1. Neutron detector for the central part of CLAS12
European
collaboration:
INFN Frascati,
INFN Genova,
IPN Orsay,
LPSC Grenoble
The Central Detector
Physics case: nDVCS
Requirements
Technical constraints
Possible solutions
Work plan
GDR Instrumentation, 8 Avril 2008
S. Niccolai, IPN Orsay

2. CLAS12: the current design
Forward Electromagnetic
Calorimeter
Forward Time-of-Flight
Low threshold
Cerenkov Counter
Drift Chambers
High threshold
Cerenkov Counter
Preshower
Calorimeter
Central detector
Full detector:
Central (CD), 40o<q<135o
Forward (FD), 5o<q<40o
Inner Electromagnetic
Calorimeter
CD + exploded view of one sector of FD
8/4/08 - S. Niccolai – IPNO

3. and/or neutron detector
Central Detector
Main coil
Central tracker
CTOF (3 cm thick)
Space for calorimeter
and/or neutron detector
(12 cm radial thickness)
Cryostat vacuum jacket
Magnetic field in the center = 5 T
8/4/08 - S. Niccolai – IPNO

4. Neutron detector Main goal: detection of recoil neutron in nDVCS
H, H, E, E (x,ξ,t) ~
x-ξ t
(Q2) e e’
gL*
x+ξ n n’ g
Detected in
forward CLAS
Not detected
ed→e’ng(p)
Detected in
FEC, IC
nDVCS: strongest sensitivity to E
PID (n or g?) via TOF + angles
to identify the final state
Ji’s sum rule
→ access to quark orbital momentum
2Jq =  x(H+E)(x,ξ,0)dx
8/4/08 - S. Niccolai – IPNO

5. Neutron kinematics in nDVCS
Neutron momentum (GeV/c)
Neutron q (°)
Average neutron momentum
GeV/c
GPD-based event generator
(free neutron) Ee = 11 GeV CD
Most of the neutrons at q~60°
Thanks to Marion Mac Cormick
8/4/08 - S. Niccolai – IPNO

6. GEANT4 simulation
estimate and maximize detection efficiency for neutrons
resolution on TOF to separate neutrons and g
or find an alternative way to separate neutrons and g x z y l R r
Realistic geometry,
following design
for CTOF: solid
composed by
trapezoids
Segmentation:
Radial, to determine interaction point
TOF  p (5 layers in r, 3 cm thick)
Azimuthal, to determine f (30 slices)
q can be obtained via z=1/2veff(t1-t2)
Dimensions
R= 39 cm
r = 24 cm
l = 50 cm
8/4/08 - S. Niccolai – IPNO

7. GEANT4 simulation
estimate and maximize detection efficiency for neutrons
resolution on TOF to separate neutrons and g
or find an alternative way to separate neutrons and g
Realistic geometry,
following design
for CTOF: solid
composed by
trapezoids
Segmentation:
Radial, to determine interaction point
TOF  p (5 layers in r)
Azimuthal, to determine f (30 slices)
q can be obtained via z=1/2veff(t1-t2)
Possible to add layers
of another material
8/4/08 - S. Niccolai – IPNO

8. Neutron efficiency Detector material: scintillator
Generated neutrons with pn= GeV/c, q=90°, f=13.5° (center of 6th f slice)
Efficiency: Nrec/Ngen
Nrec= number of events
in 6th f slice having
Edep>Ethreshold
(r: first good hit only)
Can we live with 15% efficiency?
CLAS12 will have
10 times higher luminosity
than CLAS (1035 cm-2 s-1)
→ roughly the same statistics as
what was obtained for proton DVCS
(assuming equal cross sections)
Light quenching effect taken
into account by reducing Edep for
protons by a factor 5
Efficiency increases
decreasing the threshold
Eff ≈ 15% for thr. = 5 MeV
and pn=500 MeV/c
In agreement with « thumb rule »:
1% efficiency for 1 cm of scintillator

9. TOF resolution Kyoungpook’s measurement Simulation (Raffaella De Vita
For each f, r slice:
TOF = (t1+t2)/2
t1 = tofGEANT+ tsmear+ (l/2-z)/veff
t2 = tofGEANT+ tsmear+ (l/2+z)/veff
veff=16 cm/ns (value used in GSIM)
tofGEANT= average of times of all «steps»
z = average of z positions from all steps
tsmear = smearing factor: Gaussian centered at 0 with s= t0/√Edep (MeV),
t0 = 92 ps (deduced from Kyoungpook’s measurement at 6 MeV) for 1st r layer,
for other layers t0 = 200 ps
Simulate time distribution of
the scintillator light
Introduce spread due to light transmission
in the bar
Account for transmission from the
scintillator to the pmt photocathode
(different size, lightguide...)
Account for conversion to photoelectrons
(q.e. =20%)
Include additional time spread due to PMT transit time and amplification
Kyoungpook’s measurement
Simulation (Raffaella De Vita
& Marco Mirazita)

10. TOF resolution: results
s = 90 ps
s = 122 ps
s = 138 ps
s = 175 ps
Neutrons, pn= 1 GeV/c
s = 164 ps
Threshold = 2 MeV
Resolution is worse with smaller threshold

11. TOF resolution: results
s = 54 ps
s = 82 ps
s = 82 ps
s = 76 ps
Photons, pg=1 GeV/c
s = 69 ps
n/g separation
possible up to
pn<1 GeV/c
At 3s overlap
between n and g

12. The spaghetti option: KLOE
Active material:
1.0 mm diameter scintillating fiber
Core: polystyrene, r=1.050 g/cm3, n=1.6
High sampling structure:
200 layers of 0.5 mm grooved lead foils (95% Pb and 5% Bi).
Lead:Fiber:Glue volume ratio = 42:48:10
Conceived as an electromagnetic calorimeter,
it turned out to be very efficient for neutrons
Could this solution be viable for us?
Studies underway…
50% more efficient
than a scintillator, measured with neutron beam (Uppsala) and reproduced by simulation (FLUKA)
See Anna Ferrari’s talk for details
on KLOE spaghetti calorimeter
8/4/08 - S. Niccolai – IPNO

13. Comparison of efficiencies (GEANT4)
Simulation underestimates
KLOE’s data by 5-8%: maybe
light quenching is overestimated
8/4/08 - S. Niccolai – IPNO

14. Neutron/photon separation with spaghetti calorimeter
Photon, no lead
Photon, scintillator + lead
Studied the number of hits per event
with Edep>threshold, for neutrons and photons
8/4/08 - S. Niccolai – IPNO

15. Neutron/photon separation with spaghetti calorimeter
Threshold = 2 MeV
Neutrons
Photons
But we still need good time resolution
to determine q angle of the neutron
(required angular resolution needs to be estimated)
Can it be achieved with one fiber?
Very different multiplicities for n and g!!!
“Lasagna”: 75 scintillator
layers (1.3 mm each)
+ 74 lead layers (0.7 mm each)
BUT
Eff(g) = 99%
Eff(n) = 20%
→We need to estimate COUNT RATES for n and g
8/4/08 - S. Niccolai – IPNO

16. Neutron detector: requirements
Whichever solution will be chosen for the neutron detector (layers of scintillators or spaghetti), there are the following issues:
limited space upstream and downstream, due to the presence of the light guides for CTOF
→ no space for additional light guides to “escape” from the high magnetic field
light collection in the high magnetic field
BUT, compared to CTOF, the requirement on TOF resolution is less stringent:
from preliminary simulations, a time resolution twice as bad
as the one currently achieved in KNU and Jlab measurements can still be good enough to
separate photons from neutrons for neutron momentums up to 1 GeV
Can SiPM be the solution?
We need photodetectors insensitive
to magnetic field, providing decent
timing resolution

17. SiPM: characteristics
SiPM (proposed by Sadygov and Golovin in the ’90)
matrix of tiny microcells in parallel / each micro-cell = Geiger Mode-APD + Rquench
output signal is proportional to the number of triggered microcells Al
ARC
-Vbias
Back contact p n+ 
Rquenching h
p+ silicon wafer
Front contact
Solution considered at Orsay/LAL is a
matrix of SiPMs:
larger area covered
reduce noise by requiring time coincidence of several SiPMs within the matrix
General view
Photo-sensitive side
25 µm
PROS:
Insensitive to magnetic field
High gain (106)
Good intrinsic timing resolution (30 ps/pixel)
Good single photoelectron resolution
CONS:
Very small active surface (1mm2)
→ small amount of light collected
Noise
Thanks to Nicoleta Dinu (LAL) for the images and the infos
8/4/08 - S. Niccolai – IPNO

18. Tests on photodetectors with cosmic rays at Orsay
“Trigger” PMTs (Photonis XP2020)
“Trigger” scintillators
(BC408) 1cm thick
Scintillator bar (NE102)
80cm (length) x 4 cm (width) x 3 cm (thickness)
Drawing by A. Maroni
8/4/08 - S. Niccolai – IPNO

19. The current setup
Photonis XP20D0
8/4/08 - S. Niccolai – IPNO

20. Electronics & DAQ are in place The data taking is underway
Plan:
Measuring TOF resolution with reference PMTs
Substituting PMT at one end with SiPMs (matrix of SiPMs will be
constructed, collaboration with neighbors at LAL for electronics)
…and let’s see!
8/4/08 - S. Niccolai – IPNO

21. Conclusions and to-do list
CTOF and Neutron detector could coexist in one detector, whose first layer can be used
as TOF for charged particles when there’s a track in the central tracker, while the full
system can be used as neutron detector when there are no tracks in the tracker.
Neutron detection is necessary for the measurement of nDVCS (Ji’s sum rule)
Detection of DVCS recoil neutrons with ~15% of efficiency and n/g separation for
p≤1 GeV/c seems possible from simulations, using scintillator as detector material
A spaghetti calorimeter could give higher efficiency while permitting to use hits multiplicity
as PID method, but it increases photon efficiency a lot…and angular resolution?
To do list:
use complete CLAS12 simulation and realistic event generators for signal (nDVCS)
and backgrounds (ed→enp0(p)) to define needed resolutions (q,f)
study the time resolution for the spaghetti option
evaluate count rate for signal (n) and background (g), to understand if high photon detection efficiency is a problem or not
Hardware: tests on timing with SiPM planned for the spring at Orsay, tests with ordinary PMs already underway
8/4/08 - S. Niccolai – IPNO