Ein (hoffentlich) allgemein-verständlicher Vortrag zu kosmischen Beschleunigern und Gammastrahlungs-Quellen in unserer Galaxie, der Milchstrasse. Vortrag gehalten im Haus der Astronomie, Heidelberg vom 10. März 2016. Videos sind leider nicht sichtbar im PDF.
An introduction to Galactic gamma-ray astronomy for the general public.
2. Übersicht
1. Was ist Gamma-Strahlung?
2. Wie entsteht kosmische Gamma-Strahlung?
3. Wie misst man kosmische Gamma-Strahlung?
Teleskope für Gamma-Strahlung. Fokus auf Fermi-LAT und H.E.S.S.
4. Was haben wir entdeckt?
Galaktische Quellen kosmischer Gamma-Strahlung
5. Wie geht’s weiter?
Die Zukunft — Cherenkov Telescope Array (CTA)
5. Sichtbares Spektrum
ist ca. eine Oktave
(Faktor 2 in Frequenz)
kurze Wellenlänge
hohe Frequenz
hohe Energie
lange Wellenlänge
tiefe Frequenz
niedrige Energie
Sichtbares Licht
Licht aus dem Kosmos: ein 15 meter langes Klavier!
8. Licht ist eine elektromagnetische Welle.
Licht ist auch ein Strom aus Licht-Teilchen = Photonen.
Gamma-Strahlung besteht aus den
höchst-energetischen Photonen.
Energie
9. Photon-Energie in Einheit “Elektron-Volt” = eV
Optisch
1 eV
Röntgen
1 keV = 1000 eV
Gamma
1 MeV — 1 GeV — 1 TeV
• 1 GeV = 10^9 eV = 1,000,000,000 eV
• 1 TeV = 10^12 eV = 1,000,000,000,000 eV
16. Teilchen-Spektrum
Fluss
Radio Optisch Röntgen EnergieGamma
T ~ 10^3 K
E ~ 1 eV
T ~ 10^6 K
E ~ 1 keV
Es gibt keine Objekte
im die viel heißer sind
als ein paar Millionen
Grad!
Gamma-Strahlung
ist nicht-thermisch!
40. Fermi-LAT
• Fermi-LAT ist super im Energie-
Bereich
100 MeV — 1 TeV
• Bei Energien > 1 TeV ist es zu
klein um Photonen zu sammeln.
• Vom hellste TeV-Gamma-Quelle,
der Krebs-Nebel, hat es seit 2008
nur ein paar Photonen detektiert.
• Für > TeV Astronomie brauchen
wir eine andere Methode!
1 m x 1 m
59. ~150 PulsarePulsare sind schnell rotierende Neutronen-Sterne,
mit extrem hohen magnetischen und elektrischen
Feldern in ihrer “Magnetosphäre”.
Vela-Pulsar
Die hellste Quelle
am Fermi-Himmel.
69. Stars: Galactic TeV sources outside HGPS region
Triangles: Galactic GeV sources (1FHL)
Image: Planck CO map
250
o
65
0
o
R. Hurt
NASA
H.E.S.S. Survey
70. H.E.S.S. Survey
Das erste TeV-Bild der Milchstrasse!
Das Ergebnis von 6000
Einzel-Beobachtungen von 2004 — 2014
73. Pulsar-wind-nebel
Unbekannt
Binär-System
t = 10,000 yrs 20,000 yrs
log (
1
tivistic gas in three two-dimensional simulations with
varying resolutions. We also plot the e†ective value of
for these three simulations in Figure 6. Up until aR
p
/R
1
time of D35,000 yr, all three simulations are quite similar.
Once the Rayleigh-Taylor instability kicks in, however, the
resolution-dependent e†ects of mixing become quite
evident. Independent of the numerical resolution, it is clear
that the mixing of relativistic gas and thermal gas is very
efficient.
4.2. Evolution in Nonuniform Media
We have repeated our two-dimensional simulation with
the addition of a density gradient in the ambient medium
with the goal of understanding the displacement of the
pulsar bubble seen in the Vela SNR. Following Dohm-
FIG. 8.ÈEvolution of the pulsar nebula/supernova remnant for the mod
electronic edition of the Journal for a color version of this Ðgure.]
t=10,000yrs20,000yrs30,000yrs56,000yrs
log(ρ/ρc)
0 1
threetwo-dimensionalsimulationswith
ions.Wealsoplotthee†ectivevalueof
threesimulationsinFigure6.Upuntila
0yr,allthreesimulationsarequitesimilar.
igh-Taylorinstabilitykicksin,however,the
ndente†ectsofmixingbecomequite
ndentofthenumericalresolution,itisclear
ofrelativisticgasandthermalgasisvery
EvolutioninNonuniformMedia
eatedourtwo-dimensionalsimulationwith
adensitygradientintheambientmedium
ofunderstandingthedisplacementofthe
seenintheVelaSNR.FollowingDohm-
cx
suchthatthedensitycontrast,fromaminimumatlarge
positiveztoamaximumatlargenegativez,isgivenby
(x[1)~1,andHisthecharacteristiclengthscaleover
whichthedensitychanges.Thesimulationsshownhereuse
x1.2correspondingtoadensitycontrastof5.These
simulationsrequirethecomputationofthefulldomaininz
(i.e.,noassumptionofequatorialsymmetry).Asaresult,our
standardgridof2000zonesrepresentsonlyhalfthe
resolutionusedintheprevioustwo-dimensionalmodel.
InFigure8weshowtheevolutionofthePWN/SNR
systemforanambientmediumlengthscaleofH1]1019
cm,whichcorrespondstothesizeoftheSNRatanageof
D500yr.Earlyintheevolution,wecanestimatetheasym-
metryintheSNRbyassumingonlyradialmotion,such
ionofthepulsarnebula/supernovaremnantforthemodelwithascaleheightofH1]1019cmintheambientmedium.[Seethe
theJournalforacolorversionofthisÐgure.]
10,000 yrs
No. 2, 2001 PULSAR W
FIG. 5.ÈCrushing of the pulsar nebula in a two-dimen
is normalized to take out the expansion of the outer shock
that follows the outer shock front, so that the f
of the grid is always used.
The outer shock front remains very nearly
follows the time evolution of the one-dimensio
within 0.1%. The key di†erence is the instabil
shell of ejecta and the rapid mixing driven
bility. Even before the PWN/SNR interactio
bility of the SSDW (Chevalier et al. 1992) beg
out the shell of shocked ejecta. However, gi
time between the beginning of the simulation
of the reverse shock, there is not enough time
bility to grow to signiÐcant amplitude. F
PWN/SNR collision, the deceleration of the e
the shocked ambient medium (responsible fo
instability) increases in magnitude, thereby d
rapid Rayleigh-Taylor instability. The result is
shell of mixed ejecta and ambient medium a
Ðrst two frames of Figure 5.
Much more dramatic, however, is the Ra
instability working in the opposite direction w
pressure of the compressed pulsar bubble begi
ate the ejecta shell back outwards. Fro
dimensional simulation shown in Figure 2, we
drops below D0.5, the pressure in thR
p
/R
1PWN exceeds that in the SNR and the PW
decelerate the shell of shocked ejecta. This a
the dense ejecta gas by the low-density PWN
to the Rayleigh-Taylor instability. As a result
bility, much of the ejecta gas continues toward
the SNR almost unabated. In the absence o
spherical reverse shock would reach the c
remnant at an age of D35000 yr, and inde
Figure 5 that some has reached the center by
the same process some relativistic gas is displa
center.
We see two competing e†ects of this inst
PWN crushing phase. First, as relativistic ga
from the center, it escapes the full compressio
ling ejecta seen in one dimension. Second,
turbulence driven by this instability leads to ra
the thermal and relativistic gasses. To illustrate
we compare the e†ective value of inR
p
/R
1
10,000yrs30,000yrs50,000yrs
log(ρ/ρa)
0 −1
0
0.2
0.4
0.6
0.8
1
100010000100000
Rp/R1
time(yrs)
No.2,2001PULSARWINDNEBULAEINEVOLVEDSNRs811
FIG.5.ÈCrushingofthepulsarnebulainatwo-dimensionalsimulationusingtheparametersformodelAlistedinTable1.Thelengthscaleoftheimages
isnormalizedtotakeouttheexpansionoftheoutershockfront.[SeetheelectroniceditionoftheJournalforacolorversionofthisÐgure.]
thatfollowstheoutershockfront,sothatthefullresolution
ofthegridisalwaysused.
Theoutershockfrontremainsverynearlysphericaland
followsthetimeevolutionoftheone-dimensionalmodelto
within0.1%.Thekeydi†erenceistheinstabilityofthethin
shellofejectaandtherapidmixingdrivenbythisinsta-
bility.EvenbeforethePWN/SNRinteraction,theinsta-
bilityoftheSSDW(Chevalieretal.1992)beginstospread
outtheshellofshockedejecta.However,giventheshort
timebetweenthebeginningofthesimulationandthecrash
ofthereverseshock,thereisnotenoughtimeforthisinsta-
bilitytogrowtosigniÐcantamplitude.Followingthe
PWN/SNRcollision,thedecelerationoftheejectashellby
theshockedambientmedium(responsiblefortheSSDW
instability)increasesinmagnitude,therebydrivingamore
rapidRayleigh-Taylorinstability.Theresultisabroadened
shellofmixedejectaandambientmediumasseeninthe
ÐrsttwoframesofFigure5.
Muchmoredramatic,however,istheRayleigh-Taylor
instabilityworkingintheoppositedirectionwhenthehigh
pressureofthecompressedpulsarbubblebeginstoacceler-
atetheejectashellbackoutwards.Fromtheone-
dimensionalsimulationshowninFigure2,weseethatonce
dropsbelowD0.5,thepressureinthecompressed R
p
/R
1PWNexceedsthatintheSNRandthePWNbeginsto
deceleratetheshellofshockedejecta.Thisaccelerationof
thedenseejectagasbythelow-densityPWNgasissubject
totheRayleigh-Taylorinstability.Asaresultofthisinsta-
bility,muchoftheejectagascontinuestowardthecenterof
theSNRalmostunabated.IntheabsenceofaPWN,a
sphericalreverseshockwouldreachthecenterofthe
remnantatanageofD35000yr,andindeedweseein
Figure5thatsomehasreachedthecenterbythistime.In
thesameprocesssomerelativisticgasisdisplacedfromthe
center.
Weseetwocompetinge†ectsofthisinstabilityinthe
PWNcrushingphase.First,asrelativisticgasisdisplaced
fromthecenter,itescapesthefullcompressionoftheinfal-
lingejectaseeninonedimension.Second,thevigorous
turbulencedrivenbythisinstabilityleadstorapidmixingof
thethermalandrelativisticgasses.Toillustratethesee†ects,
wecomparethee†ectivevalueofintheone-and R
p
/R
1
two-dimensionalsimulationsinFigure6.Wecomputethe
radiusofthepulsarbubbleinthetwo-dimensionalsimula-
tionbysummingupthevolumeofgaswithc1.66(i.e.,
includingthepartiallymixedgas)andcalculateane†ective
radiusassumingasphericalvolume.Priortothebounce,
theone-andtwo-dimensionalsimulationsarenearlyidenti-
cal.However,atthemomentofbounce,thePWNinthe
two-dimensionalsimulationhasavolumetwicethatofthe
PWNintheone-dimensionalsimulation.Thecompressed
PWNquicklyreboundsintheone-dimensionalsimulation,
butintwo-dimensionsthevolumeofrelativisticgascon-
tinuestoshrinkbecauseofnumericalmixing.
ThisrapiddepletionofthevolumeofthePWNinthe
two-dimensionalsimulationisanartifactofnumericaldif-
fusion;whenonenumericalzonecontainsbothrelativistic
gas(c4/3)andejectagas(c5/3),themass-weighted
averagecisdominatedbythehighdensityoftheejecta.As
aresult,themixingÈandsubsequentlossofPWN
volumeÈisstronglydependentonthenumericalresolution
ofthesimulation.Higherspatialresolutionleadstoless
numericalmixingacrossthecontactinterfacebetweenthe
FIG.6.ÈEvolutionofthepulsarnebulathroughthecrashofthereverse
SNRshockinone(solidline)andtwo(dashedlines)dimensionswithdi†er-
entnumericalresolutions.
10,000 yrs
No. 2, 2001 PULSAR W
FIG. 5.ÈCrushing of the pulsar nebula in a two-dimen
is normalized to take out the expansion of the outer shock
that follows the outer shock front, so that the f
of the grid is always used.
The outer shock front remains very nearly
follows the time evolution of the one-dimensio
within 0.1%. The key di†erence is the instabil
shell of ejecta and the rapid mixing driven
bility. Even before the PWN/SNR interactio
bility of the SSDW (Chevalier et al. 1992) beg
out the shell of shocked ejecta. However, gi
time between the beginning of the simulation
of the reverse shock, there is not enough time
bility to grow to signiÐcant amplitude. F
PWN/SNR collision, the deceleration of the e
the shocked ambient medium (responsible fo
10,000yrs30,000yrs50,000yrs
log(ρ/ρa)
0 −1
No.2,2001PULSARWINDNEBULAEINEVOLVEDSNRs811
FIG.5.ÈCrushingofthepulsarnebulainatwo-dimensionalsimulationusingtheparametersformodelAlistedinTable1.Thelengthscaleoftheimages
isnormalizedtotakeouttheexpansionoftheoutershockfront.[SeetheelectroniceditionoftheJournalforacolorversionofthisÐgure.]
thatfollowstheoutershockfront,sothatthefullresolution
ofthegridisalwaysused.
Theoutershockfrontremainsverynearlysphericaland
followsthetimeevolutionoftheone-dimensionalmodelto
within0.1%.Thekeydi†erenceistheinstabilityofthethin
shellofejectaandtherapidmixingdrivenbythisinsta-
bility.EvenbeforethePWN/SNRinteraction,theinsta-
bilityoftheSSDW(Chevalieretal.1992)beginstospread
outtheshellofshockedejecta.However,giventheshort
timebetweenthebeginningofthesimulationandthecrash
ofthereverseshock,thereisnotenoughtimeforthisinsta-
bilitytogrowtosigniÐcantamplitude.Followingthe
PWN/SNRcollision,thedecelerationoftheejectashellby
theshockedambientmedium(responsiblefortheSSDW
two-dimensionalsimulationsinFigure6.Wecomputethe
radiusofthepulsarbubbleinthetwo-dimensionalsimula-
tionbysummingupthevolumeofgaswithc1.66(i.e.,
includingthepartiallymixedgas)andcalculateane†ective
radiusassumingasphericalvolume.Priortothebounce,
theone-andtwo-dimensionalsimulationsarenearlyidenti-
cal.However,atthemomentofbounce,thePWNinthe
two-dimensionalsimulationhasavolumetwicethatofthe
PWNintheone-dimensionalsimulation.Thecompressed
PWNquicklyreboundsintheone-dimensionalsimulation,
butintwo-dimensionsthevolumeofrelativisticgascon-
tinuestoshrinkbecauseofnumericalmixing.
ThisrapiddepletionofthevolumeofthePWNinthe
two-dimensionalsimulationisanartifactofnumericaldif-
Pulsar-
Wind-Nebel
Stefan Klepser . Pulsar Wind Nebulae . ICRC . The Hague 2015
PWN Evolution in a Nutshell
Free expansion Reverse shock
interaction
Relic stage
SNR
PWN
▪ Easy & independent
▪ R ~ t6/5
▪ All the Crab wisdom, e.g.
▪ Kennel & Coroniti 1984
▪ Martín++, 2012
▪ ...
▪ Messy & depending on SNR
development
▪ Oscillative reverbations
▪ Analytically R ~ t0.3
▪ Only over-idealized and/or
numerical wisdom
▪ Swaluw++ 2001,2004
▪ ...
Pulsar
▪ More messy & more
depending on SNR dev. &
surroundings
▪ R ~ undefined
▪ Only case-by-case wisdom
2-6 kyr 20-100 kyr?
7
Die meisten Galaktischen TeV-Quellen sind PWN.
Elektronen und Positronen erzeugen die Emission.
74. SNR RX J1713-3949
Peter Eger . H.E.S.S. precision measurements of RX J1713.7-3946 . August 201
The new high-resolution H.E.S.S. map
■ exposure: 170 h
■ angular resolution: 0.05º
■ energy threshold: 250 GeV
■ Analysis: Model w/ HiRes cuts
(de Naurois & Rolland, 2007)
5
Peter Eger . H.E.S.S. precision measurements of RX J1713.7-3946 . August 2015
Mapping the magnetic field
15
Peter, March 18th
, 2014, HESS coll. meeting, Potsdam
The X-ray hotspots
B-field map
X-rays: XMM-Newton
H.E.S.S.-PSF-convolved
TeV: H.E.S.S.
War in 2004 das erste aufgelöste Gamma-Bild
von einem astronomischen Objekt.
Jetzt in 2016 haben wir genug Photonen um
ein Bild des Magnetfeldes zu machen.
76. Galaktisches Zentrum
VLA Radio Bild
TeV-Strahlung mit H.E.S.S.
Komplexe Emission.
Vieles ist noch unklar, z.B.
“Ist Teil der Halo-Emission von dunkler Materie”
“Welche Quellen erzeugen die Emission auf der rechten Seite?”
“Ist die zentrale Quelle ein SNR oder das supermassive schwarze Loch?”
“Wie hängt das Galaktische Zentrum mit den Fermi Bubbles zusammen?”
79. Super
Fermi-LAT ?
• Geht nicht gut …
• Wäre extrem teuer,
ein Fermi-LAT kostet
500 Millionen Euro.
• Aber ein Cherenkov-
Teleskop nur
~ 2 Million Euro …
1 Meter
84. Teleskop- und
Array-Optimierung
• Die letzten Jahre: lange
wissenschaftliche Diskussionen und
detaillierte Simulationen um das
beste Cherenkov Teleskope Array
zu machbaren Kosten zu bauen.
• 3 Teleskop-Typen:
• LST = Large size telescopes
• MST = Mid size telescopes
• SST = Small size telescopes
85. ~ 8 LSTs
kleiner als H.E.S.S. 2, ähnlich wie MAGIC-Teleskope
CTA LST MAGIC
86. ~ 40 MSTs
ähnlich wie VERITAS oder H.E.S.S. Teleskope
PrototypinBerlinAdlershof
MSTKamerasteilweisevonMPIKHeidelberg
87. ~ 70 SSTs
SST Prototyp Einweihung
(Dezember 2015 in Meudon. Kamera von MPIK Heidelberg.)
88. CTA Survey
Größeres Gesichtsfeld + Bessere Sensitivität
=
Survey-Geschwindigkeit
300 x schneller als H.E.S.S.
Simulation LMCLMC jetzt
92. Zusammenfassung
• Gamma-Strahlung ist hoch-energetisches Licht.
Es ist nicht-thermische Emission.
• Man kann Gamma-Strahlung mit Satelliten (z.B. Fermi-
LAT) oder von der Erde (z.B. H.E.S.S. oder CTA) messen.
• Gamma-Astronomie ist ein sehr junger Zweig der
Astronomie, vor 10 Jahren gab es noch kein gutes Bild
der Milchstraße in Gamma-Licht.
• Die Milchstrasse ist voll von kosmischen Teilchen-
Beschleunigern (z.B. Pulsare und Supernova-Überreste),
die geladene kosmische Strahlung und Gamma-
Strahlung erzeugen.
93. Danke schön
• Carolin Liefke, für die Einladung und Organisation
für der Reihe “Faszination Astronomie”.
• Kollegen, die Bild-Material und Feedback zu dem
Vortrag gegeben haben:
Werner Hofmann, Axel Donath, Johannes King,
Christopher van Eldik, Mathieu de Naurois, Rolf
Bühler, Heinz Völk, Stefan Funk, Karl Kosack,
Bernhold Feuerstein