1. Introduction

Near-infrared (NIR) quantum cutting (QC) is expected as an important method to modify the solar spectrum [1]. In a typical process, one UV-Vis (300-500 nm) photon is cut into two or more NIR (~1000 nm) photons, which perfectly match the maximum spectral response of a c-Si solar cell (Eg≈1.12eV, λ≈1100 nm). It can realize multiple electron-hole pairs generation per incident photon in the c-Si solar cell, greatly reducing energy losses due to charge thermalization and improving the solar cell efficiencies [2,3]. Trupke and co-authors theoretically demonstrated that the Shockley-Queisser limiting efficiency of 30% can be substantially increased up to 40% through a down-conversion of high-energy photons [4,5]. This work prompted worldwide research on NIR QC luminescent materials. Up to now, most of the research has focused on Ln³⁺ -Yb³⁺ couples [6–15], where Ln³⁺ ions serve as donors that utilize the UV-green fraction of the solar spectrum and Yb³⁺ ions serve as acceptors that give NIR emissions at ~1000 nm. According to the electronic configuration of rare-earth ions, these couples are classified into two groups: Ln³⁺-Yb³⁺ (Ln = Pr, Nd, Tb, Ho, Er, Tm) and Ce³⁺/Eu²⁺-Yb³⁺. The former Ln³⁺ ions have the narrow absorption peaks with low efficiency in the UV-green region [7–10]. The latter Ce³⁺/Eu²⁺ ions show broad absorption bands with high efficiency in the region [12–14, 16]. Therefore, the Ce³⁺/Eu²⁺-Yb³⁺ couple is an ideal NIR QC system as a solar spectral converter for c-Si solar cell [3].

NIR QC luminescent materials with the Ce³⁺/Eu²⁺-Yb³⁺ couples have been reported in various systems, including chloroborate phosphor, YAG ceramics, and borate and aluminosilicate glasses [13–15,17]. Cooperative energy transfer (CET) mechanism was mostly proposed to explain the quantum cutting UV-to-NIR energy transfer from Ce³⁺/Eu²⁺ to Yb³⁺ in those systems mentioned above. Figure 1 shows the typical CET processes involving Tb³⁺ and Ce³⁺ ions: Both Tb³⁺ and Ce³⁺ ions directly and respectively absorb a high energy photon via the forbidden 4f-4f transition and allowed 4f→5d transitions; then transfer the excitation energy to two Yb³⁺ ions under the assistance of phonons; finally two NIR photons are produced. Unlike the constant ⁵D₄ level (~20000 cm⁻¹) of Tb³⁺, the locations of the 5d energy levels of Ce³⁺ ions are easily affected by the crystal field. It varies significantly from 27027 cm⁻¹ in borate glasses to 26596 cm⁻¹ in YBO₃ phosphor and to 20000 cm⁻¹ in YAG ceramics [11,13,17]. The energy differences (ΔE) between the lowest 5d excited state of Ce³⁺ in these materials and twice the energy (~20000 cm⁻¹) of the ²F_5/2→²F_7/2 transition of the Yb³⁺ ions are about 7027 cm⁻¹ (ΔE₁), 6596 cm⁻¹ (ΔE₂) and 0 cm⁻¹ (ΔE₃), respectively. The QC process in Tb³⁺-Yb³⁺ couples is almost resonant cooperative energy transfer. On the contrary, the process in Ce³⁺-Yb³⁺ couples strongly depends on the energy locations of the lowest 5d excited state of Ce³⁺ ions in specific compounds. Obviously, the composition and crystal structure of host compounds, local symmetry of lattice sites, phonon frequency, etc., play key roles in the CET process involving the Ce³⁺-Yb³⁺ couples and determine the NIR quantum efficiency. Unfortunately, these issues seem to have been overlooked in previous researches on the cooperative QC process in Ce³⁺/Eu²⁺ -Yb³⁺ systems.

 

Fig. 1 Schematic energy level diagram on QC process of Tb³⁺-Yb³⁺ and Ce³⁺-Yb³⁺ couples in some specific systems: (a) borate glasses; (b) YBO₃ phosphor and (c) YAG ceramics.

Download Full Size | PDF

In this work, we aim to investigate how and to what extent the energy locations of the lowest 5d excited state of Ce³⁺ ions affect the NIR quantum efficiency in Ce³⁺-Yb³⁺ couples. We limit our investigation to the X₂-Y₂SiO₅ system with two different crystallographic sites [Y³⁺(1) and Y³⁺(2)] that Ce³⁺ and Yb³⁺ substitute to exclude the influence of other factors on the cooperative QC process, such as phonon vibration frequency, nonradiative centers and crystal structure. The influence of these mentioned factors can be considered exactly the same for the two kinds of Ce³⁺ centers in a given host lattice. The dependences of the integrated emission intensities of Ce³⁺ and Yb³⁺, decay lifetime (τ) of Ce³⁺, nonradiative energy-transfer rate (K_Ce→Yb), cooperative energy transfer efficiency (η_CET) and quantum efficiency (η_QE) on the Yb³⁺ concentration (y) were studied in details. More importantly, these results demonstrate that the closer to twice the ²F_7/2-²F_5/2 absorption energy (~20000 cm⁻¹) of Yb³⁺ ion is the 5d energy location of Ce³⁺ ion, the higher the CET efficiency in Ce³⁺-Yb³⁺ couple. We believe that this work may be of great significance for designing advanced NIR QC phosphors for Si based solar cell applications.

2. Experimental

A series of silicates compounds were prepared in solid state reactions. Stoichiometric amounts of Y₂O₃ (99.99%), SiO₂ (AR), CeO₂ (99.99%) and Yb₂O₃ (99.99%) were ground and mixed homogeneously in an agate mortar for 30 min. Then the mixtures were placed in alumina crucibles with covers and fired at 1400 °C in a reducing atmosphere (CO) for 12 h. Finally, the samples were cooled to room temperature in the furnace and reground into powders for subsequent analysis.

The phase purity of the product was examined by X-ray diffraction (XRD) using a D8 ADVANCE powder diffractometer with Cu-Kα radiation (λ = 1.54059 Å) at room temperature. The high quality XRD data for Rietveld refinement was collected over a 2θ range from 10 ° to 90 ° at an interval of 0.02 ° with a counting time of 8 sec per step. Structural refinement of XRD data was performed using the TOPAS-Academic program [18]. The photoluminescence (PL), excitation (PLE) spectra and the decay curves were obtained using a FSP920 Time Resolved and Steady State Fluorescence Spectrometer (Edinburgh Instruments) at room temperature and 10 K, which was equipped with a 450W Xe lamp, a 150w nF900 flash lamp, red sensitive PMT and R5509-72 NIR-PMT in a liquid nitrogen cooled housing (Hamamatsu Photonics K.K).

3. Results and discussion

The phase purity of all products Y_2-xCe_xYb_ySiO₅ (x = 0.002 and 0.3; y = 0, 0.02, 0.06, 0.1, 0.14, 0.2, 0.3, 0.4) was examined and their XRD patterns are similar to each other. As a representative, the experimental (crosses) XRD pattern of Y_1.64Ce_0.3Yb_0.06SiO₅ is shown in Fig. 2 . The Rietveld refinement was performed on this compound using the I12/a1 structure model reported by Maksimov et al. [19], which converged to R_wp = 3.66% and R_B = 1.47%. The refined lattice constants and cell volumes are a = 10.5267(2) Å, b = 6.7555(1) Å, c = 12.5547(2) Å, V = 869.7190 ų, respectively. The mean Y–O distances for Y³⁺(1) (CN = 6) and Y³⁺(2) (CN = 7) from the XRD refinement are estimated to be 2.2955 Å and 2.4147 Å, respectively. It was reported that there exist two different Y³⁺ sites with local symmetry C₁: Y³⁺(1) site is six-coordinated with four silicon-bonded oxygen atoms and two non-silicon-bonded oxygen atoms (free oxygen atom); Y³⁺(2) site is seven-coordinated with five silicon-bonded oxygen atoms and two free oxygen atoms [19]. Since the ionic radius of Yb³⁺ (1.01 Å at CN = 6 and 1.07 Å at CN = 7) is very similar to those of Y³⁺ (1.04 Å at CN = 6 and 1.1 Å at CN = 7) [20], it is likely that the Yb³⁺ equally distributes over the two sites. In contrast, Ce³⁺ (1.15 Å at CN = 6 and 1.21 Å at CN = 7) is signiðcantly bigger and one can imagine that Ce³⁺ more easily enters the Y³⁺(2) site which is larger and more distorted than Y³⁺(1) site. From Table 1 , one can see that the refined occupation fractions of Ce³⁺ are 0.0653 (Ce_Y1) and 0.2347 (Ce_Y2) out of 0.3, Yb³⁺ ions are 0.026 (Yb_Y1) and 0.034 (Yb_Y2) out of 0.06. These results indeed support the conclusions that more Ce³⁺ ions occupy the Y³⁺(2) site, and Yb³⁺ ions have no bias for these two non-equivalent C₁ symmetry Y³⁺ sites, which are consistent with previous results of electron paramagnetic resonance (EPR) and spectroscopic studies [21–23].

 

Fig. 2 Experimental (crosses) and calculated (red solid line) XRD patterns and their difference (blue solid line) for Y_1.64Ce_0.3Yb_0.06SiO₅. One set of tick mark shows the Bragg reflection positions of the phase Y_1.64Ce_0.3Yb_0.06SiO₅. Inset: the crystal structure of corresponding unit cell.

Download Full Size | PDF

Table 1. Final Refined Structural Parameters for Y_1.64Ce_0.3Yb_0.06SiO₅^a

View Table

Photoluminescence properties of Y₂SiO₅:Ce³⁺ have been systematically discussed in the previous research [24–26]. It was found that Ce³⁺ ions enter two Y³⁺ sites and form two emission centers. Ce³⁺ at Y³⁺(2) site (CN = 7) has lower energy of 5d level than at Y³⁺(1) (CN = 6) due to the covalence effects [24], and here they are correspondingly labeled as “Ce³⁺(2)” and “Ce³⁺(1)”, respectively. Compared to Ce³⁺(1), the excitation and emission peaks of the Ce³⁺(2) should show a red-shift. In this work, we chose two Ce³⁺ concentrations (x = 0.002 and 0.3) for the research on the CET in Y_2-x-ySiO₅: xCe³⁺, yYb³⁺ phosphors. For x = 0.002, under the excitation of 356 nm from the lowest 5d level of Ce³⁺(1) center, the emission from Ce³⁺(1) is much stronger than that from Ce³⁺(2) (Fig. 3(b) ); For x = 0.3, emission from Ce³⁺(2) is relatively stronger than that from Ce³⁺(1) when exciting the lowest 5d level at 372 nm of Ce³⁺(2) center (Fig. 3(d)). The maximum emission peak for Ce³⁺(1) is at 393 nm and Ce³⁺(2) at 466 nm, respectively, which well agrees with reported results [27,28]. Moreover, the position of the lowest 5d excited level of two Ce³⁺ centers is estimated by using the mirror-image relationship between the emission and the excitation spectra at 10 K [29]. The low temperature spectra were not given here due to their similarity with room temperature spectra (Fig. 3). From the intersection point of normalized excitation and emission spectra (10 K), the values of the lowest 5d energy level are evaluated to be at about 26740 cm⁻¹ (374 nm) for Ce³⁺(1) ion and 25970 cm⁻¹ (385 nm) for Ce³⁺(2) ion, which are close to twice the energy (20470 cm⁻¹) of ²F_7/2→²F_5/2 of Yb³⁺. It is therefore expected that CET from Ce³⁺ to Yb³⁺ would occur in Y₂SiO₅ host. Hereafter, we will examine whether the CET process happens and to what extent 5d energy locations of Ce³⁺ ion play a role in the process.

 

Fig. 3 (a), (c) Excitation, (b), (d) UV-vis and NIR emission spectra of Y_1.998-yCe_0.002SiO₅: yYb³⁺ and Y_1.7-yCe_0.3SiO₅: yYb³⁺. The visible and NIR emission intensities are not plotted on the same scale.

Download Full Size | PDF

When Yb³⁺ ions are present in the samples of the same Ce³⁺ content, regardless of x = 0.002 or 0.3, it is obviously seen that the shape of PLE spectra changes little whenever the 977 nm emission of Yb³⁺ or the blue emission of Ce³⁺ is monitored. Beside the ²F_5/2→5d absorption transition of Ce³⁺, the charge transfer band (CTB, around 260 nm) of Yb³⁺-O^2– is not obviously observed in the PLE spectra. We performed excitation at 356 nm for x = 0.002 and 372 nm for x = 0.3 to avoid the interplay of the lowest 5d levels of both Ce³⁺ centers as much as possible. The PL results show that intense NIR emission of Yb³⁺ centering at 977 nm was observed, in addition to the violet-blue broadband emission of Ce³⁺, as presented in Fig. 3(b) and 3(d). These evidences together persuaded us to believe that the energy transfer from Ce³⁺ to Yb³⁺ does take place. In order to confirm this speculation, the Yb³⁺ concentration dependence of visible and NIR emission intensity of the two series of samples were investigated in details and plotted in Fig. 4 . As the concentration of Yb³⁺ ions increases, the integrated emission intensity of Ce³⁺ remarkably decreases and the NIR emission intensity of Yb³⁺ ions initially increases and then decreases due to concentration quenching. The optimum doping concentrations (y) for Yb³⁺ were accordingly determined to be 0.2 and 0.06 in Y_1.998-yCe_0.002SiO₅: yYb³⁺ and Y_1.7-yCe_0.3SiO₅: yYb³⁺, respectively. This further proves that the absorbed energy of Ce³⁺ can be partially transferred to a pair of nearest-neighbor Yb³⁺ ions, inducing NIR emission from ²F_5/2→²F_7/2 of Yb³⁺ through CET, as expected.

 

Fig. 4 Dependences of the integrated NIR and Vis emission intensity of (a) Y_1.998-yCe_0.002SiO₅: yYb³⁺ (λ_ex = 356 nm) and (b) Y_1.7-yCe_0.3SiO₅: yYb³⁺ (λ_ex = 372 nm) on the concentration (y) of Yb³⁺. The integrated visible and NIR emission intensities are not plotted on the same scale.

Download Full Size | PDF

Like most Ce³⁺-Yb³⁺ systems, QC mechanism for Y₂SiO₅:Ce³⁺, Yb³⁺ can be generally described as Fig. 5 . Under UV light excitation, electrons on the Ce³⁺ ions are excited from the ground state (4f) to the excited state (5d) (process ①). In the excited state, part of electrons relax to the lowest 5d excited state (process ②), then return to the ground state, generating the blue emissions (process ④). The other electrons in the excited 5d state of a Ce³⁺ ion cooperatively transfer their energy to two Yb³⁺ ions by downconversion (process ③). Subsequently, electrons on the excited level (²F_5/2) of Yb³⁺ ions return to the ground state (2F_7/2), giving the NIR emissions (process ⑤). Competition between the above two processes (②→④ and ③→⑤) results in the occurrence of emissions from Ce³⁺ and Yb³⁺ simultaneously in Y₂SiO₅:Ce³⁺, Yb³⁺ samples (Fig. 4(b) and 4(d)). The entire direct CET process could be expressed as (①→③→⑤).

 

Fig. 5 Schematic energy level and cooperative energy transfer (CET) mechanism of Ce³⁺ and Yb³⁺ in Y₂SiO₅ host

Download Full Size | PDF

As we pointed out in the introduction, the QC process in Ce³⁺-Yb³⁺ couple should strongly depend on the 5d energy level of Ce³⁺ ion in some specific compounds. In the present case, the lowest 5d energy level is at about 26740 cm⁻¹ for Ce³⁺(1) ion and 25970 cm⁻¹ for Ce³⁺(2) ion. The energy differences (ΔE₁ and ΔE₂) between the lowest 5d energy level and twice the energy (20470 cm⁻¹) of ²F_5/2 level of Yb³⁺ are 6270 cm⁻¹ for Ce³⁺(1) ion and 5500 cm⁻¹ for Ce³⁺(2) ion. Comparatively, it is expected that multiple phonons-assisted CET process would be more efficient in Ce³⁺(2)-Yb³⁺ than in Ce³⁺(1)-Yb³⁺ couple, which is supported by the lifetime data, which we will discuss later.

For Y_1.998Ce_0.002SiO₅ and Y_1.7Ce_0.3SiO₅ phosphors, both Ce³⁺(1) or Ce³⁺(2) have a single-exponential decay behavior (Curves 1 and 3) as shown in Fig. 6 . The lifetimes (τ) of Ce³⁺(1) and Ce³⁺(2) ions are estimated to be about 34.2 ns and 30.7 ns, respectively. The difference in decay behavior may be due to local symmetry of the host lattice [30]. When Yb³⁺ ions are co-doped, the decay curves of Ce³⁺(1) and Ce³⁺(2) exhibit obvious nonexponential feature (Curves 2 and 4) and the lifetime of Ce³⁺ decrease gradually as Fig. 7(a) shows. This indicates there is an extra pathway for 5d electrons of either Ce³⁺(1) or Ce³⁺(2) to depopulate. The rate equation for the population densities of the excited 5d state of Ce³⁺ can be expressed as follows [31]

 

Fig. 6 Decay curves of (a) Y_1.998-yCe_0.002SiO₅: yYb³⁺ phosphor (y = 0 and 0.2) and (b) Y_1.7-yCe_0.3SiO₅: yYb³⁺ phosphor (y = 0 and 0.06)

Download Full Size | PDF

 

Fig. 7 (a) Decay lifetime (τ) of Ce³⁺, (b) nonradiative energy-transfer rate (K_Ce→Yb), (c) CET efficiency (η_CET) and (d) quantum efficiency (η_QE) as a function of Yb³⁺ concentration (y) in Y_1.998-yCe_0.002SiO₅: yYb³⁺ [Ce³⁺(1)] and Y_1.7-yCe_0.3SiO₅: yYb³⁺ [Ce³⁺(2)], respectively.

Download Full Size | PDF

Figures 7(b), 7(c) and 7(d) demonstrate the dependences of K_Ce→Yb, η_CET and η_QE on the concentration of Yb³⁺ ions, respectively. It can be clearly observed that, the CET rate and efficiencies of Ce³⁺(1)→Yb³⁺ and Ce³⁺(2)→Yb³⁺ and the quantum efficiency (η_QE) increase as the concentration of Yb³⁺ ions increases. The difference between Ce³⁺(1)→Yb³⁺ and Ce³⁺(2)→Yb³⁺ is that the increasing trends of the CET rate and efficiencies are more obvious in Ce³⁺(2)→Yb³⁺ than in Ce³⁺(1)→Yb³⁺, suggesting that the CET process between Ce³⁺(2) and Yb³⁺ is more efficient. Additionally, the quantum efficiency η_QE could be estimated by Eq. (4), where η_Ce and η_Yb stand for the quantum efficiency of Ce³⁺ and Yb³⁺, respectively. Assuming that all the excited Yb³⁺ ions and the residual excited Ce³⁺ ions decay radiatively, i.e., η_Ce = η_Yb = 1 [10], the upper limited values of the highest total DC quantum efficiency (η_QE) are calculated to be 161% for Ce³⁺(1)-Yb³⁺ couple and 194% for Ce³⁺(2)→Yb³⁺, respectively.

Since other key factors, such as phonon vibration frequency, crystal structure and Yb³⁺ ion occupation fractions, are kept constant in the same Y_2-xCe_xYb_ySiO₅ material, the CET efficiency and QE mainly depend on the energy locations of 5d orbital of Ce³⁺ in Ce³⁺-Yb³⁺ system. The as-obtained results show that the smaller ΔE between the lowest 5d orbital of Ce³⁺ and twice the ²F_5/2-²F_7/2 transition energy (~20000 cm⁻¹) of Yb³⁺ results in higher CET efficiency and QE. As well known, Ce³⁺-Yb³⁺ is an efficient donor-acceptor pair with high potential as a full spectrum converter for Si solar cells. Therefore, we believe that our effort may open a new route to the design of advanced UV/Vis-to-NIR phosphors for Si based solar cell applications.

3. Conclusions

We have systematically studied the effects of the location of the lowest 5d energy level of Ce³⁺ centers on the NIR QC process in Y₂SiO₅:Ce³⁺, Yb³⁺. The Ce³⁺(1) and Ce³⁺(2) centers dominate the emission spectra at low and high Ce³⁺-concentration in Y₂SiO₅, respectively. Phonon-assisted QC processes occur within both Ce³⁺(1)-Yb³⁺ and Ce³⁺(2)-Yb³⁺ couples, which contribute to the utilization of the UV-green fraction (300<λ<500 nm) of the solar spectrum and the consequent enhancement of the NIR emission intensity of Yb³⁺ ions. As the concentration of Yb³⁺ increases, the lifetimes of Ce³⁺(1) and Ce³⁺(2) decrease, and nonradiative energy transfer rate (K_Ce→Yb), CET efficiency (η_CET) and quantum efficiency (η_QE) increase. Additionally, the estimated highest total η_QE is 161% for Ce³⁺(1)-Yb³⁺ couple and 194% for Ce³⁺(2)→Yb³⁺, respectively. Most importantly, it was found that if the energy location of the lowest 5d excited state of Ce³⁺ center is closer to twice the ²F_7/2-²F_5/2 absorption energy (~20000 cm⁻¹) of Yb³⁺, the CET efficiency is higher. Therefore, from the point of enhancing the efficiency of silicon-based solar cells, we suggest to exploit new solar spectra-converters involving Sⁿ⁺-Yb³⁺ couple, where Sⁿ⁺ is a cation sensitizer that can efficiently absorb solar lights in the UV and visible region and has its dominating depopulation level located at around 20000 cm⁻¹.