be additional information if we decided to go 
with longitudinal segmentation, however.

Cheers,
Mickey

On 06/02/2016 12:09 PM, Sickles, Anne M wrote:
> Hi everyone,
>
> Thanks John for kicking off an interesting discussion.
>
> I think if people would like to seriously consider this route, we need a
> real model of the geometry for a reused PbSc and simulations.  The
> segmentation that John quotes is significantly worse than even the
> ganged 2x2 towers that we are considering and while the improved
> resolution would help with the e/h separation the lack of projectivity
> in both eta and phi will be a drawback.  In addition, the gaps between
> the sectors would result in a decreased acceptance for upsilons and
> photon and non-uniformities in the jets. It’s a quantitative question as
> to what the performance actually would be and since we have real
> simulations for the SPACAL, we would need real simulations for a reused
> PbSc as well. There seems to be a factual disagreement between John and
> Mickey as to what the Moliere radius of the PbSc actually is and I
> couldn’t find documentation of that in either the web page John linked
> to or the NIM paper:
>
> http://www.phenix.bnl.gov/phenix/WWW/pub/phenixnim/d_emcal/nim_4d_emcal.pdf
>
> It is really not clear to me what the impact to the schedule is of a
> reused EMCal.  It does exist, but it would take a few months to
> implement in simulations and convince ourselves that it could meet our
> physics requirements.  Given the physics aims of sPHENIX (jets and
> upsilons), I don’t see the logic in trading the EMCal performance for a
> barrel time of flight detector.
>
> The 2D projective shashlik that Craig mentions obviously deals with some
> of these issues (it could be designed to be 2D projective for example)
> and I think we had a MC model for it at some point (?).  However given
> that it is orthogonal to the last few years of R&D we have done, I agree
> with Craig’s assessment of a very long delay to the project.
>
> This is something we could certainly discuss at the EMCal meeting next week.
>
> Best,
> Anne
>
>
> > On Jun 2, 2016, at 9:41 AM, Mickey Chiu <chiu AT bnl.gov
> > <mailto:chiu AT bnl.gov>> wrote:
> >
> > Hi John, et al,
> >
> > I think this idea isn't as crazy as it sounds.  The energy resolution
> > would be much better, 8% vs 12%, which buys back some of the e/pi
> > separation (through E/p and residually through shower shape cuts).  The
> > caveat in the E-resolution argument is that we'd need to understand
> > whether the constant term can be reduced to the version we had in the
> > test beam, or the one we had in PHENIX.
> >
> > I think if you go this route you would want to optically split the PbSc
> > longitudinally, which then would provide equal or superior e/p
> > separation to the dual EMC/iHCal that is the default now.  Just reading
> > out the front and back wouldn't help very much, though in the
> > longitudinal split you certainly would want to collect the light from
> > the front and back.  I think you could start working on the
> > "construction" almost right away, with just a short R&D period to figure
> > out how to best split it longitudinally.  Longitudinal segmentation also
> > gives you a crude sort of projectivity along eta.
> >
> > One down-side that I see is the segmentation is much coarser, but this
> > can also be solved rather easily.  The PbSc Moliere radius is actually
> > 3.0 cm, a fact that is not very well known.  The fibers could be
> > rearranged to buy back some of that segmentation.
> >
> > The only other down-side I can think of are the the triangular gaps
> > between sectors.  (Well, there's also the mental down-side that this
> > won't have that new car smell...)  I'm not sure what you want to do with
> > those gaps, but I'm betting you will need to fill them somehow. However,
> > I don't agree with Craig that re-using the PbSc is going to be as
> > expensive as building brand new... I'd expect it to be _significantly_
> > cheaper, and with a better schedule.  As I mentioned above, you could
> > probably start working on the "construction" of this as soon as we get
> > the PbSc off the East Carriage around late August/September.  My guess
> > right now is that it would be some simple mods to the existing
> > calorimeter.  We have a prototype in the basement which we can get Sean
> > cranking on today if desired.
> >
> > With the money I'd expect we could save, we could consider more optimal
> > configurations of the tracking.  In addition, based on the EIC R&D I
> > have been working on, I'm currently 90% confident that I can build a
> > mRPC TOF system capable of 20 ps, which would give PID capabilities
> > similar to what we have in PHENIX, but over the whole 2pi acceptance.
> > This would cost just $4M (not counting contingency but otherwise is the
> > whole project cost).  This estimate is based on 4 times larger STAR TOF
> > cost, which was $7M, so it might be considered a conservative estimate.
> >  The remaining 10% uncertainty in my confidency level above is due to
> > whether we could create electronics capable of handling the 4 us trigger
> > latency.  I had thought I could retire that risk by this coming winter,
> > but current labor changes at BNL make that time-scale more uncertain.
> > In any case, we have a plan in place for how to make the electronics for
> > this.  Over the next year I'm hoping we can demonstrate 10 ps resolution
> > (and incidentally reduce the cost further).  If we can do that, we
> > should be able to surpass the PHENIX (and RHIC-wide) momentum reach for
> > PID by \sqrt{2}.  Of course it should be obvious that full acceptance
> > PID would significantly enhance the sPHENIX physics program...
> >
> > Cheers,
> > Mickey
> >
> > On 06/02/2016 09:23 AM, Gabor David wrote:
> > > On Thu, 2 Jun 2016, Craig Woody wrote:
> > >
> > > Hi, Craig and All,
> > >
> > > > Hi John,
> > > >   I have to say that I was thinking the same thing. When we talk about
> > > > eliminating the Inner HCAL (which is clearly like having a vital
> > > > appendage removed in my opinion...), one does have to ask, why is there
> > > > this large gap between the EMCAL and the wall of the magnet ? I agree
> > > > the main justification for going to a tungsten absorber was to keep the
> > > > EMCAL compact, thus allowing the HCAL to be kept smaller in order to
> > > > reduce the cost. However, if the entire radial space from ~ 90-140 cm is
> > > > available for the EMCAL, keeping the Outer HCAL the same, then it would
> > > > make sense to reconsider lead as an absorber again. If I recall Tom
> > > > Cormier's design, it essentially just fit into this space with ~ 18 Xo.
> > > > One could then consider a projective (even 2D projective) shashlik
> > > > design with SiPM readout. However, I don't think it makes sense to
> > >
> > > In fact, ideally dual readout (front and back), this would
> > > buy back some of the lost e/p separation
> > >
> > > > actually consider using the existing PHENIX shashlik, since it's
> > > > completely the wrong geometry and I would be the cost to rework it would
> > >
> > > Fully agree
> > >
> > > > be more than building a new one. Nevertheless, if we have the radial
> > > > space, it would be worth considering a lead absorber again in terms of
> > > > cost and performance. The only problem with doing that is that it would
> > > > set us back about two years in terms of schedule.
> > >
> > > Aye, there's the rub...  (more later, I got an appointment)
> > >
> > > Gabor
> > >
> > > > Craig
> > > >
> > > > On 6/2/2016 12:10 AM, John Haggerty wrote:
> > > > > Hello,
> > > > >
> > > > > This is not for Friday's descoping document, but I think that given
> > > > > some
> > > > > of the ideas going around now, we might have to go back to ideas that
> > > > > were dismissed many workfests ago, since they might not be as bad as
> > > > > some of the descoping options we're now talking about.
> > > > >
> > > > > The one in particular that's been bothering me the last few days is
> > > > > going to a single calorimeter inside the solenoid with lead absorber.
> > > > > If we multiplex the EMCAL to 2x2, one could ask why you need expensive
> > > > > tungsten to make a short radiation length.  Maybe lead would be good
> > > > > enough.  (I should note that Edward often asks this question, but when
> > > > > we had 1 Moliere radius towers, it was easy to say that only tungsten
> > > > > could achieve the short radiation length without going to hugely
> > > > > expensive crystals, and only tungsten would allow us to cram ~1.5
> > > > > interaction lengths inside the solenoid.)
> > > > >
> > > > > Once you think lead, you may as well think the PHENIX EMCAL, it's as
> > > > > good as any lead-scintillator calorimeter.  It's about 17.5 radiation
> > > > > lengths, about 0.85 interaction lengths in depth, which is about
> > > > > 375 mm,
> > > > > so it would fill up about the same volume as the EMCAL and Inner HCAL
> > > > > (1000 to 1400 mm in radius), and one could read it out at the inner
> > > > > radius with SiPM's.
> > > > >
> > > > > My facts com from Edward's web page
> > > > > > http://www.phenix.bnl.gov/phenix/WWW/emcal/computing/online/EmcDoc112602/Introduction.html
> > > > > and the NIM paper, by the way.
> > > > >
> > > > > A tower is 55.35 mm square, I think, so there are something like 100
> > > > > towers in phi and 60 in eta and natural numerology would make it 0.058
> > > > > (phi) x 0.037 (eta).  I think the Moliere radius of the EMCAL is about
> > > > > 55 mm, so it is matched, as is common, to the segmentation, and the
> > > > > 108*60=6480 towers takes 3 PHENIX sectors.
> > > > >
> > > > > The natural segmentation leaving the (12x12) supermodules alone would
> > > > > lead to a nine-sided calorimeter (9*12=108), odd, but not impossible.
> > > > > (If one can take apart the supermodules, one could contemplate a seven
> > > > > sided structure with 16x16 supermodules, I don't know if that's
> > > > > feasible
> > > > > or not).
> > > > >
> > > > > There are many bad features of this idea, which is why we dismissed it
> > > > > in the first place.  It's not naturally projective; there are likely to
> > > > > be sizable gaps for support; you can afford less photodetector coverage
> > > > > than you could with PMT's; the segmentation is just barely adequate.
> > > > > Some of these problems could conceivably be overcome with ingenuity and
> > > > > labor, but the segmentation and the radiation length are what they are.
> > > > >
> > > > > You can say the rework still costs a lot, but the difference in cost is
> > > > > more in design, mechanics, and modifying the existing modules, and some
> > > > > of that is common, and much of it is labor, which is not the immediate
> > > > > problem.  You lose the Inner HCAL, but the total interaction lengths is
> > > > > less than one less, from nearly 6 at zero rapidity to 5.  Part of this
> > > > > package would be to leave the Outer HCAL alone, or perhaps even augment
> > > > > it by splitting the tiles in half and reading both inner and outer
> > > > > radius.  We may lose in e-pi, but if the calorimeter's resolution
> > > > > survives intact (to be show with reduced photodetector coverage), maybe
> > > > > E/p is enough better to make up for that loss.
> > > > >
> > > > > Bad idea?  Many of us thought so, but depending on the results of the
> > > > > budget exercise, it may be worth kicking around again.
> > > >
> > > > _______________________________________________
> > > > Sphenix-emcal-l mailing list
> > > > Sphenix-emcal-l AT lists.bnl.gov <mailto:Sphenix-emcal-l AT lists.bnl.gov>
> > > > https://lists.bnl.gov/mailman/listinfo/sphenix-emcal-l
> >
> > --
> > Building 510C
> > Department of Physics
> > Brookhaven National Laboratory
> > Upton, NY 11973
> > Phone: 631-344-8428
> > _______________________________________________
> > Sphenix-emcal-l mailing list
> > Sphenix-emcal-l AT lists.bnl.gov <mailto:Sphenix-emcal-l AT lists.bnl.gov>
> > https://lists.bnl.gov/mailman/listinfo/sphenix-emcal-l
>
> --------------------------------------------------------------
> Anne Sickles
> Assistant Professor, Department of Physics
> University of Illinois at Urbana-Champaign
> --------------------------------------------------------------

--
Building 510C
Department of Physics
Brookhaven National Laboratory
Upton, NY 11973
Phone: 631-344-8428