NIR to visible upconversion in nanocrystalline and bulk Lu2O3:Er3+

Article abstract of DOI:10.1021/jp020256m

In this paper we investigate the upconversion luminescence of bulk and nanocrystalline Lu_1.98Er_0.02O₃ following excitation into the ⁴I_11/2 level using 980 nm radiation. After NIR continuous wave excit

Full text of DOI:10.1021/jp020256m

5622
J. Phys. Chem. B 2002, 106, 5622-5628
NIR to Visible Upconversion in Nanocrystalline and Bulk Lu2O3:Er3+
Fiorenzo Vetrone, J. Christopher Boyer, and John A. Capobianco*
Department of Chemistry and Biochemistry, Concordia UniVersity, 1455 de MaisonneuVe BlVd. W,
Montreal, Canada
Adolfo Speghini and Marco Bettinelli
Dipartimento Scientifico e Tecnologico, UniVersita` di Verona, and INSTM, UdR Verona, Ca’ Vignal,
Strada Le Grazie 15, 37134 Verona, Italy
ReceiVed: January 28, 2002; In Final Form: April 4, 2002
In this paper we investigate the upconversion luminescence of bulk and nanocrystalline Lu1.98Er0.02O3 following
excitation into the 4I11/2 level using 980 nm radiation. After NIR continuous wave excitation, blue, green, and
red emissions were observed for both samples under investigation with the green emission dominating the
upconversion spectrum. A power study reveals that the 4S3/2 (green) and 4F9/2 (red) emitting levels are populated
via two-photon excited-state absorption (ESA) and energy-transfer upconversion (ETU) processes. Upconversion
from the 2P3/2 level is attained via a three-step phonon-assisted energy-transfer process. A marked temperature
dependence of the excited-state lifetimes in both the bulk and nanocrystalline samples was evidenced.
1. Introduction
multiphonon relaxation and thus producing efficient upconver-
sion.
Lanthanide ions doped inorganic materials are particularly
suitable for many applications, which involve the generation
of artificial light. The emission of light from the lanthanide ions
is due mainly to electric and magnetic dipole optical transitions
within the 4fn manifold or may also be of interconfigurational
nature, involving configurations such as 4fn-15d.1 The outer,
less energetic 5s and 5p shells shield the f-electrons from the
influence of external forces and as a result, the intraconfigura-
tional f-f emission spectra consists of relatively sharp lines.2
Toward the end of the 20th century and into the new
millenium, there has been a renaissance in the study of
lanthanide-doped powder phosphors. Modern day scientists are
investigating the luminescence properties of these crystalline
materials in the nanometer scale, with particle sizes under the
100 nm threshold.3 These nanocrystals can be characterized by
markedly different properties, compared to their bulk counter-
parts. The interest stems primarily from the fact that in some
cases the luminescence efficiency is dependent on the size of
the crystallite.4 Thus, they are attractive for such applications
as light-emitting diodes, photovoltaics, and even lasers.
Rare earth sesquioxides, such as Y2O3 and Lu2O3, have
enjoyed present-day resurgence as nanocrystals owing to their
favorable physical properties (high melting point, phase stability,
and low thermal expansion). Moreover, they possess relatively
low phonon energies (phonon cutoff about 600 cm-1),8 which
are favorable for the upconversion process. The Lu2O3 lattice
crystallizes in a cubic bixbyite structure with space group Ia3.9,10
Two distinct sites are available for the lanthanide ions, one with
point-group symmetry C2 and the other with C3i symmetry.11
The C3i site has associated with it a center of inversion and
therefore, according to the selection rules, electric dipole
transitions are forbidden. It is then expected that f-f spectra
exhibit electric dipole transitions pertaining mainly to the Er3+
ions in C2 sites and magnetic dipole transitions from both sites.
In a previous paper, we have shown the emission and
upconversion (λexc ) 804 nm) spectra for bulk and nanocrys-
talline cubic Lu2O3 doped with 1 mol % Er3+.8 In this paper,
we extend the study of the upconversion by examining the
4
emission after excitation at 980 nm into the I11/2 level.
An important process for the generation of visible light is
near-infrared to visible upconversion. In this process, two (or
more) low-energy photons from the excitation source are
converted into one photon of higher energy.5 The erbium ion
(Er3+) is ideally suited to convert infrared light to visible. It
2. Experimental Section
Lu2O3 nanocrystals doped with 1 mol % Er2O3 (Lu1.98
-
Er0.02O3) were prepared using a solution combustion (propellant)
synthesis procedure.12,13 Propellant synthesis is a novel technique
capable of producing nanopowders at relatively low temperatures
and in a timely manner. The process involves the exothermic
reaction between a metal nitrate (oxidizer) and an organic fuel,
such as glycine. The stoichiometric synthesis reaction is
provides intermediate electronic energy levels (4I11/2 and 4I13/2
)
with long lifetimes, which are easily accessible with near-
infrared radiation and can be conveniently pumped with low-
cost commercial diodes.6
The upconversion efficiency is governed principally by the
nonradiative processes of the host material.7 By carefully
selecting the environment for the Er3+ ion, the dynamics of the
excited states can be controlled resulting in a reduction of the
6 M(NO3)3 + 10 NH2CH2COOH + 18 O2 f
3 M2O3 + 5 N2 + 18 NO2 + 20 CO2 + 25 H2O
where M is Lu and Er. The size of the nanopowders is greatly
influenced by the reaction temperature, which can be controlled
by adjusting the glycine-to-metal nitrate molar ratio.13 We
* To whom correspondence should be addressed. Telephone: +1-514-
848-3350. Fax: +1-514-848-2868. E-mail: capo@vax2.concordia.ca.
10.1021/jp020256m CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/14/2002

NIR to Visible Upconversion
J. Phys. Chem. B, Vol. 106, No. 22, 2002 5623
employed a glycine-to-metal nitrate molar ratio of 1.2:1 to
prepare the aqueous precursor solution. After the synthesis, the
resulting nanocrystalline material was fired at 500 °C for 1 h
to decompose any residual nitrate ions. The particle size of the
sample was determined to be about 50 nm by wide-angle X-ray
scattering (WAXS).
For comparison, a bulk sample with identical Er3+ concentra-
tion (Lu1.98Er0.02O3) was prepared by intimately mixing the
oxides, Lu2O3 (Aldrich, 99.99%) and Er2O3 (Aldrich, 99.99+%),
pressing the powders into pellets under 10 tons of pressure and
firing them in air at 1500 °C for 24 h. At this temperature, the
optimum homogeneity was obtained.
All lutetia samples were kept in air without any further
precaution.
Direct luminescence spectra were obtained by exciting the
samples at 488 nm using a Coherent Sabre Innova, 20 W argon
ion laser. Upconversion spectra were obtained by exciting at
980 nm using a Spectra-Physics model 3900 titanium sapphire
laser pumped by the 514.5 nm line of the Ar+ laser. The visible
emissions were collected and dispersed using a Jarrel-Ash
1-meter Czerny Turner double monochromator. A thermoelec-
trically cooled photomultiplier tube (Hamamatsu R943-02) was
used to monitor the signal. A preamplifier (Stanford Research
Systems, model SR 440) processed the PMT signals and a gated
photon counter (Stanford Research Systems, model SR 400)
data acquisition system was used as an interface between the
computer and the spectroscopic hardware. The signal was
recorded under computer control using the Stanford Research
Systems SR 465 software data acquisition/analyzer system.
The decay times were obtained by modulating the excitation
sources mentioned above using an optical chopper (Stanford
Research Systems, model SR 540) and were recorded using a
gated photon counter.
Figure 1. Room-temperature luminescence of (a) bulk and (b)
2
nanocrystalline Lu1.98Er0.02O3 following 488 nm excitation. (i) H11/2
f 4I15/2 (ii) 4S3/2 f 4I15/2 (iii) 4F9/2 f 4I15/2. Inset: (iv) 4I9/2 f 4I15/2 (v)
4
4S3/2 f I13/2
.
TABLE 1: Room Temperature Decay Times for Bulk and
Nanocrystalline Lu2O3:Er3+ Obtained from an Exponential
Fit of the Decay Curves upon 488 and 980 nm Excitation
decay time (µs)
bulk
nanocrystal
transition
488 nm
980 nm
488 nm
980 nm
2H11/2 f 4I15/2
4S3/2 f 4I15/2
4F9/2 f 4I15/2
4I9/2 f 4I15/2
4S3/2 f 4I13/2
63
62
74
61
63
53
55
42
47
54
190
334
246
315
A continuous flow cryostat (Janis Research ST-VP-4) was
used to acquire the low-temperature spectra, and a Lakeshore
model 330 controller was used to monitor the temperature.
talline Lu2O3, prepared by the method above, show bands
occurring at approximately 1500 and 3350 cm-1, assigned to
vibrations of the carbonate and hydroxyl groups, respectively.
The presence of these groups on the nanocrystalline surface
yields higher energy vibrational quanta as compared to the
phonons of bulk lutetia (phonon cutoff of about 600 cm-1). In
the nanocrystalline material, the presence of vibrational quanta
of about 1500 and 3350 cm-1 makes multiphonon relaxation
much more probable than in the bulk material, increases the
rate of depopulation, and therefore gives rise to a shorter
observed decay time with respect to the bulk sample.
3. Results and Discussion
On excitation with 488 nm at room temperature, the Er3+
doped bulk and nanocrystalline Lu2O3 yield distinct emission
bands because of the intra-configurational f-f transitions (Figure
1). Green emission was observed from the (2H11/2, 4S3/2) f 4I15/2
transition between 500 and 580 nm. Red emission was observed
from the 4F9/2 f 4I15/2 transition between 640 and 690 nm. NIR
emission was observed in the ranges 785-825 nm and 840-
870 nm assigned to the 4I9/2 f 4I15/2 and 4S3/2 f 4I13/2 transitions,
respectively.
The observed decay curves for the 4S3/2 f 4I15/2 and 4F9/2 f
4I15/2 transitions following 488 nm excitation were fitted using
a single-exponential function (Table 1). The decay times for
the nanocrystalline material are in general shorter than for the
bulk, which could be attributed to a higher probability for
multiphonon relaxation in the nanocrystalline material owing
to the presence of contaminants on their surface. Lu2O3 can
adsorb atmospheric CO2 and H2O;8 in fact, the combustion
process produces CO2 and H2O as byproducts of the synthesis
reaction, which could therefore be adsorbed immediately after
the formation of the nanoparticles. While the residual nitrate
ions are decomposed after firing the nanocrystals at 500 °C for
1 h, as confirmed by the absence of their characteristic bands
in the MIR and Raman spectra (not shown), we have evidence
that the heat treatment did not completely remove either the
carbon dioxide or water from the surface of the nanocrystals.
In fact, as reported previously,8 the MIR spectra of nanocrys-
Excitation of bulk and nanocrystalline Lu2O3 with NIR light
4
(λexc ) 980 nm) into the I11/2 level of the Er3+ ion produced
intense upconversion luminescence spectra (Figure 2). Blue
upconversion was observed with bands centered at 460, 475,
4
4
2
and 495 nm, which are assigned to the F5/2 f I15/2, P3/2
f
4I11/2, and 4F7/2 f 4I15/2 transitions, respectively. Similar to the
direct emission spectra, upconverted green emission was
observed from the thermalized 2H11/2 f 4I15/2 and 4S3/2 f 4I15/2
transitions centered at 525 and 550 nm, respectively. Upcon-
verted red emission centered at 660 nm was assigned to the
4
4F9/2 f I15/2 transition. The bulk and nanocrystalline samples
exhibit a visually dominant green emission following 980 nm
excitation ascribed to the spin-allowed 4S3/2 f 4I15/2 transition.
We have also observed NIR to visible upconversion in bulk
and nanocrystalline Y2O3:Er3+ 14 and Lu2O3:Er3+ 8 following
800 nm excitation. The spectral features of yttria and lutetia
bulk samples were similar save for a slight red shift (30 cm-1
)

5624 J. Phys. Chem. B, Vol. 106, No. 22, 2002
Vetrone et al.
Figure 3. Power dependence of the green and red upconversion
luminescence intensity of nanocrystalline Lu2O3:Er3+ observed follow-
ing 980 nm excitation.
Figure 2. Room-temperature upconversion luminescence of (a) bulk
and (b) nanocrystalline Lu1.98Er0.02O3 following 980 nm excitation. (i)
4
4
4
4
4
4
2H11/2 f I15/2 (ii) S3/2 f I15/2 (iii) F9/2 f I15/2. Inset: (iv) F5/2
f
2
4
4
4
4I15/2 (v) P3/2 f I11/2 (vi) F7/2 f I15/2
.
talline material, upconversion can therefore only occur via an
ESA or ETU process.
in the Lu2O3:Er3+ spectrum owing to different crystal field
effects felt by the Er3+ dopant. However, striking differences
are observed in the upconversion intensities between the two
isostructural nanocrystalline samples. Under identical experi-
mental conditions and dopant concentration, nanocrystalline
Lu2O3:Er3+ shows upconversion with luminescence approxi-
mately 102 times greater than Y2O3:Er3+. In fact, in nanocrys-
talline Y2O3:Er3+, no noticeable blue upconversion was ob-
served. It has been observed that when comparing the emission
properties of yttrium and lutetium containing oxide15 and
fluoride16 crystals, the lutetium-based crystals show stronger
luminescence. A possible explanation could be found on the
basis of an intensity-borrowing mechanism mixing the 4f and
5d orbitals of the Ln3+ ion via the lattice valence-band levels,
as proposed by Guillot-Noe¨l et al.17 In fact, in Y-based
compounds, the valence-band energy levels are due predomi-
nantly to the oxygen or fluorine 2p orbitals, whereas in Lu
crystals, the top of the valence band would be composed mainly
of Lu 4f orbitals.18 It follows that lutetium could be a more
favorable cation than yttrium for trivalent lanthanide dopant
emission.
The ESA model involves only a single ion and it is usually
the only upconversion process, which occurs in materials at low-
dopant concentrations. In this process, an incoming photon from
the pump beam will bring the ion already in an intermediate
excited level (|1>) to an upper level (|2>). In the framework
of the proposed model, the population N2 of level |2> is given
by
N0A01A12I2
τ-12 (τ-11 + A12I0)
0
N2(t) )
(1)
where N0 is the initial population of the ion in the ground state
(|0>), Aij is a characteristic constant involving the oscillator
strengths for the transitions from the initial state |i> to the final
state |j>. I0 describes the density of photons in the pump beam
-1
-1
while τ1 and τ2 are the intrinsic relaxation rates of levels
|1> and |2>, respectively. According to eq 1, the upconverted
luminescence varies quadratically with the pump beam (I0) but
varies linearly with the concentration of the emitting particle.
Using the proposed model, one photon at 980 nm from the
pump beam will excite the Er3+ ion from the 4I15/2 ground state
The process by which the excited states are populated can
be accomplished by several known mechanisms:19-21 (i) excited-
state absorption (ESA); (ii) photon avalanche (PA); and (iii)
energy transfer (ETU). To better understand the mechanism by
which the 4S3/2 and 4F9/2 levels are populated, the upconverted
4
(|0>) to the I11/2 intermediate excited state (|1>). A second
4
photon of equal energy brings the ion to the F7/2 level (|2>),
which has an energy exactly twice that of the excitation pump
beam (see Figure 4a). Multiphonon relaxation will then populate
4
luminescence intensity I of the green (4S3/2 f I15/2) and red
2
4
4
the H11/2, S3/2, and F9/2 levels.
4
(4F9/2 f I15/2) transitions was measured as a function of the
However, we cannot readily exclude the presence of the ETU
mechanism acting in conjunction with the ESA process. In
samples having a dopant concentration of 0.5 mol % or less,
the energy transfer between dopant ions can be considered as
negligible,23 as the ions are too far apart and therefore the
interaction with one another is very weak. The bulk and
nanocrystalline samples used in this study have a dopant Er3+
concentration of 1 mol % and therefore, it is conceivable that
the ETU process could be efficient. ETU proceeds according
to a scheme in which two ions in close proximity are excited
in an intermediate level |1> and are coupled by a nonradiative
process in which one ion returns to the ground state |0> while
pump power P. The slope of the curve ln(I) versus ln(P) was
1.9 and 1.6, respectively (Figure 3), so we can therefore propose
that two photons partake in the upconversion processes involved
in the population of the 4S3/2 and 4F9/2 levels. Immediately, the
PA mechanism was eliminated as a possible mechanism as no
inflection point was observed in the power study.22 In the PA
mechanism, the upconverted luminescence is weak until a
certain excitation density is achieved, after which, the emission
becomes very intense and the behavior of ln(P) versus ln(I)
exhibits a bending point. This is clearly not the case for either
bulk or nanocrystalline Lu2O3:Er3+. In the bulk and nanocrys-

NIR to Visible Upconversion
J. Phys. Chem. B, Vol. 106, No. 22, 2002 5625
Figure 5. Power dependence of the blue upconversion luminescence
intensity of bulk Lu2O3:Er3+ observed following 980 nm excitation.
the dopant concentration N0. Using the proposed model, the ETU
process (see Figure 4a) occurs as follows: One Er3+ ion is
excited to the 4I11/2 level by the pump laser beam. A neighboring
Er3+ ion also in the 4I11/2 level transfers its energy to the initial
4
ion thereby exciting it to the F7/2 level while it returns to the
ground state. Again, nonradiative phonon decay populates the
green- and red-emitting levels.
Table 1 shows the room-temperature emission lifetimes of
the 4S3/2 and 4F9/2 levels following excitation with 488 and 980
nm radiation. While the decay curves of the excited states
following 488 nm excitation were exponential, the same is not
true following 980 nm excitation, which showed a slight
deviation from single-exponential decay. In the bulk and
nanocrystalline samples, the decay times in the anti-Stokes
emission (λexc ) 980 nm) are significantly longer than the ones
measured for the same states when exciting directly (λexc ) 488
nm). Lengthening of the decay times following upconversion
pumping is a clear indication of the presence of an ETU
process.24
A power study was also performed on the blue-emitting levels
of bulk Lu2O3:Er3+. From the fitting of the ln(I) versus ln(P)
4
4
curves for the F7/2
f f
4I15/2 and F5/2 4I15/2 emission
transitions, the slopes were 1.8 and 1.7, respectively (Figure
5), and therefore two photons partake in the upconversion
processes involved in the population of the 4F7/2 and 4F5/2 levels.
Figure 4. (a) Schematic representation of the ESA and ETU processes
in bulk and nanocrystalline Lu2O3:Er3+. (b) Schematic representation
of the phonon-assisted ETU process in bulk and nanocrystalline Lu2O3:
4
The F7/2 level is therefore likely to be populated via an ESA
2
Er3+ responsible for populating the P3/2 level.
or ETU process involving two photons of 980 nm, as described
4
above (see Figure 4a). On the other hand, the F5/2 level is
the other is promoted to the upper level |2>. Using the same
symbols as above, population of level |2> can be written as
approximately 1400 cm-1 higher in energy than the 4F7/2 level
and so for this level to be populated via two photons, a phonon-
assisted process should be operative. Pure lutetia possesses a
maximum phonon energy of approximately 600 cm-1 and
therefore two or three phonons of such energy would be required
to bridge the 1400 cm-1 energy gap, resulting in a phonon-
assisted upconversion process. Moreover, as mentioned above,
carbonate ions are adsorbed on the surface of the nanoparticles,
and therefore one 1500 cm-1 phonon could possibly be involved
as well in the phonon-assisted upconversion.
(N0A01I0)2σu
τ-2 1 (τ-1 1)2
N2(t) )
(2)
where σu is the rate constant for the ETU process. Similar to
ESA, in the ETU process the population of level |2> varies
quadratically with the density of photons in the pump beam
(I0) but, at variance with ESA, also varies quadratically with

5626 J. Phys. Chem. B, Vol. 106, No. 22, 2002
Vetrone et al.
The slope of the ln(I) versus ln(P) curve for the 2P3/2 f 4I11/2
transition (shown in Figure 5) was 2.8, denoting that three
photons were responsible for the upconversion process. In bulk
Y2O3:Er3+, we have studied the blue upconversion following
804 nm excitation as a function of the Er3+ concentration14 and
have shown that ETU and ESA mechanisms were operating
simultaneously. In particular, it was observed that at low Er3+
concentrations (<1 mol %), the dominant mechanism is ESA,
the reason being that the Er3+ ions are not very close to one
another and the interionic interactions are very small. As the
erbium concentration increases more and more, the interionic
interaction increases and the ETU process becomes more and
more important.
In the present case we excited at 980 nm and thus, the
mechanism that we propose to be responsible for populating
the 2P3/2 level is slightly different. The Er3+ energy-level diagram
reveals that there is no energy level, which could be populated
by a sequential absorption of three pump photons. In fact, if
4
the F7/2 level is populated by the simultaneous absorption of
two photons, (ESA mechanism, see Figure 4a), there is no
energy level that could be populated by the absorption of another
photon. Similarly, if the ion in the ground state is excited by
one photon to the 4I11/2 level and then nonradiatively decays to
Figure 6. Upconversion luminescence spectra of nanocrystalline Lu2O3:
2
4
Er3+ at different temperatures (λexc ) 980 nm): (i) H11/2 f I15/2, (ii)
4S3/2
f
4I15/2, and (iii) F9/2
f
4I15/2 transitions. The intensities of
4
4
4
emission are normalized to the green (4S3/2 f I15/2) transition.
the I13/2 level, a second photon has no resonant energy level
for which to populate. Therefore, we propose a phonon-assisted
ETU mechanism (Figure 4b), in which one to two maximum
energy phonons are requested to compensate the mismatch in
energy (approximately 690 cm-1).
TABLE 2: Decay Times for Bulk and Nanocrystalline
Lu2O3:Er3+ Obtained from an Exponential Fit of the Decay
4
4
4
4
Curves for the S3/2 f I15/2 and F9/2 f I15/2 Transitions
upon 980 nm Excitation at Various Temperatures
The upconversion spectrum obtained at 77 K (not shown)
4S3/2 f 4I15/2
decay time (µs)
4F9/2 f 4I15/2
decay time (µs)
can be used to further illustrate that the population of the 4F5/2
2
and P3/2 energy levels can be caused by a phonon-assisted
temperature (K)
bulk
nanocrystal
bulk
nanocrystal
4
process. In fact, at 77 K, emission from the F5/2 level is not
2
77
100
130
160
190
220
250
298
414
372
349
311
277
254
226
190
447
406
393
381
353
339
266
246
423
415
409
400
389
373
352
334
692
664
628
575
553
540
450
315
observed while emission from the P3/2 level is only barely
detectable. This experimental evidence could be well explained
by the proposed mechanisms required to populate the 4F5/2 and
2P3/2 levels. In the former mechanism, two to three lattice
phonons are involved while one to two phonons are required
for the latter one. Then, the decrease of the temperature induces
4
a stronger decrease of the population of the F5/2 level with
2
respect to the P3/2 level. Therefore, a corresponding decrease
4
4
where C is a constant, k is the Boltzmann constant, T is the
absolute temperature, and ∆E is the energy gap separating levels
4S3/2 and 2H11/2. The calculated values of ln(A2/A1) as a function
of 1/T were fitted using a straight line, and from the slope an
energy gap of 689 cm-1 was obtained. This value is in good
agreement with the difference between the lowest energy 2H11/2
Stark level and the highest energy 4S3/2 Stark level determined
of the emission intensity of the F5/2 f I15/2 transition with
2
4
respect to the P3/2 f I11/2 transition is observed.
The temperature dependence of the green and red upconver-
sion luminescence for bulk and nanocrystalline Lu2O3:Er3+
following 980 nm was studied. At 77K, transitions to the ground
state originating from the 2H11/2 level are not observed. However,
as the temperature is increased to 298 K, we observed an
from the luminescence spectra (∆E ) 720 cm-1).
2
4
increase in the luminescence intensity of the H11/2 f I15/2
The decay curves of the upconverted green (4S3/2 f I15/2
)
4
4
4
transition and a decrease of the S3/2 f I15/2 intensity. This
and red (4F9/2 f I15/2) emissions were also measured as a
4
occurs because the 4S3/2 level is the feeding level for the 2H11/2
function of temperature. Table 2 presents the decay times of
2
level and as a result of the population of the H11/2 level, the
4
4
the F9/2 and S3/2 levels versus temperature for both bulk and
4S3/2 level is depopulated. As the energy gap between the 4S3/2
nanocrystalline Lu2O3:Er3+
.
and H11/2 levels is about 700 cm-1, at room temperature both
2
The decay times of the 4S3/2 and 4F9/2 levels obtained by the
upconversion process (λexc ) 980 nm) are much longer than
those obtained by direct excitation (λexc ) 488 nm) for both
bulk and nanocrystalline samples (see Table 1). This evidence
could be explained by noting that both excitation processes
are populated while at 77 K, the thermalization is very small.
4
The temperature dependence of the (2H11/2
,
4S3/2) f I15/2
transition in the nanocrystalline sample can be seen in Figure 6
(the bulk sample shows the same spectral temperature depen-
dence). If we denote as A1 and A2 the experimental integrated
luminescence intensities of the 4S3/2 f 4I15/2 and 2H11/2 f 4I15/2
4
populate the F7/2 level, which in turn decays to lower levels.
However, in the upconversion process, it could be assumed that
the decays of the emitting levels reflect the feeding from longer-
lived states.
Moreover, the decay times of the S3/2 and F9/2 levels are
longer for the nanocrystalline sample with respect to the bulk
sample. This peculiar behavior was evidenced and explained
2
transitions, respectively, the thermalization of the H11/2 level
may be expressed by the following equation:25
4
4
A
2 ) C exp(-∆E/kT)
(3)
A1

NIR to Visible Upconversion
J. Phys. Chem. B, Vol. 106, No. 22, 2002 5627
for Eu3+ doped yttria by Meltzer et al.26 In fact, the radiative
lifetime of an electronic transition of an ion embedded in a
medium was correlated with an effective refractive index neff,
which is a function of the refractive index of yttria and the
fraction of space occupied by the nanoparticles surrounded by
the media with refractive index nmed. As the refractive index of
lutetia is close to 2,27,28 a value definitely higher than that of
air (nair ) nmed ) 1), a lengthening of the decay times of the
electronic levels of Er3+ for the nanoparticles is expected, in
agreement with the present results.
continuous wave excitation. The upconverted luminescence
intensity for the nanocrystalline sample was lower than its bulk
counterpart owing to a higher probability of multiphonon
relaxation because of the adsorbed CO2 and H2O on its surface.
4
Power studies revealed that the green (4S3/2 f I15/2) and red
(4F9/2 f 4I15/2) upconversion occurred via a two-photon process.
The electronic energy level of the Er3+ ion possesses a level at
exactly twice the excitation energy (4F7/2) and therefore,
upconversion will occur via an excited-state absorption (ESA)
process. However, lifetime measurements also showed that
energy-transfer upconversion (ETU) will occur as the lifetime
of both green- and red-emitting levels are longer when exciting
From Table 2, we observe that for both bulk and nanocrys-
talline samples and for both the 4F9/2 and 4S3/2 levels the decay
times lengthen as the temperature is decreased. However, there
is a difference in the behavior of the decay times for the bulk
and nanocrystalline samples as a function of temperature. For
with 980 nm compared to direct excitation with 488 nm.
4
Blue upconversion was observed, assigned to the F5/2
f
4I15/2, 2P3/2 f 4I11/2, and 4F7/2 f 4I15/2 transitions. A power study
revealed that the 2P3/2 emitting level was populated via a three-
photon process. Population of the 2P3/2 level via an ESA process
is improbable as no resonance exists for a process involving
the sequential absorption of three photons. Therefore, an ETU
mechanism assisted by phonons was determined to be operative.
The natural extension of this work is to study the dynamics
of the upconversion process. Further experiments using pulsed
excitation are currently in progress and will be the subject of a
future paper.
4
the bulk sample, the S3/2 level decays more rapidly on
increasing the temperature with respect to the 4F9/2 level, which
is different from the nanocrystalline sample in which this
different behavior is not observed (see Table 2). The observed
rate of depopulation W of an excited state could be expressed
as the sum of the radiative WR and multiphonon transition
probabilities WMPR. While WR is independent of the temperature
for a transition between 4fn states, the dependence of WMPR from
the temperature could be written as7
WMPR(T) ) WMPR(0)(1 + neff)p
(4)
Acknowledgment. The authors gratefully thank Erica Vivi-
ani (Universita` di Verona, Italy) for expert technical assistance
and Stefano Polizzi (Universita` di Venezia, Italy) for the X-ray
measurements on nanocrystalline Lu1.98Er0.02O3. The authors
acknowledge the Natural Science and Engineering Research
Council of Canada and MURST (project 9903222581•005) of
Italy for financial support.
where
neff ) [exp(pωeff/kT) - 1]-1
(5)
is the occupancy of the effective phonon mode of energy (pωeff)
and p is the number of phonons necessary to bridge the energy
gap between the emitting level and the next lower level.
According to eq 4, W would increase and therefore the observed
decay time τ ) W-1 would decrease upon raising the temper-
ature. If we consider the energy gap between the 4S3/2 and 4F9/2
levels to be about 3000 cm-1 and thus allow five maximum
energy phonons to participate in the nonradiative relaxation,
References and Notes
(1) Dieke, G. H. Spectra and Energy LeVels of Rare Earth Ions in
Crystals; Interscience Publishers: New York, 1968.
(2) Hu¨fner, S. Optical Spectra of Transparent Rare Earth Compounds;
Academic Press: New York, 1978.
(3) Tissue, B. M. Chem. Mater. 1998, 10, 2837-2845.
(4) Goldburt, E. T.; Kulkarni, B.; Bhargava, R. N.; Taylor, J.; Libera,
M. J. Lumin. 1997, 72-74, 190-192.
(5) Scheps, R. Prog. Quantum Electron. 1996, 20, 271-358.
(6) Capobianco, J. A.; Prevost, G.; Proulx, P. P.; Kabro, P.; Bettinelli,
M. Opt. Mater. 1996, 6, 175-184.
(7) Riseberg, L. A.; Moos, H. W. Phys. ReV. 1968, 174, 429-438.
(8) Capobianco, J. A.; Vetrone, F.; Boyer, J. C.; Speghini, A.; Bettinelli,
M. Opt. Mater. 2002, 19, 259-268.
(9) Wyckoff, R. W. G. Crystal Structures; 2nd ed.; Interscience: New
York, 1964; Vol. 2.
(10) Saiki, A.; Ishizawa, N.; Mizutani, N.; Kato, M. J. Ceram. Soc. Jpn.
1985, 93, 649-654.
(11) Forest, H.; Ban, G. J. Electrochem. Soc. 1969, 116, 474-478.
(12) Tessari, G.; Bettinelli, M.; Speghini, A.; Ajo`, D.; Pozza, G.; Depero,
L. E.; Allieri, B.; Sangaletti, L. Appl. Surf. Sci. 1999, 144-145, 686-689.
(13) Ye, T.; Guiwen, Z.; Weiping, Z.; Shangda, X. Mater. Res. Bull.
1997, 32, 501-506.
(14) Capobianco, J. A.; Vetrone, F.; Boyer, J. C.; Speghini, A.; Bettinelli,
M. J. Phys. Chem. B 2002, 106, 1181-1187.
(15) Maunier, C.; Doualan, J. L.; Moncorge´, R.; Speghini, A.; Bettinelli,
M.; Cavalli, E. In Trends in Optics and Photonics; Marshall, C., Ed.;
Proceedings of Advanced Solid-State Lasers, Vol. 50; Optical Society of
America: Seattle, WA, 2001; pp 261-264.
(16) Rambaldi, P.; Moncorge´, R.; Wolf, J. P.; Pe´drini, C.; Gesland, J.
Y. Opt. Commun. 1998, 146, 163-166.
(17) Guillot-Noe¨l, O.; Bellamy, B.; Viana, B.; Vivien, D. Phys. ReV. B
1999, 60, 1668-1677.
4
then the variation in the decay time of the S3/2 level is not
well explained by eq 4 alone. In fact, the thermalization of the
4
2H11/2 level by the S3/2 level is also involved in determining
the observed behavior of the decay times, as the radiative
lifetime of the upper-lying state (2H11/2) is distinctly shorter than
the one of the lower level (4S3/2).29 Therefore, it is expected
that the lowering of the temperature induces the depopulation
2
of the short-lived H11/2 level and consequently an increase of
the observed decay time. In the present case, both mechanisms
are active; the relatively high number of phonons required for
the multiphonon relaxation probably makes the thermalization
process more important.
Also of note is the variation, with temperature, of the emission
from the 4F9/2 level. On passing from room to lower temperature,
the 4F9/2 level is populated by a weaker multiphonon relaxation
4
from the S3/2 level above. Since the energy gap between the
4
4F9/2 level and the next lower I9/2 level is about 2500 cm-1
,
and if the intrinsic decay rate of the 4F9/2 is higher than that of
the thermalized 4S3/2 level, then the multiphonon relaxation from
4
the S3/2 level will be the bottleneck, causing the decay times
from both levels to be identical. The longer observed decay
time from the 4F9/2 level may reflect a weaker decay rate from
this level.
(18) Moine, B.; Dujardin, C.; Lautesse, H.; Pedrini, C.; Combes, C. M.;
Belski, A.; Martin, P.; Gesland, J. Y. Mater. Sci. Forum 1997, 239-241,
245-248.
4. Conclusions
(19) Bloembergen, N. Phys. ReV. Lett. 1959, 2, 84-85.
(20) Chivian, J. S.; Case, W. E.; Eden, D. D. Appl. Phys. Lett. 1979,
35, 124-125.
We have reported and discussed NIR to visible upconversion
in bulk and nanocrystalline Lu1.98Er0.02O3 following 980 nm
(21) Auzel, F. C. R. Acad. Sci. (Paris) 1966, 262, 1016-1019.

5628 J. Phys. Chem. B, Vol. 106, No. 22, 2002
Vetrone et al.
(22) Joubert, M. F.; Guy, S.; Jacquier, B. Phys. ReV. B 1993, 48, 10031-
10037.
(23) van der Ziel, J. P.; Ostermayer, F. W.; Van Uitert, L. G. Phys.
ReV. B 1970, 2, 4432-4441.
(24) Capobianco, J. A.; Raspa, N.; Monteil, A.; Malinowski, M. J.
Phys.: Condens. Matter 1993, 5, 6083-6090.
(25) Shinn, M. D.; Sibley, W. A.; Drexhage, M. G.; Brown, R. N. Phys.
ReV. B 1983, 27, 6635-6648.
(26) Meltzer, R. S.; Feofilov, S. P.; Tissue, B.; Yuan, H. B. Phys. ReV.
B 1999, 60, 14012-14015.
(27) Garcia-Murillo, A.; Le Luyer, C.; Pedrini, C.; Mugnier, J. J. Alloys
Compd. 2001, 323, 74-77.
(28) Medenbach, O.; Dettmar, D.; Shannon, R. D.; Fischer, R. X.; Yen,
W. M. J. Opt. A: Pure Appl. Opt. 2001, 3, 174-177.
(29) Weber, M. J. Phys. ReV. 1968, 171, 283-291.