Spectroscopy and dynamics of rare earth doped fluorides

The defect structure of RE doped Fluorides has been studied along with the conductivity properties, using a variety of techniques, both experimental and theoretical.
Two systems have been studied in detail, which represent two kinds of defect states for RE doped SrFr.
The system SrFr:CeF, has been treated in terms of a random distribution of the Ce3* ions. Our model of this system includes the following assumptions:
i The Ce3* ions are distributed more or less statistically over the Sr2* sub-lattice.
ii The mobility of the interstitial F- ions is relatively large at NN and NNN sites and
low at the remaining sites.
iii If two Ce3* ions share NN or NNN sites the interstitial F- ions are able to move from one Ce3* ion to the other by crossing low barriers only. The two Ce3* ions
now form a cluster of size two.
Upon doping with Ce, statistical clusters of Ce3* ions are formed in accordance with the
above definition. Interstitial F- ions have a high mobility within the clusters. With increasing Ce concentration larger statistical clusters of Ce3* ions will appear. At some
concentration, the percolation threshold, a percolation cluster is formed and the conductivity is strongly enhanced. The behaviour of the DC conductivity as a function of the Ce concentration, observed by the AC impedance technique and TDR (chapters 3 and 4), is in line with this view. The conductivity increases strongly with the Ce concentration. Near the percolation threshold the rate of increase is much larger. We
have compared the experimental results of the DC conductivity with the theory of random walks in random barrier networks, which describes conductivity in a network of
high and low barriers. This theory provides a good qualitative description of the conductivity properties of Ce doped SrFr. However, it appears that the assumption that
the low potential barriers for interstitial F- motion is limited to NN and NNN sites is too crude. In reality, it is likely that the change from low to high potential barriers is more
gradual. This may explain the deviation between the experimental results and the theory,
especially in the critical region near the percolation threshold.
The MD results obtained for SrF, doped with La (a system comparable to SrFr:Ce) supports the model of statistical clustering of the large RE ions in SrF, (chapter 6). The
results show a similar behaviour for the conductivity and the activation energy as a function of the La concentration. From the ionic trajectories it can be seen that the F
ions have a high mobility in the vicinity of the La3* ions and in clusters of La3* ions.The defect structure of the system SrFr:YbF, is determined by the process of the
formation of compact clusters. The following assumptions seem to be valid:
i The RE ions form compact clusters. This process is called preferential clustering.
ii With increasing RE concentration more complex clusters are formed containing two
or more RE ions.
iii The interstitial F- ions are bound by the clusters.
iv Some cluster types are able to trap extra interstitial F- ions.
Chapter 9
Due to the formation of compact clusters the number of dipoles decreases rapidly. The
interstitial F- ions, which are responsible for the increase of the conductivity, are bound
more strongly to the clusters. Some of the clusters are able to trap extra interstitial Fions.
As a result, the increase of conductivity is only moderate compared to Ce doped
SrFr, and the behaviour with increasing Yb concentration is complex. Dissociation of
interstitial F- ions from the clusters can be observed in the DTA spectra. Above 1000 K
a (double) peak is visible, which coincides with the change to another conductivity
mechanism with a higher activation energy. This (double) peak dominates the DTA
spectrum at large Yb concentrations. Local distortions of the F- sub-lattice, due to the
compact clusters, give rise to the R,,, relaxation, The R,,, relaxation is associated with a
small activation energy of approximately 0.04 eV. Above 2 molYo Yb concentration the
R,,, relaxation contributes strongly to the dielectric constant, which indicates that above
this Yb concentration large numbers of clusters are formed with local distortion of the Fsub-
lattice. By MD simulation, we have observed that the À,,, relaxation appears in the
212, cluster. However, there may exist other cluster types, in which a similar relaxation
process, associated with a low activation energy, occurs.
The results for pure SrF, yield a new view on the conductivity mechanism at high temperatures. At moderate temperatures only very few Frenkel defects are present in
pure SrF., and the conductivity is (mainly) determined by the mobility of the vacancies at the F- sub-lattice. The appearance of a cluster-polarization-like contribution to the
dielectric constant in the TDR spectra, which increases with temperature, indicates that the mechanism for ionic conductivity becomes much more complicated at high
temperatures. A complex defect structure was found consisting of highly mobile vacancies and relatively slowly moving interstitial F- ions. The mobility of the vacancies
is enhanced within clusters due to the presence of interstitial F- ions. This is quite similar to our model of the conductivity mechanism of SrF, doped with randomly distributed Ce3* ions. In pure SrF, the defect structure is dynamic, due to the creation
and annihilation of the Frenkel pairs. This type of behaviour was confirmed by the MD simulations of pure SrFr. For the RE doped crystals the dynamic defect structure caused by Frenkel pair creation and annihilation, also exists at high temperatures, but it is mixed with the "static" defect structure of the RE dopant ions and the accompanying interstitial F- ions.
The MD simulations of chapter 6 show that this technique is very well suited to study defect structures and conductivity mechanisms. However, one has to make the
restriction that the number of thermally activated defects is not in a critical state. This restriction is related to the small size of the simulated systems. In the case of the simulation of pure SrF, the number of Frenkel defects is strongly dependent on the temperature and problems arise if very few Frenkel defects are present. For the RE doped systems this problem doesn't arise because the interstitial F- ions, which are
responsible for the enhancement of the conductivity, arc present due to the doping of the crystals. Thc Raman spectra show a strong dependence on the defect structure (chapter 7). A number of peaks appear due to the presence of the RE impurities and interstitial F ions.
By considering the symmetry properties of these peaks in combination with MD simulations (chapter 8) we were able to identify several high symmetry local modes.
The peaks, with cubic A,r, E, and Tr, symmetry, are associated with vibrational modes.
of cubes of lattice F- ions, which contain an interstitial F- ion. Compact clusters give rise to local distortions of the F- sub-lattice, which affect the cubes containing interstitials,
and the local modes vanish. The behaviour of these modes for different type of RE dopant ions is in line with our view of the defect structure of RE doped SrFr. For RE
concentration larger than 2 mol%o, the intensity of the local modes is much smaller for Yb doped samples compared to samples containing equal amounts of Ce. This indicates
that complex clusters are formed in Yb doped SrFr, which cause local distortions of the
F- sub-lattice. This result is in line with the observation of the R,,, relaxation for samples
which contain more than 2 molYo Yb. For the Ce doped samples the behaviour of the intensity of the local modes can be understood by assuming a random distribution of the
a 1 + ' Le- lons. Our investigations also open the possibility of further research. In this thesis we
have shown that the RE doped Fluorides can be simulated quite well. Therefore it seems worth while to invest large amounts of computing time into the simulation of extended systems and to use longer simulation time. In this way the defect structures can be simulated more accurately. Also, a higher accuracy of the frequency dependent (AC) conductivity can be achieved, and the observed powerlaw behaviour, believed to be
related to many particle interactions, can be studied. For pure SrFr, the use of a larger
system would reduce the problem of "quantization" of the number of Frenkel defects.
Experimentally, there are also several possibilities. Recent developments in pulse generator technology may expand the TDR frequency range to higher frequencies. An
overlap between the TDR and MD frequency domains can be achieved. The R,,, relaxation itself can be observed by TDR if low temperatures are employed. This may yield information whether or not one distinct relaxation process is responsible for the R,,, relaxation or if a collection of relaxations with low activation energies are present, as suggested by the AC impedance results. ...

Zie: Concluding remarks

Concluding remarks

Meer informatie in de catalogus
Meer informatie in Picarta