Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

The continuing exponential increase in computer power together with the recent developments of very efficient numerical procedures allow nowadays to perform predictive atomic-scale computer simulations for material science. This holds especially for advanced microelectronic devices where functional units consists more and more often of 10^6 atoms or even less. In this situation the design of new materials and devices is more and more frequently supported by atomic-scale computer simulations.

A kinetic 3D Monte-Carlo code based on stochastic probabilistic two-center cellular automaton using a double bookkeeping technique, one in the particle vector and the other in the lattice space was originally coded in PASCAL and tested on an INTEL PC, later in C and on HP workstations. This implementation should be speeded up considerably in order to undertake real scientific simulations. The only way to get the needed speedup is parallelisation.

We started with a look at the given sequential code, translated it into C and run it on our S-Class Server. While translating we had to understand a lot of things, e.g. the double booking technique, so that we could do first steps of scalar optimisation during this process.

The most simple idea of a parallel approach is distributing the lattice across the processors. But because of the physical problem modelling the growth of clusters of implanted particles in the lattice such an approach will lead to load imbalances and destroy the effect of parallelisation.

With an another approach we tried to distribute the work done in the particle space instead of the lattice space. We developed a graph-theoretical approach based on skeletons to find out a work distribution across 2^n processors that assure s a good load-balance. The disadvantage of this approach is the work needed to compute the distribution which is of order O(n^3) compared to the simulation which is of order O(n).

At this point it turned out the given model was not suitable for parallelising the algorithm. So we had feed-back with the physicist and together we developed a model of the physical process where the algorithm which solves the problem could be formulated in parallel. The new model allowed us to divide the solution into three main steps, where two of them can be done parallel and will therefor improve the speed of the algorithm.

We implemented this algorithm in a mixed language modularised program using FORTRAN for the computational part and C for the I/O part. The program runs on our S-Class server using maximum 8 processors.

In our presentation the methodology of creating an optimised parallel code for the Monte-Carlo simulation of Ostwald Ripening, the schema of the parallel code and first real physical simulation results will be shown.

In Ion Beam Synthesis (IBS) of nanoclusters the postimplantation
annealing step causes major redistribution of the implanted impurity atoms
by Ostwald ripening (OR). The diffusion-reaction equations describing OR
are highly nonlinear. Therefore, analytical studies are restricted to special
cases like the asymptotic behaviour, the diffusion or reaction controlled
limit, low cluster concentration, etc. An alternative to analytical studies
of OR is the numerical integration of the diffusion-reaction equations.
Using this method we take into account the diffusion and reaction control
of OR, i.e. we present for the first time an unified treatment for the
whole range of and in between the diffusion and reaction controlled limits.
Based on a local mean field theory, our model starts with a multipole expansion
of the concentration field, which satisfies the stationary diffusion equation.
Using appropriate boundary conditions on the precipitates we derive self-consistently
the governing equations for the evolution of the clusters. Alternatively
to Ewald's summation method we apply a Yukawa-like screening of precipitates.
The dependence of the particle radius distribution (PRD) the critical radius
and the density of nanoclusters on process parameters will be discussed.

The goal of this work was to evaluate the potential of small YAP:Ce crystals, especially designed for a high-resolution PET system. We directly compared the scintillator properties of 3x3x20 mm³ crystals YAP with those of BGO. The light output, energy resolution, detection efficiency and timing properties for the irradiation using ¹³⁷Cs and ²²Na sources were investigated. Special consideration was given to the influence of the reflector coating on the light output as well as on the overall performance of the quality of the studied crystals.

Abstract
Relying on the Luttinger-Ward theorem we derive a thermodynamically selfconsistent and scale independent approximation of the thermodynamic potential for the scalar Φ⁴ theory in the tadpole approximation. The resulting thermodynamic potential as a function of the temperature is similar to the one of the recently proposed screened perturbation theory.

Studies on the measurement of turbulent dispersion of a radiotracer in an experimental setup with natural convection liquid-gaseous flow were carried out at the Forschungszentrum Rossendorf. Liquid-gaseous bubbly flow was generated in a narrow tank by injection of pressurized air into water or by catalytic disintegration of H₂O₂. A small amount of a positron emitting radiotracer liquid ( ¹⁸F in an aqueous NaF solution) was injected instantaneously at a certain point of the tank. The Positron Emission Tomography (PET) technique was used to observe the spreading of the tracer liquid in the bubbly flow. A double head gamma detector array was used for measuring the positron annihilation rate which is proportional to the tracer concentration. The dispersion coefficient D for the tracer liquid was calculated from the experimental data assuming an isotropic spreading of the concentration profile after separation of the linear displacement of the maximum concentration point. Calculations of the two-phase flow with spreading tracer were carried out for selected experiments using the computational fluid dynamics code CFX-4. The Euler-Euler continuum approach including a homogeneous low Reynolds number K,? -model was applied for the two-phase flow. Turbulent Prandtl numbers for gas and tracer dispersion were varied. In case of higher gas injection rates into water (superficial gas velocities j_gas of 5 - 15 mm/s), a reasonable agreement was achieved between calculated and measured values of the tracer transport velocity and the dispersion coefficient. A nearly linear correlation between j_gas and D was found in agreement with other authors. However, it was necessary to take into account the significant deviation of the tracer distribution from the ideal diffusion equation solution in some cases, in particular if the tracer has been spread over upwards and downwards regions of the natural circulation velocity field. Further investigations are necessary to improve the agreement between measured and calculated values of the dispersion coefficient and tracer transport velocity in the case of H₂O₂ disintegration and low gas superficial velocities.

Das Schaumverhalten eines Stoffsystems bei Druckentlastungsvorgängen hat einen wesentlichen Einfluß auf die Druckabfallgeschwindigkeit und die ausgetragene Masse. Es wurden kleinmaßstäbliche Druckentlastungsversuche mit einem 1,9 l Gefäß durchgeführt. Das Abblasen erfolgte über eine kopfseitige Abblaseleitung mit einer Blende. Unter sonst gleichen Bedingungen wurde reines Wasser und Wasser mit Zusatz von Isobutanol abgeblasen. Gemessen wurden der Behälterdruck, die ausgetragene Masse, sowie Temperaturen und Dampfgehalte an jeweils 2 verschiedenen Höhenpositionen. Es wurden Experimente mit einem Blendendurchmesser von 2 mm und 3 mm sowie Isobutanolzusätzen von 2, 4 und 6 Gew.% durchgeführt. Für beide Abblasequerschnitte zeigte sich eine deutliche Zunahme der ausgetragenen Masse mit der Isobutanolkonzentration. Auf Grund der Schaumbildung und des dadurch vergrößerten Flüssigkeitsaustrags wird der Druckabbau zu Beginn des Druckentlastungsvorgangs behindert. Löst sich schließlich die Oberkante der Schaumschicht vom Deckel, erfolgt der Druckabbau auf Grund der dann geringeren Masse im Behälter schneller.

Die Einrichtung zur Druckentlastung (Sicherheitsventil, Berstscheibe) ist in der Praxis über eine mehr oder weniger lange Abblaseleitung mit dem Reaktor verbunden. Bei der Auslegung der Sicherheitseinrichtung wird jedoch lediglich der notwendige Ausströmquerschnitt ermittelt. Berücksichtigt wird dabei gegebenenfalls der Druckverlust in der Leitung. Die Auswirkungen auf die Dynamik des Druckentlastungsvorgangs während des zweiphasigen Ausströmens bleiben bisher weitgehend unbeachtet. Aus diesem Grund wurden verschiedene Modelle für die Simulation des gekoppelten Systems bestehend aus Druckbehälter, Abblaseleitung und Entlastungsarmatur aufgestellt. Dabei wurden mehrere Varianten von vereinfachten Leitungsmodellen untersucht. Zur Überprüfung der Modellgüte wurden dabei Ergebnisse von Experimenten genutzt, die an der Druckhalterversuchsanlage DHVA der Hochschule für Technik, Wirtschaft und Sozialwesen Zittau/Görlitz durchgeführt wurden und bei denen Massenstrompulsationen bei Abblasen auftraten. Die Schwingungen hängen sehr empfindlich von einzelnen Parametern des Schaummodells, des Modells für die Kopplung zwischen Behälter und Abblaseleitung und des Modells für die Abblaseleitung ab. Experimentelle Daten zu diesen Schwingungen stellen daher eine ausgezeichnete Grundlage für die Validierung von Codes zur dynamischen Simulation eines Blow-Down Systems dar.

Im Rahmen der Entwicklung eines Gesamtmodells für die dynamische Simulation komplexer Druckentlastungssysteme bestehend aus Reaktor, Abblaseleitung,
Entlastungsarmatur und Auffangeinrichtungen wurde ein 1-D-Behältermodell
aufgestellt. Die Transportvorgänge im Behälter werden auf der Grundlage einer
neu entwickelten Partikelmethode gelöst. Dadurch wird numerische Diffusion
vermieden, was insbesondere bei der Berücksichtigung von Diskontinuitäten, wie
z. B. dem Gemischspiegel, von Vorteil ist. Die implizite Wiedergabe der aktuellen Position des Gemischspiegels sowie ein spezielles Interface ermöglichen die
Beachtung der Entwicklung einer Schaumkrone am Übergang zwischen dem
Zweiphasengemisch und dem Gasraum. Die weitgehende Entkopplung der einzelnen
Phänomene erlaubt eine modulare Codestruktur, bei der Modelle für
Einzelphänomene leicht ausgetauscht werden können.

The EUROBALL Cluster detector is composed of seven encapsulated Ge detectors
in a common cryostat with a total volume of 2000 ccm of HP Germanium. The
development and the performance of the Cluster detector is summarized.
Up to six Cluster detectors were used in pre-EUROBALL experiments at the
S-DALINAC Darmstadt, at the tandem-postaccelerator facility of the MPI-K
Heidelberg and the UNILAC at GSI. Examples of these experiments - the excitation
of dipole modes with (gamma, gamma')-reactions and the first coincidence
spectroscopy of the N=Z nucleus ⁶⁸Se with a CLUSTER Cube - are
discussed. The development of segmented encapsulated Ge detectors for a
MINIBALL at the radioactive beam facility REX-ISOLDE has been launched.
The status of the project is presented.

Fe/Al nm-period multilayers deposited by pulsed laser deposition were investigated by conversion electron Mößbauer spectroscopy and transmission electron microscopy. No Fe-Al intermetallic phases were found in the
intermixed transition range with an Al-concentration of 20-60 at.%. A qualitative depth distribution of the impurity concentration in the transition layers between subsequent Fe and Al layers is suggested.

The safety concept of the new innovative boiling water reactor
SWR1000, which is developed by Siemens AG, is aiming at the
increase of safety margins and of the grace period for operator
actions during accidents by increasing the water inventory
inside the pressure vessel, by the arranging large water pools
in- (core flooding pool) and outside of the containment (dryer-
separator storage pool), simplifying the (emergency) cooling
systems and further decreasing the failure probabilities of
safety systems. Therefore, active safety systems are substituted
by passive ones or combined with passive safety systems in cases
where a replacement is not possible. The function of passive
systems is directly based on the physical phenomena: gravity,
natural convection and/or evaporation. Their effectiveness is in
dependent of operator actions. Therefore, smaller failure rates
are assigned to passive safety systems. The probability of
severe accidents of the new SWR is reduced significantly.

For the experimental investigation of the operation mode and the
effectiveness of these passive safety systems the multipurpose
thermohydraulic test facility NOKO was constructed at the
Forschungszentrum Jülich (FZJ) within a German collaboration
sponsored by the German Federal Ministry of Education, Sciences,
Research and Technology (BMBF) and German Utilities (EVU). The
facility has a maximum power of 4 MW for steam production and a
maximum operating pressure of 10 MPa.

In the EU BWR R&D Cluster six test series with an emergency
condenser test bundle and four test series with a single tube
were performed in 1996. Within the Physics and Thermalhydraulics
Complementary Actions to the BWR Cluster (BWR/CA) the Forschungs
zentrum Rossendorf (FZR) e.V. has performed 10 post test calcula
tions and an additional blind test calculation of NOKO bundle
experiments. The tests were selected by FZJ. The results of
these calculations are presented in this paper.

Post test calculations of NOKO experiments were performed with
an improved version of ATHLET. To calculate the heat transfer
coefficients during condensation in horizontal tubes the module
KONWAR has been developed and implemented in ATHLET. KONWAR is
based on the flow regime map of Tandon and includes several semi-
empirical correlations for the determination of the heat trans
fer coefficients. The post test calculations presented here
prove qualitatively and quantitatively the expected operation
mode and the capacity of the emergency condenser. Additionally
the blind calculation proves, that the improved version of
ATHLET (ATHLET with the extension KONWAR) is able to perform
proper calculations.

Control elements of VVER-440 type reactors consist of absorber and fuel parts. They can approximately be considered as spatial double pendulums. The length of both parts is equal to the height of the core. In the present paper the main interest is directed to the neutron noise due to random vibrations of the fuel parts (lower parts of the pendulums). It is assumed, that the fuel parts of all control elements are completely inserted into the core. A homogeneous cylindrical reactor model is used to describe the neutron flux. It is also assumed, that the control elements can be excited to flow induced random 3D transverse oscillations and pendulum motions. The transfer functions for the considered two kinds of control element motions were calculated by means of the Prompt-Response-Approximation for the Modified-One-Group-Diffusion-Equation. Using numerical results, studies have been made to clear up the contribution of the random pendulum motions to the neutron noise.

A finite-element-model describing the mechanical vibrations of the whole
WWER-440 primary circuit was established to support the early detection of mechanical component faults. A special fluid-structure module was developed to consider the reaction forces of the fluid in the downcomer upon the moving core barrel and the rector pressure vessel. This fluid-structure interaction module is based on an approximated analytical 2D-solution of the coupled system of 3D fluid equations and the structural equations of motions. By means of the vibration model all eigenfrequencies up to 30 Hz and the corresponding mode shapes were calculated. It is shown that the fluid-structure interaction strongly influences those modes that lead to a relative displacement between reactor pressure vessel and core barrel. Moreover, by means of the model the shift of eigenfrequencies due to the degradation or to the failure of internal clamping and spring elements was investigated.Comparing the frequency spectra of the normal and the faulty structure, it could be proved that a
recognition of such degradations and failures even inside the reactor pressure vessel is possible by pure excore vibration measurements.

During a hypothetical accident the reactor pressure vessel internals of boiling water reactors can be exposed to considerable loads resulting from temperature gradients and pressure waves. The finite element (FE) analysis is an efficient tool to evaluate the consequences of those loads by computing the maximum mechanical stresses in the components. 3 dimensional FE models were developed for the core shroud, the upper and the lower core supporting structure, the steam separator pipes and the feed water distributor. The models of core shroud, upper core support structure and lower core support structure were coupled by means of the substructure technique. All FE models can be used for thermal and for structural mechanical analyses. As an example the FE analysis for the case of a station black-out scenario (loss of power supply for the main circulating pumps) with subsequent emergency core cooling is demonstrated. The transient temperature distributions within the core shroud and within the steam dryer pipes as well were calculated based on the fluid temperatures and the heat transfer coefficients provided by thermo-hydraulic codes. At the maximum temperature gradients in the core shroud, the mechanical stress distribution was computed in a static analysis with the actual temperature field being the load. It could be shown that the maximum resulting material stresses do not exceed the permissible thresholds fixed in the appropriate regulations. Another scenario which was investigated is the break of a feed water line leading to a non-symmetric subpressure wave within the reactor pressure vessel. The dynamic structural response of the core shroud was assessed in a tranisient analysis. Even for this load case the maximum resulting stresses remain within the allowed limits at any time.

The coupling of advanced thermohydraulic codes with 3-dimensional neutron kinetic codes corresponds to the effort to replace conservative estimations by best estimate calculations. ATHLET is an advanced thermohydraulic code, developed by the German "Gesellschaft für Anlagen- und Reaktorsicherheit" (GRS). The DYN3D code, developed at Forschungszentrum Rossendorf (FZR) for the simulation of reactivity initiated accidents in nuclear reactors with hexagonal and Cartesian fuel element cross section geometry comprises 3-dimensional neutron kinetics, models for the thermohydraulics of the core and the thermomechanical fuel rod behaviour.

Both versions of DYN3D for hexagonal and Cartesian fuel element geometry were coupled with ATHLET according to two basically different strategies. The first way of coupling uses only the neutron kinetics part of DYN3D (internal coupling). In the second way, the whole core is cut out from the ATHLET plant model and is completely described by DYN3D (external coupling). In this case the values of pressure, mass flow rate, enthalpy and boron acid concentration at the bottom and at the top of the core have to be transferred between the codes. This way of coupling is efficiently supported by the General Control and Simulation Module (GCSM) of ATHLET.

The coupled code DYN3D/ATHLET was used to investigate the possibility of recriticality during an asymmetrical overcooling of the reactor core of a NPP with VVER-440 after a main steam line break (MSLB). This MSLB analysis was performed for hot zero power and end of fuel cycle conditions. Different coolant mixing conditions in the lower plenum of the reactor were simulated. The results show the importance of these conditions. In case of a realistic mixing model and without consideration of mixing, i.e. where each loop is connected to a particular 1/6 sector of the core, a recriticality after reactor scram was predicted. For ideal mixing only, recriticality can be avoided.

In the case of external events (e.g. earthquake or explosion pressure wave) the reactor pressure vessel (RPV) and its internal components are excited to mechanical vibrations. In such cases it is important to know the dynamic component stress for the assessment of the mechanical integrity of the system.
For this reason a theoretical global vibration model based on the finite element method (FEM) was developed for a German BWR. The model comprises the RPV including the main coolant pumps (MCP), the control element drives, the core flux measuring pipes and the RPV internals.

In 1997 a "BWR Physics and Thermohydraulic Complementary Action (BWR-CA)" was established to further assist and broaden the objectives of the "European BWR R&D Cluster for Innovative Passive Safety Systems". The activities are divided into 4 work packages (WP). The work presented in this paper is performed in WP2: Passive decay heat removal from the core region, where two different decay heat removal systems were investigated. These are isolation condensers (e.g. as operated in Gundremmingen A, Dodewaard or in the PANDA facility) or emergency condensers (e.g. as foreseen on the SWR1000 and tested in the NOKO facility).

Within the BWR Cluster the Forschungszentrum Jülich had performed 10 NO-KO tests. Six of these tests were carried out with a four tube bundle, four tests with a single tube. Several post test calculations of NOKO tests were performed in the BWR-CA by CEA with CATHARE V1.3U and FZR with an improved version of ATHLET Mod. 1.1 Cycle C. Additionally FZJ had specified data for a blind calculation. The blind NOKO calculation - as well as the other NOKO post test calculations - proves, that the modified version of CATHARE as well as the improved version of ATHLET (ATHLET coupled with KONWAR) are able to predict the global parameters in the range of the fluctuations of the measured values. In the next complementary action it is planned to perform additional post test calculations of the single tube experiments, experiments with gas/non-condensable mixtures and pretest calculations of a second optimized emergency condenser test bundle. The optimization suggest is to increase the heat transfer, which allows to decrease the number of heat exchanger tubes and the emergency condenser dimensions. This is desirable because it reduces the surface of the pressurized BWR circuit and the possibilities for leakages, which is a safety relevant task.

Loss of vacuum accidents (LOVA) in fusion reactors will
lead to an air ingress into the vacuum vessel, oxidation
of the hot armour material and a partial mobilization of
the sorbed tritium. Also burnable carbon monoxide will
be formed during these accidents. In continuation to the
testing of carbon candidate materials for the First Wall
under oxidizing conditions in oxygen and steam in
collaboration with NET the carbon compound Aerolor AO5
of Carbone Lorraine has been examined in detail in the
experiments INDEX2 and SPALEX of the Institute of Safety
Research and Reactor Technology of Forschungszentrum
Jülich. The chosen parameters for these oxidation
experiments with oxygen in the inpore diffusion
controlled regime are temperatures between 973 K and
1173 K and partial pressures of oxygen between 1140 Pa
and 21000 Pa at system pressures between and; that means
reaction gas con centrations between 1 vol.-% and 20
vol.% oxygen in argon. From these results a semi-
empirical kinetic equation for the heterogeneous primary
reaction was developed. The equation considers the
dependence of the reaction rate on temperature, partial
pressure and burn-off of the material.
For calculations of LOVA the code RALOC Mod 4.0 AD -
developed by the Gesellschaft für Anlagen- und
Reaktorsicherheit (GRS) mbH has been extended by the
model COX (Carbon Oxidation) to consider carbon
oxidation kinetics. This kinetic model is based on the
above mentioned experimental results. First calculations
have been done for leak areas of 0.01 m², 0.1 m² and 1.0
m² and for different numbers of nodes in the reaction
chamber. The comparison of these results with earlier
calculations of JRC Ispra, GRS and Technicatome showed
that the new model COX and the nodalisation scheme lead
to more reliable statements concerning the spatial
dependencies especially for the burn-off of the armour
material and for the masses of the formed gases CO and
CO2.

Es wird ein Ultraschall-Meßverfahren für Zweiphasenströmungen vorgestellt, das die simultane Messung der Volumenstromdichte der Gas- und der Flüssigphase in einer Rohrleitung ermöglicht. Ein Prototyp wurde in Zusammenarbeit mit der Universität Nishny Novgorod (Rußland) aufgebaut. Es realisiert einen gepulsten Transmissions- und Reflexionsbetrieb mit Wellenleitersensoren, die auch an heiße Rohrleitungen bis zu 350°C direkt angekoppelt werden können. Kernstück der Meßmethode ist ein Mustererkennungsverfahren, das in bestimmten Volumenstrombereichen nach einer umfangreichen Kalibrierung (Trainingsprozeß) einen Meßfehler von kleiner 10 % besitzt, wenn die thermodynamischen und geometrischen Randbedingungen nahezu konstant bleiben. Durch die Erweiterung der Trainingsmatrizen um einen, die thermodynamischen Eigenschaften determinierenden Parameter (z. B. Temperatur, Druck) kann das Klassifizierungsverfahren auch auf Fälle mit veränderlichen Stoffwerten angewandt werden. Das Mustererkennungsverfahren und die Ultraschalltechnik wurden experimentell optimiert und erprobt. Die Ultraschallsignale wurden außerdem mit Hilfe der Methode des unüberwachten Lernens klassifiziert. Die gefundenen Gruppen von Signalmustern weisen eine deutliche Ähnlichkeit zu bekannten Strömungskarten auf. Es wird gezeigt, daß das Verfahren des unüberwachten Lernens sich für die objektive Aufstellung von Strömungskarten eignet. In einem speziellen Kapitel werden die Ergebnisse von Tests zur Füllstandsmessung in einem Segment der Hauptumwälzleitung von Druckwasserreaktoren vorgestellt.

Abstract:

We study the reaction NN → NNΦ slightly above the threshold within an extended one-boson exchange model which also accounts for uud knock-out. It is shown that olarization observables, like the beam-target asymmetry, are sensible quantities for identifying a ss admixture in the nucleon wave function on the few per cent level.

New examples of characterization of vacancy-type defects
in ion implanted and annealed SiC by the established technique of slow
positron implantation spectroscopy are presented. In particular, the estimation
of the depths of damaged regions and their change (a) after post-irradiation
annealing, or (b) due to variation of substrate temperature during implantation,
is addressed.

Im Rahmen des Forschungsvorhabens 15 NU 09485 "Berechnung des passiven Notkondensators eines mit Naturumlauf arbeitenden innovativen Siedewasserreaktors (SWR600) mit ATHLET" wurden vom Forschungszentrum Rossendorf e.V. Auswertungen und Interpretationen von Messungen zu Strömungsformen bei Kondensationsvorgängen in einem umfangreich instrumentierten Einzelrohr durchgeführt. Ziel der Experimente war, die Aussagesicherheit der NOKO-Bündelexperimente zu erhöhen sowie experimentelle Daten für die Validierung von Thermohydraulikprogrammsystemen (u. a. den ATHLET-Code) bereitzustellen.
Die theoretische Modellierung des Wärmeübergangs an der Innenseite der Notkondensatorrohre erfordert die Identifizierung der im aktuell betrachteten Rohrquerschnitt vorliegenden Strömungsform sowie die Auswahl einer geeigneten (d.h. strömungsformabhängigen) Korrelation zur Berechnung des Wärmeübergangskoeffizienten. Sensitivitätsanalysen haben gezeigt, daß sich im Betrieb der Notkondensatoren in Abhängigkeit von der Wirkung der Dampfschubspannung auf den Kondensatfilm sowie dessen Schwerkrafteinfluß zahlreiche Strömungsformen (z. B. eine Ring-, Schichten, Blasen-, Schwall- und Pfropfenströmung) einstellen. Da der Flüssigkeitsanteil längs des Kondensatorrohres zunimmt, können die verschiedenen Strömungsfor-men nacheinander auftreten bzw. ineinander übergehen.

Plasmagestütztes Randaufsticken ist eine erfolgversprechende Technologie zur Verbesserung der Verschleiß- und der Korrosionseigenschaften von Bauteilen aus nichtrostendem Stahl. Bei moderaten Temperaturen (ca. 400°C) kann eine Randschicht mit einem hohem Gehalt an gelöstem Stickstoff (ca. 20at.%) erzeugt werden. Diese Randschicht weist eine deutlich erhöhte Härte und Verschleißbeständigkeit und eine hervorragende Korrosionsbeständigkeit auf. Die Prozesse, die zur Bildung dieser Randschicht führen, sind weitgehend unverstanden. Um den Einfluß der passivierenden Oxidschicht auf die Bildung der stickstoffreichen Randschicht zu untersuchen, ist ein in-situ Experiment mit einer gekreuzten Strahlgeometrie aufgebaut worden. Bei kontrollierten Restgaszusammensetzungen wird während des plasmagestützten Randaufstickens zeit- und tiefenaufgelöste Elementanalyse betrieben. Dabei wird zum Randaufsticken eine Hochstrom-Ionenquelle und zur tiefenaufgelöste Elementanalyse „elastic recoil detection analysis“ (ERDA) verwendet. ERDA ist sensitiv auf die leichten Elemente und wurde hier hinsichtlich schneller Datenaufnahme optimiert. Quantitative zeitaufgelöste Messungen der Elementverteilungen werden für verschiedene Sauerstoffpartialdrücke vorgestellt. Die Bildung der stickstoffreichen Randschicht ist stark mit der Reduzierung der Oxidschicht korreliert. Hohe Sauerstoffpartialdrücke stabilisieren die Oxidschicht. Dieses Ergebnis stimmt gut überein mit einem einfachen Transportmodell, das die Sauerstoffadsorptionsrate und die Zerstäubungsrate berücksichtigt.

The moderator temperature coefficient (MTC) of a 1400 MW pressurized water reactor (PWR) has been measured from neutron and temperature fluctuations in different reactor positions. In order to explain the deviations observed between the estimations, numerical evaluations of the MTC were performed using the 3D two energy groups DYN3D code for a prototype PWR. The differences between the experimental MTC estimations can be partially explained considering that the numerical MTC evaluations show a definite devitation of the neutronics from point kinetics.

The effect of chromium implantation and subsequent annealing
on the hardness and wear behaviour of a silicon nitride-based ceramic is
investigated. Implantation energies were 200, 400, 1000 and 2000 keV, the
fluence was 1017 Cr+ ions cm-2 in all cases. The annealing temperatures
were 800 C, 1000 C and 1200 C. The phase composition was determined with
X-ray diffraction. Annealing at 800 C does not affect the crystal structure.
At 1000 C the -Si3N4-phase is formed for 1 and 2 MeV implantation energy.
At 1200 C, Cristobalite and Keiviite are formed at the lower implantation
energies; at higher energies -Si3N4 is formed. We discuss the phase formation
for the different implantation energies and annealing temperatures and
the relationship with the observed hardness and wear.

Surfaces of SiO2, Si3N4, Ta2O5 and glass were implanted
with Ag+ ions. Studies using X-ray diffraction (XRD) and atomic force microscopy
(AFM) show that the surface morphology is dominated by metallic silver
features consisting of regularly shaped, mostly spherical, Ag particles
spanning a size range from some nanometers to some 100 nm in diameter.
The particle size metrics, distribution, shape, and their density are shown
to be controlled by the process paramters ion dose, dose rate, and ion
energy. Adjusting the energy of the incident ions results in various degrees
of submerging into the substrate including exposed Ag features anchored
onto the surface or buried Ag particles overcoated by a thin layer of the
matrix material. The substrate material differs by its ability of stabilizing
the dispersion and blocking the Ag movement. Ta2O5 and glass substrates
provide more homogeneously dispersed Ag particles, much smaller in size
and regularly shaped compared to SiO2 and Si3N4 substrate systems. Potential
application refers to surface enhanced Raman scattering.

Plasma immersion ion implantation was used to implant
oxygen into silicon with nominal doses ranging from 2 x 1016 to 2 x 1017/cm2.
Positive Ions, O+2 and O+, with a continuous energy distribution between
0 and 40 keV were implanted. The plasma parameters and the geometrical
configuration of the implantation were chosen in order to follow closely
the assumptions of a theoretical model, which allowed to calculate the
energy distribution of the incident ions. Computer simulations were performed
to derive the resulting theoretical oxygen depth distribution. The agreement
found with the experimental depth profiles, measured by elastic recoil
detection analysis, is good. The small discrepancy observed is discussed
in terms of the shortcomings of the model used.

no abstract

U LIII Polarized XAFS studies on Ba[UO2PO4]2·8H2O
C. Hennig 1, M. A. Denecke 1, A. Roßberg 1, G. Zahn 2,
T. Reich 1, H. Nitsche1
1 Forschungszentrum Rossendorf, Institut für Radiochemie,
PF 510119, D-01314 Dresden
2 Technical University of Dresden, Institute of Crystallography
and Solid State Physics, Zellescher Weg 16,
D-01062 Dresden

An experimental and theoretical study of the U LIII -edge polarization dependent XANES and EXAFS is presented for a single crystal of barium uranyl phosphate hydrate. Ba[UO2PO4]2·8H2O is built up by layers of [UO2]2+ and equatorial coordinated [PO4]3- tetrahedra. Layers are bridged by Ba2+, H+ and H3O+ cations giving an overall tetragonal (or nearly tetragonal) symmetry. The orientation of a crystal obtained by cleaving a larger crystal and used for the XAFS measurements was determined by x-ray diffraction. The layered structure of Ba[UO2PO4]2·8H2O exhibits preferred cleavage along {001} and parallel to {100}. Linear uranyl units are aligned parallel to [001]. {100} represents the least squares plane for the equatorial oxygen atoms. XAFS measurements were performed with the polarization vector aligned in the equatorial plane (parallel to [100] and [110]), as well as with the polarization vector oriented perpendicular to [010], with 90°, 60°, 45°, and 30° angles to [001].
Two different explanations of the resonance » 15 eV above the LIII -edge white-line maximum have been discussed. One explanation, based on pressure dependent measurements of UO3, attributed this feature to shake-up processes [1]. The other interpretation stems from ab initio multiple-scattering calculations using the FEFF program package. These calculations identify this feature as a multiple-scattering resonance associated with the axial U-O bonds in various uranium compounds [2], [3]. Our experimental investigations show that the relative intensity of this resonance increases if the polarization vector is aligned along the axial U-O bonds. This observation confirms the multiple-scattering assignment of this feature and is supported by our own theoretical FEFF calculations.
The U LIII -edge EXAFS show a strong polarization dependence, especially evident in the Fourier transforms at distances corresponding to uranyl axial oxygen atoms and equatorial atoms. This dependence strongly influences the effective coordination numbers, Ni, obtained in the data analysis.
The marked characteristics of such polarization-dependent XAFS spectra renders them useful as a unique indicator of uranyl orientation. For example, the specific binding sites and orientation of sorbed uranyl (actinyl) ions onto mineral surfaces can be determined with this method.

[1] Bertram, S., Kaindl, G., Jové, J., Pagès, M., Gal, J.: Phys.
Rev. B 52 (1989) 2680
[2] Hudson, E.A., Rehr, J.J., Buchner, J.J.: Phys. Rev. B 52
(1995) 13815
[3] Hudson, E.A., Allen, P.G., Terminello, L.J., Denecke, M.A.,
Reich, T.: Phys. Rev. B 54 (1996) 156

The alteration of the shape of one-dimensional, two-dimensional (2D), and three-dimensional (3D) range distributions with growing dose as well as the extension of amorphous zones formed at high doses is studied for 15 keV BF₂ ⁺ and 30 keV P⁺ implantations into (100)Si using the dynamic binary collision code Crystal-TRIM. The range and damage profiles are calculated for the area 97.74 (parallel to [010]) × 97.74 (parallel to [001]) nm² at the target surface irradiated by the ion beam and surrounded by impenetrable masks. The change of the shape of the 2D and 3D distributions in the lateral direction is less pronounced than in the depth direction. At low doses the lateral profile branches are determined by rechanneled particles, at high doses the enhanced dechanneling is the reason for the increased lateral straggling. The latter effect is responsible for the relatively large lateral extension of the amorphous zone beneath the masks. The simulated depth profiles of boron and phosphorus and the predicted thickness of the amorphous layer are in good agreement with experimental data.

In früheren Untersuchungen haben wir nachgewiesen, dass Oxorhenium(V)-Komplexe mit Ketanserin-, Cisaprid- und MDL72832-ähnlicher Struktur hohe Affinität zu Serotoninrezeptoren des Gehirns aufweisen, womit die analogen ^99mTc-Verbindungen als potentielle Radiopharmaka von Interesse sind. Ihre Struktur ließ auch inhibitorische Wirkung auf die Serotonin abbauende Monoaminoxidase (MAO) vermuten. In Untersuchungen an Rattenhirnhomogenaten mit 1mM Tryptamin als Substrat beider Isoenzyme MAO A und MAO B, bzw. 1 mM Phenylethylamin als spezifisches Substrat für MAO B hemmt die Mehrheit der untersuchten Komplexe mit IC₅₀ Werten zwischen 10^-4 und 10^-6 M das Enzym. Die höchste Affinität besitzt der abgebildete Komplex mit IC₅₀=7*10^-7±2 M.
Die räumliche Anordnung des protonierbaren Stickstoffs im Molekül ist dabei kritisch. Die Ergebnisse zeigen, daß bei der Bewertung von Radiotracern für das serotonerge System potentielle Wechselwirkungen mit der MAO berücksichtigt werden müssen.

Das P-Glycoprotein (Pgp) ist ein energieabhängiges Efflux-Transportsystem für ein breites Pharmaka-Spektrum, welches in bestimmten Tumoren, aber auch in normalen Geweben, z. B. den Endothelzellen des Gehirns exprimiert wird. Die Pgp-Bewertung in Tumoren mit lipophilen Kationen (^99mTc-MIBI, ^99mTc-Tetrafosmin) wird durch verschiedene biochemische Veränderungen (z. B. des Energiestoffwechsels) erschwert. In Multitracerexperimenten an Hirnendothelzellen untersuchten wir die Veränderungen im Membranpotential (^99mTc-MIBI, ^99mTc-Tetrafosmin), des Glukoseumsatzes ([¹⁸F]FDG) und der Pgp-Aktivität ([³H]Vinblastin) unter dem Einfluß verschiedener Pgp-Inhibitoren (u. a. Verapamil, Vinblastin,Valinomycin). Die Zellaufnahme der Pgp-Substrate ([³H]Vinblastin, ^99mTc-MIBI, ^99mTc-Tetrafosmin) korrelliert unter Kontrollbedingungen mit dem Glucosemetabolismus ([¹⁸F]FDG). Bei Hemmung des Pgp zeigte sich allerdings in den meisten Fällen eine Verminderung der Akkumulation von FDG. Das Ionophor Valinonycin löst dagegen einen dramatischen Anstieg des Glucoseumsatzes aus, gleichzeitig ist die MIBI oder Tetrafosmin-Aufnahme verringert obwohl die [³H]Vinblasin-Akkumulation unverändert bleibt. Das zeigt, daß bei der Untersuchung von Pgp-Transportprozessen Sekundäreffekte auftreten können (hier Veränderungen des Energiestoffwechsels), die bei der Interpretation der Daten zu berücksichtigen sind.

Abstract
The first series of experiments at the FOBOS detector, using beams of the U-400M cyclotron of the Flerov Laboratory of Nuclear Reactions, was devoted to few-fragment decays of equilibrated systems. To exclude deep inelastic collisions or quasifission as sources of massive fragments, hot heavy nuclei were produced in the very asymmetric reactions ⁷Li (43 AMeV) + ²³²Th, ¹⁴N (34 AMeV) + ¹⁹⁷Au, and ¹⁴N (53 AMeV) + ¹⁹⁷Au, ²³²Th. Two- and three-fragment events were analysed on the base of masses and velocity vectors measured independently for each fragment. The events were sorted into excitation energy bins according to the linear momentum transfer following the massive transfer approach. Binary events were treated as fission. Fragment mass distributions as well as total kinetic energies were studied for an excitation energy range of 100 - 500 MeV. A new TKE parametrisation is proposed extending the Viola systematics to large mass asymmetries. With rising excitation energy the mass dispersion develops unexpectedly. Two new effects have been fourid and are discussed as consquences of the cooling down during the slow fission process at moderate E*, and of a strong decrease of the fission time at large E*. Ternary events were analysed by comparing measured velocity correlations with results of Coulomb trajectory simulations. If one fragment has intermediate mass (A = 10...30), two components in the relative velocities and the Z/A ratios confirm a sequential and a neck mechanism. For events with three fragments of comperable size neither the assumption of two sequential independent fission acts nor a nlultifragnlentaion-like scenario can reproduce the data. Agreement is achieved if these three-fragment decays are characterized by a collinear intermediate state followed by two scissions separated by no more than 280 fm/c, a very short time scale compared with usual saddle-to-scission intervals.

Abstract
A method for the intrinsic energy calibration of photomultiplier-coupled CsI(Tl) detectors is described. A simple empirical model of the scintillation light pulse-shape of CsI(Tl) crystals for light charged particles has been applied to simulate the particle identification matrix as it follows from the pulse-shape analysis method. The calibration procedure for the large-area CsI(Tl) detectors of the scintillator shell of the 4 π - array FOBOS for ions with Z < 4 at energies below 100 AMeV is based on the energies of the particle punch-through points.

Cobalt disilicide layers were formed by ion beam synthesis using 35 keV Co+ focused ion beam (FIB) implantation into silicon. A strong influence of the pixel dwell time on the layer formation was found. Only for short pixel dwell-times (about 1µs) closed layers with sufficient quality for device application could be formed. To understand the dwell-time effect the as-implanted samples were examined by Rutherford backscattering (RBS) and channeling analysis. A method is presented which allows quantitative measurements of samples where the implanted areas are smaller than the diameter of the RBS beam. Evidence has been obtained that the silicon crystal damage is less for short dwell-times.

Abstract
The elastic proton-deuteron backward reaction is analyzed within a covariant approach based on the Bethe-Salpeter equation with realistic meson-exchange interaction. Lorentz boost and other relativistic effects in the cross section and spin correlation observables, like tensor analyzing power and polarization transfer etc., are investigated in explicit form. Results of numerical calculations for a compbte set of polarization observables are presented.

Different benchmarks for reactors with quadratic fuel assemblies were calculated with the code DYN3D/R. In this report comparisons with the results of the reference solutions are carried out. The results of DYN3D/R and the reference calculation for the eigenvalue keff and the power distribution are shown for the steady-state 3-dimensional IAEA-Benchmark. The results of NEACRP-Benchmarks on control rod ejections in a standard PWR were compared with the reference solutions published by the NEA Data Bank. For assessing the accuracy of DYN3D/R results in comparison to other codes the deviations to the reference solutions are considered. Detailed comparisons with the published reference solutions of the NEA-NSC Benchmarks on uncontrolled withdrawal of control rods are made. The influence of the axial nodalization is also investigated. All in all, a good agreement of the DYN3D/R results with the reference solutions can be seen for the considered benchmark problems.

In diesem Abschlußbericht wird ein Fördervorhaben beschrieben, welches auf die Unterstützung von 6 russischen Wissenschaftlern aus drei unterschiedlichen Institutionen abzielte. Ihre Arbeiten zur Anlagendiagnostik / Schadensfrüherkennung dienen der Erhöhung der Sicherheit von WWER-Reaktoren.
Das Vorhaben leistet einen Beitrag zur verbesserten Beurteilung der mechanischen Integrität von Reaktoren der russischen WWER-440 und WWER-1000 Baureihe, insbesondere zur empfindlichen Früherkennung und Lokalisation mechanischer Schädigungen an Reaktorkomponenten mit Hilfe schwingungsdiagnostischer Methoden. Zu diesem Zweck wurde das Finite-Element-Modell zur Simulation des mechanischen Schwingungsverhaltens aller Primärkreiskomponenten eines WWER-1000 erstellt. Dafür wurde das Programmpaket ANSYS® genutzt.
Das Berechnungsmodell zum WWER-440 wurde mit neueren Ergebnissen experimenteller Schwingungsuntersuchungen weiter justiert. Das Modell kann in der Hauptsache genutzt werden, um zu klären, wie sich unterstellte mechanische Defekte von Reaktoreinbauten auf die Schwingungen der Gesamtanlage auswirken. Diese Schadenssimulation ist besonders geeignet, empfindliche Meßpositionen für die on-line Überwachung zu finden und physikalisch fundierte Grenzwerte zu definieren.

As the scaling of silicon devices to 100 nm channel length requires the formation of ultra-shallow (< 60 nm) junctions, high depth resolution analytical techniques become necessary for the characterization of the dopant and damage distributions. In situ single wavelength Ellipsometric
Etch Depth Profiling (EEDP) and non-destructive Variable Angle of incidence Spectroscopic Ellipsometry (VASE) have been used to obtain accurate and quantitative information on the depth profiles of radiation damage produced by low energy, room temperature ion implantation of B⁺ into Si.

The interaction of radium with characteristic sediments of the upper Saxon Elbe river valley was investigated with the
aim to provide key data for transport modelling within the framework of mining site restoration. Adsorption
measurements therefore were carried out on site-specific sediments (different kinds of sandstone, claystone and lime
marl) using (1) acidic mine water coming from sulfuric acid residues of the former in-situ leaching process, (2)
groundwater that is inflowing into the mine and (3) mixtures of both which simulates certain stages of the mine flooding process. The distribution ratios were determined by batch experiments using an eight-week equilibration at aquifer temperature (14 °C). Radium adsorption from acidic mine water onto different kinds of sandstone is characterized by high distribution ratios (3400 mL/g). When contacted with groundwater, the sorption is decreased more than 20-fold (60-180 mL/g). For claystone and lime marl the distribution ratios amount to 130 and 480 mL/g, respectively, and they are nearly constant in the mine flooding process. The main adsorption mechanisms on the leached sandstone and the clayey sediments were identified as coprecipitation and ion exchange, respectively. The long-term migration behaviour in the aquifers of the south-east Saxon Elbe basin was calculated on the basis of Hadermanns' equation, considering retardation, dispersion and radioactive decay.

The E1 response of the semi-magic nucleus ¹⁴⁰Ce below the particle
threshold was measured in a (gamma, gamma') experiment utilizing the new
Euroball Cluster detector at the S-DALINAC. While the energy averaged data
are in good agreement with the tagged photon results, here they are resolved
for the first time into 54 individual transitions. A quasiparticle-phonon
model calculation including up to three-phonon configurations compares
well to the extracted strength distribution. The interference between one-
and two-phonon contributions is essential for a quantitative reproduction.

Excited states of the nuclei ¹²²Te, ¹²⁶Te and ¹³⁰Te
were populated via the (gamma,gamma') reaction at endpoint energies of
the bremsstrahlung between 4.5 and 5.5 MeV. Gamma rays were detected with
a EUROBALL-CLUSTER detector and a single detector. In all nuclei several
dipole transitions were identified at energies around 3 MeV. The lowest
corresponding J = 1 states are interpreted as two-phonon excitations. Quasiparticle-phonon-model
calculations predict one 1^- state arising from the coupling
of the first quadrupole and the first octupole phonon and one 1⁺
state arising from the coupling of the first and the isovector second quadrupole
phonon at about 3 MeV. The calculated transition strengths are compatible
with experimental one.

At the new heavy ion tumor therapy facility of the Gesellschaft für Schwerionenforschung at Darmstadt positron emission tomography (PET) has been implemented for in-beam and in-situ therapy control, i.e. during the tumor irradiation. The components necessary for this dedicated PET-imaging and their integration into the framework of therapy planning and quality assurance of heavy ion cancer treatments are presented. Results of the first application of this PET-method to patient treatments are reported.

The results of the 4th AER dynamic benchmark obtained by the use of the code DYN3D in the Research Centre Rossendorf are presented. The benchmark comprises the modeling of a boron dilution transient caused by switching on the first main coolant pump of a loop containing a plug of partially undiluted water. Therefore, the course of the transient is significantly affected by correct modeling of the boron transport. The results of studies on the influence of numerical diffusion during boron transport are presented. Consequently, a new numerical model based on the particle­in­cell method has been developed and implemented into DYN3D. The basic features of this model are described. It allows the complete suppression of numerical diffusion and to take into account the effect of boron enrichment in the liquid water phase in the case of boiling in a simple manner. Another important feature of the benchmark task was that each participant had to use his own macroscopic cross section library. In our analysis the MAGRU library for VVER­440 was used. A few reactor­physical key parameters had been given in the benchmark definition for optional adoptation of the cross section data. The presented analysis was carried out with a minimum amount of adoptations. Finally, only the static supercriticality of the boron dilution was given an the Doppler feedback was adopted. Some interesting thermo­hydraulic effects are discussed, which can be used for a deeper comparative analysis of the thermo­hydraulics models of the codes. Due to the low mass flow rate and high heat flux densities occuring, vapour generation during the transient is significantly affected by subcooled boiling under thermodynamic non­equlibrium conditions.

A method for the in-situ control of the heavy-ion tumour therapy by means of positron emission tomography is introduced. This method is founded on the measurement of the dynamic spatial distributions of beta+-emitters generated by nuclear fragmentation during the irradiation and their relation to the dose. In order to study this relationship and to derive the dose distribution from the measured beta+-activity distribution, a framework of model calculations is used. Results of phantom experiments with ¹²C ion beams will be presented, demonstrating good agreement between experiment and calculation as well as the possibilities and limits of the PET-technique for treatment plan verification and beam monitoring.

Excited states in ¹¹⁰Sb have been identified for the first time
in a series of gamma-spectroscopy experiments using both thin and backed
targets, including neutron-fold and recoil-mass measurements to provide
unambiguous channel identification. The three decoupled intruder bands
observed in ¹¹⁰Sb are based upon configurations involving 2p-2h
excitations across the Z=50 shell gap and show the features of smooth band
termination. The yrast intruder band, which has been connected to the low-spin
levels, is tentatively identified up to its predicted termination at I^Pi=45⁺.
Excellent agreement with configuration-dependent cranked Nilsson-Strutinsky
calculations is obtained for the high-spin states near termination.

ß-FeSi2 was produced by ion beam assisted deposition
(IBAD). The influence of the deposition parameters on the structure was
studied by Rutherford backscattering, X-ray diffraction, cross section
transmission electron microscopy, and scanning electron microscopy. The
samples grow in a columnar way with pin-holes and their surface is rough.
A soft IBAD process with low Ar energy (EAr = 200 eV) and low Ar ion to
Fe atom ratio (IAr/AFe = 0.16) improves the layer structure in comparison
to samples prepared without Ar irradiation. Less pin-holes are formed,
and the roughness shows a minimum. The roughness increases for larger EAr
and larger IAr/AFe. All samples are polycrystalline but with a pronounced
texture. The preferential orientation is also enhanced by the IBAD process.
The electrical properties of the layers were characterized by Hall effect
measurements and measurements of the I-V characteristics of simple diode
structures. The results are discussed in relation with the influence of
the ion beam.

Im Bericht werden die Potentiale zur Elektroenergieerzeugung aus verschiedenen erneuerbaren Energiequellen in Sachsen zusammengestellt. Dabei werden natürliche Potentiale und - wo möglich - technische Potentiale sowie wirtschaftliche Potentiale unterschieden. Das ermittelte technische Potential (ohne Biomasse) entspricht etwa einer Elektroenergieproduktion von 7400 GWh jährlich. Davon entfallen etwa 2/3 auf die Windenergie und der Rest zu gleichen Teilen auf die Wasserkraft und die Photovoltaik.
Die Erschließung der Potentiale seit 1990 wird im Einzelnen dargestellt. Neben der seit langem genutzten Wasserkraft entwickelt sich insbesondere die Windenergie sehr dynamisch. Die Windenergie wird noch 1997 die Wasserkraft als bedeutendste erneuerbare Energie in Sachsen ablösen. Die weitere Nutzung dieser beiden Energiequellen stößt allerdings auf zunehmende ökologische Bedenken bzw. Widerstände.
Der Anteil der erneuerbaren Energiequellen am Elektroenergieverbrauch in Sachsen erreichte 1996 etwa 1,3 %. Bis zum Jahr 2000 erscheint eine Vergrößerung dieses Anteils auf ca. 5 % möglich.

Abstract
In this article we present results concerning the extraction of light charged particle multiplicities in incomplete fusion-fission reactions with a moving source fitting procedure. To demonstrate the feasibility and reliability of the procedure we have used simulated events of known multiplicity to construct kinetic energy distributions at different angles that are fitted assuming several pre- and post-scission sources. In particular, we show that it is necessary to include at least two pre-equilibrium sources in order to extract correctly the known pre-equilibrium proton multiplicity. These two sources are characterized by high ernission temperatures, low emission barriers and high velocities, having ~ 70% and ~ 25% of the beam velocity along the beam axis, respectively The former source is naturally dominant at forward angles due to the strong focusing effect of its high velocity, Contrary to normal expectations, however, the slower pre-equilibrium source is present with considerable yield at the most backward angles θ ~ 145° where it is normally assumed only evaporative components are present. The estracted proton multiplicities are well reproduced by the moving source procedure, as well as fitting parameters with physical relevance.