Subscriber access provided by LIBRARY OF CHINESE ACAD SCI
Chemical Reviews is published by the American Chemical Society. 1155 Sixteenth
Street N.W., Washington, DC 20036
Upconversion and Anti-Stokes Processes with f and d Ions in Solids
Chem. Rev., 2004, 104 (1), 139-174• DOI: 10.1021/cr020357g • Publication Date (Web): 18 November 2003
Downloaded from http://pubs.acs.org on February 24, 2009
More About This Article
Additional resources and features associated with this article are available within the HTML version:
Links to the 29 articles that cite this article, as of the time of this article download
Access to high resolution figures
Links to articles and content related to this article
Copyright permission to reproduce figures and/or text from this article
Upconversion and Anti-Stokes Processes with f and d Ions in Solids
Franc ¸ois Auzel
GOTR, UMR 7574-CNRS, 1, Place A-Briand, 92195 Meudon Cedex, France
Received February 25, 2003
1. Introduction and Historical Background
2. Energy Transfers between RE Ions: Role of
Energy Diffusion in Up- and Downconversion
2.1. Recall of Basics of Energy Transfer with
Activator in Its Ground State
2.2. Upconversion Processes by Sequential
Energy Transfers (APTE or ETU Process):
Com parison with ESA and Typical Exam ples
3. Upconversion in a Single-Ion Level Description
for APTE (ETU) or ESA and in a Pair-Level One
(Cooperative Effects): Theoretical and
Experim ental Discrim ination
3.1. Three Different Kinds of Pair States
3.2. Fundam ental Difference for Transitions
between Single-Ion States, Dynam ical and
Static Pair States, and Cooperative Pair
3.3. Application of Cooperative Lum inescence;
Theory, and Exam ples
4. Experim ental Results and Their Im plications in
4.1. Recent Upconversion Studies in Lanthanide
(4f) and Actinide (5f) Ion-Doped Solids with
APTE (ETU) and ESA Processes
4.1.1. Pr3+(4f2) Ion
4.1.2. Nd3+(4f3) Ion
4.1.3. Gd3+(4f7) Ion
4.1.4. Dy3+(4f9) Ion
4.1.5. Ho3+(4f10) Ion
4.1.6. Er3+(4f11) Ion
4.1.7. Tm3+(4f12) Ion
4.1.8. Tm2+(4f13) Ion
4.1.9. U4+(5f2) Ion
4.1.10. U3+(5f5) Ion
4.2. Recent Upconversion Studies in
Transition-Metal (3d, 4d, 5d) Ion-Doped
Solids with APTE (ETU), ESA, or
4.2.1. Ti2+(3d2) Ion
4.2.2. Cr3+(3d3) Ion
4.2.3. Ni2+(3d8) and Mn2+(3d5) Ions
4.2.4. Mo3+(4d3) Ion
4.2.5. Re4+(5d3) Ion
4.2.6. Os4+(5d4) Ion
4.3. APTE (ETU) for Display and IR Detection
4.4. General Negative Roles Brought up by
Undesired APTE (ETU) Effects
4.5. APTE (ETU) and ESA Pum ped Lasers
5. Cross-Relaxation and the Photon Avalanche
5.1. Avalanche Process as a Positive Feedback
5.2. Conditions in Order To Observe an
5.3. Er3+−LiYF4as an Avalanche Model
5.4. Photon Avalanche in Er3+−Fluoride Glasses
in Fiber and Bulk Shape
5.5. Avalanche in Codoped System s
5.6. Upconversion Laser with
Multiphonon-Assisted Pum ping Schem e and
6. Perspectives and Future Advances
6.1. Upconversion UV-Tunable Lasers
6.2. NewMaterials for Low-Intensity IR Im aging
6.3. Upconversion Material Intrinsic Bistability
6.4. Hot Em ission and Avalanche Like Co-Doped
6.5. Biological Applications
8. Acknowledgm ents
1. Introduction and Historical Background
Before the 1960s, all anti-Stokes emissions, which
were known to exist, involved emission energies in
excess of excitation energies by only a few kT. They
were linked to thermal population of energy states
above excitation states by such an energy amount.
It was the well-known case of anti-Stokes emission
for theso-called thermal bands or in theRaman effect
for the well-known anti-Stokes sidebands. Thermolu-
minescence, where traps are emptied by excitation
energies of the order of kT, alsoconstituted a field of
anti-Stokes emission of its own. Superexcitation, i.e.,
raising an already excited electron toan even higher
level by excited-state absorption (ESA), was also
known but with very weak emissions. These types
of well-known anti-Stokes processes have been re-
viewed in classical textbooks on luminescence.1
All fluorescence light emitters usually follow the
well-known principle of the Stokes law which simply
states that excitation photons are at a higher energy
than emitted ones or, in other words, that output
photon energy is weaker than input photon energy.
This, in a sense, is an indirect statement that
efficiency cannot be larger than 1. This principle is
139Chem . Rev. 2004, 104, 139−173
10.1021/cr020357g CCC: $48.50©2004 Am erican Chem ical Society
Published on Web 11/18/2003
valid, of course, only when one excited ion system is
In this review wewill discuss anti-Stokes emissions
or upconversion processes for which emission is found
to exceed excitation energies by 10-100 times kT,
which is violating Stokes law in its basic statement.
It will be shown that coupled lanthanide and
uranide f ions and transition-metal d ions, when
embedded in solids, may deviate rather easily from
the above principle, producing upconversion emis-
sions of the anti-Stokes types under moderate to
strong excitation density.
A number of different mechanisms have been
recognized tobeinvolved in upconversion either alone
or in combination.
Besides multistep excitation due toclassical excited-
state absorption (ESA), there is the very efficient
process of upconversion by sequential energy trans-
fers which has been named by Auzel APTE effect (for
addition de photon par transferts d’energie;2this
effect was alsolater termed ETU for energy transfer
upconversion.3This last phenomenon has to be
distinguished from a third process, namely, coopera-
tiveupconversion either between twoions or between
a pair of ions and a third one. Though some aspects
of its theoretical behavior are rather analogous with
upconversion by energy transfers, its efficiency is
usually much weaker. This is because it involves
quasi-virtual pair levels between which transitions
havetobedescribed in a higher order of perturbation
due to their double-operator nature.
A fourth process will alsobeconsidered: thephoton
avalanche effect, also based on sequential energy
transfers but of the downconversion type (usually
called cross-relaxation), whereas the upconversion
step itself is due to ESA.
The various experimental techniques, which allow
distinctions between the behaviors of these various
processes, will be analyzed taking examples from the
With the advent of high energy density laser
sources, these processes have been observed in vari-
ous types of ion-doped solids such as crystals and
glasses in bulk, fiber, or waveguide form; the recent
advances will be encompassed and described there-
The whole field of upconversion in ion-doped sys-
tems can be traced back to an idea of Bloembergen
in 1959,4proposing that infrared (IR) photons could
be detected and counted through sequential absorp-
tion (ESA) within the levels of a given ion in a solid,
that is using superexcitation as a detector. This was
a short proposal for a detector called an infrared
quantum counter (IRQC). In fact, because there was
little chance with incoherent pumping that the same
single doping ion would receive two photons in
sequenceat its given position during thefirst excited-
statelifetime, theexperimental demonstration of this
effect had towait for laser excitations and fiber local
confinement. Some of the first experiments5have
been proved later to be due to energy diffusion
through energy transfers between identical ions.6
The role of energy transfers in upconversion pro-
cesses was not recognized until 1966, when it was
suggested by Auzel that energy transfers between RE
ions could take place between twoions, both of them
being in an excited stateat theenergy transfer initial
step.7Until then, all energy transfers were assumed
to take place from a first ion in an excited state to a
second onein its ground state.8Becauseupconversion
by sequential energy transfers or APTE effect is so
efficient, it could initially be obtained through black-
body excitation or spontaneous diode emission even
before laser sources became commonly available.2
Principles and applications of such upconversion
phosphors have already been presented in several
reviews up to the 1970s by Auzel,2Mita and Naga-
zawa,9Garlick,10and Wright.3Since then, laser
excitation in the IR and/or the use of fibers have
become so easy that upconversion has become a
pervading effect in all RE-doped materials under
high-density IR excitation. Also, another type of
upconversion, namely, the photon avalanche pro-
cess,11,12has been widely investigated in recent years.
Limited aspects of recent progress have partially
been reviewed through the 1980s and 1990s,13-18but
Francois Auzel, bornJuly5, 1938inRoanne(France), graduatedas both
anengineerfromISEP (InstitutSupereurd′ElectroniquedeParis) in1961
and a “Licencie ´ -es Sciences Physique” in 1962 fromthe University of
Paris. Hewas withFrance-TelecomResearchCenter(CNET) from1961
to 1999. There, in 1961, he started working on Nd-doped phosphate
glasses. In 1968, he received his Ph.D. degree fromthe University of
Otto Deutschbein and with Professor Alfred Kastler as adviser; as an
outcom e of this work, he proposed rare-earth-doped fluorophosphate
glasses as laser m aterials withweak OH interactions. During his thesis
(ETU) giving risetoupconversionofinfraredtovisiblelight, using Yb−Er
(greenem ission)andYb−Tm(blueem ission)couples.In1973, hereceived
the Foucault prize from the “Societe ´ Franc ¸aise de Physique" for his
discovery of upconversion processes by energy transfers (APTE ef-
fect)(1965−1966). In1989, the FrenchAcadem y ofScience grantedhim
dem onstration of the existence of Stokes and anti-Stokes m ultiphonon
sidebands fortrivalentlanthanides (1976), theproposalofascalarcrystal
fieldparam eterproportional tothe m axim umsplitting ofa J term(1979),
the first observation of superradiance em ission of a lanthanide (Er ions
at 2.7 µm at 10 K in weakly doped YLF crystals) (1986), the
room -tem perature avalanche effect of Er-doped glasses and crystals
(1993), andthesaturationeffectofm ultiphonondecays inglasses (1996).
He has been a part-tim e Professor at Ecole Centrale des Arts et
Manufactures (1971−99) andat Orsay University (DEA Ecole Polytech-
nique-Lab. Aim eCotton) (1990−99). Hecreatedandheadedthe“Groupe
d′OptiquedesTerresRares”, ateambelongingtobothCNRS andCNET,
until his retirem ent from CNET in 1999. He is currently a voluntary
researcher at CNRS and a consultant for active optical m aterials.
140 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
the general field has recently evolved from the rare-
earth (4f) consideration toward the use of actinide
(5f) and transition-metal (3d, 4d, 5d) ions with a
systematic use of laser excitation at precisely defined
This evolution justifies the present review.
Because it appears that the language in the up-
conversion field is still not completely fixed, possibly
inducing misinterpretation, the basic processes of
energy transfers, cooperative processes, and their
application toupconversion together with their more
recent evolutions and selected examples of applica-
tions will be presented in reference to the accepted
vocabulary proposed by the pioneers. Some of the
original papers in this field were reprinted in 1998
in a collective edition.19
2. Energy Transfers between RE Ions: Role of
Energy Diffusion in Up- and Downconversion
In the following, the mutual interactions between
ions are the key feature.
When the concentration of active ions is increased,
long before the appearance of new lines due to pairs
or modifications in radiative transition probabilities,
a migration of energy between the centers is found.
We are going to study this now, assuming that
multiphonon decays and the radiative transitions
remain one-center processes.
As single f and d ions properties are supposed to
be known, multiion processes, namely, energy trans-
fers, are now dealt with. Energy transfer occurs in a
system where absorption and emission do not take
place within the same center. It may occur without
any charge transport. Then one may distinguish
between radiative and nonradiative, resonant, and
phonon-assisted energy transfer. Theoretical ap-
proaches start from a microscopic point of view with
a macroscopic result averaged over all the centers in
the sample. In fact, an energy transfer between two
given ions cannot by itself increase efficiency; it can
only provide a new excitation wavelength range with
a reduced efficiency since it consists of the product
of twoprocesses with intrinsic efficiency less than or
equal to1. Overall efficiency improvement by energy
transfers is gained only from the spatial averaging
due to the macroscopic process of diffusion.
2.1. Recall of Basics of Energy Transfer with
Activator in Its Ground State20,21
In a schematic way, the different microscopic
energy transfer processes between two ions can be
presented as in Figure 1. Following the traditional
vocabulary of the phosphor field, the ion being first
directly excited is called a senzitizer (S); some people
would call it a donor, but because f and d ions may
alsobeimbedded in semiconductors, such vocabulary
leads to confusion and is not retained here. The ion
to which energy is transferred and which emits the
output photon is called an activator; in a synonymous
manner, it is some times termed an acceptor. To
avoid any ambiguity with the semiconductor field,
this vocabulary is not retained in the following.
One usually distinguishes radiative transfer (Fig-
ure1a), nonradiativeenergy transfer (Figure1b), and
multiphonon-assisted energy transfer (Figure 1c). S
and A may also be identical ions, and nonradiative
transfer may give rise to self-quenching by cross-
relaxation (Figure 1d).
When energy transfer is radiative (Figure 1a), real
photons are emitted by the sensitizer ions (S) and
are then absorbed by any activator ions (A) within a
photon travel distance. As a consequence, such
transfer depends on the shape of the sample.
Moreover, according to the degree of overlap be-
tween theemission spectrum of thesensitizer (S) and
the absorption spectrum of the activator (A), the
structure of the emission spectrum of the sensitizer
will change with activator concentration. Since pho-
tons are emitted anyway, the sensitizer lifetime is
independent of the activator concentration. These
three facts are the criteria used to distinguish
between radiative and nonradiative resonant energy
Probability for such transfer between two ions at
a sufficiently large distance R is found to be20
where τSis the sensitizer lifetime and σAthe absorp-
tion-integrated cross section. The integral represents
the spectral overlap between A and S. It should be
noted that thedistancedependencegoes as R-2. Such
resonant radiative transfer may permit long-range
energy diffusion between identical ions and gives rise
to photon-trapping effects of the same type as the
ones observed a long time ago in gases.22Trapping
effects increase the apparent experimental lifetime,
and τShas to be measured on thin and lightly doped
samples. These effects are particularly strong in Cr3+
Let us consider the simple case of two ions, each
with one excitable electronic state separated from its
electronic ground state by nearly equal energy; it is
the case described in Figure 1b. With suitable
F igure 1. Various basic energy transfer processes between
two ions considered before 1966: note that activator ion
(A) receiving the energy from the sensitizer (S) is initially
in its ground state. Cross-relaxation is the special case
where S is identical to A. Doubled arrows symbolize the
Coulombic interaction: (a) radiative resonant transfer; (b)
resonant nonradiative transfer; (c) phonon-assisted non-
radiative transfer; (d) cross-relaxation special case of
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 141
interaction between thetwoelectronic systems, which
is the case for nonradiative energy transfer, the
excitation will jump from one ion tothe other before
one is able to emit a quantum of fluorescence. The
mutual interactions are Coulomb interactions of the
van der Waals type between the two ions. Fo ¨rster,26
whofirst treated such a casetheoretically by quantum-
mechanical theory, considered the dipole-dipole
interaction. He assumed that the interaction is
strongest if for both transitions electric-dipole tran-
sitions are allowed.26The interaction energy is then
proportional to the inverse of the third power of the
interionic distance and the transfer probability is
HSA) electric dipole-dipoleinteraction Hamiltonian,
proportional tothe inverse third power of ion separa-
FE ) density of states provided by the vibrational
motion contributing to the line broadening of the
pSAis proportional tothe inverse sixth power of the
ion separation. The wave functions to be considered
for the matrix element describe an initial state of the
system with the sensitizer in its excited state and
the activator in its ground state, the final state
having the sensitizer in its ground state and the
activator in its excited state.
Therefore, the transfer probability can be written
whereτSis theactual lifetimeof thesensitizer excited
state, including multiphonon radiative decay, and R0
is the critical transfer distance for which excitation
transfer and spontaneous deactivation of the sensi-
tizer have equal probability.
However, Dexter pointed out27that this theory
should be extended to include higher multipole and
exchange interactions. In fact, for an isolated atom,
one can consider the transition probability as de-
creasing as (a0/λ)2n, where a0 is the Bohr radius, λ
the wavelength, and n an integer. However, in an
energy transfer process with a dependence on near-
zoneinteractions, thetransition probabilities drop off
as (a0/F)2n, whereF is theseparation of theinteracting
ions. F can be as much as 3 orders of magnitude
smaller than λ, so that the energy transfer effect
tends tobe more pronounced in systems with forbid-
den transitions.27This holds true for ions for which
transitions to first order are forbidden, such as
transition-metal and lanthanide ions.
The energy transfer probability for electric multi-
polar interactions can be more generally written as27
where s is a positive integer taking the following
s ) 6 for dipole-dipole interactions,
s ) 8 for dipole-quadrupole interactions,
s ) 10 for quadrupole-quadrupole interactions.
It should be noted that for dipole-dipole interac-
tions, the difference between radiative and nonra-
diative resonant transfer lies essentially in the fact
that for radiative transfer there is no critical R0
depending only upon concentration. The variation
goes as R-2instead of R-6, and the sensitizer lifetime
does not depend on the distance R.
Now, tobe able tocalculate effectively pSA(R), eq 4
is not very useful because R0 cannot be easily
obtained theoretically. Applying Racah’s tensorial
methods at thebeginning of thecalculation of Dexter,
eq 2, allows development of calculations analogous
to J udd’s theory for radiative transitions. The case
of themultipolar interactions was treated in this way
by Kushida28and extended by Pouradier and Auzel29
tomagnetostatic and exchange interactions, showing
that a single general formula could be used for all
types of energy transfers.
The general form obtained is then
where gS*(gA0) is the degeneracy of the S*(A0) level,
γS(E)(ΓA(E)) is the normalized line shape function of
emission (absorption) spectrum, U(l)are the tensorial
operators already seen for J udd’s theory. |Cl1l2|2can
be considered as a parameter analogous to J udd Tλ
(Ωλ) for oscillator strength.
This expression of the transfer probability has the
(1) Radial and orbital parts have been separated.
(2) Only a few reduced matrix elements need be
calculated. They are the same for the three interac-
tions we consider (for any interaction leaving spins
(3) Comparison between twodifferent interactions
can be made through comparison of Cl1l2coefficients.
They are independent of the states involved in the
transfer, and we call them El1l2, Ml1l2, and Xl1l2for
electrostatic, magnetostatic, and exchange interac-
(4) Forced electric-dipoletransitions, as calculated
by J udd’s method, can be included in eq 5.
(5) This expression also gives a single mathemati-
cal form regardless of the interaction, which is a
convenient result. The somewhat complicated ex-
pressions for the different Cl1l2of 4f electrons are
given in ref 29. However. we can note the following.
(a) For electrostatic interaction El1l2, the l1) 1 and
l2) 1 term, corresponding to dipolar-dipolar inter-
action, is zero in first order, which makes the
introduction of J udd’s Tλparameters necessary. The
El1l2values are typically between E22≈ 30 cm-1for
I )∫γS(E)ΓA(E) dE (6)
142 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
quadrupole-quadrupole intensities and E66 ≈ 3 ×
10-1cm-1, but all contain some dipole-dipole part
due to the Tλ.
(b) For magnetostatic interactions (Ml1l2), only
terms with li ) 1, 3, and 5 are nonzero. They have
the order of magnitude M11≈ 1 cm-1and M55≈ 2 ×
(c) For exchange interactions (Xl1l2), we have 1 e l1
e 6, giving estimates of 1-10-1cm-1for the coef-
Theseresults show that exchangeor magnetostatic
interactions can be found in cases of small dipole-
dipoleand quadrupoleelectrostatic interactions if the
matrix elements allow them.
If now we consider two ions with excited states of
different energies (Figure 1c), the probability for
energy transfer should drop tozerowhen the overlap
integral ∫gS(ν)gA(ν) dν vanishes. However, it has been
experimentally found that energy transfer can take
place without phonon-broadened electronic overlap
provided that the overall energy conservation is
maintained by production or annihilation of phonons
with energies approaching kΘd, where Θd is the
Debye temperature of the host matrix.30Then for
small energy mismatches (100 cm-1), energy transfer
assisted by one or two phonons can take place.31
However, for energy transfers between rare earths,
energy mismatches as high as several thousand
reciprocal centimeters are encountered. This is much
higher than the Debye cutoff frequency found in
normally encountered hosts, so multiphonon phe-
nomena have to be considered here.
Miyakawa and Dexter32showed that it is still
legitimate towrite the probability of energy transfer
in the form of eq 2, where F(E) is taken as SSA, the
overlap of the line shape functions for emission by
ion S and absorption by ion A, including the phonon
sidebands in the line shape. It is necessary to
consider each partial overlap between the m-phonon
emission line shape of ion S and the n-phonon
absorption line shape of ion A. A physical meaning
tothis mathematical assumption, criticized in ref 31,
has been given by Auzel’s experimental demonstra-
tion33of the existence of multiphonon sidebands for
trivalent rare-earth ions. Their existence could be
revealed by laser excitation spectroscopy even though
they had not been seen by usual absorption spectros-
copy because of their very small electron-phonon
Along the same lines as for vibronic sideband
studies, SSA can be expressed as follows
where S0Sand S0Aare the respective lattice coupling
constants for the ions S and A, N is the order of the
multiphonon process with N ) ∆E/pωm, ∆E is the
energy mismatch between both ions, and pωmis the
phonon cutoff frequency. σSA (0,0;E) is the zero-
phonon overlap integral between S and A. Equation
7 contains a Pekar function of the Poisson type.20
The expression for SSA with an energy mismatch
of ∆E for small S0constants and for an occupation
number n j ) (exp(pω/kT) - 1)-1, not exceeding 1 at
theoperating temperature, can beapproximated with
Stirling’s formula by
where SSA(0) is the zero-phonon overlap between S
and A in the case where there is noenergy mismatch
between the two ions. ? is given by
? ) (pω)-1log N/S0(n j + 1) - log(1 +SOA
= RS- γ ≈ RS- log 2
involving RSthe nonradiative decay parameters and
assuming identical electron-phonon coupling for ions
A and S. This exponential dependence on energy
mismatch is well substantiated by experiments.34
Up to this point we have been dealing with the
microscopic case of two ions interacting with one
another. To discuss the case of real macroscopic
samples with many ions and to obtain a link with
experimental facts, a statistical analysis of theenergy
transfer is necessary.
We have then tothink about the overlap integrals
that arisein all transfers between twoions as already
seen. In the microscopic case we are sure that the
involved line shapes can be only due to some homo-
geneous broadening even for transfer between two
identical ions in different lattices sites.
In the macroscopic case, we can measure absorp-
tion and emission spectra taking into account all
broadening processes averaged over the whole sample;
for instance, the inhomogeneous broadening process
duetoemission and absorption at centers in different
lattice sites. Then the overlap integral measured
experimentally from the usual spectra is a measure
in excess of the real overlap since we take into
account emission and absorption of centers at any
distances, even those which cannot interact. The
error is the largest for the processes occurring at
shortest interacting distances (exchange) and a con-
trario is certainly negligible for radiative transfer,
since photons can travel a much larger distance than
the spread of the spatial disorder. The error is also
smaller for systems with small inhomogeneous broad-
ening and having centers in only one type of lattice
site, that is, without disorder.
Fluorescenceline-narrowing techniques (FLN) could
give some idea about the homogeneous part of an
emission line, but the statistical analysis for the
whole sample should still be performed. Supposing
only a sensitizer-activator interaction, an averaged
transfer efficiency can be calculated.27This has been
studied in some detail by Inokuti and Hirayama.35
They considered the number of activators located at
random in a spherearound a sensitizer in such a way
that the activator concentration is constant when the
volumeof thesphereand thenumber of activator ions
considered goes to infinity. Then the averaged prob-
e-(SOS+ SOA)(SOS+ SOA)N
SSA(∆E) ) SSA(0)e-?∆E
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 143
ability for transfer from onesensitizer toany acceptor
Introducing eq 1 intothe expression for the intensity
emitted by all sensitizers, each with different activa-
tor neighborhood, they obtained thefollowing relation
for the intensity decay of the emission of the sensi-
tizer surrounded by many activators
- Γ(1 -3
where τSO is the decay constant of the sensitizer in
the absence of activators; C is the activator concen-
tration; C0is the critical activator concentration, and
s is the parameter of the multipolar interaction. The
comparison between experimental decay and this
theoretical expression has been widely used to de-
termine the index of the multipolar interaction
involved. However, because it is difficult to avoid
diffusion between sensitizers, fits of experimental
results using eq 11 have tobe taken with great care.
For example, values of s larger than 10 have been
found and it has been shown that for large s values
the multipolar result has the same limit as the
exponential behavior of an exchange process.24Yet,
onecannot infer, as is sometimes done, that exchange
coupling36is more likely than multipolar coupling.
In fact, eq 11 is valid only at the microscopic level
when thereis neither sensitizer-to-sensitizer transfer
nor activator-to-activator transfer. This formulation,
therefore, has tobe modified for high concentrations
of sensitizers and activators. Then, due tothe perfect
resonance conditions in such cases, rapid energy
migration between sensitizers or between activators
is possible. The general result is complicated,37but
Weber has shown that for large t, I(t) decays expo-
I(t) ) exp(-t
Then, two cases can be distinguished.
(i) In one case, spontaneous decay of excited
sensitizers, diffusion among sensitizers, and energy
transfer between sensitizers and activators are of
about the same order of magnitude.
For sufficiently long times and dipole-dipole in-
teractions one has38
where V ) 8πC1/4CSS3/4, NS is the sensitizer concen-
tration, NA is the activator concentration, C is the
sensitizer-activator energy transfer constant, such
that C ) (R0)s/R, and CSSis the sensitizer-sensitizer
(ii) For high sensitizer concentration, the diffusion
rate can be faster than spontaneous sensitizer decay
or sensitizer-activator energy transfer. The limiting
step is nolonger diffusion, and D appears tosaturate
with increased donor concentration; each activator
experiences the same excited sensitizer neighbor-
hood. R is taken as the minimum distance between
sensitizers as permitted by thelattice(R ) Rmin). One
with U being a constant depending on the type of
interaction as discussed earlier in this section through
Another approach to the macroscopic case is the
use of the well-known rate equations that deal with
the population of ions in a given state. This was used
as a phenomenological approach in studies of lasers.
The applicability of those equations in relation tothe
Inokuti and Hirayama statistics has been discussed
by Grant.39The basic result of Grant is that the
energy transfer probability is proportional to the
This result is the same as that obtained in fast-
diffusion studies (eq 14). The practical interest in
considering diffusion is that the decays are again
exponential, as when ions are not interacting. This
validates the use of rate equations.
Cross-relaxation terminology usually refers to all
types of downconversion energy transfers occurring
between identical ions. In such a case the same kind
of ion is both sensitizer and activator.
As shown in Figure 1d, cross-relaxation may give
rise to the diffusion process already considered
between sensitizers when the involved levels are
identical or self-quenching when they are different.
In the first case there is no loss of energy, whereas
in the second case there is a loss or a change in the
energy of the emitted photons.
Theoretically, thesametreatment is valid as in the
more general case of energy transfer. However, it
may be more difficult experimentally to distinguish
between sensitizers and activators. Thus, any of the
microscopic processes discussed above may happen
with a maximum overlap when an identical couple
of levels are involved. From the macroscopic point of
view, the diffusion-limited case predicts from eq 13
for NS ) NA ) N, and in the fast-diffusion case
A typical illustration of this is found for the self-
quenching behavior of Nd3+(4F3/2). In weak quenching
materials, such as La1-xNdxP5O14, self-quenching is
found to behave linearly with ion concentration,
whereas for strong quenching ones, such as YAG, a
quadratic behavior is obtained. This, respectively,
reflects the fast diffusion before the quenching step
in thefirst typeof materials and thelimited diffusion
before quenching in the second type of materials.20,40
I(t) ) exp -
144 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
2.2. Upconversion Processes by Sequential
Energy Transfers (APTE or ETU Process):
Comparison with ESA and Typical Examples
As said in theIntroduction, up to1966 all identified
energy transfers between rare-earth ions were of the
types summarized in Figure 1, that is the activator
ion receiving the energy from a nearby sensitizer (S)
was in its ground state. Then Auzel proposed to
consider cases where activators (A) were already in
an excited state7as shown in Figure 2. Because
activator ions usually have several (n) excited states
but a single ground state, one can understand why
n-photons may be summed up through this new
consideration. This becomes obvious when one real-
izes that only energy differences and not absolute
energy can be exchanged between ions.
Thereason for proposing such upgoing transfer was
to point out that energy transfers then used41to
improve the laser action of Er3+by pumping Yb3+in
a glass matrix could alsohave the detrimental effect
of increasing reabsorption.7,24The simple proof of
such an effect was to look for an upconverted green
(2F7/2-2F5/2) transition, which was effectively ob-
served.7,42Of course, the situation in Figure 2 could
repeat itself several times at the activator. This
meant that n-photon upconversion by energy transfer
was possible as demonstrated by the three-photon
upconversion of 0.97 µm into blue light (0.475 µm)
in the Yb3+-Tm3+couple.7Independently this IR to
blueupconversion was interpreted by Ovsyankin and
Feofilov43as a two-photon effect connected with two
excited Yb3+ions and a cooperative sensitization of
Tm3+initially in its ground state. This interpretation
originated from the law for output versus excitation,
which was quadratic instead of cubic as found in ref
7 and because energy transfers between excited
states were only being recognized independently at
the time.7The experimental discrepancy aroused
probably from a saturation in an intermediate step
in the APTE process.44
Recently a systematic analysis of the power law
governing the APTE (or ETU) process has been
performed by Pollnau,45generalizing by rate equa-
tions what had been discussed for the Yb-Tm
couple:2a Pnlaw can befound for an n-photon process
when WAPTE, the APTE (ETU) upconversion prob-
ability, is weak, whereas a P1law can be asymptoti-
cally obtained when WAPTE is large in front of other
processes depopulating the metastable state.
To make the terminology clearer, a schematic
comparison between theAPTE (ETU) effect and other
4S3/2 of Er3+) while pumping Yb3+
two-photon upconversion processes, namely, two-step
absorption, cooperative sensitization, cooperative
luminescence, second-harmonic generation (SHG),
and two-photon absorption excitation, is presented
in Figure 3 together with their respective typical
Since we are dealing with nonlinear processes,
usual efficiency, as defined in percent, has nomean-
ing because it depends linearly on excitation inten-
sity. Values arethen normalized for incident flux and
given in cm2/W units for a two-photon process. More
generally, for an n-photon process it should be in
A simple review of the energy schemes shows that
they differ at first sight by the resonances involved
for in- and outgoing photons: for highest efficiency,
photons have to interact with the medium a longer
time, which is practically obtained by the existence
of resonances. As shown, the APTE (ETU) effect is
the most efficient because it is closest to the full
However, reality is sometimes not so simple, and
different upconversion processes may exist simulta-
neously or their effects can be tentatively made to
reinforce each other. For instance, a combination of
two-photon absorption and cooperative absorption
has been theoretically investigated.46Also, SHG and
cooperative luminescence have been considered si-
multaneously in order toincreaseSHG by thepartial
resonance of cooperative luminescence.47,48
Let us consider now the role of macroscopic energy
diffusion in both APTE (ETU) and ESA upconversion
The probability for ESA in a two-step absorption
(W13) connecting a state E1toE3by the intermediate
state E2 is just given by the product of the prob-
abilities for each step (Figure 4)
F igure 2. APTE basic step: energy transfer toward an
ion already in an excited state. Nonradiative energy
transfer is either resonant or phonon-assisted with energy
mismatch ?0* 0.
F igure 3. Various two-photon upconversion processes
with their relative efficiency in considered materials.
F igure 4. Simplified energy level scheme and symbols
used in eqs 18-22.
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 145
Toobtain the same result by an APTE effect, we also
have to consider the product of two energy transfer
probabilities and calculate the equivalent rate for
populating E3by APTE (ETU); we get (Figure 4)
where WSAare the energy transfers probabilities for
each step and Ns*is the concentration of excited
sensitizers which is given by
Assuming all Wij have the same magnitude and all
WSAalso, as is typical for rare-earth ions, we have to
Clearly, the APTE (ETU) gain over one ion ESA
comes from the product NS2WSA2, which has to be as
large as possible. However, this simple quadratic
behavior with sensitizer concentration for a two-
photon upconversion though observed in the past2
can be questioned in some practical case as recently
shown by Mita;49for Yb-Tm:BaY2F8, a quadratic
behavior is first observed for a three-photon case and
then a linear one (probability is constant) at Yb
concentration above about 10%.
In any case, this points toan increase in sensitizer
concentration (NS) which leads tofast diffusion38and
allows the use of rate equations in such multiion
systems.39This validates a posteriori the implicite
use of the rate equation for establishing eqs 18-22.
Now the behavior is different for certain hosts where
ions are clustered into pairs even at low average
concentration. This has been shown by Pelle ´ and
Goldner50for CsCdBr3:Yb, Er for which an Yb opti-
mum concentration of 1% has been demonstrated, a
significant difference with the usual 15% for other
Besides nonradiative diffusion at the sensitizer
level, it has been observed that radiative diffusion
plays also a role in APTE (ETU) upconversion. The
measured time constants for the Er3+or Tm3+emis-
sion have been observed to be correlatively length-
ened by the photon-trapping lengthening of the Yb3+
lifetime, reflecting the radiative diffusion at the
sensitizer level.52Also, reducing singlecrystals doped
with Yb-Er or Tm to powder form usually produces
an intensity reduction when the grain size is less
than 100 µm. This shows that the effective radiative
diffusion length is of this order of magnitude in this
experiment. Recently, studies on nanometric size
upconversion phosphors with crystallites of Y203:Yb,-
Er, ranging from 75 to200 nm in grains between 600
and 800 nm have shown that the maximum intensity
for the blue and green emission under 632.8 nm
excitation is obtained for the maximum size of 800
nm.53,54This effect is alsomost probably in connection
with diffusion within Ybions which, though not being
directly excited, plays a role in an intermediate relay
step of the APTE (ETU) process.
Many times in the literature, when ESA is not
advocated, upconversion involving coupled ions is
referred to as cooperative effects or cooperative
energy transfers without proof when in fact APTE
(ETU) effects are involved as can be guessed from
their relative efficiencies and from the provided
description. The fact that the APTE (ETU) effect and
cooperative ones are often mistaken is due to a
number of common properties.2For instance, for two-
photon upconversion, both processes show quadratic
increases on excitation and on absorber concentra-
tions; both show an emission lifetime equal to one-
half the absorber lifetime. However, they show
different rise times; cooperative rise time is instan-
taneous as for ESA or any absorption, whereas APTE
(ETU) rise time reflects the population accumulation
at the sensitizer excited state. However, as shown
below, thedifferenceis morebasic, though sometimes
difficult toestablish experimentally except in special
cases wheresingle-ion resonances clearly donot exist
or where diffusion between ions is prohibited by a
too small concentration with still an interaction as
The basic distinction between both upconversion
processes (ETU or ESA) within a single-ion state
description and cooperativepair states is thepurpose
of the next section.
3. Upconversion in a Single-Ion Level Description
for APTE (ETU) or ESA and in a Pair-Level One
(Cooperative Effects): Theoretical and
Because in the field of upconversion pair states or
more recently dimer states are advocated to explain
some of the observed processes, it is felt appropriate
heretoprecisely definethevocabulary, which is done
in the next subsection.
3.1. Three Different Kinds of Pair States
In fact, when active ion concentration is increased,
besides the occurrence, at first, of changes in prob-
abilities for lower concentration as already pointed
above, its value may reach a point where clusters
may beformed and new levels may beexperimentally
observed. Wethink it is useful todistinguish between
three types of pair levels, according totheir different
origin and shift from their parent single-ion level.
When tworesonant systems are coupled, it is a basic
physical phenomenon that their degeneracy is re-
moved. This may be called a dynamic shift, and for
rare-earth ions it is typically 0.5 cm-1
maximum of a few cm-1; it is 2.7 cm-1for Nd3+in
CdF256and 3 cm-1in the stoichiometric compound
Cs2Yb2Br9with built-in pairs with a distance of 3.9
Å.57In a stoichiometric material this would give rise
55up to a
NS* ) NSW12
for single-ion ESA(21)
for two-ion APTE
146 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
to a Frenkel excitonic band, though for rare-earth
ions it has been shown to be rather limited by the
weakness of the interaction.58Now in divalent ma-
terials or even in trivalent hosts,59when the active
trivalent ion concentration is increased, one may
understand that the local static crystal field is
modified by thereplacement of a divalent cation with
a trivalent one or even by the ion size modification.
This gives rise to a spectral shift on the considered
single-ion levels when its concentration is increased.
Such a shift may reach up to 10 cm-1. Though still
often called a pair level,59this new single-ion level
is of completely different origin from theprevious one;
it is sometimes alsocalled a new site or a static pair
state due to the active ion concentration increase.59
It must be stressed here that one-center operators
just as for transitions between single-ion states
govern any transitions between such pair states.
Now, thethird typeof pair levels coming from what
are called cooperative processes are very different
from the twoprevious types because, as we are going
to see, they involve two-center operators and are
second order with respect tothe transitions between
pair levels of the two previous types. As we will see,
the shift from the single parent states is the sum of
theenergy of theparent states; it is several hundreds
or thousands of cm-1,60which does not represent the
interaction strength inside the pair. The cooperative
pair levels will be dealt with in more detail below
(sections 3.2 and 3.3).
3.2. Fundamental Difference for Transitions
between Single-Ion States, Dynamical and Static
Pair States, and Cooperative Pair States
When active ions are situated at sufficiently short
distances for interactions between them totakeplace,
two types of upconversion processes may occur:
summation of photon energy through energy trans-
fers2(APTE (ETU) effect) and/or cooperative effects
either by sensitization43or emission as found by
Nakazawa and Shionoya.61Both APTE (ETU) effects
and cooperative ones are often mistaken for the one
another becauseboth present several similarities and
may simultaneously occur in a given system for a
given excitation. In particular, both processes reflect
the n-photon order versus excitation density and
sensitizer concentration in the same manner.
As seen in the Introduction, upconversion by
energy transfer is a generalization of Dexter’s energy
transfer27to the case where the activator is in a
metastable state instead of being in its ground state;
this requires the interaction between S and A (HSA)
to be smaller than the vibronic interaction of S and
A, so that both ions can be described by single-ion
levels coupled to the lattice. It is generally the case
since for fully concentrated rare-earth crystals or for
clusters, pair level splitting is of the order of 0.5
cm-1;55,62in host with smaller concentrations, this
interaction can be even weaker, whereas one-phonon
or multiphonon sidebands may modulate the level
positions by several hundreds of cm-1. Further,
upconversion requires thetransfer probability for the
second step (WSA) to be faster than radiative and
nonradiative decay from the metastable level, that
is WSA > τ-1with τ measured intermediate state
lifetime for ion A. WSA is obtained from
where the wave functions are simple products of
single-ion wave functions; F(E) describes the dissipa-
tive density of states due to the coupling with the
lattice. HSAis the interaction Hamiltonian, the origin
of which may be multipolar or exchange interactions
as discussed above in section 2.1.
All cooperative processes, including the simple
cooperative absorption in PrCl3 first observed by
Varsanyi and Dieke63and the cooperative Stokes
emission as observed by Van der Ziel and Van Uitert
in EuAlO3:Cr3+ 64from pair states and called exciton-
(Eu) sidebands of localized excitons(Cr) or cooperative
upconversion emission,61have to be considered as
two-operator transitions between pair levels for both
ions as a whole. A one-center dipolar electric transi-
tion would be strictly forbidden for a two-center
transition, and as a difference with energy transfers
for which plain product wave functions are used, one
needs product wave functions corrected tofirst order
to account for interaction between electrons of dif-
as given, for example, for the ground state; s′′, a′′
denote intermediate states for S and A and δs,?a
denote their corresponding energies. Then any one-
photon transition in the cooperative description
involves already four terms in the matrix element,
which cannot be reduced to eq 5 that contains only
single-ion wavefunctions and not pair wavefunctions
as in eq 24:
APTE (ETU) upconversion or ESA, even between
static or dynamic pair levels, corresponds to a lower
order of perturbation than cooperative processes,
which involve cooperative pair states; the latter have
to be considered practically only when the first type
cannot take place.
Such is the case when real single-ion levels donot
exist toallow energy transfer; it is the case for Yb3+-
Tb3+upconversion55,66,67or when the concentration
is too small to allow efficient transfer by energy
diffusion between sensitizers. Then cooperative up-
conversion is likely within clusters.59,68One may also
look for crystal structures where the pair clustering
Becausecooperativetransitions areof second order
with respect totransitions between other pair states
or single-ion states, they are very weak, about 3-4
orders of magnitude less than one-center transitions,
and consequently they can usually be observed on
small samples only in excitation and emission spec-
tra. For example, a direct absorption spectra for a
cooperative process between Yb and an allowed OH
transition has only been observed on very long
samples of about 5 cm and high Yb concentration
Ψpair) Ψ0(S)Ψ0(A) -
δs′′- 0 + ?a′′ -0Ψs′′(S)Ψa′′(A) (24)
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 147
(from 1.6 × 1022to 4.3 × 1023cm-3)60with an
intensity ratio to single-ion transition of about 10-3.
Very recently similar results have been obtained for
cooperative Yb-Yb pairs on 1-12 cm long Yb-doped
laser crystals.70The intensity was found to be 1.3 ×
10-5of the single-ion one for a 1 cm long crystal of
Y2O3:Yb(10%), that is with a concentration of 2.8 ×
1021cm-3. Such very weak ratios demonstratetherole
of the double-operator nature of the transition. The
2 orders of magnitude difference, in both ratios given
above, reflect in part the fact that Yb-Yb pairs are
forbidden-forbidden pairs whereas Yb-OH are for-
Generally, experimental discrimination between
APTE (ETU) and cooperativeprocesses is not straight-
forward apart from the trivial cases where no real
intermediate energy level exists for the APTE (ETU)
effect totake place, even from unwanted impurities.
The weak ion concentration level alone is not a good
argument to eliminate APTE (ETU) upconversion,
knowing that RE ion clusters may exist, for instance,
in glasses, even at a doping level as low as 70 ppm.71
To illustrate the experimental difference between
APTE (ETU) and cooperative upconversion, we will
discuss an exampleof excitation line-narrowing effect
in n-photon summation as a mean to distinguish
between both processes.72,73Irradiating Er3+-doped
samples with IR radiation at 1.5 µm leads tovarious
Room-temperature IR F-center laser excitation
between 1.4 and 1.6 µm of 10% Er3+-doped vitroce-
ramics and of YF3:Er leads to emission bands from
the near-IR to the UV. Such emission may be
ascribed to multiphoton excitation, respectively, of
order 1 to 5, either of the APTE (ETU) or of the
cooperative type as depicted, respectively, with en-
ergy levels of single-ion (APTE) or cooperative pair
levels (Figure 5).72,73
Successive absorptions in Figure 5a involve a
combination of several J states. APTE (ETU) effect,
because of self-matching by multiphonon processes,
involves (Figure 5b) only J ) 15/2 and 13/2 states.72
Excitation spectra in Figure 6 show a striking
behavior: each spectrum presents the same spectral
structure with clearly an increasing narrowing with
process order. The structure reproduces the Stark
structure of the4I15/2-4I13/2first excited terms as can
be obtained by a diffuse reflectance spectrum.
Thespectral narrowing can beunderstood by a rate
equation treatment wherehigher excited populations
are neglected in front of the lower ones in order to
obtain a tractable development (weak excitation
The emitted power from an n-photon summation
is then given by
with symbols of Figure 5b and P1(λ) the line shape
The obtained excitation spectra are direct proof of
the validity of the APTE (ETU) explanation, since a
cooperative effect should show the convolution of all
J states involved in the multiple absorption between
Until the 1980s, few unquestionable experimental
examples of cooperative upconversion were demon-
strated besides the Yb-Tb cooperative sensitization
F igure 5. Cooperative (a) and APTE (b) energy scheme for n-photon (n ) 1-5) upconversion in Er3+-doped hosts.
F igure 6.
upconversion in Er-doped YF3.
Excitation spectra for n-photon (n ) 1-5)
148 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
quoted above and the cooperative luminescence in
Yb3+described in detail below.61Since then, many
more cases have been described:
The cooperative luminescence in the UV (from 405
to 270 nm) comes from two (2P0)Pr3+excited ions in
PrF374and LaF3:Pr3+; the APTE (ETU) cooperative
sensitization of the1S0stateof onePr3+ion is already
in its excited state(3P0) from the energy annihilation
of two other Pr3+ions also in their2P0excited state
as for the cooperative luminescence case.75Overall,
this is a three-ion, three-photon effect which, from
477 nm blue excitation, gives an upconversion in the
UV region (400-250 nm). Figure 7 describes the
energy schemes and mechanisms for both processes.
However, for some unknown reason, the cubic law
which should bepresent for emission intensity versus
excitation is not observed for the1S0level emission74
and a quadratic law is obtained as for thecooperative
luminescence case. The cooperative effects were
clearly discriminated from other processes through
excitation spectra investigations.
Very recently, Valienteet al. obtained upconversion
from near-IR Yb3+excitation with visible emission
from Mn2+ions in stoichiometric materials, CsMnCl3,
RbMnCl3, CsMnBr3, and Rb2MnCl4, respectively at
690, 630, 680, and 625 nm.76-79Explanations were
based on sequential absorption between dimer states
built from Yb and Mn single-ion states which are in
fact the cooperative pair states discussed above and
shown, for instance, for Er3+in Figure 6a in a
cooperative hypothesis. This explanation is coming
from the fact that emission of Mn2+is instantaneous,
as it would be for ESA between the dimer states,
which have to be considered since both ions are
involved. Along the same lines, the Yb-Tb case has
been reconsidered in SrCl2and Cs2Tb2Br9:Yb(1%),80,81
and thecooperativeeffect is found tobeESA between
cooperative pairs states and Yb cooperative lumines-
cencewith 10-6efficiency81(at 5.6 kW/cm2excitation)
for T < 100 K instead of the usual cooperative
sensitization as in ref 66 found at larger T. As rarely
given, efficiency for the cooperative sensitization
process has been determined tobe 10-4under 2.4 104
W/cm2at 300 K for SrCl2,80that is 4.2 × 10-9at a
normalized excitation level of 1 W/cm2for compari-
son, see Figure 3 and Table 1, with the values for
cooperative sensitization of 10-6in Yb,Tb:YF3and of
10-8for cooperative luminescence in YbPO4.
The Yb-Tb cooperative effect has also been revis-
ited by Strek’s group in KYb(WO4)2:Tb.82Besides the
cooperative upconversion process itself, the interest-
ing feature of a lifetime depending on the excitation
level has been observed. It has been attributed tothe
inverse of cooperative sensitization, first predicted a
long timeagoby Dexter83and only recently identified
experimentally for the first time by Basiev et al.84in
La1-xCexF3. This is different from quantum cutting,
F igure 7. Cooperative luminescence and APTE (activator
ion already in an excited state) cooperative sensitization
in LaF3:Pr3+system. (Reprinted with permission from ref
75. Copyright 1984 American Physical Society.)
T able 1. Available Measured Normalized Absolute E fficiencies for Various Upconversion Processes
matrixionsprocess order n
2.8 × 10-1
2.8 × 10-1
3.4 × 10-2
4.25 × 10-2
8.5 × 10-2
10-2to 2 × 10-4
2.5 × 10-4
5.5 × 10-2
3 × 10-7
6.4 × 10-3
8.4 × 10-4
3.5 × 10-1
3.6 × 10-3
2 × 10-6
1.7 × 10-10
8 × 10-8
1.8 × 10-8
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 149
considered by Wegh et al.,85-89which is just the
reverse of the APTE effect. Both quantum cutting
and cooperative quenching can produce quantum
efficiency larger than 1 but not with the same energy
Along analogous lines, cooperative upconversion
and downconversion processes mixed with mul-
tiphonon processes have been investigated in KYb-
(WO4)2:Eu,Tb90and in KYb0.8Eu0.2(WO4)2.91
A morecomplicated caseof cooperativeexplanation
has been given by Orlovskii et al.56,92for Nd3+in
CaF2, CdF2, and SrF2at 4.2 K. There the three kinds
of pair states discussed in section 3.1 are simulta-
neously involved in a qualitative description through
both APTE (ETU) and ESA processes. Unfortunately
no quantitative analysis with respect to the mixing
of first-order and second-order transitions has been
3.3. Application of Cooperative Luminescence;
Theory, and Examples
Because cooperative processes are less effective
than APTE (ETU) ones by 4-5 orders of magnitude,
very few applications of such processes exist except,
as we will see, the detection of RE ion clusters. We
proposed that the simplest cooperative process, the
cooperative luminescence of Yb3+,61be used as a
probe of the existence of Yb3+ions clusters in
Cooperative luminescence (or its opposite coopera-
tive absorption) is the simplest cooperative phenom-
enon, and the corresponding two-center matrix ele-
where the operator is the sum of the electric dipole
operators for ion 1 and ion 2. The wave functions for
both ions in their excited state Ψpaireand in their
ground stateΨpairgarederived from expressions given
by eq 24. The calculation of eq 26 gives four terms
that are represented in Figure 8; terms III and IV
provide the cooperative emission at twice the energy
of the single-ion excited state.
Since along the RE series nearest neighbor ions
haveanalogous chemical properties, weassumed that
Yb3+would chemically behave for the clustering
process ion the same way as for Er3+, the ion
generally used in optical amplifiers. Because Yb3+
ions have only two spin-orbit states, they are good
examples of the simple situation schematized in
Figure 8. This is one of the reasons for its use as a
cluster probe. The other reason is as follows:
In optical amplifier applications, the basic limita-
tion linked with the existence of the so-called cluster-
ing of RE ions was addressed. Clusters of RE dopant,
as found in the literature from direct fiber amplifica-
tion experiments,71,93,94are related to what could be
called interaction clusters which are much larger
than chemical clusters. Because of spatial diffusion
as shown in section 2.1, such interaction clusters
could have a spatial extension of more than 20-100
Å for nonradiative interaction clusters and up to100
µm for radiative ones. Clearly, such clusters have
nothing to do with chemical clusters, which depend
only on the chemical processes of the glass prepara-
tion. The existence of chemical clusters of spatial
extension of a few Angstroms would increase tre-
mendously ion-ion interactions of any kind. One
need to obtain a signature of such clusters that is
the only one that chemistry could eventually modify.
Cooperative luminescence of Yb3+has been proposed
as a signature of the existence of chemical clusters
Cooperative luminescence68,95-97is a phenomenon
which, requiring very close proximity of interacting
RE ions in order tobe seen in experiments, is a very
good signature of clusters constituted by ions at
distances of less than about 5 Å. Such distances, or
shorter ones, between interacting ions are also the
order of magnitude of the size of chemical clusters.
Figure 9 presents the cooperative luminescence of
Yb3+in a phosphateglass doped with Yb3+introduced
through various precursors with different Yb-Yb
shortest distances.98It has been shown that the
normalized cooperative intensities depend on such
F igure 8. Four terms for cooperative luminescence in a
two-level ion system (the Yb3+case).
F igure 9. Normalized cooperative emission spectra for
Yb3+in a phosphate glass for two doping precursors with
different Yb-Yb shortest distance Rmin.
150 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
distances. Also, it can be noted that unwanted
impurities (Er3+, Tm3+) introduced with Yb3+oxide
are revealed at ppm level by the much more efficient
APTE (ETU) effect.99
4. Experimental Results and Their Implications in
4.1. Recent Upconversion Studies in Lanthanide
(4f) and Actinide (5f) Ion-Doped Solids with APTE
(ETU) and ESA Processes
(Cooperative processes have been discussed in
sections 3.2 and 3.3 above.)
Most of the more recent published results on
upconversion under various laser pumpings have
aimed, besides cooperative effects already discussed,
at distinguishing the processes involved and mainly
at separating APTE (ETU) from plain ESA. Most of
the examples treat cases of upconversion in the now
classical 4f ion-doped solids and few in the 5f ones.
4.1.1. Pr3+(4f2) Ion
When doping fluoride glasses of the ZBLAN type
in either fiber or bulk form, Pr3+shows blue upcon-
version both of the ESA (two photon) and APTE
(ETU) (three photon) types from the3P0state when
pumping intothe1G4(1D2) state.100Analogous results
are obtained for LiKY1-xPrxF5 crystals,101but the
process is mainly APTE (ETU). Identifying the
process is based on the presence of an excitation
Germanate of general formula 60GeO2‚25PbO‚
15NbO5 as well as chalcogenide glasses of general
formula 50GeS2‚25Ga2S3‚25CsX with X ) Cl, Br, and
I,102doped with Pr3+have been investigated. IR to
blue upconversion is obtained under both ESA and
APTE (ETU) processes as revealed by the absence
or the presence of an excitation delay. The excitation
sequence is3H4to1G4then1G4to3P0(ESA) or/and
1D2(ETU).3P0gives the blue emission and1D2the
orange one. In Bi4Ge3O12:Pr3+crystal red, green, and
blue upconversion is obtained with mainly ESA and
a less important APTE (ETU) process.103From a
dynamic study of a Pr3+-doped tellurite glass, ESA
is proposed to explain the3H4 to1D2 followed by a
multiphonon process connected to a3H6 to3P1 se-
quence.104An analogous process is proposed for
LiYF4:Pr3+,105whereas a two-photon absorption is
proposed for an IR toblue upconversion in a 60ZrF4‚
33BaF3‚7LaF3 glass doped with Pr3+;106however,
because a two-photon process is likely to be less
probable than a sideband absorption,33this explana-
tion may be questioned.105
In KYb(WO4)2:Pr(0.42%),107blue upconversion un-
der red and IR excitation is obtained along now
classical APTE (ETU) schemes.
4.1.2. Nd3+(4f3) Ion
Probably due to the availability of Nd3+-doped
materials for laser research, many investigations in
Fernandez’s group with Nd3+in various kinds of
hosts have been recently performed. In germanate
glasses of composition 60GeO2‚25PbO‚15NbO5doped
with Nd3+,108the APTE (ETU) process has been
observed under CW IR excitation in the4I9/2to4F5/2
absorption. Green, red, orange emissions have been
detected from4G7/2due tothe following upconversion
energy transfer (4F3/2,4F3/2gives4G7/2,4I13/2). ESA has
also been observed with the sequence4I9/2 to4F5/2,
nonradiative decay to4F3/2, followed by ESA to2P1/2
decaying to4G7/2. In a fluoride glass of composition
0.5%Nd3+,109mostly APTE (ETU) is observed as
revealed by the delay in the emission wavelengths
ranging from red toUV and the analogy between the
absorption and their excitation spectra. The output
slope for Pnwith n ) 1.7 indicates a two-photon
In fluoroarsenate,110fluoroindate,111and in chal-
cogenide glasses,112upconversion by Nd3+ions has
been studied too, taking advantage of the weak
energy phonons of such glasses. It is mostly APTE
(ETU), with some ESA for the4G7/2 emission, that
havebeen observed from2P1/2as shown by thesimilar
spectral features in the excitation and absorption
spectra for two-photon processes.
In Pb5Al3F19:Nd3+crystals,113at 300 and 4.2 K,
APTE (ETU) is observed for visible (from4G7/2, with
n ) 1.5) and UV emission (from2P3/2, with n ) 2.2)
as indicated by similar excitation and absorption
In the stoichiometric laser material K5Nd(MO4)4,
IR to visible, blue (from4G7/2with n ) 1.6), and UV
(from2P3/2with n ) 2.4) emissions have been studied
with the very high Nd concentration of 2.37 × 1021
cm-3.114The studies, conducted in or outside the IR
lasing phases, show that the laser metastable state
is depopulated during the lasing phase as shown by
the stronger blue and orange emission. Again, up-
conversion is essentially of the APTE (ETU) type as
shown by comparing excitation and absorption spec-
In LiYF4:Nd(0.1-3%),115upconversion is found to
be due to ESA, the first step being a one-phonon
sideband absorption situated at 16563-15919 ) 644
cm-1above the2H11/2state followed by the ESA step
populating4D3/2. In the same type of fluoride crystal
as well as in YAG, the following APTE (ETU) steps
have been observed: (4F3/2,4F3/2) giving (4I15/2,4G5/2);
(4F3/2,4F3/2) giving (4I13/2,4G7/2), and (4F3/2,4F3/2) giving
(4I11/2,2G9/2)116,117tosomeextent similar with thehigh-
intensity quenching of the4F3/2 state.118BaLu2F8:
Nd3+(0.6%) crystals have alsobeen studied.119Yellow
to blue and green upconversion has been observed
due to ESA, whereas under IR excitation, green
upconversion is due to APTE (ETU) when site selec-
tion is involved at low temperature. Also, KLiYF5:
Nd3+has been investigated, and upconversion through
ESA from4F3/2has been observed.120With CaF2thin
films on LaF3crystalline sample in waveguide form,
three-photon APTE upconversion (n ) 2.9) can be
easily observed through pumping confinement.121
Very interestingly, emission at 381 nm in the UV
from4D3/2ends on4I11/2, thus providing good hope for
a UV four-level scheme laser.
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 151
4.1.3. Gd3+(4f7) Ion
Due to the fact that its lowest excited state (6P7/2)
is at very high energy (32 000cm-1), this ion is mainly
considered for downconversion studies and is in-
volved in fewer upconversion studies than the two
previous ones. However, upconversion has been
considered in the stoichiometric material with cen-
trosymmetric sites Cs2NaGdCl6.122ESA has been
advocated (maybe with some direct two-photon al-
lowed absorption) for the emissions between excited
states from the 5d to6IJ and6PJ bands at 578 and
755 nm under excimer laser excitation into the6PJ
lines. Another stoichiometric material, K2GdF3, has
also been investigated in upconversion.123Pumping
is into the6P1/2state at 312 nm. Three anti-Stokes
emissions attributed to6G7/2 and6I7/2 down to the
ground state8S7/2at 204.7, 242.0, and 2798 nm have
been observed and attributed to an APTE (ETU)
4.1.4. Dy3+(4f9) Ion
This ion had the reputation of being a poisonous
center even at trace levels for APTE (ETU) in usual
Yb-Er and Yb-Tm or Ho upconversion matrices.2
For this reason it was banished from the laboratory.
This could explain why sofew studies exist about the
upconversion properties of this ion. It is probably also
because the proximity of the lower excited states
requires low-energy phonon matrices. The level struc-
ture is also somewhat analogous to the situation for
Eu and Tb for which essentially cooperative upcon-
versions have been observed (see section 3.2). Yet
recently CsCdBr3:Dy3+(0.2%) has been studied in
upconversion.124Both APTE (ETU) and ESA pro-
cesses havebeen observed at 10 K. Near-IR excitation
is by absorption into6F5/2 at 12 338 cm-1and into
6F3/2at 13 200 cm-1. Emission is from4F9/2to6H13/2
at 17 341 cm-1. TwoAPTE (ETU) schemes are likely,
the more probable being (6F5/2,
6H13/2). At higher temperature (295 K) and large
concentration, ESA is operative from6F9/2to4F9/2. At
lower concentration, ESA appears tobe from6F5/2to
6F5/2) giving (4F9/2,
4.1.5. Ho3+(4f10) Ion
Though this ion is among the first studied in
upconversion2with Yb codoping, it has later been
studied alone. First, some years ago,125red to blue
upconversion was observed as well as a red (He-Ne
laser) pumped IRQC for 2 µm detection at 300 K was
demonstrated in HoxY1-xF3(x from 0.005% to 1). An
APTE (ETU) process, (5F5,
provides the5F3 emission at 485 nm. Much more
recently, CsCdBr3:Ho3+(0.035% and 2.25%)126has
been studied, this host differing from the previous
oneby its lower phonon maximum energy (163 versus
560 cm-1) and its pair building ability. A two-photon
process has also been found for the blue emission
from5F3with slope n ) 1.8. ESA is advocated at a
higher temperature (T > 100K), whereas APTE
(ETU) is found at T < 100K. Comparison of excitation
spectra with absorption and delay in the emission
help to separate the processes. Ho3+has also been
5I7) giving (5F3,
excited in the red at 647 nm from a gas Kr laser in
LiTaO3:Ho(0.3%).127At 15 K, a green emission with
n ) 2 is observed; it is attributed to the5S2 to5I8
transition excited by an ESA process. In a more
classical experiment, Ho is excited via Yb in YVO4:
Yb,Ho.128The near-IR excitation at 1 µm provides
both a red emission from5F5by an ESA process (n )
1.6) and a faint green one from
attributed to an APTE (ETU) process. Interestingly
and along the same lines in Yb, Ho-doped fluoro-
hafnate glasses,129APTE (ETU) IR to green upcon-
version has been measured to give an absolute
efficiency of 8.4 × 10-4cm2/W, a value directly
comparable to the one of Figure 3 and Table 1, with
10-3cm2/W for YF3:Yb,Er.
5S2 (n ) 1.6)
4.1.6. Er3+(4f11) Ion
(See also section 4.2.)
Er3+was the first ion showing upconversion,2and
it seems that thenumerous previous studies havenot
exhausted its upconversion properties. It still appears
as the most studied ion in recent times, as will be
shown in the following.
With theavailability, in the1980s, of efficient laser
diodes (LD) and tunable Ti-sapphire lasers in the
800-1100 nm range, thefield of upconversion studies
with Er3+has been renewed. In particular, Er3+has
demonstrated its capacity as a laser ion just as Nd3+
did a long time ago, and consequently, all kinds of
upconversion emissions have been observed and
studied. The role of upconversion on the CW func-
tioning of the LiYF4:Er and ZBLAN fiber lasers at
2.7 µm130-132has recently been confirmed,133and a
cascade laser at 1.72 and 2.7 µm laser have been
optimized, in particular, in a ZBLAN glass doped
with Er(0.25-8.75%) alone or with Pr(0.25-1.65%)-
Er134using the APTE (ETU) process for optimizing
the4I11/2and4I13/2lifetimes. Even classical laser hosts
have been investigated: YAG:Er,135YSGG:Er,136and
YAlO3137for which either ESA or APTE (ETU) have
In a more fundamental approach, Cs3Lu2Cl9, Cs3-
Lu2Br9, and Cs2Lu2I9doped with Er3+(1%) as well as
the stoichiometric material Cs3Er2X9(X ) Cl, Br, I)
have been investigated138,139under 1.5 µm excitation.
As in ref 72, an APTE (ETU) process describes the
observed four-photon upconversion process at higher
Er concentration, though the process is called coop-
erative energy transfer.139
In BaLu2F8:Er(1%; 4.5%), IR (0.97 µm) to green
upconversion from4S3/2is observed.140Both APTE (at
all temperature) and ESA at lower temperature from
4I11/2and4I9/2are identified by the transients of the
upconversion emission. Ba2YCl7:Er (1-100%) has
been studied141under 800 nm excitation of the4I9/2
state. Depending on excitation energy and concentra-
tion, both APTE (ETU) (for Ba2ErCl7) and ESA are
observed. In RbGd2Br7:Er (1%),142under 980 nm
discriminated by the excitation transients, are ob-
served with a ratio depending on excitation energy
and temperature. Besides the above studies, the
Gu ¨del’s group in a systematic manner also studied
BaY2F8:Er and Cs3Er2Br9143in order to compare the
4I11/2, both APTE (ETU) and ESA,
152 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
upconversion properties from near-IR togreen in two
hosts differing by their highest phonon energy,
respectively, 415 and 190 cm-1. In Cs3Lu2Br9:Er-
(1%),144a lattice with built-in pair structure, both
ESA and APTE (ETU) processes are observed but
mostly ESA when exact matching between levels and
laser excitation is obtained. Energy migration is
noted even at the relatively low 1% doping.
Similar studies have been performed in Ca3Al2-
Ge3O12:Er, where both ESA and APTE are observed
depending on the excitation wavelength.143In LiYF4:
Er(3%), Yb(20%),146five-photon near-IR to UV has
been observed with APTE (ETU) processes similarly
to72for Er alone.
A rather original and interesting result in a clas-
sical Yb(0.5%)-Er(0.1%) system is the optical am-
plification in an upconversion-pumped chalcogenide
glass (70Ga2S3-30La2O3).147Pumping is at 1.06 µm
in an anti-Stokes two-phonon sideband at 2 times 425
cm-1from the YAG:Nd laser photon energy. Ampli-
fication, which is maximum at 165 °C is at 555 nm
with a gain factor of 10. This corresponds to an
amplification efficiency of 0.012 dB/mW in the bulk
glass samplewithout optical confinement. Also, in the
sameaim of amplification in thegreen spectral region
with near-IR pumping, LiYF4:Er has been studied in
detail under an InGaAs LD pumping.148LiNbO3:Ti,-
Er waveguides149havebeen studied for upconversion-
pumped laser either at 550 nm or 2.7 µm. Both APTE
(ETU) and ESA are observed. For a 550 nm laser,
APTE (ETU) is necessary and can be obtained
essentially in Er clusters which have tobe increased
at Li+and Nb5+sites. Here Er clusters are looked
for, in contrast to the situation in Er-doped optical
fibers for 1.55 µm amplification, see section 3.3.
LiNbO3:Er waveguides, carefully pumped in a site-
selective manner150have shown ESA and APTE
(ETU) upconversion according tothepumping photon
energy. When weaker than 12450 cm-1, ESA is
observed; otherwise APTE (ETU) is obtained. A two-
photon process exists in both cases as shown by the
observed n ) 1.92 value. Pumping is into4I9/2at 800
nm, and emission is from4S3/2. ESA is from4I13/2to
2H11/2 with some involved nonradiative decays. For
APTE (ETU), after a nonradiative decay to4I11/2, the
following takes place: (4I9/2,4I11/2) gives (4I15/2,4F3/2)
providing the4S3/2excitation reached by a nonradia-
tive decay from4F3/2. In the same kind of waveguide,
traces of avalanche (see section 4) have been identi-
4.1.7. Tm3+(4f12) Ion
Tm3+is known to be one of the first ions having
shown upconversion either alone or with the help of
Yb3+.2As for other ions, the advent of lasers has
renewed its interest. Also, research of upconversion-
pumped lasers has also been an impulse to the
research. LiYF4:Tm has been studied from its spec-
troscopic parameter point of view for laser applica-
tions123with codoping with Pr3+as well.152After a
3H6to3H4excitation at 12 643 cm-1(791 nm) in the
Tm3+ion, a double excitation,3H6to1G4then to3P2
by an APTE (ETU) process in the Pr3+ion, as
revealed by a delay in the built-up transient, allows
emission from3P0to3H6(600 nm) and3H4(490 nm)
in the Pr3+ion.
Upconversion has been studied in the stoichiomet-
ric crystal TmP5O14 as well as in the amorphous
Tm0.1La0.9P5O14,.153Under red pumping, UV and blue
emission are observed from1D2with n ) 2, respec-
tively, to3H6(360 nm) and3F4(450 nm).1G4emission
at 480 nm is quenched at a concentration of 100%
down to10%. Comparing excitation spectra, stepwise
APTE (ETU) and ESA are conjectured. The amor-
phous sample provides the largest intensity at 450
nm. An emission at 347 nm from5I6 to3F4 can be
observed with n ) 2.5.
In garnets Y3Sc2Ga3O12, Gd3Ga5O12and in GdAlO3
doped with Yb(10%) and Tm(0.1%),154upconversion
to 460-500 nm can be observed under Ti-sapphire
excitation at 790 nm corresponding tothe3H6to3H4
transition in Tm3+. Then a back transfer toYb allows
the2F5/2Yb3+population and the subsequent APTE
(ETU) process from
different from the classical Yb-Tm case under a first
In Cs3Yb2Cl9:Tm, a matrix with low-energy phonons
(<280 cm-1), up to five-photon APTE (ETU) process
is observed with a scheme analogous to the one of
Figure6b, however, with a nonradiativestep replaced
by an internal APTE (ETU) step within the Tm3+
itself.155The slopes, respectively, observed are n )
1.4, 2.0, 2.6, and 3.4 for emissions from3H4,3F3,1G4,
and1D2. The respective excitation spectra are the Yb
absorption narrowed by the power law as shown in
Figure 7 for the Er case, also proving the APTE
(ETU) process. The difference here is that because
of the internal APTE (ETU) process in Tm3+, the
power law to be considered for1G4and1D2are here,
respectively, 3/2 and 4/2, as already explained for the
n ) 3/2 slope mentioned in one of the red upconver-
sion processes of Yb-Er2.
Fluorohafnate glasses doped classically with Yb
and Tm have been investigated,129and absolute
efficiency has been shown tobe 6.4 × 10-3cm2/W for
the 804 nm emission. This two-photon upconversion
efficiency is similar to the one given for Yb-Er, see
Figure 3 and Table 1.
In a silica fiber 3.5 m in length doped with Tm
alone, visible and UV upconversions at 650, 470, and
366 nm all with a slope of n ) 3 have been ana-
lyzed.156Absorption is at 8300 cm-1in the3H6to3H5
transition. Upconversion is thought to be enhanced
by the first and second Raman transitions observed
at 1120 and 1180 nm. Above a threshold at 10 mW,
line narrowing is observed and is considered as an
indication of superluminescence.
Along the same directions as in ref 51, lead ger-
manatevitroceramics doped with Yb(15%)-Tm(0.1%)
have shown APTE (ETU) upconversion with a two-
photon process at 779 and 698 nm, a three-photon
process at 478 nm(1G4), and a four-photon one at 363
nm(1D2).157Measured absolute efficiencies were 5.8
× 10-3at 779 nm and 10-6at 478 nm under a 16.5
mW/cm2excitation. In normalized units it gives
respectively for the two-photon and the three-photon
processes 3.5 × 10-1cm2/W and 3.6 × 10-3(cm2/W)2.
3H4 to1G4. This behavior is
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 153
Interestingly this IR to IR two-photon process is
much more efficient than the IR to green process in
Yb-Er. On the other hand, the IR toblue transition
is about 20 times less efficient than the efficiency
obtained in thefirst Yb-Tm-doped vitroceramics (8.5
× 10-2(cm2/W)2,158see Table 1). In the fluoride glass
BiGaZYbTZr:Tm3+,159a crossover from cooperative
sensitization to APTE (ETU) is concluded from the
time behavior changes with pulse excitation length.
4.1.8. Tm2+(4f13) Ion
This ion is considered for the first time in upcon-
version. It is isoelectronic with Yb3+and as such has
the same level structure: twospin-orbit states2F7/2
and2F5/2separated by about 8840 cm-1and parity-
allowed 4f-5d bands above15 000cm-1. Thepresence
of Tm2+is not common due to its propensity to
oxidation, but here due to the considered SrCl2
divalent host, 2% of Tm2+has been successfully
introduced160without the presence of any Tm3+. The
level structure is such that the 4f-5d bands are at
about twice the energy of the first2F5/2excited state
and has prompted Gu ¨del’s group to investigate this
new ion for upconversion at 15 K under a filtered 80
W lamp excitation at 8840 cm-1and with a pulsed
Nd:YAG laser for the transient study. The absence
of delay in the upconversion signal indicates an ESA
4.1.9. U4+(5f2) Ion
This is the first 5f ion in which upconversion has
been observed161in ThCl4and ThBr4:U4+(0.05%). The
first observation was fortuitously found at CNET on
a supposedly undoped ThBr4 sample. Under a Nd:
YAG pulsed excitation, a green SHG signal at 532
nm was looked for in order to detect the crystal
eventual noncentrosymmetry. In fact, instead of the
green spectrally narrow signal at 532 nm, we es-
sentially observed a broad red one at a luminescence
emission wavelength known for U4+. This observation
indicated that upconversion was activein U4+at very
weak concentration levels. A derived conclusion was
that the oscillator strengths were very large and
probably the energy transfers too. This induced the
first determination of U4+oscillator strengths that
showed values of =10-4 162about 2 orders of magni-
tude larger than for Ln3+and one order larger than
then known values for U3+. Recent results for U3+,
introduced for the first time by a pure chemical way
in a ZnCl2-based glass, indicate values of =10-6
which is about the same as that for Ln3+ 163and 2
orders of magnitude less than that for U4+.
With the 1 KW tungsten iode filtered lamp experi-
ment already used for the first Yb-Er and Yb-Tm
investigations (see Figure 17 in section 4.2), several
emission lines in the red and green have been
attributed to ESA either for excitation at 950 and
1170 nm separately or for excitation at 950 plus 1170
nm.161The involved levels are connected by absorp-
tion from the ground-state3H4to3H6,3F3, and3H5
and then ESA from these states to3P0,3P1, and1I6,
seeFigure10. Thelinear behavior with concentration
showed that upconversions were not due to APTE
(ETU) processes. On the other hand, as shown for
Yb-Er and Yb-Tm,51photon trapping was present
as indicated by the grain size effect on lifetimes and
on the upconversion efficiencies. It shows that radia-
tive diffusion plays an important role also in ESA
upconversion. Normalized efficiency is found to be 2
× 10-6cm2/W, see Table 1, for ThBr4grains doped
with 0.05% U4+and of 0.2-0.3 mm optimized size.161
4.1.10. U3+(5f5) Ion
Though there has been one publication on the
upconversion properties in trihalide-doped Cm3+,164
most of the upconversion studies with trivalent
actinides are with U3+from Strek’s group. In LaCl3:
U3+and LaCl3:U3+, Pr3+,165under Nd:YAG laser
excitation, ESA is found togivethe2K15/2to4I9/2green
emission in U3+alone. When coupled toPr3+, a cross-
relaxation process allows a second ESA within the
Pr3+ion giving its3P0 excitation. A refined study
indicates a more complex upconversion process with
back transfer toU3+and APTE (ETU) process within
U3+.166Under red laser pumping in the4I9/2to2K13/2
transition, green emission from2K15/2to4I9/2can be
observed. This upconversion is attributed to two
processes:167(i) a sequential absorption within one
U3+ion wherethesecond photon populates the5f5-6d
bands and thus the energy is transferred tothe2K15/2
emitting stateand (ii) an APTE (ETU) process within
an U3+pair of ions following the sequence (4F9/2,4F9/2)
w (2H9/2,2H11/2) -35 cm-1(a weak phonon energy).
Other paths for excitation have also been investi-
gated in LaCl3:U3+,168and green and red emissions
have been obtained with slopes, respectively, equal
to n ) 1.97-2.5 and 1.7-1.85, according to the
precise excitation wavelength.
In centrosymmetric elpasolites Cs2NaYBr6and in
Cs2NaYCl6 doped with U3+,169due to multiphonon
quenching, ESA upconversion has been observed only
in the bromide type.
F igure 10. Energy schemefor various ESA upconversions
involved in ThBr4:U4+.
154 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
4.2. Recent Upconversion Studies in
Transition-Metal (3d, 4d, 5d) Ion-Doped Solids
with APTE (ETU), ESA, or Cooperative Processes
The first consideration of d ions in upconversion
can befound in thework of Cresswell et al.170in 1978,
where in Cs2NaYCl6they considered Re4+, a 5d3ion,
as a replacement for Yb in the IR to blue upconver-
sion in Yb-Tm systems, with the idea of having an
APTE (ETU) two-photon process instead of a three-
photon one. This was supposed to improve overall
efficiency in diode-pumped anti-Stokes visible light
sources. Unfortunately this system proved itself to
be surprisingly inefficient. The second example was
obtained in Auzel’s group at CNET in the tunable
laser material MgF2:Ni2+.171Duetothestrong Stokes
shift experienced by d ions in solids caused by a
medium crystal field strength inducing itself medium
electron-phonon couplings, resonant diffusion is not
as effective as in lanthanides. Under a krypton gas
laser excitation at 752.5 nm, green upconversion at
80 K was observed at 500 nm coming from the1T2-
attributed tothe3T1(3F) to3T1(3P) ESA transition. A
observed to self-quench the green emission. The
determined microparameter values for this energy
transfer were consistent with an exchange interac-
tion. This upconversion was not studied in order to
improve upconversion itself but because it was seen
as a drawback tobe reduced in the IR laser function-
ing of Ni2+.
Following these twopioneering works one can say
that the field of upconversion in d ion-doped solids
has been developed by thesystematic work of Gu ¨del’s
group in recent years. Presentation of this field is the
subject of the following paragraphs.
3A2(3F) transition, see Figure 11. It was
3A2) giving (3T2,
4.2.1. Ti2+(3d2) Ion
In NaCl- and MgCl2-doped (0.1-0.2%)Ti2+crystals,
the near-IR (9400 cm-1) to visible upconversion is
analyzed at 15 K in these two hosts.172Spectral
analysis of these twocrystals reveals that NaCl and
MgCl2have crystal field strength on both sides of the
spin crossover point for the first excited metastable
state. For NaCl it is3T2with the same spin as the
ground-state3T1, whereas for MgCl2it is1T2, i.e., it
gives a spin-forbidden transition tothe ground state.
This dramatically changes the metastable state
radiative lifetime from, respectively, 1.4 to 109 ms
and consequently the relative efficiencies of the ESA
processes observed in both crystals.
4.2.2. Cr3+(3d3) Ion
In YAG173and YGG (Y3Ga5O12)174,175codoped with
(2%)Cr3+and (1%)Yb3+, upconversion of Cr3+through
a near-IR pumping of Yb3+is observed at 10 K with
an efficiency of 6% under 150 mW of excitation173-175
but at undefined energy density. The presence of a
delay in the transient of the emitted signal indicates
the presence of an energy transfer. Because there is
nometastable level below the2E Cr3+emitting state,
a cooperative sensitization process can only explain
the whole process. Further, the cooperative lumines-
cence of Yb3+is simultaneously observed. The role
of the efficient diffusion of energy among the Yb3+
ions is stressed as is generally the case with this ion.
Even three ion systems have been studied one of
them being Cr3+in YAG: (5.76%)Tm3+, (0.36%)Ho3+,
and (1%)Cr3+,176a well-known 2.1 µm laser material.
Under near-IR (720-790 nm) and red excitation
(610-660 nm), a blue emission from1G4(Tm) at 486
nm and from5F3(Ho) at 486, 489, and 497 nm can
be seen; an upconverted emission is obtained also
from the2E (Cr3+) level at 688.7 and 687.6 nm, the
R1and R2lines. All such emissions are losses for the
laser process. They come from cross energy transfers
between the three ions and ESA excitation.
4.2.3. Ni2+(3d8) and Mn2+(3d5) Ions
Besides Ni2+in MgF2,171already mentioned, Ni2+-
(0.1-10%) has been investigated as an upconversion
ion either alone in RbCdCl3,177CsCdCl3,178and Rb2-
CdCl4,179or coupled with Mn2+
CsMnCl3, and RbMnCl3,178and in Rb2MnCl4.179
Under near-IR excitation at 15 K, Ni2+alone is
found to produce a green upconversion by an ESA
process as explained above for MgF2:Ni2+ 171but with
different level attributions, maybeduetothefact that
crystal field strengths are different and that zero-
phonon lines have not been considered. Here, the
ground-state absorption is from3A2to1E and ESA
from3T2to1T2; emission is the same transition1T2
to3A2. In RbCdCl3, a pressure study modifying the
crystal field strength shows an increase in upcon-
version due an increase in the spectral overlap
between ground-state absorption and ESA.
With the Mn2+presence, a strong increase in
upconversion is observed, though Mn2+has nometa-
stable level below1T2 (Ni2+). On the contrary, the
metastable state4T1of Mn2+is in resonance with1T2
F igure 11. Energy schemefor ESA upconversion in MgF2:
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 155
(Ni2+) and in CsMnCl3 is found to emit slightly at
the expense of1T2, which is then slightly quenched.
An explanation is given by a strong enhancement of
the1E absorption intensity due to a stronger spin-
orbit interaction linked to the proximity of the3T1
and1E state. The upconversion enhancement also
found in RbMnCl3cannot be of this type because this
proximity does not exist. It is attributed to a strong
exchange coupling178between Ni2+and Mn2+, remov-
ing thespin selection rulefor the3A2to1E absorption
from ∆S ) 1 to0, and coming from themixing of both
Ni2+considered states with the Mn2+ 6A1 ground
4.2.4. Mo3+(4d3) Ion
This ion has been studied in Cs2NaYCl6and Cs2-
NaYBr6181,182at 2% doping concentration. Under
near-IR excitation from a Ti-sapphire laser, red
upconversion is observed from 10 to 150 K. Under
one-wavelength excitation, both ESA and APTE are
observed. Under two-wavelength excitation, essen-
tially ESA provides upconversion with a better ef-
ficiency. ESA upconversion implies a2T1/2E to4T2
absorption preceded by a
ground state, see Figure 12. APTE is provided by the
(2T1/2E,2T1/2E) to(4A2,4T2) energy transfer, giving the
emission from4T2. In the chloride host, the ratios for
upconversion processes are estimated to be APTE
(15%) and ESA (85%). For the bromide, it is, respec-
tively, 35% and 65%. This ion is characterized by a
very long lifetime of 67.5 ms coming from the
relatively weak oscillator strengths; it can be noted
that electron-phonon parameters are weak (S )
0.05) for the intermediate states2E/2T1 and strong
(S ) 4.5-5.7) for the final excited state very near
the emitting state (S ) 0.05). This explains the
necessity of thelow temperaturefor theupconversion
2T1/2E from the
4.2.5. Re4+(5d3) Ion
This ion, also being a d3configuration ion, has
roughly the same Tanabe and Sugano energy dia-
gram183as other better known transition metals (TM)
such as Cr3+and Mo3+. It has been studied in Cs2-
ZrCl6181,184and Cs2GeF6.185Though Re4+has thesame
level structure as Cr3+and Mo3+, as shown in Figure
12, it has been the first TM ion toshow upconversion
at room temperature.181,184Contrary toother TM ions,
Re4+shows an efficient APTE (ETU) process because
selection rules on spin are relaxed by a larger spin-
orbit coupling. The main difference with Mo3+can
be traced back to the larger oscillator strength for
the4A2(Γ8) to2T1(Γ8) by a factor 102with respect to
transfer involved being (2E/2T1(Γ8),2E/2T1(Γ8)) gives
(4A2(Γ8),2T2(Γ8)). Thus, excitation into2T1(Γ8) at about
1.1 µm (Nd:LiYF4laser at 1.047 µm) provides a red
emission at about 725 nm.
In a solution-grown Cs2GeF6:(2%)Re4+crystal,185
theupconversion luminescencedecreases down to2%
when temperatureis increased from 15 to300 K. This
is explained only partially by the larger maximum
phonon energy, 600 versus 350 cm-1in chlorides and
220 cm-1in bromides, which increases nonradiative
transitions, and mainly by a decreasing absorption
cross section at the laser excitation wavelength.
Upconversion is here also mainly an APTE (ETU)
process as shown by the time transient measure-
2T1/2E in Mo3+. The upconversion energy
4.2.6. Os4+(5d4) Ion
This 5d TM ion has been found to have also the
right sequence of levels to show upconversion, see
Figure 12. An APTE (ETU) effect is observed in Cs2-
ZrCl6:Os4+(1%)186,187below 80 K; in Cs2ZrBr6:Os4+two
ESA processes lead to upconversion;187and in Cs2-
GeF6:Os4+, no upconversion is detected.187Such dif-
ferences are traced back, in the fluoride host, to the
strong nonradiative decay from the
which would emit the visible light at about twice the
excitation energy. In the bromide188and chloride189
the level sequences allow both resonant and out of
resonance ground-state absorption, which contribute
to APTE (ETU), ESA, and avalanche upconversion
(see section 4).
In the double-doped Cs2NaYCl6:Os4+, Er3+,190up-
conversion is observed under a scheme similar tothe
pioneer work170involving a TM ion for absorption and
a lanthanide for emission. The green emission from
4S3/2(Er3+) has been found to be both of the APTE
(ETU) and ESA types (though called cooperative)
with some back-transfer from Er3+to Os4+.
4.3. APTE (ETU) for Display and IR Detection
In display technology the light-emitting material
is always in powder form, traditionally called a
phosphor. Because of various inclinations of the
crystallite external surfaces reducing total internal
reflection, more light output is extracted in a wider
view anglefrom crystallites than from theequivalent
single crystal. The upconversion phosphor field has
recently been reinvestigated191-195for the now well-
known two-photon and three-photon phosphors based,
respectively, on Er-Yb and Tm-Yb codoped materi-
als. Beyond the older light-emitting incoherent
sources,2the renewed interest stems from potential
applications ranging from simple handheld devices
used tofind IR laser beams196-198tovisible enhanced
detection of IR emissions, X-rays reusable memory
plates,199and 3-D display technologies.200
F igure 12.
configuration of Cr3+and corresponding energy schemes
involved in upconversion by ESA processes for Mo3+and
Re4+. (Reprinted with permission from ref 181. Copyright
1998 American Chemical Society.)
Tanabe and Sugano diagram for the d3
156 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
In particular, with now available tunable lasers
and fiber beam homogenizers, thevarious efficiencies
have been recently revisited in Krupke’s group191
with more refined experiments than pioneering
ones.158,201,202Results essentially confirm the previ-
ously measured efficiencies below saturation; nor-
malized efficiencies of 10-2and 2 × 10-4cm2/W, for
twophotons, havebeen obtained, respectively, for the
fluoride hosts NaYF4and Na2Y3F11, instead of 10-3
cm2/W as shown in Figure 3 and in Table 1 for YF3.
In the case of the Yb-Tm couple, the initial energy
level diagram depicting the involved processes, as
shown in Figure 13,7is confirmed191against the
cooperative sensitization scheme43of Figure 3. In
NaYF4, efficiency as high as 2% is reached under
pump excitation of 6 W/cm2at 960 nm, though a
saturation density of 4 W/cm2is estimated for the
first intermediate step (4H4). In the Yb-Er case, the
saturation is found at about 100 W/cm2for fluoride
hosts NaYF4and Na2Y3F11;191observed saturation is
explained by an excitation trapping into the long
lifetime4I13/2 state. These more recently obtained
results actually confirm thetheoretical prediction73,159
that fluorides should be the ideal hosts for green and
blue emissions with the Yb-Er and Yb-Tm couples,
see Figure 14.
In this respect, Quimby et al. studied heavy-metal
fluoride glasses203and Auzel et al.51proposed and
studied particularly efficient composite APTE (ETU)
upconversion materials in which the RE ions were
substituted into a crystalline matrix (PbF2) itself
embedded into an oxygen-based glass material. By
such means, the multiphonon processes were opti-
mized in the fluoride crystals whereas the overall
sample was obtained through classical oxygen glass
techniques. Silver nanometric particles have been
shown tocouple tolanthanide in glasses,204and such
coupling has been used in Er-Yb-doped vitroceram-
ics as a way tostudy this coupling inside a scattering
medium.205As expected, it shows a quenching of the
APTE process. The vitroceramic pioneering work has
recently been extended totransparent glass ceramics
by Ohwaki’s group.206Transparency was obtained by
reducing the crystallite (Pb0.0.5Cd0.0.5F2) size down to
about 20 nm.
The same group also recently investigated Er3+-
doped BaCl2and ErX3(X ) Br, I)207polycrystalline
phosphors for 1.5 µm detection cards,198which follow
the upconversion processes described in Figure 5b.
More recently, they demonstrated208a 1.3 µm to
visible detection in a Dy3+-Er3+codoped BaCl2
phosphor. The involved APTE (ETU) scheme is given
in Figure 15. As can be seen, Dy3+is the sensitizer
while Er3+is the activator. It should be noted that
some back transfer from Er toDy is taking place and
will be a basic limitation as it has been known for
years that Dy3+is a quencher for theEr3+emissions.2
Besides equal energy photon summation as just
seen, different energy photons may be summed up
by APTE (ETU), soproducing an effective IRQC with
visible-enhanced IR detection.209Again, with theYb-
Er couple, embedded in vitroceramics of general
composition, PbF2-GeO2-YbF3-ErF3, an IRQC has
been obtained for 1.5 µm, using an additivepump flux
at 0.96 µm. Final detection at 0.66 µm is obtained by
summation as shown in Figure 16a. The experimen-
tal test scheme is presented in Figure 16b; the pump
flux is produced by an IR GaAs:Si light-emitting
diode (Mullard GAL-10); the signal is produced by a
filtered tungsten lamp; a GaAs photocathode photo-
multiplier provides the final detection at 0.66 µm. A
noise equivalent power of 10-11WHz-1/2has been
obtained for 3 mW of pumping at 0.96 µm and useful
F igure 13.
scheme in Yb3+-Tm3+couples.
Three-photon APTE upconversion energy
F igure 14. Theoretical effective phonon energy optimiza-
tion for Yb-Er (green upconversion) and Yb-Tm (blue
F igure 15. APTE scheme for 1.3 µm to visible upconver-
sion in Dy3+-Er3+-doped BaCL2. (Reprinted with permis-
sion from ref 208. Copyright 1994 American Physical
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 157
concentration of 10%(Yb3+) and 5%(Er3+). As always
in APTE (ETU) effects, it can be noted that the
sensitizer concentration is larger than the activator
one, as predicted by eq 22. Of course, if instead of a
photomultiplier the detection is made through an
electronic image intensifier, an imaging IRQC can be
obtained for IR visualization purposes. This approach
has recently been systematically considered in
Smirnov’s group210in order to make an IR image
converter at wavelengths beyond the usual S1 or
GaAS photocathode limits. Here, IR(1.5-1.6 µm) to
-IR(0.8-0.9 µm) upconversion phosphors have been
developed with the addition of microchannel image
intensifiers. On the basis of Y0.85Er0.15O2S phosphors
and an image intensifier with a GaAs photocathode,
an IR sensitivity of 10-11W/cm2is anticipated for an
image converter screen illuminance of 0.1 cd/m2.211
In another direction, reusable X-ray memory detec-
tion plates have been made from screens based on
the same type of vitroceramics as mentioned above.
X-rays are then the source of defect centers, which
reduce the Yb lifetime, which in turn reduces the
APTE upconversion efficiency. After an X-ray ir-
radiation through the object to be investigated, the
X-ray latent imageis revealed in thevisiblespectrum
by an IR irradiation at 0.96 µm, the Yb3+excitation
wavelength. The sensitivity is increased with the
higher order upconversion processes. An image so
produced in the blue spectrum is shown in Figure
17 for a three-photon APTE (ETU) process in a Yb-
Tm-doped vitroceramic screen revealed under IR
after a 90 KV, 5 mA, 1 min 45 s X-ray irradiation.199
The screen may be reused after a heating procedure,
which bleaches the defect centers.
The inherent nonlinearity of the APTE (ETU)
process had been considered as an incoherent optical
amplifier since the first days of APTE.7Its principle,
given in Figure 18, is based on a two-beam scheme;
one of them, the pump beam, is at broad band and
CW; it fixes the bias point of the sample. The other
beam (narrow band and modulated) is the signal
beam which is tuned through the excitation band to
be investigated. Practical gain of a few factors of 10
can be so obtained. This procedure has been ef-
fectively used to obtain the excitation spectra for
upconversion both in 4f systems7and 5f systems (U4+
in ThBr4and ThCl4),161in the absence, at the time,
of synchronous electronic amplifiers.
The first proposed use of APTE (ETU) phosphors
has been the handheld laser mode visualization
screen196since then proposed again for visualization
of both 0.96 and 1.5 µm spectral regions.197,198
In ending this section, it may be useful from an
applied point of view to summarize in Table 1 the
various values for measured normalized efficiencies
as they have appeared in the literature. As it may
have been noticed, such values are rarely given but
are important in showing the relative practical inter-
est of the various proposed upconversion schemes.
4.4. General Negative Roles Brought up by
Undesired APTE (ETU) Effects
APTE (ETU) being an anti-Stokes process does
induce reabsorption from excited states. Because it
is so efficient when concentration and excitation
F igure 16. Experimental scheme for a 0.96 µm diode-
pumped IRQC at 1.55 µm with energy transfers (a);
experimental setup for IRQC detection at 1.55 µm (b).
F igure 17. Latent X-ray image of a tooth in a Yb-Tm-
doped vitroceramic, revealed by the blue emission under
an IR irradiation at 960 nm.
F igure 18. Incoherent amplifier scheme with an APTE
158 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
density are high, it is sometimes an undesired
pervading effect. Historically this effect takes its roots
in a study on reabsorption from a population-inverted
state in Yb-Er laser glass studies:24all visible light
is obtained by upconversion at the expense of the 1.5
µm laser transition. In high-concentration Nd3+laser
materials, the so-called stoichiometric lasers, APTE
effect may induce reabsorption.118Because upcon-
verted energy feeds nonradiative states, this is a
direct loss for the laser output. In NdP5O14such an
effect begins when population inversion reaches 10%,
whereas in YAG:Nd3+where the concentration of RE
ions is much lower, it cannot be seen even at high
This type of APTE effect is also just the process
that has been advocated to take place within Er3+
clusters and which limits the gain in optical fiber
amplifiers,71,93,94providing theso-called nonsaturable
ground-state absorption (see section 3.1 above). Its
involved energy scheme is given in Figure 19a.
In a CW 2.7 µm Er3+laser, it had first been thought
that APTE had the positive effect of emptying the
final4I13/2laser state, avoiding a self-terminating or
self-saturation behavior213due tothe long lifetime of
this final laser state.214An investigation of the
various upconverted emission in LiYF4:Er during the
laser action at 2.7 µm brought some proof that APTE
(ETU) was in fact emptying the laser emitting level
4I11/2, thus being a drawback rather than an advan-
tage.130Thus, the laser CW action was essentially
linked with a direct reabsorption from4I13/2 of the
pumping laser toward higher states.
Another important field in the phosphor domain
is the availability of a high-intensity projection
cathode-ray tube, which is presently universally used
for large-screen television displays. At high excita-
tion, density saturation appears that has been par-
tially attributed to APTE (ETU) for higher doping
concentration;215,216theclear analogy with thecluster
role in Er-doped amplifiers is shown in Figure 19b.
It has been the explanation for Zn2SiO4:Mn, Y2O2S:
Eu, and YAG:Tb. Furthermore, in electrolumines-
cence phosphors such as ZnS:Mn, upconversion has
been proposed to explain saturation.217
4.5. APTE (ETU) and ESA Pumped Lasers
Very early, APTE (ETU) effect has been demon-
strated to be a new way for laser pumping. For
example, in Ba(Y,Yb)2F8:Er and Ho, even with IR
flash pumping,218respective pulsed emissions have
been obtained at 0.67 and 0.55 µm as shown in Figure
20 in a three-level laser configuration at room tem-
perature. The availability now of powerful CW IR
laser diodes has rejuvenated this field. Because
powerful blue laser diodes are still lacking for Stokes
laser pumping, there are still many openings for
APTE (ETU) anti-Stokes pumping.
By using the first twosteps described in Figure 21,
APTE (ETU) has been used to excite the4I11/2of Er
in CaF2to obtain a pulsed laser at 2.8 µm between
that level and4I13/2.219Shortly afterward and for the
first time, a CW upconversion laser in the green for
the4S3/2-4I15/2transition of Er at 0.551 µm has been
F igure 19. (a) Energy scheme for cluster quenching in
an Er-doped amplifier. Reprinted with permission from ref
71. Copyright 1991 Institute of Electrical and Electronic
Engineers.) (b) Energy scheme for APTE saturation in a
cathode ray tube phosphor, note analogy. (Reprinted with
permission from ref 216. Copyright 1983 Elsevier.)
F igure 20. First operating APTE upconversion pulsed
laser-pumping schemes in Yb-Ho and Yb-Er couples.
(Reprinted with permission from ref 218. Copyright 1971
American Institute of Physics.)
F igure 21. Upconversion APTE pumping in an Er3+laser
at 2.8 µm. (Reprinted with permission from ref 219.
Copyright 1988 American Institute of Physics.)
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 159
obtained in LiYF4under laser diodepumping at 0.791
µm at 40 K;220this was followed by obtaining the
laser blue emission at 0.470 µm with a 10 mW
threshold under the same conditions.221
Upconversion, attributed totheESA type, has even
provided violet emission at 0.380 µm in LaF3:Nd3+
by summing an IR and a yellow laser pump or two
yellow dye laser pumps,222however, only at low
temperature (at 90 K). Low temperature has been a
general drawback for CW upconversion lasers. In
solids, a three-level scheme at room temperature
generally becomes a quasi-four-level scheme at lower
temperature. This explains why the first upconver-
sion lasers were working in a pulsed mode at room
temperature and continuous operation could only be
obtained at lower temperature. This drawback has
recently been overcome by using glass fibers or/and
higher density laser pumping, so providing ground-
state quasi-saturation. In particular, the versatile
tunable Ti-sapphire laser has helped a lot in this
In recent years, Huber’s group in Hamburg has
obtained many new upconversion laser results. After
an initial pulsed laser operation in KYF4:Er at 0.551
µm at room temperature,223the CW functioning has
been reached,224in both cases with an attributed ESA
upconversion pumping by a Ti-sapphire laser; for
CW operation, the laser threshold has, however, the
large value of 2.2-2.4 W. On the other hand, a very
efficient CW laser effect has recently been achieved
in a 3 mm long single crystal of LiYF4:Yb(3%)-Er-
(1%); the pumping at 0.966 µm at a level of 1.6 W
was provided by a Ti-sapphire. An upconversion
laser threshold of 418 mW was obtained, and the
nature of the pumping was proved tobe of the APTE
type both by the laser excitation spectrum analysis
replicating the Yb absorption spectrum and by a
nonlinear narrowing as described in section 3.2. The
laser useful output at 0.551 µm is 40 mW with 5%
Research on upconversion-pumped laser materials
continues to be active, and groups of laboratories
have considered GdAlO3, LiGd(MoO4)2, Y3Sc2Ga3O12,
and Gd3Ga5O12, doped with Yb, Tm, not in the usual
scheme with Yb pumping, but with Tm excitation by
the3H6to3H4absorption with forth and back transfer
with Yb.154The conclusion is that for
emission at 480 nm, the Y3Sc2Ga3O12host would be
better than the LiYF4one.
Because of their inherent small core diameter,
glass fibers easily allow one to obtain high pumping
density over long lengths. Such high densities over
long lengths cannot be provided by any lens focusing
system. Ground-statedepletion of any doped fiber can
beeasily reached with less than 100 mW pumping.227
In particular, fluoride fibers favor anti-Stokes lasers
for three reasons:
(i) the existence of long-lived metastable states
linked with the low-energy phonons of the fluoride
(ii) ground-state saturation allowing a CW laser
functioning even in a three-level laser energy scheme;
(iii) the advantage of a nonlinear pumping linked
again with the optical confinement of the fiber
Since the first demonstration in 1986228of the
feasibility of CW room-temperaturethree-level lasers
in the Er3+-doped glass fibers, one could think that
an upconversion-pumped three-level scheme could
also be used in CW operation at room temperature
for Er3+emission at 540 nm.132After preliminary
results, obtained first at 77 K with Tm3+-doped
fluorozirconatefiber, lasing at 455 and 480 nm,229the
very first CW, room-temperature upconversion laser
was demonstrated at CNET by Allain et al.230in a
three-level scheme of the Ho3+-doped fluoride fiber
laser (Figure 22). Because of the weak Ho3+concen-
tration (1200 ppm), it was believed that within the
single-ion level system ESA was taking place. How-
ever, since clustering with subsequent APTE (ETU)
effect may sometimes occur at much lower concentra-
tion (70 ppm in ref 71), some doubt is cast about the
effective pumping process, as in many of the subse-
quent upconversion-pumped fiber lasers.
Er3+-doped glass fibers have alsoshown CW room-
temperature three-level laser emission at 540 nm
when pumped at 801 nm.231Because the pumping
wavelength is in the diode laser range, there was
some hope that a compact fiber laser could be
Besides these two-photon upconversion-pumped
lasers, a three-photon pumped one has been demon-
strated in a Tm3+-doped ZBLAN fiber:232pumping at
1.12 µm, a room-temperature CW laser emitting at
480 nm, with a differential efficiency of 18%, has been
obtained with the rather low threshold of 30 mW.
Pr3+-doped fluoride fibers, because of their low
phonon energy with respect to Pr3+-emitting level
energy differences, have allowed CW room-temper-
ature anti-Stokes lasers at blue, green, and red
wavelengths in a singlefiber.233Moredetailed results
on such upconversion laser recent evolution may be
found in ref 234.
However, before closing this section, it is worth
mentioning the possibility of upconversion laser with
multiphonon pumping in theelectronic rare-earth ion
sideband transitions mentioned in section 3. It pre-
sents theadvantageof theself-adaptation of theESA
absorption to a single pump wavelength. Upconver-
sion pumping is successful through multiphonon
sideband pumping with energy mismatches as large
F igure 22.
upconversion Ho-doped fiber laser. (Reprinted with permis-
sion from ref 230. Copyright 1991 Institute of Electrical
and Electronic Engineers.)
Energy scheme for the first visible CW
160 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
as 1000 cm-1. This is the case for the Tm3+fluoride
fiber laser pumped at 1.06 µm in a three-photon ESA
process and lasing at 1.47 µm.235This provides an
efficient four-level scheme laser that allows CW
oscillation with a differential efficiency of 27% at
room temperature. On the other hand, this opens the
door tolong (kilometer) distributed amplifiers where
the losses per unit length would be compensated at
each point by the gain; the transmission line would
then be turned into a lossless line.
Multiphonon pumping is also one of the processes
involved in the avalanche upconversion process de-
scribed in the next section.
5. Cross-Relaxation and the Photon Avalanche
The most recently discovered upconversion process
is the photon avalanche effect.11Since it has not been
considered in the review of 1973,2more detailed
attention will be paid to it here.
While looking for two-step absorption (ESA) in
Pr3+-doped LaCl3and LaBr3at low temperature(<40
K) as a means to detect an IR photon by its energy
summation with a more energetic photon (IRQC) so
performing excited-state absorption (ESA), it was
found that the higher energy photon alone could, in
the same time, give rise to upconversion and reduce
the transmission of the sample above a given inten-
sity threshold;11see Figure 23. The effect was at-
tributed to an increase of population of an excited
state resulting from a cross-relaxation process. The
starting process was initially not completely deter-
mined. In the Pr3+case, the3H5
initially very weak at low temperature because3H5
is about 2000 cm-1abovetheground state(seeFigure
24); however, above about 1 mW of excitation, this
transition is increased; the cross-relaxation process
which in turn reduces thetransparency of thesample
at the (3P1-3H5) energy. Since the more the (3P1-
3H5) energy is absorbed the more the3H5population
is increased, the process was termed photon ava-
lanche.11It is clearly a way to increase ESA in a
Afterward, similar effects have been observed in
Sm3+-, Nd3+-, Ni2+-, and Tm3+-doped halide crys-
tals.236-239Recently the photon avalanche effect has
3P1 absorption is
3H5) increases the3H5 population
been obtained at room temperature for the Er3+ion
in a ZBLAN glass both in bulk and in fiber form240-243
and in a LiYF4crystal.244
The photon avalanche process is characterized by
three distinct nonlinear behaviors:
(i) transmission, (ii) emission, and (iii) rise time on
the pump power intensity with generally the exist-
ence of a critical pump threshold.
Particularly long rise times, from seconds to min-
utes,244,245have been observed.
At this point it is worth discussing the notion of
threshold for avalanche. Because of the complexity
of the phenomenon, it has been usually modeled by
a simplified three-level system.246-248
5.1. Avalanche Process as a Positive Feedback
Using the three-level simplified model of ref 246
or 247 and adding tothe initial ground-state absorp-
tion (σ1Φ ) R1) an auxiliary direct feeding into the
metastable state (σ0ΦIR) we may write the following
set of equations (see Figure 25 for explanation of
symbols which except for the trigger σ0ΦIR are the
same as in ref 246).
Being interested in the steady-state initial step of
avalanche, we assume
F igure 23.
sample under3H5-3P1pumping. (Reprinted with permis-
sion from ref 12. Copyright 1990 Elsevier.)
Decrease of transmission in a Pr3+:LaCl3
F igure 24. Simplified LaCl3:Pr3+energy scheme for the
F igure 25. Avalanche simplified three-level scheme; C is
the cross-relaxation coefficient; Wij are the spontaneous
(ii) n1) 1 - n2- n3≈ 1
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 161
Then the rate equations are simplified to
with C being the cross-relaxation parameter and b
the branching ratio, the following relationships exist
between the transition probabilities
Equation 29 can be written as
Considering amplitude and using the symbolic rep-
resentation for feedback systems, eq 30 gives block
A of Figure 26a.
In the same way, eq 27 is written
which can be symbolized in Figure 26a by block ?
and an adder with input source
Combining eqs 31 and 30 gives the classical feedback
systems scheme of Figure 26a.
Such a system is known tobe unstable for A? ) 1.
One can define a gain of the closed loop feedback
system, G, by the ratio between the green output to
a pump related input signal (R1) plus an eventual
external trigger (ΦIR)
It is well-known that in terms of the A and ? of
Figure 26a, one has
The stability condition is then written as
Its limit is just the threshold condition given by
J oubert et al.246obtained here in a simplified way as
with C > bW3for a positive feedback.
The behavior of our feedback system below thresh-
old can be described by the behavior of G(R2), Figure
The feedback black box approach has also been
considered in studying the dynamics of the above
three-level system.249It is based on the fact that the
general feedback linear theory may solve algebra-
ically time variable differential equation systems by
using the Laplace transform of the time-dependent
5.2. Conditions in Order To Observe an
Neglecting the first nonresonant absorption step
(R1) and taking intoaccount only thesecond resonant
absorption step (R2) when calculating the population
of the third level (n3) versus R2(the pumping excita-
tion) leads toa well-defined nonlinearity in n3for the
asymptotic curve (R1/R2) 0), as shown in Figure 27.
When the first step (R1) is explicitly taken into
account,248thethreshold nonlinearity is progressively
smoothed out while increasing the ratio R1/R2 as
F igure 26. (a) Positive feedback model for avalanche. (b)
Gain behavior of the model versus R2; positive feedback
condition is b > CW3; the asymptote R2ccorresponds tothe
0 ) -R1- σ0ΦIR+ bW3n3+ W2n2- Cn3
0 ) (1 - b)W3n3- W2n2+ 2Cn3+ R1+
0 ) R2n2- W3n3- Cn3
(1 - b)W3) W32;bW3) W31; W3) W32+ W31
W3+ C) n3
+C - bW3
F igure 27. Third-level normalized population, R ) n3-
(?,R2)/n3(? ) 1,R2) R2c), versus pumping term (R2) with ?
as a parameter. (Reprinted with permission from ref 248.
Copyright 1995 Elsevier.)
1 - A?
C - bW3
C - bW3
162 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
shown in Figure 24. This corresponds to a progres-
sively more resonant first step. As shown by Goldner
and Pelle ´,248practically a clear avalanche threshold
can be expected only for R1/R2e 10-4
Some of the features of the avalanche effect have
been observed at room temperaturein Tm3+:YAlO3239
and in Pr3+in silica glass fibers.251The lack of a clear
threshold in these two systems can certainly be
related to the above prediction.
The region in Figure 24, where 10-4< R1/R2< 1,
corresponds to cases for which the losses in the
feedback loop may exceed the loop gain for R2values
below R2C, so that after a number of loopings of the
excited population between level 3 and the meta-
stable level 2, the system would neither diverge nor
maintain n3 independently of R1. Such cases have
been called a looping mechanism.252We believe that
some of the reported cases of quasi-zero threshold
avalanche cases in the literature251,253belong tolarge
10-3< R1/R2< 1 cases for which, as shown in Figure
24, it is very difficult to distinguish between ava-
lanche and sequential two-photon absorption (ESA)
which occurs strictly for ? ) 1. Sometimes such cases
have been termed quasi-avalanche.248This interme-
diate behavior has been well studied in YAlO3:Er3+
under 790-810 nm excitation.254The three kinds of
upconversion, ESA, APTE(ETU), and quasi-ava-
lanche, are simultaneously found to exist according
to the precise excitation wavelength. For a 796 nm
excitation, the blue-green upconversion is attributed
for 23% to ESA and for 77% to the looping effect
Er3+, with (R1/R2) = 10-6, has shown at room
temperature all three characteristic features of ava-
lanche when doping a LiYF4 crystal or a ZBLAN
fluoride glass both in bulk and in a fiber shape (see
sections 5.3 and 5.4); even a long delay of several
seconds to a minute was observed.242-244For com-
parison, the following values for the critical param-
eter (? ) R1/R2) havebeen found for Nd3+-LiYF4, (R1/
R2) )1.7 × 10-4for avalanche at T ) 40 K;237,246for
Tm3+-Ho3+-Gd2Ga5O12, (R1/R2) ) 3.6 × 10-2for the
two-ion looping process;255for a Tm3+-BIGaZYTZr
glass, (R1/R2) ) 1.2 × 10-2 256for what was claimed
to be avalanche at 100 K.257In this last case since
the delay reaches only 16 times the metastable state
lifetime (W2-1), it looks more like a looping process
case. In Er3+, as can be seen from sections 5.3 and
5.4, the avalanche delay reaches 6 × 102to104times
W2-1, respectively, for Er-doped fluoride glass and
For Tm3+ions, in divalent fluorides, SrF2, CaF2,
BaF2, and CdF2, avalanche has been studied258in a
red toblue upconversion scheme. Avalanche is char-
acterized by the slow build-up of the signal and the
spatial spreading. In Y2SiO5:Tm3+,259avalanche, as
shown by a kick in the output slope, n, for 100 mW
excitation, is believed toexplain1G4emission, whereas
In a ZBLAN glass, doped either with (0.1-5%)Ho3+
alone or with Ho3+and (3%)Tm3+,260clear avalanche
threshold is obtained at about 140 W/cm2of 585 nm
laser excitation for the codoped sample, at both 77
1D4 is attributed to ESA and
and 300 K. Establishing times, which are the signa-
tures for avalanche, are, respectively, 250 and 30 ms.
Upconverted emission from Ho3+is at 545 nm from
5S2/3F4. The nonresonant first step is in a mul-
tiphonon sideband and for the second resonant step
in static pair levels induced on Ho3+by Tm3+as
reveled by the excitation spectrum.
As for other types of upconversion, the advent of
research with d ion-doped crystal has shown that
avalanche could alsobe obtained with such ions. Cs2-
avalanche process: slowing of the upconversion
establishment from a few milliseconds up to 1.5 s at
a threshold of 3.8 mW and nonlinearity (higher slope)
in the slope 2 line of the two-photon upconversion
process when excited asidetheresonant ground-state
absorption. From the information given, one can
estimate the R1/R2ratio to be in that case 5 × 10-3.
This is weak in principletoshow a marked threshold,
yet one can compute that the number of feedback
loops is about 1.2/1.6 × 10-3) 750, which is as large
as in an Er-doped ZNLAN glass (see above). In Cs2-
ZrCl6:(1%)Os4+,189signs of avalanche with a weak
threshold of 2.6 mW have alsobeen found from 15 to
50 K. The fitted R1/R2ratiois given tobe 3.3 × 10-3,
which alsois weak for a marked threshold. Here the
number of looping cycles is only about 0.5/20 × 10-3
) 25. This is a quasi-avalanche or looping effect.
261has shown clear signs of an
5.3. Er3+−LiYF4as an Avalanche Model
In the case of Er3+, the first step for photon
avalanche has been clearly identified and attributed
toanti-Stokes multiphonon sideband absorption240,241
(see Figure 28). Calculating the R1/R2 ratio from
mutiphonon absorption allows one to estimate a
value of 5 × 10-6,243,244as shown in Figure 29, which
displays the multiphonon sideband absorption in the
avalanche excitation region. As observed, this ex-
perimental situation provides a marked threshold
behavior in the erbium case.240
The simple theory of section 5.1 has been verified
by experimentally measuring G(R2). This was done
using the following method (see the experimental
F igure 28. Energy scheme of Er3+, and principal mech-
anisms responsible for photon avalanche cycles under
excitation at 579 and 690 nm.
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 163
setup in Figure 30): having obtained a given green
output for a pump R2 with ΦIR ) 0, R2 is reduced
while increasing ΦIRin order tomaintain a constant
green output. ΦIRis an infrared signal at 0.94 µm in
resonance with the metastable state 2 (here4I11/2)
absorbing with a cross-section σ0.
Each point is obtained after waiting for a steady
state. Because of the large ratio for σ0/σ1, this
experiment provides a good description of G(R2), as
shown by the results in Figure 31 for three temper-
atures. One can define an R2Casymptote only at 300
and 220 K, respectively, 120 and 240 mw. At 163 K
one cannot reach an asymptote (threshold) in that
experiment at the maximum available power of 250
mw. Thus, the effect of lowering temperature is
essentially to increase R2C.
The part in R2Cwhich is most sensitive to temper-
ature is C because it is related tothe phonon energy
of only 100 cm-1, whereas W2and W3are related to
phonons covering the energy gap below levels 3 and
2, that is energies >2000 cm-1. However, this re-
quires C to be of the same order as W3 or bW3;
otherwise, as long as C . W3, bW3, one has R2C= W2
and its temperature dependence is just the same as
Comparing the theoretical threshold as given by
eq 35 with experimental conditions, one can verify
the simple feedback model.
Assuming level 1 tobe4I15/2, level 2 tobe4I11/2, and
level 3 to be the aggregation of levels between2G9/2
and4S3/2 with the emission properties of4S3/2 (see
Figure 28) and taking the room-temperature data
given by ref 256 and 257, the following parameters
are found: W3) 2500 s-1; b ) 0.5; C ) 0.5 × 106s-1;
W2 ) 140 s-1, because corresponding oscillator
strengths areabout equal (0.4 × 10-6),262onecan also
assume σ2) σ0) 4 × 10-21cm2.
Using reduced population units (pure number), it
becomes R2C) 140(2500 + 5105)/(5105- 1250) ) 141
s-1= W2(at room temperature) from which it can be
At 0.578 µm it gives, for a 50 µm diameter spot, a
threshold power of Pth ) 222 mW; this value is of
thesameorder as threshold values observed for 0.578
G(R2) shows (Figure 26b) that the Er3+-doped solid
constitutes a marginally stable positive feedback
system: even below the R2C asymptote, it is known
from feedback theory that a strong input signal can
drive a system that is otherwise stable into its
instability state(existenceof a gain stability margin).
To experimentally verify this behavior, a pulsed
trigger of amplitude σ0ΦIR/W2is added to the input
signal R1/W2 given by the pump; the experimental
setup is again the same as that presented in Figure
The results at room temperature are given in
Figure 32a-c. In the absence of a trigger, with Pp )
114 mW at 578 nm incident on sample, the threshold
is reached after a very long time(>50 s) (Figure32a).
With the same pump intensity (Pp ) 114 mW) and
with a short trigger of 0.6 s, the avalanche state is
obtained quickly and maintained after trigger extinc-
tion (Figure 32b). With the same trigger but with a
reduced pump (Pp ) 99 mW), the avalanche state
cannot be reached. This behavior, as depicted in
Figure 32a-c, is obtained down to 180 K.
Below 180 K, the observed threshold increases as
shown in Figure31. However, duetothetemperature
scan cycles relatively short time constant (3 s/K from
10 to 50 K, then 21 s/K from 50 to 150 K), it is not
sure whether or not the threshold could be reached
for an avalanche delay time >50s.
From this experiment, it is understood that mea-
suring an avalanche threshold depends on the time
one is ready towait before its observation. This time
depends not only on the excited-state pumping but
alsoon theground-stateabsorption conditions. In any
F igure 29. Absorption cross section for Er3+-LiYF4taking
into account the multiphonon contribution; the heavier
lines show the anti-Stokes zones which contribute to W1
for the avalanche processes in erbium.
F igure 30.
positive feedback gain and for an external triggering.
Experimental scheme for measuring the
F igure 31. Experimental G(R2) for three temperatures:
30, 220, and 163 K; residual signal near R2) 0 comes from
direct upconversion under 0.94 µm excitation.
Φthreshold) 141/4 × 10-21) 3.5 × 1022s-1cm-2
164 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
case, below 180 K, being then limited by the pump
laser at a much lower power than the threshold, the
result of Figure32c is obtained. This triggering effect
constitutes an optical analogue of a thyratron provid-
ing an intrinsic material-based optical bistability.
The time delay behavior of the avalanche process
has been studied theoretically quite recently within
the general model of Landau for phase transition.256
The time delay at threshold, in fact a critical slowing
time, tc, proved itself tobe the most sensitive experi-
mental data when looping or avalanche takes place.
It has been shown to be given by256,263
where K depends on other spectroscopic parameters.
Equation 36 is rather well verified in the above
experiments for which avalanche delay times have
been determined for two different excitation wave-
lengths of known multiphonon anti-Stokes cross
sections: at λ ) 688 nm, with σ1) 10-24cm2, delay
is found to be 0.4 s; whereas at λ ) 579 nm, with σ1
) 2 × 10-26cm2, the observed delay is 4 s. Assuming
for σ2the same value in both cases of excitation, the
ratioof the delays is 0.1, which is in agreement with
the value of 0.14 as given by eq 36.
Above the threshold, the delay for avalanche has
been given by ref 264 as
with k ) W3-1/(1 + b)(2C + W3) and where Φc) Rc/2
is the pumping flux at threshold and Φ the effective
5.4. Photon Avalanche in Er3+−Fluoride Glasses
in Fiber and Bulk Shape
Recently, the photon avalanche effect has been
observed at room temperature in a Pr3+-doped silica
fiber251and in an Er3+-doped fluorideglass fiber.241,243
In the first case, only a nonlinearity of the transmis-
sion is observed and not the upconversion emission
threshold. It was believed that the threshold was so
low that it could not be observed.We think that this
is explained by the toolarge nonresonant toresonant
absorption ratio as mentioned in section 5.2. On the
other hand, in the second case, clear thresholds at 5
and 4 mW of incident power at, respectively, 579 and
690 nm243are observed because in these last two
cases thefirst step is a weak anti-Stokes multiphonon
absorption giving again a R1/R2ratio of about 10-6,
much below the critical value of 10-4. The involved
energy scheme for both excitations is essentially the
same as that in Figure 28; it shows both pumping
routes and the two types of involved cross-relax-
ations. In Figure 33, the typical threshold behavior
for theavalancheupconversion emission is presented.
The long delay behavior is displayed in Figure 34,
showing, near threshold, the very long time (3.5 s),
widely in excess of any of the lifetimes of the
metastable states of erbium. The observed delay
follows rather well the behavior predicted by eq 37.
As for glass fibers, thesameresults can beobtained
in bulk samples.240,241Because the first absorption
step, being of a multiphonon nature, is featureless,
the excitation spectrum for avalanche directly pro-
F igure 32.
without (a) and with (b) a trigger feeding the metastable
state and (c) much below threshold.
Avalanche behavior just below threshold
F igure 33. Upconversion emission at 550 nm showing the
existenceof theavalanchethreshold in a ZBLAN:Er3+fiber
observed from its extremity.
F igure 34. Time delay for the avalanche establishment
versus incident pump power at 579 nm in a ZBLAN:Er3+
-1/(Φ/Φc- 1) (37)
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 165
vides the ESA spectrum of the resonant second
absorption with a single excitation beam as shown
in Figure 35 for the4I11/2-2G9/2transition of Er3+; this
gives a new method242toreach ESA spectra otherwise
difficult to obtain without a double-beam excitation.
The main difference with a bulk sample is that in
the fluoride fiber case upconversion spatial domains
appear with periodic structures with periods ranging
from a few centimeters to millimeters themselves
containing substructures with a period of about 100
µm,241,243as shown in Figure 36. This behavior has
been explained by the high contrast provided by the
avalancheeffect; it then optically reveals theinternal
electric field mode structure of the fiber wave-
5.5. Avalanche in Codoped Systems
Up to this point, avalanche has been described
within a singletypeof ions. As far as energy transfers
in general are considered, the sensitizers and the
acceptors may be either of the same or of a different
nature. Thus, in the avalanche process, instead of
having two ions of the same nature participating in
the cross-relaxation step, as described to date, the
cross relaxation has also been considered between
two ions of a different type. This was the case of the
first study by Brenier et al. of theavalancheinvolving
Tm3+-Ho3+.255A complex process of cross-relaxation
within two Tm3+ions together with a back transfer
from Ho3+to Tm3+has been proposed. The first
excitation step is out of resonance with the3H6-3H4
absorption of Tm3+; it is followed by a cross-relaxation
within twoTm3+ions; which populates the3F4(Tm3+)
state, itself transferring its energy to the5I7(Ho3+)
state from which the resonant second step of excita-
cross-relaxation (energy back-transfer) between Ho3+
and Tm3+of the type3H6(Tm3+) +5S2(Ho3+) w3H4-
(Tm3+) +5I7(Ho3+) provides the feedback loop, see
Figure 37. However, the avalanche threshold is not
reached, and calculating theR1/R2ratiofrom thedata
in ref 255, the critical parameter
5S2(Ho3+) takes place. Then another
is determined. This clearly confirms, what the au-
thors have found, that the behavior of this system is
in thelooping region of Figure27. Thelooping process
is different from avalanche in the sense that it has a
reversible character that real avalanche does not
have. In the case of the codoped Yb-Pr system, real
avalanche has been reached.266In this system, see
Figure 38, the first nonresonant excitation step is in
the Yb3+ion, which transfers tothe metastable state
1G4(Pr3+) from which the resonant second step up to
3P0(Pr3+) takes place. Thecross-relaxation step within
twoPr3+ions feeds themetastablestate1G4(Pr3+) and
the back transfer to Yb3+again feeds the1G4(Pr3+)
metastable state. The ratio ? from the nonresonant
to resonant pumping was estimated to be from 10-6
to10-8 266, well within the condition toobserve a real
In YalO3:(10%)Yb, (1%)Ho,267a green emission from
5S2(Ho) is obtained under 750 and 840 nm excitation.
Yb plays a role in the back transfer which helps to
populate the metastable state
nonlinearity found in the output signal slope is an
indication of a looping mechanism.
The Ho-Tm codoped ZBLAN glass system260ob-
served in section 5.2 does not seem to enter into the
5I6 (Ho). A slight
F igure 35. Excitation spectrum for the avalanche emis-
sion at 549 nm in a ZBLAN:Er3+glass showing the
spectrum for the4I11/2-2G9/2ESA transition.
F igure 36. Spatial domains observed along thefiber above
the photon avalanche threshold: (a) dot separation about
1 mm. (b) Microscope view of a 1 mm avalanche dot. The
scale is 100 µm per division.
F igure 37. Energy scheme and mechanism for looping
effect in the Ho3+-Tm3+codoped system. (Reprinted with
permission from ref 255. Copyright 1994 Elsevier.)
? ) 6.16 × 10-22/1.73 × 10-20) 3.6 × 10-2
166 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
avalanche codoped category since the Tm ion does
not seem to play a role in the feedback loop but only
in the R2excitation term.
In analogy with the previous case, avalanche in
La1-xCexCl3:Nd3+ 268shows that for x > 0.1, Ce plays
a rolein increasing theavalancheoutput signal. This
effect is attributed to the presence of undefined pair
states in clusters.
Other systems with codoping with 4f and 5f ions
have been investigated: In LaCl3:(1%)Pr3+, (0.1%)-
U3+,269under a dye laser excitation between 615 and
617 nm, the blue emission from3P0to3H4in Pr3+is
observed at thevery low level of 2 mW and is ascribed
toa double excitation: the first one from the ground
state in U3+and the second one between excited
states in Pr3+; cross relaxation between U3+and Pr3+
provides the population of the metastable level3H6-
(Pr3+) from which takes place the resonant ESA to
3P0(Pr3+). However, becausethematching is probably
toogood between excitation and ground-state absorp-
tion, no real threshold is observed and a looping
mechanism is advocated toexplain that the presence
of U3+increases the blue output of Pr3+by 3 orders
5.6. Upconversion Laser with
Multiphonon-Assisted Pumping Scheme and
Besides pumping in the electronic RE ions transi-
tions, we have also seen that upconversion laser
pumping could be attempted in multiphonon side-
bands with energy mismatch as largeas 1000 cm-1;235
see section 4.5.
Of course, most of the avalanche-pumped lasers
alsoenter this category because of the required weak
first step absorption. It is worth mentioning the first
avalanche laser in LaCl3:Pr3+.270A CW emission in
a four-level scheme was obtained at 0.644 µm through
an upconversion avalanche process under 0.677 µm
pumping which corresponds to an ESA pumping for
the second step and probably tomultiphonon absorp-
tion for the first step, as recently observed for
amplified spontaneous emission at 0.850 µm in a
ZBLAN:Er3+fiber in an avalanche-pumped regime.271
Powerful avalanche-pumped upconversion lasers have
been obtained in a CW regime, first at low temper-
ature (7 K) in YAlO:Er3+;272,27333 mW of output at
0.549 µm has been reached with an optical efficiency
of 3.5%. Interestingly, the same crystal system may
provide laser under APTE (ETU) upconversion pump-
ing simply by tuning-in the first excitation step
(0.8069 µm). The laser threshold is then lowered
(∼100 mW) in comparison with theavalancheregime
(∼380 mW at 0.7913 µm); theoutput law is quadratic
versus pumping in the APTE (ETU) regime, while
for the avalanche regime it has a much higher power
law above threshold saturating toward a square law
at about 1 W of incident pump power. In the APTE
(ETU) regime, the laser output is 166 mW with an
optical conversion efficiency of 17%.
As already mentioned in section 4.5, fiber optical
confinement may allow one toobtain CW three-level
lasers even at room temperature. This has alsobeen
the case for avalanche-pumped lasers. A high-power
upconversion laser has been demonstrated in a
ZBLAN:Pr,Yb double-clad fiber under avalanche
pumping.274The energy scheme and avalanche pro-
cesses, which are involved, are the ones described in
the previous subsection for the avalanche process
with Yb-Pr codoping.266Laser emission is from the
3P0-3F2transition of Pr3+at 0.635 µm with a record
output of 1020 mW. The pumping is from two Ti-
sapphire lasers providing 5.51 W at 0.850 µm. This
last wavelength, being detuned from the maximum
Yb3+absorption at 0.96 µm but tuned with the1G4-
1I6 Pr3+absorption, provides the condition for the
avalanche regime of a weak ? (10-6to 10-8) as seen
6. Perspectives and Future Advances
From the most recent studies and their respective
aims as described in the above sections and from
some emerging new research, one can try to derive
some future trends and perspectives tobetter under-
stand some of the observed features. In the following
they will be divided intothe five main directions that
can be anticipated:
(i) upconversion pumped lasers;
(ii) new materials for low-level IR visualization;
(iii) intrinsic material optical bistability;
(iv) hot emission and avalanche-like codoped sys-
(v) biological applications.
6.1. Upconversion UV-Tunable Lasers
Because optically pumped lasers are originally
based on a Stokes pumping process, one basic prob-
lem is to obtain a high-density pumping source at a
shorter wavelength than their emitting wavelength.
With the tendency to go all solid state, there will be
a continuous race between the shortest wavelength
of the pumping semiconductor-based source and the
shortest solid-state laser-emitting wavelength.
Few years ago the anticipated pumping diode
emitted only in the near-IR; now that blue semicon-
ductor lasers have begun toappear, one can theoreti-
F igure 38.
codoped Yb3+-Pr3+avalanchesystem: pumping at 0.85 µm
into Yb3+sideband and in resonance by ESA into Pr3+;
cross-relaxation within Pr3+; Yb3+-Pr3+cross-relaxation;
Pr3+-Yb3+back-transfer. (Reprinted with permission from
ref 266. Copyright 2000 Elsevier.)
Energy scheme and mechanisms for the
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 167
cally anticipate that, sometimes, blue diodes could
pump visible solid-state lasers. However, because
powerful red semiconductor lasers do not exist yet
that could be used for solid-state laser pumping, one
can also anticipate that for many years to come no
UV CW semiconductor or even frequency-multiplied
semiconductor laser will not exist that will be power-
ful enough tousefully pump tunable a CW UV solid-
Clearly, upconversion pumping of a tunable UV
solid-statelaser has a roletoplay. Somuch thebetter
that presently availablegas UV laser arenot tunable.
Recent results give hints for that. As already shown
in section 4.5, UV emission has been obtained from
LaF3:Nd3+at 380 nm through a somewhat compli-
cated anti-Stokes pumping scheme.223On the other
hand, a very efficient tunable UV laser from LiLuF4:
Ce3+has recently been obtained from 305 to333 nm
under Stokes pumping provided by a frequency-
doubled copper vapor laser;275theslopeefficiency was
51%. This particular result has to be connected to
other recent results276on upconversion and energy
transfers from Pr3+(4f-5d) toCe3+(5d) tosee that a
tunable UV laser by anti-Stokes pumping can be
imagined. The recently issued collection of selected
papers on upconversion lasers232certainly comforts
6.2. New Materials for Low-Intensity IR Imaging
As pointed out in section 4.3, an interesting point
of view is the IR-IR upconversion for low-intensity
IR viewing. Though it is basically the old IRQC
concept toupconvert IR out of thephotocathoderange
to the photocathode detection range (GaAs and S1,
from 760 to 1200 nm), little research has used the
morerecent understanding of RE-doped materials for
this application. It has been recently proposed touse
low-phonon energy glasses such as chalcogenide
glasses (GeS2-Ga3S3) as 1500-1000 nm converters,
with most of the upconverted energy at the latter
wavelength.277Of course, other low-phonon energy
materials such as the stable CsCdBr3which is well-
known in this respect together with strong RE-RE
pair coupling could be designed for still farther IR
6.3. Upconversion Material Intrinsic Bistability
This is certainly the most recent fundamental
subject in the upconversion field. A few years ago it
was found in Gu ¨del’s group278,57that the cooperative
luminescence(seesection 3) of Yb3+:Cs3Y2Br9andCs3-
Yb2Br9 in the visible as well as the usual Stokes
emission in the IR clearly displayed an hysteresis
loop under variable excitation density and fixed
temperature between 11 and 31 K. Correlatively a
hysteresis loop was also observed when excitation
was kept constant and the temperature was varied
as shown in Figure 39. The physical explanation was
given in the framework of the optical coherent field
coupling between two ions in a solid that had been
theoretically proposed279,280by Heber. The problem
with this model is to what extent it can describe a
RE-doped system physical reality. This can be ques-
It has been recognized for a long time now that
optical bistability within a cavity could originatefrom
optical cooperative coupling between cavity modes
and the atomic system in a way similar tothe optical
cooperative coupling between atomic systems giving
rise to superfluorescence.281However, superfluores-
cencein RE-doped systems is very difficult toobserve
because it requires long dephasing times (T2) which
are impossible to obtain in a solid at high doping
concentration due toion-ion and ion-phonon dephas-
ing. Observation of superfluorescence in LiYF4:Er
requires both low T (<30 K) and very weak concen-
tration (<0.3%)282and a threshold of 842 W cm-2. The
rather high concentration used here in the Yb3+
experiment certainly imposes that theRabi frequency
be less than the ion line width at the considered
excitation density of 800 W cm-2, which is alsoabout
thethreshold for theobserved hysteresis loop.278This
forbids any sizable coherent field coupling.
Interestingly, in thevery first experiment on APTE
upconversion, thermal hysteresis loops wereobserved
at high temperature,283see Figure 40. Clearly, at
such high temperatures it has nothing to do with
coherent field coupling. Also, the bistability observed
F igure 39. Thermal hysteresis loop for cooperative lumi-
nescence in Cs3Y2Br9:Yb3+. (Reprinted with permission
from ref 284. Copyright 1996 American Institute of
F igure 40. First observation (from 1967 unpublished
original document) of a thermal hysteresis loop for APTE
upconversion in WO4Na0.5Yb0.5:2%Er3+. Arrows show direc-
tion of the temperature variations for two sets of experi-
168 Chem ical Reviews, 2004, Vol. 104, No. 1Auzel
in avalanche systems (see section 5.3) has been
described within the framework of population rate
equations. Then it appears more in accordance with
physical reality not toconsider coherent field coupling
as the root of observed bistability. This is implicitly
recognized for Cs3Y2Br9:Yb in a later paper284point-
ing out the analogy with avalanche bistability and
describing the effect through population rate equa-
tions. As the temperature changes, sodothe overlap
integrals ruling the RE-RE energy transfers, which
provide the necessary nonlinear feedback effect.
Clearly, it can also be the only explanation for the
high-temperature (430 K) hysteresis loops that had
been initially observed in WO4Na0.5Yb0.5:2%Er.283
From this result and the fact that the hysteresis
behavior is produced at the Yb3+ion, it can be
predicted that all upconversion systems with Yb as
a sensitizer could show thermal hysteresis. For the
last 2 years, thethermal explanation has clearly been
retained as explained by Gamelin et al.285and said
to be due to the variable thermal absorption proper-
ties of Yb3+. It is described as a thermal avalanche
with a thermal cross-relaxation in analogy with the
photon avalanche described above in section 5.1.
According toFigure41, theequivalent of R1, theweak
ground-state absorption, is a nonresonant absorption
between2F7/2to2F5/2. The resonant absorption term
R2 is from a Stark level of2F7/2 to a Stark level of
2F5/2; the cross-relaxation term, Cn1n3of Figure 25,
is produced by the heat released within the Stark
levels of2F5/2by the phonon emission, which in turn,
by absorption of phonons, populates a higher Stark
excited state of2F7/2. This is the loop of this thermal
avalanche. An external thermal triggering term,
equivalent to σ0ΦIR of Figure 25, is provided by the
external temperature variable T of the experiment.
As seen in section 5.1, one can predict that the
threshold and the hysteresis will be steeper and the
time constant longer for weaker R1/R2 ratios. Con-
sequently, we can propose here that a large crystal
field should be better for higher temperature obser-
vations and that it should be the case for hard oxides
and YAG in particular.286
At any rate, the thermal avalanche convincing
explanation certainly describes in a correct way the
published observations including the one of 1967,283
which was was thought to be due to the thermal
behavior of theoverlap integral between coupled Yb-
Er ions and had been unexplained until now!
6.4. Hot Emission and Avalanche Like Co-Doped
Here, it is interesting to discuss a not yet com-
pletely elucidated new phenomena recently observed
by Bednarkiewicz and Strek287in an upconversion
study of Nd3+-Yb3+codoped YAG nanocrystallite
ceramics. Under laser diode pumping at 976 nm into
theYbabsorption, visibleorangeantiStokes emission
is observed at 300 K with broadened features at 579
nm from4G5/2-4G7/2, 690 nm from4F9/2, 757 nm from
4F7/2-4S3/2, and 813 nm from4F5/2-2H9/2, all transi-
tions tothe Nd3+4I9/2ground state. These emissions
decrease with decreasing temperature. Those visible
emissions are described by a, Pn, law for output
versus excitation with, n linearly depending on the
energy gap above4F3/2as shown in Figure 42 There
is alsoan establishing time constant increasing with
the order parameter, n, reaching 1.5 s for n ) 4. It
was recognized that because the metastable charac-
ter of4F3/2 is reduced by back transfer to Yb, the
multiphonon process shown which could have ex-
plained the result of Figure 42 cannot be retained;
moreover, dividing the energy gap between4F3/2and
emitting states by n provides virtual phonon energies
not existing in YAG.
No real explanation is presented in ref 287, and
this result is still a question. Though not mentioned
by theauthors, wethink, however, that thelong time
transient is the clear signature of avalanche pro-
cesses which have yet to be analyzed in detail.
6.5. Biological Applications
Very recently upconversion applications of the
APTE (ETU) systems Yb-Er and Yb-Tm have been
devised by Zilmans et al.288for detection of cell and
tissue surface antigens as luminescent bioassays.
Submicrometer-sized phosphor crystals (200-400
nm) of the usual oxysulfide, fluoride, gallate, and
silicate types doped with Yb-Er and Yb-Tm couples
are considered. The main advantage is that IR-
upconverting phosphors are excited by wavelengths
that cannot excite the natural biological materials,
soproviding a better detection contrast with respect
toautofluorescence than the more usual luminescent
bioassays working in the Stokes emission mode. The
F igure 41. Yb3+simplified energy scheme according to
the simplified three-level energy scheme of Figure 25 for
avalanche with a thermal cross-relaxation process explain-
ing the thermal hysteresis loop of 285. The external
triggering term corresponds to the temperature variable
of the experiment.
F igure 42. Upconversion power law indexes versus energy
gap between emitting states and4F3/2(Nd3+) in Yb,Nd:YAG.
(Reprinted with permission from ref 287. Copyright 2002
Institute of Physics Publishing.)
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 169
upconversion method overcomes many of the limita-
tions of the common reporters used in immunocy-
A still more recent result that could be connected
toprevious application is the dissolution of nanopar-
ticles (6-8 nm) of Yb-Er- and Yb-Tm-doped LuPO4
as colloids in chloroform solutions.289Because of the
inherent high efficiency of the APTE (ETU) effect,
such colloids can show green, red, and blue upcon-
version in the liquid phase for the first time.
The general principles of upconversion have been
presented in a self-contained way together with
typical examples. Because these effects are now so
generally observed with the general use of laser
excitations, it was thought tobe important todistin-
guish them in precise ways in order for future
researchers tostart from well-established definitions
and to speak a common language.
Besides a didactical approach, I tried to present
most of all the recent important results if not in an
exhaustive way at least in a complete way for all
important turning points.
If therewas somegeneral philosophy toderivefrom
this review, it would be that upconversion is an
endless field and that some features are becoming
as common as plain Stokes luminescence. Some
aspects of this reviewed field though not really
exploited at some time may become important with
more refined experiments and availability of new
technologies. Also, the implied processes may help
understand other aspects of optical processes in RE-
An example could be the presently considered
photon-cutting effect,84-88just the opposite of APTE
(ETU) upconversion, which may open the way tonew
efficient lighting systems. Theoppositeof cooperative
luminescence, cooperative quenching, recently dis-
covered, may explain some of the yet not understood
features of concentration quenching.83
From an applied point of view, it is observed that
with the general use of lasers and the easiness in
observing visible to the naked eye upconversion, too
few people have found it necessary to measure
efficiencies in order to be able to compare quantita-
tively the various proposed upconversion systems.
This should bedonetopush upconversion beyond the
qualitativeapproach that still toooften characterizes
it. Most of the recently proposed systems can be
observed only at low temperature and no efficiency
values are provided. One can alsoverify through this
review that, as is often in science, the most efficient
systems aretheones discovered at first, heretheYb-
Er and the Yb-Tm systems.7
It is a pleasure to acknowledge a number of
researchers who have kindly sent me their reprints
and havesohelped mein writing this review. I would
like to mentioned particularly Dr. J unichi Owaki
(NTT), Pr. Gu ¨nter Huber (Hambourg University), Pr.
J ohann Heber (Darmstadt University of Technology),
Drs.Valery Smirnov and Alina Man’shina (Russian
Center for Laser Physics), Pr. Georges Boulon and
Dr. Marie-France J oubert (Universite ´de Lyon), Pr.
Hans Gu ¨del (Bern University), Dr. Markus Pollnau
(Lausanne University), Pr. Wieslaw Srek (Low Tem-
perature Physic Institute, Wrawclaw), and Pr.
J oaquim Fernandez (Universidad del Pais Vasco,
Bilbao). I would like to thank also Dr. Marco Betti-
nelli (University di Verona, Italy) for kindly pointing
to me the bioassay application and Pascal Gerner
(Bern University) for providing me with the very
recent last reference. Many thanks also for my good
friend Peter Lewis for reading over the whole text.
Last but not least, thanks are due tomy wife, Odile,
for having accepted that I divert a lot of leisure time
for that work.
(1) Leverenz, H. W. Introduction to Luminescence of Solids; Dover
Publications: New York, 1968.
(2) Auzel, F. Proc. IEEE 1973, 61, 758.
(3) Wright, J . Up-conversion and excited-state energy transfer in
rare-earth doped materials. In Radiationless Processes in Mol-
ecules and Condensed Phases; Fong, F. K., Ed.; Topics in Applied
Physics; Springer: New York, 1976; Volume 15, p 239.
(4) Bloembergen, N. Phys. Rev. Lett. 1959, 2, 84.
(5) Brown, M. R.; Thomas, H.; Williams, J . M.; Woodwards, R. J . J .
Chem. Phys. 1969, 51, 3321.
(6) Auzel, F. J . Lumin. 1984, 31/ 32, 759.
(7) Auzel, F. C. R. Acad. Sci. (Paris) 1966, 262, 1016. Auzel, F. C.
R. Acad. Sci. (Paris) 1966, 263, 819.
(8) Van Uitert, L. G.; J ohnson, L. F. J . Chem. Phys. 1966, 44, 3514.
(9) Mita, Y.; Nagasawa, E. NEC Res. Dev. 1974, 33, 61.
(10) Garlick, G. F. J . Contemp. Phys. 1976, 17, 127.
(11) Chivian, J . S.; Case W. E.; Eden, D. D. Appl. Phys. Lett. 1979,
(12) Case, W. E.; Koch, M. E.; Kueny, A. W. J . Lumin. 1990, 45, 351.
(13) Auzel, F. Semiconductor Optoelectronics; Wiley and PWN: New
York and Warszawa, 1980; Chapter 10, p 233. Auzel, F. J .
Lumin. 1990, 45, 341. Auzel, F. In Luminescence: Phenomena,
Materials and Devices; Rao, R. P., Ed.; Nova Science Pub.: New
York, 1992; p 33. Auzel, F. In Nonlinear Spectroscopy of Solids;
Advances and Applications; DiBartolo, B., Bowlby, B., Eds.;
Plenum Press: New York, 1994; p 531.
(14) Huber, G.; Heumann, E.; Sandrok, T.; Petermann, K. J . Lumin.
1997, 72/ 74, 1.
(15) Gamelin, D. R.; Gu ¨del, H. U. Acc. Chem. Res. 2000, 33, 235.
(16) Hehlen, M. P.; Phillips, M. L. F.; Cockroft, N. J .; Gu ¨del, H. U.
Encyclopedia of Materials; Science and Technology; Elsevier
Science Ltd.: New York, 2001; p 9956.
(17) Gamelin, D. R.; Gu ¨del, H. U. Top. Curr. Chem. 2001, 214, 1.
(18) Auzel, F. SPIE 2001, 4766, 179. Auzel, F. In Spectroscopic
properties of rare-earths in optical materials; Liu, G. K., J acquier,
B., Eds.; Springer-Verlag: New York, 2003; Chapter 5.
(19) Selected papers on Photoluminescenceof Inorganic Solids; Weber,
M., Ed.; SPIE Milestone Series, Vol. MS150; SPIE: Bellingham,
(20) Auzel, F. Multiphonon processes, cross-relaxation and upcon-
version in ion activated solids, exemplified by minilaser materi-
als. In Radiationless Processes; DiBartolo, B., Goldberg, V., Eds.;
Plenum Publishing Co.: New York, 1980; p 213.
(21) Henderson, B.; Imbusch, G. F. Optical Spectroscopy of Inorganic
Solids; Clarendon Press: Oxford, 1989; p 445.
(22) Milne, E. A. J . London Math. Soc. 1926, 1, 1.
(23) Varsanyi, F.; Wood, D. L.; Schawlow, A. L. Phys. Rev. Lett. 1959,
(24) Auzel, F. Ann. Telecom. (France) 1969, 24, 363.
(25) Auzel, F.; Bonfigli, F.; Gagliari, S.; Baldacchini, G. J . Lumin.
2001, 94/ 95, 293.
(26) Fo ¨rster, T. Ann. Phys. 1948, 2, 55.
(27) Dexter, D. L. J . Chem. Phys. 1953, 21, 836.
(28) Kushida, T. J . Phys. Soc. J pn. 1973, 34, 1318.
(29) Pouradier, J . F.; Auzel, F. J . Phys. (France) 1978, 39, 825.
(30) Axe, J . D.; Weller, P. F. J . Chem. Phys. 1964, 40, 3066.
(31) Orbach, R. Optical Properties of Ions in Solids; DiBartolo, B.,
Ed.; Plenum Press: New York, 1975; p 445.
(32) Miyakawa, T.; Dexter, D. L. Phys. Rev. 1971, B1, 70.
(33) Auzel, F. Phys. Rev. 1976, B13, 2809.
(34) Yamada, N.; Shionoya, S.; Kushida, T. J . Phys. Soc. J pn. 1972,
(35) Inokuti, M.; Hirayama, F. J . Chem. Phys. 1965, 43, 1978.
170 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
(36) Barthem, R. B.; Buisson, R.; Vial, J . C. J . Phys. 1985, 46, C7-
(37) Yokota, M.; Tanimoto, O. J . Phys. Soc. J pn. 1967, 22, 779.
(38) Weber, M. J . Phys. Rev. 1971, B4, 2932.
(39) Grant, W. J . C. Phys. Rev. 1958, 109, 648.
(40) Auzel, F. Mater. Res. Bull. 1979, 14, 223.
(41) Snitzer, E.; Woocock, R. Appl. Phys. Lett. 1965, 6, 5.
(42) Auzel, F.; Deuschbein, O. Z. Naturfosch. 1969, 24a, 1562.
(43) Ovsyankin, V. V.; Feofilov, P. P. Sov. Phys. J ETP Lett. 1966, 4,
(44) Hewes, R. A.; Sarver, J . F. Phys. Rev. 1969, 182, 427.
(45) Pollnau, M.; Gamelin, D. R.; Lu ¨thi, S. R.; Gu ¨del, M. Phys. Rev.
2000, B61, 3337.
(46) Rios Leite, J . R.; de Araujo, C. B. Chem. Phys. Lett. 1980, 73,
(47) Bonneville, R.; Auzel, F. Opt. Commun. 1976, 18, 51.
(48) Ovsyankin, V. V.; Fedorov, A. A. Opt. Spectrosc. 1981, 50, 565.
(49) Mita, Y.; Ide, T.; Katase, T.; Yamamoto, H. J . Lumin. 1997, 72/
(50) Goldner, P.; Pelle ´, F. J . Lumin. 1994, 60/ 61, 651.
(51) Auzel, F.; Pecile, D.; Morin, D. J . Electrochem. Soc. 1975, 122,
(52) Auzel, F.; Pecile, D. C. R. Acad. Sci. Paris 1973, 277B, 155.
(53) Silver, J .; Martinez-Rubio, M. I.; Ireland, T.G.; Fern, G. R.;
Withnall, R. J . Phys. Chem. 2001, B105, 948.
(54) Silver, J .; Martinez-Rubio, M. I.; Ireland G. R.; Withnall, R. J .
Phys. Chem. 2001, B105, 7200.
(55) Ovsyankin, V. V. Spectroscopy of Solids Containing Rare-Earth
Ions; Kaplyanskii, A. A., MacFarlane, R. M., Eds.; North-
Holland: Amsterdam, 1987; p 405.
(56) Orlovskii, Y. U.; Basiev, T. T.; Papashvili, A. G.; Voroev, I. N.;
Alimov, O. K.; Osiko, V. V.; Heber, J . SPIE 2001, 4766, 204.
(57) Gamelin, D. R.; Lu ¨thi, S. R.; Gu ¨del, H. U. J . Phys. Chem. 2000,
B104, 11045. Lu ¨thi, S. R.; Hehlen, M. P.; Riedener, T.; Gu ¨del,
H. U. J . Lumin. 1998, 76/ 77, 447.
(58) Auzel, F.; Dexpert-Ghys, J .; Gauthier, C. J . Lumin. 1982, 27, 1.
(59) Vial, J . C.; Buisson, R.; Madeore, F.; Poirier, M. J . Phys. (France)
1979, 40, 913.
(60) Dexpert-Ghys, J .; Auzel, F. J . Chem. Phys. 1984, 80, 4003.
(61) Nakazawa, E.; Shionoya, S. Phys. Rev. Lett. 1970, 25, 1710.
(62) Cockroft, N. J .; J ones, G. D.; Syme, R. W. G. J . Lumin. 1989,
(63) Varsanyi, F; Dieke, G. H. Phys. Rev. Lett. 1961, 7, 442.
(64) Van der Ziel, J . P.; Van Uitert, L. G. Phys. Rev. 1969, 180, 343.
(65) Stavola, M.; Dexter, D. L. Phys. Rev. 1979, B20, 1867.
(66) Livanova, L. D.; Saitkulov, I. G.; Stolov, A. L. Sov. Phys. Solid
State 1969, 11, 750.
(67) Ostermayer, F. W., J r.; Van Uitert, L. G. Phys. Rev. 1970, B1,
(68) Auzel, F.; Meichenin, D.; Pelle ´, F.; Goldner, P. Opt. Mater. 1994,
(69) Heber, J .; Nikitin, S. I.; Demirbilek, R.; Papashvili, A. G.;
Vorobev, I. N.; Alimov, O. K.; Orlovskii, Y. V.; Orlovskaya, E.
O. SPIE 2001, 4766, 218. Scha ¨fer, U.; Neukum, J .; Bodenschatz,
N.; Heber, J . J . Lumin. 1994, 60/ 61, 633.
(70) Noginov, M. A.; Loutts, G. B.; Steward, C. S.; Lucas, B. D.; Fider,
D.; Peters, V.; Mix, E.; Huber, G. J . Lumin. 2002, 96, 129.
(71) Blixt, P.; Nilsson, J .; Carlkna ¨s, J .; J askorzynska, B. IEEE Trans.
Photon Techn. 1991, 3, 996.
(72) Auzel, F. J . Lumin. 1984, 31/ 32, 759.
(73) Auzel, F. In Rare-Earth Spectroscopy; Trzebiatowska, B., Leg-
endziewicz, J ., Strek, W., Eds.; World Scientific: Singapore, 1985;
(74) Rand, S. C.; Lee, L. S.; Schawlow, A. L. Opt. Commun. 1982,
(75) Lee, L. S.; Rand, S. C.; Schawlow, A. L. Phys. Rev. 1984, B29,
(76) Valiente, R.; Wenger, O.; Gu ¨del, H. Chem. Phys. Lett. 2000, 320,
(77) Valiente, R.; Wenger, O.; Gu ¨del, H. Phys. Rev. 2001, B63,
(78) Gerner, P.; Wenger, O.; Valiente, R.; Gu ¨del, H. Inorg. Chem.
2001, 40, 4534.
(79) Valiente, R.; Wenger, O.; Gu ¨del, H. J . Chem. Phys. 2002, 116,
(80) Salley, G. M.; Valiente, R.; Gu ¨del, H. J . Lumin. 2001, 94/ 95,
(81) Salley, G. M.; Valiente, R.; Gu ¨del, H. J . Phys.: Condens. Matter
2002, 14, 5461.
(82) Strek, W.; Bednarkiewicz, A.; Deren, P. J . J . Lumin. 2001, 92,
(83) Dexter, D. L. Phys. Rev. 1957, 108, 630. Dexter, D. L. Phys. Rev.
1962, 126, 1962.
(84) Basiev, T.T.; Dorochenko, M. E.; Osiko, V. V. J ETP Lett. 2000,
(85) Vegh, R. T.; Donker, H.; van Loef, E. V. D.; Oskam, K. D.;
Meijerink, A. J . Lumin. 2000, 87/ 89, 1017.
(86) Wegh, T. T.; Donker, H.; Meijerink, A.; Lamminma ¨ki, R. J .;
Ho ¨lsa, J . Phys. Rev. 1997, B56, 13841.
(87) Wegh, R. T.; Donker, H.; Oskam, K. D.; Meijerink, A. J . Lumin.
1999, 82, 93.
(88) Wegh, R. T.; Meijerink, A. Acta Phys. Pol. 1996, A90, 333.
(89) Weegh, R. T.; Donker, H.; Oskam, K. D.; Meijerink, A. Science
1999, 283, 663.
(90) Strek, W.; Deren, P.; Bednarkiewicz, A. J . Lumin. 2000, 87/ 89,
(91) Strek, W.; Deren, P. J .; Bednarkiewicz, A.; Kalisky, Y., Bou-
langer, P. J . Alloys Compd. 2000, 300/ 301, 180.
(92) Orlovskii, Y. V.; Basiev, T. T.; Papashvili, A. G.; Vorobev, I. N.;
Alimov, O. K.; Osiko, V. V.; Heber, J . J . Lumin. 2002, 99, 223.
(93) Delevaque, E.; Georges, T.; Monerie, M.; Lamouler, P.; Bayon,
J . F. IEEE Photon Techn. Lett. 1993, 5, 73.
(94) Maurice, E.; Monnom, G.; Dussardier, B.; Ostrowsky, D. B. Opt.
Lett. 1995, 20, 2487.
(95) Auzel, F. J . Lumin. 1990, 45, 341.
(96) Goldner, P.; Pelle ´, F.; Meichenin, D.; Auzel, F. J . Lumin. 1997,
(97) Goldner, P.; Pelle ´, F.; Auzel, F. J . Lumin. 1997, 72/ 74, 901.
(98) Auzel, F.; Goldner, P. Opt. Mater. 2001, 16, 93.
(99) Schaudel, B.; Goldner, P.; Prassas, M.; Auzel, F. J . Alloys Compd.
2000, 300/ 301, 443.
(100) Remillieux, A.; J acquier, B. J . Lumin. 1996, 68, 279.
(101) Balda, R.; Fernandez, J .; Saez de Ocariz, I.; Voda, M.; Garcia,
A. J . Phys. Rev. 1999, B59, 9972.
(102) Fernandez, J .; Balda, R.; Mendorioz, A; Garcia-Adeva, A. J . J .
Phys. Condens. Matter 2001, 13, 10347. Balda, R.; Saez de
Ocariz, I.; Fernandez, J .; Fdez-Navarro, J . M.; Arriandaga, M.
A. J . Phys. Condens. Matter 2000, 12, 10623.
(103) J u, J . J .; Ro, J . H.; Cha, M. J . Lumin. 2000, 87/ 89, 1045.
(104) Kim, S. I.; Yun, S. I. J . Lumin. 1994, 60/ 61, 233.
(105) Malinovski, M.; J oubert, M. F.; J acquier, B. J . Lumin. 1994, 60/
(106) Hirao, K.; Higuchi, M.; Soga, N. J . Lumin. 1994, 60/ 61, 115.
(107) Deren, P. J .; Mahiou, R.; Strek, W.; Bednarkiewicz, A.; Bertrand,
G. Opt. Mater. 2002, 19, 145.
(108) Balda, R.; Sanz, M.; Fernandez, J .; Fdez-Navarro, J . M. J . Opt.
Soc. Am. 2000, B17, 1671.
(109) Fernandez, J .; Balda, R.; Mendorioz, A.; Sanz, M.; Adam, J . L.
J . Non-Cryst. Solids 2001, 287, 437.
(110) Fernandez, J .; Balda, R.; Sanz, M.; Lacha, L. M.; Oleaga, A.;
Adam, J . L. J . Lumin. 2001, 94/ 95, 325.
(111) Acioli, L. H.; Guo, J . T.; de Araujo, C. B.; Messaddeq, T.;
Aegerter, M. A. J . Lumin. 1997, 72/ 74, 68.
(112) Balda, R.; Sanz, M.; Mendorioz, A.; Fernandez, J .; Griscom, L.
S.; Adam, J . L. Phys. Rev. 2001, B64, 144101.
(113) Fernandez, J .; Sanz, M.; Mendorioz, A.; Balda, R.; Chaminade,
J . P.; Ravez, J .; Lacha, L. M.; Voda, M.; Arriandiaga, M. A. J .
Alloys Compd. 2001, 323/ 324, 267.
(114) Iparraguirra, I.; Al-Saleh, M.; Balda, R.; Voda, M.; Fernandez,
J . J . Opt. Soc. Am. 2002, B19, 1.
(115) Malinovski, M.; J acquier, B.; Bouazaoui, M.; J oubert, M. F.,
Linares, C. Phys. Rev. 1990, B41, 31.
(116) Pollnau, M.; Hardman, P. J .; Kern, M. A.; Clarkson, W. A.;
Hanna, D. C. Phys. Rev. 1998, B58, 16070.
(117) Pollnau, M.; Hardman, P. J .; Clarkson, W. A.; Hanna, D. C. Opt.
Commun. 1998, 147, 203.
(118) Bla ¨tte, M.; Danielmeyer, H. G.; Verich, R. Appl. Phys. (Germany)
1973, 1, 275.
(119) Wenger, O. S.; Gamelin, D. R., Gu ¨del, H. U. Phys. Rev. 2000,
(120) Holliday, K.; Russell, D. L.; Henderson, B. J . Lumin. 1997, 72/
(121) Zhang, X; Daran, E.; Serrano, C.; Lahoz, F. J . Lumin. 2000, 87/
(122) Deren, P. J .; Strek, W.; Krupa, J . C. Chem. Phys. Lett. 1998,
(123) Mahiou, R.; Metin, J .; Cousseins, J . C. J . Lumin. 1990, 45, 363.
(124) Wermuth, M.; Riedener, T.; Gu ¨del, H. U. Phys. Rev. 1998, B57,
(125) Pouradier, J . F.; Auzel, F. J . Phys. 1976, 37, 421.
(126) Mu ¨ller, P.; Wermuth, M.; Gu ¨del, H. Chem. Phys. Lett. 1998, 290,
(127) Ryba-Romanovski, W.; Deren, P. J .; Golab, S.; Dominiak-Dzik,
G. J . Appl. Phys. 2000, 88, 6078.
(128) Ryba-Romanovski, W.; Golab, S.; Dominiak-Dzik, G.; Solarz, P.
Appl. Phys. Lett. 2001, 79, 3026.
(129) Chamarro, M. A.; Cases, R. J . Lumin. 1988, 42, 267.
(130) Auzel, F.; Hubert, S.; Meichenin, D. Appl. Phys. Lett. 1989, 54,
(131) Ronac’h, D.; Guibert, M.; Auzel, F.; Meichenin, D.; Allain, J . Y.;
Poignant, H. Electron. Lett. 1991, 27, 511.
(132) Monerie, M.; Allain, J .-Y.; Poignant, H.; Auzel, F. In Fluores-
cence, Up-conversion, and lasing in Er-doped quasi-singlemode
fluorozirconate fibres, Proceedings of ECOC’89, Gothenburg,
Sweden, Sept 1989; paper TuB12-6.
(133) Pollnau, M.; Ghisler, C.; Bunea, G.; Lu ¨thy, W., Weber, H. P. Appl.
Phys. Lett. 1995, 66, 3564.
Upconversion and Anti-Stokes ProcessesChem ical Reviews, 2004, Vol. 104, No. 1 171
(134) Golding, P. S.; J ackson, S. D.; King, T. A.; Pollnau, M. Phys.
Rev. 2000, B62, 856.
(135) Silversmith, A. J . Lumin. 1994, 60/ 61, 636.
(136) Chen, X.; Nguyen, T.; Qui, L.; DiBartolo, B. J . Lumin. 1999, 83/
(137) Pollnau, M.; Heumann, E.; Huber, G. Appl. Phys. 1992, A54,
(138) Lu ¨thi, S. R.; Pollnau, M.; Gu ¨del, H.; Hehlen, M. P. Phys. Rev.
1999, 60, 162.
(139) Hehlen, M. P.; Kra ¨mer, K.; Gu ¨del, H.; McFarlane, R. A.; Schartz,
R. N. Phys. Rev. 1994, B49, 12475.
(140) Wenger, O; Gamelin, D. R., Gu ¨del, H. U.; Butashin, A. V.;
Kaminskii, A. Phys. Rev. 1999, B60, 5312.
(141) Riedener, T.; Egger, P.; Hulliger, J .; Gu ¨del, H. Phys. Rev. 1997,
(142) Riedener, T.; Gu ¨del, H. J . Chem. Phys. 1997, 107, 2169.
(143) Pollnau, M.; Lu ¨thy, W.; Weber, H. P.; Kra ¨mer, K.; Gu ¨del, H.;
McFarlane, R. A. OSA Topics on Advanced Solid State Lasers;
Payne, A., Polock, C., Eds.; OSA: Washington, DC, 1996; Vol.
1, p 493.
(144) Hehlen, M. P.; Frei, G.; Gu ¨del, H. Phys. Rev. 1994, B50, 16264.
(145) Zhang, X.; J ouart, J . P.; Mary, G.; Liu, X; Yuan, J . J . Lumin.
1997, 72/74, 983.
(146) Spinger, B.; Danilov, V. P.; Prokhorov, A. M.; Schwan, L. O.;
Schmid, D. SPIE 2001, 4766, 191.
(147) Felix, S. F.; Gouveia, E. A.; de Araujo, M. T.; Sombra, A. S. B.;
Gouveia-Neto, A. S. J . Lumin. 2000, 87/ 89, 1020.
(148) Tkachuk, A. M.; Razumova, I. K.; Malyshev, A. V.; Gapontsev,
V. P. J . Lumin. 2001, 94/ 95, 317.
(149) Dierolf, V.; Kutsenko, A. B.; von der Hosten, W. J . Lumin. 1999,
83/ 84, 487.
(150) Nunez, L.; Herreros, B.; Duchowicz, R.; Lifante, G.; Tocho, J .
O.; Cusso, F. J . Lumin. 1994, 60/ 61, 81.
(151) Herreros, B.; Lifante, G.; Cusso, F.; Towsend, P. D.; Chandler,
P. J . J . Lumin. 1997, 72/ 74, 198.
(152) Cokcroft, N. J .; Murdoch, K. M. J . Lumin. 1994, 60/ 61, 891.
(153) Chen, X. B.; Zhang, G. Y.; Mao, Y. H.; Hou, Y. B.; Feng, Y.; Hao,
Z. J Lumin. 1996, 69, 151.
(154) Wyss, C. P.; Kehrli, M.; Huber, T.; Morris, P. J .; Lu ¨thy, W.;
Weber, H. P.; Zagumennyi, A. I.; Zavartsev, Y. D.; Studenikin,
P. A.; Shcherbakov, I. A.; Zerrouk, A. F. J . Lumin. 1999, 82,
(155) Riedener, T.; Gu ¨del, H.; Valley, G. C.; McFarlane, R. A. J . Lumin.
1995, 63, 327.
(156) Gomes, A. S. L.; de Araujo, C. B. J . Lumin. 1991, 48/ 49, 876.
(157) Wu, X.; Denis, J . P.; O ¨ zen, G.; Pelle ´, F. J . Lumin. 1994, 60/ 61,
(158) Auzel, F.; Pecile, D. J . Lumin. 1976, 11, 321. Pecile, D. De
l’influencedela matricesur l’addition dephotons par transferts
d’energie (APTE) dans les couples Yb3+-Er3+et Yb3+-Tm3+;
The `se, Universite ´P.et M.Curie, Paris, 1976.
(159) J acquier, J . Lumin. 1994, 60/ 61, 175.
(160) Wenger, O. S.; Wickleder, C.; Kra ¨mer, K.; Gu ¨del, H. J . Lumin.
2001, 94/ 95, 101.
(161) Hubert, S.; Song, C. L.; Genet, M.; Auzel, F. J . Solid StateChem.
1986, 61, 252.
(162) Auzel, F.; Hubert, S., Delamoye, P. J . Lumin. 1982, 26, 251.
(163) Deren, P. J .; Karbowiak, M.; Krupa, J . C.; Drozdzynski, J . J .
Alloys Compd. 1998, 275/ 277, 393.
(164) Stump, N. A.; Murray, G. M.; Del Cul, G. D.; Haire, R. G.;
Peterson, J . R. Radiochim. Acta 1993, 61, 129.
(165) Deren, P. J .; Krupa, J . C.; Strek, W. J . Lumin. 1997, 72/ 74,
(166) Deren, P. J .; Feries, J .; Krupa, J . C.; Strek, W. Chem. Phys. Lett.
1997, 264, 614.
(167) Deren, P. J .; Krupa, J . C.; Yin, M.; J oubert-, M. F.; Strek, W.
Spectrochem. Acta (A) 1998, A54, 2105.
(168) Deren, P. J .; J oubert, M. F.; Krupa, J . C.; Mahiou, R.; Yin, M.
J . Alloys Compd. 2002, 341, 134.
(169) Deren, P. J .; Strek, W.; Zych, E.; Drozdzynski, J . Chem. Phys.
Lett. 2000, 332, 308.
(170) Cresswell, P. J .; Robbins, D. J .; Thomson, A. J . J . Lumin. 1978,
(171) Moncorge ´, R.; Breteau, J . M.; Auzel, F. Philos. Mag. 1985, B51,
(172) Wenger, O. S.; Gu ¨del, H. U. Inorg. Chem. 2001, 40, 5747.
J acobsen, S. M.; Gu ¨del, H. U. J . Lumin. 1989, 43, 125.
(173) Heer, S.; Wermuth, M.; Kra ¨mer, K.; Ehrentraut, D.; Gu ¨del, H.
U. J . Lumin. 2001, 94/ 95, 337.
(174) Heer, S.; Wermuth, M.; Kra ¨mer, K.; Gu ¨del, H. U. Phys. Rev.
2002, 65, 125112.
(175) Heer, S.; Wermuth, M.; Kra ¨mer, K.; Gu ¨del, H. U. Chem. Phys.
Lett. 2001, 334, 293.
(176) Lim, K., S; Lee, C. W.; Kim, S. T.; Seo, H. J .; Kim, C. D. J . Lumin.
2000, 85/ 89, 1008.
(177) Wenger, O. S.; Valiente, R.; Gu ¨del, H. U. High-Press. Res. 2002,
(178) Wenger, O. S.; Gu ¨del, H. U. Inorg. Chem. 2001, 40, 157.
(179) Wenger, O. S.; Valiente, R.; Gu ¨del, H. U. Phys. Rev. 2001, B64,
(180) Wenger, O. S.; Gamelin, D. R.; Gu ¨del, H. U. J . Am. Chem. Soc.
2000, 122, 7408.
(181) Gamelin, D. R.; Gu ¨del, H. U. J . Am. Chem. Soc. 1998, 120, 12143.
(182) Gamelin, D. R.; Gu ¨del, H. U. J . Phys. Chem. 2000, B104, 10222.
(183) Sugano, S.; Tanabe, Y.; Kaminura, H. Multiplets of Transition-
Metal Ions in Crystals; Academic Press: New York, 1970; p 109.
(184) Gamelin, D. R.; Gu ¨del, H. U. Inorg. Chem. 1999, 38, 5154.
(185) Wermuth, M.; Gu ¨del, H. U. J . Phys.: Condens. Matter 2001, 13,
(186) Wermuth, M.; Gu ¨del, H. U. Chem. Phys. Lett. 1997, 281, 81.
(187) Wermuth, Gu ¨del, H. U. J . Lumin. 2000, 87, 1014.
(188) Wermuth, M.; Gu ¨del, H. U. J . Am. Chem. Soc. 1999, 121, 10102.
(189) Wermuth, M.; Gu ¨del, H. U. J . Chem. Phys. 2001, 114, 1393.
(190) Wermuth, M.; Gu ¨del, H. U. Phys. Rev. 2001, B63, 245118.
(191) Page, R. H.; Schaffers, K. I.; Waide, P. A.; Tassano, J . B.; Payne,
S. A.; Krupke, W. F.; Bischel, W. K. J . Opt. Soc. Am. 1998, B15,
(192) Mita, Y.; Yamamoto, H.; Katayanagi, K.; Shionoya, S. J . Appl.
Phys. 1995, 78, 1219.
(193) Chamaro, M. A.; Cases, R. J . Lumin. 1990, 46, 59.
(194) Yeh, D. C.; Sibley, W. A.; Suscavage, M. J . J . Appl. Phys. 1988,
(195) Kermaoui, A.; O ¨ zen, G.; Goldner, P.; Denis, J . P.; Pelle ´, F. J .
Phys. Chem. Solids 1994, 55, 677.
(196) Auzel, F. French Patent No. 1.532.609, 1968.
(197) Tokin IR Catcher; Tokin America (155 Nicholson Lane, San J ose,
CA, 95134), 1987.
(198) Photo-Turkey-1; Sumita Optical Glass Inc. (4-7-25 Harigaya,
Urawa-City, Saitama, J apan), 1994.
(199) Breteau, J . M.; Ayral, J . L.; Micheron, F.; Auzel, F. J . Appl. Phys.
1990, 67, 1102.
(200) Downing, E.; Hesselink, L.; Ralston, J .; Macfarlane, R. Science
1996, 273, 1185.
(201) Auzel F.; Pecile, D. J . Lumin. 1973, 8, 32.
(202) Bril, A.; Sommerdijk, J . L.; de J ager, A. W. J . Electrochem. Soc.
1974, 121, 660.
(203) Quimby, R. S.; Drexhage, M. G.; Suscavage, M. J . Electron. Lett.
1987, 23, 32.
(204) Malta, O. L.; Santa-Cruz, P. A.; de Sa, G. F.; Auzel, F. J . Solid
State Chem. 1987, 68, 314.
(205) Malta, O. L.; Santa-Cruz, P. A.; de Sa, G. F.; Auzel, F. J . Lumin.
1985, 33, 261.
(206) Wang, Y.; Ohwaki, J . Appl. Phys. Lett. 1993, 63, 3268.
(207) Wang, Y.; Ohwaki, J . J . Appl. Phys. 1993, 74, 1272. Ohwaki, J .;
Wang, Y. J pn. J . Appl. Phys. 1994, 33, L334.
(208) Ohwaki, J .; Wang, Y. Appl. Phys. Lett. 1994, 65, 129.
(209) Auzel, F.; Santa-Cruz, P. A.; deSa, G. F. Rev. Phys Appl. (France)
1985, 20, 273.
(210) Kurochkin, A. V.; Mailibaeva, L. M.; Manashirov, O. Y.; Sattarov,
D. K.; Smirnov, V. B. Part 1. Opt. Spectrosc. 1992, 73, 442.
Kurochkin, A. V.; Mailibaeva, L. M.; Manashirov, O. Y.; Sattarov,
D. K.; Smirnov, V. B. Part 2. Opt. Spectrosc. 1992, 73, 447.
(211) Kurochkin, A. V.; Manshisrov, O. Ya.; Sattarov, D. K.; Smirnov,
V. B.; Tsyurupa, O. V. Svetotekhnika 1992, 5, 4 (Translated by
Allerton Press Inc., 1993).
(212) Danielmeyer, H. G.; Bla ¨tte, M. Appl. Phys. (Germany) 1973, 1,
(213) Deutschbein, O.; Auzel, F. In Quantum Electronics; Grivet, P.,
Bloembergen, N., Eds.; Dunod and Columbia University Press:
Paris and New York, 1963; p 851.
(214) Bagdasarov, K. S.; Zhekov, V. I.; Lobachev, V. A. J ; Murina, M.;
Prokhorov, A. M. Sov. J . Quantum Electron. 1983, 13, 262.
(215) Van der Weg, W. F.; Van Tol, M. W. Appl. Phys. Lett. 1981, 38,
(216) de Leeuward, D. M.; Hooft, G. W.'t. J . Lumin. 1983, 28, 275.
(217) Smith, D. J . Lumin. 1981, 23, 209.
(218) J ohnson, L. F.; Guggenheim, H. J . Appl. Phys. Lett. 1971, 19,
(219) Pollack, S. A.; Chang,D. B. J . Appl. Phys. 1988, 64, 2885.
(220) Tong, F.; Risk, W. P.; Macfarlane, R. M.; Lenth, W. Electron.
Lett. 1989, 25, 1389.
(221) Lenth, W.; MacFarlane, R. M. J . Lumin. 1990, 45, 346.
(222) Hebert, T.; Wannemacher, R.; Lenth, W.; Macfarlane, R. M. Appl.
Phys. Lett. 1990, 57, 1727.
(223) Macfarlane, R. M.; Tong, F.; Silversmith, A. J .; Lenth, W. Appl.
Phys. Lett. 1988, 52, 1300.
(224) Brede, R.; Heumann, E.; Koltke, J .; Danger, T.; Huber, G. Appl.
Phys. Lett. 1993, 63, 2030.
(225) Heine, F.; Heumann, E; Danger, T.; Schweizer, T.; Huber, G.
Appl. Phys. Lett. 1994, 65, 383.
(226) Mo ¨bert, P. E.-A.; Heumann, E.; Huber, G. Opt. Lett. 1997, 22,
(227) Auzel, F. Amplifiers and Lasers with Optical Fibers. In Defects
in Insulating Materials; Kanert, O., Spaeth, J .-M., Eds.; World
Scientific: Singapore, 1993; p 43.
(228) Mears, R. J .; Reekie, L.; Poole, S. B.; Payne, D. N. Electron. Lett.
1986, 22, 159.
172 Chem ical Reviews, 2004, Vol. 104, No. 1 Auzel
(229) Allain, J .-Y.; Monerie, M.; Poignant, H. Electron. Lett. 1990, 26,
(230) Allain, J .-Y.; Monerie, M.; Poignant, H. Electron. Lett. 1990, 26,
(231) Whitley, T. J .; Millar, C. A.; Wyatt, R.; Brierley, M. C.; Szebesta,
D. Electron. Lett. 1991, 27, 1785.
(232) Grubb, S. G.; Bennett, K . W.; Cannon, R. S.; Humer, W. F.
Electron. Lett. 1992, 28, 1243.
(233) Smart, R. G.; Hanna, D. C.; Tropper, A. C.; Davey, S. T.; Carter,
S. F.; Szebesta, D. Electron. Lett. 1991, 27, 1308.
(234) Selected papers on Upconversion Lasers; Gosnell, T. R., Ed.; SPIE
Milestone Series MS 161; SPIE Optical Engineering Press:
Bellingham, WA, 2000.
(235) Komukai, T.; Yamamoto, T.; Sugawa, T.; Miyajima, Y. Electron.
Lett. 1992, 28, 830.
(236) Krasutsky, N. J . J . Appl. Phys. 1983, 54, 1261.
(237) Lenth, W.; Macfarlane, R. M. J . Lumin. 1990, 45, 346.
(238) Oetliker, U.; Riley, M. J .; May, P. S.; Gu ¨del, H. U. J . Lumin.
1992, 53, 553.
(239) Ni, H.; Rand, S. C. Opt. Lett. 1992, 17, 1222.
(240) Auzel, F.; Chen, Y. H.; Meichenin, D., ICL’93, Storrs, CN, USA,
9-12 Aug 1993. Auzel, F.; Chen, Y. H.; Meichenin, D., J . Lumin.
1994, 60/ 61, 692.
(241) Chen, Y. H.; Auzel, F. Electron. Lett. 1994, 30, 323.
(242) Auzel, F.; Chen, Y. H. J . Non-Cryst. Solids 1995, 184, 57.
(243) Chen, Y. H.; Auzel, F. J . Phys. D: Appl. Phys. 1995, 28, 207.
(244) Auzel, F.; Chen, Y. H. J . Lumin. 1995, 65, 45.
(245) Pelletier-Allard, N.; Pelletier, R. Phys. Rev. 1987, B26, 4425.
(246) J oubert, M. F.; Guy S.; J acquier, B. Phys. Rev. 1993, B48, 10031.
(247) Kueny, A. W.; Case, W. E.; Koch, M. E. J . Opt. Soc. Am. 1993,
(248) Goldner, P.; Pelle ´, F. Opt. Mater. 1995, 5, 239.
(249) Brenier, A.; Boulon, G.; Madej, C.; Pe ´drini, C.; Lou, L. J . Lumin.
1993, 54, 271.
(250) Auzel, F. Acta Phys. Pol. 1996, 90, 7.
(251) Gomes, A. S. L.; Maciel, G. S.; de Araujo, R. E.; Acioli, L. H.; de
Araujo, C. B. Opt. Commun. 1993, 103, 361.
(252) Brenier, A.; Courrol, L. C.; Pe ´drini, C.; Madej, C.; Boulon, G. J .
Lumin. 1994, 58, 285.
(253) Dyson, J . M.; J affe, S. M.; Eilers, H.; J ones, M. L.; Dennis, W.
M.; Yen, W. M. J . Lumin. 1994, 60/ 61, 668.
(254) Brenier, A.; J urdyc, A. M. J . Lumin. 1996, 69, 131.
(255) Brenier, A.; Courrol, L. C.; Pedrini, C.; Madej, C.; Boulon, G.
Opt. Mater. 1994, 3, 25.
(256) Guy, S. L’avalanche de photons; application a ` l’ion Tm3+dans
differents mate ´riaux. The `se, Universite ´ Claude-Bernard-Lyon
I, 1995. Guy, S. S.; J oubert, M. F.; J acquier, B. Phys. Rev. 1997,
(257) J oubert, M. F.; Guy, B.; Linares, C.; J acquier, B.; Adam, J . L. J .
Non-Cryst. Solids 1995, 184, 98.
(258) J ouart, J . P.; Bouffard, M.; Klein, G.; Mary, G. J . Lumin. 1994,
60/ 61, 93.
(259) Bielejec, E.; Kisel, E.; Silversmith, A. J . Lumin. 1997, 72/ 74,
(260) Liu, G. K.; Chen, Y. H.; Beitz, J . V. J . Lumin. 1999, 81, 7.
(261) Gamelin, D. R.; Wermuth, M.; Gu ¨del, H. U. J . Lumin. 1994, 83/
(262) Hubert, S.; Meichenin, D.; Zhou, B.; Auzel, F. J . Lumin. 1991,
(263) Guy, S.; J oubert, M. F.; J acquier, B. J . Lumin. 1997, 72/ 74, 65.
(264) Case, W. E.; Koch M. E.; Kueny, A. W. J . Lumin. 1990, 45, 351.
(265) Goldner, P.; Fesquet, M.; Auzel, F. J . Opt. Soc. Am. 1998, B15,
(266) Ku ¨k, S.; Diening, A.; Heumann, E.; Mix, E.; Sandrock, T.; Sebald,
K.; Huber, G. J . Alloys Compd. 2000, 300/ 301, 65.
(267) Osiak, E.; Sokolska, I; Ku ¨k, S. J . Lumin. 2001, 94/ 95, 289.
(268) Pelletier-Allard, N.; Pelletier, R. J . Lumin. 1991, 48/ 49, 867.
(269) Deren, P. J .; Krupa, J . C.; Strek, W. J . Alloys Compd. 2000, 300/
(270) Koch, M. E.; Kueny, A. W.; Case, W. E. Appl. Phys. Lett. 1990,
(271) Chen, Y. H.; Auzel, F. Electron. Lett. 1994, 30, 1602.
(272) Scheps, R. IEEE 1994, QE30, 2914.
(273) Scheps, R. IEEE 1995, QE31, 309.
(274) Sandrock, T.; Scheife, H.; Heuman, E.; Huber, G. Opt. Lett. 1997,
(275) McGonigle, A. J . S.; Girard, S.; Coutts, D. W.; Moncorge ´, R.
Electron. Lett. 1999, 35, 1640.
(276) Nicolas, S.; Descroix, E.; Guyot, Y.; J oubert, M.-F.; Abdulsabirov,
R. Yu.; Korableva, S. L.; Naumov, A. K.; and Semashko, V. V.
Opt. Mater. 2001, 16, 233.
(277) Tverjanovich, A.; Grioriev, Y. G.; Degtyarev, S. V.; Kurochkin,
A. V.; Man’shina, A. A.; Tver’yanovich, Yu. S. J . Non-Cryst.
Solids 2001, 286, 89.
(278) Hehlen, M. P.; Gu ¨del, H. U.; Shu, Q.; Rai, S.; Rand, S. C. Phys.
Rev. Lett. 1994, 73, 1103.
(279) Heber, J . Z. Phys. B: Condens. Matter 1987, 68, 115.
(280) Heber, J . J . Alloys Compd. 2000, 300/ 301, 32.
(281) Bonifacio, R.; Lugiato, L. A. In Dissipative Systems in Quatum
Optics; Bonifacio, R., Ed.; Springer-Verlag: Berlin, 1982; p 2.
(282) Auzel, F.; Hubert S.; Meichenin, D. Europhys. Lett. 1988, 7, 459.
(283) Auzel, F. Document deTravail PCM.171, Utilisation du transfert
d’energie entre ions de terres-rares pour leur application even-
tuelles aux compteurs quantiques; addition de deux et trois
photons; CNET, Issy-les -Moulineaux (France), 1967; unpub-
(284) Hehlen, M. P.; Gu ¨del, H. U.; Shu, Q.; Rand, S. C. J . Chem. Phys.
1996, 104, 1232.
(285) Gamelin, J . Chem. Phys. 2000, B104, 11045.
(286) Auzel, F. J . Lumin. 2001, 93, 129.
(287) Bednarkiewicz, A.; Strek, W. J . Phys D. Appl. Phys. 2002, 35,
(288) Zijlmans, H. J . M. A. A.; Bonnet, J .; Burton, J .; Kardos, K.; Vail,
T.; Niedbala, R. S.; Tanke, H. J . Anal. Biochem. 1999, 267, 30.
(289) Heer, S.; Lehman, O.; Haase, M.; Gu ¨del, H. U. Angew. Chem.,
Int. Ed. 2003, 42, 3179.
Upconversion and Anti-Stokes Processes Chem ical Reviews, 2004, Vol. 104, No. 1 173