PAUL SCHERRER INSTITUT Childozsl

PAUL SCHERRER INSTITUT CHilDOZSl

JNJS-ml— 12882

1990 FOREWORD

In view of its size, the research department F3 is engaged international collaborations (EUV1TA, XMM) proceed as in an appreciable number of different directions of stud- scheduled. ies of condensed mailer, mainly involving nuclear methods and techniques. In addition, various projects are clearly application- or technology-oriented but have their roots in Nuclear methods have a wide field of applications in chem- solid-state physics or materials science. Although some istry with minute amounts of material and some of them are may criticize this diversification as leading to inefficien- used very successfully at PSl. Contributions lo experimen- cies, it is actually this broad spectrum of interests which tal environment studies arc made as well as research coping provides a very lively atmosphere and leads to a contin- with the non-trivial chemical investigations of super-heavy uous flow of new ideas, projects and collaborations. As elements. Also in these areas, distinct and pioneering ad- the reader may verify, these activities have again led to a vances in the development of experimental facilities were wealth of interesting results of which we have chosen a made. representative collection, constituting this report.

The program of defect physics entered a period of a, hope- Unfortunately, the experimental activities involving //SR fully, smooth rcoricntation. Although it is intended lo pro- had to come to a hall in 1990, because of the major shut- ceed with the PIREX program for another three years after down period of the proton accelerator, during which the 1991, new invofvements making use of the accumulated ba- installation is prepared for carrying a lot more proton cur- sic knowledge on defects in solids are evaluated. Activities rent than before. The main purpose of this enterprise is to related with material cmbrittlemcnt by neutron irradiation ensure the envisaged performance of ihe spallation source, arc declining in favour of new projects with intense positron the main project of F3. Most external //.SR users were beams. Here, both the development of new facilities and engaged in the analysis of previously obtained results and their subsequent use in new applications arc of equal im- the planning of new experiments that were submitted as portance. new proposals at the end of 1990. The y;SR group of PS1 invested most of its time in the planning of the future dedicated //SR area at the accelerator and in the design of two Most activities of the laboratory for technical physics are new, general-purpose, //SR spectrometers to be made avail- still application-oriented, anc| a large part of the efforts is able to the user community. dedicated to superconducting conductor and magnet development in the framework of EURATOM's Tusion facility program. As a new direction in applications of supercon- Particularly satisfactory is the progress made in the neutron ductivity, we envisage a study and the subsequent real- spallation source project SlNQ. Not only was the raw con- ization of an energy-storage installation based on supercon- struction of the additional buildings hosting the source and ducting solenoids, in collaboration with industrial firms and the beam lines and guides, and the non-trivial connecting future potential users. passage of the proton beam line between the accelerator and the planned target site completed, but a lot was accom- Considering the economical difficullies that we had lo cope plished in the design of the new source and its environment, with, I should like to congratulate all members of division including a promising R&D project for improving the per- F3 for their unrelenting enthusiasm for doing scientific work formance of neutron supcrniirrors. and for their achievements thai arc documented in part in this booklet. I should also like to thank the external users Considerable efforts also went into improving the experi- that are connected with F3 for their understanding and of- mental possibilities of our Tandem van dc Graaff heavy- fering help, where necessary. Their cooperation is well ion accelerator, keeping it atop of comparable installations appreciated and we hope to cover their needs for top ex- worldwide. Although the main thrust of activities is still perimental work also in the future. Of course, we are also in all kinds of applications of high-resolution accelerator grateful to many members of oihcr research divisions which mass spectromctry, new installations for serving materials- dedicate a lot of their lime for particular projets we arc inter- research projects have been and still arc being implemented. ested in. Last but not least, we owe a lot to the construction crews which made the SINQ-project so clearly visible now.

Looking back on a very active year is also the new section for applied physics, a merger of the groups involved in cryodctccior research ant) space-oriented technology. Here, the main investments for new experimental facilities were made for a versatile UHV installation dedicated to the fabrication of a number of different types of superconducting films and junctions, and a simulation chamber for experi- Professor H.R. Olt ments under space conditions. The contributions to larger Head of Department F3 Paul Scherrer Institut WUrenlingen and Villigen CH-5232ViUigenPSI Switzerland

Telephone (Exchange) 056 / 99 21 11 or 056 / 99 31 11

Telefax and Telex:

Telefax: Telex: 3000 Head of Department 056/99 32 94 82 74 19 3101 Myon Spectroscopy 056/99 32 94 82 74 19 3102 Neutron Scattering 056/99 23 27 82 74 17 3120 Applied Physics 056/99 32 94 82 74 19

3110 Accelerator Mass Spectroscopy 01/37 12 665

3200 to Chemistry Division 056/98 23 27 82 74 17 3214

3301 Defect Physics 056/98 23 27 82 74 17 056/98 23 27 82 74 17 3302 PIREX 056/99 32 94 82 74 19 3400 Technical Physics Division 056/99 32 94 82 74 19 3800 Project Spallauon Neutron Source

Editing: H.W. Gaggeler, R. Lorenzen

3.91 2000 56030/4 TABLE OF CONTENTS

Condensed Matter Research (3100)

Muon Spectroscopy ^SR (3101)

QUANTUM PHENOMENA AND SOLVENT EFFECTS ON CHEMICAL REACTIONS OF MUONIUM 2 EFFECT OF HYDROGEN IN YB^CiijO-, 4 HSR INVESTIGATION OF MAGNETIC PROPERTIES OF USb 6 MUON LOCALIZATION AND DYNAMICS IN C15 CUBIC LAVES PHASE COMPOUNDS 8 STRAIN EFFECTS ON n+ LOCALIZATION AND n+ DIFFUSION DM a-Fe 10 TC IU+ CHANNELLING STUDIES ON IRON 12

Neutron Scattering (3102)

HYDROGEN VIBRATIONAL MODES AND ANISOTROPIC POTENTIAL IN ScHx 14 ANOMALOUS HYDROGEN DYNAMICS IN RARE EARTH METALS 16 HYDROGEN ORDERING IN YHX 18 LITHIUM DIFFUSION IN THE SUPERIONIC CONDUCTOR Li2S 19 DYNAMICS OF COMPLEX CATIONS IN SODALITES 20 DYNAMICS OF WATER MOLECULES IN SODALITES 21 COMPARISON OF THE PARAMAGNETIC SPIN FLUCTUATIONS IN Ni WITH ASYMPTOTIC RENORMALIS ATION GROUP THEORY 22 DYNAMICS OF LONGITUDINAL AND TRANSVERSE SPIN FLUCTUATIONS BELOW Tc IN EuS 24 NUCLEAR AND MAGNETIC STRUCTURE OF THE PERMANENT MAGNET MATERIAL Tb2FeMC 26 NUCLEAR AND MAGNETIC STRUCTURE OF THE PERMANENT MAGNET MATERIAL Ho2Fe14C 28 LIGHT-INDUCED STRUCTURAL CHANGES IN SODIUMNITROPRUSSIDE (Na2(Fe(CN)5 NO) • 2 D2O) AT 80 K 30 SANS AND TEM INVESTIGATION OF AGE HARDENED NIMONIC PEI6 AFTER CYCLIC LOADING AT ROOM TEMPERATURE 32 SINTERING CHARACTERISTICS OF NANOCRYSTALLINE TiO2 - A STUDY COMBINING SMALL ANGLE NEUTRON SCATTERING AND NITROGEN ABSORPTION - BET 34 EARLY STAGES OF a-Co-PRECIPITATION IN DILUTE CuCo ALLOYS 36

Accelerator Mass Spectrometry AMS (3110)

THE PSI - ETH TANDEM ACCELERATOR FACILITY 38 EFFICIENCY IMPROVEMENTS WITH A NEW STRIPPER DESIGN 40 THE HEAVY ION INJECTOR AT THE ZURICH AMS FACILITY 42 AMS MEASUREMENTS OF 10Be IN POLAR ICE TO TRACE THE 11 YEAR CYCLE OF SOLAR ACTIVITY 44 HIGH RESOLUTION 10Be-STRATIGRAPHY OF ARCTIC SEDIMENT CORES 46 CHARACTERIZATION OF OPTOELECTRONIC MATERIALS WITH NUCLEAR METHODS 49

Applied Physics (3120)

EUVITA - AN EXTREME UV IMAGING TELESCOPE ARRAY ON THE SPECTRUM-X-G-SATELLITE 51 PROTON IRRADIATION TEST OF THE 14 bit/100 kHz COMPLETE SAMPLING ADC AD 779 KD 57 TEST CHAMBER FOR THE SIMULATION OF OUTER SPACE 59 ON THE PERFORMANCE OF SUPERCONDUCTING TUNNELING JUNCTIONS AS DETECTORS FOR NON-THERMAL PHONONS IN SINGLE CRYSTALLINE SILICON 60 UHV-DEPOSITION OF SUPERCONDUCTING T(H)IN FILMS 62 II

Chemistry (3200) Geochemistry (3211)

ANNUAL BIOGEOCHEMICAL CYCLES IN RIVERBORN GROUNDWATER AT GLATTFELDEN, SWITZERLAND 66 THE APPLICATION OF THE Rn-222 TECHNIQUE FOR ESTIMATING RESIDENCE TIMES OF ARTIFICIALLY RECHARGED GROUNDWATER: WATER CATCHMENT AREA HENGSEN, DORTMUNDER STADTWERKE AG 68 SAMPLE PREPARATION FOR AMS MEASUREMENTS OF 32Si 70

Trace Elements (3212)

CHANGES OF THE CONTENTS OF MICRO- AND MACROELEMENTS IN SPRUCE NEEDLES WITH THE NEEDLE AGE 71

Aerosol Chemistry (3213)

DIRECT MEASUREMENT OF MASS TRANSFER TO AGGLOMERATE AEROSOLS 73 IN-SITU MEASUREMENT OF SIZE, SURFACE AND MASS OF SILVER AGGLOMERATES 74 RELATION BETWEEN MASS, FUCHS SURFACE AND MOBILITY DIAMETER OF PARTICLES IN COMBUSTION EXHAUST 76 BIOGENIC HYDROCARBONS AS AEROSOL PRECURSORS: AN OUTDOOR SMOG CHAMBER STUDY 78 AEROSOL MONITORING WITH THE EPIPHANIOMETER AT REMOTE LOCATIONS 79 DEPOSITION OF AIR POLLUTANTS AT AN ALPINE SNOW FIELD 80

Heavy Elements (3214)

THERMOCHEMICAL CHARACTERIZATION OF BINARY TELLURIUM-METAL SYSTEMS 81 THE NEW NUCLIDE f^Ha 82 GASCHEMISTRY EXPERIMENTS WITH BROMIDES OF NIOBIUM, TANTALUM AND ELEMENT 105 83 CHEMICAL PROPERTIES OF ELEMENT 105 IN AQUEOUS SOLUTION: FURTHER STUDIES OF THE HALIDE COMPLEX EXTRACTION INTO TRIISOCTYL AMINE 84 ON THE VOLATILITY OF PROTACTINIUM BROMIDE 85

Material Sciences (3300)

Defect Physics (3301)

DEFECT PHYSICS - POSITRON PHYSICS - A PROGRAMME 88 PRELIMINARY NEUTRON SPECTRUM TAILORING AND DPA CALCULATIONS FOR IRRADIATION EXPERIMENTS IN THE POOL REACTOR SAPHK 90 PHASE SEPARATION OF COPPER IN IRON-BASED ALLOYS UNDER IRRADIATION: A SMALL ANGLE NEUTRON SCATTERING STUDY OF MODEL STEELS 92 ENERGETICS OF INTERSTITIAL MUONS IN BISMUTH: A STUDY OF Bi32(H) CLUSTERS 94 POSITRON LIFETIME MEASUREMENTS ON NEUTRON IRRADIATED IRON AND IRON-COPPER ALLOYS 96 LAB-SCALE POSITRON BEAM PRODUCTION AT PSI 98 DEVELOPMENT OF 1 Ci-58Co POSITRON SOURCES 100 POSITRON REPOLARIZATION 101 TRANSMISSION ELECTRON MICROSCOPY OF PRECIPITATES IN A THERMALLY AGED Fe-Cu MODEL ALLOY 104 GENERAL RELATIONS BETWEEN THE RESTRICTIONS IMPOSED BY DIFFERENT POINT GROUPS ON THE FORM OF PROPERTY TENSORS 105 Ill

PIREX, Irradiation Damages, Fusion (3302)

THE EFFECT OF ELECTRONIC ENERGY LOSS ON THE DYNAMICS OF THERMAL SPIKES IN Cu 106 RADIATION DAMAGE CASCADES: A LIQUID DROPLET MODEL OF SUBCASCADE INTERACTIONS 108

Technical Physics (3400)

STUDY PROJECTS 112 SUPERCONDUCTING PROPERTIES OF Bi2Sr2CaCu20yAy WIRES 115 HYSTERETIC CRITICAL INTERGRAIN TRANSPORT CURRENT IN SUPERCONDUCTING YBa2Cu307 AS A BASIS FOR NEW SWITCHING - AND DATA STORAGE EFFECTS 116 ON THE REDUCTION OF FLUX-CREEP IN SUPERCONDUCTING ACCELERATOR MAGNETS 119

Project Spallation Neutron Source (3800)

PROGRESS AT THE SPALLATION NEUTRON SOURCE OF PSI, A PICTORIAL REPRESENTATION 124 THERMOFLUID DYNAMICS OF LIQUID METAL TARGET OF SINQ [ 1 ] 128 SHUTTER SYSTEM FOR THE SINQ NEUTRON GUIDES 131 TECHNICAL CONCEPT OF THE LIQUID D2-COLD MODERATOR FOR SINQ 133 INVESTIGATION AND GROWTH OF MULTILAYERS 136 TAYLOR VORTEX FLOW IN A ROTATING COUETTE SYSTEM 138

List of Publications 143

Contributions to Conferences and Workshops 154

Lectures and Courses 162

Org. Chart Department F3 164

Scientific Committee Research Department F3 165 Condensed Matter Research (3100)

Muon Spectroscopy p.SR (3101)

Neutron Scattering (3102)

Accelerator Mass Spectrometry AMS (3110)

Applied Physics (3120) QUANTUM PHENOMENA AND SOLVENT EFFECTS ON CHEMICAL REACTIONS OF MUONIUM

E. Roduner*5, I.D. Reidf, P.W.F. Louwrier*, G.A. Brinkman1, D.M. Garner5, D.J. Arseneau^, M. Senba*, D.G. Fleming5 * Physical Chemistry Institute, University of Zurich, CH-8057 Zurich, Switzerland t Paul Scherrer Institute, CH 5232 Villigen PSI, Switzerland | NIKHEF-K, P.O.B. 4395, NL-1009 AJ Amsterdam, The Netherlands § Chemistry Department, University of British Columbia and TRIUMF, Vancouver, B.C., Canada V6T 1Y6

1 Introduction tors this is a clear indication for a strong contribution of tunnelling of the lighter isotope [2], Kinetic isotope effects have been widely used to test theories of absolute rate constants and thus to obtain a more detailed understanding of the principles of chem- log(k/M-'s-') ical reactions in the gas as well as in the condensed II- phase. A major fraction of the work has been concerned Mu with abstraction reactions involving hydrogen isotopes since the mass ratio of 2:1 between D and H can give rise to large effects. More recently, studies of muonium (Mu = n+e~), an exotic atom with the chemical properties of a hydrogen isotope, have been demonstrated to 10- lead to even larger effects since the mass of Mu is only one-ninth the mass of II [1]. There have been far fewer examples of isotopic studies of addition reactions. The questions of interest relate to the shape of the barriers to chemical reactions, to the effects of zero- 9- point vibrational energy and the contribution of tunnelling. The influence of solvation has been studied well for polar reactions but very little for elementary radical additions. A further aspect is diffusion: isotope effects have obviously been observed in solids and in the gas phase, but in liquids diffusion coefficients are generally believed to be mass-independent. We have chosen benzene and 2,3-dimethylbutadiene- 1,3 (DMBD) as substrates for our study. Both molecules react with Mu by addition. H adds to the former in an activated reaction while addition to the diene is prob- 7- ably encounter-controlled. Both reactions were mea- 1000/T —1— 1—"• sured, using the technique of muon spin rotation [1], in 2.0 2.5 3.0 3.5 the gas phase and in several solvents [2]. Caption Figure 1: Arrhenius plot of rate constants for the reactions of Mu 2 Results and discussion with 2,3-dimethylbutadiene, of H with butadiene, and of Mu, H, and D with benzene in the gas pbaso. The 2.1 Gas phase size of the symbols corresponds to a statistical error of one standard deviation. The rate constants for the reactions of Mu in the gas phase are displayed and compared with literature data for corresponding reactions with H and D in Figure 1. Addition of Mu is obviously considerably faster than 2.2 Liquid solutions that of the heavier isotopes for both substrates. For 2.2.1 A mass effect on diffusion the reactions of Mu and H with the dienes the slopes are small and reflect essentially the temperature depen- The rate constant for Mu addition to DMBD was mea- dence of the frequency factor, confirming that the re- sured in n-hexane over a temperature range 185 < T < action is encounter-controlled. Addition to benzene is 333 K. It varies by a factor of nine and scales with the in- clearly activated for the cases of H and D but far less so verse viscosity as expected based on the Stokes-Einstein for Mu. Together with the corresponding frequency fac- relation for a diffusion controlled reaction. However, its absolute value at a viscosity of 1 cp amounts to 5 • 1O10 M~Is~1 for Mu and is thus a factor of five higher than than for comparable reactions of H. This can only mean that Mu diffuses significantly faster than H, which represents the first clear evidence for a mass effect on diffusion in liquids. The thermal de Broglie wave length at room temperature is 0.1 nm for H but as much as 0.3 nm for Mu, indicating that the Mu wave package extends over a region comparable in size to a solvent molecule. It is therefore not unexpected that diffusion of these light particles is affected by quantum phenomena even in liquids.

2.2.2 A solvation effect on activated reactions The rate constants for the reactions of Mu with benzene are below 1010 M^'s"1 und thus significantly below the diffusion controlled limit. However, they are faster in the liquid than in the gas by a factor of 3-13 depending on the solvent. For H, this effect amounts even to a factor of 25. For solvents other than water this acceleration is due to unfavourable solvation of the hydrogen isotopes [2]. For water, there is an interesting mass dependence of the solvation effect. It is tentatively ascribed to the dynamics of the disappearence during reaction of the clathrate-like cage around the H or Mu atom [2].

2.2.3 Selectivity, a measure of thermalisation Comparison of the rate constants in n-hexane at room temperature reveals that Mu adds to DMBD a factor of 40 faster than to benzene. The Mu lifetime is of the order of one microsecond under the conditions used. The selectivity can also be measured by looking at the relative yields of the two product radicals formed upon addition of Mu in binary mixtures of the two substrates or concentrated solutions thereof. These conditions correspond to a Mu precursor lifetime of the order of ten picoseconds, and the .selectivity observed is only a factor of five, much let's than in more dilute solutions. The difference has probably to be ascribed to the energy of Mu which just after formation is still several eV. Thermalisation is not complete after 10 ps, but after one microsecond the system has equilibrated with its surrounding. Further measurements are needed to support this tentative interpretation.

References [1] E. Roduner, Prog. React. Kinetics 14 (1986) 1. [2] E. Roduner et al., Ber. Bunsenges. Phys. Chem., in press. EFFECT OF HYDROGEN IN YBa2Cu3O7

H. Gliicklcr*, Ch. Nicdcrmaycr', G. Nowitzkc", E. Recknagel"5, J.I. Budnick', A. Golnik:, A. Weidinger^5

* Fakultat fur Physik, Univcrsitiit Konstanz, D-7750 Konstanz, FRG t Physics Department, University of Connecticut, Storrs, CT 06268, USA t Institute of Experimental Physics, University of Warsaw, PL-00681 Warsaw, , oland § Bcrcich Kcm- und Strahlenphysik, Hahn-Meilner-Institut Berlin, D-1000 Berlin 39, FRG

uaOy can be changed from an antiferromagnetic to a superconducting state by "arying the oxygen sloichiom- ctry from y = 6.0 to y = 7.0 [1],[2]. We have investigated the effect of hydrogen on fully oxygenated YBa2Cu3Oy o „ HxYBa 2Cu samples using the muon spin rotation technique. The con- o undoped ductivity of YBa Cu O7 is of p-type (holes) and therefore 0 2 3 ~"-<3- H it should be possible to change the physical properties by rL o 0(7 counterdoping with an electron donor in a similar way as by A ^0.31 - o o the removal of oxygen from this system. In the present ex- D 3 a periment we have studied the effect of hydrogen doping on 0.5 per formula unit YBa2Cu307 [3]. We have reported these results in the previous annual report. o 2. Measurements in an applied external magnetic field. Q. These experiments were performed on weakly hydrogen- a \ doped samples YBa2Cu307Hz (i- < 0.5) where no long- range magnetic ordering occurs. The purpose of this study was to get informal'on on the density of superconducting carriers via the magnetic field distribution in lhe flux lattice stale. After cooling the samples in an external magnetic \\ field (Bexl = 0.3T) below the transition temperature 7J- a poo OA lattice of flux vortices (vortex state) is built up which gen- 0 erates inhomogeneiiies of the local fields at the muon sites. 0 20 L0 60 80 100 120 These inhomogeneities lead to an increase of the depolar- Temperature [Kl ization rate a of the muon signal in the superconducting state. For the analysis of the data we assumed a Gaussian 2 2 Figure 1: Depolarization rate a for weakly hydrogen- function for the damping, P(t) = exp(-

Figure 2: Relationship between the depolarization rate

We would like to thank Dr. D. Hcrlach for making the Stuttgart /zSR setup available for the TF-//SR measurements.

References [1] N. Nishida et ah, Jpn. J. Appl. Fhys. 26 (1987) L1856 [2] J.H. Brewer et al., Phys. Rev. Lett. 60 (1988) 1073 [3] Ch. Niedcrmayer ct al., Phys. Rev. B 40 (1989) 11386 [4] YJ. Uemura et al., Phys. Rev. Lett. 62 (1989) 2317 //SR INVESTIGATION OF MAGNETIC PROPERTIES OF USb

L. Asch 'f, G.M. Kalvius 's, A. Kratzcr ', K.H. Miinch \ FJ. Liners! !, A. Schcnck \ F.N. Gygux K. Mattenbcrgcr ", O. Vogt ",

Fhysik Department, Tcchnischc Univcrsiliil Munelicn, D-8046 Garching, FRG. SckLion Physik, Univcrsitat Miinchcn, D-8000 Munchen, FRG. Instilut ftir Mctallphysik, Tcchnische Univcrsiliit Braunschweig, D-33OO Uraunschwcig, FRO. Institul fur Miltclcncrgiephysik, ETH Zurich, CH-8093 Ziirkh, Switzerland. f Labor fur Fcstkorpcrphysik, ETH Zurich, CU-8093 Zurich, Switzerland.

This work is part of a comparative /(SR study of static zero field. Two of these spectra show spontaneous spin and dynamic properties of the uranium monopnictkics UX precession (with differcn'. frequencies), the third none. The with X = N, P, As, Sb, and Bi. These compounds all spectral typo in both magnetic phases can be explained con crystallize in the NaCl structure. They arc fairly localized and strongly anisotropic anlifcrromagncts, but some dclo- calization can occur on account of hybridization with the 200 K ligands [1]. in addition to the question of 5f localization, the uranium monopnictides are of interest on account of their unique magnetic properties which are nowadays discussed in terms of "Kondo lattice" and "heavy fermion" behavior [2]. These features arise out of an intricate in- terplay of various coupling mechanisms between the magnetic ions like short-range anisotropic CoqbJin-Schrieffer and iong-range RKKY exchange, quadrupolar interaction and crystalline electric field anisotropy [3]. One of the more apparent results is the formation of rather complex ordered spin arrangements known as multiple k structures 14). UAs is the compound studied by us in most detail within the series UX. The present report is aimed to emphasize the similarities and differences of our findings in USb with respect to UAs. Hence a brief summary of the / axes). At 62 K it changes into a type IA-2k arrangement (a non-collincar + + -- sequence with the spins pointing along face diagonals). Our main results were (a) Paramagnetic regime, T > 180 K. Zero field spectra revealed a Gaussian relaxation function corresponding to TIME ((is) firms ^ 1 G. This reflects the depolarization of /i spins by quasi-static nuclear dipolcs of 7DAs and shows the muon to Figure 1: Zero field spectra of USb at different temperatures be localized. In a longitudinal field of 10 G only a weak below TN. exponential relaxation with A « 0.01/JS"1 remains. It is 13 due to the fast fluctuating (l/r5y « 10 Hz) moments on sistcntly by assuming the muon to be localized at the inu.i- uranium. slices in the center of the cube formed ay two As and two (b) Paramagnetic regime, 180K > 7' > 124 K. An increase U atoms. The nearly static Lorcnlzian Kubo-Toyabe of the of relaxation rate and a change in the shape of the zero field type I-lk structure implies a rather perfect spin lattice with spectra arc observed. They arc considered evidence for the only highly dilute faults and the al nee of low frequency formation of a magnetic precursor phase. spin excitations. It should be emphasized that the static (c) Anliferromagnelic regimes. Zero field spectra for the response is seen right below 7'N aid docs not change with type I-lk structure show no spin precession but a nearly lowering the temperature. static (1/r w 0.2 MHz) Lorcnlzian Kubo-Toyabc relaxation We now turn 'o USb. This compound exhibits one function arising from a rather narrow field distribution of second-order magnetic phase transition at 7N = 214K ss 8G. Accordingly, full decoupling can be achieved in where a non-collincar type l-3k spiii arrangement (a f -+- a longitudinal field of 100 G. In contrast, we observe fcr sequence wit.h spins aligned along one of the body diag- the type IA-2k structure an overlay of three subspectra in onals) develops. Fig.l shows zero field spectra at three temperatures below TN- They were fitted to a dynamic Lorentzian Kubo-Toyabe function with static width A and modulation rate 1/r (strong collision model) as physical parameters. Common with UAs in the type I structure is the absence of spontaneous spin rotation and the Lorentzian field distribution. Its width, however, is about four times wider. Of this, a factor of ~ 2.3 is simply due to the difference in ordered moment on the uranium (n = 1.9^B in UAs and /i = 2.85/JB in USb). Thus distortions of the spin lattice seems somewhat larger in USb, but they are still dilute. A slight sample dependence is observed. Most significant is the difference in dynamical properties of the zero field spectra in UAs and USb. The fluctuation rate 1/r changes between 1/r as 1 MHz for the spectrum just below TN and 1/r « 0.5 MHz for the one at low temperatures. Weak low energy excitations are present which freeze out for T —* 0. Fig. 2 shows the temperature dependence of width and rate. The width first increases with temperature, then passes through a maximum at 140 K. In this temperature range the fluctuation rate also starts sud-

-6.0 USb

•5.0 Figure 3: Longitudinal field decoupling spectra slightly below the Ndel temperature and at 4 K. -(.0 •/ \ '' *"'""

-3.0 / "/•' O stays constant in field at 200K, the situation is different at / o •2.0 —Width/ 4 K where the width decreases by about 15% between zero

1.0J • • - x . -»-• - 1.0 and 10 mT longitudinal field. In addition, the relaxation rate is slightly field dependent: the spectrum becomes more 0 50 0 Itmperolure IK) " static wilh higher fields. All these effects are not completely u-iderstood at this stage and need further investigations. Figure 2: Temperature dependence of width and rate of the Lorentzian Kubo-Toyabe spectra seen in the antiferromag- References netic regime. [1] M.S.S. Brooks and D. Glotzcl, Physica B102 (1980) denly to increase. By neutron scattering [6] a collapse of 54 spin wave spectra into broad inelastic scattering peaking only at zero frequency is observed at about the same tem- [2] T. Kasuya ct al., "Seminar on strongly correlated perature. metallic systems" Schloss Ringbcrg, 1990 (unpub- The following interpretation for the difference in /iSR lished) spectra between Ik UAs and 3k USb is proposed: At low [3] B.R. Cooper ct al., in HdbK. Phys. Chem. Actinidcs, temperatures the phases of the three magnetic components (A.J. Freeman, G. Lander, eds.) Vol.2, p.435 in the 3k structure are locked. As the temperature rises magnetic excitations de-lock these phases and produce phase 14] J.Rossat-Mignod ct al., in Hdbk. Phys. Chem. Ac- irregularities. These magnetic pseudo excitations do not tinides, (A.J. Freeman, G. Lander, cds.) Vol.1, p.415 propagate like spin waves but are of diffusive nature. Phase de-locking disturbs the regularity of the spin lattice and in- [5] L. Asch ct al., Europhys. Lett. 10 (1989) 673 creases the field distribution width. The onset of diffusive [6] M. Hagen ct al., Phys. Rev. 37 (1988) 1846 motion then produces narrowing. This interpretation hinges on the assumption that the muon is static over the whole temperature region of interest. As mentioned above, a localized muon has been proven to exist in UAs up to at least 250 K. It is likely that this tesuK can be extended to USb, but. m certain. The crystal structure is of course the same. However, the lattice constant is about 7% larger in USb (6.209 X vs. 5.779 A). Finally, longitudinal field spectra are presented in Fig.3. In comparison to UAs larger decoupling fields are needed in accordance with ihe broader field distribution in USb. Unusual are the following observations: Whilst the width 8

MUON LOCALIZATION AND DYNAMICS IN C15 CUBIC LAVES PHASE COMPOUNDS

15 5 O. Hartmann', R. Wappling*, G.M. Kalvius , L. Asch*. A. Kratzer', K. Aggarwal*, F.J. Littcrst h A. Yaouanc', P. Dalmas de Reotier", B. Barbara", F.N. Gygaxtt, B. Hittitt, E. Lippelt", A. Schenck"

* Institute of Physics, University of Uppsala, S-75121 Uppsala, Sweden t Physik Department, Technische Universitat Miinchen, D-8046 Garching, FRG I Sektion Physik, Universilat Miinchen, D-8000 Miinchen, FRG § Institut fur Metallphysik, Technische Universitat Braunschweig, D-33OO Braunschweig, FRG f DRF/SPh, CENG, F-38041 Grenoble-Cedex, France ** Lab. Louis Ned CNRS, F-38041 Grenoble-Cedcx, France ft Institut fQr Mittclenergiephysik, ETH Zurich, CH-8093 Zurich, Switzerland

Several /tSR studies of cubic Laves phase compounds with the C15 structure have appeared in the literature. These comprise compounds of the type RA12 (with R = rare earth element) [l]-[4], RNi, [3],[4], RFe2, RCo2 [5], .20 T = 69 K AnAl2 (with An = actinide element U or Np) [6],[7] and YMn2 [8],[9]. The detailed interpretation of those measurements suffers from the fact that the stopping site of the . IS muon in these materials has not been known. The reason _ • 10 is that the samples used were polycrystalline powders and T = 136 K 7 that in most cases the signal in the magnetically ordered state is lost, at least at low temperatures where one expects a .os the muon to be localized. In the C15 structure the rare earth (or actinide) atoms form a diamond lattice while the Al (or transition element) atoms form regular tetrahedra in the interstices between the rare earths. Prime candidates for muon stopping sites are tetrahedral and octhedral interstices. Neutron scattering on hydrogen loaded ReFe2 compounds have shown that only the tctrahedral sites are occupied [10]. Of those one can distinguish interstices surrounded by 2A1 and 2R (2-2 site), by 3AI and 1R (3-1 site), or by 4A1 (4-0 site) atoms. The above-mentioned hydrogen loading 3-1 experiments have shown that for low concentration the 2- 2 sites are occupied, then, as the concentration increases, a successive occupation of the 3-1 and then the 4-0 sites 30 60 90 120 ISO 1BO occurs. Rotation Angle [*] In order to obtain information on the location of stopped fi+ in the C15 structure we have carried out ^SR spectroscopy on a single crystal sphere of CeAl2. The trans- Figure 1: Angular dependence of the transverse muorj verse field damping rate was determined as a function of damping rate in CeAl2 with the crystal being rotated around crystal orientation at different temperatures within the para- the [110] direction. An angle of 0° corresponds to the ap- magnetic regime. Certain features in the temperature de- plied field along the [110] direction, an angle of 90° along pendence of the damping rate for CeAl2 could also be found [100]. Top: Experimental result taken at the /*E4 beam of in paramagnetic UAI2 and diamagnetic LaAl2. This gives PSI. Bottom: theoretical calculation for the three interstitial strong evidence that the result for CeAl2 is typical at least sites discussed in the text. It is assumed that the influence for the other Laves phase di-aluminides. of Ce atomic moments is negligible due to very fast fluctu- CcAl2 is an antiferromagnet with a Neel temperature ation rates and that therefore the damping arises only from of ~ 3.9 K [11]. This means that one expects no marked 27Al nuclear moments. temperature dependence of the depolarization rate due to paramagnetic cluster formation [1] above ~ 20K. Fig. 1, top shows the angular dependence of the damp- The temperature dependence of the damping rate for ing rate measured in a transverse field of 0.15 T at two CeAl2 is depicted in Fig. 2, middle. Besides the strong temperatures. Only at the lower temperature (69 K) a vari- rise near I'N which is typical for the approach to a second ation of the damping rate with angle is seen. It agrees well order magnetic phase transition [1] one observes a definite with the calculated curves for the 2-2 site (Fig. 1, bottom). drop in the damping rate around 80 K and a weaker one around 200 K. Together with the data presented in Fig. 1 References this puts into evidence that the muon is stationary below 80 K and located at the 2-2 site. At higher temperature [1] O. Hartmann et al.. J. Phys. F: Met. Phys. 16 (1986) muon motion sets in since the orientation symmetry is lost. 1493 Probably within the range up to ~ 200 K the diffusion is related to some trapping or change of lattice position. Free [2] B.A. Gradwohl cl a]., Hyperfinc Interact. 31 (1986) diffusional motion sets in above 200 K. 319 As can be seen from Fig. 2 a quite similar temperature [3] J. Chappert et al., Hypcrftnc Interact. 32 (1986) 331 dependence of damping rate is observed for non-magnetic [4] P. Dalmas de Rcoticr el al., Hyperfine Interact. 63-65 (1991) 383 [5] S. Barlh et al., Phys. Rev. B 33 (198*) 430 0.15 [6] A. Kratzcr et al., Hyperfine Interact. 32 (1986) 309 LaAl, 0.10 [7] K. Aggarwal et al., Hypcrfine Interact. 63-65 (1991) 401 0.05 + + + [8] L. Asch et al., Hyperfine Interact. 63-65 (1991) 435 \ 0.00 [9] R. Cywinski et al., ilyperfine Interact. 63-65 (1991) _ 0.20 CeAl, 427

0.1C

RAT E [10] G.E. Fish et al., J. Appl. Phys. 50 (1979) 2003

0.1C [11] B. Barbara et al.. Solid State Commun. 24(1977)481 [12] L. Asch, Physica B 161 (1989) 299 o.os kRIZATIO N Necl [13] R.H. Heffner, Hyperfine Interact. 63-65 (1991) 497 o.or IPOLt a 0.20 UA1,

0.15 ' *. * • *•* 0.10 • , .*#

0.05

0.00 50 100 150 200 250 300 TEMPERATURE ( K )

Figure 2: Temperature dependence of the transverse muon damping rate in LaAl2 (top) CeAl2 (middle) and UA12 (bottom). The measurements on LaAI2 were carried out at the /x-ISIS facility of the Rutherford Appleton Laboratory (UK).

LaAl2 and paramagnetic UA12. Hence we conclude that the muon gets de-localized in these materials between 60 and 80 K. It might be mentioned that the absolute value of the damping rate in LaAl2 at low temperatures also agrees roughly with the expected number for the 2-2 site.

For the strong spinfluctuator UA12 the comparison with the CeAl2 data proves that the observed variation in damping rate has no connection with magnetic properties [12]. The fluctuation rate of the magnetic moment on the uranium atoms remains very fast (w 1013 Hz) down to the lowest temperatures measured (« 2 K). Hence there is no evidence for the formation of static, randomly oriented weak electronic moments as have been observed in several heavy fermion compounds [13]. 10

Strain Effects on fi+ Localization and /i+ Diffusion in a-Fe

RA-85-06, STUTTGART - PSI

C. Baines', A. Fiitzsche", M. Hampeie', R. Henes", D. Herlach*, M. Krenke1, K. Maier", J. Major' L. Schimmele', A. Seeger-t, W. Staiger", W. Tempi"

* Max-Planck-Institut fur Metallforschung, Institut fur Physik, Postfach 800665, D-7000 Stuttgart 80, Germany f Universitat Stuttgart, Institut fur Theoretische und Angewandte Physik, PfafFenwaldring 57, D-7000 Stuttgart 80, Germany t Paul-Scherrer-Institut, CH-5232 VUligen PSI, Switzerland

In previous interpretations of/i+SR experiments on magnetization M of the domains or the sample, respec- a-Fe the fi+ energy was assumed to be independent of tively, will always be parallel to an easy ((100)) direc- + + the /J. site. Since the possible /i sites, i.e., tetrahe- tion. In this case B'dip,B?ermi, and BLor,.Ilt,, are paral- dral (T) or octahedral (O) sites have tetragonal sym- lel or antiparallel to M. The same holds for jBappi and + metry, the distortion cloud surrounding the fi will be •Bdemag in the cases considered below. All fields may l anisotropic in general. The anisotropy is characterized then be replaced by scalars, especially B Aip by Bd'ip or by the tetragonal axis i of the site ( i || to one of the B/Ji , which are the dipolar magnetic fields felt by the three (100) directions). In what follows it is sufficient /x+ at sites with tetragonal axis parallel or perpendic- to consider either O or T sites and therefore we simply + ular to M, respectively. The cubic symmetry of the speak of i sites. If the (i interacts with an anisotropic, crystal leads to homogeneous strain field caused by external or internal stresses interstitial sites with differently orientated nil jjp (2) tetragonal axes become energetically distinguishable. ^dip To study systematically the influence of strains on In a very good approximation this relation also holds in the fj,+ SR signals, the difference in the thermal expan- the strained samples despite the slight deviation from sion of a-Fe and Al (W) was used by preparing Fe-Al-Fe, cubic symmetry. If the /i+ hopping rate is sufficiently Fe-W-Fe sandwich structures. After cooling the sam- high, but also in the other cases considered below for ples below 100 K strains are built up which are almost which this condition is not necessarily fulfilled, one gets 8 I 1 temperature independent below that temperature and (7M = 8.516- 10 radT- s~ ) of the order of 10~3. In the Fe-Al-Fe sandwich the a-Fe platelets (normal of the platelets approximately parallel to (100)) are compressed in the platelet plane; in the Fe-W-Fe sandwich, they are stretched however. + /i SR experiments on both samples have been per- (3) formed in magnetic fields applied parallel to the normal of the platelets (-Bappi = 2.2 T) as well as in zero field (£ai(I,i = 0). The temperature dependences of the /x+-spin pre- where pi = Pi{T) are the occupation probabilities of the cession frequencies w^ observed on two strained sam- sites (2,3= i Pi = !)• ples are shown in Fig. 1. They show very clearly the If all sites are energetically equivalent or if their en- strong influence of strains. Measurements in jBappi = 0 ergy differences are small compared to fcrjT one gets cannot determine the sign of wti. Thus essentially two Pi = 1/3 for all sites i. From (2) it follows then that different plots ox the data are possible. Various consid- Aui vanishes. Since at low temperatures all magnetic eration? suggest that the plot marked by 1 (in Fig. 1) fields appearing in (3) are T independent, a tempera- of the Bappi = 0 data gives the correct w (T). M ture independent u),,(T) results in these instances. This From the experiments performed on the two samples behaviour was found in unstrained samples. [1] we determined [2] the fi+ site, the dipolar magnetic + For the discussion of the strained samples we chose field Baip felt by the /i at this site, and sign as well as the 3-axis to be parallel to the normal of the platelets magnitude of the anisotropy of the muon elastic dou- and the 1- and 2-axis parallel to the other two (100) ble force tensor. The results and the essentials of the directions lying in the planes of the platelets. Since reasoning leading to them are given in the following. + it can be assumed that the Al or W platelets, which The local magnetic field at the /i site is given by are thick compared to the a-Fe platelets and elastically isotropic, are almost isotropically distorted we get eJI ss €22 —'• f ^ *33 f°r the distortions in the a-Fe platelets. Bdnmag, BPcrmii and BLOKHU are the demagnetization For the difference in the fi+ energies Ei, we obtain field, the Fermi contact field, and the Lorentz field. In therefore E^ - E2 w 0 and (1) only the dipolar field Bdip depends on the orientation of the sites tetragonal axis. In the following the AE = E3 - = (A~ B)(\ (4) 11

A denotes the ii-component and B the jj-component (j/ Our low temperature longitudinal spin relaxation i) of the elastic double force tensor of a /J+ located at experiments (down to 20mK) [1] can be understood if a site i. if is a positive constant of order I. The sign broad distribution of internal strains is assumed which of ( changes if one switches from the Fe-Al-Fe to the leads to a distribution of/x+ hopping rates and therefore Fe-W-Fe sample. to a distribution of relaxation rates. For \AE\ > k&T the occupation probabilities of the This work was funded by the Ministerium fiir For- sites become temperature dependent where pi as pi •£ schung und Technologie, Bonn, Federal Republic of Ger- Pi holds. For sufficiently low but not too low tempera- many, under contract no. 03-SE2STU-0. tures either pi ss p2 «a 1/2, ps s» 0 or pi « p2 as 0, P3 ss 1 holds, depending on the sign of {A — B)e. Aw(T) then reaches the plateau which is determined by the dipolar fields at the occupied sites. For M parallel to (0 the 3-direction B,\\ at 1- and 2-sites is given by B^ at v ip 00 T 3-sites by B^-. This case is realized for samples mag- in 40 © $ T e netically saturated in 3 direction by a high JB i- The « app O 1 relation between pi and pi is irrelevant in this case for 20 1 the interpretation of the low temperature plateau of AOJ 0 as long as |A£| is larger than \Ei - E2\. fe 9n + cD 8 T Calculations for a strongly localized fi wavefunc- \ . tion, neglecting lattice distortions and possible changes 40 e «j e « | 1 e of the magnetic moments of the iron ions in the vicin- 3*- ao B •'• ity of the fj,+ show that the dipolar fields have different ao B 8 signs at O- and T-sites (B|p = + 1.86 T at O-sites, B\ip 1000 100 10 - - 0.52 T at T-sites). The validity of these signs is also T[K) assumed for the more general O- or T-site configurations discussed in the following and is used to determine the SO B sites. From the sign of Au for, say, the Fe-Al-Fe sand- 8 8 80 e wich (only measurements in Bapv\ = 2.2 T are used in the following) as determined from a comparison of high (0 40 e e e ID m temperature plateau (w,») and low temperature plateau O 20 1 {ti}^ + AUJ(T)) sign and absolute value of JBdip is deter- 9 mined. From its sign it can be concluded whether 1,2, or 0 8 B 3 3-sites are occupied. This has to be done assuming O- ?0 $ as well as T-site occupation. This also determines the e © 40 e e e e e sign of A — B, again for the assumption of both O and T 31- so site occupation. Changing the sign of e, i.e., going from the Fe-Al-Fe to the Fe-W-Fe sample, leads according to eo (4) to the occupation of the respective other sites. From 1000 100 10 Aw again B,iip at these sites is determined. r[K]

Thus from both experiments Bjjip and B^ip are obtained, one pair assuming O-site occupation, the other for T-site occupation. For T-sites the symmetry rela- Figure 1: Spin precession frequency iv^ versus tempera- tion (2) is strongly violated (instead of 0 one obtains ture of the Fe-W-Fe sample (a) and the Fe-Al-Fe sample 1 T) whereas for O-site occupation (2) is obeyed within (b) at Bappi = 0 (o) and Bappi = 2.2 T D. Since the the error bars. Thus we conclude that O-sites are oc- sign of a;,, is unkown at Bxpl,\ = 0, two possibilities are cupied by the (i+. The sign of A — B is compatible considered at low T (1: sign change, 2: no sign change with this conclusion, too. The absolute value of A — B follows from the temperature variation of u>ti(T) if e is known. The behaviour of T2 is understood within the same model as well as that of u>tl(T) for J3appi = 0. In the latter case the difference of the domain structures References in the bulk for tensile and compressional strains has to [1] C. Baines, A. Fritzsche, M. Hampele, R. Henes, be taken into account. In summarizing the preceding D. Herlach, M. Krenke, K. Maier, J. Ma- results we found: jor, L. Schimmele, A. Seeger, W. Staiger, and 1) The /J.+ occupies O-sites (at least for T < 100 K). W. Tempi, PSI Annual Report 1989, Annex III, (1990) p. 12 2) A - B « + 2 eV. Below 1 K therefore even mag- netoelastic strains are important. [2] A. Fritzsche, M. Hampele, D. Herlach, K. Maier, J. Major, L. Schimmele, A. Seeger, W. Staiger, 3) B,dip + 0.7 T. W. Tempi, and C. Baines, Hyperfine Interact. 64 (1990) 691 The comparison of B'\ip ss -1-0.7 T with the theoretical value given above points to an extended fi+ wavefunction and/or a strongly relaxed environment of the fi+. 12

TT+ / fi+ Channelling Studies On Iron

RA-79-02, STUTTGART - PSI

W. Staiger", E. Widmann' D. Herlach', M. Krenke', K. Maier", J. Major', A. Seeger'l

* Max-Planck-Institut fiir Metallforschung, Institut fur Physik, Postfach 800665, D-7000 Stuttgart 80, Germany f Universitat Stuttgart, Institut fiir Theoretische und Angewandte Physik, Pfaffenwaldring 57, D-7000 Stuttgart 80, Germany } Paul-Scherrer-Institut, CH-5232 Villigen PSI, Switzerland

The question as to which sites are occupied by positive muons (M+) in metals with body-centered-cubic (bcc) lattice structure has been answered only very recently for a-Fe by /i+SR experiments on strained samples [1]. This knowledge is important, e.g., for the understanding of quantum diffusion of the fi+. The most spacious interstices in the bcc lattice ate the tetrahedral (T) and octahedral (O) sites. The O site was found to be occupied by the n+ at least below about 100 K [1]. The w+ which has the same charge and a mass similar to that of the n+ is expected to occupy the same sites. In order to test this expecta- tion and to deteimine the TT+ decay site at temperatures above 100 K we have performed ir+ /n+ channelling experiments on a-Fe. Unfortunately, in bcc lattices both T and O sites give rise to muon flux enhancement along (100) as well as (less pronounced) (111) and (110) directions. Likewise, the planar channelling along {110} planes observed in both orientations (c.f. arrows in Fig. 1 and Fig. 2) does not allow to distinguish between these sites. The relatively strong muon flux en- Figure 2: Muon flux density measured at 293 K on Fe oriented along (111).

hancement of about 25 % observed in our experiments along (100) (Fig. 1) and (111) (Fig. 2) shows that the pions are well localized on one of these sites in the temperature interval between 3.3 K and 300 K. In order to obtain channelling profiles that are qualitatively different for ir+ decaying at T or O sites, a new method based on the different ranges of channelling and blocking effects was developed. The basic idea is to supplement the information contained in the angular distribution of the decay muons (which immediately after the pion decay all have the same energy 2?k;,, = 4.12 MeV) by using the energy loss on their way through the crystal as an additional parameter. This energy loss reflects the distance a ji+ has travelled from the site of its creation (the 7r+ decay site) to the sample surface. Whereas channelled muons starting from comparatively deep below the crystal surface may still contribute to flux enhancement, blocking can only be observed on jx+ Figure 1: n+ flux density distribution from the decay + starting close to the surface because this information is of 7r implanted into Fe oriented along (100) measured lost after a short travelling distance due to large-angle at 25 K by means of a position-sensitive detector. Dark scattering. Therefore this method enables us to distin- areas correspond to high muon flux density. guish between different sites if in certain lattice directions one site leads to a superposition of a channelling 13

Figure 3: Projection of T sites and O sites in the bcc structure along (100) and (111) directions. °3 and a blocking pattern whereas the other one gives rise

to pure channelling. A look at the T and O sites pro- 01 jected along (100) and {111) directions (Fig. 3) shows that this is the case for O sites but not for T sites in, e.g., the (100) orientation. In constructing the "channelling" profiles only /i+ with energies above a low-energy cut off, i.e., with a maximum energy loss AE^ are counted. By varying AE^ we may thus switch, if both effects occur (which is the case for O site occupation), between a situation in which blocking plays no role (large AE^) and a situation in which blocking and channelling give equivalent contributions (small AE,A). The latter result in apparently narrower "channelling" patterns due to 200 300 400 900 the compensating effect of the smaller but much broader blocking dip on the shoulder of the genuine channelling A E [keV] peak. Fig. 4 shows that at low temperatures the chan- Figure 4: Dependence of the full width at half maximum (PWHM) of the muon channelling peak on the muon nelling peak becomes narrower with smaller AE/L as predicted for O sites whereas at room temperature the energy loss AEp for Fe oriented in (100) direction. peak width remains essentially constant. The measure- n: T = 4.7 K, o= T - 9 K, A: T = 293 K. ments for a (111) orientation (neither T nor O sites show blocking in this direction) can be taken as reference for the absence of blocking effects. We conclude -a from these observations that the pions occupy O sites at low temperatures and T sites at room temperature. We have also performed time-differential channelling measurements. As in earlier experiments with (111) orientation [2] a strong dependence of the channelling profiles on the time the 7T+ has sper.l in the sample was observed in (100) orientation. Between 20 K and 0J 70 K the "young" pions (decay between 0 and 20 ns) always give rise to high axial channelling peaks whereas channelling is strongly reduced for "old " pions (20 - 60 ns), with a minimum at around 30 K [3]. This 400 300 points either to an occupation of sites in a distorted AE [keV] environment or to a delocalized pion wavefunction. For the time being, the question on whether this is related + s n Figure 5: Dependence of FWHM on AE,L for Fe ori- to the anomaly observed in the n P' relaxation rates ented in (111) direction (T = 8K). in the same temperature range remains open. This work was funded by the Bundesministerium fur Forschung und Technologie, Bonn, Federal Republic of [2] S. Connell, G. Fabritius, R. Feldmann, G. Flik, Germany, under contract no. 03-SE2STU-0. D. Herlach, M. Koch, M. Krenke, K. Maier, A. Seeger, W. Staiger, and E. Widmann , PSI An- References nual Report F 3 1988, Annex III, 12 [3] W. Staiger, E. Widmann, S. Connell, G. Fabritius, [1] A. Fritzsche, M. Hampele, R. Henes, D. Herlach, K. Maier, and A. Seeger, Hyperfine Interact. 64 M. Krenke, K. Maier, J. Major, L. Schimmele, (1990) 701. A. Seeger, W. Staiger, W. Tempi and C. Baines, Hyperfine Interact. 64 (1990) 691 14

HYDROGEN VIBRATIONAL MODES AND ANISOTROPIC POTENTIAL IN SCHL

T.J. Udovic", JJ. Rush', I.S. Andcrsonf * National Institute of Standards and Technology, Gaithcrsburg, MD 20899, USA f Paul Scherrcr Institut, CH 5232 Villigcn PSI, Switzerland

Numerous studies have been carried out in recent years H concentration fx=0.05 (at 4K), 0.16 (9K), and 0.25 (4K)]. to elucidate the structure and dynamics of hydrogen in sev- The spectra indicate, first, that the peak position shifts from eral rare earth metals (eg. Sc, Y) which demonstrate the ca. 102.9 meV at the higher concentrations, x=0.25 and remarkable property of maintaining an o-MHr solid solu- 0.16, to ca. 101 meV at the lowest concentration x=0.05. tion phase with high H concentration down to low temper- Second the peak width (FWHM) decreases from 13 meV atures. Recent diffuse neutron scattering measurements in for x=0.25 to 12.4 mcv for x=0.16, which corresponds to these systems have revealed the existence of strong short intrinsic line widths of 12.8 and 12.1 meV respectively. It range order, which has been modeled by zig-zag chains of is clear that, despite the broad nature of the c-axis mode at H-M-H pairs along the c-axis. This essentially one dimen- all H concentrations, no distinct splitting is visible within sional ordering persists even to room temperature, while at the 2.5 meV resolution of the instrument low temperatures (and high concentrations) some evidence The overall widths of the a-ScH,. spectra are broader than of the onset of three-dimensional ordering is observed, ap- that of a-YHQ 18 and arc consistent with the presence of H parantly involving inter-chain interactions. Previous inelas- pairing across Sc atoms along the c-axis. Moreover the lack tic neutron scattering (INS) measurements on single crystals of observed splitting in the a-ScHr spectra is not inconsis- of d-YHj [1] showed that the local vibrational potential is tent with this pairing mechanism. Rather it suggests that highly anisotropic, producing a local mode rather softer the individual components of the c-axis spectra are more along the c-axis (Ec=101 meV) than the modes in the basal broadened than in a-YH0,ig due to a larger distribution of plane (EaJ=148 mcV). Furthermore higher resolution mea- H-pair environments (ie. less extended ordering of H pairs surements of the local vibrations polarised along the c-axis in the c direction). Thus it would appear that the hydrogen showed a clearly resolvable peak splitting of about 4 meV pairs in a-ScHx are more disordered than in a-YHr. which was explained in terms of coupled harmonic oscil- lations of H pairs giving rise to 'local' optic and accoustic modes. We have extended the INS measurements to a-ScHr where References there is evidence that the short range ordered chains arc shorter than in the yttrium case and where low temperature [1] I.S. Anderson, J.J. Rush, T.J. Udovic, and J.M. Rowe, H tunelling has recenUy been observed [2]. Both high and Phys. Rev. Lett. 57, 2822 (1986) low resolution INS spectra of »-ScH (x=0.05, 0.16, and r [2] I.S. Anderson, N.F. Berk, J.J. Rush, T.J. Udovic, R.G. 0.25) were collected on the BT4 triple-axis spectrometer at Barnes, A. Magerl, and D. Richter. Phys. Rev. Lett. the NIST reactor using a filter analyser. 65, 1439(1990) Figure 1 shows the low resolution vibrational spectra of o-ScH0 ]6 as a function of the angle i/> between Q and the c-axis allowing the selection of vibrations polarised in different directions. The results show that the c-axis vi- bration peak (at ca. 103 meV) is over 30% lower in energy than the doubly degenerate basal plane modes (at ca. 148 mcV) and possesses considerable anharmonicity, as reflected by the ratio (1.76) of the second- to first-excited-state energies. These results are similar to those observed for the comparable yttrium system however the magnitudes of the anisotropy and anharmonicity in the scandium case arc somewhat larger (25% and 1.8 respectively in Y). Further- more comparison of the peak shapes with the instrumental resolution shows that although the basal plauc mode is more or less resolution limited, the c-axis mode has an intrinsic line width of ca. 14 meV. In view of this, the line shape of the c-axis vibrational peak for a-ScHr was investigated more closely under high resolution conditions. In Figure 2, high resolution INS spectra at low temperature for a-ScH^, are plotted as a function of 15

a-ScH 016 (10K) Afn a-ScH, (<10K) • 4 c-axis mode

(a) 90°

(b)45° ,«u •• o "•„ (c) 25°

(d)0° 90 100 110 60 100 140 180 220 Energy Loss (meV) Energy Loss (meV)

Figure 1: Lew-resolution IINS spectra of a-ScHoie K as Figure 2: High-resolution IINS spectra of the c-axis mode a function of ip, the angle between Q and the c-axis. Posi- at T < 10 K for a-Sc//x (with 4' = 25°) as a function of tions of multiphonon scattering sidebands are marked with H concentration. The instrumental resolution (FWHM) is arrows. illustrated by the horizontal bar beneath the spectra. 16

ANOMALOUS HYDROGEN DYNAMICS IN RARE EARTH METALS

N.F. Berk', T.J. Udovic', }.}. Rush', I.S. Anderson'

* National Institute of Standards and Technology, Gailhersburg, MD 20899, USA t Paul Scherrer Instilut, CH 5232 Villigcn PSI, Switzerland

Hydrogen dissolved in hep rare-earth metals is known to factor (EISF) to be independent of T, while as seen in Fig. exhibit unusual short-ranged ordering, which apparently sta- 2 the EISF is actually strongly T dependent. bilizes the a phase to much higher concentration at low temperature than in other mclal hydrogen systems, and which leads to a variety of interesting physical properties. In particular we have made quasiciastic-ncutron-scattering studies of proton "hopping" rates between near neighbour H sites along the c-axis in both o-ScHr [1] and a-YHx. The measured temperature dependence of the quasielastic line width F is shovvu for a-ScHole in Fig 1.

1000 •?

Figure 2: The elastic incoherent structure factor (EISF) vs Q, determined from fits of the QENS data on ScHoia at temperatures: 275 K (circles), 218 K (squares), and 175 10 100 1000 K (triangles). The solid lines are fits to the EISF yield- T(K) ing a hopping distance d=l .02 A which corresponds to the near neighbour tetrahedral site separation in scandium. Thp Figure 1: The fitted quasielastic linewidths for ScHoie. dashed line indicates the expected EISF calculated from a The solid line is a Jit of the symmetric nonadiabatic T independent model. weak-coupling model to the data below 100 K.

At all temperatures, the hopping rate exceeds 1011 s"1 This suggests lhat the observed fast moving protons be- and reaches 1012 s"1 at the lowest temperature. Such rapid long to a labile population of hydrogens which is distinct motion at low T implies a relatively free hopping entity. from a second population moving so slowly as to appear The. most dramatic feature of T vs T, of course, is the to be "non-labile" within the timcscalc of the QENS mea- hopping rate minimum near 100 K and the sharp rise be- surements (ic. hopping rates smaller than 3xl010 s"1). We low the minimum which can be fitted with an approximate indentify the non-labile species as protons involved with 1 T' law. A similar minimum is also observed in ScH0 Os. pairing in the short range ordered chains and whose motion ScH025, YHQOI and YH0!. We attribute this behaviour is likely to be determined by strongly asymmetric T-site to the nonadiabatic effects of weak coupling of the pro- potentials. In contrast the fast moving "labile" protons are ton to the metal conduction electrons, as first discussed by identified with the temperature-dependent population not in- Kondo [2]. Increasing T in this regime causes a decrease volved in pairing and thus moving relatively freely between in the relevant density of coherent electron-hole excitations approximately symmetric T-sitc wells. A simple statistical responsible for proton delocalization and thus to an anoma- model for the temperature dependence of this labile popula- lous slowing down of the proton. Eventually (near 100 K in tion based on the assumption of a well defined energy dif- this case) proton-latlice effects dominate, and the hopping ference between labile and non-labile configurations leads rate increases with T producing the minimum. to a reasonable description of the temperature dependence Analysis of r at low T with the nonadiabatic model, in- of the EISF as is shown in Fig. 3 for a-YH* (x=0.0l dicates lhat these rapidly moving protons are subject to an and 0.1). It is interesting to note that the fitted energy dif- essentially symmetric two-site potential. However in this ferences compare well with binding energies for ordering case one would expect lhat the clastic incoherent structure reported in earlier resistivity studies [3]. 17

References

[I] I.S. Anderson, N.F. Berk, 1 J. Rush, T.J. Udovic, R.G. Barnes, A. Magcrl, and D. Richicr. Phys. Rev. Lett. 65, 1439 (1990)

[2J J. Kondo, Physica B 125, 279 (1984); ibid 126, 377 (1984); ibid 141, 305 (1986)

[3] J.P. Burger, J.N. Daou, A. Lucasson, P. Lucasson, and P. Vajda, Z. Phys. Chcm. NF143, Sill (1985)

100 200 300 T(°K)

Figure 3: Temperature dependence of the EISF deiermined at fixed underline Qfor ct-Y/lT at two concentrations. 18

HYDROGEN ORDERING IN YH3

I.S. Anderson', C.M.E. Zcyenf, J.J. Rush'

* Paul Schcrrcr Institul, CH 5232 Villigen PSI, Switzerland t Institut Lauc Langcvin, 38042 Grenoble, Cedex, France $ National Institute of Suindards and Technology, Gaithcrsburg, MD 20899, USA

The unusual ordering of hydrogen isotopes in the em- Krexner, J. Pleschiulschnig, J.N. Daou, and P. Vajda, phases of some hep rare-earth metal hydrogen systems such Phys. Rev. B 39, 5605 (1989); O. Blaschko, J. Pleschi- as YDX, ScDx, and LuDr has been extensively studied by utschnig, P. Vajda, J.P. Berger, and J.N. Daou, Phys. diffuse elastic neutron scattering (DENS) [1,2], following Rev. B 40, 5344 (1989) the observation of resistivity anomalies over ten years ago. The DENS measurements have been performed mainly at [2] M.W. McKcrgow, D.K. Ross, J.E. Bonnet, I.S. An- high deuterium concentrations in the a-phase and reveal derson, and O. Scharpf, J. Phys. C 20, 1909 (1987) essentially'two dimensional intensity distributions perpen- [3] I.S. Anderson, N.F. Berk, J.J. Rush, and T.J. Udovic, dicular to the c-axis which can be modeled by a chain-like Phys. Rev. B 37, 4358 (1988) arrangement parallel to the hexagonal axis. The basis of these chains are pairs of H(D) atoms located on next nearest neighbour tctrahcdral sites along the c-axis and seperated 250 by a metal atom, which are ordered in a zig-zag arrangement with pairs on a neighbouring c-axis. Moreover the observation of localised intensities in the DENS measure- 200 - ments indicates that inter-chain correlations exist so that the ordering has three dimensional characteristics. 150 - Recent inelastic neutron scattering (INS) studies on YHr [3], however, have related the concentration dependence of the splitting of the hydrogen local mode vibrations to a 100 - breakdown in long-range ordering as the concentration is •«•*• lowered. Thus we have made a preliminary investigation of the diffuse scattering from a single crystal of YDo.os on the four 50 • 1 •—i ' 1 '—I ' 1 ' 1 •" circle diffractometer, D10, at the Institut Lauc Langcvin, -0.4 0 0.4 0.8 1.2 1.6 2 2.4 Grenoble, to study the effect of concentration on the ordering. Fig. 1 shows the measured diffuse intensity in a scan along the hexagonal direction [0.65,-0.325,£] at 10 K Figure 1: Diffuse scattering intensity for YDooi along the indicating clear diffuse maxima at approximately £ = 0 and [0.65,-0.325,CI direction at 10 K. C=1.3. The maximum at C='>3 corresponds to the two dimensional diffuse ridge observed previously at higher concentrations and is a signature of the D pairing. The intensity at £=0 would be expected from the localised intensities corresponding to three dimensional ordering, however a scan along the [2£, —£, 0] direction, shown in Fig. 2, demon- strates that this intensity is not localised but extended perpendicular to the c-axis. Thus, although further measurements are in progress to determine the modulation of the o diffuse ridges, these preliminary results indicate, in agreement with the conclusions from the INS studies, that long range order breaks down in YD.,; as the concentration is lowered. In fact the diffuse pattern shown in Fig. 1 resem- bles the simple cosine function that would be expected from individual pairing of D atoms along the hexagonal axis.

References Figure 2: Diffuse scattering intensity for YDaos along the {2(, -C 0j direction at 10 K. [1] O. Blaschko, G, Krexner, J.N. Daou, and P. Vajda, Phys. Rev. Lett. 55, 2876 (1985); O. Blaschko, G. 19

LITHIUM DIFFUSION IN THE SUPERIONIC CONDUCTOR Li2S

F. Altdorfer", W. Biihrer', I.S. Anderson', O. Schiirpf1, H. Bill§, P. Carrot, H.G. Smith"

* Labor fiir Ncutronenstreuung ETHZ, CH 5232 Villigcn PSI, Switzerland t Paul Scherrcr lnstitut, CH 5232 Villigcn PSI, Switzerland | Institut Lauc Langcvin, F-38042 Grenoble GSdcx, France § Department dc Chimie Physique, CH-1211 Geneve, Switzerland % Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA

Li2S shows a high ionic conductivity which might be a consequence of the generation of Frenkel defects due to the mobility of lithium ions. It is expected that such a defect structure would give rise to diffuse scattering which is quasiclastically broadened due to the diffusion. Since lithium has significant coherent and incoherent cross-sections for neutrons, an investigation of the microscopic diffusion process requires the determination, or scpcralion, of the resulting coherent and incoherent quasielaslic components. Such a seperation can be achieved in the case of spin in- coherence by performing neutron polarisation analysis and comparing the spin-flip and non spin-flip scattering. We have performed such an experiment to study the lithium diffusion in an isotopically pure (to reduce absorption), sin- 7 gle crystal of Li2S. The measurements were made on the Time of Flight (TOF) spectrometer D7 at the Institut Lauc Langevin, Grenoble, at temperatures of 300 K and 1200 K. The clastic energy resolution of 200 /JCV was achieved at a wavelength of 4.8 A by selecting a 5mm chopper window and polarisation analysis was made using supermirror ben- ders in the incoming and scattered beams. Figure 1: Raw TOF spectra from Li-jS at 300 K and 1200 Extended elastic scans at 300 K showed no Q dependent K. structure in the diffuse scattering, and furthermore the width in energy was resolution limited. On the contary, at 1200 K significant quasiciastic broadening was observed as shown in Fig. 1, where raw TOF spectra at 300 and 1200 K are compared at a momentum transfer Q=2.53 X"1. Polari- sation analysis at this temperature yielded the interesting result that the observed quaielastic scattering is almost en- tirely incoherent, arising from tracer-like lithium diffusion. This observation, besides indicating that co-operative diffusion effects are not present will allow a simple analysis of the diffusion process to be made. 20

DYNAMICS OF COMPLEX CATIONS IN SODALITES

W. Biihrcr, I.S. Andersont, j. Felsche1, P. Sieger1, F. Ricutord§ * Labor fur Neuironcnsireuung ETHZ, CH 5232 Villigcn PSI, Switzerland t Paul Schcrrcr Institut, CH 5232 Villigen PSI, Switzerland t Universilat Konstanz, Fakultat fiir Chemie, D-7750 Konstanz, Germany § Institut Lauc Langevin, F-38042 Grenoble Ccdex, France

The hydroxy-sodaiile Nas(AlSiO4)6.(OH)2 is a ralhcr simple member of the big zeolite family in which the sodium cations are located on the vertices of a tetrahedron each bonded to three six-ring framework oxygens. The oxygen of the (OH)-group occupies the central position of the sodalite cage and is tctrahcdrally surrounded by the four sodium ions. The hydrogen is also located on the vertices of a tetrahedron, statistically yielding a di-tetrahedral configuration with respect to the sodium. The bonding distance to the central oxygen is 1.09 X. Preliminary experiments have shown that although the hydrogens are statistically frozen in at low temperature, at higher temperatures thermally activated jumps occur between the four sites corresponding to 'rotations' of the rigid OH-dumbcll about it's 'heavy' part, the oxygen atom. We have made a detailed investigation of the dynamics of this (OH)-rotational diffusion by means of quasielastic neutron scattering experiments performed both at Saphir and the IN5 speedometer at the Institut Laue Langevin, Greno- Figure 1: The EISF determined from Naa(AlSiO4)e.(OH)2 ble. Measurements were performed on INS with incoming as a function of momentum transfer, Q.for different tem- wavelengths of 3.5, 5, and 8 X in the temperature range peratures and incident wavelengths. between 150 and 620 K. The resulting elastic incoherent structure factor (EISF) is plotted as a function of momentum transfer, Q, in Fig. 1. It is clear that all the data follow the same curve, in- depcndenl of temperature and incident wavelength (resolution), leading us to the conclusion that there is only one thermally activated process in the temperature range investigated. Specifically there is no evidence of a relaxation or 'memory effect', which might arise from a deformation of the surrounding Na tetrahedron, or such a relaxation occurs with the same time constant as the basic jump process. The comparison of the measured EISF with the theoretical curve, based on a jump model with 4 equivalent sites, is displayed in Fig. 2. Theory and experiment disagree with respect to the modulation; in other words, the jump distance of 1.78 A, given by the crystal structure and the (OH) bond length, o i is too long. Furthermore, a second drop in the EISF is expected near Q=4 A"1, a momentum transfer which is beyond the range of IN5. Further experiments are in progress to re-examine the struc- Figure 2: Comparison of the experimentally determined ture, in particular the O-H bond-disiancc, and to determine EISF, circles, and a model calculation, solid line, for jumps the EISF at large momentum transfer, Q > 4 A. between tetrakedrally coordinated sites seperated by a distance of').78 A. 21

DYNAMICS OF WATER MOLECULES IN SODALITES

W. BUhrer", I.S. Anderson', J. Felsche1, P. Siegert, p Ricutord5 * 1 abor fur Neutronenstreuung ETHZ, CH 5232 Villigen PSI, Switzerland t Paul Scherrer Institut, CH 5232 Villigen PSI, Switzerland t Universitat Konstanz, Fakultat fiir Chemic, D-7750 Konstanz, Germany § Institut Laue Langevin, F-38042 Grenoble Cedex, France

The hydro-sodalite Na6(AlSiO4)6-8H2O has an average structure which cubic, in which the 3 sodium cations and formally one Na-vacancy are located on the vertices of 975- a tetrahedron. According to the structural data, the oxygen * *. of the water molecules shows positional disorder and occu- 950-

pation of different H positions can partly be accounted for 925- by re-orientations of the water molecules (the refinement was made without constraints). We have made a detailed 9QD- investigation of the dynamics of these water re-orientational S75-J motions by means of quasielastic neutron scattering. Measurements were performed on IN5 spectrometer at the Institut Laue Langevin, Grenoble, at incident wavelengths of 5 and 8 A in the temperature range 125 to 350 K. The results for the elastic incoherent structure factor (EISF) and the quasielastic linewidth are shown in Figs. 1 and 2 respectively as a function of momentum transfer, Q, for various temperatures and incident wavelengths. Since the EISF is

always larger than 0.8, we are led to assume that only one Figure 1: EISF for Na$(AlSi04h.8H2O as function ofmo- water molecule can perform re-orientational motions, the mentum transfer, temperature and incident wavelength other 3 being frozen in by the nearby Na-vacancy. Addi- tionally the EISF varies as a function of temperature, sug- gesting that re-orientations occur in a 'two step' process. This assumption is further supported by the observation of a resolution (wavelength) dependent linewidlh indicating the presence of two quasielastic components of differing witivhs. These observations suggest the following picture: oniy the water molecule which is opposite to the Na-vacancy per- forms re-orientational motions. The first step (T < 200 K) corresponds to the movement of only one 'leg' of the H2O; with increasing temperature, a second movement, where both 'legs' are involved, sets in. The present structural data only accounts for hydrogen positional disorder of the second process, i.e., there are no • 250 hydrogen sites reachable by a 'one leg" movement of a • 300 quasi-rigid molecule. Thus a re-examination of the structure including refinement with constraints, allowing only translation and/or rotation of the quasi-rigid H2O molecule, is in progress. O [reciprocal

Figure 2: Quasielastic linewidlh forNa6(AlSi0U)^-8H2O as function of momentum vanrfzr, temperature and incident wavelength. 22

Comparison of the Paramagnetic Spin Fluctuations in Ni with Asymptotic Renormalisation Group Theory

P. BOnito, G. Shirane', J.L. Martinez50, H.A.

f Paul Schetrer Institute, CH-5232 Villigcn PS1 t Brookhaven National Laboratory, Upton, New York 11973 § Institute Laue Langevin, F-38042 Grenoble Cedex, France % Solid State Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830

Many years ago Mook et al. [1] measured the differen- Ni, when compared with purely exchange coupled systems, tial neutron scattering cross section in paramagnetic Ni and most likely due to the conduction electrons. discovered well defined peaks at finite momentum transfer Research at Brookhaven is supported by the Division <7o when the energy transfer (hu > 12 meV) from the neu- of Material Sciences, U. S. Department of Energy under tron to the spin system was kept fixed (constant-E scans). contract DE-AC02-76CH00016. These peaks seemed to define a dispersion like curve and were thus interpreted as spin waves persisting into the paramagnetic phase. A re-investigation of the paramagnetic References scattering revealed, however, that constant-E peaks are a general property of isotropic ferromagnets [2] and that their [1] H. A. Mook, J. W. Lynn, and R. M. Nicklow, Phys. Rev. Lett. 30, 556 (1973). location qa is well reproduced by hua = ]-27Aq^, z = 2.5, independent of the degree of localization of the magnetic [2] Y. J. Uemura, G. Shirane, O. Steinsvoll, and J. P. moments (A is a spin diffusion constant) [3]. Wicksted, Phys. Rev. Lett. 51, 2322 (1983). This observation was at first rather puzzling because a Lorentzian spectral weight function yields at Tc for the [3] P. Boni and G. Shirane, J. Appl. Phys. 57, 3012 peak position hu>o = 3F. The observed discrepancy be- (1985). tween prediction and experiment was shown by asymptotic renormalisation group theory to be caused by the non- [4] R. Folk and H. Iro, Phys. Rev. B 32, 1880 (1985). Lorcntzian shape of the spectral shape function [4]. The [5] R. Folk and H. Iro, Phys. Rev. B 34, 6571 (1986). extension of this theory to T > Tc by Folk and Iro [5] and Iro [6,7] predicts that i) the position q0 is almost inde- [6] H. Iro, Z. Phys. B - Cond. Malt. 68, 485 (198/). pendent of T and ii) the width Sq of the constant-E peaks increases with T. [7] H. Iro, J. Magn. Magn. Mater. 73, 175 (198?). In order to test the proposed spectral weight function and the underlying physics we performed neutron scattering measurements in the paramagnetic phase in Ni in the range 1 < hu> < 45 meV and for temperatures ranging from Tc lo 1.48 Tc {Tc = 631 K). The experiments were conducted at the HFBR at Brookhaven National Laboratory. The inset of Fig. 1 shows a typical scan for hu = 1 meV as measured in 64Ni at 1.19 Tc. Near Tc the peak is well defined. Its intensity decreases and the peak broadens with increasing T. Initially, the peak position docs not change much as expected on the basis of the theoretical work. However, at 1.19 Tc the peak shifts to smaller q in contrast to the theoretical prediction. In order to compare our data with the theory in more detail we have plotted in Fig. 1 the scaled position q$ — OA OA qo(A/hu) and width Sg^ = 6qo{A/hu) versus the scaling variable r* -i0 \og(1iuj/AK\5) (r* -» oo for T —> Tc), and compare them with the results of Ref. [7). First of all the data exhibits scaling behaviour with respect to position and width over a large range in T and u (note the logarithmic scale), indicating that the scaling variables used are reasonable. Second, there are serious, systematic discrepancies for r* < 0, i.e. at high T and small u, as observed already in the raw data (inset). A detailed investigation shows that the main reason for the disagree- ment is the additional damping of the spin fluctuations in * also at Brookhaven National Laboratory, Upton, N.Y. 11973. U.S.A. 23

a) o 60Ni low energy

* 60Ni high energy

1 1.02 1.04 1.06 1.08 1.1 reduced wavevector

0123 r=10log(E/Ax2 5)

Figure 1: Scaling pious for a) position q^ andb) width Sqg of constant-E peaks versus r*. The bold solid lines are the predictions of asymptotic renormalisation group theory, the thin solid lines are based on the parametrizations of the dynamical scaling function. The dotted lines correspond to the Lorentzian approximation for the spectral weight function. The inset shows a constant-E scan for hw = 1 meV conducted in 64Ni. The bold solid line is the prediction of asymptotic re •.. malisation group theory and the thin solid line represents a parametrization. At high temperature the asymptotic renormalisation group theory cross section is equivalent to the Lorentzian approximation. 24

Dynamics of Longitudinal and Transverse Spin Fluctuations Below Tc in EuS

P. Boni", D. Gorlitz', J. Kolzlcr', J.L. Marn'ncz1 * Paul Scherrer Institute, CH-5232 Villigen PSI t Universitat Hamburg, D-2000 Hamburg 36, Germany t Institute Laue Langevin, F-38042 Grenoble Ccdex, France

The availability of polarized neutrons has enabled in- [2] F. Mczci, Phys. Rev. Lett. 49, 1096 (1982). vestigations of dipolar effects on the order parameter fluctuations above the Curie temperature Tc. These (demag- [3] E. Frey and F. Schwabl, Z. Phys. B 71, 355 (1988), netisation) effects are present in all ferromagnets. They E. Frcy, F. Schwabl, and S. Thoma, Phys. Rev. B 40, heavily suppress the criticality of the longitudinal compo- 7199 (1989). nent, SSq || q, below the dipolar wavevector q < qd, which [4] J. Kotzlcr, Phys. Rev. B 38, 12027 (1988). was recently delected arount a finite Bragg reflection of EuS and EuO [1]. Mezci's observation of violation of dy- f5] P. Boni, G. Shirane, H. G. Bohn, and W. Zinn, J. namical scaling for the transverse fluctuations in Fe [2] has Appl. Phys. 61, 3397 (1987). prompted a mode-coupling theory [3], which by taking full account of the dipolar forces could no! only explain ihe salient features of Mezei's data but also the critical dynamics of the homogeneous magnetization [4]. The goal of our experiment was a novel check of the mode-coupling theory which predicts relaxation rates and lincshapcs for the longitudinal fluctuations being distinctly different form the transverse ones. The longitudinal fluctuations can be separated from the transverse fluctuations by measuring the spin flip cross sec- a) EuS tions in a vertical field and in a horizontal field, respec- 00 - q = 0l9 A1 - tively, at a point in reciprocal space which is displaced A T = 1 006 T transverse from a Bragg peak. In Fig. 1 typical spectra for j longitudinal the longitudinal and the transverse fluctuations for q - 0.10 50 •i J 1 X~ (qd = 0.22 A" ) arc shown. Obviously the intensity j/ of the longitudinal fluctuations is significantly suppressed 0 in agreement with previous work [1], and their lincwidth I 1 1 is larger. The transverse data agree equally well with the I J predicted transverse line shape and the Lorcntzian profile. b) In contrast, the longitudinal data is incompatible with the A 00 predicted longitudinal line shape, which is loo square near E = 0. To provide a direct comparison of the dynamics with 50 the mode-coupling results we have normalized all fitted Lorcnlzian widths with the line width measured at Tc, 0 J which is given by T = Aq2 5, A = 2.1 ±0.3 meVA2 5[5), and plotted them versus the scaled inverse correlation length < i I ! >c/q in Fig. 2. The solid lines indicate dynamic scaling -O.3 -O2 -0 1 O 0] 02 03 04 functions of the mode-coupling theory. This representation E (meV) does not involve any free parameters and obviously, within experimental error, theory and experiment arc in excellent agreement. This is true in particular for the longitudinal data for which our experimental conditions were optimized. Figure 1: Constani-q scans probing paramagnetic fluctu- We are indebted to M. Koch and R. Thut of the Lab- ations with q = 0.19 A~' very close to Tc. Data were oratory for Neutron Scattering ETH at PSI for the skillful filled alternatively to longitudinal and transverse spectral construction of the sample holder. It enabled us to align weight functions of mode-coupling theory (thick lines) and the 40 single crystallites within 0.8°. to Lorenlzians (dashed lines). Thin lines depict the uncon- volutcd mode-coupling line shape. References

[1] J. Kotzlcr, F. Mezci, D. Gorlitz, and B. Farago, Euro- phys. Lett. 1, 675 (1986). 25

EuS b)

1 006T 1 • J I 038T (a I 1 1267" ( * J

x/q

Figure 2: Comparison of the longitudinal and transverse line widths of Lorentzian fits normalized to Aq2i, A = 2.1 meVA25with the dipolar dynamic scaling function predicted by mode-coupling calculations /3J. 26

Nuclear and Magnetic Structure of the permanent Magnet Material Tb2Fe14C

Ch. Hellwig', K. Girgis', J. Schefer*, P. Fischer*, K.H.J. Buschow?

* Instiiut fur Kristallographie und Petrographie, ETH Zurich, CH-8092 Zurich t Solid Slate Physics (F3), Paul Scherrer Institute, CH-5232 Villigen PSI t Labor fiir Neutronenstreuung, ETHZ, c/o PSI CH-5232 Villigen PSI § Philips Research Laboratories, NL-5600 JA Eindhoven

In the course of investigations of new cobalt free per- diffractomeier at reactor SAPHIR using the 110 reflection. manent magnet materials H^Fe^C has been studied met- The Tc determined by this method is about 25 K higher allographically, by X-ray and by neutron diffraction me- than determined by macroscopic measurements [3]. thods. TboFe^C crystallizes in the tetragonal space group This work has been supported by "Swiss National Sci- P42/mnm (No. 136) with 68 atoms per unit cell. Neutron ence Foundation", Nationalfonds zur Forderung der wis- diffraction measurements were carried out at 23 K, 295 K, senschaftichen Forschung. 585 K (near the Curie-Temperature), and at 723 K (above Tc). The measurements were performed on the DMC- speclromcter [1] at the reactor SAPHIR (PSI). Nuclear and References magnetic parameters of the Rietveld refinement (c.f. figure 1) are summarized in table 1 (23 K only). Structure param- [1] 1. Schefer et al., Nucl. Instruments Methods A288, eters are comparable to Nd2Fei4C [2]. 477 (1990). The magnetic moments of the Tb-atoms are orderd an- [2] R.B. Helmholdt and K.H J. Buschow, J. Less.-Comm. liparallel to those of Fe-atoms. The sample contains traces Met. 144, L33 (1988). of iron and another magnetic phase Tb2Fei7 (space group No. 166), therefore a three phase Rietveld calculation has [3] M. Gueramian, A. Bczinge, K. Yvon and J. Mullcr, been carried out. Solid State Commun. 64, 639 (1987). The magnetisation curve was measured on the two-axis

i III » miii iiIII i III so. i i II i mi in in II 11 II in laiiiniiimiii iiiHinffliiiiiHimiiiiii in

45.

40. obs h 35- calc dlff ;• ao. - I 25. - Tb2Fe,.C, 23 K,

g 20. - 1.707 A, P^/mnm I «•"

10.

5. -

0. —'iW*—--v

40 60 BO 100 120 140 2-Theta (Degrees)

Figure 1: Observed and calculated neutron diffraction pattern of Tb%Fe\AC at 23 K. Neutron wavelength X = 1.707 A. Refinement parameters are summarized in table 1. Additional peaks are due to iron (middle row) and to Tb2Fen (lower row). 27

Table 1: Nuclear and magnetic parameters of Tb2Fe14C at 23 K

atom pos. Tb 1 4f 0.2592(6) 0.2592(6) 0.000 -9.2(2) 0.23(14) Tb2 4g 0.1428(6) 0.8572(6) 0.000 -9.1(2) 0.23(14) Fe 1 16k 0.2257(5) 0.5635(6) 0.1217(4) 2.6(1) 0.39(2) Fe2 16k 0.0346(5) 0.3588(6) 0.1748(4) 2.8(1) 0.39(2) Fe3 8j 0.0968(5) 0.0968(5) 0.2041(5) 3.1(2) 0.39(2) Fe4 8j 0.3156(4) 0.3156(4) 0.2458(5) 3.7(2) 0.39(2) Fe5 4e 0.000 0.000 0.6084(7) 2.0(1) 0.39(2) Fe6 4c 0.000 0.500 0.000 2.6(1) 0.39(2) C 4g 0.3715(11) 0.6285(11) 0.000 0.0 0.98(26) - The second phase Tb2Fei7. Tb 1 6c 0.000 0.000 0.342(4) -9.0 0.23(14) Fe 1 6c 0.000 0.000 0.103(3) 2.177 0.39(2) Fe2 9d 0.500 0.000 0.500 2.177 0.39(2) Fe3 18f 0.284(2) 0.000 0.000 2.177 0.39(2) Fe4 18h 0.501(2) 0.499(2) 0.162(3) 2.177 0.39(2) cell constants: Tb2Fei4C: a,b = 8.78(1) A, c = 11.86(11) A Tb2Fe]7: a,b = 8.60(2) A, c = 12.48(2) A Occupation (Tb): 93(2)% rest: 100% = ^•nuclear = 3.1%; ^magnetic 3.5%; R-profile = 5.(Wo; Rexpected = 2.2% 28

Nuclear and Magnetic Structure of the permanent Magnet Material Ho2Fei4C

Ch. Hellwig", K. Girgis', J. Schefcrt, P. Fischer*, K.HJ. Buschow$

* fnstitut fiir Krisiallographie und Pcuographie, ETH Zurich, CH-8092 Zurich t Solid Scale Physics (F3), Paul Scherrer Institute, CH-5232 Villigen PSI t Labor fiir Neutronenstreuung, ETHZ, c/o PSI CH-5232 Villigen PSI § Philips Research Laboratories, NL-5600 JA Eindhoven

In the course of investigations of new cobalt free per- at 50 K. We found no spin reoricntation at temperatures manent magnet materials Ho2Fei4C has been studied mct- below 35 K as predicted in [3]. allographically, by X-ray and by neutron diffraction meth- This work has been supported by "Swiss National Sci- ods. Ho2Fci4C crystallizes in the tetragonal space group ence Foundation", Nationalfonds zur Forderung dcr wis- P42/mnm (No. 136) with 68 atoms per unit cell. Neutron senschaflichen Forschung. diffraction measurements were carried out at 10 K, 50 K, 295 K. The measurements were performed on the DMC- spectrometcr [I] at the reactor SAPHIR (PSI). Nuclear and References magnetic parameters of the Rietveld refinement (c.f. figure 1) are summarized in table 1 (50 K only). Structure [1] J- Schefer et al., Nucl. Instruments Methods A288, parameters are comparable to Nd2Fe]4C [2]. 477 (1990). The magnetic moments of the Ho-atoms are orderd an- [2] R.B. Hclmholdt and K.H.J. Buschow, J. Less.-Comm. tiparallcl to those of Fe-atoms. The sample contains a Met. 144, L33 (1988). second magnetic phase Ho2Fei7 (space group No. 194), therefore a two phase Rietveld calculation has been carried [3] F.R. dc Boer, R. verhocf, Z.-D. Zhang, D.B. de Mooij out. and K.H.J. Buschow, J. Magn. Magn. Mater 73, 263 The refinement parameters ?l 10 K are similar to those (1988).

30. i in ii mi lining iriiiniiiii i i II 31 nil i II i nun; raiim man IO ii:nninmrai iioiu minicmminin IIMIIIII.'EIII

25. -

obs 20. - calc dirt

c 15. - Ho2Fe,,C, 50 K,

1.706 A, P42/mnm

1.0 - 0.0 1.0 60 80 120 140 2-Theia (Degrees)

Figure 1: Observed and calculated neutron diffraction pattern of Ho2FeuC at 50 K. Neutron wavelength A = 7.706 A. Refinement parameters are summarized in table 1. Additional peaks are due to Ho

Table 1: Nuclear and magnetic parameters of Ho2Fei4C at 50 K

atom pos. Ho 1 4f 0.2586(5) 0.2586(5) 0.000 -9.8(1) 0.41(8) Ho 2 4g 0.1432(5) 0.8568(5) 0.000 -9.8(1) 0.41(8) Fe 1 16k 0.2254(5) 0.5629(5) 0.1210(4) 2.5(1) 0.54(2) Fe2 16k 0.0345(4) 0.3583(5) 0.1748(4) 2.9(1) 0.54(2) Fe3 8j 0.0977(5) 0.0977(5) 0.2020(4) 3.0(1) 0.54(2) Fe4 8j 0.3164(4) 0.3164(4) 0.2448(5) 3.7(1) 0.54(2) Fe5 4e 0.000 0.000 0.6090(7) 2.0(1) 0.54(2) Fe6 4c 0.000 0.500 0.000 2.3(1) 0.54(2) C 4g 0.3705(10) 0.6295(10) 0.000 0.0 0.69(21) The second phase Ho 1 2b 0.000 0.000 0.250 -9.6(2) 0.41(8) Ho 2 2d 0.333 0.667 0.750 -9.6(2) 0.41(8) Fe 1 4f 0.333 0.667 0.100(1) 1.9(1) 0.54(2) Fe2 6g 0.500 0.000 0.000 1.9(1) 0.54(2) Fe3 12j 0.337(1) 0.965(1) 0.250 1.9(1) 0.54(2) Fe4 12k 0.166(1) 0.332(1) 0.992(1) 1.9(1) 0.54(2) cell constants: Ho2Fe14C: a.b = 8.760(4) \ c= 11.803(6) X Ho2Fei7: a,b = 8.526(5) K c = 8.353(5) A Occupation (Ho): 93% rest 100% ^nuclear = 4.2%; ^•magnetic ~ 4.4%; ^•profile = 5.f1%: Kxvecled = 3.0% 30

Light-Induced Structural Changes in Sodiumnitroprusside (Na2(Fe(CN)5 NO) 2 D2O) at 80 K

J. Schefcr', M. Riidlingert, T. Vogt*, T. Woikc5, S. Haussiihl§, H. Zollner§, H.U. Giidel", N. Furrcr", P. Schweiss", G. Heger", G. Chevrier"

* Solid State Physics (F3), Paul Scherrer Institute, CH-5232 Villigcn PSI t Laboratory for Neutron Scattering, ETH Ziirich, CH-5232 Villigcn t Insitute Laue Langevin, F-38042 Grenoble, France § Institute for Crystallography, University, D-5000 Koln, Germany f Inorganic Chemistry, University, CH-3006 Bern ** Laboratoire Leon Brillouin, CEN, F-91191 Gif sur Yvettc, France

Sodiumnitroprusside can be transformed into an extreme- along the N1-C1-FC-N4-O1 bond. The bond distance of ly longliving (r > 107 s) metastablc state [1] at tempera- Fe-N4 changed by 0.019(2) X and simultanously the one tures below 190 K, using polarized light of wavelengths of Fe-Cl by 0.012(3) X, see fig. 1. These changes arc 3500 X < A < 5600 X. Beside the groundstatc (GS) there significant as they arc approximately 10 times larger than are two known excited states (MS; and MS/;) [2). Popu- the estimated errors. AH other interatomic distances for the lations of both states are strongly dependent on wavelength two stales are unchanged within standard errors. and polarization direction of the light. With the wavelength For a calculation of the interatomic distances on the ba- used in this study (A=4880 X) not more than 45% of the sis of the Badger rule [4] corrected by Herschbach & Laurie molecules can be transformed into the excited state MS;. (5] we used the relation Two data sets were collected at the ILL: Set 1: GS and set 2 (GS/MS;). The metastable state has been populated (1) using monochromatic laser light of A = 4880 X with the polarization vector E parallel to c during 36 hours with a 150 mW/cm . The population was done at 80 K. The re- where rg, and rms are the equilibrium bond lengths in sulting change in the optical absorption was checked using the two states, fc;j is a constant for bonds between atoms a photodiode, situated behind the crystal. The meiastable from row j to row j of the periodic table [51 and Vs,, Vm, state MS;;of data set 2 was quenched by increasing the frequencies of the vibrational modes. temperature to 160 K for 10 minutes without affecting the population of MS;. Vibrational frequencies have been measured by Raman and infrared spectroscopy [6]. Bond lengths of the excited state The refinement used anisotropic temperature factors for rms,ob.< have been linearly extrapolated from values ob- all atoms. 96 parameters have been refined simultaneously. served in the groundstale and the mixed state (55% GS, A limited number of constraints has been introduced, such 45% MS;). Neutron, Raman and infrared data are in fair as identical deutcration of both D-sitcs and identical B- agreement (cf. table 2). factors for the two Natrium and the two Deuterium atoms, A full publication of our results is in press [7]. respectively. No absorption correction was applied to both data sets. An isotropic extinction correction of type I according to the Bccker-Coppens formalism [3] was used. As expected structural changes in the metastablc state Table 1: Interatomic distances in [A]. were small. It was therefore important to determine the structures at the same temperature (80 K) and under identi- Data Set 1 2 cal experimental conditions. Data set 1 has been collected States GS GS/MS; after quenching the mctastablc stale MS/Of data set 2 by Temp. [K] 80 80 increasing the temperature to 250 K for 10 minutes. Data N4-O1 1.1293(36) 1.1333(28) collection of sets 1 and 2 were performed under identical Fe - N4 1.6643(24) 1.6828(21) instrumental conditions . Fe-Cl 1.9204(30) 1.9087(27) Fe-C2 1.9294(21) 1.9261(17) The refinement of the structure of the excited state is limited by the fact, that expected changes arc small and that Fc-C3 1.9367(21) 1.9377(17) 1.1576(26) at least 50% of the molecules are still in the ground stale. Nl -Cl 1.1593(31) We found to our knowledge for the first time small but sig- N2-C2 1.1541(22) 1.1538(18) nificant structure changes in bond distances and temperature N3-C3 1.1542(22) 1.1538(18) factors in an electronically excited state. Interatomic dis- O2-D1 0.9428(88) 0.9539(68) tances and selected bond angles are tabulated in table 1. It O2-D2 0.9481(83) 0.9530(68) should be emphasized that in the case of the metastablc state Dl -D2 1.495(11) 1.5140(92) we are always talking about average positions of the atoms. As expected from Raman data we found the most significant changes between GS and MS;on the bond lengths 31

Table 2: Vibrational frequencies of the ground- and metasiablc state MS/ observed by Raman and infrared spectroscopy [6] rm,,Caic are the interatomic bond distances calculated from rg, (table 1) using equation (1), ;•„„,o63 has been extrapolated from our observed neutron diffraction data listed in table 1 (GS, GS/MSi), assuming a 45% N! population of the mctastable state.

Bond bij[S\ [cm-1] [cm-'] •- •. c- 1*1 TA] [A] Nl N4-O1 1947 1835 0.470 1.15 1.14 N4-Fe 657 565 0.600 1.74 1.71 Cl-Fe 427 454 0.600 1.89 1.89

GS References [1] T. Woike, W. Krasser and P.S. Bechthold, Phys. Rev. Lett. 53, 17 (1984) [2] H. Zi»)ner, T. Woike, W. Krasscr and S. HaussiiM, Z. fur Kristallographie 188, 139 (1989) [31 PJ. Becker and P. Coppcns, Acta Crystallogr. A30, 129 (1974) I i\i [4] R.M. Badger, J. Chcm. Phys. 2, 128 (1934) [5] D.R. Herschbach, V.W. Laurie, J.Chem.Phys. 35, 458 (1961)

[6] W. Krasser, T. Woike, T. Woike, S. Haussiihl, J. Kuhl I GS/MS, and A. Breitschwerdt, J. Raman Spectr. 17, 83 (1986) S'jilininnitroprussiclL1 [7] M. Rudlinger, J. Schefer, T. Woike, T. Vogt, P. Schweiss, O. Chevrier, G. Hegcr, H.U. Gudel, Figure 1: ORTEP-plot of the average positions and H. ZOllner, S. Haussiihl and N. Furer, Z. Physik B anisotropic temperature factors of the prusside ion in (a) - Condensed Matter (1991, in press) the groundstatc (GS, data set 1) and (b) the mixed state (GS/MS/, data set 2). 32

SANS and TEM Investigation of Age Hardened Nimonic PE16 after Cyclic Loading at Room Temperature

W. Wagner*, A. Wiedenmann', W. Chen*, R.P. Wahi', M. Sundararamant, W. PetryS

* Paul Scherrer Institute, CH-5232 Villigen PSI t Hahn-Meinter-Institute, Glienickcr Sirasse 100, D-1000 Berlin 39 t B.A.R.C., Phys. Mctall. Dcpt., Bombay 400 085, India § ILL Grenoble, B.P. 156X, F-38042 Grenoble, France

Machine components such as blades and discs of a formed during loading, they would contribute with negative turbine roior are exposed to complex loading conditions values to the intensity difference at the location in Q which resulting from time dependent and superimposed thermal, corresponds to their size. mechanical and corrosive loading. The microscopic mech- Since in the whole Q-range under investigation no neg- anisms of deformation and damage in the component ma- ative difference was detected, it can be concluded that the terial essentially determine its life. An evaluation of these formation of new precipitates during deformation with a mechanisms is therefore necessary for a basic understand- radius larger than 0.5 nm can be ruled out. Instead, the ing of the material and component behaviour under such SANS results reveal that a relative volume fraction of 6 to loading conditions. 8 % of the initially present precipitates have disappeared Nickel base, age hardenable alloys, like the NIMONIC by complete dissolution during deformation. group, are nowadays the most frequently used materials Among the other possible explanations, neither the dis- for such components. They are strcngthed by precipitates ordering nor the incomplete slicing into small fragments of the Llo-ordercd 7'-phase of composition Ni3 (Al, Ti). (below the resolution limit of TEM) can be relevant in the Fig. 1 presents TEM micrographs of a specimen which present case. The former would not change the SANS in- had undergone low-cyclc-fatigue (LCF) loading. The mi- fens'ty of precipitate scattering, the latter would give rise crographs show that a portion of the bright imaging, spher- to negative differences at larger Q. In contrast, in the range ical 7'-prccipitates have been cut in slices or pieces during of Q > 0.3 nm"1, small but overall positive intensity dif- deformation. fei Ciices are measured. In the undcraged and peak aged conditions these alloys arc found to exhibit cyclic softening which is usually associated with localisation of plastic strain within the so-called References persistent slip band (PSBs) or deformation bands. In the [1] M. Wilhelm, Mater, Sci. Engng., 48, 91 (1981) TEM image (Fig. 2) these deformation bands appear free of precipitates. However, it is also possible that precipitates [2] C.A. Slubbington, Acta metall., 12, 931 (1964) arc present but invisible due to disordering. Both mechanisms, i.e. precipitate dissolution by sharing and fragmenta- [3] D. Sterner, R. Beddoe, V. Gerald, G. Kostorz and R. tion [1,2,3] and precipitate disordering [4,5,6], arc possible Schmelezer, Scripta metall., 17, 733 (1983) causes of softening, controvcrscly discussed in literature. [4] C. Calabrese and C. Laird, Mater. Sci. Engng., 13, Since TEM-invcstigations can not satisfactorily contribute 141 (1974) to a solution of this question, small angle neutron scattering (SANS) was applied in complement. [5] R.E. Stoltz and A.G. Pineau, Mater. Sci. Engng., 34, SANS-samples were prepared from fractured LCF spec- 275 (1978) imens by cutting slices perpendicular to the loading axis. From each LCF specimen, two of such slices were pre- [6] B.A. Lcrch and V. Gcrold, Acta metall., 33, 1708 pared, one from the gauge which had undergone the fatigue (1985) deformation, and the other from the unfatigucd head portion which served as reference sample. Since most of the 7'-precipitates survive the fatigue treatment (see Fig. 2), the original SANS-pattern from both portions of the LCF- specimen looked very similar. The differences become obvious after subtraction of the scattering data of the fatigued portion of the specimen from those of their unfatigucd reference. This yields the scattering intensity differences shown in Fig. 3. By definition positive intensity differences represent the structure function of precipitates which have disappeared during deformation. Merely disordering without dissolution would not originate intensity differences in the SANS curves. Further, if a new class of precipitates would have 33

Figure 1: TEM dark field mio-ograph of sheared precipitates of 7''-phase in Nimonic PE16 after LCF deformation

1.5

1.0

0.5

v—' 0.0 CO <\^^\ _ .5 CD o -1 .0

_ -I .5

-2.0 -1.0 -0.5 0.0

log (Q/nm~1)

Figure 3: SANS result on fatigue treatment of Nimonic PE16, given as intensity difference between a specimen after Figure 2: An overview of the same specimen as in Fig. 1 LCF~loading and an unloaded reference specimen. showing slip bands without ^'-contrast. 34

Sintering Characteristics of Nanocrystalline TiO2 - A Study Combining Small Angle Neutron Scattering and Nitrogen Absorption - BET

W. Wagner", R.S. Averback;, II. Halm', W. Petry', A. Wicdcnmann'

* Paul Scherrcr Institute, CH-5232 Villigcn PSI t Department of Materials Science and Engineering and Materials Research Laboratory, University of Illinois, 1304 W. Green Street, Urbana, 111 61801, USA } lnsilule Lauc Langevin, 156X Centre dc iri, F-38042 Grenoble, France ij Hahn-Mcintcr-lnsiilute, Glienicker Strassc 100, D-1000 Berlin 39

The structures of nanophasc materials arc rather unique ordinatc scale is given i:i units of m3/(g-nm) in order to since grain boundaries comprise a significant fraction of Ihc facilitate direct comparison with the BET results shown in total volume of the material. It is noi surprising, therefore, Fig. 3. that ihc properties of nanophase materials are oncn very dif- In general, the results of the two techniques are in good ferent from those of their counterpart polycrystallinc mate- quantitative agreement in regards to both the size distribu- rials having larger grain sizes. In nanophase ceramics, me- tion and the absolute magnitude of N(R). This agreement chanical properties are of particular interest since it has been suggest that both methods, in principle, arc suitable for a reported that diffusional creep and superplastic behaviour quantitative analysis of pore size, volume distribution, and are possible at low temperatures 11,2,3]. Small grain sizes, volume fraction. however, are not the only important tnicrostructura! fea- On a finer scale, clear differences arc observed between ture in nanophase ceramics that has a strong influence on the BET and SANS results. For the specimen compacted mechanical properties; flaws, such as residual porosity, also at room temperature, a significant volume fraction of pores affect these properties [4]. The microstructurc of nanophasc with diameters between 2 and 8 nm are detected by SANS, r (n-) TiOv d>;ri ig sintering has previously been studied us- but not by BET. Most likely these pores arc enclosed and ing a variety of techniques: gravimclry (using Arehimcdc's unacccssiblc to nitrogen adsorption or condensation. The principle), BET nitrogen absorption, scanning electron mi- portion of closed porosity in the sample compacted at room croscopy and x-ray diffraction [5|. Although this investiga- temperature amounts to a volume fraction of about 10%, tion provided an overview of the microstructural develop- i.e., one third of the total porosity. A fraction of these small ment in n-TiGs during sintering, a detailed understanding pores survive compaction at 290°C, but they arc completely of the pore structure and its evolution in nanophasc ceram- removed during compaction at 413°C. The removal of these ics is still lacking. For the purpose to better characterize pores during compaction and sintering at low temperatures nanophasc ceramics, a study using small angle neutron scat- is expected since small pores have the highest sintering rates tering (SANS) was initiated, complemented by BET, x-ray 18]. diffraction and gravimctry On the lari>c pore end of the distribution, a tail in the The samples employed for these experiments were pro- specimen compacted at 25"C is detected by both tech- duced using a three step process which has been described niques; it extends to about 100 nm in diameter, while the elsewhere [5,6,7]. In the final step of the process, the sam- average pore diameter is 21 nm. Compaction at cither ples arc highly compacted under an applied pressure of ~ 290°C or413°C removes a major portion of this coarse 1.0 GPa at specified temperatures between 25 and 550°C, porosity, presumably by grain boundary sliding processes providing samples of similar grain size but different volume that arc thermally activated. The pores that survive these fractions of porosity. The SANS scattering intensities for compaction have a comparatively narrow size distribution, a specimen compacted at room temperature ("as prepared") extending from about 4 to 20 nm in diameter, with the aver- and specimens compacted at higher temperatures arc plot- age at 14 nm, instead of 21 nm after the initial preparation. ted in Fig. 1 as a function of scattering vector, Q, Double When the results of SANS and BET arc in dose agree- logarithmic scales arc employed since the data extend over ment, it is evident that the pores structure is open. It sig- several decades. Compared to the specimen compacted at nifies that ail pores larger than 8 nm in diameter in the room temperature, the scattering from samples compacted specimen compacted at room temperature, and essentially at elevated temperatures, which have many fewer pores, is all pores overcoming compaction at 290°C and 413°C, arc significantly reduced, showing that a large contribution of cither at the surface, or completely interconnected. the SANS scattering docs, indeed, derive from pores. A quantitative evaluation of the SANS data was carried out using a computer algorithm which calculates S(Q) on References the basis of modelled microstructurc, considering contributions from grains and intcrfacial regions, surface roughness, 111 H. Hahn, J. Logas, H.J. HSflcr and R.S. Avcrback, pores, and cavities. Mats.Rcs.Soc. Proc. vol (1989) The best fits for S(Q) derived from our scattering model are plotted in Fig. 1 as solid curves. The evaluation results 12] H. llalin and R.S. Avcrback, Acta. Mctall et Matcria, in the pore volume distributions shown in Fig. 2. The (in press.) 35

[3] H. Karch, R. Birringcr and H. Gleitncr, Nature, 330, 556 (1987) [4J HJ. HOfler and R.S. Averback, Scripta Meiall et Ma- 1GPe/25°C teria (in press.) 1GPa/290°C 1GPa/4i3°C [5] H. Hahn, J. Logas and R.S. Avcrback, J. Mater. Res. 5 (1990) 609 [6] R. Birringer, U. Hcrr and H. Gleitner, Trans. Jpn. Inst. Met. 27 (supp) 43 (1986) [7] R.W. Siegel, R. Ramasamy, H. Hahn, Li Zongquan, Lu Ting and R. Gronsky, J. Mat. Res. 3, 1367 (1988) [8] R.J. Brook, Proc. Brit. Ccram. Soc. 32, 7 (1982)

100 pore diameter (nm)

Figure 2: Pore size distribution in nanocrystalline TiO? after compaction at room temperature and elevated temperatures as indicated, derived from the SANS results.

6 • 1GPa/25°C I A 1GPa/290°C o 1GPa/413°C E ci -1 0 C 4 log (Q/nrrr1) o 3 > Figure 1: SANS intensity data of nanocrystalline TiO2 af- O o ter compaction at room temperature ("as prepared") and Q ^ elevated temperatures as indicated. The solid curves are 75 modelled structure functions fitted to the data.

o £ 0 10 100 pore diameter (nm)

Figure 3: Pore & e distribution in nanocrystalline TiO2 after compaction at temperatures as indicated, determined by BET. 36

Early Stages of a-Co-Precipitation in Dilute CuCo Alloys

W. Wagner*

* Paul Schcrrcr Institute, CH-5232 Villigcn PSI

The nucleation of coherent, spherical precipitates of the a-phase requires more than 90 at.% Co, and the equilibrium Co-rich tt-phase (c > 90 at.% Co), as imaged in Fig.l, volume fractions are between 0.3 and 0.6 % for the alloys from dilute solutions of Co in Cu is extensively treated and presented here. discussed in literature: Several investigations are reported Hence the SANS results give evidence for precursor using electrical resistivity measurements 11], transmission fluctuations small in amplitude but large in extent as com- electron microscopy (TEM) [2] and Field Ion Microscopy pared to the subsequently forming a-precipitales. They (FIM) [3]. In spite of controvcrsis in the discussion [4,5J cover a considerable volume fraction, i.e. more than 10 the studies come to the general conclusion that CuCo gives times the equilibrium volume fraction of a-precipitates. strong support for the validity of classical homogeneous Further, the precursor fluctuations are stable for finite re- nuclcation theory [6] in metallic systems. This conclusion action periods. After the transformation to a-precipitates implies that stable nuclei form by casually occuring fluctu- the particular characteristic is observed that all precipitates ations in Co-concentration which in amplitude exceed the are essentially equal in size. Their average radius is R = equilibrium concentration o( the a-phase (> 90 at.% Co) 3 nm, and the distribution width AR/R does not exceed over a region of at least the critical nuclei size. It fur- 5% (FWHM) (see Fig. 3). This is reflected in the second- ther implies that fluctuations smaller in size or smaller in order maximum in the structure function, which is clearly amplitude are unstable and decay. resolved at Q around 2 nm"1 (see Fig. 2). The unusually The above mentioned conclusions on nuclcalion were narrow size distribution requires a sharp and well defined predominantly drawn from the kinetics in number density of stability limit for a-precipitates at a radius of 3 nm, inde- the a-prccipitates. Stages before the completion of individ- pendent on the proceeding history of compositional fluctu- ual nuclei were not accessible in these studies. In order to ations. obtain both, information on nucleation kinetics and access In the presence of precursor fluctuations, the conditions to the pre-nuclcation stages. Small Angle Neutron Scatter- for classical nucleation theory are certainly disturbed. Con- ing (SANS) was applied, complemented by TEM and FIM sequently, a quantitative agreement between theory and ex- [7-12]. The present paper gives a brief review on the find- periment can hardly be expected, although it is strongly ings of these studies, which draw a quite different picture of propagated in literature. Indeed, comparing the theoret- early stage decomposition in CuCo than commonly slated ically expected number densities of precipitates with the in literature. experimental results, large discrepancies arise [11]: For tne Fig. 2 shows, as example, a sequence of scattering smallest supersaturation analysed (Ac = 0.1 at.%), precipi- curves. The solid lines represent calcuated structure func- tation rates are found enhanced by more the 40 (fourty) tions modelling a poly-disperse array of spherical a-pre- decades. Further, the size of initially forming a-Co preci- cipitates, with a size distribution parameterized by log- pitates is increased up to a factor of 10 compared to the normal function. A fit of these structure functions allow theoretical predictions of the critical nuclei size. A recent a quantitative determination of the relevant precipitate pa- study by analytical field ion microscopy (APFIM) corrobo- rameters "average size", "volume fraction" and "number rates the SANS-findings and yields microstructural details density", as given in [10 I 'J on the precursor aggregation of Co-atoms [12]. Several measuremer t sequences revealed that the scattering data of the short reaction periods (c.f. for 10 min. in Fig. 2) do not follow the modelled structure functions of References a-precipilatcs. The scattering curves are rather characterized by a continuous and approximately straight intensity [1] I.S. Servi and D. Tumbull, Acta Metall. 14, 161 (1966) increase towards small scattering vectors Q. Straight lines drawn through the data have slopes between -I and -2, i.e. [2] F.K. LeGoues and H.I. Aaronson, Acta Metall. 32, significantly smaller than Porods's Q~4 law for spheres. 1855 (1984) Fourier transformation of the scattering, as presented in Refs. [9,10], as well as the size distribution functions [3] H. Wcndt and P. Haascn, Scr. Metall. 19.1053 (1985). shown in Fig. 3, illustrate that the scattering contributions at [4] F.K. LeGoucs, T.L. Dcvitt, and H.I. Aaronson, Scr. the early stages are originated by compositional fluctuations Metall 20, 1305 (1986) of broad distribution in size, extending from 1 to about 15 nm, i.e. to dimensions more than five times as large as [5] H. Wendt, P. Haasen, T. Al-Kassab, L. v. Alvensleben, the average diameter of the later forming a-prccipitatcs. R. Griine, A. Hiittcn and M. Oehring, Scr. Metall. 20, An evaluation of the integral scattering cross section, as 1311 (1986) outlined in Ref. [10], allows an estimate of the average compositional amplitude of the fluctuations to 4 to 10 at.%, [6] R. Becker and W. Docring, Ann. Phys. 24, 719 (1935); and the volume fraction to about 10%. For comparison, the R. Becker Ann. Phys. 32, 128 (1938) 37

[7] W. Wagner, J. Piller, H.-P. Dcgischcr and H. Wollcn- bergcr, Z. Mctallkde. 76, 693 (1985) [8] W. Wagner, J. Phys. F 16, L 239 (1986) 19] W. Wagner and W. ^etry, Physica B 156&157, 65 (1989) [10] W. Wagner, Z. Metallkde. 80, 873 (1989) [11] W. Wagner, Acta mctall. malcr. 38, 2711 (1990) [12] X. Jiang, W. Wagner and H. Wollenbergcr, accepicd by Z. Metal Ikdc. (1990).

Log (Q/ntn 'I

Figure 2: SA/Vj results for Cu-0.8 at.% Co after solution treatment (WQ) and subsequent annealing at 833 K. The solid curves are calculated structure functions fitted to the data, which model the scattering due to the phase transformation in addition to the isotope incoherent scattering Sinc and a Q'""-dependent scattering of geometric sample inhomogeneities.

0 • jl Cu - 0.8 at% Co

06 - 833 K Figure 1: FIM-micrographs ofCu-2 at.% Co, after ann.al- I 05 ingfor 1 hat 833 K, imaged in a mode of atomic resolution .20 mm (a) and precipitate resolution (b). The bright areas in (b) are precipitates of a-Co. £ 03 2 10 min 1 V \ 60 mm 02 h \ • 01 I ^i i i m 4 6 R Inm) Figure 3: Normalized she distribution functions resulting from the fit of the model structure functions to the SANS data of Fig. 2, for a period of dominating precursor fluctuations (hatched area, curve labeled 10 min) and subsequently arising a-Co precipitates (20 and 80 min). 38

THE PSI - ETH TANDEM ACCELERATOR FACILITY

G. Bonani', B. Dittrich*, M. Dobeli', U. Fischer*, H.J. Hofmann", T. Niklaus", M. Sutcr', H. Riihl.', H.A. SynaT W. WoMi' and U. Zoppi*

* Institute for Intermediate Energy Physics, ETH-H5nggerberg, 8093 Zurich, Switzerland t Paul Schcrrcr Institute, c/o ETH-Honggcrberg, 8093 Zurich, Switzerland

Operation to suppress strong isobaric interferences such as :iJS in the case of •r-'Si measurements. The total suppression needed The 6 MV tandem accelerator facility located at the ETH in this example is 8 to 10 orders of magnitude to be a'ulc Honggcrberg, has mainly been used for Accelerator Mass to detect :!2Si on natural levels. Spcctromctry (AMS) and applications in Materials Sciences. The use of the facility is summarized in Table I and com- Tests with '-'"I have shown that routine measurements are pared with the corresponding data from the previous year. possible after a few minor adoptions of the system. For The total operating lime was about 13 % less than in the "1iCa measurements the sample preparation techniques arc year before. This can be explained by two facts: explored.

1. The transmission improvements made in 1989 and 1990 reduced the beam time for AMS measurements considerably. Applications of

2. The regular maintenance period last summer was longer Accelerator Mass Spectrometry than usual, due to the installation of a stronger pumping system in the high voltage terminal. The total number of samples measured has increased by about 10% compared to the previous year. About 10% of samples were standards for proper calibration and stability Technical Developments control. About the same number of samples were blanks to determine the background. In order to maintain a facility with excellent performance and to enable research at the frontier of AMS, we continued The most important isotope is 14C, for which the number of our systematic studies of understanding the background, the unknown samples is 49%. The fraction of 10Be is almost errors and the limits of efficiency. Based on these studies one third. The number of 26AI and 3<5CI samples arc in the the equipment is improved and new components arc added. order of 10 % each. During 1990 a new ion source and injector system with About 60% of the samples arc studied in the context of a high resolution magnet and electrostatic analyzer became long term collaborations with the University of Bcmc (Prof. operational. These components open the possibility to anal- H. Ocschger) and EAWAG (Dr. J. Beer) (22%), Uni- yse heavy ions up to Uranium (sec separate contribution in versity of Heidelberg (Prof. K.O. Miinnich and PD A. this report). It also gives more flexibility and reliability to Mangini) (15%), Columbia University in Lamont (Prof. the facility, because both injectors systems arc completely W.S. Broeckcr) (6%) and the University of Koln (Dr. U. independent, allowing to service one source during the op- Hcrpcrs) (17%). eration of the other. An overview of the field of applications is given in Table II. The systematic investigations of the charge changing pro- A significant portion of the research is related to ocean stud- cesses of fast moving ions have been continued. These ies. The main goal of these projects is to gel information studies clearly show the advantage of me tandem acceler- on the ocean, its circulation pattern and its inllucncc on cli- ator with voltage ranging up to 6 MV, compared to the matic changes. Through investigations of l0Bc in polar ice new generation of smaller uindcm accelerators which arc interesting correlations between solar activity (e.g. 11 year primarily designed for ion beam applications in materials cycle) and the radioisotopc production have been found. sciences. Stripping efficiencies for 10Be and HC are sig- Methane studies in the atmosphere give important informa- nificantly higher at energies in the 6 McV range. These tion on the origin of this greenhouse gas. MC studies in studies also show the importance of the good vacuum con- lake sediments arc also linked to climatic changes in the ditions in the accelerator tubes for reducing beam losses past. They arc of special interest, when annual structures and background due to charge changing processes. arc found in the sediment, this allows to derive absolute age differences similar to tree ring studies. There is the hope Extensive studies of the slopping process of the fast ions M in these studies to extend the C calibration curve further in the gas counter indicate that the isobar separation is pri- back . For the interpretation of radioisotopc concentrations marily limited by energy and angular straggling processes. in meteorites, their production mechanism and rates have to With an appropriate design of the detector and an adequate be known. For this reason production cross section studies data analysis the separation can be improved. have been performed during the last years, by irradiating Further developments of the technique of gas filled mag- targets at various high energy accelerators and subsequent nets arc in progress. The primary goal of this technique is AMS-mcasurcments of the radioisotopc contents. 39

Table I: Operation of the- Tandem Futility in l'JS'J ami l'J'JO Field Hours ci> Samples 1990 1989 1990 1989 1990 1989 AMS luBc 215 275 12 14 830 910 14C 663 759 37 37 1319 1228 -6A1 116 52 7 3 309 114 :)CC1 109 121 6 6 253 250 Tests 209 187 12 9 - Subtotal 1312 1394 74 69 2711 2502 Materials Science 195 298 10 14 Conditioning, development 266 351 16 17 Total 1773 2043 100 100

Table II: List of samples analysed by AMS in the calcmlar year l'J'JO

lu 1J Field Bc MC - AI ^Cl TOTAL Oceanography 222 225 447 (21%) Polar ice (and precipitations) 148 155 303 (14%) Limnology 120 120 ( 6%) Atmosphere studies 51 51 (3%) Others 69 252 38 28 387 (18%) Earth Sciences (Total) 439 648 38 183 1308 (62%) Meteorite and production cross-sections 186 180 2 368 (18%) Archeology 340 340 (16%) Others, Tests 44 45 1 90 ( 4%) 100 % Subtotal 669 1033 219 185 2106 (78%) Standards 74 126 26 51 277 (10%) Blanks 87 160 64 17 328 (12%) Total 830 1319 309 253 2711 (100%)

Materials sciences The analytical methods Rutherford backscaiicring spectromctry (RBS) and nuclear reaction analysis (NRA) have been A substantial amount of beam time was spent for ion beam applied to the investigation of oplo-clcctronic materials pro- analysis and modification of materials by irradiation. Thin duced by the Institut fiir Quantcnelcktronik, ETH-H6nggcrbcrg layer activation (TLA) projects in collaboration with Gcbr. and PSI Zurich. A five axes goniometer with a temperature Sulzer AG and Eidg. Pulverfabrik have been continued. controlled sample holder for ion beam channeling experi- An irradiation chamber for ion implantation of 4 inch Si ments is being tested now and will be operational in the wafers (in collaboration with Asea Brown Boveri Ltd.) has first half of 1991. been tested and a few irradiations have been done. Tests for the modification of the optical absorption of InP by ion implantation have been conducted with PSI Zurich. 40

EFFICIENCY IMPROVEMENTS WITH A NEW STRIPPER DESIGN

G. Bonani", P. Ebcrhardf, HJ. Hofmann" Th.R. Nik1aus\ M. Suterf, H.A. Synal' and \V. Wolfli' * Institute for Intermediate Physics, ETH-Honggcrbcrg, 8093 Zurich, Switzerland t Paul Scherrcr Institute, 5232 Villigen PSI, Switzerland

For precise and efficient measurements in accelerator mass a length of 680 mm with an inner diameter of 8.1 mm. A spectromctry (AMS) it is important to have a high trans- Leybold Turbo Vac 150 is connected to the stripper housing. mission and efficient electron stripping in the terminal of The lurbomolccular pump runs at about 70% of the nomi- the accelerator. Thus, for gas stripping additional pumping nal speed (150 t/s), i.e. it has a pumping speed of 110 t/s systems were installed in accelerators. Around 1982 tur- for N;,. Passive cooling is sufficient at 18 bar. The pump bomolccular pumps were first used for recirculation of the rccirculatcs the gas from both ends of the stripper channel stripper gas. At present, terminal pumps arc standard fea- via a volume of high conductance. The pump discharge is tures of newly designed tandems. Installing such a pumping fed back over the gas inlet line to the center of the stripper system in an EN-Tandem as used at our laboratory the fol- channel through a backing line. This line is big enough lowing problems have to be considered: all components not to introduce a significant pressure drop and is equipped have to withstand the high pressure of the insulating gas with a sorption trap (Micromazc7M/) to reduce the partial (18 bar), the space available to install gas and foil stripper pressure of hydrocarbons from the grease lubricated ball and pumping system is very narrow and the electric power bearings of the turbomolecular pump. If the pump is not available is limited. running, the backing line is closed by an electromagnetic Detailed measurements of the stripping process on differ- valve. The present design of the stripper allows a recircu- ent gases and Toils [1][2J13] performed in the last few years lation of about 90 % of the gas. The stripper gas pumped showed that a stripping efficiency of about 70% can be away over the tubes is replaced over the gas inlet line from reached with Ar stripper gas selecting charge state C1+ (in- a reservoir through a fine Ihcrmolcak valve. The pressure stead of C3+) at the appropriate terminal voltage. is measured in the stripper housing with a commercial cold The stripper thickness required to obtain the equilibrium cathode gauge (Penning), which was equipped with a ce- charge state distribution (C1+) is about 1.5 /;g/cm2. The ramic feedthrough for the cathode to make it suitable for resulting gas flow, pumped only through the accelerator operation in an ambient pressure of 18 bar. tubes, results in a high pressure in the housing surrounding In the new stripper arrangement the foil stripper is placed the stripper channel (about 10~4 Torr) and in the HE-tubc. directly behind the gas stripper (sec Fig. 1). The accep- Such a high pressure both degrades the performance of the tance is mainly defined by the HE-tubc and is about five accelerator tubes and causes significant beam losses due to times higher than before, where it was defined by the stripper channel. With this arrangement and thinner C-stripper multiple scattering and charge exchange. The new stripper 2 arrangement with a gas rccirculation pump allows a higher foils (3 fig/cm ) the transmission for all isotopes for which gas density in the stripper and improves the vacuum con- foil strippers arc used was significantly enhanced. ditions in the accelerator tubes. • H1"1]""!1"1!111'!11"}'"1!11"!""!"11!1111!11"!11"!11"!"11!"11!"11!"11!1"'!""! The new stripper was designed in such a way that when the pump is not running or fails, the pressure conditions Stripper Old Stripper through the accelerator can be maintained the same as with New Stripper the old stripper arrangement. A schematic view of the new A stripper design is shown in Fig. 1. The stripper channel has Loir Energy Tube Bifh Energy Tube ~

-4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 DISTANCE [meter] Figure 2 Pressure characteristics through the accelerator of the old and new stripper arrangements. In the new system, the stripper density could be enhanced by a factor Figure 1 of five. The corresponding stripper thickness is about Sclicmatic design of the new stripper configuration. 1.6 fig/cm2. 41

1 The major improvemenl over previous setups is the change :" " 1 ' i "'l""l'" '""1 1 of the pressure profile in the acceleration tubes and the stripper channel. These profiles along the accelerator are shown in Fig. 2 for the old and new stripper. They were :..£+.. -•^ +. ±J " 1 m - 14 :

derived based on the conductances, the estimated pumping " 1 1' speeds and the measured pressures in the terminal and at both ends of the acceleration tubes. As shown in the figure, -j the stripper density could be enhanced by a factor of five, 1 • ....+ corresponding to a stripper thickness of about 1.6 ug/cm-. o K' m -= 13 : At the same time, the vacuum in the lubes could be im- « (m) proved by at least a factor of two. This results in a lower I' F(m) 4 charge exchanging rate in the tubes, yielding a higher trans- '•- -. mission of the chosen charge state (C+). 1, For various gas pressures, a standard sample was measured '•.... i.. ,..i..,.l... 0.4 0.6 1.2 l.O 2.0 2.< 20 using our regular data aquisition and the fast beam switch- _. . STRIPPER THICKNESS [jug/cm2] Ar ing system. With the position feed back system the termi- figure 4 nal voltage was automatically adjusted to compensate the Model calculation and measured absolute fractional ion pressure dependent energy losses in the stripper. No other for mass 13 and 14 as a function of the stripper thick- settings were adjusted during the measurements. From the recorded t2C currents at the low and high energy side the ness.

transmission was determined. Based on the measured cur- H 12 rents ('-C,13C), the 14C counting rate and the known iso- 4 % for C/ C was applied to fit the calculated curves to topic ratios the mass fractionation was calculated. the experimental data. The remaining mass fractionation is probably due to ion optical effects. Above 1.4 fty/cur the fractionation is well described by the model. The deviation ' ' ' J at low pressures can be explained as mentioned above. Other ions than carbon arc stripped with foils, resulting 50 — < —* in higher charge states. For Be a combination of gas and foil stripping is used. The BcO~ molecules arc dissociated _ 40 -i at low gas pressure and post-stripped in the carbon foil, leading to the highest possible efficiency. With the new arrangement and thinner stripper foils the transmission could 30 model calculation -: be improved by about a factor two. l2 + 7 H —H C - data -; With the new stripper arrangement we obtained a transmis-

20 sion of 30% for beryllium, 50% for carbon, 18% for alu- V,, , = 5.2 MV ~: Terminal : minium, and 14% for chlorine. More details are given elsewhere [4J. A further improvement of the vacuum conditions : to ...,i,..,l,,,,i,... I....I....I l...,i...,l,,,, and thus of the transmission could recently be achieved by installing a turbon. 'ccular pump with higher speed (Ley- STRIPPER THICKNESS Og/cm2] Ar bold TurboVac 360, running at 80% of the nominal speed). figure 3 Detailed investigation of the upgraded stripping system is Model calculation and experimental data of the trans- currently under way. mission for a 12C'1+ beam through the accelerator as a function of stripper thickness.

Figure 3 shows the measured transmission of the 12C beam through the accelerator vs. the stripper thickness in /.tg/cm2 argon. The transmission increases until the losses due to charge exchange processes in the LE acceleration tube and angular straggling become dominant. The measured maximum transmission for charge state 4+ is about 50% in the thickness range 1.4 to 1.6 fig/cm2 (compared to a transmis- References sion of about 27% for charge state 3+ with the old stripper). [1] Ch. Stollcr, M. Sutcr, R. Himmcl, G. Bonani, M. Nessi For radiocarbon dating it is highly desirable to work in a and W. Wolfli, IEEE Trans. Nucl. Sci. NS-30 (1983) region, where fractionation is independent of stripper thick- 1074 ness. Thus, we measured the fractionation, as defined in [2] H.J. Hofmann, G. Bonani, E. Morcnzoni, M. Ncssi, Fig. 4, of 13C and 14C as function of the stripper thickness M. Sutcr and W. Wblfli, Nucl. Instr. and Mclh. US using a standard with well known isotopic ratios. (1984) 254 The mass fractionation is shown in Fig. 4. Experimental [3] H.J. Hofmann, G. Bonani, M. Sutcr and W. Wolfli, data points were derived from the measured isotopic ra- Nucl. Instr. and Mclh. B29 (1987) 100 tios and the nominal values of the standard. The model [4] G. Bonani, P. Ebcrhardt, H.J. Hofmann, Th.R. curve was normalized to match the data at higher pres- Niklaus, M. Sutcr, H.A. Synal and W. Wolfli, Nucl. sures. The model indicates a smaller mass fractionation. Instr. and Mclh. B52 (1990) 338 For the isotopic ratios a correction of 2 % for 13C/12C and 42

THE HEAVY ION INJECTOR AT THE ZURICH AMS FACILITY

H.-A. Synal", G. Bonani*, R.C. Finkcl', Th. R.Niklaus', M. Sutert and W. Wo'fli'

• Institute of Intermediate Energy Physics, ETH-HOnggcrbcrg, 8093 Zurich, Switzerland t Paul Scherrcr Institute, 5232 Villigen PSI, Switzerland

At the ETH/PSI-AMS facility a new high resolution injector has been designed and set up. The system consists of a spherical electrostatic deflector which acts as an energy analyzer and a stigmatic magnetic spectrometer to perform the mass separation. Sufficiently high mass resolution of an injecting system for an accelerator mass spectrometer cannot be achieved simply by selling narrow slits in ihc object and image plane of the injection magnet. Since the ions to be analyzed must pass several filtering steps, nearly 100% transmission for the rare isotope is needed at each step. The minimum slit sizes are then determined by the emission of the primary ion beam. Another important point is that requirements for a reliable normalization of the rare to stable isotope ratio have to be met by the system. It has been shown with the old injector that this is possible with a fast beam switching system [1], if the injection magnet is designed to have a symmetrical point-to-point mapping. A double focusing device (achromatic system) as used in standard mass spectrometry would not meet these requirements and has therefore not been considered. In our system the stigmatic magnet is placed after the stigmatic electrostatic deflector. The image of Uic electrostatic deflector is used as the object of the magnetic spectrometer (Fig. 1). The ion source is of the same type as the one in the old system. The new source has, in addition, the possibility to work with independent extraction and sputter energies. The electrostatic dc/lector design is based on two spherical seg- ments in a concentric mounting. The effective field length Figure I is determined by two diaphragms, which perform a correc- Details of the experimental setup at. the low energy side tion in a first order approximation [2]. In the object plane of of the ETH/PSI-AMS facility. the electrostatic deflector an analyzer chamber is installed. Here, apertures can be set to limit the energy spread of the ion beam. A retractable Faraday cup to measure the ex- quadrupole lens and an electrostatic Einzel lens. With re- tracted ion current and electrostatic steercrs are placed in spect to the total space available for the setup of the sys- this chamber. The magnetic spectrometer is a stigmatic 90" tem, a maximum in mass separation can be achieved for dipolc magnet. A second analyzer is installed in its object a magnet radius of 56 cm and a radius of the electrostatic plane. With a variable aperture, mounted on the beam axis, deflector of 75 cm. For a 3 mm beam spot an energy res- the momentum acceptance of the spectrometer can be lim- olution E/AE = 500 and a mass resolution m/Am = 370 ited. Three adjustable Faraday cups arc placed inside this is achieved. This means that beams of ions having a mass vacuum chamber. of more than 200 amu can be separated. The experimental setup has now been completed and the With these cups the stable isotope currents can be mea- new injecting system has been operated under routine con- sured, while the rare isotopes arc injected into the tandem. ditions for 36C1 measurements. :!6C! measurements of sam- An additional retractable Faraday cup for the measurement ples with natural isotopic compositions has been possible of the injected ion current and elcrtrostatic steercrs for beam with the old injecting system, despite its poor stable iso- alignment are also installed in the chamber. The accelera- tope reduction rate. To reach a sufficient suppression of the tion gaps of the beam switching system are placed near the stable isotope an additional mass selective filtering step, a focal points in the object and image planes of the magnet. time of flight spectrometer, has been used. First measure- The vacuum chamber of the magnet and the beam pipes ments with the new system have shown that the isotope between the two analyzer chambers are insulated. Beam suppression at the low energy side of the accelerator has transport into the tandem is performed with an electrostatic been improved by at least 3 orders of magnitude. There- 43

fore, time of flight measurements are no longer needed. The new heavy ion injector at the ETH/PS1-AMS facility is This means that the efficiency and stability of 3SC1 mea- operational now. It will be used mainly for measurements surements can be significantly improved. of heavier cosmogenic radioisotopes, while the old injector will be used further on for the detection of inBc and MC. Besides interferences from stable isotopes, isobaric ions Since the new system will also allow to measure these two also cause background which limits the sensitivity. For isotopes, it will improve the availability and simplify the 36C1 measurements the reduction of 36S is achieved by maintenance of the whole facility. In case of XC\ measure- measuring the different energy loss characteristics of both ions in an absorber medium. The extent to which 36S can ments strong improvements in isotope separation have been made. Further investigations will be focused on the devel- be reduced is therefore strongly correlated to the energy l2 J available. At the ETH/PSI EN-Tandem accelerator the ter- opment of ' I measurements under routine conditions as well as on the development of the technique for "Ca and minal voltage is limited to 6 MV and the maximum energy 230 is 48 McV for ions in the 7+ charge state. At this en- Th isotope measurements. ergy an isobaric reduction factor of 5 x 10~5 is possible [3]. Background measurements with the new system have been performed without using our time of flight spectrometer. References Blank levels between 2xlO~15 and 5xlO~15 have been [1] M. Sutcr, J. Beer, D. Billctcr, G. Bonani, H.J. Hof- obtained, depending on the 3GS contamination of the sam- mann, H.-A. Synal and W. Wol/li, Nucl. Instr. and ples (less than 1 ppm). Recently, measurements of 36C1 in Mcth. B40/41 (1989) 734-740 the 8+ charge stale with a beam energy of 54 MeV have been performed. A sulfur reduction factor of lxlO~5 has 1G [2] H. Wollnik, Focusing of Charged Particles cd. A. Sep- been measured and a background ratio of 7xlO~ has tier, Academic Press, Chapter 4.1 (1967) 163-202 been obtained. This shows that 36C1 measurements at the 10~lr' level arc possible with accelerators that are limited [3J H.-A. Synal, ETH Dissertation no. 8987 (1989) to a maximum terminal voltage of 6 MV, when isotopic background is eliminated by an appropriate mass spectrometer design.

First 1231 test runs have been made with the new system. Typically a 10 //A 127I current could be extracted from the ion source. We used the 8+ charge state at a terminal voltage of 5 MV. Standards with 129I/I isotopic ratios between 5xlO~10 and 5xlO~12 have been analyzed. I29I could clearly be identified and the isotopic ratios measured agree well, within the uncertainties, with the nominal isotope ratios. For the development of 4iCa measurements at ETH, current tests of a CaH2 sample have been made with the new injector. 40 We obtained about 300 nA of CaH3 current. With such currents and a transmission similar to the one obtained in 36C1 measurements, it should be possible to detect 41Ca at the 10"13 level.

1 L"'l "I "1" ''"" "T"'1""! mini 1.0E-6 4°CaH, D 3 !l CaH 40 1.0E-7 r-1 'r n. CaH2 ." • _ '• i 44 "CaH _; l.OE-B n • CaH 3 1.0E-9 ,- • • • co2 f : ": .' 1.OE-1O V /"I • * , t.OE-11 T 1: V. ; •• " t V l.OE-12 "// J;

In ,i I,,,I, Ml,,,, 43 43 11 45 40 47 Mass [AMU]

Figure 2 Negative ion current extracted from a 40CaJi2 sample in the mass region between 40 and 48 amu. 44

AMS MEASUREMENTS OF 10Be IN POLAR ICE TO TRACE THE 11 YEAR CYCLE OF SOLAR ACTIVITY

J. Beer', G. Bonani', R.C. Finked, H.J. Hofmannt, B. Lehmann1, H. Ocschgcr', B. Stauffcr1, M. Suter11 and W. Wollli' * Environmental Physics, Institute for Aquatic Sciences and Water Pollution Control, ETH Zurich, c/o EAWAG, 8600 Diibendorf, Switzerland t Institute for Intermediate Energy Physics, ETH-H6nggerberg, 8093 Ziirich, Switzerland I Physics Institute, University of Bern, 3012 Bern, Switzerland § On leave from Lawrence Livcrmore National Laboratory, Livermorc, CA, 94550, USA 1 Paul Sclierrer Institute, 5232 Villigen, Switzerland

The solar activity, expressed as sun spot numbers or auroral 2.0 events, has been recorded quite consistently since 1700. Al- though great effort has been expended to extend the sunspot oi 1-5 © and auroral record further back, the data before 1600 be- — 1.0 come more and more incomplete and uncertain. One way m to overcome this problem is to rely on indirect data such 2 0.5 as records of cosmogenic radioisotopes which are produced 0.0 by the interaction of cosmic ray particles with the atmo- 1.6- sphere. From neutron monitor measurements over the last 1.4- few decades it is known that the cosmic ray (lux reaching CT 1-2 the atmosphere is modulated by solar activity. Magnetic "2 1.0- fields frozen in the solar wind dellcct the cosmic ray panicle £ 0.8- o 11 ux with greatest effect at low energies, and lead to a re- " 0.6- 200 duction of the production rate of cosmogenic radionuclidcs. 0.4 0 Although the details of the modulation process arc not yet 1.4 understood, there is a clear anticorrclation between solar 1.2 10 •10 activity and cosmic ray flux reaching the Earth. Be is S" 1.0 •20 produced by nuclear spallation reactions with the principal 5 0.8 a™ °'6 components of the atmosphere, becomes fixed to aerosols •30 and is removed finally from the atmosphere mainly by pre- 0.4 cipitation. The overall atmospheric residence time is about 0.2 40 one year. In spite of the short-term fluctuations caused by 1.4 0 10 the removal process, Be turns out to be the most promis- ~ 1.2- •20 'a! ing cosmogenic radioisotopc with which to detect the 11 y •40 solar cycle. 'o 1.0- /V\ 60 S 0.8 80 10 \/ t Even though first detected in 1956 cosmogenic Be only 100 0.6- SUN SPOTS became a potential tool to reconstruct solar activity with 0.4 120 the development of accelerator mass spectromctry (AMS) 1780 1800 1820 1840 I860 1880 1900 1920 1940 1960 1980 2000 through which the detection sensitivity was increased by YEAR [ADJ several orders of magnitude over decay counting [1]. Figure 1 Comparison of the 1GZ?e ice core record with sunspot- We have analysed the upper part of a 300 m ice core from and Aa-index. 10 Greenland [2]. The seasonal variations of H2O2 were used a) Measured Be concentrations (step (unction). To to determine the individual annual layers and to establish remove short-term climatic noise the data have the timcscale. To confirm the time scale stratigraphic hori- been smoothed with a, 3 point binomial Filter. zons were also used: the bomb peak of tritium in 1963 b) Comparison of the smoothed i0Be record with and the acidity peaks caused by the volcanic eruptions of the annual sunspot numbers for the period 1783 - Tambora (1815) and Laki (1783). First the core was cut into 1985. annual samples with typical weights of 2 kg. The samples iD were then melted, 0.5 mg of sBe carrier was added and the c) Comparison of the smoothed Be record with the annual Aa-index for the period 1868 - 1985. volume was reduced by evaporation. Finally the Bc(OH)2 was precipitated and converted to BcO. The measurements d) Comparison of the long-term trend between wDe were carried out at the AMS facility of ETH/PSI [3]. and sunspot numbers. Doth curves have been smoothed with a 19 point binomial filter. 45

The results are shown in Fig. 1 for the period 1783 - 0.2 1985. To remove the rather large short-term fluctuations most likely caused by deposition processes, a 3 point binomial filter was applied to the measured 10Be data (Fig. la). In Figure lb the filtered data are compared with the sunspot o numbers. There is a clear aniicorrclation between wBe and < LJ. sunspots, both with respect to the 11-year solar cycle and z the long term trend. The main discrepancy is related to the o fact that, while the sunspot number minima are always close

to zero and therefore do not show much variation, the cor- UJ responding 10Be maxima vary considerably. This indicates a: cc that sunspots and 10Be are not identical indicators of over- o -0.4 i all heliomagnetic activity. The relationship breaks down o especially at solar minimum, where very low sunspot numbers may occur in conjunction with considerable residual heliomagnetic activity. This can be seen by examining the -30 -20 -10 Aa-indcx, another indirect solar parameter, which is based on solar wind induced short term (up to 3 hours) variations LAG (Y) Figure 2 of the geomagnetic activity rfleasured at two antipodal sta- I0 tions. The Aa-index has been recorded only since 1868. It Cross-correlation of the measured Be data (Fig. Jaj agrees well with the I0Be record not only with respect to with the sunspot numbers (Fig- lb). The highest an- the 11 y solar cycle, but also reflects the decreasing trend of ticorrelation is reached at a time lag of about 1 year 10 corresponding to the mean atmospheric residence time the Be maxima and minima since about 1900 (Fig. lc). 10 In Fig. 2 the cross-correlation between the measured 10Be of Be. The cross correlation clearly shows (.he exis- data and the sunspot record is displayed as a function of lag tence of the 11 y solar cycle. time. The highest anticorreU>tion is observed for a lag of about 1 year, which corresponds nicely to the mean atmospheric residence time of 10Be. The cross-correlation with References a significance of > 99.99% clearly proves the existence of [1] W. Wolfli, Nucl. Instr. and Meth. B29, 1-13 (1987) the 11 -year solar cycle: the correlation factor changes pe- riodically, reaching minima after time lags of 11, 22, 33... [2] J. Beer, et al., Nature 347, 164-166 (1990) years. The cross-correlation curve between 10Bc and the Aa-indcx is similar to the one between 10Bc and sunspots [3] M. Sutcr, ct al., Nucl. Instr. and Meth. B40/41, 734- (Fig. 2) and is therefore not shown. 740 (1989)

For easier comparison of the long-term trend both records have been filtered with a 19 point binomial filter and are displayed in Figure Id. Except for the last 20 years, the two curves show the same pattern of variation. The data of Fig. 1 and 2 show a clear anticorrelation be- vies™ ' ° Be and sctai aclWfry "Wilh vtve. wpecved phase, teg of about 1 year. The remaining discrepancy can be explained by fluctuations during the removal of lcBe from the atmosphere and by the fact that solar modulation of cosmic rays, sunspot numbers and Aa-indices are not identical rep- resentations of solar activity. The main uncertainty in the interpretation of 10Be data in terms of solar activity modulation is therefore the extent to which fluctuations induced by 'climatic noise' (atmospheric mixing, precipitation rale, scavenging efficiency) effect the polar ice cores which we have studied. AMS measurements of 10Be in polar ice open up the possibility of studying the long term behavior of solar activity and the history of solar-terrestrial relationships. In particular several interesting questions can be addressed such as: 1. Is the 11 y solar cycle driven by processes in the convection zone or by a precise chronometer deep in the sun? 2. How does the heliomagnctic activity behave during periods of solar sunspot minima (Maunder, Sporer etc.) with respect to periodicity, amplit, !c and phase?

3. How did the solar activity change in the past and how docs the climate system respond to such changes? 46

HIGH RESOLUTION 10Be-STRATIGRAPHY OF ARCTIC SEDIMENT CORES

A. Eisenhauer', G. Bonani', H.J. Hofmann', A. Mangini", R. Spiclhagcn', M. Sutcr5, J. Thiede* and W. W6IfiiJ

* Heidclberger Akademie der Wissenschaften, Im Neuenheimer Feld 366, 6900 Heidelberg, Federal Republic of Germany t GEOMAR, Wischhofstrassc 1 - 3, 2300 Kiel 14, Federal Republic of Germany t Institut fur Mittelcnergiephysik der ETH Zurich, ETH-H6nggcrbcrg, 8093 Zurich, Switzerland § Paul Scherrcr Institute, 5232 Villigen PS1, Switzerland

Introduction In this paper we present lcBe profiles of three cores from Northern high latitudes on a transect Recently, important progress in dating arctic sediments was made by applying the AMS ^C-dating method [1] which requires Only a few milligrams of biogenic carbonate for from 70°N Norwegian Sea 70°18,3O'N,O0.4°01,3O'E.2281m the measurement of one sample [2]. However, this method via78°N Fram Strait 78°51,55'N,00.1°18,59'E,2456m lo 86°N Arciic Ocean 86°21,80-N,26.0°12,90>E,3646m is restricted to the uppermost centimeters of the sediment column, representing the last 35 kyrs and including the Core locations arc shown in Fig. 1. Holocenc and Termination 1. Older important paleoclimatic events, e.g. the transition from glacial oxygen isotope stage 6 to intcrglacial stage 5 (Termination II, at 127 kyrs B.P.), cannot be dated by this method. Many sections in arctic Late Quaternary sediments contain small to negligible amounts of biogenic carbonate, rending 1SO-stratigraphy difficult. Also, the oxygen isotope records derived from planktonic foraminifcra are difficult to interpret because this signal is strongly influenced by the input of isotopically light meltwatcr [3]. Interpretation of these records is possible only if other methods offer a reliable stratigraphic framework. Most successful have been paleomag- nctic stratigraphy [4] and the •""Thexcess-mcthod, which can be applied for the dating of Late Quaternary sedimenb on a lime scale up to 350 kyrs B.P., corresponding to the last 9 glacial/intcrglacial stages. Ku and Broecker [5], and in a more recent work Somayajulu ct al. [6], evaluated mean sedimentation rates of arctic sediments by the exponential 230 fits of the decrease of the Thexcess-concentration with depth to be in the order of some mm/kyr. The application of this technique was complicated by the observation of a 23C "non-uniform" behaviour of the Thex(:ess-concentrations [5], [7]. In a detailed 2-Tracer-Study [8] we measured concentra- 230 10 tions of Thexcess and of Be in polar sediment core 23235-2 (Fram Strait) in high resolution. We found a strong non-uniform behaviour of 10Be and 230Th isotope:, in the sediment, and we were able to show that these nonuniformities can be correlated to the climate of the past. Sections of the core corresponding to interglacial periods reveal high 10Be and 23DTh concentrations whereas the sections of the core reflecting glacial periods have low isotope concentrations. Figure 1 Map showing the core locations. 47

200 400 Depth (cm) Figure 2 In Fig. 2A,B,C the concentrations of M0Th, l0Bc and '^O-vatucs in core 23059 are plotted against depth. 200 400 Eight climatic stages can be identified in Fig. 2C apply- Depth (cm) ing the lsO-stratigraphy. The interglacial stages (}, 3, 5, 7) arc characterized by higher and the glacial stages (2, A, 6, 8) by lower 1SO- values. The transition from one climatic stage to another are marked by grid lines in the figure. The core sections corresponding to inter- Figure 3 i0 glacial stages show high Bc-concentrations (Fig. 2B). In Fig. 3A,D,C the 10Bc-conccntralions in 3 cores from The major climatic transitions (oxygen isotope stages a transect at 70°N, 78°N and 86°N are plotted against 1/2, 4/5, 5/6, 6/7, 7/8) ere marked by a drastic change depth. The 10Bc,-conccntrations of all cores vary be- l0 of the Bc-transition from glacial stage 6 to interglacial tween values lower than 0.05-109 atoms/g up to approx. 10 stage 5 (Termination U) at 210 cm depth the Be- 2-109 atoms/g. The grid lines connected the core sec- 9 concentrations increases from approx. 0.3 10 atosns/g tions ofKAL 364 and 23235 with drastic changes of the 230 10 to 1.1 10" atoms/g. At this transition also the Thex Be-concentrations with comparable sections of core concentrations shows a marked increase from approx. 23095. That such a link of the l°Bc-concentrntions of 0.5 dpm/g to 3.5 dpm/g. In the core sections (>) no re- the 3 cores yields a correct stratigraphy was confirmed 730 l6 lationship of the Th-conccntration to the O-curve.is by other independent methods. Therefore, maxima and found due to radioactive decay, whereas climatic stage minima of the l0Bc-concentration represent time mar- ]0 7 (270 - 410 cm) has a Bc.-conccntration as high as kers which can be used as a dating tool just like the interglacial stage 5. laO-stratigraphy. 48

Results References

The cores (23235 and KAL364) from northern localities [1] W. Wollli: Advances in Accelerator Mass Specirom- do not contain sufficient carbonate for 18O analyses on the etry, Nucl.Inslr. and Mcth. B29 (1987) 29 entire core profile. However, a detailed iaO-record has been established far core 23059 and may be used to cafi- m B. Kramer, C. Pflcidercr, P. ScMosser, I. Levin, K. M brate the I0Be-profiIe in this core (Fig. 2., [9]). The com- Munnich, G. Bonani, M. Suter, W. Wolfli: AMS C- parison shows that interglacial periods in core 23059 are Measurcmcnt of Small Volume Oceanic Water Sam- characterized by high and the glacial periods by low 10Bc- ples: Experimental Procedure and Comparison with conccntrations. The palcoclimatic transitions in the core Low-Level Counting Technique, Nucl.lnstr. and Mcth. arc marked by drastic changes of the 10Bc-conccntration. B29 (1987) 302 l0 The similar pattern of the Be-distribution observed in all [3] S.E.I. Kohler, R. Spiclhagen: The Engima of Oxygen 3 cores, namely a sequence of maxima and minima, can be Isotope 5 in the Central Fram Strait, U. Blcil and J. used for correlating the profiles in cores 23235 and KAL364 Thicdc (eds). Geological History of the Polar Oceans: to the palaeoclimatic transitions detected in the record from Arctic versus Antarctic, Kluwcr Academic Publishers core 23059. (1990)489 As shown in Fig. 3, most of the last 8 climatic transitions [4] N.R. Nowaczyk, M. Baumann: Combined High Res- during the Late Quaternary, covering a lime span of approx. olution Magnctosiratigraphy and Nannofossil; Bios- 250 kyrs can be identified in these cores. Palcomagnclic iraligraphy for Late Quaternary Arctic Ocean Sedi- stratigraphy [4], biological stratigraphies [10],[11] as well ments. Deep Sea Res., in press 23O as Th<,sccss-dating confirm this correlation. [5] Ku and Brocckcr: Rates of Salimentation in the Arc- tic Ocean, in M. Scars (cd.), Progress in Oceanogra- Conclusions and Further Work phy, vof. 4, Pergamon Press, London (1967) 95 Whether the drastic changes of the 10Bc-conccntrations are [6] B.L.K. Somayajulu, P. Sharma, Y. Herman: Thorium caused by variation of the amount of dilution with dctric and Uranium Isotopes in Arctic Sediments, in Y. Her- material, or are due to variable 10Be-fluxes into the sedi- man (cd.), The Arctic Seas, Rcinhold Company, New ments is subject of further studies. Our preliminary results YOrk H989)571 suggest that the profiles of the 10Be-concentration make [7] R. Finkcl, S. Krishnaswami, D.L. Clark: 10Be in Arc- possible a so called "10Be-stratigraphy". which might be tic Ocean Sediments. Earth and Planetary Science Let- used for dating of selected cores from the Arctic ocean with ters 35 (1977) 204 18 a precision similar to that of O-stratigraphy. [8] A. Eiscnhaucr, A. Mangini, A. Botz, P. Walcr, J. Beer, G. Bonani, M. Sutcr, H.j. Hofmann, W. Wolfli: 10Bc and '-'30Th Stratigraphy of Late Quaternary Sedimcnls from the Fram Strait (Core 23235), in U. Blcil and j. Thiedc (Eds); Geological History uf the Polar Oceans; Arctic versus Antarctic, Kluwcr \cadcmic Publishers (1990) 475 [9] J.C. Scholtcn, R. Botz, A. Mangini, H. Paetsch, 2:in P. Staffers, E. Vogelsang: High Resolution Thcx Stratigraphy of Sediment from High Latitude Areas (Norwegian Sea, Fr?rr, Strait), Earth Planet. Sci.Lcti. (in press) [10] G. Gard: Lato Quaternary Calcarous Nannofossil Bio- zonation, Cnronology ana Palco-Occanography in Ar- eas North of the Facroe-lccland Ridge: Quat.Sci.Rcv. 7 (1988) 65 [11] M. rfaumann: Coccolilhs in Sediments of the Eastern Arctic Basin, in U. Blcil and J. Thicclc (eds), Ge- ological History of the Polar Oceans, Arctic versus Antarctic, Kluwer, Academic Publishers (1990) 437 49

CHARACTERIZATION OF OPTOELECTRONIC MATERIALS WITH NUCLEAR METHODS

M. Dobeli", U. Fischer*, P. Giinter', R. Gutmann', J. Hulliger', H.W. Lehmann', S. Schwyn", M. Suter", U. Zoppi! and W. Wolfli1

* Paul Scherrer Institut, 5232 Villigen PSI, Switzerland t Institut fiir Quanteneleklronik, ETH-Honggerberg, 8093 Zurich, Switzerland t Institut fiir Mittelenergiephysik, ETH-H6nggerbcrg, CH-8093 Zurich, Switzerland

Rutherford Backscattcring Spectrometry (RBS), Particle In- a spinel substrate. The sample has been prepared at PSi duced X-ray Emission (PIXE) and Nuclear Reaction Anal- Zurich [3]. The comparison with simulated spectra yields ysis (NRA) have been used at the PSI/ETH Tandem labora- the thickness and elemental composition of the layer (simu- tory to characterize thin film and bulk samples of the non- lation performed according to rcf. [4]). In this case a severe linear optical materials KTa!_rNbrO3 (KTN) and KNbO3. deficiency of potassium in the approximately 170 nm thick In the following we demonstrate the potential of these nu- layer is revealed (the potassium to niobium atomic rau'o clear analysis tools with a few examples. turns out to be 0.42 ± 0.03). Nonuniformities in the film composition and/or diffusion into the substrate can be detected by the same method. Fig. 2 shows the RBS spectrum Rutherford Backscattering Spectrometry RBS is a fully quantitative and nondestructive technique for the analysis of the elemental composition profile of sur- Energy (MeV) faces and thin films. It relies on the fact that the energy of 1.8 1.9 2.0 2.1 2. 2 2.3 120 i I 1 1 I I

^.NNb Energy (MeV) 100 — - — s \ 1.0 1.5 To ~~\ Au 80 - Yiel d

60 _ ize d

40 - Norma l

20 1 -

• _ n

Figure 2: Solid line: 2.4 MeV Alle RBS spectrum of a KTN layer on KTaO-j. Dashed Line: simulated spectrum for an approx. 60 nm thick uniform layer of KTao-xNba 2i,O3. Dashed-dotted line: simulation of the same layer with the 4 Figure 1: 2 MeV He RBS spectrum of a thin KNbO3film on Nb concentration falling off linearly to zero across the first a spinel substrate. Solid line: simulated spectrum for a uni- 180 nm of the substrate). The edge marked with "Ta" is form layer of KoA2^b03. Dashed line: simulated spectrum caused by the increase of the Ta concentration at the sub- for standard sloicniometry of KNbO3. strate surface. Normalization of the simulations is arbitrary. an MeV ion that has been elastically backscattered from a (2.4 MeV 4Hebcam) of a thin layer ofKTa^NbjA) with target material is a function of the target atom mass and of a nominal x of 0.29 deposited by liquid phase epitaxy onto the depth at which the scattering took place [1]. Routinely a KTaO3 substrate. The sample has been prepared at the 2 MeV 4He beams are used. The backscattered particles Instilut fur Quantcnclektronik, ETH-Honggcrberg [5]. The are detected with silicon surface barrier detectors. In some remainders of a thin gold layer that has been evaporated cases different ion beams (protons, heavy ions) and energies onto the sample earlier and wiped of again are clearly vis- are used to enhance the target depth and mass resolution and ible. The shape of the Nb peak and of the Ta edge suggest to optimize the accessible depth [1,2]. The depth resolu- that the layer substrate-interface is fuzzed and/or the thick- tion at the sample surface is a few pg/cm2. Fig. 1 shows ness of the layer is nonuniform. In this case the method a RBS spectrum of a sputter deposited layer of KNbC>3 on cannot discriminate between the two possibilities. 50

Particle Induced X-Ray Emission (PIXE) Nuclear Reaction Analysis (NRA) In NRA the beam particles make a resonant nuclear re- In this method inner shell vacancies arc produced in the action with the target atoms and Hie reaction products (-,, target atoms by bombardment with MeV protons [6]. The n, p, a) arc detected. The number of detected events is emitted characteristic X-rays arc measured with a Si(Li) proportional to the target atom concentration. The ^nethod detector. The major advantage compared to an electron can be calibrated absolutely with samples of known com- microprobc is the absence of the electron bremsstrahlung position. Since the reaction is resonant, a depth profile background which greatly enhances the sensitivity (up to can be measured by changii.g the incident beam energy. three orders of magnitude) to trace elements. The sensi- The method is mostly applied to the detection of light el- tivity for most elements heavier than Al is of the order ements in a heavy matrix where RBS fails. We have used of 10"c. The method is only scmiquantiiative (X-ray ab- the 1GO(

References [1] W.K. Chu, J.W. Mayer, M.A. Nicolet in: Backsc.ai- tering Spectrometry, Academic Press, 1978. |2] M. Dobcli, P.C. Haubcrt, R.P. Livi, S.J. Spicklemirc, D.L. Weathers, and T.A. Tombrello, Nucl. Instr. and 500 Metl). IJ47 (1990) 148. 0 13] Annual Report, PSI Zurich c/o Laboratories RCA Ltd., Zurich, Switzerland (1989). Zr K, 4000 [4] L.R. Dooliltlc, Nucl. Instr. and Mem. B15 (1986) 227.

3000 [5] R. Gutmann and J. Hulliger, Crystal Properties and Preparation 32, 117 (1991). [6] e.g. L.C. Fcldman, J.W. Mayer in Fundamentals of Surface and Thin Film Analysis, North Holland 1986.

16 17 18 19 20 Energy [keV]

Figure 3: Top: RBS spectrum of a KTa^x^yNbxZry03 sample. The Nb and Zr edges cannot be resolved. Bottom: PIXE spectrum of the same sample. Ka and Up lines of Zr and Nb are clearly separated and background is nearly absent.

Fig. 3 PIXE has been combined with RBS for the analysis of a KTa,_r_5,NbxZrj,O3 sample. Nb and Zr cannot be resolved with Hc-RBS (their atomic mass differs only by about 1 amu) and hence only the sum of the Zr and Nb concentrations can be determined to x + y — 0.41 ± 0.01. The PIXE analysis, however, yields a Zr to Nb ratio of 0.99 •*• 0.07. In addition the analysis beam spot can be fo- cussed down to a diameter of approx. 20 ^m and the beam can be rastercd across the sample digitally. Tiis allows to image the -'.emental concentration at the sample surface or to scan the composition depth profile along a cross section of the sample. 51

EUVITA AN EXTREME UV IMAGING TELESCOPE ARRAY ON THE SPECTRUM-X-G-SATELLITE

The EUVITA Collaboration:

A. Zehnder* , V. ArefeT', J. Bialkowski', P. Biihler*, P. Charkobarti5, L. Chesalin', T. Courvoisicrt, V. Drcmin', R. Henneck', K. Lund", A. Orr', N. Schlumpf*. W. Schoeps*, O.H.W. Siegmund§, R. Sunayev*, K. Thomson', P.W. Vedder5, J. VellergaS, G. Walker'

* Paul Scherrer Insiitule, 5232 Villigen PSI, Switzerland t Observatory of Geneva University, 1290 Sauverny, Switzerland I USSR Academy of Sciences, Space Research Institute (IKI), 117 296 Moscow, USSR § University of California, EAG, Berkeley, CA 94720-0002, USA f University of British Columbia, Vancouver BC V6T 1W5, Canada ** National Research Council, Space Division, Ottawa, K1A 0R6, Canada

ABSTRACT The instrument sensitivity for continuum sources is proportional to the band-pass of the mirror of about 10%, thus EUVITA is a set of eight extreme UV telescopes, each sen- providing a good compromise between well defined spec- sitive in a wavelength band of about 10% centered at wave- tral windows and sensitivity. lengths between 50 A and 250 A. EUVITA is to be flown on the Russian mission SPECTRUM X-G. The present pa- The combination of effective area and very long exposure per includes an overview of the scientific objectives, as well time will provide a very high sensitivity when compared as a description of the instrument configuration, the optical to that expected from other space experiments, and will and detector characteristics, and requirements on thermal allow a detailed study of a number of sources covering a stability and attitude reconstruction. wide range of astrophysical problems (see section 2). In addition, the relatively small bandwidth of each instrument will provide a clear deconvolulion between the emission in 1 INTRODUCTION the different bands. Although, the experiment will not be EUVITA consist of an array of eight extreme UV imaging a true spectroscopic instrument, it will be able to charac- telescopes with incidence reflection multilayer coated mir- terize the shape of continuum emission much better than rors with central wavelengths of the individual telescopes most current EUV instrumentation, and determine source between 50 A and 250 A. EUVITA will be part of the pay- temperatures by observing the relative strengths of emis- load of the Soviet SPECTRUM-Roentgen-Gamma satellite, sion lines dominating the emission in the different bands. to be launched in 1993. At present, little is known from observation and theory on the emission in the extreme UV domain, due to the very The Spectrum Roentgen-Gamma (in short spectrum-X-G) few experiments that have been flown and the large inter- satellite, shown in Fig. 1, will be launched in a highly stellar extinction expected close to 100 A. However, it has eccentric orbit with a period of 4 days, allowing long un- recently been discovered that the interstellar medium is very interrupted observations. The scientific payload is aimed at inhomogeneous, leaving relatively large regions of the sky a broad range of X-ray astrophysics. It includes SODART, with small hydrogen column densities of 1018Atoinscm"2 a Soviet Danish grazing incidence telescope of 8 m focal for objects within 100 pc ani of 1O20Atoms • cm"2 out of length optimized for high resolution spectroscopy of weak the galaxy. The all sky surveys by the WFC on ROS AT and sources; JET-X, a mainly British grazing incidence X-ray by EUVE will provide essential information on the number telescope equipped with cooled CCDs as focal plane de- and approximate position of sources that can be observed tectors, sensitive between 0.5 keV and 10 keV; MART, a with EUVITA, and hence will give important input for the coded mask hard X-ray image; ART-SP, a wide angle hard preparation of the observation program. X-ray image installed on a steering platform; GITA, a grazing incidence X-ray telescope also on the steering platform; EUVITA is an international undertaking, including the USSR, and EUVITA, with 6 telescopes on the main platform and Switzerland, USA and Canada. 2 on the steering platform. The basic design of each EUVITA telescope consists of a 2 SCIENTIFIC OBJECTIVES spherical mirror of 20 cm diameter, with a EUV-detector in the focal plane. The EUV photon detector is a microchan- The interstellar absorption cross section in the EUV, given nel plate (MCP) with a position sensitive readout. The by Morrison and McCammon (1983), is shown in Fig. 2. telescope will have an effective area of few square cen- It can be seen that the absorption cross section decreases timeters in each of the wave bands proposed with a spatial as E"3 in most of the wave band considered for EUVITA. resolution of about 10 arcsec. The field of view is about Thus the smallest absorption is expected for the telescopes one degree. sensitive to the shortest wavelengths. Only the nearby tar- 52

Figure 1 Spectrum X-G space apparatus developed by the Lavochkin SUE. 1 Astrophysical module; 2 Mechanical link to platform; 3 Platform for scientific equipment; 4 SODART; 5 JET-X; 6 MART; 7 and 14 EUVITA; 8 MOXE X-ray monitor; 9 SPIN gamma-burst detector; 10 Steerable platform; 11 GITA; 12 Star tracker; 13 ART-SP

gets (200 pc) being observable beyond 150 A. OSAT or will be possible with ROSAT, or the EUVE scanning telescopes. This "spectral" capability will provide an , , EUVITA Bandpass unprecedented diagnostic tool in the EUV domain. 1000. It is currently planned to have 6 telescopes (see position 7, Fig. 1) co-aligned with the JET-X and SODART X instru- 900. ments. This combination will allow a very wide spectral coverage of the objects to be observed, a very important feature to understand the emission mechanisms of various suu. Ca sources. This will also be a great help in deconvolving the - Ary emitted spectral shape from the interstellar effects. Con- 400. - S f versely, EUVITA data will be of great help in determining - Si, the column density towards sources and thus constrain the models relevant to the X-ray emission. 300. - No Grains- LlJ The instruments on the steering platform (ART-SP, GITA 7 and EUVITA, see position 14, Fig. 1) provide also a com- b Fe-Lj Grains 200. plement, characterized by a wide spectral coverage. They y will provide a very efficient tool of EUV and X-ray obser- 100. vations from about 0.120 keV to 30 keV. ^H +He The central wavelengths for the instruments on me main —' yH Only " ^—~^_ platform will be around 50 A, 100 A, 130 A, 175 A, 250 n 77771 ~1 i—, i t 111n 1 . 0.1 1.0 10. A and possibly in the band of 1300 - 2000 A. The latter PHOTON ENERGY (keV) wavelength band could be used for attitude information by Figure 2 observing hot stars (see sect. 4.3). The central wavelengths Photoelectric absorption cross section per hydrogen of the telescopes on the steering platform will be around 60 atom as a function of energy A and 110 A. The exact wavelengths will be determined by more detailed simulations and will depend on the available coatings. The relatively narrow bandwidth of each telescope will allow a much cleaner spectral analysis in the EUV than was Since all EUVITA telescopes will be operated simultane- possible with the broad-band overlapping filters of EX- ously, variability of the spectral properties of sources will 53

be continuously monitored. '1 us is not possible with instru- of bands will help determine the angular distribution of the ments in which wavelengths are selected by filter wheels. gas and the ionization state of the plasma in the different directions. 2.1 Planetary Science 2.5 Interacting Binaries The interaction between the gaseous outer layers of plan- ets and the interplanetary space, and in particular the solar SS Cygni, a cataclysmic variable, was observed in the EUV wind, is a region of intense activity. This activity is ob- domain while in outburst. Cataclysmic variables emit most served as Aurorae and emits in the EUV domain. Many of their flux in the soft X-ray and EUV domains. CVs are ions of planetary atmospheres have transitions in the EU- relatively near objects and numerous. They are the objects VITA pass bands, EUVITA will be capable of detecting for which detailed accretion physics can be tested. Soft X- Jovian Aurorae in about 10000s. For Aurorae, EUVITA ray and X-ray observations of cataclysmic variables have will be capable of measuring the relative intensity of the provided rich information on the instabilities of accretion different ions, and with its excellent spatial resolution pro- discs. Extending the spectral domain in which cataclismic vide information on the spatial distribution of activity. variables can be monitored will allow the measurement of accretion instabilities in temperature regimes which were 2.2 Stars not accessible until now. CVs are variable on many different timescales, from tens of seconds to years. The long It is unlikely that the atmosphere of normal stars will be ob- uninterrupted observations of EUVITA will allow to probe served by EUVITA, as their characteristic temperature lies several decades of variability timescales. considerably below the region in which the instrument will be sensitive. However, it is well known that the coronae of A revival of interest in symbiotic stars began with the IUE late-type stars are sources of intense soft X-rays and EUV satellite. The nature of the hot component in these double radiation with characteristic temperatures of the order of star systems, some of which show nova-like outbursts, is 1-10 Million degrees. Multi-band EUV observations will still an unsolved problem. Theoretical investigations are provide important temperature diagnostics, based on the ra- strongly dependent on the spectral energy distribution on tio of flux detected in various band passes. In addition to both sides of the He II 228 A ior.ization limit. the long exposure times, the instruments will provide important data on the time-dependent activities of stars, and 2.6 Galaxies and Clusters of Galaxies allow to include the correlation of other stellar parameters, such as the rotation period (measured in other spectral do- UV images at 2000 A and X-rays observations of galaxies mains). have provided a very different picture from that obtained in the visible domain. It is possible that EUVITA could 50% of the galactic soft X-ray sources could be late M detect the hot gas in galaxies which produce the observed dwarf flare stars. One can expect that characteristic rise X-ray emission. This would provide important temperature times during flares are wavelength dependent in the EUV diagnostic information. domain. EUVITA, with its capacity to measure simultaneously in different bands with high time resolution, will help Hot gas and cooling flows have been detected in several understanding the heating and acceleration processes. clusters of galaxies. Observation of these features in the EUV will help to follow the fate of the flows as they cool Hot stars are the source of powerful winds which interact down to temperatures of about 1 million degrees. with the surrounding medium or indeed with the wind of other hot stars in symbiotic stars. These interactions are the source of strong shocks, again powerful emitters in the 2.7 Active Galactic Nuclei EUV domain. Temperature diagnostics on these sources will likewise give important new contributions. Although AGN are extragalactic objects, some are located in regions of the sky with a low absorption column density and are expected to be observable by EUVITA. 23 White Dwarfs Approximately one third of AGN has been shown to have HZ43 is the brightest source in this energy range. The pa- a component contributing to the EUV emission in excess of rameters governing the emission from while dwarfs are the the extrapolation of the canonical X-ray power law. This temperature, composition and interstellar absorption. EU- component has been named "soft excess", its spectral shape VITA with its 8 spectral bands will provide information on is at present very poorly determined, due in large part to the these parameters for HZ43 and many other white dwarfs. small spectra] information available from the broad filters It is expected that the ROSAT WFC will detect several of EXOSAT. It is thus at present not possible to determine whether this component is the high frequency tail of the thousands white dwarfs, thus providing a rich sample from blue bump dominating the UV emission or whether it is an which more detailed EUVITA observations will be selected. addition to the already known components. The relatively narrow filters of EUVITA in the range between 50 A and 2.4 Interstellar Medium 130 A will be a crucial source of information on this soft excess component. Observations of white dwarfs and nearby stars will also provide crucial information on the local structure of the in- The soft X-ray spectral domain is thus very complex in terstellar medium. The present picture suggests that the sun AGN. The complementary nature of the X-ray instruments is embedded in a 3pc cloud surrounded by a very low den- and EUVITA will be very powerful to deconvolve the var- sity hot intercloud medium. EUV observations in a number ious emission components. Both types of instruments will 54

also cover a wide spectral domain, and thus help to under- mounted directly onto the the front of the detectors. These stand the relationship between the emission components. will be chosen to match the bandpass of each mirror type, Since the soft excess componem has been shown to be and will employ well understood materials (Vallerga et al., highly variable in several sources, EUVITA will be able 1986). The main function of the filters is to attenuate the lo follow this variability separately in the different wave atmospheric airglow lines (Helium and Hydrogen) in the bands. EUV and FUV which will be the main background sources. The prime focus assembly, including the detector, will be The telescopes on the movable platform could well be used 10 cm in diameter. to perform additional monitoring measurements, and thus contribute to the study of the lime variability of the soft There will be a grid biased at +28 V across the detector excess emission. aperture to provide an ion rejection. The broad band filters attached to the detectors provide a high level of charged particle rejection as well. The cylindrical prime focus con- 3 INSTRUMENT CONFIGURATION tainer which holds the detector and electronics also acts as 3.1 Telescope and Multilayer Mirrors a magnetic shield to eliminate any magnetic field perturba- tions that may affect the detector imaging. Mir Detector Box An opaque photocathode layer will be deposited directly onto the front of the top MCP in the stack to enhance the quantum detection efficiency. A fine pitch mesh mounted in front of the MCP stack biased to provide a repelling field R2600_ to reflect outgoing photoclcctrons back onto the MCP will provide further QDE enhancement. 1400 The EUVITA requirement for a 10 arcsec resolution over a Figure 3 1° field of view indicates that a detector resolution of about EUVITA telescope with EUV detector 50 microns FWHM over the 30 mm detector aperture would be adequate. The detector will then not be the dominating Figure 3 gives a schematic view of one EUVITA telescope term in the telescope resolution budget. This performance with its EUV detector. The eight telescopes are identical, may be obtained with a standard wedge and strip design. differing solely in the multilayer coating of the mirrors, the filters and the photocaihode material, the elements defining the wavelength domain of each individual telescope. 4 INSTRUMENT SENSITIVITY Multilayer coatings in the required wavelength range can 4.1 Effective Area be fabricated as artificial Bragg crystals consisting of alter- nating layers of different refractive index materials (usually on Si substrates). The thicknesses of the layers arc chosen in such a way as to enhance the overall reflectance in the particular wavelength band. Coalings with reflectivities of 10% to 40% have been produced in the USSR (Platonov ct al.), the UK and in the USA. The mirror bandpasses should be as narrow as A/A \ ~ 10, yielding a moderate energy resolution. The diameter of the mirrors is 20 cm with a § E focal length of 140 cm. EUVITA (100.000 s) 1 The field of view is 1.1 degrees. The quality of the optics EUVE Survey (2000 s). is aberration limited and should be better than 10 arcsec. The degradation of the point spread function towards the outer regions of the optics and detector is to be defined. The thermal and pointing stability of the spacecraft are currently I ... I under investigation (sec sect. 4.3). zoo

3.2 MCP Detector Figure 5 Sensitivity of EUVITA and EUVE survey band-passes The detectors will be mounted at the on axis prime focus in a housing supported by a spider arrangement. The For our present estimation of the effective area, we assume prime focus assembly will also contain the detector front- for the mirror reflectivity 10%, for the filter transmissivity end electronics (preamplifiers, amplifiers and ADC's). The 80% and for the detector quantum efficiency 25%. This proposed detector aperture is 30 mm diameter, which can gives an effective area of 236 cm2-0.1-0.8-0.25 = 4.7 cm2 be easily achieved with standard microchanncl plate sizes. for each individual telescope. Since the telescope is F~ 7 the focal plane curvature is small (

4.2 Background The MCP configuration will produce a background count rate (at sea level, room temperature) of less than 0.5 events sec^cm"2, or less than 3.5 events sec"1. Experience with previous experiments (e.g. EXOSAT) shows that the total background in orbit is about 25 events sec"! per telescope, dominated by cosmic rays. This rate corresponds to a background of B = 8 events in a field of 50 micron2 (the expected FWHM resolution) during 105sec integration time. Including this background, the minimum detectable 3 sigma signal S is therefore S(3u) = 5 4-^/(25 + 105), or about 15 events. The arrangement of the detector is, however, different than on the previously Figure 6 flown missions, where the MCP detector was pointed to Functional block diagram of the Detector Electronics wards open space. The EUVITA MCP detector is better Box (sec text for details) shielded and the background may be somewhat less than the EXOSAT background, although the orbits are similar. The background will need detailed modelling. A prelimi- termination circuits, designed for CDHS to DEB distances nary analysis of the achievable photon sensitivity is shown of up to 10 m. in Fig. 5 including EUVE survey results. Clearly, the EU- VITA telescope has a superior performance as well as a Each detector is read-out via two redundant links by a Com- better energy resolution. mand and Data Handling System (CDHS), which serves 2 detectors. Each CDHS has two microprocessors including 0.5 Mb of memory. Figure 6 shows the functional block 4.3 Attitude Reconstruction diagram of a Detector Electronics Box. The charge of the wedge, strip and zig-zag anodes is integrated, RC-shaped The attitude reconstruction requirement is driven by the and digitized in 14-bil ADC's. The raw data is read by precision of the wedge and strip readout and the optical the microprocessor and corrected for ADC non-linearities; performance. In order not to degrade the point spread the event position (X,Y) is calculated and stored in a mass function, the attitude reconstruction must be better than memory. Attitude and on-board parameters (house keep- 10 arcsec during the exposures. The spacecraft design ing) at time intervals to be defined are also stored in the states that the drifts will be about 10~4 degrees per sec. EUVITA mass memory. in a 2.5 x 2.5 arcscc2 region. This jitter will therefore not influence the sensitivity of the instrument. However, ther- The hold signal for the ADC's is obtained from the Lower mal deformation of the platforms may have an amplitude Level and the Upper Level Discriminators (LLD, ULD) of 3 arcmin. during one observation time period, which which are controlled by the sum of the signals W,S and Z. is typically about 6 to 8 hours. This will impose an atti- The same hold signal is used to control the fourth ADC tude reconstruction every 5-10 minutes. Possible methods digitizing the sum signal. After conversion a Look At Me to obtain the attitude information are under investigation. (LAM) signal is generated and the ADC data should be One method would consist in implementing a long wave read-out by the CDHS. length telescope identical to the EUV telescopes, but sensi- The house keeping functions arc defined to be the follow- tive around 1300 - 2000 A, in a spectral region where pho- ing: tons are much more numerous, and where hot stars could • threshold adjustment of LLD and ULD discriminator be used to monitor the thermal drifts of the EUVITA struc- levels by CDHS commands, ture. A dedicated star tracker for the EUVITA array is also being considered. Clearly, all EUVfTA telescopes must be • measurement of the event input rate (rate at the ULD rigidly mounted, and time-dependent distortions of the in- output) and dividual telescopes with respect to the star tracker must be • measurement of the dcUclor chamber and the elec- kept under 5 arcsec. tronics chamber temperatures The same requirements on pointing stability exist for the For position calibration purposes and for dead time correc- telescopes on the movable platform. However, it is planned tions 4 pulsers are integrated on the wedge and strip anode. to use a dedicated star tracker for attitude reconstruction. Firing these pulscrs with a fixed frequency gives 4 well defined event positions with a known event rate. 5 DETECTOR ELECTRONICS AND In addition to the detector readings, house keeping data and COMMAND, the total count rate of each detector will be collected at a frequency to be defined. The front end electronics will be AND DATA HANDLING SYSTEM (CDHS) designed to handle 1000 events per second and detector. The block diagram of the EUVITA electronics is shown in It is foreseen to build in the CDHS a two-dimensional X-Y Fig. 6. A Command and Data Handling System (CDHS) histogram with reduced pixel resolution. This histogram al- and the corresponding Detector Electronics Boxes (DEB) lows to monitor unexpected high count rates in individual communicate through EUVITA Detector Communication pixels and could apply scaling factors to those pixels, in Channels (MIL-1553B). To reduce power requirements, all order not to fill the mass memory with uninteresting data. lines of the MIL-1553B link will be implemented without Also it allows to mask (with a 16 x 16 mask) unwanted 56 fields of view areas. REFERENCES One solid state mass memory is foreseen for all detec- Morrison, McCammon, AstrophysJ. 270, 119 (1983) tors. The required memory, assuming one data dump to Platonov Y.Y., Polushkin N.I., Salashenko N.N., Fraerman ground per day is given by the expected background and A.A., Zh.Tekh.Fiz. £7, 2192 (1987) source rale. The amount of memory space should be about 120 Megabyte. Following the good experience with a Winch- Sicgmund O.H.W., Everman E., Vallcrga J., and Lampion ester disk drive on the GRANAT spacecraft, we plan to M., Proc. SPIE 86§, 18 (1987) space-qualify such a device for possible use as a back-up Siegmund O.H.W., Chakrabarti S., Cotton D., and Lampton storage medium. M., IEEE Trans.Nucl.Sci. NS-36, 916 (1989) Vallerga J.V., Sicgmund O.H.W., and Jelinsky P., Proc. SPIE 689, 138(1986) 57

PROTON IRRADIATION TEST OF THE 14 bit/100 kHz COMPLETE SAMPLING ADC AD 779 KD

P. BUhlcr', N. Schlumpf*, A. Zehnder*

* Paul Scherrer Institute, 5232 Viiligen PSI, Switzerland

alihrjilion residuals fiDC RD779KD Introduction

The AD 779 is a complete, multipurpose 14-bit monolithic analog-to-digital converter, consisting of a sample-hold am- plifier, a microprocessor compatible bus interface, a voltage reference and clock generation circuitry and is fabricated on Analog Devices BiMOS process. This device will be used in the MCP readout system of the EUVITA telescopes. These telescopes are one part of the payload of the Spcctrum- Rocntgen-Gamma satellite mission. We report the preliminary proton irradiation test results of one single ADC unit with the 62 MeV proton beam in the OPTIS [1] area of PS). 0.25 0.S 0.75 BMC-PB'J pulse generator [-1 *E5 Figure 1 Experimental Setup

6? MeU nroton irradiated ADC RD779KD: {(p):I.3E.*7/cni6ec We used the standard OPTIS proton beam setup as described in [1] with the last quadrupolc magnet QLA 18 switched off. The proton flux at the ADC position was measured with a ionization chamber (diameter = 6mm, thickness = 2mm, current = 0.04 nA). Using the ionization potential of 37.0(9) cV/ion pair [2] we obtain a proton flux of 1.3 107 protons/cm2sec. This corresponds to a dose of 1.84 rad/scc in silicon. During the irradiation tests the ADC worked with about 250 conversions/sec, the output data were accumulated by the TANDEM Data Acquisition system and analyzed with the PAW software. The power supply currents were measured during the irradiation sessions of ~ 103 sec. acc. Dose in Silicon (k raril Figure 2 Results 42 HeU urDton irradiated ADC AD779KD: B

-1 -3E»//cn'sec r First we present the calibration of the ADC, done prior to ''7 the irradiation, see Fig. 1. \ gfl Figure 2 shows the observed slight increase of the digital output in function of the accumulated dose (5 chan- ncls/krad). This increase could be explained by thermal drifts of the ADC. After the accumulated dose reached 3 krad, more than one line occurred in the ADC histogram. This is a clear significance for an irreversible damage of the ADC. The measured increase in power consumption is shown in Fig. 3. At the beginning of the irradiation the power consumption was: + 15 V/19.4 mA; -15 V/24.6 mA; +5 V/5.7 mA. acc. Dose In Silicon (k rail) The occurrence of the multi-line pattern in the histogram is Figure 3 indicated by the dashed line. 58

Acknowledgements References

We arc very grateful to Ch. Markowts and Dr. E. Egger {l] ch MarkovitS) s Jaccard and Ch. Pcrrct, 12th Int. for making the OPTIS area available for our lesis and for Conference on Cyclotrons and their Applications, their assistance during the measurements. BerIin 1989 ^ p^^ Bcam Facj|jty OPTIS for the Therapy ol Ocular Tumors. [2] J.M. Denis, 1. Slypcn, I. Tilquin, J.-P. Mculdcrs, Av- erage lonizalion Energy, w, for 65 MeV Protons in Nitrogen, EPAC 90, Proceedings of the 2nd European Particle Conference, Vol. 2. 59

TEST CHAMBER FOR THE SIMULATION OF OUTER SPACE

K. Thomson*. T. Kontci*

• Paul Scherrer Institute, 5232 Villigen PSI, Switzerland

PSI contributes to the ESA cornerstone project XMM (X- Ray Multi-Mirror Mission) by developing the detector box of the RGS (Reflection Grating Spectrometer) instruments including the cooling system for their CCD detectors. In order to verify the envisaged concept of cooling by dissi- pating heat via a radiator facing cold space a test chamber has been designed and constructed. With substantial support from the FI Research Department the XMM Space Test Chamber has been taken into operation during the last months and initial calibration measurements are underway. The test chamber is designed for performing thermal tests, especially concerning radiative heat transfer. In the heart of the apparatus we simulate a "piece of empty space", i.e. there is a specially coated absorber which can be cooled down to 10 Kelvin and which is going to absorb any thermal radiation from items under test. The first component to be tested will be the XMM radiator which has to provide cooling to the RGS detector box and its CCDs. At 20 Kelvin operating temperature the cooling power of the refrigerator in the test chamber is in the order of 20 Watts. Together with the applied black coating on the absorber this allows for very realistic conditions in terms of absorb- ing thermal radiation emitted towards the absorber. Figure 1 shows the top flange and pan of the interior of the thermal lest chamber; inside a first heat shield which itself is cooled to about 80 Kelvin resides the absorber. It has the form of a box, a cylinder which, depending on the types of measurements performed, is closed at ieast on one side. The outside is covered with a space qualified aluminum tape to reduce parasitic thermal load to the coldest parts, on the inside a two-layer, two-component coating ensures Figure 1 a total emissivity of the surface in the order of 0.9 for the Top flange of the XMM cold test chamber with the outer whole range of radiation from the visible to 100 pm wave heat shield visible length. After tests of the present XMM radiator design simulations Along with these tests, mainly concerned with XMM, the of complete orbits of the space craft will be performed to facility will be used to determine thermal material's prop- gain an understanding of the actual performance of the cur- erties, especially total cmissivities of surface coatings. rent available bread board model of the XMM RGS detector For this and possibly other applications interest in using the box and to provide an input to ongoing numerical simula- thermal test chamber has been stated from research insti- tion efforts. tutes concerned with the development of satellite experiments and also from space industry. For the future it is intended to use the thermal test chamber for the qualification tests up to the actual XMM flight models. 60

ON THE PERFORMANCE OF SUPERCONDUCTING TUNNELING JUNCTIONS AS DETECTORS FOR NON-THERMAL PHONONS IN SINGLE CRYSTALLINE SILICON

W. Rothmund', C.W. Hagen', A. Zehnder"

* Paul Schcrrcr Institute, 5232 Villigen PSI, Switzerland

INTRODUCTION to the o-absorption spot detects always more energy than In earlier work we reported on the detection of a-radialion the other detector. This can be clearly seen in Figs. 4a by means of Sn/SnO^/Sn superconducting tunneling junc- and b. In Fig. 4a the slit was positioned parallel lo the tions (STJ) evaporated onto a single crystalline Si absorber junction leads, on the side of STJ2, and at 200 fim distance [1]. Coincidence measurements on a pair of junctions re- from the midpoint between the junctions. Only events with vealed a strong correlation between the energies deposited more energy in STJ2 arc visible. Figure 4b displays events in the two detectors, which is basically due to the anisotropic with largest energy in STJ 1 for the slit moved over to the phonon propagation in the Si absorber. However, the ob- side of STJ1, at the same distance of 200 /an. In Fig. lc served energy correlation could not be explained solely by the slil was just above the junction pair. The substructure, assuming that the phonons propagate ballistically in the ao- which is seen at higher energies in the energy scatter plots sorbcr. Boundary scattering and isotope scattering were in- for slit positions above the leads, is fully absent here. This cluded in our simulations but the calculated energy deposits indicates already that the current leads are an essential part in the junctions always came out lower than the experimen- of the detector. From the rise time measurements this be- tally observed ones. In order to explain this discrepancy comes more evident. For the measurements presented in we performed additional experiments in which the Si ab- Fig. 1 the corresponding rise times of the signals of STJ1 sorber was only partly exposed to a-radiation. From the versus energy El are shown in Fig. 2. For perpendicular results of these experiments, which will be presented here, slit positions only one branch is observed which extends it is concluded that the current leads of the junctions are to higher energies when the slit is closer to the junction. an important part of the detector. Quasiparticles which are Rise time measurements for slit positions on either side ot generated in the leads by incoming phonons diffuse into the junction pair do not reveal exactly the same behaviour the tunneling volume and contribute to the total detected (Figs. 2b and d). This reflects the difference in granular- energy. This leads to the result that the detector, consist- ity and thickness of both current leads. These differences ing of a tunneling junction plus current leads, has a large show up clearly in the rise time scatter plots for parallel slit effective sensitive area and is location selective. positions, Fig. 3. Two branches are observed now, each originating from one of the current leads. EXPERIMENTAL SETUP The observed features can be understood in the following Two Sn/SnO^/Sn superconducting tunneling junctions were way. The phonons generated in the silicon substrate by evaporated 100 /*m apart on lop of a 10x13x0.33 mm3 sin- the absorption of an Q-particle propagate mainly along the gle crystalline <100>-Si substrate. The detectors consisted principal axes of the cubic crystal. This effect, which is of a 400 /;m2 junction area and two current leads of 1 mm due to the anisotropy of the clastic constants in Si, is called length and 5 /im width. The leads were parallel to one of phonon focusing [3]. After an o-particlc absorption the the principal axes of the silicon substrate. The lower film highest phonon dcnsiiy at the surface of the substrate will of the junction (i.e. the one in contact with the substrate) be found at the point just above the absorption spot and is typically 200 nm thick and consists of small grains. The along lines extending from this point in the direction of the upper film has a typical thickness of 500 nm and is made main crystal axes. When phonons hit the superconducting up of much larger grains [2]. The substrate was irradiated Sn leads they will create quasiparticles which diffuse into with 5.5 McV a-particlcs from a 2/|IAm-source through a the junction area. From solid angle considerations one ex- 70 )im wide and 2 mm long movable slit which was either pects that the quasiparticlc density in the leads is highest parallel or perpendicular to the current leads. The pulse at locations which are nearest to the o-absorption spot. At height and rise time of correlated events in the junction these locations the quasiparticle dcnsiiy is even further en- pair were determined for various slit positions and for front hanced due to the phonon focusing effect. The quasiparticlc side as well as back side irradiation. diffusion in the current leads will increase the risetimc of the signals and reduces the detected energy as a result of RESULTS AND DISCUSSION quasiparticlc loss processes. As the structure and the thickness of the leads arc different tnc quasiparticfc diffusion and In Figs, la to d the detected energy in STJ2 (E2) is plot- loss processes will also differ. Therefore for parallel slit po- ted versus the energy detected in STJ1 (El) for four slit sitions, when quasiparticlcs arc created in both leads, two positions, as indicated in the figures, perpendicular to the branches in ihc risctime plots arc observed. High-energy junction leads. The substrate was irradiated from the back signals, arising from phonons propagating directly into the side at a temperature of 0.6 K. All plots reflect the sym- junction area, arc not influenced by the properties of the metry of the slit-detector arrangement. The detector closest 61

1 2 1 1 1 2 1 2

la b a

STJ 2 (AR B U ) i j :m •jh 2 2 1 '2 / ' 2 .NERG V

2a b c t 3

• 2 d> k cr < t / 1 2 ' 1 2 1 2

b .5a . b •3 m \ 2 - 1 I

• / , 2 1 2 / .

Figures 1 to 5: ENERGY STJI (ARB. U.) The energy correlation between STJ1 and STJ2 is shown in Fig. 1 for perpendicular slit positions as niuicaicd in the insets. The corresponding risetirnes of STJl are shown in Fig. 2. Figure 3 displays the risetimes of STJl for parallel slit positions. In Figs. 1 to 3 the slit was positioned at a distance of: a) -400 fim; b) -200 fim; c) 0 fim; and d) +200 jim. Two energy scatter plots for parallel slit positions are shown in Figs. 4a (-200 fim) and 4b (+200 fim). Figures 5a resp. 5b are obtained from series of perpendicular resp. parallel slit measurements. For more details see text.

current leads, which explains why ihe rise time branches The results of these simulations will be discussed in a forth- coincide at higher energies. However, in the situation of coming paper together with the front side irradiation mea- perpendicular slit experiments, quasiparticles will be cre- surements and the results for substrates of different thick- ated mainly in the particular lead above which the slit is nesses. positioned. Accordingly only one branch in ihe risetime plots is obtained. The high-energy events in these single- CONCLUSIONS branch plots can be identified as originating from absorption It has been shown that the current leads of the junction spots just below the lead. By collecting the rise times only detector contribute significantly to the total detected energy. for these high-energy signals for 14 perpendicular slit po- This has the advantage of an increased sensitive area of sitions, ranging from -1000 ^m to +300 nm, one simulates the detector. Furthermore it is shown that the detector is actually a measurement with a parallel slit positioned just position selective in the sense that events on either side of above the current lead. The result is plotted in Fig. 5b. The the detector pair show up clearly separated in the energy two branches are clearly recognized, and agree very well correlation plots. with the parallel slit result of Fig. 3c. The same procedure can be used for the high-energy events of the parallel slit experiments which are due to phonons going directly into References the junction area, thereby simulating a perpendicular slit [1] W. Rothmund, C.W. Hagen, and A. Zehnder, in Low directly above the junction pair. This results in the single Temperature Detectors for Neutrinos and Dark Matter branch which is shown in Fig. 5a, which compares quite HI, edited by L. Brogialo, D.V. Camin and E. Fiorini, well to Fig. 2c. Editions Frontieres, 201(1990) The observed behaviour of the detector can be simulated [2] W. Rothmund and A. Zehnder, in Proceedings of the very well within a model based on phonon focusing includ- Workshop on Superconductive Particle Detectors, To- ing isotope scattering and boundary scattering as well as rino, ed. by A. Barone, World Scientific Publ. Singa- the quasiparticle diffusion and quasiparticle loss processes pore, 52(1987) in the junction leads. [3] B. Taylor, HJ. Maris and C. Elbaum, Phys. Rev. Lett. 23, 416(1969) 62

UHV-DEPOSITION OF SUPERCONDUCTING T(H)IN FILMS

C.W. Hagen*, F. Finkbcincr*, A. Jaggi', A. Sarro§, M. Schcrschel', A. Zehndcr* S. Zhaof, * Paul Schcrrcr Institut, 5232 ViJligen PSI, Switzerland t Balzcrs Ltd., 9496 Balzcrs, FL Liechtenstein t Laboratorium fur FcSfkorperphysik, ETH-Honggerberg, 8093 Zurich, Switzerland § IFCAI/CNR, via Mariano Stabile 178, 90100 Palermo, Italy

First results arc presented on the fabrication of supercon- ii) three effusion cells which can be healed up to ]4()0"C ducting thin films in an ultra-high-vacuum system. The ultimate goal of this study is the development of high- iii) a thermal (boat) evaporation source resolution radiation-and/or particle-detectors. Such detec- iv) three 1.5 inch (600 W) magnetron sputter sources tors may consist of either a superconducting radiation ab- which can be used for AC or DC sputtering. sorber in combination with a superconducting tunnel junction, or a narrow superconducting strip. The principle of Substrates can be cleaned by means of a fast-atom-beam operation of both types of detectors is based on the genera- source. tion of excess quasiparticies caused by energy deposition in Tin films were evaporated out of an effusion cell on to fused a superconductor. In the junction detector the excess quasi- quartz substrates. The background pressure was 4 • i0~ U1 particies are collected in the junction area and counted by mbar and the pressure during evaporation did not exceed measuring the tunnel current. These detectors operate at 5 • 10~9 mbar. The deposition rate was typically 3 A/s very low temperatures (T ~ Tc/10, where T,; is the su- but could be increased up to 12 A/s The substrates were perconducting transition temperature) and low bias current frc'io al room temperature. This is required by the use of (Ii, <1 /(A). A strip detector operates at temperatures closer photolithographic 'ift-off techniques for patterning the films to Tc and has a bias current somewhat smaller than the crit- (photoresist layers do not withstand extreme high or low ical current. After radiation absorption a small normal (hot) temperatures). The deposited films displayed a strong is- spot is created in the superconductor, thereby increasing the land growth. The size of the islands increases with film current density. When the current density exceeds the crit- thickness. For a 4000 A film the islands are typically a ical ci'. t density the strip is driven normal and a voltage few micrometers in size. Due to the island growth all across uu strip is developed, which can be measured. The films were electrically non-conducting. Connected films, performance of both types of detectors depends strongly on and thus conducting, can be obtained by introducing pure the relaxation of the initially created excess quasiparticlc oxygen into the vacuum system during deposition (1 j. This density or, in other words, on the quasiparticlc diffusion. increases the sticking probability of the Sn atoms to the The latter depends on the microstructure of the supercon- substrate surface. We found that, in order to obtain conducting films. In very granular films the quasiparlicle dif- ducting Sn films, a minimum Oj pressure of 3-10~' mbar fusion is slow which means that the initial hot spot volume is required when the deposition rate is 3 A/s and the sub- is small but contains a large energy density. This would strate at room temperature. For this pressure very pure films cause a strip detector to flip over to the normal state easily. are obtained with RRR ~ 35. Increasing the O2 pressure However, in a junction detector the quasiparticlcs need to results in films with higher room temperature resistivities 5 travel as fast as possible to the junction barrier, in order and smaller RRR. At 10~ mbar O2 the room temperature to detect them efficiently. Therefore, granular films are ad- resistivity is about 5 times larger and RRR decreases by vantageous for strip detectors but not for junction detectors. an ordci af magnitude. Above 31O~'1 mbar connected but For the latter one would athcr need epitaxial films. This almost insulating films arc obtained. By means of a 0-2O motivates the present investigation on thin film growth in X-ray diffractomcier the lattice constants were determined an ultra-high-vacuum system. for a conducting and a non-conuueting tin film. Both films showed a tetragonal structure with a = 5.816 ± 0.007 A As a wealth of experience on Sn/SnOj/Sn junctions exists and c = 3.177 ± 0.004 A. These values are 0.3% smaller in our group, we concentrated initially on the growth of than the values for bulk Sn, which is attributed to lattice- pure Sn films. strains. These ncasurcmcnLs also revealed a preferential growth direction with the a-axis perpendicular to the sub- Until 1990 tin films and junctions were g.own in a high- strate. vacuum system by beams of thermal evaporation. At a 6 background pressure of 10" mbar and with deposition The possibility of adjusting the RRR and the room tem- tales ranging from 10 to 30 A/s very granular films were ob- perature resistivity of the films by means of the O-, pres- tained, with residual resistivity ratios (RRR) of typically 10. sure enables us to fabricate junction detectors with different In the beginning of 1990 our Balzcrs UMS 630 ultra-high- quasiparticle diffusion properties. The junction fabrication vacuum system became operational. This system, which is is now in progress. shown in Fig. 1, cjntains a variety of evaporation sources: Parallel to (he work described above the growth of thin i) a 15 kW electron gun AI and Nb films, by means of DC sputtering, was started. 63 H-7L

Figure J The Balzers UMS-630 ultra-high-vacuum system is shown. The small chamber on the right-hand-side contains the sputter sources. The other evaporation sources are installed in the large main chamber.

This study has to result in the fabrication of Nb/A10x/Nb In the near future an additional evaporation technique will tunnel junctions. The advantage of the^e junctions is that be available: physical laser deposition. By means of a they arc more resistant to thermal cycling between helium 1 J excimcr laser thin films of high-Tc superconductors and and room temperature than Sn/SnO^/Sn junctions, and they refractory metal superconductors are planned to be made. have a larger Tc which allows a higher operating temperature of the junctions. By now 2 ,um wide strips have been Acknowledgements produced with Tc = 9.25 K, R(300 K) = 25 ftQcut and RRR = 4. X-ray diffraction measurements to determine the We are grateful to Dr. F.A. SaroU and T. Grabcr for per- lattice constants are in preparation. forming the X-ray measurements. In addition several smaller projects were carried out in the UHV-system. A wedge-and-strip-detector was made, con- References sisting of a complicated, photolithographically obtained pattern of Cu wedges and strips on a fused quartz substraic. A [1] H.J. Trumpp, thesis, Stuttgart, Germany 1976 better slicking of the Cu film to the substrate was achieved with a thin interlayer of Ti. From other projects experience was gathered in the evaporation of Au and Cr and the sputtering of PL 65

Chemistry (3200)

Geochemistry (3211)

Trace Elements (3212)

Aerosol Chemistry (3213)

Heavy Elements (3214) 66

ANNUAL BIOGEOCHEMICAL CYCLES IN RIVERBORN GROUND- WATER AT GLATTFELDEN, SWITZERLAND

H.R. von Gunten't, G. Karametaxas*, U. KrahenbuhT, M. Kuslys*, E. Hoehnt, R. Keilt, R. Giovanoli*

* Laboratorium fur Radiochemie, Universitat Bern, CH - 3012 Bern t Paul Scherrer Institut, CH - 5232 Villigen PSI * Laboratorium fur Elektronenmikroskopie, Universitat Bern, CH - 3012 Bern

In the perialpine belt of central Europe and in S 1«B6 1987 1968 BE& geologically similar regions a large part of the groundwater SO • •

15 is recharged from rivers. This situation is uniquely suited to • study chemical changes and processes during infiltration /v\

'•-•,•';- through a saturated aquifer. The impact of contaminated 6 i'' Si! ^3 Si rivers on groundwater quality is of major concern. For an 200 estimation of the migration of contaminants in an aquifer 1 150 • ^ and for an efficient water management it is essential to I 100 i understand physical, chemical and biological processes 50 : •';. •r~?x __ I \ which retard or remobilize water constituents. 0 . . '-CL The study site is located in the River Glatl Valley near ,' i NITRATE j *00 Zurich, Switzerland and is equipped with a series of •; !

sampling wells [1], It has been used extensively [e.g. C j^ 300 j

2,3,4]. The aquifer in the Glatt Valley consists of tightly ; • ! ^ - packed glaciofluvial outwash deposits of gravel and interbedded layers of poorly sorted sands containing 500 ';, \ quartz, calcite/dolomite and aluminium silicates. Water 5 « i from the river penetrates the uppermost part of the 10 D- :'' i saturated aquifer and travels with an average flow velocity • : ' I of about 5 md"l [5], The river water is a calcium carbonate water, but contains also about 15 % water from several biological sewage treatment plants. Water samples were collected from river and wells " J' 'x JW'A &' M • between 1984 and 1989. 0.45nm-filtered samples were used to determine cations and anions. Precautions were taken to minimize trace metal contamination. The investigated river/groundwater infiltration system is inherently complex due to changing discharge rates of the river, anthropogenic activities and varying groundwater levels, etc. Despite this complicated situation, the properties of samples taken at corresponding months Fig. I: Annual cycles of temperature, O2, NOf, Mn and showed similar trends during all years of investigation. Cd in the groundwater at 5 m distance from the river. Temperature and temperature-variations of almost 20 °C are key parameters for the processes which occur The chang s in biological, physical and chemical during infiltration of river water to the aquifer. They properties in the river and aquifer are fully or partly control biological activity at the water/sediment interface responsible for the behaviour of trace elements. One in the river and influence solubilities of chemical observes every summer a considerable increase of compounds, e.g. CaCO . dissolved manganese (Fig. 1). Its speciation in the 3 concentration peaks is not known, but calculations indicate Upon infiltration to the aquifer, the pJH decreases by 2+ >0.5 units. Along the further flow path it remains very that it is most likely Mn . We were able to show that the constant due to buffering by calcite. The pH-drop between main source of Mn is located within the river sediments. the river and the first sampling well is mainly due to The differences in amplitudes of the Mn peaks may be due degradation of organic matter. This process is also to climatic conditions, variable amounts of river sediments responsible for a significant decrease in the concentrations or changing groundwater levels. At greater distances from of oxygen and nitrate during summer (Fig. 1). the river and in winter, the concentrations of dissolved Mn begin to decrease because biological activity is reduced and/or the Mn deposits in the river sediments start to 67

exhaust. The river sediments need to be replaced to enable threefold P-reduction after 1985. In the groundwater a a new cycle. steady decrease of P with increasing flow distance is evident for 1984/85. In contrast, after 1985, P concentrations increase in the aquifer with increasing distance due to dissolution of ^sphates (memory effect), but are lower than before the ban of P. The concentrations of the measured cations and anions are in general lower than drinking water limits. Mn, Cu and NO3' being most critical. Furthermore, Mn, Cd and Cu reach toxicity limits for aquatic organisms in the peak concentrations. Work partly supported by me Swiss National Science Foundation.

References

[1] E. Hoehn et a)., Oas-Wasser-Abwasser 6_3_, 401-410 (1983).

[2] J. Zobrist el al., Gas-Wasser-Abwasser 5J>, 97-114 10 -6r-:T____l^ (1976).

[3] R. Schwarzenbach et al., Environ. Sci. Tecnn. .17. 472^79, (1983).

2.5 5.0 7.0 [4] L. Jacobs et al., Geochim. Cosmochim. Acta 5.2,2693- 2706 (1988). Distance from River Gtatt (m)

Fie. 2: Mean concentrations of phosphates in the River [5] E. Hoehn and H.R. von Gunten, Water Resour. Res. Glalt (0 m) and in welts at 2.5, 5, 7 w distance from the 25_,1795-1803 (1989). river. Upper curves: 1984185; lower curves 1986187; solid lines: measured; dashed lines: predicted. [6] C.C. Fuller and J.A. Davis, Geochim. Cosmochim. Acta 5J., 1491-1502(1987). Copper concentrations were in general somewhat higher in summer than in winter. Copper is mobile in the [7] J.A. Davis et al., Geochim. Cosmochim. Acta 5_i, aquifer and its conccnlrations do not change much along 1477-1490(1987). the infiltration path. The mobility reflects the stability of organic Cu complexes. The main source for zinc is the River Glatt (industry). Zinc concentrations are generally at a maximum in late summer, again due to decomposition of organic matter. One notices a fast reduction in Zn concentrations near the river and a slower decrease in the more distant aquifer due to sorpu'on. Cadmium, like Mn, shows significant annual concentration cycles (Fig. 1). The Cd peaks are not, or only partly related to the behaviour of Mn. Cadmium results to a large extent from decomposition of organic matter in the river sediments. Its further increase along the flow path (up to ~ 7 m from the river) is related to the dissolution of CdCO3 from surfaces of aquifer material. Beyond this distance, Cd concentrations decrease, probably due to sorplion on calcite surfaces and/or precipitation [6,7] when alkalinity and total inorganic carbon increase as the water moves further [4]. In 1985 phosphates were banned from i^tergents and were partly replaced by NTA. The concentration patterns of P in the river and ground water in 1984/85 and 1986/87 are presented in Fig. 2. In the river, one observes a 68

THE APPLICATION OF THE Rn-222 TECHNIQUE FOR ESTIMATING RESIDENCE TIMES OF ARTIFICIALLY RECHARGED GROUND- WATER: WATER CATCHMENT AREA HENGSEN, DORTMUNDER STADTWERKE AG

E. Hochn*, H.R. von Gunlcn't, U. Willme*, R. Hollcrung*. U. Schultc-Ebbcrl*

* Paul Schcrrcr Inslilut, CH - 52.32 Villigcn PSI t also Laboratorium fiir Radiochemic, Universilat Bern, CH - 3012 Bern * Instilut fiir Wasscrforschung GmbH Dortmund, D - 5840 Schwcrte 6

Recently a new technique was established to estimate Waters with residence limes of less than aboul four days groundwaler residence times of up to about 15 days [1|. were of particular interest because they showed mixing o( The technique assumes that the ingrowth of Rn-222 upon naturally infillraicd water with younger water. In two cases infiltration and movement in the ground can be described leaky underground water pipes were detected. In one case, by the growth law of radioactivity. Rn-222 emanates from a strong artificial recharge from the slow sand fillers was mineral grains by alpha recoil or by diffusion. It dissolves traced. in the ground-water and migrates in the aquifer without The radon technique proved to be very useful in ihe interactions. This was confirmed at two sites of naturally groundwalcr protection area al Hcngscn, where ground- infiltrating rivers and at a canal where the saturated aquifer water is used for drinking purposes. The knowledge of is recharged. Here, we give results from the application of residence times of up to about 15 days is of special inicrcsi the method to a water catchment area with artificial in the light of the decay kinetics of many fecal bacteria and groundwatcr recharge. the microhial consumption of oxygen in groundwatcr. Since 1936, the water supply of the city of Dortmund (FRG) relics on artificially recharged groundwaler (sec Figure). Raw wale is taken from River Ruhr and diverted to the reservoir Lake Hcngscn. After prcfillration in gravel References basins water from the lake is routed to slow sand filters at different rates. In addition, water infiltrates naturally from fl| E. Hochn, H.R. von Gunten, Radon in groundwatcr: the lake into a 5 - 10 m thick quaternary sand and gravel A tool to assess infiltration from surface waters to aquifer with a hydraulic conductivity in the order of 10"^ to aquifers, Water Rcsour. Res., 25(8), (1989) 1795- IO'TII/SCC. The groundwatcr of both artificial and natural 1803. recharge arc extracted from this aquifer by a horizontal filter drain, at variable mixing ratios and pumping rates |2], |2] U. SchOttler, H. Sommer, Hydrogeologische Unier- The groundwatcr at this site suffers from the reduction of suchung des Ruhrtalcs im Bcrcich dcr Wasscrgewin- iron, manganese, and sulfatc. The knowledge of the nungsanlagcn dcr Dortmundcr Stadtwerke AG, Band groundwatcr's residence time is of importance in studies of 5: Wassergcwinnungsanlagc Hcngscn - Arbcilsbc- the behavior of the rcdox-dependent species. richt des Institutes fiir Wasscrforschung GmbH In an area of groundwatcr recharge for public drinking Dortmund (1987). water, the use of artifical tracers to measure groundwatcr residence limes is prohibited. Groundwalcr flow velocities were calculated from How models. In a mixture of waters of an equal oxygen and hydrogen isotopic composition, the Rn-222 technique proved to be best to estimate groundwalcr residence times. The flow was .studied in groundwater observation wells of a diameter of 2 inches, along two transects of the site (sec Figure). The transects represent more or less the direction of the groundwatcr How. Most observation wells consist of several boreholes with piezometer tubes at different depths, which allow sampling in three dimensions. Since the noble gas radon is subject to outgassing, we sampled the groundwatcr with submersible pumps. Water from the reservoir lake infiltrates through the lake bed, rather than through the dam of a small hydraulic conductivity. In general, the radon concentrations increased with flow distance and confirmed the ingrowth model. 69

Lake Hengsen

O PVC, old DDavrviDon wtllt

• PVC, obacrvatlon wall totally icfka

•' PVC. 3 ot>«arvatlon walls pertly scr

* SUinir *« *t«ti, 3 odiarviuort walli partly tcr«sn«

-r SUinl*t» «MI. 6 Q6«rvtUon wvlla partly icraanad

WATERCATCHMENT AREA HENGSEN OBSERVATION WELLS

Figure: Plan View of Water Catchment Area 70

SAMPLE PREPARATION FOR AMS MEASUREMENTS OF32Si

D. Zimmcrmann

Paul Schcrrcr Instilut, CH - 5232 Villigcn PSI

32 The cosmogcnic radionuclide Si has potential the outside to 800-90CTC; trough a capillary die SiH4 is applications for various gcochcmical and cosmochcmical directed onto the hot Ta disk, where it decomposes. While

problems. The use of this nuclide has been hampered by the silicon is deposited on the Ta disk, the product H2 is the poor knowledge of its half-life and by its low continuously pumped off. 32 production rate, which leads to very low Si/Sitol ratios in Starting with BaSiF6, yields of Si were about 60%. As natural samples. Therefore, tons of material have to be test measurements by AMS have shown, suitable targets processed to determine 32Si via its daughter 32P by decay may be obtained this way. However, the rcproducibilily of counting. the deposition step is still unsatisfactory. Therefore, several Recently, we have redctcrmined the half-life of 32Si by parameters in the production of the targets have to be absolute AMS measurements to be (133 ± 9) y fl]. For investigated in more detail. 32 these measurements, samples of K2SiF6 enriched in Si 32 8 ( Si/Silol = 10 ) were mixed with tantalum powder and Acknowledgments pressed into a hole in the target holder to give targets for The author thanks G. Bonani, H.J. Hofmann, M. Smcr, the sputter ion source of the accelerator. In order to detect W.Wfjlfli and U. Zoppi for their careful performance of the 32Si in natural samples, the sensitivity has to be increased AMS measurements. by several orders of magnitude. Improved Si targets may contribute to this by giving high and stable ion currents. In addition, targets of very high chemical purity are required to keep contamination with the interfering isobar 32Si as Reference low as possible. Test measurements have shown, that 111 H.J. Hofmann, G. Bonani, M. Sutcr, W.Wolfli, D. semiconductor silicon targets contain less sulfur and yield Zimmermann and H.R. von Guntcn, Nucl. Instr. and about ten times higher currents than the K2SiF targets of 6 Mcih.B52, 544 (1990). above. Thus, a procedure for obtaining suitable deposits of about 1 mg of elemental silicon is presently developed. It consists of three steps which are performed in a vacuum system (Fig. 1):

1) Production of SiF4. For enriched samples (standards) this is conveniently

done by decomposing BaSiFg or K2SiF6 at 600-700'C in AL tube A; the evolving SiF4 is collected in trap B cooled with liquid nitrogen. From natural samples SiF4 may be generated by treating with HF/HC1O4.

B C D E lo HID pump 2) Reduction of SiF4 with LiAIH4.

LiAlH4 in ether is put into trap C and cooled with liquid nitrogen; trap B is kept at -78°C and SiF4 is condensed into the reaction vessel C. While the rcactants arc slowly Fig. 1: schematic representation of the vacuum system

brought to room tcmpciature, the formation of SiH4 takes A: stainless steel tube place. Trap C is then cooled to -130°C and the SiH4 BE: cold traps product gas is condensed into trap D, which contains dry F: lube furnace benzene at -196°C. The trap is warmed to room temperature and allowed to stand for about one hour, when G: piezoelectric pressure transducer the benzene absorbs the ether vapours. After cooling trap D II: capillary to -160°C, SiH4 is collected in trap E with liquid nitrogen. I: tantalum disk K: heating coil 3) Decomposition of SiH4. L: water cooled copper rod With valve 6 closed the SiH4 is admitted to the 1-8: valves deposition chamber, the flow being adjusted with valve 5. The deposition chamber consists of a silica tube in which a Ta disk (10 mm diameter, lmm thick) is placed on a water- cooled copper rod. The Ta disk is heated inductively from 71

CHANGES OF THE CONTENTS OF MICRO- AND MACROELEMENTS IN SPRUCE NEEDLES WITH THE NEEDLE AGE

A. Wyttenbach, L. Tobler, S. Bajo

Paul Scherrer Institut, CH - 5232 Villigen PSI

Most conifers retain their needles for several years and the needle age class, and by a content of age c'ass 1 that is thus have on the same branch needles of successive age equal to the mean increase per year, lne dynamic classes, which differ in age by one year. For the 2 to 6 behaviour thus does not depend on the supply to each tree essential macroelements that are usually measured, it is known that their contents differ between age classes. However, the behaviour of (essential and nonessential) trace elements is largely unknown, and it is also unknown which groups of elements behave similarly. Although it has been postulated that knowledge of the dynamic behaviour of elements is important for judging the nutritional status of the tree as well as the underlying physiological processes [1,2], there is up to now no systematic investigation of these effects. In the present work, the dynamic behaviour in the last three needle age classes was investigated in a population of 9 Norway spruce trees, all of similar age (50 years) and growing on an area with uniform pedological, climatic and immission characteristics (Schluchsee, Black Forest, BRD). c 5 Methodology Trees wore sampled in November by helicopter, taking branches from the 4th whirl. Needles were separated according to their age classes, cleaned from aerosols residing on their surface, and dried. Most elements were determined by instrumental thermal neutron activation [3]. As this technique has multi- element capabilities, simultaneous information on a variety of elements is obtained from the same sample. Moreover it has a very good sensibility and practically no blank and thus gives access to the microelements that are difficult or impossible to determine by other methods. The advantage of the method used in this work may be seen from Fig. 1, where the dynamic behaviour of Cr as determined by NAA and by a nonnuclear method is compared. For the 6 / Eu determination of Si, an instrumental method working with cpiihermal neutrons was developed [5] and found to give Figure 1: Cr content in needles as determined by 1NAA very reliable results. As, Cu, Br, and Ba were measured by (this work) and by ICP-AES [4). The values given are activation with subsequent radiochemical separation. N means ± SE. As the trees are growing on the same site *us determined with a C-N-analyzer. (Alpthal, SZ), equal values are expected. The observed differences in levels, dispersion and dynamic, behaviour Results must be due to analytical difficulties. It was surprising to find that for a given element the individual trees have a very similar dynamic behaviour. As but is governed by plant physiological processes which do an example, Fig. 2 gives the variation of the Al content of not vary among the individual tree. Since Si behaves three trees. Although each tree has its individuality as far exactly as Al, it is likely that with these two elements the as the content is concerned (what is probably due to a dominant processes are simply upward transportation and slightly different supply to each tree), their dynamic accumulation, but no retranslocation. The close connection behaviour is identical. In the present case this behaviour is between Si and Al was not known before. It is also evident characterized by an increase that is strictly proportional to from Fig. 2 that an appropriate normalization will reduce 72

the values of each tree to a unique curve which is quantitative and valid description of the siatus of the tree characteristic for the whole investigated population and than the concepts presently in use, and thus to a more maybe also to the species itself. efficient detection of abnormal conditions.

1 2 needle age class

Figure 2: Al content in needles of individual trees at different age classes. The values for the tree with the minimum and the maximum content and an intermediate tree are shown. It should be noted that Al contents at the site studied (Black Forest, BRD) are larger by a factor of 15 compared to most other sites. needle age class

Not all elements show the same type of dynamic behaviour as Al and Si. Some have no significant changes between needle age classes, and some even decrease with Figure 3: The dynamic behaviour of the alkali metals differ needle age; examples for all three types are given in Fig. 3. fundamentally. Cs (not shown) behaves exactly as Kb. The Constant or decreasing contents can only be explained by values (means +. SE, n=9) are normalized to the contents an out/lux from the needle, either by rctranslocation or by of needle age class 1 (Na 3.4, K 7655, Rb 147. Cs21 \Xg/g). leaching by acid fog or rain. The dynamic behaviour of mosL decreasing elements can be described by an Acknowledgement: asymptotic approach to a constant value. A classification of The permission to work on the experimental ground of elements that is simply based on the direction of the change the University of Freiburg i.B. in the Black Forest and the between age classes gives the following results: Increasing logistic support provided is highly appreciated. Thanks are are Al, As, Br, Cr, Fe, Hg, La, Na, Sb, Sc, Si, constant are also due to the staff of the Saphir reactor for providing and Ba, Ca, Cu, Sr, decreasing arc Cl, Co, Cs, K, Mg, Mn, N, maintaining the irradiation facilities, and to the Swiss P, Rb, Zn. Federal Institute for Forest, Snow and Landscape Research Some pairs of elements show a remarkably similar (WSL) for providing the C-N-analysis. dynamic behaviour and thus similar involved processes. While some of these pairs could be expected on grounds of similar chemical characteristics of the involved ions (Rb- References: Cs, Sr-Ba, Sc-La), others are quite unexpected (Si-Al, Zn- Mg, P-K). That chemical similarity does not always lead to [1] H.J. Fiedler, W. Nebe, F. Hoffmann, Forstliche a similar dynamic behaviour is shown by the different Pflanzenernatirung, Fischer, Jena (1973). behaviour of the alkali metals (Fig. 3). [2] H.J. Fiedler, BeitrSge Forstwirlschaft 22 (1988) 61. It is assumed that comparing these results with similar [3] A. Wyttenbach, S. Bajo, L. Tobler, J. Biol. Trace observations from experimental fields on other geological Elem. Res. 26/27, 213 (1990). bedrocks will show if the present behaviour for a given [4] M Stark, private communication. clement is universal or if it reacts to changes in the supply. 15] L. Tobler, V. Furrer, A. Wyttenbach, J. Radioanal. It is further assumed that this will lead to a more Nucl. Chetn., in press. 73

DIRECT MEASUREMENT OF MASS TRANSFER TO AGGLOMERATE AEROSOLS

U. Baltenspergeri, S.N. Rogak*, R.C. Flagan*, A. Portmann*

* California Institute of Technology, Pasadena, CA 91125, USA t Presently at Paul Scherrer Institut, CH - 5232 Villigen PSI t Institute of Inorganic Chemistry, University of Zurich, CH - 8057 Zurich

Mass transfer to aerosol particles is an important molecules forming a cluster with lower density and process in many natural and industrial systems. For diffusivity compared to the lead atom with a diameter of example, the enrichment of atmospheric aerosols with 0.35 nm. heavy metals has been attributed to condensation of metal vapors on particle surfaces in combustion systems. The transport of organics to a particle is an important mechanism of soot formation. Furthermore, mass transfer of ultrafine particles plays an important role in powder synthesis, the attachment of radon progeny to aerosols and in the measurement of atmospheric aerosols with the recently developed epiphaniometer f 1]. Theoretical descriptions of mass transfer to particles have traditionally been based on the assumption that ihe particles are dense spheres. However, many combustion systems emit fumes consisting of agglomerates of tiny spherules. The structures of these agglomerates has been described as "fractal" or "self-similar" [2]. For the free molecular regime where the particle • "IOO diameter is small compared to the mean free path X of the Mobility Dtamat* diffusing species, it has been shown that the mass transfer radius is the same as the mobility equivalent radius [3]. In Figure 1: Lead count rale f of deposited aerosols as a this work, we compared mass transfer and mobility function of mobility diameter, for titanium dioxide equivalent diameters for spherical particles (ammonium agglomerates, polystyrene latex spheres and ammonium sulfate and polystyrene latex) and titanium dioxide sulfate spheres. The best fit quadratic curve is compared to agglomerates in the transition regime, where the particle the best fit of the sphere data. The transfer rate is expected diameter is comparable to the mean free path. Detailed to be proportional to the coagulation coefficient K, procedures are described elsewhere [4,5]. For each aerosol, computed for a lead cluster diameter of 1.5 nm. a narrow mobility range was selected with a differential mobility analyzer (DMA). The attachment rate of radioactive lead atoms to these monodisperse aerosol particles Acknowledgement: This work was supported by the Swiss was measured with the epiphaniometer [1], giving an National Science Foundation and by the International Fine indication of the relative mass transfer rates of the lead Particle Research Institute. atoms to the spheres and agglomerates. The results are summarized in Fig. 1. The measured References quantity / is the epiphaniomcier signal, divided by the [1] H.W. Gaggeler, U. Baltensperger, M. Emmenegger, aerosol concentration and thus a measure for the average D.T. Jost, A. Schmidt-Ott, P. Haller and M. Hofmann, numrxr of lead atoms per panicle. It can be seen that J. Aerosol Sci. 20,557-564 (1989). spheres and agglomerates with the same mobility have [2] B.M. Smimov, Phys. Rep. 188,1-78 (1990). nearly the same / value. This is reasonable because the [3] A. Schmidt-Ott, U. Baltensperger, H.W. Gaggeler and nwss and momentum transfer processes are analogous, at D.T. Jost, J. Aerosol Sci. 21,711-717 (1990). least if the mean free paths of the diffusing species and the [4] S.N. Rogak, U. Baltensperger and R.C. Flagan, J. gas molecules are similar. Aerosol Sci. 21 in press (1990). [5] S.N. Rogak, U. Baltensperger and R.C. Flagan, Aerosol The experimental results were compared with the Sci. Technol. in press (1991). coagulation theory for spheres based on the transition [6] B. Dahneke, in: Theory of Dispersed Multiphase Flows regime interpolation by Dahneke [6] as well as the attachment coefficient theory of Porstendtfrfer et al. (7j. (R.E. Meyer,cd.), Academic Press, New York, 7-138 Good agreement of the curve shape was found between (1983). theory and experiment for a Pb diameter of 1.5 nm. This [7] J. Porstendorfer, G. Robig and A. Ahmed, J. Aerosol means that the lead atom is surrounded by water or other Sci. 10,21-28(1979). 74

IN-SITU MEASUREMENT OF SIZE , SURFACE AND MASS OF SILVER AGGLOMERATES

A. Weber", U. Baltensperger*. H.W. Gaggcler*. R. Kcil*. L. Tobler*, A. Schmidl-Ottt

* Paul Scherrer Institut, CH - 5232 VilJigcn PSI t Process- and Aerosol Measurement Technology, University of Duisburg, W - 4100 Duisburg 1, Germany

In order to describe dynamics, optics, or the chemical However, the experimental method applied gives correct behaviour of diffusion grown agglomerates information on values for dj only if df>2 . Additional experiments have to be their structure is needed. Especially for chemical processes performed to verify a devalue significantly higher than 2. in-situ measurements are often required. One possibility to Silver agglomerates formed by heating by a HF field describe agglomerates is the concept of fractals. It has been showed a higher fractal dimension of 2.60 + 0.15 indicating a shown in experiments [1] and simulations [2] that more compact structure (Fig. 1). This was corroborated by agglomerates grown by diffusional coagulation of primary transmission electron micrographs, which also showed larger particles have a fractal-like structure. This means that over a primary particles with 18 nm diameter. These increases are certain size range the following relation between the number due to the tempering effect of the hot surrounding gas as it of primary particles N of an agglomerate (or its mass M) and was also shown by [10]. a static radius (e.g. the radius of gyration Rg) holds [3] : Secondly this new method was combined with a N~M~R/f (1) simultaneous measurement of the exposed surface of an where dj is a structural parameter of the agglomera'e and agglomerate by the epiphaniomctcr. The epiphaniometer is called the fractal dimension. measures the attachment rate of neutral radioactive 211Pb It has been shown that also dynamic radii fulfill atoms, which is a measure for the exposed surface of an relation (1) in the free molecular regime for d^l and in the agglomerate [11]. This rale can be described by the Fuchs hydrodynar.^c regime fordy>l [4,5]. coagulation theory for spherical particles as well as for First we introduce a new method to determine the fractal deeply invaginated agglomerates [5,12]. In the free molecular dimension in-situ and on-line by measuring simultaneously regime the mobility equivalent diameter Dp is related to the the mobility equivalent diameter and a mass proportional number TV, of the primary particles, which are on the surface signal obtained by inductively coupled plasma atomic of the agglomerate and exposed to the surrounding gas, by : emission speclromeiry (ICP-AES) [6,7]. Ni-Df (2) For silver agglomerates produced in a spark discharge [8] Between the total number N of primary particles and the we found a fractal dimension of 2.19 + 0.16 in agreement number Nt of the primary particles on the surface of the with the value of 2.18 reported by Schmidt-Ott. The primary agglomerate the following relation holds in the free particle size was 6.0 + 1.3 nm. molecular regime [5]: (3) 3.2T

1.6 1.8 :, so log Dp [nm] diameter [ run ] Figure 1: Mass proportional ICP signal M' as a function of the mobility equivalent diameter for silver agglomerates Figure 2: 3-dimensional plot of mobility equivalent diameter, produced by heating. Linear regression yields a slope of 2.60 exposed surface and mass of silver agglomerates produced in for the fractal dimension. spark discharges.

The cluster-cluster aggregation model [9] yields Fig. 2 shows a 3-dimensional logarithmic plot of silver

two parameters is shown by the projection to (he according plane. For the absolute calibration of the agglomerate mass samples have been taken for neutron activation analysis (NAA). The experimental data confirm relation (2) and give a value for dj of 2.20 from (1) and 2.18 from (3). These values agree well with the one giver: above from a former experiment.

Acknowledgement This work was supported by the Swiss National Science Foundation.

References [1] S.R. Forrest, T.A. Wittcn, J. Phys. A 12, L109 (1979). [2] T.A. Wittcn, L.M. Sander, Phys. Rev. Lett. 47, 1400 (1981). [3] B.B. Mandelbrot,. The Fractal Geometry of Nature , Frecmann, San Francisco (1982). [4] A. Schmidt-Olt, Appl. Phys. Lett. 52, 954 (1988). [5] A. Schmidt-Ott, U. Baltcnsperger, H.W. Gaggeler, D.T. Jost, J. Aerosol Sci. 21,711 (1990). [6] U.K. Bochert, W. Danncckcr, J. Aerosol Sci. 20, 1525 (1989). |7] H. Kawaguchi, N. Fukasawa, A. Mizuike, Spectrochim. Acta41B, 1277(1986). [8] S. Schwyn, E. Garwin, A. Schmidt-Ott, J. Aerosol Sci.19,639 (1988). [9] P. Meakin, Computer simulation of growth and aggregation processes. In: On Growth and Form , edited by H.E. Stanley and N, Ostrowsky, Martinus Nijhoff, Hingham, MA, (1984). [10] A. Schmidt-Olt, J. Aerosol Sci.19, 553 (1988). [ 11 ] H.W. Gaggeler, U. Baltensperger, M. Emmenegger, D.T. Jost, A. Schmidt-Ott, P. Haller, M. Hofmann, J. Aerosol Sci. 20, 557 (1989). [12JS.N. Rogak, U. Baltensperger, R.C. Flagan, Aerosol Sci. Tcchnol., in press (1991). 76

RELATION BETWEEN MASS, FUCHS SURFACE AND MOBILITY DIAMETER OF PARTICLES IN COMBUSTION EXHAUST

H. Burtscher*, A. Leonardi*, D. Steiner*, U. Baltenspergert, A. Webert

* Labor fur FestkOrperphysik, ETH-H6nggcrberg, CH - 8093 Zurich t Paul Scherrer Institut, CH - 5232 Villigen PSI

Particles produced by combustion of organic material Fig. 1 shows the size distributions, obtained for the two are agglomerates of spherical primary particles. The engines at different engine loads. In both cases the panicle primary particle diameter is in the range of 10 to 40 nm, size increases with increasing engine load. the agglomerates have a chain- or grape-like shape. The Comparison of size distribution and mass allows to main components of the particles are elemental carbon obtain structure information. For the SI engine a linear (EC), a large number of hydrocarbons and some inorganic relation was found between mass concentration and the species as sulfates and water. This means, the particles integral [5] consist of a solid 'core' and a volatile fraction. The ratio between these components depends significantly on the J00 n(R)R2dR. kind of fuel and the characteristics of the combustion. It may vary from almost only EC to mainly volaliles. According to [6] this means that the fractal dimension In the following some results, obtained by applying a df of the agglomerates is < 2. number of different methods for particle mass and size determination to panicles from a Diesel and a four-stroke spark ignition (SI) engine are presented. Three instruments are used to determine the mass concentration: a tapered element oscillating microbalance (TEOM) [1], a P-absorption meter [2] and an aethalometer [3]. Whereas the TEOM directly responds to the mass, the other two methods measure absorption of radiation, which is then transformed into a mass. The aethalometer uses visible light. This means that it mainly responds to EC, which has an absorption coefficient much higher than the other constituents of the particles. (3-absorption also is material dependent, but much less than light absorption. Comparing TEOM or p-absorption and aeihalomcter data therefore allows to estimate the EC fraction. A differential mobility analyzer (DMA) is used for the size measurement, which means that the mobility equivalent diameter of the particles is obtained. Finally the Fuchs surface respectively the attachment coefficient of lead atoms to the particles is measured by an epiphanio- mcter [4J. Preliminary results of the mass measurements with the three methods mentioned show that • the values obtained by TEOM and P-absorption are quite similar and show no systematic deviations • in idle the ratio of aethalometer and P-meter data is about 0.25 for the Diesel engine, indicating an EC content of 25 %. At an engine power of 50 % of nominal power the ratio increases to almost 100 % • for the SI engine an EC content of 2-5 % is obtained, which increases to about 50 %, if volatile material is diameter luml evaporated by heating the exhaust gas to 200 °C (for unleaded fuel). Figure 1: Size distribution of a) the spark ignition engine, These results are still more qualitative and substantial operated with leaded fuel and b) the Diesel engine. The artefacts due to absorption of material from the gas phase on the filter cannot be excluded. distributions are normalized for easy comparison. 77

A similar result is obtained for the Diesel engine by comparing epiphaniometer results and mass concentration (Fig. 2). These experiments show a linear relation between Fuchs surface and particle mass, also indicating a df < 2. The daia at a mass concenuaiion of about 43 mgjem^ and very low epiphaniometer signals correspond to idle conditions. The mass measurement there is most probably adulterated by unbumed fuel, absorbed from the gas phase by the filter of the 0-meter. Another possibility to obtain information on the attachment coefficient is to measure the charge, particles acquire by ion attachment. However, in this case electrostatic forces may also influence the attachment probability, whereas in the case of neutral atoms as used in the epiphaniometer only diffusion is relevant Fig. 3 shows 300 400 500 600 the epiphaniometer signal, plotted versus particle charge due to attachment of ions, produced in a corona discharge. The fairly linear relation indicates that the electrostatic forces are not significant in this case. This means that both Figure 3: Epiphaniomeier signals versus panicle charge methods yield the came information, the ion attachment by attachment of ions, produced by a corona discharge for with a much simpler setup, the epiphaniometer with very Diesel particles. much higher sensitivity.

References

[1] H. Patashnick and G Rupprecht; American Laboratory, July (1986).

[2] B. Georgi and A. Kasenbrink; in Aerosols Science, Industry Health and Environment, S. Masuda and K. Takahashi (eds), Pergamon Oxford, 528-531 (1990).

[3] A.D.A. Hansen, H. Rosen and T. Novakov; Sci. Tot. Environ. 36,191-196 (1984).

[4] H.W. Gaggeler, U. Baltensperger, M. Emmencgger, D.T. Jost, A. Schmidl-Ott, P. Hallcr and M. 0 10 2030405060 70 8090 3 Hoffmann; J. Aerosol Sci. 20, 557-564 (1989). mass !nigr'n> J [5] A. Leonardi, H. Burtscher, U. Baltenspcrger and A. Weber; J. Aerosol Sci., in press (1990). Figure 2: Epiphaniometer signal versus mass concentration of the Diesel engine, determined by ^-absorption. The [6] A. Schmidt-Ott, U. Baltenspergcr. H.W. Gaggclcr engine load is varied at a constant speed of 3000 RPM. and D.T. Josl; J. Aerosol Sci. 21, 711-718 (1990). 78

BIOGENIC HYDROCARBONS AS AEROSOL PRECURSORS: AN OUTDOOR SMOG CHAMBER STUDY

U. Baltensperger*'. SJM. Pandis*. S.E. Paulson*. J.H. Seinfeld*, R.C. Flagan*, EJ. Palent, D.T. Allent, C. Schaffner*, W. Giger*, A. Portmann§

* California Institute of Technology, Pasadena, CA 91125, USA t University of California, Los Angeles, CA 90024-1592, USA t EAWAG, CH - 8600 Diibendorf § Anorganisch-chemisches Institut, UniversitSt Zurich, CH - 8057 Zurich 1 Presently at Paul Scherrer Institut, CH - 5232 Villigen PSI

Biogenic hydrocarbons are produced by vegetation in pinene, where formation of condensible aerosol products quantities comparable to anthropogenic emissions. were observed for p-pinene concentrations as low as 20 Biogenic hydrocarbons are oxidized in both urban and rural ppb. environments by a combination of hydroxyl radical and Scanning electron micrographs showed expected liquid ozone [1]. Recent aerosol measurements in Southern aerosol particles on all filters. However, filters from all California showed that a substantial fraction of fine aerosol experiments also coniained some agglomerates, comprised carbon is modem and suggested that the highly reactive of small particles of 40-60 nm size, indicating that a biogenic hydrocarbons may be important aerosol fraction of the aerosol formed consisted of solid material. precursors. Monoterpenes, the dominant biogenics from FTIR spectra showed absorbances of organic nitrates, C=O conifers, have been recognized as potential sources of (alkyl carbonyls and organic acids), and hydroxyl groups. organic aerosol since at least 1960 [2], but only a few GC-MS foi P-pinene aerosol samples revealed the presence studies have been carried out. In view of the limited data, of about 11 major species and several minor ones, with particularly under conditions similar to the atmosphere, molecular weights between 138 and 200 amu indicating which include both OH radicals and ozone as well as NOX) mainly mono- and dioxygenated products. From this large this study was undertaken to investigate the importance of variety of condensible components only nopinone and biogenic hydrocarbons as aerosol precursors. possibly pinocamphone could be identified so far. The The experiments took place in the Caltech outdoor isoprene-derived aerosol was also a mixture of several smog chamber facility. The smog chamber is a 60 m3 batch species with molecular weights ranging from 86 to 154 reactor constructed of Teflon film. Isoprene and P-pinene amu, with most of the compounds at around 112 amu. were used as reacting hydrocarbons. The desired mixture of NOx/biogenic hydrocarbon with optional seed (NH4)2SO4 Acknowledgement aerosol was injected sequentially into the covered chamber, This work was supported by the Swiss National Science time was allowed for mixing, and then the chamber was Foundation. exposed to sunlight. The gas-phase parameters measured included concentrations of O3, NO and NOX, isoprene, (3- pinene and major products, smog chamber temperature, and tola! and UV solar radiation. The aerosol particle size References distribution was monitored using the Scanning Electrical Mobility Spectrometer (SEMS) for the 0.01 - 0.2 pm size [1] R. Atkinson, Atmos. Environ. 24A (1990) 1-41. range and an optical particle counter for the 0.12 - 5 jim size range. The Fuchs surface of the aerosol particles was [2] F.W. Went, Nature 187(1960) 641 -643. measured using the epiphaniometer. Nuclepore filters from the epiphaniometer were subjected to scanning electron [3] S.N. Pandis, S.E. Paulson, U. Baltensperger, J.H. microscopy. Additional aerosol samples were collected on Seinfeld, R.C. Flagan, E.J. Palen, D.T. Allen, in: the eight stages of a Hering low pressure impactor. The Aerosols, Science, Industry, Health and Environ- aerosol deposits on each stage were characterized using ment, S. Masuda and K. Takahashi (eds), Pergamon infrared spectroscopy. The average saturation vapor Press, 1990, Oxford, pp. 974-977. pressure of the condensing species was measured with the Tandem Differential Mobility Analyzer (TDMA). Aerosol [4] S.E. Paulson, S.N. Pandis, U. Baltensperger, J.H. samples for GC-MS analysis were collected on annealed Seinfeld, R.C. Flagan, E.J. Palen, D.T. Allen, C. quartz filters. Details of the procedures are published Schaffner, W. Giger, A. Porimann, J. Aerosol Sci. elsewhere [3-5], 21 (1990) in press. From the aerosol yields at various initial concentrations it can be concluded that aerosol formation from the [5] S.N. Pandis, S.E. Paulson, J.H. Seinfeld, R.C. isoprene photooxidation is negligible under ambient Flagan, Atmos. Environ. 25A (1991) in press. isoprene concentration levels (0.1-30 ppb), in contrast to P- 79

AEROSOL MONITORING WITH THE EPIPHANIOMETER AT REMOTE LOCATIONS

D. T. Jost, H. W.Gaggeler, U. Baltensperger, M. Ei.. nenegger, W. Nageli, J. A. Kovacs, A. Weber, M. Schwikowski, D. Zimmermann

Paul Scherrer Inslitut, CH - 5232 Villigen PSI

The epiphaniomeier is a device used lo continously comparable to the 0.9 g/m3 measured in winter time on measure aerosol surface concentrations [1]. A battery Jungfraujoch. Jungfraujoch is therefore IM winter an ideal operated unit has been successfully run on Colle Ginfetti site for monitoring atmospheric background parameters since August 1988 [2], The demonstrated high sensitivity pertinent to our latitudes. and reliability have lead to numerous requests to use epiphaniometers in several field campaigns and at remote 3000 monitoring stations. Permanently installed epiphaniometers are located at Colle Gnifetti (4450m a.s.l., 45.9"N 7.9'E), Jungfraujoch (3450m a.s.l., 46.5 "N 7.9"E Switzerland), Sonnblick (3100m a.s.l., 47.0'N 13.0°E Austria) and Manna Loa (3400m a.s.]., 19.5"N 155.6°W Hawaii). For the ALPTRAC field campaign 1990 an additional instrument was installed at the Glacier dc la Girose (3360 m a.s.l. 45.0'N 6.3°E France). The ALPTRAC program, a subproject of the environmental EUREKA project EUROTRAC, is designed to investigate the occurrence and 20 30 the deposition of atmospheric constituents in the high Dale Alpine region. Monitoring aerosol concentrations v/ith the time resolution of half an hour provided, together with air Fig. 1: Epiphaniometer measurements of August 1990 at mass trajectory calculations, a means to identify air masses. the GISP 2 site. The left hand scale is in counts of the used The measurement programs at all 3 participating sites radioactive tracer, the right hand scale is an approximate (Glacier de la Girose, Jungfraujoch and Sonnblick) calibration using a correlation of epiphaniomeier and total included the determination of chemical components in the suspended paniculate matter measurements. air and in the snow. Mauna Loa is one of the four stations for monitoring Acknowledgments: global climatic changes (CMDL Climate Monitoring We are greatly indebted to various persons who helped Diagnostic Laboratory). It is considered to be free of direct operating and maintaining the epiphaniometers: Mr. and anthropogenic influence, a so called background site, and Mrs. Kuster and Biuischi at the high Alpine research for this reason ideal for monitoring global atmospheric station Jungfraujoch, the group of Dr. H. Puxbaum, parameters such as CO2 (The by now classical curve on the University of Vienna, at the Sonnblick observatory, Elmer CO2 increase in the atmosphere is from measurements at Robinson and Alan Yoshinaga, at the Mauna Loa MLO). This site is ideal to calibrate and correlate the observatory, and Jean-Luc Jaffrezo of the Carnegie Mellon epiphaniometer measurements with other relevant University, Pittsburgh, and Markus Leuenberger, atmospheric and meteorological data. University of Bern at the GISP 2 site. Another background site is the central region of Greenland (3210m a.s.l. 72.3°N 38.3'W). The American GISP (Greenland Ice Sheet Project) and its European counterpart GRIP (Greenland Ice core Project) use this site References to investigate past global climatic changes preserved in the ice. The 3000m deep GISP 2 ice core will yield a 200'000 [1] H.W. Gaggeler, U. Baltensperger, M. Emmen- year history of global climatic changes. An important egger, D.T. Jost, A. Schmidt-Ott, P. Haller, M. aspect of this project is to study todays atmosphere and its Hofmann, J. Aerosol Sci., 20(5), 557, (1989). signature in the ice record in order to be able to interpret the record. Figure 1 shows the epiphaniometer [2] U. Baltensperger, H. W. Gaggeler, D. T. Jost, measurements of aerosol concentrations at this site from M. Emmenegger, W. Nageli, Aunos. Environ., August 1990. The single sharp peaks lasting only a few in press (1991). hours are most likely due to local emissions from the drill camp. Longer lasting events have to be compared to the meteorological data and other measurements to identify the source. A first result however is, that the average concentration of about 1.0 g/nP at the GISP site is 80

DEPOSITION OF AIR POLLUTANTS AT AN ALPINE SNOW FIELD

M. Schwikowski*, U. Ballensperger*, D.T. Jost*, H.W. Gaggeler*, B. Lehmannt, M. Lehmannt, U. Siegenthalert

' Paul Scherrer Insdiut. CH - 5232 ViUigen PSI t Physikalisches Institut, Universiiat Bern, CH - 3012 Bern

ALPTRAC, a subproject of the European EUREKA Determined concentrations in the snow and air samples program EUROTRAC [1] is aimed at gaining more insight were used to calculate the mass-basis scavenging ratios into the interrelations between atmospheric and snow 1 concentrations of chemical constituents as well as of where Cs = concentration in snow (ng g ), Ca = concen- 3 natural tracers at alpine locations. tration in air (ng nv ) and pa = density of air. Similar In a coordinated program with French, Austrian and scavenging ratios as in the study on Weissfluhjoch were Swiss groups, simultaneous aerosol and snow found. Figure 1 shows the scavenging ratios for the measurements were performed on Glacier de la Girose, measured ionic components from both studies as a function 3360 m a.s.l., France, on Sonnblick, 3100 m a.s.l., Austria, of amount of precipitation during the snowfall. An inverse and on Jungfraujoch, Switzerland (3450 m a. s. 1.) from 1 correlation with snowfall amount can be seen. The values to 29 March, 1990. At our site (Jungfraujoch) we used for 210Pb agreed well with these ionic components. filler stacks (teflon filters followed by nylon and im- pregnated cellulose filters) to determine Cl", NO2",NO3', 2 + + 2+ 2+ 210 \ H SO4 ", Na , NH4 , K+, Ca , Mg , Pb, HNO3, and SO2 with a time resolution of 12 hours. In addition, we operated 20000- the following continuous aerosol instruments: a condensation particle counter, an optical panicle counter with 46 channels for particle diameters dp from 0.1 to 7 |im, an epiphaniometer which measures the Fuchs surface concentration of the aerosol particles, all with a time I oooo -i*"-. \ resolution of 30 minutes, and an aethalometer to determine the black carbon concentration (time resolution 2 hours). soco-f? Furthermore, radon daughters were determined by a- spectroscopy with a time resolution of 4 hours. Continuous measurements of SO2, NO2, O3, and H2O (by differential 20 40 60 SO 100 120 MO icO optical absorption spectroscopy) as well as of H2O2 were performed. Two additional epiphaniometers at the other Snow Fell Amount (mm Waterequivaleni) two investigation sites were operated in order to + 2y differentiate between long-range transport phenomena and Figure I Scavenging ratios ofSOf', NH4 , Ca , Na+, as local emission events. well as the sums ofHNOj and NO f, and Cl' and HCl, as a function of amount of precipitation during snowfall. Data The same ions and isotopes as mentioned above for the from Jungfraujoch as well as from Weissfluhjoch [4] are air samples were measured in the snow samples. included. Additionally, 618O was determined. More details are published elsewhere [2]. Acknowledgements Very good agreement was found between the epi- We thank R. Gehrig and B. Buchmann, EMPA Dubendorf phaniometer signal and the aerosol number concentration for the NABEL data, and the personnel of the high alpine (dp > 0.1 \vm), with a correlation coefficient of r = 0.97. research station Jungfraujoch (Mr. and Mrs. Kuster and This is an indication for an aged aerosol with a rather Bartschi) for their valuable technical support. This work constant size distribution in the accumulation mode (both was supported by the Swiss National Science Foundation. the optical particle counter and the epiphaniometer are most sensitive to particles in the accumulation mode). References Several correlated peaks of the epiphaniometer signals [1] EUROTRAC Annual Rep. 1989, Garmisch-Partenkirchen. at the three stations gave evidence for long-range transport [2] M. Schwikowski, H.W. Gaggeler, M. Gloor, R. Keil, events. Similar findings were made earlier for stations with D.T. Jost, J. Aerosol Sci. 21 (1990) in press. smaller horizontal distances [3]. [3] U. Baltensperger, H.W.Gaggelcr, D.T Jost, M. Emmen- Snow concentrations in new snow and pit samples did egger, W. NSgeli, Atmos. Environ.25A (1991) in press. not show the good agreement found in an earlier study on [4] U. Baltensperger, H.W. Gaggeler, D.T. Jost, M. Schwi- Weissfluhjoch [4]. Obviously, high winds and percolation kowski, U. Siegenthaler, A. Neftel, D. Wagenbach, K. due to unusually high temperatures had destroyed the Geis, J. Beer and W. WOlfli, in: Aerosols, Science, original stratigraphy of the snow layer. Industry, Health and Environment, S. Masuda and K. Takahashi(eds), Pergamon Press, 1990. Oxford, p.1078. 81

THERMOCHEMICAL CHARACTERIZATION OF BINARY TELLURIUM- METAL SYSTEMS

B. Eichler*, H. Rossbach*, H. Gaggelert

* Zentralinstitut fiir Kernforschung, Postfach 19,0 - 8051 Dresden, Germany t Paul Scherrer Institut, CH - 5232 Villigen PSI

Tellurium reacts with nearly all metals. So far, reaction calculated formation enthalpies for tellurides are products have been characterized, mainly on the basis of summarized in the table [1] and exhibit a well-pronounced structural investigations. Thermodynamic or thermoche- periodicity within the periodic table of elements. Most mical parameters are known for only a few combinations. stable f;llurides are found to exist with trivalent rare earth Belter knowledge of tellurium-metal interactions, as well and actinide elements, followed by the divalent metals and as their stability, is needed for an improved understanding tetravalent transition metals. of the systematics of chalcogenides. Application of such The calculated formation enthalpies correlate with the information is in the fields of the synthesis and analytics of structural parameters of the lellurides. The results allow semiconductors, for nuclear purposes (fission product estimates of the volatility or reactivity with other materials behaviour, optimum fuel cladding materials, radiochemical and also provide information on the best choice of separations by sorption on metal surfaces etc.) and also for preparative methods. modelling the gcochemical behaviour of this element. In this work, the formation enthalpies A//f(s) of solid, ordered binary tellurium compounds with transition metals and f elements are calculated using the Miedema model. References The parameter electron density, necessary for the calculation, was determined via a variation calculation [1] B. EichJer, H. Rossbach, H. Gaggeler, J. Less based on experimental data for A//f from the literature. Common Metals I£i 297 (1990). Other attempts to calculate the electron density with semi- empirical correlations were not successful. The resulting

Table: Calculated standard formation enthalpies (-AH((kJ mol-1) of tellurides

a M,_,Te,ij-<:5i>0i H Li Be B C \ O F \ Na ME Al Si P s a A 87" 105' K Cii Sc Ti V Cr Ml Fe Co Si Co Zn fJu Ge As -Se Br k 28- 1.14 148 122 59 30 56 22 2y 12 IS 611' 62' 16' Rb Sr V Zr Nb Mo Tc Ru Rh Pd Cd In Sn Sb Te I Xc 139 165 166 S3 24 24 21 4.1 69 14 50' 36' 30' O Ba La Hf Ta W Re Os lr PI Au He TI Pb Bi Pu Al Ri 144 161 147 79 11 11 15 .1(1 5 5 16 16' 35' Fr Ra Ac 16.1

Ce Pr Nd Pm Sm Eu Cd Tb D> Ho Er Tm n Lu 155 162 166 16.1 163 136:11! 166 164 163 162 161 120'11, 120'II! 139illl; 160111; 150,111) Trl P« V NP Pu Am Cm Bit Cf Es Fm Md No Lr 177 18.1 131 119 140 158 159 155 1 ?* III 1.18 128 129 120 159 ifS'Illi

(b! M,. ,Te,(i-< )66(>7) H He Li Be B C N O F Ne Na M« Al Si P s Cl Ar K Ca Se Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Sc Br Kr 107 11(1 S7 41 21 .19 15 20 22 12 Rb Sr Y Zr Nb Mo Te Ru Rb Pd H Co In Sn Sb Te 1 Xe 116 129 12.1 6(1 17 17 15 30 50 III - Cs Ba La Hf Ta YV Re O* Ir Pi Hs TI Pl> Bi Po Al Rn 125 129 HIM 57 11 22 39 12 Fr Ra Ac IM

Ce Pr Nd Pm .Sm Eu Cd Tb lit Ho Er Tm Yb Lu 117 128 m I2S 127 11211, I'll 127 127 126 124 9^11 97 11 121 123 111. 123111 136 III Th Pa I! Np Pu Am Cm Bk Cf Es Fro Md No Lr 141 136 94 K.I 1112 i:i 122 11* Mil 11 115 JO2 103 97 122 153 111

'Reference 2(1. 'Estimated in anali• B) 10 VItoibiurn 82

THE NEW NUCLIDE

J.V. Kratz*, M.K. Gober*. H.P. Zimmermann*, M. Schadel1, W. Brilchle1, H. Gaggeler*, D. Jostf, J. Kovacs*. U.W. Sclierer*. A. Weber*, K.E. Gregorich§, A. Tiirler5, B. Kadkhodayan5, K.R. Czerwinski5, N./. Hannink§, D.M. Lee§, M.J. Nurmia§, D.C. Hoffman§

Institut fiir Kernchemie, Universitat Mainz, W - 65 Mainz, Germany f Gesellschafi fur Schwerionenforschung mbH, W - 61 Dannsladt, Germany * Paul Scherrer InsUtut, CH - 5232 Villigen PSI § Lawrence Berkeley Laboratory, Berkeley, CA 94720, USA

In the course of a series of experiments investigating observed, compatible with the 23 % SF branch [2] in 259Lr. further the chemical properties [1] of element 105, The frequency at which various event types associated with hahnium, using the well-known isotope 43-s ^jHa the decay of 263Ha/259Lr were observed, is consistent with produced in the 249Bk (18O,5n) reaction, we have deve- the detection efficiency and with a SF branch in 263Ha of loped and used a very efficient aqueous phase chemical separation procedure for hahnium: Ha5+ ions were com- The kinetic energy distribution of the 18 SF events plexed in 0.05 M a-hydroxy-iso-butyric acid (oc-HIB) and suggests that the average kinetic energy is 207 McV. Both eluled in 50 jxl from 1.6 x 8 mm cation exchange columns. the value of and the apparents symmetry of the The effluent was quickly evaporated to dryness an Ta distribution are consistent with SF systcmaiics [3]. discs. Alpha and SF fragment pulse height analyses were performed on each sample using a system of ten 300 mm2 passivated ion-implanted planar silicon detectors. Typical ot-encrgy resolutions were 60 keV (FWHM). Each event was stored along with the time after start of counting and the detector identification. Start of counting was 40 s after the end of collection and the total counting time was 450 s. The sum spectrum of all a-particle spectra containing a-events with 8.3 < Ea < 8.7 MeV obtained at 99 MeV bombarding energy is shown in Fig. 1. Apart from a contamination by 249Cf sputtered from the target (this linrii material is not dissolved in the weakly acidic a-HIB solution and is washed mechanically through the column), and by a small Bi and Po activity produced by transfer Figure 1: Sum spectrum of all a-particle spectra containing events with 8.3

GASCHEMISTRY EXPERIMENTS WITH BROMIDES OF NIOBIUM, TANTALUM AND ELEMENT 105

H.W. Gaggeler*, D.T. Jost*, J. Kovacs*. U.W. Schercr*, A. Weber*, D. Vermeulen*, J.V. Kratzt, M.K. Gobert H.P. Zimmermannt, M. Schadel*, W. Briichle*, I. Zvara§, A. Tiirler', K.E. Gregorich', R.A. Henderson', K.R. Czcrwinski', B. Kadkhodayan', D.M. Lee', M. Nurmia', D.C. Hoffman'

* Paul Scherrer Institut, CH - 5232 Villigen PSI t Inslitut filr Kemchemie, Univcrsitat Mainz, W - 54 Mainz, Germany i Gescllschafl fiir Schwcrionenforschung mbH, W - 61 Darmstadt, Germany § Joint Institute for Nuclear Research, Dubna, USSR ' Lawrence Berkeley Laboratory, Berkeley, CA 94720, USA

Recently, considerable interest has been drawn to the signal the activity behind the quartz chromatography study of the chemistry of the transaclinide elements column was measured as a function of the temperature of (atomic number > 103) from both experimental and the tube. The temperature at the quartz wool plug was kept theoretical chemists. At these very high atomic numbers constant at about 900 °C. the inner electrons arc subjected to such large nuclear Fig. 1 depicts the chemical yields of bromide molecules 166 charges that they attain relativistic velocities. Relativislic formed with »gNb (T1/2 = 15 s), Ta (Tlf2 = 35 s) and 262 effects may not only alter the electronic structure of inner Ha (T1/2 = 35 s), respectively. From Fig. 1 we read that electrons, but also affect the configuration of valence the volatilities of Nb and Ta bromide arc very similar. Ha- electrons to such an extent that the chemical properties of bromide, however, has a significantly lower volatility these elements may not longer be extrapolated from their relative to those of Nb and Ta. lighter homologs in the periodic table. Recent extensive theoretical calculations [1, 2, 3] have now provided chemists with predictions on how the relativislic rearrangements of valence electrons affect the chemical properties of the heaviest elements and in which compounds these effects should become noticeable. The practical investigation of the chemical properties of the .2 iransactinide elements, however, is extremely difficult. These elements can be produced in heavy ion fusion reactions at accelerators only at a rate of a few atoms per j? hour of beam time. In addition, even the longest-lived o 0.5 known isotopes of these elements have half-lives of only one minute or less, which further complicates chemical I studies. Due to the low production rates and short half-lives jn very fast chemical procedures, which work continuously or DC 200 400 600 800 can be performed with a high repetition rate, must be Temperature C°C] devised to study the chemistry of these elements. We have developed an p.n-Hne gaschemisiry apparatus Figure 1: Relative chemical yield* (maximum yield = 1) of (OLGA) which allows to continuously separate volatile Nb-[4j, Ta- and I la-bromide molecules behind an empty, atoms or molecules of different volatility (for details see isothermal quartz chromatography column. 14]). This device was used to investigate the volatility of the group VB pentabromide molecules NbBr5, TaBr5 and HaBr5 by isothermal gaschromatography in empty quartz columns. Short-lived isotopes (T1/2 < 1 min) of those References elements were produced at the reactors SAPHIR (Nb, see Rcf. 4) and at the 88-inch cyclotron in Berkeley (Ta, Ha) in [1] V.A. Glebov etal., Radiochim. Acta4£, 117 (1989). the fusion reactions of 120 MeV 20Ne + 147Sm and 100 MeV 18O + 249Bk, respectively. Collection and transporta- |2| B.L. Zhuikov et al., Radioanal. Nucl. Chem., M3_, tion of reaction products was made with a He/KCl gas-jet 103 (1990). (4). After collecting the KC1 particles on a quartz wool p'ug the bromide molecules of Nb, Ta and Ha were formed |3] E. Johnson et al., J. Chem. Phys., in press. by adding 100 ml/min HBr gas (Nb, Ha) or 100 ml/min HBr gas saturated with BBr3 vapour (Ta). As chemical [4] Ya Nai-Qi ct al., Radiochim. Acta 47,1 (1989). 84

CHEMICAL PROPERTIES OF ELEMENT 105 IN AQUEOUS SOLUTION: FURTHER STUDIES OF THE HALIDE COMPLEX EXTRACTION INTO TRIISOCTYL AMINE

H.P. Zimmcrmann*, J.V. Kralz*, M.K. Gobcr", M. SchadcJt, W. Briichlct, E. Schimpf*, H. Gaggelci4, D. Josl*, J. Kovacs*, U.W. Schcrcr*, A. Wcbcr*, K.E. Grcgorich5, A. Tiirlcr5, B. Kadkhodayan5, K.R. Czcrwinski5, N.J. Hannink5, D.M. Lee5' M.J. Nurmia5, D.C. Hoffman5

Inslitut fur Kcrnchcmic, Univcrsitiit Main/, W - 65 Mainz, Germany t Gcscllschaft fur Schwerioncnforschung mbH, W - 61 Darmsladl, Germany * Paul Schcrrer Inslilut, CH - 5232 Villigcn PS1 § Lawrence Berkeley Laboratory, Berkeley, CA 94720, USA

Previously, we have performed anion exchange sepa- we observed 8 ^Ha decays (38 °/t), while in the "late Pa rations of 34-s f^Ha from HCI and mixed HCI/HF so- fraction", containing about 75 % of the Pa tracer activity lutions |IJ using revcrsed-phasc micro-chromaiographic we observed 13 ^i;H^i; a decays (62 <7i). These arc the results columns incorporating triisooctyl aminc (TIOA) on an inert of 228 experiments. Within the error limits, the hahnium support in the computer-controlled liquid chromatography distribution is not significantly different from the Pa apparatus, ARCA II |2]. ^fjHa was shown to be adsorbed distribution, even though one might recognize that on the column from cither 12 M HCI/0.02 M HF or 10 M hahnium clutes slightly earlier than Pa. HCI solutions like its homologs Nb and Ta, and like Pa. In The clution positions can be transformed into dis- clutions with 4 M HC1/0.02 M HF (Pa-Nb fraction), and tribution coefficients, the latter arc expressed as fraction with 6 M HNGyO.015 M HF (Ta fraction), the f^Ha contained in the organic phase in Fig. I, together with the activity was found in the Pa-Nb fraction showing thai the respective data for Zr, Hf, Nb, Ta, and Pa 11 ]. The results anionic halidc complexes arc different from those of Ta, from our previous work at 10 M HCI and 4 M HCI, and the and arc more like those of Nb and Pa. In separate elutions results of the present work at 0.5 M HCI indicate that the with 10 M HCI/0.025 M HF (Pa fraction), and 6 M distribution coefficients for hahnium follow in detail the

HNO3/0.015 M HF (stripping of Nb), the f^Ha was found ups and downs in the coefficients for Nb and Pa, and that to be equally divided between the Pa and Nb fractions. The the extraction behavior of hahnium is very different from non-tantalum like halidc compilation of clement 105 is that of Ta over a wide range of HCI concentrations. indicative of the formation of oxohalidc or hydroxohalidc In summary, the close analogy [ 1 ] of the chemical complexes, like [NbOCl4V and \PaOCl4Y or behavior of hahnium to protactinium, and to niobium, is \Pa(0ll)2Cl4\~, in contrast to the pure halidc complexes of confirmed also lor low HCI concentrations. This strength- Ta, like [TaCl6\: ens our case for the formation of oxo- or hydroxohalidc We have extended the clulion studies of the group-VB complexes by hahnium. elements from TIOA columns in ARCA II to 0.5 M _'•'••. •••• • • IOC HCI/0.01 M HF as the effluent. 34-s ]$Ha from the 24yBk (l8O,5n) reaction at 99 McV beam energy was produced at 9D T the LBL 88-inch cyclotron and transported on-line by a 7 J Hc/KCl jet to the chemistry apparatus where it was / j ( 1 i \ND \ collected on a polyethylene frit. After a collection time of £ 50 / ; \ i 60 s the frit was washed with 12 M HC1/0.01 M HF which u / / dissolved the activities, and assured chloride complcxing of i the group-VB elements. This solution passed through one Organ ] / / ;/ s j of the TIOA-Voltalcf mini columns (1.6 x 23 mm) where S 30 / 1 s' I 7 the homologs Nb and Ta, as well as the pscudo homolog , \Pa / . Pa, were extracted, while the group-JVB clement and the j Conten t actinides were clutcd from the column. 2r.H In 345 elutions with 0.5 M HC1/0.01 M HF (Pa DC1 Cl fraction) and with 4 M HCI/0.02 M HF (Nb fraction), if- HCl-Conccntiation (Mol/1) Kg Ha activity was found in the Pa fraction. Ta is not clutcd Figure I: Fraction extracted into TIOA vs lit. concen- from TIOA under these conditions. In order to determine tration. The vertical bars represent the behavior of dement the clution position of clement 105 more accurately, we 105. have cut the Pa fraction after 4.5 s into an "early Pa fraction" and a "late Pa fraction" (clution with 0.5 M References HCl/0.01 M HF for another 7.0 s). In the "early Pa II) J.V. Kraizcial.,Radiochim.Acta,48,121 (1989) fraction" containing about 25 % of the Pa tracer activity, [2] M. Schadcl ct al., Radiochim. Acia, 48, 171 (1989). 85 t:\C

ON THE VOLATILITY OF PROTACTINIUM BROMIDE

D.T. Jost*, H.W. Gaggeler*. A. Weber*, A. Kovacs", U. Scherrer*, M. Schwikowski*, W. Briichlet, M. Schadcli, G. Schimpft, J.V. Kratz*, H.P. Zimmermann*, M.K. Gobert, B. Eichler§

* Paul Scherrer Institut, CH - 5232 Villigen PSI t Gesellschaft fiir Schwerionenforschung mbH, W - 61 Darmstadt, Germany * Inslitut fiir Kemchemie, Universitat Mainz, W - 65 Mainz, Germany § Zentralinstitut fur Kcmforschung, Postfach 19,0 - 8051 Dresden, Germany

In the course of experiments to investigate the chemical properties of element 105, hahnium, it was found thai the volatility of HaBr5 is significantly lower than those of the group VB homologs TaBrs and NbBr5. It was argued thai possibly Ha behaves more like the pseudo-group V element Pa. However, detailed knowledge on the volatility of PaBr5 is lacking. We have studied the gaschromatographic behaviour of protactinium bromide in quartz columns in a similar way as U! for the group VB elements Nb, Ta and Ha |1]. The short- lived isotope 226Pa (Tj^ = 2.6 min) was used as tracer, produced in the reaction 58 MeV p on a 100 fig/cm2 232Th Lift* target. For continuous transportation of 226Pa a He/KCl gas-jet system was used. ProducLs were Lransported along a 700 m polyethylene capillary to the on-line gaschemistry Log (Time in Seconds) apparatus OLGA. In order to maintain a flow rate of about 1 1/min a pressure in the target chamber of abouc 4,5 bar was needed. 100 ml/min pure HBr or HBr saturated with Figure 1 a (left): Time correlations between a-events from BBr3 vapour was added to the position of the quartz wool 226Paand222Ac. plug in the chromatography column. Behind the column, 222 218 molecules were rcattached onto new KC1 particles and Figure 1 b (right): Between Ac and Fr. transported to the tape counting system. 226Pa was measured with PIPS detectors via its cc-dccay chain. For 100 the first time the data were stored in an event-by-event mode with the PSI-TANDEM system and analyzed with I- i the CERN-PAW software. This allowed to identify also o> 1 I a-correlations. Figures la and b show measured time 1 l I correlations between a-events from the 6.47 MeV decay 1 ! 226 10 1 line of Pa and those of the 4,2 s daughter 7.01 MeV 1 / 222 222 * I Ac and from Ac and those of the 700(xs daughter 1 7.87 MeV 218Fr, respectively. 1t Fig. 2 depicts the measured chemical yields as a / 1 ( I K/ function of the oven temperature using different reactive 1 I ! T 1 gases. Only with HBr/BBr3 reasonable maximum yields of 1 I 0• about 40 % vere observed above 300 °C. With pure HBr / / and HC1 the formation of the pcnia-halides seems to be a hindered. The experiments have to be repealed in order lo • H& f »HB

References Figure 2: Chemical yields for protactinium halides behind an isothermal empty quartz column, using different [11 H. Gaggelcr ct al., this annual report. reactive gases. 87

Material Sciences (3300)

Defect Physics (3301)

PIREX, Irradiation Damages, Fusion (3302) DEFECT PHYSICS - POSITRON PHYSICS - A PROGRAMME

Paul Scherrer Institute, CH-5232 Villigen-PSI / Switzerland W.B. Wacber

Lattice defects created by neutron irradiation [ 1 ] in Fe- New theoretical results with respect to /^SR-measure- based alloys constitute the main subject of the defect physics ments on Bi single crystals have been obtained in collabo- programme. The underlying fundamental research has been ration with PSIZ: Calculations predict a shift of the muon's continued on the neutron embrittlement phenomenon in low equilibrium position off the interstitial symmetry centres alloy steels. The thermal evolution of defects and their [3]. They allow a more precise interpretation of corre- mutual interactions are being studied theoretically as well sponding experiments. The technique used in these cluster as experimentally by means of nuclear solid-state meth- calculations is the density functional formalism. Similar ods like neutron scattering [2], muon spectroscopy [3] or methods can be applied to positrons. positron annihilation [4]. Participation in a positron beam experiment in Brookha- The application of positrons as a probe for the study ven, BNL, allowed a deeper insight into the PAES technique of solids represents the main field in our solid-state and (Figure 1). It has been shown to be an ultra selective sur- materials research activities [5]. In this respect a marked trend to ever increasing intensity [6] is to be noted in the CONVKNTIONAf. AVUIM lilliflKON use of monoenergctic slow positron microbeams. Further A.K.S. 1M,,W) technological developments of positron sources [7] and new beam concepts [8], |9] are being realized in collaboration VACUUM with several PSI research departments and with external H-RMI SUKl-ACI- researcr centres. This research is partially supported by VAI.HNCE HAND the Swiss National Science Foundation. ,*s5>:

1 Neutron Irradiation Damage A highly versatile irradiation facility [lj presently under

construction will allow investigations of the defect mor- POSITRON A.E.S. AUGiiR ELECTRON phology, for example, in ultra pure metals at low temper- (M,3VV)

atures: As a function of irradiation temperature, neutron t VACUUM energy spectrum and in situ uniaxial mechanical stress, the FERMI SURFACE processes of defect creation can be 'separated' from intermediate and !;:gh temperature thermal evolution of the de- VALENCE BAND fects, thus allow ing a deeper insight into irradiation damage mechanisms. Microhardness measurements and small angle neutron scattering experiments in collaboration with the Research ••«••- - CORE LEVEL Centre GKSS, Geesthacht, have been carried out on a series of irradiated laboratory steels characterized by systematically varied impurity contents [2]. The purpose of such Figure 1 [13]: The positrons implanted at low mergy investigations is the understanding of irradiation embrittle- diffuse to and get trapped at the surface. A few per- ment mechanisms. They are being complemented by high cent annihilate with core electrons leaving the atom in resolution TEM studies f 10] on iron-based alloys Fe(Cu.Ni) an excited state. The atom relaxes via the emission and by theoretical cluster calculations, in collaboration with of an Auger electron (first demonstrated [hi]). PAES ETH Zurich and PSIZ respectively. TEM diffraction results shows an enhanced surface sensitivity involving in large indicate some kind of super structures existing in the nm- only the top most atomic layer, because, of the trapped sized copper precipitates. positron surface states ~ 95% annihilate with the top layer in Cu(WO). Positron AES shows a vastly de- creased secondary electron background under the A uger 2 Positron Solid-State Research peak. By using low energy positrons the energy dose to the sample surface can be reduced by several orders of As a solid-state probe slow positron beams essentially magnitude as compared to EAES. Hence it is advanta- provide a means for studying bulk and surface electronic geous to use PAES to look at fragile systems that arc structure [6],[11). Our goal is the physics of surface states damaged or desorbed by keV electrons. Future research (positronium formation [12], Positron annihilation induced may include studies of ultrathin film growth, studies of Auger Electron Spectroscopy, PAES [14]), but also the segregation at alloy surfaces, high resolution studies of physics of positrons in perturbed crystal lattices. In both A ugcr from top layer, spin polarized PA ES or the mea- cases theoretical considerations have lo complement exper- surement of ionization cross sections. iments. 89

face technique (topmost atomic layer) [14]. Measurements The development of 58Co-sourccs [7] combined with on MgO and GaAs single crystals qualitatively showed the solid rare gas moderators to be used in planned positron applicability of this technique also to semiconductors and laboratory-based beams [5| has been started in collabora- insulators [15]. The preparations of planned beam experi- tion with the Department 'Physique dc la Maticre Con- ments at PSI have been started. Plans are being set up for a densce' of the University of Geneva. higher resolution PAES spectrometer. The purpose of this First experiments for testing the efficiency of repolariza- technique is aimed at investigations of the specific surface tion of slow positron beams interacting with spinpolarizcd, electronic structure of metals and semiconductors as well atomic hydrogen it low temperatures and high magnetic as of surface magnetism (spinpolarized PAES). fields [9] are being planned in collaboration with the Kur- chatov Institute for Atomic Energy, Moscow. 3 Positron Materials Research

The characterisation of defect structures at surfaces References and interfaces of semiconductor devices is still traversing a phase of increasing significance. Defect analysis in such [1] M. Caro, this volume. environments is as important as are the crystal structure or [2] G. Solt, F., Frisius, W.B. Wacbcr, this volume. the chemical analysis, which can be determined by conventional techniques. Monocnergctic slow positron beams [3| G. Solt, E. Lippelt, B. Dellcy, this volume. represent the most adequate defect probes to be used in such inves jgations. |4J U. Zimmcrmann cl.al., this volume. Investigations of a similar nature have been performed [5] W.B. Wacbcr, this volume. on Cu-precipitates in irradiated Fc(Cu) by means of positron lifetime and ACAR measurements, partially in collab- (6| W.B. Wacbcr ed., Helvetica Phy.sica Ada, 63 (1990) oration with the Technical University of Helsinki [4|. Life- 377-470. time measurements have essentially confirmed the ACAR results. In one specific case voids have been detected that |7] U. Zimmcrmann ct.al., this volume. have not been resolved in the ACAR measurements. [8] W.B. Wacbcr D. Taqqu, U. Zimmermann, G. Solt, Positron microscopy is an encouraging nc r technique 'Proposal for an Intense Slow Positron Beam Facility which promises new possibilities in the field of meaterials at PS!', PSl-Report No.68, May 1990. research, especially in defect physics. A proposal for such an experiment has been formulated [81. However, micro- [9] D. Taqqu, this volume. scopic imaging with reasonable 'exposure times' asks for highest intensity monoencrgetic positron beams. [10] H.K. Grimmer, this volume.

Ill] W. Brandt and A. Dupasquier, Eds., '1 positroni nclla 4 Positron Beam Development fisica dci solidi', North Holland (1983).

The proceedings of the international workshop at PSI (12) M.H. Weber, S. Tang, S. Berko, B.L. Brown, K.F. on 'Intense Slow Positron Beams' in November 1989 have Canter, K.G. Lynn, A.P. Mills, Jr., L.O. Roc'!;g and appeared in Helvetica Physica Acta [6]. They reflect the A.J. Viescas, Phys. Rev. Lett., 61 (1988) 2542. preseni state of positron beam development and many new ideas and applications. In a proposal |8J it has been shown 113] A.H. Weiss, privat communication. how at PSI a very high intensity positron factory could be installed and be operated as a large-scale user facility in [14] A.H. Weiss, R. Mayer, M. Jibaly, C. Ley, D. Men! the domains of solid-state, materials and atomic physics and K.G. Lynn, Phys. Rev. Lett., 61 (1988) 2245. research. The proposed project has been incorporated in [15] Collaboration with A. Weiss, A. Schwab and Y. Kom. the recent mid-term planning document of PSI [16]. 116] 'Miltclfristigc Planung 1992 - 1995', Report PSl-Dir- No 1, November 1990. 90

PRELIMINARY NEUTRON SPECTRUM TAILORING AND DPA CALCULATIONS FOR IRRADIATION EXPERIMENTS IN THE POOL REACTOR SAPHIR.

Paul Scherrer Institute, 5232 Villigcn-PSI / Switzerland

M. Caro, W.B. Waeber

100 Neutron irradiation experiments are planned at PS I to study radiation damage in a fundamental way, under the scope of the Defect Physiscs research programme. Pure iron samples will be irradiated at the pool research reactor SAPHIR using a new irradiation facility SPIRTS (SPec- wm > 3 Mev •• 3 MeV - 1 MeV trum, IRradiation Temperature, Stress) [1], which allows CD 1 MeV- 0.1 MeV C3 0.1 MuV -4 eV variable temperature, neutron environment and in situ strain- •• < 4 ev ing. The SPIRTS irradiation facility is designed to support three specimens in a cylindric container. The samples arc placed into operationally independent irradiation blocks at three different height levels and radial positions of the SAPHIR reactor. Each specimen is surrounded with a different material which acts as a tailoring clement (TE) affecting the neutron flux, DPA and DPA-ratc spectrum behaviour in the sample. Thus, the neutron environment changes with the selection of diflcrcnt TE. Several neutron transport calculations have been carried out using different materials to select the best candidates. In all these calculations, the SAPHIR reactor was supposed to be critical with control rods half inserted and operating at a maximum power of 10 MW. The 2D code TWODANT [2], was used to model the Loading 555 of SAPHIR including fuel elements, con rol rods, BeO reflector, moderator, sample and target TE, see Tungsten Aluminium Water Fig. 1. The reactor core was simulated with an isotropic space and energy dependent neutron source. The spatial Figure 2: Comparison of the DPA-rale induced in the source is defined using a spatial power distribution obtained sample using three different tailoring elements. Values frjm a previous detailed calculation [3|. The energy de- are prcscnteil in percent for live main groups of energy. pendent source corresponds to a U235 fission spectrum. TRAMIX [4] was used to derive the 165 neutron group P5 cross-section library from the basic MAT175 data file |5] for these calculations. MATI75 is a JEF-1 based multi- group cross-section library in the VITAMIN-J structure for fast reactor and fusion related calculations. It contains ra- 1 2 3 4 5 6 / a y 10 11 diation displacement energy cross sections which can be 1 11 Be Be Be Be Be Be B.3 easily convened to displacement cross sections [6|. 2 BeO BeO Neutron flux and DPA spectrum were evaluated in the 2 3 BeO BeO middle of a 0.7 x 1.0 cm pure iron sample , placed at position 6-11 of SAPHIR reactor and surrounded by TE 4 r BeO w, material in positions 4-10 to 8-10 and 5-11 to 7-11. The 5 BeO results of this first parametric study show that distinct DPA- 6 BeO W m rate spectrum using tungsten, aluminium and water as TE, 7 BeO can be obtained, sec Fig. 2. Using tungsten, 61% of the total DPA-ratc in the sample comes from the contribution 8 of neutrons in the energy range between 0.1 and I McV. 99 9 91 In the case of aluminium, the contribution in percent of the group corresponding to Lhe energy range between 1 and 3 McV amounts to 44% of the total DPA-rate. DPA rales of Figure 1: Loading 5.W of SAPHIR reactor with tlio 36% at higher energies arc obtained using water as TE. As iron sitniplr at position (i-ll. The hutched region cor- shown for the case of aluminium in Fig. 3, the DPA-ratc responds to the tailoring element (TK) material. spectrum per unit of lethargy can be obtained with finer details since the group structure used contains 105 groups 91

2.0 10' As concluding remarks it can be said that a first step in the selection of TE has been done. Further studies using a combination of TE materials and/or varying the irradialion position and height should proceed till an ideal optimization is achieved. A fol) optimization procedure will also require the interaction with the design of the irradialion blocks and TE. This in turn, will result in a new geometry which will give rise lo a final spectrum through an iteration process. The author gratefully acknowledges D. Mathews and S. Pclloni for numerous fruitful discussions and critical com- ments during this work.

References

(1) W. B. Wacbcr, M. Caro, j. Sicpanck, "Irradiation Fa- cility with Variable Temperature, Neutron Environment and In Situ Straining", [G-RDM Meeting, 7-11 November, Helsinki, (1989) and Progress Report, An- nexe III, (1989), PS1.

[2| R. E. Alcouffc, F. W. Brjnkley, Jr., D. R. Marr and R. D. O'Dell, "User's Guide for TWODANT: A Code Package for Two-Dimensional Diffusion-Accelerated, Neutral-Particle Transport", Los Alamos National Lab- oratory, LA - 10049 - M, Revision 1, (1984). Figure 3: Dl'A spectrum -HT unit of lethargy obtained in the sample :it position (> ]/ of SAI'HIH reactor he- [3] M. Caro, S. Pclloni, "Nuclconic Calculations for Pos- hind an a/uuiiniiiiii TK. sible Irradiation Experiments in SAPHIR", PSI-Rcport Nr. 55, (Januar 1990). with energies above 0.1 Mcv. [4] J. Stcpanck, R. E. MacFarlanc, "TRAMIX: A Code Finally, total DPA values at high energy (E > 0.1 McV) :l for Interfacing MATXS Cross -section Libraries to Nu- of 1 • 10~\ 3 10"'-' and 7 • 10~ arc obtained for the clear Transport Codes for all Type Fission as Well as specimens surrounded by tungsten, aluminium and water Fusion Reactor System Analysis", PSI-Rcport, (to be respectively, after six montiis of irradiation under a fast published). llux (E > 0.1 McV) which varies depending on the case: 12 2 12 2.6 • 10 n/cm s for tungsten, 4.0 10 for aluminium [5] P. Vomobcl, S. Pclloni, "JEF/F.FF Based Nuclear Data 2 and 5.4 • 10" n/cm s for water. These values may vary Libraries", EIR Report 63f>, (December 1987). within loading configurations of SAPH1R and control rod positions. |61 "Annual Book of ASTM Standards", Vol. 12.02, (September 1984,1. 92

PHASE SEPARATION OF COPPER IN IRON-BASED ALLOYS UNDER IRRADIATION: A SMALL ANGLE NEUTRON SCATTERING STUDY OF MODEL STEELS

G. Solf, F. Frisiust, W.B. Waeber"

* Paul Scherrcr Institute, CH-5232 Villigcn-PSI t Research Centre GKSS, D-2054 Gecsthacht, Germany

Irradiation generated (or enhanced) phase separation in nificantly narrower than the one deduced from the scattering metastable solid solutions is a rapidly developing field of of purely nuclear origin. both experimental and theoretical research [1]. The case of precipitation of copper from its oversaturated solution in iron for T < 300C under irradiation by fast neutrons [2,3,4] A 5*10" has, moreover, a technical aspect, since this phenomenon is at the origin of the embrittlcmcnt of copper containing ^ o ;oo steels used in nuclear technology [5]. E While the phase separation process is relatively well understood for pure Fe(Cu) and promising results have been obtained also for Fc(Cu.Ni) alloys, available microphys- ical studies on different industrial steels refer to materials with widely diverse composition and thermal history, which makes statements on the role of different parameters rather difficult.To obtain basic information on materials of such 'technical' complexity, one obviously needs a systematic variation of some selected composition parameters on the fixed 'background' of the relevant mulu'alloy system. With 0 01 0.10 l.OC this aim, representative 'laboratory steels' with varying con- momentum tronsfer K. [A~'] tent of Cu, P and Ni have recently been manufactured and made available by the IAEA to interested research groups, Figure 1: Differential cross sections due to the irra- for neutron irradiation and subsequent physical investiga- diation damage in sample JPA (see text). The aver- tions. The concentration variables i, (i=CuJ>Ii>P) are cho- age radius of *he damage centres is ft — 11.6 ± 0.1.4, sen, since nickel is known to facilitate separation of the the data are fitted by assuming logarithmic-normal size copper-rich phase [5], forming presumably an interface be- distributions as shown in the insert. Nuclear (n) and tween the iron solvent and the nucleated copper-rich parti- magnetic+nuclear ((m)+(n)) cror,s sections are sepa- cles [3], and radiation embrittlement has proved to be very rately measured by two detectors. Thf deduced size sensitive to the phosphorus content. distributions for 'magnetic' and 'nuclear' defects have A series of eight 'model' steels, JPA,...H, were ex- different widths, SR/fl = 0.22 and 0.27 ± 0.005, re- posed to fast neutron irradiation at the STILO facility of spectively; the density of the magnetic 'holes' is A' = the SAPHIR reactor in the PSI, Wurenlingen, for two flu- (8.35 ± 0.2) -1017/cm3. ences 5 • 1018 and 5 • XO^n/cm2. First, preliminary results on the irradiation defect structure are reported here, This feature, observed for other compositions as well obtained by the analysis of small angle neutron scattering but to a different extent, is an indication that magnetization (SANS) data. The experiment was carried out at the new penetrates some part of the nucleated centres which become cold source facility of the GKSS, Geesthachl. In Fig.l the thereby magnetically more or less 'invisible'; this tendency 'additional' scatifting intensity after irradiation, i.e. the is seen to depend sensitively on variation of jrjvt and xp. A cross section due to the damage centres, is plotted against detailed account of the results will be published elsewhere. the neutron momentum transfer K = {Air j \)sin(0 /2) for the high-copper 'model' steel JPA (xCu =0.33 wt%, xNi = 0.82 wt%,xP =0.018 wt% + 'fixed' components) for the References fiuencc 5 • 1019n/cm2; here A ~ 5.25/1 is the average neutron wavelength and 0 is the angle of scattering. The fairly [1] R.A. Johnson and A.N. Orlov, Physics of Radiation good resolution, much better than attained earlier in similar Effects in Crystals, North-Holland 1986 experiments [2,6], has been made possible by the higher intensity at the new facility. [2] F. Frisius, R. Kampmann, P.A. Beaven and R. Wagner, The insert shows that the size distribution of the radia- Proc. BNES Conf, "Dimensional Stability and Mechan- tion induced 'magnetic' holes, acting as scattering centres ical Behaviour of Irradiated Metals and Alloys", vol.1., in the otherwise homogeneously magnetized sample, is sig- p. 171, BNES London !984, p. 169. 93

[3] J.T. Buswell, C.A. English, M.G. Hetherington, WJ. Pylhian, G.D.W. Smith and G.M. Worrall, Effect of Radiation on Materials, 14th Int. Symposium, vol.11., eds. N.H. Packan, R.E. Stoller, and A.S. Kumar, p.127, ASTM Philadelphia, 1990 [4] G.R. Odette and G.E. Lucas, ibid, p.323. [5] J.R. Hawthorne, Effects of Radiation on Materials: 11th Conference, ASTM STP 782, ASTM Philadelphia 1982 [6] G. Soil, F. Frisius, W.B. Waeber and W. Biihrer, Effects of Radiation on Materials, 14th Int. Symposium, vol.11., eds. N.H. Packan, R.E. Stoller, and A.S. Kumar, p. 154, ASTM Philadelphia, 1990 94

ENERGETICS OF INTERSTITIAL MUONS IN BISMUTH; A STUDY OF Bi i2(H) CLUSTERS

G. Solf, E. Lippelt', and B. Delley*

* Paul Schcrrer Institute, CH-5232 Vi)ligcn-PSI f Institut fiir Mittclcnergiephysik dcr ETH Zurich, c/o PSI CH-5^32 Villigcn t Paul Scherrcr Instilut Zurich, Badenerstrasse 569, CH-8048 Zinch

To interpret results cf muon jpin rotation QiSR) experiments, the location of" the implanted ;x+ probe in the lattice is of primary importance. Recent calculations on small Bis(H) clusters [1] predicted a static position for ihe hydrogen-like //+ particle lying outside of the centre of the Bis frame, an indication that the positions of energy a. -77.0-^ minimum for muons, stopped in crystalline bismuth at low (T1 < 15A) temperature, may also be off-centre in the interstitial cage. In fact, the usual assumption of a muon occupying an interstitial symmetry centre can be brought into agreement with the data for bismuth only by invoking an unprecedented, strong local lattice contraction, whereas site 2 by admitting off-centre muon positions a substantially better description of the data is possible both for '[' — 11 A' and between 85/\- I00A' [1,3,6]. The theoretical argument based on our quoted results for Bis(H) has been, however, -2-10 I 2 necessarily a qualitative one, since this cluster is too small to reproduce a crystal like electronic structure and the forces displacement of y.' along c-axls [a.u.] experienced by the bismuth atoms in this unit refer more to Figure 1: Total energy ofB'3->(li) for unrelaxed frames a 'surface' than to a bulk environment. modeling interstitial sites 1 anil 2 m the bismuth crystal, In view of the unusual physical situation implied by as a function of the muou displacement along the c-nxis off-centre muon positions and the success of this hypothe- (symmetry centre: z,L = 0) [6]. sis in explaining the experiment, the calculations have now been extended to the much larger systems Bi32(H), obtained from Bi (H) by inclusion c>f the six neighbouring interstitial 8 H-like particle, extending up to rM ~ ±1.2/1 about the cages having a common face with (he central B'ig unii. The symmetry contre, which is to be compared with the coordi- electronic structure of sucH a cluster is certainly^ better ap- nates zi - ±3.096/1 of the two neighbouring Bi atoms situ- proximation to that of the bulk and, in addition, by fixing ated on the same axis. One expects that letting the bismuth the position of the 24 surface atoms, a first model for the atoms to relax will strongly modify this quasi-degenerate relaxation of the inner atomic shells is obtained. As in a pre- energetics, and this can be seen indeed in Fig.2, where vious work [1], the electronic energy of Bin(H), modeling Etot is shown for two different atomic arrangements. The either of the two kinds of interstitial sites in the rhombo- first of these, 'fixed equ.', refers to an atomic configuration, hedral bismuth lattice, was calculated by the local density kept unchanged during the variation of the muon displace- functional algorithm 'Dmol' [4]; no dynamics was consid- ment, which corresponds to the static equilibrium of the ered, thus the calculated electronic structures hold equally inner eight atoms within the empty cluster, while the outer for clusters with the ;J+ particle or with its heavier isotope 24 atoms are still fixed at their crystal positions. Unlike H+: from the point of view of static equilibrium positions for an infinitely large cluster, this 'pre-relaxed', equilib- Bin(H) and Bin(/i+) are equivalent. The total energy 'hy- rium structure differs already from crystalline order; in the persurface' £1<,t(r/1,r<) has been analyzed by varying the cluster calculation it is the obvious starting configuration muon position vector r^ in fixed bismuth frames and also for a study of the impurity generated relaxation. The 're- by allowing the atomic positions r\ to relax towards static laxed' structure in Fig.2 corresponds to the energetically equilibrium. optima] position of the inner eight atoms in the presence of the muon^ with atomic coordinates adjusting themselves The sutface. E,o, as a function, of the nuian displacement corresponding to each value of the muon position zh. zM along the C3v axis (crystal c-axis) is shown for the two unrelaxed frames, site 1 and 2, in Fig.l; all bismuth atoms One sees that both the prC-relaxed and relaxed atomic in the Bi32 frame are fixed according to the unperturbed configurations generate side minima at the muon potential crystal structure at T = 78A' [5]. at about z^ = ± 1.2/1 in the case of site 2, and also for site 1 Cage 2, obiong in the ±z directions, provides a par- a slight dip is found at ;„ ~ ±0.55/1. Interestingly, results ticularly wide and flat 'square well' like potential for the for the small Bis(H) cluster [1] have predicted potential 95

. , . . 1 . . . . 1 . . . . 1 . — 79 6- In conclusion, the calculated potential surfaces for a 1 i Site \2 H-like impurity in Bi3;i(H) clusters support the prediction, based on a previous study of the BiB(H) system, that the •—, (ixod ; impurity is not localized at the centre of cither of the two . 8qu. i interstitial cages of the bismuth lattice. The predicted off- 1 centre positions at distances of ~ I ;iA along the c-axis _5 13 for site 2 occur for the relaxed cluster; the possibility of a similar but much weaker trapping about half so far from

\\ // site 1 is also indicated. The role of lattice rearrangement ? around the H+ particle is important in forming the trap, but 5 -80 0- the static equilibria arc different for cluster and crystal and, 5 relaxed 3 therefore, a direct extrapolation of the relaxed cordinatcs to the lattice is not possible. The analysis of [/.Sit daia at '/' < 15A' [1,3,61 give strong support to the predicted muon displacements and gives the best agreement for occupied — 80 2 ' • ' 1 • ' ' • 1 • ' • *- ' • ' ' 1 • ' • ' 1 ' ' ' ' 1 ' off-centre positions at : = ±().

Figure 2: 7bf a/ energy of lSi;i2(ii) a.s in /-Vg-. i but for two different, atomic configurations [6J. The dashed line is References for the structure obtained by relaxation to static equi- [1] G. Solt, E. Lippelt and B. Dcllcy, Hypcrlinc Interac- librium of the inner eight atoms in the empty /Ji'32 clus- tions 1990 ter (a structure kept fixed for aft z^); the solid line is for the relaxed configuration of the inner eight utorus as [2] F.N. Gygax, B. Hitti, E. Lippclt, A. Schcnck and S. optimized for each muon position zfi. Barth, Z.Phys.B 71, 473,

[3] E. Lippclt, thesis ETH 9249, ETH Zurich, and E. Lip- traps at :„ = ± 1.1514 for site 2 and shallower ones at zM = pell et al., to appear ±Q.8°A for site 1, but the traps found here, for the large cluster, occur only al some relaxed atomic configurations [4] B. Dcllcy, J. Chem. Phys. 92 508, 1990 and arc levelled out for the ideal crystal. [5] R.W.G. Wyckoff, Crystal Structures, Intcrscience, New York I960, vol.1, p.32

|6] G. Solt, E. Lippelt and B. Dclley, to appear 96

POSITRON LIFETIME MEASUREMENTS ON NEUTRON IRRADIATED IRON AND IRON-COPPER ALLOYS

U. Zimmermann", W.B. Wacbcr", K. Saarincn', P. Hauiojiirvi'

* Paul Scherrcr Institute, CH-5232 Villigen-PSI t Laboratory of Physics, Helsinki University of Technology, SF-02150 Espoo 15, Finland

The behaviour of positrons in Fe-Cu alloys after different heat treatments and neutron irradiations as well as post irradiation annealing has been described in f I] and (2J. The

results could be explained assuming thermally induced and 1.0- irradiation enhanced precipitation of Cu-panicles in a su- LIFETTIME SPECTRA . r persaturated solid solution of copper in iron. These results "c •!"o ° f e-0.7%Cu inhom. »B°OO • Fe onneoied were obtained by means of the 27-angular correlation tech- o »• nique (ACAR). Since the positron lifetime technique (I_T) o CD is known 1.0 reveal more detailed information on the nature _O 06- •'•••X of the positron trapping sites, this teennique was applied to a ••••.•. °*= the same samples as in the ACAR experiments. E 5 The LT-technique has been applied successfully to de- o °4' 3 ""Si*!, fects in iron and other metals at the Helsinki University

of Technology (HUT) and the LT-spectrometers developed •: :o JO «0 bO 60 ^o 3D 90 100 and operated at HUT were used for our experiments. channels Two questions were of special interest:

• Are the positrons directly trapped by the Cu-particles Figure 1: J?aiv lifetime spectra of pure iron (full dots) due lo the positron affinity to Cu, which could be and inhomogencous Fc 0.7% Cu (open nrclex) contain- expected from the positron work functions of iron ing Cu as particles of 12 16 nrn diameter and copper? In this case, the corresponding lifetime should be that of pure Cu (120 ps). Or is the interface lifetime technique and r, > T), indicates additional trap- responsible for trapping due to "open volume" defects uik ping sites. at the surface of the precipitates? In this case, the Irradiated samples corresponding lifetime should be larger, i.e. ~ 170 The occurence of lifetime components with lifetimes ps or more, as for vacancies or dislocations in iron larger than ~ 200 ps clearly indicates small vacancy clus- 13) and copper [4J. ters. The limit for annihilation of free positrons at a metal surface is 500 ps. The existence of large voids is evident • Are the lifetimes in the irradiated samples compatible with those of voids? In this case, lifetimes of 200 to especially in the samples with the higher neutron dese. Va- 500 ps should occur in the LT spectra [51. cancy clusters (~ 6 vacancies) arc already clearly detected in the homogeneous Fe-Cu sample at the lower neutron dose. These clusters were not found in the earlier ACAR Unirradiated samples: experiments. The observed lifetime of 115 ps in pure iron (table 1) The short lifetime components in the samples irradiated agrees well with the expected value. The in homogeneous to the higher neutron dose were difficult to obtain and arc Fe-Cu sample showed a second lifetime component of 187 not very reliable. These LT-spectra were strongly disturbed ps which is close to that for dislocations and vacancies in by the background radiation which was due to the ""Co iron and copper but far from that of 115 ps for the defect- content of these samples (60Co has a -> -cascade which leads free metals. It is therefore concluded that this result is to a prompt component in the LT-spcctra). consistant with our earlier interpretation that the positrons Conclusions can be trapped in "misfit-dislocations" at the interfaces be- The LT-measurcmenis, to a large extent, confirmed our tween the copper particles (fee) and the iron matrix (bcc). interpretation of earlier ACAR measurements. It also show- It should however be mentioned that the decompositions ed that LT gives more reliable information about the na- into various LT-componcnls (Table 1) are not consistant ture of the defects. It can also be expected that further with the trapping model. This is already indicated by the LT-experiments including post-irradiation annealing would values of r,, which are in many cases higher than the bulk give more information about the still unresolved trapping lifetime. The trapping model predicts Tbull! > TU (l/r(,,lll sites indicated in the above measurement. = li/n + I2/72 + •••)• A decomposition of the spectra into more conponcnts than indicated in Table 1 was possible in some cases but only with poor convergence. It is to be References expected that the first component in reality consists of two components which can not be resolved with the positron [I] K. Ghazi-Wakili, U. Zimmermann, J. Brunncr, P. Tip- 97

n(ps) n>(ps) ^(ps) !,(%) I:t(%) unirradiated: pure Fe 115±1 100 Fc-Cu horn. I26±I 253 ±7 83±2 17±2 Fe-Cu inhom. 122±8 187±5 41±9 59±9 0.5 K J0l;ln/cm-: pure Fe 93±2 237 ±4 535 ±20 39±2 57±1 4±1 Fe-Cu hom. 157 ±4 350±30 610±120 58±5 37 ±3 5 ±4 Fc-Cu inhom. 144±5 290±30 560±60 5S±4 35±3 7±3 5 x 101'1 n/cm': pure Fc ~ 22 208±8 537±17 ~28 53±1 19±2 Fc-Cu hom. -80 211±1! 546±12 ~ 19 54±3 27 ±2 Fc--Cu inhom. ~ 70 205 ±7 532±17 ~~ 24 62±2 Mil

Table 1: Lifetime components in pure iror. ami Fc 0.7S? C'N before mul after neutron irradiation. Specimen* (for details sec [2]): pure Fc: nnnenled al 750 °C. Fe 0.7% Cu liom.: quenclied from solid solution (8-10 "C) Fe O.T'X f.'u inhom.: annealed 120 h / 500 "(' after quenching. Irradiation (SAPHlIt): T = 290 "<", flux = 2.3 x 10r' fast n/cirrs (E > I MeV)

ping, W.B. Waeber and F. Heinrich, phys. slat. sol. (a) [4] P. Hautojarvi, P. Jauho, Acta Polytcchnica Scand. Ph 102. 153 (1987). 98 (1973). [2] K. Ghazi-Wakili, Ph. Tipping, U. Zimmermann and |5] M.J. Puska. phys. stat. sol. (a) J02, 11 (1987). W.B. Wacbcr, Z. Phys. B, Condensed Matter 79, 39 (1990). [31 A. Vehanen, P. Hautojarvi, J. Johansson, J. Yli- Kauppila and P. Moscr, Phys. Rev. B 25, 762 (1982). 98

LAB-SCALE POSITRON BEAM PRODUCTION AT PSI

Paul Scherrer Institute,, CH-5232 Villigcn-PSI/Swiuerland

W.B. Wacbcr, U. Zimmcrmann

Besides the general use of positron beams in fundamen- B magn. Feld tal research [1,2], preferentially in the domains of atomic physics or on typical positron interaction processes wilJi solids in the bulk and at the surface (electronic structure, defects), positron beams also represent an excellent probe for studying materials properties (defect probe, microscopy, M T diffraction, spectroscopy, clc)|3]. Theoretical and experimental aspects and bask details of tunable, monoencr- gctic slow positron beam production have been well established for many years [4,5,6]. The type of beams discussed here and planned at PSI arc high flux beams (>Uf c+/s), isolopc-produccd. general purpose slow positron beams, where the slow positron production technique is appropriate to UHV conditions [7,H{, In contrast, the proposed al- tra high intensity large-scale positron beam factory (>l()l(l c+/s) [9] is based on a conceptually different slow positron production scheme (see also [10|. Manip. The basic elements needed to perform slow positron beam experiments arc

• The beam production and handling system, including the primary source of high-energy positrons, the moderator which Compresses the source spectrum into a narrow energy band at cV energies and the M hentn transport system which may consist of extraction fillers, rcmodcriuion stages, beam forming and transporting optics, uccclcralion sections, etc, trans- fcring the moderated positrons to the target and giving Lead schield the beam the final space, time and energy characteristics. Figure 1: Central .source clmniher (schematic) includ- • TTie detection system and data processing including ing J soiirce/motlerxlor positions X furnishing differ- the specimen chamber equipped with positron and/or ent stow positron beams. Lab A: magnetically guided gamma detectors, on line computing system, etc. general purpose hean\ line, Lab B: electrostatic heam line, Lab C: Posit mniuni production beam line. I — ion • The sevices including pumping stations at least at pumping system for VIIV conditions, R = refrigerator the source end and the target end of the beam line facility (liquid He temperatures), G = %as inlet system producing good vacuum conditions depending on the with mass spectr,imetry for solid rare gas moilerators, type of moderator or the type of experiments per- T = ''*('o source transfer and transport flasc with ma- formed (UHV for example surface physics investiga- nipulator M, L = km) shielding, B = magnetic field tion), various conventional surface analytical or con- solenoids, S = siwplies. ditioning instruments for surface characterization, etc.

Space, time and energy characterization of a slow posi- d) the beam's lime structure. If the beam is produced by tron beam may be summarized into the following param- radioactive sources it is continuous in time and it is eters which also determine the limils and feasibility of a spin polarized due to the natural helicity of positrons given class of beam experiments [6,4]: (sec however the unconventional scheme proposed in [9,10]). There is also the possibility of beam bunch- a) The beam intensity at the target and ing to get a pulsed beam structure (sec e.g. [1,11,12]).

b) the beam geometry with cross-section and angular Depending on the type of beam experiments, two major divergence. The combination of a and b leads to the classes of beam experiments can be distinguished brightness of the beam. 1) experiments with magnetic beams. Magnetic sole- c) The beam energy and its energy width at the target, noids are used to transport positrons from the mod- and finally erator to the target. 99

2) electrostatic beams arc used for angle resolved stud- 1991: Lab A, general purpose beam line concept-engineering ies of scattered positrons or positron induced sec- and design, start of construction of the central source ondary electrons. chamber (middle part of Figure 1). Development and production of 1 Ci-"*Co sources' 116). It is planned to Hence positron recmission microscopes or positron annihi- slan designing a PAES spectrometer with improved lation microprobes as well as LEPD (Low Energy Positron energy resolution |17]. Diffraction) experiments 113] or experiments where c+ polarization is important [14,10], call for electrostatically fo- 2nd step: Lab B, electrostatic beam line (1H|. cusscd beams. 3rd step: Lab C, Ps beam line [I5|. The main purpose of the planned laboratory-scale positron beam facility at PSI is to provide slow positrons mainly The mentioned realization phases will heavily depend on for carrying out experiments with solid surfaces. With view future activities with view lo the very high intensity positron to a possible very high intensity positron factory it would factory at PSI. also serve as a means to prepare new types of experiments or to the development of components for beam formation, like the development of advanced brightness enhancement References stages, etc. Such a laboratory beam facility should be de- |1] A.P. Mills jr., in I positoni nella lisica dei solidi, W. signed according lo the following practical requirements: Brandt and A. Dupasquicr, cdts., North-Holland, Am- • It should be based on state of the art technology, thus sterdam, (1983). p. 432. saving time for design and construction. |2] 8th Int. Conf. on Positron Annihilation, L. Dorikens- Vanprael, M. Dorikcns and D. Segers, ails.. World • For more than one beam line a combination of the Scientific, Singapore, (1989). source chamber into one central source-moderator assembly (one active /.one only) may be advantageous, [3] W.B. Wacbcr. cdt.. Helvetica Physica AcUl 63, (1990). whence saving additional UHV pumping or refrigera- 377. tor systems, although independence of different beam [4] K.F. Canter and A.P. Mills jr.. Can. J. Phys. 60. lines must be guaranteed. (1982), 551. • The complete facility should be designed for step by |5] A.P. Mills jr., in Positron Scattering in Oases, J.W. step realization in terms of modules. Humbcrslon and M.R.C. McDowell, edls.. NATO AS1 • The design of specimen chambers for surface studies Scries B:107, Plenum Press, New York, (1984), p. should be done in dose collaboration with surface 121. physics croups. |6| A. Dupusquicr and A. Zecca, Kiv. Nuovo Cimcnto, 8, (1985), 1. Figure 1 shows the planned central source/moderator facility leading to 3 beam lines: [7] M. Peter and A.A. Manuel, in (3| p. 45.S.

A) General purpose beam line, magnetically guided for [8] P.J. Schulty. and K.G. Lynn. Rev. Mod. Phys.. 60. experiments where all scattered charged particles (1988). 701. should be confined like PAES (Positron annihilation |9] W.B. Wacbcr, D. Taqqu, U. Ziminermann and G. Soil, stimulated Auger Electron Spcctroscopy) or exper- cdLS., PSI-Bcricht Nr.68, (1990). iments like defect profiling or ACAR experiments, etc. [10] D. Taqqu, this volume.

B) Electrostatic beam line for diffraction and scattering Ill] W. TrifLshauscr, O. Kogcl. K. Schreckenbach and B. at surfaces, in experiments like positron microscopy, Krusche in |3] p. 378. LEPD, or positron rcpolarization, etc. |12| A.P. Mills jr., E.D. Shaw, R.J. Chichcsier and D.M. C) A beam line for positronium formation in order to Zuckcrman! Rev. Sci. Instnim.. 60. (1984), 825. use a Ps beam to study clean surfaces or adsorbants [13| IJ.Rosenberg. A.M. Weiss, and K.F.Canter. Phys. Rev. on surfaces [15|. Letters 44, (19K0), 1 139. A central workstation with terminals would allow for the (141 J. Van House and P.W. Zit/owitz, Phys. Rev. A29, beam line surveillance and controlling, the data acquisition (1984), 96. and the data processing. The planning, the concept-engineering, the design and [151 M.H. Weber, S. Tang, S. Bcrko, B.L. Brown, K.F. Canter, K.G. Lynn, A.P. Mills jr., L.O. Rocllig and construction of the presently discussed laboratory beam fa- A.J. Vicscas., Phys. Rev. Letters, 61, (1988), 2542. cility will be realized in several steps: 116] U. Zimmcrmann ct al., this volume. 117] R. Mayer, D. Becker, A. Schwab and A. Weiss, Rev. Sci. Insiruni. 61, (1990), 42. [18] K.F. Canter, in Positron Studies of Solids, Sufaccs, and Atoms, A.P. Mills jr., W.S. Crane and K.F. Can- ter, cdts., World Scientific, Singapore, (1986), p. 102. 100

DEVELOPMENT OF 1 Ci- 8Co POSITRON SOURCES

U. Zimmermann", R. Schwar/.bach", 1. Huszar", W.B. Waeber", R. Wcinrcich', F. HcgcdiJs', L. Hoffmann', A. Manuel', A. Shukla', P. Bisson' and M. Peter'

* Paul Schcrrcr Inslitulc, CH-5232 Villigcn-PSI t DiSparlemcnt de Physique de la Matierc Ccndcnscc, Univcrsitd dc Geneve, CH-1211 Gen6ve 4

The Development of intense .-f+-sources (> 100 niCi Test runs at SAPHIR using Ni-foil from CBNM/Gccl ;i+) at PS1 is needed for both new and exisl'r.g experimental (< 0.1 ppm Co) and extraction tests using existing equip- applications in positron physics at PSI ynd at the University ment at PSI at low activity (10 mCi) have shown that the of Geneva: extraction efficiency is better than 60 c/i activity. Spectral interference with other nuclides was very low (< 100 ppm • Installation of a slow-positron laboratory-beam fa- ''"Co). Care must be taken to avoid contamination with cility at PSI |1] iron. The Ni-conlcm of the Co-fraction is acceptable. The final extraction product is a CoClv-solution. • 2D-ACAR for electronic properties of solids |2] at A prototype electrolytic cell was built and is now ready Uni/Gcncva for lest runs. The source will be able to withstand temperatures from liquid He to 300"C. Controlled diffusion of the • Tests of proposed procedures for slowing down positrons source material into the substrate (< 5 //m) al high temper- and conditioning slow-positron beams (repolari/a- ature will eventually be used to improve the adherence and tion, phase-space shifter) [3] at PSI. wear resistance. • 1D-ACAR defect studies |4) al PSI. Since the shut- Kquipmenl: A closed-cycle apparatus in which all down periods of the SAPH1R reactor were increased steps of extraction and electroplating can be controlcd by 1 significantly, a long-lived J' -emitter is desirable to remote operation in a safe way is necessary. To a certain compensate for the reduced availability of short-lived extent, an existing equipment used for '-'Fc-production can 1 'Cu which was used in the past. be duplicated. The whole apparatus must be placed in a hot cell. Unfortunately there is no free space available in Among the frequently used isotopes '"'Cu, --Na and an existing hot cell at PSI and even no free space for a new r + '"Co for strong .i -sources, "'To is preferred because: hot cell in an existing B-typc active area! Therefore it was necessary to use an existing C-typc laboratory which will M r h • The half-life of ' Cu is only 12.7 h ( ' Co: 71 d). be cquiped with additional filters to meet the requirements This is too short for 2D-ACAR and positron-beams. of a B-typ laboratory. The hot cell is at present under construction. Practically all necessary parts such as ma- • Commercially available --'NaCl-sourccs (2.7 y) arc nipulators, lcad-glas windows, glove-box and lead-bricks very expensive and often not reliable (;^ already existed at PSI and will be used to build a new hot cell. • -'-Na-sourcc production at PSi would imply an expensive dcvclopmcnl for the deposition of '-'-'NaCl of the desired activity, since this work must be done in a hot cell [51. Deposition of r"%Co by electroplating is the proper choice for remote handling in a hot cell. References • Reactor activated :'*Co from r'"Ni (n, p):iSCo can be separated from Ni by means of an ion exchanger, i.e. |1| W.B. Waebcr, D. Taqqu, U. Zimmermann, G. Soil, high specific activities can be obtained (31.7 Ci/mg). PSI-Bericht Nr. 68 (1990). A similar procedure is used at PSI for r'-Fc. [2] A.A. Manuel, M. Peter, Helvetica Physica Acta 63, 397 • It can be deposited on a substrate in metallic form by (1990). electroplating. This leads to better adherence and per- [31 D. Taqqu, Helvetica Physica Acta 63, 442 (1990), sec formance at high and low temperatures as compared also appendix 2 in (1|. to 22NaCl. Scaling of the sources behind ;Mhin win r dows is not essential for '*Co. [4] K. Ghazi-Wakili, Ph. Tipping, U. Zimmcrmann, W.B. Wacbcr, Z. Phys. B, Condensed Mailer 79, 39 (1990). Prototype design: 1 Ci •''"Co-sources will be produced by means of reactor activation, chemical separation and |5] H. Huomo, R. Jones, J. Hurst, A. Vehanen, J. Throwe, clcctrodcposition. This means 150 mCi /^-activity (15 % S.G. Usmar, K.G. Lynn, Nuel. Instr. and Meth. A284, /?+). The source material will be deposited on a Cu-shcct 359 (1989). (0,3 mm thick) in an active area of 8x3 mrrr. The source will be mounted in a Ti-capsule (16 mm diameter, 5 mm high) behind a 5 )im Ti-window. 101

POSfTRON REPOLARIZATION

Paul Scherrer Institute, 5232 Villigen-PSI / Switzerland D. Taqqu

With the increasing general interest in the applications M = 1. The oiher slates formed will have population p and of slow positron beams, positron factories providing high lifetime r,.-M (computed taking inio account (he mixing of intensity beams with various properties are being consid- the M = 0 slates in a magnetic field of 9 T |9]) as follows: ered all around the world [I], in contrast with standard laboratory sources where polarized beams based on rf+ Triplctt F= 1 M =-1 p = 50% r, -1 = 142 ns polarization can be easily obtained, the presently planned Triplet! F = 1 M = 0 p = 25% r, „ = .4 ns positron factories lead to intrinsically unpolarizcd slow po- Singlctl F = 0 M = 0 p = 25% r,, „ = .18 ns sitrons. This is naturally the case when the positrons are The positronium atom interacts with the hydrogen gaz produced via bremsstrahlung of high energy -, 's as it is done either by making elastic collisions with a cross-section in Livermore [2] and planned in Japan [3J where an electron 12 all possible kinds of beam features the positron factories cV (however not with polarization dcpcndancc) can be ex- will be able to provide, one of the most important properties trapolated to obtain an estimate of the M dependent uoss- - the polarization - is doomed to remain inaccessible and all seciions aM. what will be offered to experimentalors requesting polarized Now, the basic rcpolarization feature results from the beams is a standard low intensity beam of the kind used very short annihilation lifetimes n « and r« i, of both un- until now in various laboratories [6|. polarizcd (M = 0) positronium states so that they decay The purpose of the letter is to present a method to over- much faster than the break-up rate A,, = ptaa (with >• become this limitation of positron factories and achieve in- ing the Ps velocity). On the other hand the polarized (M tense slow positron beams with a high degree of polariza- = -1) state, because of its slow decay rate (t/n,_, < A, tion. The basic idea is to rcpolarizc the slow positrons by = rv(Ti)< prcdomincntly breaks up and regenerates a fully letting them interact adequately with spin-polarized atomic polarized low-energy free positron. hydrogen. Gaz cells of atomic hydrogen with 100 % po- The outcoming positron/posiironium yield Y,, and polarization have been produced a decade ago [7] by trapping larization P,, are: atomic hydrogen in a high magnetic jSeJd Ba ~ 10 Tal Jow temperatures (T ~ .3°K). In a solenoidal field with axial 1 + A,n,_i 1 + A r, „ 1 + A,,- distribution B(z) and maximum BOl the H atoms in two n lowest hyperfine states (designated by H1) are confined by •r'Ain.i ,._] a potential well of 7°K so that a window-free gaz cell can be obtained with a density that falls off axially from a max- 1 + -V..-. ° " 2 imum p0 at Bo like p- pa Cxp[-0.7°K(Bn - B)/ T Bo). In »Vith cxlrapolatcd at ~ 1 A- and a,, - 2.5 X at a Ps 17 3 the first experiments maximal densities p = 1O at/cm energy of 9 eV, one computes for p = 5 1015 at -n3 a 16 3 have been achieved and very stable operation at 10 at/cm yield Zn = 47% and a polarization Po = 95%. could be obtained. This intrinsically high polarization resulting from po- For the present application we consider such a cell sitronium formation can be significantly reduced by con- placed in a solenoidal field distribution where B = Bn is tamination from positrons having lost their energy by ex- constant over a length / and is made to fall off as sharply citation or ionization of the hydrogen atoms. The relevant as possible at the edges. The slow positron beam is injected cross-sections arc shown in Fig. ]. By taking into account axially from one side and extracted at the other side. For positronium losses resulting from backscattcring and accep- an average injection energy Eo ~ 16 eV the predominant tance limitations the effective Yn yield is smaller than the interaction is positronium (Ps) formation: value given above, and the competing processes mix into the beam a sufficient amount of unpolarizcd positrons to + 1 e + If —. Ps(E = Eo- 6,8eV) + H* (I) reduce the final polarization to about 50%.

2 An improved final polarization will be obtained if some with a cross-section of aPl ~ 3.9 X [8J. means for separating the polarised positrons from the oth- As only electrons with spin projection se- = -1/2 arc ers can be implemented. A first highly selective method present, only spin slates \sr+,St- > = |l/2,-1/2 > and makes use of the fact that, in the high magnetic field in 1-1/2,-1/2 > can be formed. This inhibits the formation which the operation takes place, the free positrons have cy- of the hyperfine state with total spin F = 1 and projection clotron radii of the order of a few microns so that their 102

the /: x H drift region [5] separates them from the incoming beam. The value of the cut-off energy c AV can be op- umized to obtain maximal beam repolanzation. Positrons loose 13.6 cV in positronium formation and break-up while those having excited an hydrogen atom lo the n = 2 level 4 1- loose thereby only 10.2 cV. For an incoming beam with maximal energy I:,,, and less than a lew eV energy width, most of Ihc backscaltercd positrons will be eliminated by selling AV = [•:,„ 13.(i eV (sec Fig. 1) without having any loss of the rcpolari/cd positrons. This selective feature achieves full •jlliciency by using a thick gay. cell {/ ~ 6 cm at /•,, = 5 K)1'1 at/cm'-') enhancing thereby the fraction of positrons whose first exit from the gaz cell is in the hack- ward direction. In this way accepted unpolarized positrons reduce practically to those having lost energy hy ionization (or excitation to n > 2) which according to Fig. 1 are quite a small fraction at E,> = 16 cV. The resulting beam polarization can be estimated to be around HO'/r with a yield close to 40%.

In comparison will) the polarization and the yields that have been achieved with standard radioactive sources, the tigures given afrave reprcsevrf a significant rmprovemem. This indicates that available laboratory beams would also benefit from such a repolarization scnemc. However the realisation of ihc window-free spin jxilan/cd gaz cell in- volves a technical effort that is best invested within the 5 10 15 20 25 frame of a positron factory. Furthermore, the oulcoming e+ Energy (eV) polarized positron beam, with a minimal diameter of a few mm at 9 T, belongs lo the class of "magnetic" beams which cannot be extracted into the free space without an adequate

Figure 1: Cross section fur e*~ -// /imnwrs; I lie ;>osit ro compensation of the angular momentum component of the field |I5|. Such a compensation can be achieved in lhe iiiinii fi'riiiutiiHi is taken fr,,;,/ ;'i^ (; (,f (f,f. ,s' (/frst /?o/-/j ;ii>l>ro.sini:ition). (total) elastic cross scctiim ;unl reccniiy proposed "phase spate shifter" which is planned excitation cross section* (wliiclt go like l/i?) from lief. to be devcloppcd as an extraction device for the intense I'.l ;UK1 the ionr/;tt it >n cros>: ^.a-tiou from I Iw most recent slow positron source proposed last year at PSI [16]. It ap- theoretical work [It]- pears therefore that the scheme presented here will find its most optimal implementation in the form of a rcpolarization stage that can be inserted between the converter and the phase space shifter of the RSI proposal. diffusion perpendicular to the field lines is cssenially zero. This contrasts with the transverse diffusion of the neutral This re-polarization concept can be investigated experi- positronium atom between formation ami break-up which ln :i mentally by inserting a slow positron source in an existing is of the order of \lp^Jc,v\ ~ 5 mm at /< = 5.10 at/cm . spin-polarized apparatus. By variing the source voltage The resulting signilicant widening of the oulcoming polar- and the cell density, the energy dependant cross-sections ized positron beam can be used to select Uic polarized com- for positronium formation, clastic and inelastic processes- ponent by introducing a cache over the field lines which arc can be measured. Together with the determination of the the continuation of the lines encircling the incoming beam. break-up cross -section (extracted from the lime distribution This method is optimal for small size beams as those oh- of the annihilation signal) optimization of the rcpolarizaiion Uiincd after a first rcmodcration stage. It should rcsull in a efficiency can be achieved. Ultimately, direct measurement repolari/.ation P = 98% with a yield Y = 25%. of rcpolarizaiion and yield can be obtained by introduction Another method can be also used for reduction of the of a suitable polarimclcr arrangement [o|. unpolari/.ed background in association lo the positron source based on the high rmxleralion efficiency scheme |5|. One An experimental program in that direction is presently introduces a cylindrical electrcxlc around ihe beam axis in under consideration in a collaboration between the PSI slow the low magnetic (icld region near the entrance of ihc spin- positron group and a group al the Kurchatov Institute in polarized gaz cell, With AV being the voltage difference Moscow which works in the field of spin-polarized hydro- between the electrode and the g;iz cell, positrons rcemit- gen. With source and detectors prepared at PSf and the tcd from the ga/ cell in I he backward direction will be apparatus available at the Kurehatov Institute a first ex- reflected by the electrode back into gaz cell if their energy periment putting together these two very different research is less than eAV. In this case they may travel Inck and fields may be realised in the near future. forth until diffusion allows ihem lo cross the gaz cell and come out (low s:ream. On die olher hand higher energy The author is indebted to W.B. Waebcr for his strong positrons pass ilic electrode and return lo the source where support and encouragement in pursuing Uiis line of research. Refercn es |8] i.C. Siralon, Phys. Rev. A35 M««7) 3725. 11 ] M. Peter and A.A. Manuel. Hclv. Phys. AcUi 63 (1990) 19] A. Rio.. Rev. Mod. Phys. 53 (1981) 127. 45X. |i')| J.W. Humberslon, Adv. Al. Mol. Fhys. 22 (19X6) 1. (2) R.H. Howcl) cl al, Nwl. In.str. ;ind Meth. in Phys. Res. ,' 111 L.W. Anderson c! al, Phys. Rev. Lcll. 52 (19X4) 609. BIO (i985) ?73. 112] A.M. Ermolacv and B.H. Brandscn, to be publishc;!. [3| T. Akahane an'S P. Chi ha in Positron Annihilation, World Scientific, Singapore, 1988, p. 592. ]13] R.J. Drachman, in Positron Scaticri.ig in Gazes, Plenum Press, New York, 1984, p. 203. (41 W. TrilLshauscr, G. Kogel, K. ichrcckcnbach, B. Kr- uschc, Hclv. Phys. Ada 63 (19s.')) yiX. [14] K.K. Mukhcrjec, N.R. Singh and P.S. Mazumdar, J. Phys. B 22 (1989) 99. [5| D. Taqqu, Hclv. Phys. Ada 6.3 (1990) 4-12. 115] D. Taqqu, Ret. 16, p.75 [6| J. Van House and P.W. Zitzcwiiz, Phys. Rev. A29 (1984) 96. [16| Porposal lor an Intense Slow Positron Beam Facility at PS1. PSl-Rcpon Nr. 6X, 1990. W.B. Wacbe-, D. [7] R.W. Clinc, D.A. Smith, P.J. Circytak and D. Klepp- Taqqu, U. Zimmermann, G. Soil, F.d. ncr, Phys. Rev. Lett. 45 (1980) 2117. 104

TRANSMISSION ELECTRON MICROSCOPY OF PRECIPITATES IN A THERMALLY AGED FE-CU MODEL ALLOY

H. Grimmer" and H. Caldcmn}

* Paul Scherrcr Institut, CH-5232 Villigen PSI t Inslitul fiir Angcwandie Physik, ETH-Honggcrberg, CH-8093 Zurich

Samples of Fc-0.8 wt% Cu were heat treated for 24 h at fim in diameter. The electron diffraction pattern showed 840 °C and quenched in saline solution at 15 °C producing that the precipitate was a Fe3O4 crystal viewed in [2 4 11] a supersaturated homogeneous solid solution of copper in direction. This interpretation was confirmed by EDX analy- a-iron. Copper precipitation was induced either by neutron sis, which gave 45 al% Fe and 55 at% O. A possible reason irradiation or by thermal aging lor 120 h at 500 °C . We for the formation of this oxide is the cutting of the alloy by describe TEM investigations of the aged samples. clcctrocrosion after thermal aging. X-ray diffraction of the Cutting circular disks with 3 mm diameter out of the cut surface showed that a layer of ->-Fc of approximately 0.3 mm thick sheet, which had been obtained by elcctrocro- 1 /im thickness had developed at the surface and it also sion, proved to be a problem due to the britilcness of the showed a weak signal of Fc:!O4. it remains unclear how material caused by the precipitates. Therefore, the disks it was possible for such an oxide to develop more than will be cut before aging or neutron irradiation in the future. 0.1 mm below the surface. The disks were thinned using a Tcnupol apparatus with an electrolyte of 200 g NaaCrO4 in 1000 ml acetic acid. Transmission electron microscopy was done with a Phi- lips CM30 at 300 kV. The diameter of the precipitates shown in Fig. 1 is typically 10 nm. The Moire fringes arc caused by the lattice mismatch between matrix and precipitates that try to minimize imcrphasc boundary energy by a definite orientation relationship with respect to the matrix.

"fe'r

^ 50 nm

Figure I: aright ficltl imiige of precipitate* in the n-iron matrix. Figure 2: Dark field image of Fe^O* precipitate. Dark field images taken with a {110} spot of o-iron show that this spot coincides with a spot of many precip- The dark and bright parts of the precipitate shown in itates, which is compatible with the hypothesis of Cu pre- Fig. 2 belong to different subgrains. Dislocation arrays cipitates in Kurdjumov-Sachs orientation. The diffraction (marked by arrows) separate further subgrains. The weak pattern showed further spots due lo ihc precipitates that lay beam conditions prevailing in the dark parts map disloca- clc .cr to the transmitted beam than expected for copper, tions as lines consisting of a succession of bright points. indicating an ovcrstructurc of the precipitates or periodic It is planned to study the small precipitates in more de- defects in the inlcrphasc boundary. tail using high resolution electron microscopy and to con- Analysis of the chemical composition using the EDX tinue by investigating radiation induced precipitation. facility of the CM30 gave a copper content of the matrix of 0.5 wl%. Jl was not possible to analyze the composition of the precipitates due to their small size. A sheet of the thermally aged material was analyzed by X-ray diffraction. It showed trie «-iron and gave a faini indication of copper, the strongest reflection of which overlaps with the strongest reflection of n-iron. It was a surprise lo discover a large precipitate over 2 105

GENERAL RELATIONS BETWEEN THE RESTRICTIONS IMPOSED BY DIFFERENT POINT GROUPS ON THE FORM OF PROPERTY TENSORS

Paur Scherrer Institui, 5232 Villigcn PSI / Switzerland

H. Grimmer

Anisotropic properties of crystals play an important role has been illustrated in |2| by deriving the form of the c- in modem solid state physics and its technological applica- tensor describing quadratic clectrojsyration from the form tions. Nonlinear response of crystals to external stimulation of Uic clasto-optic tensor, which is an i-tensor of fourth described by tensors of higher rank is gaining importance rank, symmetric in its first und second pairs of indices. with the availability of strong sources of stimulation (e.g. The relations (I) provide a quick test for existing tables of lasers). This note presents research aimed at describing the forms of property tensors that change sign under 1 and the connection between the forms of property tensors for will be helpful for the compulation of new tables for higher different material symmetries in a particularly lucid man- order effects that become accessible to measurement with ner, which is of interest not only for the determination of improved experimental techniques. the form of tensors of high rank but also for the choice of Our result becomes much more powerful if Shubnikov materials for a given application. point groups arc considered instead of ordinary point groups, Tensors describing material properties arc inyarianl un- i.e. if rotations may be combined not onjy wjih J but also der the point group of the material under consideration. with time inversion 1' and space-time inversion I'. A tensor This requirement puts restrictions on the form of the ten- is cither invariant under 1, 1' and I' (i-tensor) or it is invari- sor. They depend on its rank, internal symmetry, and us be- ant for one of them and changes sign for the other two: an haviour under space inversion 1. Ordinary tensors of even s-tensor is invariant under spate inversion, a t-tensor under rank and pscudo tensors of odd rank are invariant under 1 time inversion and a u-tensor under space-time inversion. (i-tensors); ordinary tensors of odd rank and pscudo ten- The form of an i-tensor depends only on the Lauc class sors of even rank change sign under 1 (c-tensors). Eleven (The 122 Shubnikov CPGs fall into 11 Lauc classes, each among the 32 crystallographic point groups (CPGs) con- of which contains one of the 11 CPGs consisting of pure tain only rotations, 11 are direct products of rotation groups rotations only.) The forms of tensors for these 11 groups with 1, and 10 contain I only in combination with rotations. together with the forms derived from them by (1) arc suf- Assume that the form of a tensor of given rank and inter- ficient to describe the forms also of s-, t- and u-lcnsors in nal symmetry has been determined for the 11 pure rotation any of the 122 Shubnikov CPGs if the orientations of these groups. These forms arc Uie same for c- and i-tensors. groups with respect to a Cartesian coordinate system arc There are no other forms for i-tensors because the form of chosen appropriately, as explained in |2|. an i-tensor depends only on the Lauc class of the CPG. A Analogous results for the limiting point groups {which c-tensor vanishes for the groups containing 1 and is given contain oc-fold rotation axes and often arc called Curie by the following formulae for the 10 remaining groups [ 1 ]: groups) have been derived in [i|. Whereas CPGs describe tfic point group symmetry of single crystal's. Curie groups Ti =7^-7:, often appear as symmetry groups of textured polycrystaliinc

7mm2 = 7o — T?22 7G =73-71; materials, composite materials, polymers and liquid crystals. 14mm = M ~ J 42'2 j2rn = 7 222 — 7 .) (1) The results can be generalized also to the non-crystallo- Vim = 7.-12 - Tfi-2 graphic point groups that describe the point symmetry of quasi-cryslals, e.g. pentagonal, hcpiagonal or icosahedral symmetry. where Ta denotes the form of a c-tensor of given rank and internal symmetry that is invariant under the point group References G. The first tensor on the right hand side can be uniquely written as a sum of the tensor on the left hand side and of the (1] II. GRIMMKR, llelv. Phys. Acia (in print). second tensor on the right. To give an example: any tensor of form T3 can be uniquely written as a tensor of form T3m [2] II. GitlMMf.R, Ada Cryst. A (in print). and one of form 732. This implies that tensors l\m and 732 have only the vanishing tensor in common. The Hermann- (31 International Tables for Crystallography. (1983). Mauguin symbols for the point groups fix the orientation of Vol. A. Dordrecht: Rcidel. trre symmetry efcrnents as stated in Table 12.2 of (31 with c unique for the monoclinic point groups. (T,n = 1\ - T2 is valid also for b unique.) A proof of the above relations for the CPGs can be found in [2]. The formulae (1) can be used for example to determine all forms of a c-tensor if the forms of the i-tensor of the same rank and internal symmetry are known. This 106

THE EFFECT OF ELECTRONIC ENERGY LOSS ON THE DYNAMICS OF THERMAL SPIKES IN CU

Nr , PS1 - LLNL

S. Pronneckc", A. Caro', M. Victoria", T. Diaz dc la Rubia', M. W. Guinan'

+ Paul Schcrrcr Institute, 5232 Villigcn, Switzerland t L-644, Lawrence Livermore National Laboratory, Livcrmore, CA 94550, USA

The characterization of the primary state of radiation dam- 1.00 i age in irradiated material has been a problem of interest to the physics and materials research community for the last fifteen years. Component degradation under irradiation in fission and fusion reactors, as well as ihc use of mod- 0.95 - em of ion beam techniques for materials modification have motivated large research efforts aimed at achieved a basic understanding of this problem. Of fundamental impor- 0.90 - tance in the characterization of the primary stage of damage CO is understanding the role played by energetic displacement cascades. OJ 0.85 Displacement cascades are initiated when an energetic Total energy ion or neutron collides with an atom of the solid, thereby ^ Coupling constant driving the system to a highly non-equilibrium state con- fcfl 1.00 r 8 taining elevated defect and energy concentrations and, in % \ 5.0 KeV the case of irradiated alloys, disorder. C In the last few years, molecular dynamics computer sim- ft ulations (MD) have been shown to be a useful tool to de- H 0.95 scribe the dynamics and structure of energetic displacement o cascades. For primary knock-on atom (PKA) energies up to several kcV previous studies [1] (2] have shown that af- 0.90 - ter a brief collisional phase lasting of the order of 10"13 sec, a cooling phase ensues in which atom trajectories arc diffusive as opposed to ballistic. This phase, commonly termed a thermal spike, lasts several picoseconds and has 0.85 been shown to be of critical importance in determining the amount of atomic rearrangement that takes place in displacement cascades. Most of the MD simulations has been done up to now Time (ps) using a pair potential to describe the ion-ion interaction. A new vectorized code, named MOLDY-CASK, which con- Figure 1: Logarithm of the ratio of the total energy of tains a many-body potential based on local electronic den- the crystal for the case with electronic losses to thai sity formalism [3] , has been used for our simulations. A of the reference run versus time for both spike events, simple model for the ion-electron interaction has been built Y(t). On the right ordinate, the inverse of the derivative with this formalism to describe the transfer of energy from ofY(t) is plotted, which is equal to the eleclron-phonon ions to a bath of electrons [4]. coupling constant. In order to evaluate the effect of the ion-eleciron interaction on the dynamics of the thermal spike, two gaussian radial distribution of 2.5 and 5 keV of kinetic energy has the losses included in the simulation. To avoid including been distributed in a cell containing 55296 atoms of cop- in the analysis the effect of the artificial damping imposed per [5]. This metal was choosen because its weak coupling at the boundaries, which describe the dissipation of en- between ions and electrons which validates the assumption ergy out of the cell into an infinite continuum, the quan- thai electron temperature remains constant. Two runs ai tity Y(t) = \o%\Et(t)/E0{t)] has also been defined. In each of the two energies considered have been performed: order to obtain the time dependence of the coupling con- one with electronic losses and one without them, the latter stant between the electrons and the lattice throughout the providing the reference frame. spike lifetime, W{t) - [dY(t)/dl]-' has been evaluated. To investigate the influence of the electronic coupling In fig. 1 we plot Y(t) versus lime, which shows that the on the total energy, Ee(t) and E0(t) have been defined total energy does not decrease linearly in the logarithmic as the total crystal energy for the cases with and without scale. This effect is due to the temperature dependence 107

° c miiii; :•':::: :•;-.••! v,"'r'. ^' '"^•>:':->;;.';:;:*

d Hi b S

M'M'M^'M'.'j'X'lvi'/M1!'/!"'1!1 iiiiiiiiii Illiftlilfl iillii i

Figure 2: Cross sectional views of the 5 keV heat spikes Figure 3: Radius of the regions where the temperature in at two different times. The plots represent cuts through the spike is above the experimental melting temperature a (001) plane, (a) and (b) no electronic losses at t = 1 of copper for the four events considered, (a) and (b) 5 and 4 ps, (c) and (d) with the electronic losses at t = 1 keV events with and without the electronic losses, (c) and t = 4 ps. and (d) 2.5 keV events, same as before.

of the electron-phonon coupling. On the right axis of fig. temperature is no longer valid and a model for the elec- 1, we plot the value of W(t), the lifetime of the fluctua- tron heat transport must be included in the calculation. We tions. This ranges from 2 to 6 picoseconds, the last value conclude by pointing out that for higher energy events the corresponding to the choosen value at low temperature. lifetime of the cascade will increase, up to the subcascade The disordering tendency of the 2.5 and 5 keV spike breaking point, and therefore a significant effect may be event in copper is demonstrated in fig. 2, where instan- expected even in the case of weakly-coupled copper. taneous atomic configurations in a (010) plane through the center of the spike are shown at two different times. As can be seen from fig. 2, the atomic rearrangement lasts of the References order of 4.5 picoseconds in the case where the electronic [1] M.W. Guinan and J. Kinney, J. Nucl. Mater. 103/104, losses are included and more than 5 picoseconds without 1319 (1981). them. To further quantify the dynamics of the cascade, we [2] T. Diaz de la Rubbia, R.S. Averback, and Horngming have evaluated the temperature as a function of distance Hsieh, J. Mater. Res. 4, 579 (1989). from the center of the spike region and from this, we have extracted information regarding the radius of the region that [3] T. Diaz de la Rubbia and M.W. Guinan, J. Nucl. Mater. 174, 151 (1990). remains at a temperature above the melting point of copper, Rm(t). This is shown in fig. 3. While the melted zone has [4] A. Caro and M. Victoria, Phys. Rev. A40, 2287 (1989). already disappeared in the case where electronic losses are present by about 3 picoseconds, there is still a melted region [5] S. Pronnecke, A. Caro. M. Victoria, T. Diaz de la Ru- of about 12 A in radius in the case without the losses. bia, M.W. Guinan, accepted for publication in J. Mater. Therefore the present simulation has shown the impor- Res. tant effect that inelastic energy losses might have in the evolution of energetic displacement cascades. We have observed a moderate decrease of the disordered zone lifetime when electronic losses are included. This effect is present in copper, the most unfavorable example as the ion-electron coupling is weak in this case. In other materials such as nickel, where the ion-electron interaction is stronger, the difference is expected to become important already for events in the low-keV range of energies. In this case however, the approximation of a fixed electron bath 108

RADIATION DAMAGE CASCADES: A LIQUID DROPLET MODEL OF SUBCASCADE INTERACTIONS.

M. Alurraldc*t, A. Caro*. M. Victoria*.

* Paul Sclicrrer Institute 5232 Villigcn - PSI, Switzerland. t Comision Nacional dc Energfa Alomica, Libertador 8250, (1429) Buenos Aires Argentine.

Present day computer capabilities does not allow to are intended to provide information related to simulate molecular dynamics cascades with energies fluctuations of the derived quantities. above 25 keV [1]. One of the most important results of Our analysis of the influence of the initial parameters MD simulations is the prediction of the cascade core of the PKA ( position and direction of motion), on the melting in Cu and Ni [2-4]. derived results exhibit a standard deviation of only 15 In a previous investigation we assumed that the % for the volume and 13 % for the damage energy melting picture for Cu and Ni is correct and applied despite of shape differences as show in Figures la and simple thermodynamics to predict the behavior of lb. cascades [5]. Previous attempts to solve the cascade evolution using continuum heat equations gave only qualitative information because both the energy deposition profile and the thermodynamic factors entering ihc equation, are in fact unknown [6,7]. The ?im of this work is to apply the ideas presented in Ref. [5], namely the coupling between the binary collision approximation (which gives the profile of the energy deposition), and the simplified version of the heat equation, to get a description of the cascade behavior in several materials, over a large energy range, correctly accounting for the geometric aspects . This description provides an estimation of subcascade interactions, in an energy range inaccessible to MD with present-day computational performances. To perform these calculations we have used the Marlowe-12 code [8J. It is based on the binary collision approximation, in which the cascade develops inside a crystalline solid as a sequence of two-body collisions. The output configuration contains the positions and kinetic energies of all atoms set into motion. It represents the end of the collisional phase of i o. the cascade. The problem then is how to relate the spatial distribution of this amount of energy to the real deposited energy density. The volume of the melt strongly depends on this density. A simple way to solve this problem is to apply a multiplicative factor, EpKA^k't0 me energy of each of these atoms. As we showed in [5], this approximation does not significantly affect the energy density profile. We start then with the energy density at the end of the Figure I - 500 keV cascades in Cu for two different PKA directions: collisional phase. This energy will be transformed into a- (0, direction. All these cases subcascade regime. It is well known that if the volume 109

of the cascade is extracted from the nuclear stopping It is interesting to compare these results with the linear cross section, which decreases with energy, ihc energy size of the cascade obtained by measuring the distance density of the cascade decreases with increasing EJ.^A- between ihe two more distant defects surviving close However, the volume of the melt depends on pairs recombination, Rcf. [9). There it is shown that additional variables. In a very qualitative way, one this dimension increases with energy as E°-7. might think that, at high energy, the volume of the Compatibility of this two results requires that the cross melt is just the volume of a single subcascadc times section of the cascade be almost constant, varying as the number of subcascades and therefore, since the E0 2. volume of a single subcascadc is independent of E|>KA, Let's now analyze Figure 2b. Although the volume of the total volume of the melt has to be proportional to the melts show no influence of the breakdown into Ei'KA (slope 1 in Fig. 2a). At low energy, below the subcascades, their life-limes T show a clear break. threshold of subcascade formation, the energy density Once again, in a very qualitative way, it can be said may be supposed to decrease, let's say like 1/Ea; then that at high energy, assuming independent assuming uniform spatial density, pV = E = V/E(I, we subcascades, the life .imc has to lie constant and equal get V = E1+H, i. e. for any positive a the slope in to the life lime of a single subcascade. On the other Figure 2a has to be grater than 1 and, on going from hand, at low energy the life time has to relied the low to high energy, the slope 1 has to be reached from surface to volume ratn. Therefore, on going from low above, giving a negative second derivative. However, to high energy the slope has to decrease from some from the figure it is absolutely clear that the volume of positive value to 0. Figures 21) show that the slope at the melt docs not behave in that way: The volume of low energy is approximately 0.5 and at high energy it the melt is proportional, to a high accuracy, to E"yu is slightly larger than zero. This non-zero slope is a measure of the .subcascadc interaction: the life lime is larger than the simple non-interacting prediction because the subcascades cannot quench efficiently due to their proximity. The remaining parameter plotted in Figure 2b is the time to reach ihc maximum volume. For very low and very high energies the mell .starts to decrease soon after the end of the collisional phase, at the first step of the diffusion process; but at intermediate energies the energy density induces ihc melt to expand. Assuming a Gaussian shape for the energy density, it is easily seen that the volume increases provided ihc width of the Gaussian is smaller than a critical value or, equivalcntly, the average energy density is larger than a critical value. It is very important to point out that the MD simulations on 25 kcV in Cu show the same behavior namely, the volume of the mcit, dcicnui.uxi by counting the number of atoms with kinetic energy larger than 3kTm, has a maximum at some time well after the end of the collisional phase (1 ]. The variation of the life-time versus E is reflected in the ion mixing which shows a break at the subcascade threshold energy, Figure 2c. According to the most commonly used thermal spike mixing models 110,11) the mixing for an spherical cascade is proportional to 5ri Ed and the mixing density of a cylindrical cascade is proportional to (Ej/L)2, that is, to the square of the energy density (L is the cascade length). Multiplying 15 by L lo get the total amount of mixing one gets the IM 10 2 proportional to Ed /L. Taking the longitudinal 1 10 100 1000 7 dimension of a cascade from Rcf. [9] as Ed°- we get Energy (keV) 3 IM proportional to Ed'- .

I-igurc 2 • Results for cascades in Cu: a) maximum volume of die According to Figure 2c the IM at low energies is 3/2 melt, b).: life-lime of the liquid (straight lines arc guides to the eye), approximately proportional lo EPKA . From our 09 " : lime at which the maximum volume of the melt appears, c) ion Marlowe outputs Ed is proportional to EpKA for all mixing in units of A2 slcp/scc. * : molecular dynamics results 111. four materials; therefore the slopes we get for low energy spikes, E '-35, is 20 % smaller than the 0 92 os d for Cu and Ag, to E - for Fe, and 10 E '-> for Ni, spherical gaussian prediction. Al high energy our 0 8 0 7 with correlation better than 0.999. results show IM proportional to E,.KA - or Ed - 110

which is only one half of the value predicted by the ions, for example light and heavy, at the same energy; cylindrical model. We believe that our more precise then the initial energy deposition wiil be very different description of the cascade geometry makes our but the diffusion process will be similar since the prediction more reliable. target is ihc same. With our procedure, the ratio of IM The IM efficiency per unit damage energy, i. c. the in the two cases can be obtained, once again without derivative of the IM with respect to the energy, free parameters. decreases wi'h increasing energy, above the The number of liquid droplets can be assimilated, as subcascade threshold. The efficiency changes from expressed above, to the number of subcascades or the more than linear to Jess than linear. In other words: number of defects clusters resulting from their below the threshold a cascade of energy E is more quenching. A first comparison made with the results efficient to mix than two cascades of energy E/2, reported in Rcf. [12] shows that thresholds for whereas above the threshold the reciprocal is true. The subcascade formalion arc comparable: in Cu il is in ihe slope of the IM is determined by both the slope of the region of 20-50 keV according to the present life-lime and the slope of the volume. In fad. the ratio calculation, compared to 30-50 kcV in the IM/VT, which is independent of the unknown lime observations. The corresponding values in Ag arc also unit, is found to be fairly constant over ihe whole comparable: 100-200 kcV from the calculations and energy range, in particular for Ag: IM/VT = 1.5 10n 100-200 kcV in the TEM observations. A more (A sec)-' with a standard deviation of 21%. Similar detailed comparison with these and oilier electron calculations give IM/VT = 4.3 10", 7.0 1011, and 3.2 microscope observations is being prepared. 1011 (A sec)"1, with standard deviation of 18%, 24%, In conclusion, this simple analysis in terms of the and 14% for Cu, Ni and Fe respectively. It is liquid model for the thermal spike correctly includes interesting to note that, within a factor of 2, this the geometry of the cascades, giving .satisfactory quantity appears to be constant for all four materials. quantitative results and providing a qualitative view Finally, figure 3 shows the number of clusters versus over a large energy scale as well as over different PKA energy for the four elements. The thresholds for materials. These characteristics arc far from being subcascade formalion appear clearly between 20 and fulfilled by present-day molecular dynamics 50 keV for Cu and Fc, between 50 and 100 kcV for Ni, possibilities. and between 100 and 200 kcV for Ag. Unfortunately the point at 1000 keV for Fe could not be determined References due to the very large size of the cascades which creates difficulties in the discretization procedure. As [1] T. Diaz de la Rubia, private communication. mentioned before, according to the liquid picture of [2] T. Diaz dc la Rubia, R. S. Avcrback, R. Bcnedck, cascades, these numbers should correspond to the number of clusters of defects surviving the cascade. and W. E. King, Phys. Rev. Lett. 59 (1987) 1930. 13] T. Diaz dc la Rubia, R. S. Avcrback, H. Hsich, and R. Benedck, J. Mater. Res. 4 (1989) 579. Also H. Hsieh, T. Diaz de la Rubia, R. S. Avcrback, and R. Bcnedck, Phys. Rev. B 40 (1989) 9986. [4] T. Diaz de la Rubia, 'The Structure and Dynamics of Energetic Displacement Cascades in Cu and Ni. A Molecular Dynamics Computer Simulation Study', Ph. D. thesis, SUNY at Albany, 1989. (5] M. Caro, A. Ardclea, and A. Caro, J. Mater. Res. 5, 1990. (to appear) [6] G. H. Vineyard, Radiation Effects, 29 (1976) 245. [7] R. Kelly, Radiation Effects, 32 (1977) 91. Figure 3 - Number of clusters as a function of PKA energy. These [8] M. T. Robinson and I. M. Torrcns, Phys. Rev. B 9 numbers are determined by visual inspection of outputs. (1974) 5008. 19] H. L. Hcinisch, J. Nucl. Mater. 103 & 104 (1981) We can compare results on the same material at 1325. different PKA energy or at ihe same energy but [10] G. H. Vineyard, Rad. Effects 29 (1976) 245. different type of projectile. For the different PKA [11] J. B. Sanders, Rad. Effects 51 (1980) 43. energies we choose to compare with preliminary [12] M. Kiritani, T. Yoshii, S. Kojima and Y. Satoh. results on 25 keV MD simulations in Cu [1], which, Proceedings of the Workshop on "Effects of Recoil together with the 5 keV simulations used in [5], are Energy Spectrum and Nuclear Transmutations on the added to Fig. 2. The agreement at 5 keV is of course good because it was used as support of the present Evolution of Microstructure", Eds. W. Green, M. approach in [5]. The excellent agreement at 25 keV Victoria, T. Leffers and B. Singh; Rad. Effects and has to be considered as a strong support for our model. Defects in Solids 113 (1990) p. 75. For the case of different projectiles, it would be very interesting to have experimental results of IM on a given material bombarded with two unlike type of Ill

Technical Physics (3400) 112

STUDY PROJECTS

C. Marinucci* • Paul Scherrer Institute, CH-5232 Villigen PSI

Magnetic energy storage The quench characteristics along the conductor length are within acceptable limits (metal tem- The Paul Scherrer Institute is studying, in collabora- perature «s 100 K, helium pressure =s 34 MPa, tion with Asea Brown Boveri, a 50 kWh superconduc- helium mass flow rate « 75 g s"1).

tive magnetic energy storage model (referred to as the •;rric-'^ ciodcl Inr.ff. cefe '••0--.--13

SMES-ck Modefy. Goal of this test device is to demon- 1 16 31 strate the feasibility of a 1 MWh SMES concept (referred to as the SMES-ch Prototype), designed to level 1 ! the peak electrical loads required, for example, by Rail- — ways applications. ! — i

• • tib-ch maximum sw>re

Figure 1: Helium pressure time-history at He inlet (1), The SMES-ch Model proposed by PSI is a solenoid He outlet (31) and conductor half-length (16) of 2.6 m outer diameter and 1.8 m length, operated at a current of 5 kA and at a magnetic field of 7.2 Tesla, with forced-flow cooling. This model, using NbTi • A minimum-volume coil configuration operated at superconductor in a cable-in-conduit (CIC) conductor, low magnetic field (e.g. 4 to 5 Tesla) is feasible can be discharged to 10% of the maximum stored en- as far as all quench characteristics are concerned. ergy with pulses of « 2.5 minutes. During the prelimi- Some design parameters require careful selection. nary design phase, now in progress, a number of detailed • The dump delay, which is the time between the analyses have been performed in the LTP Study Group detection of the noim»l xoi\« aivd tfre sVaiV of \Are [l][2][3][4][5]. Objective of these studies is to contribute dumping process, is a critical parameter for the to the definition of an optimal conductor geometry and coil safety folio.-ing a quench. coil layout which satisfies refrigeration and fabrication constraints, within realistic scenarios for the operation of the SMES system. Discharge performance analysis Goal of the SMES-ch system is to be operated with Quench transient analysis fast (ss 2.5 min.) discharge pulses at full power. During the discharging process, and subsequent charging one, The basic process of quench in superconducting mag- heat is generated due to ac losses. This, and other heat nets is the conversion of stored electromagnetic energy loads are cooled by force-flow of supercritical helium, into heat. This process is characterized by the irre- so that the temperature in the coil remains below safe versible propagation of the normal zone, its detection limits. Each pulse generates a temperature wave which and the consequent current decay. This analysis, whose propagates, with helium, along the conductor hydraulic goal is to demonstrate that during a quench the coil will channel; a sequence of pulses generates a sequence of not be subject to conditions leading to its damage, was temperature waves, whose effect are superimposed. The performed using the 1-dimensional code QUBIG, devel- discharge performance envelope defines the characteris- oped at PSI for the Large Coil Project. The parameters tics of pulses which can be eventually applied to the used ior trie analysis ot ttie SMES-ch Model are based coil in an unlimited sequence. In this study the com- on the reference configuration proposed by PSI, and on plex problem of defining the performance envelope of a new set of minimum-volume configurations. The ef- the SMES-ch Model was solved by dividing it in three fect of variation of some coil and conductor parameters, steps: still undefined or open for optimization, was also investigated. The main conclusions of this analysis, reported 1. an ac losses analysis, which provides heat loads in detail in [2], are: used in the 113 '.'J.'i S - CM model

2. 1-dimensional steady-state flow analysis of a single pulse, which provides temperature increment and helium velocity along the conductor length, used in the

3. multi-pulsr analysis, which defines the performance envelope.

A detailed description of models and results can be found elsewhere [3]. The main conclusions, assuming a rectangular pulse at full power (1 MW for 154 s), are:

• The total heat load due to ac losses is 20x4.5 W in the complete coil. Conductor optimization may cc: isoo realistically lead to a 50 % reduction of this load. -:..;.=, role [S] Figure 3: Summary of discharge performance, using an optimistic Tmi and a pessimistic Tm2 temperature limit

The SMES-ch Model proposed by PSI is a solenoid with current density of 40 A mm"2 in the winding pack, ?= 5 -00 - • inner and outer radius R^n = 1.11 m and R,mt = 1.30 m, respectively (a = ROTXt/R,n = 1.17). The selection of the outer dimension was constrained by an existing epoxy impregnation equipment. The solenoid length Z UJ3.00 - was left variable to match the specified stored energy of 180 MJ. Due to radial and axial symmetry, only the ,,2.00 - upper-right quarter of the coil was investigated, using cylindrical coordinates (R,Z). The magnetic field calculation has shown that although the total field Bt decreases rapidly outside the coil, absolute values remain n-TTTn relatively high near the solenoid. At a radial distance of O. 1 .OO P. -OO 3.00 4 .00 5-00 6.00 7.00 8.00 5 m from the coil axis (R — 5, Z = 0 m) the approximate MRGNETI C F IELD C T] value is 50 mT (1 mT = 10 Gauss), and 100 mT at Z - 5 m. At R =s 18 m Bt is below the tolerance level of 1 Figure 2: Time average of total ac-losses, as a func- mT. Shielding of this stray field may become necessary tion of initial magnetic field. Hysteresis losses (P#), in order to reduce its eventual damaging effects on bio- coupling losses in wire (Pc,w) and subcable (Pc,s) are logical systems and electronic equipment. The shielding also included. analysis of the SMES-ch Model will be performed in a next development, when the evaluation of a site at PSI is completed. • Following a single discharge pulse, the maximum conductor temperature is 5.3 K, at the outlet of the 1035 m long hydraulic channel (0.6 K due to ac losses, 0.3 K due to throttling).

• A continuous coil operation require dead times between pulses. A discharge-charge cycle at full power is possible, depending on the temperature limit, every 20 to 50 minutes, as shown by a con- servative model (Fig. 3). Considerably shorter cycle times are predicted by a more realistic set of assumptions, based on recent conductor developments.

• A coil with layer-wound subcoils and a squared conductor cross section are recommended.

Electromagnetic analysis 0 500 1000 1500 ?000 Figure 4: Contour plot of the total magnetic field gen- This analysis, reported in detail in [4], includes: erated by the SMES-ch Model 1. magnetic field calculation of the SMES-ch Model, and The 1 MWh SMES-ch Prototype proposed by PSI is a NbaSn solenoid with overall current density in the 2. shielding analysis of the SMES-ch Prototype. winding of 20 A mm"2, operated at a current of 17 kA 114 and a magnetic field of 12 T. Inner radius, outer radius References and length are 2.15, 2.85 and 3.0 m, respective! \ Goal of this shielding analysis is: (a) to investigate .afferent [1] C. MARINUCCI - Latent developments in the pre- iron shielding geometries, and (b) to access their impact liminary design studies of the SMES-ch. Model- PSI on the stray (=leaking) magnetic field in the vicinity Report LTP-90.06 (1990) of Jhe superconducting coil. Thick and thin shielding, close and at a distance from the coil, in open and closed [2] C. MARINOCCl - Preliminary quench analysis of configurations, have been investigated. The preliminary the Swiss superconductive magnetic energy storage results have shown: model- PS1 Report LTP-90.08 (1990) [..-] C. MARINUCCI - Preliminary discharge perfor- • A thin shielding at a distance from the coil presents mance analysis of the Swiss superconductive mag- an advantage with respect to a thick one close to netic energy storage model- PSI Report LTP-90.10 the coil. A careful assessment of the magnetic (1990) forces is necessary. [4j C. MARINUCCI - Preliminary electromagnetic • An interesting cost/benefit compromise is offered analysis of the Swiss superconductive magnetic en- by a shielding without the bottom part, whose ergy storage - PSI Report LTP-90.11 (1990) weight is =s 5.6 MN. An optimization is necessary to match the tolerated limit of 1 mT. [5] C. MARINUCCI - Auslegung Bahnspeicher-Modell. Felder, Krdfte und Kulung - PSI Report LTP-90.14 • An iron shielding represents a non-negligible frac- (1990) tion of the total cost of a SMES-ch Prototype. A system without shielding may prove to be adequate if enough vertical and horizontal clearance .is available. 115

SUPERCONDUCTING PROPERTIES OF Bi2Sr2CaCu2Oa!/Ay WIRES

R. Wesche" B. Jakob" G. Pasztor' * Paul Scherrer Institute, CH-5232 Villigen PSI