L. Ziegeler, J. Roth, P. Schaaf, and H. Metzner. II. Physikalisches Institut, Universität ... J. Olivier-Fourcade and J. C. Jumas. Laboratoire de Physico-Chimie des ...
PHYSICAL REVIEW B
VOLUME 58, NUMBER 17
1 NOVEMBER 1998-I
Perturbed angular correlation study of the thiospinel b -In2 S3 L. Aldon Laboratoire de Physico-Chimie des Materiaux Solides, Universit de Montpellier 2, 34095 Montpellier Cedex 5, France
M. Uhrmacher II. Physikalisches Institut, Universita¨t Go¨ttingen, Bunsenstrasse 7/9, D-37073 Go¨ttingen, Germany
C. Branci Laboratoire de Physico-Chimie des Materiaux Solides, Universit de Montpellier 2, 34095 Montpellier Cedex 5, France
L. Ziegeler, J. Roth, P. Schaaf, and H. Metzner II. Physikalisches Institut, Universita¨t Go¨ttingen, Bunsenstrasse 7/9, D-37073 Go¨ttingen, Germany
J. Olivier-Fourcade and J. C. Jumas Laboratoire de Physico-Chimie des Materiaux Solides, Universit de Montpellier 2, 34095 Montpellier Cedex 5, France ~Received 1 April 1998! The electric field gradients ~EFG’s! of 111In(EC) 111Cd nuclei at the different crystalline sites in spinel b -In2 S3 have been measured, using perturbed angular correlation spectroscopy. The radioactive 111In tracers were introduced into the samples by means of ion implantation or during the chemical synthesis using natural indium doped with 111In. The radiation damage after the implantation was annealed by heating the samples to above the transition temperature T5693 K where the phase transition to cubic a -In2 S3 occurs. In contrast to previous PAC measurements, three electric field gradients were found. Their temperature dependences were measured in the temperature range between 8 and 743 K. The crystalline structure was checked by x-ray diffraction and refined by means of a Rietveld analysis. Fully consistent with the refined structure, the three observed EFG’s were attributed to probes residing in the different sulphur octahedra and tetrahedra of b -In2 S3 . The EFG values are reproduced by the point charge model. A strong damping of the perturbation functions was observed in the temperature range between 150 and 370 K, which was attributed to dynamical hyperfine interactions caused by aftereffects of the electron capture decay of 111In. @S0163-1829~98!05341-7#
I. INTRODUCTION
The perturbed gg -angular correlation method ~PAC! has often been used as a tool to study, via the hyperfine interaction of a radioactive probe nucleus, the properties of solids such as magnetic and structural phase transitions, chemical reactions, and defects on an atomic scale. Measuring the electric field gradient ~EFG! in semiconductors offers the possibility to change the electron density by adding electrically active impurities ~e.g., Li! or by varying the temperature. The EFG’s at 111Cd probe nuclei on substitutional cation sites for binary metal oxides have been studied systematically in Go¨ttingen,1–5 including bixbyites (M 2 O3 ) ~Ref. 6! and ternary oxides with either M 2 Cu2O5 structure7 or delafossite (ABO2 ) structure.8 The PAC experiments for the last three classes of compounds gave information on the local oxygen configuration in terms of the cation size. In all of these compounds the 111In(EC) 111Cd probe was found to replace the central cation of the oxygen octahedra. The structure of the spinels generally is of great interest as it provides the possibility to use the free space in this lattice for storage of foreign atoms such as lithium. Either their exact position, their number, or their chemical binding are of major importance. The large band-gap semiconductor b -In2 S3 is also interesting as a window material in thin-film CuInSe2 photovoltaic devices.9 0163-1829/98/58~17!/11303~10!/$15.00
PRB 58
Spinels are composed of octahedra and tetrahedra. Only a few examples are known where the 111In probe was safely attributed to the central site in an oxygen tetrahedron; in Mn3O4 with a high EFG value,3 in Fe3O4 ,10 and in CdIn2S4 ~Ref. 11! with a low value. In b -Ga2 O3 , which is also composed of octahedra and tetrahedra, only the octahedral site is occupied by the probe atom.12 In2S3 exists in three phases.13,14 In the defective spinel b -In2 S3 ~stable up to 693 K!, the regular sites of indium atoms are the centers of both disturbed sulphur-octahedra and tetrahedra. In addition, the probe atom does not represent an impurity but belongs to the constituting elements of the compound. Therefore, this compound offers the unique possibility to observe the different types of local environments ~octahedra and tetrahedra! under otherwise identical conditions. Furthermore, it will be interesting to find out if the general rules found for oxides will hold for chalcogenes, too. II. SYNTHESIS OF b -In2 S3
All the samples were prepared by solid state reactions. The binary thiospinel b -In2 S3 was synthesized by means of a direct reaction from stoichiometric mixtures of the elements in evacuated tubes ( P51023 Pa). Afterwards, this mixture was heated up to 573 K for one day. Then, the temperature was raised to finally reach a value of 1373 K which remained 11 303
©1998 The American Physical Society
11 304
L. ALDON et al.
constant for three hours. As an annealing treatment, the temperature was then kept at 1073 K for eight days. Pellets made from each substance were implanted with 111In1 ions at an energy of 400 keV and a total dose of about 1012 ions, using the Go¨ttingen ion implanter IONAS.15,16 Then the samples were annealed under the conditions described below in order to remove the radiation damage. One sample was synthesized using indium metal, into which the radioactive 111In tracer atoms were introduced by electrolysis. The deposition of 111In was performed in an electrochemical cell with a natural indium cathode and a platinum anode. The electrolyte was a mixture of Na2SO4 , an ammonium salt, and 111InCl3 ;6H2O. A voltage of 5 V was applied to the cell loading to a current of 140 mA. Finally, the electrochemical reaction was cathode:
In31 13e 2 → 111In0 ,
111
2H1 12e 2 →H2 , anode:
~1!
PRB 58
containing all the information on the hyperfine interaction. In the case of static electric interaction, G 22(t) is defined as 3
G 22~ t ! 5
(
n50
S 2n ~ h ! cos~ v n t !
3exp@ 2g 2n ~ h ! d t # d @ g 2n ~ h ! n Q t, t R #
~3!
with v n being the primary transition frequencies and S 2n their respective amplitudes; their values are defined in Ref. 20. The parameter d denotes the width of a Lorentzian distribution of the EFG and the function d @ g 2n ( h ) n Q t, t R # accounts for the damping of G 22(t) due to the finite time resolution t R of the apparatus. The PAC experiments were carried out using a conventional slow-fast setup with four NaI~Tl! detectors in 90° geometry or a fast-fast setup21 with four BaF2 detectors positioned in the same geometry. Then, the twelve time spectra obtained from all possible combinations of the four detectors were used to calculate the experimental perturbation function R(t), given by
4OH2 →O2 14e 2 12H2O.
5
R ~ t ! 52
N ~ 180,t ! 2N ~ 90,t ! 5A 22 f i G i22~ t ! , N ~ 180,t ! 12N ~ 90,t ! i51
(
~4!
After the electrolysis, 0.41955 g of indium, containing an activity of 0.41 mCurie of 111In tracers, were placed in a tube together with the sulphur, to react to b -In2 S3 . Subsequently, the same thermic program was performed as in the case of the nondoped samples.
where f i denotes the relative fractions for different EFG’s contributing to the PAC spectrum and G i22(t) the corresponding perturbation factors.
III. PAC METHOD AND APPARATUS
IV. EXPERIMENTAL RESULTS
The time-differential perturbed angular correlation technique with radioactive 111In probe atoms was used to measure the EFG’s at the different crystallographic sites of b -In2 S3 . A detailed description of this method can be found in the literature.17–19 The short overview given here will demonstrate how, in principle, the EFG in a solid compound can be measured. The EFG is a tensor V i j defined by the second derivative of the electric potential V(r) at the probe site. Its nine components can be reduced to the three diagonal elements V xx , V y y , and V zz , arranged as V xx RT.
111
PRB 58
Cd in annealed b -In2 S3 , measured in vacuum at different measuring tempera-
cated in Figs. 2 and 3. The temperatures were chosen in random order. All the PAC spectra had to be fitted with a set of three different frequencies. As the result, two of them ( n Q2 and n Q3 ) agree quite reasonably with the ones found in the previous PAC experiments.22,23 The third one is a lower frequency, characterized by the EFG parameters n Q1 535 MHz and h 1 50.6. In all the Fourier spectra, the first peak of this triplet is either clearly separated from the larger one at about 60 MHz or at least visible as a shoulder on the left side of the large peak. In fact, the fractions f 1 and f 2 of these two EFG’s amount to 75%. This value was fitted in
Ref. 23 for the dominant EFG with n Q 565 MHz. It seems that in the measurements of Refs. 22,23, the experimental resolution did not allow one to detect the lowest frequency. Furthermore, the temperature region from 150 to 300 K was not shown or discussed in Refs. 22,23. The low-temperature measurements given in Fig. 3 show a dramatic damping of the perturbation function in that temperature region, whereas at lower temperatures the three frequencies reappear. A similar behavior was observed in some oxides, especially in La2O3 ~Ref. 24! and also in In2O3 ~Ref. 25!. In these cases, it could be proven by using 111m Cd probes, that the after
PRB 58
PERTURBED ANGULAR CORRELATION STUDY OF THE . . .
FIG. 3. PAC spectra and their Fourier transforms for tures T m
