The Physics of High

Energy Reactions
Alfredo Ferrari and Paola R. Sala

Istituto Nazionale di Fisica Nucleare, Sezione di

Milano
Via Celoria 16, 20133 Milano, Italy

Lecture given at the

Workshop on Nuclear Reaction Data and Nuclear

Reactors

Physics, Design and Safety

International Centre for Theoretical Physics

Miramare{Trieste, Italy, 15 April { 17 May 1996

Proceedings in press
 The Physics of High Energy Reactions
Alfredo Ferrari, and Paola R. Sala
INFN, Sezione di Milano, Via Celoria 16, 20133 Milano, Italy
Abstract
The basic aspects of particle nuclear interactions in the energy range
from a few tens of MeV up to several hundreds GeV, are presented, with
particular emphasis on the intermediate energy range (from 20 MeV to
 1-2 GeV). All topics concerning with hadron-nucleon interactions are
discussed mainly on a phenomenological basis, while nuclear eects are
presented in a more quantitative way. For both, the lecture is focussing
on the general aspects, rather than going into details. A particular eort
is made to illustrate the general features of the processes through the
discussion of models of common use in practical calculations.

1 Introduction
The importance of shower simulations in many elds of present day particle
physics has grown considerably during the last years, in parallel with the rapid
increase in available CPU power. However there has been no corresponding
development concerning the physical models used for such simulations, despite
the strong impact that simulation studies have on the analysis of running
experiments and on the design of future detectors.
A great importance has been attached instead to the development of so-
phisticated informatics packages to describe and visualize complex detectors
and to provide the generic user with powerful interactive tools for detector
modelling and data analysis. Perhaps the best example of such philosophy is
the geant code 1 , developed at CERN and widely used in the high energy
physics community.
Besides physics experiment, there is an increasing interest for applications
of accelerator beams. A new generation of intermediate energy proton and elec-
tron accelerators is under construction or planned in the near future, spanning
a variety of applications, ranging from energy production, waste transmuta-
tion, synchrotron radiation to radiotherapy. Such applications call for more
and more rened simulations tools, to be able to design and properly operate
these facilities.
A good knowledge of radiation transport is critical also for other activities
which apparently have very little in common with the new medical and indus-
trial accelerators. Radiation background in the large experiments which are
currently planned to be installed at the future LHC proton-proton collider will
1
be dominated by particle uxes which can only be estimated by simulation of
the whole hadronic cascade from several TeV down to thermal energies. Most
of the particle production will take place at energies below 1 GeV. A similar
situation arises in the assessment of the radiation dose aecting the crew of
commercial airplanes or of space stations. The interplanetary radiation envi-
ronment is indeed one of the major challenges of a possible manned mission
to Mars, and the optimization of the spacecraft in order to maximize shielding
while minimizing the weight is top priority. In general all aspects connected
with the dosimetry of medium and high energy particles are still waiting for a
fresh and more systematic treatment.
Detailed physical models are also required when designing and operating
experiments based on calorimetry. A detailed comprehension of active device
responses to subGeV hadrons will be a key issue for many of the future ex-
periments, both at accelerator beams, or in underground laboratories. The
recent question about the Kamiokande result on the decit of atmospheric -
neutrinos is a typical example. According to some calculations 2 , the \signal"
could be explained as due to 0 s produced by energetic neutrons generated
in the surrounding rock by high energy muons. A precise knowledge of the
full chain of muon photonuclear interactions, hadron showers, and nally pion
production in the intermediate energy range is required to settle the question.
Despite they heavily rely on MonteCarlo modelling when designing and
understanding their devices, and when analyzing experimental data for back-
grounds and kinematical cuts, high energy physicists tend to consider all the
complex phenomena of hadronic and electromagnetic showers occurring in their
experimental apparatus as well understood physics, without recognizing the
still large uncertainties connected with the physical description of nuclear in-
teractions. While QCD inspired models are very powerful in predicting and
describing the rare interesting phenomena searched for by high energy experi-
ments, eorts in describing the bulk of high energy interactions, which cannot
be understood in terms of perturbative QCD, are relatively rare.
Nuclear physicists are often working on a small subset of \interesting"
phenomena related to nuclear interactions, and usually do not like to spend
time on more general models which can be used for whichever application, par-
ticularly for technological ones. Very interesting physics researches are going
on in the description of nucleon induced interactions below the pion threshold
(see the reviews 3;4 ), however little or no work is done in the understanding of
interactions at higher energies, with possibly the exception of some interest-
ing developments in the pion sector 5;6;7 and of Quantum Molecular Dynamics
(QMD) models 8;9 when extended to the few GeV range. Anyway, new or
updated models are seldom formulated in such a way as to allow a general
2
application. As a consequence, most applications must rely on models, like the
glorious Bertini IntraNuclear Cascade code, which are 20 to 30 years old, and
which are no longer up-to-date with the present knowledge.
As soon as the energy of a primary hadron beam exceeds few tens of MeV,
inelastic interactions start to play a major role and secondaries have enough
energies to trigger further interactions, giving rise to a hadronic shower. When-
ever the beam energy is such that signicant pion production can occur (the
pion production threshold for nucleons interacting with stationary nucleons
is around 290 MeV), an increasing fraction of the energy is transferred from
the hadronic (HAD) to the electromagnetic (EM) sector due to production of
mesons (mainly 0 and ) which quickly decay into EM particles (e+ , e , and
). Hadron and electromagnetic showers are very complex phenomena, whose
description in terms of basic physical interactions requires a lot of knowledge.
There are two basic dierences between hadronic and EM showers. The rst is
that, while energetic hadronic showers are always giving rise to signicant EM
ones (and such EM component is more and more important with increasing
primary energy), EM showers develop independently without further hadronic
particle production, forgetting for a while the (small) probability of electro and
photonuclear interactions. The second dierence is that, while EM interactions
are in principle well understood (see however 10) and described by QED, the
same does not apply to hadronic nuclear interactions, where such a complete
theory does not exist and one has to resort to suitable models to have some
insight into the physics of the processes.
The development of hadron initiated showers is determined both by atomic
processes (dE/dx, multiple Coulomb Scattering etc), which take place very
frequently, and by the relatively rare nuclear interactions (both elastic and
nonelastic). EM showers are determined by the same atomic processes (dE/dx,
multiple Coulomb Scattering etc), plus other atomic processes (Bremsstrah-
lung, pair production, Compton scattering etc), which are specic of e and
photons, while nuclear interactions play a minor role, and whenever the inter-
est is not in the small amount of hadrons produced by EM particles, they can
be safely neglected.
Therefore this lecture a will restrict mainly to hadron induced nuclear inter-
actions with little or no discussion of those induced by real or virtual photons.
The description of hadron-nucleus interactions is presented in sections 5,6. In
particular section 5 describes hadronic inelastic interactions in the framework
of the (Generalized) IntraNuclear Cascade approach, whose basis are presented
in section 3. A discussion of practical implementations of (G)INC models, and
a
This lecture is partially based on a similar one given at the 1995 Frederic Joliot Summer
School in Reactor Physics 19 .
3
examples of performances are given in section 7.
Before going on with the proposed scheme, it is important to make the
reader aware of a few important warnings and comments. The present lecture
is based on the experience and feelings of the authors, who work since several
years in the eld of shower modelling, mainly for applications connected with
high energy accelerators and experiments. Their opinions are not necessarily
those of the majority of people working in the eld.
Whenever presented, plots of cross sections etc are to be intended as in-
dicative. Usually the data used for such plots are taken from the tabulations
used by the author code (fluka). Whenever experimental data are plotted,
the source is always indicated. Most of the examples have been computed again
with fluka , just for convenience: indications whether they can be considered
or not as typical examples are given every time.
No discussion is given about hadron interactions below, say, 20 MeV, both
because they are outside the aim of this lecture (and they are perfectly covered
by other authors at this school 11;12), and because for most practical applica-
tions they are essentially restricted to neutron interactions, for which a huge
and accurate body of experimental informations, sometimes supplemented by
sophisticated modelling does exist in ocial compilations like ENDF-B, JEF,
JENDL etc. It is always assumed that (nucleon) interactions below a few
tens of MeV can be accurately described either with the models developed for
this energy range 11;12;13;14;15, or making use of such (experimental) informa-
tions. The extension of these models 16;17;18, to higher energies close to the
pion threshold, is of little or no interest for many applications since correla-
tions among reaction products are forcibly lost due to the inclusive nature of
the computed distributions. Such correlations are of fundamental importance
whenever exclusive events are required, as for example in experiment analysis
and design. The most popular solution for preserving the internal correlations
of individual events are MonteCarlo based models. To the knowledge of the
authors, no exclusive (MonteCarlo) model based on the rened physical basis
of those models has been yet developed.
Furthermore, only single hadrons will be considered as possible projec-
tiles. The extension to very light nuclei (deuterons and alphas), is relatively
straightforward, at least for projectile energies much in excess of the binding of
these nuclei, but will not be explicitly described. Models for complex nucleus-
nucleus interactions will not be considered at all. Such interactions have indeed
little or no interest for technological applications (with the possible exception
of cancer therapy with ion beams), and for high energy experiment analysis
(unless a primary ion beam is used).
The description of hadron-nucleon interactions above few GeV will be
4
forcibly limited to few remarks, because of space reasons and of the relative
scarce interest of this energy range for accelerator technological applications.
Nuclear eects will be presented in a more quantitative way, without however
going into too many details. Anyway a comprehensive bibliography has been
included in order to allow the reader to access more informations if required.
Finally, the title itself is rather misleading. \High energies" for people
working in the nuclear eld usually mean several tens of MeV, or few hundreds
of MeV at most. They will probably feel rather exotic all parts dealing with
interactions at several tens of GeV. On the contrary high energy physicists will
be disappointed by the low energy tri es which ll a good fraction of this text.

2 Generalities
Before discussing specic aspects of hadronic interactions, there are a few com-
ments which help in focussing the problem. Assuming one is dealing with some
technological application of an accelerator beam (energy production, waste
transmutation, cancer therapy etc), which kind of informations does he need?
Without claiming that these are all the possible informations required, a rea-
sonable list could be the following:
 Elastic and nonelastic cross sections as a function of projectile, energy
and target
 Elastic scattering angular distributions
 Secondary particle yields as a function of energy and angle, following
nonelastic interactions
 Residual nuclei produced by nonelastic interactions
 Internal correlations among produced particles in the same event
Ideally, our tool, from now identied with some MonteCarlo program, should
be able to give us accurate predictions about all the above points, for any
desired spatial mesh. Usually the degree of accuracy required for the dierent
kind of informations is not the same. For example, residual nuclei are seldom
required to an accuracy better than one order of magnitude, while neutron
uence can well be requested to be within a factor 10% or better.
In order to better focus our needs in the description of nuclear interactions,
some general features of high energy showers are reported in the following
(see 19 for details).
5
 a) Energetic particles (often called shower particles) are concentrated
mainly around the primary beam axis, regardless of their identity. Their
dE=dx and the EM cascades associated with 0 constitute the \core" of
the energy proles.
b) Neutral particles (! neutrons, since these are the only neutral hadrons
with long enough lifetime) dominate at energies such that charged par-
ticle ranges become shorter than the interaction length. The energy de-
position associated with \low" energy neutron interactions, both recoils
and photons, constitute the long tails in the energy deposition proles.
c) Most of the interactions are due to particles (mainly neutrons) of mod-
erate energy, ! a good description of this energy range is mandatory.
d) On the contrary the longitudinal shower development is ruled by shower
particles, which carry a good fraction of the energy and have a longer
interaction length. Taking into account that any approximation or inac-
curacy in the rst interactions cannot be recovered with a better physics
in the following ones, this means that a good description of energetic
particle interactions is also mandatory.
e) Pions can be only produced by shower particle interactions, so they are
the real \tracer" of the high energy cascade. Neutrons and to less extent
protons are copiously produced also in the nal (evaporation) stages of
nuclear interactions down to projectile energies comparable with their
nuclear binding energy (see paragraph 5.8).
The term shower particles comes from the early experiments of high energy
physics, where nuclear emulsions were often used as recording media. Charged
particle tracks are therefore customarily classied in weakly ionizing, or shower
tracks, medium ionizing, or grey tracks, and heavily ionizing, or black tracks,
just according to their ionization rate. In practice, shower tracks correspond
to (charged) particles with  = vc  0:7, grey tracks to 0:25   < 0:7, and
black tracks to  < 0:25. Forgetting the original meaning, but just retaining
their velocity interval interpretation, such denitions are sometimes used also
for neutral particles.  = 0:7; 0:25 corresponds to  400 MeV and 30 MeV for
nucleons, and to  50 and 5 MeV for pions. Therefore black tracks are a good
indicator of evaporation products, while grey tracks are often associated with
nucleons emitted during intranuclear cascading. The bulk of shower tracks is
due to secondaries produced during projectile primary interactions.
6
3 (Generalized)IntraNuclear Cascade
Hadron-nucleus non-elastic interactions will be described mostly in the frame-
work of the IntraNuclear Cascade (INC) model. This model was developed at
the very beginning (the original ideas go back at the end of the 40's) of the
history of energetic nuclear interaction modelling, but it is still very valid and
in some energy range it is the only available choice. The model is intrinsically
a MonteCarlo model, well suited for numerical applications, while no closed
analytical expression can be derived without severe approximations. Therefore
INC models became more and more rened and widespread with the evolu-
tion of computer codes; currently available models can reach 100,000 lines of
program.
In the energy range going from the pion production threshold ( 290 MeV
for a free nucleon, down to 200 MeV for nucleons in nuclei because of the
Fermi motion) to high energies, INC models are practically the only available
tools to model hadron-nucleus interactions. At lower energies, a variety of
preequilibrium models can do a very good job, with physics foundations which
become surely more robust than those of INC ones as the energy is going down.
However one of the fundamental requirements for a model describing nu-
clear interactions to be applied in practical calculations, is speed. Indeed there
are two possible approaches to the speed problem: the former is to develop so-
phisticated models, not necessarily MonteCarlo ones, possibly very accurate
and maybe slow, and to produce comprehensive tabulations of energy-angle
spectra of all emitted particles for a ne mesh of energies of possible projec-
tiles. Such an approach, very similar to that adopted for neutron calculations
below 20 MeV, becomes soon very impractical. Indeed, assuming 20 points
per energy decade for the energy mesh (still  10% inaccuracy), 3 energy
decades (1-1000 MeV for example), 10 deg steps for angular distributions (!
18 points), tabulated data for at least p; n; + ;  ; 0 , and , it can be easily
computed that  5-10 Mbytes of dynamic memory will be required for each
(target) isotope present in the problem setup. Even assuming possible com-
pression schemes, such a (still rough) mesh cannot be actually used without
exploding memory requirements. Furthermore this kind of tabulations, while
possibly suitable for many calculations, cannot be used for problems where in-
ternal correlations among dierent particles emerging from the same collision
are important. Indeed problems like detector eciencies, resolutions, coinci-
dences background studies for rare events, etc, do often require the simulation
of proper correlations among many particles inside each event, and therefore
cannot be studied with approaches based on precomputed data, where such
correlations are forcibly lost.
7
 The second approach (used by the vast majority of practical applications
in the energy domain of interest), is to use the MonteCarlo method, simulating
at run time every interaction. One of reasons of the long dating success of INC
models is just their ability to model in reasonable time, almost whichever tar-
get nucleus with whichever projectile, with no or small need for external input
informations or preliminary calculations. The other great advantage is that all
correlation among the dierent quantities and particles are fully reproduced.
Of course, the speed of such a model is a key feature, since a large number
of interactions must be simulated within reasonable time. Actually, most INC
codes do their job fast enough that they do not represent any signicant limi-
tation to the CPU required, which is instead dominated by the transport time.
Until this situation will continue, the push for complex and huge pre-tabulated
libraries has no serious foundation, unless the physics is much better. Further-
more, while a code can be easily updated and/or corrected, every change in
the underlying physical models will require a complete recalculation of the
whole library, for all isotopes, particles and energies, making \de facto" such
an hypothetical library a quickly obsoleting tool.
3.1 Basic assumptions of IntraNuclear Cascade (INC) models
Classical INC codes 20;21;22;23 are based on a more or less accurate treatment
of hadron multiple collision processes in nuclei, the target being assumed to
be a cold Fermi gas of nucleons in their potential well 24;25. The hadron-
nucleon cross sections used in the calculations are free hadron{nucleon cross
sections. Usually, the only quantum mechanical concept incorporated is the
Pauli principle. Possible hadrons are often limited to pions and nucleons, pions
being also produce or absorbed via isobar (mainly
33) formation, decay, and
capture. The Fermi motion is taken into account when considering elementary
collisions, both for the purpose of computing the interaction cross section, and
to produce the nal state particles. The basic assumptions of INC models can
be summarized as follow:
1. Hadrons propagate like free particles in the nuclear medium, with in-
teraction probability per unit length given by free space cross sections,
properly averaged over the Fermi motion of the target nucleons, times
the local nuclear density.
2. The particle motion is formulated in a classical way. It can be subject
to an average nuclear mean potential, which must be added to the free
particle kinetic energy when tracking through the nucleus. The radial
and energy dependence of such eld are model and particle dependent.
8
 3. The eect of the nuclear mean eld on the particle motion can either be
null or can produce curved trajectories in a semiclassical approach, ac-
cording to energy and momentum conservation, depending on the model.
The curvature eects induced by the nuclear mean eld are usually re-
ferred to as refraction and re ection eects.
4. Interactions occur like in free space in the Center of Mass System of the
two colliding hadrons. Of course, because of the Fermi motion, the lab
frame will not coincide with the frame where the target nucleon is at
rest, but suitable Lorentz boosts (see Appendix A) have to be applied to
transform back the secondary particles in the lab frame.
5. Interactions occur in a completely incoherent and uncorrelated way. No
coherence or diractive eect is included. No multibody or cluster pro-
cess is included, with the possible exception of pion absorption (see next
sections).
6. Quantum eects are mainly limited to Pauli blocking (see section 5):
only few codes contain further quantum eects (see again section 5 for a
discussion).
7. Secondaries are treated exactly like primary particles, with the only dif-
ference that they start their trajectory already inside the nucleus.
An obvious requirement arising from the previous points is that the wavelength
associated to hadron motion must be much shorter than the hadron mean free
path inside the target nucleus, and also much shorter than the average distance
among two neighboring nucleons. That is:
h = 2 p
h 1
hN  (1)

h = 2 h   3 1=3 (2)

p 4
The nucleon density at the center of nuclei is typically   0:17 fm 3, therefore
the latter condition would require a particle momentum in excess of 1 GeV/c.
For a projectile nucleon, again at central density, the former condition is ful-
lled only starting from 200 MeV (see g. 1 for evaluating hN ). From these
back of the envelope estimates it would appear that INC models cannot work
at all, unless the projectile energy is above a few hundreds MeV. Furthermore,
the transport and reinteraction inside the nucleus of secondary particles, which
are at lower energies, cannot easily comply with the above requirements. Fortu-
nately the situation is not so bad as it could appear. Pauli blocking and other
9
eects contribute to increase particle mean free paths. In particular Pauli
blocking is more eective for low energy particles, in such a way partially
compensating the increase in nucleon-nucleon cross sections at low energies.
Furthermore, hadronic interactions are mostly surface eects (as can be easily
checked both from mean free paths and realizing that hadron-nucleus cross
sections scale with the target atomic mass approximately like A2=3). Hence
the nuclear reactions mostly occur at densities signicantly lower than the cen-
tral one, therefore partially mitigating the requirements on particle energies.
However it is clear that the physical foundations of INC are not very sound
for primary or secondary particle momenta below a few hundreds of MeV/c.
According to these considerations, the quality of results which can be obtained
by INC codes is somewhat surprising.
3.2 The Steps of an INC simulation
A typical INC code usually follows the following logic:
 Target nucleus description, typically realized through a few concentric
spheres of dierent density and Fermi energy
 Geometrical cross section, corresponding to the nuclear radius or to the
maximum possible impact parameter
 Impact parameter selection with a constant probability over the geomet-
rical cross section area. More than one selection can be required if the
particle crosses the nucleus without interacting
 Interaction point selection and projectile tracking through the nucleus,
according to Fermi motion averaged hadron-nucleon cross sections and
possibly to the nuclear mean eld, including the Coulomb eld
 Target nucleon selection according to hp ; hn and local Fermi energy
 Interaction simulation according to free hN interactions, local Fermi en-
ergy and Pauli blocking
 Secondary tracking into the nucleus, until interaction, escape, or energy
cut-o
 (Possible) preequilibrium stage, whenever all excited nucleons are below
a given energy threshold (typically a few tens of MeV). This stage is
included only in the most recent developments
10
  Evaporation stage whenever the preequilibrium stage is nished, or all
particles are below a given threshold (usually of the order of the binding
energy), and the system can be assumed to be equilibrated
 Final deexcitation stage when the excitation energy is below the thresh-
old for particle emission and it is spent through photon emission
A description of the physics involved in all stages can be found in the
next sections. It is important to remark that in principle INC codes are able
to compute reaction cross sections by themselves. Indeed, the reaction cross
section is given by the geometrical one times the probability of interaction.
The latter information can be easily derived from the ratio of the impact
parameter selections which gave rise to interactions and the total number of
selected impact parameters.
3.3 Advantages and Limitations of INC models
A few of the strong points of INC codes have been already touched in the
introduction to this section, however they are repeated here for completeness:
 No other model available for energies above the pion threshold (with the
exception of QMD models, which however include a substantially similar
approach for including hadron-nucleon nonelastic interactions)
 No other model for projectiles other than nucleons
 Easily available for on-line integration into transport codes
 Every target-projectile combination, without any extra information
 Particle-to-particle correlations preserved
 Equally valid on very light as well as on very heavy nuclei (with some
question mark on evaporation)
 Capability of computing reaction cross section where unknown
Before going into details about weak points of INC codes, it must be remarked
that many of the historical weaknesses have been mitigated or even completely
solved in some of the most recent developments 26;27;28. In the following list,
points agged with an asterisk refer to glorious but old models, typically the
Bertini model 21;29;30, but are partially or completely solved in state-of-the-art
codes.
  Low projectile energies (E  100 200 MeV) are badly described
11
   Binding energy: the commonly used assumption of a constant binding
energy makes the end points, shapes, and absolute intensities of emitted
particle spectra increasingly meaningless for energies below 100 MeV,
particularly for reactions like (p; xn) or (n; xp)
  Backward angle emission poorly described (see 31;32;33;34;35)
  Coulomb barrier eect: neglected when refraction/re ection are not
modelled or no Coulomb potential is included
  Cross section calculations: crazy when no potential eect is included
for E <100 MeV, and anyway at low energies. This point is particularly
critical for codes, like hetc 44, which make use of INC computed cross
sections for particle transport
  Secondary transport threshold: down to energies much lower than any
safe limit for the theory
 Quasielastic peaks above 100 MeV are usually too sharp when compared
with experimental data 36
 Coherent eects as well as direct transitions to discrete states are not
included
 Nuclear medium eects which can alter interaction properties are not
taken into account (see 37;38;39 for example)
 Multibody processes, like interaction on nucleon clusters, are not in-
cluded in INC approaches (with the possible exception of pion absorp-
tion)
 Composite particle emission (d, t, 3 He, ) cannot be easily accomodate
into INC codes, but for the evaporation stage.
It must be stressed that a relevant part of the improvements achieved by recent
developments is due to the inclusion of a preequilibrium stage, which prevents
the INC part from working in an unsafe energy range, and strongly improves
the modelling of reactions at the lowest energies. Since most of the interactions
for a real shower occur in this energy range, it is highly questionable the use
of codes which are still based on sloppy physics in this region.
12
4 Hadron-Nucleon interactions
The basic assumption in calculations of intranuclear cascades is that nuclear
reactions involving incident particles of high energy can be described in terms
of hadron-nucleon collisions within the nucleus. Models for such elementary
interactions are supposed to be available from phenomenology complemented
by theoretical analysis. Therefore a description, mainly phenomenological, of
hadron-nucleon interactions is essential in understanding the following more
complex treatment of nuclear interactions and is given in this section.
4.1 Preliminary
We remind here a few concepts and formulae of general utility.
We will deal with elastic and inelastic scatterings, thus the quantities to
be calculated , and sampled, will be mainly total or dierential cross sections.
Plenty of books exist on this subject. We re-write here well known formulae:
the dierential cross section as a function of scattering angle and centre of
mass energy can be written in terms of a scattering amplitude f(; s) :
d 2
d
() = jf()j (3)
The scattering amplitude is a complex quantity, and it is determined by the
interaction matrix. A very useful theorem is the optical theorem, that links
scattering amplitude and total (elastic scattering + reaction) cross section:
T = 4k Imf(0) (4)
The scattering amplitude can be separated in partial amplitudes corre-
sponding to the dierent values of the angular momentum l; in the case of
spinless particles one has:
X
f() = k1 (2l + 1)fl Pl (cos ) (5)
l
where the Pl are the Legendre Polynomials. More details about partial wave
expansions and the corresponding impact parameter representation of scatter-
ing amplitudes can be found in Appendix B.
The experimental observation of charge invariance of nuclear forces sug-
gested the introduction of the isotopic spin vector T. For instance, the proton
and the neutron can be described as the two states of a single entity, the nu-
cleon, having total isospin T = 21 , and possible projections Tz =  12 . In the
13
case of the nucleon, the charge is given by q = 12 + Tz . From charge inde-
pendence, it follows that nuclear interactions depend only on the total isospin
value, and not on its projection. Cross sections and other matrix element
can be separated in isospin channels; Coulomb forces must then be treated as
a correction. The isospin formalism is analogous to the spin formalism, and
the Pauli principle is taken into account by (anti)-symmetrization of the wave
function also with respect to isospin.
4.2 Nucleon-Nucleon elastic interactions
Below the pion production threshold, the only possible interaction between two
nucleons is just elastic scattering. The cross sections for p{p (or n{n) and n{p
total and elastic scattering are plotted in g. 1. There are two basic features
of these cross sections (which are the total ones up to roughly 300 MeV). The
former is the rapid increase of both cross sections with decreasing energy, while
the latter is the factor three at the lowest energies between the n{p and the p{p
cross sections, as expected on the basis of symmetry and isospin considerations.
The total cross section for the twonpisospin components are shown in g. 2,
according to T1 = pp and T0 = 2T Tpp . The same decomposition can be
shown to apply for the elastic and the reaction cross sections too.
The angular distributions of the p{p reaction is fairly isotropic up to 
300 MeV, while the n{p one starts to exhibit a signicant anisotropy at much
lower energies 47.
As soon as the energy exceeds a few hundred MeV, both reactions start
to develop strong forward peaks, while the backward \charge exchange" peak
of n{p becomes progressively smaller than the forward one, on the contrary of
what can be observed at lower energies. Such a behaviour is partially re ected
also in nuclear collisions, where nucleon-nucleon elastic interactions are the
dominant process for nucleon induced reactions below 1 GeV. Above few GeV
the behaviour of the bulk of the angular distribution is well described by a
standard diraction-like expression (see Appendix C):
d = B eBt (6)
0
dt
where t is the four-momentum transfer (see Appendix A).
4.3 Pion-Nucleon Elastic and Charge Exchange Interactions at Intermediate
Energies
Inelastic processes in Pion-Nucleon collisions remain negligible up to a pion
laboratory energy of about 250 MeV, although the kinematic threshold for
14
Figure 1: Total and elastic cross section for p{p and p{n scattering, together with experi-
mental data

pion production is at 170 MeV. Up to these energies the most important pro-
cesses are elastic and charge exchange scattering. Elastic, charge exchange
and inelastic + cross sections extracted from tabulations in 48 are shown in
g. 3. All show sharp peaks, distinctive of resonances. These peaks are most
conveniently examined in the Isotopic Spin formalism. The total isotopic spin
for the pion is T = 1, and the three charge states correspond to the three
values of Tz . Thus, in the pion-nucleon system two values of T are allowed :
T = 21 and T = 32 , and two independent scattering amplitudes, A 21 and A 32 ,
enter in the cross sections. From Clebsch-Gordan coecients one obtains the
scattering amplitudes A in the dierent charge states:


A + p ! + p = A 23 (7)
15
Figure 2: Total cross section for T = 1 and T = 0 isospin channels for nucleon-nucleon
scattering

 
A  p !  p = 31 2A 12 + A 23

p p 
A  p ! 0 n = 13 2A 32 2A 21

 
A 0 p ! 0 p = 13 2A 32 + A 21


A 0 p ! + n = A  p ! 0n




A  n !  n = A + p ! + p




A + n ! + n = A  p !  p




A + n ! 0 p = A  p ! 0n


16
 Figure 3: Cross sections for + p and + n reactions


A 0n ! 0 n = A 0 p ! 0 p




A 0 n !  p = A 0 p ! + n


Using these relations all dierential cross sections can be derived from

the
three measured ones:  (+ p ! + p),  ( p !  p) ,   p ! 0 n . Care
must be taken to include the eect of the mass dierences between charged
and neutral pions, and between neutron and proton 49. At low energies also
Coulomb eects should be taken into account.
From 7 and from the optical theorem 4 the total cross sections for the
T = 21 and T = 32 states can be isolated:
3
T2 = T + p


17
Figure 4: Total cross section for the T = 32 and T = 21 channels for pion-nucleon reactions

T2 = 32 T  p 12 T + p
1
(8)
These are plotted in g. 4 and show the dominant role of the
(1232) (also
called
33 or simply
) resonance in the T = 32 channel at  180 MeV lab.
energy, and the presence of several N  resonances in the T = 21 channel at
higher energies.
The same relations for the two possible isospin channels shown in eq. 8
for the total cross sections, can be demonstrated to hold also for the elastic +
charge exchange and non-elastic (reaction) cross sections.
3
r2 = r + p


18


r2 = 32 r  p 12 r + p
1
(9)

3
el2 +cx = el + p




el2 +cx = 32 el  p + cx  p
1
 1  + p
(10)
2 el
Detailed phase shift analysis have been performed for + p and  p scat-
terings up to centre of mass energies of 2.5 GeV 48 . For kinetic energies below
300 MeV the interaction is strongly dominated by s- and p-waves. The

resonance occurs in the l = 1, JP = 32 + , T= 23 channel (where J is the total an-
gular momentum, P the parity and T the isospin), at a centre-of-mass energy
of 1232 MeV and with a width of 120 MeV. The branching for decay into
 N is 100%. The cross sections for + on p and n for the elastic, charge
exchange, and inelastic (pion production) channels are shown in g. 3.
In the
region the angular distribution of pions after elastic or charge
exchange scattering closely follows the theoretical expectation:
P(cos )d
/ (1 + 3 cos2 )d

For energies larger than 1 GeV, the elastic scattering angular distribution be-
comes progressively more peaked, approaching the usual exponential behaviour
in the 4-momentum transfer t described in Appendix C, as for the nucleon-
nucleon case.
4.4 Pion production at Intermediate Energies
Pion production is the rst non-elastic channel (particle production) to be
open both in pion-nucleon and nucleon-nucleon interactions, obviously because
of the small pion mass. The reaction N1 +N2 ! N10 +N20 + has its threshold
around 290 MeV, and it starts to be important around 700, while the reaction
 + N ! 0 + 00 + N 0 opens at 170 MeV. The dominance of the
resonance
and of the N  resonances at higher energies, in the  N channel, suggest to
treat both reactions in the framework of the isobar model, that is to assume
that they all proceed through an intermediate state containing at least one
resonance. For instance, in the case of nucleon induced single pion production,
the following channels can be considered:
N1 + N2 ! N10 +
(1232) ! N10 + N20 + 
N1 + N2 ! N10 + N  (1440) ! N10 + N20 + 
19
  
where the N  (1440) is a T(JP )= 12 21 + baryon resonance with sizeable decay
channels into one or two pions plus one nucleon.
The situation for pion induced reactions is a bit dierent due to the possible
direct production of s-channel resonances (that is  N going into one resonance,
like the
), which are not possible in the N N system since no dibaryon
resonance has ever been discovered. Pion induced single pion production can
therefore be described by (among all possible channels):
+N ! N  (1440) ! (770) + N 0 ! 0 + 00 + N 0
+N !
(1600) ! 0 +
(1232) ! 0 + 00 + N 0
+N ! (770) + N 0 ! 0 + 00 + N 0
+N ! 0 +
(1232) ! 0 + 00 + N 0
where the rst two reactions are examples of s-channel direct resonance produc-
tion. (770) is a T(JP )=1 (1+ ) meson resonance with 100% decay branching
into .
The isobar model easily accommodates multiple pion production, for ex-
ample allowing the presence of more than one resonance in the intermediate
state 50 . Double pion production opens already at 600 MeV in nucleon-nucleon
reactions, and at about 350 MeV in pion-nucleon ones. In case of nucleon-
nucleon reactions, two pion production can be obtained considering for exam-
ple (many more channels are possible):
N1 + N2 !
1 (1232) +
2(1232) ! N10 + 1 + N20 + 2
N1 + N2 ! N10 + N  (1440) ! N10 + 1 +
(1232) ! N10 + 1 + N20 + 2
N1 + N2 ! N10 +
(1600) ! N10 + 1 + N  (1440) ! N10 + 1 + N20 + 2
and for pion-nucleon:
 + N !
(1600) ! 0 + N  (1440) ! 0 + 00 +
(1232)
! 0 + 00 + 000 + N 0
 + N ! !(782) + N 0 ! 0 + 00 + 000 + N 0
 + N ! (770) +
(1232) ! 0 + 000 + 00 + N 0
Summarizing, all reactions can be thought to proceed through channels
like:
h + N ! X ! x1 + ::: + xn (11)
h + N ! X + Y ! x1 + ::: + xn + y1 + :::: + ym (12)
20
where X and Y can be real resonances, or stable particles (n, p, ) directly.
Resonances which appear in the intermediate states can be treated as real
particles, that is, in a MonteCarlo code they can be transported and then
transformed into secondaries according to their lifetimes and decay branching
ratios.
These denitions can be obviously extended to elastic and charge exchange
scattering too, for example:
+ + n !
+ ! + + n
+ + n !
+ ! 0 + p
p+n ! p+n
Reactions described by eq. 11 are examples of s-channel direct resonance pro-
duction, and therefore they show up in the corresponding isospin cross section
as bumps around the centre-of-mass energy corresponding to the resonance
nominal mass (see g. 4). The reactions proceeding like in 12, due to the
presence of two particles in the intermediate state with the associated ex-
tra degree of freedom of their relative motion, do not exhibit a resonant be-
haviour,pbut rather a relatively fast increase starting from a centre-of-mass
energy, s  MX + MY , followed by a smooth behaviour. N N reactions are
all of this latter type and therefore no resonant structure can be nd in N N
cross sections (see g. 2).
For a practical use, all hadron-nucleon cross sections must be decomposed
into channels like those of eqs. 11 and 12, and the relative angular distribu-
tion of the two resonances of the latter case must be known, together with the
decay branching ratios and possible anisotropic decay matrix elements for all
considered resonances. Unfortunately the experimental information about ex-
clusive channels is far from being complete. However, resorting again to isospin
decomposition both for deriving quantities for charge states dierent from the
experimentally known ones, and for correctly isolating the contributions of the
dierent resonances, together with the constraints given by the measured in-
clusive cross sections for one-, two- and many pion production, a reasonable
description can be achieved. Example of analysis and parametrizations of nu-
cleon induced pion production in terms of isospin can be found in 51;52. An
example of a model aiming to describe all exclusive channels up to few GeV is
given in 50.
However, as soon as the incident hadron energy exceeds 3-4 GeV, the
description of nonelastic interactions via quasi two-body reactions with forma-
tion and decay of resonances starts to become dicult. The number of possible
channels (and consequently of resonances) to be considered grows very rapidly,
21
the relevant resonances are often not well established and sometimes their de-
cay channels are unknown. More fundamentally, an inspection of experimental
data clearly shows how the produced particles are no longer associated with
the projectile/target fragmentation region in the centre-of-mass frame, but
they start to preferentially populate the \central" region (see Appendix A),
in contrast with the naive picture of two excited objects with some transverse
momentum in the CMS, decaying into particles in the forward and backward
emisphere.
Fortunately as soon as the energy is beyond the resonance region (above a
few GeV), dierent models can be used. These models are also more appealing
since their theoretical basis are stronger, and their dependence on experimental
phenomenology weaker than for the isobar model.
4.5 Hadron-Nucleon High Energy Inelastic and Diractive Interactions

Figure 5: Invariant cross section spectra, as a function of Feynman x of negative (left),

and positive (right) pions emitted for + on protons at various momenta. Symbols are exp.
F

data, dashed histograms are the fluka result

Hadron nucleon interactions at high energy can be described by QCD based
models only at relatively large transverse momenta. Up to fairly high energies
(several tens of TeV in the lab, where minijet production can no longer be
neglected), the bulk of hadron-nucleon inelastic collisions is due to processes
with small transverse momentum. In this regime, QCD calculations become
impractical because of the large coupling constant, which prevents the use of
22
Figure 6: p spectra of + and  produced Figure 7: Feynman x spectra of positive
particles and  produced by 250 GeV/c +
T
by 16 GeV/c  incident on an hydrogen
F

target. Symbols are exp. data, the dashed incident on an hydrogen target. Symbols are
histogram is the fluka result exp. data, the dashed histogram is the fluka
result

perturbation methods, and because transverse momenta well in excess of 1 GeV

are required to prevent scattered quarks to undergo multiple interactions with
the spectator quarks.
The produced particle distributions essentially re ect the motion of the
internal constituents, valence and sea quarks. Therefore the longitudinal dis-
tribution of produced particles is intimately linked with the longitudinal mo-
tion of the projectile and target parton constituents, while the pT distribu-
tion has a roughly gaussian shape, typical of multiple collisions. The scale
of the pT distribution is directly linked to the uncertainty principle, taking
into account that hadrons are composite objects with a rms radius  0.8 fm,
pT  0h:8  200 300 MeV/c. The shape of the pT distribution is fairly inde-
pendent from the projectile energy up to at least several TeV in the lab, and
can be well described in terms of invariant cross section (see Appendix A) by:
3
inv = E dd3p~ / e BMT (13)
The longitudinal momentum distributions of produced particles approximately
scale with the incident energy. Such a scaling is often quoted as the hypothesis
of limiting fragmentation 53 (see Appendix A for details). A consequence of this
23
approximate scaling is that the producedpparticle multiplicity
p does not increase
linearly with the system invariant mass, s, (ps / Ebeam, for Ebeam  M),
but increases roughly logarithmically with s.
An example of the invariant cross section distribution, F(x) = 1 2EpCMs dxdF ,
of negative pions produced in the + + p !  + X reactions at 3.7, 7, and
18.5 GeV is presented in g. 5. A similar plot for + + p ! + + X is also
shown in the same gure. All the experimental data have been taken from the
compilation 54. Both gures show the characteristic concentration of particles
in the central region. The scaling violation occurring in that region (the scaling
hypothesis originally proposed by Feynman for high energy particle production
would predict that the invariant cross section when plotted as a function of the
Feynman x variable does not depend on the projectile energy in the central
region too) is also evident in the  plot. Diractive production for + at
large xF is apparent from that gure.
In order to describe these features, more or less phenomenological models
have to be used, even though motivated by physical ideas. Such models usually
work at the constituent level, that is with the constituent quarks (often called
valence quarks), and possibly also with the so called sea quarks which are
also present inside real hadrons. The momentum or energy distribution of
the constituents must be given (it is the equivalent for soft collisions of the
structure functions used for hard collisions), and proper hadronization models
must be provided.

The Dual Parton Model

One of the most successful models in this eld is the so called Dual Parton
Model (DPM), originally developed in Orsay in 1979 55. It provides the the-
oretical framework for the model developed for fluka in order to describe
hadron-nucleon interaction from several GeV onwards. Only few words will be
given here, since the main purpose of this lecture is to introduce the reader
to the aspects specic of nuclear interactions. Of course hadron-nucleon in-
teractions are an essential basis for nuclear interactions, however they are not
our main goal. The interested reader can nd all relevant details in a recent
review 56 .
In principle one would like to derive all features of \soft" interactions (low-
pT interactions) from the QCD Lagrangian, as it is done for hard processes.
Unfortunately the large value taken by the running coupling constant prevents
the use of perturbation theory.
Indeed, in QCD, the color eld acting among quarks is carried by the vector
bosons of the strong interaction, called gluons, which are \colored" themselves.
24
Figure 8: Leading two-chain diagram in Figure 9: Leading two-chain diagram in
DPM for p p scattering. The color (red, DPM for p p scattering. The color (red,
blue, and green) and quark combination blue, and green) and quark combination
shown in the gure is just one of the allowed shown in the gure is just one of the allowed
possibilities possibilities

Figure 10: Leading two-chain diagrams in DPM for + p scattering. The color (red, blue,
and green) and quark combination shown in each gure is just one of the allowed possibilities

Therefore the characteristic feature of gluons (and QCD) is their strong self-
interaction, on the contrary of what occurs for example in QED, where the
force carriers, the photons, are chargeless and therefore do not interact among
25
themselves. If we imagine that quarks are held together by color lines of force,
the gluon-gluon interaction will pull them together into the form of a tube
or a string. Since quarks are conned, the energy required to \stretch" such
a string is increasingly large until it suces to materialize a quark-antiquark
couple from the vacuum and the string breaks into two shorter ones, with still
quarks at both ends.
Therefore it is not unnatural that because of quark connement, theo-
ries based on interacting strings emerged as a powerful tool in understanding
QCD at the soft hadronic scale, that is in the non-perturbative regime. An
interacting string theory naturally leads to a topological expansion. At high
energies, such an expansion was developed already before the establishment of
QCD, that is the Reggeon-Pomeron calculus in the framework of perturbative
Reggeon Field Theory (for a review of Regge theory applied to high energy
scattering see 57 ). Regge theory makes use explicitly of the constraints of ana-
lyticity and duality, and the Dual Parton Model is built introducing partonic
ideas into a topological expansion which explicitly incorporates the constraints
of duality and unitarity.
In DPM a hadron is a low-lying excitation of an open string with quarks,
antiquarks or diquarks sitting at its ends. In particular mesons (all naturally
occuring mesons are explained as colorless combination of a quark and an
antiquark qq) are described as strings with their valence quark and antiquark at
the ends. (Anti)baryons (all baryons are colorless combinations of three quarks,
qqq) are treated like open strings with a (anti)quark and a (anti)diquark at
the ends, made up with their valence quarks.
At suciently high energies, when all Reggeon (IR) exchange amplitudes
become negligible (see Appendix C), the leading term in high energy scattering
corresponds to a Pomeron (IP) exchange (a closed string exchange), which has
a cylinder topology. When an unitarity cut (think about the optical theorem)
is applied to the cylindrical Pomeron two hadronic chains are left as the sources
of particle production. While the partons (quarks or diquarks) out of which
chains are stretched carry a net color, the chains themselves are built in such a
way to carry no net color, or to be more exact to constitute color singlets like
all naturally occuring hadrons. In practice, as a consequence of color exchange
in the interaction, each colliding hadron splits into two colored system, one
carrying color charge c and the other c. These two systems carry together the
whole momentum of the hadron. The system with color charge c (c) of one
hadron combines with the system of complementary color of the other hadron,
in such a way to form two color neutral chains. These chains appear as two
back-to-back jets in their own centre-of-mass systems.
The exact way of building up these chains depends on the nature of
26
the projectile-target combination (baryon-baryon, meson-baryon, antibaryon-
baryon, meson-meson): examples are shown in gs. 8,9 and 10, and explained
in the following.
In the case of baryon-baryon scattering, indicating with qpv the valence
quarks of the projectile, and with qtv those of the target, and assuming that
the quarks sitting at one end of the baryon strings carry momentum fraction
xvp and xvt respectively, the resulting chains are qtv qpv qpv and qpv qtv qtv , as
shown in g. 8.
For meson-baryon scattering, indicating with qpv the valence quarks of the
projectile, and with qtv those of the target, and indicating with xvp and xvt
respectively the energy/momentum fractions carried by the two quarks sitting
at the chain ends, the resulting chains are qtv qpv and qpv qtv qtv , as shown in
g. 10.
For antibaryon-baryon scattering, adopting the same notation, the chains
are qtv qpv and qpv qpv qtv qtv , as shown in g. 9.
For all cases, the energy, and momentum in the centre-of-mass system of
the collision, as well as the invariant mass squared of the two chains, can be
obtained from:
ps

Ech1  2 (1 xvp + xvt )
ps
 
Ech (1 xvt + xvp )
2 2
ps

pch1  2 (1 xvp xvt ) = pch2 (14)
sch1  s(1 xp )xt v v
sch2  s(1 xvt )xvp
The single Pomeron exchange diagram is the dominant contribution, how-
ever higher order contributions with multi-Pomeron exchanges become impor-
tant at energies in excess of 1 TeV in the laboratory. They correspond to more
complicated topologies, and DPM provides a way for evaluating the weight of
each, keeping into account the unitarity constraint. When cut, every extra
Pomeron exchanged gives rise to two extra chains which are built using two
qq couples excited from the projectile and target hadron sea respectively. The
inclusion of these higher order diagrams is usually referred to as multiple soft
collisions.
Two more ingredients are required to completely settle the problem. The
former is the momentum distribution for the x variables of valence and sea
quarks. Despite the exact form of the momentum distribution function,
27
P(x1; ::; xn), is not known, general considerations based on Regge arguments
allow to predict the asymptotic behaviour of this distribution whenever each
of its arguments goes to zero. The behaviour turns out to be singular in all
cases, but for the diquarks. A reasonable assumption, always made in practice,
is therefore to approximate the true unknown distribution function with the
product of all these asymptotic behaviours, treating all the rest as a normal-
ization constant.
Under this approximation, indicating with xsea sea, the ener-
q i , and xqsea
i
gy/momentum fractions carried by the sea quarks and with Xi the sum of
xsea sea
q i and xq i , the total momentum distribution function for a(n) (anti)baryon
in the case of nIP -cut Pomerons can be written as:
1 3 Y1
nIP 1 sea 1
P(x)dx  Cb xq 2 xqq2 (Xisea ) 1 (xsea
q i ) 2 (xq i ) 2
i
nX
IP 1

(1 xq xqq Xisea ) dx (15)
i
Y1
nIP

x  xq
xqq xsea sea
q i
xq i
i
where Cb is a normalization factor. The momentum distribution function for
a meson reads:
1 1 Y1
nIP 1 sea 1
P(x)dx  Cm xq 2 xq 2 (Xisea ) 1 (xsea
q i ) 2 (xq i ) 2
i
nX
IP 1

(1 xq xq Xisea ) dx (16)
i
nY
IP 1

x  xq
xq xsea sea
q i
xq i
i
The latter ingredient is of course a hadronization model, which must take
care of transforming each chain into a sequence of physical hadrons, stable
ones or resonances. The basic assumption is that of chain universality, which
assumes that once the chain ends and the invariant mass of the chain are given,
the hadronization properties are the same regardless of the physical process
which originated the chain. Therefore the knowledge coming from hard pro-
cesses and e+ e collisions about hadronization can be used to fulll this task.
28
There are many more or less phenomenological models which have been devel-
oped to describe hadronization (examples can be found in 58;59). In principle
hadronization properties too can be derived from Regge formalism 60;61.
Summarizing, DPM provides recipes for performing the following tasks:
 determining the number of cut Pomerons, and therefore the number of
chains contributing to the reaction
 forming the chains using the valence and possibly sea quarks of the two
colliding hadrons
 determining the energy and momentum carried by each chain, according
to the momentum distribution functions of the two colliding hadrons
 hadronizing each chain producing the nal hadrons, stable ones or reso-
nances
The last step is not exactly a part of DPM, but rather DPM is factorized in
such a way that it can be accomplished using whichever hadronization scheme.
Indeed DPM is intrinsically factorized, and this strongly constrains the model.
In principle there is little or no freedom in each individual step, therefore
strenghtening the predictive power of the model. Actually, in the energy range
of interest for experiments, that is with showers in the apparatus with ener-
gies from few GeV up to several hundreds GeV, threshold eects are still very
important. While DPM is assumed to be valid in the asymptotic regime, and
treats massless partons at energies large enough to neglect hadron masses (the
limit in which eq. 14 is derived), most practical implementations deal with
chains with invariant masses so small that only very few particles can be pro-
duced out of the chain itself. In this regime the treatment of nite mass eects
both when building chains, with the possibility of cutting-o secondary chains
because of insucient energy, and during hadronization, is very important. As
an example, the centre-of-mass energy for 450 GeV protons on protons (the
maximum laboratoryp energy obtainable with the CERN accelerator complex
on xed targets) is s  30 GeV . For p xvq  0:2 which is a typical value, the
two chains have invariant masses of s1=2  12 GeV each. Therefore, already
6 GeV in the global centre-of-mass frame are \wasted" into chain motion, and
each chain consists of two back-to-back jets with  6 GeV each, which cannot
produce much more than 2-3 particles/resonances, taking into account that
typical resonance masses are of the order of 1 GeV and still some motion be-
tween resonances created out of the same jet must be provided. Moreover, a
centre-of-mass energy of 30 GeV is not obviously negligible with respect to the
two proton masses, of  1 GeV each. Of course such nite mass and threshold
29
eects are even more important at lower energies, where however we still would
like to use DPM due to the lack of valid alternatives. Indeed with a proper
treatment of these eects, models based on DPM can be successfully used at
lower energies, as shown in the examples reported in this section.
It is possible to extend DPM to hadron-nucleus collisions too 55;56, mak-
ing use of the Glauber-Gribov approach (see the next section). Furthermore
DPM provides a theoretical framework for describing hadron diractive scat-
tering both in hadron-hadron and hadron-nucleus collisions. The approach to
diractive scattering will not be touched here, however general informations on
diraction can be found in 62 and details as well as practical implementations
in the DPM framework in 56;63;64.
At very high energies, those of interest for cosmic ray studies
p (10{105 TeV
in the lab), or for collisions at present and future colliders ( s=0.2{20 TeV),
hard processes cannot be longer ignored. They can be included in DPM
through proper unitarization schemes which consistently treat soft and hard
processes together. Again we are not going to give any detail on this sub-
ject, but the interested reader can nd more informations as well as practical
implementations and results in 56;65;66;67.
It must be stressed that DPM is not the only model which has been de-
veloped an successfully compared with experimental data for high energy in-
teractions. Other models are available, but most of them share an approach
based on string formation and decay. For example, the Quark Gluon String
Model 68 has been developed more or less in parallel with DPM. This model
shares most of the basic features of DPM, while diering for some details in
the way chains are created and in the momentum distribution functions.
No eort will be made to illustrate the many successes of DPM in predict-
ing experimental observables. The quoted references include a vast amount of
material showing the capabilities of the model when compared with experimen-
tal data. Only examples computed with the DPM based model implemented
in the fluka code have been included.
One of the few models available for high energy interactions 69 , which is
used in many avors in hetc88 45, lahet 46 , hermes 43 , and geant, has been
developed in the DPM framework, originally for the fluka code. fluka does
now include a much more advanced version, which however is still based on the
same physical basis. It includes the leading order contribution (one pomeron
exchange) as well as diraction, and for the hadronization it makes use of an
advanced version of the model described in 59 , with particular emphasis on a
correct description down to the lowest energies (a few GeV in the lab).
As an example of pT distributions, the experimental data (and the MC
simulation) of the pT spectra of pions produced by 16 GeV/c  on Hydrogen
30
are shown in g. 6 (experimental data from 54). The longitudinal distribution of
positive particles and negative pions produced by 250 GeV/c + on Hydrogen
are shown in g. 7 (experimental data from 70). Again the diractive peak in
the positive particle distribution is clearly visible in g. 7.

5 Nonelastic Nuclear Interactions

Hadronic nuclear interactions are by far the most complex topic of this lecture.
A thorough discussion of their characteristics and of the relevant physics is out-
side the aims of this lecture (and perhaps of the knowledge of the lecturer).
The approach followed is just to describe the basic ideas underlying the most
successful models used in practice to describe such interactions. In particular
the intranuclear cascade model will be described, mainly because it is still the
only one able to cover the full energy range of interest for hadron interactions.
However plain INC models like those developed in the seventies become unre-
liable both at the lower (< 100-200 MeV) and higher (> 2-3 GeV) ends of the
energy scale, and show limitations (see 3.3 and the examples reported in 7)
also when used in the proper energy range.
Once suitable models for describing hadron-nucleon interactions are avail-
able, the high energy regime can be properly handled provided the spacetime
characteristics of high energy interactions and multiple primary collisions ac-
cording to the Glauber approach (see after) are taken into account. Models
which includes these features are often referred to as Formation Zone IntraNu-
clear Cascade or Glauber Cascade approaches (see for example 71;72).
However important changes to the original INC approach must be done
also at the lowest energies, and are mainly related with quantum nuclear eects
and multibody interactions, besides the introduction of a preequilibrium stage.
IntraNuclear Cascade approaches which make use both of the high energy
and low energy extensions can be called (Generalized) IntraNuclear Cascade
models, and will be referred as such in the next sections.
Most of the description of the next paragraphs will follow the layout of the
(G)INC model implemented in the fluka code, whose low-mediumenergy part
(up to few GeV) is called peanut (for PreEquilibrium Approach to NUclear
Thermalization). A few words about the latter model are given in the following
paragraph in order both, to allow the reader to understand how the model is
built, and to present a practical example of INC model.
After this brief introduction, a description of the main physical ingredients
will be presented; the physical grounds will be discussed in general, giving some
details about the actual implementation in peanut .
31
5.1 peanut : Generalities
Presently, peanut handles interactions of nucleons, pions, kaons, rays, and
stopping negative muons with nuclei from about 2.5 GeV down to reaction
threshold (or 20 MeV for neutrons). As concerns photonuclear reactions which
will not be discussed in the next paragraphs, details can be found in 73.
The reaction mechanism is modelled in peanut by explicit intranuclear
cascade smoothly joined to statistical (exciton) preequilibrium emission 4;74.
The full INC treatment stops, and the statistical treatment starts, when all
the secondary nucleons have energies smaller than 50 MeV (as calculated
in the continuum). To ensure continuity, however, secondary nucleons with
10<E<50 MeV are transported by the INC algorithm till they either escape or
reinteract in the nucleus. In the latter case no explicit interaction is performed:
only an average Pauli rejection factor 75 is applied and the exciton number is
increased, leaving to the preequilibrium stage the further development of the
conguration.
In both stages, INC and exciton, the nucleus is modelled as a sphere with
density given by a symmetrized Woods-Saxon 76 shape for A>16,
sinh(R0=a)
(r) = 0 cosh(r=a) (17)
+ cosh(R0=a)
 
 0
1 + exp r aR0
and by a harmonic oscillator shell model for light isotopes (see 77 ). It is radially
divided in 16 zones of constant density, and its boundary is set at the radius
(Rnuc) where the density is one hundredth of the central one. Six radial zones
are added to allow a suitable description of the nuclear potential ouside the
nucleus, and nally 10 extra radial bins are there for charged particles to
describe the long range eect of the Coulomb potential. Primary and secondary
particles are transported inside these radial zones according to their nuclear
mean eld and to the Coulomb potential.
Proton and neutron densities are generally dierent, according again to
shell model ones for A<16, and to the droplet model 78;79 for heavier nuclei.
Binding Energies (Ben ) are obtained from mass tables, depending on parti-
cle type and on the actual composite nucleus, which may dier from the initial
one in the case of multiple particle emission. Relativistic kinematics is applied,
with accurate conservation of energy and momentum, and with inclusion of the
recoil energy and momentum of the residual nucleus.
32
5.2 Nucleon Fermi Motion
Some properties of nuclear matter can be explained in a very simple way by
the Fermi gas model (for descriptions of this and other nuclear models see
for instance 80;81;82 ). Nucleons are described as non-interacting fermions con-
strained in a cubic box of linear dimension a. The connement is somehow
equivalent to the introduction of an attractive interaction among nucleons that
keeps them inside the nucleus. The nucleon wave functions are those of a free-
particle with the condition (x; y; z) = 0 at the nuclear boundary, thus with
constraints on the three components of the momentum:
Kx a = nx  Ky a = ny  Kz a = nz  (18)
where the n's are positive integers. Each set of integers denes a solution of
the Schrodinger equation with energy given by:
h2 jK j2
E(nx ; ny ; nz ) = 2M (19)
jK j2 = (Kx2 + Ky2 + Kz2 ) (20)
Nucleons must obey the Pauli principle, thus for each set of n's, that is for
a given momentum state, there can be at most four nucleons, namely two
protons and two neutrons, each pair with opposite spin. The ground state of
the nucleus will be that with all the lowest-lying states occupied. The number
of possible states can be easily calculated in momentum space: from the above
quantization
it follows that each solution occupies a box in momentum space
given by a 3 ; thus the number of states with momentum comprised between
K and K + dK is given by the volume of the spherical shell of radius K and
thickness dK divided by the volume occupied by each single solution, and
divided again by eight since we consider only positive values of nx; ny ; nz :
dN = j2Kj2
dK;
= a3
2 (21)
re-expressed as a function of energy :
p
dN =
m 3 2mEdE (22)
2 h
2
Integrating eq. 21 up to a given momentum jKF j, one obtains the total number
of states with momentum smaller than jKF j:
n(KF ) = 4 KF3 (23)
3 8(=a)3
33
These n levels must be occupied by N neutrons and Z protons. If we take
N=Z=A/2, and place 4 nucleons on each level, we get:
2
K 3
A = 3 (24)
2 F
it follows that the momentum of the highest occupied state depends only on
the nuclear density  = A=
:
 = 32 2 KF3 (25)
From the observed density of nuclei,  = 1:7  1038 nucleons/cm3, one obtains
KF = 1:36 fm 1 EF = 38 MeV (26)
These are called the Fermi momentum and Fermi Energy. In nuclei with
N 6= Z, two dierent values of the Fermi energy can be dened:
n = N  = (KFn )3
A 32
p 3
p = ZA  = (K3F2) (27)
The so dened Fermi energies are obviously kinetic energies, that is energies
counted from the bottom of a potential well that in this model must be input
from outside. The total depth of the potential should be such to place the so-
called Fermi level, i.e. the last occupied level, at a total energy corresponding
to the separation energy for the nucleus considered. This gives an average
potential depth of about 38+8=46 MeV. The Fermi kinetic energy distribution
can be used with success in many applications. The model is also able to predict
the order of magnitude of some empirical nuclear parameters, such as surface
energy and asymmetry energy.
The model can be extended to account for realistic nuclear density distri-
butions, which are not uniform, but present a zone of gradual decrease called
nuclear skin. The Fermi energy of eq. 25 can be made radius-dependent in a
straightforward way, through the so called local density approximation:
(r) = 32 2 KF3 (r) (28)
Since the density distribution is experimentally well determined, this last for-
mulation of the Fermi energy could provide the radial behaviour and value of
the nuclear potential. This is of widespread use in intranuclear cascade codes.
34
Moreover, the same assumption, i.e. nuclear matter as a gas of noninteracting
nucleons, linked with thermodynamics, leads to the denition of nuclear tem-
perature and to a general expression for the density of nuclear excited states as
a function of excitation energy. An other important quantity that can be cal-
culated in this model is the so-called correlation function, that must be taken
into account in intranuclear cascade codes, expecially when dealing with multi-
body processes. Due to the antisymmetrization of the fermion's wave function,
given a nucleon in a position ~r in a nucleus with density 0 , the probability of
nding another like nucleon in a position ~r 0 is decreased for small values of
the distance r = j~r ~r 0 j by a factor
  2
g(x) = 1 21 x32 sinx x cos x (29)

where x = KF
r, and the factor 21 in front of the parenthesis accounts for the
two possible spin orientations.
5.3 Nuclear Potential
The approximation of independent motion of nucleons inside a common single-
particle potential is possible due to the fact that the average inter-nucleon
distance is large with respect to the range of the nucleon-nucleon interaction.
This is partly ensured by the Pauli principle, and partly by the presence of a
repulsive core in the nucleon-nucleon interaction. The single particle potential
can be thought as the mean eld resulting from all the two-body interactions
among nucleons. Many possible shapes of this potential have been proposed,
either on the basis of self-consistency (Hartree-Fock like) or of computational
usefulness linked with agreement with experimental quantities (harmonic os-
cillator, Nilsson anharmonic oscillator, Woods-Saxon, all with spin-orbit cou-
pling). These potential are able to reproduce the observed level spacings and
ordering, and other nuclear quantities. They are, however, less useful in scat-
tering or reaction processes. In these application, the so called Optical Po-
tential is introduced. It is a complex potential that enters in the Schrodinger
equation for the projectile. The real part describes the scattering, the com-
plex part describes the absorption: to illustrate this, let us suppose that the
optical potential V (~r) = [U(~r) + iW(~r)] has the shape of a square well. The
wave functions can then be described by plane waves, and the solution of the
Schrodinger equation
 h 2 
2M r 2+V E =0
K (30)
35
inside the range of the potential will be
K~ = AeiK~
~r hi Et
2K 2
E = h2M U iW (31)
since E is real and equal to the energy before the collision, K~ must be a
complex vector, thus if one looks at the dependence of the particle probability
density j j2 as a function of ~r, and assuming for simplicity K k z, one obtains
j j2 / e 2ImK
z (32)
That means that the particle can be absorbed in the nucleus with a mean free
path
1
 = 2ImK
From eq. 31 one has ImK = 2M W
h2 2ReK , and, assuming that the absorptive po-
 1
tential W is small with respect to E + U, ReK  h2 (E + U) 2 , from which
2M
 M  12
 1  2W (33)
h 2(E + U)
this becomes
 1  2Wvh (34)
where v is the particle velocity. The nucleon mean free path in a nucleus
should be given, at least in rst approximation, by the cross section on a
single nucleon multiplied by the nuclear density:  1 = 
NN , thus a link
between the absorptive nuclear potential and the nucleon nucleon cross section
is immediately derived:
W = h2 vNN (35)
We shall see in sec. 5.6 that many eects invalidate this simple relation, which
however remains a rst order estimate.
The optical potential can be derived from the convolution of the interac-
tions between the projectile and all the nucleons in the nucleus, or can be tted
to the scattering data with preliminary assumptions about its radial depen-
dence. The real part of the optical potential is also sometimes used to describe
bound states. As life is always complicated, both the optical potential and the
self consistent nuclear potential turn out to be energy dependent, but we stop
here.
36
5.4 Pion Nuclear Potential

Figure 11: Nuclear potential for  in a Pb nucleus as a function of radius. Dierent curves
refer to dierent pion kinetic energies (MeV). Coulomb potential is included
For pions, a standard nuclear potential exists 83. It must be written in rel-
ativistic form, because the pion mass is small (about 139 MeV) and can be
comparable with its kinetic energy. In coordinate space, and in the frame in
which the nucleus is at rest, this is written as (the upper and lower signs refer
to + and  , respectively, while  s are not sensitive to N-Z asymmetry):
! r2 (!; r) r

2!Uopt (!; r) = (!; r) + 2M
 !   1 + g(!;
 r) r (36)
 = 4 1 + M b0 (!)  b1 (!) N A Z (r)+
37
  !  B (!)2 (r)i
1 + 2M 0 (37)
 1  
 = 4 1 + ! c0(!)  c1(!) N A Z (r)+
1
M 
2
+ 1 + ! C0 (!) (r) (38)
2M
where ! is the rst component of the pion 4-momentum, M is the nucleon
mass, Uopt is the optical potential, b0 ; b1; c0; c1; B0; C0 are complex parame-
ters, g is the Lorentz-Lorenz correction parameter 83 (here and in the following
the natural system of units, hc = 1 , is used). The functional form of this
potential derives from that of the pion-nucleon scattering amplitude: b 's are
related to s-wave pion-nucleon interaction, while c 's are related to p-wave
interaction. The terms in 2 account for two-nucleon mechanisms. This po-
tential contains a divergence operator, thus it is non-local, that means that it
depends on the pion velocity besides on position, or, equivalently, the potential
in a given position depends on the value of the wave function at dierent po-
sitions. A semiclassical approximation of the particle trajectory is in principle
not possible for a non-local potential. However, many localization procedures
have been already devised for it 84;86. We do not go into details, but just give
the result to understand its properties :
2!Uopt (!; K) =  K 2 1 +g + 2M ! r2 (39)
where K is the pion wave number inside the nucleus and the nuclear density
that enters in  and  is the local one. It can be seen that the dependence
on momentum is strong. In the literature, the Klein-Gordon equation for the
pion in the nuclear optical potential is always written as:
h i
(! Vc )2 2!Uopt K 2 = m2 (40)
From eq. 40 applying energy conservation and calling k0 the linear momentum
of the pion outside the potential range, one obtains an equation for K 2 :
K 2 = k02 + Vc2 2!Vc2 2!Uopt (!; K) (41)
substituting eq. 39
k2 + V 2 2!Vc2 +  2!M r2
K2 = 0 c 1  (42)
 = 1 +g
38
The real part of K 2 gives then the real part of the potential from eq. 41. As
pointed out by Johnson and Bethe 87, the denominator of eq. 42 approaches
zero for normal nuclear densities. In ref. 87;88 they suggested a correction which
takes into account the eects of correlated scattering centers, and reduces the
values of the potential as the density increases. The values of the optical po-
tential parameters at low energies (below 50 MeV) are known 83;89 from pionic
atoms and scattering data. At higher energies, however, only an extrapolation
based on pion-nucleon phase shifts is available 89 . For the dependence of the
p-wave parameters on energy a resonant shape can also be assumed, following
the theoretical behaviour of the pion-nucleon scattering amplitude, which is
dominated by the
resonance channel.
The resulting potential has again a resonant shape, with strong depen-
dences on nuclear density. Its depth reaches several tens of MeV, therefore
its introduction, although complex, is essential to correctly describe the pion
transport and interaction in nuclear matter. An example is shown in g. 11
for pions in Lead, as a function of pion energy and nuclear radius. In making
this plot, a nuclear density of the form of eq. 17 has been used.
5.5 Multibody Absorption of Pions

Figure 12: Charged pion absorption cross section for innite symmetric nuclear matter at
two dierent density values ( 1 = 
21  = 
 = 
 )
abs abs abs pro abs neu

Negative pions at rest can be captured on atomic orbits, forming pi-mesic

atoms, jump on more internal orbitals with the emission of X-rays and nally
be captured by the nucleus. This process is a long-studied one 84 , and led to
39
the construction of very detailed models of low-energy pion-nucleus interac-
tions. But this is not the only possible absorption: the in- ight one exists,
and its cross section can amount to about one third of the total pion-nucleus
cross section, and about one half of the reaction one. For free pions and nu-
cleons the non-radiative absorption process does not occur since energy and
momentum cannot be conserved, and the radiative absorbtion is a very weak
process. Conversely, pion absorption in complex nuclei can proceed via mult-
inucleon processes: the basic process is absorption on two nucleons, thus the
simplest case to be examined is the absorption on the deuteron. Both total
cross sections and angular distribution of emitted nucleons have been exten-
sively measured in the d ! pp reaction, and show again the dominance of the

(1232) resonance 83;85 . At very low energies, the s-wave non-delta interaction
is important and the cross section rises as the inverse of the relative velocity.
Turning to absorption on an arbitrary pair of nucleons, this dominance of non-
resonant s-wave and resonant p-wave absorption is the guide to describe the
process. Again, the isospin decomposition turns out to be a very useful tool.
The outgoing nucleon pair can have T = 0 or 1, thus these are the two isospin
channels allowed. The initial state is composed by a pion, and again an isospin
I = 0 or 1 nucleon pair. If we label the amplitudes as fIT , some passages of
momentum combination algebra help to demonstrate that, given a pion with
isospin projection , and a nucleon pair with isospin projection i,
X
(; t) / j(Ii1jTt)j2
jfIT j2 ; t = i +  (43)
IT
If, now, we consider the intermediate state, in order to extract the
contribu-
tion, we can use the Racah coecients to recouple the isospin and explicitate
the pion-nucleon intermediate state, always in a denite total T state . In this
case the amplitudes will be denoted by g 23 ;T and g 12 ;T , and the link between
the f and the g is given by
p 
f01 = p1 2g 23 ;1 + g 12 ;1 (44)
3
 p 
f11 = p1 g 32 ;1 + 2g 21 ;1
3
f10 = g 12 ;0
If we assume
dominance, and put g 12 ;T =0, from eqs. 44 and 43 we obtain that
for charged pions the absorption cross section on a (p,n) pair is ve times that
on a (p,p) (or (n,n)) pair. Moreover, the isospin 0 channel has no contribution
from the
resonance.
40
 The formal treatment of this process inside a real nucleus is quite com-
plex, because the pion can interact with other nucleons before being absorbed,
it acquires a potential energy due to its average interaction with nucleons and
to the Coulomb eld, and the
properties, in particular the partial widths,
are modied by nuclear matter. Inside nuclear matter, the
can either de-
cay, resulting in elastic scattering or charge exchange, or interact with other
nucleons, resulting in pion absorption: its width is broadened with respect
to the free one 5;6;7;90. Modern calculations 6 take these eects into account
by calculating energy and nuclear density dependent partial widths for the

decay:
1 1
2 T = 2 F Im
(45)

= Q +2 +3 , where Q , 2 and 3 are the partial widths for quasielastic
scattering, two body and three body absorption. The nuclear density depen-
dence of two and three body absorption turns out to be of the type  and
 , respectively with  < 2 and < 3, thus dierent from the simple 2
(3 ) dependence one expects for a two-body (three-body) process. In this way
pion scattering, charge exchange, two-body and even three-body absorption
can be treated in a consistent way as two-step processes, the rst one being
the excitation of a
from a pion-nucleon interaction, the second being the
propagation and rescattering or decay of the resonance in the nucleus. The
inclusion of this process in an INC code allows to keep into account processes
that are called FSI (nal state interaction) and ISI (initial state interaction),
that are absorption preceeded (or followed) by scattering of the projectile or
of the secondaries, and the eect of the Coulomb eld. These processes repre-
sent a signicant complication when theoretical calculations of absorption are
compared to experimental data in medium-heavy nuclei.
In addition, a two-body s-wave absorption cross section can be derived
from the optical model 83 as
sA (!) = 4 !
p (1 + 2m )ImB0 (!) (46)
where ! is the pion energy in the lab frame and B0 is a parameter of the pion
optical potential.
As a result, the absorption cross section on nuclei still resembles in shape
that on the deuteron, with a resonance-like behaviour, but the resonance is
smoothed, and an enhancement of the  cross section with respect to the +
one arises, both due to the Coulomb eld and to the neutron excess in heavy
nuclei. The absorption cross section, both 2 and 3 body resonant, and 2 body
non resonant, is shown in g. 12, for two dierent values of the nuclear density.
41
The gure is based on the approach outlined in 27. The values must be taken as
indicative, mainly because no dierence between neutron and proton densities
has been included.

5.6 Hadron Mean Free Paths in the Nucleus and Pauli Blocking
As already mentioned, the plain use of hadron nucleon-nucleon cross sections
leads to values of the mean free path of hadrons which are by far too short
with respect to reality. There are many eects that in uence the in-medium
cross sections. The rst, and better known, eect, is Pauli blocking. Any
secondary nucleon created in an intranuclear interaction must obey the Pauli
exclusion principle, thus it must have enough energy to jump from the Fermi
sea where it lies before the interaction to an unoccupied state, above the Fermi
level. As a result, interactions with small momentum transfer are partially
inhibited. Here two points must be mentioned: the former is that the kinetic
energy of the projectile inside the nucleus is dierent with respect to the free
one because of the presence of the nuclear potential. The latter is that the
target nucleons possess themselves a kinetic energy, distributed accordingly to
the Fermi distribution 22. The interaction kinematics must be applied in the
correct frame, that of the projectile-target nucleon center of mass, to get the
correct cross section and outgoing particle distribution.
Analytical calculations for the combined eect of Fermi motion and Pauli
blocking on the in-nucleus nucleon-nucleon cross section can be performed un-
der a few approximations, such as isotropy of the angular distribution and ei-
ther independence on or inverse proportionality to energy of the cross section 75.
The reduction of the cross section with respect to the free one depends on the
projectile energy and on the Fermi energy. For instance taking EF = 40 MeV,
the reduction factor is  0:3 at 20 MeV incident Energy (measured outside the
nucleus) and  0:6 at 100 MeV. Even with these corrections, the mean free
path of a 60 MeV nucleon results as small as 2 fm.
In a MonteCarlo cascade code, the aforementioned approximations can
be avoided, by explicitly checking after each interaction that the momenta of
all secondary nucleons are above the Fermi level. Pauli blocking is of course
eective also in pion-nucleus interactions, both scattering and absorption.
Mechanisms other than Pauli blocking are eective in increasing the par-
ticle mean free path in nuclear medium. These mechanisms are important to
prevent the well known problems 91 met by INC codes which includes refrac-
tion and re ection because of strong secondary absorption in the nucleus core,
and to match at low energies the mean free paths obtained from optical model
analysis.
42
  The formation zone 92 concept after pion or nucleon inelastic (pion pro-
duction) interactions
 Nucleon antisymmetrization eects 82 , which decrease the probability for
secondary particles to reinteract on a nucleon of the same type very close
to the production point (see eq. 29)
 Nucleon-nucleon hard-core correlations (see for example 93) which also
prevent secondary particles to collide again too close to the production
point. Typical hard-core radii used are in the range 0.5-1 fm
 \Coherence" length after elastic or charge exchange hadron-nucleon scat-
terings. In analogy with the formation zone concept, such interactions
cannot be localized better than the position uncertainty connected with
the four-momentum transfer of the collision. Reinteractions occurring at
distances shorter than the coherence length would undergo interference
and cannot be treated anyway as independent interactions on other nu-
cleons. It must be stressed that this mechanism is rather selective on the
outgoing direction. Forward scattered particles are preferably emitted as
well as backward scattered ones if the colliding particles are identical, pp
or nn for example.
5.7 Preequilibrium Emission
The intranuclear cascade model is the most straightforward approach when
writing a MonteCarlo code, and has been successfully used for decades in
the intermediate energy region. However, its physical foundation becomes
approximate at low energies as seen in the previous sections, resulting in a
decrease of its accuracy. Moreover, it can be very time consuming, since many
particles must be followed down to very low energies.
On the other hand, a description based simply on direct reaction plus com-
pound nucleus evaporation is surely not sucient. Already in 1966 Grin 74
described the spectra following nucleon-induced reactions in terms of a pree-
quilibrium model, that is, a transition between the rst step of the reaction
and the nal thermalization.
Since then, many models have been developed (see 4 for an exhaustive re-
view). The two leading approaches (with many dierent implementations) are
the quantum-mechanical multistep model 94, which has a very good theoretical
background, but is complex and poses some diculties to the description of
multiple nucleon emission, and the exciton model 4;74;95, which relies on sta-
tistical assumptions. This makes it very simple and fast, but of course leads
to some limitations, especially for medium-high energy projectiles.
43
 The approach described here is that adopted in peanut and used for the
examples reported at the end of this lecture. Before describing the actual
preequilibrium model, it is important to discuss how the transition from the
INC part to the preequilibrium stage occurs.
The INC step goes on until all nucleons are below 50 MeV (with the further
specications discussed before) and all particles but nucleons (typically pions)
have been emitted or absorbed. At the end of the INC stage a few particles may
have been emitted and the nuclear conguration is characterized by the total
number of protons, Zpre 0 , and neutrons, N 0 , by the number of particle-like
pre
excitons (nucleons excited above the Fermi level), np (np = npro +nneu), and of
hole-like excitons (holes created in the Fermi sea by the INC interactions), nh ,
by the \compound" nucleus excitation energy (actually the nucleus is not yet at
all in apequilibrated state and the term \compound" is somewhat incorrect), E
(E = s MN 0 ;Z 0 , where ps is the center of mass energy of the system), and
by the \compound" nucleus momentum components, pi comp . All the above
quantities can be derived by proper counting what occurred during the INC
stage and they represent the input conguration for the preequilibrium stage.
It must be stressed that in our approach the typical problems of INC codes
with binding energies and reaction Qs are completely solved, thanks to the use
of the \running" binding energy, which evolves with the reaction evolution and
accounts at every emission stage for the proper Q.
The exciton formalism employed in peanut follows that of M. Blann and
coworkers 96;97;98;99, called Geometry Dependent Hybrid Model (GDH). Indeed
there are a few modications, regarding mainly the way the nuclear geometry
is accounted for, inverse cross sections, and exciton reinteraction rates.
The preequilibrium process in the exciton model is described as a chain
of steps, each step corresponding to a certain number of \excitons", where an
exciton can be either a particle above the Fermi surface or a hole below the
Fermi surface. The statistical assumption underlying the exciton model states
that any partition of the excitation energy E among n, n = nh + np , excitons
has the same probability to occur. The nucleus proceeds in the chain through
nucleon-nucleon collisions which increase the exciton number by two units,
thus assuming that the probability of having an interaction that decreases the
exciton number or lets it unchanged can be neglected (the so called \never
come back" approximation). The chain stops, and equilibrium p is reached,
when either the exciton number n is suciently high (n = 2gE), where g is
the single particle level density, or the excitation energy is below any emission
threshold.
The initial number of excitons depends on the reaction type and on the
cascade history.
44
 At each step there is a denite probability Px;n() of emitting a nucleon
of type x and energy  in the continuum. This probability can be factorized
in two parts, one giving the fraction of n-exciton states in which one exciton is
unbound and has energy  in the continuum, the other giving the probability
for the exciton to escape from the nucleus during its mean lifetime 96:
Px;n()d = n (U; )gd rc ()
n (E) rc () + r+ () (47)
where npx is the number of particle-like excitons which are of type x, g is
the single-particle state density, U is the residual nucleus excitation energy
(U = E  Ben ), n (E) is the density (MeV 1 ) of exciton states, and is
given by:
g(gE)n 1
n (E) = n!(n (48)
1)!
rc () is the rate of emission in the continuum, and is related to the cross section
of the inverse process (inv ) by the detailed balance principle (see par.5.8),
rc = inv g (2s +h1)3 8m (49)
x
and r+ () is the exciton reinteraction rate. The r+ () can be calculated from
the nucleon mean free path in nuclear matter 75 (V = EF + Ben , fPauli is the
Pauli blocking suppression factor, see 75 for explicit expressions),
 + V ) 1=2
r+NN = fPauli (; EF ) [pro xp + neuxn] 2( m (50)
or from the optical model (W is the imaginary part of the optical potential).
r+W = 2Wh (51)
The value of the single particle density is often taken as g=Z/14 (N/14)
MeV 1 for protons(neutrons), although it could be a complex function of en-
ergy 100.
This formulation has been rened in the GDH 97 to account for the ex-
perimentally established importance of peripheral collisions. This has been
accomplished through two mechanisms, both applied to the rst step of the
exciton chain. As rst, all position dependent parameters (density, Fermi en-
ergy, imaginary part of the optical potential etc) entering the quantities in
45
eq. 47, 49,50(51) are no longer constant, but depend on the impact parame-
ter and are chosen according to an average carried out along a straight line
crossing the nucleus at constant impact parameter. The relative importance of
dierent impact parameters is then established through the use of transmission
coecients obtained from optical models.
Secondly, the exciton state density is modied from eq. 48 by the assump-
tion that any hole-like exciton cannot carry an excitation energy larger than
the (local! impact parameter averaged) Fermi energy. Explicit formulae for
the constrained exciton state densities can be found in 99. Both mechanisms
contribute to the enhancement of the hardest part of the emitted particle spec-
tra.
Our approach is similar but with a few dierences. First of all geome-
try dependent quantities are no longer averaged as a function of the impact
parameter, but real point like values are used, since the position of the rst
interaction is known from INC part where anyway the projectile is tracked into
the nucleus until the rst interaction occurs (remember that the interaction
can or can be not modelled through the INC stage depending on the projec-
tile energy, but tracking is always performed). Therefore when entering the
preequilibrium stage peanut knows the actual values of the nuclear density
and of the (local) Fermi energy corresponding to the hole(s) generated in the
INC step at positions ~xi. At the rst step only one hole has been created and
therefore the local nuclear density and Fermi energy are those corresponding
to the interaction position. However, since the preequilibrium stage follows the
intranuclear cascade step, at suciently high energy it can be reached when
already two or more holes have been created. In this a proper average must be
taken. In general for nh holes already present when reaching the preequilibrium
stage:
Pnh (~x )
loc
nh = i=1 i (52)
PnhnhE (~x )
EFlocnh = i=1 F i (53)
n h
The local Fermi energy dened in eq. 53 is then used when computing the
exciton state densities for constrained exciton states, since these are truly local
quantities. Constrained exciton state densities are in fact used in peanut for
the congurations 1p-1h, 2p-1h, 1p-2h, 2p-2h, 3p-1h and 3p-2h, which cover
most of the emission spectra.
However the Fermi energy and the nuclear density enter also the denition
of r+ if the nucleon-nucleon cross section approach is used (and it is used in
peanut ). For such purpose the use of point values would not be correct, since
46
a possible subsequent reinteraction will take place at some distance. Therefore
the following quantities are introduced and used when computing reinteraction
rates as representative of the nuclear matter in the neighborhood of the point
where the nucleon originated,
nh loc
nh + 
ave
nei
nh = nh + 1 (54)
n E loc + E ave
EFneinh = h Fn nh+ 1 F (55)
h
ave ave
where  and EF are the nuclear density and the Fermi energy respectively
averaged over the whole nucleus. At each step of the preequilibrium stage the
exciton number is increased and nh ! nh + 1, and of course the position of
the last interaction is not known. Our prescription is to use for the new local
values
loc nei
nh +1 = nh (56)
loc
EF nh +1 = EF nhnei (57)
which implicitly assumes that the density and the Fermi energy of the last
generated hole are given by ave and EFave respectively. It is clear that as soon
as the number of steps becomes large all quantities naturally converge to the
nucleus averaged values. Summarizing our treatment of geometry dependent
eects is much more \local" for the rst interaction(s), where not only im-
pact parameters but complete positions are explicitly selected and taken into
account, and becomes less and less local, in contrast with the original GDH
approach as soon as the number of interactions increases, according to a de-
scription where larger and larger volumes of the nucleus are involved in the
reaction.
For the exciton reinteraction rate nucleon-nucleon cross sections cor-
rected 75 for Fermi motion and Pauli principle, have been used. Further correc-
tions connected to nucleon correlations and to coherence considerations after
scattering events have been introduced when computing the reinteraction rate,
for consistency with what is done in the INC part. These corrections proved
to be very useful and prevented the need for arbitrary reduction factors of the
nucleon-nucleon cross sections, which were often required in similar models to
match the experimental data.
In our approach an energy dependent form for g is usually assumed, which
follows the main features described in 100 for simple cases.
In the exciton model the angular dependence has to be somehow added,
since it is not intrinsic in the formulation. Many dierent approaches have been
47
recently developed, most of which rely on the fast particle approximation. A
non-isotropic angular distribution has been implemented in peanut , following
the fast particle approximation 101, as implemented by Akkermans et al. 102. In
this model the angular orientation of the nucleus at each step is dened by the
direction of the fast particle, which changes gradually in a series of two-body
collisions. The transition rate between dierent exciton states is supposed to
be factorizable in an energy-dependent and an angle-dependent factor. The
same factorization holds then for the resulting emission probability :
dPx;n; ()d = P ()d
X am P (cos()) (58)
d
x;n l l
l
where Px;n()d is the angle-integrated emission probability from an n-exciton
state as given by GDH, the aml are coecients that depend on the number
of steps m in the exciton chain and  in the original formulation is the polar
angle with respect to the direction of the projectile. Besides the fast particle
one, other approximations are introduced in the calculations leading to eq. 58:
in the binary collisions the Fermi motion of target nucleons and the Pauli
exclusion principle are neglected, as are re ection/refractions at the nuclear
boundary.
Since in peanut the exciton stage may be reached after one or more
cascade steps, possibly with particle emission, care must be taken in dening
the initial step number and the reference direction. The adopted choice has
been to dene the reference axis as that of the residual momentum of the
system at the step under consideration (it coincides with that of the projectile
if no particles have been yet emitted). If this total momentum is comparable
with the average Fermi momentum, the angular distribution is assumed to be
isotropic. As for the step number m, it is taken equal to the number of holes
below the Fermi level.
For projectile energies below 30 MeV, where only tracking up to the rst
interaction is performed in the INC stage, the preequilibrium model must
provides also the angular distribution for nucleons scattered only once. The
adopted approach is to consider an angular distribution kernel obtained assum-
ing isotropic nucleon-nucleon scattering and folding with the Fermi motion of
the target nucleon and with (average) refraction and re ection eects in the en-
trance and exit channels. Such a kernel is properly correlated with the selected
nucelon emission energy.
At the end of the preequilibrium stage, a true compound nucleus is left
with Zres and Nres , moving with ~pres , and with excitation energy U. The
Evaporation/ssion/fragmentation stage is then performed starting from this
conguration.
48
5.8 Evaporation/Fragmentation
At the end of the exciton chain, or of the INC part whenever no preequilibrium
stage is included in the calculation, the residual nucleus is supposed to be left
in an equilibrium state, in which the excitation energy U is shared by a large
number of nucleons. Such equilibrated compound nucleus is supposed to be
characterized by its mass, charge and excitation energy with no further memory
of the steps which led to its formation. The excitation energy can be higher
than the separation energy, thus nucleons and light fragments (, d, 3H, 3He)
can still be emitted: they constitute the low-energy (and most abundant) part
of the emitted particles in the rest system of the residual nucleus, having an
average energy of few MeV. The emission process can be well described as
an evaporation from a hot system. The treatment starts from the formula of
Weisskopf 103, that is an application of the detailed balance principle, linking
the probabilities Pi!f to go from a condition i to a condition f and viceversa
through the density  of states in the two systems:
Pi!f (i) = Pf !i (f) (59)
Pf !i is the probability per unit time of capturing a particle and form a com-
posite nucleus, and is given by the product of the compound nucleus cross
section inv times the particle ux. Restricting to a volume
, the particle
ux is
v , and
Pf !i = inv
v
Thus the evaporation probability for a particle of type j , mass mj , spin Sj
h
and kinetic energy E is given by
Pj (E)dE = (2Sj +2 1)m j
inv f (U f)
EdE (60)
 h 3 i (Ui )
where 's are the nuclear level densities (f (Uf ) for the nal nucleus, i (Ui ) for
the initial one), Ui  U is the excitation energy of the evaporating nucleus,
Uf = U E Qj that of the nal one, Qj is the reaction Q for emitting a
particle of type j from the original compound nucleus, and inv is the cross
section for the inverse process. The non-relativistic phase space density for a
free particle of spin S in a volume
has been used (eq. 22 with a spin term)
and entered in the nal state density.
Eq. (60) must be implemented with a suitable form for the nuclear
level density and the inverse cross sections. Many recipes have been sug-
gested for both. In the original work of Dostrovsky 106, where the evapora-
tion process has been translated into a MonteCarlo code for the rst time,
49
 p
(U)  C exp (2 aU), with a = A=8 MeV 1 has been used for the level den-
sity dependence on the excitation energy U, an expression inspired to the Fermi
gas one. This has led to a simple form for the evaporation probability:
p
2 a(U E Qj )
Pj (E)dE = (2Sj +2 1)m j e
inv p EdE: (61)
 h3 e2 aU
In the same work, the inverse cross sections have been parametrized in a
very simple way, so that expression (61) can be analytically integrated and
used for MC sampling. The same formulation is used in the examples reported
in this work, with, however, a dierent choice of a, taking into account that the
Fermi expression for the nuclear density must be rened, essentially allowing
the value of a to depend both on nuclear species and on excitation energy (see
below). As a rst steppaU the so-called backshifted level 2density
pa(U
)should be used,
where (U) = Ce 2 is substituted by (U) = Ce ,
being the
pairing gap in the nucleus considered.
To get a rst idea of the spectrum of emitted nucleons, one can isolate the
dominant dependence on the outgoing particle energy, in the case of neutrons,
where one can assume an almost constant inv:
2
pa(U E Qj )
Pj (E)dE / E e p dE
p
e2 aU


2 aU 1 12 E+Qj
 Ee
U
p dE
e2 aU
 Ee TE dE (62)
from which the evaporative Maxwellian
q shape of the spectrum is evident, once
the nuclear temperature T  Ua is introduced (this denition of nuclear tem-
perature follows from the thermodynamical denition T = [d(ln(U))=dU] 1
). Taking for instance U=20 MeV in an A=80 nucleus (a = 10), the emit-
ted neutron spectrum is peaked at about 2.8 MeV, that is quite a low energy.
The emission of protons is reduced and the spectrum is shifted toward higher
energies due to the Coulomb barrier.
The total width for neutron emission can be found by integrating Eq. (60)
between zero and the maximum possible ejectile energy (U Qj )
(2S + 1)m Z (U Qj )
j=
j
2
j inv (E) f EdE (63)
 h
2 0 i
50
The same applies to charged particles, where the integration actually goes from
some eective Coulomb barrier where inv drops to zero, up to the maximum
energy.
The evaporative process is in competition with another equilibrium pro-
cess, that is ssion 110. A fraction of the excitation energy may be spent to
induce a collective deformation. As the nucleus shape departs from sphericity
the surface energy increases but the Coulomb energy decreases. The poten-
tial energy reaches a maximum at a deformation stage that is called \saddle
point". The height of the potential energy over the ground state is the ssion
barrier Bf . Once a nucleus reaches the saddle point, the ssion occurs, and the
nucleus separates, most of the times into two heavy fragments. The height of
the ssion barrier can be in rst order calculated with the liquid drop model,
although shell eects have to be incorporated. For the ssion probability, a
statistical method can be used 103;111: it will be equal to the probability of
reaching the saddle point, because this is a no-way-back point. Given the ex-
citation energy U, this will be divided between a relative kinetic energy of the
two fragments E and a residual excitation. All the procedure is like the one
for evaporation, where now the spectrum is that of the kinetic energy at the
saddle point, the Pf !i is simply one times the ux of \ingoing" fragments, but
the system is conned to one dimension because of the path to saddle point.
The total ssion width can be expressed as:
1 1 Z (U BF )
F = 2  (U) F (U BF E)dE (64)
i 0
p
where BF is the ssion barrier, and F (UF )  C exp (2 aF UF ), the level
density of the ssioning nucleus at the saddle point, where the excitation energy
UF is given by the initial one minus 2the ssion barrier. Fission barriers vary
roughly with the ssility parameter ZA : measured values range from about 25-
30 MeV at ZA2  30 to 6 MeV for ZA2 > 34. Fission for nuclei with Z < 70 can
be neglected in most practical calculations. In other nuclei, the ssion width
has to be compared with the evaporation widths to determine the relative
importance of the two processes. The level density at the saddle point F
is dierent from that of the nucleus in its ground state. From comparison
to experimental data, it turns out that aF is greater than the a used for
evaporation of about 10% at low excitation energies, and the two a's become
equal at large excitation energies. Moreover, a~ = a=A, and a~F = aF =A, are
found to be all but constant parameters: they possess a dependence on A and
Z, due to shell and deformation eects, and a dependence on excitation energy.
Both eects have been experimentally observed, and have been subject of many
phenomenological and theoretical investigations (see 100;104;105;107;108;109).
51
 After scission, the fragments acquire kinetic energy from the Coulomb
repulsion between them. Their mass distribution depends on Z and on the
excitation energy of the ssioning nucleus: for low Z ssioning system the
mass distribution is centered around the half mass, while for high Z nuclei, the
mass distribution has a symmetric component and an asymmetric one in which
one of the fragment has always a mass around 140. The relative importance
of the symmetric component grows with growing excitation energy.
In the examples presented in the following paragraphs and obtained with
the fluka code, the prescriptions of Atchison 112 have been used to calculate
the quantities entering in Eq. (64), except, again, for the level density param-
eter aF , and for the omission of the \ad hoc" reduction factor for the ssion
width.
In both F (U) and (U) the so-called backshifted level density have been
used, using U
rather than U, where
is the pairing energy. For a; aF ,
the N and Z dependence of ref. 109 has been used, and complemented with the
energy dependence prescription of Ignatyuk 107;108.
a = A
[a
f(U) + ~a
(1 f(U))]
a = a0 + 9:17  10 3
[SZ (Z) + SN (N)] (65)
a~ = 0:154 6:3  10
A 5

f(U) = 10:054 e 0:054
(U
)

(U
)
where according to 109, a0 is given by 0.142 and 0.12 MeV 1 for undeformed
and deformed nuclei respectively, and SZ (Z) and SN (N) are the shell correction
terms for protons and neutrons. The unit of energy used throughout eq. 65
are MeV.
We use aF  1:08a, with a smooth A dependence. After ssion occurs,
the two fragments are treated like independent residual nuclei with their own
excitation and can possibly emit further particles.
For light nuclei, the statistical assumptions and the sequential emission
scheme underlying the classical evaporation models become less and less sound,
because:
 Already moderate excitation energies can represent a substantial fraction
of the (total) binding energy of such nuclei.
 The level structure of such nuclei is usually highly specic and anyway
level spacings can be comparable with the excitation energy.
 The \evaporation" of light fragments other than p or n becomes mean-
ingless, since the mass of the \evaporated" fragment can be comparable
52
 or even larger than the mass of the residual nucleus.
Therefore other deexcitation mechanisms are more suitable for these light (typ-
ically A16) residual nuclei. A possible choice (adopted for all presented ex-
amples) for this calculations is the so called Fermi Break-up model 113;114,
where the excited nucleus is supposed to disassemble just in one step into two
or more fragments, with branching given by plain phase space considerations.
In particular, the probability for disassembling a nucleus of N neutrons, Z pro-
tons, and U excitation energy (total mass M  = U + MA;Z ) into n fragments
(n  2) of the same total charge and baryon number, is given by:
 Vbr n !3=2
g
W = G (2h)3
1 1 Y n (2)3(n 1)=2 E 3n=2 5=2
M  i=1 mi ( 32 (n 1)) kin
(66)
where the spin factor g, and the permutation factor G are given by (nj is the
number of identical particles of jth kind)
Y
n Y
k
g= (2Si + 1); G= nj ! (67)
i=1 j =1
and Ekin is the total kinetic energy of all fragments at the moment of break-up.
Vbr is a volume of the order of the initial residual nucleus volume. Therefore,
the nal state can be conveniently selected by means of a MC procedure,
by evaluating such an expression for all possible combinations of fragments
energetically allowed and making a random selection.
In the examples reported in this paper, all combinations formed by  6
fragments have been considered, unless the residual \nucleus" is composed by
A like particles (p or n), in which case it is disintegrated into A fragments
according to phase space. All particle stable states with A  16 have been
included, plus the particle unstable levels with sizeable decay branching ra-
tios. Also a few known particle unstable isotopes, like 8 Be, have been included
and, if produced, are let to decay according to the experimental branching.
Once the nal state conguration has been selected, the kinematical quanti-
ties of each fragment are chosen according to n-body phase space distribution.
Such a selection must be performed having care to subtract from the available
energy the Coulomb repulsion of all charged particles: the Coulomb energy is
then added back to the charged particles alone, to simulate properly the eect
of the Coulomb repulsion. In practice Ekin at disassembling will be given by:
X
n !
Ekin = U mi c2 MA;Z c2 BCoul (68)
i=1
53
where it must be recalled that the emitted fragments can be in an excited state.
The total Coulomb barrier BCoul of the selected conguration is distributed to
charged particles after disassembling, in their own c.m. system.
According to the picture of the compound nucleus like an equilibrated
system determined only by its mass, charge and excitation energy, with no
memory of previous steps of the interaction, Fermi Break-up is activated in
the model every time the current compound nucleus has mass number A  17,
including possible light ssion fragments.
A similar process occurs also in medium-heavy nuclei, although it becomes
important only at multi-GeV proton induced reactions or in heavy ion induced
reactions. This is called nuclear fragmentation and is characterized by the
emission of low energy complex fragments with A  5 from a thermalized
nucleus. Reviews of this topic can be found in 117;120;118. Fragmentation is
expected to occur when the excitation energy per nucleon exceeds about 3
MeV 115, that in A=100 mass nucleus correspond to 300 MeV of excitation
energy after all direct and preequilibrium emissions. As an example, the cross
sections for emitting Sodium or Magnesium isotopes in proton induced reaction
on gold are or the order of 0.03 mb at 0.5 GeV, 0.5 mb at 1 GeV and saturate
at values between 3 and 10 mb for incident energies larger than 10 GeV. This
is not a process that will in uence the development of a shower or the average
energy deposition, but it is the only one that can produce residual nuclei very
far both from the target mass and from the ssion product distribution, thus
it must be taken into account for a detailed analysis of residual nuclei.
Theoretical descriptions of fragmentation model it as a two-stage process,
the rst being the production of a suciently hot compound nucleus, the
second being the fragmentation itself. For the rst stage, both intranuclear
cascade 119;115, and Fluidodynamical 116 models have been employed. The
second stage can be described in terms of statistical methods or in terms of
the uctuations of the nuclear mean eld. The statistical approach of 115 and
that of 121 are among the few examples of implementation in a MonteCarlo
code.

5.9 Residual Nucleus Deexcitation

The evaporation process continues till it is energetically possible, and leaves
generally the residual nucleus in an excited state. The residual excitation
energy is then dissipated by emission of rays. Actually emission occurs
even during evaporation, as a competing process, but with a small branching
ratio.
The evaporation stage ends when the nuclear excitation energy becomes
54
Figure 13: Photon spectrum following the re- Figure 14: as in g.13 for Tungsten. Experi-
action Ti(n,x) at 19 MeV. The dashed his- mental points are from 129
togram represents MonteCarlo results with
errors. Dots are experimental data from 128

lower than all separation energies for nucleons and fragments. This residual
excitation energy is then dissipated through emission of photons. In reality,
gamma emission occurs even during the preequilibrium and evaporation stages,
in competition with particle emission, but its relative probability is low, and
it can safely neglected in most practical applications.
Gamma deexcitation proceeds through a cascade of consecutive photon
emissions, until the ground state is reached. The cascade is assumed to be
statistical as long as the excitation energy is high enough to allow the denition
of a continuous nuclear level density. Below a (somewhat arbitrary) threshold,
set at the pairing gap value in the presented examples, the cascade goes through
transitions between discrete levels.
The statistical model formulation for the gamma ray emission probability
is again similar to those for evaporation and ssion 122;123:
(Uf ) X f(E ; L)dE
P(E )dE = f (U (69)

i i) L

where L is the multipolarityof the transition. The strength functions f(E; L)

can be either derived from photoabsorption cross sections or calculated from
single-particle estimates of transition strengths. The former approach is more
55
sophisticated but requires the knowledge of the resonance parameters for all
isotopes; the latter approach is easier and sucent for a rst order estimate
of the spectral distribution. In the following the procedure used to compute
the examples and implemented in the peanut code is brie y described. We
assume
f(E ; L) = cL
FL (A)
E (2L+1) (70)
where E (2L+1) is the energy dependence for multipolarity L. For the cL coe-
cients the Weisskopf single particle estimates124can be used. The FL(A) factors
have been included to partially account for the many eects that bring to devi-
ations from the single particle estimates; they are rough A dependent averages
of the hindrance and enhancement factors given in 125. Only E1, M1 and E2
transitions have been considered. The assumed level density is the same as
in the evaporation part, but the ratio of exponentials coming from the level
densities has been approximated as rst order expansion around E = 0. This
is equivalent to the assumption of a constant nuclear temperature at low exci-
tation energies, that is often used in the analysis of gamma emission following
neutron capture122;126
As a result, one obtains the expressions for the emission probabilities for
the considered multipoles. Since competition between photon and particle
emission is neglected at the present status of the model, only the relative
values are of interest:
E
P(L; E )dE = C~LE (2L+1) e T dE (71)
~CM 1 2 FM 1 (A)
~CE 1 = 0:31A 3 FE 1 (A)
C~E 2 = 7:2
10 7A 23 FE 2 (A) MeV 2
C~E 1 FE 1 (A)
T is the nuclear temperature at the initial excitation energy U, taken as U

= aT 2 , a being the usual level density constant and
the pairing energy.
A rst sampling is performed on the integrated gamma emission proba-
bilities to choose the character (electric or magnetic) and the multipole order
of the emitted photon, and a second sampling is performed to determine the
emission energy according to the selected multipolarity. For both steps the full
energy range 0  E  U is used, even though the intrinsic limit of validity
would be 0  E  (U
). After emission, all parameters are updated
on the basis of the new excitation energy, and another statistical emission is
performed, until the excitation energy falls below the preset "discrete level
threshold". This threshold has been set to the pairing energy for even-even
56
or odd mass nuclei; for odd-odd nuclei the threshold corresponds to the rst
excited level.
For many isotopes the experimentally determined values of the rst and
second excited levels have been tabulated in the code, for the others a rotational
like structure is assumed, with level energy given by :
2
UI = 2hI I(I + 1) (72)
where I is the level spin (integer for even-mass, half integer for odd mass nu-
clei, 0 or 1/2 for the ground state), and I is the nuclear moment of inertia,
taken as 0.4 times that of a rigid body . The last steps of the cascade consist
of I=2 transitions among these rotational levels, down to the ground state.
When known levels are tabulated, the cascade is forced to pass through them.
All photons are emitted isotropically, since from the evaporation stage no in-
formation on the residual nucleus spin and polarization is available; it would
be possible to implement angular distributions based on angular momentum
estimates and on average experimental values of substate distributions 127.
This deexcitation model has been developed and tested within the pea-
nut 27 model. Anyway the process is supposed to depend only on the residual
nucleus mass, Z and excitation energy after evaporation, and not on the details
of the preceeding interaction history. Two examples of its performances are
shown in gs. 13 and 14. Both the total gamma multiplicity and the shape of
the spectra are well reproduced.
5.10 High Energy Interactions: Generalities
Once suitable models are available for the description of hadron-nucleon (hN
from now) interactions (see paragraph 4.5), there are still two basic ingredi-
ents to be added for a sensible description of hadron-nucleus (hA from now)
interactions at high energies. Both stem from experiments which show some
distinct features of these interactions when compared with hN interactions at
the same energy, or with hA interactions at lower energies, and are supported
by convincing theoretical modelling. The former is the increasing number of
shower particles produced with increasing mass number of the target. The lat-
ter is the apparently reduced cascading into the nucleus when compared with
naive expectations based on hN cross sections and hN multiplicities.
5.11 Glauber Model for High Energy Hadron-Nucleus Interactions
To better appreciate the increase of shower particle multiplicity as a function of
the target mass number, g. 15 shows the experimental rapidity distributions of
57
Figure 15: Rapidity distribution of charged Figure 16: Ratio of rapidity distribution of
particles produced in 200 GeV proton colli- charged, positive, and negative particles pro-
sions on Hydrogen, Argon, and Xenon tar- duced in 200 GeV proton collisions on Xenon
gets, symbols exp. results 133 , histo fluka and Hydrogen, symbols exp. results 133 , histo
results fluka results

charged particles produced in interactions of 200 GeV protons with Hydrogen,

Argon, and Xenon. The ratios of the rapidity distribution of charged, positive,
and negative particles produced on Xenon and Hydrogen are shown in g. 16.
Both experimental data, taken from 133 (which are actually pseudorapidity
spectra above about 1 GeV) and fluka results are presented. The peak at
low rapidity in the Argon and Xenon data is due to grey particles emitted
following cascading of secondaries produced in primary interactions, which are
mainly slow protons produced by cascading into the target nucleus.
These results cannot be interpreted as due to increased cascading in heavier
nuclei of secondaries produced in the projectile-nucleon primary interaction.
There are two main diculties connected with this naive interpretation.
 the rapidity distributions on medium-heavy nuclei are indeed somewhat
shifted toward lower values with respect to hN ones, but the amount of
this shift and the overall shape of the distributions are not in agreement
with those expected for the simple reinteraction hypothesis. Indeed,
looking at the p{p distribution in g. 15, it is clear how most of the
secondaries in p{p collisions are produced in the central region, around
mid-rapidity. For p{p collisions at 200 GeV, mid-rapidity (the rapidity
of the centre-of-mass system) is around 3, while the projectile rapidity
58
 is twice that value, around 6. Particles created in the rst interaction,
when undergoing further collisions, will produce particles with rapidity
distributions similar in shape to that for p{p at 200 GeV, but extend-
ing from 0 to the rapidity of the reinteracting particle ! typically
around 3. Clearly the distributions of particles created in the rst gen-
eration of reinteractions will be centered around 1.5, that of the second
around 0.75 and so on. The net result would be a distribution steadily
increasing towards low rapidities, with no plateau, and showing when
compared to the p{p one a depletion in all the forward emisphere (from
mid-rapidity upward), with an overall spectrum much softer than the p{
p one. Actually the experimental evidence is very dierent, showing an
increase with respect to the p{p case already at rapidity 4-5, with only
a slight decrease in the extreme projectile fragmentation region, with a
well dened plateau just slightly below mid-rapidity. Furthermore there
is no evidence of the large number of slow particles which are foreseen
by the reinteraction hypothesis
 the reinteraction of fast particles produced by primary collisions is stron-
gly inhibited by Lorentz contraction and uncertainty considerations.
These arguments will be developed in more details in paragraph 5.12.
Therefore, the standard interpretation of these results is that the projectile
itself undergoes multiple primary interactions with one or more target nucleons,
and that the number of such interactions is increasing with the target mass.
Experimentalists have developed suitable methods to \count" these primary
interactions 133;134, and their outcomes for the average number of \primary"
collisions are in agreement with:
<  >= A 
hN (73)
hA
where s are to be intended as absorption (reaction) cross sections. A rig-
orous demonstration of eq. 73 in the framework of the Glauber model will be
given in the following paragraphs, together with the exact denition of absorp-
tion cross section. Since high energy hA absorption cross sections scale well
with the mass number:
hA = 0
A (74)
with 0 usually some 20-30% higher than hN and  very close to 2/3, it is
easy to verify that the average number of primary collisions scales roughly like
<  >/ A1=3, with the proportionality constant slightly smaller than unity.
Therefore <  > can be signicantly larger than unity for heavy nuclei, giving
59
rise to a substantial increase of shower particle multiplicity with respect to the
hN case. The theoretical explanation of this multiple collision approach is the
so called Glauber-Gribov cascade 135;136;137;138;139;141;140;142. Within such an
approach multiple collisions of primary particles are naturally accounted for,
and their distribution as a function of the impact parameter can be computed
once hN is known. It is important to stress again that the term \multiple
collisions" must not be interpreted as several successive interactions of the
primary particle, or maybe of the leading particle coming from the previous
interaction but rather as an expansion of the interaction cross section into
several diagrams, the rst one representing the exchange of \something" in
between the primary particle and one target nucleon, the second one with two
nucleons etc.
The Glauber formalism provides also the most powerful approach for cal-
cuting total and elastic hadron-nucleus cross sections, and elastic dierential
cross section from few hundreds of MeV up to the highest energies.
In the following a derivation of the original Glauber approach is sketched
out, and the most relevant formulae for non-elastic interactions are demon-
strated.

Basic Assumptions and Formulae

According to eq. 187 in Appendix B, the nuclear scattering amplitude can be
written as:
ik Z
f(~q; s) = 2 d2~b ei~q
~b hA (~b; s) (75)
where the nuclear prole function, hA , is dened as a function of the complex
phase shift hA , by (see again Appendix B):
hA (~b; s) = 1 ShA (~b; s) = 1 eihA (~b;s) (76)
and the total, reaction and elastic cross sections can be expressed as:
Z
hA T (s) = 2 d2~b Re hA (~b; s)
Z h i
hA r (s) = d2~b 1 jShA(~b; s)j2 (77)
Z
hA el (s) = d2~b j hA(~b; s)j
2

Under the following assumptions:

60
  the phase shifts hA (~b; s) are mainly imaginary ! the scattering ampli-
tude f mainly imaginary as well and mostly absorptive
 the elastic scattering is mostly forward peaked
 the momentum transfer to individual nucleons is negligible
 there is no correlation among nucleons (but for the obvious constraint
on the centre of mass position)
it is reasonable to assume that the nuclear phase shift, hA , is the sum of
the phase shifts out of individual nucleons, that is changes in the amplitude
are mutiplicative. For a purely absorptive amplitude this assumption simply
tells that the overall attenuation factor due to the nuclear target is simply
obtained as the convolution(! multiplication) of the attenuation due to indi-
vidual nucleons. Such a behaviour strictly resembles those of light rays in an
absorbing/diracting medium, where the net eect can be obtained as the mul-
tiplication of the eects of the layers constituting the medium. This phase shift
additivity hypothesis despite its simplicity and naturality, has consequences
which go far beyond. Indeed, under such hypothesis alone, the Glauber ap-
proach allows to derive all properties connected with hadron-nucleus scattering,
making use only of hadron-nucleon interaction and nuclear ground state prop-
erties. Suppose we have a specic conguration of the nucleons, characterized
by the position vectors ~rj (~rj ? when projected onto a plane perpendicular to
the incident beam direction), then the phase shift additivity hypothesis trans-
lates into:
X
A
~hA (~b; s) = hN (~b ~rj ?; s) (78)
j =1
Y A
S~hA (~b; s) = ShN (~b ~rj ?; s) (79)
j =1
A h
Y i
~ hA (~b; s) = 1 S~hA (~b; s) = 1 1 hN (~b ~rj ?) (80)
j =1
ik Z d2~b ei~q
~b ~ (~b; s)
f~hA (~q; s) = 2 (81)
hA
4 Z
~hA T (s) = k ImfhA(0; s) = 2 d2~b Re ~ hA (~b; s)
~ (82)
where tilded values indicate that all quantities refer to that specic nuclear
conguration and quantities labelled hN refer to elementary hadron{nucleon
61
scattering. In principle such quantities depend on the considered nucleon too
(for example p or n) and therefore an index j should be also present. Such index
is not explicitly indicated both in order to simplify the notation and because
h n and h p scatterings are very similar at high energies. Whenever the
(in)equality of h p and h n scattering will be explicitly used, it will be
explicitly mentioned.
Now, in order to compute the observable total cross section, one needs to
average the scattering amplitude over all posibble nucleon congurations of the
specic initial and nal nuclear states under consideration. For this purpose
let us introduce:
g~hA (~q; ~b; s) = ei~q
~b ~ hA (~b; s) (83)
~
ghA (~q; ~b; s) = ei~q
b hA (~b; s) (84)
and write the scattering amplitude as:
ik Z ~ ik Z
fhA (~q; s) = 2 d b e hA (b; s) = 2 d2~b ghA (~q; ~b; s)
2 ~ i~
q
b ~ (85)
then the (averaged) scattering amplitude corresponding to the transition from
an initial nuclear state i to a nal one f can be obtained in terms of the
respective wave functions, i and f .
ghA fi (~q; ~b; s)  < f jg~hA ji >=
ZY
A
= d3~rj f (~r1 ; :::;~rA) i (~r1 ; :::;~rA)~ghA (~q; ~b; s) (86)
j =1
ShA fi (~b; s)  < f jS~hA ji >=
ZY
A
= d3~rj f (~r1 ; :::;~rA) i (~r1 ; :::;~rA)S~hA (~b; s) (87)
j =1
hA fi (~b; s)  < f j ~ hA ji >=
ZY
A
= d3~rj f (~r1 ; :::;~rA) i (~r1 ; :::;~rA) ~ hA (~b; s) (88)
j =1
The i ! f transitions include also transition to excited nuclear states besides
the elastic channel (f = i). A cross section for the transition from the i
(ground) state to the f nal state can be dened according to:
Z 0
hA fi (s) = d
k0 kk jf(k; k0)j2 (89)
62
where k0 is the momentum of the outgoing hadron. Since we are neglecting
k k0, and using formula 184 from Appendix B, it turns out:
1 Z
hA fi (s) = k2 d2~q jfhA  (~q; s)j2 (90)
where using eq. 86:
ik Z ~ ik Z
fhA fi (~q; s) = 2 d2~b ~
hA  (b; s)e
i~q
b = 2 d2~b < f j~ghA  (~q; ~b; s)ji >
(91)
In order to evaluate hA fi and to give a closed expression for ShA fi and
hA fi it is convenient to simplify the notation, introducing:
~u  (~r1; :::;~rA) (92)
Y
A
d3~u  d3~rj (93)
i=1
Making use of eqs. 79,80 together with 87 and 88:
Z Y
A
ShA fi (~b; s) = d3~u f (~u) i (~u) ShN (~b ~rj? ; s) (94)
j=1
Z 8 A 9
~
<
3~u  (~u) (~u) 1 Y 1
h ~
i=
hA fi ( b ; s) = d f i : hN ( b ~
r j? ; s) ;(95)
j=1
Z Z
j hA fi (~b; s)j2 = d3~u d3~u 0 f (~u) f (~u 0) i (~u 0) i (~u)
( A h
Y i)

1 1  (~b ~r 0 ; s)
hN l?
8 l=1 9
< A h
Y i=

:1 1 hN (~b ~rj ?; s) ; = (96)
j =1
Z Z
= d3~u d3~u 0 f (~u) f (~u 0) i (~u 0) i (~u)
" Y
A #2 YA
3

1  (~b ~r 0 ; s) 41
ShN ShN (~b ~rj ?; s)5
l?
l=1 j =1

63
Total, Elastic, Quasi{Elastic, and Absorption Cross Section Deni-
tion
Making use of the closure relation 136 for the nal states f:
X
f (~u) f (~u 0) =  3A(~u ~u 0) (97)
f
one can write, introducing 's as in Appendix B, and summing over all possible
nuclear states:
X X Z 2~
hA f (s)  hA fi (s)  d b hA  (~b; s) =
ZfX
f
= d2~b j hA  (~b; s)j
2=
Z Zf
= d2~b d3~u j i (~u)j2j ~ hA (~b; s)j2 =
Z Z 2 A 32
 Y 
= d2~b d3~u j i (~u)j2 41 ShN (~b ~rj? ; s)5 (98)
 j=1 
Z
hA el (s)  hA ii (s)  d2~b hA el (~b; s) =
Z Z 2 A 32
 Y 
= d2~b  d3~u j i (~u)j2 41 ShN (~b ~rj? ; s)5 (99)
 j=1 
where hA f represents the cross section for both elastic scattering and in-
elastic scattering to excited states, and does not include contributions from in-
dividual hN non-elastic interactions. The total cross section can be expressed,
using eqs. 77,94, as:
Z Z
hA T (s)  d b hA T (b; s) = 2 d2~b Re hA ii (~b; s) =
2 ~ ~
Z Z
= 2 d2~b d3~u j i (~u)j2Re ~ hA (~b; s) = (100)
Z Z 2 A 3
Y
= 2 d2~b d3~u j i (~u)j2 41 ReShN (~b ~rj? ; s)5
j=1
These equations show how the knowledge of elementary hadron{nucleon scat-
tering and of the nuclear ground state is sucient to compute all relevant cross
sections without any further assumption and/or information.
64
 It is customary to dene a cross section for quasi{elastic scattering as the
dierence between hA f and the elastic cross section:

hA qe (s)  hA f (s) hA el (s) (101)

It is worthwhile  R out that hA qe  0, since for whichever function f

R to point
it holds that  dx f   dx jf j.
The physical interpretation of the quasi{elastic cross section is the follow-
ing: besides the \true" elastic scattering (often called coherent{elastic) there
are contributions coming from interactions where the nucleus is no longer left
in the ground state, but still no elementary inelastic hadron{nucleon inter-
action occurred. Therefore such interactions do not give rise to direct par-
ticle production (particle emission could always occur during the following
nuclear deexcitation process), leaving the target nucleus in an excited state.
Naively speaking, quasi{elastic scattering stems from elastic hadron{nucleon
interactions which leaves the nucleus excited; for this reason it is also called
incoherent{elastic. From an experimental point of view, quasi{elastic scatter-
ing is dicult to discriminate from genuine elastic scattering, since high energy
hadron{nucleon scattering, as we already discussed, is strongly forward peaked
and the resulting momentum transfer is often too small to be discriminated
from the \true" elastic one within experimental resolutions. From a practical
point of view, quasi{elastically scattered hadrons behave in a very similar way
to elastically scattered ones and both are not the dominant ingredients of high
energy cascades which are mostly ruled by absorption reactions.
Absorption reactions are those reactions where at least one hadron{nucleon
non{elastic interaction (! particle production) occurs. It is common practice
among high energy physicists to quote the absorption cross section as the
\reaction" or \non{elastic" cross section, even though the last two terms should
be referred to all non-elastic channels including those contained in the quasi{
elastic cross section. Let us dene the absorption cross section as:

hA abs(s)  hA T (s) hA el (s) ZhA qe (s) = hA T (s) hA f (s) =
= hA r (s) hA qe (s)  d2~b hA abs (~b; s) (102)

At this stage the equivalence of eq. 102 with our denition of absorption cross
section is still to be proved.
65
Absorption Cross Section and Multiple Collisions
Using eqs. 98,100 and 102
Z h i
hA abs(~b; s) = d3~u j i (~u)j2 2Re ~ hA (~b; s) j ~ hA(~b; s)j2 =
Z h i
= d3~u j i (~u)j2 1 jS~hA(~b; s)j2 (103)
Using eq. 79 and the denition of r given in eq. 194 in Appendix B:
Y
A
jS~hA (~b; s)j2 = jShN (~b ~rj ?; s)j2 =
j =1
Y A n h io
= 1 1 jShN (~b ~rj ?; s)j2 =
j =1
Y A h i
= 1 hN r (~b ~rj ?; s) (104)
j =1
one can nally derive the fundamental formula:
Z
hA abs(s)  d2~b hA abs (~b; s) = (105)
Z Z 8 A 9
< Y h i=
= d2~b d3~u j i (~u)j2 :1 1 hN r (~b ~rj? ; s) ;
j=1
The probabilistic interpretation of eq.Q105 is straightforward and consistent
with our denition of hA abs . Indeed Aj=1 [1 hN r ] is the probability that
the projectile impinging along impact parameter ~b will escape inelastic colli-
sions will all A nucleons for the givenQAnucleon conguration. Therefore the
absorption probability is given by 1 j =1 [1 hN r ], and eq. 105 is nothing
else that the average of such probability over all impact parameters and nu-
cleon congurations. The fact that the phase shift additivity hypothesis alone
implies this result is striking and gives a further conrmation of the validity
of the Glauber approach.
Let us assume that the ground state wave{function, i , can be expressed
as a product of independent particle wave{functions, ji:
2A 3 0A 1
Y X
i (~r1 ; :::;~rA) = 4 ji (~rj )5  3 @ ~rj A (106)
j =1 j =1
66
 j (~rj )  ji (~rj )
ji(~rj ) (107)
Z
d3~r j (~r) = 1 (108)
The delta function expresses the centre-of-mass constraint for the A nucleon
positions. For all nuclei but the lightest ones, such term can be safely neglected:
for the lightest nuclei there are exact ways of keeping track of it using shell
model wave{functions, and approximate ones for heavier nuclei (for details
see 138). Under these approximations, and writing (using eq. 194):
hN r (~b; s)  hN r (s)
hN (~b; s)
~b; s)j2 hN r (~b; s)
hN = 1 jShN ((s) =  (s) (109)
hN r hN r
where, by construction, hN (~b; s) is real and:
Z
d2~b hN (~b; s) = 1 (110)
expression 105 for hA abs transforms into:
A h
Y i
hA abs(~b; s) = 1 1 hN rj (s)Trj (~b; s) (111)
j =1
Z
Trj (~b; s)  d3~r j (~r) hN j (~b ~r? ; s) =
Z +1 Z
= dz d2~s j (z;~s)hN j (~b ~s; s) (112)
1
d3~r = dz d2~s = dz sds d
where the thickness function for non-elastic reactions, Trj , has been introduced.
Note that the index j has been explicitly written in this case to stress both
that densities can be dierent for dierent nucleons (i.e. for light nuclei when
using shell model densities), and that hN r and hN depend on the nucleon
type (proton or neutron). Trj (~b; s) obviously represent the amount of nuclear
matter seen by the incident hadron travelling along the impact parameter ~b,
when properly folded with its prole function.
The thickness function has an important normalization property, that de-
rives from the normalization of j and hN , and from the property that the
integral of a convolution of two functions is the product of the integrals. This
67
result is a direct consequence of two properties of the Fourier transforms,
namely that the transform of a n-dimensional convolution is the product of
the transforms, and that the n-dimensional integral of a funtion is equal to its
transform evaluated in the origin. Therefore:
Z Z +1 Z Z
d2~b Trj (~b; s) = dz d2~b d2~s j (z;~s) hN j (~b ~s; s) =
1
Z +1 Z Z 
= dz d2~s j (z;~s) d2~s hN j (~s; s) =
Z 1
= d3~r j (~r) = 1 (113)
For  and  depending only on the modulus of ~r and ~b ~s respectively,
which is a very common situation, and omitting the dependence on the center-
of-mass squared energy s to prevent confusion with ~s:
Z +1 Z p p
Trj (b) = dz d2~s j ( z2 + s2 ) hN j ( b2 + s2 2bs cos ) =
Z +11 Z p
= dz d2~s 0 j ( z2 + b2 + s02 2bs0 cos ) hN j (s0 ) (114)
1
Eq. 114 is a useful starting point for practical calculations. Simple exponential
amplitudes like those depicted in Appendix C are well adequate in order to
parameterize hadron{nucleon scattering, and to make calculations rather com-
fortable. In order to further understand how multiple collisions are described
by eq. 105, let us make the further assumption that j does not depend on
the nucleon index, and that hn r = hp r  hN r . Then (again omitting the
dependence on the centre-of-mass energy):
hA abs(b) = 1 [1 hN r Tr (b)]A =
= [1 hN r Tr (b) + hN r Tr (b)]A [1 hN r Tr (b)]A =
X A  
= A [ T (b)] [1  T (b)]A  +
hN r r hN r r
 =0 
[1 hN r Tr (b)]A =
X A  A

=  [hN r Tr (b)] [1 hN r Tr (b)]A  (115)
 =1
A  
X
= A Pr (b) [1 Pr (b)]A 
 =1 
68
 X
A
= Pr  (b)
 =1
Pr (b)  hN r Tr (b)
Pr  (b)  A P  (b) [1 Pr (b)]A  (116)
 r
Z +1 Z p p
Tr (b) = dz d2~s ( z2 + s2 )hN ( b2 + s2 2bs cos ) =
Z +11 Z p
= dz d2~s 0 ( z2 + b2 + s02 2bs0 cos )hN (s0 ) (117)
A 1

 = !(AA! )!
Since Pr (b) is the probability of getting one specic nucleon hit (for our as-
sumptions it is the same for all nucleons) and there are A possible trials,
Pr  (b) is exactly the binomial distribution for getting  successes out of A
trials, with probability Pr (b) each. Therefore the absorption cross section is
just the integral in the impact parameter plane of the probability of getting
at least one non-elastic hadron-nucleon collision. The average number of non-
elastic hadron-nucleon for a given impact parameter b can be easily obtained
observing that we have A independent trials with probability Pr (b) each:
< (b) >= A Pr (b) (118)
and the overall average number of collision is given by:
R d2~b < (b) >
<  > = R 2~
d b hA
Z abs (b)
A
=  hN r d2~b Tr (b) = A 
hN r (119)
hA abs hA abs
where the integral has been estimated according to eq. 113. Therefore the
result 73 has been demonstrated under very general assumptions in the frame-
work of the Glauber model. The extension of the previous equations to the
case of dierent densities and/or cross sections for protons and neutrons is
straightforward and left to the reader. The nal result for the average number
of collisions reads:
<  >= Zhpr + Nhn r (120)
hA abs
In order to make a quick evaluation of Tr (b), which can be required in
order to compute hA abs or for example in a MonteCarlo model to select the
69
number of collisions occuring for a randomly sampled impact parameter, one
can resort to the so called nuclear thickness approximation (sometimes also
called optical limit). It consists in making the substitution hN (s0 ) !  2 (0)
in eq. 114 and 117, which is justied whenever hN takes signicant values
over a spatial extension << of the nuclear radius. According to Appendix C,
the typical spatial extent for hN is of the order of 1 fm, and therefore this
approximation is reasonably justied for medium and heavy nuclei. In this
case:
Z +1 p
Tr (b)  dz ( z2 + b2) (121)
1
with an obvious geometrical interpretation.
When computing absorption cross sections a small correction 143;144, known
as inelastic screening, should be taken into account. This correction arises be-
cause expression 80 contains only contributions from elastic hadron-nucleon
scattering. There are additional contributions in second and higher order dia-
grams from intermediate states in which the incoming hadron is diractively
excited on one target nucleon, and reverts to its ground state in a later inter-
action.
The same approach used to derive the absorption cross section can be
extended at least partially to the total one, making the substitutions:
hN j (~b) ! hN j (~b)
hN rj ! 1 2i0 hN Tj
Trj (~b) ! TTj (~b)
where the new quantities are dened by:
~
hN (~b)  (1 2 ihN)(b) (122)
0 hN T
0  Ref hN(0)
Imf (0) (123)
8hN A 9
< Y h  i=
hA T (~b) = 2 :1 Re 1 (1 i0 ) hN2 Tj TTj (~b) ; (124)
j=1
Z
TTj (~b)  d3~r j (~r) hN j (~b ~r? ) =
Z +1 Z
= dz d2~s j (z;~s) hN j (~b ~s) (125)
1
70
The denition of hN is such that, using the denition 193 of Appendix B, it
holds:
Z
d2~b hN T (~b) = 1
Im hN (0) = 0
In general the imaginary part of hN can be dierent from zero for b 6= 0.
hN is real for every ~b when the ratio of the imaginary-to-real part of the
hadron-nucleon scattering amplitude is constant and equal to 0 for whichever
momentum transfer. The simplied scattering amplitudes described in Ap-
pendix C fulll this requirement. Whenever hN is not completely real little
can be done in expanding eq. 124 in a way similar to the one adopted for the
absorption cross section. However the real part of high energy hadron scatter-
ing amplitudes is usually small and therefore can be safely neglected. Under
this approximation (0  0, hN (~b) real) it is easy to show that:
A  h
X i h iA 
hA T (b) = 2 A

hN T
2 TT (b) 1 hN2 T TT (b) (126)
 =1
XA  
= 2 A P  (b) [1 P (b)]A 
 T T
 =1
X
A
= PT  (b)
 =1
PT (b)  hN T T (b)
2 T 
PT  (b)  2 A PT (b) [1 PT (b)]A  (127)
Z +1 Z p p
TT (b) = dz d2~s ( z2 + s2 ) hN ( b2 + s2 2bs cos ) =
Z +11 Z p
= dz d2~s 0 ( z2 + b2 + s02 2bs0 cos ) hN (s0 ) (128)
1
where the meaning of the various terms is obvious. Of course the physical
interpretation of eq. 126 is much less obvious than in the absorption case. A
part the overall factor of 2, It describes a multiple scattering expansion, where
individual scatterings are governed by hN T =2. This is a direct consequence
of the wave nature of the whole problem and in particular of the optical theo-
rem, since the total cross section is linked to the forward scattering amplitude
71
and not to an amplitude squared. In the nuclear thickness approximation
( hN (s0 ) !  2(0)), it holds:
Z +1 p
TT (b) = Tr (b)  dz ( z2 + b2)
1
It is worthwhile to point out that for whichever b:
PT (b)  Pr (b)
The proof of this inequality is straightforward in the nuclear thickness approx-
imation, for a purely imaginary amplitude with zero phase shifts l (in this
case hN T  hN r  hN T =2).
Diagram Interpretation of Multiple Collisions
Most of the interest of the Glauber multiple collision model lies in the possibil-
ity of formulating it as a eld theory (the so called Glauber-Gribov model 139)
based on general principles such as unitarity and analyticity. The various mul-
tiple collision terms can be shown to be in a one-to-one correspondence with
the various Feynman graphs describing the interaction. In this approach, for
example, at high energies where the scattering amplitude can be assumed to
be dominated by exchanges in the t-channel with the quantum numbers of vac-
uum (the so called pomeron, see Appendix C), the projectile partons exchange
one or more pomerons according to the multiple scattering formula, with one
or more target nucleons.
In the DPM language, neglecting the contributions from higher order di-
agrams (those called multiple soft collisions in the DPM paragraph) that is
considering at most one pomeron exchange with each target nucleon, a colli-
sion where the projectile is undergoing interactions with n target nucleons is
described by 2n chains, out of which, two result from the combination of the
projectile valence quarks with the valence quarks of one target nucleon (!
2 valence-valence chains), and 2(n 1) chains are stretched among 2(n 1)
couples q q of sea quarks of the projectile and the valence quarks of the
remaining 2(n 1) target nucleons (! 2(n 1) sea-valence collisions).
The whole procedure for describing hA absorption interactions can be
summarized as follows, striclty following what can be done in a MonteCarlo
implementation:
 an impact parameter b is randomly selected over the geometrical area of
the target nucleus
72
  the distribution for the number of collision is then computed according
to eq. 116, and the actual number of collisions randomly sampled from
such distribution. If no collision is selected the procedure restarts from
the rst step
 the position and Fermi momenta of the hit nucleons are sampled
 the momentum fractions of the projectile and target partons as well as
chain ends are selected according to DPM prescriptions. Sea{valence
chains which result in too small invariant masses are discarded
 all chains are hadronized into stable hadrons or resonances
One of the most critical points is connected with the treatment of low
energy chains. The Glauber approach has widely demonstrated its validity
at projectile energies as low as few hundreds MeV in the laboratory for what
concerns cross section calculations and elastic scattering distributions. How-
ever the multiple scattering regime is approached only at much higher energies.
Indeed the number of primary interactions extracted from the experiments sat-
urates at the level given by Glauber approach (<  > is fairly independent on
energy, due to the slow variation of both hN and hA ) at several tens of GeV.
At lower energies the average number of collisions start to decrease, eventually
approaching one at few GeV (see for example ref 145 where exp. data for p-Ta
interactions at 8 GeV/c are reported and discussed). This behaviour is not
in contradiction with the Glauber approach, but rather re ects that more and
more interactions with decreasing energy do not possess enough invariant mass
to give rise to particle production, and do not show up as real interactions,
therefore reducing the eective number of primary collisions. Models can ten-
tatively take into account these eects by setting proper thresholds for the
minimum energy carried by a chain and for the minimum momentum fraction
carried by a parton. These thresholds are a sort of free parameters, which
re ect our ignorance in treating the whole business with nite masses and en-
ergies, and they are usually set in such a way to achieve agreement with the
experimental data in the transition region from ordinary INC (with just one
primary interaction) to the full Glauber cascade.
5.12 Formation Zone
It has been known for many years that a plain treatment of INC at high en-
ergies, on the basis of elementary hadron-nucleon cross sections, overestimates
measured particle yields, if the incident energy exceeds a few GeV 22;21. In par-
ticular, experiments support the eect of some mechanism in limiting the rein-
teractions of energetic particles. In emulsion experiments for example, while
73
Figure 17: Shower, grey, and black tracks multiplicities for  (left) and protons (right) inci-
dent on emulsion, as a function of the projectile momentum. Open symbols are experimental
data, full symbols are fluka results

shower particle multiplicities increase steadily with the energy of the primary
particle, the multiplicities of grey and black tracks rapidly saturate at few tens
of GeV, and stay constant (see g. 17, the exp. data are taken from 146;147).
The physical mechanism which is believed to be at work in limiting high energy
secondary reinteractions in nuclei, is the so called \formation zone" 92;148;149.
This concept has a strong analogy with the Landau-Pomeranchuk-Migdal 150
(LPM) eect which reduces electron bremsstrahlung and photon pair produc-
tion at very high energies.
Naively, it can be understood considering that hadrons are composite ob-
jects and that the typical time of strong interactions is of the order of 1 fm
(please note that space units are used for time in all this paragraph, x = ct).
If one thinks about the hadrons emerging from an inelastic interaction, it re-
quires some time to them to \materialize" and be able to undergo further
interactions. The amount of time can be qualitatively estimated considering a
secondary particle of mass M and of momentum components pk and pT with
respect to the original projectile direction. Making a Lorentz transformation
to the frame where the longitudinal momentum is null (remember that pT is
invariant under Lorentz transformations), the uncertainty principle tell us:

t =
t  Eh = p 2 h 2 (129)
T pT + M
74
This time interval can be translated into the particle proper time, , and the
lab frame time, t, taking care of the Lorentz dilation among the various frames:
 = EM t = p2 h+MM 2 (130)
T T
tlab = EElab t = EMlab  = p2hE+lab
M2 (131)
T T
The time interval in the lab frame can be also expressed as a function of the
particle rapidity, y, by:
tlab = tcosh y = p 2 h 2 cosh y (132)
pT + M
If such an interaction takes place inside a nucleus, the condition for having
(possible) reinteractions of our secondary particle is expressed by :
v
t  RA  r0A 31 (133)
Such an equation can be used to dene a critical rapidity, above which particles
have no chance to reinteract and will \materialize" already outside the nucleus.
Inserting typical values for pions inside eq. 132, ET  0:3 GeV, E  10 GeV,
it can be easily checked that the formation time can easily exceed heavy nuclei
radii. All this derivation must be taken with a lot of care: among all other
uncertainties, it is not clear whether the particle entering the formulae must
be the nal one, or rather the \mother" resonance. Nevertheless the concept is
very powerful, and using a scale factor for the formation time as a free param-
eter, very nice agreement for hadron-nucleus and nucleus-nucleus interactions
have been obtained 149;151. Examples of complete models including all eects
described above can be found in 152;153.
6 Elastic Nuclear Interactions
The elastic scattering of nucleons on nuclei is normally described in terms of
the optical model. An optical potential as described in sec. 5.3 is used to model
the nucleus as a scattering centre. The potential is most frequently chosen as a
sum of a central and a spin-orbit part, with a Woods-Saxon radial dependence
for the real part, and a more periphery-peaked shaped imaginary part. The
parameters of the optical potential are determined by tting the existing data,
after corrections for the Coulomb eect on proton scattering: they are found
to vary both with A and with projectile energy.
75
 The behaviour of the dierential elastic cross section exhibits a shape sim-
ilar to a diraction pattern, and becomes more strongly forward peaked as the
projectile energy increases. In rst approximation, this pattern can be repro-
duced with the expression of the scattering from a grey disk. The formulae
relevant for scattering from a grey disk can be obtained from Appendix B,
for a purely real prole function (! a purely imaginary scattering amplitude),
and assuming that:
(b) = 1 a; b  R
(b) = 0; b > R
where R is the scattering sphere radius, and a gives the opacity of the sphere.
Under these assumptions it is easy to show starting from the expressions re-
ported in Appendix B, that:

d = (1 a)2R2 J1(2kR sin =2) 2

d
2 sin=2 (134)
and:
el = R2(1 a)2
r = R2(1 a2 )
T = 2R2(1 a)
For complete absorption a = 0, and the reaction cross section becomes equal
to the geometrical one, and one half of the total one.
The same diractive pattern is found in pion-nucleus scattering. At res-
onance, the diraction minima and the total cross section can be very well
described by the black disk formulation with a = 0, due to the strong pion-
nucleus interaction.
At higher energies (> 1 GeV), the pion and nucleon elastic scattering on
nuclei is more and more forward peaked: the rst minimum of eq.134 is at
  0:61 pRh , that means that for a 1 GeV nucleon on Carbon it is situated at

about 12 . Eventually, the usual exponential form becomes appropriate. One
remark to be made is that these forward-peaked scatterings involve very small
energy losses, thus the importance of elastic scattering on particle propagation
becomes increasingly small at high energies.
Elastic Scattering in the Glauber Model
The formulae developed in the previous paragraphs can be used to derive a
closed expression for the elastic scattering amplitude in the Glauber approach.
76
With a bit of algebra, and using approximation 106 with  and hN scattering
amplitudes not depending on the nucleon type, it can be shown that:
ik Z 
fhA el (~q; s) = 2 d2~bei~q
~b 1 +
 1 Z 0 ~
A )
1 2ik d2~q 0 e i~q
b fhN(~q 0 ; s)F(~q 0) (135)

where F is the nuclear form factor, that is the Fourier transform of the nuclear
density: Z
F (~q 0) = d3~rei~q 0
~r (~r) (136)
In case of charged particle scattering, the previous expression should be prop-
erly modied to take into account interference with Coulomb scattering 138.
Similar expressions can be derived also for the quasielastic scattering ampli-
tude.
Experimental examples of high energy hadron-nucleus elastic scattering
and of its analysis in terms of Glauber scattering can be found in 130;131;132.

6.1 Examples of hadron-nucleus cross sections

Figure 18: Total, elastic and inelastic cross Figure 19: Total, elastic and inelastic cross
section for negative pions on Carbon section for neutrons on Copper
77
Actually hA cross sections at intermediate and high energies resemble the
behaviour of the corresponding hN ones (see gs. 1 and 3), and this is not
unexpected due to the strong relationship between hA and hN (see para-
graph 5.11). The main dierences are in the resonance region, where cross
section features are smoothed by the Fermi motion, and at very high energies
where the increase of hA cross sections is slower than in the corresponding hN
ones. This feature can be easily explained in the Glauber model, observing
that both total and absorption cross sections are determined by the proba-
bility of having at least one collision in the corresponding multiple scattering
expansions. Moderate increases in the elementary hN cross section are there-
fore eective only for large impact parameters where this probability is small,
while for more central collisions they simply result in an increase of the average
number of collisions rather than in an increase of the cross section. Typical
behaviours of cross sections for neutrons on Copper and for negative pions
on Carbon are reported in gs. 18 and 19, together with some representative
experimental data.

7 Practical Examples of (G)INC calculations

In the previous sections, a brief description of the processes occurring during
hadronic inelastic interactions in the energy range of interest has been given.
In the present section, a few representative comparisons between model results
and experimental data on particle production in the energy range of interest are
presented. All the presented results have been obtained with the last version
of the fluka 40;41;27;42 code, and most of them rely on the model used for the
intermediate energy range 27 and called peanut (PreEquilibrium-Approach-
to-NUclear-Thermalization). As already explained, this model combines both
an INC part, and a preequilibrium part. Nuclear potential eects (refraction
and re ection) are modeled into the code, as well as quantum eects, like Pauli
blocking, nucleon-nucleon correlations, fermion antisymmetrization, coherence
length and formation zone. The results obtained with such a code are not at
all representative of the majority of the INC models, usually based on some
avour of the Bertini INC code 21;29;30, or similar approaches. Among the
available INC codes, possibly only cem92 28 and the lahet version with the
preequilibrium model included 26 can be compared with this model for nucleon
induced interactions. For pion induced interactions, the approach of peanut
is very dierent from all other mentioned codes, none of which includes a
complex optical potential or absorption eects computed according to modern
approaches 5;6;7 including three body absorption (for example, cem92 154 de-
rives 2-body absorption cross sections just multiplying the deuteron one by a
78
Figure 20: 90 Zr(p,xn) at 80.5 MeV, plain Figure 21: 90 Zr(p,xn) at 80.5 MeV, plain
INC (see text) calculation INC plus preequilibrium (see text) calcula-
tion

Figure 22: 90 Zr(p,xn) at 80.5 MeV, peanut Figure 23: 90 Zr(p,xn) at 80.5 MeV, full
(see text) calculationwith no quantum eect, peanut (see text) calculation
but Pauli blocking

tunable factor, and uses a square well potential of xed depth for pions).
79
Figure 24: 90 Zr(p,xp) at 80.5 MeV, plain Figure 25: 90 Zr(p,xp) at 80.5 MeV, plain
INC (see text) calculation INC plus preequilibrium (see text) calcula-
tion

Figure 26: 90 Zr(p,xp) at 80.5 MeV, peanut Figure 27: 90 Zr(p,xp) at 80.5 MeV, full
(see text) calculationwith no quantum eect, peanut (see text) calculation
but Pauli blocking

7.1 From plain INC to more Complex Models

In paragraph 3.3 a list of known weaknesses of INC codes has been presented.
Here some examples are given, trying to illustrate the sources of such decien-
80
Figure 28: 208 Pb(p,xn) at 25 MeV, peanut Figure 29: 208 Pb(p,xn) at 45 MeV, at several
calculation emission angles

Figure 30: 12 C(p,xn) at 256 MeV, at several Figure 31: Pb(p,xn) at 800 MeV, at several
emission angles emission angles

cies. Examples of the problems met by classical INC codes based on the plain
Bertini model or similar approaches can be found in 36, PSI and Julich contri-
butions, where the results of an intercomparison carried by the Nuclear Energy
81
Figure 32: 209 Bi(p,xp) at 62 MeV, angle in- Figure 33: 90 Zr(p,xn) at 80.5 MeV, angle in-
tegrated spectrum preequilibrium only tegrated spectrum preequilibrium only

Agency are reported. Results on 208Pb(p,xn), and 90Zr(p,xn), 90Zr(p,xp) re-

actions where the topic of the intercomparison. The equivalent set of data
computed with peanut , 1994 status, can be found in 155. It is important to
remind that the Bertini model does not include re ection or refraction, neither
includes any preequilibrium stage. Furthermore an average binding energy is
used throughout all cascading process without account for actual Q values of
the dierent reaction channels.
In order to illustrate the eect of the various ingredients, the same projec-
tile-target combination, 80.5 MeV protons on 90Zr, has been computed under
dierent assumptions and the results compared with experimental data 156;157.
Four dierent trials have been performed, always using peanut with all or
only a few of the ingredients at work. The results of the exercise are presented
in gs. 20,21,22,23, for (p,xn), and in gs. 24,25,26,27, for (p,xp). In all these
gures as well as in following ones, experimental data are plotted as full symbols
joined by a line, while model results are given or by symbols with error bars,
or by histograms with shaded areas representing the statistical error.
The rst attempt (gs. 20,24) has been performed using a plain INC ap-
proach with no preequilibrium stage, no refraction and re ection, and only
Pauli blocking activated among the quantum eects. The transport threshold
for secondary nucleons was set equal to the average binding energy. As can be
easily seen results are very poor, with an overestimation of the forward peak
82
and a strong underestimation of backward angle emission. These results have
been obtained with assumptions which are quite consistent with those used by
the Bertini model, but the binding energy which, as usual in peanut , has
been correctly set at each emission stage to match the actual reaction Q-value.
This is the explanation, maybe paradoxically, of the slightly worse results of
this trial at medium/large angles with respect for example to the PSI contri-
bution in 36 . The use of a constant binding energy value in 36 overestimates
the available ejectile energy of about 6.9 MeV, shifting the spectra of the same
amount towards larger energies.
The second trial does not yet use refraction and re ection, but the preequi-
librium stage is there. There is a clear improvement in the results, particularly
at backward angles, but still there are large discrepancies.
The third trial has been performed switching on all the refraction and
re ection business, but still limiting the quantum eects to Pauli blocking
alone. There is a great improvement, with still discrepancies at the forward
angles, but a reasonable description of backward ones. The reason for these
last deciencies is related to the eect of particle curvature in the nucleus which
both increases the particle track length and hence the reinteraction probability,
and at the same time \pushes" particles towards the nucleus centre, again
increasing the interaction probability.
The fourth and last trial has been performed with all quantum eects
on, that is with coherence length, nucleon correlation, and fermion antisym-
metrization eects on. Clearly these are eective in increasing particle mean
free paths in the nuclear medium resulting in a quite reasonable description of
the whole spectrum at all angles (see also 93 for the eect of nucleon correla-
tions).
The eect of particle curvature on interaction rates are typical of INC codes
which include refraction and re ection 23. The reason for the good results, at
least for angle integrated spectra, of INC models based on straight trajectories
(like the Bertini INC) probably lies in the compensating eect of the lack
both of mechanisms able to lengthen particle mean free paths, and of the
curvature eect which operates in the opposite direction. The price is the
poor description of angular distributions, at least for energies not much larger
than typical potential energies. Such a picture is consistent with what has
been already reported in the literature 31;33;34;35.
Other examples of complete calculations of neutron double dierential
spectra are reported in gs. 28,29,30, and 31. The experimental data have
been taken from ref. 33;35;158;160. It is particularly interesting to note that in
the peanut model at 25 MeV incident energy no INC stage is performed and all
the emission is due to preequilibrium. At 45 MeV the INC stage starts to con-
83
 Al Au,Bi
absorption cr. sect., mb

absorption cr. sect., mb

Pion Energy (MeV) Pion Energy (MeV)

Figure 34: Pion absorption cross section on Figure 35: Pion absorption cross section on
Aluminum as a function of energy Gold or Bismuth as a function of energy

tribute (compare g. 29 with ref. 35 to appreciate the improvement over plain

INC models), while at higher energies it dominates the most energetic part of
the spectra. Neutron double dierential spectra following C(p,xn) at 256 MeV
(exp. data from 33), and Pb(p,xn) at 800 MeV (exp. data from 158;160) are pre-
sented in gs. 30,31. In the latter gure three experimental curves are drawn,
two come from 160 and correspond to two possible choices for the eciencies
of the scintillators used in the experiment, the third one comes from 158. The
relative disagreement among these curve can give an idea of the experimental
uncertainties in this energy range. An inspection of the literature data for
113, 256, 597, and 800 MeV proton energies 31;33;158;159, shows how \classical"
INC are more and more reliable with increasing energy. However, it must be
taken into account that the onset of the eects discussed in paragraph 5.10,
and the need for physical models other than those based on isobar production
and decay (see paragraph 4.4,4.5), limit the validity range to a maximum of
1.5{3 GeV.
Two examples of angle integrated spectra computed with the preequilib-
rium model alone (no INC) are reported in g. 32, and 33, for 209Bi(p,xp) at
62 MeV and 90 Zr(p,xn) at 80.5 MeV (exp. data from refs. 34 and 156 respec-
tively). Again a comparison with the Bertini results reported in the original
reference shows the superior capabilities of preequilibrium models in this en-
ergy range. Examples about pion absorption are presented in gs. 34,35
and g. 36. The former two gures show the experimental 162;161 and calcu-
lated absorption cross section for incident + and  on Aluminium, and on
84
Figure 36: 58 Ni(+ ,xp) at 160 MeV, at sev- Figure 37: Double dierential distribution
eral emission angles. Symbols exp. data of negative pions in the reaction p+Be at
from 163 164 , dahsed histogram peanut cal- 730 MeV. Dashed histograms are peanut re-
;

culations sults, points are experimental data 165

Figure 38: Angular distribution of positive (right) and neutral (left) pions following
58 Ni(+ ,+ x) and 58 Ni(+ ,0 x)

Bismuth and Gold. The comparison is made on an absolute scale, since the
calculated results have been obtained making use of the capability of peanut
(as of all other INC models) of computing reaction cross sections. The latter
85
gure presents the double dierential emission spectrum of protons, following
+ interactions on 58Ni at 160 MeV (exp. data from 163;164). The contribution
of absorption in the energetic part of the spectra is evident.
An example of pion production at intermediate energy is reported in g. 37,
where calculated and experimental 165 double dierential negative pion yields
are compared for the reaction 9 Be(p, x) at 730 MeV. An example of pion
inelastic and charge exchange scattering in the
region is reported in g. 38,
where the calculated and experimental 166;167 angular distributions of positive
and neutral pions for 58Ni(+ ,x) are shown.
Finally two examples for high energy interactions, besides those already
discussed in the previous sections, are presented in gs. 39 and 40. The former
presents the multiplicity distribution of negative shower particles emitted by
250 GeV positive kaons on Aluminium and Gold targets 134: this distribution
if of course mainly sensitive to the details of the Glauber cascade undergone by
the projectile into the target nucleus. The latter is instead related to intranu-
clear cascading and shows the mutual correlations of black and grey tracks 168.

Figure 39: Multiplicity distribution of neag- Figure 40: Mutual correlations ( < n > vs
tive shower particles for 250 GeV/c K+ on n and < n > vs n ) between black and grey
g

b b g
Aluminium and Gold targets. Symbols exp. charged tracks for 400 GeV/c p on emulsion.
results 134 , histo fluka results Full symbols are exp. data from 168

7.2 Residual Nuclei

The problem of a reliable description of individual isotopes produced in spal-
lation reactions is still an open one. While the general features of the resid-
ual nuclei distributions are usually well reproduced by INC with the possible
86
 Cross Sec.(mb)
Residual nuclei distribution

A
Cross Sec.(mb)

Figure 41: Residual nuclei mass distribution. Experimental data are from 169 for silver,
and 170 for gold.

exception of fragmentation products, (see paragraph 5.8), predictions on indi-

vidual isotopes can be out of large factors. Many reasons contribute to this
uncertainty:
 Slight inaccuracies in excitation energy spectra reaching the evaporation
stage can results in small shifts in particle multiplicities, but in substan-
tial shifts among close isotope yields
 The lack of spin and parity dependent (Hauser-Feshbach) calculations in
the last stage can also bias the very last steps and hence the nal residual
nucleus
 Charged particle emission channels of low probability can be badly re-
produced because of sub-barrier eects etc
 No prediction about metastable states is possible
87
  Fragmentation is usually not included
Besides these physical reasons, there are also \technical" ones. What is often
required when looking for residual nuclei production, is the production cross
section of some specic isotopes, for example because of its toxicity or long life
time etc, which however can be a very small fraction of the total reaction cross
section. Therefore a good agreement over the gross part of emission spectra is
not at all a proof of a similar agreement on some specic and relatively unlikely
isotope production channel. Sticking however at our previous statement, that

Fission Cross Section

Cross Section (mb)

E(GeV)
Cross Section (mb)

E(GeV)

Figure 42: Neutron induced ssion cross sections on Uranium. Experimental data are
from 171
is that the general features of isotope production are reasonably reproduced,
in g. 41 the computed and measured mass distributions of residuals after
300 GeV proton interactions on Silver, and 800 GeV protons on Gold are
shown. The agreement is fairly good in the spallation region close to the
target mass, and still reasonable down to very light masses where the lack of
fragmentation in the model shows up clearly.
Fission models developed for INC codes often contain \ad hoc" parameters
88
 Isotope Production
Cross Section (mb)

MeV

Figure 43: Residual nuclei excitation functions. Experimental data are from 114 172 with
;

various symbols, open stars are peanut results

to adjust the calculations, like the unphysical reduction factor that was applied
to the ssion width in the original Atchison work 155 for the Bertini model.
This excitation-dependent factor was introduced in hetc to cut o the ssion
process at high excitation energies and bring the calculations in agreement
with measured data. Our explanation is that an INC without preequilibrium
emission leads to an average overestimation of the nuclear excitation energy
at the equilibrium stage. This is not the case for peanut , and the agreement
with experiment is nice without any arbitrary factor, as shown in g. 42 for
Uranium (exp. data from 171).
Regarding light nuclei, the inclusion of Fermi Break{up allows reasonable
predictions about residual nuclei also for light nuclei, which are known to be
\dicult" targets. In g. 43 the excitation functions for the production of
dierent isotopes by proton bombardment of Carbon are shown (exp. data
from 114;172. The evaporation model alone could never have explained the
abundance of isotopes like 7 Be, while it would have grossly overestimated the
89
Figure 44: Residual nuclei distributionproduced by proton cascades on Pb and U at dierent
energies. Only residuals produced by all particles but neutrons below 20 MeV are presented.
The abscissa is the neutron excess, N-Z, and the ordinate the atomic number Z. The stability
line is also shown.

emission of neutrons  particles.

As an example of the inventory of residual nuclei produced by whole high
energy cascades, in g. 44 the residual nuclei distributions produced by pro-
tons of 1.5 GeV on a depleted Uranium target, and of 1.5, 15, and 150 GeV
on a natural Lead target are presented. The isotopes shown, are only those
produced by particles other than low energy neutrons (those below 20 MeV).
A few important features can be seen in these plots, taking into account
that neutron-rich isotopes sit to the right of the stability line, and proton-rich
90
ones to the left:
 The ssion zone in the U plot, and to a lesser extent also in the Pb,
1.5 GeV one
 The spallation zone, with the characteristic peak close to the original
target A and Z, populated mainly by proton-rich isotopes
 The increasing extension of the spallation zone with increasing energy
The production of proton-rich isotopes is one of the distinct features of spal-
lation induced by energetic particle. It is mainly due to the reduced emission
probability of charged particles in the evaporation stage, due to the Coulomb
barrier. A last word must be spent about the life times of these impressive
inventories of nuclides: a good fraction of them are very short half-life ones,
created relatively far from the stability line. Codes able to model the evolution
with time of such inventories are therefore required for a proper evaluation of
incineration rates or other eects of interest. However a bit of caution must
always be taken when considering INC code results on residual nuclei, partic-
ularly for isotopes far from the original target.
Acknowledgments
The authors are greatly indebted to A. Fasso and J. Ranft for the many stim-
ulating discussions and the continuous support, and to dr. G. Reo for the
invitation to give this lecture.
Appendix A: useful kinematical quantities
It is customary in high energy physics to make use of a few \strange" kine-
matical variables to describe particle kinematical properties. Let us assume
to deal with a particle, c, of mass M, total energy E, and momentum
~p = (p sin  cos ; p sin sin; p cos ), possibly emerging from an interaction
of the kind a + b ! c + X. From, now variable with an asterisk (e.g. E  ; p )
are supposed to be evaluated in the Centre of Mass System (CMS) of the col-
lision, variables labelled with L are suppose to be computed in the lab frame
(that is the frame where particle b is at rest), and variables with no particular
labelling can be in whichever frame. Four-momenta are indicated by tilded
characters, and are dened by:
p~a  (Ea ; ~pa ) (137)
p~a
p~b = Ea Eb ~pa
~pb (138)
91
 p~a2  p~a
p~a = Ma2 (139)
The invariant mass squared of the system, s, (ps  the total energy of the
system in the CMS frame) is given by:
s = (Ea + Eb )2 = (Ea + Eb)2 (~pa + p~b )2 = Ma2 + Mb2 + 2Mb EaL (140)
In particular, for a+b ! c+d, it is customary to use the Mandelstam invariants:
s  (~pa + p~b)2 = (~pc + p~d )2 (141)
t  (~pa p~c )2 = (~pd p~b)2 (142)
u  (~pa p~d )2 = (~pc p~b)2 (143)
which are related by:
s + t + u = Ma2 + Mb2 + Mc2 + Md2 (144)
The 4-momentum transfer t in case of elastic scattering is simply related to
the 3-momentum transfer ~q = ~pc ~pa and to the scattering angle  in the
centre-of-mass frame by:
t = ~q 2 = 2p 2(1 cos  ) (145)

q = 2p sin 2 (146)
The Lorentz transformation required to go from the original frame to an-
other one moving with velocity ~v with respect to the rst one, are given by:
~ = ~vc (147)
= q 1 (148)
1 ~2
~ = ~ ; ( 2 ~ 2 = 1) (149)
p~ 0 = ~p ~ E ~+
p~1 (150)
E 0 = E ~
~p (151)
The inverse transformation is readily obtained exchanging non-primed vari-
ables with primed ones, and ~ ! ~ . In particular, for a system of particles
92
with total energy Etot, and total momentum ~ptot, the transformation to the
Center of Mass System can be accomplished using:
CM = Eptots (152)
~CM = ~pptots ; (s = Etot
2 ~ptot2 ) (153)
The single particle inclusive cross section for particle c is dened as:
3 p  p 
inv = Ec dd3~pab s; pk; pT  Fabc s; pk; pT (154)
c
where the labels a; b; c are there just to remind that this is the inclusive distri-
bution of particle c after an interaction with particle a impinging on particle
3
b. Such a distribution is often called the invariant cross section, since dEp~ can
be shown to be Lorentz invariant, and therefore:
3 3
Ec dd3~p = Ec dd3~p (155)
c c
The transverse momentum pT and the transverse mass ET or MT (which are
all invariant under Lorentz transformation), are dened by:
p
pT = p sin  = E 2 M 2 sin  (156)
q2
MT = ET = pT + M 2 (157)
The longitudinal momentum, pk is instead frame dependent. A very useful
quantity connected with longitudinal momentum is the so called x Feynman
variable, dened as the ratio of the particle longitudinal momentum and the
maximum possible longitudinal momentum, in a given frame, usually assumed
to be the CMS (but sometimes also the lab frame is used):
pk 2p
xF = p (Max)  psk (158)
k
Please note that 1  xF  1. The regions xF > 0 and xF < 0 are usually
called the projectile fragmentation region, and the target fragmentation region,
the region xF  0 is called the central region. The hypothesis of limiting frag-
mentation 53 supposes that in the projectile and target fragmentation regions
the following scaling holds (see eq. 154):
p  p !1 c 
Fabc s; pk ; pT s! Fab (xF ; pT ) (159)
93
A stronger version of the scaling hypothesis, due to Feynman, assumes that
such a scaling holds also in the central region. As a consequence Fabc (0; pT ))
is energy independent, and the produced particle multiplicity can be shown to
scale like < nab >/ n0 +n1 ln s. Both predictions are violated, but they can be
used as order of magnitude guesses without too much danger. Other variables
very frequently used are the rapidity, y (dy is again Lorentz invariant, that is
ys dier in dierent frames only of a constant factor), and the pseudorapidity,
, whose major (and unique) advantages are that it can be easily measured
and it coincides with rapidity for pT  M
E pk 1  E + pk 
y = acosh( M ) = atanh( E ) = 2 ln E p (160)
k
 T  1  p + pk 
 = ln tan 2 = 2 ln p p (161)
k

Appendix B: partial waves and cross sections

Let us start from the standard partial wave expansion of the scattering ampli-
tude f for a spinless scattering:
X
1 2il
fcm (t; s) = k1 (2l + 1)Pl (cos ) l e 2i 1 (162)
cm l=0
where cm indicates the centre of mass system, kcm is given by
kcm = k = pcm h = 
1 (163)
and l , l are real with 0  l  1. The last condition follows from ux
conservation and it is often called the unitarity condition. The physical inter-
pretation of l 's (often called inelasticities) is the following: they are linked to
the fraction of the incident wave which survives the scattering (therefore they
must be  1) or conversely 1 l2 is the fraction of the incident ux which is
removed by inelastic processes. Indeed l ! 0 maximizes the reaction cross
section for the lth partial wave, while l ! 1 implies purely elastic (or charge
exchange when possible) scattering. Introducing:
2il
al = l e 2i 1 = Sl 2i 1 (164)
Real = l sin22l = Im(S
2
l) (165)
94
 Imal = 1 l2cos 2l = 1 2ReSl (166)
(167)
the expansions for the scattering amplitude and the total, elastic and reaction
cross sections, T , el and r = T el , can be written as:
X
1
fcm () =  (2l + 1)al Pl (cos ) (168)
l=0
X
1
=  (2l + 1) Sl 2i 1 Pl (cos )
l=0
X
1 2il
=  (2l + 1) l e 2i 1 Pl (cos )
l=0
X
1
T = k4 Imfcm (0) = 4 2 (2l + 1)Imal (169)
cm l=0
X
1
= 2 2 (2l + 1)(1 ReSl )
l=0
X
1
= 2 2 (2l + 1)(1 l cos 2l )
l=0
Z X
1
el = d
jfcm()j2 = 4 2 (2l + 1)jal j2 (170)
4 l=0
X
1
=  2 (2l + 1)j1 Sl j2
l=0
2X
1
=  (2l + 1)(1 + l2 2l cos 2l )
l=0
X
1
r = T el = 4 2 (2l + 1)(Imal jalj2) (171)
l=0
X
1
=  2 (2l + 1)(1 jSl j2 )
l=0
2X
1
=  (2l + 1)(1 l2)
l=0
When a large number of partial waves are involved, it is convenient to re-
95
sort to the so called impact parameter representation, which is readily obtained
from the previous formulae with the following substitutions:
  
lim P (cos )
l>>1;or<<1 l
= J0 (2l + 1) sin 2 (172)
l + 12 1
b = k = (l + 2 ) (173)
f() ! f(q; s) (174)
al ! a(b; s) (175)
l ! (b) = e Im(s;b) (176)
l ! (b) = Re(s;
2
b) (177)
hence
kb = l + 12
(2l + 1) sin 2 = qb
(2l + 1)Pl (cos )  2kb J0 (qb)
X
1 Z1 Z1
! dl = k db
l=0 0 0
Z1
f(q; s) = k d2 b J0 (qb) a(b; s) (178)
Z1
0
h i
= ik bdb J0 (qb) 1 ei(b;s)
0
where we used the momentum transfer q (for a denition of q and of the four-
momentum transfer t see Appendix A), and the complex phase shift (b; s),
according to:
i(b;s)
a(b; s) = e 2i 1 (179)
and we made explicit the dependence on the centre{of{mass squared energy s.
Furthermore, making use of one of the many explicit denitions of the Bessel
function J0 Z 2
1
J0 (z) = 2 eiz cos  d (180)
0
96
one can write:
ik Z h i
f(~q; s) = 2 d2~b ei~q
~b 1 ei(~b;s) ; d2~b = bdbd (181)
where ~b is a vector perpendicular to the incident momentum ~k which is the
analogue of the classical impact parameter between the incident particle and
the scattering centre, and the assumption ~q
~b = qb cos  has been made.
Eq. 181 is usually called the impact parameter representation of the scattering
amplitude. The relations between q,
and t are the following (being  and 
the azimuthal and polar angle of the scattered particle):
d
= d cos d = 2d sin2 2 d = qdqd k2 (182)
d
=  (183)
dt Zk2 Z
el (s) = d
jf j2 = k12 d2 q jf(~q; s)j2 (184)
in which d2~q = qdqd is a two-dimensional element of integration in the plane
perpendicular to ~k. It is also customary to introduce the prole function (~b; s),
dened by:
(~b; s) = 1 ei(~b;s) (185)
Comparing eqs. 181 and 185, it is obvious that the prole function is the
scattering amplitude in impact parameter space, that is the Fourier transform
of the scattering amplitude f(~q; s):
(~b; s) = 2ik 1 Z d2~q e i~q
~b f(~q; s) (186)
Z
ik d2~b ei~q
~b (~b; s)
f(~q; s) = 2 (187)
Introducing the analogue of Sl :
S(~b; s) = ei(~b;s) = (~b; s)e2i(~b;s) (188)
the following relations dening the S-matrix approach, hold;
Z
S(s) = d2~b ei(~b;s) (189)
S(~b; s) = 1 Z (~b; s) (190)
ik d2~b ei~q
~b h1 S(~b; s)i
f(~q; s) = 2 (191)
97
and the unitarity condition translates into:
jS(~b; s)j  1 (192)
where the equality holds when all inelasticities go to zero, that is Im(~b; s) = 0,
while in general eq. 192 simply states that part of the incident ux is removed
by the inelastic channels.
The expressions for the total, elastic and reaction cross section can be
readily obtained from 187 or 191:
Z Z
T (s) = 2 d2~b Re (~b; s)  d2~b T (~b; s) (193)
Z
h i Z 2~ ~
2~ ~
r (s) = d b 1 jS(b; s)j  d b r (b; s)
2 (194)
Z Z
el (s) = d2~b j (~b; s)j2  d2~b el (~b; s) (195)
The physical interpretation of T , r , and el is rather obvious: they represent
the probability of undergoing an interaction, or an inelastic one or an elastic
one, for a particle impinging along the impact parameter ~b.
Assuming that /S depend on b and not on its azimuthal orientation (as it
is for most practical situation, or anyway when averaging over the initial spins
and summing over the nal ones), the scattering amplitude can be written as:
Z1
f(q; s) = ik bdb J0 (qb) (b; s) (196)
Z01
f(q; s) = ik bdb J0 (qb) [1 S(b; s)] (197)
0

Appendix C: Toy Model of High Energy Hadron-Nucleon Scattering

Whenever the elastic scattering distribution is a simple exponential function
of the 4-momentum transfer t, ddtel / exp (Bsl t), the scattering amplitude for
high energy hadron-nucleon scattering can be written as (in this paragraph the
natural unit system h = c = 1 will be used unless explicitly indicated):
f(t; s) = [(s) + i] k4
T(s) Bsl2(s)t
e
~ s) = [(s) + i] T(s) e Bsl2(s)t
f(t; (198)
4
 s) = s[(s) + i]T(s) e Bsl2(s)t
f(t; 4
98
which fullls by construction the various possible denitions of the optical
theorem:
T (s) = 4 k Imf(0; s) =
= 4 Im~f(0; s) =
= 4 
s Imf (0; s)
and where the slope parameter, Bsl , and the  parameter are dened in the
usual way:
Bsl (s)  d [ln (ddt
el =dt)] (t = 0)

Ref(0; s)
(s)  Imf(0; s)
The dierential and integrated elastic scattering cross sections are given
by:
del (t; s) = 2
dt [1 + 2 (s)] 16T (s) eBsl (s)t = (199)

k2 jf(t; s)j
= 2

= ~ s)j2
jf(t;

kZ 2 jf(t; s)j
= 2
0 T2 (s)
el (s) = dt deldt(t; s) = [1 + 2 (s)] 16B
1 sl (s)
Therefore, summarizing, the following relations among the elastic cross section,
the total cross section, the ratio, , of the real{to{imaginary part of the forward
scattering amplitude, and the slope parameter, Bsl , hold:
s
T (s) = 16B sl (s)el (s)
1 + 2 (s) (200)

el (s) = [1 + 2 (s)] 16B T2 (s) (201)

sl (s)
Bsl (s) = [1 + 2 (s)] 16 T2 (s) (202)
el (s)
s
(s) = 2 (s)
T 16Bsl (s)el (s) (203)
T2 (s)
99
Applying eq. 186 of Appendix B, the prole function for the scattering am-
plitude of eq. 198 can be readily obtained as (t = q2 = 2k2 (1 cos ) for
elastic scattering):
T(s) e 2Bbsl2(s)
(b; s) = [1 i(s)] 4B (204)
sl (s)
Therefore the gaussian scattering amplitude in the 3-momentum transfer q
described by eq. 198 translates into a gaussian prole function in the impact
parameter b, with their respective variances given by (with now all c and h
factors):
q = p 1 (205)
Bsl (s)
p
b = (hc) Bsl (s) (206)
For example, a typical slope parameter of the order of 10 [GeV/c] 2 corre-
sponds to an eective interaction radius of 0.7 fm, which gives us the order of
magnitude of the \range" of the interaction between two hadrons. The S(b; s)
function for our simplied scattering amplitude would read (eq. 190):
S(b; s) = 1 [1 i(s)] 4B T(s) e 2Bbsl2(s) (207)
sl (s)
Finally the total, elastic and inelastic scattering probabilities for a given impact
parameter b, according to eqs. 193,195,194 can be written as:
2 b2 3
e 2Bsl (s) 5
T (b; s) = T (s) 4 2B (208)
sl (s)
2 b2 3
e Bsl (s) 5
el (b; s) = el (s) 4 B (209)
sl (s)
2 b2 3 2 b2 3
e 2Bsl (s) e Bsl (s) 5
r (b; s) = T (s) 4 2B (s) 5 el (s) 4 B (210)
sl sl (s)

where all terms within square brackets are normalized to unity when integrated
over d2~b.
Suppose now that the total cross section for high energy hadron{nucleon
scattering can be expressed as a function of the squared centre-of-mass energy
100
s as:
X  s i X
T (s) = ~i  Ti (s)
s0 (211)
i i
where s0 is a scale factor, with a corresponding scattering amplitude f given
by, as a function of s and of the 4-momentum transfer t = q2 :
X X Ti(s)
f(t; s) = fi (t; s) = k 4 (i + i)i (t; s)
i  s i i
X k ~i s0
 s  (t)
= 
(i + i)i (t) s (212)
i 4 0

where the functions (t) and (t) satisfy:

Rei (0) = 1
Imi (0) = 0
Im i (t) = 0
i (0) = 0 (213)
Under these assumptions, eqs. 211 and 212 automatically fulll the optical
theorem, and the real-to-imaginary ratio of the forward scattering amplitude,
(s), is given by:
P  Imf (0; s) P   (s)
(s)  Ref(0; s)
Imf(0; s) =
Pi Imf
i i
(0; s) =
Pi i Ti(s) (214)
i i i Ti
where
i  Ref i (0; s)
Imf (0; s) (215)
i
are assumed to be independent of s.
Using dispersion relations it is possible to derive an approximate explicit
expression for the ratio of the real-to-imaginary forward scattering ampli-
tude 173:
1
   d 
   (s) tan 2 dlns t(s) =
T "
1 X1   2k+1 d2k+1  (s) #
T
=  (s) 2 2k+1 (216)
T k=0 dln s
101
which gives the following result when applied to Ti :
 
i = tan i 2 (217)
Eqs. 211,212, and 217 have been introduced in such a way, that they match
the corresponding results of Regge theory 57 for the scattering of two hadrons
mediated by n Reggeons IR or pomerons IP, with trajectories, i (t), (see 57
for the denition of trajectory) given by:
i(t)  0i + 0i t = 1 + i + i (t)
Therefore, identifying the trajectory intercept, 0i, with 1 + i , and i (t) with
0it, these equations describe Regge amplitudes and cross sections.
Regge theory applications to high energy scattering are based on crossed
symmetry between reactions occurring with exchange of mediators in the t-
channel, a + b ! c + d, and the crossed reaction, a + c ! b + d, occuring
via s-channel intermediate states. The same scattering amplitude is supposed
to describe both processes when suitable prolonged under analyticity require-
ments into the unphysical region of one process, which corresponds to the
physical region of the crossed one. Crossed symmetry is at the basis of the
duality concept.
High energy elastic scattering is supposed to be mediated by the exchange
of a variety of neutral mesons with the proper quantum numbers. However,
the trajectory associated with these mesons (Reggeons) has a typical inter-
cept 0IR  0:5 and thereforepgives rise to cross sections which decrease with
energy approximately like 1= s. The asymptotic behaviour of all hadronic
cross sections is indeed dominated by a trajectory, the pomeron, with the vac-
uum quantum numbers and intercept slightly larger than one, 0IP = 1 + IP ,
IP << 1, which is not associated to any physical particle. The exact na-
ture of the Pomeron is still a matter of debate. The name comes from the
observation that this trajectory is dominating the high energy behaviour of
cross sections, and since it has the vacuum quantum numbers it couples in the
same way to a particle and its antiparticle, therefore assuring the fulllment of
the Pomeranchuk theorem which states that asymptotically particle and an-
tiparticle cross sections must be equal. These simple formulae give astonishing
results when used to t all known hadron-hadron cross sections with very few
parameters (see 174 for details).
References
1. R. Brun et al., CERN DD/EE/84-1(1987).
102
 2. O.G. Ryazhskaya, report LNGS 94/110 1994.
3. R. Bonetti, A.J. Koning, J.M. Akkermans, and P.E. Hodgson, Phys. Rep.
247, 1 (1994).
4. E. Gadioli, and P.E. Hodgson, Pre-equilibrium Nuclear Reactions,
(Clarendon Press, Oxford, 1992).
5. M.J. Vicente Vacas, and E. Oset, Nucl. Phys. A568, 855 (1994).
6. E. Oset, and L.L. Salcedo, Nucl. Phys. A468, 631 (1987).
7. L.L. Salcedo et al., Nucl. Phys. A484, 557 (1988).
8. J. Aichelin, Phys. Rep. 202, 233 (1991) and references therein.
9. K. Niita et al., Phys. Rev. C52, 2620 (1995).
10. A. Ferrari, and P.R. Sala, Improvements to the Electromagnetic part of
the FLUKA code, to be published; P. A. Aarnio, A. Fasso, A. Ferrari,
J. -H. Moehring, J. Ranft, P.R. Sala, G.R. Stevenson, and J.M. Zazula,
Proc. of the MC93 Int. Conf. on Monte Carlo Simulation in High-
Energy and Nuclear Physics, Feb. 22-26, 1993, (Ed. P. Dragovitsch, S.L.
Linn, M. Burbank, World Scientic, Singapore, 1994) 100.
11. M. Herman, and G. Reo, Recent Progress in Nuclear Reaction Theory
and Models, lecture given at the Workshop on Nuclear Reaction Data
and Nuclear Reactors - Physics, Design and Safety, International Centre
for Theoretical Physics, May 15{17 1996, Trieste, these proceedings.
12. P. Young, E.D. Arthur, and M.B. Chadwick, Comprehensive Nuclear
Model Calculations: Theory and Use fo the GNASH Code, lecture given
at the Workshop on Nuclear Reaction Data and Nuclear Reactors -
Physics, Design and Safety, International Centre for Theoretical Physics,
May 15{17 1996, Trieste, these proceedings.
13. P. Young, Comprehensive Nuclear Model Calculations: Theory and Use
fo the GNASH Code - Part II, lecture given at the Workshop on Nuclear
Reaction Data and Nuclear Reactors - Physics, Design and Safety, Inter-
national Centre for Theoretical Physics, May 15{17 1996, Trieste, these
proceedings
14. M. Blann, Recent Progress and Current Status of Preequilibrium Reac-
tion Theories and Computer Code ALICE, Proc. of the Workshop in
Computation and Analysis of Nuclear Reaction Data Relevant to Nu-
clear Energy and Safety, Truste, 1992, (M.K. Mehta and J.J. Schmidt
eds., World Scientic, Singapore 1993) 517.
15. A.J. Koning, and J.M. Akkermans, Phys. Rev. C47, 724 (1993).
16. M. Blann, in Intermediate Energy Nuclear Data: Models and codes, Proc.
of a specialits' meeting, Issy-les-Moulineaux, May 30th - June 1st 1994,
(published by OECD/NEA 1994), p. 23.
17. M.B. Chadwick, P.G. Young, and F.S. Dietrich, in Intermediate Energy
103
 Nuclear Data: Models and codes, Proc. of a specialits' meeting, Issy-
les-Moulineaux, May 30th - June 1st 1994, (published by OECD/NEA
1994), p. 67.
18. A. Koning, O. Bersillon, and J.P. Delaroche in Intermediate Energy Nu-
clear Data: Models and codes, Proc. of a specialits' meeting, Issy-les-
Moulineaux, May 30th - June 1st 1994, (published by OECD/NEA 1994),
p. 87.
19. A. Ferrari and P.R. Sala, Physics of Showers induced by Accelerator
Beams, in Proceedings of the 1995 Frederic Joliot Summer School in Re-
actor Physics, August 22{30 1995, Cadarache, France, (CEA ed., Vol. 4
1996), Lecture 5b.
20. N. Metropolis et al., Phys. Rev. 100, 185 (1958); N. Metropolis et al.,
Phys. Rev. 100), 204 (1958).
21. H.W. Bertini, Phys. Rev. 131, 1801 (1963).
22. V.S. Barashenkov et al., Sov. Phys. Usp. 16, 31 (1973); V.S.
Barashenkov et al., Nucl. Phys. A222, 204 (1974).
23. K. Chen et al., Phys. Rev. 166, 949 (1968); K. Chen et al., Phys. Rev.
176, 1208 (1968).
24. R. Serber, Phys. Rev. 72, 1114 (1947).
25. M.L. Goldberger, Phys. Rev. 74, 1269 (1948).
26. R.E. Prael and M. Bozoian, Los Alamos report LA-UR-88-3238 (1988).
27. A. Ferrari, and P. R. Sala, Proc. of the MC93 Int. Conf. on Monte-Carlo
Simulation in High-Energy and Nuclear Physics, Tallahassee, Florida,
22-26 february (1993), (World Scientic ed. 1994), p. 277; A. Fasso,
A. Ferrari, J. Ranft, and P. R. Sala, Proc. of the Specialists' Meeting
on Shielding Aspects of Accelerators, Targets & Irradiation Facilities,
Arlington, April 28-29 1994, (published by OECD/NEA 1995), p. 287;
A. Fasso, A. Ferrari, J. Ranft, and P. R. Sala, in Intermediate Energy
Nuclear Data: Models and codes, Proc. of a specialist' meeting, Issy-
les-Moulineaux, May 30th - June 1st 1994, (published by OECD/NEA
1994), p. 271.
28. S.G Mashnik and S.A. Smolyansky, JINR preprint E2-94-353 , Dubna
(1994)
29. H.W. Bertini, Phys. Rev. C1), 1801 (1970).
30. H.W. Bertini, Phys. Rev. 6(2), 631 (1972).
31. M.M. Meier, D.A. Clark, C.A. Goulding, J.B. McClelland, G.L. Morgan,
and C.E. Moss, Nucl. Sci. and Eng. 102, 310 (1989).
32. M.M. Meier et al, Nucl. Sci. and Eng. 104, 339 (1990).
33. M.M. Meier, W.B. Amian, C.A. Goulding, G.L. Morgan and C.E. Moss,
Nucl. Sci. and Eng. 110, 289 (1992).

104
34. F.E. Bertrand, and R.W. Peelle, Phys. Rev. C8, 1045 (1973).
35. A. Galonsky et al., Phys. Rev. C14, 748 (1976).
36. International Code Comparison for Intermediate Energy Nuclear Data,
(M. Blann, H. Gruppelaar, P. Nagel and J. Rodens eds., OECD/NEA
1994).
37. G.Q. Li, R. Machleidt, and Y.Z. Zhuo, Phys. Rev. C48, 1062 (1993).
38. G.Q. Li, and R. Machleidt, Phys. Rev. C48, 1702 (1993).
39. G.Q. Li, and R. Machleidt, Phys. Rev. C49, 566 (1993).
40. A. Fasso, A. Ferrari, J. Ranft, P. R. Sala, G. R. Stevenson, and J. M. Za-
zula, Proc. of the workshop on Simulating Accelerator Radiation Envi-
ronment, SARE, Santa Fe, 11-15 january (1993), (A. Palounek ed., Los
Alamos LA-12835-C 1994), p. 134.
41. A. Fasso, A. Ferrari, J. Ranft and P. R. Sala, Proc. of the IV Inter-
national Conference on Calorimetry in High Energy Physics, La Biodola
(Elba), September 19-25 1993, (A. Menzione and A. Scribano eds., World
Scientic 1994), p. 493.
42. A. Fasso, A. Ferrari, J. Ranft and P. R. Sala, Proc. of the 2nd work-
shop on Simulating Accelerator Radiation Environment, SARE2, CERN-
Geneva, October 9{11 1995, Yellow report CERN in press.
43. P. Cloth et al., HERMES, a Monte Carlo program system for beam-
materials interaction studies, Report KFA/Jul{2203 (1988).
44. K.C. Chandler, and T.W. Amstrong, Operating instructions for the
high-energy nucleon-meson transport code HETC, ORNL{4744 (1972);
K.C. Chandler, and T.W. Amstrong, Nucl. Sci. and Eng. 49, 110
(1972).
45. R.G. Alsmiller, F.S. Alsmiller and O.W. Hermann, Nucl. Instr. Meth.
A295, 337 (1990).
46. R. E. Prael and H. Lichtenstein, LA-UR-89-3014 Los Alamos (1989).
47. W. O. Lock and D. F. Measday, Intermediate Energy Nuclear Physics
(Methuen & co, London 1970).
48. Phase shift solutions KH78, KH80, in Landolt-Bornstein , new series ,
Vol. 9, part II, (Springer-Verlag 1983).
49. C.J. Joachain, Quantum Collision Theory (North-Holland, Amsterdam
1975).
50. K. Hanssgen, and J. Ranft, Nucl. Sci. Eng. 88, 537 (1984).
51. J.Bystricky et al., Energy dependence of nucleon-nucleon inelastic total
cross sections, DPhPE 87-03 (1987).
52. B.J. VerWest, and R.A. Arndt, Phys. Rev. C25, 1979 (1982).
53. J. Benecke, T.T. Chou, C.N. Yang, and E. Yen, Phys. Rev. 188, 2159
(1969).
105
54. M.E Law et al., A Compilation of Data on Inclusive Reactions, Internal
Report LBL-80 (1972).
55. A. Capella, U. Sukhatme, and J. Tran Thanh Van, Z. Phys. C3, 329
(1980); A. Capella, and J. Tran Thanh Van, Phys. Lett. B93, 146
(1980).
56. A. Capella et al., Phys. Rep. 236, 225 (1994).
57. P.D.B. Collins, An Introduction to Regge Theory & High Energy Physics,
(Cambridge University Press, Cambridge 1977).
58. T. Sjostrand, CERN Report CERN-TH 6488/92 (1992).
59. S. Ritter, Comput. Phys. Commun. 31, 393 (1984); J. Ranft, and S. Rit-
ter, Acta Phys. Pol. B11 259 (1980).
60. A.B. Kaidalov, and O.I. Piskunova, Z. Phys. C30, 141 (1986).
61. A. Capella, A. Kaidalov, A. Kouider Akil, C. Merino, and J. Tran Thanh
Van, Z. Phys. C70, 507 (1996).
62. K. Goulianos, Phys. Rep. 101, 169 (1983).
63. S. Roesler, R. Engel, and J. Ranft, Z. Phys. C59, 481 (1993).
64. J. Ranft, and S. Roesler, Z. Phys. C62, 329 (1994).
65. K. Hahn, and J. Ranft, Phys. Rev. D41, 1463 (1990).
66. F.W. Bopp, R. Engel, D. Pertermann, and J. Ranft, Phys. Rev. D49,
3236 (1994).
67. P. Aurenche et al., Phys. Rev. D45, 92 (1992).
68. A. Kaidalov, Phys. Lett. B117, 459 (1982); A. Kaidalov, and K.A. Ter-
Martirosyan, Phys. Lett. B117, 247 (1982).
69. J. Ranft and S. Ritter, Z. Phys. C27, 413 (1985).
70. M. Adamus, Z. Phys. C39, 311 (1988).
71. A. Ferrari, P.R. Sala, J. Ranft, and S. Roesler, Z. Phys. C70, 413 (1996).
72. A. Ferrari, P.R. Sala, J. Ranft, and S. Roesler, Z. Phys. C71, 75 (1996).
73. A. Fasso, A. Ferrari, and P. R. Sala, Proceedings of the 8th Interna-
tional Conference on Radiation Shielding, Arlington, April 24-28 1994,
(American Nuclear Society ed., 1994), Vol. 2, 643.
74. J. J. Grin, Phys. Rev. Lett. 17, 438 (1966).
75. K. Kikuchi and M. Kawai, Nuclear Matter and Nuclear Interactions,
(North-Holland, Amsterdam, 1968).
76. M.E. Grypeos, G.A. Lalazissis, S.E. Massen and C.P. Panos, J. Phys.
G17, 1093 (1991).
77. L.R.B. Elton, Nuclear Sizes, (Oxford University Press, Oxford 1961).
78. W.D. Myers, Nucl. Phys. A204, 465 (1973).
79. W.D. Myers, Droplet Model of Atomic Nuclei, (IFI/Plenum Data Com-
pany, New York 1977).
80. A. deShalit and H. Feshbach Theoretical Nuclear Physics, Vol. 1: Nu-
106
 clear Structure, (Wiley 1974).
81. A.G. Sitenko and V.K. Tartakovskij,lektsii po teorii iadra, (Atomizdat,
Moskow 1972).
82. A. Bohr, B.R. Mottelson, Nuclear Structure Vol. 1, (W.A. Benjamin,
Inc. 1969).
83. T. Ericson and W. Weise, Pions and Nuclei, (Clarendon Press, Oxford,
1988).
84. M. Krell and T.E.O. Ericson, Nucl. Phys. B11, 521 (1969).
85. H. J. Weyer, Phys. Rep. 195, 295 (1990).
86. J.N. Ginocchio and M.B. Johnson, Phys. Rev. C21, 1056 (1980).
87. M.B. Johnson and H.A. Bethe, Nucl. Phys. A305, 418 (1978).
88. M.B. Johnson and B.D. Keister, Nucl. Phys. A305, 461 (1978).
89. K. Stricker, H. McManus, and J.A. Carr, Phys. Rev. C19, 929 (1979).
90. J.N. Ginocchio, Phys. Rev. C17, 195 (1978).
91. K. Chen et al., Phys. Rev. 166, 4, 949 (1968).
92. L. Stodolski, Proc. of the Vth Intern. Coll. on Multiparticle Reactions,
Oxford, 577 (1975).
93. K. Chen et al., Phys. Rev. C4, 2234 (1971).
94. H. Feshbach, A. M. Kerman, S. Koonin, Ann. Phys. 125, 429 (1975).
95. M. Blann, Ann. Rev. Nucl. Sci. 25, 123 (1975).
96. M. Blann, Phys. Rev. Lett. 27, 337 (1971).
97. M. Blann, Phys. Rev. Lett. 28, 757 (1972).
98. M. Blann and H.K. Vonach, Phys. Rev. C28, 1475 (1983).
99. M. Blann, Phys. Rev. C28, 1648 (1983).
100. S. Shlomo, Nucl. Phys. A539, 17 (1992).
101. G. Mantzouranis, D. Agassi, and H. A. Weidenmuller, Phys. Lett. 57B,
220 (1975).
102. J. M. Akkermans, H. Gruppelaar and G. Reo, Phys. Rev. C22, 73
(1980).
103. V.F. Weisskopf, Phys. Rev. 52, 295 (1937).
104. A.S. Iljinov and M.V. Mebel, Nucl. Phys. A543, 517 (1992).
105. S.G. Mashnik, Acta Phys. Slov. 43, 86 (1993).
106. I. Dostrowsky, Z. Fraenkel and G. Friedlander, Phys. Rev. 116, 683
(1959).
107. A.V. Ignatyuk, G.N. Smirenkin and A.S. Tishin, Sov. J. Nucl. Phys. 21,
255 (1975).
108. A.V. Ignatyuk et al., Sov. J. Nucl. Phys. 21, 612 (1975).
109. A. Gilbert and A.G.W. Cameron, Can. J. Phys. 43, 1446 (1965).
110. R. Vandenbosh and J. R. Huizenga, Nuclear Fission, (Academic Press,
New York 1973).
107
111. N Bohr and J.A. Wheeler, Phys. Rev. 56, 426 (1939).
112. F. Atchison, Meeting on Targets for neutron beam spallation sources, (ed.
G. Bauer, KFA Julich Germany, Jul-conf-34 1980).
113. E. Fermi, Prog. Theor. Phys. 5, 1570 (1950).
114. M. Epherre and E. Gradsztajn, J. Physique 18, 48 (1967.
115. A.S. Botvina et al., Nucl. Phys. A475, 663 (1987).
116. S. Leray et al., Nucl. Phys. A511, 414 (1990).
117. W. G. Lynch, Ann. Rev. Nucl. Part. Sci. 37, 493 (1987).
118. J. Richert, report CRN 93-11.
119. X. Campi et al., Phys. Lett 142, 8 (1984).
120. L. G. Moretto and G. J. Wozniak, Ann. Rev. Nucl. Part. Sci. 43, 379
(1993).
121. P.A. AArnio and M. Huhtinen, Proc. of the MC93 Int. Conf. on Monte
Carlo Simulation in High-Energy and Nuclear Physics, Feb. 22-26, 1993,
(Ed. P. Dragovitsch, S.L. Linn, M. Burbank, World Scientic, Singapore,
1994), p. 1.
122. G.A. Bartholomew et al., Advances in Nuclear Physics 7, 229 (1975).
123. I. Bergqvist and N. Starfelt, Progress in Nuclear Physics 11, 1 (1970).
124. D.H. Wilkinson, in Nuclear Spectroscopy B, (ed. F. Ajzenberg-Selove ,
Academic Press, New York 1960).
125. P.M. Endt, Atomic Data and Nuclear data Tables 26, 46 (1981).
126. W.Dilg, W.Schantl and H.Vonach, Nucl. Phys. A217, 269 (1973).
127. H. Morinaga and T. Yamazaki, In-Beam Gamma-Ray Spectroscopy,
(North Holland 1976).
128. G.L. Morgan, report ORNL-5563 (1979).
129. J.K. Dickens , T.A. Love and G. L. Morgan, report ORNL-4847 (1973).
130. G.D. Alkhazov et al., Nucl. Phys.bf A381, 430 (1982).
131. A. Schiz et al., Phys. Rev. D21(11), 3010 (1980).
132. G. Bellettini et al. Nucl. Phys. 79, 609 (1966).
133. C. De Marzo et al., Phys. Rev. D26, 1019 (1982).
134. I.V. Ajinenko et al., Z. Phys. C42, 377 (1989).
135. R.J. Glauber, Phys. Rev. 100, 242 (1955).
136. R.J. Glauber, in Lectures in Theoretical Physics, Vol. 1 (1959).
137. R.J. Glauber, in High Energy Physics and Nuclear Structure, (North
Holland 1967), 311.
138. R.J. Glauber and G. Matthiae, Nucl. Phys. B21, 135 (1970).
139. V.N. Gribov, Sov. Phys. JETP 29, 483 (1969); V.N. Gribov, Sov. Phys.
JETP 30, 709 (1970); L. Bertocchi, Nuovo Cimento 11A, 45 (1972).
140. K. Zalewski, Ann. Rev. Nucl. Part. Sci. 35, 55 (1985).
141. A. Capella, and A. Krzywicki, Phys. Rev. D18, 3357 (1978).
108
142. A. Capella, Nucl. Phys. A25, 133c (1991).
143. V.A. Karmanov and L.A. Kondratuk, JETP Lett. 18, 266 (1973).
144. D. Diamond and B. Margolis, Phys. Rev. D16, 1365 (1977).
145. N. Fujiwara et al., Nucl. Phys. A404, 509 (1983).
146. A. Tufail et al., Phys. Rev. D42, 2187 (1990).
147. S. Fredriksson et al., Phys. Rep. 144, 187 (1987).
148. Z. Pengfei, and C. Weiqin, Nucl. Phys. A552, 620 (1993).
149. J. Ranft, Z. Phys. C43, 439 (1988).
150. L. Landau, I. Pomeranchuk, Dokl. Akad. Nauk SSSR 92, 535 (1953);
A.B. Migdal, Phys. Rev. 103, 1811 (1956).
151. H.-J. Moring, J. Ranft, Z. Phys. C52, 643 (1991).
152. J. Ranft, Phys. Rev. D51, 64 (1995).
153. G. Battistoni, C. Forti, and J. Ranft, Astroparticle Physics 3, 157 (1995).
154. S.G Mashnik, JINR preprint E2-93-470 (1993).
155. Intermediate Energy Nuclear Data: Models and codes, Proc. of a special-
its' meeting, Issy-les-Moulineaux, May 30th - June 1st 1994, (published
by OECD/NEA 1994).
156. M. Trabandt et al., Phys. Rev. C39, 452 (1989).
157. A.A. Cowley et al., Phys. Rev. C43, 678 (1991).
158. W.B. Amian et al., Nucl. Sci. Eng. 112, 78 (1992).
159. W.B. Amian et al., Nucl. Sci. Eng. 115, 1 (1993).
160. S. Stamer et al., Phys. Rev. C47, 1647 (1993).
161. K.Nakai et. al., Phys. Rev. Lett. 44, 1446 (1979).
162. D. Ashery et al., Phys. Rev. C23, 2173 (1981).
163. W.J. Burger et al., Phys. Rev. C41, 2215 (1990).
164. R.D. McKeown et al., Phys. Rev. C24, 211 (1981).
165. D.R.F. Cochran et al., Phys. Rev. D6, 3085 (1972).
166. S.M. Levenson et al., Phys. Rev. C28, 326 (1983).
167. D. Ashery et al., Phys. Rev. C30, 946 (1984).
168. E. Stenlund, and I. Otterlund, Nucl. Phys. B198, 407 (1982).
169. N.Y.T. Porile, G.T. Cole and C.Rudy, Phys. Rev. C19, 2288 (1979).
170. L.Sihver et al., Nucl. Phys. A543, 703 (1992).
171. P.W. Lisowski et al., in Proc. Int. Conf. on Nuclear Data for Science
and Technology, Julich, Germany, (Springer Berlin 1992), p. 732.
172. A.S. Iljinov et al., (Landolt-Bornstein, Springer{Verlag, New Series, Vol.
13a 1991).
173. U. Amaldi, M. Jacon, and G. Matthiae, Ann. Rev. Nucl. Sci. 26, 385
(1976).
174. A. Donnachie, and P.V. Landsho, Phys. Lett. B296, 227 (1992).

109