1. Introduction

The application of solar cells is of great significance. As is widely known, in recent years energy-related issues have become a common concern of mankind. Solar energy is a novel type of green energy; and it is inexhaustible for practical purposes. It is important to improve the utilization of solar energy by means of quantum cutting (QC), in order to enhance the conversion efficiency of solar cells. The Rare earth (RE) ions are the ideal candidate to realize the QC, in that these ions generally have plenty of well-shielded 4f states that can emit fluorescence from ultraviolet to infrared [1–5]. Therefore, these RE doped materials have generated much interest in visible QC, infrared (IR) down-conversion (DC) and visible up-conversion (UC) [1, 2, 6].

Although, QC phosphors via DC have been investigated for decades for their use in the lighting and display industry, only recently DC through a cooperative energy transfer (ET) process has attracted growing attention in enhance the efficiency of photovoltaic cells [2, 7–10]. The most widely used solar cells are based on crystalline silicon (Si). The thermalization of electron–hole pairs, generated by the absorption of high-energy photons, is one of the major energy loss mechanisms in a conventional solar cell [2, 11]. Thermalization losses can be reduced by using DC whereby, for example, a photon with twice the energy of the band gap transforms into two photons with energy is just above the band gap of a single junction Si (Eg = 1.12 eV, 1100 nm). Theoretically, for a single junction Si-based solar cell in conjunction with an ideal DC layer, the conversion efficiency can be up to 40%, which is a significant improvement over the limiting efficiency of 30.9% for conventional solar cells [6–11].

Ion pairs of Re³⁺-Yb³⁺ (Re = Tb, Pr, Er or Tm) have been demonstrated with optical spectroscopy for cooperative DC in various hosts, including glasses and crystalline phosphors [2, 8, 11, 12]. However, these DC materials are still far from practical application, because the absorption of the sensitizer ion (Tb³⁺, Pr³⁺, Er³⁺ or Tm³⁺) arises from the parity-forbidden 4f-4f transitions, which are naturally weak in intensity and narrow in bandwidth. On the contrary, Eu²⁺ ion might be an ideal broadband sensitizer for Yb³⁺, as its 4f-5d transition covers a broad spectral range and more importantly, the energy of its 4f-5d transitions can be tuned by changing the crystal field strength as well as the covalency of the host [13, 14].

In the present work, we chose the low silica calcium aluminosilicate glass (LSCAS) as the host lattice, which have high concentration quenching limit [14–17]. The host is proper for Eu²⁺, in which the energy of the 4f-5d transition of Eu²⁺ locate in the visible region and matches well with the twice energy of the ²F_5/2→²F_7/2 transition of Yb³⁺. Besides, the absorption region also matches the high-energy part of the solar spectrum well. Therefore, the Eu²⁺-Yb³⁺ dual ions combination is a promising system to realize NIR QC, and we observed ET from Eu²⁺ to Yb³⁺ by a cooperative process.

2. Experiment

The compositions of the host used in this experiment were 8%SiO₂-58%CaO-27%Al₂O₃-7%MgO in mole percentage. The Eu²⁺-Yb³⁺ doped glasses are referred to as LSCAS:Eu,Yb. Analytical reagent of SiO₂, CaCO₃, Al₂O₃, MgCO₃, Eu₂O₃, and Yb₂O₃ were used as starting materials. Detailed glass-preparation procedure was described as follows. The homogenously mixed batch with 20g was melted at 1480°C in a graphite crucible for 2 hours in vacuum atmosphere. The glass melt was poured on the pre-heated steel mold and then annealed for 2 hours near the glass transition temperature to enhance the mechanical property by releasing the stress induced during the quenching process. Finally the samples were then cut into 15mm × 15mm × 1.5mm sizes and polished for optical measurements.

Optical absorption spectra at room temperature were measured using a Lambda 35 spectrophotometer. Excitation and emission spectra in the ultra-violet, visible and infrared wavelength ranges were recorded on a FLS920 fluorescence spectrophotometer equipped with a tunable excitation source. Photoluminescence decay measurements of Eu²⁺ were performed by using an FLS920 fluorescence spectrophotometer. Decay curves for Eu²⁺ were recorded at 600nm under 405nm excitation. In order to reduce the test error, we put every sample at the same position and keep them on same angle related to excitation light. All of measurements were carried out at room temperature.

3. Results and discussion

Part of the energy-level diagram of Eu²⁺ and Yb³⁺ in LSCAS glass is shown in Fig. 1 to illustrate the possible cooperative ET from Eu²⁺ to Yb³⁺. In this system Eu²⁺ was doped as a sensitizer and Yb³⁺ was doped as an activator, whereby to convert one visible photon into two NIR photons. The doped Eu²⁺ ions absorb energy from the excitation source and are excited to the 4f⁶5d¹ level. Then, parts of the energy of the Eu²⁺ ions are relaxed to the split excited state, along with a Stokes-shift. The emission occurs in terms of the parity-allowed transition from the excited state to the ground state (4f⁷). Meanwhile, another part of the Eu²⁺ ions releases the energy absorbed by activating two NIR photons due to the Yb³⁺:²F_5/2 → ²F_7/2 transitions. The Eu²⁺:4f-5d transition is located at approximately twice the energy of the Yb³⁺:²F_5/2-²F_7/2 transition and Yb³⁺ has no other levels up to the UV region [2, 8–10]. Therefore, there would not be resonant radiation transfer, resonant non-radiation transfer or cross relaxation transfer. Non-resonant ET maybe takes place by the assist of phonons. While the low phonon property of the host and the large difference band gap between the ground and excited states of Eu²⁺ and Yb³⁺ makes phonon assist non-resonant ET negligible. Therefore, the cooperative DC process may become the dominant relaxation process in this system, which is opposite to the UC phenomenon [18].

 

Fig. 1 Schematic energy-level diagram of LSCAS: Eu²⁺,Yb³⁺ shows the concept of NIR QC with visible excitation at 405nm.

Download Full Size | PDF

The emission band of Eu²⁺ ion is due to the 5d→4f transition which can vary from long-wavelength ultraviolet to yellow by choosing different glass matrix. The host lattice influence on the emission color of the Eu²⁺ ion is mainly determined by several factors such as covalency effect and crystal field effect [14]. The influence of the surroundings on Eu²⁺ ion is expected to be large indeed, despite the 4f electrons are well shielded from the surroundings by completely filled 5s and 5p orbits, however the 5d electrons of Eu²⁺ are exposed to the surroundings and the influence is large [3,14]. If the crystal field is strong and the amount of covalency high, the lowest component of the 4f⁶5d configuration of the Eu²⁺ ion may shift to some low energy level, that the excitation source have a red-shift and the absorption band focus much more energy on the blue range. And the energy difference between Eu²⁺ excited state and twofold Yb³⁺ excited state will become much smaller with the redshift of Eu²⁺ excited state. The smaller the difference the bigger the ET efficiency will be. Therefore, the Eu²⁺-Yb³⁺ co-doped LSCAS glass have potential advantage for solar cells down-conversion layer.

Figure 2 compares the absorption spectra of un-doped and co-doped glasses. The curve b is the net absorption of rare-earth ions without the influence of host. It shows that the matrix glass has no absorption except the UV absorption sideband. The broadband absorption peak from 300nm to 500nm is due to Eu²⁺ ions. The intrinsic absorption of Eu³⁺ at 394nm cannot be observed here. The infrared peaks are the Yb³⁺ ions absorption. Since the Eu²⁺ ions are the main absorbers and sensitizers. Besides, there is no apparent effect of Eu³⁺ ions on IR luminescence from Yb³⁺ ions. Thus we focus the discussion on the effect of Eu²⁺ ions. The ion valence in glass can be tuned by varying the reducing atmosphere in the melting process, tuning the optical basicity of the host and adding reducing agent. In this work, the vacuum atmosphere is good enough to maintain the europium valence. It is obviously, this material has excellent transmission efficiency in the visible and infrared region. Thus it can be made into DC layer and covered at solar cells front surface.

 

Fig. 2 Absorption of the matrix glass and the Eu²⁺-Yb³⁺ co-doped glass.

Download Full Size | PDF

Figure 3 presents the excitation spectra of LSCAS:Eu²⁺,Yb³⁺. It can be seen from Fig. 3 that the excitation spectra shape monitored the Yb³⁺ emission at 980nm and the Eu²⁺ emission at 460nm matches quite well. This similarity in the shape of the excitation spectra can be considered as direct evidence of ET from Eu²⁺ to Yb³⁺. Besides, there is no absorption transition for Yb³⁺ ion in blue range. The broad peak ranging from 300nm to 500nm is ascribed to the first allowed 4f-5d transition of Eu²⁺ [13, 14]. This broad excitation band indicates that the glasses are suitable for excitation by blue and violet sun light.

 

Fig. 3 Excitation spectra of Eu²⁺-Yb³⁺ co-doped samples monitored at 600nm and 980nm compared in Fig. 3. And the visible and near infrared emission spectra of Eu²⁺-Yb³⁺ co-doped borosilicate glasses excited by 405nm are shown in Fig. 3. The Eu²⁺ fixed concentration is 0.2mol% and Yb³⁺ varied concentration are 0.2, 0.4 and 0.6mol% for sample a, b and c, respectively.

Download Full Size | PDF

The Eu²⁺-Yb³⁺ dual ions combination is one of the promising systems to realize NIR QC, since the Eu²⁺ ion 4f-5d transition is located at approximately twofold the energy of the Yb^{3+ 2}F_7/2-²F_5/2 transition (~1000 nm). Moreover, the ²F_5/2-²F_7/2 emission is just above the band-gap energy of Si solar cells. It will be favorable to increase the solar cell conversion efficiency.

Figure 3 also shows the visible and NIR photoluminescence emission spectra of LSCAS: Eu²⁺,Yb³⁺ glasses. Excited by 405nm lights these samples gave rise to emission both from Eu²⁺ and Yb³⁺ ions. The emission spectrum ranging from 460nm to 780nm is due to the 5d-4f transition of Eu²⁺. The ⁵D₀-⁷F_1,2 emission of Eu³⁺ ions was totally covered by the strong emission of Eu²⁺. The Eu³⁺ concentration is nearly negligible compared to Eu²⁺ concentration.

It is also noticed that, from sample (a) to sample (c), the emission intensity of Eu²⁺ at 600nm decreases monotonically, while the intensity of the NIR emission at 980nm increases rapidly with increasing Yb³⁺ concentration. This phenomenon proved the ET rate of this process strongly depends on the ions distance. In the case of the Eu²⁺ and Yb³⁺ system with the Yb³⁺ concentration increase, the distance between Eu²⁺ and Yb³⁺ decreased, which alleviate the ET efficiency and Yb³⁺ emission. In addition, this also facilitates to form ions group with one Eu²⁺ ion and two Yb³⁺ ions, that is to say the probability of Eu²⁺ donor have two Yb³⁺ acceptors in critical distance increased. Using the Dexter’s theory the ET rate can be expressed as a function of the distance between donor and acceptor, so that the ET probability depends upon the distance between donor and acceptor. The distance dependence varies with the interaction type. For exchange interaction it is exponential, while it is of the type R^-n (where R is the distance between donor and acceptor) for the multipolar interaction. Therefore, the ET efficiency and the QE rose with the increased Yb³⁺ concentration [3–5].

Since the spectral range of VIS-to-NIR conversion is reflected by the excitation spectra of Yb³⁺ emission in DC materials, a broad excitation bandwidth for NIR emission is desirable and needed to be studied [2]. In view of practical application and solar spectrum the more high-energy photons of the solar spectrum especially in the range of 300nm–500nm converted into NIR photons the higher the conversion efficiency will be [13]. In this sense, the broadband VIS-light excitable LSCAS:Eu²⁺,Yb³⁺ reported here is comparatively better than previous DC materials doped with an RE³⁺-Yb³⁺ pair (RE = Tb, Pr, Tm), because the bandwidth of the 4f-5d transition of Eu²⁺ is much broader than that of the 4f-4f transitions of those RE ions [11–14].

Figure 4 shows the global solar spectrum (AM1.5G, blue background), the spectral response of Si solar cells (a), the excitation spectrum of Yb³⁺ at 980nm (b) and the IR emission spectrum excited at 405nm (c). As the shown in Fig. 4 the single junction Si solar cells photoelectric transformation efficiency is the highest near the band gap (1.1 eV). While the response become weak in the high energy region, due to thermalization loss. Especially in the 300nm–500nm region, while the solar spectrum is still very strong. Therefore, the excitation spectrum monitored at NIR emission of DC material should located in the above mentioned region as shown in curve (b). The curve (c) in Fig. 4 is the NIR emission, which matches quite well with the response spectrum of Si. By using the glasses developed in this work, the broad region 300nm–500nm light can be converted to around 1000nm, where the response is highest. This process was realized by the second-order DC mechanism. In general one photon with twice the energy of the Si band gap absorbed by Eu²⁺ ions, and then transfers the energy simultaneously to two Yb³⁺ ions with the emission of exactly the energy of the Si solar cell band gap, resulting in a great improvement in conversion efficiency of Si solar cells.

 

Fig. 4 The blue background is the solar spectrum (AM1.5G) in visible and NIR region. The white slash area is the maximum fraction available for DC. The curves a, b, and c are spectral response of single junction Si, excitation spectrum monitored at 980nm and IR emission spectrum excited at 400nm of our Eu²⁺-Yb³⁺ co-doped sample, respectively.

Download Full Size | PDF

Assuming that the interaction is multipolar type and no migration takes place, the Eu²⁺ decay curves could be analyzed using the well-known direct quenching mechanism. There is no energy migration effect or less energy migration effect in the 5d level because it is obviously suppressed by faster relaxation. Usually, the interaction scheme involves both the dipole-dipole and the dipole-quadrupole interactions, which can be described by Inokuti-Hirayama formula [3–5]. The rare earth ions doped in oxide glass tend to accumulate, therefore the concentration of rare-earth ions is different everywhere. Thus parts of Eu²⁺ ions doped in oxide glass clustered and these ions have no or less contribution to the emission. The non-clustered Eu²⁺ ions can efficiently emit fluorescence, and can be classified into two types due to the distance from Eu²⁺ to Yb³⁺ ions, one isolated and with no interaction with Yb³⁺ ions, another with energy transfer by multipolar interaction. Therefore, the fluorescence from the former should decay exponentially and the latter should decay according to Inokuti-Hirayama formula. So the resultant fitting equation could be formulated as [3]:

 

Fig. 5 Decay curves of the Eu²⁺ 5d-4f emission (600nm) in the LSCAS glasses with 405nm excitation. The red lines are the fitted decay curves. The concentration of Yb³⁺ is 0.2mol% (a), 0.4mol% (b) and 0.6mol% (c), respectively. The fixed Eu²⁺ concentration is 0.2mol%.

Download Full Size | PDF

Table1. Parameters for the fitted equation, decay life time, energy transfer efficiency and quantum efficiency of samples with different Yb³⁺ concentration.

View Table

It is very interesting to note that the ratio of C/C₀ nearly equates to the concentration ratio of Yb³⁺/Eu²⁺. The C/C₀ means the number of acceptors in donor’s sphere of influence, in this work that is the numbers of Yb³⁺ in a Eu²⁺ around. And this proved our hypothesis about the ions group formation with one Eu²⁺ ion and two Yb³⁺ ions, when the Yb³⁺ concentration increase.

In summary, Eu²⁺–Yb³⁺ co-doped LSCAS glasses were investigated as a downconversion layer candidate to enhance silicon solar cell efficiency. Excitation and emission measurements indicate the occurrence of cooperative energy transfer from Eu²⁺ to Yb³⁺ ions. The maximum value of QE approaches 163.8% in this work. The results show efficient conversion of broad 300nm–500nm light, which is not fully utilized by the existed solar cells, to 1000nm IR light, which can be efficiently absorbed by silicon solar cell and converted into electric energy. In addition, the excitation and luminescence due to f-d transition of Eu²⁺ is determined by the ligand field around Eu²⁺, it is possible to modify the excitation spectrum which fits well with solar spectrum below 500nm through controlling glass matrix. Besides, the modification of glass matrix can also raise the NBOs number, then decrease CQ effect, shorten ions distance, heighten ET efficiency and further promote QE. Our approach may open an effective way for the realization of efficient spectral modification of solar spectrum.

4. Conclusions

In conclusion, the LSCAS glasses impregnated with Eu²⁺ and Yb³⁺ ions were successfully prepared as efficient visible to NIR converters. Luminescence spectra, excitation spectra and quantum efficiency of Eu²⁺-Yb³⁺ co-doped glasses were measured and studied. The decay curves were well fitted and the energy transfer process was considered to be multipolar interaction. It is suggested that this dual ions system is one of the most efficient NIR DC systems. In view of their strong and broadband absorption in the range of 300nm-500nm and the efficient NIR emission, the DC glass system has a potential application in solar cell as DC phosphor converter layer.