Search for axion-like particles using a variable baseline photon regeneration technique

0.18 MB·English

by  A. S. Chou

FERMILAB-PUB-07-555-A-CD-E-TD
Search for axion-like particles using a variable baseline photon regeneration technique
A. S. Chou1,2, W. Wester1, A. Baumbaugh1, H. R. Gustafson3, Y.
Irizarry-Valle1, P. O. Mazur1, J. H. Steffen1, R. Tomlin1, X. Yang1, and J. Yoo1
1Fermi National Accelerator Laboratory, PO Box 500, Batavia, IL 60510
2Center for Cosmology and Particle Physics, New York University, 4 Washington Place, New York, NY 10003
3Department of Physics, University of Michigan, 450 Church St, Ann Arbor, MI 48109
(Dated: February 2, 2008)
We report thefirst results of the GammeV experiment, a search for milli-eV mass particles with
axion-likecouplingstotwophotons. Thesearchisperformedusinga“lightshiningthroughawall”
technique where incident photons oscillate into new weakly interacting particles that are able to
pass through the wall and subsequently regenerate back into detectable photons. The oscillation
baseline of the apparatus is variable, thus allowing probes of different values of particle mass. We
8
find no excess of events above background and are able to constrain the two-photon couplings of
0 possible new scalar (pseudoscalar) particles to be less than 3.1×10−7 GeV−1 (3.5×10−7 GeV−1)
0
in the limit of massless particles.
2

n PACSnumbers: 12.20.Fv,14.70.Bh,14.80.Mz,95.36.+x
a
J

6 Recently, the PVLAS experiment reported a positive New low-mass particles with masses smaller than the
1 signal in a photon oscillation experiment [1]. In their electronmass haveyet to be directly detected, but mod-

“disappearance” experiment, a polarized 1064 nm laser els containing such light degrees of freedom have been
]
x beamwassentintoanFabry-Perotcavitywhichalsocon- motivatedbythesmallneutrinomassdifferences[9][10],
e tained a high transverse magnetic field. By measuring by the vacuum dark energy density of (2 meV)4 [11],
-
p a net rotation of the laser polarization vector upon ex- and rather generically by string theory [12] and theo-
e iting the cavity, PVLAS found a non-zero relative at- ries with large extra dimensions [13]. Couplings of light
h tenuation of laser polarizationcomponents transverse to particles to photons are tightly constrainedby star cool-
[
and parallel to the external magnetic field. This result ing considerations, including the experimental limits set

2 was interpreted as evidence for the production of new by the CAST axion helioscope [14] whose recent limit
v spin-0 particles from one of the polarizationcomponents g <8.8×10−11 GeV−1 strongly excludes PVLAS. Vari-
3
[2]. Newer PVLAS results reported at conferences [3] ousmethodshavebeenpostulated,however,toevadethe
8
included the measurements of non-zero polarization ro- star cooling limits [11][15][16].
7
3 tation with 532 nm green light, and non-zero ellipticity InthisletterwereportthefirstresultsfromGammeV,
. at both wavelengths, where the relative phase delay is a photon regeneration experiment similar to that origi-
0
1 induced by loop effects. nally suggested in [17]. Our apparatus is specifically de-
7 The effective axion-like interaction Lagrangian for a signed to quickly probe the region in parameter space
0 pseudoscalar particle is: suggestedby the PVLAS results, at modest expense. As
:
v L =−gφF Fµν =gφ(E~ ·B~) (1) an “appearance” experiment rather than a “disappear-
i pseudoscalar 4 µν ance” experiment, a positive signal would yield an un-
X
ambigious new interpretation of oscillations of photons
while that for a scalar particle is:
r e
a g into new, weakly interacting particles.
L =− φF Fµν =gφ(E~ ·E~ −B~ ·B~). (2)
scalar µν
4
WhilePVLASwasnotabletodeterminetheparityofthe
EXPERIMENTAL DESIGN
new particle, their four measurements gave a consistent
picture of a low scalar mass m ∼1.2 meV and a rather
φ
large two-photon coupling g ∼ 2.5×10−6 GeV−1. The The key to this experiment is the short 5 ns, 160 mJ
suggested PVLAS region of interest was not previously pulses of 532 nm light emitted with a repetition rate
excludedbythepioneeringBFRTlaseroscillationexper- of 20 Hz by our light source, a frequency-doubled Con-
iment[4][5]. However,forreasonsunknown,thePVLAS tinuum Surelite I-20 Nd:YAG laser. As described be-
signaldisappeared after the experimental apparatuswas low, the small duty cycle enables a large reduction in
later rebuilt in an effort to improve the detection [6]. A the detector noise via coincidence counting. The laser
number of experimental efforts world-wide have begun light is vertically polarizedand when needed, a halfwave
to test the hypothesis that the anomalous PVLAS ob- plate is usedto obtainhorizontalpolarization. The laser
servationsaredue to anew axion-likeparticle[7],andin pulses are sent through a vacuum system (diagrammed
particulartheBMVexperimenthasrecentlyexcludedthe in Fig. 1) designed around an insulating warm bore in-
pseudoscalar interpretation of the PVLAS signal[8][23]. serted into a 6 m Tevatron superconducting dipole mag-
2
net. The magnet produces a 5 T vertical field uniform
across the aperture of the 48 mm inner diameter warm
bore. A “wall” consisting of a high-power laser mir-
ror on the end of a long (7 m) hollow stainless steel
“plunger” is inserted into the warm bore. The mirror
maybe placedatvariouspositions withinthe magnetby
sliding the plunger. The plunger mirror projects the re-
flectedwaveintoaphotonstateandthetransmittedwave
into a (pseudo-)scalar state, provided that the scalar is
sufficiently weakly-interactingto pass throughthe mate-
FIG. 1: Diagram (not to scale) of the experimental appara-
rial of the mirror. The mirror also functions to reflect tus. Theinitialvacuumchamberconsistsofa10minsulating
the incident laser power out of the magnet to prevent warm bore which is offset by 1.6 m from the end of the 6 m
heating of the magnet coils. The mirror is mounted on magnetic field region, and is sealed tothesliding plungervia
a welded stainless steel cap on the end of the plunger a double o-ring assembly. The sliding plunger has a range
of motion of 1.9 m, and contains an independent vacuum
in order to prevent stray photons from passing through.
chamber. The vacuum window at the far end slides within
Thus, the beam passing through the end of the plunger
a stationary, long dark box.
is a pure scalar beam. These scalars can then oscillate
back into photons through the remaining magnetic field
region inside the 35 mm inner diameter hollow plunger.
region.
Upon exiting the magnetic field region, the interaction
As can be seen from Eqn. 5, the meter scale baseline
ceasesandthephoton-scalarwavefunctionisfrozen. This
combined wavefunction then propagates ∼ 6 m into a providedbytypicalacceleratormagnetsiswell-suitedfor
probing the milli-eV range of possible particle masses.
dark box where a Hamamatsu H7422P-40 photomulti-
This fact can be a curse as well as a boon because for a
pliertube(PMT)moduleisusedtodetectsinglephotons
monochromatic laser beam, a fixed magnet length may
incoincidence with the laserpulses. As describedbelow,
accidentally coincide with a minimum in the oscillation
a high signal to noise ratio is achieved due to the very
ratherthanamaximum. Indeed,thisisapossiblereason
short pulses emitted by the laser.
whytheBFRTexperiment[5]didnotseethePVLASsig-
The photon-scalar transition probability may be writ-
nal even though they had similar sensitivity. GammeV’s
ten in convenient units as:
plunger design allows us to change the oscillation base-
Pγ→φ = M42(B∆2mω22)2 sin2 ∆m4ω2L (3) lminaessanindtthheusmsiclalin-etVhrroaunggheawllitphoosustibalenyvarleugeisonofstohfedsicmailnar-
(cid:18) (cid:19)
4B2ω2 m2L ished sensitivity. The total conversion and regeneration
≈ sin2 φ (4) probabilitycontainstwofactorsofEqn.4,corresponding
M2m4 4ω
φ ! to the pre-mirror and post-mirror magnetic field regions
= 1.5×10−11(M/10(B5/GTeeVsl)a2)(2m(ωφ//e1V0−)23 eV)4 osifn2lenmgt42φhωLs1L1sina2ndmL42φω2L.2 TwhheetroetLal1p+roLb2a=bil6itym.varies as
×sin2 1.267(mφ/10−3 eV)2(L/m) (5) To(cid:16)detec(cid:17)t rege(cid:16)nerated(cid:17) photons we use a 51 mm di-
(ω/eV) ameter lens to focus the beam onto the 5 mm diame-
(cid:18) (cid:19)
ter GaAsP photocathode of the PMT. The alignment is
where B is the strengthof the externalmagnetic field, ω
performed using a low power green helium-neon align-
is the inital photon energy, L is the magnetic oscillation
ment laser and a mock target. The alignment is verified
baseline. Themass-squareddifferencebetweenthescalar
both before and after each data-takingperiod by replac-
mass and the effective photon mass, ∆m2 = m2 −m2,
φ γ ing the sealed plunger with an open-ended plunger, re-
characterizesthe mismatch of the phase velocities of the
establishing the vacuum, and firing the Nd:YAG laser
photonwaveandthemassivescalarwaveanddetermines
ontoaflashpapertarget. Anopticaltransportefficiency
the characteristic oscillation length. While the photon
of 92% is measured using the ratio of laser power trans-
doesnotreallygainamassinanormaldielectricmedium,
mitted throughthe open-endedplunger and throughthe
the phase advance may be modelled with an effective
various optics and vacuum windows, to the initial laser
imaginarymassm2 =−2ω2(n−1), where n is the index
γ power, using the same power meter in both cases to re-
ofrefraction[18]. Boththewarmboreandtheinteriorof
move systematic effects. The quantum efficiency of the
the plunger are pumped to moderate vacuum pressures
photocathode is factory-measuredto be 38.7% while the
ofless than 10−4 Torr, and a conservativeestimate gives
collection efficiency of the metal package PMT is esti-
−m2γ < 10−4 eV. Therefore, starting with Eqn. 4 we mated to be 70%. The PMT pulses are amplified by
aqssume that the contribution from the effective photon 46 dB and then sent into a NIM discriminator. Us-
massisnegligibleforlargervaluesofm nearthePVLAS ing a highly attenuated LED flasher as a single photon
φ
3
3 Configuration # photons Est.Bkgd Candidates g[GeV−1]
2 Horiz. pol., center pos. Horiz.,center 6.3×1023 1.6 1 3.4×10−7
Horiz.,edge 6.4×1023 1.7 0 4.0×10−7
1 Vert.,center 6.6×1023 1.6 1 3.3×10−7
0 Vert.,edge 7.1×1023 1.5 2 4.8×10−7
Horiz. pol., edge pos.
2
s
5 n 1 TABLE I: Summary of data in each of the4 configurations.
2
er 1. 0 Vert. pol., center pos.
p
s 2 time as the straight-through photons since milli-eV par-
e
s
ul 1 ticles are also highly relativistic. For coincidence count-
p
T 0 ing between the laser pulses and the PMT, a 10 ns wide
M Vert. pol., edge pos.
P 2 window is chosen and includes 99% of the measured
photon time distribution shown in Fig. 2. The coinci-
1
dent dark count rate can be estimated to be R =
noise
6000 Leaky mirror 20 Hz×130 Hz×10 ns=2.6×10−5 Hz. This noise rate
400 is negligibleto the expected signalrateof∼2×10−3 Hz
200 estimated from the central values of the PVLAS param-
0 eters.
0 25 50 75 100
Time(PMT photon relative to Laser) (ns)
DATA COLLECTION
FIG. 2: PMT trigger times for the four run configurations,
shown relative to the expected time distribution of photons
as calibrated from the leaky mirror data. To cover the entire PVLAS range of m , 20 hours
φ
of data are collected in each of four configurations of
plunger position andpolarization. The plunger is placed
source, the discriminator threshold is optimized to give either in the center of the magnet with L = 3.1 m and
1
99.4% efficiency for triggering on single photo-electron L = 2.9 m or near the far edge of the magnet with
2
pulseswhilealsoefficientlyrejectingtheloweramplitude L = 5.0 m and L = 1.0 m. With each plunger posi-
1 2
noise. By studying the trigger time distribution, the tion, data are taken separately with both vertically and
deadtime fraction due to possible multiple rapid PMT horizontally polarized laser light in order to test pseu-
pulses is found to be negligible (0.001%). Thus, we esti- doscalar and scalar couplings, respectively. The central
mate the total photon transport and counting efficiency plungerpositioncoversmostofthe PVLASsignalregion
to be (25±3)%. Using this threshold, and the built-in butthecorrespondingoscillationprobabilitygoestozero
cooler to cool the photocathode to 0◦C, we measure a in Eqn. 5 at m ≈ 1.4 meV where the argument of the
φ
typical dark count rate of 130 Hz. eachof the sine factors approachesπ. These zeroes indi-
To perform the coincidence counting we use two cate regions of diminished sensitivity as the scalars have
Quarknet boards [19] [20] with 1.25 ns timing precision, oscillated back into photons upon reaching the plunger
referenced to a GPS clock. The Quarknet boards de- mirror. Movingtheplungertothefarpositionsimultane-
termine the absolute time of the leading edge of time- ously changes the baseline for both the initial oscillation
over-threshold triggers from the PMT and from a mon- and the regeneration, and shifts the two regions of di-
itoring photodiode that is located inside the laser box. minished sensitivity away from m =1.4 meV. The two
φ
The clocks on the laser board and on the PMT board plungerpositions thus coverthe entire PVLASsignalre-
are synchronized using an external trigger from a signal gion.
generator. The operating conditions are continuosly monitored
The absolute timing between the laser pulses and the during each run. The reflected beam from the plunger
PMT traces is established by removing the plunger with mirrorisslightlyoffsetfromtheincidentbeam,andisdi-
the mirror, and allowing the laser to shine on the PMT rected into a calorimetric power meter by a pick-off mir-
throughseveralattenuationstagesconsistingoftwo par- ror. The number of incident photons is determined with
tiallyreflective(“leaky”)mirrors,apinhole,andmultiple 3% accuracy from these measurements. The alignment
absorptivefiltersmounteddirectlyontheapertureofthe of the laser is monitored using a fast solid state camera
PMT module. The 1019 photons per second emitted by which takes 30 Hz of images of the reflected laser spot
the laser are thus attenuated to a corresponding PMT on the pick-off mirror. The total pathlength to the pick-
trigger rate of less than 0.1 Hz for this timing calibra- off mirror is comparable to the pathlength to the PMT,
tionandtoprovideanin situtestofthedataacquisition andsotransversedeviationsseenintheimagesareclosely
system. matchedtothedeviationsatthePMT.Small∼mmscale
The regenerated photons should arrive at the same transversedeviationsareseenduringthecourseofatyp-
4
RESULTS
-1-1]V]V BFRT
GeGe Regeneration We count the PMT triggers within the 10 ns coinci-
[ [ggscalarscalar1100--55 ddeantac.e tTimhee wexinpdecotwe,ddbeaficnkegdroaupnrdioirsimbyeatshuereledakuysinmgirtrhoer
dark counts in time bins outside of the coincidence win-
PVLAS dow. No excess counts above background are observed
in any of the four configurations. The data are summa-
1100--66 rized in Tab. I. The PMT timing data, along with the
leaky-mirrorcalibration data, are shownin an expanded
time scale in Fig. 2. No excess counts above background
GammeV are observed in any time bins near the laser pulse. For
the central values of the PVLAS parameters one would
1100--77
00..44 00..55 00..66 00..77 00..88 00..99 11 22 expect ∼150 excess counts.
mmff [[mmeeVV]]
We use the Rolke-Lopez method [21] to obtain limits
FIG.3: 3σ limitcontoursforscalarparticles. Thesolidblack on the regeneration probabilities, and use Eqn. 4 to ob-
line is the combined limit using data taken at both the cen- tain the corresponding3σ upper bounds onthe coupling
tral(reddot-dashed)andtheedge(bluedashed)plungerpo- g asafunctionofmφ. Therelativesystematicuncertain-
sitions. ThePVLASrotationsignal(pink/darkgrey)andthe ties of 12% on the photon transport and detection, and
BFRT regeneration limit (tan/light grey) are also shown. 3% on the laser power measurement are incorporated in
the limits. TheGammeVlimits areshowninFigs.3and
4 along with the PVLAS 3σ signal region, the BFRT 3σ
regenerationlimits, and the 99.9%limit on pseudoscalar
-1-1]V]V BFRT couplings from BMV. As expected, the regions of insen-
GeGe Regeneration sitivityforoneplungerpositionarewell-coveredbyusing
[ [ggpseudopseudo1100--55 tshiteioontshaerrepcloumngbeirnepdosaintidona.naDlyazteadfrjooimntlbyotthopprluodnugecrepthoe-
combined limit curve. The weakly-interacting axion-like
PVLAS particle interpretation of the PVLAS data is excluded
BMV
at more than 5σ by GammeV data for both scalar and
1100--66 pseudoscalarparticles. Theasymptotic3σ upperbounds
on g for small m for each configuration are listed in
φ
Tab.I,andthecombinedanalysisgives3.1×10−7 GeV−1
GammeV
(3.5 × 10−7 GeV−1) for the scalar (pseudoscalar) cou-
plings. The GammeV exclusion region extends beyond
1100--77
00..44 00..55 00..66 00..77 00..88 00..99 11 22 the previous best limits and sets limits in regions where
mmff [[mmeeVV]]
BFRT had reduced sensitivity.
FIG. 4: 3σ limit contoursfor pseudoscalar particles. Acknowledgements: We would like to thank the staff
of the Fermilab Magnet Test Facility for their tireless
efforts, and the technical staff of the Fermilab Particle
Physics Division design group who aided in the design
ical 5 hour run, due to small changes in the orientation and construction of the apparatus. JS thanks the Brin-
oftheplungermirrorastheplungerslowlycoolsthrough son Foundation for their generous support. This work is
heat leaks in the warm bore insulation. The deviations supportedby the U.S.Department ofEnergyunder con-
are small enough that, were they due to actual changes tract No. DE-AC02-07CH11359. AC is also supported
in the laser alignment, the beam would still clear the by NSF-PHY-0401232.
aperture of the plunger. In addition, the focussing lens
at the PMT makes the final light collection system in-
sensitiveto these potentialsub-mradangulardeviations.
Nevertheless, the alignment is double-checked using the
open-ended plunger after collecting data in each config- [1] E. Zavattini et al. [PVLAS Collaboration], Phys. Rev.
Lett. 96, 110406 (2006) [arXiv:hep-ex/0507107].
uration, and no misalignment has ever exceeded our tol-
[2] G. Raffelt and L. Stodolsky, Phys. Rev. D 37, 1237
erances. The operation of the PMT is monitored using
(1988).
its dark rate. In addition, an LED flasher fires every
[3] U.Gastaldiet al.[PVLASCollaboration], Proc.XXXIII
5 minutes to verify the integrity of the light collection Int. Conf. on High Energy Physics, Moscow, Russia
system. (2006).
5
[4] R.Cameron et al.,Phys. Rev.D 47, 3707 (1993). and F. Takahashi, Phys. Rev. D 75, 013004 (2007)
[5] G. Ruoso et al.,Z. Phys. C 56, 505 (1992). [arXiv:hep-ph/0610203].
[6] E. Zavattini et al. [PVLAS Collaboration], [16] A. K. Ganguly, P. Jain, S. Mandal and S. Stokes, Phys.
arXiv:0706.3419 [hep-ex]. Rev. D 76, 025026 (2007) [arXiv:hep-ph/0611006].
[7] LIPSS, OSQAR, ALPS, BMV presentations at the [17] K. Van Bibber, N. R. Dagdeviren, S. E. Koonin,
3rd Joint ILIAS-CERN-DESY Axion-WIMPS Training A. K. Kerman and H. N. Nelson, Phys. Rev. Lett. 59,
Workshop,UniversityofPatras,Greece,19-25June2007. 759 (1987).
[8] C. Robilliard, R. Battesti, M. Fouche, J. Mauchain, [18] K. van Bibber, P. M. McIntyre, D. E. Morris and
A. M. Sautivet, F. Amiranoff and C. Rizzo, Phys. Rev. G. G. Raffelt, Phys.Rev. D 39, 2089 (1989).
Lett.99, 190403 (2007) [arXiv:0707.1296 [hep-ex]]. [19] S.Hansenet.al,IEEETrans.Nucl.Sci., vol.51,no.3,pp.
[9] R.Fardon,A.E.NelsonandN.Weiner,JCAP0410,005 926-930 (2004).
(2004) [arXiv:astro-ph/0309800]. [20] H.G.Berns,et.al,IEEETrans.Nucl.Sci.,vol.2,pp.789-
[10] D. B. Kaplan, A. E. Nelson and N. Weiner, Phys. Rev. 792 (2003).
Lett.93, 091801 (2004) [arXiv:hep-ph/0401099]. [21] W.A.Rolke,A.M.LopezandJ.Conrad,Nucl.Instrum.
[11] P. Brax, C. van de Bruck and A. C. Davis, Meth. A 551, 493 (2005) [arXiv:physics/0403059].
arXiv:hep-ph/0703243. [22] P. Pugnat et al. [OSQAR Collaboration],
[12] P. Svrcek and E. Witten, JHEP 0606, 051 (2006) arXiv:0712.3362 [hep-ex].
[arXiv:hep-th/0605206]. [23] OSQARhasalsorecentlyreportedresults[22]whichmay
[13] D.Maity,S.RoyandS.SenGupta,arXiv:0709.3940[hep- not be valid due to a matching of group velocities but
ph]. not phase velocities of photon and scalar waves in their
[14] S. Andriamonje et al. [CAST Collaboration], JCAP residual gas technique.
0704, 010 (2007) [arXiv:hep-ex/0702006].
[15] J. Jaeckel, E. Masso, J. Redondo, A. Ringwald