You are on page 1of 19

ARTICLE IN PRESS

Solar Energy Materials & Solar Cells 90 (2006) 1189–1207


www.elsevier.com/locate/solmat

Luminescent layers for enhanced silicon solar cell


performance: Down-conversion
B.S. Richards1
Centre of Excellence for Advanced Silicon Photovoltaics and Photonics, University of New South Wales,
Sydney, New South Wales 2052, Australia
Received 8 July 2004; received in revised form 2 December 2004; accepted 1 July 2005
Available online 29 August 2005

Abstract

A down-converting (DC) layer placed on the front side of a silicon solar cell has the
potential to generate more than one low-energy photon for every incident high-energy photon.
Recent theory has predicted that DC in conjunction with a silicon solar cell can achieve a
conversion efficiency of up to 38.6%. This paper examines the application of rare-earth-doped
inorganic materials for achieving external quantum efficiencies greater than unity and
enhancing the conversion efficiency of silicon solar cells. Such DC mechanisms have been
applied in similar phosphor materials used by the lighting and display industries. The
opportunities for the DC of high-energy solar photons to multiple photons with energy greater
than the silicon bandgap are discussed.
r 2005 Elsevier B.V. All rights reserved.

Keywords: Luminescence; Down-conversion; Quantum cutting; Rare-earths; Solar cells; Photovoltaics

Abbreviations: CB, conduction band; CRT, cathode ray tube; DC, down-conversion; EQE, external
quantum efficiency; II, impact ionisation; IQE, internal quantum efficiency; IR, infrared; MP,
multiphonon; NIR, near infrared; PDP, plasma display panel; PV, photovoltaic; QC, quantum cutting (a
down-conversion process that occurs on one ion); PCE, photon cascade emission (alternative name for
quantum cutting); RE, rare-earth; UV, ultraviolet; VB, valence band; VUV, vacuum ultraviolet
E-mail addresses: b.richards@unsw.edu.au, bryce.richards@anu.edu.au.
1
Present address: Centre for Sustainable Energy Systems, The Australian National University,
Canberra, ACT 0200, Australia.

0927-0248/$ - see front matter r 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.solmat.2005.07.001
ARTICLE IN PRESS
1190 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

1. Introduction

The thermalisation of charge carriers, generated by the absorption of high-energy


photons, is one of the major loss mechanisms by which collected energy is
underutilised in a conventional solar cell. The method investigated here for reducing
these energy losses is via down-conversion (DC). Each high-energy photon
impinging on a DC layer, which overlies the solar cell, results in the emission of
more than one low-energy photon. The primary motivation for the application of the
purely optical DC process to photovoltaics (PV), at least initially, is to enhance the
conversion efficiency of existing silicon solar cells. Many decades of research have
gone into optimising silicon devices, and it is anticipated that further advances can
be made by modifying the input spectrum, rather than the electronic properties of
the device.
While DC phosphors have been investigated for decades in the lighting industry,
the possibility of down-converting sunlight to enhance the performance of solar cells
was only recently treated theoretically by Trupke et al. [1]. These authors
demonstrated that the optimum underlying solar cell bandgap was very close to
1.1 eV. Using a silicon solar cell with an ideal DC, a conversion efficiency of 38.6%
could be achieved under unconcentrated sunlight. This is in comparison to the
limiting efficiency of 30.9% for the same solar cell without a DC layer present. In
practice, much research on luminescence DC with an external quantum efficiency
(EQE) of less than unity has been performed for shifting the shorter wavelengths to
longer ones, which we refer to here as down-shifters. The application of down-
shifting layers in PV include luminescent solar concentrators [2–4] and measures to
overcome limitations in the front surface of some solar cell designs [5]. However, as
this paper focuses primarily on DC mechanisms that exhibit an EQE greater than
unity, down-shifting devices will not be discussed further here.
Rare-earth-doped luminescent materials are extensively used in the lighting
industry [6–8], as well as in the fabrication of cathode ray tubes (CRT) and plasma
display panel (PDP) technologies [9,10]. The motivation for using rare-earths is that
this family of elements luminesce over a wide range, from the near-infrared (NIR),
through the visible (vis) to the ultraviolet (UV). Their optical transitions involve 4f
orbitals, which are well shielded from their local environment by the outer
completely-filled 5s2 and 5p6 orbitals. Transitions between the different f levels are
parity forbidden, and hence the absorption coefficients are low (typically a few cm1)
and the slow emission rates result in long-lived, line-like emission.
Since the 1950s, rare-earth-doped phosphors have been optimised to achieve a
balanced blue, green and red output from the high-energy emission lines of low-
pressure mercury (Hg) discharge lamps. These UV emissions (185, 254, and 365 nm)
then excite carriers from the ground state to excited states within rare-earth atoms,
and radiative recombination between the levels produces light in the desired
wavelength (l) range. Commonly used fluorescent lamp phosphors today utilise
europium (Eu3+) ions for blue and red emission, terbium (Tb3+) ions for green
emission, and all exhibit an EQE) of 90–93% and strongly absorb the lexc ¼ 254 nm
Hg excitation line [8]. The energy conversion efficiency of the low-pressure Hg lamp
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1191

is close to 30% [8], being primarily limited by the efficiency of the Hg plasma (75%).
However, alternative plasma discharges that could substitute the Hg plasma are
being widely investigated today, as Hg is harmful to the environment, and the time it
takes Hg to evaporate results in a delay in the emission of light, preventing the use of
Hg lamps in special applications which require immediate start-up [11]. The xenon
(Xe) excimer plasma is the most promising candidate and is also used in PDPs;
however, its energy conversion efficiency is lower at 65% and the main Xe emissions
are in the vacuum ultraviolet (VUV) at 147 and 172 nm [8]. This precludes the
use of existing phosphors designed for 254 nm excitation, thus requiring the
development of new luminescent materials. In addition, since the Xe emission occurs
at higher energies than for Hg, the overall efficiency of visible light generation from
the lamp is reduced due to the increased energy gap between the excitation and
emission levels. Therefore, one promising solution to increase the efficiency is
to design phosphors that emit more than one visible photon per absorbed VUV
photon via a DC process known as quantum cutting (QC) or photon cascade
emission (PCE) process.
A typical inorganic phosphor consists of a host lattice doped with a few mol% or
less of an activator ion, such as the Eu3+ ion mentioned above, which is responsible
for the luminescence. As with many of the lanthanide ions, the energy levels of this
activator ion involved in the emission process exhibit only weak interactions with the
host lattice, and the optical transitions occur solely between the 4f levels that are well
shielded from their chemical environment by outer electrons. In the above-
mentioned Tb3+-activated phosphor, the Tb3+ ion is only absorbing below
230 nm, thereby not absorbing the main 254 nm Hg emission line. For this reason,
Ce3+ is added as a sensitiser ion as it is strongly absorbing from 250 to 350 nm and
can transfer the absorbed energy to the Tb3+ with near unity efficiency [12].
Energy transfer from one rare-earth ion (or the host lattice) to another rare-earth
ion is known to take place via several mechanisms [8,11]: (i) migration of electric
charge (electrons and holes); (ii) migration of excitons; (iii) resonance between atoms
with sufficient overlap integrals; and (iv) reabsorption of photons emitted by another
activator ion or sensitiser. Energy transfer is often required to overcome the small
absorption coefficients of the rare-earth ions, and energy absorbed in the host
material can be transferred to the desired activator. However, this may not be a one-
way process and energy already absorbed in a rare-earth ion can also migrate to host
lattice defects where it can recombine non-radiatively. Therefore, luminescent host
materials should be highly crystalline, have as few lattice defects and impurities as
possible, and the surface area of the crystals should be minimised [8].
Three luminescence DC mechanisms have been identified in the literature and
these will be discussed below: (1) QC using host lattice states or (2) on single ions,
and (3) DC using ion pairs. The term QC is usually used to describe the generation of
multiple photons using a single activator species, while DC involves energy transfer
between a sensitiser and activator species. Subsequently, the potential application of
these DC mechanisms to PV will be investigated, with an emphasis on luminescence
in the 900–1100 nm range where silicon solar cells exhibit their greatest spectral
response (A/W).
ARTICLE IN PRESS
1192 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

2. Down-conversion mechanisms

2.1. Quantum-cutting using host lattice states

QC on host lattice sites is based on the impact ionisation process, whereby multiple
lower energy electron–hole (e–h) pairs are generated per incident higher-energy
photon. Green [13] has recently reviewed the potential of the impact ionisation (II)
mechanism, for improving the efficiency of solar cells. The difference in this scenario
to other luminescent processes described here is that multiple electron hole pairs are
both generated and collected within the semiconducting material of the solar cell, and
not an overlying layer. Theory predicts that a threshold energy for II to occur (Eii)
equal to 1 or 1.5 times the bandgap of the semiconductor is required before a
conduction band (CB) electron is able to excite another electron from the valence
band (VB), depending on the effective masses of electrons and holes [14]. However, in
practice the required energy is greater due to the non-parabolic form of the energy
bands and the non-constant effective masses. The creation of a second e–h pair in
silicon (E g ¼ 1:12 eV) is not observed for an initial electron energy less than E ii 
2:6 eV [14]. Correspondingly, the gains in internal quantum efficiency (IQE) achieved
with silicon solar cells by this mechanism have been small, with an IQE of 105%
being reported at l ¼ 350 nm [15], while at l ¼ 250 nm the IQE increased to 150%
[16]. While bandgap engineering may possibly enhance the efficiency of the II process
by preventing carriers from relaxing via other means [13,17], this would require the
redevelopment of a solar cell based on the new material, and even then an increase in
the overall cell efficiency would not be guaranteed.
A few years ago, Nozik [18] suggested that the II process might be enhanced in
semiconducting nanocrystals (NC). Recently, this has been demonstrated experi-
mentally [19] using lead selenide (PbSe) NCs with E g ¼ 0:8121:00 eV. An impact
ionisation threshold of E ii  3E g was observed, with a peak response of 218% being
recorded for 3.8 eV incident photons. The latter result indicates that for some
interactions a third e–h pair is being generated, increasing the theoretical conversion
efficiency of a photovoltaic cell operating under concentrated sunlight to 48.3%,
compared to 43.9% for a cell without II [19]. Such devices, though, are far from
being realised, and there are only a few photons in the solar spectrum with high
enough energy to achieve an IQE4200%.
While the NCs described above are very inefficient light emitters due to the
dominance of the reverse process, Auger recombination, an extension of the II
mechanism has been investigated by the lighting community using solid-state
materials doped with luminescent centres. As shown in Fig. 1, this quantum-cutting
mechanism is performed via host lattice states and result in incident photons with
excitation energy E exc 42E g to generate two e–h pairs, each with energy EEg. This is
an interband Auger effect, where an electron is excited from the VB to an energy far
into the CB, and the excess energy can excite another electron over the bandgap,
such that one exciting photon produces two e–h pairs as shown in Fig. 1.
Subsequently, the e–h pairs recombine on the luminescent centres, yielding two
emitted photons for every photon absorbed.
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1193

inter- relax-
band ation emit
absorb
conduction Auger to two
single
band effect lum. low-
high-
centre energy
energy
photons
photon

valence lum.
band centre

Fig. 1. Interband Auger mechanism for the generation of two low-energy photons from each high-energy
incident photon within a solid-state material doped with luminescent centres.

For lighting applications, the bandgap of the host material must be larger than
3.0 eV to remain transparency to visible light, thus requiring E exc 46:0 eV, which is
possible using a Xe discharge. Although certain luminescent materials, such as
europium-doped yttrium oxide (Y2O3:Eu3+), have been found to exhibit very high
visible EQEs of up to 240%, these have required extremely high excitation of E exc ¼
23 eV [20]. Furthermore, E exc 415 eV was required to achieve an EQE greater than
unity [20]. The poor performance of phosphors in the near-bandgap region has been
attributed to surface recombination effects [21]. Non-radiative recombination rates
increase with decreasing penetration depth of the incident radiation until a minimum
EQE is reached at about 2.5Eg. For E exc 42:5E g the EQE increases with increasing
photon energy due to Auger interband transitions [20]; however by this stage a Xe
lamp can no longer be used as an effective excitation source and therefore the
practical relevance of this QC mechanism is limited [7].

2.2. Quantum-cutting on single rare-earth ions

The second QC mechanism occurs on single rare-earth ions, where well-separated


energy levels can be used to generate more than one visible photon out of one high-
energy photon. This was first reported by two research groups in the mid-1970s
[22,23] using praseodymium (Pr3+) as the activator ion. The initial absorption of a
185 nm photon takes place in the YF3 host lattice, and the energy is transferred into
the underlying 1S0 state via a 4f–5d transition, as shown in Fig. 2. From there a blue
photon (408 nm) is emitted from the 1S0- 1I6 transition, and red photon (620 nm)
from the 3P0- 3H6 transition with an EQE of 140715% [22].
Research has also shown that it is important that the 5d band lies above the 1S0
level, otherwise non-radiative decay into the 1S0 level and subsequent emission will
not occur. In this case, the IQE of the phosphor cannot exceed 100% [27,28]. The
EQEs for Pr3+ were calculated using Judd–Ofelt theory, which agreed with the
experimental results (150%), while the EQE increased to 176% when UV emissions
from the 1S0 level were taken into account [25]. However, modelling performed on
other rare-earth ions has not revealed visible EQEs greater than unity [24–26].
A generalisation of this quantum-cutting scenario is shown in Fig. 3(a), with
unwanted emissions in the IR and UV (thin lines) and non-radiative recombination
ARTICLE IN PRESS
1194 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

5d band

50
1
S0

40

Energy (103 cm-1)


30

3
P
3 2 1
P I6
3 1
20 P0

1
10 G4

3
FJ
3
0 HJ

Fig. 2. Quantum-cutting on YF3:Pr3+, where a 185 nm photon is absorbed in the YF3 host and the energy
is transferred into the 1S0 state. The first ( 1S0- 1I6) and second ( 3P0- 3H6) radiative transitions result
in the emission of a blue (408 nm) and red (620 nm) photon, respectively, with an EQE of 140715%.

Fig. 3. Hypothetical quantum cutting mechanisms for rare-earth ions: (a) two photon emission from a
single ion; and down-conversion with ion pairs by (b) cross-relaxation from ion A to ion B (A) and energy
transfer from ion A to ion B (B) with emission from ion B, (c,d) cross-relaxation followed emission from
ions A and B (adapted from Wegh et al. [31]).

(wavy lines) also competing with the desired emission of two visible photons (thick
lines) [11]. A higher EQE of 210% has been achieved by exciting the 6F4 level of
holmium (Ho3+) doped lanthanum chloride (LaCl3), however only one-sixth of the
photons exhibited visible emission wavelengths, with more than half the photons
being emitted in the IR at 2 mm [29].
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1195

2.3. Down-conversion using rare-earth ions pairs

The third QC mechanism occurs on ion pairs of physically interacting rare-earth ions.
In these materials, one of the ions in the pair has the ability to show QC. By interaction
with the other ion in the pair, part of the energy is transferred to this latter ion. In this
manner, the previous IR and UV losses encountered with a single ion can be avoided by
selecting the appropriate ions [11]. The possibility of achieving two-photon emission via
energy transfer was predicted by Dexter in the 1950s [30] and is the opposite of the
‘‘Addition de Photons par Transfert d’Energie’’ up-conversion mechanism discovered
by Auzel [29] in 1966, and is therefore referred to as DC [11]. The DC term will be used
throughout the remainder of the paper to refer to two-photon emission.
Fig. 3(b–d) illustrates generalised energy level diagrams for three DC mechanisms
that involve energy transfer between two different rare-earth ions (A and B). Ion A
exhibits emission from a high-energy level, while ion B supports energy transfer. Fig.
3(b) indicates two photon emission from ion pairs by cross-relaxation from ions A to
B (denoted by A) and energy transfer from ions A to B (denoted by B) with emission
from ion B. Fig. 3(c,d) depicts a cross-relaxation mechanism followed the emission
of photons from both ions A and B.
An example of the DC process for the phosphor LiGdF4:Eu3+ is shown in Fig. 4(a)
[11], where after excitation of the gadolinium (Gd3+) ion, part of the energy is
transferred to Eu3+ by cross-relaxation (A), while the remaining energy is transferred
to Eu3+ by energy transfer (B). This system results in emission from a single Eu3+
peak at 612 nm with a demonstrated IQE of almost 200% [11,31]. The EQE of this
system though is quite poor at 32% due to, firstly, the weak absorption of the 6GJ level
of the Gd3+ ion and, secondly, the strong absorption in the LiGdF4 host [32], which is
not transferred to the 6GJ level. However, by taking advantage of the VUV absorption
of the host, the DC phosphor was improved by using the 4f–5d absorption of the
erbium (Er3+)ion in LiGdF4:Er3+,Tb3+ [31,33] as shown in Fig. 4(b). QC occurs on
the Er3+ ion, with energy being initially transferred by cross-relaxation to Gd3+,
which subsequently transfers its energy to Tb3+, resulting in green emission. The Er3+
ion exhibits green emission after the cross-relaxation step, and the resulting maximum
EQE is 130%, although as part of the Tb3+ emission being situated in the UV part of
the spectrum the actual upper limit for the visible EQE is 110% [33].

3. Non-radiative recombination mechanisms

The minimum prerequisite for generation of luminescence by any material is the


presence of at least two metastable excited states. For efficient DC to be observed,
the metastable excited states must have lifetimes sufficiently long so that ions have a
high probability of resulting in luminescence, rather than relaxing by non-radiative
mechanisms. The lifetime of the excited state can be measured by monitoring the
exponential decay of its luminescence intensity I,
IðtÞ ¼ Ið0Þektot t , (1)
1196
ARTICLE IN PRESS

Fig. 4. Energy level diagrams of two down-conversion mechanisms with a potential quantum efficiency of greater than unity. The two-step energy transfer
from Gd3+ to Eu3+ in the LiGdF4:Eu3+ phosphor system are shown in (a) and the steps are indicated by A (cross-relaxation) and B (transfer of remaining
energy on Gd3+ to Eu3+). Down-conversion is also possible by (b) energy transfer upon excitation in the 4f10–5d state of the Er3+ ion of LiGdF4:Er3+,Tb3+
(adapted from Wegh et al. [31].
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1197

where ktot, the decay rate constant, is the sum of the contributing radiative (krad) and
non-radiative (knr) decay processes and leads to the definition of the effective lifetime
(t1
eff ) of the excited state

ktot krad þ knr ¼ t1


eff . (2)
Non-radiative transitions of rare-earth ions in a glass host can occur via three
different mechanisms: (i) multiphonon (MP) relaxation; (ii) cross-relaxation; and,
(iii) co-operative up-conversion [34]. Of these possible non-radiative processes, MP
relaxation generally makes the largest contribution to ktot. MP relaxation occurs by a
simultaneous creation of several phonons, which suffice to equal the energy of the
transition between the excited level and the next lower level. Since the probability for
MP decay decreases exponentially with the number of required phonons to bridge
the energy gap, it is desirable to surround the active ion by a matrix that possesses
low vibrational energies. The highest energy longitudinal optical phonon modes that
dominate in glasses are typically stretchings of the anions against the glass forming
cations, whose frequency varies with the glass composition. The highest vibrational
energies in various types of inorganic glasses are listed in Table 1.
Due to the high shielding of their f electrons, rare-earth ions are characterised by
an extreme insensitivity to their environment, in particular having very small
electron–phonon coupling strengths in their excited states. The dependence of the
MP relaxation rate constant across an energy gap DE between f–f states on the
vibrational energy is well described by the energy gap law [35]
kmp ¼ Cepb , (3)
where C and b are positive constants characteristic of the material. The minimum
number of maximum energy phonons required for a transition between states
separated by an energy gap DE is given by
DE
p¼ . (4)
_nmax
As a rule of thumb for 4f systems, MP relaxation processes will dominate up to
about p ¼ 526 [36]. Thus, the same rare-earth luminescence centre in a borate or

Table 1
Maximum vibrational energies :nmax (cm1) in several inorganic materials

Material _nmax (cm1)

Borate 1400
Phosphate 1100
Silicate 1000–1100
Germanate 800–975
Tellurite 600–850
Fluoride 500–600
Chalcogenide 200–300
Bromide 175–190
Iodide 160
ARTICLE IN PRESS
1198 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

fluoride glass may exhibit very different luminescent properties due to their
dissimilar MP relaxation rates. This is demonstrated by the absence of emission from
the 3P0 level of Pr3+ in the borate glass LaMgB5O10 and, hence, QC not being
observed [26].
In addition, the MP relaxation rate is greatly enhanced by the presence of residual
hydroxyl (OH) ions in the glass host, due to coupling between the rare-earth states
and the high energy stretching vibrations (typically 3200 cm1) of the OH groups.
For example, this contribution to non-radiative decay is very important in the case
of the Er3+ ion, where two OH phonons are sufficient to bridge the energy gap
between the 4I13/2 and 4I15/2 levels [34]. Thus, the residual OH content of the glass
must be kept to a minimum. This is of primary importance to sol–gel derived glasses,
where organic contaminants may remain and need to be removed by high-
temperature annealing, or in zeolites [8].
The two remaining non-radiative processes, cross-relaxation and co-operative up-
conversion, both involve rare-earth ion pair interactions, and the luminescence
intensity decreases as the rare-earth ion concentration increases. Cross-relaxation
may take place between any two closely spaced rare-earth ions that happen to have
two pairs of energy levels separated by the same amount. One of the ions, in an
excited state, gives half of its energy to a ground state ion, so that both ions end up in
the intermediate level. From this level, they both relax quickly to the ground state,
via MP relaxation. The cross-relaxation process is believed to be the dominant factor
in the concentration quenching of Nd3+ in glass matrices. As shown in Fig. 3(b–d),
cross-relaxation can be an important mechanism facilitating DC, and some degree of
control is possible via careful selection of the rare-earth dopants.
The spontaneous co-operative up-conversion process may occur when two
neighbouring ions are in an excited state: one of them, A, gives its energy to the
other, B, thus promoting B to a higher level, while A relaxes to the ground state.
From this higher energy level, the B ion relaxes rapidly, radiatively or non-
radiatively. This co-operative up-conversion process is believed to be the major cause
of concentration quenching in Er3+-doped glasses. Er3+ has no intermediate states
between the 4I13/2 metastable level and the 4I15/2 ground state, thus cross-relaxation
between an excited ion and one in the ground state cannot occur.

4. Limitations of down-conversion for lighting, display and photovoltaic applications

Many challenges remain before down-converting phosphors with an EQE of


4100% will be implemented in lighting and display applications, let alone the PV
industry. The luminescent properties that are required for the implementation of DC
luminescent materials in these industries are compared in Table 2. The DC
phosphors described above have not met some of these requirements. For example,
the YF3:Pr3+ QC phosphor described above, although the red emission is located at
almost exactly the right wavelength for the human eye, the blue emission is too
deeply blue and poor colour rendering results [6]. While some of these above
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1199

Table 2
Comparison of properties of luminescent DC materials required for successful implementation into the
lighting, display and photovoltaics industries

Luminescent Description Lighting Displays Photovoltaics


property

Emit only visible — | | 


light
Good chromaticity Well-balanced emission to | | 
and saturation allow the accurate
reproduction of all colours
Brightness and Critical to readability in x | 
contrast ambient light
High colour Broadband emission such that |  
rendering illuminated objects have a
natural appearance
Maintenance Degradation of the phosphor | | (|)
by high energy photons or
during device fabrication
sequence
Efficiency Brightness per unit input | | |
powera (lm/W)
Choice of material Ease of manufacture, stability, | | |
environmental concerns
Minimal reflection Need to couple light into | | |
phosphor
Cost High price of rare-earth ions | | |
a
For lighting, the luminous efficiency is also a function of the optical response of the human eye.

problems can be avoided in PDP applications by using colour filters [6], this adds
further to the cost of the final product and reduces the efficiency.
The majority of these limitations are not relevant to the application of rare-earth
phosphors to PV devices, which are able to absorb any photon that possesses energy
greater than the bandgap of the semiconductor material Eg. In the case of silicon,
this means photons with a wavelength of less than 1100 nm. The constraints that
then remain, which are relevant for all phosphor applications are, firstly, that
phosphors should exhibit a high emission lifetime, such that radiative recombination
process are allowed to dominate, resulting in a high EQE.
Secondly, the phosphors should exhibit good stability, including ionic bombard-
ment of the host material under, surface amorphisation, photo-oxidation of the rare-
earth centres, and trapping of energy by impurities [37]. One reason for the less-than-
ideal stability of the VUV-excited phosphors in PDPs is the high energy and
extremely small penetration depth of the photons. The formation of defects and
traps in the host material results in the energy being absorbed in the host instead of
being transferred to the activator ion. This results in the phosphor performance
degrading by a few tens of per cent after several thousand hours of operation [37].
Oxidation of a phosphor (e.g. from Eu2+ to Eu3+) can occur during the baking
process, which is required after coating a PDP with phosphors [37]. Therefore,
ARTICLE IN PRESS
1200 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

luminescent layers for PV deposited in the same manner must also be compatible
with baking temperatures of typically 400 1C. However, the damage caused by UV
photons from the sun should be much less with the higher energy excitation used in
lighting and PDPs.
Thirdly, the high price of some of the rare-earth minerals and the necessary
purification processes for lanthanides [8] remains a barrier to implementation. In the
lighting industry, phosphors contribute tens of percent to the final price of a
fluorescent lamp [6]. However, it is anticipated that the use of deposited thin films
and the incorporation of rare-earths into polymer or zeolite hosts may further reduce
cost as well as overcoming the low-absorptivity. In addition, a PV module is a much
higher value and longer lasting product than a fluorescent lamp and the contribution
of rare-earths to the final cost of the product will be small.
Fourthly, the phosphor host may be required to exhibit minimal reflection in order
to couple light into the excited levels of the rare-earth ion, unless energy transfer
from the host lattice to the rare-earth ion is capitalised upon.
Finally, the selection of the host material is a trade-off in some applications. While
fluoride-based phosphors exhibit excellent performance in fluorescent lamps, several
disadvantages with fluoride hosts have been identified [27]. Fluorides involve
significance environmental and safety issues in their production and there should be
no possibility for the reaction of the fluoride host with water vapour in the process.
Fluorides are also known to be damaged by VUV radiation, resulting in the
formation of defects, which will reduce the EQE of the luminescence. For this
reason, zeolite (aluminosilicate) hosts are popular for phosphors due to their
enhanced stability. The first two issues will also be of concern in a PV environment;
however, it remains to be seen if UV radiation in the solar spectrum can lower the
performance of fluoride phosphors.

5. Down-conversion mechanisms for PV applications

Considering the three QC or DC mechanisms outlined above—host lattice states,


on single ions, and ion pairs—we now focus on the possible application of these
mechanisms to enhancing PV cell performance.

5.1. Quantum-cutting using host lattice states

Applying the same treatment as that previously discussed for QC on host lattice
states for lighting applications, for a QC material with E g ¼ 1:1 eV, tuned to emit for
a silicon solar cell, a minimum EQE should be observed at about 2.75 eV and a
EQE41 would only be expected for E exc 45:5 eV (226 nm). However, there are no
photons of this energy in the global AM1.5 spectrum received on earth. Two slight
possibilities for improvement in this area are improved surface passivation
techniques from PV that can be applied to the QC phosphors, and enhanced
luminescence from semiconducting NCs.
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1201

5.2. Quantum-cutting on single rare-earth ions

To the author’s knowledge there have been no experimental studies focussed on


achieving the emission of two NIR photons from UV or visible excitation via QC in
phosphors doped with a single rare-earth ion. Although the possibility of the
emission of two visible photons per incident high energy photon has been ruled out
[6], the relaxed requirements for the emission of two NIR photons on single rare-
earth ions or on rare-earth ion pairs suggest that there is still some hope for this QC
mechanism and further studies are required in this area. A close examination of the
Dieke diagram [38] of rare-earth levels did not reveal an obvious ion that was
capable of being excited in the range E exc ¼ 2:224:1 eV and may produce two
emissions with lo1100 nm.

5.3. Down-conversion using rare-earth ions pairs

The theory for cooperative DC as shown in Fig. 3(b) was developed by Dexter
[30], who predicted that EQEs greater than unity could be achieved. In addition,
Dexter suggested that the emission would increase linearly as a function of both
excitation intensity and activator ion concentration, up until a critical concentration
of about 1%. However, rather than DC, the reverse of this mechanism was first
observed experimentally via up-conversion of NIR to visible radiation. Several
researchers [39,40] determined that cooperative excitation was a possible up-
conversion process in Tm3+–b3+ co-doped systems. However, an additional up-
conversion process called step-wise excitation is more probable due to the better
match between Tm3+and Yb3+ energy levels [40]. The step-wise excitation involves
an MP cross-relaxation from the 2F5/2 level of the Yb3+ ion to the 3H5 level of the
Tm3+ ion. Thus, the drawback is that from this level it can relax back down to 3H4
and will most likely emit at a wavelength of 1700 nm, not possessing enough energy
to be collected by a silicon solar cell. More recently, several groups realised DC of
UV or visible light to NIR light using rare-earth ion pairs [41–43] and these
possibilities are discussed in more detail below.
Tanner et al. [41] reported two photon emission from Cs2LiTmCl6 under 476.5 nm
argon ion laser excitation which populates the 1G4 levels of thulium (Tm3+). Visible
emission from this level was not observed due to cross-relaxation quenching and
only NIR emission in the range 12534–10991 cm1 (798–910 nm) was observed. The
higher-energy luminescence is a result of 3F4- 3H6 transitions in Tm3+, while the
lower-energy transitions were attributed to 4F3/2- 4I9/2 transitions in trace
neodymium (Nd3+) impurities. The EQE of this DC mechanism were not stated,
and it is believed that some NIR transitions on the Tm3+ ion may occur at
wavelengths of 1700 nm ( 3H5- 3H6).
Str˛ek et al. [42,43] investigated cooperative DC using Tb3+-doped KYb(WO4)2.
The crystal was excited with a 308 nm laser, designed to only excite Tb3+ ions. In
addition to visible emission (16,000–21,000 cm1) from the Tb3+ ions ( 5D4- 7FJ),
NIR emission from the Yb3+ ions ( 2F5/2- 2F7/2) is observed as well, as shown in
Fig. 6. The intensity of the Yb3+ emission is significantly greater at room
ARTICLE IN PRESS
1202 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

temperature, suggesting that the energy from phonons is required to assist with
energy transfer between Tb3+ and Yb3+ ions. However, the energy level diagram for
Tb3+ and Yb3+ ions, illustrated in Fig. 5(b), indicates that an energy mismatch of
DE  4200 cm1 exists for the 5D4- 7F0 and 2F5/2- 2F7/2 transitions, respec-
tively. The maximum phonon energy in the KYb(WO4)2 crystal is about 1070 cm1,
and any cross-relaxation process would require at least four maximum energy
phonons to bridge the energy gap DE (Fig. 6).
Str˛ek et al. [43] determined that the Yb3+ emission intensity increases with power
more slowly than the Tb3+ intensities. This is assumedly due to the fact that only
some of the excited Tb3+ ions transfer their energy to Yb3+ ions. In addition to the
phonon-assisted process described above, a DC process has also been observed,
where the energy of the excited Tb3+ ion is distributed between two different Yb3+
ions, 5 D4 !2 F 5=2 þ2 F 5=2 . This is a DC mechanism of the type shown in Fig. 3(b), and
the transitions for the Tb3+–Yb3+ ions are illustrated in Fig. 5(b). This is the
opposite of the cooperative sensitisation up-conversion mechanism investigated by
Salley et al. [44] in Tb3+–Yb3+ doped compounds. DC emission has also been

Fig. 5. Down-conversion facilitated by energy transfer from (a) one Tm3+ ion to two Yb3+ ions, and (b)
one Tb3+ ion to two Yb3+ ions, resulting in the emission (thick arrows) peak at 980 nm. However, other
radiative recombination paths also exist, thus reducing the efficiency of the down-converted emission.

Fig. 6. Luminescence emission spectrum of Tb3+-doped KYb(WO4)2 at room temperature, demonstrat-


ing NIR light emitted from Yb3+ ion at about 10,000 cm1 (adapted from Str˛ek et al. [43]).
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1203

observed in a Tb3+–Yb3+ co-doped trinuclear complex, by exciting the 5D4 level


directly at 488 nm [45]. The observed 980 nm luminescence was attributed to the
Yb3+ ion; however, more investigations are still to be performed to gain a better
understanding of the energy transfer processes.
The intensity of DC light could be expected to exhibit a linear increase as a
function of excitation power density [1]. However, Mishra et al. [46] point out that a
mathematical error in the work of Dexter [30] means that the simultaneous transfer
of excitation energy from a sensitiser to two activators can, at best, be a second-order
process. The cooperative DC process competes with other recombination mechan-
isms, such as the non-resonant phonon-assisted energy transfer. Energy from the of
2
F5/2 state of Yb3+ ion may be transferred to two Tb3+ ions in their lower-energy
states 7FJ, 2 F 5=2 !7 F J þ7 F J . From these levels all carriers will recombine non-
radiatively. Measured emission lifetimes from Tb3+–Yb3+ pairs have also been
observed to significantly decrease under higher excitation powers [43]. At increased
temperature, the probability of the DC process was found to increase, most likely
because of the better matching of the energy levels due to the assistance of phonons.
Based on the non-resonant energy transfer model presented by Miyakawa and
Dexter [47], Str˛ek et al. [43] expressed the rate equations for this three-level DC
system and found that the emission from the Yb3+ ion is expected to be proportional
1=2
to the square-root of the incident light intensity, I 0 , due to its dependence on the
occupation levels of three states. This demonstrates that the cooperative DC
mechanism is indeed non-linear, and exactly the same intensity of light that is
expected via the non-linear up-conversion process [48,49]. Mishra et al. [46] clarify
that two-photon emission may occur as a first order process via an interband Auger
process or a modified cooperative DC process.

6. Down-conversion materials for photovoltaic applications

If a DC with an EQE4100% can be found, there are several possibilities and


challenges for incorporating the DC into a PV device or module. Encapsulated solar
cells have three transparent layers above the silicon device: an antireflection (AR)
coating; the encapsulant, most commonly ethylene vinyl acetate (EVA); and a glass
cover sheet. Therefore, a DC layer could either be inserted between, or incorporated
into, any of the above layers.
The ideal down-converting material for a silicon solar cell would absorb all
photons greater than 2Eg, and re-emit these at an energy slightly greater than Eg,
while remaining transparent to photons in the range Eg–2Eg. As there appears to be
no likely QC mechanism to generate two NIR photons from UV/blue radiation
using single ions, the most likely scenario will be some form of sensitised
luminescence based on the suitable emission of the Yb3+ ion. If the example of
energy transfer from Tb3+ or Tm3+ ions to the Yb3+ ion is used, the 5D4 level of
Tb3+ and 1G4 level of Tm3+ reside in an energy range of 470–485 nm (2.56–2.64 eV).
However, due to the low absorption coefficient of rare-earth ions, the most suitable
host candidate would be a wide bandgap semiconductor with a bandgap of greater
ARTICLE IN PRESS
1204 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

than 2.7 eV. This would ensure strong absorption of photons above the bandgap and
the carriers can relax from the CB down to the excited level of the rare-earth ion via
energy transfer from the host. Known semiconductors with a bandgap in this range
include zinc selenide (ZnSe, 2.7 eV), silicon carbide (6H–SiC, 3.0 eV) and titanium
dioxide (TiO2, 3.0–3.2 eV).
Of these materials, crystalline ZnSe is commonly used for the fabrication of high
refractive index optics, while TiO2 has been commonly used as an AR coating [50].
Motivations for choosing TiO2 could include its high-refractive index, excellent
transparency to visible/NIR light, low maximum phonon energy (_nmax p650 cm1 ),
TiO2 should also exhibit minimal degradation from high-energy photons due to its
high chemical bond energy, and that energy transfer from the TiO2 lattice to rare-
earth activators has been reported in the literature [51,52]. The latter would allow
photons with energy greater than the TiO2 bandgap (E g ¼ 3:05 for the rutile phase)
to be absorbed and their energy efficiently transferred to a rare-earth ion that has an
excited state lying just below the TiO2 CB. The energy transfer mechanism observed
by Frindell et al. [51,52], the energy transfer mechanism occurred from TiO2 defect
states lying at 2.35–2.59 eV rather than the TiO2 bandgap; however, it is anticipated
that with larger crystal sizes transitions from the TiO2 CB will dominate.
The primary challenge then remaining is to capture all of the isotropically emitted
luminescence from the DC with the solar cell. Trupke et al. noted that for a DC
located on the front surface of the solar cell, only the fraction of photons that are
emitted towards solar cell contribute to the photocurrent of the cell, requiring a high
refractive index for the solar cell and DC to achieve a high conversion efficiency [1].
This is because the escape cone angle (w) for luminescence generated within the DC
layer increases with n, given by w ¼ sin1 (1/n), which is 421 at n ¼ 1:5, but only 161
at n ¼ 3:6. Although a high n DC layer will maximise the amount of luminescence
collected by the underlying solar cell, the amount of incident light reflected off the
front surface will be over 30% for the scenario of n ¼ 3:6 as used by Trupke et al. [1].
Naturally, this value could be reduced by the addition of an AR coating; however,
this would also have the undesired effect of assisting more luminescence from
escaping out the front of the device. The author is investigating the combination of a
wavelength selective filter, and down-converted or down-shifted luminescence as one
possible solution to this problem [4].

7. Conclusions

In summary, there would appear to be limited options for the successful


application of rare-earth luminescent DC to silicon PV devices. QC using host lattice
states, based on Auger interband transitions, typically requires photon energies
much greater than those available in the solar spectrum, while the author has not
been able to identify a single rare-earth ion that might exhibit QC of one UV/blue
photon into two NIR photons with an energy greater than 1.1 eV. Cooperative DC,
which capitalises upon energy transfer between rare-earth ion pairs of Yb3+ and
either Tm3+ or Tb3+ has been observed. While this would be well-suited to silicon
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1205

PV, with the Yb3+ emission at 980 nm, the emission intensity from this level has been
1=2
shown to vary with the square-root of the incident light intensity, I 0 ,
demonstrating that cooperative DC is not a linear process. Finally, due to the
trade-off between increased front surface reflection losses and luminescence not be
collected due to the escape cones, it would appear that an EQE significantly greater
than unity would be required before any benefit would be seen in PV device
performance.

References

[1] T. Trupke, M.A. Green, P. Würfel, Improving solar cell efficiencies by down-conversion of high-
energy photons, J. Appl. Phys. 92 (3) (2002) 1668–1674.
[2] A. Goetzberger, W. Gruebel, Solar energy conversion with fluorescent collectors, Appl. Phys. 14 (2)
(1977) 123–139.
[3] P.S. Friedman, C.R. Parent, Luminescent Solar Concentrator Development Final Report, SERI
subcontract no. XE-2-02145-01, 1987, 193pp.
[4] B.S. Richards, A. Shalav, R.P. Corkish, A low escape-cone-loss luminescent solar concentrator,
in: 19th European Conference on Photovoltaic and Solar Energy Conversion, Paris, 2004,
pp. 113–116.
[5] T. Maruyama, A. Enomoto, K. Shirasawa, Solar cell module colored with fluorescent plate, Sol.
Energy Mater. Sol. Cells 64 (2000) 269–278.
[6] C.R. Ronda, Phosphors for lamps and displays: an applicational view, J. Alloys Comp. 225 (1995)
524–538.
[7] C.R. Ronda, Luminescent materials with quantum efficiency larger than 1, status and prospects,
J. Lumin. 100 (2002) 301–305.
[8] T. Jüstel, H. Nikol, C. Ronda, New developments in the field of luminescent materials for lighting
and displays, Angew. Chem. Int. Ed. 37 (1998) 3084–3103.
[9] P.H. Holloway, T.A. Trottier, B. Abrams, C. Kondoleon, S.L. Jones, J.S. Sebastian, W.J. Thomas,
Advances in field emission displays phosphors, J. Vac. Sci. Technol. B 17 (2) (1999) 758–764.
[10] P.D. Rack, P.H. Holloway, The structure, device physics, and material properties of thin film
electroluminescent displays, Mater. Sci. Eng. R21 (1998) 171–219.
[11] R.T. Wegh, H. Donker, K.D. Oskam, A. Meijerink, Visible quantum cutting in LiGdF:Eu3+ through
downconversion, Science 283 (1999) 663–666.
[12] T. Jüstel, D.U. Wiechert, C. Lau, D. Sendor, U. Kynast, Optically functional zeolites: evaluation of
UV and VUV stimulated photoluminescence properties of Ce3+- and Tb3+-doped zeolite X, Adv.
Funct. Mater. 11 (2) (2001) 105–110.
[13] M.A. Green, Third Generation Photovoltaics, Springer, Berlin, 2004 (Chapter 7).
[14] K.W. Böer, Survey of Semiconductor Physics: Electrons and Other Particles in Bulk Semiconductors,
Van Nostrand Reinhold, New York, USA, 1990.
[15] F.J. Wilkinson, A.J.D. Farmer, J. Geist, The near ultraviolet yield of silicon, J. Appl. Phys. 54 (1983)
1172–1174.
[16] M. Wolf, R. Brendel, J.H. Werner, Quantum efficiency of silicon solar cells at low temperatures, in:
Tech. Digest of the International PVSEC-9, Miyazaki, Japan, 1996, pp. 519–520.
[17] S. Kolodinski, J.H. Werner, H.J. Quessier, Potential of Si1xGex alloys for Auger generation in
highly efficient solar cells, Appl. Phys. A 61, 535–539.
[18] A.J. Nozik, Quantum dot solar cells, Physica E 14 (2002) 115–120.
[19] R.D. Schaller, V.I. Klimov, High efficiency carrier multiplication on PbSe nanocrystals: implications
for solar energy conversion, Phys. Rev. Lett. 92 (18) (2004) 186601.
[20] J.K. Berkowitz, J.A. Olsen, Investigation of luminescent materials under ultraviolet excitation
energies ranging from 5 to 25 eV, J. Lumin. 50 (1991) 111–121.
ARTICLE IN PRESS
1206 B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207

[21] E.L. Benitez, D.E. Husk, S.E. Schnatterly, C. Tarrio, A surface recombination model applied to large
features in inorganic phosphor efficiency measurements in the soft X-ray region, J. Appl. Phys. 70 (6)
(1991) 3256–3260.
[22] W.W. Piper, J.A. DeLuca, F.S. Ham, Cascade fluorescent decay in Pr3+-doped fluorides:
achievement of a quantum yield greater than unity for emission of visible light, J. Lumin. 8 (4)
(1974) 344–348.
[23] J.L. Sommerdijk, A. Bril, A.W. De Jager, Two photon luminescence with ultraviolet excitation of
trivalent praseodymium, J. Lumin. 8 (4) (1974) 341–343.
[24] R. Pappalardo, Calculated Quantum Yields for Photon-Cascade Emission (PCE) for Pr3+ and Tm3+
in fluoride hosts, J. Lumin. 15 (1976) 159–193.
[25] K.J.B.M. Nieuwesteeg, The dependence of quantum yields on the Judd-Ofelt intensity parameters in
Europium-, Terbium-, and Thulium-activated solid materials, Philips Journal of Research 44 (1989)
385–406.
[26] A.M. Srivastava, D.A. Doughty, W.W. Beers, Photon cascade luminescence of Pr3+ in LaMgB5O10,
J. Electrochem. Soc. 143 (12) (1996) 4113–4115.
[27] A.A. Setlur, H.A. Comanzo, A.M. Srivastava, W.W. Beers, W. Jia, S. Huang, L. Lu, X. Wang, W.M.
Yen, Advances in the development of quantum splitting phosphors, Mater. Res. Soc. Symp. Proc. 667
(2001) G1.6.1–G1.6.6.
[28] J.F. Porter, H.W. Moos, Direct evidence for and transition rates of ion-pair interactions in
LaCl3:Ho3+, Phys. Rev. 152 (1) (1966) 300–305.
[29] F. Auzel, Compteur quantique par transfert d’energie entre deux ions de terres rares dans un
tungstate mixte et dans un verre, Comp. Rend. Acad. Sci. B 262 (1966) 1016–1019.
[30] D.L. Dexter, Possibility of luminescent quantum yields greater than unity, Phys. Rev. 108 (3) (1957)
630–633.
[31] R.T. Wegh, H. Donker, E.V.D. van Loef, K.D. Oskam, A. Meijerink, Quantum cutting through
downconversion in rare-earth compounds, J. Lumin. 87–89 (2000) 1017–1019.
[32] C. Feldmann, T. Jüstel, C.R. Ronda, D.U. Weichert, Quantum efficiency of down-conversion
phosphor LiGdF4:Eu, J. Lumin. 92 (2001) 245–254.
[33] K.D. Oskam, R.T. Wegh, H. Donker, E.V.D. van Loef, A. Meijerink, Downconversion: a new route
to visible quantum cutting, J. Alloys Comp. 300–301 (2000) 421–425.
[34] M.C. Gonc- alves, L.F. Santos, R.M. Almeida, Rare-earth-doped transparent glass ceramics, C. R.
Chimie 5 (2002) 845–854.
[35] J.M.F. van Dijk, M.F.H. Schuurmans, On the nonradiative and radiative decay rates and a modified
exponential energy gap law for 4f–4f transitions in rare-earth ions, J. Chem. Phys. 78 (9) (1983)
5317–5323.
[36] H.U. Güdel, M. Pollnau, Near-infrared to visible photon upconversion processes in lanthanide doped
chloride, bromide and iodide lattices, J. Alloys Comp. 303–304 (2000) 307–315.
[37] B. Moine, G. Bizari, Rare-earth doped phosphors: oldies or goldies?, Mater. Sci. Eng. B 105 (2003)
2–7.
[38] G.H. Dieke, Spectra and Energy Levels of Rare Earth Ions in Crystals, Interscience Publishers, New
York, 1968.
[39] V.V. Ovsyankin, F.F. Feofilov, Cooperative sensitization of luminescence in crystals activated with
rare earth ions, JETP Lett. 4 (11) (1966) 317–318.
[40] T. Miyakawa, D.L. Dexter, Cooperative and stepwise excitation of luminescence: trivalent rare-earth
ions in Yb3+-sensitized crystals, Phys. Rev. B 1 (1) (1970) 70–80.
[41] P.A. Tanner, C.S.K. Mak, W.M. Kwok, D.L. Phillips, M.-F. Joubert, Luminescence from the 3P2
state of Tm3+, J. Phys. Chem. B 106 (2002) 3606–3611.
[42] W. Strek, P. Dereń, A. Bednarkiewicz, Cooperative processes in KYb(WO4)2 crystal doped with
Eu3+ and Tb3+ ions, J. Lumin. 87–89 (2000) 999–1001.
[43] W. Strek, A. Bednarkiewicz, P. Dereń, Power dependence of luminescence of Tb3+-doped
KYb(WO4)2 crystal, J. Lumin. 92 (2001) 229–235.
[44] G.M. Salley, R. Valiente, H.U. Güdel, Luminescence up-conversion mechanisms in Yb3+–Tb3+
systems, J. Lumin. 94–95 (2001) 305–309.
ARTICLE IN PRESS
B.S. Richards / Solar Energy Materials & Solar Cells 90 (2006) 1189–1207 1207

[45] S. Faulkner, S.J.A. Pope, Lanthanide-sensitized lanthanide luminescence: terbium-sensitized


ytterbium luminescence in a trinuclear complex, J. Am. Chem. Soc. 125 (2003) 10526–10527.
[46] K.C. Mishra, J.K. Berkowitz, E.A. Dale, T.P. Das, K.H. Johnson, Cooperative two-photon
luminescence, J. Lumin. 46 (1990) 209–215.
[47] T. Miyakawa, D.L. Dexter, Phonon sidebands, multiphonon relaxation of excited states, and
phonon-assisted energy transfer between ions in solids, Phys. Rev. B 1 (7) (1970) 2961–2969.
[48] A. Shalav, B.S. Richards, T. Trupke, K.W. Krämer, H.U. Güdel, M.A. Green, The application of
NaYF4: Er3+ up-converting phosphors for increased silicon solar cell efficiencies in the near-infrared,
Appl. Phys. Lett. 86 (2004) 013505.
[49] M. Pollnau, D.R. Gamelin, S.R. Lüthi, H.U. Güdel, Power dependence of upconversion
luminescence in lanthanide and transition-metal-ion systems, Phys. Rev. B 61 (5) (2000) 3337–3346.
[50] B.S. Richards, Comparison of TiO2 and other dielectric coatings for buried-contact solar cells: a
review, Prog. Photovolt. 12 (4) (2004) 253–281.
[51] K.L. Frindell, M.H. Bartl, A. Popitsch, G.D. Stucky, Sensitized luminescence of trivalent europium
by three-dimensionally arranged anatase nanocrystals in mesostructured titania thin films, Angew.
Chem. Int. Ed. 41 (6) (2002) 959–962.
[52] K.L. Frindell, M.H. Bartl, M.R. Robinson, G.C. Bazan, A. Popitsch, G.D. Stucky, Visible and near-
IR luminescence via energy transfer in rare earth doped mesoporous titania thin with nanocrystalline
walls, J. Solid State Chem. 172 (2003) 81–88.

You might also like