1. Cosmic-ray iron and electron detection with H.E.S.S.
Rolf Bühler • ACKS Seminar • 28 of January, Stanford
Astrophysics Colloquium

2. Outline Introduction to cosmic rays The H.E.S.S. telescopes
Measuring the iron spectrum
Measuring the electron spectrum
Summary and Outlook

3. Cosmic Ray Discovery Discovered (beyond doubt) by Victor Hess
“The result of these observations seems best explained by a radiation of great penetrating power entering our atmosphere from above..”
Phys. Zeitschriften 1912
High energy particles reaching Earth at a rate of ≈1000 s-1m-2

4. Energy Spectrum Remarkably featureless energy spectrum
Well described by power-law with softening at ≈4 PeV (the “knee”)
Confined to the galaxy below the knee
Total energy density ≈1 eV cm-3
γ ≈ 2.7
“knee”
~4 PeV
Nuclei (98%)
Electrons (2%)
γ ≈ 3.0

5. Composition Similar to solar but:
Enhancement below C-N-O and Fe → Spallation, traversed ≈40 g cm-2 at 1 GeV
C-N-O Si Fe
Engelmann et al. 1990
Radioactive “clocks” → confinement of ≈10 Myrs at 1 GeV
Yanasak et al. 2001
Meyer et al. 1997
Less H and He → Less high ionization energy or high volatility elements
Normalized to Silicon
At 1 TeV

6. Composition Index independent of element
→ Hints at common origin
Spallation elements have softer spectrum
→ Energy dependent escape from galaxy
Swordy et al. 1990
Tracer & CRN Ave et al. 2008
CREAM II Ahn et al. 2009
Compilation Wieble Sooth 1998

7. Where do they come from?
Isotropic flux, deflected by magnetic fields, no directional information left
Options:
Measure spectrum and composition and model source/propagation
Use neutral tracers (photons, neutrinos)
Everything points to Super Novae Remnants (below the knee) →

8. Why Super Novae Remnants?
1) Photon observations:
Non-thermal spectrum, consistent with origin from pion decays at high energies
Aharonian et al. 2004, Abdo et al. 2010,
Ellison et al. 2010
RXJ 1713 above 200 GeV
2) Cosmic-ray spectrum:
Power law of index 2 result from Fermi I acceleration. Index of 2.7 from propagation effects
Knee could correspond to maximum particle energy (gradually light to heavy nuclei break away)
Bell 1978
Hoerandel 2004

9. Why Super Novae Remnants?
𝑃≈ 𝑉ϱ τ ≈ 𝑒𝑟𝑔 𝑠 −1
Assume local cosmic ray
density in galaxy
≈ 107 years
(from spallation and
radioactive isotopes)
𝑃 𝑠𝑢𝑝𝑒𝑟𝑛𝑜𝑣𝑎𝑒 ≈ 𝑒𝑟𝑔 30𝑦𝑒𝑎𝑟𝑠 ≈ 𝑒𝑟𝑔 𝑠 −1
Supernovae rate from
similar galaxies
3) Energetics:
Helder et al. 2009
They do efficiently release energy into CR
The sources of cosmic-ray electrons:
Are not constrained by (1), could also be pulsars, which also fulfill arguments (2), (3)
Should be local ( ≈1kpc) for ≈1 TeV electrons due to fast energy loss
Kobayashi et al. 2004

10. H.E.S.S. Telescopes Located in Namibia (1800 a.s.l.)
Sensitive between ~0.1 to 100 TeV
Field of View of 5º diameter

11. Gamma-ray detection Image shower Cherenkov light
High cosmic-ray background Rejection of ~99%, hadron showers are wider
Remaining background from regions off the source
≈ 30 km
EAS-light
Not possible for diffuse signal

12. Shower reconstruction
Shower-light
≈ 30 km
γ-ray
Resolution:
Direction 0.1°
Core position 20m
Energy 15%
≈ 2º
Shower direction
Energy from total
intensity and core distance
≈ 100 m

13. Iron detection Detection of Cherenkov Light before first interaction Z
DC-light
Shower-light
≈ 2º
DC-light
Shower direction
Shower-light
≈ 100 m

15. DC-Light detection DC-light ~ Z2 Shower intensity ~ EFe
Shower outshines
DC-light ~ Z2
Shower intensity ~ E
Iron detection >13 TeV (high Z and flux)
Cherenkov
threshold
Kieda et al. 1999

16. Dataset & Charge Reconstruction
Effective exposure of ≈107 m2 sr s
→ In total 1899 events with DC-light in 2 telescopes (background-free)
Charge reconstruction over DC-light intensity.
Fit iron fraction in five energy bins
1.5 < lg( E/TeV ) <1.7
𝑍=𝑘 θ,𝐸 𝐼 𝐷𝐶

17. Iron Spectrum Good agreement with other experiments
Hadronic model ≈20% on normalizarion (smaller than at higher energies)
Power-law Index γQGSJET= γSIBYLL=
Among most precise
Proof of principle
Aharonian et al. 2007

18. Electron Detection Electrons (positrons) induce narrow EM-showers
Analysis done by Kathrin Eggberts
Electrons (positrons) induce narrow EM-showers
No off-source region → background from simulations (SIBYLL 2.1 and QGSJET II)
“Electron likeness”  from random forest resulting in 10-4 hadron rejection in  > 0.9
Large effective exposure of ≈2·107 m sr s 
Simulated background
Data
Electron excess

19. Electron Detection
Fit electron contribution in energy bands in >0.6 region (contribution of heavier elements negligible)

20. Gamma-ray Background?
Only extra-galactic sky off sources considered, still similar showers, so diffuse gammas?
Gammas interact 7/9 rad. length lower. Fit of Xmax distribution → gamma-rays less than 50%
Low level of gamma-ray background expected due to pair creation on photon background

21. Electron Spectrum Spectral softening at ≈1 TeV ( γ ≈ 3→4.1 )
Extends up to 4 TeV → source within ≈1 kpc
ATIC peak disfavoured (yet not excluded)
Fermi & HESS spectrum can be modelled including Klein- Nishina effect and source cutoff → No “exotic physics” required.
Aharonian et al. 2008, Acero et al. 2009
Stawarz et al. 2009, Schlickeiser et al. 2009

22. Atmosphere Uncertainties
Error on energy scale of 15% from:
Uncertainty of atmospheric density profile (showers could be closer/nearer, ≈3 g cm2 at Xmax)
Uncertainty in dust and ozon absorbtion
No temporal variations considered
Optical efficiency of detector and opacity low atmosphere known though muons.

23. Hadronic-model Uncertainties
SIBYLL and QGSJET results in ≈20% difference in flux normalization and ≈0.2 in index, comes from:
Electrons
How often does a proton look like an electron?
Iron
At which depth does the nuclei interact?
Which particles are created? p π0 γ Fe N

24. Conclusions Iron measurement
One of the most precise between TeV
Agreement with independent technique
Proof of Principle for DC-light detection
Electron measurement
Extension of spectral measurements to 4 TeV
Spectral cutoff around 1 TeV
ATIC-peak disfavored
Proof of principle of ground based detection

25. Outlook
AGIS / CTA increase in exposure by ~30 with respect to H.E.S.S.
→ Iron spectrum to ~PeV
→ Electron spectrum ~15 TeV
Lower energy threshold of ~100 GeV for electrons. Maybe already with H.E.S.S. II or MAGIC II
CTA / AGIS (~2014)
MAGIC II (2009)
H.E.S.S. II (~2011)

26. Outlook Improvement of systematics Hadronic Models Atmospheric →
Will be highly constrained by LHC experiments testing forward direction reactions (LHCf, TOTEM)
Will reach lab energies of few PeV (Already sufficient: ~10 TeV p on N → ECM~50 GeV)
Atmospheric
Future instruments will have atmospheric monitoring
Dova et al. 2007 →
Great prospects for cosmic-rays measurements

27. Backup slides..

28. Simulated flux assumes composition of
Dataset & Background
Simulated flux assumes composition of
Hoerandel et al. 2003
Effective exposure of ≈107 m2 sr s
→ In total 1899 events with DC-light in 2 telescopes (background-free)
DC and shower light yield prediction depends on hadronic model
→ Estimate error from using QGSJET 01 / SIBYLL 2.1

29. Charge reconstruction
𝑍=𝑘 θ,𝐸 𝐼 𝐷𝐶 *
1.3 < lg( E / TeV ) < 1.5
DC-intensity depends on:
- first interaction height
- energy (const > Ethreshold) →
Allows measurement of the iron fraction in the data.
Reconstructed charge:
(k(E,θ) normalizes Z* to iron)