Precise Measurement of the Absolute Yield of Fluorescence Photons in Atmospheric Gases

0.75 MB·English

by  AIRFLY Collaboration

1
1
Precise Measurement of the Absolute Yield of Fluorescence Photons in
0
2 Atmospheric Gases

n
a AIRFLY Collaboration: M. Avea, M. Boha´ˇcov´abc, K. Daumillera, P. Di Carlod, C. Di Giulioe,
J P. Facal San Luisb∗, D. Gonzalesf, C. Hojvatg, J. R. Ho¨randelh, M. Hrabovsky´i, M. Iarlorid,

9 B. Keilhauera, H. Klagesa, M. Kleifgesj, F. Kuehng, M. Monasorb, L. Noˇzkac, M. Palatkac, S. Petrerad,
1 P. Priviterab, J. Ridkyc, V. Rizid, B. Rouill´e d’Orfeuilb, F. Salamidaa, P. Schov´anekc, R. Sˇmidaa,

] H. Spinkak, A. Ulrichl, V. Verzie, C. Williamsb
M
aKarlsruhe Institute of Technology, IK, Postfach 6980, D - 76021 Karlsruhe, Germany
I
.
h bUniversity of Chicago, Enrico Fermi Institute & Kavli Institute for Cosmological Physics,
p 5640 S. Ellis Ave., Chicago, IL 60637,USA
-
o
cInstitute of Physics of the Academy of Sciences of the Czech Republic,
r
t Na Slovance 2, CZ-182 21 Praha 8, Czech Republic
s
a
[ dDipartimento diFisicadell’Universit`ade l’AquilaandINFN, ViaVetoio,I-67010Coppito,Aquila,Italy

1 eDipartimento di Fisica dell’Universit`a di Roma Tor Vergata and Sezione INFN,
v Via della Ricerca Scientifica, I-00133 Roma, Italy
9
9 fKarlsruhe Institute of Technology, IEKP, Postfach 3640, D - 76021 Karlsruhe, Germany
7
3 gFermi National Accelerator Laboratory, Batavia,IL 60510,USA
.
1
hIMAPP, Radboud University Nijmegen, 6500 GL Nijmegen, The Netherlands
0
1
iPalacky University, RCATM, Olomuc, Czech Republic
1
:
v jKarlsruhe Institute of Technology, IPE, Postfach 3640, D - 76021 Karlsruhe, Germany
i
X kArgonne National Laboratory,Argonne, IL 60439,USA
r
a lPhysik Department E12, Technische Universit¨at Muenchen,
James Franck Str. 1, D-85748 Garching, Germany
We have performed a measurement of the absolute yield of fluorescence photons at the Fermilab Test Beam.
A systematic uncertainty at 5% level was achieved by the use of Cherenkov radiation as a reference calibration
light source. A cross-check was performed by an independent calibration using a laser light source. A significant
improvement on theenergy scale uncertainty of Ultra-High Energy Cosmic Raysis expected.
1. Introduction an integral part of the Pierre Auger [2] and the
Telescope Array [3] experiments. Excitation of
Fluorescence detection of Ultra High Energy
the atmospheric nitrogen by the charged parti-
Cosmic Rays (UHECRs) is a well established
cles in the extensive air shower induces the emis-
technique, pioneered by Fly’s Eye [1], and today
sion of fluorescence photons, mostly in the 300-
400 nm range. A Fluorescence Detector (FD)
∗Correspondingauthor: [email protected]
1
2
records this radiation to infer the cosmic ray en-
ergy and the particle type. For this purpose, the
fluorescence light yield from the charged parti-
cles in the showermust be knownfor every emis-
sion point along the shower axis. A correction
is then appliedto accountfor atmosphericeffects
between the shower and the telescope, enabling
an accurate, quasi-calorimetric, primary energy
determination.
The uncertainty on the fluorescence light yield
isoneofthemainsystematicuncertaintiesonthe
cosmic ray energy determination by experiments
that employthe fluorescencetechnique (e.g. 14%
overatotal22%uncertaintyforthePierreAuger
experiment). The AIRFLY collaboration has al-
ready performed a very precise measurement of
the fluorescence spectrum and its pressure de-
pendence [4], as well as the dependence of the
emission on the temperature and humidity [5].
AIRFLY measurements over electron kinetic en-
ergies rangingfrom keVto GeV using severalac-
Figure 1. Layout of the experimental apparatus
celeratorshavealsoproventhe proportionalityof
used at the Fermilab Meson Test Beam
thefluorescenceyieldwiththeelectronenergyde-
posit [6].
The final step in the precise characterization
of the nitrogen fluorescence light emission is the
measurement the absolute value of the yield for
the main emission line at 337 nm. The AIRFLY
strategyto reducethe systematic uncertaintiesis
to calibrate the experimental apparatus in situ,
using photons emitted by a well know process:
Cherenkov radiation [7]. A second calibration
method, with nearly independent systematic un-
certainty, is based an absolutely calibrated laser
light source. In this paper, we present prelimi-
nary results from a series of dedicated measure-
ments atthe FermiNationalAcceleratorLabora-
tory (Fermilab).
2. Experimental method
The measurements were performed at the Fer-
milab Test Beam Facility. Most of the mea-
surements were carried out using the 120 GeV
Figure 2. AIRFLY chamber with the integrating
proton beam, extracted from the Main Injector.
sphere fitted inside.
Secondary beams of 32 GeV pions and 8 GeV
positrons were also used.
A sketch of the apparatus is shown in Fig.
3
gle(andthusmaximizesthe signaltonoiseratio)
and it works as equalizer of the detection effi-
ciency of the set-up for the isotropic fluorescence
light and for the highly directional Cherenkov
light. The integrating sphere (Fig. 3) was built
from two hollow aluminum half spheres coated
with a materialof very high diffusive reflectance.
Lightproducedinside thesphereiscollectedover
4πsolidangleandatthesametimeisotropizedby
several diffusion bounces, so that the lambertian
light output is independent of the original light
distribution. A total of 4 ports were machined
in the sphere, one was the detection port, two
wereforthe beamentranceandexit,andthelast
one was opened at the top of the sphere. Two
remotely controlled shutters could close the top
Figure3. Theintegratingsphereusedinthemea- andexitportsoftheintegratingspherewithplugs
surements. The diffusive coating applied to the coated with the same diffusive material material
interior can be seen through two of the ports. usinginsidethesphere. Withtheexitportclosed
theCherenkovisdiffusedbackintothesphereand
canreachthephotondetector. Withtheexitport
open the Cherenkov is absorbed by the chamber
1. A fluorescence chamber made of a 3 mm lining and thus only fluorescence can reach the
thickstainlesssteelwasplacedinthebeampath, photon detector. The top port compensates the
withthecorrespondingflangesforwindows,shut- open/closedposition of the exit port to maintain
ters, gauges, gas inlet and pump-out. Both the alwaysthreeopenportsandthustheopticalchar-
entrance and exit windows are 0.1 mm thick acteristics of the sphere. i
aluminum, with the entrance window before a Thewholechamberwasairtightandaremotely
18 cm long tube to provide additional length controlled system for gas and vacuum handling
for Cherenkov light production. An integrating was used. Pure nitrogen and a dry-air mixture
sphere was used to collect light produced inside were used for the measurements and additionally
the chamber (Fig. 2). One of the ports of the heliumandargonwereusedforbackgroundruns.
sphere was open to a a gas-tight window fitted The pressure, temperature and humidity inside
witha337nmfilterandthencoupledtothepho- the chamber were monitored remotely using the
ton detector. A Hamamatsu H7195P photomul- appropriate sensors.
tiplier (PMT) tube, with good single photoelec- A set of particle counters was used for beam
tron resolution, was used for photon detection. monitoring. At the entrance of the chamber a 10
The opticalfield of view was defined by a 40 mm mm diameter finger counter was used for beam
diameter acceptance cylinder placed between the tagging. At the exit the beam was tagged with
integrating sphere’s port and the filter, and by aCherenkovcounter,a10mmdiametercylindri-
circularaperturesofthesamesizeplacedinfront cal rodmade of UV-transparentacrylic material.
ofthe PMT photocathode. Amechanicalshutter The rod was 30 mm long and allowed very good
remotely controlled allowed to take background singleparticleresolutionwithfasttiming. Before
measurements. The chamber was internally cov- and after the chamber two big scintillator pads
eredwithablackUV-absorbingmaterialtoavoid with a central 10 mm diameter hole for beam
stray light. passage provided a veto for off-axis particles.
The purpose of the integrating sphere is The chamber and the counters were mounted on
twofold: it increases the light collection solid an- an optical breadboard for precise mounting and
4
alignment, and placed on a remotely movable ta- a 4 s beam spill, particles were grouped in trains
ble that allowed the alignment of the appara- of bunches. Trains were separated in time by 10
tus with the beam by maximizing the finger and µs,andbuncheswithinatrainwereseparatedby
Cherenkov counters rates. The beam profile was 19 ns. Typical conditions for data taking with
monitored by wire chambers placed before and the proton beam were 30 bunches per train, and
after the AIRFLY apparatus and was typically 3 a multiplicity of 2·105 particles per spill. The
mm x 4 mm wide. trigger logic was built from the coincidence of a
traintriggergate,issuedincorrespondenceofthe
arrivalof eachtrainof bunches,and a single par-
ticletriggergate,comingfromabeammonitoring
scintillator counter. Both triggers were provided
by the Test Beam Facility. Whenever a trigger
was issued the signals from the scintillator coun-
ters and photomultipliers were digitized by a 12-
bit 500 MHz FADC and 600 samples (equivalent
to 1.2 µs, containing the entire train of bunches)
were saved in the FADC memory. The data for
the whole spill was stored in the FADC internal
memory and then readout and saved to disk in
about 40 s between spills. As an example, the
ADC trace of the beam counters for one trigger
is shown in Fig. 4.
Data during the test beam was taken in a dif-
ferentconfigurations. Foreachconfigurationruns
of up to half hour (i.e. 30 spills) of data were
acquired. Runs in the same conditions were re-
peated periodically to improve statistics and to
assure redundancy and consistency.
3. Data analysis and results
Offline data analysis for one run proceeds se-
lecting single particles that cross the fiducial vol-
umeofthecamera,hencearetaggedbythebeam
Figure 4. The signal of a train of bunches for countersplacedattheentranceandtheexitofthe
the Cherenkov rod (top plot) and the two veto chamber (Fig. 5). For these selected particles we
counters (middle and bottom plots). Two par- requirethatnosignalispresentinanyoftheveto
ticles in two different bunches can be observed: counters. Additional analysis cuts are placed to
the first one passes through the rod while the discard trains that have unusually large number
second one passes through both veto counters of particles passing throughthe veto counters,as
(and leaves also a small signal in the rod when this has been shown to improve the background
it hits the PMT glass producing a small amount conditions without excessively penalizing statis-
of Cherenkov in it). tics. Once the clean single protons have been se-
lected the PMT signal is analyzed and the pho-
tons in coincidence with the selected protons are
counted (Fig 6). The signal S, in units of pho-
The trigger and DAQ were designed consider- toelectronsper beamparticle,pbp, is then calcu-
ingthecharacteristicsofthebeamtiming: within lated.
5
Figure 6. PMT spectrum for a run. The events
inside the single photoelectron peak are counted
Figure 5. Spectrum of the Cherenkov rod. to define the signal.
From left to right the three peaks correspond to
pedestal, particles hitting the PMT window and
singleparticlespassingthroughtheCherenkovra-
diator. The fourth rightmost smaller peak corre-
was measured by AIRFLY as a function of pres-
sponds to two particles in the same bunch.
sure in [4] and was cross-checkedin our Fermilab
experimental apparatus using a 241Am radioac-
tive source, giving 7.45±0.10 at 1000 hPa.
We measure∆S =(16.83±0.13)·10−4 pbp.
FL
Themeasuredsignalintheafluorescencemea-
FromEq.3,wederivethebackground-subtracted
surement taken in a given gas, Sgas(meas), is
FL fluorescence signal:
given by:
Sgas(meas)=Sgas+Bgas, (1) SFNL2 =(19.44±0.15)·10−4 pbp, (4)
FL FL FL
where Sgas is the signal from the 337 nm band, with the background (Eq. 1) estimated as a 3%
FL
andBgas isattributabletobackground. Thebest of the signal.
FL
determination of the overall background can be In the Cherenkov calibration measurement,
obtained combining measurements from different both Cherenkov and fluorescence emission in the
gases. FromEq. 1,thedifferenceofthemeasured gas contribute to the measured signal:
signal in N2 and air is given by: Sgas(meas)=Sgas+Sgas +Bgas+B , (5)
CH FL CH FL CH
∆S =SN2 −Sair +BN2 −Bair. (2)
FL FL FL FL FL where Sgas is the signal from Cherenkov light in
CH
Since the beam related background and the pri- the 337nm band emitted in the gas under study,
mary interactions are the same and secondary and B takes into account background origi-
CH
particleproductionisverysimilarinnitrogenand nating from the interactionof the beam particles
air backgrounds cancel in Eq. 2. Thus, in the exit port plug (that is independent of the
1 gas filling the chamber). BCH can be estimated
∆S =SN2 1− , (3)
FL FL(cid:18) r (cid:19) directly from the difference in the vacuum mea-
N2
surement in fluorescence and Cherenkov modes
where r is the ratio of the 337 nm fluorescence
N2
in pure nitrogen to the signal in air. This ratio B =(2.63±0.23)·10−4 pbp, (6)
CH
6
a value that is ∼ 10% of the Cherenkov signal. the 5% uncertainty in the absolute calibration of
We made several cross-check of this background, the laser probe.
for example changing the exit port plug material
to a thin Mylar foil where no light production is 4. Outlook
expected.
We measure SN2 (meas) = (32.89 ± 0.15) · The preliminary results on the absolute flu-
10−4 pbp. Using ECqH. 5 we obtain orescence yield reported in this work are com-
patible with the current values used in Fluores-
SN2 =(10.27±0.23)·10−4 pbp, (7) cence Detector analysis (see Ref. [8] for a review
CH
of the different measurements.) We expect a fi-
and from it and Eq. 4 we derive
nal measurement with a systematic uncertainty
RN2 = SFNL2 =1.893±0.045, (8) below 5%, a significant improvement over pre-
SN2 vious measurements, which will correspondingly
CH
improve the uncertainty on the energy scale of
the ratio of fluorescence to Cherenkov 337 nm
UHECR measurements.
photons produced inside the chamber.
In order to derive the absolute 337 nm yield
REFERENCES
in air, we performed a full Monte Carlo simula-
tionofthesetup,whereallindividualcomponents 1. R.M.Baltrusaitisetal.,Nucl.Instrum.Meth.
aresimulatedaccordingtomeasurementsdonein A 240 (1985) 410.
thelaboratory. Theabsolutefluorescenceyieldin 2. J. A. Abraham et al. [The Pierre Auger
the simulation, Yair , which determines a corre- Collaboration], Nucl. Instrum. Meth. A 620
MC
spondingexpectedFluorescence/Cherenkovratio (2010) 227 [arXiv:0907.4282[astro-ph.IM]].
Rair ,canbescaledtomatchthemeasuredvalue 3. Y.Tamedaetal.,Nucl.Instrum.Meth.A609
MC
of Eq. 8 and obtain our measurement of the ab- (2009) 227.
solute fluorescence yield in air: 4. M. Ave et al. [AIRFLY Collabora-
Yair =5.60±0.13 γ /MeV, (9) tion], Astropart. Phys. 28 (2007) 41
337nm
[arXiv:astro-ph/0703132].
where the uncertainty is statical only. Our pre- 5. M. Ave et al. [AIRFLY Collaboration],
liminary estimate for the systematic uncertainty Nucl. Instrum. Meth. A 597 (2008) 50
ofthemeasurementisbetterthan5%,dominated [arXiv:0711.4583[astro-ph]].
by the uncertainty inthe wavelengthdependence 6. M. Ave et al. [AIRFLY Collaboration], Nucl.
ofthe PMTquantumefficiency andthe 337filter Instrum. Meth. A 597 (2008) 46.
transmission,accountingforthedifferentspectral 7. M. Bohacova et al. [AIRFLY Collabora-
distributions of the fluorescence and Cherenkov tion], Nucl. Instrum. Meth. A 597 (2008) 55
signals. The detailed analysis of the uncertain- [arXiv:0812.3649[astro-ph]].
ties is still, however,ongoing. 8. J. Rosado, F. Blanco and F. Arqueros, As-
In the laser calibration method, we use a tropart.Phys.34(2010)164[arXiv:1004.3971
337 nm nitrogen laser and an NIST calibrated [astro-ph.IM]].
probe to determine the overall efficiency of the
set-up,thenumberofdetectedphotoelectronsper
photon entering the sphere. Using this efficiency,
the number in Eq. 4 and a detailed Monte Carlo
simulation of the laser calibration the value
Yair =5.56±0.07 γ /MeV, (10)
LASER 337nm
is obtained, which has an uncertainty nearly in-
dependentoftheonefromtheCherenkovcalibra-
tion. Thesystematicuncertaintyisdominatedby