Post on 30-Dec-2015
description
18 May 2006 D W Miller Chicago 1
High-Energy Axion Detection Probing hadronic axions with
the CAST calorimeter
I Origin of hadronic axion models and constraints
II Axion emission by nuclear de-excitation
III The CAST γ-calorimeter amp results
Juan I Collar David W Miller Joaquin D VieiraThe University of Chicago
18 May 2006 D W Miller Chicago 2
KSVZTree-level couplings to quarks
suppressedrarr Evades constraints of DFSZ model
rarr Nucleon coupling still possible
Origin of hadronic axions
bull The PQ Weinberg amp Wilczek axion was pinned to the weak scale resulting in ma~ 01-1 MeVndash Quickly ruled out experimentally
(eg quarkonium and kaon decays)
bull New models developed to evade such constraintsndash Adjustment of the PQ scale (fa rarr gt106 GeV)ndash Modification of couplings
DSFZAxion couples to Fermions and photons
rarr Less experimentally favored
18 May 2006 D W Miller Chicago 3
Hadronic axion window
bull SN1987A hadronic axion constraints (via gaNN)ndash For ~10-2 lt malt ~2-5 eV axions
would have carried away too much energy from SN1987a
bull The so-called ldquohadronic axion windowrdquo is ~2-5 lt malt ~20 eV
ldquoExperiments probing distinct properties of axions
even if redundant in the context of a particular
model provide independent constraints on theories incorporating the PQ
symmetryrdquo[1]Telescopelab limits in
gaγγ vs ma space[1] F T Avignone III et al PRD37 618 (1988)
18 May 2006 D W Miller Chicago 4
Axion-nucleon coupling gaNN
bull gaNN derives from axion-light meson mixing
ndash independentindependent of the U(1)PQ charges that are implicit in KSVZ (ie hadronic) axion models
bull As a pseudoscalar object carries JP = 0- 1+ 2- hellipndash M1 nuclear transitions could (should) emit axions
bull Axion-nucleon coupling explicitly as
Where g0 and g3 are the isoscalar and isovector axial-current contributions respectively
aggi
agiL
NN
NaNNNaNN
)(
3305
5
18 May 2006 D W Miller Chicago 5
Nuclear de-excitation via axionsbull For anyany (eg thermally)
excited M1 nuclear transition
drrrNDrdE
dΦ sunR
aEa
a 2
02 4 )(
4
1
]s [g 1 1-1-
τBNN a
Na
2
10
10
3
)(212
1
gg
k
kaa
DD = Gaussian DopplerBroadening term
Axion emission per gram solar mass per sec
BB = the Boltzmann factor for thermally excited nuclear state
NNNN = density of the isotope
Axion-photon branching ratio
ττγγ = lifetime of excited nuclear state
ALL of the decay channel-specific axion physics is
contained in β and η rarr
nuclear structure dependent terms
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 2
KSVZTree-level couplings to quarks
suppressedrarr Evades constraints of DFSZ model
rarr Nucleon coupling still possible
Origin of hadronic axions
bull The PQ Weinberg amp Wilczek axion was pinned to the weak scale resulting in ma~ 01-1 MeVndash Quickly ruled out experimentally
(eg quarkonium and kaon decays)
bull New models developed to evade such constraintsndash Adjustment of the PQ scale (fa rarr gt106 GeV)ndash Modification of couplings
DSFZAxion couples to Fermions and photons
rarr Less experimentally favored
18 May 2006 D W Miller Chicago 3
Hadronic axion window
bull SN1987A hadronic axion constraints (via gaNN)ndash For ~10-2 lt malt ~2-5 eV axions
would have carried away too much energy from SN1987a
bull The so-called ldquohadronic axion windowrdquo is ~2-5 lt malt ~20 eV
ldquoExperiments probing distinct properties of axions
even if redundant in the context of a particular
model provide independent constraints on theories incorporating the PQ
symmetryrdquo[1]Telescopelab limits in
gaγγ vs ma space[1] F T Avignone III et al PRD37 618 (1988)
18 May 2006 D W Miller Chicago 4
Axion-nucleon coupling gaNN
bull gaNN derives from axion-light meson mixing
ndash independentindependent of the U(1)PQ charges that are implicit in KSVZ (ie hadronic) axion models
bull As a pseudoscalar object carries JP = 0- 1+ 2- hellipndash M1 nuclear transitions could (should) emit axions
bull Axion-nucleon coupling explicitly as
Where g0 and g3 are the isoscalar and isovector axial-current contributions respectively
aggi
agiL
NN
NaNNNaNN
)(
3305
5
18 May 2006 D W Miller Chicago 5
Nuclear de-excitation via axionsbull For anyany (eg thermally)
excited M1 nuclear transition
drrrNDrdE
dΦ sunR
aEa
a 2
02 4 )(
4
1
]s [g 1 1-1-
τBNN a
Na
2
10
10
3
)(212
1
gg
k
kaa
DD = Gaussian DopplerBroadening term
Axion emission per gram solar mass per sec
BB = the Boltzmann factor for thermally excited nuclear state
NNNN = density of the isotope
Axion-photon branching ratio
ττγγ = lifetime of excited nuclear state
ALL of the decay channel-specific axion physics is
contained in β and η rarr
nuclear structure dependent terms
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 3
Hadronic axion window
bull SN1987A hadronic axion constraints (via gaNN)ndash For ~10-2 lt malt ~2-5 eV axions
would have carried away too much energy from SN1987a
bull The so-called ldquohadronic axion windowrdquo is ~2-5 lt malt ~20 eV
ldquoExperiments probing distinct properties of axions
even if redundant in the context of a particular
model provide independent constraints on theories incorporating the PQ
symmetryrdquo[1]Telescopelab limits in
gaγγ vs ma space[1] F T Avignone III et al PRD37 618 (1988)
18 May 2006 D W Miller Chicago 4
Axion-nucleon coupling gaNN
bull gaNN derives from axion-light meson mixing
ndash independentindependent of the U(1)PQ charges that are implicit in KSVZ (ie hadronic) axion models
bull As a pseudoscalar object carries JP = 0- 1+ 2- hellipndash M1 nuclear transitions could (should) emit axions
bull Axion-nucleon coupling explicitly as
Where g0 and g3 are the isoscalar and isovector axial-current contributions respectively
aggi
agiL
NN
NaNNNaNN
)(
3305
5
18 May 2006 D W Miller Chicago 5
Nuclear de-excitation via axionsbull For anyany (eg thermally)
excited M1 nuclear transition
drrrNDrdE
dΦ sunR
aEa
a 2
02 4 )(
4
1
]s [g 1 1-1-
τBNN a
Na
2
10
10
3
)(212
1
gg
k
kaa
DD = Gaussian DopplerBroadening term
Axion emission per gram solar mass per sec
BB = the Boltzmann factor for thermally excited nuclear state
NNNN = density of the isotope
Axion-photon branching ratio
ττγγ = lifetime of excited nuclear state
ALL of the decay channel-specific axion physics is
contained in β and η rarr
nuclear structure dependent terms
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 4
Axion-nucleon coupling gaNN
bull gaNN derives from axion-light meson mixing
ndash independentindependent of the U(1)PQ charges that are implicit in KSVZ (ie hadronic) axion models
bull As a pseudoscalar object carries JP = 0- 1+ 2- hellipndash M1 nuclear transitions could (should) emit axions
bull Axion-nucleon coupling explicitly as
Where g0 and g3 are the isoscalar and isovector axial-current contributions respectively
aggi
agiL
NN
NaNNNaNN
)(
3305
5
18 May 2006 D W Miller Chicago 5
Nuclear de-excitation via axionsbull For anyany (eg thermally)
excited M1 nuclear transition
drrrNDrdE
dΦ sunR
aEa
a 2
02 4 )(
4
1
]s [g 1 1-1-
τBNN a
Na
2
10
10
3
)(212
1
gg
k
kaa
DD = Gaussian DopplerBroadening term
Axion emission per gram solar mass per sec
BB = the Boltzmann factor for thermally excited nuclear state
NNNN = density of the isotope
Axion-photon branching ratio
ττγγ = lifetime of excited nuclear state
ALL of the decay channel-specific axion physics is
contained in β and η rarr
nuclear structure dependent terms
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 5
Nuclear de-excitation via axionsbull For anyany (eg thermally)
excited M1 nuclear transition
drrrNDrdE
dΦ sunR
aEa
a 2
02 4 )(
4
1
]s [g 1 1-1-
τBNN a
Na
2
10
10
3
)(212
1
gg
k
kaa
DD = Gaussian DopplerBroadening term
Axion emission per gram solar mass per sec
BB = the Boltzmann factor for thermally excited nuclear state
NNNN = density of the isotope
Axion-photon branching ratio
ττγγ = lifetime of excited nuclear state
ALL of the decay channel-specific axion physics is
contained in β and η rarr
nuclear structure dependent terms
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 6
Decay channels and signal
bull A few M1 transition decays available from solar core p + d rarr 3He + a (55 MeV) 7Li rarr 7Li + a (048 MeV) 23Na rarr 23Na + a (044 MeV)
bull Give mono-energetic axions at the decay energyndash Detector signal will of course
include effects such as escape peaks (for Ea gt 12 MeV)
bull γ-ray energy axions have improved probability of conversion for a helioscope
22
22 )cos(21
)2(aaa EqL
q
BgP
Low-E axions(5 keV)
High-E axions(1-60 MeV)
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 7
The CAST high-energy calorimeter
Detector design goalsbull Maximize sensitivity to γ-ray
conversion photonsndash A dense crystal works
well
bull Maintain minimalist design due to CAST constraintsndash Minimal lead shielding +
muon veto
bull Search for several decay several decay channels and other possible channels and other possible new pseudoscalar bosonsnew pseudoscalar bosonsndash So maintain good photon
efficiency amp dynamic range for E gt 100 MeV
Chicago calorimeter
adjustable platform for alignment
MicroMegas Detector (transparent above
~102 keV)X-ray Telescope
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 8
Large scintillating crystal (CdWO4)
Very pure amp high efficiency Low-background PMT Offline particle identification Env radon displacement Plastic muon veto Thermal neutron absorber Low 02 MeV energy threshold ~100 MeV dynamic range Compact XIA Polaris - Digital
Gamma Spectrometer
CWO PMT Power
Lead shielding
Low-backgroundAncient Lead
Plastic scintillatorMuon veto
22 cm58 cm
The CAST high-energy calorimeter
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 9
Front ViewFront View
Side ViewSide View
~4~4Plastic Muon Plastic Muon
VetoVeto
CWO CWO CrystalCrystal
Light Light guideguide
Brass support Brass support tubetube
Shielded Shielded electronicselectronics
Ultra-low bckg PbUltra-low bckg Pb
GammasGammas(from magnet bore)(from magnet bore)
Pb shieldingPb shieldingMuon veto PMTMuon veto PMT
Characteristic pulse
~16srate~4 Hz
Radon Radon displ amp displ amp borated borated thermalthermal
n absorbern absorber
Large CWO (good Large CWO (good particle ID radiopure particle ID radiopure efficient)efficient)
Low-BCKG Low-BCKG PMTPMT
22 cm
58 cm
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 10
Particle identification
α γ separation from detector calibrations
αγ
We actually expect very
few α events since CWO so pure
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 11
A Data acquisitionndash Digital waveform acquisition 40 MHzndash Muon veto coincidence rejection (95
of μ events)
B Offline processingndash Livetime calculation via LED pulserndash Particle identification cutsndash Correction for detector systematics
(temp position)ndash BCKG 210-6 cm-2 s-1 kev-1
C Background subtractionD Limits on possible anomalous events
ndash Look for Gaussian signals at low energies
ndash Look for complex signal shape (including photon escape peaks) at higher energies
Calorimeter data and operation
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 12
bull Signal Gaussian peaks E lt 10 MeVndash E gt 10 MeV the
functional changes due to photonuclear reactions
bull Obtain 95 CL (2σ) for allowed anomalous events at each energy
bull Any signal after subtraction could be a hint towards new physicshellip
Tracking dataBackground data
Best fit (signal)Best fit (bckg)Best fit (sig+bckg)
95 CL peak
Looking for evidence in the data
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 13
bull This is trickyhellipndash Production mechanism
(generally) relies on nuclear coupling
ndash Detection mechanism relies on photon coupling
bull It is possible but unlikely that there are cancellations in the PQ symmetry that cause gaγγ~0 but gaNN ne 0 hellipin which case we can stop talking now [2]
aaa
γaγγ
mPg
)(
To derive this limit we must (a)(a) calculate the expected axion flux or (b)(b) make an assumption about
what it is
aγγaaaγ gP m 2)(
Limits on axion physics
[2] D B Kaplan Nucl Phys B260 215 (1985)
(a) In order to calculate Фa
a
total
M12
1
4
1
Ea r
Гa Гγ most difficult p + d rarr 3He + a we have
concentrated most on this reactionhellipassumption (b) used now direct calc
7Li rarr 7Li + a 23Na rarr 23Na + a
Calcs in progress
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 14
Improved ldquolimitsrdquo using
max Φa assumption for 55 MeV axion
Self-consistent improved limits possible for several energies of interest eg 55 MeV by using Фa MAX hellip but we are in
the process of calculating Фa for several channels
(b) Assume Фa MAX aaa
γaγγ
mPg
)(
Results from the calorimeter 55 MeV(a) Of course we want an
expression for Фa bull η β almost impossible to calculate (3-body nuclear decay)bull Perhaps we can simplify the expression Not sure yet
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 15
Conclusions
bull Axion-nucleon coupling will give rise to axion-emission through M1 transitions
bull We have built a gamma-ray detector that is sensitive to several axion decay channels
bull In a generic search in the range 03 ndash 60 MeV no statistically significant signal has been seen
bull To correctly set limits on the axion parameters (fPQ) more careful calculations of nuclear structure dependent parameters are necessaryhellipIN PROGRESS
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 16
Backup slides
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 17
Details for this data set
Data taking period (2004) 1509 ndash 0811
Total Running TimeTotal Running Time 1257 hrs (53 days)1257 hrs (53 days)
Tracking TimeTracking Time 603 hrs (25 days)603 hrs (25 days)
Total Background Time 898 hrs (37 days)
Normalized Background TimeNormalized Background Time 1173 hrs (5 days)1173 hrs (5 days)
Systematics Time (valves open quencheshellip) 299 hrs (12 days)
Ratio Norm BCKG to Total BCKG 013
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 18
Data processing and results
Data treatmentData treatment
ResultResult
data data keptkept
BCKG BCKG Count rate Count rate
(Hz)(Hz)
Integ Integ ΦΦ
03-50 MeV03-50 MeV (cm(cm-2 -2 secsec-1-1))
Raw data 100 382 0263
Anti-coincidence with muon veto 634 242 0167
Recursive 40K peak gain shifting
(temperature correlated)634 242 0167
PSD analysis and cuts
(incl removal of livetime pulser)374 143 01
FULL DATA TREATMENTFULL DATA TREATMENT 374374 143143 0101
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 1901
1
1 10
MCNP calculated full-energy (peak)efficiency for collimated axion-induced gammas
CdWO4(r=215l=12)CdWO4(r=215l=75)CdWO4(r=215l=5)
pe
ak
effi
cie
ncy
E (MeV)
chosen size (cost and self-absorption also an issue)0
50
100
150
200
250
300
350
400
16 18 2 22 24 26 28
CWO crystal (l=75cm) in Pb shielding (R=475cm)response to 1 MeV isotropic bckg
and collimated axion beam signal (MCNP MonteCarlo)
peak bckgtotal bckgsignal
in
cre
ase
re
lativ
e to
R=
17
5 c
m
crystal radius (cm)
(slig
htly
larg
er
than
bor
e)
signal noiseworse for larger r -gt
001
01
1
1 10
efficiency for different crystals
BGO(r=215l=12)C6F5I(r=215l=20)BaF2(r=215l=12)YAGCe(r=215l=12)YAPCe(r=215l=12)
GSOCe(r=215l=12)NaI(r=215l=12)CdWO4(r=215l=12)LSO(r=215l=12)PWO(r=215l=12)
full-
ener
gy e
ffici
ency
(1
= 1
00
)
gamma energy (MeV)
(beh
ind
LSO
)
Crystal selection and Monte Carlo
bull Tested CWO BGO BaF2
bull MC CWO BGO BaF2 PWO YAG LSO NaIhellip
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 20
Detector Parameters
Resolution versus energy Efficiency for full energy deposition
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 21
Energy Spectrumbull Without position normalized background
datandash Apparent agreement butbut systematic effect systematic effect
due to the pointing positiondue to the pointing position of the magnet as in TPC
bull With position normalizationndash Error bars increase by factor x2ndash Systematic effect of position is removed effect of position is removed
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 22
Software cuts
bull Use γ calibrations to determine software cutsndash Keep 997
bull Of the events above 300 keVhellipthreshold set due to noise events + BCKG
bull Set cuts forndash Energy
ndash Shape of Pulsebull PID = pulse identification
parameter
ndash Pulse rise time
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 23
Look for evidence buried in databull Signal mono-energetic peaks
below 10 MeVndash Width determined by measured
detector resolutionbull Obtain 95 CL (2σ) for
allowed anomalous events at each energy
bull Above 10 MeV the functional form used becomes more complex (effect of photonuclear reactions)
bull Dedicated search for specific energy including escape peaks in fit yields better sensitivity than simple gaussian and full-E efficiency (a first-order approach)
Example illustrated here iExample illustrated here is s emission at 55 MeV emission at 55 MeV
(p + d (p + d He + a hellip a He + a hellip a )) G Raffelt and L Stodolsky Phys Lett B119 323 (1982)
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 24
Residual spectraDifference between signal (solar tracking) and background
Low energy03 ndash 3 MeV
Mid energy3 ndash 10 MeV
High energy10 ndash 50 MeV
3 energy regions to allow for different binning based on detector resolution
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts
18 May 2006 D W Miller Chicago 25
From counts to photons
bull Use the 95 CL allowed counts at each energy and convolve withndash Detector efficiency (ε)
ndash livetime (t)
γt
detlivetime
counts