Festkörper- und Clusterphysik. Jahresbericht 1999

SOLID STATE and CLUSTER PHYSICS

Annual Report 1999

Universität Konstanz

Fachbereich Physik

350 500 650

GaN:Pt, 4 K

Pt₁ 15 meV

17 d FWHM ~ 6 meV

YL

PLintensity(arb.units)

wavelength (nm) 3.5 3.0 2.5 2.0

16 h4 d D^oX-LO

D^oX

energy (eV)

840 870 x 40

1.47 1.43 0 20 40 60 80 100 120 140 160

3 4 5 6 7 8 9 10

T(K) l(nm) 20 146(3) 50 169(4) 70 223(4) 80 348(6)

FieldB(z)(mT)

Depth z (nm)

Universität Konstanz Fachbereich Physik

Annual Report

1999

Solid State and Cluster Physics

Universität Konstanz, Fachbereich Physik D-78457 Konstanz, Universitätsstraße 10 Tel. (07531) 88-2415; Telefax (07531) 88-3888

E-mail: fakultaet.physik@uni-konstanz.de

http://www.uni-konstanz.de/FuF/Physik/

Cover pictures (from top):

· LPE layer for solar cell application grown on a polycrystalline substrate (article 1.12)

· Monte Carlo-configuration of a pure CO-monolayer on graphite (article 1.31)

· Field profile inside a YBCO film in the Meissner state (article 1.1)

· Structures produced by colloid monolayer lithography (article 1.15)

· Photoluminescence spectra of ¹⁹¹Pt doped GaN (article 1.6)

· Al₇₀ cluster on graphite (article 2.3)

Editing: M. Deicher, Ch. Niedermayer, T. Blasius, A. Maier Printing: Fabian, Konstanz

© 2000 Universität Konstanz, Fachbereich Physik

This report can be downloaded in Adobe

^®

Portable Document Format (PDF) from

http://www.ub.uni-konstanz.de/serials/phyfest.htm

By this annual report we present a comprehensive survey of our department’s activities in solid state physics and cluster physics in 1999. This report was possible by the large amount of interesting contribu- tions from our colleagues. We thank all co-workers from the secretaries’ offices, central services, laborato- ries and workshops of the university contributing to the results of the previous year.

It is a great pleasure to report the recognition given to the Konstanz research in form of awards in 1999, namely the ‘Ecology Design Award’ (sunways company) for Dr. Fath (LS Bucher) and the ‘Dornier Special Award’ (Dornier company) for colloid masks and nano-structuring (LS Leiderer).

We gratefully acknowledge the generous support given by several research institutions, institutes and companies, in particular the ‘German Research Society’ (DFG), the ‘European Union’ (scientific projects:

ACE, ASCEMUS, HIT, Fast-IQ, SIMU), the ‘State of Baden-Württemberg’, the ‘German Ministry of Education, Science, Research and Technology’ (BMBF), the ‘East European Office of the BMBF’, the ‘Ger- man-Israeli-Foundation’ (GIF, Jerusalem), the ‘Paul-Scherrer-Institute’ (Villingen/Switzerland), the ISOLDE/CERN (Geneva/Switzerland), and the companies ASE, Bayer, centrotherm, Ersol, EKRA, Merck, Solon, sunways, Winter, Zeiss and BP Solarex (USA/GB), DISCO HiTec (J), Elkem (N), Evergreen Solar(USA), Eurosolare (I), GT Solar (USA), Helios (I), Photowatt (F), Shell Solar (NL).

Konstanz, July 2000

Peter Nielaba

Contents

I. Preface iii

II. Research Reports 1

1. Solid State Physics 1

1.1 Direct measurements of the penetration of a magnetcic field into a

superconductor in the Meissner state ...1

1.2 Temperature dependence of the magnetic penetration depth in an YBa

₂

Cu

₃

O

_7-d

film ...2

1.3 Low temperature vortex structures of the mixed state in underdoped Bi

₂

Sr

₂

CaCu

₂

O

_8+d

(Bi-2212) single crystals...3

1.4 Ultrafast magneto-optical studies...5

1.5 Electrical characterization of

¹¹¹

Ag doped GaN ...7

1.6 Near-infrared-photoluminescence in GaN doped with radioactive platinum...9

1.7 Observation of free H after annealing of GaN implanted with

¹¹⁷

Cd(

¹¹⁷

In)...11

1.8 Behavior of doping atoms in CdCr

₂

Se

₄

...13

1.9 Crystalline silicon solar cells – new materials ...15

1.10 Compound semiconductors for photovoltaics / materials for thermoelectric applications ...17

1.11 Industrial solar cell development ...20

1.12 Thin film silicon solar cells on low-cost metallurgical silicon substrates...23

1.13 Novel solar cells ...24

1.14 Solar cell characterization ...29

1.15 Colloidal mask lithography ...31

1.16 Surface structuring with ultrashort laser pulses ...32

1.17 Metallic nanostructures on metaldichalcogenides...33

1.18 Self-organized growth of epitaxial CoPt

₃

nanostructures on WSe

₂

...34

1.19 Co islands on WSe

₂

...35

1.20 A low energy muon study of the dipolar fields produced by an assembly of iron nanoclusters in silver ...36

1.21 Dimensional cross-over in AuFe spin-glass studied by low energy muons ...37

1.22 Magnetic anisotropy and chemical long-range order in epitaxial ferrimagnetic CrPt

₃

films ...38

1.23 Structure of Pt/Mn superlattices grown by molecular beam epitaxy ...40

1.24 Modified growth of Co on Cu(111) using In as an interlayer ...42

1.25 Laser cleaning of silicon surfaces ...44

1.26 Range of low energy muons in Cu films ...46

1.27 Response of the electric field gradient on an external electric field...47

1.28 Diffusion of muons in metallic multilayers...49

1.29 Energy dependence of muonium formation in solid Ar, N

₂

, Xe, and SiO

₂

...50

1.30 Progress on the low energy m

⁺

apparatus ...51

1.31 Phase transitions in two dimensional (adsorbed) layers at low temperatures ...52

1.32 Phase transitions in alloys with elastic interactions ...54

1.33 Enrichment of surfaces in contact with stable binary mixtures ...56

1.34 Elastic constants from microscopic strain fluctuations and melting of hard disks in two dimensions ...58

1.35 Low temperature properties of molecular solids...60

1.36 Structures, phases, and phase transitions in solids in reduced geometry ...62

1.37 Correlations and dynamic ordering processes near solid surfaces ...64

1.38 Transport processes in glasses...68

2. Cluster Physics 70 2.1 The structure of medium sized silicon cluster anions ...70

2.2 Time resolved dynamics of electronic excitations in C

₃^-

...72

2.3 Deposition of mass selected aluminum clusters...74

2.4 A new experimental setup for the in situ investigation of the electronic, vibrational and chemical properties of monodisperse supported clusters ...76

2.5 Sputtering with cluster ions...78

III. Publications and Talks 80 1. Publications ...80

2. Conference contributions ...85

3. Lectures ...91

4. Theses...93

IV. Staff and Guests 96

II. Research Reports

1. Solid State Physics

1.1 Direct measurements of the penetration of a magnetcic field into a superconductor in the Meissner state

E.M. Forgan, T.J. Jackson and T.M. Riseman (University of Birmingham, Birmingham B15 2TT, United Kingdom)

in collaboration with

H.Glückler, E. Morenzoni and T. Prokscha (Paul-Scherrer-Institut,CH-5232 Villigen, Schweiz) Ch. Niedermayer, M. Pleines and G. Schatz

M. Birke, J. Litterst and H. Luetkens (IMNF, TU Braunschweig, D-38106 Braunschweig) Low energy muons extend the utility of mSR to thin

films and multilayers, and to studies of near surface properties distinct from bulk behavior ¹⁾. In this report we describe the use of low energy muons to measure di- rectly the in plane penetration depth l_ab of an YBa₂Cu₃O_7-@ (YBCO) superconducting film in the Meissner state. The depth of muon implantation was controlled between 20 and 150 nm by tuning the in- coming muon energy from 3 to 30 keV.

The YBCO film was a 700 nm thick film grown epi- taxially on a 50 mm diameter LaAlO₂ substrate by H.

Kinder's group at TU München. The magnetic field B₀ of approximately 10 mT was applied using a permanent magnet assembly as shown in ²⁾.

Fig. 1: Field profile inside the YBCO film in the Meiss- ner state at different temperatures. Inverse triangles: 20 K, triangles: 50 K, diamonds: 70 K, and squares: 80 K.

The field was applied perpendicular to the YBCO c- axis after zero field cooling the sample through the su- perconducting transition, so that the field was screened by supercurrents flowing within the ab-planes. Twinning within the ab planes precluded experiments with the field applied along a unique in-plane axis. The depth de- pendent field profile B(z) within such a semi-infinite su- perconducting slab follows from the well-known Lon- don exponential decay law, and is described by

0

( ) cosh

cosh

ab ab

d z B z B

d

where d is the half thickness of the film ³⁾. The results of our measurements are shown in Fig. 1, in which the solid lines represent fits of equation 1 to the experimen- tal data, shown by the points. The muon implantation profile was determined from simulations ⁴⁾ and the field at the most probable depth determined from Maximum Entropy analysis ⁵⁾ of the decay histograms.

Equation 1 describes the data well, yielding a value of lab at 20 K of 146(3) nm, with which other less direct determinations agree ⁶⁾. This method offers unique pos- sibilities for measurements of the field penetration in other geometries and materials where the London law is not expected to apply.

(1) E. Morenzoni et al., Physica B in press

(2) H. Glückler et al., Progress on the low energy m⁺ appa- ratus, PSI Scientific Report, Vol. 1 (1999)

(3) D. Schoenberg, Superconductivity, (Cambridge Univer- sity Press, Cambridge,1952)

(4) H. Glückler et al., Physica B in press

(5) T.M. Riseman and E.M. Forgan, Physica B in press (6) Ch. Niedermayer et al., Physical Review Letters 83

(1999) 3932 0 20 40 60 80 100 120 140 160

3 4 5 6 7 8 9 10

T(K) _l(nm) 20 146(3) 50 169(4) 70 223(4) 80 348(6)

FieldB(z)(mT)

Depth z (nm)

1.2 Temperature dependence of the magnetic penetration depth in an YBa

₂

Cu

₃

O

_7-@@@@

film

M. Pleines, Ch. Niedermayer and G. Schatz in collaboration with

H.Glückler, E. Morenzoni and T. Prokscha (Paul-Scherrer-Institut,CH-5232 Villigen, Schweiz) M. Birke, J. Litterst and H. Luetkens (IMNF, TU Braunschweig, D-38106 Braunschweig) E.M. Forgan, T.J. Jackson and T.M. Riseman

(University of Birmingham, Birmingham B15 2TT, United Kingdom) In many of the potential applications of high tem-

perature superconductors thin films will play an impor- tant role. A proper characterization of the supercon- ducting parameters of these films is therefore of great interest. One of the fundamental parameters is the mag- netic penetration depth lab. We have used the low en- ergy muon source ¹⁾ to study the temperature depend- ence of lab of an YBa₂Cu₃O_7-@ (YBCO) film.

The measurements were performed on a 700 nm thick, c-axis orientated YBCO film (T_c = 87.5 K), grown epitaxially by thermal coevaporation on a LaAlO₃ substrate. The film was characterized by a spa- tially resolved J_c - measurement using an inductive technique ²⁾. The J_c(77 K) map of the investigated film, displayed in Fig.1, shows that the critical current density is very homogeneous over the whole sample area of the 2 inch wafer.

Fig. 1: Map of the critical current density J_c. In the area of the film (2 inches diameter) the critical current den- sity varies between 2.5×10⁶ A/cm² and 3×10⁶ A/cm².

An external field of 10.4 mT was applied parallel to the c-axis of the film. The sample was then field-cooled into the superconducting state. The energy of the in- coming muon beam was fixed at 29 keV.

YBCO is an extreme type II superconductor. When a magnetic field H_ext > H_c1 is applied, a vortex lattice of the magnetic flux is formed. The field variation B(r) within the vortex lattice produces an asymmetric field distribution p(B) with a cusp which corresponds to the most probable field B_sad.

The measured field distribution was obtained from the time evolution of the muon spin polarization P(t) via

Maximum Entropy Technique. As it shown earlier ³⁾ there is a good agreement between measured and theo- retical field distribution.

The penetration depth can be derived from the shift between the mean field B_ext and the cusp field B_sad:

0

( )

2

ext sad

ab

B B c H

Figure 2 shows the temperature dependence of the magnetic penetration depth l_ab. The solid curve repre- sents a fit to a power law of the form:

^{1 2}

( ) (0) 1

ⁿ

ab T ab T Tc

with l_ab = 137(10) nm and n = 1.7(3)

Fig. 2: Temperature dependence of the magnetic pene- tration depth l_ab of a 700 nm thick YBCO film. For comparison the dashed lines with n = 2 and n = 4 are shown, where n = 4 is obtained from the two fluid model and observed for classical superconductors.

Our value of the penetration depth lab(0) is in good agreement with results from conventional mSR meas- urements ⁴⁾ on bulk samples. The inset shows lab(T)/lab(0) below 55 K in more detail together with data points from conventional mSR measurements on YBa₂Cu₃O_6.95 single crystals ⁴⁾. There is a very good agreement between both sets of data. The observed lin- ear temperature dependence was taken as evidence for an unconventional symmetry (d-wave) of the supercon- ducting order parameter.

(1) E. Morenzoni et al., Appl. Magn. Reson. 13 (1997) 219 (2) H. Kinder et al., Physica C 107 (1997) 282

(3) Ch. Niedermayer et al., Phys. Rev. Lett. 83 (1999) 3932 (4) J.E. Sonier et al., to be published

-24 -24 -12 0 +12 +24

-12 0 12

Y[mm]

X[mm]

0 0.6 1.2 1.8 2.4 3.0

24

J [10 A/cm ]

c62

0 20 40 60 80 100

100 200 300 400 500 600 700

n = 4

n = 2

Temperature [K]

0 10 20 30 40 50 0.90

1.00 1.10 1.20 1.30

Temperature [K]

λλ(T)/(0)[arb.units]

penetrationdepth[nm]λ

1.3 Low temperature vortex structures of the mixed state in underdoped Bi

₂

Sr

₂

CaCu

₂

O

_8+@@@@

(Bi-2212) single crystals

T. Blasius and Ch. Niedermayer in collaboration with

J.L. Tallon and D.M. Pooke (New Zealand Institute for Industrial Research, Lower Hutt, New Zealand) A. Golnik, C.T. Lin, and C. Bernhard (Max Planck Institut für Festkörperforschung, Stuttgart, Germany) One of the striking features of the high - T_c supercon-

ductors (HTSC) is the influence of their electronic and magnetic anisotropy on the nature of the mixed state.

The vortex lines in these systems reveal, contrary to conventional superconductors, a discreteness due to the layered structure of the HTSC. With the magnetic field applied along the c-axis direction (perpendicular to the CuO₂ planes) the vortex lines can be viewed as stacks of pancake vortices. These quasi-two-dimensional vortex currents reside within the superconducting planes and are connected across the insulating spacer layers via electromagnetic interaction and Josephson coupling.

Depending on the anisotropy gs = lc / lab, where lc and lab are the out-of-plane and in-plane penetration depths, the CuO₂ layer spacing s and lab, a wide variety of coupling scenarios from dominating electromagnetic coupling, (lab < gs s), to dominating Josephson coupling (lab > gs s) have been reported ^1,2).

Here we present transverse-field muon spin rotation (TF-mSR) data on underdoped Bi-2212 single crystals, which were grown by a floating-zone technique as de- scribed elsewhere ³⁾. In TF-mSR experiments, spin-po- larized muons are implanted into the bulk of the crystal.

An external magnetic field H_ext is applied perpendicular to the initial polarization of the muon spin and parallel to the c-axis of the crystals. The muons come to rest at interstitial locations, r, which are randomly distributed on the length scale of the magnetic penetration depth lab. Their spins start to precess in the local magnetic field with the Larmor frequency w_m = g_m B(r), where g_m is the gyromagnetic ratio of the muon with g_m / 2p = 135.54 MHz / T. The time evolution of the muon spin polarization P(t) is measured by monitoring the decay positrons, which are preferentially emitted along the muon spin direction at the instant of decay.

The probability distribution of the local magnetic field n(B_m) is extracted from P(t) via Fast Fourier Transform or Maximum entropy techniques and contains detailed information on the vortex structure.

For a static flux line lattice, n(B_m) is asymmetric ⁴⁾ with a pronounced tail towards high fields due to muons that stop near the vortex cores, a cusp that corresponds to the field at the saddle point between adjacent vortices and a cutoff on the low field side corresponding to the field minimum at the point that is most remote from the vortex cores.

A typical example for such a field distribution is shown in Figure 1, which has been measured at 5 K after field cooling in an external field of 8 mT. The additional feature at the external field value is due to a small frac-

tion of muons (~ 5%) not stopping in the sample which is included in the data analysis by a Gaussian distribu- tion. The solid line is a fit to the data as described in ³⁾. Imperfections in the vortex lattice and instrumental resolution have been accounted for by a Gaussian con- volution with a smearing of ~8% in n(B_m). The best result has been realized with lab = 2000 (50) Å. The asymmetry of n(B_m) can be characterized by the dimen- sionless parameter a

^< DB³ >^1/3 / < DB² >^1/2, where

< DBⁿ > are the n^th central moments of n(B_m) [4]. A value of a ~ 1.2 is typical for a static well ordered flux line lattice, whereas 1 > a > 0 either indicates a disor- dered static vortex structure or else vortex dynamics in excess of the typical mSR time scale of t

^{10-6 s 4,5).}

Fig. 1: Magnetic field distribution (open squares) at T = 5 K determined in a field-cooled TF-mSR experiment on underdoped Bi-2212 single crystals (T_c = 77 K) in an applied field of m_oH_ext = 8 mT. The solid line represents a fit to the data as described in [1].

Figure 2 shows observed field distributions at 5 K after field cooling in 8 mT and 10.3 mT. Figure 3 sum- marizes the results for the parameters a and B_sh of the field distribution n(B_m) for the measured field depend- ence at 5K after field cooling. The sudden reduction of a and B_sh at B*

8.5 mT indicates a strong change in the vortex structure. The pinning properties of the different vortex states have been tested by reducing the applied field after the field-cooling process. For a rigidly pinned vortex state n(B_m) will not follow the change of the applied field, whereas for a depinned vortex array n(B_m) will follow the changes of H_ext. The finding that on the time scale of hours n(B_m) at T = 5 K is unaffected by the reduction of the external field by several mT indicates that the observed changes in n(B_m) are of static rather than dynamic nature.

-4 -2 0 2 4 6

moH_ext = 8.0 mT, T = 5 K Bi2212

underdoped, T_c = 77 K

n ( Bmmmm ) [ arb. units ]

B_mmmm - mmmmoH_ext [ mT ]

Fig. 2: Magnetic field distributions from field-cooled TF-mSR experiments at T = 5 K on underdoped Bi-2212 single crystals (T_c = 77 K).

As shown by E.H. Brandt ⁴⁾ for the case of a static vortex state, the above described behavior can be ac- counted for only by disorder in the vortex lines along the c-axis direction on a length scale of lab. For anisotropic systems with lab < gs s, as for example underdoped Bi- 2212, such a behavior has been related to a 'dimensional crossover' at B_cr

f_o / (lab )² [5]. To attribute the sudden change in the vortex structure at B*

^{8.5 mT}

to B_cr

f_o / (lab )², lab ~ 5000 Å is required in strong disagreement with the measured value of lab ~ 2000 Å.

However, as proposed by A.E. Koshelev, L.I. Glazman and A.I. Larkin ²⁾, A.E. Koshelev and V. Vinokur ²⁾ and E.H. Brandt ⁴⁾, the observed changes in the internal field distribution are due to random displacements of the pancake vortices within the individual vortex lines induced by the pinning potential. The corresponding transition with increasing field is then a disorder- induced destruction of the vortex lattice, which is more pronounced in systems with large anisotropy due to the softening of the vortex lines.

Interestingly, we observe an increase of a and B_sh for fields moH_ext

350 mT = B_cp (Figure 3). Such a partial restoration of the coupling between the pancakes across the insulating spacer layers has been observed up to now only for increasing temperatures [6]. For the static case, this behavior may be related to the change in the pinning properties from single vortex pinning to bundle pinning [1] as the vortex density and thus the interaction be- tween the vortices increases. Additional experiments probing in detail the pinning properties have to be done to check for this possibility.

In summary, we find by TF-mSR experiments on un- derdoped Bi-2212 single crystals (T_c = 77 K) that the strong changes in the low-temperature vortex state at B*= 8.5 mT are consistent with a disorder-induced destruction of the vortex lattice. In addition, we obtained for the first time evidence for a low temperature resto- ration of the coupling in the c-axis direction for moH_ext

^{350 mT.}

Fig. 3: Field dependence of a and B_sh of n(B_m ) at T = 5 K determined from field-cooled TF-mSR experi- ments on the underdoped Bi-2212 sample (T_c = 77 K).

(1) G. Blatter, V. Feigel'man, V.B. Geshkenbein, A.I. Larkin and V. Vinokur, Rev. Mod. Phys. 66 (1994) 1125 (2) A.E. Koshelev, L.I. Glazman and A.I. Larkin, Phys. Rev.

B 53 (1996) 2786, A.E. Koshelev and V. Vinokur, Phys.

Rev. B 57 (1998) 8026

(3) T. Blasius, PhD-thesis, Universität Konstanz 2000, http://www.ub.uni-konstanz.de/kops/volltexte/2000/405/

and references therein

(4) E.H. Brandt, J. Low Temp. Phys. 73 (1988) 355, ibid Phys. Rev. Lett. 63 (1989) 1106 and 66 (1991) 3213 (5) S.L. Lee, P. Zimmermann, H. Keller, R. Schauwecker,

M. Warden, M. Savic, D. Zech, R. Cubitt, E.M. Forgan, P.H. Kes, T.W. Li, A.A. Menovsky, and Z. Tarnawski, Phys. Rev. Lett. 71 (1993) 3862, C.M. Aegerter, S.L. Lee, H. Keller, E.M. Forgan, and S.H. Lloyd, Phys.

Rev. B 54 (1996) R15661

(6) T. Blasius, Ch. Niedermayer, J.L. Tallon, D.M. Pooke, A. Golnik and C. Bernhard, Phys. Rev. Lett. 82 (1999) 4926

-4 -2 0 2 4 6

a ~ 0.2

a ~ 1

m_oH_ext = 8.0 mT m_oH_ext = 10.3 mT

Bi-2212 underdoped, T_c = 77 K

n ( B mmmm ) [ arb. units ]

B_mmmm - _mmmmoH_ext

0.2 0.4 0.6 0.8 1.0

==== ( m oH ext = 5.8 mT )==== ( moH ext )

10 100 1000

0.2 0.4 0.6 0.8 1.0

B

_cp

B*

B sh ( m oH ext = 5.8 mT )

mmmm_oH_ext [ mT ] B sh ( m oH ext )

1.4 Ultrafast magneto-optical studies

B.-U. Runge, U. Bolz, B. Böck, C. Häfner and P. Leiderer

1. Investigations of flux avalanches in high-T_c superconductors

An ultra fast magneto-optic pump-probe technique has been used to trigger and image a flux instability in high-temperature superconducting thin films. Snapshots of the dendritic flux avalanche spreading into the film could be obtained with a time resolution in the picosec- ond range.

The dynamics of magnetic flux avalanches in high- temperature superconductors (HTSC) is of great interest not only from a fundamental point of view, but also with respect to the application of these materials e.g. as cur- rent limiters in the field of electrical power distribution.

Previous studies ¹⁾ have shown that much of the flux dy- namics in YBa₂Cu₃O_7-@ takes place well below the nanosecond range. Therefore it is necessary to improve the time resolution into the picosecond regime to get more detailed information about the processes involved.

All samples studied were 330 nm thick epitaxially grown c-axis oriented YBa₂Cu₃O_7-@ films deposited by thermal evaporation onto SrTiO₃ ²⁾. The experiments were carried out in a small continuous flow cryostat, which had two optical windows with a diameter of 25 mm. For detecting the magnetic field penetrating the superconductor we used a doped ferrimagnetic iron gar- net layer grown onto gadolinium-gallium-garnet sub- strate by liquid phase epitaxy with an additional alumi- num layer ³⁾. This magneto-optical layer was placed just above the YBCO film. By using a home-built polariza- tion microscope the local Faraday rotation of the linearly polarized light caused by the local magnetic field H_z in the magneto-optical layer was made visible with nearly crossed polarizer and analyzer as an intensity contrast and imaged with a 12 bit slow-scan CCD camera.

The YBCO film was zero field cooled down to 10 K.

After reaching a stable temperature an external magnetic field B_ext perpendicular to the sample surface was ap- plied. Magnetic flux penetrated into the superconducting film first from the edges until a local equilibrium of the flux distribution due to the pinning force and the mag- netic force was reached. This induces a current distribu- tion in the superconducting film. That current distribu- tion can be disturbed to initiate a magnetic instability.

For this purpose we used a single pulse of a Ti:sapphire laser (l = 800 nm, half width t = 150 fs) which was fo- cused onto the film from the substrate side to a spot di- ameter of about 50 µm. The sample temperature in the focus could not be measured directly, but from the pulse energy we estimate that the temperature rises well above the critical temperature.

If the perturbation is sufficiently strong, this triggers a magnetic instability, in which a magnetic flux ava- lanche penetrates into the film. In order to record snap- shots of the flux moving into the sample a beam splitter

is used to separate part of the trigger pulse and send it through a delay line for illumination of the sample at a well defined time after the trigger event. This time can be varied from below zero (illumination before trigger) to several 100 ns with an accuracy in the picosecond range.

Fig. 1: Snapshots of the flux penetration at delay times of 3.2 ns, 13.5 ns and 41.2 ns after the trigger event.

Sample temperature T = 10 K, external magnetic flux density B_ext = 19 mT. The size of the images shown is 1.8´1.8 mm².

Fig. 1 shows typical snapshots at delay times of 3.2 ns, 13.5 ns and 41.2 ns for a sample temperature of T = 10 K and an external magnetic flux density of B_ext = 19 mT. In order to have reproducible starting conditions, for each image the sample was heated above the critical temperature and zero field cooled again to 10 K. The form of the dendritic structure is found to be similar from image to image and for all delay times used, although the individual dendrites are not identical for subsequent laser pulses. The width of the dendrites and their mutual distance remains about constant during the process. A “typical spreading velocity” of the den- drites was calculated measuring the distance between the starting point of the avalanche (i.e. the laser focus) and the tip of a typical dendrite in the center of the ava- lanche. The average of this velocity over the first 41.2 ns is found to be 3.2(2)×10⁴ m/s which is far above the velocity of sound in YBCO. In Fig. 2 we show the time dependence of the length of a typical dendrite.

In conclusion we have improved the time resolution for the magneto-optical observation of magnetic flux avalanches in high-T_c superconductor thin films from nanoseconds to picoseconds by using laser pulses with a half width of less than one picosecond. This allows a much more precise determination of the spreading ve- locity of the dendrites. We find 3.2(2)×10⁴ m/s. This value is slightly lower than the value of 5(2)×10⁴ m/s reported in an earlier study ¹⁾ which is probably due to differences in the sample preparation. The increased time resolution will allow us to study the very beginning of the flux avalanche and other phenomena caused by perturbation of superconducting films.

The authors would like to thank H. Kinder and K.

Numssen for providing the YBCO films as well as H.

0 10 20 30 40 50 660

680 700 720 740 760 780 800 820

time [s]

detectorsignal[mV]

Dötsch and E. Il'yashenko for providing iron garnet layers.

Fig. 2: Length of a typical dendrite close to the center of the dendritic flux structure plotted as a function of time after the trigger event. The dotted line is a guide to the eye.

2. Pulsed magneto-optic Kerr measurements of thin magnetic cobalt films

The magnetic behavior of cobalt and its alloys is of special interest, as magnetic ordering can be observed in sufficiently supercooled melts ⁴⁾. Up to now these stud- ies have been carried out using an electromagnetic levi- tation technique. This limits the cooling rate to about 100 K/s. Much higher cooling rates (and therefore lower temperatures of the supercooled melt) can be achieved using laser annealing on the nanosecond time scale. As an important first step towards the observation of the magnetic behavior of thin cobalt films during laser an- nealing and melting we performed magneto-optic Kerr (MOKE) measurements using single pulses of a Nd:YAG laser system. As the Co film is molten by the pulse and the melting time depends strongly on the pulse energy, it would be desirable to have absolutely repro- ducible pulse energies. But typically the energy varies by 5-10% from pulse to pulse not allowing averaging of the signal over subsequent pulses. It is therefore neces- sary to record the magneto-optic signal as well as the pulse power for each individual pulse with a time reso- lution of about 1 ns. Fig. 3 shows first results of such a measurement. In this case the sample was not really molten to create a change in the magnetic properties but

rather a sinusoidally slowly varying external magnetic field was applied in the plane of a Co film of 30 nm thickness. The resulting magneto-optic signal was de- tected and normalized to the pulse energy for each indi- vidual pulse (half width »10 ns) at a repetition rate of 10 Hz. The magnetic response of the film looks rather square than sinusoidal, indicating that the magnetic field was strong enough to drive the film into saturation. Our result shows the feasibility of single pulse Kerr meas- urements with our setup. Currently measurements are under way in which we melt the sample by each pulse and simultaneously record the thermal radiation in order to relate the magnetic properties to the temperature of the sample.

Fig 3: Pulsed MOKE measurement of 30 nm thick Co film. The strength of the external magnetic field was varied sinusoidally with a frequency of about 0.07 Hz and an amplitude of 10 mT. The film is driven into satu- ration, as can be seen from the rather square form of the detector signal.

(1) P. Leiderer, J. Boneberg, P. Brüll, V. Bujok and S. Herminghaus, “, Phys. Rev. Lett. 71 (1993) 2646 (2) P. Berberich, W. Assmann, W. Prusseit, B. Utz, and

H. Kinder, J. of Alloys & Compounds 195 (1993) 271 (3) M. Wallenhorst, Herstellung und Charakterisierung

magnetooptischer Eisengranatfilme für nichtreziproke Wellenleiter und magnetooptische Sensoren, PhD thesis, Universität Osnabrück (1998)

(4) C. Bührer, Der Flüssige Ferromagnet - Kritisches Ver- halten am magnetischen Phasenübergang der flüssigen Phase von Co₈₀Pd₂₀, PhD thesis, Universität Bonn (1998)

0 10 20 30 40

0 500 1000

dend rit e le ngt h [µm ]

delay time [ns]

1.5 Electrical characterization of

¹¹¹

Ag doped GaN

A. Stötzler, M. Dietrich and M. Deicher in collaboration with

ISOLDE Collaboration (CERN, Geneva) The great interest on GaN is mainly based on the po- tential applications in optoelectronics and high tem- perature electronics. Most of the work during the last years have been focused on the optical properties and due to the difficulties in interpreting the gathered data there are only limited reports on the electrical properties of GaN ^1,2). In spite of the fast development of new growing techniques, GaN is still a highly defective ma- terial. Especially after ion implantation and annealing it is extremely hard to distinguish between the intrinsic de- fects present in the GaN layer, defects caused by an- nealing or the levels introduced intentionally or uninten- tionally by implantation. Another problem is, that due to the small thickness of the epitaxial layer the measured values are strongly distorted by degenerate layers at the GaN/sapphire interface. The usual way of determining activation energies by temperature dependent Hall measurements often fails since low temperatures are re- quired and the lower the temperatures the greater the in- fluence of this distortion effect ³⁾.

Defects are getting distinguishable if one uses radio- active dopands. The concentration of a defect containing a radioactive atom is changing with the element specific half-life of the decay while the concentrations of all other defects remain unchanged. Therefore, the fraction of activated ions after annealing, the carrier type or the influence on resistivity and mobility can be determined directly from exponential fits to the data.

Photoluminescence (PL) experiments on ¹¹¹Ag-doped GaN have shown that Ag and Cd are introducing strong PL signatures in the PL spectra of GaN ⁴⁾. In order to in- vestigate the electronic properties of Ag and Cd in GaN Hall measurements on ¹¹¹Ag-doped GaN were performed.

For the Hall measurements a 1.5 µm epitaxial GaN layer grown on an AlN/c-sapphire substrate with a size of 5 ´ 5 mm² was used. The nominally undoped layer was n-type with a free carrier concentration of 5 ´ 10¹⁶ cm^-3 prior to the annealing. The sample was doped by ion implantation with the radioactive isotope ¹¹¹Ag at the on-line mass separator facility ISOLDE at CERN.

This isotope decays within a half-life of 7.45 d into sta- ble ¹¹¹Cd. The implantation energy was 260 keV and a maximum dose of 1 ´ 10¹³ cm^-2 was used. During im- plantation the sample was covered with an aperture of 5 mm diameter in order to achieve a symmetric and ho- mogeneous implantation area. The Gaussian shaped im- plantation profile is centered at 56 nm depth with a width of 13 nm. The peak concentration is about 4 ´ 10¹⁸ ions/cm². To reduce the implantation induced damage the sample was annealed 600 s in an evacuated quartz ampoule together with a piece of elementary Si at 1573 K ⁵⁾. During annealing a SiN_x layer develops at the

GaN surface which prevents the epitaxial layer from de- composition usually observed at such high annealing temperatures. After annealing the SiN_x layer was re- moved by etching in a solution of 2 HF: 5 HNO₃ : 2 C₂H₄O₂. The electrical characterization was performed by van der Pauw Hall measurements using alloyed (473 K, 60 s) Indium contacts at the corners of the sample.

Fig. 1: Resistivity r (upper graph) and carrier- concen- tration n of ¹¹¹Ag-doped GaN (10¹³ cm^-2, 260 keV) plotted as a function of temperature. The solid square spectra were recorded 3 h after annealing and the open square spectra were recorded 25 d after annealing.

Fig. 1 shows the net carrier-concentration n = (N_D- N_A) and the resistivity r of the ¹¹¹Ag-doped sample 3 h (solid squares) and 25 d (open squares) after implanta- tion and annealing as a function of temperature. For clarity only two of the 15 measurements are shown. One can clearly observe, that during the 25 d the carrier-con- centration n decreases and that the resistance r in- creases. GaN is a highly n- type material and the higher the temperatures the greater the influence of the native donors on the resistivity and carrier concentration. At lower temperatures the effect of the native donors and activated intrinsic carriers can be minimized and there- fore the changes in the carrier-concentration and espe- cially the resistivity are more pronounced.

3 4 5 6 7 8 9 10 11

0.8 1.0 1.2 1.4 1.6 1.8 2.0

3 h 25 d

n (1017 /cm3 )

1000/T_M (K^-1) 0.4

0.6 0.8 1.0 1.2

3 h after annealing 25 d

H (9cm)

300 200 150 100

Temperature (K)

In Fig. 2 the resistivity and carrier-concentration re- corded at 130 K are shown as a function of time after annealing. The solid lines correspond to exponential least square fits using the following equations:

1 / 2

1 / 2

(ln 2) / 0

(ln 2) / 0

( ) ( )

t t n

t t

n t n A e

t A e

The fits on the carrier concentrations yield an average half-life of t_1/2 = (5.78 ± 0.72) d in rough agreement with the nuclear half-life of the isotope while the fits on the resistivities yield the correct average half-life of t_1/2 = (7.39± 0.34) d. In contrast to the resistivity meas- urements the carrier concentration strongly depends on the positioning of the sample in the magnetic field and therefore the deviation of the half-life determined from the carrier concentrations is reasonable.

Fig. 2: Carrier-concentration (circles) and resistivity (squares) of ¹¹¹Ag-doped GaN (10¹³ cm^-2, 260 keV) re- corded at 130 K as a function of time after annealing.

Ag is a group Ib element and from valence arguments should act as a double acceptor. Cd, being a group IIb element should form a single acceptor. Therefore during the transmutation of Ag (two holes) to Cd (one hole) the number of additional holes should decrease and the con- centration of the majority carriers (electrons, n-type material) should increase in contrast to the observed be- havior. From our results we conclude that Ag does not form a double acceptor level in GaN, or that this accep- tor levels are not ionisized up to 300 K.

The fits on the carrier-concentrations yields an aver- age amplitude of A_n = 1.04 ´ 10¹⁶ cm^-3 carriers. This means, that at annealing temperatures of 1573 K only 18.6% of all implanted ions are activated during the an-

nealing procedure, since the concentration of implanted ions in the sample volume was 5.6 ´ 10¹⁶ cm^-3. Addi- tionally, those results are an explanation of the low frac- tion of Cd-H-pairs in GaN found with PAC measurements ⁶⁾.

Fig. 3: Hall mobility of 111Ag-doped GaN plotted as a function of temperature. The spectra were recorded 3 h and 25 d after implantation and annealing.

In Fig. 3 the hall-mobility is shown as a function of temperature. The spectra were taken 3 h and 25 d after implantation and annealing. An increasing mobility can be observed only at temperatures above 150 K presum- able due to a decrease in phonon scattering. If one as- sumes an increasing concentration of ionized impurities (Ag⁰ ® Cd^-) the mobility should decrease in contrast to the results.

Our results show that Cd acts as an acceptor in con- trast to Ag. However, annealing temperatures above 1573 K are required to activate all of the implanted dopands and achieve p-type GaN.

(1) S.J. Pearton, C.R. Abernathy, C.B. Vartuli, J.C. Zolper, C. Yuan and R.A. Stall, Appl. Phys. Lett. 67 (1995) 1435 (2) J.C. Zolper, H.H. Tan, J.S. Williams, J. Zou,

D.J.H. Cockayne, S.J. Pearton, M. Hagerott Crawford and R.F. Karlicek, Appl. Phys. Lett. 70 (1997) 2729 (3) D.C. Look and R.J. Molnar, Appl. Phys. Lett. 70 (1997)

3377

(4) A. Stötzler, R. Weissenborn, M. Deicher and the ISOLDE Collaboration, Physica B 273-274 (1999) 144 (5) A. Burchard, E.E. Haller, A. Stötzler, R. Weissenborn,

M. Deicher and the ISOLDE Collaboration, Physica B 273-274 (1999) 96

(6) A. Burchard, M. Deicher, D. Forkel-Wirth, E.E. Haller, R. Magerle, A. Prospero and A. Stötzler, in: Defects in Semiconductors 19, eds. G. Davies and M.H. Nazare, Materials Science Forum Vol. 259-263 (Trans Tech Pub- lications 1997) p. 1099

0 5 10 15 20 25 30

0.94 0.96 0.98 1.00 1.02 1.04

n (1017 /cm3 )

Time (d) 0.76

0.78 0.80 0.82 0.84 0.86 0.88

T_M=130 K

Fit: <t_1/2> = (7.39 ± 0.34)d

T_M= 130 K

Fit: <t_1/2> = (5.78 ± 0.72) d

H (9cm)

100 150 200 250 300

60.0 70.0 80.0 90.0 100.0

3 h 25 d Mobility (cm2 /Vs)

Temperature (K)

1.6 Near-infrared-photoluminescence in GaN doped with radioactive platinum

A. Stötzler and M. Deicher in collaboration with

ISOLDE Collaboration (CERN, Geneva) Since the development of blue diodes, GaN has attracted great attention. Most of the investigations on the optical properties of GaN have focused on shallow levels and only little information about deep levels in GaN was gathered. Since the last years the interest on deep levels in GaN is increasing not only due to the po- tential applications of Er-doped ¹⁾ GaN for optical fiber communications or the discovery of near-infrared (NIR) luminescence in V-doped ²⁾ GaN. It also has been shown ^3,4), that the entire visible spectrum can be cov- ered by doping GaN with the rare-earth elements Tm, Pr or Er. Additionally, the transition metals Cr and Fe were found to introduce luminescence in the near infrared re- gion ⁵⁾. Another interesting property of transition metals in semiconductors is that their energy levels are ap- proximately equidistant to the vacuum levels for differ- ent materials ⁶⁾. This can be used to estimate the band discontinuities of a heterojunction, as shown for AlN/GaN ⁷⁾.

Platinum, being a transition metal as well, is known to create luminescence centers in Si ⁸⁾, where Pt is often used to control the carrier lifetimes. In GaN no Pt-re- lated luminescence has been reported up to now. The aim of this experiment was to investigate the optical properties of GaN doped with Pt. In contrast to the re- sults described above, the chemical assignment of the optical transitions was not performed by a comparison with the corresponding energy level schemes consider- ing crystal field splitting but by the use of radioactive isotopes. If an optical transition is due to a recombina- tion center in which the parent or daughter isotope is in- volved, the concentration of that defect will change with the element specific half-life of the radioactive decay.

The changing defect concentration then shows up in the changing PL intensity of the corresponding transition.

The GaN sample used was a 1.5 µm epitaxial layer grown on AlN/c-sapphire by metal organic vapor phase epitaxy (MOVPE) purchased from Cree Research. The sample was doped by ion implantation with the radioac- tive isotope ¹⁹¹Pt at the on-line mass separator ISOLDE at CERN. The implantation energy was 60 keV and a maximum dose of 3 ´10¹² ions/cm² was used. To serve as a reference, a small part of the sample was not im- planted. The implantation induced damage was reduced by annealing the sample at 1300 K for 20 min in sealed quartz ampoules filled with nitrogen gas at a pressure of 1 bar at room temperature. The isotope ¹⁹¹Pt transmutes with a half-life of t_½ = 2.9 d ⁹⁾ into stable ¹⁹¹Ir. This chemical transmutation was monitored by photolumi- nescence spectroscopy (PL).

The PL experiments were carried out at 4 K using a He flow cryostat. The luminescence was excited with the 325 nm line of a HeCd-laser, dispersed with a 0.75m

monochromator and detected either with a cooled GaAs- photomultiplier in the visible and UV range or a liquid- N₂ cooled Ge detector in the infrared region.

The UV-VIS part of the luminescence recorded after ion implantation and annealing is shown in Fig. 1.

Fig. 1:Photoluminescence spectra of ¹⁹¹Pt-doped GaN recorded with a PMT at 4 K. The spectra were recorded within 17 d after implantation and annealing at 1300 K.

All spectra are normalized to the PL intensity at 2.2 eV.

All spectra are normalized to the same PL intensity at 2.2 eV. The spectrum recorded 16 h after ion implanta- tion and annealing shows a peak at 3.468 eV (Fig. 1).

The intensity of this donor bound exciton emission (D₀X) ¹⁰⁾ drops by a factor of about 30 in intensity com- pared to the reference part, showing that some implan- tation damage still remains after the annealing procedure and this decrease is not caused by the annealing proce- dure itself. However, the full width at half maximum (FWHM) of about 6 meV of the DX transition indicates still an overall good material quality. A longitudinal phonon replica ¹¹⁾ of the (D₀X) transition at 3.376 eV and the yellow luminescence ¹²⁾ band (YL) at 2.2 eV were also observed. At 1.461(3) eV a new transition la- beled Pt₁ can be observed accompanied by a second line centered at 1.446(3) eV. As shown in Fig. 1 the PL in- tensity of these transitions decreases and finally vanishes within 17 days after annealing. The apparent strong drop in intensity below 1.43 eV is caused by the decreasing sensitivity of the photomultiplier tube.

Since nothing else is changing but the decreasing Pt concentration due to the radioactive decay this clearly shows that these transitions must be caused by a recom- bination center involving Pt.

Fig. 2 shows 8 of the 15 recorded PL spectra covering the infrared range of ¹⁹¹Pt-doped GaN taken within

350 500 650

GaN:Pt, 4 K

Pt₁ 15 meV

17 d FWHM ~ 6 meV

YL

PL intensity (arb. units)

wavelength (nm) 3.5 3.0 2.5 2.0

16 h4 d D^oX-LO

D^oX

energy (eV)

840 870

x 40

1.47 1.43

27 days after annealing. All spectra were normalized to the PL intensity at 1.3 eV. The first PL spectrum re- corded 16 h after annealing shows new intense NIR lu- minescence labeled with Pt₂. Six single transitions, each separated by (15 ± 1) meV, are detectable starting with the first transition centered at 1.273(1) eV. Within the following 27 d the PL intensity of all six transitions de- creases continuously. Finally, after 27 d no NIR lumi- nescence can be detected any longer. Therefore we con- clude that these transitions have to be Pt-related as well.

Fig. 2:Photoluminescence spectra of ¹⁹¹Pt-doped GaN recorded with a Ge detector at 4 K within 27 d after im- plantation and annealing at 1300 K. All spectra are normalized to the PL intensity at 1.3 eV. The lowest spectrum (reference) corresponds to the unimplanted part of the sample.

In Fig. 3, the PL intensities of both Pt related PL- bands are plotted as a function of time after implantation and annealing. The solid lines correspond to exponential fits to the data using the following equation:

1 / 2

(ln 2) /

( )

0

Pt Pt t t

I t I e

The open squares correspond to the integral PL inten- sity of the Pt₁ transitions. The fit yields a half-life of (2.6 ± 0.6) d. The solid circles show the integral PL in- tensity of the Pt₂ transitions. In this case the exponential fit to the data yields a half-life of (3.1 ± 0.3) d. Both values are in good agreement with the nuclear half-life of ¹⁹¹Pt (t_½ = 2.9 d). These results clearly show, that the corresponding recombination center contains exactly one Pt atom. The involvement of more than one Pt atom in this defect would show up in a smaller time constant.

As shown in Fig. 2, the unimplanted reference part of the sample does not show any of these transitions after annealing and therefore annealing effects causing these transitions can be excluded. Furthermore, we do not ob- serve any other changes in the whole spectral region between 0.9 eV up to 3.5 eV, in particular no new lines coming up, so we conclude that iridium does not intro- duce any optically active recombination centers in GaN.

Fig. 3: Normalized integral PL intensities of the Pt-re- lated transitions Pt₁ and Pt₂ in GaN as a function of time after annealing. The solid lines correspond to ex- ponential fits to the data using equation 1.

Using the radioactive isotope ¹⁹¹Pt, we have shown for the first time that Pt introduces near-infrared photo- luminescence consisting of two single transitions (Pt₁) at 1.461 eV and 1.446 eV followed by a set of seven addi- tional transitions (Pt₂) starting at 1.273 eV each sepa- rated by 15 meV in energy. The time dependence of the Pt-related PL intensities clearly shows, that the corre- sponding recombination center involves exactly one Pt atom, but the exact recombination mechanism responsi- ble for the several transitions is not clear yet, and addi- tional investigations are needed to clarify this.

(1) J.D. MacKenzie, C.R. Abernathy, S.J. Pearton, U. Hömmerich, J.T. Seo, R.G. Wilson and J.M. Zavada, Appl. Phys. Lett. 72 (1998) 2710

(2) B. Kaufmann, A. Dörnen, V. Härle, H. Bolay, F. Scholz and G. Pensl, Appl. Phys. Lett. 68 (1996) 203

(3) A.J. Steckl, M. Garter, D.S. Lee, J. Heikenfeld and R. Birkhahn, Appl. Phys. Lett. 75 (1999) 2184

(4) R. Birkhahn, M. Garter, and A.J. Steckl, Appl. Phys.

Lett. 74 (1999) 2161

(5) R. Heitz, P. Thurian, I. Loa, L. Eckey, A. Hoffmann, I. Broser, K. Pressel, B.K. Meyer and E.N. Mokhov, Appl. Phys. Lett. 67 (1995) 2822

(6) J.M. Langer and H. Heinrich, Phys. Rev. Lett. 55 (1985) 1414

(7) J. Baur, M. Kunzer, K. Maier, U. Kaufmann and J. Schneider, Appl. Phys. Lett. 65 (1994) 2211

(8) G. Armelles, J. Barrau and J.P. Noguier, Phys. Rev. B 33 (1986) 1243

(9) Table of Isotopes, CD ROM Edition, Version 1.0 ed.

V.S. Shirley, S.Y. Frank Chu (John Wiley, New York, 1996)

(10) A.K. Viswanath, J.I. Lee, S. Yu, D. Kim, Y. Choi and C. Hong, J. Appl. Phys. 84 (1998) 3848

(11) R.L.Bergman, D. Alexson, P.L. Murphy, R.J.Nemanich, M. Dutta, M.A. Stroscio, C. Balkas, H. Shin and R.F. Davis, Phys. Rev. B 59 (1999) 12977

(12) H.M. Chen, Y.F. Chen, M.C. Lee and M.S. Feng, Phys.

Rev. B 56 (1997) 6942 940 960 980 1000 1020 1040 1060

GaN:Pt, 4 K

15 meV

reference 27 d 16 h

PL intensity (arb. units)

wavelength (nm)

1.31 1.29 1.27 1.25 1.23 1.21 1.19 1.17 Pt₂

energy (eV)

0 5 10 15 20 25 30

Pt₁: t_1/2 = (2.6 ± 0.6) d Pt₂: t_1/2 = (3.1 ± 0.3) d

PL intensity (arb. units)

time (d)

1.7 Observation of hydrogen in GaN doped with

¹¹⁷

Cd(

¹¹⁷

In)

M. Dietrich, A. Stötzler, R. Weissenborn and M. Deicher in collaboration with

ISOLDE-Collaboration (CERN, Geneva, Switzerland) The current research on GaN is driven by its promis-

ing applications as a material for LEDs and laser diodes in the blue and UV region ¹⁾. One aspect to be investigated is the behavior of H in GaN. During different steps of device processing, like crystal growth or chemical etching, H can be introduced into the material unintentionally. It may change the electrical properties by saturating dangling bonds, passivating shallow and deep level dopants and impurities or causing new, H-related levels in the band gap. Therefore the understanding of the behavior of H in GaN is essential for optimization of different processing steps.

We have recently reported on studies of Cd-H pairs in GaN with perturbed g-g-angular correlation spectroscopy (PAC) ²⁾. Implanted ^111mCd substitutes the Ga-site and can act as an acceptor after annealing of the implantation damage. H is implanted with low energy and trapped by Cd. The orientation of two different Cd- H pairs as well as their dissociation energy have been determined ²⁾.

The same Cd-H pairs should form using the isotope

117Cd. H will be released after the nuclear disintegration

117Cd ® ¹¹⁷In because In is isoelectronic to the host element Ga and the probe does not act as an acceptor anymore. The initially captured H may then leave the probe and start to diffuse. This onset of diffusion is investigated with PAC. Similar studies in InP and GaAs have been carried out successfully in the past ³⁾.

A prerequisite of these studies is the activation of the dopants implanted. Due to the high melting point of GaN and its tendency to loose N from the surface at temperatures above 1200 K, only a partial removal of implantation defects by annealing is possible. This an- nealing of the sample after implantation of ^111mCd has been studied in detail. After treatment of the crystal at 1323 K under N₂-atmosphere, 60 % of the probes sub- stitute an undisturbed lattice site ⁴⁾. At higher tempera- tures GaN starts to degrade. The possibility to reach temperatures up to 1573 K by adding a small amount of Al to the sample has been observed recently ⁵⁾.

We report on experiments that use this new technique to study the annealing of GaN after implantation of

117Cd(¹¹⁷In). GaN grown on sapphire has been im- planted with ¹¹⁷Ag that decays to ¹¹⁷Cd with a half-live of 73 s at the on-line mass separator ISOLDE at CERN.

The implantation was carried out with an energy of 60 keV and a dose of 1.2×10¹² cm^-2. Most PAC-meas- urements have been carried out at ambient temperature with the c-axis of the crystal pointing between two de- tectors at 45°.

Fig. 1 shows the fraction of ¹¹⁷Cd(¹¹⁷In) probe atoms at identical lattice sites in GaN (top), the quadrupole

coupling constant n_Q (center) and the width of its distri- bution Dn_Q in dependence on the annealing temperature (bottom). After thermal treatment at 773 K, 51(1) % of the probes are exposed to a unique electric field gradient (EFG). The fraction increases with increasing annealing temperature to 81(1) % at 1073 K and remains mainly constant to the final temperature of 1558 K. Thus, about 80 % of the probes are at sites with no defects present in the nearest neighborhood. Nevertheless, there are de- fects further away that are not directly correlated with the probe atoms. These are responsible for the distribu- tion of EFG represented by Dn_Q. The remaining 20 % of probes experience correlated defects in their nearest neighborhood creating strong non-unique EFG.

Fig. 1: Fraction of ¹¹⁷Cd(¹¹⁷In) atoms at identical lat- tice sites in GaN (top), quadrupole coupling constant n_Q (center) and width of its distribution Dn_Q (bottom) in dependence on the annealing temperature. The samples were annealed in evacuated quartz ampoules with ad- ditional Al.

The quadrupole coupling constant after annealing at 1073 K has a value of n_Q = 20.9(1) MHz. From this value one can calculate the EFG that a Ga-atom would experience at the same position with help of the Stern- heimer correction. This gives the largest component of

0 20 40 60 80 100

fraction (%)

20.0 20.5 21.0 21.5

n Q (MHz)

600 800 1000 1200 1400 1600 0

1 2

DnQ (MHz)

annealing temperature (K)