Minutes (Hours) from May 6 Meeting
Discussion today focused on the summaries of the
new intiatives workshop that we just survived (May 2-4),
in particular what the detector issues are. One new detector technology
that was discussed at the workshop came from Ken Heller and consists of
a fine-grained water-liquid scintillator detector. The absorber is cheap,
and adds stability to the liquid scintillator, which is itself known to be
significantly cheaper than solid scintillator.
Ken Heller's talk in the working group session can be found in
Mayda reported that Michal was looking into what happens if you take
the steel--solid scintillator--air design they have now and change it
such that the air gap is filled with water, and the scintillator is now
liquid scintillator and not solid. They will have new performance results
for that detector soon, although it was noted that if there's 0.45mm of
steel combined with 3cm of water and still only one readout plane,
then the total mass of the detector per readout plane
has gone up by almost a factor of 1.7
and the segmentation has been increased a little above 1/3 a
Michal is also looking into the performance of Ken's design, which would
only contain water and scintillator, and different transverse segmentation
than the initial design. Both sets of results should be very interesting...
The summary talks from the detector working group at the New Initiatives
session can be found in
newinit.ps (Debbie's talk) and
Detectorsummary.pdf (Ken's talk).
The discussion at the workshop after these two talks (and at today's meeting)
centered around the following two approaches:
On the subject of how well delta m2 has to be known before you would pick
a site, Mayda sent email to the group last month with this plot
fom.eps showing the figure of merit
(signal/sqrt(signal+background))for the fine-grained steel-scintillator
detector as a function of delta m^2
(delta m2 ranging between 2e-3 and 4e-3eV2). Things don't change very
quickly within that range, for several different angles and baseline
distances. A similar plot can be found basevdm2.eps
, which shows the limit versus baseline length for dm2 = 2,3,4eV^2.
However, there are several problems with these kinds of plots, as pointed
out by Kevin but iterated by others (I added my own capitalization...):
- chose a detector technology in the next 6 months
and write a detailed experimental proposal
in the next 12 months, with the idea being that this detector technology
had better be one whose performance and cost are pretty well understood?
The idea here is that we'd pick a site and a detector technology before MINOS
turned on, so we wouldn't know delta m^2 very precisely, and we also wouldn't
know how such a detector operates in a 2GeV neutrino beam either, unless
a new hall at MiniBooNE got built, or we managed to get something into the
K2K beamline. It was argued that certainly we could put such a detector
in an electron and a pion test beam, at the least, before MINOS starts.
- take more time
to do more detector r&d, find out what delta m2 is to
30%, and maybe chose something that looks different from what we would
chose if we had to chose in the 6 month time scale? By having more time to
figure things out, it's assumed we would also be in a better position to
pick something that would not pose a problem for the next stage of
experiments: going a factor of 10 or more in statistics past the first
stage of this experiment.
Many believe that there is a strong case for going ahead quickly now and
proposing something that has a reach of 10x the CHOOZ limit, and trying
to get funding such that the detector can be built ASAP and start running
as early in the MINOS run as possible. While we all agree that the first
measurement of a non-zero theta_13 is exciting and interesting, we also all
agree that it's only the first step. The sooner we go ahead with an experiment
and a detector technology choice, the sooner we get to that first step, but
it might be argued (Debbie's editorializing here) that the sooner we make
the detector decision on the first step the less likely it is that the
second step, namely the upgraded detector, looks identical to the first one.
On the other hand, maybe the NC cross sections and rejection factors we are
using now turn out to be correct, and the next detector looks almost identical
to the first one. Also, in the model that Stan has suggested where we just
start building and just keep adding more detector as the funding allows,
this would make changing detector technologies in the middle really painful.
- We DON'T KNOW that dm2 = 2,3, or 4eV^2!!!
- Even if we do know it at that range, the limits at 900km are really
really different depending on delta m2, but if you were trying to go for
an expandable program which could do matter effects, 900km would be
- There is currently a
LARGE UNCERTAINTY ON THE OVERALL NC CROSS SECTION
(I think a factor of 2 was even suggested as an uncertainty)
and what these plots look like depends on what that cross section
is. Also, it's not clear that all of these detector background studies
are even using the same nc cross section model, so comparisons between
the detectors aren't necessarily all that valid unless the same cross section
model is used. For the record, all the fine-grained detectors are simulated
using GMINOS which uses NEUGEN and GEANT-FLUKA for neutrino interactions.
- The nature of the experiment will change dramatically if in reality delta
m2 is low and your NC background is high: you're going to start caring a lot
more about systematic uncertainties, and you would be rushing to build a
near detector, if you hadn't built one yet.
Finally, Maury lead a discussion on comparing detector costs--both for the
"first" and "second" generation detector, where the assumption is that
the second detector is a factor of 5 or 25 in mass past the first detector.
Executive summary: 25 times anything is a big number
The idea is: if the proton driver upgrade is something like 300M$ and
gives you a factor of 5 in intensity, how much do you spend for upgrading
your program in two scenarios: making a detector a factor of 5 more massive
plus a proton driver upgrade, or making a detector a factor of 25 more
massive with no proton driver upgrade? Although the answer of which
option you would prefer is not surprising to anyone, the magnitude of the
factor by which a proton driver upgrade is cheaper was surprising to most of
us in the room. Steve Geer commented that in some of these detectorx25
options, you're starting to talk about an experiment that is as expensive
as a neutrino factory. (At which point many of us in the room said "yes,
that's why we've wanted a neutrino factory all along!). The other important
point everyone has to remember of course is that with a proton driver you get
lots of other physics too, whereas with a more massive detector near the
surface you're still not going to be able to do proton decay, solar
In comparing different detector technologies, the cost of the detector
for a given sensitivity in Ue3^2 (i.e. you need less mass for an ICARUS-type
detector, more mass for a water cerenkov detector) ranges from 30M all the
way to 100M. (Aside: when Jeff Nelson used the spread sheet for what
MINOS would cost for a 20kton detector, but with 0.45mm steel segmentation
he got 313M, which includes labor, materials, etc. There are cheaper readout
options being explored by Bruce King, and also Michal is now adding water
and liquid scintillator, as mentioned above). But anyway, if the detector
is 30M to 100M, then a 5 times more massive detector is 150M to 500M
So a 5x more massive detector plus a PD upgrade is (very very roughly)
450M to 800M. Just an aside for the water cerenkov fans (who probably
know this already): 5 to the 2/3
power is only 3, and 25 to the 2/3 power is only 9.
Compare this with a x25 more massive detector, which would
be 750M to 2.5B!!! Of course with most of these detector technologies,
with a factor of 25 more exposure you do have to start worrying about
the systematic errors, so a factor of 25 in exposure might not get you a
factor of 5 in overall reach or precision!
As noted above, this last region
is where the neutrino factory with its "measly"
50kton minos-type detector starts looking comparable in cost.
Last modified: Thu May 9 12:41:28 CDT 2002