Cooperative radiative and nonradiative effects in K2NaScF6 codoped with V3+ and Er3+.

Article abstract of DOI:10.1063/1.1640335

K2NaScF6 crystals codoped with V3+ and Er3+ exhibit some novel cooperative near-IR to visible upconversion processes at cryogenic temperatures. V3+ mainly acts as a broadband sensitizer. The V3+ 3T1g --> 3T2g excitation between 13,500 and 15,500 cm(-1), after fast relaxation to V3+ 1T2g, can be transferred to Er3+ 4I(11/2), and then upconversion takes place. Four upconversion mechanisms are identified and characterized. For narrow-band laser excitation the overall efficiency of the upconversion processes is low. However, at 12 K for broadband excitation, such as in a lamp, between 12,000 and 14,500 cm(-1) the number of emitted visible photons is roughly doubled by codoping V3+ in addition to Er3+. Copyright 2004 American Institute of Physics

Full text of DOI:10.1063/1.1640335

Cooperative radiative and nonradiative effects in K 2 NaScF 6 codoped with V 3+ and Er
3+
Christine Reinhard, Karl Krämer, Daniel A. Biner, and Hans U. Güdel
Citation: The Journal of Chemical Physics 120, 3374 (2004); doi: 10.1063/1.1640335
View online: http://dx.doi.org/10.1063/1.1640335
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/120/7?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Effect of cerium codoping in Er 3 + , Ce 3 + : Na La ( Mo O 4 ) 2 crystals
J. Appl. Phys. 97, 123531 (2005); 10.1063/1.1938276
Spectroscopic analysis of the Er 3+ (4f 11 ) absorption intensities in NaBi ( WO 4 ) 2
J. Appl. Phys. 95, 1180 (2004); 10.1063/1.1637139
Luminescence and visible upconversion in nanocrystalline ZrO 2 : Er 3+
Appl. Phys. Lett. 83, 4903 (2003); 10.1063/1.1632020
Green luminescence of Er 3+ in stoichiometric KYb(WO 4 ) 2 single crystals
Appl. Phys. Lett. 80, 4510 (2002); 10.1063/1.1486477
Upconversion dynamics of Er 3+ doped RbGd 2 Br 7
J. Chem. Phys. 107, 2169 (1997); 10.1063/1.474597
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

JOURNAL OF CHEMICAL PHYSICS
VOLUME 120, NUMBER 7
15 FEBRUARY 2004
Cooperative radiative and nonradiative effects in K2NaScF6 codoped
with V3¿ and Er3¿
a)
¨
¨
Christine Reinhard, Karl Kramer, Daniel A. Biner, and Hans U. Gudel
¨
¨
Departement fur Chemie und Biochemie, Universitat Bern, Freiestrasse 3, CH-3000 Bern 9, Switzerland
͑Received 7 July 2003; accepted 17 November 2003͒
K2NaScF6 crystals codoped with V3ϩ and Er3ϩ exhibit some novel cooperative near-IR to visible
upconversion processes at cryogenic temperatures. V3ϩ mainly acts as a broadband sensitizer. The
3ϩ 3
3
V
T
1g
→ T2g excitation between 13 500 and 15 500 cmϪ1, after fast relaxation to V3ϩ 1T2g , can
be transferred to Er3ϩ 4I11/2 , and then upconversion takes place. Four upconversion mechanisms are
identified and characterized. For narrow-band laser excitation the overall efficiency of the
upconversion processes is low. However, at 12 K for broadband excitation, such as in a lamp,
between 12 000 and 14 500 cmϪ1 the number of emitted visible photons is roughly doubled by
codoping V3ϩ in addition to Er3ϩ
. © 2004 American Institute of Physics.
͓DOI: 10.1063/1.1640335͔
I. INTRODUCTION
takes place. And finally there must be a spin-allowed absorp-
tion band in the red/NIR part of the spectrum. V3ϩ in a
fluoride or chloride environment fulfills the conditions, and
we have already demonstrated the principal feasibility of the
concept in V3ϩ and Mo3ϩ doped Cs2NaYCl6 .10 In the
present study we choose Er3ϩ, one of the most efficient and
best studied UC ions, and we choose a fluoride instead of
chloride, because fluorides are generally much less air sensi-
tive than chlorides. For the particular combination of V3ϩ as
The phenomenon of photon upconversion, i.e., long-
wavelength excitation followed by shorter wavelength emis-
sion, has been widely observed and investigated in lan-
thanide containing materials.1–3 Lanthanide ions in solids,
thanks to the shielding of the 4f electrons, often exhibit
more than one metastable excited f-f state. This is one of the
conditions for upconversion. Another one is a sizeable ab-
sorption cross section at the excitation wavelengths, usually
in the near-infrared ͑NIR͒.4 f-f transitions are usually weak
a sensitizer and Er3ϩ as an upconverter a fluoride lattice has
3ϩ 1
the essential advantage that V
in contrast to the other halides.
T
lies above Er3ϩ 4I11/2
,
2g
and the corresponding absorption bands sharp. Laser excita-
tion into the absorption maximum of, e.g., Er
3ϩ 4
I
around
11/2
The aim of the present paper is not the development of a
material which can be used to increase the visible output of
lamps. Rather, it represents a very first step towards this long
term objective. The chemistry and upconversion photon
physics of compounds codoped with a lanthanide and a TM
ion is essentially unexplored.11,12 So we have to start with
questions such as: Is the V3ϩ/Er3ϩ couple a suitable one?
Can a fluoride containing both ions be synthesized? Does
upconversion occur and with which efficiency and by which
mechanisms? Figure 1 shows the energy level schemes of
Er3ϩ and V3ϩ in an octahedral fluoride environment. The
absorption and luminescence properties of V3ϩ doped
K2NaScF6 have been studied and described in the
980 nm still leads to intense green 4S3/2→ I15/2 upconversion
͑UC͒ emission in many Er3ϩ containing materials.5,6 But
higher UC yields can be obtained by sensitization. Yb3ϩ has
proven to be an excellent UC sensitizer for the most impor-
tant UC ions Er3ϩ ͑green͒ and Tm3ϩ ͑blue͒ in halide
lattices.7,8
4
The idea in the present paper is to use broadband NIR
sensitization of Er3ϩ and thus induce upconversion. This is
driven by academic curiosity on the one hand, but also by the
application potential of a device based on such a material.
Present day lamps, particularly incandescent lamps, emit a
great deal of their output in the NIR, where it is lost. If part
of this output could be recycled by NIR to visible ͑VIS͒
upconversion, the lamp efficiency could potentially be im-
proved. Spin-allowed d-d bands of transition-metal ͑TM͒
ions are broad, and properly situated in the spectrum, might
suit the purpose of light absorption, followed by energy
transfer ͑ET͒ to Er3ϩ, where the UC process could take
place.9 There are a number of restrictions for the choice of
literature.13 The broad absorption T1g→ T2g is situated be-
3
3
tween 13 500 and 15 500 cmϪ1. The lowest energy excited
1
state T2g from which emission takes place at low tempera-
ture is situated at 10 413 cmϪ1. Er3ϩ upconversion processes
have been studied in many fluoride crystals and glasses.14,15
4
The origin of the I11/2 excited state lies at about 10 200
cmϪ1 in such an environment, i.e., just about perfect for ex-
the TM ion. Its first and hopefully metastable excited state
3ϩ 1
citation transfer from V
T
.
2g
3ϩ 4
must lie above Er
I
, so that ET to Er3ϩ must be pos-
11/2
sible. The TM ion must have negligible absorption in the
II. EXPERIMENTAL SECTION
A. Crystal growth
3ϩ 4
4
green, around 550 nm, where the Er
S
→ I15/2 emission
3/2
K2NaScF6 :Er3ϩ
:
The appropriate stoichiometric
a͒
Author to whom correspondence should be addressed. Fax: ϩ41 31 631 43
99; Electronic mail: hans-ulrich.guedel@iac.unibe.ch
amounts of Sc2O3 ͑REacton® Alfa Aesar 99.99%͒, Er2O3
0021-9606/2004/120(7)/3374/7/$22.00
3374
© 2004 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Cooperative radiative and nonradiative effects in K2NaScF6
3375
lymerizable resin and then cut and polished. The samples
were mounted in an open copper cell and cooled with a
closed cycle cryostat.
Continuous-wave luminescence and excitation spectra
were performed using a tunable Ti:sapphire laser ͑Spectra
Physics 3900S͒, pumped by
a
frequency doubled
Nd3ϩ:YVO4 laser ͑Spectra Physics Millenia Xs͒, as an ex-
citation source. Wavelength control was achieved by using
an inchworm driven ͑Burleigh PZ-501͒ birefringent filter,
and the wavelength was monitored with a wavemeter ͑Bur-
leigh WA2100͒. The visible sample upconversion lumines-
cence was dispersed by a 0.85 m double monochromator
͑Spex 1402͒ with 500 nm blazed 1200 grooves/mm gratings.
The visible luminescence was detected by a cooled photo-
multiplier tube ͑Hamamatsu 3310-01͒ using a photon count-
ing system ͑Stanford Research SR400͒. The laser beam ͑di-
ameter: 1 mm͒ was focused on the sample with a fϭ53
mm focal lens.
The near-infrared sample luminescence was dispersed by
a 0.75 m single monochromator ͑Spex 1702͒ equipped with a
750 nm blazed 600 grooves/mm grating. It was measured
with a liquid nitrogen cooled Ge detector ͑ADC 403L͒. The
detector signal was processed by a lock-in amplifier ͑Stan-
ford Research DS 345͒.
FIG. 1. Energy level scheme for Er3ϩ and V3ϩ in an octahedral fluoride
environment with some relevant excitation and relaxation processes.
Straight and wavy arrows indicate radiative and nonradiative processes, re-
spectively.
The spectra represent the number of photons versus
wave numbers. All the luminescence spectra were corrected
for the wavelength dependent response of the monochro-
mator and detection system, and for the refractive index of
air ͑vacuum correction͒. The upconversion excitation spec-
trum was corrected linearly for the wavelength dependent
output power of the Ti:sapphire laser. The luminescence
spectra measured using near-infrared and visible sensitive
detectors were scaled to each other by measuring the sample
luminescence around 10 000 cmϪ1 with both detectors.
For luminescence lifetime measurements in the NIR,
rectangular pulses of the Ti:sapphire were generated by using
an acousto-optic modulator ͑Coherent 305͒ connected to a
function generator ͑Stanford Research DS 345͒. The tempo-
ral behavior of the luminescence intensity was detected using
a multichannel scaler ͑Stanford Research SR430͒ for decay
curves. For all the luminescence measurements, cooling of
the crystals was achieved using the helium-gas flow tech-
nique.
͑Johnson Mattey & Brandenberger͒, KCl ͑Merck Supra-
pur®͒, and NaCl ͑Merck Suprapur®͒ were dissolved in hot
HCl ͑30% Suprapur͒. The solution was evaporated, and the
resultant white solid was ground and dried. The white
K2NaScCl6 :Er3ϩ powder was fluorinated at 550 °C in a HF
gas stream for 64 h. The HF gas stream was produced by
decomposing KHF2 ͑Fluka, puriss p.a.͒ at 420 °C. The
K2NaScF6 :Er3ϩ powder was enclosed in a tantalum am-
poule, which had been sealed into a quartz ampoule. Crystal
growth of K2NaScF6 :Er3ϩ was achieved by melting the
powder at a furnace temperature of 1050 °C and slow cooling
to room temperature over a period of two weeks. The Er3ϩ
concentration in the starting mixture was 0.5%, but only
0.05% were actually incorporated into the crystal, as esti-
mated from absorption spectra.
VF3 : Dark green VF3 was obtained by fluorination of
VCl3 ͑Atomergic Chemetals 99%͒ in a HF gas stream for 14
h at 400 °C.
K2NaScF6 :Er3ϩ; V3ϩ: First, the same procedure was
followed as for K2NaScCl6 :Er3ϩ, except for the fluorination
step. The white K2NaScCl6 :Er3ϩ powder was mixed with
the appropriate amount of VF3 and fluorinated in a HF gas
stream at 400 °C for 24 h. Single crystals with an average
size of 1ϫ1ϫ0.5 mm were obtained, with the method de-
scribed for the growth of K2NaScF6 :Er3ϩ. The elpasolite
structure was verified by powder X-ray diffraction. The V3ϩ
and Er3ϩ concentration in the starting mixture was 2.5% and
0.5%, respectively, but only 0.2% and 0.05%, respectively,
were actually incorporated. This was estimated from the d-d
and f-f absorptions in the VIS/NIR.
III. RESULTS
Figure 2͑a͒ shows the 12 K luminescence spectrum of
K2NaScF6 :Er3ϩ excited into the maxima of the 4I9/2 absorp-
tion band at 12 525 cmϪ1 for the visible and at 12 534 cmϪ1
for the near-IR part, with a laser power of 121 mW. We
4
4
observe the dominant NIR emissions I11/2→ I15/2 and
4I13/2→ I15/2 located around 10 000 and 6500 cmϪ1, respec-
4
2
4
4
4
4
tively. In addition, the H9/2→ I15/2 , S3/2→ I15/2 and F9/2
4
→ I15/2 emissions are observed in the VIS part of the spec-
trum, but with smaller intensities. The 4I11/2 and 4I13/2 origins
are located at 10 198 and 6503 cmϪ1, respectively. The pho-
4
4
4
4
ton ratio of the I11/2→ I15/2 and I13/2→ I15/2 emissions is
B. Spectroscopy
44/56 at this temperature.
Absorption spectra were measured on a Cary 5E ͑Varian͒
spectrometer. Samples for absorption were embedded in po-
Figure 2͑b͒ shows the 12 K luminescence spectrum of
K2NaScF6 :Er3ϩ; V3ϩ excited into the 3T2g absorption band
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

3376
J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Reinhard et al.
FIG. 2. ͑a͒ 15 K luminescence spectra of K2NaScF6 :Er3ϩ excited into 4I9/2
at 12 525 cmϪ1 for the visible and at 12 534 cmϪ1 for the near-IR part ͑laser
power 121 mW͒. ͑b͒ 15 K luminescence spectra of K2NaScF6 :Er3ϩ; V3ϩ
excited into V3ϩ 3T2g at 14 175 cmϪ1 ͑laser power 150 mW͒. Note the
FIG. 3. ͑a͒ 7 K absorption spectrum of K2NaScF6 :Er3ϩ in the region of the
4
4F7/2
,
4F5/2 , and F3/2 transitions. The true energy axis of this absorption
*
scaling factors of the VIS part. The asterisk ͑ ͒ denotes a peak which is due
spectrum is at the top of the graph. The energy axis at the bottom represents
to the second order of the pump laser.
a redshift of 10 198 cmϪ1, the energy of the I11/2 origin in the absorption
4
spectrum. ͑b͒ 15 K excitation spectrum of K2NaScF6 :Er3ϩ monitoring the
Er3ϩ 4S3/2→ I15/2 upconversion luminescence at 18 315 cmϪ1. ͑c͒ 15 K ex-
4
4
citation spectrum of K2NaScF6 :Er3ϩ monitoring the Er3ϩ 4I11/2→ I15/2 NIR
at 14 175 cmϪ1 with a laser power of 150 mW. In the near IR
around 10 000 cmϪ1 we observe a superposition of
luminescence at 10 081 cmϪ1. The excitation spectra are corrected linearly
for the output of the Ti:sapphire laser. The y scales are different for the
energy ranges to the left and right of the dotted line. The letters A, B, C, and
V
T
→ T1g emission, with an origin at 10 413 cmϪ1
,
3ϩ 1
3
2g
4
4
4
4
D stand for the excitations I15/2→ I11/2
,
4I11/2→ F7/2
,
4I15/2→ I9/2 and
and a weak Er3ϩ 4I11/2→ I15/2 emission with an origin at
4
4I9/2→ F3/2 respectively, see Fig. 6͑a͒.
4
10 198 cmϪ1. Besides this dominant feature there is a strong
Er3ϩ 4I13/2→ I15/2 emission around 6500 cmϪ1 and very
4
2
4
4
4
4
weak H9/2→ I15/2
,
4S3/2→ I15/2 and F9/2→ I15/2 emis-
thus able to catch the most important part of the excitation
spectrum. Figure 4͑a͒ shows a section of the single-crystal
absorption spectrum of K2NaScF6 :Er3ϩ. The proper x axis
is shown at the top. It is shifted with respect to the x axis of
the Figs. 4͑b͒ and 4͑c͒ by 6503 cmϪ1, the origin of
Er3ϩ 4I13/2 in this lattice.
sions in the VIS.
Figures 3͑b͒ and 3͑c͒ show 15 K NIR excitation spectra
4
4
of K2NaScF6 :Er3ϩ monitoring the S3/2→ I15/2 UC lumi-
nescence at 18 315 cmϪ1 and the 4I11/2→ I15/2 NIR lumines-
4
cence at 10 081 cmϪ1, respectively. The I15/2→ I11/2 and
4
4
4I15/2→ I9/2 multiplets are seen in both spectra, but with
4
In Fig. 5 we compare the excitation spectrum of
4
K2NaScF6 :Er3ϩ in the region of the 4I15/2→ I11/2 absorption
clear differences in the fine structure. For future reference,
Fig. 3͑a͒ shows a section of the single-crystal absorption
spectrum of K2NaScF6 :Er3ϩ. The proper x axis is shown at
the top. It is shifted with respect to the x axis of the Figs.
3͑b͒ and 3͑c͒ by 10 198 cmϪ1, the origin of Er3ϩ 4I11/2 in this
lattice.
3ϩ 1
3
with the
V
T
→ T1g emission spectrum of
2g
K2NaScF6 :Er3ϩ; V3ϩ at 15 K. The spectral overlap is seen
to be very good. The inset of Fig. 5 shows the temperature
3ϩ 1
dependence of the V
T
lifetime.
2g
Figures 4͑b͒ and 4͑c͒ show 15 K excitation spectra of
IV. DISCUSSION
4
K2NaScF6 :Er3ϩ; V3ϩ monitoring the Er3ϩ 4S3/2→ I15/2 UC
A. V3¿ and Er3¿ luminescences
emission at 18 315 cmϪ1 and the V3ϩ 1T2g→ T1g NIR emis-
3
sion at 10 198 cmϪ1, respectively. They both show bands
Our primary goal in this study is the search for coopera-
tive new processes in V3ϩ and Er3ϩ codoped K2NaScF6
crystals. Before doing so, see Sec. IV B, we have to briefly
characterize and discuss the ‘‘normal’’ luminescence pro-
cesses of V3ϩ and Er3ϩ, respectively, in this lattice.
corresponding to Er3ϩ 4I15/2→ I11/2 ͑around 10 300 cmϪ1
͒
4
and I15/2→ I9/2 ͑around 12 500 cmϪ1͒ excitations, but with
very different relative intensities and fine structures. Above
13 500 cmϪ1 the excitation spectrum of the V3ϩ 1T2g
4
4
3
3
→ T1g emission in Fig. 4͑c͒ shows the onset of the T1g
The host lattice K2NaScF6 is cubic with a lattice con-
stant of 8.47 Å. It belongs to the family of cubic face-
3
→ T2g absorption. In the UC excitation spectrum in Fig.
3Ϫ
4͑b͒ there are relatively strong sharp features above 14 000
cmϪ1 superimposed on a broadband shown in gray. Our
Ti:saphhire laser is tunable up to 14 500 cmϪ1, and we were
centered elpasolites and contains discrete ͓ScF6͔
units.
The ionic radii of Sc3ϩ, V3ϩ, and Er3ϩ for octahedral coor-
dination are 0.75, 0.64, and 0.89 Å, respectively.16 We at-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Cooperative radiative and nonradiative effects in K2NaScF6
3377
FIG. 5. Luminescence spectrum at 15 K of K2NaScF6 :Er3ϩ; V3ϩ excited at
14 175 cmϪ1 ͑bottom, V3ϩ 1T2g→ T1g) and 15 K excitation spectrum of
3
K2NaScF6 :Er3ϩ monitoring the I11/2→ I15/2 emission at 10 081 cmϪ1
͑top͒. The areas below the curves are normalized to one. The inset shows the
V3ϩ 1T2g luminescence lifetime in K2NaScF6 :Er3ϩ; V3ϩ as a function of
temperature.
4
4
FIG. 4. ͑a͒ As in Fig. 3, but with a shift of 6503 cmϪ1, the I13/2 origin,
between the top and bottom axes. ͑b͒ 15 K upconversion excitation spec-
4
4
trum of K2NaScF6 :Er3ϩ; V3ϩ monitoring the Er3ϩ 4S3/2→ I15/2 lumines-
and 4I13/2→ I15/2 emission in the NIR, to efficient upconver-
4
cence at 18 315 cmϪ1. ͑c͒ 15 K excitation spectrum of K2NaScF6 :Er3ϩ
;
sion emission in K2NaScF6 :Er3ϩ at 15 K. The dominant UC
V3ϩ monitoring the NIR luminescence at 10 198 cmϪ1. Spectra ͑b͒ and ͑c͒
are corrected for the linear output power of the Ti:sapphire laser. The y
scales are different for the energy ranges left and right of the dotted line. The
letters refer to excitations defined in Fig. 6.
4
4
4
bands are due to S3/2→ I15/2 ͑green͒ and 4F9/2→ I15/2 ͑red͒
transitions, with a very small intensity also in the violet
2H9/2→ I15/2 emission, when excited into 4I15/2→ I9/2 . They
are all due to two-photon and three-photon processes as evi-
denced by their power dependence, which was experimen-
tally determined.
4
4
tribute the fact that only small amounts of both V3ϩ and
Er3ϩ could be incorporated to the differences in ionic radius
between host and guests.
Figure 6͑a͒ shows a scheme with the UC mechanism in
K2NaScF6 :Er3ϩ for I15/2→ I11/2 and I15/2→ I9/2 excita-
tion, respectively. These were determined from the excitation
spectra in Fig. 3. A sequence of ground-state absorption
͓GSA, steps A and C in Fig. 6͑a͔͒ and excited-state absorp-
tion steps ͓ESA, steps B and D in Fig. 6͑a͔͒ lead to 4F7/2 and
4
4
4
4
3ϩ 1
3
The
V
T
→ T1g luminescence spectrum of
2g
K2NaScF6 :Er3ϩ; V3ϩ shown in Fig. 2͑b͒ is slightly con-
taminated by the underlying Er
3ϩ 4
4
I
→ I15/2 emission. It
11/2
is very similar in line positions and relative intensities to the
13
4
4
4
4
published spectrum of K2NaScF6 :V3ϩ
.
There is a signifi-
4F3/2 population upon I15/2→ I11/2 and I15/2→ I9/2 laser
excitation, respectively. In our K2NaScF6 :Er3ϩ we see from
a comparison of Figs. 3͑a͒ and 3͑c͒ an energetic coincidence
of the multiplets A and B. A similar coincidence occurs for
transitions C and D. These are well established processes in
Er3ϩ doped fluorides.5,19 As a result of these mechanisms the
actual excitation multiplets around 10 400 and 12 600 cmϪ1
in Fig. 3͑b͒ correspond to the products of the profiles A B
and C D, respectively. This explains the difference to Fig.
*
3͑c͒, in which the pure GSA steps A and C are probed.
From the relative photon counts in the various emission
bands in Fig. 2͑a͒ we get important information about non-
radiative relaxation processes. The integrated photon ratio
cant difference in the 15 K lifetimes, however, we determine
␶
15Kϭ68 ms, whereas a value ␶15Kϭ25 ms was reported for
this temperature in Ref. 13. Since the presence of Er3ϩ in our
1
crystal cannot possibly lead to a longer T2g lifetime, we
conclude that the sample used in Ref. 13 was suffering from
nonradiative loss processes not present in our sample. The
steep drop of ␶ between 15 and 80 K, see inset of Fig. 5, was
also observed in Ref. 13. It is thus typical of V3ϩ in this
lattice and not primarily resulting from the presence of Er3ϩ
in our codoped crystals. In Ref. 13 this thermal quenching
was attributed to thermally activated multiphonon relaxation
processes.
*
4
4
4
4
There are hundreds of papers in the literature dealing
with NIR to VIS upconversion in Er3ϩ doped fluorides.17,18
In upconversion phosphor and laser materials, in which sharp
for the I11/2→ I15/2 and I13/2→ I15/2 emissions at 15 K
4 4
upon I15/2→ I9/2 excitation is 44/56. The most likely non-
radiative I11/2 4I13/2 process is multiphonon relaxation.
4
2
4
4
line excitation is used, Yb3ϩ 2F7/2→ F5/2 excitation around
The energy gap between the I11/2 and I13/2 multiplet is
10 307 cmϪ1 ͑970 nm͒ is often used as an efficient sensitizer.
about 3700 cmϪ1. The highest phonon energy in K2NaScF6
As shown in Figs. 2 and 3 excitation into I15/2→ I11/2 be-
is estimated to be 560 cmϪ1 13,20
.
Thus six quanta of the high-
4
4
tween 10 200 and 10 500 cmϪ1 and I15/2→ I9/2 between
est energy vibration are needed to bridge this energy gap.
Since the I15/2↔ I11/2 oscillator strength is very small, the
4
4
12 300 and 12 800 cmϪ1 leads, in addition to I11/2→ I15/2
4
4
4
4
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

3378
J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Reinhard et al.
FIG. 6. Schematic representation of the relevant upconversion mechanisms in K2NaScF6 :Er3ϩ ͑a͒ and K2NaScF6 :Er3ϩ; V3ϩ ͑b͒–͑d͒. The relevant GSA and
ESA transition steps are labeled by the letters A–G. Straight and wavy arrows stand for radiative and nonradiative processes, respectively.
4
nonradiative I11/2 4I13/2 deactivation by multiphonon re-
nitude. This is mainly due to V3ϩ 1T2g 3T1g multiphonon
relaxation processes. The part which is transferred to Er3ϩ
laxation is competitive. Looking at the VIS part of Fig. 2͑a͒,
2
4
4
all ends up on I13/2 , because Er3ϩ 4I11/2 4I13/2 mul-
we see that the H9/2→ I15/2 emission is extremely weak. In
4
tiphonon relaxation also dominates the 4I11/2 depletion at this
temperature. The net result is a photon ratio of 22/78 for
V3ϩ (1T2g)/Er3ϩ(4I13/2), respectively, at 200 K. From the
above numbers we conclude that V3ϩ to Er3ϩ energy trans-
this case multiphonon relaxation down to S3/2 is absolutely
2
4
dominant, since the largest gap between H9/2 and F3/2 is
only 2300 cmϪ1. In Er3ϩ containing UC phosphors the green
(4S3/2) and red (4F9/2) photon ratio can be tuned by varying
the Er3ϩ ͑and Yb3ϩ) concentrations as well as the tempera-
ture. Besides multiphonon relaxation there are processes
such as cross-relaxation that are likely to contribute to
4S3/2 4F9/2 relaxation.
fer is competitive with the other relaxation processes of
3ϩ 1
V
T
, particularly at lower temperatures.
2g
3ϩ 1
The energy transfer occurs from V
T
, after fast
2g
3T2g→ T2g multiphonon relaxation. The accepting state is
1
For 121 mW laser excitation into 4I9/2 at 15 K we obtain
a ratio of integrated VIS (4S3/2ϩ4F9/2) to NIR (4I11/2
ϩ4I13/2) emitted photons of 13%. This is a high ratio, com-
parable to the best known Er3ϩ based UC phosphors.21
Er3ϩ 4I11/2 . The relevant spectral overlap is thus between the
3ϩ 1
3
V
T
→ T1g emission profile and the Er3ϩ 4I15/2
2g
4
→ I11/2 excitation profile. These experimental profiles, both
normalized to one, at 15 K are shown in Fig. 5. The overlap
is substantial and this explains the observed phenomena.
3ϩ 1
B. Cooperative processes in V3¿ and Er3¿
codoped K2NaScF6
From the V
T
branching ratio kET /kradϭ42/58 and
2g
k
ETϭktotϪkrad , see Fig. 7 and the measured V3ϩ 1T2g life-
time at 15 K of 68 ms (ktotϭ14.7 sϪ1) we derive a value for
the macroscopic energy transfer rate constant kETϭ6.2 sϪ1
1. V3¿ to Er3¿ energy transfer
.
Excitation into the spin-allowed V3ϩ 3T1g→ T2g ab-
3
sorption band between 13 500 and 15 500 cmϪ1 in the
codoped K2NaScF6 :Er3ϩ; V3ϩ crystal leads, in addition to
the normal V3ϩ emission around 10 000 cmϪ1, to Er3ϩ emis-
sion both in the NIR and the VIS. This is seen in Fig. 2͑b͒.
3ϩ 4
4
The Er
I
→ I15/2 band strongly overlaps with the
11/2
V3ϩ 1T2g→ T1g band, see Figs. 2͑a͒ and 2͑b͒. The
3
3ϩ 4
4
Er
I
13/2
→ I15/2 emission around 6500 cmϪ1, on the other
hand, is well separated.
From the measured integrated photon counts at 15͑200͒
K in the various emission bands we derive the ratios shown
in Fig. 7. At 15 K we can assume that there is no mul-
3ϩ 1
3
tiphonon relaxation from V
T
to T1g . We then find
2g
that 58% of the V3ϩ excitation is emitted as 1T2g to 3T1g and
42% is transferred to Er3ϩ. This is subdivided into
19% 4I11/2→ I15/2 and 23% 4I13/2→ I15/2 . At 200 K mul-
tiphonon relaxation processes become much more important.
The total of emitted photons decreases substantially, and the
ratios change. The total loss of V3ϩ excitation due to nonra-
diative processes is reflected in the decreasing lifetime of the
4
4
FIG. 7. Schematic representation of the relevant radiative ͑straight arrows͒
and non-radiative ͑wavy arrows͒ processes in K2NaScF6 :Er3ϩ; V3ϩ. The
numbers represent the measured emission photon ratios relevant at 15 and
͑200͒ K.
1T2g→ T1g luminescence between 10 and 200 K shown in
3
the inset of Fig. 5. The decrease is about one order of mag-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Cooperative radiative and nonradiative effects in K2NaScF6
3379
steps involved, one ͑G͒ on V3ϩ and the other one ͑F͒ on
This is a small value considering the almost optimal spectral
overlap, and we ascribe it to the low concentration of both
dopant ions in this lattice.
Er3ϩ, the excitation profile corresponds to the product G F
*
of the two profiles.
The second mechanism, which experimentally manifests
2. Cooperative upconversion mechanisms
itself in the structured peak labeled G E in Fig. 4͑b͒, is more
*
unusual and novel. It is a truly cooperative mechanism, in
which both ions are involved in the upconversion step itself.
We use a simple V3ϩ-Er3ϩ dimer picture to represent it in
Figure 2͑b͒ shows that upconversion emission can be
induced in K2NaScF6 :Er3ϩ; V3ϩ by exciting into the
3
V3ϩ 3T1g→ T2g absorption band in the far red. There is no
Er3ϩ absorption at the laser wavelength used in this experi-
ment, and a control experiment with the singly Er3ϩ doped
crystal does not show any upconversion emission. The three
emission bands in the near UV, green and red spectral re-
gions are very similar to the UC bands in the singly doped
Fig. 6͑c͒. While the first step G is not unusual, the second
3ϩ 1
step E is. It is a radiative step with V
T
as the initial
2g
and Er3ϩ 2H9/2 as the final state. This follows from the ener-
gies of the peaks in the G E band in Fig. 4͑b͒. They corre-
*
spond to energy differences between the Er3ϩ 2H9/2 levels
and 10 413 cmϪ1, the metastable lowest excited state of V3ϩ
.
2
4
sample ͓Fig. 2͑a͔͒ and assigned to H9/2→ I15/2
,
4S3/2
4
4
4
In the dimer picture the intermediate state in the upconver-
→ I15/2 and F9/2→ I15/2 transitions, respectively. The
sion process is a dimer state mainly localized on V3ϩ
,
4
4
4
power dependence of the H9/2→ I15/2
,
4S3/2→ I15/2 and
4F9/2→ I15/2 upconversion emission intensity was found to
be very close to quadratic, confirming the two-photon char-
acter of the process.
4
whereas the upper emitting state is a dimer state with pre-
dominant Er3ϩ character. In a number of mixed lanthanide or
mixed transition metal/lanthanide systems we have recently
identified similar cooperative upconversion processes, and in
all of these we ascribed them to an exchange mechanism.23,24
In the present case we have no direct contact between V3ϩ
and Er3ϩ through a common bridging ligand, and the situa-
tion is less clear. We also considered another possible and
less exotic mechanism, namely G on V3ϩ, followed by en-
We note that the Er3ϩ bands in Fig. 2͑b͒ are not as well
resolved as in Fig. 2͑a͒. The individual lines are broader, and
this is ascribed to an inhomogeneous broadening due to the
relatively high V3ϩ doping on the one hand and to small
shifts and splittings of the Er3ϩ levels due to V3ϩ-Er3ϩ ex-
change interactions on the other. The green emission is vis-
ible by eye at temperatures up to 200 K for 96 mW laser
excitation at 14 175 cmϪ1 into V3ϩ 3T2g . But, as seen in Fig.
2͑b͒, the visible emissions are very weak compared to the
NIR emission bands. A comparison of Figs. 4͑b͒ and 4͑c͒
4
ergy transfer to Er3ϩ 4I11/2 and relaxation to I13/2 and then
4
4I13/2→ F7/2 on Er3ϩ as the second excitation step. Adding
6503 cmϪ1, the Er3ϩ origin, to the peak positions in the band
system labeled G E in Fig. 4͑b͒, we do not get the known
*
energies of the 4F7/2 level, see also Fig. 4͑a͒. In Fig. 6͑d͒ we
show the sensitization and upconversion scheme responsible
for the underlying broad band between 13 500 and 14 500
cmϪ1 in Fig. 4͑b͒. Two quanta of V3ϩ 1T2g excitation are
consecutively injected by energy transfer into Er3ϩ. The re-
3
confirms that V3ϩ 3T1g→ T2g broadband excitation above
13 500 cmϪ1 is most efficient in inducing Er3ϩ NIR (4I11/2
4
4
4
→ I15/2ϩ4I13/2→ I15/2) emission. Direct Er3ϩ 4I15/2→ I11/2
excitation, on the other hand, is unimportant.
The absorption and upconversion excitation spectra
shown in Fig. 4 allow for an unambiguous identification and
characterization of three distinct mechanisms. They are sche-
matically represented in Figs. 6͑b͒, 6͑c͒, and 6͑d͒ and will
now be discussed in this order. The first step is the
sulting excitation spectrum corresponds to G G and is thus
*
broadband.
Let us finally return to the original question: is the title
compound a broadband sensitizing upconverter? The answer
is yes, but with an extremely low efficiency, particularly at
3ϩ 3
3
V
T
→ T2g excitation. It is common to all three mecha-
1g
nism, and we label it G in Fig. 4 and 6. The onset of the 3T1g
temperatures of 200 K and above. At 15 K for narrow-band
3ϩ 3
to T2g absorption band is around 13 500 cmϪ1, and we can
3
laser excitation with 150 mW into the V
T
absorption
2g
follow it up to about 14 500 cmϪ1, the limit of our Ti:sap-
phire tuning range. In Fig. 4͑c͒ it can be seen that we reach
about the maximum of the absorption band. This is in good
agreement with the absorption spectrum of V3ϩ in
K2NaScF6 in Ref. 13, in which this absorption band extends
from 13 500 to about 15 500 cmϪ1. In the mechanism de-
we get a VIS:NIR photon ratio of about 0.1%. For broadband
excitation, such as in a lamp, the situation would be much
more favorable. Figure 4͑b͒ shows that excitation peaks la-
beled C D and the sum of G E, G F and G G are of com-
*
*
*
*
parable intensity. This is mainly the result of the high absorp-
3
3
tion cross section G across the T1g→ T2g absorption
3ϩ 3
3
profile.
picted in Fig. 6͑b͒, this V
T
→ T2g excitation relaxes to
1g
1T2g and is then transferred to Er3ϩ 4I11/2 ͑Sec. IV B 1͒. The
V. CONCLUSIONS
next step is an excited-state absorption ͑ESA͒ 4I11/2 to H9/2
2
on Er3ϩ ͓process F in Fig. 6͑b͔͒. The experimental evidence
K2NaScF6 codoped with Er3ϩ and V3ϩ is a candidate
for broadband sensitized upconversion from the far red and
near-infrared into the green and red. The present study
proves the principal feasibility. At 15 K broadband excitation
between 12 000 and 14 500 cmϪ1 will roughly double the
visible light output in an Er3ϩ and V3ϩ codoped crystal of
K2NaScF6 compared to single Er3ϩ doping. Even though
for this is the peak labeled G F in the upconversion excita-
*
tion spectrum Fig. 4͑b͒. Its position corresponds to the en-
ergy difference between the 2H9/2 levels and 10 198 cmϪ1 the
4
4
2
lowest I11/2 level of Er3ϩ. The I15/2→ H9/2 GSA is very
2
weak, see Fig. 3͑a͒, but the H9/2 position in a fluoride envi-
ronment is well established.22 We conclude that for excita-
tion energies between 14 300 and 14 500 cmϪ1 this is the
main UC mechanism. Since there are two radiative excitation
4
green Er3ϩ 4S3/2→ I15/2 luminescence can be observed by
3ϩ 3
3
eye upon V
T
→ T2g excitation in the full temperature
1g
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01

3380
J. Chem. Phys., Vol. 120, No. 7, 15 February 2004
Reinhard et al.
6
¨
¨
¨
range 15–200 K, the overall upconversion efficiency is very
low. At temperatures above 50 K efficient multiphonon re-
laxation processes, mainly on the V3ϩ, open up efficient loss
channels. These are intrinsic processes of V3ϩ in an octahe-
dral fluoride environment.13 In higher halides such as chlo-
rides and bromides, multiphonon relaxation would be less
M. Pollnau, W. Luthi, H. P. Weber, K. Kramer, H. U. Gudel, and R. A.
McFarlane, Appl. Phys. B: Lasers Opt. 62, 339 ͑1996͒.
7 B.-C. Hwang, S. Jiang, T. Luo, J. Watson, G. Sorbello, and N. Peygham-
barian, J. Opt. Soc. Am. B 17, 833 ͑2000͒.
T. Riedener, H. U. Gudel, G. C. Valley, and R. A. McFarlane, J. Lumin.
8
¨
63, 327 ͑1995͒.
9
10
11
12
13
¨
¨
P. Gerner, K. Kramer, and H. U. Gudel, J. Lumin. 102–103, 112 ͑2003͒.
O. S. Wenger and H. U. Gudel, J. Phys. Chem. B 106, 10011 ͑2002͒.
A. D. Kirk, N. Furer, and H. U. Gudel, J. Lumin. 68, 77 ͑1996͒.
C. Reinhard and H. U. Gudel, J. Lumin. 102–103, 373 ͑2003͒.
C. Reber, H. U. Gudel, G. Meyer, T. Schleid, and C. A. Daul, Inorg.
Chem. 28, 3249 ͑1989͒.
efficient, but in these materials the V3ϩ 1T2g state lies below
¨
3ϩ 4
¨
Er
I
and the sensitization cannot work. In addition, we
11/2
are limited in the amounts of V3ϩ and Er3ϩ ions that can be
incorporated into this lattice. This is ascribed to differences
in the ionic radii between host and guests. The inability to
incorporate a higher concentration of V3ϩ ions is particularly
detrimental to our objective of using this ion as a sensitizer.
Lattices based on Ga3ϩ such as GaF3 would provide a better
¨
¨
14 J. L. Sommerdijk, J. Lumin. 6, 61 ͑1973͒.
15 S. R. Bullock, B. R. Reddy, P. Venkateswarlu, S. K. NashStevenson, and J.
C. Fajardo, Opt. Quantum Electron. 29, 83 ͑1977͒.
16 R. D. Shannon, Acta Crystallogr., Sect. A: Struct. Crystallogr. Cryst.
fit for V3ϩ but might be problematic for Er3ϩ
.
Chem. 32, 751 ͑1976͒.
17
¨
E. Heumann, S. Bar, H. Kretschmann, and G. Huber, Opt. Lett. 27, 1699
͑2002͒.
ACKNOWLEDGMENTS
18 P. Xie and S. C. Rand, Appl. Phys. Lett. 63, 3125 ͑1993͒.
19 T. Catunda, L. A. O. Nunes, A. Florez, Y. Messaddeq, and M. A. Aegerter,
This work was financially supported by the Swiss Na-
tional Science Foundation and the Portland-Cement Stiftung.
Phys. Rev. B 53, 6065 ͑1996͒.
A. Al-Abdalla, L. Seijo, and Z. Barandiaran, J. Mol. Struct.: THEOCHEM
20
´
451, 135 ͑1998͒.
1 M. F. Auzel, C. R. Seances Acad. Sci., Ser. B 262, 1016 ͑1966͒.
21 R. H. Page, K. I. Schaffers, P. A. Waide, J. B. Tassano, S. A. Payne, W. F.
Krupke, and W. K. Bischel, J. Opt. Soc. Am. B 15, 996 ͑1998͒.
22 G. H. Dieke, Spectra and Energy Levels of Rare Earth Ions in Crystals,
edited by H. M. Crosswhite and H. Crosswhite ͑Intersience, New York,
2
¨
¨
S. R. Luthi, M. Pollnau, H. U. Gudel, and M. P. Hehlen, Phys. Rev. B 60,
162 ͑1999͒.
3 M. F. Joubert, S. Guy, S. Cuerq, and P. A. Tanner, J. Lumin. 75, 287
͑2001͒.
4
1968͒.
¨
D. R. Gamelin and H. U. Gudel, Topics in Current Chemistry ͑Springer-
23
¨
C. Reinhard and H. U. Gudel, J. Phys. Chem. B 106, 10051 ͑2003͒.
P. Gerner, O. S. Wenger, R. Valiente, and H. U. Gudel, Inorg. Chem. 40,
Verlag, Berlin, Heidelberg, 2001͒, Vol. 214.
24
5
¨
¨
O. S. Wenger, D. R. Gamelin, H. U. Gudel, A. V. Butashin, and A. A.
Kaminskii, Phys. Rev. B 60, 5312 ͑1999͒.
4534 ͑2001͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.111.185.72 On: Fri, 05 Dec 2014 11:43:01