Professional Documents
Culture Documents
SPECTRA FORMATION
Mathias Th. Keil and Georg G. Raelt
Max-Plan
k-Institut f
ur Physik (Werner-Heisenberg-Institut)
F
ohringer Ring 6, 80805 M
un
hen, Germany
Hans-Thomas Janka
Max-Plan
k-Institut f
ur Astrophysik
Karl-S
harzs
hild-Str. 1, 85741 Gar
hing, Germany
ABSTRACT
The neutrino
ux and spe
tra formation in a supernova
ore is studied by using a Monte Carlo
ode. The dominant opa
ity
ontribution for is elasti
s
attering on nu
leons N ! N,
where always stands for either or . In addition we swit
h on or o a variety of pro
esses
whi
h allow for the ex
hange of energy or the
reation and destru
tion of neutrino pairs, notably
nu
leon bremsstrahlung NN ! NN , the pair annihilation pro
esses e+e ! and ee !
, re
oil in elasti
nu
leon s
attering, elasti
s
attering on ele
trons e ! e and elasti
s
attering on ele
tron neutrinos and anti-neutrinos e ! e and e ! e. The least
important pro
esses are neutrino-neutrino s
attering and e+e annihilation. The formation of
the spe
tra and
uxes is dominated by the nu
leoni
pro
esses, i.e. bremsstrahlung and elasti
s
attering with re
oil, but also ee annihilation and e s
attering
ontribute on a signi
ant
level. When all pro
esses are in
luded, the spe
tral shape of the emitted neutrino
ux is always
\pin
hed" and
an be represented by a nominal Fermi-Dira
fun
tion with an ee
tive degenera
y
parameter in the range 1{2, depending on the details of the ba
kground model. In all of our
ases
we nd that the average and energy ex
eeds the average e energy by only a small amount,
10% being a typi
al number. Depending on the density, temperature, and
omposition prole,
the
avor-dependent luminosities Le , Le , and L
an mutually dier from ea
h other by up
to a fa
tor of two in either dire
tion.
Subje
t headings:
1. INTRODUCTION
In numeri
al
ore-
ollapse supernova (SN) simulations, the transport of - and -neutrinos has
re
eived s
ant attention be
ause their exa
t
uxes
and spe
tra are probably not
ru
ial for the explosion me
hanism. However, the re
ent experimental eviden
e for neutrino os
illations implies
that the
avor-dependent
uxes and spe
tra emitted by a SN will be partly swapped so that at
any distan
e from the sour
e the a
tual
uxes
and spe
tra
an be very dierent from those originally produ
ed. In prin
iple, this ee
t
an be
1
the neutrino transport only in a very approximate way and do not provide spe
tral information. The
al
ulations performed by the Livermore
group also yield robust explosions (Totani et al.
1998). They in
lude a mixing-length treatment of
the phenomenon of neutron-nger
onve
tion in
the neutron star, that in
reases the early neutrino
luminosities and thus enhan
es the energy transfer
by neutrinos to the postsho
k medium (Wilson &
Mayle 1993). Whether neutron-nger
onve
tion
a
tually o
urs inside the neutrinosphere and has
ee
ts on a ma
ros
opi
s
ale, however, is an unsettled issue.
We will follow here an alternative approa
h to
full hydrodynami
al simulations, i.e. we will study
neutrino transport on the ba
kground of an assumed neutron-star atmosphere. While this approa
h la
ks hydrodynami
al self-
onsisten
y, it
has the great advantage of allowing one to study
systemati
ally the in
uen
e of various pie
es of
mi
ros
opi
input physi
s and of the medium prole. The goal is to develop a
learer pi
ture of the
generi
properties of the SN neutrino spe
tra and
uxes and what they depend upon.
To this end we have adapted the Monte Carlo
ode of Janka (1987, 1991) and added new mi
rophysi
s to it. We go beyond the work of Janka
& Hillebrandt (1989a,b) in that we in
lude the
bremsstrahlung pro
ess, nu
leon re
oils, ee pair
annihilation into , and s
attering of on e
and e. With these extensions we investigate the
neutrino transport systemati
ally for a variety of
medium proles that are representative for dierent SN phases. One of us (Raelt 2001) has re
ently studied the spe
tra-formation problem
with the limitation to nu
leoni
pro
esses (elasti
and inelasti
s
attering, re
oils, bremsstrahlung),
to Maxwell-Boltzmann statisti
s for the neutrinos,
and plane-parallel geometry. Our present study
omplements this more s
hemati
work by in
luding the leptoni
pro
esses, Fermi-Dira
statisti
s,
and spheri
al geometry. In addition we apply our
Monte Carlo
ode to the transport of e and e
and thus are able to
ompare the
avor-dependent
uxes and spe
tra.
In Se
. 2 we rst assess the relative importan
e
of dierent pro
esses in terms of their energydependent \thermalization depth". In this
ontext we introdu
e a number of stellar ba
kground
models. In Se
. 3 we perform a Monte Carlo study
3
of transport on the previously introdu
ed ba
kground models in order to assess the importan
e
of dierent pie
es of input physi
s. In Se
. 4 we
ompare the
uxes and spe
tra with those of
e and e . We
on
lude in Se
. 5 with a dis
ussion
and summary of our ndings.
2. THERMALIZATION DEPTH OF
ENERGY-EXCHANGE PROCESSES
2.1. Simple Pi
ture of Spe
tra Formation
One of our goals is to assess the relative importan
e of dierent neutrino intera
tion
hannels with the ba
kground medium of the SN
ore.
As a rst step it is instru
tive to study the thermalization depth of various energy-ex
hange pro
esses. Within the transport sphere, the neutrinos are trapped by elasti
s
atterings on nu
leons,
N ! N , whi
h are by far the most frequent
rea
tions between neutrinos and parti
les of the
stellar medium. (Unless otherwise noted \neutrinos" always refers to any of , , or .) Assuming for the moment that these
ollisions are
iso-energeti
(no nu
leon re
oils), it is straightforward to dene for a neutrino of given energy
the lo
ation (\thermalization depth") where it
last ex
hanged energy with the medium by a rea
tion su
h as e ! e . Following Shapiro &
Teukolsky (1983) we dene the opti
al depth for
energy ex
hange or thermalization by
Z 1 s
1
1
1
0
dr
therm (r) =
+ (r0 ) :
E (r0 ) T (r0 )
E
r
(1)
Here, E is the mean free path (mfp) for the relevant energy-ex
hange pro
ess and T the transport mfp, i.e. the mfp
orresponding to the
ross
se
tion for momentum ex
hange in the N ! N
rea
tion. The quantities therm, E and T are
all understood to depend on the neutrino energy
. The main philosophy of Eq. (1) is that a neutrino trapped by elasti
s
attering has a
han
e to
ex
hange energy
orresponding to its a
tual diusive path through the s
attering atmosphere (for a
dis
ussion, see Suzuki 1990). The thermalization
depth Rtherm is given by
2
(2)
therm (Rtherm ) = ;
3
where Rtherm depends on the neutrino energy .
In order to determine the lo
ation of the thermalization depth for dierent pro
esses we need to
dene our assumed neutron-star atmospheres. As
a rst example we use a model taken from a full
hydrodynami
al simulation. This model is representative for the a
retion phase; hen
eforth we
will refer to it as the \A
retion-Phase Model I"
(Fig. 1). It was provided to us by O. E. B. Messer
and was already used in Raelt (2001) for a more
s
hemati
study. Based on the Woosley & Weaver
4
Table 1
Chara
teristi
s of power-law models.
Steep
p
10
q
2.5
q=p
0.25
0 [1014 g
m 3 2.0
T0 [MeV
31.66
r0 [km
10
As another self-
onsistent example (A
retionPhase Model II) we obtained a 150 ms postboun
e
model from M. Rampp (personal
ommuni
ation)
that uses a very similar progenitor (s15s7b2). The
simulation in
ludes an approximate general relativisti
treatment in spheri
al symmetry as des
ribed by Rampp & Janka (2002). The three
neutrino
avors are transported with all relevant
intera
tions ex
ept ee pair annihilation to
10
[1013 g/cm3]
1
0.1
0.01
0.001
0.0001
14
12
10
8
6
4
2
0.1
0.01
0.001
15
10
5
0.5
0.4
0.3
Ye
Ye
20
T [MeV]
T [MeV]
[1013 g/cm3]
15 M progenitor model labeled s15s7b, the Newtonian
ollapse simulation was performed with the
SN
ode developed by Mezza
appa et al. (2001).
The snapshot is taken at 324 ms after boun
e when
the sho
k is at about 120 km, i.e. the star still a
retes matter. In this simulation the traditional
mi
rophysi
s for transport was in
luded, i.e.
iso-energeti
s
attering on nu
leons, e+e annihilation and e s
attering.
Shallow
5
1
0.2
0.2
20.0
10
0.2
0.1
0.0
20 30 40 50 60 70 80 90 100
Radius [km]
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
20 30 40 50 60 70 80 90 100
Radius [km]
[MeV]
[MeV]
[MeV]
60
50
40
30
20
10
0
60
50
40
30
20
10
0
60
50
40
30
20
10
0
b
p
n
s
urca
20 30 40 50 60 70
Rtherm [km]
[MeV]
b
p
n
s
urca
[MeV]
[MeV]
[MeV]
[MeV]
[MeV]
60
50
40
30
20
10
0
60
50
40
30
20
10
0
60
50
40
30
20
10
0
20 30 40 50 60 70 80 90
Rtherm [km]
60
50
40
30
20
10
0
60
50
40
30
20
10
0
60
50
40
30
20
10
0
Ye = 0.00
Ye = 0.05
Ye = 0.50
12 13 14 15 16 17 18 19 20
Rtherm [km]
other possibility is the mean energy of the neutrino
ux, in parti
ular the mean energy of those
neutrinos whi
h a
tually leave the star. For our
power-law atmospheres this is always around 20{
25 MeV. Therefore, the pro
ess with the largest
Rtherm in this energy band is the one most relevant for determining the emerging neutrino spe
trum. It appears that at least for steep proles
pair annihilation is never
ru
ial on
e bremsstrahlung is in
luded, i.e. we would guess that in
luding
pair annihilation will not ae
t the emerging neutrino spe
tra. The relevan
e of ele
tron s
attering
is far more di
ult to guess. On the one hand
it surely is more important than re
oil in nu
leon
s
atterings for some of the relevant energies, on
the other hand we are not able to dene an energy
sphere for nu
leon re
oils be
ause this pro
ess is
dierent from the others in that neutrinos transfer
only a small fra
tion of their energy per s
atter7
[MeV]
[MeV]
[MeV]
60
50
40
30
20
10
0
60
50
40
30
20
10
0
60
50
40
30
20
10
0
Ye = 0.00
( ) = 1 + exp(
=T )
f ;
and therefore
7
hi = 180
4
Ye = 0.05
15 20 25
Rtherm [km]
3:1514 T :
(6)
Ye = 0.50
10
(5)
30
hi
a 2
This
hara
teristi
spe
tral energy is useful for estimating the energy transfer from neutrinos to the
stellar medium in rea
tions with
ross se
tions
proportional to 2. For thermal neutrinos with
vanishing
hemi
al potential we nd
r
hirms = 930
441 T 4:5622 T : (13)
With Eq. (6) this
orresponds to hi 0:691hirms.
</T>
1.02
1
0.98
0.96
0.94
0.92
0.9
4.00
3.80
3.60
3.40
3.20
3.00
-2
-1
or o bremsstrahlung (b),+ nu
leon re
oil (r), s
attering on ele
trons (s), e e pair annihilation (p),
and ee annihilation (n). We never in
lude inelasti
nu
leon s
attering NN ! NN as this pro
ess is never important relative to re
oil (Raelt
2001). Likewise, we ignore s
attering on e and e
whi
h is always unimportant if e is in
luded
(Buras et al. 2002). We also negle
t or
s
attering even though su
h pro
esses may have
a larger rate than some of the in
luded leptoni
pro
esses. Pro
esses of this type do not ex
hange
energy between the neutrinos and the ba
kground
medium. They are therefore not expe
ted to ae
t
the emerging
uxes and should also have a minor
ee
t on the emitted spe
tra.
luminosity by a sizable amount without ae
ting mu
h the spe
tral shape. This suggests that
bremsstrahlung is important as a sour
e for
pairs, but that the spe
trum is then shaped by the
energy-ex
hange in s
attering with e. In the next
row we swit
h o e s
attering (bp) so that no
energy is ex
hanged ex
ept by pair-produ
ing pro
esses. The spe
tral energy indeed in
reases signi
antly. However, the biggest energy-ex
hange
ee
t in the s
attering regime is nu
leon re
oil.
In the next two rows we in
lude re
oil (brp) and
then additionally e s
attering (brsp), both lowering the spe
tral energies and also the luminosities.
The pi
ture of all relevant pro
esses is
ompleted by adding ee pair annihilation (brspn),
whi
h is similar to e+e pair annihilation, but a
fa
tor of 2{3 more important (Buras et al. 2002).
The luminosity is again in
reased, an ee
t whi
h
is understood in terms of our bla
kbody pi
ture
for the number and energy spheres. In the lower
panel of Fig. 3 we see that Rtherm moves to larger
radii on
e \n" is swit
hed on, the radiating surfa
e
of the \bla
kbody" in
reases and more pairs are
emitted. For both \p" and \n" Rtherm is strongly
energy dependent and therefore it is impossible to
dene a sharp thermalization radius.
To study the relative importan
e of the dierent pair pro
esses, we swit
h o the leptoni
ones
(row \brs") and
ompare this to only the leptoni
pro
esses (row \rspn"). In this stellar model both
types
ontribute signi
antly. Comparing then
\brsp" with \brsn" shows that among the leptoni
pro
esses \n" is
learly more important than \p".
The last row \brsnpn" in
ludes in addition to
all other pro
esses s
attering on e and e. It was
already shown by Buras et al. (2002) that this pro
ess is about half as important as s
attering on e
and its in
uen
e on the neutrino
ux and spe
tra
is negligible. We show this
ase for
ompleteness
but do not in
lude s
attering on e and e for any
of our further models.
A retion-Phase Model I
Our rst goal is to assess the relative importan
e of dierent energy-ex
hange pro
esses for
the transport. As a rst example we begin with
our A
retion-Phase Model I. The results from our
numeri
al runs are summarized in Table 2 where
for ea
h run we give hi
ux, h2 i
ux, and the pin
hing parameter p
ux for the emerging
ux spe
trum, the temperature and degenera
y parameter of an ee
tive Fermi-Dira
spe
trum produ
ing the same rst two energy moments, and the
luminosity.
The rst row
ontains the muon neutrino
ux
hara
teristi
s of the original Boltzmann transport
al
ulation by Messer. To make a
onne
tion to these results we ran our
ode with the
same input physi
s, i.e. e s
attering (s) and
e+e annihilation (p). There remain small dieren
es between the original spe
tral
hara
teristi
s
and ours. These
an be
aused by dieren
es in
the implementation of the neutrino pro
esses, by
the limited number of energy and angular bins in
the Boltzmann solver, the
oarser resolution of the
radial grid in our Monte Carlo runs, and by our
simple bla
kbody lower boundary
ondition. We
interpret the rst two rows of Table 2 as agreeing
su
iently well with ea
h other that a detailed
understanding of the dieren
es is not warranted.
Hen
eforth we will only dis
uss dierential ee
ts
within our own implementation.
In the next row (bsp) we in
lude nu
leon bremsstrahlung whi
h has the ee
t of in
reasing the
3.3.2.
As another example we study the steep powerlaw model dened in Eq. (3) and Table 1. This
model is supposed to represent the outer layers
of a late-time proto-neutron star but without being hydrostati
ally self-
onsistent. It
onne
ts dire
tly with Raelt (2001), where the same pro-
10
Table 2
Monte Carlo results for A
retion-Phase Model I.
1.1
0.3
2.2
1.9
1.6
1.1
1.2
0.4
0.6
1.2
1.2
1.7
1.2
L
14.4
15.8
19.1
20.1
18.6
14.8
17.6
20.2
13.1
17.0
13.0
16.8
17.9
11
the neutrino spe
trum and therefore to a visibly smaller mean
ux energy and lower ee
tive
spe
tral temperature, but higher ee
tive degenera
y. Many nu
leon s
atterings, however, are
needed to downgrade the high-energy neutrinos
(dierent from e s
attering). Therefore neutrinos stay longer at high energies and experien
e a
larger opa
ity and a larger amount of ba
ks
attering. This suppresses the neutrino
ux signi
antly.
In the runs in
luding both s
attering rea
tions
(brsp), we nd a mixture of the ee
ts of e and
nu
leon s
atterings and an enhan
ed redu
tion of
the mean
ux energy.
Finally, adding the neutrino pair pro
ess yields
almost no
hange in energy and pin
hing, but an
in
reased luminosity as expe
ted from the analogous
ase in Se
. 3.3.1. Although this prole is
rather steep, leptoni
pair pro
esses are still important (Fig. 5).
In order to estimate the sensitivity to the exa
t
treatment of nu
leon bremsstrahlung we have performed one run with the bremsstrahlung rate arti
ially enhan
ed by a fa
tor of 3, and one where
it was de
reased by a fa
tor 0.3. All other pro
esses were in
luded. The emerging
uxes and
spe
tra indeed do not depend sensitively on the
exa
t strength of bremsstrahlung.
3.3.3.
For the shallow power law almost the same dis
ussion as for the steep
ase applies. As we
an
already infer from Fig. 6, leptoni
pro
esses are
more important. This leads to a mu
h higher in
rease of the neutrino
ux on
e \p" or \n" are
in
luded, and to stronger spe
tral pin
hing when
e+e annihilation is swit
hed on. S
attering on
e downgrades the transported neutrino
ux by a
larger amount.
3.4. Summary
12
Table 3
Monte Carlo results for the steep power-law model.
Energy ex
hange
b
{ { {
b
r { {
b
{ { p
b
{ { p
b
{ { p
b
{ s p
b
{ s p
b
{ s p
b
r { p
b
r s p
b
r s p
b
r s p
b
r s p
b3 r s p
b
r s p
b0:3 r s p
b
r s p
Ye
{
{
{
{
{
{
{
{
{
{
{
{
n
n
n
n
n
|
|
0
0.05
0.5
0
0.05
0.5
0.5
0
0.05
0.5
0
0.05
0.05
0.05
0.5
hi ux h2 i ux
25.8
19.5
25.4
25.6
25.5
24.2
23.8
21.3
20.0
20.3
20.3
19.6
20.7
20.3
20.6
20.7
19.8
962.
487.
890.
908.
917.
787.
753.
591.
507.
518.
518.
488.
535.
522.
530.
534.
499.
p ux
1.11
0.98
1.06
1.06
1.08
1.03
1.02
1.00
0.98
0.97
0.97
0.98
0.96
0.97
0.96
0.96
0.97
|
6.0
|
|
|
|
|
6.8
6.0
5.9
5.9
5.9
5.8
6.0
5.9
5.8
5.9
L
|
0.7
|
|
|
|
|
0.3
1.0
1.4
1.4
1.1
1.8
1.3
1.7
1.8
1.2
21.0
14.5
23.8
23.2
21.6
24.5
24.5
23.1
16.8
19.7
19.5
18.7
23.9
24.2
23.8
23.4
21.4
Table 4
Monte Carlo results for the shallow power-law model.
Energy ex
hange
b { { { {
b r { { {
b { { p {
b { { p {
b { { p {
b { s p {
b { s p {
b { s p {
b r { p {
b r s p {
b r s p {
b r s p {
b r s p n
b r s p n
b r s p n
Ye
|
|
0
0.05
0.5
0
0.05
0.5
0.5
0
0.05
0.5
0
0.05
0.5
hi ux h2 i ux
27.7
20.1
27.7
27.9
28.3
25.5
25.4
23.5
22.5
22.3
22.2
21.7
22.2
22.4
21.8
13
1120.
521.
974.
990.
1019.
830.
815.
706.
624.
612.
609.
585.
608.
615.
587.
p ux
1.12
0.99
0.98
0.98
0.98
0.98
0.97
0.98
0.95
0.95
0.95
0.95
0.94
0.94
0.95
|
6.3
8.3
8.3
8.5
7.6
7.5
7.1
6.1
6.1
6.1
6.1
6.0
6.1
6.1
|
0.4
1.0
1.1
1.0
1.1
1.2
1.0
2.2
2.1
2.1
1.9
2.2
2.3
1.9
L
20.3
13.4
43.1
43.3
38.3
46.2
46.3
44.8
33.1
39.6
39.1
39.2
54.7
54.9
51.3
4. COMPARING DIFFERENT
FLAVORS
4.1. Monte Carlo Study
Table 5
Comparing Monte Carlo results for different flavors.
Ye
A retion-Phase Model I
Original
,
e
e
Our runs
, (\sp")
, (all pro
esses)
e
e
A retion-Phase Model II
Original
,
e
e
Our runs
,
e
e
= 2:5
,
e
e
,
e
e
q = 3:0
,
e
e
q = 3:5
,
e
q
h2 i
ux
hi
ux hhi
ux
e i
ux
p ux
L
|
|
|
|
|
|
|
17.5
14.6
12.5
16.6
14.3
14.0
11.8
1.20
1
0.86
1.19
1.02
1
0.84
388.
253.
190.
362.
260.
237.
175.
0.97
0.91
0.93
1.01
0.97
0.93
0.97
5.2
3.5
3.2
5.3
4.3
3.6
3.4
1.1
3.4
2.8
0.3
1.2
2.7
1.4
14.4
29.2
30.8
15.8
17.9
31.7
31.9
|
|
|
|
|
|
17.2
15.8
12.9
15.7
15.4
13.0
1.09
1
0.82
1.02
1
0.84
380.
300.
207.
317.
283.
207.
0.98
0.92
0.96
0.98
0.92
0.95
5.2
4.0
3.7
4.8
3.8
3.6
0.8
3.0
1.7
0.8
3.2
2.1
32.4
68.1
65.6
27.8
73.5
73.9
20.4
18.5
12.7
20.4
17.9
13.4
17.7
15.5
10.5
15.8
13.0
9.4
22.0
19.3
14.7
1.10
1
0.69
1.14
1
0.75
1.14
1
0.68
1.22
1
0.72
1.14
1
0.76
525.
413.
198.
521.
383.
218.
393.
289.
132.
310.
207.
103.
596.
440.
262.
0.96
0.92
0.94
0.97
0.92
0.93
0.96
0.93
0.92
0.95
0.94
0.90
0.94
0.91
0.93
5.9
4.6
3.4
5.9
4.4
3.4
5.0
4.0
2.6
4.4
3.5
2.1
6.0
4.5
3.8
1.5
3.0
2.4
1.5
3.1
2.9
1.8
2.8
3.0
2.1
2.3
3.9
2.2
3.7
2.7
23.5
23.5
12.8
23.3
11.7
24.4
12.7
8.8
6.6
7.9
4.3
4.1
53.9
85.7
56.5
0.15
0.15
0.15
0.2
0.2
0.2
0.1
0.1
0.1
0.07
0.07
e
0.07
Shallow Power Law p = 5, q = 1
,
0.3
e
0.3
e
0.3
Note.|
15
ing results from the original runs. For the transport of our inner boundary is Rin = 16 km,
while for e and e we use Rin = 24 km. For e
and e the
harged-
urrent pro
esses (ur
a) keep
these neutrinos in LTE up to larger radii than pair
pro
esses in the
ase of . With our
hoi
e of Rin
the neutrinos are in LTE within the innermost radial zones.
The results are similar to the A
retion-Phase
Model I. The luminosities are not equipartitioned
but instead follow roughly Le Le 2 L .
The ratios of mean energies are he i : he i :
h i = 0:82 : 1 : 1:09 in the original run and
0:84 : 1 : 1:02 in our run.
In summary, both a
retion-phase models agree
reasonably well in the he i : he i ratio for all runs.
Moreover, using traditional input physi
s one nds
something like he i : h i = 1 : 1:20. Depending
on the implementation of the new input physi
s
and depending on the model one nds results between he i : h i = 1 : 1:02 and 1 : 1:09. The
higher ratio in Rampp's simulation
ould be due
to the in
lusion of weak magnetism whi
h tends
to raise h i more than he i.
In order to estimate the
orresponding results
for later stages of the proto-neutron star evolution
we employ our steep power-law model. We vary
the power q of the temperature prole within a
reasonable range so that q=p = 0:25{0.35, with q
and p dened in Eq. (3). Ye is xed by demanding
roughly equal number
uxes for e and e be
ause
a few se
onds after boun
e deleptonization should
be essentially
omplete. The
uxes of these neutrinos depend very sensitively on Ye so that this
onstraint is only rea
hed to within about 30%
without tuning Ye to three de
imal pla
es. However, the mean energies are rather insensitive to
the exa
t value of Ye. This is illustrated by the
steep power-law model with q = 2:5 where we
show results for Ye = 0:15 and 0.20. The number
uxes of e and e dier by less than 30% for
Ye = 0:15, but dier by a fa
tor of 3 for Ye = 0:2.
At the same time, the average spe
tral energies
barely
hange.
The ratios of mean energies are not very dierent from those of the a
retion-phase models. Of
ourse, the absolute
ux energies have no physi
al
meaning be
ause we adjusted the stellar prole in
order to obtain realisti
values. For the luminosities we nd Le < L , dierent from the a
retion
[MeV]
[MeV]
[MeV]
ux parameters are measured at this radius be
ause farther out Doppler ee
ts of the original
model would make it di
ult to
ompare the results. Keeping in mind that we use very dierent numeri
al approa
hes and somewhat dierent
input physi
s, the agreement in parti
ular for e
and e is remarkably good. This agreement shows
on
e more that our Monte Carlo approa
h likely
aptures at least the dierential ee
ts of the new
mi
rophysi
s in a satisfa
tory manner.
In Fig. 8 we
ompare our
al
ulations for the
A
retion-Phase Model II with those of the original simulation. The step-like
urve again represents the mean energy of neutrinos in LTE for zero
hemi
al potential. The smooth solid line is the
mean energy hi from our runs, the dotted (lower)
line gives hi
ux. The
rosses are the
orrespond45
40
35
30
25
20
15
10
45
40
35
30
25
20
15
10
45
40
35
30
25
20
15
10
flux
20 30 40 50 60 70 80 90100
Radius [km]
phase. The steep power law implies that the radiating surfa
es are similar for all
avors so that
it is not surprising that the
avor with the largest
energies also produ
es the largest luminosity.
We nd that h i always ex
eeds he i by a
small amount, the exa
t value depending on the
stellar model. During the a
retion phase the energies seem to be almost identi
al, later they may
dier by up to 20%. We have not found a model
where the energies dier by the large amounts
whi
h are sometimes assumed in the literature. At
late times when Ye is small the mi
rophysi
s governing e transport is
loser to that for than
at early times. Therefore, one expe
ts that at late
times the behavior of e is more similar to than
at early times. We do not see any argument for
expe
ting an extreme hierar
hy of energies at late
times for self-
onsistent stellar models.
We never nd exa
t equipartition of the
avordependent luminosities. Depending on the stellar
prole the
uxes
an mutually dier by up to a
fa
tor of 2 in either dire
tion.
Table 6
Flavor dependent flux
hara
teristi
s from the literature.
h i
L
Le Le
tpb he i he i h i
he i
1.0
12 24 22 0:50 : 1 : 0:92 20 20 20
0.3
12 15 19 0:80 : 1 : 1:26 20 20 20
10
11 20 25 0:55 : 1 : 1:25 0.5 0.5 1
0.5
10 12 25 0:83 : 1 : 2:08 3
5 16
0.13 11 13 24 0:85 : 1 : 1:85 30 30 16
0.3
8 14 16 0:57 : 1 : 1:14 30 220 65
1
9.5 13 15 0:73 : 1 : 1:15 4
4 3
20
8 10
9 0:80 : 1 : 0:90 0.3 0.3 0.07
1
9.5 13 15 0:73 : 1 : 1:15 3
3 3
15
8
9 9.5 0:89 : 1 : 1:06 0.4 0.4 0.3
1
9 12 13 0:75 : 1 : 1:08 3
3 3
15
7
8
8 0:88 : 1 : 1:00 0.3 0.3 0.3
0.32 13 15 18 0:86 : 1 : 1:20 31 29 14
0.32 12 14 14 0:84 : 1 : 1:02 32 32 18
0.15 13 16 17 0:82 : 1 : 1:09 66 68 32
0.15 13 15 16 0:84 : 1 : 1:02 74 74 28
0.5
0.5
16
19
19
21
24 0:84 : 1 : 1:26 25
24 0:90 : 1 : 1:14 30
We give the time post boun e (tpb) in s, hi in MeV, and L in 1051 erg s 1.
Note.|
18
25
30
8
10
xed ba
kground model, not self-
onsistent simulations. Taking into a
ount the dierent mi
rophysi
s the mean energies are
onsistent with our
present work. The mean energies of e were somewhat on the low side relative to e and the e luminosity was overestimated. Both
an be understood
by the fa
t that the stellar ba
kground
ontained
an overly large abundan
e of neutrons, be
ause
the model resulted from a post-boun
e
al
ulation
whi
h only in
luded ele
tron neutrino transport.
Suzuki (1990) studied models with initial temperature and density proles typi
al of protoneutron stars at the beginning of the KelvinHelmholtz
ooling phase about half a se
ond after
boun
e. He used the relatively sti nu
lear equation of state developed by Hillebrandt & Wol
(1985). In our table we show the results of the
model C12. From Suzuki (1991) we took the
model labeled C20 whi
h in
ludes bremsstrahlung. The model C48 from Suzuki (1993) in
ludes multiple-s
attering suppression of bremsstrahlung. Suzuki's models are the only ones from
the previous literature whi
h go beyond the traditional mi
rophysi
s for transport. It is reassuring that his ratios of mean energies
ome
losest
to the ones we nd.
Over the past few years, rst results from Boltzmann solvers
oupled with hydrodynami
al simulations have be
ome available, notably the unpublished ones that we used as our A
retionPhase Models I and II. For
onvenien
e we in
lude
them in Table 6. Moreover, we in
lude two simulations similar to the A
retion-Phase Model I,
one by Mezza
appa et al. (2001) and the other
by Liebendorfer et al. (2001). These latter papers
show rms energies instead of mean energies. Re
alling that the former tend to be about 45% larger
than the latter these results are entirely
onsistent
with our A
retion-Phase Models. Moreover, the
ratios of hirms tend to exaggerate the spread between the
avor-dependent mean energies be
ause
of dierent amounts of spe
tral pin
hing, i.e. different ee
tive degenera
y parameters. To illustrate this point we take the rst two rows from
Table 5 as an example. The ratio of mean energies for Fermi-Dira
spe
tra with temperatures
T1 = 5:2 and T2 = 3:5 and degenera
y parameters
1 = 1:1 and 2 = 3:4 is h1 i=h2 i = 1:19, whereas
the ratio of rms energies equals 1.30.
To summarize, the frequently assumed exa
t
equipartition of the emitted energy among all
avors appears only in some simulations of the Livermore group. We note that the
avor-dependent
luminosities tend to be quite sensitive to the detailed atmospheri
stru
ture and
hemi
al
omposition. On the other hand, the often-assumed extreme hierar
hy of mean energies was only found
in the early simulations of Bruenn (1987) and of
Myra & Burrows (1990), possibly a
onsequen
e
of the neutron-star equation of state used in these
al
ulations.
If we ignore results whi
h appear to be \outliers", the pi
ture emerging from Table 6 is quite
onsistent with our own ndings. For the luminosities, typi
ally Le Le and a fa
tor of 2{3
between this and L in either dire
tion, depending on the evolutionary phase. For the mean
energies we read typi
al ratios in the range of
he i : he i : h i = 0:8{0:9 : 1 : 1:1{1.3. The
more re
ent simulations involving a Boltzmann
solvers show a
onsistent behavior and will in future provide reliable information about neutrino
uxes and spe
tra.
5. DISCUSSION AND SUMMARY
We have studied the formation of neutrino spe
tra and
uxes in a SN
ore. Using a Monte Carlo
ode for neutrino transport, we varied the mi
ros
opi
input physi
s as well as the underlying
stati
proto-neutron star atmosphere. We used
two ba
kground models from self-
onsistent hydrodynami
al simulations, and several power-law
models with varying power-law indi
es for the density and temperature and dierent values for the
ele
tron fra
tion Ye, taken to be
onstant.
The transport opa
ity is dominated by
neutral-
urrent s
attering on nu
leons. In addition, there are number-
hanging pro
esses (nu
leon bremsstrahlung, leptoni
pair annihilation)
and energy-
hanging pro
esses (nu
leon re
oil,
e s
attering). The spe
tra and
uxes are
roughly a
ounted for if one in
ludes one signifi
ant
hannel of pair produ
tion and one for energy ex
hange in addition to N s
attering. For
example, the traditional set of mi
rophysi
s (isoenergeti
N s
attering, e+e annihilation, and
e s
attering) yields
omparable spe
tra and
uxes to a
al
ulation where pairs are produ
ed by
nu
leon bremsstrahlung and energy is ex
hanged
19
by nu
leon re
oil. The overall result is quite robust against the detailed
hoi
e of mi
rophysi
s.
However, in state-of-the-art simulations where
one aims at a pre
ision better than some 10{20%
for the
uxes and spe
tral energies, one needs
to in
lude bremsstrahlung, leptoni
pair annihilation, neutrino-ele
tron s
attering, and energy
transfer in neutrino-nu
leon
ollisions. Interestingly, the traditional e+e annihilation pro
ess is
always mu
h less important than ee annihilation, a point that we previously raised with our
ollaborators (Buras et al. 2002). None of the rea
tions studied here
an be negle
ted ex
ept perhaps the traditional e+e annihilation pro
ess and
e and e s
attering.
The existing treatments of the nu
lear-physi
s
aspe
ts of the NN ! NN bremsstrahlung pro
ess are rather s
hemati
. We nd, however, that
the
uxes and spe
tra do not depend sensitively on the exa
t strength of the bremsstrahlung
rate. Therefore, while a more adequate treatment
of bremsstrahlung remains desirable, the nal results are unlikely to be mu
h ae
ted.
In
luding all pro
esses works in the dire
tion of
making the
uxes and spe
tra of more similar
to those of e
ompared to a
al
ulation with the
traditional set of input physi
s. During the a
retion phase the neutron-star atmosphere is relatively expanded, i.e. the density and temperature gradients are relatively shallow. Our investigation suggests that during this phase h i is
only slightly larger than he i, perhaps by a few
per
ent or 10% at most. This result agrees with
the rst hydrodynami
al simulation in
luding all
of the relevant mi
rophysi
s ex
ept ee annihilation (A
retion-Phase Model II) provided to us by
M. Rampp. For the luminosities of the dierent
neutrino spe
ies one nds Le Le 2 L . The
smallness of L is not surprising be
ause the ee
tive radiating surfa
e is mu
h smaller than for e.
During the Kelvin-Helmholtz
ooling phase the
neutron-star atmosphere will be more
ompa
t,
the density and temperature gradients will be
steeper. Therefore, the radiating surfa
es for all
spe
ies will be
ome more similar. In this situation L may well be
ome larger than Le . However, the relative luminosities depend sensitively
on the ele
tron
on
entration. Therefore, without
a self-
onsistent hydrostati
late-time model it is
di
ult to
laim this luminosity
ross-over with
onden
e.
The ratio of the spe
tral energies is most sensitive to the temperature gradient relative to the
density gradient. In our power-law models we used
/ r p and T / r q . Varying q=p between
0.25 and 0.35 we nd that he i : h i varies between 1 : 1:10 and 1 : 1:22. Noting that the upper range for q=p seems unrealisti
ally large we
on
lude that even at late times the spe
tral differen
es should be small; 20% sounds like a safe
upper limit. We are looking forward to this predi
tion being
he
ked in a full-s
ale self-
onsistent
neutron-star evolution model with a Boltzmann
solver.
The statements in the previous literature fall
into two
lasses. One group of workers, using the
traditional set of mi
rophysi
s, found spe
tral differen
es between e and on the 25% level, a
range whi
h largely agrees with our ndings in
view of the dierent mi
rophysi
s. Other papers
laim ratios as large as he i : h i = 1 : 1:8 or
even ex
eeding 1 : 2. We have no explanation for
these latter results. At least within the framework
of our simple power-law models we do not understand whi
h parameter
ould be reasonably adjusted to rea
h su
h extreme spe
tral dieren
es.
In a high-statisti
s neutrino observation of a future gala
ti
SN one may well be able to dis
over
signatures for
avor os
illations. However, when
studying these questions one has to allow for the
possibility of very small spe
tral dieren
es, and
onversely, for the possibility of large
ux dieren
es. This situation is orthogonal to what often
has been assumed in papers studying possible os
illation signatures.
We have always assumed that and behave identi
ally. However, weak-magnetism ee
ts
render the N and N s
attering
ross se
tions
somewhat dierent (see, e.g., Horowitz 2002). To
study the resulting dieren
es of the and
uxes and spe
tra requires a modied setup of our
Monte Carlo
ode be
ause and have to be
transported simultaneously. The dierent transport
oe
ients will
ause a
hemi
al potential
to build up. While we expe
t the dieren
es of
uxes and spe
tra to be small, we intend to investigate this issue quantitatively in the near future.
20
ACKNOWLEDGMENTS
Our Monte Carlo
ode is based on that developed by Janka (1987) where a detailed des
ription of the numeri
al aspe
ts
an be found. The
ode was rst applied to
al
ulations of neutrino
transport in supernovae by Janka & Hillebrandt
(1989a,b) and Janka (1991). It uses Monte Carlo
methods to follow the individual destinies of sample neutrinos (parti
le \pa
kages" with suitably
attributed weights to represent a number of real
neutrinos) on their way through the star from the
moment of
reation or in
ow to their absorption
or es
ape through the inner or outer boundaries.
The
onsidered stellar ba
kground is assumed to
be spheri
ally symmetri
and stati
, and the sample neutrinos are
hara
terized by their weight fa
tors and by
ontinuous values of energy, radial position and dire
tion of motion, represented by the
osine of the angle relative to the radial dire
tion.
The rates of neutrino intera
tions with parti
les of
the stellar medium
an be evaluated by taking into
a
ount Fermion blo
king ee
ts a
ording to the
lo
al phase-spa
e distributions of neutrinos (Janka
& Hillebrandt 1989b).
As ba
kground stellar models we use the ones
des
ribed in Se
. 2.2. They are dened by radial proles of the density , temperature T , and
ele
tron fra
tion Ye, i.e. the number of ele
trons
per baryon. The
al
ulations span the range between some inner radius Rin and outer radius Rout.
These bound the
omputational domain whi
h is
divided into 30 equally spa
ed radial zones. In
ea
h zone , T , and Ye are taken to be
onstant.
Rin is
hosen at su
h high density and temperature that the neutrinos are in LTE in at least
the rst radial zone. Rout is pla
ed in a region
where the neutrinos essentially stream freely. At
21
During a time step the path of ea
h test parti
le through the stellar atmosphere is followed by
Monte Carlo sampling. With random numbers we
de
ide whether it
ies freely or intera
ts. If it intera
ts it
an s
atter or it
an be absorbed; in
this
ase we turn to the next parti
le. For s
attering we determine the new momentum and position
and
ontinue with the pro
ess until the time step
is used up. Parti
les leaving through the lower or
upper boundaries are eliminated from the transport.
After a
ertain number of time steps (typi
ally
around 15,000) the neutrino distribution rea
hes
a stationary state and further
hanges o
ur only
due to statisti
al
u
tuations. At that stage we
start averaging the output quantities over the next
500 time steps.
Sre oil !; k
exp
! T
k
!k
(B3)
4T !k
with !k = k2=2m.
Multiplying Eq. (B1) with the density of nu
leons, ignoring phase spa
e blo
king of the essentially nondegenerate nu
leons, yields the dierential rates that
an then be integrated for obtaining the required energy and angular dierential
rates. In the
ase of re
oil the numeri
al integrations are rather tri
ky be
ause Eq. (B1) is strongly
forward peaked. In our
ode we employ the \reje
tion method" for obtaining the integrated rates
(Press 1992).
B.2. Bremsstrahlung
B. NEUTRINO PROCESSES
B.1. Neutrino-Nu
leon S
attering
parti
les as indi
ated in Fig. 9. The weak intera
tion
onstants for e+e annihilation are
1 + 2 sin2 W ; CA = 1 (B6)
CV =
2
2
while for ee annihilation they are
1
CA = C V = :
(B7)
2
e, e
k1
k3
k2
k4
e+,
spins
nbaryons
e, e
k1
k3
k2
k4
e, e
Dutta, G., Indumathi, D., Murthy, M. V., & Rajasekaran, G. 2000, \Neutrinos from stellar
ollapse: Comparison of the ee
ts of three and
four
avor mixings," Phys. Rev. D, 62, 093014
Freund, M., Huber, P., & Lindner, M. 2001,
\Systemati
exploration of the neutrino fa
tory
parameter spa
e in
luding errors and
orrelations," Nu
l. Phys. B, 615, 331
Fryer, C. L. 1999, \Mass limits for bla
k hole formation," Ap. J., 522, 413
Fryer, C. L. & Warren, M. S. 2002, \Modeling
ore-
ollapse supernovae in 3-dimensions," Ap.
J. Lett., in press, astro-ph/0206017
Fuller, G. M., Haxton, W. C., & M
Laughlin,
G. C. 1999, \Prospe
ts for dete
ting supernova
neutrino
avor os
illations," Phys. Rev. D, 59,
085005
Fuller, G. M., Mayle, R., Meyer, B. S., & Wilson,
J. R. 1992, \Can a
losure mass neutrino help
solve the supernova sho
k reheating problem?,"
Ap. J., 389, 517
Giovanoni, P. M., Ellison, D. C., & Bruenn,
S. W. 1989, \Neutrino transport during the
ore
boun
e phase of a type II supernova explosion,"
Ap. J., 342, 416
Hannestad, S. & Madsen, J. 1995, \Neutrino de
oupling in the early universe," Phys. Rev. D,
52, 1764
Hannestad, S. & Raelt, G. 1998, \Supernova neutrino opa
ity from nu
leon nu
leon bremsstrahlung and related pro
esses," Ap. J., 507, 339
Hillebrandt, W. 1987, \Stellar
ollapse and supernova explosions," in: High Energy Phenomena
around Collapsed Stars," Pro
s. NATO ASI, 2{
13 Sept. 1985, Cargese, Corsi
a, Fran
e, edited
by D. Pa
ini (Reidel, Dordre
ht), p. 73
Hillebrandt, W. & Wol, R. G. 1985, \Models
of type II supernova explosions," in: Nu
leosynthesis: Challenges and New Developments,
edited by W.D. Arnett and J.W. Truran (University of Chi
ago Press, Chi
ago), p. 131
Horowitz, C. J. 2002, \Weak magnetism for antineutrinos in supernovae," Phys. Rev. D, 65,
043001
Barger, V., Marfatia, D., & Wood, B. P. 2001, \Inverting a supernova: Neutrino mixing, temperatures and binding energies," hep-ph/0112125
Bruenn, S. W. 1987, \Neutrinos from SN1987A
and
urrent status of stellar-
ore
ollapse,"
Phys. Rev. Lett. 59, 938
Bu
ella, F., Esposito, S., Gualdi, C., & Miele, G.
1997, \Supernova neutrino energy spe
tra and
the MSW ee
t," Z. Phys. C, 73, 633
Buras, R., Janka, H.-T., Keil, M. Th., Raelt,
G. G., & Rampp, M. 2002, \Ele
tron neutrino pair annihilation: A new sour
e for muon
and tau neutrinos," submitted to Ap.J. (astroph/0205006)
Burrows, A. 1988, \Supernova neutrinos," Astrophys. J., 334, 891
Burrows, A. & Sawyer, R. F. 1998, \Ee
ts of
orrelations on neutrino opa
ities in nu
lear matter," Phys. Rev. C, 58, 554
Burrows, A. & Sawyer, R. F. 1999, \Many-body
orre
tions to
harged-
urrent neutrino absorption rates in nu
lear matter," Phys. Rev. C, 59,
510
Burrows, A., Young, T., Pinto, Ph., Eastman, R.
& Thompson, T. A. 2000, \Supernova neutrinos
and a new algorithm for neutrino transport and
some appli
ations," Astrophys. J., 539, 865
Carter, G. W. & Prakash, M. 2002, \The quen
hing of the axial
oupling in nu
lear and neutronstar matter," Phys. Lett. B, 525, 249
Cervera, A., Donini, A., Gavela, M. B., Gomez
Cadenas, J. J., Hernandez, P., Mena, O., &
Rigolin, S. 2000, \Golden measurements at a
neutrino fa
tory," Nu
l. Phys. B, 579, 17 [Erratum ibid. 593, 731 (2000)
Chiu, S. H. & Kuo, T. K. 2000, \Ee
ts of neutrino temperatures and mass hierar
hies on the
dete
tion of supernova neutrinos," Phys. Rev.
D, 61, 073015
Dighe, A. S. & Smirnov, A. Y. 2000, \Identifying the neutrino mass spe
trum from the neutrino burst from a supernova," Phys. Rev. D,
62, 033007
24
Rampp, M. & Janka, H.-T. 2002, \Radiation hydrodynami
s with neutrinos: Variable Eddington fa
tor method for
ore-
ollapse supernova
simulations," submitted to Astron. Astrophys.
(astro-ph/0203101)
Rampp, M., Buras, R., Janka, H.-T., Raelt,
G. 2002, \Core-
ollapse supernova simulations:
Variations of the input physi
s," in: Pro
s. 11th
Workshop on \Nu
lear Astrophysi
s," Ringberg Castle, Tegernsee, Germany, Feb. 11{16,
2002, MPA-Report P13 (Max-Plan
k-Institut
fur Astrophysik, Gar
hing), p. 119 (astroph/0203493)
Shapiro, S.L. & Teukolsky, S.A. 1983, \Bla
k
holes, white dwarfs and neutron stars: The
physi
s of
ompa
t obje
ts" (John Wiley &
Sons, New York)
Smirnov, A. Y., Spergel, D. N., & Bah
all, J. N.
1994, \Is Large Lepton Mixing Ex
luded?,"
Phys. Rev. D, 49, 1389
Suzuki, H. 1990, \Neutrino burst from supernova explosion and proto neutron star
ooling,"
(Ph.D. Thesis, University of Tokyo)
Suzuki, H. 1991, \Neutrino emission from protoneutron star with modied Ur
a and nu
leon
bremsstrahlung pro
esses," Num. Astrophys.
Japan, 2, 267
Suzuki, H. 1993, \Supernova neutrinos|Multigroup simulations of neutrinos from protoneutron star," in: Pro
. International Symposium
on Neutrino Astrophysi
s: Frontiers of Neutrino Astrophysi
s, 19{22 O
t. 1992, Takayama,
Japan, edited by Y. Suzuki and K. Nakamura
(Universal A
ademy Press, Tokyo)
Takahashi, K. & Sato, K. 2002, \Earth ee
ts on
supernova neutrinos and their impli
ations for
neutrino parameters," arXiv:hep-ph/0110105.
Takahashi, K., El Eid, M. F. E., & Hillebrandt, W.
1978, \Beta transition rates in hot and dense
matter," Astron. Astrophys., 67, 185
Thompson, T. A., Burrows, A., & Horvath, J. E.
2000, \Mu and tau neutrino thermalization
and produ
tion in supernovae: Pro
esses and
times
ales," Phys. Rev. C, 62, 035802
26