A new measurementof the Cosmic-Ray

Electron andPositron Spectrum

with the LargeArea Telescope

Alberto ManfredaUniversita and INFN–Pisa

albertom@pi.infn.it

on behalf of the Fermi-LATcollaboration

Pisa, October 17, 2016

More than an update

Previous Fermi measurement of the spectrum from 7 GeV to 1 TeVpublished in 2010 [Ackermann et al., Phys. Rev. D 82, 092004]

The analysis is new in many respectsI Almost seven times the amount of data availableI Completely revised event reconstruction provided by the Fermi

collaboration (Pass 8)I Upper energy range extended to 2 TeVI Improved selection algorithmsI Better understanding of the environment in which the LAT

operatesI Better modelling of the systematic uncertaintiesI New detailed study of the geomagnetic field effects in the GeV

energy rangeAlberto Manfreda (UNIPI) Pisa, October 17, 2016 2 / 14

Fermi LAT

I Launched by NASA on June 2008 at 565 km altitude, 25.6◦ inclination

I Large field of view (20% of the sky), full sky observed every ∼ 3 h

I Designed to detect EM showers, naturally including e+ and e−

Tracker/Converter (TKR)I 18 x-y planes of silicon

strip detectorsI ∼ 10 k sensors,

∼ 80 m2 of siliconactive area, 1M readoutchannels

Anti-Coincidence Detector (ACD)I Segmented (89 tiles) as to

minimize self-veto at high energyI 0.9997 average MIP detection

efficiency

Calorimeter (CAL)I 1536 CsI(Tl) crystal (96 per tower); 8.6

radiation lengths on-axisI Hodoscopic, 3D shower profile

reconstruction for leakage correction

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 3 / 14

Datasets

Two slightly different analyses, for two different datasets:I High Energy (HE): 42 GeV - 2 TeV using events from the

standard on-board filterAccept all events with more than 20 GeV released in the CAL(fully efficient for electrons above ∼ 40 GeV)

I Low Energy (LE): 7 GeV - 70 GeV using events from thediagnostic (DGN) filterUnbiased sample of all triggers, pre-scaled by a factor 250

Overall live time is 4.68 years - outside SAA and in standard opera-tion modesField of view reduced to ≤ 60◦ from the instrument zenith (avoidEarth limb photons contamination).

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 4 / 14

CRE selection

Energy [GeV]210 310

sr]

2Ac

cept

ance

[m

0

1

2

3

4

5

PRELIMINARY

Trigger & FilterTrack found & energy

Minimal PSF quality

in the CAL0> 8 X

Alpha removal

I Event-quality cuts (good direction and energy reconstruction)I Removal of particles with Z > 1, tagged by ionization in ACD

and TKRI Boosted Decision Trees (BDTs) selection, for proton rejectionI LE: additional orbital selection for geomagnetic field effects

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 5 / 14

Proton removal selection

I Exploits differences betweenleptonic and hadronic eventsin the detector:

I Shower development inthe CAL

I Number of δ-raysproduced in the TKR

I A few variables have beenindividually rescaled, toimprove their data/MCagreement

Shower transverse size10 20 30 40 50 60 70 80 90 100

Rat

e [H

z]0

0.001

0.002

0.003

0.004

= 0.44--1.00]θ[112.2--144.5 GeV, cos

±MC r/w eMC r/w pMC sum

PRELIMINARY

Example:

I The BDT combines all the information into a single estimatorof ’leptonicity’ (PCRE ).

I 8 BDTs used for HE analysis, optimized for different energyranges (plus 1 for LE)

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 6 / 14

BDT output

CREP-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0

Cou

nts

0

100

200

300

400

500

600

700

800

900

1000

Data

MC total

MC electron

MC protonPRELIMINARY

[866 − 1000] GeV

I Cut on PCRE optimized in each energy binI Normalization of Monte Carlo templates fitted to data for residual

background estimation

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 7 / 14

Performance

I Selection increasingly difficult at higher energyI Contamination below 20% in the whole energy range

E [GeV]10 210 310

sr]

2A

ccep

tanc

e [m

0

0.5

1

1.5

2

2.5

3

Res

idua

l con

tam

inat

ion

0

0.1

0.2

0.3

0.4

0.5

0.6LE acceptance (x 250)

HE acceptance

LE contamination

HE contamination

PRELIMINARY

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 8 / 14

LE Orbital Selection

McIlwain L and rigidity cut-off values across the LAT orbit

McIlwain L1 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Live

time

frac

tion

0

0.2

0.4

0.6

0.8

1

1.2

7.0 GeV7.8 GeV8.7 GeV9.7 GeV10.8 GeV12.0 GeV

13.3 GeV

14.8 GeV16.4 GeV

PRELIMINARY

I GeV range: shielding effect of theEarth magnetic field

I LAT orbit: 0.98 - 1.73McIlwain L, vertical rigiditycut-off from ∼ 6 GeV to∼ 14 GeV.

I Energy dependent selection: ineach bin accept only orbitalregions where the whole bin isabove the cut-off

I Optimized directly on data, byfitting the count spectrum inseveral McIlwain L bins

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 9 / 14

Geomagnetic correction

I Cut-off is smooth: residualparticle loss even after theselection

I Recovered with a tracingtechnique: simulated realistic fluxof e+ + e− in the LAT, tracedback in a model of the Earthmagnetic field

I Estimated correction factors(up to 40%) applied to theobserved flux

I Separate corrections for first yearof data (due to rocking angleincrease from 35 to 50 degree in2009)

Top view

10 GV/c e-

Top view

10 GV/c e+

Top view

25 GV/c e-

Top view

25 GV/c e+

longitude°0E

W

longitude°90WE

longitude°180W

E

longitude°270EW

Top view

(vertical cutoff)10.99 GV/c e-

longitude°0E

W

longitude°90WE

longitude°180W

E

longitude°270EW

Top view

(vertical cutoff)11.66 GV/c e+

longitude°0E

W

longitude°90WE

longitude°180W

E

longitude°270EW

forbidden

forbidden

allowed

allowed

E [GeV]8 9 10 20 30 40 50 60 70

Cor

rect

ion

fact

or

1

1.05

1.1

1.15

1.2

1.25

1.3

1.35

1.4nominal cut-offnominal cut-off + 30%

year, nominal cut-offst1 year, nominal cut-off + 30%st1

PRELIMINARY

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 10 / 14

Systematics uncertaintiesI Acceptance uncertainty:

I Studied with a scan on the selection cut

I Residual background uncertainty:I Given by MC simulation uncertainties in reproducing the most

electron-like hadronic showers.

I Energy scale bias uncertainty:I At 10 GeV given by comparison of observed CRE cut-off with

tracer predictionI At 1 TeV study of data/MC disagreement of key energy

reconstruction variables

I Individual variable calibration uncertainty (HE analysis):I Estimated by bracketing with different sets of corrections

I Geomagnetic corrections uncertainty (LE analysis):I Studied by varying the orbital selection and recomputing the

corrections

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 11 / 14

Results: LE spectrum

E [GeV]7 8 9 10 20 30 40 50 60 70

]2 G

eV-1

sr

-1 s

-2 In

tens

ity [m

× 3E

60

80

100

120

140

160

180

200

220

240

260

Fermi HE (2010)

Fermi LE (2010)

AMS-02 (2014)

Fermi Pass 8 - LE selection

PRELIMINARY

I Significant difference wrt the previous Fermi publicationI Geomagnetic correction (new)I ’Ghost’ signals effect not included in previous simulation

(corrected in Pass 8)

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 12 / 14

Results: HE spectrum

I Shaded band is sum in quadrature of systematic uncertaintiesI Energy scale uncertainty not includedI Spectrum is harder than that reported by AMS-02I Our systematic errors do not allow to confirm or exclude the cut-off

observed by H.E.S.S.Alberto Manfreda (UNIPI) Pisa, October 17, 2016 13 / 14

Conclusions

I New measurement of the CRE spectrum, pushing the FermiLAT to its limits in all metrics

I New event-level analysis (Pass 8)

I Energy range extended to 2 TeV (first direct measurementabove 1 TeV)

I Better handling of the geomagnetic effects in the evaluationof the low-energy spectrum

I Better characterization of the systematic uncertainties

I Paper submitted to the journal (stay tuned)

I CRE selection will be used for anisotropies study update,currently under publication (again, stay tuned!)

Alberto Manfreda (UNIPI) Pisa, October 17, 2016 14 / 14