LAr Detectors for Neutrino
Physics
Gary Barker
University of Warwick
UCL, 04/02/11
1
Outline






Liquid argon time projection chamber
Neutrino physics programme
Detector requirements/options
LArTPC R&D/ challenges
Current status
Outlook and conclusion
2
Bubble Chambers
 How to keep topology information of the bubble
chamber in a (high mass) neutrino detector?
3
Time Projection Chamber
 Charpak(1969), Nygren(1974) introduce TPC
 Drift electron-charge image of event to (x,y)
electrode array to give (x,y,z) image with
drift time
4
Liquid Argon TPC (LArTPC)
 (1977) Carlo Rubbia proposes a TPC based on
LAr as both n-target and detection medium.
Advantages:
1. Reasonably dense (1.4 g/cm3)
2. Does not attach electrons (much) => long drift times
3. High electron mobility (500 m2/Vs)
4. Easy to obtain, cheap (liquefaction from air)
5. Inert and can be liquefied by liquid nitrogen
6. Charge, scintillation light and Cherenkov light readout
possible
5
LAr Properties
 LAr has many similar properties to freon
CF3Br used in Gargamelle bubble chambers:
Argon
CF3Br
Nuclear
collision length
53.2 cm
49.5 cm
Absorption
length
80.9 cm
73.5 cm
dE/dx,
minimum
2.11 MeV/cm 2.3 MeV/cm
Radiation
length
14 cm
11 cm
Density
1.40 g/cm3
1.50 g/cm3
 Can expect event-development in
LAr/bubble chamber is very similar
6
Ionisation Charge
LAr ionisation: We=23.6 ±0.5 eV low detection thresholds and 6k
ionisation electrons per m.i.p.
 Some electrons will recombine – suppressed by Edrift (absent for mip’s at
Edrift ≥ 10 KV/cm)
 Drift velocity parametrised, Vdrift(E,T),
and measured in LArTPC’s

Vdrift~2 mm/ms @ Edrift=1 KV/cm
DVdrift/ DT Vdrift ~ -1.7%/K
 Oxygen (nitrogen) impurities capture free
electrons:
te [ms] ~300/r [ppb]
(t is electron lifetime, r is O2 concentration)
 clearly a crucial issue for LAr
 Diffusion effects are small e.g. for Edrift
O1 KV/cm: transverse ~ mm’s and
longitudinal « uncertainty on Vdrift
7
Light Production
 LAr is an excellent scintillator: Wg=19.5 eV giving approx. 5000
photons/mm/m.i.p
 Singlet (t=6ns) and triplet (t=1.6ms) excimers both give spectrum
peaked at l=128nm
 Light at this wavelength not energetic enough (9.7 eV) to cause
secondary ionisation/excitation  transparent to scintilation
light
 Light propagation governed only by Rayleigh scattering (assuming
high purity LAr) and the l4 –dependence ensures propagation of
visible light over 100’s of meters
 Furthermore, LAr has similar Cherenkov imaging capability to
water : H20(LAr), n=1.33(1.24)
8
9
A. Marchioni (ETHZ)
ICARUS
Max. Drift 1.5m (0.5 kV/cm),
to 3 electrode planes
Prompt scintillation
light detected by
WLS PMT’s and used
as a `t0’
10
ICARUS TPC
11
Proof of Principle
The ICARUS project has proven the
principle of the LAr TPC:
 Tracking device with precise event topology
reconstruction
 dE/dx with high density sampling (2% X0) for particle
ID
 Energy reconstruction from charge integration (fullsampling,
fully
homogeneous
calorimeter):
s/E=11%/√E(MeV)+2% : Michel electrons ‹E›=20MeV
s/E=3%/√E(MeV): electromagnetic showers
s/E=30%/√E(MeV): hadronic showers
12
Neutrino Physics Programme
 Neutrino oscillation physics:
atmospheric, solar, neutrino beams
 Proton decay
 Astrophysics: type-II supernovae, early
universe relic neutrinos
 Geo-neutrinos
13
Neutrino Oscillations
 Neutrino mixing: q23, q13, q12, d
 Goals of next oscillation measurements:
-measure q13 (improve on T2K, Nova,)
-measure CP violation in neutrinos
-measure neutrino mass hierachy
14
Neutrino oscillations
e.g. Measuring the `golden channel’
A = 2GF ne is the matter potential;  = Dm21 / Dm31
2
2
Contains information on all parameters we
want to measure (up to degeneracies!)
15
Neutrino Oscillations
Fit oscillation signal as
function of energy – requires
coverage of 1st and 2nd
oscillation peak for required
sensitivity
L=1300km
or
`Counting’ experiments:
Not sensitive to d=0o,
180o
16
Oscillation Facilities

Super beam: p  N  p   X and p   m  n m
to study n m n x next generation long
baseline.
USA(FNAL to Homestake), Japan (T2K
upgrade), CERN to ?
 Beta-beam: n e /n e from high-g beta emitters
(6He,18Ne,8Li,8B), pure flavour, collimated beam,
well understood flux
 Neutrino Factory: muon storage ring,
well understood flux, electron and muon
m   e   n e n m
flavours :
17
Neutrino Factory
CP violation
`Ultimate’ n-oscillation
facility
 12 oscillation processes
available:
Superbeam experiments are only competative
for large q
i.e. sin 2 2q  103
13
13
due to irreducible contamination of nm
beam with ne
18
Detector: General Requirements
• High rates -> scalable to > 10kt
• Reconstruction of charged current
interactions
• Particle identification: leading lepton (e,m) in
CC interactions and separate from pions
nℓ+N→ℓ+hadrons
• Energy resolution: En=Eℓ+Ehad
• Low thresholds
19
Detector: Specific Requirements
Regardless of facility (Superbeam, beta-beam or N F) the ideal detector
would reconstruct all oscillation channels:
 n( )  n( ) ; n( e)  n( e)
m
m
( )
( )
( )
( )
 n m  n e ; n m nt
()
disappearance
appearance
( )
 n e  n m appearance (Golden channel)
 n e  n t appearance (Silver channel)
( )
()
Will probably also need to be multipurpose:
 Proton decay (p->e+ + p0 ; p->K+ + n), supernova neutrinos etc
Highly isotropic: exposure to long baseline oscillations expts. from below,
particle astrophysics expts. from above, p-decay expts. from within
Affordable i.e. simple and scalable
Probably underground (engineering, safety issues)
20
Detector: Specific requirements
 Detectors must be able to discriminate m+/m- and e+/e=> magnetisation!
 e.g. The NF Golden Channel signal is wrong-sign muons:
Major issue for all large-scale detector options
(iron calorimeter, LAr, scintillator) and rules out
water Cherenkov as a NF option
21
Realistic Options
 Water Cherenkov
 Tracking Calorimeter
 Emulsion?
 Liquid argon TPC
Plastic base
1 mm
t
n
Pb
Emulsion layers
22
Water Cherenkov
Electron-like
For:
 Proven technology
 Excellent e-muon separation
Against:
 Only a low En option (0.2-1GeV)
 How to magnetise?
 Relatively poor En resolution
 Rates too high for use as Near Det.
 Kaons below Cherenkov threshold in
p->K+ + n
 Cost – maybe up to 1Mton would be
needed (x20 SuperK)
Muon-like
23
Magnetised Iron Neutrino
Detector: MIND
Iron-scintillator sandwich
(like 9x MINOS)
For: relatively little R&D
Against: Detector optimised for
golden channel at high-E neutrino
factory only (relatively high
thresholds, no electron ID)
L>75 cm
L>150 cm
L>200 cm
24
Totally Active Scintillator Detector:TASD
For:
Tried and trusted
15 m
Like a larger Nova/Minerva
Few mm transverse spatial resolution
Relatively low thresholds (100MeV)
Against:
15 m
1.5 cm
3 cm
Large number of channels –> cost
Magnetise?
R&D needed to prove coextrusion/light levels
Event reconstruction can get complicated –
must match 2D measurement planes
m efficiency
A. Bross et al. arXiv:0709.3889
25
LArTPC:Particle ID
Detector ideal to discriminate e/m/p to
low thresholds
e.g. e/p0 discrimination in appearance: um n e
NC p0 background rendered almost
negligible
1.5GeV electron
1.5GeV p0
26
LArTPC: Proton Decay
 Two main channels:
 LAr is only way to include the kaon
channel to reach ~1035 year limits where
several theoretical models could be
tested
27
A. Marchionni, NP08
LAr/H2O Physics Reach
Study for the FNAL-Homestake (LBNE) project found
~6:1 mass equivalence between water:LAr
28
For:
Liquid Argon TPC’s
 Multipurpose + will deliver oscilln.
program at Superbeam and NF
 True 3D imaging with pixel
size~(x,y,z)=(3mmx3mmx0.3mm)
 High granularity dE/dx sampling - e/g
separation >90% (p0 background to
electrons negligible)
 Total absorption cal sE/E <10%
 Low energy threshold (few 10’sMeV)
 Continuously live
(A. Rubbia NuFact’05)
 Charge and scintillation light readout
Against:
R&D needed:scalability,engineering,purity,
B-field?
29
(FLARE LOI hep-ex/0408121)
Towards Large-Scale LAr TPC’s
* Conclude that LArTPC’s are the best match to the
physics requirements of the next generation of
experiment – but can they be built on scale required?
LArTPC’s that are 50-100 kton (i.e. ~100 times larger than
ICARUS) require:
 Recirculation and purification systems capable of achieving
few x 10’s ppt electronegative impurities
 Longer ionisation charge drift lengths to keep down number
of readout channels per unit volume and dead space (readout
planes and cathodes) => demands HV systems producing drift
fields 0.5-1 kV/cm
 Huge cryogenic vessels leak tight enough to maintain purity
and suitable for underground construction/operation
30
R&D 1: Readout
 ICARUS scheme: 3 wire planes at
different angles, all in liquid phase
 Difficult to scale up without charge
amplification: want S/N ~10 but long wires
give large capacitance, mech. issues etc
Alternatively amplify charge in
argon vapour above the liquid
volume with TGEM/LEMS
 S/N~60, gain of 10 achieved
Cosmic muons
31
R&D 1: Readout Light Imaging TGEM
Planes
 Idea is to optically image the TGEM plane
(array of photosensors, pixel detectors,..)
 LAr test-stands in Sheffield and Warwick
 Shown that SiPM’s work in LAr: JINST 3
P10001(2008)
 Shown that luminescence light produced
based on a single TGEM hole: JINST 4
P04002(2009)
Next steps:
 Demonstrate tracking
 Investigate pixel devices (e.g. fast
CMOS sensors coming out of the LC
effort?)
 NB could reduce readout channels
dramatically and be largely free of
electronics noise -> scalability
32
R&D 2: Electronics
 Per-channel cost of electronics for huge detectors
could be show-stopper
0.35 mm CMOS amp.
working at cryo. temps
(IPNL, Lyon)
Advantages to having front-end digitisation take place
inside cryostat: short connections => lower Cap./noise,
low temp => lower noise)
 Push to develop CMOS ASIC amplifiers
(cheap)
that operate in the LAr at 87K : minimise distance
from readout electrodes to amplifier for S/N~10
 Expect in future digitisers and multiplexers to also be
inside cyrogenic vessel => demands low heat dissipation!
33
R&D 3:LAr Vessels
 Standard stainless steel vacuum dewars not scalable to >10,000 m3
 Huge LNG cryo. vessels with small surface/volume ratios use perlite
or foam glass insulation up to 200,000 m3
 Boiling point LAr and CH4 similar =>boil-off only 0.04%/day for 100
kton vessel
 Ar-gas purging of air (at ppm level) needed before filling: tests
happening at KEK, FNAL (20 t, LAPD) and CERN (6 m3)
Stainless/invar LN2
underground tank
34
R&D 4: LAr Purity
 Impurities enter from outgassing from
contact surfaces : material test facility@FNAL
 Large volumes  all liquid phase circulation
and filtering needed
 ICARUS have demonstrated >10 ms electron
lifetime (needed for 10’s m drift) over several
weeks using commercial Oxysorb/Hydrosorb
filters
 Can this scale? Many pumps in parallel with
small thermal losses
NIM A527 (2004) 329
35
R&D 5:Long Drift
ICARUS drift field 0.5 kV/cm, drift velocity 1.6 mm/ms
ArgonTube:
over 1.5 m – to get 2.0 mm/ms requires 1 kV/cm
To get 1 kV /cm over 10 m drift requires ~ 1 MV
5m drift test
feedthroughs!
@Uni. Bern
ArDM(RE18) generates up to 4 kV/cm internal to LAr
volume and should be scalable
High voltage and purity tests currently under way with
long drift tests at Bern and CERN
LANDD: 5m
drift test
@CERN
36
First Operation LAr TPC in B-Field
37
A. Marchionni, NP08
B-Field
A significant challenge on this scale
 Conventional room temp. magnets too expensive
(power consumption)
 Coventional super-conducting magnets also
probably too expensive due to enormous cyrostats
FNAL investigating use of
superconducting transmission
line technology developed for
VLHC superferric magnets
38
Do We Need a B-Field?
Maybe we can take advantage of the fact that n m  n m !
m+/m- lifetimes in matter different
due to m-capture and no Michel decay
electron. Already used by MiniBooNE
(n) and Kamiokande (cosmic muons)
MiniBooNE hep-exp/0602051
Muon angle wrt neutrino direction
sensitive to nhelicity (used by MiniBooNE)
Outgoing nucleon (p or n)
39
LAr Proton Tagging
Ability of a LArTPC to reconstruct
protons already proven by 50L
(ICARUS) prototype exposed to the
CERN WANF neutrino beam
physics/0609205 (2006)
40
Do We Need a B-Field?

Only statistical separation of n /n possible but moderate efficiencies and
purities up to 90% should be possible (combine variable?):
Huber, Schwetz, Hep-ph/08052019
P=0(1) no(perfect) n
discrimination
/n
Iron calorimeters are no use at a low energy NF (muon tracks too short at
E<5GeV) and magnetisation puts a limit on the ultimate size of detectors 
these methods may enable a huge multipurpose LAr detector with a broad
programme (oscillations, p-decay, astrophysics etc)
Needs detailed simulation study !
41
100kt Main
concept- designs
42
Main Design Concepts I: GLACIER
43
44
45
Reconstruction
 Algorithms not well developed –
partly historical but also, it’s not
so easy!
 Tracks and showers develop
side-by-side in the same volume
 topologically complicated
 No well defined start point for
what initiated the event
 Very high density of
information: mm-scale energy
deposits, delta-rays, vertices,
kinks etc
 Multiple scattering occurring
continuously throughout volume
ICARUS, arxiv:0812:2373
46
c.f. Accelerator experiments
 Relatively sparse space/pulseheight data points radiating from
this point
 Tracks and showers develop in
separate, optimised, sub-detectors
 Well defined interaction point
 Multiple scattering happening
mostly at well-defined boundaries
between sub-detectors
 Track search within a welldefined model (circle or helix) to
decide on associated hits
47
Past developments: Hough transform
 Part of effort to automate bubble chamber reconstruction
(Hough: Proc. Conf. High Energy Acc. Instr., CERN 1959)
y = mx  c
 cos    r 
y = 
x  

 sin    sin  
r = y sin   x cos 
 A sinusoid in (r,q)
points
(x,y) intersecting at the
parameters of straight line
structures
A
 Found wide application as feature extraction technique
48
for image analysis
Clustering
 DBSCAN* algorithm: the `density-
neighbourhood’ (e) around each point in
the cluster must contain at least Nmin
other points
 Cellular automaton*: 3D
implementation for chargedcurrent interactions in LAr
Raw hits
um  n  m   p
Kinga Partyka (Yale/ArgoNEUT)‫‏‬
Eu = 0.7GeV
Clustered
* Sander et al., Data Mining and knowledge Discovery 2, pp169-194 (1998)
* Warwick group
49
Examples/issues
um  n  m   p
Hough transform:
end-points
DBSCAN: high density
clustering
Delta electrons
50
Corner Finding
 Harris-Stephens/Plessey
Function*: `cornerness’ measure
where size/direction governed by the
Eigenvalues/Eigenvectors of
structure tensor, A
 Dx 
H ( x, y ) = (Dx Dy )A 
 Dy 
 ( / x )2
( / x )( / y ) 

A = 
2
( / y ) 
 ( / y )( / x )
HSP Transform
Picks out:
Vertex AND Track Ends!
* C. Harris, M. Stephens, proc. 4th Alvey Vision Conf. pp147-151 (1988)
51
Corner Finding
 GENIE generated nm CCQE events in 3T LAr TPC:
Vertex picked out
Proton Stop
Delta Electron ID!
Ben Morgan, Warwick, JINST 5 P07006 (2010)
52
Technology Choice
• These studies to feed into the international
programme for next generation project: IDS-NF,
LAGUNA-LBNO, LBNE etc
Phys. Rev. D81 073010 (2010)
53
Latest Results – Neutrino’10
ArgoNeuT:
175L prototype in
NUMI beam
infront of
MINOS
ICARUS T600: starting to collect events
in CNGS beam – analyses to find t’s
250 L prototype
in 340 MeV/c K
beam @JPARC
54
Europe
55
56
EU projects
• EUROnu(FP7 design study for neutrino oscillation facility in Europe):
- Machine and target R&D
– Detector simulation studies: MIND (Neutrino Factory), water
Cherenkov detector (for Super-Beam and Beta Beam), scintillator
and near detector (all facilities)
– No detector R&D funded
• LAGUNA(FP7 design study for large proton decay + neutrino
astrophysics):
– Large underground chambers for Liquid Argon or Water Cherenkov
detectors: site evaluation and construction
– Detector studies: water Cherenkov, liquid scintillator, liquid argon
- LAGUNA-LBNO proposal: includes CERN superbeam R&D
– No detector R&D funded
• AIDA(Euro Integrating Activity Project):
– test beam infrastructure at CERN for neutrino detector
prototyping (MiniMIND)
57
58
Status of UK Activity
 Some small-scale LAr test-stand R&D at Liverpool, Sheffield
and Warwick (all unfunded)
 Important to keep all options open at present until decisions on
the next generation of neutrino project are made – this may be
around 2013 coinciding with:
- final reports of international studies such as IDS-NF and
EuroNu
- results from T2K and reactor experiments on the size of q13
(a `large’ value would boost superbeam projects)
 UK groups very active in the European initiatives: IDS-NF,
EuroNu, LAGUNA-LBNO, AIDA concerning machine studies,
underground site development, physics studies etc
 Close links also maintained with the US LBNE programme (LAr
software, electronics) and in Japan (T2K, 250L prototype
reconstruction)
 A measurable q13 would see momentum grow for : T2K upgrade in
Japan or LBNE in the USA or LBNO in Europe , all hopefully
59
incorporating a LArTPC!
Outlook+Conclusion
 A ~100kt LArTPC is the best-performing
detector option for a next generation
neutrino facility
 Possible only only one huge detector will be
built => important it is multipurpose (n
oscillations, p-decay and astrophysics) LArTPC in good position to deliver
 Whether it gets built depends on solving
remaining tech. challenges before deadlines
like IDS in 2012-13 and whether value of q13
60
warrants building superbeam or NF
Current Global Status
 Europe/Japan (base on IDS report),
including European initiatives, CERN
beam upgrade etc
 US (base on IDS report)
61
62
Large Electron Multiplier LAr TPC
63
A. Marchionni, NNN09
LEM LAr TPC Performance
64
A. Marchionni, NNN09
The ArDM Dark Matter Experiment @
CERN
65
A. Marchionni, NNN09
66
lllkkkkkkkkkkkkkkkkkkkk
67
kkkkkkkkkkkkkkkkkk
ArDM Status
 Filled in 2009 with 1 tonne LAR
(no fields, charge readout or
recirculation)
 Light readout worked well
 Scint. Light slow component time
constant confirmed at ~ 1.5 ms
indicating excellent purity without
circulation
To come this year:
Top TGEM
 Purity measurement in drift fields up to
1 kV/cm
 Recirculation filtering
 Double phase LEM TPC readout scaled up
to 10x10 cm2, 16 strips 6mmm wide
68
LAr Scale-up: Pros and Cons
 A modular approach is usually more expensive and
each module having its own fiducial volume
complicates matters
 A two-phase readout in a single, large, LNG volume
with drift up 20 m will need substantial charge gain to
replace impurity losses and Ar± build-up at gas-liquid
interface will affect the electron ionisation signal
 Motivation for approach based on single phase,
relatively short drift distance, imaging luminescence
light emitted from the TGEM holes?
69
Engineering
• Rubbia venice doc has details: LNG gas
tanks, cavern more difficult than the tank
construction(site selection details of
laguna in NNN09 talk), operated in boiling
mode with good insulation you only need
to top up the evaporative losses….
70
A. Rubbia J. of Phys. 171 (2009) 012020
Challenges and status
 Large engineering
 Long drift/ Purity
 B-field – maybe not needed, quote the
Huber paper
 Reconstruction
71
Long Drift
Test signal attenuation and diffusion over 5m drift
Simulate `very long’ drift (~ 20m) by reduced E-field and purity
High voltage test (up to 500 kV)
Measure Rayleigh scattering length and attenuation length vs purity
LANDD 5m@CERN
ArgonTube 5m @Bern
University
Infrastructure ready
Will take data this year
Will take data this year
72
Other contenders: Water Cher.,
scintillator, iron cal.
•
•
•
•
•
Use my slides from PPAP to show disadvantages of other techs
Use soderberg numbers for lar cf H2o equiv. Mass (aprox. 6:1 mass
advantage fro Lar over H2O for CP, mass hierarchy, theta_13)
Iron cal can’t go down to low thresholds(LENF doc)
Huber non B-field points out TASD has huge num. Readout channels and
similar issue with b-field as Lar
Tech choice in 2012-13 ?
73
74
75
76