close

Вход

Забыли?

вход по аккаунту

?

Applications of SEPIL to the Solid State Defect Chemistry of Fluorites and Ultra-trace Inorganic Analysis [New analytical methods (16)].

код для вставкиСкачать
[I971 J . Buter, S. Wassenaar, R. M. Kellogg, J. Org. Chem. 37, 4045 (1972).
11981 R. M. Kellogg, S. Wassenaar, Tetrahedron Lett. 1970, 1987; D. H. R. Barron, B. J. Willis, J. Chem. SOC.Perkin Trans. I 1972, 305.
[I991 K. T. Ports, D. M. McKeough, J. Am. Chem. SOC.96, 4276 (1974).
[200] M. P. Caua, M. Y Lakshmikanfham, Acc. Chem. Res. 8,139 (1975).
[201] K . T. Potfs, D. R. Choudhury. A . J. Ellioff, U. P. Singh, J. Org. Chem. 41.
1724 (1976).
[202l J. D. Bower, R. H . Schlessinger, J. Am. Chem. SOC.91. 6891 (1969).
[203l R. Criegee, Angew. Chem. 87,765 (1975); Angew. Chem. Int. Ed. Engl. 14,
745 (1975) references cited therein.
[?-MI P. S. Bailey, A. Rustaiyan, T. M. Ferrel, J. Am. Chem. Soc. 98, 638 (1976);
C. W Giilies, ibid. 99, 7239 (1977); P. S. Boiiey, Th. M. Ferrel, ibid. 100, 899
(1978), references cited therein.
12051 W G. Alcock, B. Mile,J. Chem. Soc. Chem. Commun. 1976, 5; B. Mile. G.
M. Morris, ibid. 1978, 263.
[206] K. L. Gallaher, R. L. Kuczkowski, J. Org. Chem. 41, 892 (1976).
(2071 D. P. Highley, R. W Murray, J . Am. Chem. Soc. 98, 4526 ( I 976).
[208] D. R. Kerur, D. G. M. Diaper, Can. J. Chem. 55, 588 (1977).
12091 K. Griesbaum, P. Hofmann, J. Am. Chem. SOC. 98. 2877 (1976).
[210] R. Criegee, G. Schroder, Chem. Ber. Y3, 689 (1960).
[211] F. L. Greenwood, L. J. Durham, J. Org. Chem. 34, 3363 (1969).
12121 R. P. Lartimer, R. L. Kuczkowski, C. W Gillies, J. Am. Chem. SOC.96, 348
(1974); cf. also R. A. Rouse, ibid. 96, 5095 (1974).
(213) P. C. Hiberfy, J. Am. Chem. SOC.
YX. 6088 (1976); see also P. C. Hiberty. .
I
P. Deuidal, Tetrahedron 35, 1015 (1979).
[214] S. D. Razumouskri, L. V. Berezoua, Izv. Akad. Nauk SSSR 1, 207 (1968).
12151 W.R. Wndt, W. A . Goddard I I I , J. Am. Chem. SOC. 97, 3004 (1975): see
also L. B. Arding, W. A . Goddard 111,ibid. 100, 7180 (1978).
(2161 H. E. O'Neal, C. Blumstern, Int. J. Chem. Kinet. 5, 397 (1973); N . J. Pifrs,
B. J. Finlayson, Angew. Chem. 87, 18 (1975); Angew. Chem. Int. Ed. Engl.
14. I (1975).
12171 S. Flmrir. M. Granger, J. Am. Chem. Soc. 91, 3330 (1969); 92, 3361 (1970);
S. Fliszdr. J. Renard, Can. J. Chem. 4X, 3002 (1970); S. Fliszdr, J. Renard.
0. Z . Simon, J. Am. Chem. SOC.93, 6953 ((971).
[218] M. Schulz, N . Grossmann, J. Prakt. Chem. 318, 575 (1976).
12191 M. Bertrand, J. Carles. S. Fliszar, Y. Rousseau, Org. Mass Spectrom. 9. 297
(1974).
(2201 G. Klopman, C. M. Joiner, J. Am. Chem. SOC.97, 5287 (1975).
Applications of SEPIL to the Solid State Defect Chemistry of Fluorites and Ultra-trace Inorganic Analysis
methods (16)
By John C. Wright, D. R. Tallant, F. J. Gustafson, M. V. Johnston, M. P. Miller, D.
S. Moore, L. C. Porter, and J. R. Akse[*]
Although it has been recognized for many years that the spectra of lanthanoid ions can provide
useful information about short range phenomena in the neighborhood of these ions in a solid
material, lanthanoid spectroscopy has only infrequently been used for studying complex materials because of the problem of line-sorting the complex spectra that are obtained. The advent
of convenient tuneable lasers has eliminated this problem. By selectively exciting probe ion luminescence (SEPIL), it is possible to obtain fluorescence and excitation spectra from a single
kind of crystallographic environment. Two applications of this method are discussed in this paper. The first application is the study of the defect chemistry of fluorite materials (compounds
with CaF2lattice). It is shown how this method can provide unique information about the solid
state chemistry, thus clarifying many of the unexplained behaviors of this important class of
material. The second application shows how ultra-trace analysis can be carried out by causing
an association between an analyte ion and a fluorescent probe ion. The unique crystal field levels of a probe ion associated with a particular analyte can be selectively excited so that traces of
the ion to be analyzed can be detected with very high selectivity and with very low detection
limits.
1. Introduction
The optical spectroscopy of lanthanoid-doped materials
has been studied for many years"]. Early efforts were concerned with understanding the physical mechanisms for the
characteristic sharp line spectra and strong fluorescence of
lanthanoids (rare earth metals)I2.'1. The discovery that many
lanthanoid compounds function very well as lasers[41brought
about a sudden surge in interest in the spectroscopy of this
group of elements, particularly with a view to understanding
the relaxation mechanisms and dynamic processes which determine the efficiency of the quantum electronic applicat i o n ~ ' ~This
' . work has led to a clearer understanding of both
['I
Prof. Dr. J. C. Wright, Dr. D. R. Tallant ['*I, Dr. F. J. Gustafson, M. V.
Johnston. Dr. M. P. Miller, D. S. Moore, L. C. Porter, Dr. J. R. Akse
Department of Chemistry, University of Wisconsin
Madison. Wisconsin 53 706 (USA)
['*I
738
Present address: Sandia Laboratories, Albuquerque, NM 87 115 (USA).
0 Verlag Chemie, GmbH, 6940 Weinheim, 1979
the static and dynamic aspects of lanthanoid spectroscopy
and in many cases has furnished a degree of predictive power.
It was realized in the early spectroscopic studies that because the line positions and transition probabilities depend
upon the immediate environment of the lanthanoid ion, one
had at ones disposal a probe for the local crystal field'']. By
measuring the polarization of the transition, the line positions and/or the Zeeman splitting tensor, the local symmetry
of a lanthanoid ion could be determined in many
cases[2.3.6 '01. some workers have succeeded in gaining information about the static and dynamic properties of lattice
phonons["-l5' and about exciton propagation[l6].The majority of the work has been carried out on simple materials,
usually single crystals in which the lanthanoid ion occupies a
unique crystallographic site. Fundamental investigations of
more complex materials have usually been avoided because
of the dificulty in characterizing the site of the lanthanoid
0570-0833/79/1010-0738 $ 02.50/0
Angew. Chem. Inr. Ed. Engl. 18. 738-752 (1979)
ion or in interpreting the spectroscopic data. If a lanthanoid
ion can occupy several different kinds of crystallographic
sites, the spectroscopy is much more difficult since each lanthanoid site contributes a unique set of lines to the overall
spectrum observed. Each line in the spectrum has to be studied individually in order to measure the polarization and the
intensity as a function of the dopant concentration, the crystal treatment, the temperature, the symmetry of the Zeeman
All of the lines can then
splittings, and/or the lifetime[7-t0~17].
be divided into groups that have the same behavior and correspond to the lanthanoid sites that are present. This process
has been successfully used by a number of groups but it is
generally time-consuming, tedious, and subject to error.
Even after such a procedure is carried out, the results can still
be ambiguous. Two groups studying KCI :Sm2 came to diametrically opposed views and wrote a series of passionate articles about their different c o n c l ~ s i o n s ~ " ~ ~ .
+
In order to function efficiently as probes of short range order in more general situations where one is not dealing with
simple materials, it is necessary to use techniques that can
simplify the spectroscopy of a complex material (i. e. of a material where there are several environments encountered by a
lanthanoid ion). A key idea for surmounting the problems
was published in 1966 by Voron'ko et uZ.lfn1.They used a
mercury lamp and a monochromator with a 3 A bandpass to
selectively excite a specific absorption line of a specific lattice site of the lanthanoid ion and obtained fluorescence
from that site alone. By exciting other absorption lines, they
could obtain the fluorescence spectrum of each of the other
sites. Using this method, they were able to classify the lines
in the spectrum of CaF2:Er3+ into three different Er3+ sites.
The method was not used for later studies, however, because
the fluorescence intensities encountered with this excitation
method were low and the bandwidth could not be made sufficiently narrow. The advent of tuneable dye lasers has eliminated these problems and made the selective excitation
to obtaining fluoresmethod quite p r a c t i ~ a l [ ' ~In. ~addition
~~.
cence spectra from single sites by exciting a specific absorption line, one can also obtain single site excitation spectra by
monitoring a specific fluorescence line as the laser excitation
wavelength is changed. A complex absorption spectrum can
therefore be split into the spectra of individual sites by monitoring different fluorescence transitions and obtaining excitation spectra. The problem of line classification then becomes quite simple (Fig. 2). Figure l a shows the absorption
spectrum of CaF, :0.05 mol-% Er. This absorption spectrum
is reproduced as the upper trace in Figures Ib-d. The bottom trace in these three figures is the excitation spectrum
that results when different fluorescence lines are monitored.
Figures 1b-d show how the absorption lines can be assigned
to three lattice sites. The excitation line from the site shown
in Figure I d is in the same position as a line in Figure l b , yet
both excitation spectra can be obtained without interference
from each other because the fluorescence lines monitored do
not overlap.
We believe that selective laser excitation methods (also
called site selective spectroscopy by later workers["') will extend the application of lanthanoid ions from studies aimed at
understanding the spectroscopy of the lanthanoids to studies
of more general interest where the lanthanoids are used as
probes of short range environments in complex materials. In
Angew. Chem. Inr. Ed. Engl. 18, 738-752 (1979)
0.3r a)
bl
I
Ir
cl
1
dl
I
I
1
I [nmlFig. 1. a) Absorption spectrum of the Z+F(41,,,2+2HI,,2) transition in
CaF2:0.05mob% Er'+. This spectrum is reproduced as the top trace in Fig. I bd. b-d) Excitation spectrum of the Z-F transition obtained while monitoring
the fluorescence line at 540.0 nm (b), 539.15 nm ( c ) or 537.98 nm (d).
this paper, we present two examples of how the methods
might be used-one example is a fundamental study of the
solid state chemistry which occurs in charge compensated
crystals, while the second example is the development of
practical techniques for the ultra-trace analysis of inorganic
ions.
2. Solid State Chemistry of Fluorites
One of the most important problems in materials science is
the description of point defects. In most cases the defects in a
material determine its overall propertiedz2*"I. Compounds
having the fluorite structure have played an important role
in understanding defects both because of their practical importance and their use as model systems. A great deal of
work has been carried out with compounds having a fluorite
structure, particularly as regards dielectric relaxation, ionic
conductivity, thermal conductivity, diffuse and Bragg neutron scattering, diffusion, density measurements, NMR,
ESR, ENDOR, UV, visible and IR spectroscopy[27441. In this
paper, we shall restrict our discussion to the fluorites CaF2,
SrF2, and BaF2.
It is generally accepted that the predominant defects in
CaF,, SrF,, and BaF, are fluoride interstitials (F:)[45' and
fluoride vacancies (V;) which exist in equal numbers because of charge neutrality r e q u i r e m e n t ~ [ * ~If
. ~trivalent
~].
cations like Er3+ are introduced into a lattice such as
CaF,(Er&), additional fluoride interstitial ions are required
for charge compensation. There will be Coulombic interactions between Ere, and F: promoting association into single
pairs (Ere,. F,)" which can have different symmetries depending upon the relative positions of Er,, and F:. The two
most important symmetries are tetragonal and trigonal[lol.
The relative numbers of the different symmetry pairs is determined by the relative free energies of association and the
temperature of the crystal[471.In addition, the Er& and F:
can be distant from each other and occupy sites of cubic
symmetry. The free F: that are found then force a decrease in
the V;. concentration. The model can be described in terms
of two equilibria analogous to solution equilibria.
739
Perfect CaF, lattice + F: + V;
t
anomalous changes in the ratio of the associated to non-associated pairs.
Ere.
41
(Er,;F:)
This simple model has been extensively employed in many
studies with apparent success. Particularly convincing evidence was provided by EPR measurements, which conclusively identified the tetragonal and trigonal site symmetries
expected for the lanthanoid ion in an associated pair and the
cubic symmetry of the site expected for a lanthanoid in an
unassociated
ENDOR measurements identified
the presence of the FI in the tetragonal and trigonal
SiteS14’.501,
There were several observations however that did not
agree even qualitatively with this m 0 d e 1 [ ~ ~ -The
~ ~ l .ratio of
associated pairs to non-associated pairs (or trigonal + tetragonal to cubic pairs) was measured by EPR spectroscopy as a
function of the crystal temperature[531.CaF2:G d 3 + samples
were heated in an oxygen-free atmosphere until the distribution of site populations reached equilibrium and then were
quenched rapidly to freeze the high temperature distribution.
The simple model would predict that the ratio of associated
to non-associated pairs would decrease as the effective crystal temperature was raised in order to maximize the entropy
of the system. Instead, exactly the opposite behavior was observed: The number of associated pairs increased as the temperature was raised. The ratio was also measured as a function of the Gd3+ ~ o n c e n t r a t i o n ~ ~According
~ - ~ ~ l . to the simple model the ratio of associated to non-associated pairs
should increase as the dopant concentration is raised, but the
measurements actually showed that the ratio decreases.
Problems were also encountered in the interpretation of
Bragg neutron diffraction and diffuse neutron scattering data
in highly doped samples of CaF2:Y +.These experiments
revealed two types of F: ions, those with distortions along the
(110) direction and those with distortions in the (111 j direction from the central (‘A, ‘A, ‘A) cubic position of a normal FI
site12x,2y1.
Fluoride vacancies could also be observed. The
relative concentrations of these three types of defects could
be explained by postulating that the two types of F: ions and
the V; form a cluster designated as a 2:2:2 cluster since it
contains two of each type of defects. There are also a number
of variations in the structure and composition of this basic
cluster. The Y;., ions in the sample could not be observed directly; though there was evidence from the diffuse neutron
scattering measurements that the Y & were only loosely associated with the clusters. It was further argued that the cation
mobility in CaF, should be sufficiently small that one would
~ .there
~ ~ ] . is only a
not expect Y;, to cluster e x t e n ~ i v e l y [ ~If
loose association of Y;., with the 2 :2: 2 clusters, the clusters
would possess a net negative charge which would favor their
dissociation. What stabilizes the formation of negatively
charged clusters? Furthermore, the distance required between two FI with (1 10) distortions in the 2: 2 :2 cluster was
2.0 A, much closer than the 2.3 A separation expected theoretically[2x-30’.
In order to explain these two problems, Cutlow
proposed that an additional covalent interaction existed between F: which both stabilized the negatively charged 2 :2 :2
clusters and produced the 2.0 A separation between F:[591.
We shall see later how Cutlow’s proposal also explains the
140
2.1. Experiments on Selective Excitation in CaFz
We have investigated materials having a fluorite lattice by
the selective laser excitation technique. This technique is the
only one that has permitted the examination of all the different sites occupied by a trivalent dopant. Most of our work
has been concentrated on CaF,: Er3+119.20,ho 621, although we
have found that SrF21631
and BaF2Ih4]exhibit a similar behavior. The excitation spectrum that one observes for a CaF2:0.2
mol-% Er3+ sample when fluorescence from all Er sites is
monitored is shown in Figure 2. The lines in this excitation
450
1 Lnml-
445
Fig. 2. Excitation spectrum of CaF2:0.2 mol-X Er’+ that results on monitoring
the fluorescence of all E-Z (4S1/2+41,r,2)transitions with a broad bandwidth
monochromator while a dye laser is scanned over the possible Z-H (‘1,vl4F5/z)transitions.
spectrum can be divided into groups corresponding to the
different Er sites by monitoring a single fluorescence line
from an individual site. The single-site excitation spectra obtained in this way are shown in Figure 3. Each line in Figure
3 corresponds to a line of the excitation spectrum in Figure 2.
I
/I
L.
L.
b’
O’
450 445
445
C’
dl
450
l
ei
Fig. 3. Single site excitation spectra of CdFl:0.2 mol-X Er’ * (see Frg. 2) (Z-H
(41,s,2-4F5/2) transitions]. The spectra are observed by monitoring the fluorescence at: a) R35.4 nm ( A site), b) 835.8 nm (B site), c) 836.1 nm (C site), d) 843.1
nm (D(1a) site), e) 654.4 nm (D(2a) site).
Only two representative spectra (Fig. 3d and 3e) of the many
that make up the bands shown in Figure 2 have been shown.
It was possible to show that at least 15 sites contributed lines
to the bands, 11 which had spectra similar to Figure 3d and
four which had spectra similar to Figure 3e. The site whose
spectrum is shown in Figure 3a corresponds to an Er3+ ion
with a F’, in the nearest possible interstitial position producing a tetragonal site symmetry. This identification could be
made because Rector et u1.[’1 had previously shown that the
lines from this site had a Zeeman splitting tensor that possessed tetragonal symmetry. Similarly, the spectrum in Figure 3b corresponds to an Er3+ ion with a Fi that produces a
Angew. Chem. Inl.
Ed. Engl. 18, 738-752 (1979)
trigonal site symmetry. The spectra in Figures 3c-e were
shown to correspond to different clusters of Er3+ ions whose
symmetry remains undctermincd. I t i \ clcitr Itom Figure 2
that these clusters cause the dominant features in the overall
spectrum at the 0.2 mol-% concentration level. A great deal
of effort was spent in showing that clusters of Er3+ actually
cause these spectral features, both because of their importance in the optical spectrum and because many workers did
not believe that Er3+ clusters could be important[" "I or in
fact even be formed because of low Er3+ mobilities[591.A
prime reason for this disbelief was the inability of EPR
measurements to identify any features that would correspond
to Er3+ ion clusters[501.It should, however, be pointed out
that many optical spectroscopists have observed the lines and
correctly assigned them to Er3+ clusters because of their rapid increase with dopant concentration[". 17-20.31-33.651. In
1966, Voron'ko et al. were particularly successful in unraveling the optical spectrum into four sites which are essentially
those represented in Figure 3['01 and in attributing two of the
sites to clusters of Er3+ ions.
2.2. Proof of Clustering in Fluorites
Two experiments show quite conclusively that clusters of
Er3+ ions are responsible for the features shown in Figures
3c-e. If a particular electronic multiplet within the 4f"electron configuration of a particular lanthanoid ion is excited,
resonant or non-resonant energy transfer can occur with another lanthanoid ion[60.661.Non-resonant energy transfer is
generally quite weak if the initial and final energy states differ by more than a few hundred cm-'. However, if the two
ions involved in the energy transfer are quite close to each
other, as is the case for the clusters, the non-resonant energy
transfer rate can become efficient. In fact, non-resonant energy transfer is generally the dominant non-radiative relaxation mechanism for the clusters. A series of measurements
were performed on CaF2 in order to determine how the
transfer rate depends upon the energy mismatch between the
initial and final energy states. The results are shown in Fig-
I
.
,
.
1000
.
2000
Energy Gap
1cm"I
electronic level can relax non-radiatively to a lower electronic level by dissipating the energy difference as lattice phonons. Although multiphonon relaxation is more rapid than
non-resonant energy transfer for a particular value of energy dissipation, there are generally enough different electronic levels for two arbitrary lanthanoid ions that a combination
of lower levels will lie closer in energy to an excited level
than the gap appropriate for multiphonon relaxation. Nonresonant energy transfer is therefore very important in clusters. No energy transfer has been observed for the tetragonal
and trigonal Er sites which are isolated from other Er ions.
The energy transfer causes some interesting phenomena.
Up-conversion of excitation energy can be particularly efficient if two ions in the clusters are excited[*".6n1.
One ion
transfers its excitation energy to the neighboring excited ion
and causes a fluorescence from a level much higher in energy
than that originally pumped. Thus, one can visually observe
green fluorescence from a CaF2:Er3+ crystal when the Er
clusters are being excited in the red. If one monitors the
green fluorescence, the red excitation spectrum obtained will
only contain lines from the Er clusters. The single pairs (i. e.
the tetragonal and trigonal sites) cannot undergo the energy
transfer and consequently cannot produce the up-converted
fluorescence. This method provides a convenient way in
which to discriminate between the single (Erca.F,) pairs and
Er3 clusters.
The second experiment which shows the presence of clusters involves the double doping of CaF2 samp1esl"'l. In a
sample of CaFz:0.2 mol-% Yb3+,0.01 mol-% Er3 , the site
distribution is determined by the Yb3 concentration, which
is large enough that the optical spectrum is dominated by the
clusters. The Er3+ will also be distributed among the various
possible sites, and some Er3+ ions will form mixed clusters
with Yb3+. The Er3+ concentration is sufficiently low that
pure Er clusters are improbable. However, if one monitors
the Er transitions, mixed Er --Yb clusters can be seen. The
lines are shifted slightly from those observed in CaF, : Er'+
samples because Yb3+ has a different ionic radius from that
of Er3 . An example of the differences observed is shown in
Figure 5 . Since it is clear that Yb3+ is sufficiently close to
Er3+ to affect the Er spectrum, it follows that these transitions must arise from clustering of the cation dopants.
+
+
+
+
-
'
.
moo
Fig. 4. The points of curve a represent the non-radiative rate for multiphonon relaxation as a function of the energy gap between the emitting level and the next
lowest level. The points of curve b represent the non-radiative rate due to non-resonant energy transfer between two ions in a cluster. All of the data is for CaFl
with Er' as dopant.
+
ure 4 as the isolated points. It can be seen that there is a
strong dependence upon the energy mismatch. The multiphonon relaxation rate was also measured for CaF, as a
function of the amount of energy lost by the non-radiative
relaxation (line (a) in Figure 4). In this process, an excited
Angew. Chem. Ini. Ed. Engl. 18, 738-752 (1979)
538
540
542
544
1 Inml+
Fig. 5. Single site fluorescence spectrum for the E-Z(4S,,2+41,s,2) transition of
the C site in a) CaF,: Er" and b) CaF2:Er". Yb" . Note the shift of the lines
due to the influence of the Yb'+ on the clusters.
141
tions and must be included in any description of the solid
state equilibria.
2.3. Concentration of Clusters
Secemski and Low pointed out that the observation of
clustered ions in the optical spectrum does not necessarily
mean that clusters are important[501.The ions in the clusters
may simply have much more favorable radiative transition
probabilities and quantum efficiencies than single pairs and
therefore may look brighter in the fluorescence spectra.
Their actual concentration could still be quite low and thus
might prevent EPR spectroscopic detection. The authors
point out that transition probabilities for EPR are much
simpler than for optical transitions. The measurement of the
cluster concentrations therefore constitutes an important experimental step in solving the solid state chemistry of the
fluorites.
The only technique that permits observation of the cluster
species is optical spectroscopy. However, the relationship between the fluorescence intensity or absorption coefficient
and concentration is complex. One can follow how any given
site will change in samples of different concentrations but
the transition probabilities must be known in order to determine an absolute concentration. The transition probability
for an absorption transition can be obtained if that same
transition can be observed in fluorescence where a fluorescence lifetime can be determined[']. The fluorescence lifetime determines the total relaxation rate for the level. The radiative quantum efficiency must be measured in order to obtain the total radiative relaxation rate for the level. Finally,
the radiative relaxation rate for a particular transition of interest can be derived by measuring what fraction that transition represents of the total fluorescence from the level['"1.
The actual procedure for obtaining all the concentrations is
more involved than described here (for further details see
referenceI6']). The results thus obtained are shown in Figure
6, where the site concentrations are plotted as a function of
total dopant concentration in the system CaF2:Er3+.The C
-3
-2
Ig c t r [ M o l - % l
-1
Fig. 6. Absolute concentrations of the Er sites In CaF2:Er'
Er' dopant concentration.
-0
as a funcllon of
and D(2) designations refer to Er clusters that have the spectra shown in Figures 3c and e. The concentration of the
unassociated or cubic Er site does not appear in Figure 6 because it cannot be observed: electric dipole transitions are
forbidden in a cubic crystal field[21.It should be clear from
this data that the clusters are a dominant feature of the site
distribution at almost all commonly used dopant concentra142
2.4. Discussion
As shown in Figure 6, at very low concentrations the single
pair sites (the tetragonal and trigonal sites) change linearly
with concentration. As soon as the clusters appear, the number of single pair sites changes more slowly until their absolute concentration actually begins to decrease. A decrease in
the absolute concentration of any site is unexpected, since a
simple equilibrium of defects will allow only changes in the
rate of increase of a site population with dopant concentration, but never a
Our explanation for this behavior requires an extension of
the simple model presented eariier[4'.61'.
Perfect CaFz lattice
e
F:
+
+ V;
Er;.,
41
(Erca'F,)" + (Ercd-F,)"$(2 Erc;2
+
F,)" +
F:
41
EL 3FJ'
The clustering has been shown explicitly as an additional
equilibrium. The additional covalent interaction between F:
ions that was proposed by cat lo^['^' has been represented by
an equilibrium in which (2 Ere,. 3 F,)', a charged cluster, is
formed. This last type of equilibrium is the key to the explanation of the anomalous behavior of CaFz and may well be
the equilibrium that controls the solid state equilibria in the
fluorites. At high temperatures, the Er,, and F: are both free
to move about in the lattice. As the temperature is lowered,
the Er& lose mobility and the distribution becomes frozen.
This event fixes the total number of Ere, that exist as single
ions, dimers, trimers, eic. However, the F: are still mobile
and can be distributed among the different Er;, sites to form
different species. In particular, if the bonding of a F: to make
a charged species like (2ErC;3 Fi)' is very favorable, the
concentration of free FI would be forced lower. However, the
equilibrium between unassociated Ere, and the single associated pairs (Ere,. F,)" must follow a simple mass action relation
which determines the ratio of unassociated to associated Erea.
If [F:] is small because of the formation of (2Erca.3F,)',
the ratio [Er;,]/[(Ercl.F,)"] will become large as the associated single pairs are forced to dissociate. Higher dopant
concentrations produce more clusters which form the
(2Erc,-3FJ' type of species that removes F: from the equilibria. Thus, the absolute concentrations of the tetragonal
and trigonal sites will decrease, as is observed experimentally
(Fig. 6). The same argument provides an explanation of the
anomalous behavior observed in EPR measurements: the
cubic site concentration increases relative to the associated
single pair sites as the total dopant concentration is increased.
Angew. Chem. Int.
Ed. Engl. 18, 738-752 (1979)
3.1. Analysis of Lanthanoids
This model also provides an understanding of the temperature dependence observed by Franklin and M a r z ~ l l o [ ~ ~ ~ ,
We found that coprecipitation from aqueous solution was
who found that the concentration of cubic sites decreases rea
suitable
way to incorporate a n a l y t e ~ ~Treatment
~~].
of a solative to the concentration of tetragonal sites as the temperalution
containing
unknown
amounts
of
different
lanthanoid
ture of the sample is raised. At high temperatures, species
ions with Ca(N03)2and then with NH4F leads to coprecipisuch as (2 Ere,. 3 Fi)' will dissociate and raise the concentratation of the lanthanoids in the resulting CaFz microcrystals.
tion of free Fi. The free FI ions can then recombine with
We have measured the amount of coprecipitation that occurs
Er3+ ions in cubic sites to form the associated single pairs.
as a function of the total amount of CaF2 precipitated and
The results from Bragg neutron diffraction and the diffuse
have found that the distribution law for lanthanoid copreciscattering experiments relate very directly to this model. The
pitation is very favorable. The precipitation therefore perselective laser excitation studies showed clearly that contrary
forms several functions:
to some expectations, there is extensive clustering of Er3
1) it incorporates the analyte into a suitable matrix,
ions. The presence of several Er3+ ions in a cluster would
2)
it serves as a preconcentration step,
help to stabilize it against the dissociation discussed for the
3)
it
serves as a separation step, especially since other ions
(2 :2 :2) cluster model presented by Cheetham et al. [2x,291,
with
much different ionic radii are not likely to coprecipiSteele ef al. and CaflowrS91.
The cluster of Er3 ions would
tate.
be further stabilized by the additional covalent bonding of F:
Although it might be expected that the quality of the miproposed by C a t l ~ w l ~This
~ ] . interaction also produces the
cro-crystals would be too poor to show sharp line structure,
charged species which determines the defect equilibrium disthe lanthanoid transitions are reasonably sharp. An example
cussed.
of the excitation spectrum of a CaF2:Er3+ precipitate is
This model is supported by density measurements carried
out by Franklin as a function of dopant c ~ n c e n t r a t i o n ~ ~ ~ l . shown in Figure 7a. There are lines from both the tetragonal
Franklin observed that the density of CaF2:Gd3+ increased
too rapidly with concentration to be explained by having a
single Gd3+ charge compensated by a single Fi. He sugbl
gested that HF from the atmosphere around the crystal during its growth had entered the lattice and formed associations with (Gdc; Fi)" pairs. No evidence could be found for
the F H F groups that would be formed. However, the
additional HF could be brought into the lattice in association
with the cluster species and result in the anomalously high
C)
d)
values for the density.
Although this model provides a self-consistent explanation
for the behavior of CaF,, it remains a speculative model that
must be proven. We are presently engaged in experiments
446
448
that will monitor the populations of all the sites as the temA [nm]
perature of the crystal is changed. We expect to see the popuFig. 7. Excitation spectra of the Z-H(41,s,2+4Fs,1) transition for a CaF1:Er"
lation of species like (2 Ere,. 3 F,)' decrease as the temperaprecipitate dried at different temperatures. A monochromator with a very wide
ture is raised while the (2 Er,, .2 F,)" and (Ere,. F,)" concenbandpass was used to monitor the fluorescence in order to include fluorescence
trations increase correspondingly. These experiments are not
from all the sites present. a ) Room temperature (A. B see text). b) 350°C. c )
5M)"C, d) 1000°C.
complete but it is clear that temperature changes cause large
redistributions in site population.
and trigonal Er sites present which are labeled A and B respectively. If the precipitate is dried at 350°C, dramatic
changes occur in the spectrum as shown in Figure 7b. Lines
that have the same appearance as the cluster site lines shown
in Figure 3e become prominent while the lines from single
3. Ultra-trace Chemical Analysis
pair sites almost disappear. This behavior is quite unexWhen we began our experiments in selective laser excitapected and cannot be explained with any degree of certainty.
tion, we were struck by both the excellent signal/noise ratio
It would appear to indicate that the Er3+ ions initially copreand ability to selectively excite a specific site without intercipitate in isolated positions in the lattice and do not have
ference from other sites. These two features are also of great
sufficient mobility at room temperature to reach the equiimportance for a good trace analysis method. One would like
librium distribution. As the temperature is increased, they
to have analytical methods that can measure very low confinally achieve sufficient mobility (in precipitates at least) to
centrations with a high degree of selectivity for the analyte of
move and form clusters with each other. At higher temperainterest. Could the trace analysis of lanthanoids be carried
tures (Fig. 7c), the spectrum changes once more. All the preout using the laser methods? Our systems at that time were
viously observed lines and bands disappear and are replaced
single crystals, not particularly interesting samples for an
by lines from a single site-labeled GI. The fluorescence inanalytical determination. In order to use our selective laser
tensity from this site is several orders of magnitude higher
excitation methods, we had to find a convenient way of inthan those of the previously discussed sites. Other workers
corporating lanthanoid analytes from a material of analytical
have studied CaF2 under very similar conditions to ours and
interest into a material suitable for our methods.
have discovered the reason for the appearance of the new
+
+
+
Angew. Chem. Ini. Ed. Engl. 18, 738-7S2 (1979)
143
The fluoride charge is compensated by oxygen anions instead of by fluoride ions, because F, reacts with water
vapor:
2F:+H20
+
2HF+0"
The oxygen anion replaces one of the nearest neighbor fluoride ions in the lattice and produces an Er site with trigonal
symmetry. The spectrum continues to change as the precipitate is heated to 1000 "C (Fig. 7d). Lines associated with sites
labeled G2, G3 and G4 increase in intensity up to 1000°C
but then become weaker again at higher temperatures. These
sites are probably the result of additional oxygen incorporation. For example, the site labeled G4 is believed to result
from replacement of seven nearest neighbor fluoride ions
and the fluoride interstitial charge compensation by four
oxygen anions[7o1.At very much higher temperatures, the
spectrum changes to that of CaO["I.
The sites that are present below 500 "C are of no analytical
value. As one lowers the concentration of the rare earth in
solution, the line intensities drop non-linearly at a rapid rate
until they cannot be detected at a concentration of 1.25 pg/
ml. It is assumed that the non-linear behavior is caused by
the dissociation of the single pairs of Erea and FI at very low
concentrations. The lines of the single pairs disappear and
are not replaced by spectra from the dissociated pairs because the cubic Er site of a dissociated pair does not have allowed electric dipole transitions. The temperature region between 500" and 700°C is optimum for an analysis because
the site distribution has collapsed into a single site which is
intensely fluorescent and is stable against dissociation at very
low concentrations. It is particularly advantageous to have
only one site because the peak intensity is higher and the system becomes more insensitive to other influences that could
change the site distribution.
Any lanthanoid that fluoresces in CaFz can be determined
by the procedure described. In order to fluoresce, an ion
must have excited electronic levels within the 4f" electron
configuration (transitions to other configurations have much
broader lines because of the additional interactions with the
lattice) and must have an excited electronic level whose nonradiative relaxation is slow in comparison with its radiative
relaxation[51.Of the 15 lanthanoid ions, only La, Ce, and Lu
do not have excited electronic levels. (Ce actually has one excited level but it is too close to the ground state to be used.)
Gd has many excited levels but they lie in the UV beyond
310 nm where it is difficult to obtain convenient tuneable
lasers. The non-radiative relaxation rate is determined by the
strength of multiphonon processes that convert electronic energy into vibrational energy. The rate for this process depends exponentially upon how much energy an excited level
must dissipate in the lattice. If there is another electronic level that has a lower energy than the excited level, a non-radiative transition to the lower level can occur and the energy
difference is dissipated as several phonons. Figure 4 describes the relationship between relaxation rate and the energy gap for CaF2. If the typical radiative transition rate is l o 3
s - I, an energy gap greater than 2600 cm is required before
a level will generate sufficient fluorescence. All lanthanoid
ions having excited electronic levels-Pr3 +, Nd3 , Pm3+,
Sm3+,Eu3+, Gd3+, Tb3+, Dy3+, Ho3+, Er3+, Tm3+, and
~
+
744
Yb3+-also have at least one level that can fluoresce efficiently. We have experimentally demonstrated the feasibility
of analyzing all of these ions excepting Pm3+, Gd3+, and
Yb3+. Pm is radioactive, while Gd and Yb have electronic
levels in a spectral region that is not accessible with our lasers. As will be discussed later, there are methods that can be
used for the analysis of those ions which cannot be detected
by this method, namely La3+, Ce3+, Gd3+, Yb3+, and
Lu3+.
10-l
I
10
102
103
104
c [nglrnll-
Fig. 8. Fluorescence intensity (I)of a CaF2:Er" precipitate as a function of
Er'+ concentration (c) in the original solution (calibration curve).
The analytical working curve for Er analysis is shown in
Figure 8. The concentrations are referenced to the original
solution before precipitation. At high concentrations the
curve becomes non-linear; at these concentrations additional
lines also appear in the spectrum which have been proven to
be associated with cluster formation (see later discussions).
The spectrum of the cluster site is shown in Figure 9. Note
the double line structure that is associated with exchangesplittings between the two lanthanoids. The change in the
site distribution accompanying cluster formation causes the
non-linearity of the calibration curve (Fig. 8) at high concentrations.
CoF2: 0.1%Er
II
Fig. 9. Excitation spectrum of the Z+H(41,5,2+4FS,,) transition obtained with
wide bandpass monitoring of a CaF2:0.1 mob% Er3+ precipitate. By comparing
this spectrum with that in Figure 7c, the additional lines that are formed at high
concentrations are readily recognizable. The three lines from Figure l c are
marked.
The limit of detection for this method depends upon the
number of photons available from the laser excitation source.
One of the advantages of a pulsed laser is the ability to use
time resolution techniques or gating to discriminate between
signal and background. The lanthanoids have relatively long
fluorescence lifetimes, usually 200 bsec or longer, and discrimination becomes simple. On the other hand, the lifetimes
are short enough to allow photomultiplier dark noise fluctuations to be greatly reduced by gating. Thus, the limit of detection is determined more by the excitation source intensity
than by the fluctuations in source intensity and by the residual dark noise fluctuations. With currently available instruAngew. Chem. Int. Ed. Engl. 18, 738-752 (1979)
mentation a detection limit of 25 fg/ml (25 x
g/ml) of
Er3+ can be achieved, calculated by extrapolation of the signal/noise ratio at 1 pg/ml. The absence of clean-room facilities prevented us from reaching the detection limit of the
method experimentally. The instrumentation used by us included a N,-pumped dye laser with an average power of 1
milliwatt and an uncooled S-20 photomultiplier. With a
cooled photomultiplier and a laser power of a modest 50 milliwatts, the detection limit would be 25 x
g/ml.
The sources of interference in this method can be divided
into different classes:
1) formation of other precipitates
2) changes in efficiency of coprecipitation
3) line-broadening
4) changes in site distribution
5 ) quenching of fluorescence.
The first class of interferences are obvious ones. If we assume the lanthanoid analytes are in solution as we begin the
procedure, problems can arise if there are substances present
which precipitate when either Ca(N03), or NH4F is added.
Insoluble fluorides would provide the main problem since
most anions that precipitate Ca will have already precipitated the lanthanoid ions. The insoluble fluorides include
such compounds as SrF,, PbF,, BaF2, and BiF3. Generally if
such substances are present in concentrations large enough
to affect the amount of CaF, precipitate that is formed, this
will cause more severe interferences of the other classes mentioned above. Interferences of class 2 have not been identified or studied.
Fig. 10. Excitation spectrumof the Z-H(41,5/2-t4F5/2)transition for a CaF2:0.1
mob%Er’ precipitate after addition of SO ppm of Na +;for usual conditions see
Figure 7c. It can he seen that Na’ changes the spectrum and therefore acts as an
interference.
+
We have examined examples of each kind of ion that can
act as a potential interference. As typical monovalent cations, Li +,Na + and K were added to the lanthanoid-containing solution. Na + and Li produce marked changes in
the spectrum (Fig. 10). They affect the site distribution by accelerating charge compensation by oxygen anions instead of
by fluoride anions. Additional lines appear in the spectrum
that are probably associated with lanthanoid-ion sites which
are compensated by the monovalent cation.
The spectrum obtained when K + is added to the solution
is shown in Figure 11. The site distribution is completely
changed, the intensity of the lines is dramatically lowered,
the lines are broadened, and the background is increased.
These changes are very severe and would drastically limit the
applicability of the technique if they could not be eliminated.
Traces of other ions have also been observed to severely
change the site distribution.
+
+
Angew. Chem. Inl. Ed. Engl. IS, 738-752 (1979)
448
450
452
454
1 Inml+
Fig. 11. Excitation spectrum ofthe Z-+H(41,5/2-4Fs/2)transition for a CaFI:O.I
mol-% Er3+ precipitate after addition of 0.1 mol/l of K +.Comparison with Figure 7c shows the profound influence of K +.
We believe that these problems are best interpreted in
terms of the solid state equilibria in the precipitated microcrystals. Any aliovalent ion that enters the lattice will require
a charge compensation and must therefore affect all of the
defect equilibria. If all the aliovalent ions are present at trace
levels, the position of the defect equilibria can fluctuate as
other trace contaminants enter the lattice. The situation is
analogous to an unbuffered aqueous solution which can fluctuate in pH as trace quantities of different ions enter the solution. A buffer is required to fix the equilibria positions and
remove the sensitivity to traces of other ions. Li and K were
chosen to act as the solid state buffer. The solution containing Ca(N03), and the lanthanoid analytes is treated with
LiN03 and KN03 before the precipitation is carried out. The
precipitate is heated at as low enough a temperature that
only sites of type G1 are formed. This step eliminates the
sensitivity of the system to any of the monovalent cations
and has been made an integral part of the procedure. All of
the results described in the following sections were obtained
on adopting this procedure.
1 [nml
-
Fig. 12. Excitation spectrum of the Z-+H(41,s,2+4F5/2)transition for a CaF2:O.l
mob% Er’+ precipitate after addition of 0.005 mol/l of Sr2+. Comparison with
Figure 7c shows the line-broadening effect of Sr2+ and the additional sites that
appear.
Divalent cations do not interfere at low concentrations,
but do so at higher concentrations (ca. 5 x
mol/l). An
example of a spectrum obtained with Sr(NO& present in the
original solutions is shown in Figure 12. The lines are all
broadened and therefore result in a lower measured intensity. Sr ions can enter the lattice substitutionally for Ca and
cause disruptions in the lattice order because of their different ionic radii. The lanthanoid ions are therefore present in
many different environments, resulting in both broadened
lines and a background fluorescence.
Trivalent cations include the lanthanoid ions themselves
and many other similar ions. The trivalent cations interfere
at concentrations above cu.
mol/l by the same clustering mechanism that causes non-linearities in the working
curve (see Fig. 8). This interference might be expected, because trivalent cations having properties similar to those of
lanthanoid ions should cause a site distribution characteristic
745
of a higher lanthanoid concentration in the cation lattice.
The lanthanoids themselves could cause additional interference by participating in energy transfer processes that
quench the fluorescing level. These processes have a particularly strong influence if the two levels lie quite close to each
other[661.However, we have not found any evidence of lanthanoid fluorescence being quenched by other lanthanoids
over the entire linear region of the working curve. This lack
of interference by quenching means that all of the lanthanoids can be determined in a single sample without any need
for prior separation.
Interference by transition metals does not generally occur
until concentrations greater than ca.
mol/l are reached.
Fe3+,however, interferes particularly strongly and its concentration must be kept below ca. 4 x
mol/l. Interferences by transition metals belong to the class 5 interferences
(quenching of lanthanoid fluorescence).
We have also examined a number of the common anions
such as Br-, C1-, SO:-, N O j and PO:- and have not
found any interference at concentrations below ca.
mol/l. We are presently extending our study of interferences
to include many other ions. In addition, we are engaged in
applying this technique to samples of geological interest in
order to test whether interferences will arise which we had
not anticipated and to demonstrate the practicability of the
method in cases where lanthanoid analysis is important.
-
Ho
Er
bronic sideband for a single pure electronic transition in
CaF,:Eu3+ is shown in Figure 14. The peak of the sideband
is ca. lo3 times weaker than the parent electronic transition
shown on the right-hand side in Figure 14. This selectivity is
available in both the excitation and fluorescence spectrum
I
562
566
570
574
1 "n--
Fig. 14. The structure of the vibronic sideband for the 'F,,-5D,I transition of
CaF2:Eu3+.The sensitivity at the lower wavelengths has been increased by a
factor of 167 to make the vibronic sideband features visible.
and therefore results in a lo6 rejection factor between ions
that have close lines. If there are lines that lie outside of the
region of vibronic sidebands (a region determined by the
maximum phonon energy of the lattice; for CaFz this is ca.
370 cm-'), the rejection factor is larger. This selectivity is
more than adequate for performing multiple-ion analysis on
samples containing all of the lanthanoids.
3.2. Analysis of Non-fluorescent Trivalent Ions
The clustering behavior that is observed at high lanthanoid concentrations suggests a way for extending the method
to other ions that cannot fluoresce. If one wanted to perform
a La3 analysis (La3 is a lanthanoid ion that has no excited
electronic levels in the 4f" electron configuration), a
CaF2:Er3 precipitate could be formed in the presence of a
high Er3+ concentration and an unknown concentration of
the La3+ analyte. Large numbers of Er3+ clusters would
form, a few of which would also contain a La3+ ion. Since
La3+ has a different ionic radius from Er3+,the Er3+ ions in
the clusters containing La3 would have slightly different
crystal field splittings from those of the Er3+ ions in pure
Er3 clusters. By selectively exciting the crystal field levels of
the La3+-containing clusters, one can obtain a spectrum of
only these clusters. This procedure would establish a 1 :1 relationship between the non-fluorescent and the fluorescent
ion. The sensitivity of the technique depends upon how well
one can measure the fluorescent ion. We have already seen
that very low detection limits can be achieved by the methods described. The selectivity will depend upon how well the
analyte's spectral lines can be resolved from the lines of other
sites.
Naturally, this analytical method presents a problem. The
success of the method rests upon the ability to discriminate
between the weak lines of the clusters containing analytes
and the much stronger lines of the pure clusters. Since the
separations between the lines are not likely to be very large,
one can expect that the lower detection limit is fixed by the
selectivity of excitation. This limitation can be overcome by
taking advantage of the efficient non-resonant energy transfer that occurs between ions in clusters. The rates of these
processes for fluoride compensated clusters can be estimated
from Figure 4. The energy mismatch can be estimated by
constructing the double ion energy levels which show all the
possible energy states for two ions. The double ion levels for
Er are shown in Figure 15. The gap below the D level is large
+
+
+
d [nrnlFig. 13. Excitation spectrum of the Z-H(411s/2+4F5/z) transition for a mixed
precipitate of CaF2:Er3+,Ho'+. The individual Er'+ and Ho3+ transitions are
indicated. The bandwidth of the fluorescence monitor was sufficiently wide to
include both the Ho'+ and Er'+ fluorescence.
+
+
The technique is highly selective regarding the analysis of
specific lanthanoid ions. This selectivity is demonstrated for
a CaF2 precipitate containing Ho3+ and Er3+, two lanthanoid ions with very similar energy levels. The spectrum in
Figure 13 shows that the excitation transitions are close to
each other. The fluorescence transitions are also quite close.
However, the lines are all sufficiently sharp, so that either Ho
or Er can be excited individually. The resulting fluorescence
spectrum contains only transitions from the ion excited. The
selectivity is ultimately limited by the weak vibronic sidebands that accompany any electronic
The vibronic sidebands occur because the electronic states of the
lanthanoids are loosely coupled to the lattice, and changes in
the vibrational state of the lattice can happen simultaneously
with an electronic change. Vibronic processes result in the
absorption or emission of a lattice phonon. If the experiments are performed at low temperatures, as is usually the
case in order to obtain narrow line profiles, then only phonon emission processes need to be considered. Vibronic sidebands are much weaker than the pure electronic transition
and their shape reflects the density of phonon states at different energies and the amount of coupling between a phonon
of particular symmetry and the electronic levels. The vi746
Angew. Chem. I n f . Ed. Engl. 18, 738-752 (1979)
I I:=
201 G-
The technique retains the sensitivity of the direct lanthanoid
method and can separate the different analyte peaks nicely
by selective excitation. It has been used successfully for the
analysis of Sc3+, Y 3 + , La3+,Ce3+,Gd3+, and Lu3+ when
Er3+ was used to monitor the fluorescence. These ions are
the ones that could not be determined directly by the previously discussed method. The ions that can be determined
directly will quench the Er3+ fluorescence if incorporated
into a dimer and thus will not interfere in this method. Thus,
this method is complementary to the direct method. All these
studies are very new and there are still many factors which
must be investigated. We are very pleased with these initial
results and expect that the method will be suitable for practical applications.
-
-
ARY
-
0Lz1 Er
YRY
-
3.3. Analysis of Non-fluorescent Ions Using Charge Compensation
2 Er
Fig. 15. Left: Electronic energy levels for a single Er’+ ion. Right: Possible levels
for two Er” ions (“double ions”).
I
bl
Fig. 16. Excitation spectrum of the Z+H(41,5,2+4Fs,2) transition for a CaF2
precipitate containing 0.02 mol-16 Er’+ and a) 0.02 mob% La’+ or b) 0.02 mol-16
Ce”. The lines due to Er-La or Er-Ce clusters are seen by comparison with Figure 7c. A third line from the Er-Ce cluster is hidden by the line at 44.4 nm.
enough for D to fluoresce, while the gap below the E level is
too small to allow eficient fluorescence. A cluster that contains two Er3+ ions should therefore not exhibit fluorescence
from the E level but will fluoresce from the D level. If one of
these Er3 ions is replaced by a La3 ion which has no electronic levels, the double ion levels of the Er-La dimer are
identical to a single Er3+ ion’s levels and both the E and D
levels can fluoresce. The fluorescence spectrum from the E
level can only contain contributions from the Er3+ ions that
are paired with La3+, while the fluorescence of the pure
Er3 cluster will be quenched.
Examples of typical excitation spectra for the detection of
La3+ and Ce3+ are shown in Figure 16 when fluorescence
from the E level is monitored with a wide bandpass instrument. Note the absence of lines from pure Er clusters. The
relative line positions and intensities are similar to those of
the pure Er cluster spectra shown in Figure 9 except that the
exchange splittings are not observed and the line positions
are shifted. There does not appear to be any simple correlation between the shifts and the ionic radius of the analyte.
+
+
+
Angew. Chem. In1 Ed. Engl. 19, 738-752 (1979)
The idea of achieving a 1:1 association between the analyte and a fluorescent ion, whose fluorescence is modified by
the analyte, can be extended to other metal ions if one uses
charge compensation to promote an association. In an arbitrary binary compound M,X, where the cation M is not
trivalent, a trivalent lanthanoid dopant will require a charge
compensation. This can be either an intrinsic compensation,
such as a lattice vacancy, or an extrinsic compensation.
There are several ways of achieving extrinsic compensation.
If M is a divalent cation, the charge of the lanthanoid ion can
be compensated by replacing another M ion by a monovalent cation. In this case only those monovalent cations are
suitable whose ionic radii are close to that of the lanthanoid
ion because both are substituting at the M ion lattice site.
The charge of the lanthanoid ion can also be compensated
by replacing an X ion by another anion that has one unit
more negative charge. If M is a tetravalent cation, the lanthanoid ion can be compensated by replacing another M ion by
a pentavalent cation with an ionic radius similar to that of
the lanthanoid ion or by replacing an X ion by an anion with
one unit smaller negative charge.
A less restrictive situation arises when a tertiary compound, M,A.X,, is used. If M is again a non-trivalent ion
which is replaced by a trivalent lanthanoid ion, the same
kinds of charge compensation are possible as in the binary
system. In addition, the A ion can be replaced by another ion
which has a similar radius. Since the identity of A can be
varied widely, the ions of most elements of the periodic systems are potential charge compensators. If M is a divalent
cation and A is a cation, the lanthanoid ion can be charge
compensated by replacing an A ion by an ion with one fewer
positive charges. If M is a tetravalent cation, charge compensation is achieved by replacing A by an ion with one more
positive charge.
Both the lanthanoid ion and its charge compensating ion
have equal and opposite effective charges relative to the lattice and therefore tend to associate because of their Coulombic attraction. When the charge compensating ion enters the
neighborhood of the lanthanoid ion, it changes the crystal
fields at the lanthanoid site in a unique way. The new crystal
field levels can be selectively excited or monitored in the
presence of many other sites. If the charge compensator is an
analyte, one can use the line positions to qualitatively indi747
cate the presence of the analyte and the line intensities to
quantitatively measure its concentration. This method has
been given the acronym SEPIL (selective excitation of probe
ion luminescence).
Procedures and guidelines had to be developed to aid the
search for proper systems and techniques in order to put
these ideas into practice. One of the first problems is the difficulty of carrying out a study of many materials, each of
which can be prepared in different ways. Moreover, an arbitrary material generally contains several sites, each having
their individual set of lines. Although selective excitation
enables the lines in a spectrum to be classified according to
the sites, the method requires a great deal of time, especially
because one does not know where the lines are. Since the
lines are sharp in both fluorescence and excitation, time must
be spent at the beginning of studying a material, merely in
order to determine the positions of the lines.
The problem of line sorting can be alleviated by using
Eu3+ as the probe ion. Eu3+ has a singlet ground state (7F0)
and a singlet excited state (5Do)[21.
Only one transition can
occur between these two states in either fluorescence
(5Do+7Fo)or absorption (7Fo+5D0).
If one monitors the fluorescence from a material with a very wide bandpass so that
all fluorescence can be seen and scans the excitation wavelength of a narrow bandwidth dye laser over the region of the
'FO+'Do transition, the resulting excitation spectrum will
contain one line from each site. Thus in a single scan, one
can obtain all of the information about the sites present in a
material and an indication of their concentration. There is
the possibility, however, that the lines for two sites accidentally appear at the same wavelength. Such an occurrence can
be detected by reference to the many other transitions of
Eu3 . Of course it is not necessary to restrict the probe ion to
Eu3 after having carried out the initial experiments. A lanthanoid ion can be chosen which gives the best sensitivity
and selectivity for the particular analysis.
There are a number of restrictions on the choice of material that can be used in guiding survey experiments to find
suitable materials:
+
+
1) The short range order about a lanthanoid ion dopant must
be sufficiently high that sharp line transitions occur.
2) Both the lanthanoid ion and the analyte ion must be incorporated into the lattice.
3) Both the lanthanoid ion and the analyte ion must be associated with each other.
4) The material must not quench fluorescence from the lanthanoid.
Our preliminary experiments have always been aimed at
determining whether the particular material under study
meets these requirements. If the material quenches fluorescence, there is little one can do with the material. In the preparation of a sample there are many variables which can affect whether a material meets the other requirements. These
variables include the method of sample preparation (precipitation from solution, solid state reaction, cooling of a melt,
molten flux reaction, solvent evaporation, etc.), the temperature of annealing, the type of reagents used and the valence
state of the dopants, the method of incorporating the dopants
(coprecipitation, diffusion, solid state reaction), the reaction
vessel material used, and the atmosphere above the material
during its preparation. Contamination must be carefully
748
guarded against in the initial studies before one knows what
ions a given material is sensitive to. Once a method of preparation is found which gives sharp lines and good intensities,
different possible analytes are introduced into the material
and the 7Fo+SDoexcitation spectrum is obtained to see if
new sites are formed as a result of association between the
lanthanoid ion and the analyte ion. The preparation variables, particularly the temperature of annealing and the type
of reagents used, will also affect the association. The identification and development of suitable materials are therefore
involved procedures. It is hoped that once experience has
been gained on a number of representative materials, a more
efficient method for obtaining suitable materials can be developed.
We have further restricted our survey experiments to materials that can be heated, so as to reduce the number of lattice defects, to permit diffusion of ions into association with
each other, and to provide a range of annealing temperatures
over which association can take place. We have used materials whose cations have ionic radii similar to those of the lanthanoids, and selected analyte ions whose ionic radii are
close to the radius of the lattice ion which they replace.
The first example of a suitable system was BaS04:Eu3+,
PO:-. EuCI3 and Na2S04were added to a solution containing trace concentrations of PO:- and BaSO, was precipitated by adding BaC1,1731.The Eu3+ and PO:- coprecipitated with the BaS04. The excitation spectrum of the undried precipitate is complex and indicates the presence of at
least 9 sites. There are two sites in the spectrum that are associated with presence of the PO:-- ions which compensate
the Eu3+ -charge. The spectrum is poorly resolved however
and cannot be easily used for PO:' analysis. If the precipitate is heated, the spectrum and the site distribution change.
The sites that correspond to associates with PO:- disappear, as do many of the other sites. At higher temperatures
between 800-1000 "C, the spectrum becomes sharper and a
new, strong line appears that is proportional to the PO:concentration. The excitation spectra of a BaSO, precipitate
ignited at 940°C without and with the PO:- analyte are
shown in Figure 17a and b, respectively. The line at 578.2
nm corresponds to the site containing PO:-. It should also
be noted that the presence of PO:- causes other sites to disappear, e. g. the site whose line appears at 574.9 nm. This effect is believed to be caused by the shift in the solid state
equilibria that must occur when PO:- is added (this will be
discussed in more detail below).
bI
h
Fig. 17. The excitation spectrum of the 'Fo-5D, transition of BaS0,:Eu" a)
without PO:- and b) with 4 ppm PO:- in the Na2S04solution.
Angew. Chem. I n ( . Ed. Engl.
IS, 738-752 (1979)
The methods described earlier in the paper can be used to
isolate the transitions of the PO:- site from those of the
other sites in the spectrum. Remembering that the PO:line at 578.2 nm in Figure 17b represents a 7Fo+5Doexcitation transition that can equally well be observed in fluorescence (5DO-+7F0),
we can tune a monochromator to 578.2 nm
and scan the dye laser wavelength over the spectral region of
the electronic 5D2-multiplet. The 5DZstate, with J=2, will
transition can
have ( 2 J + 1) crystal field levels. The 7F0-+5Dz
therefore be expected to have five transitions for each site.
The experimentally observed spectrum is shown in Figure
18. It contains exactly five lines, and there are no indications
of lines from other sites as might be expected from the wellresolved lines of Figure 17. Thus Figure 18 demonstrates
once again the high selectivity of the technique for specific
analytes.
463
bl
,il
A
y10Z
580.6
581 0
581 4
A Inrnl-
Fig. 20. The 'Fo-r5D, excitation spectrum obtained by wide bandwidth monitoring of a CdMo04 sample containing a) Eu7+ and b) Eu' and N b 5 + .
+
compensation, the intensity of the main line increases dramatically and the weaker line disappears. Clearly the As5
has an effect on the site distribution-but no new lines are
formed.
+
465
464
1 [nrnl-
bl
Fig. 18. Single site excitation spectrum of the 7Fo-5D, transition in BaS04:
EU' +,PO: obtained by monitoring the 5Do-7Fo transition at 578.2 nm for the
P O 2 site.
Other systems have also been found that are suitable materials for this procedure. Two examples of such materials are
shown in Figures 19 and 20: the 7Fo-+5Do
excitation spectrum of CazNb207:Eu3+both undoped and doped with
Zr4+,and of CdMo04:Eu3+ both undoped and doped with
Nb5+.In the spectra new lines appear that are characteristic
of the presence of Zr4+ and NbS+,respectively, and can be
used for their analysis. Other systems have been found that
give analyte lines for Li+, N a + , VS+,Ta5+, Cu', Ag+,
Bi3+, Sb5', As5+, Hg+, and S2-. Work is continuing on
these systems.
a1
i
bI
I
580
-581
A
[nm 1
Fig. 21. 'F,-5D0 excitation spectrum obtained by wide bandwidth monitoring
of a PbMo04 sample containing a) EuZ+and b) Eu' ' and A s S + .
Even more dramatic changes are observed in the case of
%SO4:Eu3+. Figure 22a shows the spectrum of SrS04: Eu3
without extrinsic charge compensation. The Eu3 occupies
one intrinsic site. If PO:- is added as a charge compensator,
a completely new set of lines forms and the intrinsic line
disappears (Fig. 22b).
+
+
1 l n m l --?I
Fig. 19. 'F<,-+'Do excitation spectrum obtained by wide bandwidth monitoring
of a Ca2Nb2O7sample containing a) Eu' and b) Eu' and Zr4 + .
+
+
The addition of an analyte ion to a system can have other
effects. In many cases, lines characteristic of an analyte-lanthanoid site fail to appear, while lines for an intrinsically
compensated lanthanoid site disappear upon adding a charge
compensating ion. Such an example is shown in Figure 21
for PbMo04:Eu3+. If no analyte ions are introduced as
charge compensators, the two lines in the 7Fo+5Doexcitation spectrum of Figure 21a indicate there are two major
sites intrinsic to the material. If As5+ is added for charge
Angew. Chem. Int. Ed. Engl. 18, 738-7S2 (1979)
576
577
570
A [nmIF-+
579
Fig. 22. 'Po+ 5Doexcitation spectrum obtained by wide bandwidth monitoring
of a SrS04 sample containing a) Eu' and b) Eu' and 5 &ml PO:+
+
All of the examples given are unsuitable for practical applications at the present time because the intensity of the
749
analyte lines is a non-linear function of concentration. At
high concentrations the intensity becomes constant, while at
low concentrations it plummets so rapidly that one cannot
work below concentrations of about 50 ppb in most cases. If
this problem were eliminated, the characteristics of the technique would be very similar to the excellent methods described earlier in Sections 2.1 and 2.2 for lanthanoid analysis.
In order to get an insight into the reasons for the observed
behaviors, let us consider the solid state equilibria that
should be involved in this problem. Assume we have a binary compound MX, in which M and X are divalent, and that
the dominant defects of the compound are the Shottky defects, i. e. cation and anion vacancies. A trivalent lanthanoid
ion, R3', and a trivalent anion analyte, A3-, are present as
dopants. The relevant equilibria are the following:
where the notation of Kroger- Vink has been used[45'.Notice
the similarity between these equilibria and the familiar set of
EDTA equilibria of analytical importance. The intrinsic vacancies V, and V; play the role of H + and OH- which are
formed on dissociation of the H 2 0 solvent. The (2RM.VM)X
species is analogous to the metal hydroxide complexes that
form in aqueous solution, while the (2Ax.Vx)" species is
analogous to the different protonated species of the EDTA
weak acid. The (RM .A,)" species is the analytically important complex we are interested in observing and is analogous
to the metal-EDTA chelate. It is well known that the key to
using EDTA as a chelating agent lies in controlling the pH.
Too acidic a solution results in an unfavorable conditional
formation constant causing dissociation of the chelate, while
too basic a solution precipitates the metal hydroxide. One
might expect the same ideas would apply to the defect equilibria, where the V, and V, concentrations must be controlled. These concentrations were not controlled in the examples given. Addition of analyte ion would raise the concentration of V, by an amount that depended upon the stability of the (2Ax.Vx)x pair and, as a result, the concentration of V; would decrease. The (2RM.VM)X
species would
then tend to dissociate. The addition of analyte ions has two
effects in this situation-the shift in defect equilibria and the
pairs. An
formation of the analytically interesting (RM.
analytical procedure, however, should have the relevant
equilibria fixed at optimal values by suitable buffers while
the analyte is added. It is believed that the failure to buffer
the equilibria is the cause of the non-linear working curves
and the plummeting of the intensity at low concentrations.
The situation is analogous to the non-linear behavior that
has been observed for the fluoride compensated lanthanoid
ions in CaF,; the kind of non-linearity is the same. We are
presently studying the control of these equilibria by the atmosphere above the reaction vessel and by addition of other
aliovalent ions.
This picture also enables a description of the different behavior observed in materials after addition of the analyte ion
A3-. If the equilibrium constant for the formation of
750
(RM.Ax)' is much larger than the constants of the other
equilibria, the addition of A3- will lead to formation of the
species (RM .A,)" and force dissociation of the (2 RM.VM)
species. A behavior such as that shown in Figure 22 would
then be observed. If the formation constants for both the species (RM.Ax)" and (2RM.VM)X
are large, addition of A3will cause new lines to form without destroying the lines
from the intrinsic sites. A behavior such as that shown in
Figures 19 and 20 might then be expected. If the formation
constants of the (RM .A,)" and (2 Ax. V,)" species are small,
the concentration of VG will be raised as A3- is added to the
system; this will lower the V, concentration and cause dissociation of (2RM.V#. In this case the system might behave similarly as shown in Figure 21.
The picture presented above is quite speculative and is not
meant as a definitive explanation of experimental observations. It is certainly oversimplified and omits many important factors such as the interaction with other defects or impurities, the equilibria with atmospheric species and the
reaction vessel, the distribution coefficients into the various
phases, and the possibility of having several types of defects.
It is certainly inadequate for explaining the complex behavior of the BaS04 system shown in Figure 17. Instead, the picture is meant to focus on the key parts of this method of analysis to provide a framework for organizing previous observations and guiding the ideas of future work. It is a tentative
picture at best and can be expected to change appreciably as
further work is done in this area.
3.4. Analysis with Pure Lanthanoid Compounds
There are other methods of causing an analyte ion to associate with a fluorescent probe ion. One of the simplest is to
form a pure, fluorescent lanthanoid compound in the presence of an analyte ion. Anything that enters the lanthanoid
lattice will have to perturb many lanthanoid ions and give
rise to new lanthanoid sites which can be identified spectroscopically. Again, selectivity is the key to a successful technique. The site associated with an analyte ion must be measured in the presence of a much larger population of intrinsic
sites. There is a severe limitation in the choice of lanthanoid
as lattice cation, since rapid exciton migration to quenching
ions and energy transfer processes prevent fluorescence from
most pure
Generally, fluorescence can be obtained from Eu3+, Tb3+, and Yb" compounds (and occasionally from Gd3+ compounds) whose large energy gaps
prevent efficient energy transfer.
Fig. 23. Bottom: 7Fa-.5Do excitation spectrum of a Eu2(S0& sample formed
mol/l PO:-. Top: Fluorescence spectra
from a solution containing 5 x
that result if the dye laser is tuned to either of the marked lines (5D,-7F2 transition).
Angew. Chem. In[. Ed. Engl. 18, 738-752 (1979)
E u ~ ( S O(:~PO:-)
)~
was prepared by rapidly evaporating a
solution of E U * ( S O ~in) ~the presence of PO:-. The 7F0-+
5Doexcitation spectrum of this material is shown in Figure
23 (below). It contains one very intense and sharp line and a
band at higher wavelength whose intensity varies from sample to sample. Two additional weak lines appear at short
wavelengths with intensities which are proportional to the
PO:- concentration. Although it would appear difficult to
measure these lines at much smaller concentrations because
of the very large intensity of the neighboring intrinsic peak,
it is possible to selectively excite these lines and obtain fluorescence spectra that are free of fluorescence from the main
intrinsic site. This is illustrated in Figure 23 (above left) for
the weak line at 578.5 nm; also illustrated (above, right) is
the spectrum obtained on excitation of the intrinsic line at
578.1 nm. The selective excitation allows one to follow the
fluorescence of this PO:- site down to 0.3 ppm PO:-, at
which point it is obscured by a background fluorescence.
A difficulty that was encountered in this work was the inability to selectively excite the PO:- site using other excited
levels. For example, if any of the lines occurring in the fluorescence shown in Figure 23 were monitored as an excitation
spectrum of the 7Fo-+5D2transition was obtained, all of the
excitation spectra would be identical. Excitation anywhere in
the 5Dz-multiplet results in exactly the same fluorescence
spectrum. This behaviour suggests that rapid energy transfer
can occur from either the 'D2 or 5D, excited states to the
Eu3+ site which has the lowest 'Do energy, where it becomes
trapped and must fluoresce. The lower part of Figure 23
shows that the PO:- sites have 5Do states whose energy is
higher than that of the other intrinsic sites (since they appear
at shorter wavelength). Therefore they can transfer energy to
the intrinsic sites. If the 5Dostate is excited directly, this energy transfer between sites does not occur, probably because
the density of phonon states available at these small energy
differences is too small to allow efficient non-resonant energy transfer. This explanation immediately suggests that if the
introduction of an analyte ion produces a Eu3+ site with a
lower 5Dostate than the intrinsic sites, energy transfer to this
site should be very efficient and would greatly enhance the
fluorescence intensity for such a site. Such a system might be
of great analytical interest.
3.5. Analysis of Complexing Agents
Another method of achieving an association between a
lanthanoid probe ion and an analyte ion is to form com-
81 3
61.
616
818
817
1 InmlFig. 24. 5Do-7F2 fluorescence spectrum of a) a Eu(NO,), solution and b) of a
EuCI, solution.
Angew. Chem. Inf. Ed. Engl. 18, 738-752 (1979)
plexes of the two ions in aqueous solutions. One is again restricted in the choice of lanthanoid ions that can be used for
such experiments since the high vibrational energies of water
strongly favor multiphonon processes. Only the ions with
very large energy gaps, such as Eu3+,Tb3+,and Gd3+,can
fluoresce.
Figure 24a shows the 5Do+ 'F2 fluorescence spectrum of a
EuC13 solution while Figure 24b shows the 5D0+'F2 fluorescence spectrum of a E u ( N O ~ solution.
)~
Although the lines
are quite broad, the Eu(NO,)~fluorescence can be distinguished from the fluorescence of the uncomplexed Eu3+.
Figure 25 shows the 'Fo+'D2 excitation spectra of uncomplexed and complexed Eu3'. Again the lines are broad but
separated sufficiently enough to distinguish between complexed and uncomplexed Eu3+.Thus despite the large line-
b'
f
Fig. 25. 7Fo-5D2 excitation spectrum of a) a EuCh solution and b) a Eu(N03),
solution.
widths of Eu3+ in solution, it is still possible to obtain good
selectivity for NOT because of the large separations of its
lines in both the excitation and fluorescence spectra.
4. Conclusions
In this paper, we have shown how the laser excitation
techniques provide a way of obtaining direct information
about the microscopic details of the clustering processes in
fluorite materials. EPR methods have failed in this respectfor reasons which are still unknown. The many different
bulk measurements that have been made require theoretical
modeling in order to glean microscopic information, and
they lack the selectivity required to study the many different
kinds of sites. The selective excitation of lanthanoid probe
ions offers a new approach to the problem. The sensitivity
and selectivity of laser excitation methods provide a way of
examining sites in a material whose concentrations differ by
several orders of magnitude. It is expected that these advantages will result in laser and lanthanoid spectroscopy playing
a far greater role in the quest to understand the basic chemical processes which control the solid state chemistry of a material.
We have also outlined a very different method of performing chemical measurements at the trace level. The direct excitation of a lanthanoid ion has resulted in the ability to detect 25 parts in 10'' of a lanthanoid in aqueous solution with
very high selectivity for the specific lanthanoid of interest.
We have shown how non-fluorescent ions can be measured
with a similar sensitivity and selectivity by forcing the system
751
to dimerize and causing a 1:1 association between the nonfluorescent analyte ion and a fluorescent lanthanoid probe
ion. Extension of this idea to other ions of the periodic system can be accomplished by using charge compensation to
cause the 1 : l pairing between analyte ion and fluorescent
lanthanoid ion. The procedures for lanthanoid analysis work
quite nicely and we believe they presently constitute one of
the best ways of performing analytical measurements on lanthanoids. There is still much work that needs to be done before practicable methods for the real analytical determination of other ions in the periodic system are available, but the
method has great promise.
This paper has outlined two applications of the idea of using lanthanoid probe ions with selective laser excitation
methods to study complex materials. We believe there are
many applications that remain to be tried. The spectra of the
lanthanoids provide information at the atomic level about
the short range environments of a material, indeed with a
sensitivity and selectivity that are difficult to match with other techniques. These are the very properties that modern researchers need in their effort to understand fundamental
processes and to further improve the limits of detection.
This research was supported by the National Science Foundation under grants CHE74-24394 A - l and DMR77-0776.5.
Received: July 31, 1978 [A 289 IE]
German version: Angew. Chem. 91, 765 (1979)
[I] J. Becquerel, Z . Phys. 58, 205 (1929).
(21 G. H. Dieke: Spectra and Energy Levels of Rare Earth Ions In Crystals. Interscience, New York 1968.
[3] B. G. Wybourne: Spectroscopic Properties of Rare Earths. Interscience, New
York 1965.
[4] R. B. Cheder, J. E. Geusic in F. T. Arecchi, E. 0. Schulr-Dubois: Laser
Handbook 1. North-Holland, Amsterdam 1972, p. 325.
[5] L. A. Riseberg, M . J. Weber, Prog. Opt. 14, 90 (1976).
[6] B. DiBartolo: Optical Interactions in Solids. Wiley. New York 1968.
[7] C. W. Rector, B. C. Pandey, H. W. Moos, J . Chem. Phys. 45, 171 (1966).
[8] R. H. Heist, I? K. Fang. Phys. Rev. 8 1 , 2970 (1970).
[9] R. E. Bradbury, E. Y. Wong, Phys. Rev. 8 4 , 690, 694, 702 (1971).
(lo] Yu. K. Voron’ko, A. A. Kaminskii, V. K Osiko, Sov. Phys. JETP 23, 10
(1966).
[I I] A. A. Kaplyanskii, S. A. Basoon, V. A. Rachin, R. A . Titou, Pisma Zh. Eksp.
Teor. Fiz. 21, 438 (1975).
(121 A. A. Kaplyanskii, S. A. Basoon, V. A. Richin, R. A . Titov, Fiz. Tverd. Tela
17, 3661 (1975).
[13] K. F. Renk, J. Deisenhofer, Phys. Rev. Lett. 26, 764 (1971).
(141 W. E. Bran, W Grill, Phys. Rev. B16, 5303, 5315 (1977).
[IS] R. S. Meltzer, J. E Rives, Phys. Rev. B15, 2442 (1977).
[16] P. M. Selzer, W. M. Yen, Phys. Rev. Lett. 38, 858 (1977).
1171 Yu. K. Voron’ko, A. A. Kaminskii, V. V. Osiko, Sov. Phys. JETP 22, 501
(1966).
118) Yu. K. Voron’ko, A. A . Kaminskii, V. Y. Osiko, A. M. Prokhorov, JETP Lett.
1 , 120 (1965).
(191 D. R. Ta/lant, J. C. Wright: Proceedings of the 11th Rare Earth Research
Conference 1974.
(201 D. R. Taflant, J. C. Wright, J. Chern. Phys. 63, 2074 (1975).
1211 L. Dubicki, J. Ferguson, G. A. Osborne, I. Trabjerg: Spectroscopie des Elkrnents de Transition et des Elements Lourds dans les Solides. No. 255.
CRNS, Paris 1977.
152
(22) W Yan Goal: Principles o f Defect Chemistry of Crystalline Solids. Academic Press, New York 1966.
[23] W. Hayes: Crystals with the Fluorite Structure. Clarendon Press, Oxford
1974.
[24] N. B. Hannay: Treatise on Solid State Chemistry. Plenum Press, New York
1975.
[25] C. N . R. Rao: Solid State Chemistry. Marcel Dekker. New York 1974.
126) J. B. Goodenough. M. S. Whittingham: Solid State Chemistry of Energy
Conversion and Storage. American Chemical Society, Washington 1977.
(271 A. D. Franklin, J . Nonmetals I, 27 (1972).
[28] A. K. Cheetham, B. E. F. Fender, D. Steele, R. I. Taylor, B. T. M. Willis, solid State Commun. X. 171 (1970).
[29] A. K. Cheetham, B. E. F. Fender, M. J. Cooper, J. Phys. C4, 3107 (1971).
[30] D. Steele, P. E. Chirds, B. E. F. Fender, J. Phys. CS, 2677 (1972).
1311 J. Makousky, Phys. Rev. Lett. 15, 953 (1965).
(321 I. 8 . Airenberg, M. S. Orlov, A. L. Stolov, Sov. Phys. Solid State 15, 1240
(1973).
[33] P. P. Yaney, D. M. Schaeffer, J. L. WOK Phys. Rev. B11. 2460 (1975).
1341 N. E. Kask, L. S. Kornienko, M. Fakir, Sov. Phys. Solid State 6. 430
(1964).
(351 Kh. S. Bagdasarov, Yu. K. Voron’ko, A. A. Kaminskii, L. K Krotooa, Y Y
Osiko, Phys. Status Solidi 12, 905 (1965).
[36] Yu. K. Voron’ko, A. A. Kaminskii, V. V. Osiko, Sov. Phys. JETP 22, 295
(1966).
(371 J. Fontanella, C. Andeen, J. Phys. C9, 1055 (1976).
I381 A. Edgar, H. K. Welsh. J . Phys. C 8 , L336 (1975).
[39] J. H. Chen, M. S. Donough, Phys. Rev. 185, 453 (1969).
(401 C. Burefy, I? Gonzales-Jimenez, P. t’mbert, 15: Varret, J. Phys. Chem. Solids
36, 683 (1975).
[411 M. R Mustafa, W E. Jones, B. R. McCarvey, M. Greenblatt. E. Banks, J.
Chem. Phys. 62, 2700 (1975).
(421 E. L. Kim, M. Ikeya, J. H. Crawford, Phys. Rev. B8, 5840 (1973).
1431 A. Edgar, C. A. Freeth, G. D. Jones, Phys. Rev. B15, 5023 (1977).
[44]C. Andeen, D. Lank, J. Fontanella, Phys. Rev. B 16, 3762 (1 977).
1451 Throughout the paper, we will use the notation of Kroger-Yink 1221. The
subscript indicates the lattice site, the symbol represents the type of occupancy of that site, and the superscript indicates the effective charge relative
to the lattice where ’ is negative, * is positive, and x is neutral.
(461 F. K. Fang, Prog. Solid State Chem. 3, 135 (1966).
(471 R. H . Heist, F. K. Fang, Phys. Rev. B1, 2970 (1970).
I481 M. J. Weber, R. W. Bierig, Phys. Rev. A 134, 1492 (1964).
I491 U. Ranon, J. S. Hyde, Phys. Rev, 141, 259 (1966).
[50] E. Secemski, W. Low,J . Chem. Phys. 64, 4240 (1976).
I511 J. M . O’Hare, T P. Graham, G T. Johnston, J . Chem. Phys. 64, 4242
(1976).
[52] F. K. Fang, J. Chem. Phys. 64, 4143 (1976).
I531 A. D. Franklin, S. Marrullo, Proc. Brit. Ceram. Soc. 19, 135 (1971).
[54] A. D. Franklin, U.S.N.T.I.S. AD Rep. 748259 (1971).
1551 J. M . O’Hare, J . Chem. Phys. 57, 3838 (1972).
1561 C. K. Miner, T. P. Graham, G, T. Johnston, J . Chem. Phys. 57, 1263
(1972).
[57] M. R. Brown, K. G. Roots, J. M. Williams, W S. Shand, C. Grater, H. F.
Kay, J. Chem. Phys. SO. 891 (1969).
[58] A. D. Franklin, J. Chem. Phys. 64, 1509 (1976).
1591 C. R. A. Catlow, J. Phys. C9. 1859 (1976).
[60] D. R. Tallant, M. P. Miller, J. C. Wright. J. Chem. Phys. 65, 510 (1976).
(611 D. R. Tallant, D. S. Moore, J. C. Wright, J. Chem. Phys. 67, 2897 (1977).
1621 J. C. Wright: Spectroscopie des Elements de Transition et des Elements
Lourds dans les Solides No. 255. CRNS, Paris 1977.
I631 M. D. Kurz, J. C. Wright, J. Lumin. 15. 169 (1977).
[64] M. P. Miller, J. C. Wright, J. Chem. Phys. 68, 1548 (1978).
(651 J. B. Fenn, J. C. Wright, F. K. Fong, J. Chem. Phys. 59,5591 (1973).
[66] J. C. Wright: Radiationless Processes in Molecules and Crystals. Vol. 15.
Springer, Berlin 1976.
(671 F. J. Gustafson, J. C. Wright, Anal. Chem. 49, 1680 (1977).
[68] G. M . Zuereu, A. I. Smirnou, Sov. Phys. Solid State 9, 1586 (1968).
[69] A. F. Lrung, J . Phys. C 6, 2234 (1973).
[70] C. Yang. S. Lee, A. J. Bevelo, Phys. Rev. B12, 4687 (1975).
[71] T. R.r. Reddy, E. R. Dauies, J. M. Baker, D. N. Chambers, R. C. Neumann.
B. Uzbay, Phys. Lett. A36, 231 (1971).
(721 J. M. Baker, E. R. Dauies, T. Rs. Reddy, Contemp. Phys. 13, 45 (1967).
(73) J. C. Wright, Anal. Chem. 49, 1690 (1977).
Angew. Chem. Int. Ed. Engl. 18, 738-752 (1979)
Документ
Категория
Без категории
Просмотров
1
Размер файла
1 573 Кб
Теги
analytical, inorganic, solis, application, ultra, method, state, new, chemistry, sepic, fluorite, trace, analysis, defects
1/--страниц
Пожаловаться на содержимое документа