NOTIZIARIO Neutroni e Luce di Sincrotrone - Issue 4 n.1, 1999

NOTIZIARIO Neutroni e Luce di Sincrotrone Rivista del Consiglio Nazionale delle Ricerche

SOMMARIO

Cover photo: Contour plot of the intensity measure from a graphite crystal versus the strip position and the neutron wavelenght. The two plots correspond to a different sample-todetector distance.

EDITORIALE . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2

F.P. Ricci

RASSEGNA SCIENTIFICA Resonant Magnetic Scattering of Polarized Soft X-Rays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 M. Sacchi

The Silicon/Gadolinium Detector for Thermal Neutrons: Current Status and Future Perspectives

....................

14

...............................................................

23

C. Petrillo

Muons at ISIS Il

NOTIZIARIO Neutroni e Luce di Sincrotrone

R. De Renzi, T. Shiroka

è pubblicato a

cura del C.N.R. in collaborazione con il Dipartimento di Fisica dell’Università degli Studi di Roma “Tor Vergata”.

DOVE LUCE DI SINCROTRONE

F.P. Ricci

The VUV Beamline of ELETTRA: an Example of High Resolution Photoemission Core Level Spectroscopy on Sb/Si(001)c4x2 and Ge/Sb/Si(001)2x1 Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

COMITATO DI DIREZIONE:

C. Quaresima et al.

Vol. 4 n. 1 Giugno 1999 Autorizzazione del Tribunale di Roma n. 124/96 del 22-03-96 DIRETTORE RESPONSABILE:

M. Apice, P. Bosi COMITATO DI REDAZIONE:

C. Andreani, L. Avaldi, F. Boscherini, U. Wanderlingh

DOVE NEUTRONI Design and Performance of the Novel Multidetector Neutron Spin Echo Spectrometer Span at BENSC

SEGRETERIA DI REDAZIONE:

D. Catena HANNO COLLABORATO A QUESTO NUMERO:

M. Celli, D. Colognesi, M. Zoppi GRAFICA E STAMPA:

....................

39

C. Pappas et al.

VARIE

..............................

44

om grafica via Fabrizio Luscino 73 00174 Roma

SCUOLE E CONVEGNI

.....................

49

Finito di stampare nel mese di Giugno 1999

CALENDARIO . . . . . . . . . . . . . . . . . . . . . . . . . .

50

SCADENZE

...........................

52

FACILITIES

...........................

53

PER NUMERI ARRETRATI:

Paola Bosi, Tel: +39 6 49932468 Fax: +39 6 49932456 E-mail: bosi@server.dcag.cnr.it.

PER INFORMAZIONI EDITORIALI:

Desy Catena, Università degli Studi di Roma “Tor Vergata”, Dip. di Fisica via della Ricerca Scientifica, 1 00133 Roma Tel: +39 6 72594364 Fax: +39 6 2023507 E-mail: catenadesy@roma2.infn.it

Vol. 4 n. 1 Giugno 1999

EDITORIALE

partire da questo numero il Notiziario assume una veste editoriale diversa, si arricchisce di un Comitato di Direzione, di cui fanno parte il Dott. M. Apice e la Dott.ssa P. Bosi del CNR. Inoltre si avvarrà della consulenza scientifica del Dott. L. Avaldi e Dott. U. Wanderlingh che, nel Comitato di Redazione, sostituiscono rispettivamente la Dott.ssa R. Camilloni e il Prof R. Caciuffo.

Questi avvicendamenti sono nella logica di stimolare la partecipazione dei colleghi più giovani alla gestione delle attività della nostra comunità di luce di sincrotrone e di neutroni.

Ringrazio i colleghi Camilloni e Caciuffo per l'attività svolta nel Comitato di Redazione e mi auguro che la loro collaborazione con il Notiziario continui, sia con articoli scientifici che di rassegna e riflessione sulla politica italiana in questo campo.

Auguro ai colleghi Avaldi e Wanderlingh un ottimo lavoro nel nuovo Comitato di Redazione in cui, sono sicuro, entrambi apporteranno sia competenza scientifica che entusiasmo giovanile.

Venendo poi alla situazione internazionale voglio rendere noto alla nostra comunità che il prof.

Marcello Fontanesi è stato nominato rappresentante del CNR nell'ESS R&D Council, e la prof.ssa Carla Andreani suo advisor. Le linee di sviluppo dell'ESS R&D Council sono riportate in un articolo in questo numero. Voglio concludere questo Editoriale esprimendo a nome di tutta la nostra comunità, i più vivi ringraziamenti al Comitato Scienze Fisica ed in particolare al suo Presidente, prof. Marcello Fontanesi, nel momento in cui il Comitato di Fisica è stato sciolto a causa del rinnovo della struttura del CNR. Questa gratitudine non è solo dovuta all'aiuto da loro dato ai singoli gruppi di ricerca italiani operanti nel campo della luce di sincrotrone e della spettroscopia neutronica, quanto soprattutto per aver spinto il CNR a stabilire accordi internazionali che hanno consentito ai ricercatori italiani l'accesso a grandi sorgenti neutroniche internazionali. Senza il prestigio e l'infaticabile attività del prof. Fontanesi, questa realizzazione sarebbe stata impossibile. Ringraziamo infine il prof. Marcello Fontanesi per il lavoro che svolgerà nell'ESS R&D Council e che sarà a nostro parere di grande valore per il prestigio italiano.

A

tarting from this issue Notiziario will have a new graphical style, a wider Editorial Committee which will now include Dr. M. Apice and Dr. P. Bosi of CNR and will benefit from the scientific expertise of Dr. L. Avaldi and Dr. U. Wanderlingh, who substitute, respectively, Dr. R. Camilloni and Prof. R. Caciuffo.

The rationale behind these substitutions is to encourage the participation of younger colleagues to the organizational activities of the synchrotron and neutron communities. I would like to thank Dr. Camilloni and Prof. Caciuffo for their work in the Editorial Committee and I hope that their collaboration with Notiziario will continue, both with scientific papers and with contributions on the organizational aspects of Italian research. I wish Dr. Avaldi and Dr. Wanderlingh every success in their new function in the Editorial Committee, to which I am sure they will contribute with their scientific competence and enthusiasm.

With regard to the international situation I would like to announce that Prof. M. Fontanesi has been nominated representative of CNR in the R&D Council of the ESS with Prof. C. Andreani as his Advisor.

The lines of action of the ESS R&D Council are outlined in a paper of this issue.

I would like to take this opportunity to thank, on behalf of all our community, the Comitato Fisica of CNR, and in particular its President, Prof. M. Fontanesi, at this time when, due to the restructuring of CNR, this committee ends its activity. Not only has Comitato Fisica supported individual groups working with synchrotron radiation and neutron sources but it has led CNR to sign international agreements which have opened access for the Italian source to important international neutron sources. Without the prestige and indefatigable activity of Prof. Fontanesi this would have been impossible. Lastly, I would like to thank Prof.

Fontanesi for his prestigious contribution to the ESS Scientific Council, which will be of great value for Italian community.

S

F.P. Ricci

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

2

•

Vol. 4 n. 1 Giugno 1999

F.P. Ricci

RASSEGNA SCIENTIFICA

Articolo ricevuto in redazione nel mese di Marzo 1999

RESONANT MAGNETIC SCATTERING OF POLARIZED SOFT X-RAYS M. Sacchi LURE, BP 34, Centre Universitaire Paris-Sud 91898 Orsay (France)

Abstract Resonant elastic scattering of polarized x-rays is a powerful technique for the study of the magnetic properties of solids. Its recent extension to the soft x-ray energy range has been driven by applications in the field of artificially structured magnetic devices, like multilayers and superlattices. This article reviews recent elastic scattering experiments using synchrotron radiation, performed at the 2p core resonances of transition metals in solids, thin films and ordered multilayers.

Introduction Elastic scattering (Bragg diffraction, specular reflecivity, or diffuse scattering) of X-rays from solids has been for decades the technique of choice for studying structural order at both atomic and microscopic levels [1]. In a standard approach, one works at a given photon energy, introducing the optical constants of the sample as fixed parameters, and is interested mainly in the correlation between sample structure and angular distribution of the scattered intensity.

A different point of view consists in the introduction of the photon energy as a variable of the experiment, in order to obtain information about both optical response and structure of the target. Varying the photon energy across core resonances of one of the sample components leads to the so called anomalous diffraction [2] or, as defined here, resonant scattering. The important point to underline is that now we are dealing with a spectroscopic technique, whose results can yield information about the ground and excited electronic configurations of a given element in the sample, in addition to average charge density distributions.

Elastic X-ray scattering is also influenced by the magnetic order of the sample, as theory [3,4] and experiments [5] pointed out long ago. The applications to magnetic studies, though, have been rather sparse, mainly because of the very weak intensity that characterizes magnetic Xray scattering [6-9].

About ten years ago, first experiments were performed combining the Resonant and Magnetic labels for X-ray Scattering (XRMS), showing that the weak magnetic signals can be boosted of several orders of magnitude by the resonant term when the photon energy is tuned at a core excitation [10]. The comparison between resonant

and non-resonant terms in the magnetic scattering amplitude has been discussed, e.g., in ref. [11].

Experiments that take advantage of the resonant enhancement of the magnetic scattering have been performed mainly under Bragg diffraction conditions [12-14], which implies, for typical lattice spacings of a few Å, the use of hard X-rays (hν ≥ 3 keV, or λ £ 4 Å). Resonant magnetooptic effects, though, are stronger when the magnetic orbitals are directly involved in dipolar excitations, like (3p,2p → 3d) for the transition metals of the first row, or (4d,3d → 4f) in rare-earths. These resonances are all located in the so-called soft X-ray range, whose extension can be roughly identified with the energy interval from 40 to 2000 eV, or 6 to 300 Å. As a drawback, these long wavelengths prevent a general application of Bragg scattering to the study of single crystals. Therefore, resonant magnetic scattering of polarized soft X-rays has been developed along two main lines: i) the study of Bragg diffraction from artificially structured materials of appropriate periodicity (e.g. multilayers and superlattices), and ii) the analysis of specular reflectivity, which contains analogous information but has no constraints related to a lattice spacing. In the following, examples will be given of both approaches.

Basic formulae for resonant scattering Clear and complete formulations of resonant magnetic scattering have been given by several authors [15,16], and only a few useful formulae in their simplified expression will be reported in the following.

The scattering amplitude is related to the real dispersive (1-δ) and imaginary absorptive (β) parts of the frequency dependent index of refraction of the material. Following the notation of Hannon et al. [15], the core resonance modifies both real and imaginary parts of the index, contributing an additional component to the scattering amplitude that, in the dipole approximation, can be written as the sum of three terms of order 0, 1 and 2 in the magnetic field unit vector m :

fres = e *f ⋅ ei (F+1 + F–1) – i (e *f × ei) ⋅ m (F+1 – F–1) + (e *f ⋅ m) (ei ⋅ m) (2F0 + F+1 – F–1)

Vol. 4 n. 1 Giugno 1999

•

(1)

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

3

RASSEGNA SCIENTIFICA

where ei and ef are the polarization vectors of the incoming and outgoing photons, and F∆M is related to the oscillator strength for a transition with a change ∆M(=+1,0,-1) in the magnetic quantum number

following we will assume that the magnetic ordering induces circular but not linear dichroism in absorption, i.e. that the dielectric tensor has equivalent diagonal elements. This simplifying hypothesis is acceptable for 3dTM in their metallic form, but has no general value.

2 ′

′

J M | Q1,p | J M F∆M ∝

EJ ′ – EJ – hω – i Γ 2

(2)

| JM > and | J’M’ > are the initial and final states of energy EJ and EJ’ for the transition, and Q1,p is the electric dipole operator, where p = ∆M = M’ - M is defined by the polarization of the light.

The first term in eq.1, independent of the magnetization, gives the resonant contribution to the charge scattering amplitude. The second term, linear in m, consists of a polarization dependent geometrical factor (e*f ¥ ei ) • m multiplying the difference between the resonant optical response of the medium for opposite magnetization-to-helicity orientations. The term (F+1 - F-1) is directly related to the off-diagonal terms of the dielectric tensor and, in turn, to the observation of dichroism in the absorption of circularly polarized X-rays. In a scattering experiment, though, (e*f × ei ) • m can be non-vanishing for both linear and circular polarization of the light. For a sample magnetized in its surface plane, we can consider for instance the two following experimental geometries: a) circularly polarized photons, magnetic field in the scattering plane, b) linearly p-polarized photons, magnetic field normal to the scattering plane. In both geometries, the inversion of the magnetization produces a mirror image, with the mirror plane perpendicular to the magnetization axis. And in both cases the resonant scattering amplitude contains the (F+1 - F-1) term, with an angular dependence cosθ for circular polarization and sin2θ for linear polarization (2θ being the scattering angle).

The third term in eq. 1 is related to the existence of linear dichroism in absorption, i.e. to the difference between spectra mesured with the electric vector of the linearly polarized light either parallel or perpendicular to the quantization axis. If we consider for instance the scattering of s-polarized light from an in-plane magnetized sample, we can orient the magnetization either parallel to the electric vector (normal to the scattering plane) or perpendicular to it (in the scattering plane). The former geometry will give fres = 2F0, the latter fres = F+1 + F-1.

The variation is directly related to linear dichroism in absorption and, in turn, to the fact that the diagonal elements of the dielectric tensor are non equivalent. In the

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

4

•

A Computational method The magnetic circular dichroism in the X-ray absorption (XMCD) at a core resonance can be represented by a polarization dependent dissipative index β± The dispersive part δ±, related to β± through the Kramers-Kronig (KK) transformations, will also depend on the polarization, and the complex optical index will be written as n± = 1 δ± + β±. Having defined ε± = (n± )2 and assuming the absence of linear dichroism, the dielectric tensor ε has the following elements:

ε y ′z ′ = – εz ′y ′ = – 2i (ε + – ε –) ε x ′x ′ = ε y ′y ′ = εz ′z ′ = 12 (ε + + ε –) ε x ′z ′ = εz ′x ′ = ε x ′y ′ = ε y ′x ′ = 0 where x’ identifies the magnetization axis. We evaluate the scattered intensity in the specular direction by solving Maxwell’s equations in the form: k2x 0 k xk z det k ε – (k + k )I + 0 0 0 k xk z 0 k 2x 2 0

2 x

2 z

=0

(3) where the coordinate system is defined by x, the intersection between sample surface and scattering plane, y, normal to the scattering plane, and z, normal to the sample surface. In eq. 3, I is the identity matrix, ε the dielectric tensor defined above written in the xyz coordinate system, θ is the grazing angle, kx (= k0 Cosθ) and kz are the x and z components of the wavevector in the material, and k0 is the modulus of the wavevector in vacuum.

Eq. 3 gives a polynomial of fourth degree in kz, whose four roots can be split in two couples of solutions according to the sign of the real part of kz: positive for the two waves propagating in the negative z direction (into the sample), negative for the two along the positive z direction (towards vacuum). The electromagnetic field within the sample is written as a linear combination of the four waves with coefficients determined by imposing the condition of continuity of the parallel components of E and B at the vacuum interface. Finally, the reflectivity is obtained by imposing the condition of outgoing waves in the substrate.

This description refers to the case of a semi-infinite solid,

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

element, from experimental absorption curves measured for parallel and antiparallel alignment of photon helicity and sample magnetization. The imaginary part β± of the helicity dependent refractive index can be obtained directly from the absorption coefficient (β± =(λ/4π) µ±), while the real part δ± is obtained from the Kramers-Kronig transformation of β±. The optical constants for Fe, Ni and Cu are reported in Fig. 1. Before applying KK transformations, the experimental absorption curves were scaled to match off-resonance atomic values [18].

Fig. 1. Optical constants of Fe, Ni and Cu over the energy range 600-1000 eV used for reflectivity calculations. Top panels present the magnetization averaged decrement to the real part (δ) and imaginary part (β) of the index of refraction, used to construct the diagonal elements of the dielectric tensor. Bottom panels report the corresponding magnetization dependent contributions for Fe and Ni, used to construct the off-diagonal elements of the dielectric tensor.

Resonant magnetic scattering from a semi-infinite solid A semi-infinite magnetic solid is a simple test case for the adopted description of XRMS and for the corresponding model calculations, since the presence of only one interface greatly reduces the number of parameters. Fig. 2 reports selected spectra from a study of the magnetization dependent scattering at the 2p edges of Ni in a Ni(110) single crystal [19,20], performed at beamline SA8 of SuperACO (LURE). Experimental results are reported

Fig. 2. Experimental (symbols) and calculated (lines) asymmetry curves for the magnetic scattering from a Ni(110) crystal. The photon energy range includes the 2p resonances of Ni. Curves in a) and b) refer to elliptically polarized light. The magnetic field is reversed along the [1 1 1] direction in the sample surface and in the scattering plane.

Panels c) and d) refer to linearly p-polarized light. The field is still along the [1 1 1] direction, but normal to the scattering plane. 2θ is the scattering angle.

a model that applies, e.g., to the reflectivity from a bulk crystal. The same method has been developed and integrated in a computer code for a general stacking along the z direction of (magnetic) layers of finite thickness [17].

The examples of reflectivity calculations reported in the following are based on dielectric tensors built, for each

in the form of asymmetry ratio curves, i.e. the difference between spectra measured for opposite magnetizations divided by their sum. Fig. 2 includes the results of experiments performed in both a) and b) geometries discussed in Sec.II, i.e. using elliptical (Fig. 2a,b) or linear (Fig. 2c,d) polarization of the light. Full lines are calculated

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

5

RASSEGNA SCIENTIFICA

advantage of the angular range below 2 degrees.

This behaviour can be qualitatively understood remembering that, given a certain grazing angle θ, we can define a critical value nc of the real part of the index of refraction for having total external reflection. Since n = 1 for vacuum, nc is always lower than 1 and decreases with θ. Around 2.3 degrees, for instance, nc is about 0.9992 : according to Fig.1 total external reflection in the region of the L3 edge of Ni can only occur above 854 eV, i.e. above the absorption maximum. Hence at this angle the reflectivity will be relatively suppressed below 854 eV and enhanced between 854 and 860 eV. The same qualitative argument holds for explaining the enhanced magnetic effects in the satellites energy region over a certain angular range.

Fig. 3. Magnetic part of the reflectivity of elliptically polarized photons from a Ni(110) crystal as a function of the scattering angle 2θ. The curves are arbitrarily scaled for a better comparison.

according to the method outlined in Sec.IIA, using the optical constants of Ni given in Fig.1. Fig. 2 shows experimentally that the same magnetic properties, expressed in eq.1 through the term (F+1 - F-1) and measured in absorption by circular dichroism, can be investigated by XRMS using, on the same footing, either circular or linear polarization.

Fig. 3 compares the difference in reflectivity curves obtained at five different angles by switching the orientation of the magnetization (elliptical polarization of the light). The presence of interference terms containing an angular dependence makes the reflected intensity a very sensitive probe of the constituting transitions forming the 2p edges. One can remark in Fig.3 the existence of a sort of optimized angular range over which specific structures are emphasized. For instance, structure B of XMCD in absorption [21] is usually no more than a shoulder to the main peak, but in reflectivity and for θ values around 2 degrees it can represent the strongest and sharpest peak of the dichroism spectrum. On the other hand, the analysis of the magnetic effects in the pre-edge region (840-850 eV), which are related to the magnetic part of δ (see Fig.1), can take

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

6

•

Ni layers on Cu Magnetism of surfaces and interfaces is a field of research that has taken full advantage of the possibilities offered by new synchrotron radiation sources in terms of photon beams with high energy resolution and flux, and selectable polarization state. In view of the interesting characteristics of resonant X-ray scattering for magnetic studies, we have explored the extension of this technique to surface science. A first example concerned the investigation of the magnetic properties of clean uncapped layers of Ni grown in situ on a Cu(110) single crystal [22,23]. The samples were prepared in the UHV chamber of the dutch-french beamline SA8 of SuperACO (LURE).

Fig. 4a compares experimental and calculated reflectivity curves at 2θ = 2.2 degrees for a single atomic layer of Ni deposited on Cu(110). The optical constants used in the calculation are the same as in Fig.1, i.e. the same magne-

Fig. 4. Comparison between the experimental and calculated reflectivity curves (left panel) and corresponding magnetic signals (right panels) for 1 ML Ni / Cu(110). Right panel reports the asymmetry ratio, defined as the difference between reflectivity curves for opposite magnetizations divided by their sum.

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Fig. 5. Same as Fig. 4, for 10 ML Ni / Cu(100)

tization dependence as for bulk fcc nickel is assumed.

The polarization of the light is elliptical (with ~ 40% circular polarization rate), giving, when the magnetic field is switched along the [1 1 1] in-plane direction, the asymmetry ratio reported in Fig. 4b, full line. In the experiment, though, no magnetization dependence is observed (within Âą 0.2% noise level), indicating that, at room temperature, the average in-plane magnetic moment per Ni atom in a monolayer is vanishingly small. The same holds for Ni coverages up to 6 ML, and, as proposed before for the growth on other copper surfaces, we interpret our results in terms of a reduced Curie temperature of thin layers [24,25]. When the Ni thickness is increased to 10 ML, both reflectivity and asymmetry ratio curves are well reproduced by calculations based on Ni fcc optical constants (Fig. 5). Contrary to previous results obtained on Ni/Cu(001) [24-29], we observe a complete remanent alignment of the Ni magnetic moment in the surface plane also for layers thicker than 10 ML. For a meaningful comparison, though, one should remember that the [111] directions are the easy axes of magnetization in bulk fcc Ni, and it is along the [1 1 1] direction laying in the surface plane that we apply the external magnetic field to the Ni films in our experiment. The analysis of resonant scattering, together with LEED observations, absorption spectra and hysteresis curves, suggests that the structure and electronic properties of Ni are not strongly modified in our films compared to the bulk. We conclude that, for the (110) growth, the magnetic anisotropies characteristic of the bulk are recovered, at least in terms of easy axes, already for very thin layers. This favours a magnetic ordering of Ni/Cu(110) films along the [111] easy axes giving rise, in particular, to in-plane magnetization over the thickness range 10-30 ML that we investigated, a situation different from what observed in the Ni/Cu(100) system.

The high sensitivity illustrated by the results reported in Fig. 4 is not limited to surfaces: thanks to the element selectivity and the large field of view, it applies also to buried interfaces, as well as diluted systems [22]. An example is given in Fig. 6: the reflectivity is measured at fixed photon energy (L3 edge of Ni) and scattering angle (3.8 degrees) as a function of the applied magnetic field. The sample is a 10 ML Ni/Cu(110) layer, capped by an extra 20 ML Fe film. Even in presence of other magnetic elements and of strong and time dependent fields, we can still probe the magnetic behaviour of the buried Ni layer.

Our work on Ni layers shows that XRMS has the potential for being considered as a surface science technique.

Data for 1 ML were collected in about one hour, i.e. a duration perfectly compatible even with the strict constraints on time scales in surface physics. Moreover, to fully appreciate the sensitivity of the technique for surface magnetism, one should also take into account that the new generation of storage rings and beamlines can easily deliver, with respect to the conditions for this experiment, photon beams with approximately three to four orders of magnitude higher flux and double circular polarization rate. Compared to other spectroscopies, XRMS has a complementary sensitivity to morphology that can add information about, e.g., film thickness, surface roughness and interdiffusion processes.

Resonant scattering from magnetic superlattices The magnetic properties of artificially structured devices (thin layers, superlattices or arrays of microdots) can be

Fig. 6. Reflected intensity as a function of the applied field for a 20 ML Fe/10 ML Ni/Cu(110) sample. The photon energy corresponds to the L3 edge of Ni.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

7

RASSEGNA SCIENTIFICA

tailored to specific needs using appropriate elements, growth modes or periodic structures. Metallic multilayers, consisting of magnetic and non magnetic elements, are preminent among such devices [30], featuring oscillating magnetic coupling between ferromagnetic layers and giant magnetoresistance effects. Not only are these systems under study by theoreticians and experimentalists [31,32], but they are also employed in the development of new magnetic recording devices [33].

Fe/V superlattices present in general a whole variety of magnetic behaviour, depending on the thickness values tFe and tV, and on the crystal orientation. For the (001) orientation, ferromagnetic coupling between successive Fe layers was reported [34] for tV up to about 12 ML (1ML = 1 Å) while antiferromagnetic coupling was observed [35] for thin Fe layers (about 3 ML) when the vanadium spacers exceed 12 ML. In both cases, the samples had two in-plane easy axes along the [100] and [010] directions. Coupling influences the global properties of the superlattice, particularly in terms of magnetoresistance. When grown in the (110) orientation, the Fe/V system exhibits uniaxial in-plane anisotropy and a magnetic coupling between Fe layers that still depends on the V thickness [36]. Until recently, the magnetic properties of V in these layers were subject to controversy. Va-

nadium is known to acquire a magnetic moment when in contact with Fe, e.g. when diluted in an Fe matrix [37] or deposited on an Fe crystal [38]. In superlattices and multilayers it is difficult to detect and quantify the magnetic moment of vanadium which is small, in presence of large amounts of Fe, and element selective techniques based on electron analysis are unable to probe buried interfaces. This explains why recent experiments have been performed taking advantage of the element selectivity and relatively large probing depth of magnetic dichroism in the absorption of circularly polarised soft x-rays [39,40]. A recent paper [41] also addressed the interesting and indeed crucial problem of interdiffusion at the Fe/V interfaces and of its influence on the induced magnetism in vanadium, comparing experimental data with model calculations.

In this section we will concentrate on Fe/V superlattices [42] of the type (5 ML Fe / n ML V) m grown on MgO(001) [35,36]. Three samples will be considered, with (n = 1, m = 80, 2d = 16.5 Å), (n = 2, m = 60, 2d = 22.2 Å) and (n = 5, m = 40, 2d = 30.6 Å). The easy axes of magnetization were determined for each sample to be oriented along the [110] axes of the substrate, i.e., [100] in-plane directions of the superlattice.

X-ray scattering and absorption measurements were

Fig. 7. Reflectivity spectra across the V and Fe 2p resonances at a scattering angle of 20 degrees (elliptically polarized light). The sample is a (1 ML V / 5 ML Fe)80 superlattice, magnetized along a [100] direction in the sample surface and in the scattering plane. Full lines are magnetization averaged curves, dashed lines are difference curves. Right panel reports the corresponding asymmetry curves, on an arbitrary energy scale that matches the spin-orbit splitting of the 2p core hole for V and Fe.

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

8

•

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

performed on the soft x-ray metrology beamline 6.3.2 at ALS (Berkeley) [43], using out-of-plane elliptically polarized radiation. The samples, mounted in a θ/2θ reflectometer, were magnetized along an easy axis, parallel to its surface and in the scattering plane. Absorption spectra were recorded by measuring the drain current from the sample. We measured energy dependent resonant scattering at both the Fe and V 2p edges for different scattering angles 2θ up to 30 degrees. θ/2θ angular scans were also performed for photon energies across the 2p resonances, including resonant Bragg diffraction spectra.

All resonant spectra showed magnetization dependence.

Fig. 7 shows for example the energy dependent reflectivity curves for the n = 1 sample : the left panels refer to the vanadium and iron 2p edges, measured at θ = 10 degrees. Full and dashed lines are the sum and the difference of reflectivity curves obtained for opposite magnetizations. In the right panel, the corresponding asymmetry ratio curves are reported, drawn on different energy scales so as to give a visual match of the 2p spin-orbit separations of Fe and V. This comparison suggests an antiparallel alignment between Fe and V magnetic moments, though such a conclusion can be reached safely only after a thorough analysis of the complete set of data, including angular dependence. Experimental data have been analysed following the method outlined before, optimizing the offdiagonal terms of the dielectric tensor on the experimental asymmetry ratio curves. This procedure allows to apply sum rules to the polarization dependent scattering in order to obtain a quantitative estimate of the element specific magnetic moments, just as in absorption spectroscopy [44,45]. For each sample, the fit on the ensemble of the reflectivity curves at the Fe 2p resonances, including Bragg diffraction spectra, gave values of the magnetic moment per Fe atom between 2.1 and 2.3 µB, i.e. there is no measurable change in the Fe average magnetic moment as a function of V thickness and with respect to bulk Fe. This also implies the ferromagnetic alignement of all the Fe layers. By analysing the data at the V 2p edges, and with the limitations associated to sum rules [42], we could determine separately the orbital and spin moments carried by the V 3d electrons. For the 1 ML V sample, we obtained <lz > =- 0.08 µB < sz> = 0.37 µB These values are reduced to 60% and then to 40% when n goes from 1 to 2 and then to 5, so that we finally obtain: < µz >V1 ML = 0.66 ± 0.20 µB

Fig. 8. Calculated magnetic asymmetry ratio in the resonant Bragg scattering from a (5 ML V / 5 ML Fe)40 superlattice. The four curves a) to d) refer to different magnetization profiles within each V sheet of five atomic layers (see bar diagrams). The same average magnetic moment per V atom (0.26 µB) is assumed.

< µz >V2 ML = 0.40 ± 0.12 µB < µz >V5 ML = 0.26 ± 0.08 µB These values represent an average per vanadium atom.

It is interesting to consider how the magnetic moment is distributed within, say, the five atomic planes forming each vanadium layer in the 5 ML Fe / 5 ML V sample.

An average value of 0.26 µB can be represented by the four diagrams in Fig. 8, but also by many others. Resonant scattering is an ideal technique for studying magnetization depth profiles, as it is sensitive to changes in the optical constants along the sample normal, and magnetic properties affect these optical constants [46]. Reflectivity curves calculated over the 5-15 degs. range assuming the four magnetic profiles of Fig.8 were not significantly different for the 5 ML Fe / 5 ML V sample. On the other hand, curves in Fig.8, calculated under resonant Bragg scattering conditions, show a variation of a factor 2 in the asymmetry ratio according to the specific magnetization profile. This looks like a promising direction for future developments in soft x-ray scattering experiments applied to the study of magnetic multilayers.

Optical constants at core resonances For decades, x-ray absorption spectra have been used to determine local structure and chemical environment element-specifically [47]. The use of polarized x-rays has widened this field to include the study of anisotropic systems, and in particular magnetic materials [48], with, at the forefront, interest in tailor-made magnetic multilayers. Modeling these requires simultaneously

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

9

RASSEGNA SCIENTIFICA

taking into account the spin polarization of the conduction electrons, changes in the dielectric constant at interfaces, and interlayer magnetic coupling. Comparison with experiment, at this stage of the development of first principles calculations, is imperative. The energy dependence of the optical constants at a resonant excitation represents a privileged standpoint for such a comparison [49], yet in the x-ray region only the imaginary part β (i.e., absorption) of the refractive index is used.

Model calculations, on the other hand, may deal with the real and imaginary parts on the same footing. The problem is simply an almost complete lack of experimental data even for standard materials, let alone for

fects are stronger when the core excitations produce dipolar transitions to a final state that directly involves the magnetic orbitals, which implies that the most interesting resonances for studying magneto-optics effects are all located in the soft x-ray region. This means that metallic multilayers with artificial periodic structures with 2d spacings in the range 10 to 200 Å are ideally suited to performing XRMS with soft x-rays [53-55].

In this section we show that the real part δ of the index of Fe across its 2p resonances (dominated by 2p → 3d transitions) may be determined by analyzing the Bragg diffraction from a metallic superlattice constituted of alternate Fe and V layers deposited on MgO [55]. For this

Fig. 9. θ/2θ scans around the first order Bragg peak of a (5 ML Fe / 5 ML V)40 superlattice, for photon energies across the 2p edge of Fe. Panels a) and b) are for parallel and antiparallel magnetization/helicity orientations. Panel c) reports the corresponding difference curves.

multilayer structures. While it is true that the real part of the index provides similar information to the imaginary part, it contributes a further set of independent data for refining theoretical models [50]. It has even been shown that interference between real and imaginary parts (as in resonant reflectivity) can facilitate the investigation of spectroscopic features, such as satellites, that are weak but of major importance for understanding some ground state properties [20,51].

Most of the few direct experimental determinations of δ at resonance (i.e., without using KK relations) have been performed on crystals by measuring the Bragg peak displacement when the photon energy is scanned through an absorption edge [52]. The Bragg law imposes the relation between the photon wavelength λ and the crystal spacing 2d, and this explains that XRMS experiments have been performed mainly at high energies. However, as pointed out in the introduction, x-ray magneto-optic ef-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

10

•

experiment we chose a system where Fe is known to have closely the same magnetic properties as in bulk form [41]. A future step will be to correlate magneto-optics constants and magnetic properties in multilayer devices presenting specific magnetic anisotropies as a function of environment (e.g. metastable phases stabilized by epitaxial growth [30]).

Resonant scattering spectra are reported in Figs. 9(a,b) as a function of θ and photon energy for the two magnetization to helicity orientations. These are raw data, simply normalized to the incoming photon flux. Fig. 9c shows the difference between curves (a) and (b). When approaching the L3 and, to a lesser extent, the L2 edges of Fe, spectra show strong variations that depend on the magnetization of the sample. To better appreciate this energy dependence, Fig. 10a shows the displacement of the maximum of the Bragg peak position θ as a function of photon energy which is partly related to the change in

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

wavelength. Data in Fig. 10a, though, show large deviations from a simple arcsine behavior. These deviations are related to the energy dependence of δ which is something we will bring out in what follows. It is apparent, moreover, that there is a difference in the curves for opposite magnetizations, giving us access to the magnetic part of δ for Fe. A simple way of obtaining δ from the data in Fig. 10a is to consider the average decrement given by

δ = γδ + (1 – γ)δV = (Sin θ M – λ ) Sin θ M 2d

(4)

Fig. 10. a) Angular position of the Bragg peak maximum as a function of the photon energy. Continuous and dashed lines refer to topposite magnetizations of the sample along a [100] direction. b) Imaginary part (β) and decrement to the real part (δ) of the Fe refractive index. Plus and minus signs refer to opposite magnetization orientations.

where the subscript V relates to the (non resonant) optical constants of vanadium. Eq. 4 accounts for refraction effects, but is only valid for a real index. A further correction term should be introduced to account for the effect of absorption on the Bragg peak position. Following the model of Rosenbluth and Lee [56] one has:

δ = (Sin θ M – λ ) Sin θ M – δV /D + δV 2d

(5)

where

D=γ–

(β – β V) Sin 2(πγ) π 2[γβ + (1 – γ)β V

(6)

To determine δ from Eqs. 5 and 6 we have scaled the experimental absorption spectra to the calculated values over the 650-680 eV and 750-780 eV energy ranges, i.e., below and above the 2p resonances. Absorption correction is found to have little, but not negligible, effect on δ for Fe. Finally we have corrected the δ and β curves for the incomplete polarization of the light, and the β curves also for the incomplete alignment between the photon propagation axis and the magnetization. The results are shown in Fig. 10b: these four curves, according to the relations given in Sec.IIa, are all we need to construct the complete dielectric tensor ε, and to obtain a totally experimental determination of the optical constants of Fe over the energy range covering the 2p resonances.

Resonant magnetic scattering from a superlattice makes it possible to obtain the real part of the refractive index without using KK transformations. In a previous work, Kortright et. al. [57] used another method to the same end, i.e., they measured the Faraday rotation of the polarization upon transmission through a thin magnetized Fe layer. Both techniques present some experimental difficulties and require a certain degree of data reduction, but we want to stress that, even at this stage where they are not yet optimized, both give access to information difficult to attain with the same reliability by other methods.

As far as comparisons with the previous results are possible, we can only say that we observe a much larger magnetic component of δ than reported in Ref. [57]. From Fig. 10b one obtains the magnetic part of δ, which, if expressed in electrons as in Ref. [57], gives values as large as 20 electrons around 707-708 eV, i.e., at the steep rise of the L3 edge. Such a high value is consistent with the KK transformation of the dichroism in the imaginary part β.

This application of XMRS to the region of the Fe 2p edges especially demonstrates feasibility and points the way to future developments where the advantages of a photon-in/photon-out experiment and a large probing depth are at a prime. We consider that the experimental determination of the dielectric tensor including its offdiagonal terms is an important step towards a better refinement of theoretical and computational models that describe the electronic and magnetic ground state proper-

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

11

RASSEGNA SCIENTIFICA

ties of materials. To this end, it is essential to develop several independent experimental methods in addition to the standard KK approach. Polarization rotation (upon transmission or reflection), interferometry and resonant magnetic scattering constitute together a sound basis of complementary techniques for future work.

and the energy separation of resonant features that permits element and even chemical selectivity. On the other hand, soft x-ray wavelengths match the typical 2d spacings of magnetic multilayers and superlattices, and are ideally suited to performing resonant Bragg diffraction experiments on these systems.

Conclusions Examples of recent applications of x-ray scattering to magnetism have been reported, with particular attention to experiments performed at core excitations located in the soft x-ray range. To summarize, one may underline the importance of each single word introduced in the title of this review: Resonant Magnetic Scattering of Polarized Soft X-rays. X-ray Scattering is the well established base of these recent developments. It is a photon-in/photon-out technique, well adapted to working in presence of strong magnetic fields (see, e.g., Fig. 6), it has a large probing depth for studying bulk properties and buried layers, and it has a high sensitivity to structural parameters.

The importance of the Resonant character of these experiments is at least twofold. First, it gives the technique a highly valuable element selectivity (see Fig. 5), and it turns it effectively into a spectroscopy. On the other hand, it makes it possible to increase the (magnetic) signals of orders of magnitude, leading to a much higher sensitivity to small amounts of material (see Fig. 4). One should also mention that angle and energy dependent resonant scattering spectra contain a large amount of experimental information that help to refine parametrized model calculations.

The use of Polarized x-rays imposes a defined symmetry to the probe. This allows one to analyse the presence of a specific symmetry in the sample under investigation, and eventually to quantify the physical parameters relating to it. Polarization dependent x-ray scattering under resonant conditions becomes just as powerful as x-ray dichroism in absorption. Moreover, the same fundamental properties related to Magnetic ordering can be investigated using either linearly or circularly polarized photons, on the same footing (see Fig. 2).

Finally, the choice of the Soft X-ray range for magnetic studies can be supported by two kinds of remark. On the one hand, the dipolar transitions directly involving the relevant orbitals for the most common magnetic elements are all located in the 40-1600 eV range. For comparison, magnetic effects just as large can be obtained in the visible range, but the element selectivity is lost to a large extent. On the other hand, hard x-rays easily provide element selectivity, but magneto-optics effects are reduced of one or two orders of magnitude. In the soft xray range we have an excellent compromise between the intensity of the magneto-optics effects (up to 70-80 %)

Acknowledgements The activity developed at LURE in the field of soft x-ray resonant scattering has involved several people, and many collaborations with external groups. Coryn F. Hague (Lab. Chimie-Physique, Paris) inspired and oriented most of this research. Alessandro Mirone (LURE, Orsay) developed the computational methods and the software for data analysis. I also want to thank Jan Vogel (CNRS, Grenoble), Stefano Iacobucci (CNR, Roma), Flàvio C. Vicentin (LNLS, Campinas), Luca Pasquali (INFM, Modena), Eric. M. Gullikson and James H. Underwood (ALS, Berkeley).

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

12

•

References 1. e.g. R. W. James, The Optical Principles of the Diffraction of X-rays (OxBow, Woodbridge, 1982).

2. e.g., F.J.Walker, E.D.Specht, and R.A.McKee, Phys. Rev. Lett. 67, 2818 (1991).

3. O.Klein and Y.Nishina, Z.Phys. 52, 853 (1929).

4. P.M.Platzman and N.Tzoar, Phys.Rev. B 2, 3556 (1970).

5. F.deBergevin and M.Brunel, Phys.Lett. 39A, 141 (1972).

6. F.deBergevin and M.Brunel, Acta Crystallog. Sect. A 37, 314 (1981).

7. D.Gibbs, D.E.Moncton, K.L.D'Amico, J.Bohr, and B.Grier, Phys. Rev. Lett. 55, 234 (1985).

8. C.Vettier, D.B.McWhan, E.M.Gyorgy, J.Kwo, B.M.Buntschuh, and B.W.Batterman, Phys. Rev. Lett. 56, 757 (1986).

9. V.Fernandez, C.Vettier, F.deBergevin, C.Giles, and W.Neubeck, Phys.Rev. B 57, 7870 (1998).

10. see, e.g., D.B.McWhan, J.Synchrotron Radiat. 1, 83 (1994) and references therein.

11. D.Gibbs, G.Grübel, D.R.Harshman, E.D.Isaacs, D.B.McWhan, D.Mills, and C.Vettier, Phys. Rev. B 43, 5663 (1991).

12. D. Gibbs, D. R. Harshman, E. D. Isaacs, D. B. McWhan, D.Mills, and C. Vettier, Phys. Rev. Lett. 61, 1241 (1988).

13. C. Tang, W. G. Stirling, G. H. Lander, D. Gibbs, W.Herzog, P. Carra, B. T. Thole, K. Mattenberger, and O. Vogt, Phys. Rev. B 46, 5287 (1992).

14. J.P.Hill, C.-C.Kao, and D.F.McMorrow, Phys. Rev. B 55, R8662 (1997).

15. J.P.Hannon, G.T.Trammell, M.Blume and D.Gibbs, Phys. Rev. Lett. 61, 1245 (1988).

16. M.Blume and D.Gibbs, Phys.Rev. B 37, 1779 (1988).

17. A.Mirone, Computer code ONDA (LURE, 1998), unpublished.

18. B.L.Henke, E.M.Gullikson and J.C.Davis, Atomic Data and Nuclear Data Tables 54, 181 (1993).

19. M.Sacchi, J.Vogel and S.Iacobucci, J.Mag.Mag.Mater. 147, L11 (1995).

20. M.Sacchi and A.Mirone, Phys.Rev. B 57, 8408 (1998).

21. C.T.Chen, F.Sette, Y.Ma, and S.Modesti, Phys. Rev. B 41, 9766 (1990).

22. M.Sacchi, Surf.Rev.Lett. 4, 343 (1997).

23. M.Sacchi, A.Mirone, and S.Iacobucci, unpublished 24. W.L.O'Brien and B.P.Tonner, Phys.Rev. B 49, 15370 (1994).

25. W.L.O 'Brien, T.Droubay, and B.P.Tonner, Phys.Rev. B 54, 9297 (1996).

26. R.Naik, C.Kota, J.S.Payson, and G.L.Dunifer, Phys.Rev. B 48, 1008 (1993).

27. R.Jungblut, M.T.Johnson, J.aan de Stegge, A.Reinders, and F.J.A.den Broeder, J.Appl.Phys. 75, 6424 (1994).

28. G.Bochi, C.A.Ballentine, H.E.Inglefield, C.V.Thompson, and R.C.O'Handley, Phys.Rev. B 53, R1729 (1996).

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

29. J.Lee, G.Lauhoff, M.Tselepi, S.Hope, P.Rosenbusch, J.A.C.Bland, H.A.Dürr, G.van der Laan, J.P.Schillè, and J.A.D.Matthew, Phys.Rev. B 55, 15103 (1997).

30. see for instance the Proceedings of the International Symposium on Magnetic Ultrathin Films, Multilayers and Surfaces, J.Mag.Mag.Mater. 121 (1993).

31. P.Bruno and C.Chappert, Phys.Rev.Lett. 67, 1602 and 2592 (1991).

32. S.Y.Hsu, A.Barthèlèmy, P.Holody, R.Loloee, P.A.Schroeder, and A.Fert, Phys.Rev.Lett. 78, 2652 (1997).

33. G.Prinz, Physics Today 48, 58 (1995).

34. P.Granberg, P.Nordblad, P.Isberg, B.Hjövarsson, and R.Wäppling, Phys.Rev. B 54, 1199 (1996).

35. P. Isberg, B. Hjörvarsson, R. Wäppling, E. B.Svedberg, and L. Hultman, Vacuum 48, 483 (1997).

36. P. Isberg, P.Grandberg, E. B.Svedberg, B. Hjörvarsson, R. Wäppling, and P.Nordblad, Phys. Rev. B 57, 3531 (1998) 37. M.Finazzi, E.Kolb, J.Prieur, C.Boeglin, K.Hrikovini, G.Krill, C.Chappert, and J.-P.Renard, J.Mag.Mag.Mater. 165, 373 (1997).

38. P.Fuchs, K.Totland, and M.Landolt, Phys.Rev.B 53, 9123 (1996).

39. G.R.Harp, S.S.P.Parkin, W.L.O 'Brien, and B.P.Tonner, Phys.Rev. B 51, 3293 (1995).

40. M.Finazzi, F.Yubero, P.Benock, F.Chevrier, K.Hrikovini, F.Ciccacci, and G.Krill, J.Mag.Mag.Mater. 165, 78 (1997).

41. M.M.Schwickert, R.Coehoorn, M.A.Tomaz, E.Mayo, D.Lederman, W.L.O'Brien, B.P.Tonner, T.Lin, and G.R.Harp, Phys.Rev.B 57, (1998).

42. M.Sacchi, A.Mirone, C.F.Hague, L.Pasquali, J.-M.Mariot, P.Isberg, E.M.Gullikson, and J.H.Underwood, unpublished.

43. J. H. Underwood, E. M. Gullikson, M. Koike, P. J.Batson, P. E. Denham, K. D. Frank, R. E. Tackaberry, and W. F. Steele, in Abstracts of the 9th National Conference on Synchrotron Radiation Instrumentation (Argonne National Laboratory, Argonne, IL), Rev. Sci.Instrum. 67, 3343 (1996); J. H. Underwood and E. M. Gullikson, J.Electr.Spectrosc.

92, 265 (1998).

44. B.T.Thole, P.Carra, F.Sette and G. van der Laan, Phys.Rev.Lett. 68, 1943 (1992).

45. P.Carra, B.T.Thole, M.Altarelli and X.Wang, Phys.Rev.Lett. 70, 694 (1993).

46. L.Sève, J.-M.Tonnerre, M.Arend, D.Raoux, W.Felsch, F.Bartolomè, A.Rogalev, J.Goulon, J.F.Bèrar, and C.Gautier, Phys. Rev. B, submitted.

47. See, e.g., B. T. Thole, G. van der Laan, J. C. Fuggle, G. A. Sawatzky, R. C. Karnatak, and J.-M. Esteva, Phys. Rev. B 32, 5107 (1985); F. M. F. de Groot, J. C. Fuggle, B. T. Thole, and G. A. Sawatzky, Phys. Rev. B 42, 5459 (1990).

48. C. T. Chen, N. V. Smith, and F. Sette, Phys. Rev. B 43, 6785 (1991); T. Jo and G. A. Sawatzky, ibid. 8771 (1991); P.Kuiper, B. G. Searle, P. Rudolf, L. H. Tjeng, and C. T. Chen, Phys. Rev. Lett. 70, 1549 (1993); M. Tischer, O. Hjortstam, D. Arvanitis, J. Hunter Dunn, F. May, K. Baberschke, J. Trygg, J. M. Wills, B. Johansson, and O. Eriksson, Phys. Rev. Lett. 75, 1602 (1995).

49. See, e.g., M. Alouani, L. Brey, and N. E. Christensen, Phys.

Rev. B 37, 1167 (1988); Z. H. Levine and D. C. Allan, Phys. Rev. Lett. 63, 1719 (1989).

50. N. Mainkar, D. A. Browne, and J. Callaway, Phys. Rev. B 53, 3692 (1996).

51. C. C. Kao, J. B. Hastings, E. D. Johnson, D. P. Siddons, G.C. Smith, and G. A. Prinz, Phys. Rev. Lett. 65, 373 (1990).

52. J.Vacinovà, J.L.Hodeau, P.Wolfers, J.P.Lauriat, and E.Elkaìm, J.Synchrotron Radiat. 2, 236 (1995).

53. J.-M. Tonnerre, L. Sève, D. Raoux, G. Soulliè, B. Rodmacq, and P. Wolfers, Phys. Rev. Lett. 75, 740 (1995).

54. M. Sacchi, C. F. Hague, E. Gullikson, and J. Underwood, Phys. Rev. B 57, 108 (1998).

55. M.Sacchi, C.F.Hague, L.Pasquali, A.Mirone, J.-M.Mariot, P.Isberg, E.M.Gullikson, and J.H.Underwood, Phys. Rev. Lett. 81, 1521 (1998) 56. A. E. Rosenbluth and P. Lee, Appl. Phys. Lett. 40, 466 (1982).

57. J. B. Kortright, M. Rice, and R. Carr, Phys. Rev. B 51, 10240 (1995).

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

13

RASSEGNA SCIENTIFICA

Articolo ricevuto in redazione nel mese di Ottobre 1998

THE SILICON/GADOLINIUM DETECTOR FOR THERMAL NEUTRONS: CURRENT STATUS AND FUTURE PERSPECTIVES C. Petrillo Istituto Nazionale per la Fisica della Materia Unità di Perugia, Dipartimento di Fisica Università di Perugia. Via A. Pascoli, I-06123 Perugia, Italy

Abstract Recent developments in design and production of thermal neutron detectors based on a crystalline Si diode coupled to a Gd converter are reviewed. The most significant results of neutron test measurements carried out on prototype systems and pointing out the performances of the device are reported. Progress through various detection schemes and technological solutions for the production of a real time neutron counter for operation under intense neutron fluxes are outlined.

Introduction Demand for intense neutron sources to be built in Europe and U.S.A. has been growing during the last years brought about by either scientific need in the neutron scattering research field or the expected shutdown of many neutron sources in the near future. The Spallation Neutron Source (SNS) in U.S.A. is now in planning and design stage and the European Spallation Source (ESS) is the next major planned neutron facility in Europe. A remarkable intensity gain is expected for next generation neutron sources, a provisional estimate of the gain in peak neutron flux at the high-flux ESS yields a factor about 30 over the most powerful present sources (ILL, ISIS)[1]. The increase of the intensity available at the sample will make possible higher resolution experiments or investigations of small size samples. Quite generally, the intensity gain will be exploited in high data rate experiments with shortening of the collection time against the same statistics. Improving the source performances demands for design and development of future generation instrumentation among which a prominent role is played by neutron detection. In particular, access to higher data rates and/or higher resolutions will require substantial improvements, over the present standards, in detector capabilities like speed and space resolution. Moreover, improved performances in both shape flexibility and total coverage surface will be relevant for fully exploiting the neutron source possibilities by new spectrometer designs. A further aspect, no more negligible under intense neutron fluxes, is the detector aging which deserves an accurate investigation.The present status and the new concepts in neutron detector

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

14

•

technologies have been recently reviewed at the Workshop on Neutron Detectors for Spallation Sources held in Brookhaven National Laboratory in September 1998[2].

The requirements for a high-flux spallation source detector, emerging from an open discussion on technologies to pursue for the next generation instruments, are reported in Table I. Traditional neutron detectors, routinely operated in most of the existing neutron spectrometers are 3He gas counters, in the extended concepts of Multi-Wire-Proportional-Chamber (MWPC) and Micro-Strip-Gas-Chamber (MSGC), and solid scintillators. Position resolution and counting rates achievable by this class of detectors can considerably vary depending on the specific detector design and construction parameters. However, typical values of position resolution can be set in the range of a few millimeters at usable data rates of the order of 105 s-1 over a 30 x 30 cm2 area of a standard Position Sensitive Detector (PSD). As to time resolution, other than the intrinsic timing properties of the detector, the readout scheme is what dominates the effective time response of the system under operation. Typical values of the pulse-pair resolution are in the range of a few ms for real-time counters. Higher values in both space resolution and counting rate can be achieved by means of image plate detectors[2,3] whose use is, however, ruled out at pulsed sources when time-of-flight techniques are exploited.

Table I. Spallation source detector requirements (Ref. 2).

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Considering that the technology in gas counters and solid scintillators is well assessed and the intrinsic limitations of these devices have been investigated, it is worthwhile to examine alternative solutions which could extend the capability range of the current detectors to the values desired for high intensity operation. In the field of semiconductor detectors based on crystalline Si diode sensors, the technological development has been impressive with advances in growing and ion-implantation techniques coupled to progress in Very Large Scale Integration (VLSI) readout electronics. Large area Si detectors can be assembled from microstrip or pixel sensors and they are commonly employed as particle trackers in High Energy Physics (HEP) for their unrivalled space resolution at high counting rates. Resolution values as high as a few µm (2-5 µm) are achievable with Si detectors whereas operation at 1 MHz count rate is rather standard. The intrinsic capabilities of the Si sensor, coupled to an advanced frontend electronics, are rather appealing and make it an interesting candidate for neutron detection purposes.

Use of an appropriate neutron converter is however necessary and this raises the problem of the neutron detection efficiency that, in principle, should be as high as possible over a wide range of neutron energies.

Investigation of the performances of the Si sensor as a novel neutron detector and optimization of the system features to match with the ideal detector characteristics were first pursued under the European Project ENNI (European Network for Neutron Instrumentation) and are currently running under the European Community Project XENNI (The 10-Members European Network for Neutron Instrumentation). Various detection schemes have been designed and prototyped. In the present paper the main achievements are outlined and the characteristics of the detector, as resulting from neutron test measurements on the prototype systems and from the Monte Carlo simulation of the detection process, are presented.

The results of the test measurements clearly point out at both possible improvements and intrinsic limitations of the detector capabilities, which, in turn, define the ideal application of this device. Owing to the very high position resolution, the Si sensor is well suited for neutron transmission applications (neutron radiography and tomography) as well as for neutron reflectometry or residual stress analysis. Moreover, exploitation of the high time resolution, at similar or somewhat relaxed position resolution, makes the Si sensor a suitable detector for single crystal diffraction on small samples. Operation of a properly designed Si detector in conjunction with focusing neutron optical devices could open the possibility of microdiffraction studies with thermal and cold neutrons.

The crystalline Si sensor The Si device is basically a p-n junction diode operated

at reverse bias and working as a solid state ionization chamber. The region depleted of free charge carriers represents the sensitive active volume for detection of ionizing particles or radiation. Electron-hole pairs created inside the depletion region by the ionizing particles are swept by the electric field out of the depletion region to the electrodes and give rise to a net current signal. Reverse biasing of the junction by an external voltage reinforces the built-in electric field and greatly improves the performances as a detector[4]. Basically, incomplete charge collection due to trapping and recombination is reduced and the thickness of the depletion region is increased. This latter effect results in a lowering of the junction capacitance and hence in a reduction of the electrical noise.

In Si sensors charge collection is particularly efficient: due to the narrow band gap, a relatively small amount of energy is necessary to create an electron-hole pair (~3.6 eV at room temperature to be compared with ~30 eV in a typical gas filled detector). Therefore, for a given energy deposited by the ionizing particle the number of charge carriers is relatively high with benefit of the attainable energy resolution. The response of the Si diode is intrinsically quite fast and, for detection of light particles, dominated by the charge transit time, namely the time required for complete migration of the carriers from their creation points to the opposite sides of the depletion region. The transit time depends on the electric field, the depletion width and the carrier mobilities. The advantage of high mobilities and rather long lifetimes of both electrons and holes in Si is exploited. The pulse rise time actually observed from a Si detector-preamplifier combination is influenced by the preamplifier characteristics which should be designed to held the rise time down to that determined by the charge collection time. Finally, because of the rather short ranges of charged particles inside Si, the creation of electron-hole pairs is confined to regions quite close to the actual path of the ionizing particle. The intrinsic limit in the achievable position resolution is thus very low, typically ≤1 µm.

Nowadays, the fabrication of Si diode detectors combines the technique of ion-implantation and photolithography and takes advantage of the methods first developed to produce integrated circuits. Complex electrode geometries can be obtained like, for example, double-sided microstrips or pixel sensors. Schematically, p+ and n+ layers are obtained by ion-implantation into a standard 300 µm thick high-purity (111) Si wafer which is mildly n-type. Details of the fabrication technique can be found in [5,6]. A schematic of a double-sided microstrip sensor, with p+-side strips perpendicular to the n+-side strips, is shown in Fig. 1. Double-sided microstrips are a current version of a 2d-PSD whereas a 1d-PSD is obtained by the strip segmentation of just the p+-side of the

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

15

RASSEGNA SCIENTIFICA

Fig. 1. Schematic of a double-sided Si microstrip.

sensor.The strip pitch can be varied from 25 µm to a few mm, 50 µm being a standard in HEP Si trackers, and the typical sensor size is of the order of some cm2. In standard operation, the whole n-bulk of the detector is fully depleted. The quality of a Si detector which aims for very good space and energy resolution at high data rates depends to a large extent on the performances of the frontend electronics. Recently a number of circuits have been developed for low noise Si strip detector readout, based on charge-sensitive preamplifier and shaper system[7,8]. Quite more complex VLSI readout chips, able to drive up to 128 output channels, are commercially available and currently employed for serial readout in large area Si trackers.

In summary, intrinsic characteristics of the Si detector like the short collection time (tens of ns), the high space resolution (~1 µm ), the low bias voltages (≤100 V for 300 µm Si thickness) and the shape flexibility could be favourably exploited in thermal neutron detection. Of special interest is the maximum instantaneous data rate capability that influences the detector dead-time. A detector operating under an intense and pulsed neutron flux should have a dead-time notably less than 1 µs and the capability of managing the time-information contained in the white neutron pulse. In order to operate the Si device as a neutron detector, a neutron converter is necessary. Ideally, the converter should have a high neutron absorption cross section over a wide neutron energy range (from a few µeV to tens of eV), a quantum yield nominally equal to 1 and a thickness allowing the ionizing particle produced by neutron capture to escape and impinge the Si sensor. In practice, the optimization of these parameters for real converters involves compromise.

Commonly employed neutron converters are 10B and 6Li where the neutron capture is accompanied by emission of 1.5 and 2.1 MeV α particles, respectively. A severe restriction to the use of these converters in conjunction with the Si sensor arises from the limited range of the α particles inside the converter itself. The very small converter thickness necessary for escaping of the α particles (~3 µm) results in an extremely low neutron detection efficiency. In Gd, which is a less common neutron conver-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

16

•

ter, the absorption process is followed by emission of low energy γ-rays (keV) and conversion electrons with an energy spectrum ranging from ~30 keV to ~250 keV. A quantum yield of 0.79±0.03 for the neutron-conversion electron process was measured in [9]. The neutron absorption cross section both in natural and isotopic Gd takes quite high values in the range of cold and thermal neutron energies, even though it decreases very sharply for energies ≥0.1 eV. The ranges of the conversion electrons in Gd vary from 9.2 µm for the 70 keV electron line to 62 µm for the 220 keV electron line while the corresponding values in Si are 35 µm and 210 µm. The combination of a rather large thickness and a high absorption cross section makes Gd an appropriate converter for cold and thermal neutrons.

Neutron detector prototypes and measurements A number of neutron detector prototypes have been developed with designs optimized to match with the specific characteristics to be measured. Progress through various detection schemes originated from the primary need of demonstrating the capability of the Si sensor as a real neutron detector through measurements of neutron detection efficiency, background sensitivity and dead-time. These fundamental tests were carried out on the most elementary detector design, namely a diode sensor, 5 x 4 mm2 size, coupled to VLSI readout electronics, in the simplified version of a low-noise charge-sensitive preamplifier and shaper system, and conventional electronics. Being the basic performances well assessed, the next step was the development of position sensitive devices to prototype both 1d and 2d detectors and to measure the achievable position resolution. Two detector designs were developed, namely a low-resolution 4 x 4 diode matrix as a simple 2d-system for beam monitor applications and a 64 channel microstrip sensor as a medium resolution 1d-detector. Parallel readout of both the 16 and the 64 channels was carried out by the 8-channel version of the commercial VA chip (Ide As, Norway) in conjunction with auxiliary amplification stages based on conventional electronics. Natural Gd was employed as a converter in all cases since it gives the best compromise between cost and detection efficiency. A summary of the detector type and related readout electronics is given in Table II. The last three entries of this table refer to detector prototypes which have been developed quite recently and not yet tested. Basically, they are advanced versions of the 1d and 2d detectors based on both a microstrip sensor of larger size and larger number of channels and a Si wafer with 70 x 22 pixels. In all cases, the parallel readout of a relatively large number of channels required the development of new electronics based on both VLSI and SMD components to produce a fully digital output for direct interface to the computer.

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Fig. 2. Time of flight spectrum of the Pb single crystal as measured by the Si/Gd diode detector. Data collected by the ZnS scintillator are showin the inset. The intensity scales as the detector areas, namely 20x20 mm2 ZnS and 5x4 mm2 Si.

Table II. Prototypes developed under the E.U. projects.

Neutron measurements were carried out on the steadystate neutron source TRIGA Reactor (ENEA-CRE Casaccia, Rome, Italy) and the pulsed source ISIS (RAL, U.K.).

Neutron pulse measurements and pulse height spectra of the conversion electrons produced by neutron capture in Gd are reported in [9]. Typical values of the rise time were of the order of 1 µs, but this value was lowered down to 50 ns using a fast version of the preamplifier chip[9]. It must be remarked that rise time values as low as 50 ns are presently hardly obtainable by standard gas detectors or scintillators and would enable the detector to work under data rates of the order of 5 x 106 s-1mm-2.

As to the energy spectrum, since the relevant features are confined over the energy region 40-250 keV, electronic discrimination of the output signal, by means of a window discriminator, was found to be advantageous to cut off electrons produced by γ-ray absorption and having energies outside the window[9]. The γ-ray sensitivity was measured by exposing the Si detector box to a 60Co source operated at 2.5 cm distance with an activity of 32 mRem/hr. An efficiency of ~10-5 was found. Lower values of γ-ray efficiency are desirable and improvements are possible by a careful choice of the detector shielding materials. Indeed, low energy photons produced by collisions of the primary high-energy g-rays, present in the source environment, within the shielding materials can be directly detected by the Si sensor.

Time of flight spectra and neutron detection efficiency The Si diode coupled to a 10 µm thin natural Gd converter was extensively tested as a neutron detector. Results obtained at the steady-state source are reported in [9].

Further measurements were carried out on the PEARL beam line at ISIS where the diode was operated simulta-

neously with a standard ZnS(Ag)/6Li scintillator, 20 x 20 mm2 size. Diffraction patterns of some reference samples were measured, namely an incoherent scatterer, a single crystal, a powder and an amorphous sample. The scattering angle was fixed at 90o for all the measurements with a moderator-to-sample distance of 12.6 m and a sample-to-detector distance of 0.4 m. Background measurements were performed using either B4C or 6Li glass as neutron absorbers. As an example, the diffraction pattern of a cylindrical Pb single crystal, 2.5 cm diameter and 5 cm height with the vertical axis parallel to the [001] direction, is shown in Fig. 2. Bragg reflections extending from (2 0 0) to (10 0 0) were observed with the ZnS scintillator, whereas the highest order observed with the Si detector was (8 0 0). This corresponds to an inco-

Fig. 3. Time of flight spectrum of Fe powder as measured by the Si/Gd diode detector. Data collected by the ZnS scintillator are also shown.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

17

RASSEGNA SCIENTIFICA

ming neutron wavelength λ = 0.87 Å. Such a value represents the lower operational limit of the diode below which the neutron detection efficiency of the Gd converter sharply decreases. Because of the high counting rate, dead-time corrections were applied to the intensity collected by the ZnS scintillator (~2 µs). The diffraction pattern of the Fe powder sample, contained into a cylindrical thin walled vanadium can (0.1 mm thickness, 5 mm diameter, 80 mm length), is shown in Fig.3. The highest order Bragg reflection detected by the Si sensor is (3 2 1) corresponding to λ = 1.08 Å.

The whole of the Bragg peak data collected on polycrystalline and single crystal samples was used to deduce the neutron detection efficiency of the Si detector using the known ZnS efficiency as a reference, that is ~20% at λ = 1 Å, with a 1/v scaling law for its absorption cross section. A description of the data analysis can be found in [10]. In Fig. 4 the measured efficiency values of the Si detector are given as a function of neutron wavelength. An inspection of this figure shows that an efficiency of ~4% is still observed at λ = 0.87 Å and saturation values of ~20% are found at λ≤ 2 Å. The experimental data are shown in comparison with the efficiency curve calculated by the Monte Carlo simulation described in Section

Fig. 5. Contour lines of the neutron beam as measured by the 4x4 Si/Gd diode matrix. Plots correspond to a different alignement of the detector in the beam.

Fig. 4. Measured neutron detection efficiency of the Si/Gd detector.

Experimental data are from Pb single crystal (dots), Fe powder (triangles) and BaF2 polycristal (circles). The error on the Pb data is of the same size as the dot. The full line is the Monte Carlo curve calculated for the specific experimental configuration.

4. A sensible increase in detection efficiency can be easily obtained by an optimized Gd-Si-Gd sandwich configuration, that is with the converter on both sides of the sensor. Efficiency values of the order of 40% at λ≤ 2 Å are expected.

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

18

•

Beam shape measurements The 4 x 4 Si diode matrix, 2 x 2 mm2 pad size and 10 µm thin Gd, was employed as a low resolution 2d-sensor for measurements of the neutron beam profile. Potential application of the device as an area position sensitive beam monitor is quite interesting. In this case the efficiency has to be low and good space resolutions can be easily achieved. In principle, the position sensitive monitor could be used in place of Polaroid films when details of the sample centering in the incoming beam are required or in neutron radiography applications. The beam shape measurements were carried out on the thermal monochromatic neutron beam at TRIGA Reactor. The beam line was equipped with a neutron attenuator and an adjusta-

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

ble slit. The Si matrix was mounted on a translation stage that could be stepped in front of the incoming beam.

In Fig. 5 the contour lines of the measured beam profile are shown for two different translation position of the detector.

Space resolution measurements The Si sensor was a double-sided microstrip with 50 µm pitch and 20 x 20 mm2 size, operated as a single-sided microstrip for being a linear PSD. Signals were collected from the front-side (p-side) strips while the backplane (nside) strips were electrically connected together to give a common signal through a resistive load. The Gd converter (250 µm thickness for reflection operation) was mounted in contact with the backplane. Since not extremely high resolutions are required for normal operation as neutron detector, groups of strips were connected together via ultrasonics microbonding. A 64 channel sensor configuration, with an effective pitch of 400 µm and the central portion at 200 µm pitch, was produced. The possibility of varying the pitch between 200 µm and 400 µm on the same sensor was preferred in this prototype version to test the performances obtainable with different configurations in terms of efficiency and noise. Each group of strips was coupled to one channel of the readout chip (8-channel VA chip) by integrated capacitors (16-channel, 100 µm pitch). For the detector to be spaceexpandable, the VA chips were mounted on both sides of the sensor with an independent readout of the 200 µm pitch central portion on both sides. The resulting layout is schematically shown in Fig. 6. For the first series of test measurements, the 64 output signals were separately sent to a PC equipped with a PHA (Pulse Height Analyser) card (ADC, 8000 channels; Silena, Italy) and a Scope card (ADC, 100MHz, 8-Bit; Matec Instruments, USA).

Space resolution measurements were carried out on the monochromatic thermal neutron beam of the TRIGA reactor. Two different experimental setups were employed. In both cases a narrow incoming neutron beam was produced by means of appropriate Cd slits and Soller collimators and the spectra were collected as a func-

tion of the relative slit-detector translation position. The main difference between the two experimental configurations was the use of a single monochromator against a double-crystal setup which guarantees a minor γ-ray contamination of the incoming beam. A detailed description of these measurements can be found in Refs. 11 and 12.

In Fig. 7 the spectra of two alternate strips belonging to the 200 µm pitch portion of the sensor, as collected with the single monochromator setup and integrated from 10 keV to 70 keV, are shown versus the translation position.

As expected, the distance between the peak positions is 400 µm and this figure is representative of the effect expected from a simultaneous readout of two alternate strips. The curves superimposed to the experimental data in Fig. 7 were obtained from a Monte Carlo simulation of the detection process in the specific experimental configuration. In particular, the incoming neutron beam was modeled by a triangular function, 0.55 mm Full-Widthat-Half-Maximum (FWHM), and the profile of the direct γ-ray beam was taken into account[12]. In Fig. 8 the spectra collected with the double-crystal setup and integrated over the full energy range are shown versus the translation position. The effect brought about by the γray beam is apparent by the comparison with Fig. 7 where extended intensity tails are observed. The full curve, also shown in Fig. 8, is the calculated convolution of the incoming neutron beam profile, 0.57 mm FWHM, with the sensor response function[12].

In both the experimental configurations the width of the incoming neutron beam was of the order of 0.55-0.6 mm, that is a value in excess of the strip pitch, and the measured curves were found to have those width values.

This result shows that the sensor response does not introduce any sizeable broadening at this level of space resolution values which, in any case, are extremely good in comparison with the standard few millimeters resolution of the gas or scintillator counters. However, further measurements of the position resolution, making use of an incoming neutron beam with a FWHM comparable to or lower than the strip pitch, have to be carried out.

Diffraction spectra of small samples The operation of a linear position sensitive Si/Gd detector as a diffraction counter was tested on the pulsed neutron beam at ISIS, ALF beam line. The sensor was the same as described in Section 3.c. The detector box was mounted at 77 mm distance from the sample position and it was not shielded. No collimator was inserted between sample and detector. The scattering plane was vertical and a total scattering angle range of 15o was covered by the sensor with 0.27o step. The diffrac-

Fig. 6. Layout of the 64 channel microstrip detector board.

tion patterns from two standard samples were measured, namely a pyrolitic graphite single crystal, 24 mm length, 3 mm width and 0.8 mm thickness, and a poly-

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

19

RASSEGNA SCIENTIFICA

Fig. 7. PHA spectra integrated from 10 KeV to 70 keV versus translation position for two alternate strips of the 200 µm pitch portion of the microstrip sensor. The experimental data (dots) are compared with the results (full lines) on the Monte Carlo simulation.

Fig. 8. PHA spectra integrated over the full energy range versus translation position for a reference strip from the 400 µmpitch portion of the microstrip sensor. Experimental data (dots) collected in the doublecrystal configuration. The full line is the results of the numerical convolution of the incoming beam profile with the sensor resposnse function [12]

crystalline Ni wire, 10 mm length and 2 mm diameter.

Both the graphite crystal and the Ni wire were mounted with the long side parallel to the strip length. A BN shielding was mounted all around the samples. The diffraction spectra from graphite were collected at two different sample positions, in order to observe the displacement of the Bragg peak intensity through different strips, i.e. portions of the detector. As expected with a single crystal sample, only a number of strips, corresponding to the angular width of the Bragg peak, was collecting neutrons. In Fig. 9 the contour plots of the intensity from the graphite crystal versus the strip position, i.e. scattering angle, and the neutron wavelength are shown for the two sample positions. As ap-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

20

•

parent from this figure, the region of bulky intensity corresponds to the projected size of the graphite crystal. The diffraction spectra from the Ni wire are shown in Fig. 10 for some selected strips and versus the neutron wavelength. In this case, being the sample a polycrystal, the wavelength position of the Bragg peaks shifts in going from one strip to another. Even with this non-optimized geometry, an intrinsic detector resolution ∆λ/λ = 4 x 10-3 was obtained and a ratio ∆d/d = 2 x 10-2, with d the lattice spacing, was observed on the powder diffraction spectra. This latter value has to be compared with the typical values ∆d/d ranging from ~4 x 10-4 to 2 x 10-2 obtainable at the high-resolution neutron powder diffractometer HRPD (ISIS). It is also apparent from these results that, although investigation of relatively small size samples is feasible, a gain in intensity is absolutely necessary. At present, this could be achieved operating the detector in conjunction with neutron optical elements with focusing of the intensity at the sample position.

Monte Carlo simulation The Monte Carlo analysis is an efficient tool to outfit the detector design with those construction parameters which optimize a selected set of capabilities. An extended Monte Carlo analysis of the detection process was carried out in Ref. 9 pointing at the estimate of the neutron detection efficiency versus wavelength and converter thickness for various geometric configurations of the Si sensor and the Gd converter. The reliability and the accuracy of the numerical routine were tested against the measured electron energy spectrum and detector efficiency, accounting for the experimental configuration of the system, namely converter and sensor dimensions and relative distance (see Fig. 4). The Monte Carlo code provides also the simulation of the space resolution of the detector as a function of both the sensor geometry and the incoming beam characteristics and allowance for the motion of the sensor relative to an incoming beam of variable size was considered.

The neutron flux inside the converter was modeled by the function: Φ(x,y; λ) = Φι(y) exp[−x µ(λ)]

(1)

where µ(λ) is the linear attenuation coefficient of the converter. The incoming neutron flux was described by the function Φι(y) which accounts for the (xz) section of the beam shape. In the linear strip detector, the relevant variation of the beam is only that along the y-direction perpendicular to the strip length. Several analytic options for Φι(y) were possible, namely, Gaussian, Lorentzian, triangular and trapezoidal functions. The tabulated values of the neutron cross sections in Gd from Ref.

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Fig. 11. Monte Carlo simulation of the space resolution of the Si microstrip detector.

Fig. 9. Contour plots of the intensity measured from a graphite crystal versus the strip position and the neutron wavelenght. The two plots correspond to a different sample-to-detector distance

Fig. 10. Diffraction spectra of the Ni wire versus neutron wavelenght and for selesctedstrips.

13 were employed in the calculation. Only conversion electron production, upon neutron absorption in Gd, was considered as reaction channel, that is γ-emission and production of secondary electrons in Gd were neglected. The quantum yield of the neutron/electron reaction was set to 0.8, according to Ref. 9. The angular distribution of the electrons at the creation points was assumed to be isotropic. The electron creation points inside the Gd converter were randomly generated with a distribution proportional to Φι(y) along the y-direction and an exponential decreasing distribution along the xaxis, corresponding to the neutron beam attenuation with depth according to Eq. 1. The conversion electron specific energy loss due to collision inside the Gd converter was calculated using the Bethe formula, adapted to the case of fast electrons[4]. The specific energy loss due to radiative processes, approximately given by ZE(MeV)/700 times the collisional loss, was neglected amounting to ~2% of the collisional term for the most energetic electrons produced in Gd. The energy loss formula, applied to the conversion electron groups with initial kinetic energies centered at ~70 keV, ~150 keV and ~220 keV, was integrated to deduce the energy-distance relation. Being known, from the random sampling procedure, the coordinates of the creation points and the exit direction (escape angle), i.e. the path inside the converter, it was possible to deduce the energy of those electrons reaching the interface between Gd and Si and entering Si. From the residual electron energy, the total number of electron-hole pairs in Si can be deduced. In order to deduce the space resolution of the sensor, it would be necessary to follow the electron-hole pairs along their paths from the creation points to the closest electrode (strip). Therefore, the collisional energy loss formula was applied again to deduce, upon integration, the energy-distance relation inside

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

21

RASSEGNA SCIENTIFICA

Si for all the electrons escaped from Gd with a distributed initial kinetic energy. This amounts to know the number of the created electron-hole pairs as function of the distance. Being known the path direction and the total range of the electron inside Si, it was possible to deduce the space distribution of the released charge among different adjacent strips.

The calculated space resolution curve is shown against the experimental data in Fig. 7. The overall good agreement between the data and the simulation results is an indication of the reliability of the program which was employed to simulate the response of a Silicon sensor with variabile strip width to a narrow incoming neutron beam. The calculation was carried out at 1 Å incoming neutron wavelength for a 10 µm FWHM triangular beam profile. The width of the sensor was ranging from 10 µm to 500 µm and the thickness of the backside natural Gd converter was 250 µm. The calculated curves obtained by integration of the energy-dependent spectra from 10 keV up to the maximum energy of secondary electrons are shown versus translation position and strip width in Fig. 11. The broadening introduced by the sensor response to the 10 µm wide incoming beam is quite small and of the order of 50 µm FWHM, independently of the width of the Si strip. The contribution to the counts collected on a reference strip due to charge released by electrons travelling far away the strip, which is accounted for by the Monte Carlo simulation, is low and does not produce a remarkable resolution degradation.

Conclusions The neutron test measurements carried out on the prototype systems have defined advantages and operational limits of the Si/Gd device. Improvements in some of the detector capabilities, like the neutron detection efficiency and the γ-ray sensitivity, are possible. The detection efficiency can be easily increased by use of a double converter and/or use of isotopic Gd. The possibility of selective coating of the sensor by isotopic Gd oxide is presently addressed. The γ-ray sensitivity can be reduced by an accurate choice of the shielding material: substitution of the Al detector box by blackened LiF box is expected to reduce the interactions of high-energy γ-rays and the fast neutron activation. At present, the main limitation arises from the converter which confines the application to cold and thermal neutrons. Certainly the values of pulse rise time and space resolution, as obtained on the prototype systems, are much better than those of standard detectors.

Aiming at the development of a large area detector as a modular assembly of single sensor units, requires the increase of the sensor size simultaneously to the capability of high speed parallel readout of a large number of chan-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

22

•

nels. Serial readout techniques, typically exploited in HEP Si trackers, are not suitable for pulsed neutron applications. The last year activity in this field was focused on these two specific aims. As mentioned in Section 3., a 70 x 40 mm2 Si/Gd microstrip detector with 600 mm pitch was designed and prototyped. Readout of the 128 output channels was fully digital and based on both VLSI and SMD components. Electrical tests of the device are in progress and neutron test measurements are scheduled by June 1999. Based on the design of the 128 channel readout system, a new chip, VA-TAN, for use in high-rate neutron counting applications has been designed and delivery of the first prototype is expected by April 1999. The chip is in a 32 channel version, containing preamplifier&shaper, window discriminator and priority encoder blocks, yielding a strobe signal and a 5-bit address with a timing of the address adjustable between 1 MHz and 20 MHz. Coupling of the Si microstrip sensor to the fast, parallel and integrated VA-TAN chip would allow to build a second-generation detector of larger size and increased number of channels which exploits the speed and resolution capabilities of the Si sensor at best.

Finally, the possibility of operating the Si/Gd detector in conjunction with a high-efficiency microfocusing optical device is presently addressed.

References 1. ESS. A Next Generation Neutron Source for Europe, Vols. 1-3, edited by J. Kjems, A. D. Taylor, J. L. Finney, H. Langeler and U. Steigenberger (1997).

2. Proceedings of Workshop on Neutron Detectors for Spallation Sources, Brookhaven National Laboratory, September 24-26 (1998).

3. New Tools for Neutron Instrumentation, Journal of Neutron Research, Special Issue, 4 (1997).

4. G. F. Knoll, Radiation Detection and Measurement, (John Wiley & Sons, New York) (1989).

5. G. Batignani, F. Bosi, L. Bosisio, A. Conti, E. Focardi, F. Forti, M. A. Giorgi, G. Parrini, E. Scarlini, P. Tempesta, G. Tonelli and G. Triggiani, Nuclear Instrum. and Methods A 277, 147 (1989).

6. G. Lutz, Nuclear Instrum. and Methods A 367, 21 (1995).

7. E. Nygard, P. Aspell, P. Jarron, P. Weilhammer and K. Yoshioka, Nuclear Instrum. and Methods A 301, 506 (1991).

8. O. Toker, S. Masciocchi, E. Nygard, A. Rudge and P. Weilhammer, Nuclear Instrum. and Methods A 340, 572 (1994).

9. C. Petrillo, F. Sacchetti, O. Toker and N. J. Rhodes, Nuclear Instrum. and Methods A 378, 541 (1996). See also: C. Petrillo, F. Sacchetti, O. Toker and N. J. Rhodes, Journal Neut. Res. 4, 65 (1996).

10. C. Petrillo, F. Sacchetti, N. J. Rhodes and M. W. Johnson, Technical Report, RAL-TR-97-010 (1997).

11. C. Petrillo, Il Nuovo Cimento D 20, 931 (1998).

12. C. Petrillo, F. Sacchetti, G. Maehlum and M. Mancinelli, Nuclear Instrum. and Methods A, (1999).

13. C. D. Garber, R. R. Kinsey, Neutron Cross Sections, Vol. II, BNL 325 (1976).

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Articolo ricevuto in redazione nel mese di Maggio 1999

MUONS AT ISIS R. De Renzi, T. Shiroka Dipartimento di Fisica, Parco Area delle Scienze, 7A 43100, Parma, Italy

Introduction The scope of this short article is to familiarize the reader with the µSR activity which is pursued at ISIS since the beginning of the spallation neutron source. In the following we shall briefly review the relevant muon properties and the principle of the technique, with a list of applications and a few examples of recent results.

We shall then describe the existing ISIS facility and its history (a few more details are available from the ISIS web pages, under http://www.isis.rl.ac.uk/muons/).

The last part of this article is devoted to very promising technical developments: extensions to thin films and (possibly) interface studies, and synchronous excitations, i.e. ideas for exploiting the pulsed nature of the ISIS facility. µSR in its essence is an exotic version of magnetic resonance (although, technically, no true resonance is strictly required).

Positive muons are spin I=1/2 particles which may be implanted in any sample. They either stop in a lattice interstice or bind chemically to a host molecule, whereby they provide a spin probe analogous to the nuclear species of the host material. Although the production and detection techniques are borrowed from particle physics, the use of muons in condensed matter research is very closely related to NMR. In particular both spectroscopies are complementary to neutron scattering, in the strict sense that the measured quantities correspond to local response functions, which may be expressed, by Fourier theorem, in terms of weighted averages over q space of the corresponding scattering cross sections.

Muons are unstable particles which decay into a positron and two neutrinos. The muon spin dynamics may be monitored directly thanks to two phenomena, both provided by parity violation in the weak interaction decays: firstly a very large spin polarization is easily obtained, and, secondly, a large correlation results between the directions of the muon spin and of the decay positron momentum. Hence the muon is a magnetic dipole probe, a tiny magnet inside matter which can measure local fields. We shall see that this feature does not restrict applications to magnetism, since even electron mobility may be measured in this unconventional way, and to an advantage!

The muon and µSR in a nutshell Table I summarizes the relevant muon properties. Spin and magnetic moment characterize its NMR-like behaviour, setting the scale between local magnetic field and spin precession frequency. The lifetime of roughly 2 µs determines the time window for the signal observation, which is typically 15 µs. The mass, 1/9 of the proton mass, and the positive charge dictate that muons behave in condensed matter like light hydrogen isotopes. We shall disregard negative muons, which in principle may be employed in much the same way, since their fate is heavily dependent on the nuclei of the host material, due to the very high cross section for nuclear capture.

For this reason their use is scarce.

Figure 1 shows the two decays governing the implementation of µSR. Positive pions are produced by impinging the 800 MeV protons of ISIS onto the intermediate transmission target - a graphite slab placed half way between the accelerator and the spallation source. Pions decay into a neutrino and a muon, as shown in fig.1 (top). Parity violation amounts to the fact that only lefthanded neutrinos exist, i.e. they all arise from the decay with negative helicity (spin antiparallel to momentum).

Since the pion is an I=0 particle, in the center-of-mass frame also the muon must have negative helicity by angular momentum conservation. Hence by focusing with a small angular acceptance on pions at rest on the target surface one obtains a 100% spin polarized muon beam of well defined energy (E ≈ 4.2 MeV, p ≈ 28 MeV/c), which can be transported to the sample. Thermalization of muons inside condensed matter takes place in very short times - fractions of picoseconds - and very often without loss of polarization.

Parity violation and angular momentum conservation lead in a similar fashion to a correlation between the direction of the muon spin and that of the positron emis-

Table I. Muon properties Mass mµ/me

Lifetime τµ(µs)

Spin I

Gyromagnetic ratio γµ/2π (MHz/T)

Production

Decay channel (main)

205.

2.196

1/2

135.5

π+ → µ+ + νµ (τπ = 26ns)

µ+ → e+ + νe + νµ

Vol. Vol.3 4 n.n.21 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

23 23

RASSEGNA SCIENTIFICA

Fig. 1. Top: the pion decay; bottom: the muon decay. The blue lobe, with centre on the muon, represents the polar plot of the probability for the positron direction of emission. Short arrows show the spin of each particle.

sion. The correlation is not full because the muon decays into three bodies. Averaging over positron energies it amounts to a 1/3 enhanced probability for emission in the muon spin direction, as indicated by the lobe in the polar plot of fig.1 (bottom), which looks almost like a circle displaced in the direction of Iµ. With a fixed arrangements of detectors the asymmetry in the positron count rate may be measured in each of them as a function of time from the instant of implantation. The time evolution of the asymmetry effectively monitors the spin dynamics of the ensemble of the implanted muons. Timing is then particularly straightforward at a pulsed source like ISIS, since the instant of the muon arrival (the centre of the bunch) may be obtained from the accelerator.

To be more specific the positron count rate in a given direction, making an angle φ with the initial muon direction is: N(t)=N0e-t/τ (1+A G(t) cos φ)

(1)

where the spin autocorrelation function G(t)= 〈I(0)I(t)〉 modulates the asymmetry A of the muon exponential decay. This general expression is perhaps clarified by two specific examples.

First imagine to have just a static external magnetic field B, perpendicular to the initial muon spin direction. It is the so-called transverse field geometry in which I precesses around B at frequency ω = γµB. The anisotropy lobe of the decay positron momentum probability follows the spin precession. The correlation function in this case is Gxx(t)cos φ =cos(ωt +φ): the resulting positron count rate consists in a sinusoidal modulation superimposed to the muon decay curve and it is straightforward to determine B from it. Both geometry and count rate are shown in fig.2 (top).

A second example is when a field is applied parallel to the initial muon spin direction, or even when no field at all is applied. In this case any reduction of the initial asymmetry must be attributed to the so-called spin lattice

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

24

relaxation: muon spin flips due to creation or destruction of an excitation which couples to I (often both a creation and a destruction take place, in a second order Raman process). The experimental condition is referred to as zero or longitudinal field geometry, and it is depicted in fig.2 (bottom). Detectors are placed in the direction of the incoming muon spin polarization (forward) and in the opposite one (backward). The relaxation function Gzz(t)=exp(-t/T1) appears in the difference between the count rates in the forward and in the backward detectors.

Relaxation is not exclusive of the longitudinal geometry.

A similar relaxation function may also appear in the transverse geometry, due to both static and dynamic losses of coherence in the precession (spin-spin relaxation).

Since, besides spin polarization, also detection efficiency is close to 100%, total of 108 implanted muons provide already a high statistics experiment. At ISIS these events are collected in the order of an hour, with less then a

•

Fig. 2. Top: Transverse field experiment and positron count rate in a detector placed in the precession lane Bottom: Zero or longitudinal field experiment; positron count rates in detectors forward and backward relative to the initial spin direction I(t=0). The relaxation function G(t) may be obtained as (F(t)-B(t)) / (F(t)+B(t)).

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

Fig. 4. Time window for the muon as a probe of dynamics, compared to other techniques.

Fig. 3. Spectral density, J(w), of the fluctuating local fields, under e.g. thermally activated dynamics. Three different temperatures are shown, both in linear and in logarithmic scales (inset): blue, low T; green, intermediate; red, high T. Relaxation is due to fields which oscillate at resonance with the muons, at ω=ω0 (yellow arrow). The area under the curve is 〈b(0)b(0)〉 = b2, constant, hence an estimate of the relaxation rate is J(ω0) ≈ b2/Ω.

thousand muons coexisting in the sample at a time. Therefore total radiation damage is absolutely negligible and probe-probe interactions are absent.

In many applications the local field, or the local field distribution are directly the quantity of interest. This is obviously the case of magnetic materials, both long range ordered systems and spin glasses, where the temperature dependence of spontaneous local fields yields the order parameter. It is also the case of type II superconductors, where the application of an external magnetic field produces a flux lattice, whose local field distribution is directly mapped by muons.

Also the relaxation function G(t) is of interest since it may be related to the electronic response functions, for instance to the electronic spin autocorrelation function in an ordered magnet: G(t) ∝ 〈S(0) S(t)〉

(2)

For example this quantity gives direct access to critical phenomena, or to the excitation spectrum of the system.

In this respect it is a muon specialty, among magnetic resonance techniques, to give readily access to spin relaxations in zero external magnetic field. NMR, for instance, normally relies on an intense external field to provide the nuclear polarization, while muons can do without. This is extremely advantageous in magnetic critical studies, where an external field may couple to the order parameter.

Muons, like nuclei in NMR, may also probe very efficiently the properties of non magnetic materials. A whole chapter is provided by insulators - molecular systems among them - in which the muon forms a paramagnetic bound state, either a muonium atom (the muonic equivalent of hydrogen) or radicals, by muonium addition to an unsaturated molecule. These composite species are themselves extremely sensitive probes (spin labels), since the muon local field is amplified via the hyperfine interaction with the much larger magnetic moment of the bound electron. A distinctive set of precession frequencies, corresponding to transitions between the hyperfine levels of the paramagnetic centre may be detected in suitable experimental conditions. Applications range from radical chemistry itself, to molecular dynamics (fullerenes, metallorganic complexes, polymers, zeolites), to studies of hydrogen centers in semiconductors, where the muon plays the role of an ideally diluted hydrogen analogue, up to the equivalent studies in metals and hydrides. In this last case muon diffusion in itself is the focus and the tiny nuclear random fields provide the interaction which allows the detection of a slow dynamics.

The specific time window for dynamical processes offered by muons depends on the strength of the dominant spin interaction. This is estimated considering a rough approximation to the relaxation function time constant, the relaxation time T1, illustrated schematically in fig.3.

Let us assume that the interaction is given by an effective instantaneous local field b, of spectral density J(ω) =

∫

∞

−∞

dt

〈 b(0)b(t)〉 iωt e 〈 b(0)2〉

extended over a frequency width Ω=1/τc. This amounts to saying that τc is the characteristic time of the dynamics. Then, if the dynamics is fast, i.e. the instantaneous

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

25

RASSEGNA SCIENTIFICA

Larmor frequency ω0=γµb is much less than Ω, one has 1/T1 ∝ J(ω0) ∝ ω2 τc

(3)

0

This relation is qualitatively justified by fig.3 (see the figure caption). Note that, as motion becomes faster, the relaxation - hence the frequency linewidth - decreases.

This is called motional narrowing, and it is the opposite of what takes place in scattering, where the faster is the motion the broader are the spectra. Actually muons and neutrons probe the same correlation function, which broadens as the motion becomes faster, but the muon relaxation is governed by the height, not the width of the function.

Figure 4 illustrates the relative position of the µSR time window in comparison to that of other techniques, including neutrons. Since measurable relaxation rates range from 103 to 108 s-1, the time window may be estimated from eq.(3) assuming an upper limit for ω0=2.4 1010 s-1 from contact hyperfine interactions in muonic atoms (109 is a more typical value for free muons in magnetic materials) and a lower limit of 105 s-1 from nuclear dipoles. The figure shows that muons bridge a very interesting gap between neutrons and other techniques, such as NMR and Mössbauer, only partially covered by neutron spin echo and Perturbed Angular Correlation Spectroscopy (PACS). Notice further that for each system, characterized by a well defined interaction strength, a typical span of four orders of magnitude inτc is easily accessible.

We mentioned earlier the measurement of electron mobility, which is a recent breakthrough of muons [1]. In insulators and semiconductors it is very difficult to obtain the true intrinsic value, since the outcome of macroscopic experiments is often entirely determined by trapping at defects or impurities.

Not so for µSR. We mentioned that in both insulators and semiconductors a fraction of the implanted muons may bind to a radiolysis electron to form paramagnetic species (either muonium or adduct radicals). This may happen also in a delayed process: epithermal muons still travel fast for some distance after they have produced the last electron by ionization. When they finally localize they are swept by the diffusing radiolysis cloud at some later time, on the pico to nanosecond scale, and a fraction may form a paramagnetic centre. However in these conditions the paramagnetic signal cannot be detected because, due to the delayed formation, the precessions are not time coherent and their asymmetry averages to zero.

This correspond to an easily measured missing fraction.

The application of an electric field reveals a peculiar asymmetry: an increasing field parallel to the incoming muon momentum drastically reduces the missing fraction, while an antiparallel field has a much smaller ef-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

26

•

fect. The field biases the electron diffusion and the parallel geometry removes the cloud from the muon site, leaving an enhanced bare muon signal. The antiparallel alignment drives the electrons past the muon, with an initial increase of the paramagnetic missing fraction. The electric field dependence of the measured precession amplitudes does indeed afford the determination of the electron mobility. This is not a universal method, since it relies on delayed muonium formation: quartz, semiconductors and cryocrystals are a few examples of recent applications. In other compounds muonium may not form altogether, or it may result from extremely rapid “hot” chemistry reactions.

The ISIS facility The duty cycle of the ISIS accelerator, repeated at 50 Hz, makes it ideal as a pulsed muon source. Muons very clo-

Fig. 5. ISIS muon passband: the blue curve is essentially the Fourier transform of the muon beam profile in time. Precession frequencies higher than the cutoff and relaxation times shorter than its inverse cannot be detected. The red curve shows the improved passband obtained by pulsed field methods - limited only by the time resolution of the electronics.

sely follow the proton time profile, constituted of two bunches, each ca. 50 ns wide and separated by 350 ns. As a consequence muons are implanted in bunches and their time of arrival starts a separate clock for each µSR detector. Detected positrons are directly recorded as a function of the time elapsed, accumulating events bunch after bunch.

An important advantage of this pulsed scheme is the extremely low background of uncorrelated events, whose main source is the beam itself: at ISIS the beam becomes silent after the bunch arrival and muon decays can be cleanly detected up to more than ten muon lifetimes, affording the measurement of very slow relaxation rates.

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

A second potential is the possibility to synchronize different perturbations with the muon implantation. Since decay positrons are detected for 20 µs every 20 msec his allows a 10-3 duty cycles for the perturbation as well, which is beneficial, e.g. for intense laser or radio frequency excitation.

An important drawback must be mentioned, though.

Muon spin precessions are detected inasmuch as they start coherently - at the same time, with the same initial phase. However, since the muon burst has a FWHM duration of dT=70 ns, precession frequencies are limited to a maximum of 1/dT=13 MHz. At higher frequencies the passband function of the spectrometers, shown by the blue curve in fig.5, falls to zero, since early and late muon signals interfere. In practice 6 MHz is the upper limit, to be compared with at least 200 MHz at the continuous muon sources. Detection of fast relaxations is likewise limited. Therefore ISIS is somewhat complementary to TRIUMF, Canada, and PSI, Switzerland, which are the major continuous beam facilities.

Figure 6 shows the present ISIS muon facility and beam line, together with the two instruments, MuSR and EMU, regularly open to experimental scheduling. The third beam line, DEVA, accommodates several temporary experimental setups at subsequent times.

MuSR was the first muon beam at ISIS, opened to regular scheduling since 1986. The instrument on MuSR is based on a spectrometer built in Parma, Dizital, which is designed to allow both transverse and longitudinal field experiments. The change of geometry is fast and a single experiment may exploit both setups. The spectrometer is composed of 32 positron detectors - plastic scintillators on photomultipliers - in two circular arrays around the sample position. The large number of detectors is needed to minimize the count loss due to the finite dead time after each positron trigger (this is common to all instruments at a pulsed source). It is equipped with a pair of Helmholtz coils providing a maximum field of 0.2 T. It may accommodate a furnace, a closed cycle refrigerator, an standard Orange He cryostat, a dilution fridge a sorption cryostat or a furnace, with sample temperatures ranging from 50 mK to 800K. In the early times the two subsequent muon bunches were both fed onto MuSR, with a further reduction of the frequency response of the instrument.

An EEC project resulted in the construction of a second spectrometer, EMU, and in the tripartition of the original beam. In the present scheme the first muon bunch is let through to MuSR, while the activation of a fast kicker magnet slices the second burst in halves and sends them to the other two twin areas, EMU and DEVA. EMU compensates the halved intensity, with respect to MuSR, by a factor two in solid angle coverage around the sample.

This was accomplished by specializing EMU for the longitudinal geometry. There are 32 detectors also in EMU.

The temperature range available is 300 mK to 800 K and larger longitudinal fields of 0.4T may be reached. For fifty percent of the experiments run at ISIS the two instruments are virtually equivalent.

The ISIS beam cross section is somewhat larger than that of the competitor facilities at TRIUMF and PSI: 90% of the beam is within an ellipse of 25x17 mm2. The ideal sample shape is that of a thin slab (4 MeV muons stop in 1 mm water and 200 µm Cu). Samples with smaller cross section may be run in the so-called fly-past scheme, where special care is taken to avoid thick material in the neighbourhood of the sample. In this way muons not hitting the specimen are let through and out of the spectrometer.

The same transmission target station along the Extracted Proton Beam delivers muons onto three other areas on the opposite side of the ISIS muon facility. They belong to the Riken-RAL Facility, sponsored and run under a UK-Japan agreement, which entails a separate selection panel. One of them is permanently dedicated to a µSR spectrometer, ARGUS. A small percentage of the ARGUS beam time is allocated by the IBM6 panel, in parallel with the entire EMU and MuSR scheduling.

New developments The challenge at ISIS has always been that of fully exploiting its pulsed nature. Synchronous laser, current, electric field excitations and radio frequency (rf) resonance have already produced specific successful experiments. The regular implementation of rf methods, which open the vast field of the NMR spin manipulation tricks, is the subject of a dedicated EEC project [2], which we shall not cover here.

We concentrate instead on two other very recent technical developments studied at ISIS: • The commissioning of a new pulsed low energy muon beam; • The use of a pulsed magnetic field for MuSR experiments.

Both of them extend µSR beyond its present limits: the first one opening new perspectives for using muons in meso- to nano-scale structure investigations, whereas the second is a workaround to the frequency passband of fig.5 and a tool to study delayed muonium formation.

Let us start from the low energy muon beam. We have already mentioned that the standard muon beams deliver particles with a kinetic energy of 4.2 MeV. When implanted, they stop after 150 mg/cm2 of material, with a distribution of implantation depths of width ≈20 mg/cm2 (typical figures). This clearly makes them unsuitable for studies of thin films, surfaces and interfaces.

Several ideas were put forward in the past to obtain a beam of much lower kinetic energy. The most promising was proposed by Harshman et al. [3] from TRIUMF and pursued at PSI by Morenzoni et al. [4]. It takes advanta-

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

27

RASSEGNA SCIENTIFICA

ge of the particular mechanisms of slowing down that charged particles, once they have passed through a premoderator, will undergo in thin layer (300 nm) of a Van der Waals solids (typically Ar and N2 cryosolids). These are insulators where considerable electronic gaps (Eg>10 eV) coexist with low frequency phonons (hω=5 meV). In them muons may lose energy by ionization until the gap threshold is reached, while below this threshold all scat-

source low energy beam cannot be used to explore very thin layers and interfaces. For these applications the use of a pulsed source is mandatory. This being the motivation, a new pulsed beam of low energy muons was developed in a collaboration among Heidelberg, ISIS, University College London and Parma [5].

The layout of the apparatus is shown in fig.7, where three main components are recognizable: the moderator,

Fig. 6. Layout of the ISIS muon facility with the beam and the three areas, EMU, MuSR and DEVA. The two main spectrometers are also shown (picture taken from the ISIS web pages).

tering cross sections drop drastically. Therefore the moderator becomes transparent to low energy muons, while it still acts as an energy degrader for faster particles.

This results in a relative intensification (an accumulation) of the beam spectrum around energies of 15±10 eV. Up to now, the highest low energy muon production efficiency is about 10-4, a factor 103 more than in other materials. This amounts to saying that the low energy muon production takes effectively place in the top layer of the moderator, of thickness a few thousand Å, which is the so-called escape length.

The efficiency is certainly low, but sufficient for dedicated experiments. The pulsed ISIS source is less intense than the two major continuous sources of PSI and TRIUMF. This is a disadvantage, which is however balanced by a very important advantage: the a-priori knowledge of the implantation time. A µSR experiment may be thus performed without detecting the incoming muons. Since detection implies scattering, hence a minimum spread of the particle energy spectrum of order 1 keV, the continuos

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

28

•

the electrostatic acceleration and transport section, made of three segments, and finally the sample area, where Low Energy µSR (LE-µSR) experiments are performed.

The moderator is made up of a frozen layer of Ar on an electrically insulated Al substrate, kept at 10 K by a liquid He flow cryostat. By raising the substrate to 10kV the very slow muons generated inside the Ar moderator are extracted and delivered into the transport section, consisting of three electrostatic lenses and two electrostatic mirrors. The first mirror act also as a filter, separating the low energy muons from the non moderated ones.

The total path of the low energy muons is one and a half meter. A variable voltage decelerator before the sample area provides then a tunable epithermal muon beam.

The sample area is surrounded by conventional µSR spectrometer, equipped with a simple solenoid magnet.

The whole apparatus operates under ultra-high vacuum (UHV) conditions (<10-10 mbar), essential to preserve a stable moderator.

A first characterization of the beam is provided by the ti-

Vol. 4 n. 1 Giugno 1999

RASSEGNA SCIENTIFICA

me-of-flight spectrum of fig.8, which shows a large peak due to the low energy muons at times appropriate for the transport energy. Demonstrative experiments were performed on a sample consisting of a 200 Å Cu layer deposited on a quartz substrate. Muons in quartz form predominantly paramagnetic muonium, and by performing an intermediate magnetic field (4.2 mT) experiment the diamagnetic muon fraction may be selectively detected.

We finally mention the use of pulsed magnetic fields to extend the frequency instrumental response, which is limited by the muon beam time width. The basic idea consists in applying the transverse magnetic field in a µSR experiment synchronously but delayed in time with respect to the muon bunch arrival. Then all muons will precess in phase, despite the jitter in their time of arrival.

This technique is particularly relevant for the observa-

Fig. 7. Schematics of the Low Energy Muon apparatus from ref. [5]. A noble gas cryosolid is placed in the primary muon beam.

The moderated low energy muons are accelerated monochromatically to 10 keV and deflected by electrostatic mirrors.

The last section is shown rotated by 90 degrees (it actually points out of the picture plane). It contains a conventional µSR spectrometer.

The signal amplitude as a function of energy should then decay as muons penetrate farther into the quartz substrate. This is shown in fig.8 where the two top histograms reveal transverse field precessions with different amplitudes at two distinct implantation energies. The bottom panel shows the entire amplitude dependence on muon energy E, which demonstrates the expected behaviour.

Furthermore, despite the low statistics, an initial reduction of amplitude around E=500 eV is visible. This is attributed to muons stopping in the top oxide layer of the thin film. Since CuO is magnetic, muons in that layer do not contribute to the precession at the frequency appropriate for 4.2 mT. The continuous curve in the figure represents the simulation performed with a standard Monte Carlo program (SRIM), which represents both features.

Fig. 8. Time-of-flight spectrum of the moderated beam. The low energy muon peak is apparent.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

29

RASSEGNA SCIENTIFICA

since the signal was observed in detectors at 90 degrees from the initial muon spin direction. The top panel in the same figure shows the precession simulated by numerically solving the spin dynamics in a time dependent field. The pulsed field data fall on the improved red passband of fig.5. Note that in principle there is no intrinsic passband with the pulsed field method and the finite frequency width of the red curve is due entirely to the finite resolution of the clock employed.

This new scheme is very promising in particular for paramagnetic centre investigations, since it allows to measure directly a signal from a delayed muonium fractions, which would loose the needed precession coherence in a conventional transverse field experiment.

Muons in condensed matter research have a scientific history dating back thirty years. µSR became an established technique just around the opening of the ISIS facility, thanks both to the instrumental advances and to the particularly successful contributions to high Tc superconductor studies. The developments pursued at ISIS and in the other facilities witnesses that it is a mature technique, and still vital.

Fig. 9. First Low Energy Muon results. Top: transverse field precessions of monochromatic muons, at 1 and 8 keV respectively, implanted in a thin Cu film on quartz substrate. Bottom: muon asymmetry as a function of muon energy, i.e. implantation depth. Above 4 keV muons start to enter quartz, where the muon fraction is small due to muonium formation. A slight asymmetry reduction at 500 eV is probably due to a similar phenomenon in the top CuO layers.

tion of paramagnetic species [6], which have larger effective gyromagnetic ratios due to the hyperfine coupling to the unpaired electron spin. In this case an additional static longitudinal field must be applied to preserve the muon spin polarization until the transverse field is switched on. The pulsed field is produced by a loop of laminar currents flowing in two planes around the thin sample. The field device may be constructed with a thin enough front plane to let muons in through it. This geometry minimizes eddy currents since the field, uniform in the sample region, vanishes outside.

The bottom panel in fig.10 shows the measured precession pattern obtained in quartz, as it appears from the function G(t) (the muon decay curve has been removed).

At low fields the muonium signal from inter-triplet transitions is observed at frequency ωT ≈ γe/2B, where γe ≈ 28 GHz/T is the electron gyromagnetic ratio. The frequency reaches 28 MHz, appropriate for B ≈ 2 mT. The data show clearly the initial increase of the precession frequency, due to the delayed pulsing of the transverse field. Notice also the muon spin polarization which starts from zero,

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

30

•

Fig. 10. Demonstration of the pulsed field method: observation of a muonium precession up to 28 MHz (b, cf. the passband of fig.6) and its ODE simulation (a). Notice the variable precession frequency in the beginning, during the field ramp.

References 1. V. Storchak et al., Phys. Rev. Lett. 76, 2969 (1996).

2. S.P. Cottrell et al., Appl. Magn. Reson. 15, 469 (1998).

3. D.R. Harshman et al., Phys. Rev. Lett. 56, 2850 (1986); D.R. Harshman et al., Phys. Rev. B 36, 8850 (1987).

4. E. Morenzoni et. al., Phys. Rev. Lett 72, 2793 (1997); E. Morenzoni et al., J. Appl. Phys. 81, 3340 (1997).

5. C. Bucci et al., to appear on Physica B - Condensed Matter; K. Traeger, Doctoral Thesis, Univ. of Heidelberg (1999).

6. T. Shiroka et al., subm. to Phys. Rev. Lett.

Vol. 4 n. 1 Giugno 1999

DOVE LUCE DI SINCROTRONE

Articolo ricevuto in redazione nel mese di Marzo 1999

THE VUV BEAMLINE OF ELETTRA: AN EXAMPLE OF HIGH RESOLUTION PHOTOEMISSION CORE LEVEL SPECTROSCOPY ON Sb/Si(001)c4x2 AND Ge/Sb/Si(001)2x1 INTERFACES C. Quaresima, P. Perfetti, P. De Padova, C. Ottaviani CNR-ISM, Via del Fosso del Cavaliere 00133 Roma, Italy

R. Larciprete ENEA, Dip. INN/FIS, Via E. Fermi 45 00044 Frascati (RM), Italy

Abstract We present the High Energy Resolution Photoemission VUV beamline at ELETTRA. After an introductory description of the beamline features, we report on the most outstanding results recently obtained. Sb-dimer-induced Si(001)c(4x2) relaxation was studied on the Si2p core level by high-resolution photoemission spectroscopy. Two surface components, S* and C* were identified in the

Si2p core level measured on the Sb/Si(001)2x1 surface at 1 monolayer (ML) Sb coverage. The S* was attributed to the whole first, half second, and half third Si layers, whereas only half of third layer contributes to C* component. By using the Sb-Ge site exchange process, Ge was epitaxially grown on Sb/Si(001) producing a perfect Si bulk-like structure. A surface component S'* was also

Fig. 1. Picture of the Ge heteroepitaxial growth on Si(001) (a), and assisted by Sb surfactant (b).

Vol. 4 n. 1 Giugno 1999

â&#x20AC;˘

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

31

DOVE LUCE DI SINCROTRONE

HIGH ENERGY RESOLUTION PHOTOEMISSION BEAMLINE

Fig. 2. Scheme of the VUV beamline.

found in the Ge3d core level obtained from the Sb/Ge/Si(001) heterostructure.

Introduction In very recent years, the experimental techniques based on synchrotron radiation have received a great impulse, due to the construction of the third generation storage rings having unprecedented high brilliance. Amng these techniques, photoemission has taken great advantage in terms of energy resolution and photon flux due to the birth of new undulator beamlines.

Photoemission is one of the most widely used experimental techniques, both using conventional sources (i. e. ultra-violet and x-ray lamps) and synchrotron radiation.

Its popularity mainly derives from the quite direct interpretation of energy distribution curves which reproduce (in first approximation) the density of electronic occupied states in the sample. The photoemission technique using the soft x-ray energy range (hν≤1000 eV) of synchrotron radiation is quite attractive because it offers the possibility of studying surface versus bulk contributions in solids as a consequence of the strong energy dependence of the electron escape depth. Moreover, the tunability of the photon energy gives the possibility to perform absorption experiments at the K-edge of some chemically interesting low-Z elements (e.g. C, N, O, Cu, Ni) thus obtaining information on filled and empty sta-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

32

•

tes coupling. Information about the surface and interface structure can also be obtained by means of experimental techniques such as photoelectron diffraction.

High energy resolution and high photon flux were the major requirements made by scientists working with synchrotron radiation to the constructors of new third generation sources. Once a high brilliance, low emittance source has been made available, the building of a beamline achieving a high energy resolving power over such a wide energy range is a delicate task, due to the fact that monochromatization is obtained by using optical gratings, whose slope errors play a major role in determining the total resolving power. Thus, this important goal has stimulated a strong effort by many scientists to find the best and easiest technological solutions to achieve high energy resolution. Great performance were obtained at the VUV Photoemission beamline [1,2], designed and built by our group in collaboration with Sincrotrone Trieste at the ELETTRA third generation storage ring in Trieste. This beamline completed the commissioning period at the end of 1995, since then offering to many users and to our collaborators the possibility to perform leading experiments in many research fields, such as metal and semiconductor surfaces and interfaces and gas phase photoemission [3,4,5,6,7,8,9]. In the present paper we will report on some recent results trying to focus the attention on the importance of high energy

Vol. 4 n. 1 Giugno 1999

DOVE LUCE DI SINCROTRONE

resolution and brilliance. The large interest developed in the last years on V group atoms (As, Sb, and Bi) adsorption on Si(001), mainly resides in the surfactant properties exhibited by these atoms in assisting the epitaxial growth of Ge/Si(001) [10].

For the Ge/Si heteroepitaxy, the interfacial free energy, the chemical interaction and the lattice mismatch govern the growth mode. Due to the large lattice mismatch of 4.2% between Si and Ge, the Ge heteroepitaxy on Si is determined by the Stranski-Krastanov (SK) growth modality [11]. A picture of this growth is drafted in Fig.1a.

On the SK growth the formation of pseudomorphic bidimensional (2D) layer is followed by the nucleation of tridimensional (3D) islands which, relaxing the elastic energy accumulated in the few layers compressively strained, cause the formation of defected film.

It is well known that the use of surfactant atoms (As, Sb, and Bi) suppresses the Ge/Si(001) SK growth, by means of the site-exchange process [12], and allows the deposition of a continuous 2D smooth film. The picture of the mechanism is drafted in Fig.1b.

Several theoretical and experimental investigations have been carried out on the interface formation of Sb on Si(001) surface in order to determine the atomic configuration, surface reconstruction and electronic band structures. Our recent high resolution photoemission investigation of the Sb/Si(001)2x1 interface [13] has demonstrated that the Si2p peak, in addition to the bulk component, contains two surface components attributed to the deeper Si layers, in a qualitative agreement with the surface core level calculated by final-state pseudopotential theory [14].

We also performed the Ge/Sb/Si(001) photoemission study which showed a perfect bulk-like configuration for the Si substrate atoms [13]. In this paper we report a description of the VUV beamline of ELETTRA, and the high resolution photoemission study of Sb/Si(001)c4x2, and Ge/Sb/Si(001)2x1 interface formation.

Beamline features and experimental details The VUV photoemission beamline uses the light emitted by the U12.5 undulator mounted on the straight section 3.2 of the third generation storage ring ELETTRA in Trieste. This undulator has 36 periods (three independent sections of 12 periods each) and a period length of 12.516 cm. When the working energy of the storage ring is 2 GeV, it covers the energy range hν~17÷900 eV using the first three odd harmonics and gap values between 32 and 93 mm. The light beam central cone is selected by a pinhole whose section is approximately 1x1 mm2 .

The VUV photoemission beamline [1,2] uses a Spherical Grating Monochromator (SGM) of the DRAGON type [15, 16]. The optical scheme of the beamline is shown in Fig. 2. This monochromator has real entrance and exit slits while all the optical elements, i. e. the prefocusing

mirrors and the gratings, are spherical, except for the last toroidal refocusing mirror, placed before the experimental chamber. This implies that the focal position for the exit slit is different for different wavelengths. Therefore, the exit slit is moved along the optical path contemporary to the grating rotation, in order to follow the focal point. The SGM concept was derived from the TGM (Toroidal Grating Monochromator) design and it is a simplified version of the CEM (Cylindric Element Monochromator) development. The breakthrough of the DRAGON design in obtaining extraordinary performance in terms of energy resolution was due to the use of spherical optical elements, which can be manufactured very accurately allowing the achievement of state-of-the-art in terms of surface smoothness and figure error accuracy. Our results, obtained using the described beamline, have confirmed the extremely high energy resolution, and the average resolving power obtained in all entire energy range (17÷900 eV) is of the order of 104.

The beamline optics We may divide the beamline into three parts: the prefocusing mirrors, the monochromator (entrance slit, gratings and exit slit) and the refocusing mirror.

Prefocusing mirrors The prefocusing mirrors M0 and M1 are spherical and made of gold coated SiC; they are mounted in the Kirkpatrick-Baez configuration, i. e. they focus the source horizontally and vertically on the exit slit and entrance slit respectively. Each mirror focuses individually in one plane and their optical functions are completely decoupled.

The first mirror has the additional function to absorb the major part of the power load. In fact the figure errors induced by the heat load on this mirror are not critical for the beamline performance. Anyway, the cooling system of this two mirrors has been accurately designed and the temperature gradient on these elements is less than 0.1 ˚C. Actually, during the first four years of beamline operation we never observed changes in the beam intensity and/or in the beamline performance due to heat load problems, despite the beam current accumulated in the storage ring was increased from the initial 10 mA to 300 mA now available. To achieve the required precision in the mirror alignment, the mirror holders act as cardan joints allowing adjustment of the roll and pitch movements with a precision of 0.1 µrad.

Monochromator The monochromator is a Spherical Grating Monochromator (SGM) of the Dragon type. The entrance slit consists of two blades made of tungsten carbide in contact with water cooling copper tubes. The minimum aperture is about 5µm and the maximum is about 80µm (the si-

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

33

DOVE LUCE DI SINCROTRONE

the focal position for different photon energies. Although a cooling system identical to the entrance slit one is mounted, we never had to really cool this element, as the heat load after diffraction is very low.

The refocusing mirror M4 (see Fig. 2) is made of pure silica coated with gold; it has a toroidal shape and a magnification of 0.2 which leads to a calculated image size on the sample (130 x 12)µm2.

The VUV beamline energy resolution has been tested by photoabsorption experiments [2] performed on several gases at different energies. The table 1 reports the energy esolution of the VUV beamline for high and low photon energies.

The Fig.s 3 report the low energy range of the double excitation Rydberg series of He vs photon energy. Fig. 3(a) shows the odd and even Rydberg states, while Fig 3(b), which is an expansion of the spectrum reported in Fig 3(a), shows clearly the Rydberg states up to n=20. The instrumental resolution limit, estimated by the number of resolved He Rydberg states, corresponds to 3 meV. Table 1.

Gas

Helium Argon Nitrogen Neon

Fig. 3. Photoionization spectra of the doubly excited He: (a) He Rydberg odd and even series converging to the n=4+; (b) expansion of the (65.24 65.4) eV energy range of the spectrum reported in Fig. 3(a). The even series up to n=20 is clearly visible.

ze of the light beam, at the entrance slit, depends on the undulator gap width), but it is usually ranging between 5µm (high energy limit) and 20µm (low energy limit).

The gratings chamber is equipped with five interchangeable gratings which allow to cover the whole source energy range. All gratings are spherical and made of gold coated SiC. The first three gratings cover the energy range 100÷900 eV with a grating included angle of 173˚. The other two gratings cover the low energy range hν~17÷120 eV, and to avoid the use of gratings with very low line densities, i. e. high limit to the resolution due to diffraction, the included angle was reduced to 160˚. This was accomplished by a pair of removable plane mirrors M2 and M3 placed between the entrance slits and the gratings. After the two plane mirrors, the optical path is the same for both high and low energies.

The scanning mechanism is made with a double elastic coupling between the rotating bar and the push rod. The exit slit can be moved along the optical path to follow

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

34

•

Photon energy (eV)

65 244.4 400.9 867.1

Resolution (meV)

3 19 28 191

The experimental station The experimental chamber of the VUV beamline is described in Fig. 4. This system is made of two UHV chambers separed by a gate valve and equipped with several tools for surface preparation and/or surface diagnostics. This configuration avoids to break the UHV of the analyser chamber during the turnover of the experiments. A pumping system, made of 1000 l/s turbo and a 400 l/s ion pump with integrated titanium sublimator, carries out a background pressure better than 1*10-10 mbar. The analysis chamber and the experimental chamber communicate through a sample manipulator, which permits to move the sample in the x,y, and z spatial coordinates, ϑ, ϕ, and χ rotation angles. Fig.4 reports the analysis chamber on the left, and the preparation chamber on the right side. The UHV analysis chamber is equipped with two emispherical electron energy angle integrated (R=125 mm, acceptance angle =±8°), and angle resolved (mean radius R=50 mm, acceptance angle=±0.5°) analysers. A µ-metal shield, placed inside the experimental chamber, minimizes the path distortion of the electrons caused by the earth magnetic field. By using the smaller electron analyser, which is mounted on a goniometer system at two axis, it is possible to perform photoemission angle-resolved spectroscopy.

Vol. 4 n. 1 Giugno 1999

DOVE LUCE DI SINCROTRONE

Fig. 4. Schematization of the experimental chamber.

Legend: 1), 2) the analisys and the experimental chambers; 3) UHV valve the chambers separation; 4) Manipulator manipulator; 5) Angular resolving electron analyser; 6) angular integrating electron analyser; 7) Sample fast entry; 8) Ar Ion Gun; 9) crystal cleaver; 10) LEED optics; 11) Evaporation sources system; 12) oscillating quartz crystal balance; 13) Samples reservoir; 14) RHEED electron gun; 15) Gas inlet gasses immission

The hemispherical electron energy analyser, OMICRON, Mod. EA 125 HR U5, allows an energy resolution of 6 meV, while the angle resolved energy analyser, VSW, Mod. HA 54, allows an energy resolution of ≈30 meV. The preparation chamber is equipped with a load-lock system, which permits to change the sample without breaking the UHV (background pressure of 10-10 mbar).

It contains the Ar+ ion sputtering system, the cleaver, the scraper, the gas-line, the Low Energy Electron Diffraction, the Resolved High Electron Energy Diffraction and Auger analysis for surface diagnostics. A molecular beam epitaxy (MBE) system, having four effusion cells, allows the "in-situ" growth of metal and semiconductor films. The nominal thickness is monitored with a quartz balance. The manipulator transfers the sample from the preparation chamber to the focal point of the electron energy analysers placed into the analysis chamber, and it can be cooled down to 90 K, and heated up to 1300 K.

Experimental The Si(001) substrate (p-type, 5.5 Ω·cm) was loaded in the UHV chamber (base pressure 8.5 10-11 mbar). Oxide was removed by repeatedly flashing the substrate to 850˚C by resistive heating, being careful that the pressure remained below ~2x10-10 mbar during the heating. Sb was evaporated at a rate of 0.3 ML/min from a quartz pipe, while the substrate was kept at 500˚C. Ge was evaporated at a rate of 0.1 ML/min with the Si substrate at 350˚C. Adsorbate coverage was measured using a quartz microbalance; the error in determining the Ge thickness was estimated to be 10%. The sample temperature was measured by an infrared pyrometer and a thermocouple.

Spectra were measured after cooling the sample to ~120˚C. The photoemission spectra were acquired using the angle-resolved electron energy analyzer, set at emission angles of ϑ=60˚ (surface sensitive) and ϑ=0˚ (bulk

sensitive). The angle between the photon beam and the normal to sample surface was 10˚. The photon energies were hν=135 eV, and hν=230 eV, while the total energy resolution was better than 50 meV. Results and discussion

Sb/Si(001)c4x2 Figure 5 reports the Si2p core-level spectra measured at surface (ϑ=60˚)- and bulk (ϑ=0˚)-sensitive emission an-

Fig.5. Surface-sensitive (left column), and bulk-sensitive (right column) convoluted Si2p core levels for clean Si(001)c(4x2), and after the adsorption of 1 ML of Sb.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

35

DOVE LUCE DI SINCROTRONE

Fig.6. (a) Comparison between surface-sensitive (hν=135eV), and bulksensitive (hν=230 eV) Si2p spectra collected at ϑ=60˚, after the adsorption of 1 ML of Sb. (b) Comparison between surface-sensitive (ϑ=60˚), and bulk-sensitive (ϑ=0˚) Si2p spectra collected at photon energy of 135 eV, after the adsorption of 1 ML of Sb.

gles for the clean Si(001)c(4x2) surface, and for clean surface covered by 1 ML of Sb. The deposition of Sb on Si substrate held at 500˚C has been shown to give a saturation coverage of almost 1ML in a 2x1 surface reconstruction [17]. The binding energies (BEs), reported for all the figures, are measured relative to the position of the corelevels bulk component B. The Fig.5 clearly shows that the clean Si peak is completely modified during the Sb/Si(001) hetero-interface formation. It shows the loss of the structures related to the Si surface reconstruction, and the arising of two new peaks. Quantitative information is obtained by convoluting the spectra with spin-orbit split Voigt functions. A least-square fitting procedure was used to separate the doublet components of the spectra having spin-orbit (S.O) splitting = 606 meV and branching ratio= 0.53. Lorentzian width was allowed to vary between 0.085 and 0.13 eV for the surface components, while in the case of the bulk a smaller width equal to 0.020 eV was used.

In addition to the bulk component peak convolution for the Si2p core level of Si clean surface was carried out according to ref. [18], where four components related to the c(4x2) reconstruction were identified: exactly, the up

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

36

•

(U) and down (D) dimer atoms, the second layer atoms (S), and the subsurface layers or defects (C). The energy shifts found in our case for the clean Si2p were U=-489 meV, C=-200 meV, D =60 meV, and S=200 meV, in close agreement with the values reported in [18]. The Si2p peak measured after the adsorption of 1 ML of Sb shows that the U and D components disappear, in agreement with scanning tunneling microscopy investigations [17], which showed that 1 ML of Sb removes Si(001)c(4x2) reconstruction and forms symmetric dimer rows rotated by 90˚ with respect to the original Si dimer rows. In this case the Si2p lineshape convolution gives the S* and C* doublets, located at BE of +200±5 meV and -192±5 meV, in addition to B. In order to confirm this result, enhancing the bulk -or the -surface sensitivity, we report in Fig.6(a) the comparison between surface-sensitive and bulk-sensitive Si2p spectra collected at ϑ=60˚, obtained by using a photon energies of hν=135 eV and hν=230 eV; and in Fig. 6(b) the comparison between surface-sensitive and bulk-sensitive Si2p spectra collected at photon energy of 135 eV, obtained by using the emission angles of ϑ=60˚ and ϑ=0˚, after the adsorption of 1 ML of Sb. All these spectra, measured at surface- and at bulk- sensitive configurations, definitively show in addition to B component, the S*, C* contributions. The peak intensities of each component in Fig.s 6 are altered since varying the photon energies or the emission angles, the probed

Fig.7 (a) Side-view configuration of Si(001)c4x2. The different Si atoms, which are assigned the U, D, S, B, and C Si2p components (see Fig.5), are represented by different symbols. (b) Side-view configuration of Sb/Si(001)2x1 interface. Arrows represent the atomic distortions due to the Sb/Si(001)2x1 interface formation, while the different Si atoms, which are assigned the S*, C* and B Si2p components, are represented by different symbols.

Vol. 4 n. 1 Giugno 1999

DOVE LUCE DI SINCROTRONE

fig.7(b), S* includes the contribution of whole first, half second, and half of the third layers, whereas only half of the third layer contributes to C*. These results are in qualitative agreement with theoretical calculations performed using the final-state pseudopotential theory [14] on the Sb/Si(001) interface to calculate, including the substrate relaxation effects, the surface core level shifts for the Si2p core level .

Fig. 8. Ge3d and Sb4d core level spectra collected at surface-sensitive (υ=60°), and bulk-sensitive (υ=0°) configurations, after the deposition of 2.5 ML of Ge on the Si(0011) surface covered by 1 ML of Sb.

depths is changed carring out the contribution originated by different atomic layers. This is, in fact a way to differentiate the information originated in the surface structures by the bulk contribution during the heterostructure formation.

Fig. 7a reports a side-view configuration of Si(001)c4x2 clean surface, while Fig. 7b reports a side-view configuration of Sb/Si(001)2x1 interface. Different symbols represent each type of Si atom to which is assigned the Si2p component reported in fig. 5, while the arrows represent the atomic relaxation due to the Sb/Si(001)2x1 interface formation [14].

We showed in Ref. [13], that S* and C* components are attributed to the deeper Si layers, and as schematized in

Fig.10. Si2p core levels taken after the adsorption of 0.7, 1.8, and 2.5 MLs of Ge collected at surface-sensitive ϑ=60˚ (left column), and bulk-sensitive ϑ=0˚ (right column) configurations.

Fig.9. Ge3d core levels taken after the adsorption of 0.7 ML and 2.5 MLs of Ge on the Sb/Si(001)2x1 surface.

Ge/Sb/Si(001)c4x2 Figure 8 reports the Ge3d and Sb4d core level spectra at normal (ϑ=0˚), and grazing (ϑ=60˚) emission angles after the deposition of 2.5 MLs of Ge on the Si(001) surface covered by 1 ML of Sb. The absence of surface reconstruction features on the Ge3d core level spectra [19] , and the enhanced intensity of the Ge/Sb ratio collected at surface-sensitive configuration with respect to the bulk-sensitive configuration, indicate that the Ge atoms are completely covered by Sb atoms. The Sb-Si interface has been substituted by Ge-Si interface during the Sb-Ge site-exchange process [12]. In this case the Ge atoms, that typically dimerize on Si(001) surface [20], form an intralayer between Sb and Si atoms.

Figure 9 reports the least square fit of Ge3d core levels at 0.7 ML and 2.5 MLs of Ge coverage on the

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

37

DOVE LUCE DI SINCROTRONE

determine in Si2p core level two surface components S* and C*, attributed to the first, half of the second, and half of the third Si relaxed layers, and half of the third layer respectively, while for Ge/Sb/Si(001) interface that the Ge-Sb site exchange process takes place giving to Si substrate atoms a bulk-like configuration.

These results show that high resolution photoemission spectroscopy used on properly fabricated systems can be used as an accurate diagnostics for the characterization of complex heterostructures.

The authors thank the staff of ELETTRA, of the VUV beamline, and the MADESS II "Disposotivi optoelettronici discreti ed integrati in Silicio" project. S. Priori and M. Capozi are gratefully acknowledged for their invaluable technical assistance.

References

Fig.11. Side-view configuration of Ge/Sb/Si(001)2x1. The different Ge atoms, which are assigned the S*' and B components (see Fig.9), are represented by different symbols. Si atoms (gray-color) are in bulk-like configuration.

Ge/Sb/Si(001) interface. At 0.7 ML Ge coverage on Sb/Si(001) all the Ge atoms, which form a single layer, contribute to the S*' component. The presence of two components, S*' and B, at 2.5 MLs Ge on Sb/Si(001), is indicative of the fact that the Ge atoms are in two inequivalent configurations: B represent the bulk-like contribution, whereas S*' is due to the presence of the Sb dimers.

The Ge3d core level convolution was obtained considering a S.O splitting=0.55 eV, and a Brancing ratio=0.6, in close agreement to ref. [19]. It is worth noting that this component has the narrowest (full width at half maximum of 280 meV) line shape never reported for the Ge3d core level [13].

The Si2p core level spectra collected at ϑ=0˚, and ϑ=60˚, after the deposition of 0.7, 1.8, and 2.5 MLs Ge on Sb/Si(001) are reported in Fig.10. The least square fit of the Si2p spectra obtained accordingly to Ref. 18, shows the evolution of the heterointerface formation.

At 2.5 MLs Ge coverage, the Si2p mainly shows the B doublet: all the Sb/Si surface contributions disappear giving the Si atoms in a bulk-like configuration. The Ge atoms buried beneath the Sb dimers form a pseudomorphic crystal on Si which probably relieving the Si lattice distortion, put the Si atoms in a bulk-like configuration. A picture of a side view of the Ge/Sb/Si(001) heterointerface is reported in Fig. 11.

In conclusion, we found that Sb adsorbed on Si(001)c4x2

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

38

•

1. C. Quaresima, C. Ottaviani, M. Matteucci, C. Crotti, A. Antonini, M. Capozi, S. Rinaldi, M. Luce, P. Perfetti, K. C. Prince, C. Astaldi, M. Zacchigna, L. Romanzin and A. Savoia, Acta Physica Polonica 4, 487, (1994).

2. C. Quaresima, C. Ottaviani, M. Matteucci, C. Crotti, A. Antonini, M. Capozi, S. Rinaldi, M. Luce, P. Perfetti, K. C. Prince, C. Astaldi, M. Zacchigna, L. Romanzin, A. Savoia, Nucl. Instr. and Meth. in Phys.

Res. A 364, 374, (1995) 3. C. Ottaviani, M. Capozi, C. Quaresima, M. Matteucci, C. Crotti, P. Perfetti, C. Astaldi, M. Zacchigna, C. Comicioli, M. Evans, and K. C. Prince, J. of Electron Spectroscopy and Related Phenomena, 76, 139, (1995) 4. M. Zacchigna, C. Astaldi , K. C. Prince, M. Satry, C. Comicioli, R. Rosei, C. Quaresima, C. Ottaviani, C. Crotti, A. Antonini, M. Matteucci, P. Perfetti, Phys. Rev. B 54, 7713, (1996) 5. R. Camilloni, M. Zitnik, C. Comicioli, K. C. Prince, M. Zacchigna, C. Crotti, C. Ottaviani, C. Quaresima, P. Perfetti, and G. Stefani, Phys.

Rev. Lett. 77, 2646, (1996) 6. P. Segovia, E. G. Michel and J. E. Ortega, Phys. Rev. Lett. 77, 3455, (1996) 7. A. Mascaraque, C. Ottaviani, M. Capozi, M. Pedio, and E. G. Michel, Surf. Sci. 377-379, 650, (1997) 8. G. Le Lay, M. Gothelid, V.Yu Aristov, A. Cricenti, C. Hakansson, C. Giammichele, P. Perfetti, J. Avila, and M. C. Ascensio, Surf. Sci., 377379, 1061, (1997) 9. A. Cricenti, C. Ottaviani, C. Comicioli, P. Perfetti, G. Le Lay, Phys.

Rev. B 58, 7086, (1998) 10. M. Copel, M. C. Reuter, M. Horn Von Hoegen, and R. M. Tromp, Phys. Rev. B, 42, 11682, (1990) 11. D. J. Eaglesham and M. Cerullo, Phys. Rev, Lett. 64, 1943, (1990) 12. F. K. LeGoues, M. Copel and R. M. Tromp, Phys. Rev. B 42, 11690 (1990) 13. P. De Padova, R. Larciprete, C. Quaresima, C. Ottaviani, B. Ressel, P. Perfetti, Phys. Rev. Lett. 81, 2320, (1998) 14. J.-H. Cho, M.-H. Kang, K. Terakura, Phys. Rev. B, 55, 15464 (1997) 15. C. T. Chen, Nuclear Instr. and Meth. A256, 595,(1987).

16. C. T. Chen and F. Sette, Rev. Sci. Instrum. 60, 1616, (1989) 17. M. Richter, J. C. Woicik, J. Nogami, P. Pianetta, K. E. Miyano, A. A. Baski, T. Kendelewicz, C. E. Bouldin, W. E. Spicer, C. F. Quarte and I. Lindau, Phys. Rev. Lett., 65, 3417 (1990) 18. E. Landemark, C. J. Karlsson, Y.-C. Chao, and R. I. Uhrberg, Phys.

Rev. Lett. 69, 1588 (1992) 19. A. Goldoni, and S. Modesti, V. R.Dhanak, M Sancrotti, A. Santoni, Phys Rev B, 54, 11340 (1996) 20. L. Patthey, E. L. Bullock, T. Abukawa, S. Kono, L. S. A. Johansson, Phys. Rev. Lett. 73 , 2538, (1195)

Vol. 4 n. 1 Giugno 1999

DOVE NEUTRONI

DESIGN AND PERFORMANCE OF THE NOVEL MULTIDETECTOR NEUTRON SPIN ECHO SPECTROMETER SPAN AT BENSC C. Pappas, G. Kali BENSC, HMI Berlin, Glienicker Str. 100 D-14109 Berlin, Germany

F. Mezei BENSC, HMI Berlin, Glienicker Str. 100, D-14109 Berlin, Germany and Los Alamos National Laboratory, Los Alamos NM 87545 USA

Abstract The new spectrometer SPAN at BENSC combines neutron-spin-echo (NSE) and time-of-flight (TOF) operation and covers a uniquely large dynamic energy transfer range of 5 orders of magnitude. In the TOF configuration the neutrons pass 4 choppers. A typical TOF resolution is 100 µeV at 7 Å. At this wavelength, the present NSE configuration of the spectrometer covers the dynamic range from 51 µeV to 170 neV. The NSE precession field configuration is a novelty in the design of SPAN. It is created by three pairs of coils with diameters 1, 3 and 4.8 m, respectively, which allows for simultaneous NSE measurements over a wide angular range. Each pair of coils is mounted in a Helmholtz-like fashion, with the electric currents oriented oppositely to each other, one coil above and one coil below the horizontal scattering plane. The centers of all coils are placed on the vertical axis, which crosses the sample position.

The maximum magnetic field integral is 0.06 T m. The NSE results confirmed the high axial symmetry of the precession field and show that the new magnetic field configuration of SPAN is particularly appropriate for multidetector NSE over a wide angular range.

The inelastic neutron scattering spectrometer SPAN at BENSC operates with cold neutrons and can either be run as a medium resolution time-of-flight (TOF) or as a high solid angle neutron spin echo 1 (NSE) spectrometer. The combination of these two inelastic scattering techniques leads to a particularly large dynamic energy transfer range of 5 orders of magnitude. For this reason, SPAN is particularly well suited for the study of the dynamics of systems with a broad distribution of relaxation times, such as glasses and spin-glasses, or of systems with strongly q-dependent relaxation rates, such as ferromagnets near their critical temperature Tc. Indeed, the understanding of the dynamics of these systems often requires the combination of spectra obtained by different inelastic neutron scattering techniques, such as TOF and NSE 2,3.

In NSE operation, the precession field is a novelty in the design of SPAN and it optimizes the conditions for multidetector NSE. The magnetic field is created by three pairs of coils with diameters of 1 m, 3 m and 4.8 m, respectively. The set-up has a symmetry axis, which is perpendicular to the scattering plane and crosses the sample position. The resulting magnetic field does not change with the scattering angle.

General description of the spectrometer A view on SPAN is given in Fig. 1. Fig. 2 shows a schematic drawing of the spectrometer seen from the top, with the velocity selector, the NSE coils, and the detector banks, as well as the choppers, necessary for the TOF operation.

SPAN is located at one end of a beam-splitter polarizing guide 4 . The incident wavelength range (2.4 Å ≤ λ ≤ 10 Å) is very large and extends from the cold neutrons to the spectrum of thermal neutrons. The guide is 58 mm wide and 100 mm high.

A monochromatization of 15% FWHM is obtained by a mechanical velocity selector located next to the first chopper at a break in the guide, 5 m from the sample position. The selector is used in both the TOF and the NSE configuration.

The scattered neutrons are detected by 96 3He detectors and a 32x32 cm2 3He multidetector.

The sample-detector distance is 3.5 m. The single detectors (ø 5 cm, height 15 cm) are arranged on three banks with 32 detectors each and an opening of 25.7o per bank. The banks and the multidetector move around the sample at scattering angles ranging from 30o to 150o.

Supermirrors located in front of the detectors analyze the polarization of the scattered neutrons. Presently 10 detectors are equipped with analyzers covering 8o. In the near future 5 more detectors will be equipped with analyzers bringing the overall opening for simultaneous NSE measurements to 12o. In the TOF configuration the analyzer can be replaced by a radial collimator.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

39

DOVE NEUTRONI

Fig. 1 : View of the instrument showing the arrangement of the precession coil pairs: one above and one below the horizontal scattering plane. The coils in an aluminum housing have ø 4.8 m and the brownish coils are the main precession coils with ø 3 m. The phase coil around the incoming beam and the flipper coils are also seen.

TOF configuration In the TOF configuration behind the velocity selector the neutrons pass through two single and one double chopper. The choppers rotate at a maximum speed of 10000 rpm. The disc diameter of 700 mm was chosen because of the large dimensions of the incident beam.

The flight path of 4 m between the first and last chopper is almost equal to the path between the last chopper and the detectors. In order to improve the resolution and flexibility of the set-up, the double chopper lo-

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

40

•

cated just in front of the sample is made of two counterrotating discs.

This design offers the possibility to chose between two different sets of windows. The repetition rate can be varied by a slow chopper placed between the first and the double chopper (see Fig. 2). The TOF resolution typically amounts to 0.1 meV FWHM at 7 Å.

TOF measurements are usually done with all detectors.

It is also possible to combine polarization analysis and TOF, but in this case only those detectors, which are

Vol. 4 n. 1 Giugno 1999

DOVE NEUTRONI

equipped with analyzers, can be used. In the NSE configuration the double chopper, which is located in the area of the NSE precession field, is removed and the other choppers are blocked in open position.

NSE configuration Fig. 3 shows the NSE set-up seen from the side with the precession coils and the flipper coils (π/2 and π). The π/2 flippers, which are located 2.5 m from the sample, mark the beginning and the end of the NSE precession field area. Before the first π/2 flipper the polarization is parallel to the magnetic field (guide field). The π/2 flipper turns the polarization by 90o, so that it becomes perpendicular to the magnetic field (precession field) and the Larmor precessions can start. The π flipper marks the reversal point between the precessions of the incoming and the scattered beams. In the symmetric configuration, where the magnetic field integral of the incoming beam equals that of the scattered beam and the scattering of the sample is purely elastic, the precessions cancel out and the polarization focuses at the position of the second

π/2 flipper. This coil turns the polarization by 90o, so that it becomes again parallel to the magnetic field, and can thus be reflected by the supermirrors of the analyzer.

Fig. 3 also shows the symmetry adjustment or phase coil, which is placed at the side of the incoming beam. In a defined way, this coil modifies the balance between the magnetic field integrals of the incoming and the scattered beams so that the NSE signal, i.e. the amplitude of the Larmor precessions, can be recorded.

The magnetic field configuration of the spectrometer SPAN is new and optimizes the conditions for multidetector NSE. The precession field is created by three pairs of coils with diameters of 1 m, 3 m and 4.8 m shown in Fig. 3. Each pair is mounted in a Helmholtz-like fashion, one coil above and one coil below the scattering plane.

The set-up has a vertical symmetry axis, which crosses the sample position. The magnetic field does not change with the scattering angle and allows for simultaneous NSE measurements over a wide angular range.

The maximum magnetic field integral typically is 0.06 T m, i.e. ~1/3 of IN11. The present NSE configuration covers an energy range of 300 µeV ≤ ω ≤ 1 µeV at

Fig. 2 : Schematic view of SPAN from the top with the NSE precession coils (ø 1 m, ø 3 m and ø 4.8 m) and the choppers used for the TOF measurements. The double chopper is removed in the NSE configuration.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

41

DOVE NEUTRONI

Fig. 3 : Schematic view of the NSE configuration of SPAN from the side with the precession coils (ø 1 m, ø 3 m and ø 4.8 m), the phase coil and the flippers.

3.8Å and 24 µeV ≤ ω ≤ 80 neV at 9 Å respectively. The corresponding Fourier time range ranges from 2.2 ps ≤ t ≤ 0.66 ns at 3.8 Å to 0.027 ns ≤ t ≤ 8.2 ns at 9 Å respectively. The magnetic field configuration of SPAN is described in detail in 5.

The main precession magnetic field is created by the coils with a diameter of 3 m, which have antiparallel electric currents. The resulting magnetic field is horizontal with the required axial symmetry in the scattering plane and becomes zero at the sample position.

The coils with a diameter of 1 m shape the magnetic field around the sample. They produce a weak and homogeneous vertical field at the sample position, which maintains the axial symmetry required in the magnetic configuration. The transition from the vertical field at the sample to the horizontal field of the main coils occurs at a radius of 30 cm to 50 cm.

The low magnetic field around the sample allows for using the basic NSE configuration of SPAN for normal as well as for paramagnetic spin echo experiments. In the latter case the π-flipper is not needed, because the spin flip part of the paramagnetic scattering also behaves like a π-flipper. The amplitude of the paramagnetic scattering, necessary for the normalization of the NSE signal, can be determined from separate 3 directional polarization analysis measurements. The coils with a

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

42

•

diameter of 4.8 m shape the axial component of the magnetic field at the position of the π/2 flippers. In order to keep the axial symmetry of the magnetic field, remanent supermirrors produced at PSI 6 were used as analyzers. These supermirrors perform in the stray magnetic field of the main precession coils and require no additional magnetic field.

Due to the high symmetry of the set-up, the position of the echo slightly depends on the Larmor precession field as shown in Fig. 4. The data of Fig. 4 were collected at λ = 3.8 Å, on the direct beam, for two values of the precession field integral: I = 0.015 T m and 0.03 T m , i.e. 1/4 and 1/2 of the maximum field integral, respectively. The abscissa in Fig. 4 shows the current required in the phase coil, which has 98 turns.

The symmetry of the precession field was confirmed by first NSE measurements 7, which showed that the echo position, located at 2.15 A for the measurements shown in Fig. 4, changes slowly with the scattering angle and that this change is weak enough to allow for multidetector NSE over a large angular range.

In order to further optimize the conditions for multidetector NSE, the π/2 flippers, which are located in front of the detector banks, can tilt and move radially, which reduces the phase shift between neighboring detectors to a minimum. This effect is illustrated in Fig. 5, which

Vol. 4 n. 1 Giugno 1999

DOVE NEUTRONI

shows the NSE spectra of a reference quartz probe. The spectra were simultaneously collected by the ten detectors equipped with supermirrors and show no phase shift from one detector to the next. This feature is of particular importance when the signals are weak and the dynamics are slightly q-dependent, because it allows for directly adding the signal of several detectors prior to the determination of the amplitude of the echo, and thus considerably reduce the counting time.

Concluding remarks After the commissioning of the TOF option, first NSE measurements showed that the new magnetic field configuration of SPAN meets the expectations and is particularly appropriate for multidetector NSE. SPAN has a large maximum scattering angle and a relatively short lowest incident wavelength for a cold neutron spectrometer. The wavelength and momentum transfer (q) range of SPAN is uniquely large for a NSE spectrometer. NSE can be done ranging from cold neutrons to the thermal neutron spectrum and for large q values, which were not accessible for NSE experiments before (at λ = 3 Å : 0.08 Å-1 ≤ q ≤ 4 Å-1).

SPAN is now entering routine operation. It is designed to be a versatile and flexible instrument. TOF measurements can be done with and without polarization analysis. Full 3 directional polarization analysis is implemented and the instrument can be used for normal as well as for paramagnetic NSE experiments.

Fig. 4 : NSE measurements on the direct beam at λ=3.8 Å and for two magnetic field integrals : I = 0.015 T m (closed circles) and 0.03 T m (opened circles). There is no significant change in the position of the echo although the magnetic field integral varies by a factor of two. The continuous line is a fit to the data measured for I = 0.03 T m.

References 1. F. Mezei in Neutron Spin Echo, Lecture Notes in Physics, Vol. 122, Springer Verlag Berlin, 1980.

2. F. Mezei, J. Magn. Magn. Mat. 31-34 (1982) 1096 W. Knaak, F. Mezei, B. Farago, Europhys. Lett. 7 (1988) 529 3. F. Mezei, Phys. Rev. Lett. 49 (1982) 1096 C. Pappas, M. Alba, C. Lartigue, F. Mezei, Physica B 180-181 (1992) 359 4. Th. Krist, C. Pappas, Th. Keller, F. Mezei, Physica B 213&214 (1995) 939 5. C. Pappas and F. Mezei, J. Neutron Research, Vol. 5 (1996) 35 6. P. Böni, D. Clemens, M. Senthil Kumar and C. Pappas, to appear in Physica B, 267-268 (1999) 7. C. Pappas, G. Kali, P. Böni, R. Kischnik, L.A. Mertens, P. Granz and F. Mezei, to appear in Physica B, 267-268 (1999) 285

Fig. 5 : NSE spectra of a reference quartz probe, collected simultaneously by the ten detectors equipped with supermirrors (angular range 8o). There is no noticeable change of the position of the echo from one detector to the next. This result was obtained after adequate tilting of the π/2 flipper, which is located in front of the detectors. The incident wavelength was 4.5 Å and the magnetic field equaled I = 0.03 T m.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

43

VARIE

Stato del progetto TOSCA: primi risultati sperimentali La sezione in back-scattering dello spettrometro TOSCA è stata consegnata al Rutherford Appleton Laboratory, nei termini stabiliti, il 16 Febbraio 1998. Successivamente, e per la durata di qualche mese, sono iniziate le opere per la messa in esercizio dello strumento. Queste opere sono consistite nella realizzazione degli opportuni schermaggi, connessioni elettriche, circuiti di trasporto dell'elio, linee a vuoto ancillari, elettronica di acquisizione, etc.

La zona del sample-environment è stata dotata di un criostato in grado di ospitare lo stick di un orange-cryostat, ma il cui potere refrigerante è fornito da tre teste fredde di altrettanti refrigeratori a circuito chiuso (CCR) di elio. In tale configurazione, il sistema è in grado di raffreddare campioni fino a 15K in tempi molto rapidi.

Lo strumento è stato installato nella sua posizione durante lo shutdown di MarzoAprile 1998 ed ha visto i primi neutroni durante il ciclo di Giugno 1998. In questa faFig. 1 se, sono state effettuate le necessarie misure di commissioning. Queste misure, sebbene siano state ridotte al minimo, su precise indicazioni da parte della dirigenza di ISIS a causa dell'elevata pressione da parte degli users, hanno comunque evidenziato i vantaggi di TOSCA sul vecchio strumento (TFXA) che TOSCA ha sostituito.

Sebbene il netto miglioramento in termini di potere risolutivo sia basato essenzialmente sullo spostamento dello strumento dagli attuali 12 metri di distanza dal moderatore ai previsti 17 metri (spostamento che verrà attuato durante la fase-2, 1999-2000)

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

44

sono comunque osservabili notevoli progressi già nella configurazione attuale. In pratica, la situazione è evidenziata nella Fig. 1 che riporta il calcolo del potere risolutivo di TFXA, quello di TOSCA nella fase attuale, e quello di TOSCA nella configurazione finale. Vale la pena sottolineare che nel passaggio da TFXA a TOSCA-1, il fattore migliorativo è

determinato soltanto dalla riduzione dello spessore dei rivelatori dagli originali 6 mm di TFXA agli attuali 2.5 mm di TOSCA e dalla riduzione in lunghezza attiva degli stessi.

I punti sperimentali che sono riportati in Fig. 1 sono stati ottenuti confrontando lo spettro di un calibrante, il di-iodio-tiofene, misurato in condizioni analoghe sia su TFXA che su TOSCA. Spettri tipici, che sono stati usati nel confronto, sono riportati in Fig. 2. Risalta immediatamente la maggior risoluzione di TOSCA nella

•

Vol. 4 n. 1 Giugno 1999

zona a bassa energia. Infatti, il picco a ~10 meV risulta di ampiezza quasi doppia rispetto al corrispondente di TFXA mentre il doppietto a ~28 meV appare risolto molto meglio che nel caso precedente. Naturalmente, spostandosi ad alta energia, le differenze tendono ad attenuarsi e sarà necessario attendere lo spostamento dello strumento a 17 metri per ottenere la risoluzione finale prevista. E' interessante notare, comunque, che i calcoli sono confortati dal dato sperimentale (cf. Fig. 1) e che pertanto niente lascia presagire che la risoluzione finale sarà inferiore alle aspettative.

Durante i primi mesi di attività dello strumento, sono stati effettuati alcuni esperimenti proposti da gruppi italiani. Tali esperimenti sono elencati di seguito in ordine di esecuzione e, per semplicità, riportiamo soltanto il nome e la sede del proponente principale: 1 Alberto Albinati (Istituto Chimico Farmaceutico Tossicologico dell’Università di Milano), 2 Roberto Caciuffo (Dipartimento di Scienza dei Materiali e della Terra, Università di Ancona), 3 Piero Baglioni (Dipartimento di Chimica, Università di Firenze), 4 Francesco Aliotta (Istituto di Tecniche Spettroscopiche, CNR, Messina), 5 Valerio Rossi (Istituto di Struttura della Materia, CNR, Roma), 6 Marco Zoppi (Istituto di Elettronica Quantistica, CNR, Firenze).

Tutti gli esperimenti hanno avuto un esito positivo e tutto lascia presagire che le misure effettuate, che nella maggior parte dei casi devono essere intese come esplorative, avranno un seguito importante nel futuro di TOSCA. Vale la pena aggiungere che, sebbene lo schema di progetto dello stru-

VARIE

mento preveda un time-focusing su tutta la superficie dei detectors, la lunghezza d'onda della riflessione di Bragg sulla grafite dipende dal punto di incidenza e quindi, in definitiva, dall'angolo di scattering. La configurazione attuale di TOSCA permette l'analisi dei dati sperimentali operando una partizione arbitraria sui rivelatori, in maniera cioè da ridurre a piacere (ovviamente fino al limite del singolo detector) lo spread sulle energie finali e quindi sul momento trasferito.

misure hanno portato interessanti risultati ma hanno anche evidenziato la presenza nello spettro degli ordini superiori generati dallo scattering di Bragg della grafite.

La soppressione di questi picchi spuri ha notevolmente impegnato il comitato di progetto. In effetti, nella parte in back scattering, tale soppressione è ottenuta per mezzo dei filtri di Berillio, i quali però divengono inefficaci quando l'angolo di scattering sulla grafite diminuisce al di sotto del limite imposto dalla so-

Sempre nel gennaio 1999, è stato effettuato ad ISIS un meeting per discutere delle problematiche inerenti lo spostamento dello strumento a 17 metri, nonché la localizzazione e le caratteristiche della stazione sperimentale italiana che dovrà essere costruita a valle dello strumento. Nel corso della riunione (presenti: D. Colognesi, C. Petrillo, F. Sacchetti e M.Zoppi) è stato definito un calendario di incontri che dovrà accompagnare la fase realizzativa di TOSCA2 fino all'installazione che è prevista

glia del berillio. La sezione in forward scattering di TOSCA, invece, era stata pensata in una configurazione constant-Q nella quale il berillio diventa inefficace per filtrare gli ordini superiori. In questo caso, diviene interessante la possibilità di usare un filtro a paraidrogeno liquido. Lo studio di fattibilità di una tale soluzione è stata oggetto di un secondo test effettuato, sempre su ROTAX, nel gennaio 1999 da parte di F. Sacchetti, C. Petrillo, M Zoppi e D. Colognesi.

in occasione dello shutdown di febbraio dell'anno 2000.

Fig. 1: The spectrum of 2.5 diiodothiophene.

Nell'ambito della determinazione della geometria ottimale per la fase-2 di TOSCA, sono state effettuate misure di test atte ad evidenziare pregi e difetti delle possibili configurazioni proposte. Una prima misura di test è stata effettuata sullo strumento ROTAX nel Gennaio 1998 (F. Sacchetti e C. Petrillo, Dipartimento di Fisica, Università di Perugia). In questa fase, sono state controllate le configurazioni geometriche fondamentali per lo scattering in avanti, usando campioni standard. Queste

Vol. 4 n. 1 Giugno 1999

•

M. Zoppi, M. Celli Consiglio Nazionale delle Ricerche Istituto di Elettronica Quantistica, Firenze, Italy D. Colognesi Consiglio Nazionale delle Ricerche ISIS-Rutherford Laboratory, Chilton, Didcot, U.K.

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

45

VARIE

Attività di ricerca della comunità italiana presso la sorgente di neutroni Isis (UK) Percentuale di utilizzo del tempo macchina.

ISIS Instruments

Days Requested

HRPD POL SXD SANDALS CRISP SURF LOQ HET MARI PRISMA eVS TOSCA IRIS MuSR EMu ARGUS DEVA PEARL ROTAX OSIRIS TOTAL

% Requested

8 5 0 13 4 0 32 0 11 0 14 10 9 14 7 0 0 0 0 0

7,0 4,2 0,0 11,2 3,5 0,0 21,9 0,0 6,5 0,0 12,5 11,2 7,3 14,3 7,7 0,0

127

Days Allocated

% Allocated 9,7 4,2 0,0 15,9 5,3 0,0 6,3 0,0 0,0 0,0 11,8 12,0 10,7 11,7 8,9 0,0

0,0 0,0 0,0

7 3 0 10 4 0 5 0 0 0 9 6 8 9 7 0 0 0 0 0

6,6

68

5,6

0,0 0,0 0,0

ITALY 25 20 15 10 5

Total

OSIRIS

ROTAX

PEARL

DEVA

ARGUS

EMu

MuSR

IRIS

TOSCA

eVS

PRISMA

MARI

HET

LOQ

SURF

CRISP

SANDALS

SXD

POL

HRPD

0

ITALY 35 30 25 20 15 10 5

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

46

•

Vol. 4 n. 1 Giugno 1999

OSIRIS

ROTAX

PEARL

DEVA

ARGUS

EMu

MuSR

IRIS

TOSCA

eVS

PRISMA

MARI

HET

LOQ

SURF

CRISP

SANDALS

SXD

POL

HRPD

0

VARIE

ISIS Round

ITALY Requested Days

%

ITALY Allocated Days

%

222,3 203,0 198,3 198,8 170,3 176,0 169,3 135,3 167,4 172,0 158,6 112,3 123,6 190,4 141,8 119,7 143,7 102,0 164 123 189 127

11,8 10,3 11,4 10,7 8,8 7,9 8,1 4,7 7,3 6,1 6,7 4,6 5,0 8,2 5,7 5,3 5,9 4,6 7,5 5,7 8,8 6,6

100,1 55,1 64,8 33,0 53,5 61,8 57,4 49,8 61,3 45,9 51,9 39,4 51,1 72,4 56,5 53,3 57,7 56,3 71 52 92 68

10,5 6,0 9,8 5,5 7,6 7,9 5,8 4,6 7,5 4,9 5,1 4,2 4,9 6,5 4,7 4,6 4,7 4,5 6,0 4,1 6,6 5,6

88/1 89/1 89/2 90/1 90/2 91/1 91/2 92/1 92/2 93/1 93/2 94/1 94/2 95/1 95/2 96/1 96/2 97/1 97/2 98/1 98/2 99/1 TOTAL

3507,8

1304,3

ITALY

% time

14 12 10 8 6 4 2

Vol. 4 n. 1 Giugno 1999

•

99/1

98/2

98/1

97/2

97/1

96/2

96/1

95/2

95/1

94/2

94/1

93/2

93/1

92/2

92/1

91/2

91/1

90/2

90/1

89/2

89/1

88/1

0

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

47

VARIE

European Spallation Source (ESS) Negli ultimi anni un consorzio di istituzioni Europee e di enti di ricerca ha condotto, grazie ad un finanziamento comunitario, uno studio preliminare per avviare la costruzione di una nuova sorgente neutronica europea e prodotto alla fine del 1996, un progetto di fattibilità per una nuova sorgente di neutroni, con potenza del fascio di 5 MW, circa 30 volte più intensa di ISIS. Il documento, in tre volumi, è disponibile sul sito webhttp://www.kfajuelich.de/ess/PRO/ Council.html).

In seguito, nel 1997, è stato costituito l’ESS R&D Council con l’obiettivo di predisporre un progetto tecnico definitivo, sia della sorgente che della strumentazione, da presentare ai governi dei paesi della Unione Europea.

Le Aree individuate dal Council per le attività di Ricerca e Sviluppo (R&D) sono: • Area 1: Linac, anelli di accumulazione e linee di trasferimento dei protoni • Area 2: Targhette e moderatori • Area 3: Sviluppo di Strumentazione Originariamente l’ESS R&D Council è stato costituito a seguito della firma di un “Memorandum of Understanding” (MoU) da parte di cinque istituzioni scientifiche appartenenti a diversi paesi europei. Successivamente sono entrati a fare parte del Council anche altri Istituzioni e Laboratori europei ed attualmente i componenti firmatari del Memorandum sono undici: CIEMAT Madrid (Spagna) CEA (Francia) CLLRC-ISIS (UK) CNR (Italia) Forschunszentrum Jülich, GmbH (Germania) HMI Berlino (Germania) INFM (Italia) IRI Delft (Olanda) Paul Scherrer Institute (Svizzera) Risø (Danimarca) Università di Uppsala (Svezia)

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

48

I rappresentanti nel Council nominati dal CNR e dall’INFM sono rispettivamente il prof. M. Fontanesi ed il prof. F. Barocchi.

Per quanto riguarda il CNR l’impegno dell’ente fin qui svolto è relativo alle attività di R&D previste nell’Area 3. Infatti il CNR svolge fin dal 1985 attività di ricerca nell’ambito della spettroscopia di neutroni: – garantendo l’accesso alla comunità italiana alla sorgente di neutroni ISIS, con una percentuale di utilizzo della stumentazione disponibile ad ISIS pari al 5%/anno, attraverso un accordo con il Central Laboratory of the Research Council (CLRC); – finanziando direttamente lo sviluppo di strumentazione, progettata e costruita presso i proprio organi di ricerca (nel periodo 1985/1995 Progetto PRISMA presso l’ISM (Frascati) e nel periodo 1996/2001 Progetto TOSCA presso l’IEQ (Firenze), con il supporto di personale esterno dalle università.

Le attività di ricerca relative al progetto TOSCA sono visibili al sito http://laser.ieq.fi.cnr.it/ projects/tosca/staff.htm Nel periodo 1985-1995 l’impegno complessivo del CNR è stato, in media, di 4 uomini anno/anno ed ha comportato un investimento pari a 1.5 MECU (Progetto PRISMA e sviluppo di cristalli monocromatori). In aggiunta l’accesso alla sorgente ISIS ha comportato un investimento pari a 10 MECU. Nel periodo 1996-2001 si prevede che l’impegno complessivo del CNR sarà in media di 4.5 uomini anno/anno con un investimento complessivo pari a 1.5 MECU (Progetto TOSCA e sviluppo di cristalli monocromatori). In aggiunta l’accesso alla sorgente ISIS comporterà un investimento pari a 6 MECU. Il contributo italiano nel suo complesso è il risultato di attività di ri-

•

Vol. 4 n. 1 Giugno 1999

cerca svolte da gruppi CNR in collaborazione con gruppi INFM localizzati presso le università. I fondi destinati a tali attività in ambito ESS R&D derivano direttamente dal CNR e dalla Comunità Europea. Le attività di ricerca italiana CNR ed INFM in ambito ESS possono essere così riassunte: Ente CNR :

Attività di ricerca TOSCA, strumentazione per studi di spettroscopia molecolare.

INFM: Sviluppo di rivelatori a stato solido.

INFM: VESUVIO, strumentazione per spettroscopia di neutroni agli eV. CNR/INFM: Sviluppo di cristalli monocromatori a grande area.

L’attività italiana in ambito ESS è stata valutata in 11.5 uomo anno/per anno corrispondente ad un investimento complessivo pari a 2.6 MECU. Le decisioni più rilevanti assunte dal Council, nelle ultime due riunioni di Madrid (28/10/1998) e Villigen (18/5/1999), si possono così riassumere: – si è deciso di definire il progetto realizzativo dell’ ESS entro il 2003; – di procedere in tempi brevi all’assunzione di un project manager per il progetto ESS. – di predisporre un documento, che illustri lo sviluppo temporale del progetto ESS e le specifiche opzioni tecniche della sorgente, da presentare alla riunione dei ministri europei nell’ambito della Conferenza Intergovernativa dell’OECD Megascience prevista in Giugno 1999.

La prossima riunione del Council avrà luogo a Delft (Olanda) il giorno 11 Novembre.

C. Andreani

SCUOLE E CONVEGNI

SILS (Società Italiana di Luce di Sincrotrone)

organizza la

V Scuola Nazionale di Luce di Sincrotrone Santa Margherita di Pula (Cagliari) 27 settembre - 8 ottobre 1999 I Direttori della Scuola Prof. Settimio MOBILIO e-mail: mobilio@lnf.infn.it Prof. Gilberto VLAIC e-mail: vlaic@elettra.trieste.it

Scopi e programma della Scuola La Scuola intende offrire a persone già operanti nel campo della Luce di Sincrotrone o interessate ad entrarvi una panoramica attuale delle caratteristiche e potenzialità dell’uso della stessa. Le possibilità di ricerca con L.S. saranno affrontate sia da un punto di vista teorico che sperimentale e viste nella loro connessione a varie discipline (chimica, fisica, biologia, scienze della terra) e a diversi tipi di materiali. La Scuola si articolerà in circa 70 ore di lezione. Sarà organizzata una sessione poster informale in modo da favorire gli scambi di idee e la nascita di collaborazioni. Il numero di partecipanti alla Scuola è limitato a 50 persone. Le domande verranno accettate in ordine cronologico di arrivo alla Segreteria Organizzativa. La tassa di iscrizione è fissata in Lire 500.000.

Sede della Scuola e sistemazione alberghiera La scuola si svolgerà presso la sala congressi dell’Hotel Flamingo, Santa Margherita di Pula (CA). I partecipanti saranno alloggiati presso l’Hotel Flamingo o presso l’Hotel Mare e Pineta, situati all’interno di un parco privato di pini marittimi e di euca-

liptus e distanti tra loro circa 5 minuti a piedi. Il complesso è dotato di piscina, spiaggia privata, minigolf e campi da tennis.

Prezzi di pensione completa per persona in camera doppia con servizi (bevande escluse): Hotel Flamingo**** 118000 lire/giorno Hotel Mare e Pineta*** 98000 lire/giorno Supplemento camera singola 20000 lire/giorno

Le condizioni di miglior favore praticate ai partecipanti saranno valide anche nel fine settimana precedente e in quello successivo alla Scuola.

Hotel Flamingo e Mare e Pineta Hotel SS 195 KM. 33.800 I-09010 S. Margherita di Pula (CA) Tel. 0709208361 Fax 0709208359 www.tiscalinet.it/flamingo e-mail: flamingo@tiscalinet.it

Riduzione tripla/quadrupla 25%

Scheda di iscrizione Alla segreteria Organizzativa Carla DI GIALLORENZO Paola BOSI CNR, DAS - Reparto III Progetto Sincrotroni Via Tiburtina 770 , 00159 ROMA Tel. 0649932467, 0649932468 Fax 0649932456 e-mail: p.bosi@dcas.cnr.it

Nome e cognome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Indirizzo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Telefono . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Fax

.........................................................................................

E-Mail

....................................................................................

Eventuale richiesta motivata di finanziamento (*) ...............................................................................................

Prevedo di partecipare alla sezione poster con un lavoro dal titolo: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...............................................................................................

Verserò la somma di lire 500.000 quale tassa di iscrizione appena ricevuta conferma della mia ammissione da parte degli organizzatori) e provvederò a inviare la scheda di prenotazione alberghiera (che mi sarà fornita dalla Segreteria Organizzativa). (*) La SILS mette a disposizione alcune borse di studio intitolate alla memoria di Carla Cauletti. L’entità delle borse è di L. 500.000; inviare curriculum vitae al Prof.

Gilberto Vlaic.

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

49

CALENDARIO

14-18 giugno 1999

CATANIA, ITALY

INFMeeting http://www.infm.it

1-3 luglio 1999

L’AQUILA, ITALY

GRANADA, SPAIN

IV Liquid Matter Conference Prof. Dr. Roque Hidalgo Elvarez, Depart. de Fisica Aplicada, Facultad de Ciencias, Universidad de Granada, Campus de Fuentenueva, E-18071 Granada (Spagna) Tel: +34 958 243213; Fax +34 958 243214 e-mail: liquid99@ugr.es http://www.ugr.es/~liquid99

26-30 luglio 1999

1-4 settembre 1999

5-7 settembre 1999

20-24 settembre 1999

BERKELEY, CALIFORNIA, USA

GLASGOW, SCOTLAND

ZUOZ, SWITZERLAND

7th Summer School of Neutron Scattering “Neutron Scattering in the Next Millennium” Renate Bercher, Paul Scherrer Inst., CH-5232 Villigen PSI, Switzerland Tel: +41 563103402; fax: +41 563103131 e-mail: bercher@psi.ch

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

50

•

STUTTGART, GERMANY

7th International Conference on Quasicrystals ICQ7’99, Institut für Theoretisch und Angewandte Physik, Universitat Stuttgart Tel: +49 711 685 5253 / 5254; Fax: +49 711 685 5271 icq@itap.physik.uni-stuttgart.de

18th IUCr Gen. Assembly and International Congress of Crystallography http://www.chem.gla.ac.uk/iucr99

7-13 agosto 1999

SCHWAEBISCH GMUND, GERMANY

International Conference on Solid State Spectroscopy http://cardix.mpi-stuttgart.mpg.de/icss/

International Conference on X-ray Microscopy e-mail: xrm99@lbl.gov

4-13 agosto 1999

BUDAPEST, HUNGARY

Second European Conference on Neutron Scattering (ECNS ‘99) Dr. Tamos Grûsz, Neutron Physics Laboratory Research Institute for Solid State Physics and Optics, H-1525 Budapest, P.O.B. 49, KFKI, Bldg. 10, Hungary Tel. +36 1 395 9220/1738; Fax +36 1 395 9165 e-mail: ecns99@sunserv.kfki.hu http://www.kfki.hu/ecns99/

VANCOUVER, CANADA

6th International Conference on the Structure of Surfaces (ICSOS-6) e-mail: karm@chem.ubc.ca

1-7 agosto 1999

CHICAGO, USA

X-99 18th International Conference on X-ray and inner shell processes http://www.phy.anl.gov/X99

Settimo Conv. della Società Italiana Luce Sincrotrone e-mail: sils99@aquila.infn.it http://www.chem.uniroma1.it/~dicastro/Convegno.htm

3-7 luglio 1999

23-27 agosto 1999

Vol. 4 n. 1 Giugno 1999

CALENDARIO

21-24 settembre 1999

VIENNA, AUSTRIA

ECOSS-18, 18th European Conference on Surface Science Institut für Allgemeine Physik, U Wien, Wiedner Haupstr. 8-10/134 e-mail: ecoss18-secretary@iap.tuwien.ac.at

23-25 settembre 1999

26-29 gennaio 2000

International Workshop on Dynamics in Confinement I. Volino, Institut Laue-Langevin, B.P. 156, F-38042 Grenoble Cedex 9, France Tel: +33 4 76207060; Fax: +33 4 76483906 e-mail: confit@ill.fr http://www.ill.fr/Events/confit.htlm

FRASCATI, ITALY

SRRTNET Workshop Workshop on theory and computation for synchrotron radiation e-mail: wsp99@lnf.infn.it http://www.sis.lnf.infn.it/SRRTNET99/Welcome.html

10-12 febbraio 2000

29 settembre - 2 ottobre 1999

DUBNA, RUSSIA

2nd International Seminar on Neutron Scattering at High Pressure (NSHP-II) D.P. Kozlenko, Frank Laboratory of Neutron Physics, Joint Institute for Nuclear Reserach, 141980 Dubna, Moscow Reg., Russia Tel: +7 09621 65644; Fax: +7 09621 65882 e-mail: denk@nf.jinr.ru http://nfdfn.jinr.ru/~denk/NSHPII/

4-8 ottobre 1999

SEVILLA, SPAIN

8th European Conference on Applications of Surface and Interface Analysis e-mail: ecasia@uam.es

STANFORD, CA, USA

National Synchrotron Radiation Instrumentation Conference e-mail: barrett@slac.stanford.edu

28-29 ottobre 1999

GRENOBLE, FRANCE

ESRF Users’ Meeting http://www.esrf.fr

20-24 marzo 2000

13-15 ottobre 1999

GRENOBLE, FRANCE

MINNEAPOLIS, MN, USA

APS March Meeting http://www.aps.org

24-28 aprile 2000

SAN FRANCISCO, CA, USA

Materials Research Society Spring Meeting http://dns.mrs.org

20-23 maggio 2000

BARCELLONA, SPAIN

7th European Powder Diffraction Conference (EPDIC7) e-mail: gcasanova@uex.es

21-25 agosto 2000

BERLIN, GERMANY

7th International Conference on Synchrotron Radiation Instrumentation http://sri2000.tu-berlin.de

ROMA, ITALY

V Convegno Nazionale Materiali Nanofasici e-mail: ianni@axcasp.caspur.it http://nanofasici.univaq.it

29 novembre - 3 dicembre 1999

BOSTON, USA

Materials Research Society Fall Meeting http://dns.mrs.org

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

51

SCADENZE

Scadenze per richieste di tempo macchina presso alcuni laboratori di Neutroni

Scadenze per richieste di tempo macchina presso alcuni laboratori di Luce di Sincrotrone

ISIS

ALS

La scadenza per il prossimo call for proposals è il 16 ottobre 1999 e il 16 aprile 2000

Le prossime scadenze sono il 1 dicembre 1999 e il 1 giugno 2000

ILL

BESSY

La scadenza per il prossimo call for proposals è il 1 marzo 2000

Le prossime scadenze sono il 15 agosto 1999 e il 15 febbraio 2000

LLB-SACLAY

DARESBURY

La scadenza per il prossimo call for proposals è il 1 ottobre 1999

La prossima scadenza è il 24 novembre 1999

BENSC

ELETTRA

La scadenza è il 15 settembre 1999 e il 15 marzo 2000

Le prossime scadenze sono il 31 agosto 1999 e il 28 febbraio 2000

RISØ E NFL La scadenza per il prossimo call for proposals è il 1 aprile 2000

ESRF Le prossime scadenze sono il 1 settembre 1999 e il 1 marzo 2000

GILDA (quota italiana) Le prossime scadenze sono il 1 novembre 1999 e il 1 maggio 2000

HASYLAB (nuovi progetti) Le prossime scadenze sono il 1 settembre e il 1 dicembre 1999 e il 1 marzo 2000

LURE La prossima scadenza è il 30 ottobre 1999

MAX-LAB La scadenza è approssimativamente febbraio 2000

NSLS Le prossime scadenze sono il 30 settembre 1999, il 31 gennaio e il 31 maggio 2000

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

52

•

Vol. 4 n. 1 Giugno 1999

FACILITIES

LUCE DI SINCROTRONE SYNCHROTRON SOURCES WWW SERVERS IN THE WORLD (http://www.esrf.fr/navigate/synchrotrons.html)

DAFNE INFN Laboratori Nazionali di Frascati, P.O. Box 13, I-00044 Frascati (Rome), Italy tel: +39 6 9403 1 fax: +39 6 9403304 http://www.lnf.infn.it/ Tipo:P Status: C

ALS Advanced Light Source MS46-161, 1 Cyclotron Rd Berkeley, CA 94720, USA tel:+1 510 486 4257 fax:+1 510 486 4873 http://www-als.lbl.gov/ Tipo: D Status: O AmPS Amsterdam Pulse Stretcher NIKEF-K, P.O. Box 41882, 1009 DB Amsterdam, NL tel: +31 20 5925000 fax: +31 20 5922165 Tipo: P Status: C

DELTA Universität Dortmund,Emil Figge Str 74b, 44221 Dortmund, Germany tel: +49 231 7555383 fax: +49 231 7555398 http://prian.physik.uni-dortmund.de/ Tipo: P Status: C

APS Advanced Photon Source Bldg 360, Argonne Nat. Lab. 9700 S. Cass Avenue, Argonne, Il 60439, USA tel:+1 708 252 5089 fax: +1 708 252 3222 http://epics.aps.anl.gov/welcome.html Tipo: D Status: C

ELETTRA Sincrotrone Trieste, Padriciano 99, 34012 Trieste, Italy tel: +39 40 37581 fax: +39 40 226338 http://www.elettra.trieste.it Tipo: D Status: O

ASTRID ISA, Univ. of Aarhus, Ny Munkegade, DK-8000 Aarhus, Denmark tel: +45 61 28899 fax: +45 61 20740 Tipo: PD Status: O

ELSA Electron Stretcher and Accelerator Nußalle 12, D-5300 Bonn-1, Germany tel:+49 288 732796 fax: +49 288 737869 http://elsar1.physik.uni-bonn.de/elsahome.html Tipo: PD Status: O

BESSY Berliner Elektronen-speicherring Gessell.für Synchrotron-strahlung mbH Lentzealle 100, D-1000 Berlin 33, Germany tel: +49 30 820040 fax: +49 30 82004103 http://www.bessy.de Tipo: D Status: O

ESRF European Synchrotron Radiation Lab. BP 220, F-38043 Grenoble, France tel: +33 476 882000 fax: +33 476 882020 http://www.esrf.fr/ Tipo: D Status: O

BSRL Beijing Synchrotron Radiation Lab. Inst. of High Energy Physics, 19 Yucuan Rd.PO Box 918, Beijing 100039, PR China tel: +86 1 8213344 fax: +86 1 8213374 http://solar.rtd.utk.edu/~china/ins/IHEP/bsrf/bsrf.html Tipo: PD Status: O CAMD Center Advanced Microstructures & Devices Lousiana State Univ., 3990 W Lakeshore, Baton Rouge, LA 70803, USA tel:+1 504 3888887 fax: +1 504 3888887 http://www.camd/lsu.edu/ Tipo: D Status: O CHESS Cornell High Energy Synchr. Radiation Source Wilson Lab., Cornell University Ithaca, NY 14853, USA tel: +1 607 255 7163 fax: +1 607 255 9001 http://www.tn.cornell.edu/ Tipo: PD Status: O

EUTERPE Cyclotron Lab.,Eindhoven Univ. of Technol, P.O.Box 513, 5600 MB Eindhoven, The Netherlands tel: +31 40 474048 fax: +31 40 438060 Tipo: PD Status: C HASYLAB Notkestrasse 85, D-2000, Hamburg 52, Germany tel: +49 40 89982304 fax: +49 40 89982787 http://www.desy.de/pub/hasylab/hasylab.html Tipo: D Status: O INDUS Center for Advanced Technology, Rajendra Nagar, Indore 452012, India tel: +91 731 64626 Tipo: D Status: C

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

53

FACILITIES

KEK Photon Factory Nat. Lab. for High Energy Physics, 1-1, Oho, Tsukuba-shi Ibaraki-ken, 305 Japan tel: +81 298 641171 fax: +81 298 642801 http://www.kek.jp/ Tipo: D Status: O Kurchatov Kurchatov Inst. of Atomic Energy, SR Center, Kurchatov Square, Moscow 123182, Russia tel: +7 95 1964546 Tipo: D Status:O/C

SOR-RING Inst. Solid State Physics S.R. Lab, Univ. of Tokyo, 3-2-1 Midori-cho Tanashi-shi, Tokyo 188, Japan tel: +81 424614131 ext 346 fax: +81 424615401 Tipo: D Status: O SRC Synchrotron Rad. Center Univ.of Wisconsin at Madison, 3731 Schneider DriveStoughton, WI 53589-3097 USA tel: +1 608 8737722 fax: +1 608 8737192 http://www.src.wisc.edu Tipo: D Status: O SRRC SR Research Center 1, R&D Road VI, Hsinchu Science, Industrial Parc, Hsinchu 30077 Taiwan, Republic of China tel: +886 35 780281 fax: +886 35 781881 http://www.srrc.gov.tw/ Tipo: D Status: O

LNLS Laboratorio Nacional Luz Sincrotron CP 6192, 13081 Campinas, SP Brazil tel: +55 192 542624 fax: +55 192 360202 Tipo: D Status: C LURE Bât 209-D, 91405 Orsay ,France tel: +33 1 64468014; fax: +33 1 64464148 E-mail: lemonze@lure.u-psud.fr http://www.lure.u-psud.fr Tipo: D Status: O

SSRL Stanford SR Laboratory MS 69, PO Box 4349 Stanford, CA 94309-0210, USA tel: +1 415 926 4000 fax: +1 415 926 4100 http://www-ssrl.slac.stanford.edu/welcome.html Tipo: D Status: O

MAX-Lab Box 118, University of Lund, S-22100 Lund, Sweden tel: +46 46 109697 fax: +46 46 104710 http://www.maxlab.lu.se/ Tipo: D Status: O NSLS National Synchrotron Light Source Bldg. 725, Brookhaven Nat. Lab., Upton, NY 11973, USA tel: +1 516 282 2297 fax: +1 516 282 4745 http://www.nsls.bnl.gov/ Tipo: D Status: O NSRL National Synchrotron Radiation Lab. USTC, Hefei, Anhui 230029, PR China tel:+86 551 3601989 fax:+86 551 5561078 Tipo: D Status: O Pohang Pohang Inst. for Science & Technol., P.O. Box 125 Pohang, Korea 790600 tel: +82 562 792696 f +82 562 794499 Tipo: D Status: C

SRS Daresbury SR Source SERC, Daresbury Lab, Warrington WA4 4AD, U.K. tel: +44 925 603000 fax: +44 925 603174 E-mail: srs-ulo@dl.ac.uk http://www.dl.ac.uk/home.html Tipo: D Status: O SURF B119, NIST, Gaithersburg, MD 20859, USA tel: +1 301 9753726 fax: +1 301 8697628 http://physics.nist.gov/MajResFac/surf/surf.html Tipo: D Status: O TERAS ElectroTechnical Lab. 1-1-4 Umezono, Tsukuba Ibaraki 305, Japan tel: 81 298 54 5541 fax: 81 298 55 6608 Tipo: D Status: O UVSOR Inst. for Molecular ScienceMyodaiji, Okazaki 444, Japan tel: +81 564 526101 fax: +81 564 547079 Tipo: D Status: O

Siberian SR Center Lavrentyev Ave 11, 630090 Novosibirsk, Russia tel: +7 383 2 356031 fax: +7 383 2 352163 Tipo: D Status: O SPring-8 2-28-8 Hon-komagome, Bunkyo-ku ,Tokyo 113, Japan tel: +81 03 9411140 fax: +81 03 9413169 Tipo: D Status: C

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

54

•

D = macchina dedicata; PD = parzialmente dedicata; P = in parassitaggio.

O= macchina funzionante; C=macchina in costruzione.

D = dedicated machine; PD = partially dedicated; P = parassitic.

O= operating machine; C= machine under construction.

Vol. 4 n. 1 Giugno 1999

FACILITIES

NEUTRONI NEUTRON SCATTERING WWW SERVERS IN THE WORLD (http://www.isis.rl.ac.uk)

BENSC Berlin Neutron Scattering Center, Hahn-Meitner-Institut, Glienicker Str. 100, D- 14109 Berlin-Wannsee, Germany Rainer Michaelsen; tel: +49 30 8062 3043 fax: +49 30 8062 2523 E - Mail: michaelsen@hmi.de http://www.hmi.de BNL Brookhaven National Laboratory, Biology Department, Upton, NY 11973, USA Dieter Schneider; General Information: Rae Greenberg; tel: +1 516 282 5564 fax: +1 516 282 5888 http://neutron.chm.bnl.gov/HFBR/ GKSS Forschungszentrum Geesthacht, P.O.1160, W-2054 Geesthacht, Germany Reinhard Kampmann; tel: +49 4152 87 1316 fax: +49 4152 87 1338 E-mail: PWKAMPM@DGHGKSS4 Heinrich B. Stuhrmann; tel: +49 4152 87 1290 fax: +49 4152 87 2534 E-mail: WSSTUHR@DGHGKSS4

ISIS The ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot Oxfordshire OX11 0QX, UK Richard Heenan; tel +44 235 446744 E-mail: RKH@UK.AC.RUTHERFORD.DEC-E Steve King; tel: +44 235 446437 fax: +44 235 445720; Telex: 83 159 ruthlb g E-mail: SMK@UK.AC.RUTHERFORD.DEC-E http://www.nd.rl.ac.uk JAERI Japan Atomic Energy Research Institute, Tokai-mura, Naka-gun, Ibaraki-ken 319-11, Japan.

Jun-ichi Suzuki (JAERI); Yuji Ito (ISSP, Univ. of Tokyo); fax: +81 292 82 59227 telex: JAERIJ24596 http:// neutron-www.kekjpl

IFE Institut for Energiteknikk, P.O. Box40, N-2007 Kjeller, Norway Jon Samseth; tel: +47 6 806080 fax: +47 6 810920 telex: 74 573 energ n E-mail: Internet JON@BARNEY.IFE.NO ILL Institute Laue Langevin, BP 156, F-38042, Grenoble Cedex 9,France Peter Timmins; tel: +33 76207263 E-mail: TIMMINS@FR ILL Peter Linder; tel: +33 76207068; E-mail: LINDER@FR ILL Roland P.May;tel:+3376207047; E-mail: MAY@FRILL fax: +33 76 48 39 06 telex: ILL 320-621 http://www.ill.fr IPNS Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439-4814, USA P.Thiyagarajan,Bldg.200,RM. D125; tel :+1 708 9723593 E-mail: THIYAGA@ANLPNS Ernest Epperson, Bldg. 212;

tel: +1 708 972 5701 fax: +1 708 972 4163 or + 1 708 972 4470 (Chemistry Div.) http://pnsjph.pns.anl.gov/ipns.html

JINR Joint Institute for Nuclear Research, Laboratory for Neutron Physics, Head P.O.Box 79 Moscow, 141 980 Dubna, USSR A.M. Balagurov; E-mail: BALA@LNP04.JINR.DUBNA.SU Yurii M. Ostaneivich; E-mail: SANS@LNP07.JINR.DUBNA.SU fax: +7 095 200 22 83 telex: 911 621 DUBNA SU http://www.jinr.dubna.su KFA Forschungszentrum Jülich, Institut für Festkörperforschung, Postfach 1913, W-517 Jülich, Germany Dietmar Schwahn; tel: +49 2461 61 6661; E-mail: SCHWAHN@DJUKFA54.BITNET Gerd Maier; tel: +49 2461 61 3567; E-mail: MEIER@DJUKFA54.BITNET fax: +49 2461 61 2610 telex: 833556-0 kf d

Vol. 4 n. 1 Giugno 1999

•

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

55

FACILITIES

LLB Laboratoire Léon Brillouin, Centre d’Etudes Nucleaires de Saclay, 91191 Gif-sur-Yvette Cédex France J.P Cotton (LLB); tel: +33 1 69086460 fax: +33 1 69088261 telex: energ 690641 F LBS+ E-mail: COTTON@BALI.CEA.FR http://bali.saclay.cea.fr/bali.html NIST National Institute of Standards and TechnologyGaithersburg, Maryland 20899 USA C.J. Glinka; tel: + 301 975 6242 fax: +1 301 921 9847 E-mail: Bitnet: GLINKA@NBSENTH Internet: GLIMKA@ENH.NIST.GOV http://rrdjazz.nist.gov ORNL Oak Ridge National Laboratory Neutron Scattering Facilities, P.O. Box 2008, Oak Ridge TN 37831-6393 USA G.D. Wignall; tel: +1 615 574 5237 fax: +1 615 576 2912 E-mail: GDW@ORNLSTC http://www.ornl.gov/hfir/hfirhome.html PSI Paul Scherrer Institut Wurenlingen und Villingen CH-5232 Villingen PSI tel: +41 56 992111 fax: +41 56 982327 RISØ EC-Large Facility Programme, Physics Department, Risø National Lab.P.O. Box 49, DK-4000 Roskilde, Denmark K. Mortenses; tel: +45 4237 1212 fax: +45 42370115 E-mail: CLAUSEN@RISOE.DK or SANS@RISOE.DK NFL-Studsvik in Sweden E-mail: mcgreevy@studsvik.uu.se

NOTIZIARIO NEUTRONI E LUCE DI SINCROTRONE

56

•

Vol. 4 n. 1 Giugno 1999