NUCLEAR

PHYSICS A
ELSEVIER Nuclear Physics A 625 (1997) 325-346

In-medium kaon production at the mean-field level

Jtirgen Schaffner-Bielich a'b, I g o r N . M i s h u s t i n a,c, J a k o b B o n d o r f b
a The Niels Bohr Institute, Blegdamsvej 17, DK-2100 Copenhagen, Denmark
b Nuclear Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
c The Kurchatov Institute Russian Research Center, Moscow 123182, Russia

Received 23 April 1997

Abstract
The in-medium mass and energy of kaons and antikaons are studied within the relativistic mean-
field approach compatible with kaon-nucleon scattering data and are compared with predictions
from chiral models. Implications for the subthreshold production of kaons and antikaons in heavy-
ion collisions are discussed. On the mean-field level we find only small in-medium corrections for
the relevant kaon production processes. The production of kaons is even less favourable at high
density due to repulsive vector interactions. We conclude that one has to go beyond the mean-
field level and take fluctuations and secondary production processes via resonances into account
to explain the recently observed enhancement of kaon production at subthreshold energies. The
situation is different for antikaons where in-medium effects strongly enhance their production
rates. We also see strong in-medium modifications of the absorption processes for antikaons and
A's which might be visible in flow measurements. We predict that at high density the threshold
energy for antikaon and A production as well as for the annihilation become equal leading to
similar numbers of antikaons and A's in the dense zone of a relativistic heavy-ion collision.
(~ 1997 Published by Elsevier Science B.V.

1. Introduction

Kaons are an interesting tool to investigate the properties of hadrons inside nuclear
matter. Kaon beam experiments on nuclear targets give information about the kaon
optical potential as well as about the properties of hypernuclei. The status on this field,
theoretically and experimentally, has been summarized by Dover and Walker in [1].
I n - m e d i u m properties o f hadrons have received considerable attention again recently in
relation to relativistic heavy-ion collisions. Charged kaons ( K +) seem to be a quite
promising tool for probing the dense interior of the collision zone as their mean tree
path is long enough to escape without further interactions. Kaplan and Nelson first

0375-9474/97/$17.00 (~) 1997 Published by Elsevier Science B.V. All rights reserved.
P1!S0375-9474(96)00369-2
326 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346

proposed that a kaon condensed phase may be formed in dense matter [2]. Further
studies within the Nambu-Jona-Lasinio model [3], chiral perturbation theory [4] and
the one-boson exchange model [ 5 ] showed that the kaons ( K +) see a repulsive potential
in the medium and will not condense.
On the other hand, the antikaons ( K - ) feel a strong attraction in the medium,
as demonstrated by recent calculations taking into account the contribution from the
A(1405) resonance just below the K - p threshold [6,7]. It was then predicted [8] that
antikaons can form a condensate in the dense interior of a neutron star. The calculations
were performed within the chiral perturbation approach, consistent with scattering data
[4,9] and kaonic atoms [ 10]. In this approach the scalar density is identified with the
baryon density and higher order terms in density are neglected. These shortcomings
were criticized in [5,11 ] using the relativistic mean-field (RMF) model consistent with
bulk properties of nuclear matter. As shown in Ref. [ 12], the appearance of hyperons
also shifts the onset of kaon condensation to higher density. When non-linear terms
in density and hyperon-hyperon interactions are taken into account [ 13] the antikaon
condensed phase does not appear inside the neutron stars.
Here we extend our work [13] to symmetric nuclear matter and apply it for the
situation in heavy-ion collisions at threshold energies. Subthreshold K + production rates
were recently measured in heavy-ion reactions at GSI [ 14]. Earlier works showed that
the in-medium modifications of kaons and antikaons might be observable in heavy-ion
collisions near the threshold. For example, it was shown that the kaons are sensitive to
the equation of state (EOS) of nuclear matter [ 15-17]. More kaons are expected in the
case of a soft EOS compared to a hard one. On the other hand the rate also depends on
the parametrization of the K + N cross section [ 18,19]. It was also shown that N-body
collisions [20] and the high-momentum tail of the nucleon momentum distribution [21]
can not alone explain the data. The influence of rescatterings and formation of resonance
(A) matter was studied within microscopic models like QMD [16] and RBUU [22].
It was demonstrated that they are essential to explain the data. The RBUU model was
also used to study the kaon effective energy in medium and kaon production [23]. The
results of the transport simulations [23] are essentially similar to the ones obtained
without medium modifications [22], because the in-medium kaon mass used in [23]
is quite close to its free mass. But there exist other observables which might be better
suited for extracting in-medium effects. The flow of kaons might be a promising tool for
measuring the kaon potential in dense matter [24]. And more pronounced in-medium
effects are expected for the case of K - [ 25 ]. Indeed, enhanced production rates for K -
have been seen in recent GSI experiments [26].
In this paper we investigate the possible influence of a dense nuclear environment on
the properties of kaons and antikaons. We show that the in-medium effects on the mean-
field level can not explain the measured enhanced production rates of kaons in contrast
to the conclusion drawn in Ref. [ 25 ]. We discuss two different approaches: first, an one-
boson exchange model consistent with nuclear properties and second, a chiral approach
where the parameters are fixed by s-wave scattering lengths and the low-density theorem.
In-medium effects for A's are also studied by linking them to hypernuclear data. We
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 327

show that the phase space in the medium for the processes N N --~ N A K and also for
secondary processes such as 7rN ---* A K and NA ~ N A K does not change considerably.
On the contrary, effects non-linear in density even cause an enhanced repulsion for
these processes at high densities. Hence, subthreshold production of kaons can not be
explained by the potentials generated in the dense region of an heavy-ion collision. On
the other hand, in-medium effects are essential for explaining the enhanced production
of antikaons. We show that the process N N ---, NNKg; is enhanced in the medium
while the absorption process K N ---, ATr is suppressed. We also find that the process
A N --+ N N K is essentially enhanced in dense matter and might be equally important as
the antikaon absorption. This may lead to equal numbers of antikaons and A's in the
dense zone of a relativistic heavy-ion collision.
The paper is organized as follows: first we introduce the RMF model and extend
it to include A's. In Section 3 we discuss two different approaches for the interaction
of kaons with nuclear matter, one based on the one-boson exchange model and the
other on chiral perturbation theory (ChPT). The parameters are fitted to the s-wave K N
scattering lengths. Results for the in-medium effects on kaon and antikaon production
are presented in Section 4. The last section is devoted to conclusions and an outlook.

2. The R M F model

It has been demonstrated by many studies that the RMF model gives a good description
of nuclear matter in bulk as well as of properties of nuclei [27,28]. We start from the
Lagrangian

1] = ~ N ( i ~ O t ~ -- m u ) l I t u + 101~0-0#0- -- U ( o')
_ l_t'~./~v('~. Ira2 vl'tv 1-Fgl~Vl~ 1. 2 r q ~ n
4~ ~ # u -~" 2 " ~ m - - ,,u. -- 4 u ~tx ~ ~- ~ m p l t l t #
--g~rNatt N~trNO" -- gtoNlff NytZtlt NVlz -- gpN~lt N~/tZ T a l t u R # , ( 1)

where the nucleons interact via an attractive scalar (0-) and repulsive vector ( V ~, R u)
meson fields. The term U(o-) stands for the scalar self-interaction

U(o-) ~m0o1-. 2 2 -- b
~ 0 -3 - ~ 4 °-4 (2)

introduced by Boguta and Bodmer [29] to get a correct compressibility of nuclear

matter (for another stabilized functional form see Ref. [30] ). The parameters of this
Lagrangian can be fixed to bulk properties [29] or to the properties of finite nuclei
[31,32]. A general discussion about the scalar self-interaction terms can be found in
[33]. Bodmer proposed an additional self-interaction term for the vector field [34]

£v~ = ¼d(GW') 2, (3)

which leads to a soft equation of state at high densities in agreement with Dirac-
Brtickner calculations [34,35]. Fits to the properties of nuclei with this new term are
quite successful [36]. In the following we take mostly the two parameter sets: N L - Z
328 J. Schaffner-Bielich et aL /Nuclear Physics A 625 (1997) 325-346

[31] which is the commonly used set NL1 with a better zero-point energy correction,
and the recent set TM1 [36] with vector self-interaction terms. The former one gives a
rather stiff equation of state while the latter one a rather soft one.
The implementation of A hyperons proceeds through the additional Lagrangian

~--"A = q-t A ( i'Y/z a l~ -- m A ) lltA -- g ~rA ~ A ~rtA Or -- g o~A ![t A TIZ ~ltA Vtz • (4)

The two new coupling constants can be fixed to hypernuclear data [37]. These data
suggest that the depth of the A-potential in saturated nuclear matter is about
u(N)
A = g o-AO'eq" + g~°AV~)q" ~' --30 MeV (5)

which already fixes one coupling constant of the A, say go-a [ 38,39]. The vector coupling
constant gwA is then given by SU(6) symmetry

g~,A = 2 g~oN (6)

(the quark model, see e.g. Ref. [40]). The SU(6) symmetry is also consistent with the
fact that the spin-orbit force is negligibly small and there is no experimental evidence
for a spin-orbit splitting for hypernuclear levels. It was first shown by Jennings [41]
that the contribution of the vector terms (gore and f, ou) to the spin-orbit term nearly
cancel that of the scalar term when taking into account SU(6) symmetry. This differs
from the case of the nuclear spin-orbit splitting where the vector and scalar contributions
are coherent to each other.
The in-medium energy of nucleons and hyperons is then given by

EN(p) = v/(mu + go-NO') 2 + p2 + gcouVo + gpNTORo,o, (7)

EA(p) = v/(mA + go-AO) 2 +p2 + goJAVo. (8)

It is important to note that the parameters here are connected to properties of nuclear
matter at normal density. Therefore, at this density the medium modifications of nucleons
and A's are known and should be taken into account when extrapolating to higher
densities. The in-medium energy of nucleons for zero momentum E u ( p = 0) is plotted
in Fig. 1 for various parametrizations. Note that the minimum for the potential is at a
little bit higher density than the saturation point because the Fermi momentum is set
to zero. For high temperature we expect that there is no Pauli blocking and hence all
nucleons will feel then a potential like the ones shown in Fig. 1.
As pointed out by Millener, Dover and Gal [42] three-body forces are also important
for explaining hypernuclear data. As these forces are repulsive, the hyperon potential
shows a non-linear behavior with density and changes sign at higher density. The
Schr6dinger equivalent A-nuclear potential is defined as
1
USEV = g~.4o" + goJaVo + ~ma ((g~rAO')2 -- (g~°AVo)2) (9)

is shown in Fig. 2 for different parameter sets and compared with the findings of the non-
relativistic approach [42]. The overall behavior is quite similar despite of the different
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 329

, ~ ' , , h '~ I ' ~ _

.,, ,~ .¢, ;

70.0 ........ N L , ,. , //,,4 -

c, - - e NL-SH ~: ? ~ "
50.0 PL-.o
÷ .... + P L - Z ~-,~ ':"if,
,x,"~::~
30.0
* - - - ~ TM1 +"E~f~,/,=,
> 10.0

-10.0

-30.0 X
-50.0

-70.0

-90.0

-110.0
0.00 0.10 0.20 0.30 0.40
Density [fm -~]

Fig. I. The relativistic potential for the nucleons for several parameter sets of the RMF model as a function
of density.

+
90.0
c- ~ NL-Z /
70.0 e ..... -o NL1 ~ d~'
~*--~NL-SH ' __J'~%
.
~-- -~ PL-40 S~4~
50.0 + .... + PL-Z .~ ~ ~,,z~,o~.

>
30.0
MDG _d
i
D 10.0

-10.0

-30.0

-50.0 , ,_ k , ,
0.00 0.10 0.20 0.30 0.40
Density [fm -a]

Fig. 2. The SchrOdinger equivalent potential of the A for several parameter sets of the RMF model as a
function of density. The curve labeled MDG is the non-relativistic potential fit to hypernuclear data of Dover,
Millener and Gal [42].
330 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346

parametrizations used. The non-linear behavior of the scalar field with density simulates
the repulsive three-body force of the non-relativistic approach. It also demonstrates that
it is crucial to make a difference between scalar and vector (baryon) density to get a
saturation point. The turning of the hyperon potential will be quite important for our
further discussion of the kaon production in the medium.

3. K a o n interactions

The case of the kaon is quite different from that of the A. There does not exist
any kaon-nuclear states similar to hypernuclei as the KN interaction is known to be
repulsive. Taking the (real) isospin averaged KN scattering length [43]
1 ,"-~ /=1 /--0
~lKN = ~tJa 0 + a 0 ) = - 0 . 2 5 5 fm (10)
and using the low-density theorem one gets a repulsive optical potential at normal
nuclear density of about

UopKN = _ 27r 1+ aXUpN ~ +29 MeV Pu/Po (11)

mK

compatible with kaon ( K +) scattering on nuclear targets [40]. Here we have taken the
saturation density to be P0 = 0.15 fm -3. Note that the K + optical potential depth is
repulsive and just opposite to the one of the A.
The isospin dependent potential in nuclear matter can be estimated from the isovector
1 / I=1 1=0
scattering length aiso = ~ , a 0 - - a 0 ) = - 0 . 0 5 5 fm and the low density theorem

uislCN= ___27r 1 + aisoPiso ~ +6 MeV Piso/PO, (12)

rnK

where piso is the isovector density of the system. For lead one can estimate piso
(2Z - A ) / A p N ,,~ --0.21pu, which gives about 1 MeV correction at normal nuclear
density.
For antikaons the absorption processes

k + N ---~Y + 7"r (Y = A, 27) (13)

gives a big imaginary part for the scattering amplitude. At first glance the experimental
situation seems to be controversial: The available K - N low energy scattering data
indicate a repulsive interaction while the K - atomic data demand an attractive potential.
The situation can be remedied by accounting for the existence of the A(1405) resonance
just below the K - p threshold. Recently an improved fit of K - atomic data was carried
out assuming a non-linear density dependence of the effective t-matrix [44]. It has been
shown that the real part of the antikaon optical potential can be as attractive as

Uogpu ~ - 2 0 0 + 20 MeV (14)

at normal nuclear matter density while being slightly repulsive at very low densities
in accordance with the K - p scattering length. The change of sign and the non-linear
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 331

density dependence results from the A(1405) resonance. Also another family of solutions
has been found with a moderate potential depth around - 5 0 MeV. Note that also the
standard linear extrapolation gives only values of about - 8 5 MeV [44]. These latter
two solutions are not getting repulsive at low densities, i.e. not fulfilling the low-density
theorem.
The K - N scattering data can be explained by vector meson exchange models where
the A(1405) is a quasi-bound state in the t-channel [45,46]. The coupled channel
analysis of Siegel and Weise [45] has been also applied for interaction terms coming
from chiral perturbation theory [47]. The coupled channel formalism automatically
generates the A(1405) and successfully describes the low energy K - p scattering data.
In the following we adopt the meson-exchange picture and the chiral approach for the
K N interaction on the mean-field level and fit the parameters to the K N scattering length.
The in-medium interaction of antikaons is then obtained by a G-parity transformation
which simply changes the sign of the vector potential term. This simple treatment does
not include the A(1405) resonance. But there exist some indications that this resonance
dissolves at higher density and the antikaon energy is shifted down. In Ref. [6] a
separable potential was applied for the K - p interaction at finite density. Indeed, it was
found that the mass of the A(1405) is shifted upwards and exceeds the K - p threshold
already at densities of about p ~ 0.4p0. This picture is supported recently within a
similar analysis using a chiral approach [7], where this resonances dissolves already
at very low densities p ~ 0.2p0 due to the Pauli blocking. If the A(1405) indeed
dissolves at low density, the use of the mean-field potentials may be justified at higher
densities. Hence, we simplify our calculation by neglecting the contributions coming
from the A(1405) in the medium and treat the problem on the tree-level using G-parity
transformation. Nevertheless, the results presented below for the N N --+ N N K f ( case
should be taken with some caution. More elaborate models are needed to draw final
conclusions about the in-medium properties of antikaons.

3.1. One-boson-exchange approach

In the kaon sector, we start from the following Lagrangian [ 13]:

£KN = D ~KD*- ~*K - m2KRK - g~KmKRKcr -- gaKmKRrK~ (15)

with the covariant derivative

D u = Ou + ig~xVu + igpxrR~. (16)

Here, for completeness we add isospin-dependent terms which couple to an isovector-

scalar field (3) and an isovector-vector field (R~,). Note that interaction terms of the
form

•KNA = --gKNA (,gIrTsAK + 71rysNR) (17)

do not contribute on the mean-field level since they are off-diagonal. We will come to
this point later in more detail.
332 J. Schaffner-Bielichet al./Nuclear PhysicsA 625 (1997) 325-346

The coupling constants to the vector mesons are chosen from the SU(3) relations
assuming ideal mixing

2go, K = 2gpK = g~',rp = 6.04, (18)

where g,r,rp is fixed by the p decay width. The scalar coupling constants can be fixed
to the s-wave K N scattering lengths [ 13]. The isospin averaged scattering length in the
tree approximation is given by [48]

1 I--0 , 3 I=1 = mK (go-Kgo-N 2g,oKg,o.____.__~NI'X = --0.255 fm,

glXN=4ao ~-'4aO 4rr(l+mK/mN)\ m2 m2 J
(19)

where only the isoscalar terms contribute. This can be used to fix g,~x for known
go,x = 3.02. The K N scattering lengths for a given Isospin I on the tree level are then
given by [48]

I=l InK (go'Kgo'u gsKgaN __ 2go, Kgo,N 2gpKgpN'~ (20)

a° =4rr(l+mx/mN)\ -~ + m~ m2 m2 J'

1=0 mK ( g o - x g o - N 3gaKgaN 2 g°)Kg°m + 6 gpxgpN'~ (2l)

a° =4rr(l+mK/mu) \ m2o- m~ m2 m 2 / I"

Recent experimental values are a/=1 = 0.31 fm and a~=° = - 0 . 0 9 fm [431. The im-
portance of the 6-meson exchange contribution can be seen by looking at the a~--°
scattering length. The vector terms largely cancel each other as go,xgo, N ~ 3gaKgpN.
Hence, without the contribution from the g-exchange one gets

I=O mK ( g,yxg~rN ~
a° = 4rr (1 + mK/mN) \ m~ i} "~ 0.4 fm (22)

in contradiction with experiment (here we used g,,u = 10, go)u = 13 as standard values
for the RMF model). Including the g-meson term and using gaN = 5.95 from the Bonn
model [49] we can fit both scattering lengths nicely for

g,~K ~ 1.9-2.3, g~x "~ 5.6-6.4 (23)

for the various nucleonic parameter sets used in the literature (see Table l ). Note that the
values of g,~K significantly deviate from the simple quark-model (simple quark counting
gives g,~r = grru/3 "~ 3.3). The coupling of the kaon to the &meson is quite strong.
Therefore, we expect some noticeable effects for isospin-asymmetric systems which we
will discuss later.
We define the antikaon optical potential in nuclear matter (p = 0) as

- = o)2k _ m 2 = g,~ro'mr -- 2go, rw~Vo -- (go,xV0) 2.

2mKUKt (24)

The fit based on K N scattering lengths gives f~Rpt = - - ( 8 5 - 1 0 0 ) MeV at normal nuclear
density for the parameter sets used. These values are lower than the ones quoted in our
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 333
Table 1
The coupling constants of the parameter sets used. The vector coupling constant for the A are taken from
SU(6) relations. The coupling constants of the kaons to the or- and 6-meson are fixed by the s-wave KN
scattering lengths. The vector coupling constants for the kaon are given by SU(3) relations. The parameters
for the scalar and vector self-interactionterms are not given,they can be found in the correspondingreferences

Set NL-Z NL-SH PL-Z PL-40 TM 1 TM2

Ref. [31l 132] [301 1301 [361 [361

g,rN l 0.0553 10.4440 10.4262 10.0514 10.0289 I 1.4694

gwN l 2.9086 12.9450 13.3415 12.8861 12.6139 14.6377
gOu 4.8494 4,3830 4.5592 4.8101 4.6322 4.6783
g,rl 6.23 6,47 6.41 6.20 6.21 7.15
g,o,~ 8.61 8.63 8.89 8.59 8.41 9.76
g,TK 1.85 2.05 2.20 2.27 1.93 2.27
go,K 3.02 3.02 3.02 3.02 3.02 3.02
gpK 3.02 3.02 3.02 3.02 3.02 3.02
g~K 6.37 5.59 5.89 6.31 5.87 5.94

previous work [ 13] since we use here the vacuum kaon mass mx instead of the reduced
mass IZKN. This is more consistent with the parametrization used in the study of kaonic
atoms [44[, and now our value is much closer to the standard fit based on the linear
extrapolation, UoXpt = --85 MeV. Note that the optical potential as defined in (24) is
always lower than the relativistic potential the antikaon feels at normal nuclear density
which is about

U~j. = w k - m x ~-~ - - ( 9 5 - 1 10) MeV. (25)

This definition corresponds to the sum of scalar and vector potentials as discussed in
[50]. Nevertheless, the scalar and also the vector potentials are much lower than the
ones deduced from simple quark model counting as used in [50]. The reason is that our
coupling ratios g~K/g~N ~ 1/5 and gcoK/gcoN ~ 0.23 (see Table 1) significantly deviate
from the simple quark model value of 1/3.
We have also studied the off-shell terms and found that they have only small influences
on the in-medium behavior of kaons (see Ref, [ 13] ). Note that off-shell terms are not
needed for describing the s-wave K N scattering lengths correctly. On the other hand,
they are essential in the chiral approach which we will discuss in the next section.
The equation of motion for kaons in the mean-field approximation in uniform matter
reads

{c~O ~ + m2K + g,rKmXO" + g~KmK~'060 + 2(go, xVo + goKToRo,o)ic ~u

--(g~oKVo + gpKT"oRo,o) 2 } K = 0. (26)

Note that terms quadratic in the vector fields appear in Eq. (26). The importance of the
isospin-dependent terms can be estimated from the equation of motions for the vector
fields in uniform matter

mZVo + dVo3 =g~oUPN, (27)

334 J. Schaffner-Bielich et al,/Nuclear Physics A 625 (1997) 325-346

m2Ro,o = gON (Pp -- Pn), (28)

where pp and p , are the densities of protons and neutrons, respectively. For lead one
gets pp - - P n ~ (2Z -A)/ApN ~ -0.21pN, and the isovector contribution is then about
m2
gpxRo,o ,~ 2 1 % g p N ~ ~ 8%,
go, K Vo g~oNmep

if one neglects the vector field self-interaction. Hence, the isovector contributions are
expected to be small for the densities considered here (p < 3p0).
Decomposing the kaon field into plane waves, one obtains the equation for the kaon
(antikaon) energy w and momentum k

--(0 2 ~- k e + m2x + g,~xmxo" + gaxmx~'o~o

+2(go, xV0 + gpxroRo.o)w - (g,oxVo + gpx~'oRo.o) 2 = O.

Now one obtains the following dispersion relation for kaons (upper sign) and antikaons
(lower sign) in uniform matter composed of nucleons only:

wx, • = V/mK + k 2 i (go, xVo + gpxroRo,o) , (29)

where the effective kaon mass is

m*x = ~/m2x + m x (g~xo" + g s x r o ~ ) (30)

(note that the effective mass is sometimes also defined as m~ = w K ( k = 0)). The scalar
field o- reduces the effective mass of the kaon in the medium, i.e. the scalar interaction
is attractive. The isovector-scalar field 6 shifts the effective mass if there is an isospin
asymmetry in the system. Note that for kaons as bosons the dependence on the scalar
potential is different from that for nucleons (fermions): it adds to the mass squared
for kaons but linear to the mass for baryons (see Eq. ( 7 ) ) . Moreover we point out
that the scalar potentials always follow the scalar density as required by the Lorentz
invariance. The scalar density is saturating in dense matter to ensure the existence of the
ground state minimum in symmetric nuclear matter. As shown in [5,11 ] these non-linear
effects are important already at a moderate density and responsible for a saturation of
the effective kaon mass at high density. We can say that there exists a minimum effective
kaon mass. Since in the RMF model the nucleon effective mass approaches zero at high
density (m~v ---, 0), the maximum scalar field is O ' m a x = - - m N / g ~ N . This leads to the
minimum value of the effective kaon mass

mK, min + mKgaxO'max = mK 1 mNg~rK ~ 390 MeV (31)

mKg~rN
for the parameters of Table 1.
Note that due to the covariant derivative coupling scheme (16) the vector term appears
linearly in the kaon energy, Eq. (29). The vector field is repulsive (attractive) for the
kaon (antikaon) and will dominate the behavior at very high density, where the kaon
 J. Schaffner-Bielichet al./Nuclear Physics A 625 (1997) 325-346 335

(antikaon) energy is then increasing (decreasing) like pl/3 because the vector field is
scales according to pl/3 if one takes into account the vector field self-interaction term
(see Eq. (27)).

3.2. Chiral approach

We follow the procedure outlined in [4] starting from the kaon-nucleon chiral La-
grangian (next-to-leading order in the chiral perturbation theory)

L2chiral i [ ( ~ ) ( ~ )]
KN = - - ~ K 3,gTygN R 0 g K + l?~'ygN R~" 0 ~' K

C -
+ ~-~x lqNB2K + -f~ N~'NKTK
~ ~)l _

+ - ~ N N (O~tKO~K) + -~xN~'N (Og[£~'O~'K) , (32)

Jx
where f ~ ~ 93 MeV is the kaon decay constant and £XN is the KN sigma term. The first
two terms correspond to the Tomozawa-Weinberg vector interaction, which is repulsive
(attractive) for kaons (antikaons). The next two terms are scalar interactions which will
shift the effective mass of the kaon and antikaon. The last two terms, which sometimes
are called off-shell terms, also modify the scalar interaction, Here one encounters striking
similarity with the RMF model as the KN interaction is governed by scalar and vector
terms (see Ref. [50] for a discussion about this point). In the original paper [4] the
authors chose XxN ~ 2m~ in accordance with the Bonn model [46]. More recently the
value £XN -----450 zL30 MeV is favored according to lattice gauge calculations [ 51 ]. The
constant C can be fixed by the Gell-Mann-Okubo mass formula to C = 33.5 MeV. The
KN scattering lengths corresponding to the Lagrangian (32) are
l
afN = 4~r/2 (1. -+-m~:lmN)I--rex+ X x N + C + (/) +/)')m~] ,

4~'i 2 (1 -+-mxlmN) [+-~Ku -- 3C + (Z) - 3/)') m~] . (33)

Now one can determine the constants b a n d / ) ' for a given XKN via the relations

/) ~ 0.33/rex - XKN/m 2, b ' ~ O.16/mx - C/rn~. (34)

Note that the off-shell terms involving the constants/) and/)~ are essential for a correct
description of the scattering lengths (see Ref. [4] for details). The equation of motion
for the kaon field in the mean-field approximation and in uniform matter reads

XKN C iso
3~3 '~ + m~ - ----T-ps - --~roPs
JK JK
D' .
-- r ^'s°o 3u 3i __j__t T iso~ "~
+ f 2psOuOg+~-K JK ot.'s u +5-~PNOtK+4f 4f20PNOt) K=0, (35)
336 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346

where p.i~° = ps,p - Ps,n is the scalar-isovector density and p~O = pp _ pn is the vector-
isovector density which are simply the difference of the corresponding densities of
protons and neutrons.
The magnitude of the isovector terms can be estimated from the isospin asymmetry
of nuclei. For a lead nucleus one has pp -- Pn "~ (2Z - A ) / A p N ~ -0.21pN, i.e.
about 21/3 = 7% correction due to the vector-isovector term of Eq. (35). This is in
accordance with our estimate for the one-boson exchange model in the previous section.
In the following we will neglect the isovector contributions. Plane wave decomposition
of the equation of motion yields

-,:o :2 + k 2 + H ( w , k; PN) = --092 -q'-k 2 + m~ - -~KN

-7~--p s
Jx
/3 3
- f-~x ps ( w 2 - k 2) - - ~ x w pN = 0 . (36)

Here we have introduced the kaon self-energy l l ( w , k; PN) which depends in general on
the kaon energy and momentum. The energy of a kaon/antikaon in the nuclear medium
is

rex, g = ( m ~ 2 + k 2) ( 1 + - - ~ x p s ) + (-~KpN) i 87KP N q- ~KPS ) ,

(37)

where m~. is the kaon effective mass

m~ = + ~ . (38)

We emphasize again that there exists a minimum effective kaon mass corresponding to
a maximum of the scalar density which is about Ps,max ~ 2p0 in the Walecka model
(see e.g. Ref. [30] ). This gives a minimum kaon effective mass of 350-400 MeV
depending on the kaon-nucleon sigma term. In the high density limit the kaon energy
is growing linear with density while for the antikaon the energy saturates at m,~ as the
vector contributions cancel each other.
The optical potential for antikaons at normal nuclear density
1
UoKpt-- 2ff/K H ( o ) k , k = 0; t90) ~ - 6 8 M e V (39)

is rather moderate while the relativistic potential is about - 7 5 MeV. This is in contrast to
the findings of Brown and Rho [50] who obtained a rather deep potential of - 2 0 0 MeV.
There are several reasons for this discrepancy: first, BR scaling is not taken into account
here (which gives an additional factor of 5/3 at p0, i.e. a potential of - 1 2 5 MeV);
second, Brown and Rho neglect the off-shell terms and do not take into account the K N
scattering lengths; third, they use an approximation that the scalar and vector densities
are equal, which is unrealistic at p ~ 3p0.
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 337

700.0 i i
m

600.0
K+ _

> 500.0

400.0

(D
c- 300.0
t- RMF " " " ~ " ' = ....
O
200.0 ............ ChPT (Y=270 MeV)
.... ChPT (I;;450 MeV)
- - - ChPT (D=O)
100.0
- -- coupled channel

0.0
0.0 1.0 2.0 3.0
Density P/Po

Fig. 3. The energy of kaons and antikaons in nuclear matter as function of density for the soft EOS ( parameter
set TMI ).

Recently, the dynamics of the A ( 1 4 0 5 ) has been studied in nuclear matter using a
coupled channel formalism [6,7]. The most important finding is that the effects coming
from the A ( 1 4 0 5 ) vanishes at rather low densities ( p < 0.25,00). The optical potential
for the antikaon is about - 1 0 0 M e V [6] and - 1 0 7 MeV [7] corresponding to a kaon
energy o f oJk = 380 M e V and w£ = 372 MeV at normal nuclear density, respectively.
These values are in accordance with the ones calculated in the mean-field approximation
in the previous sections within the relativistic mean-field ( R M F ) model and the chiral
perturbation theory ( C h P T ) .

4. Results

4.1. K a o n e n e r g y in m a t t e r

In the following we calculate the energy of kaons and antikaons in nuclear matter
using a soft (set TM1 of Table 1) and a hard (set N L - Z of Table 1) equation of state.
Fig. 3 shows the energy of kaons (upper curves) and antikaons (lower curves) with
the soft EOS for the R M F model (Eq. ( 2 9 ) ) , ChPT (Eq. ( 3 7 ) ) and the results of the
coupled channel analysis o f Waas et al. [7]. For the ChPT we consider three cases:
(i) a sigma term o f EKN = 2m~ as used in [4], (ii) a sigma term of EKN = 450 MeV
as derived from recent lattice data [51], (iii) vanishing off-shell terms (denoted as
/3 = 0) and a sigma term of EKN = 2m~ as used as input for the R B U U calculations
[ 17,23-25 ].
338 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346

700.0 , ,
NL-Z
600.0 K÷

> 500.0

400.0
>,

300.0
a)
c-
o - - RMF
200.0 ............ C h P T ( E = 2 7 0 M e V )
.... ChPT (X=450 MeV)
100.0 - - - ChPT (D=0)
coupled channel
0.0 ~ i ,
0.0 1.0 2.0 3.0
D e n s i t y P/Po

Fig. 4. The same as Fig. 3 for the hard EOS (parameter set NL-Z).

As seen in Fig. 3, all models show a quite similar behavior for the kaon energy at
low density except for the case /) = 0. This behavior follows from the low density
theorem and is a generic feature when the coupling constants are fixed to the K N
scattering lengths. Neglecting the off-shell terms, i.e. setting /) = 0, violates the low
density theorem. This gives a slower raise of the kaon energy with density and the
kaon energy nearly stays constant for a wide range of densities. Note that this latter
parametrization for the kaon energy is used in the dynamical calculations [23]. For
higher density the other curves also deviate from each other. The ChPT gives a higher
kaon energy, i.e. more repulsion than in the RMF model. The results of the coupled
channel calculation follow more closely the ones of ChPT. At p = 3p0 the kaon energy is
~ox = 585 MeV for the RMF model, (OK = 630 MeV for the coupled channel calculation
[7] and O)K = 640-670 MeV for ChPT, so they deviate for about 85 MeV from each
other.
The antikaon energy (lower curves) predicted by the different models is always
attractive, except for the low density region for the coupled channel calculations due to
the A(1405) resonance. The latter one gives the largest attraction with oJk = 217 MeV
at p = 3p0, followed by the RMF model with ~ok = 263 MeV and the ChPT with
o)k = 280-300 MeV. The curve for the case of/7) = 0 used in [25] closely follows the
one for the RMF model. All the curves for the antikaon energy are lying surprisingly
close together. Note that the ChPT predictions are rather insensitive to the choice of
Xmv but rather sensitive to the off-shell terms, especially for the kaon energy.
In Fig. 4 the results for the hard EOS are shown. Now the curves of the kaon energy
go very close together, even at higher densities. This is due to the fact that the vector
potential in the RMF model is now raising linear with density as in the ChPT in contrast
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 339

tO the soft EOS where it raises like pl/3 due to the vector self-interaction terms. The
energy of the kaon is now oJr = 630-670 MeV at p = 3p0. Without the off-shell terms,
the kaon energy significantly deviates from the other curves and stays rather constant
up to 1.5p0. Note that the overall difference between the hard EOS and the soft EOS
are quite moderate, especially when using ChPT, and only show up at higher density.
The predictions for the antikaon energy seem to split into two regimes: the results for
the ChPT give an antikaon energy of about wg = 300 MeV at p = 3p0 rather independent
of the off-shell term and the choice of ,SKN, while the RMF model and the coupled
channel analysis get around wR = 200 MeV at p = 3p0. The antikaon energy within the
RMF model is now much deeper due to the stronger vector potential compared to the
soft EOS. We want to point out again that Dirac-Brtickner calculations seem to favor
the soft EOS [34,35]. Nevertheless, we see that the differences in the kaon/antikaon
energy due to the EOS are well within the differences of the model predictions.

4.2. Threshold energy f o r kaon p r o d u c t i o n in matter

In the following we discuss the shift of the threshold energy of various processes for
heavy-ion collisions due to medium modifications.
Near the threshold, kaons are mainly produced via the process N N --~ N A K . The
threshold energy in vacuum is Q ( N A K ) = mA + m x -- mN ~ 671 MeV. In the medium,
the threshold is shifted to

Q(NAK) = E A ( p = O) + oJK(k = O) - E N ( p = 0), (40)

where we assume that the outgoing nucleon is not Pauli-blocked in the hot zone of the
collision. Hence, the subthreshold production of kaons is sensitive to three different in-
medium effects: the EOS ( E u ) , the A potential (EA) and the kaon energy (~ox). These
effects will partly cancel each other as the kaon feels a repulsive potential of 29 MeV
(Eq. (11)) while the A sees an attractive potential of - 3 0 MeV at P0 (Eq. (5)).
Therefore, subthreshold kaon production seems to probe mainly the EOS. If all nucleons
participating in the reaction N N --~ N A K feel the same attractive potential of about
- 6 0 MeV the threshold will be shifted upwards at normal nuclear density by this
amount. Therefore, the production of kaons will be reduced in the medium. This is
indeed be seen in Fig. 5 which shows the threshold energy Q ( N A K ) as a function
of density. The similar behavior of the different curves at low density is due to the
low-density theorem. At p = 3/9o the value of Q ( N A K ) reaches about 800 MeV for
the RMF model and about 860 MeV for ChPT which is quite insensitive to the value
of the K N sigma term. Without off-shell terms, the threshold energy is underestimated
in the medium, and we expect that the production rates for kaons calculated in [23,25]
are overestimated. Note, that the calculations ignoring in-medium effects [ 16,22] will
also give a too high production rate for kaons. As seen in Fig. 6, the behavior of the
threshold energy in medium is quite similar for the hard EOS. Again, the low density
limit more or less fixes the shape of the curves of the kaon energy independent of
the EOS used. The curves for the RMF model and ChPT are lying closely between
340 J. Schaffner-Bielich et aL /Nuclear Physics A 625 (1997) 325-346

,OOOl0

800.0 22-

> TM1
600.0 - - RMF
............ ChPT (E=270 MeV)
400.0 .... ChPT (E=450 MeV)
t~
>, -- - ChPT (D=0)
(2
200.0 " ~ ~
- :a-zz.~ :-'LL".L ......

0.0

-200.0 , r
0.0 1.0 2.0 3.0
Density P/Po

Fig. 5. The Q-values of the production processes of kaons and antikaons (the two upper bunches of curves)
and the annihilation processes of antikaons and A's (lower bunches of curves) versus the density for the soft
EOS.

r - - ~ r T '
+ -

1000.0

800.0

NL-Z
>~ 600.0 - - RMF
............ ChPT (~=270 MeV)
.... ChPT (E=450 MeV)
-~ 400.0
-- - - ChPT (D=0)
6
200.0

0.0
......... KN Arc
- - L J
-200.0
0.0 1.0 2.0 3.0
Density P/Po

Fig. 6. The same as Fig. 5 for the hard EOS.

8 0 0 - 8 3 0 M e V at p = 3po. Especially, the curve for the R M F m o d e l does not change

significantly b e t w e e n the hard E O S and the soft one. All curves seem to saturate for
the hard E O S and are lying within the uncertainties o f the different m o d e l s used for
calculating the kaon energy.
A d d i t i o n a l uncertainty is associated with the space-time d y n a m i c s o f a h e a v y - i o n
 J. Schaffner-Bielich et al./Nuclear Physics A 625 (1997) 325-346 341

collision. For instance, one can imagine that the kaons are predominantly produced on
the first-chance collisions of the nucleons initially located in the surface regions of the
colliding nuclei. Such nucleons feel a much weaker mean field than the nucleons in
the dense regions. If we assume also that the produced kaons, A's and nucleons (N r)
are already in the hot and dense zone, then the attractive potential acting on N r will
in fact reduce the threshold energy. Therefore, the kaon production might be enhanced
in this case. This remark applies also for the antikaon case discussed below. A definite
conclusion whether or not subthreshold kaon production probes the EOS can not be
drawn until the in-medium properties of the kaon as well as the heavy-ion reaction
mechanism are determined more precisely.
Antikaons are created in heavy-ion collisions first by the process N N --~ N N K K . The
vacuum threshold value of Q ( K f f ) = 2mx ~ 988 MeV is modified in the medium by
the sum of the kaon and antikaon energy Q ( K K ) = oJK(k = 0) + wg(k = 0). In this
sense, subthreshold antikaon production probes the in-medium properties of kaons and
antikaons. Of course, rescattering effects might be even more important as the reaction
TrY --~ K N due to its large cross section [52]. The in-medium effect on the shift of
mass and energy we are discussing here will mainly depend on the scalar potential the
kaon feels in the medium, as the vector potential cancels out approximately. The upper
curves in Fig. 5 show that indeed Q ( K K ) is reduced in the medium in all models
discussed here. ChPT predicts an in-medium reduction of about - 5 6 MeV at maximum
compared to the vacuum and then the curves go up again for higher density. The reason
is that the sum of the kaon and antikaon energy contains a term coming form the
Tomozawa-Weinberg interaction

fm b 2 3 2 b -r

(41)

which is repulsive and dominates at higher density. On the other hand, the RMF model
gives a reduction of about - 1 4 0 MeV at p = 3p0. The Q-value is steadily decreasing as
the sum of the kaon and antikaon energy,

w x ( k = O) + w R ( k = O) = 2m*x = 2v/m2x + mxg~Ko', (42)

depends on the attractive scalar potential only. The curve used in RBUU calculations
with a soft EOS [25] is lying even lower and hence, the production rates of antikaons
seem to be overestimated. With the hard EOS (Fig. 6) the situation does not change
significantly. The Q-value is now reduced by - 1 6 0 MeV at p = 3p0 for the RMF model.
The curves for the ChPT go up stronger at higher density compared to the soft EOS as
they are sensitive to the strength of the vector potential (i.e. to the behavior of the EOS
at high density) in contrast to the RMF model.
As an interesting fact, the Q-values for associated strangeness ( N N K K and N A K
final states) in N N collisions lie close together in the RMF model. Note that this does
not mean that the numbers of produced kaons and antikaons are the same inside the