Near-infrared down-conversion in Er³⁺–Yb³⁺ co-doped transparent nanostructured glass ceramics for crystalline silicon solar cells

Abstract

A two-step energy transfer was achieved in Er³⁺–Yb³⁺ co-doped transparent glass ceramics containing CaF₂ nanocrystals, which involved down-conversion of an absorbed visible photon to two emitted near-infrared photons. Therefore, the Yb³⁺ emission centered at 980 nm was efficiently enhanced in response to the strongest absorption of crystalline silicon solar cells. The mechanisms of the two-step energy transfer mechanism were verified based on spectral and lifetime measurements, and the maximal energy transfer efficiency and corresponding quantum yield obtained were as high as 75.3% and 150.6%, respectively. As a result, near-infrared quantum cutting transparent glass ceramics will open a route to enhance the energy efficiency of silicon solar cells.

---

Introduction

Crystalline silicon (c-Si) solar cells, which are extensively used in everyday life, have been particularly appealing.^1,2 However, a large part of the solar energy is dissipated during the photovoltaic process owing to the thermalization of charge carriers whose energy exceeds the c-Si energy gap (E_g = 1.12 eV), significantly limiting the conversion efficiency of solar cells.³ Adapting the solar spectrum through down-conversion (DC) is an optimal scheme for reducing the energy loss caused by the thermalization, which involves the conversion of one ultraviolet (UV)-visible photon into two near-infrared (NIR) photons.^4–6 Recently, NIR DC in RE³⁺/Yb³⁺ (RE = Tb,^7–9 Pr,^8,10,11 Tm,^8,10,12,13 Ce,^14,15 and Eu^16,17) co-doped systems have been intensively investigated, on which RE³⁺ is excited by Yb³⁺ with a UV or visible photon and two NIR photons ∼980 nm are emitted, which can be efficiently absorbed by c-Si solar cells through a DC process.

However, most of the DC between RE³⁺ and Yb³⁺ is achieved via a one-step cooperative energy transfer (ET) process, during which the system exhibits low emission intensity owing to concentration quenching and low energy transfer efficiency (ETE) between RE³⁺ and Yb³⁺.^12,18,19 For more efficient DC, a two-step ET mechanism has been presented for Nd³⁺–Yb³⁺,²⁰ Pr³⁺–Yb³⁺,²¹ and Er³⁺–Yb³⁺ (ref. 22) lanthanide ion couples (to our knowledge). This two-step DC energy transfer is effective only when the donor has a suitable intermediate state to assist the sequential energy transfer to the acceptor. Herein, we focus on the Er³⁺–Yb³⁺ couple, which may suit for this condition (⁴I_11/2 for Er³⁺). Nevertheless, efficient DC by a two-step process cannot be easily implemented for the Er³⁺–Yb³⁺ co-doped system due to the fast multi-photon relaxation from the ⁴F_7/2 level.²³ Some studies have been performed for Er³⁺–Yb³⁺ co-doped halides with low photon energy, but their applications are limited because of the poor thermal stability and adverse light scattering.^22,24

Recently, oxyfluoride glass ceramics (GCs) have attracted considerable attention because of their low phonon energy and high chemical and mechanical stability,^25–27 which can suppress multi-photon relaxation processes between the adjacent energy levels in Er³⁺ and enable efficient two-step ET from the ⁴F_7/2 level of Er³⁺. Therefore, nanostructured GCs have been chosen as the host for the Er³⁺–Yb³⁺ couple to compensate for the shortcomings of the aforementioned hosts in our report. In addition, evidence for the two-step ET process between the Er³⁺ and Yb³⁺ ions is presented based on spectral and lifetime measurements of Er³⁺–Yb³⁺ co-doped GCs. Cross relaxation (CR) participates in the first ET step, leading to an Er³⁺ population in the ⁴I_11/2 level and Yb³⁺ in the ²F_5/2 state. Subsequently, Er³⁺ will transfer another part of the energy to Yb³⁺ via a direct ET process in the second step. Both steps result in a NIR photon due to Yb^{3+ 2}F_5/2 → ²F_7/2 transition. Finally, the calculated energy transfer efficiency (ETE) and the quantum yield (QY) of nanostructured GCs were as high as 75.3% and 150.6%, respectively. Therefore, the Er³⁺–Yb³⁺ co-doped transparent nanostructured GCs make it possible to enhance the efficiency of c-Si solar cells as a DC layer.

Experimental

Synthesis of oxyfluoride GCs

The precursor oxyfluoride glass was prepared by the conventional melt-quenching method using high-purity reagent powders. The composition of the host precursor glass was 45SiO₂–15Al₂O₃–20LiF–20CaF₂ in mol%. For each batch, about 20 g of mixed original materials were melted in a covered corundum crucible under a dry argon atmosphere at 1500 °C for 1 h, and then cast into a brass mold to form the precursor glass (PG). RE³⁺ was introduced in the form of Er₂O₃ and Yb₂O₃. The Er³⁺ concentration was fixed to 0.5 mol%, while the Yb³⁺ concentrations were set to 0, 0.5, 1.0, 2.0, and 5.0 mol%. The PG samples were subsequently heat-treated at 580 °C for 12 h to obtain GCs through crystallization. The GCs doped with different RE³⁺ concentrations are denoted as GC0, GC1, GC2, GC3 and GC4, respectively.

Characterization of materials

To confirm the crystallization phase of the oxyfluoride glass, X-ray diffraction (XRD) analysis was performed using a D/Max-3C diffractometer with Cu Kα radiation (1.5405 Å, 40 kV, 60 mA). The sizes and shapes of the glass nanocrystals were characterized by high resolution transmission electron microscopy (HRTEM, JEM2100). The visible-NIR absorption spectra were obtained using a Perkin-Elmer UV/vis/NIR Lambda 900 spectrophotometer. The excitation spectra and the emission spectra, both in the visible and NIR regions, were recorded on an FLSP920 spectrofluorometer (Edinburgh Instruments, Britain). The emission and excitation measurements were performed using a 450 W Xe lamp as the excitation source with different detectors. The decay curves for the Er³⁺ 486 nm and 948 nm emissions upon 460 nm excitation were recorded using a μF900 pulse xenon lamp.

Results and discussion

Structure behaviour

The XRD patterns of PG and GC are shown in Fig. 1(a). The curve for the PG sample exhibits representative broad humps, indicating the amorphous structure of the material. After the thermal treatment at 580 °C for 12 h, the characteristic XRD peaks were observed at 2θ = 28.21°, 47.01°, and 55.81°, confirming the orthorhombic CaF₂ (JCPDS 35-0816). The nanocrystal size was evaluated to be 10 nm using the Scherrer formula. According to TEM image in Fig. 1(b) and scanning transmission electron microscopy (STEM) image in Fig. 1(c), nanoparticles with a size of 8–10 nm were distributed homogeneously in the glass matrix, which is in accordance with the XRD result. Fig. 1(d) shows a high resolution TEM (HRTEM) image of individual CaF₂ nanocrystal. The measured lattice spacing is 0.32 nm, which can be indexed to the (111) lattice plane of orthorhombic CaF₂, which is consistent with the XRD patterns in Fig. 1(a).

Fig. 1 (a) XRD patterns of PG and GC; (b) TEM image of GC containing 0.5Eu²⁺/0.5Yb³⁺:CaF₂ (GC1); (c) STEM image of GC containing 0.5Eu²⁺/0.5Yb³⁺:CaF₂ (GC1) and (d) HRTEM image of individual nanocrystal.

Optical properties

The absorption spectra of the Er³⁺ single-doped (GC0) and Er³⁺–Yb³⁺ co-doped containing 0.5Eu²⁺/0.5Yb³⁺:CaF₂ (GC1) samples are presented in Fig. 2. The UV absorption edge that limits the glass transparency at shorter wavelengths is caused by the electronic transitions in the host glass.²⁰ Fig. 2 shows that the peaks observed in the absorption spectrum of the GC0 are centered at 400, 460, 488, 520, 543, 650, 796, 948, and 1520 nm; these peaks were ascribed to the absorption transitions from the ⁴I_15/2 ground state to every excited state of the Er³⁺ ion, including ²H_9/2, ⁴F_5/2, ⁴F_7/2, ²H_11/2, ⁴S_3/2, ⁴F_9/2, ⁴I_9/2, ⁴I_11/2, and ⁴I_13/2, respectively. With the addition of Yb³⁺ ions (GC1), the peak wavelengths of all the absorption bands were almost identical for the GC0 sample, while the absorption band at ∼980 nm became wider and stronger owing to overlap between the Yb^{3+ 2}F_7/2–²F_5/2 and Er^{3+ 4}I_15/2–⁴I_11/2 transitions.

Fig. 2 Absorption spectra of Er³⁺ single-doped (GC0) and Er³⁺–Yb³⁺ co-doped GC containing 0.5Eu²⁺/0.5Yb³⁺:CaF₂ (GC1).

Luminescence of Er³⁺ single doped and Er³⁺–Yb³⁺ co-doped GCs

The photoluminescence excitation (PLE) and photoluminescence (PL) spectra in the GCs were measured to investigate whether the ET occurred from Er³⁺ to Yb³⁺. The PLE spectra of Fig. 3(a) and (b) reveal sharp excitation bands peaked at 460 nm, corresponding to the Er³⁺:⁴I_15/2 → ⁴F_5/2 transition, which were recorded by monitoring both the Er³⁺:⁴F_7/2 → ⁴I_15/2 transition at 486 nm and the Yb³⁺:²F_5/2 → ²F_7/2 transition at 980 nm, verifying the existence of ET from Er³⁺ to Yb³⁺. Furthermore, in the PL spectra of Fig. 3(c), a series of sharp emission peaks in the visible-NIR region can be observed upon 460 nm excitation, which decrease monotonously with increasing Yb³⁺ concentration, while the intensity of Yb³⁺ 980 nm emission increases greatly simultaneously; this further elucidates the ET process from Er³⁺ to Yb³⁺. However, as the concentration of Yb³⁺ was increased from 2 mol% to 5 mol%, the emission intensity of Yb³⁺ at 980 nm becomes weaker than that of GC3, revealing the occurrence of cross-relaxation between the Yb³⁺ ions with increasing Yb³⁺ concentration and reduced luminescence yield.

Fig. 3 (a) PLE spectrum monitored by the Er³⁺:⁴F_7/2 → ⁴I_15/2 emission (486 nm) and (b) the Yb³⁺:²F_5/2 → ²F_7/2 emission (980 nm) in the GCs; (c) PL spectra of Er³⁺ in the GCs with different Yb³⁺ concentrations upon excitation at 460 nm.

Mechanisms of ET from Er³⁺:⁴F_7/2 to Yb³⁺ in Pr³⁺–Yb³⁺ co-doped GCs

To further examine the ET mechanism of DC, the photoluminescence spectra from 900 to 1100 nm for GC0 to GC4 are demonstrated in Fig. 4(a). The emission spectra centered at 980 nm for GC1 to GC4 can be decomposed into two independent peaks according to Gaussian deconvolution, which are ascribed to the Er³⁺:⁴I_11/2 → ⁴I_15/2 (948 nm) and Yb³⁺:²F_5/2 → ²F_7/2 (980 nm) transitions. In Fig. 4(b), the emission spectra of Er³⁺ 948 nm have been normalized by the intensity of Er³⁺:⁴F_9/2 → ⁴I_15/2 (650 nm), and the latter does not participate in the ET process. The result shows that the relative emission intensities of Er³⁺ 948 nm to 650 nm increase obviously with increasing Yb³⁺ concentration.

Fig. 4 (a) Photoluminescence (PL) spectra between 900 and 1100 nm in GCs and corresponding Gaussian deconvolution; (b) change in relative emission intensity of Er³⁺ 948 nm (normalized by the intensities of ⁴F_9/2 → ⁴I_15/2 emissions).

According to the abovementioned analysis of Fig. 4(a) and (b), the following conclusion can be drawn. First, the emission intensities of Er³⁺ 948 nm increase gradually from GC0 to GC4 because more photons were populating the ⁴I_11/2 energy level; subsequently, the emission intensities of Yb³⁺ are enhanced except for the concentration quenching in GC4. The results make it possible for cross relaxation (CR) from Er³⁺:⁴F_7/2 to Yb³⁺:²F_7/2. Second, it is also possible that the Er^{3+ 4}F_7/2 state transfers all the excitation energy directly to Yb³⁺. However, direct ET will only populate the ²F_5/2 state of Yb³⁺ and cannot influence the change in the relative intensities of Er^{3+ 4}I_11/2. On the other hand, for CR, it will fill the excited states of Yb³⁺ and simultaneously increase the population of the Er^{3+ 4}I_11/2 state with increasing Yb³⁺ concentration. Therefore, the increase in the relative intensities of Er³⁺ 948 nm to 650 nm should be attributed to CR: Er³⁺ (⁴F_7/2 → ⁴I_11/2)–Yb³⁺ (²F_7/2 → ²F_5/2), and the CR from Er^{3+ 4}F_7/2 to Yb^{3+ 2}F_7/2 dominates the first-step ET process.

Mechanisms of ET from Er³⁺:⁴I_11/2 to Yb³⁺ in Er³⁺–Yb³⁺ co-doped GCs

To determine the second-step ET mechanism between Er³⁺ and Yb³⁺, the typical luminescent dynamics of Er^{3+ 4}F_7/2 and ⁴I_11/2 emissions in GCs were plotted, as shown in Fig. 5. Fig. 5(b) displays the decay curves of the Er^{3+ 4}I_11/2 state, it can be seen that the Er³⁺ single-doped GC shows a nearly single exponential decay. On the other hand, for Er³⁺–Yb³⁺ co-doped GCs, the decay curves obviously deviate from single exponentials with increasing of Yb³⁺ concentration. Because the Er³⁺ concentration is fixed to 0.5 mol% in the four different GCs samples, the decrease in the lifetime should not be ascribed to concentration quenching of Er³⁺ but to direct ET from Er^{3+ 4}I_11/2 to Yb^{3+ 2}F_5/2, which indicates that a part of energy of the Er^{3+ 4}I_11/2 is transferred to Yb³⁺ instead of being emitted in the Er³⁺:⁴I_11/2–⁴I_15/2 transition completely. However, the rise component was not observed, which may be due to the rapid depopulation process of ⁴I_11/2 state. Therefore, the direct ET from Er^{3+ 4}I_11/2 to Yb^{3+ 2}F_5/2 is the second-step ET process according to Fig. 5(b).

Fig. 5 Decay curves of (a) Er³⁺:⁴F_7/2 → ⁴I_11/2 transition under 486 nm (b) Er³⁺:⁴I_11/2 → ⁴I_15/2 transition (948 nm) under 460 nm excitation.

Fig. 5(a) exhibits the luminescence decay curves of the Er^{3+ 4}F_7/2 state. It can be clearly seen that the lifetime of Er^{3+ 4}F_7/2 (486 nm emission) decreased rapidly, which can be explained by the introduction of an extra relaxation pathway (CR ET process from Er^{3+ 4}F_7/2 to Yb^{3+ 2}F_5/2, as reported above). The mean lifetime τ_m of 486 nm emission upon 460 nm excitation was calculated as follows:

where I(t) is the luminescence intensity and I₀ is the maximal value of I(t) that occurs at time t₀. The ETE is defined as the ratio of the number of donors that are depopulated by the ET to the acceptors over the total number of excited donors. The overall theoretical ETE for the GC samples can be calculated using the following equation:

As shown in Fig. 6, the τ_m and ETE were plotted vs. the concentration of Yb³⁺ ions, for GC0 to GC3. The lifetime decreased from 0.78 μs to 0.62 μs with increasing Yb³⁺ content from 0 mol% to 2 mol%, while the ETE increased to 75.3% with x = 2 mol% composition of Yb³⁺ content.

Fig. 6 Mean decay lifetimes of the Er³⁺:⁴F_7/2 → ⁴I_11/2 transition and ETE between the Er³⁺ and Yb³⁺ as a function of Yb³⁺ concentrations in GC0 to GC3.

In addition, the QY of Yb³⁺ is related to η_ETE as follows:

QY = 2η_Ybη_ETE

where η_Yb is the quantum emission efficiency of Yb³⁺, which is normally about 100% owing to the low multi-photon relaxation rate and large energy gap between the ²F_7/2 and ²F_5/2 energy states.²⁸ The corresponding QY for the GC1, GC2, and GC3 were 70.2%, 136.6%, and 150.6%, respectively.

The two-step ET mechanism

In Fig. 7, the energy levels of Er³⁺ and Yb³⁺ are schematically illustrated, and a possible DC mechanism is also proposed for explaining the two-step ET between Er³⁺ and Yb³⁺. When Er³⁺ is excited into the ⁴F_5/2 state upon 460 nm excitation and subsequently relaxes to the ⁴F_7/2 state, the ⁴F_7/2 state depopulates through a two-step sequential ET from Er³⁺ to Yb³⁺, with Er³⁺:⁴I_11/2 acting as the intermediate state. In the first step (indicated by ①), a part of the Er³⁺ energy is transferred to Yb³⁺ according to the CR: Er³⁺:⁴F_7/2 + Yb³⁺:²F_7/2 → Er³⁺:⁴I_11/2 + Yb³⁺:²F_5/2, the first NIR photon is acquired and Er³⁺ is still in the excited state ⁴I_11/2. In the second step (indicated by ②), another part of the energy of Er³⁺ is transferred to the second Yb³⁺ ion according to a direct ET process and the second NIR photon is emitted, which is shown as Er³⁺:⁴I_11/2 + Yb³⁺:²F_7/2 → Er³⁺:⁴I_15/2 + Yb³⁺:²F_5/2. The emission of the Er³⁺:⁴I_11/2–⁴I_15/2 transition may occur and acquire another NIR photon (indicated by ③).

Fig. 7 Schematic energy level diagram of Er³⁺–Yb³⁺ co-doped GCs, showing the two-step ET mechanism for the NIR emission under 460 nm excitation.

Conclusions

In summary, the Er³⁺/Yb³⁺ co-doped GCs containing CaF₂ nanocrystals were obtained, and efficient NIR DC took place via a two-step ET process, which was verified experimentally. The maximal ETE and QY values were estimated to be as high as 75.3% and 150.6%, respectively. According to the ET process, one visible photon can be converted into two NIR photons, which can be efficiently absorbed by c-Si solar cells. Furthermore, GCs are advantageous owing to their higher thermal stability, transparency and low photon energy, which can adequately utilize the solar spectrum and increase the ETE of the Er³⁺–Yb³⁺ couple. Therefore, the Er³⁺–Yb³⁺ co-doped GCs have potential applications as a DC layer to improve the conversion efficiency of c-Si solar cells.

Acknowledgements

This study was supported by National Hi-Tech Research and Development Program (863) Key Project of China (No. 2012AA050301-SQ2011 GX01D01292) and Xi'an Industrial Technology Innovation Project-technology transfer promoting program (No. CX1242, CXY1123-5, CX12182-2, CX12182-3, CXY1421, ​CXY1511(9)).

Notes and references

1. M. Ye and X. Wang, Chin. Opt. Lett., 2014, 12, 053501.
2. J. Fan, B. Jia and M. Gu, Photonics Res., 2014, 2, 111–120.
3. C. Strümpel, M. McCann, G. Beaucarne, V. Arkhipov, A. Slaoui, V. Švrček, C. del Cañizo and I. Tobias, Sol. Energy Mater. Sol. Cells, 2007, 91, 238–249.
4. T. Trupke, M. A. Green and P. Würfel, J. Appl. Phys., 2002, 92, 1668–1674.
5. B. S. Richards, Sol. Energy Mater. Sol. Cells, 2006, 90, 1189–1207.
6. B. S. Richards, Sol. Energy Mater. Sol. Cells, 2006, 90, 2329–2337.
7. S. Ye, B. Zhu, J. Chen, J. Luo and J. Qiu, Appl. Phys. Lett., 2008, 92, 141112.
8. Q. Y. Zhang, G. F. Yang and Z. H. Jiang, Appl. Phys. Lett., 2007, 91, 051903.
9. X. Liu, S. Ye, Y. Qiao, G. Dong, B. Zhu, D. Chen, G. Lakshminarayana and J. Qiu, Appl. Phys. B, 2009, 96, 51–55.
10. G. Lakshminarayana and J. Qiu, J. Alloys Compd., 2009, 481, 582–589.
11. Y. Xu, X. Zhang, S. Dai, B. Fan, H. Ma, J. I. Adam, J. Ren and G. Chen, J. Phys. Chem. C, 2011, 115, 13056–13062.
12. L. Xie, Y. Wang and H. Zhang, Appl. Phys. Lett., 2009, 94, 061905.
13. S. Ye, B. Zhu, J. Luo, J. Chen, G. Lakshminarayana and J. Qiu, Opt. Express, 2008, 16, 8989–8994.
14. X. Liu, Y. Teng, Y. Zhuang, J. Xie, Y. Qiao, G. Dong, D. Chen and J. Qiu, Opt. Lett., 2009, 34, 3565–3567.
15. H. Lin, S. Zhou, H. Teng, Y. Li and W. Li, J. Appl. Phys., 2010, 107, 043107.
16. Y. Tai, G. Zheng, H. Wang and J. Bai, J. Solid State Chem., 2015, 226, 250–254.
17. K. Biswas, S. Balaji, D. Ghosh, A. D. Sontakke and K. Annapurna, J. Alloys Compd., 2014, 608, 266–271.
18. Q. Y. Zhang, C. H. Yang, Z. H. Jiang and X. H. Ji, Appl. Phys. Lett., 2007, 90, 061914.
19. D. Q. Chen, Y. L. Yu, Y. S. Wang and P. Huang, J. Phys. Chem. C, 2009, 113, 6406–6410.
20. D. Chen, Y. Yu, H. Lin, P. Huang, Z. Shan and Y. Wang, Opt. Lett., 2010, 35, 220–222.
21. B. M. van der Ende, L. Aarts and A. Meijerink, Adv. Mater., 2009, 21, 3073–3077.
22. J. J. Eilers, D. Biner, J. T. van Wijngaarden, K. Krämer, H.-U. Güdel and A. Meijerink, Appl. Phys. Lett., 2010, 96, 151106.
23. L. Aarts, B. M. van der Ende and A. Meijerink, J. Appl. Phys., 2009, 106, 023522.
24. S. R. Lüthi, H. U. Güdel and M. P. Hehlen, J. Chem. Phys., 1999, 110, 12033–12043.
25. L. H. Huang, T. Yamashita, R. Jose, Y. Arai, T. Suzuki and Y. Ohishi, Appl. Phys. Lett., 2007, 90, 131116.
26. L. H. Huang, G. S. Qin, Y. Arai, R. Jose, T. Suzuki, Y. Ohishi, T. Yamashita and Y. Akimoto, J. Appl. Phys., 2007, 102, 093506.
27. D. Q. Chen, Y. S. Wang, Y. L. Yu, F. Liu and P. Huang, Opt. Lett., 2007, 32, 3068–3070.
28. J. Zhou, Y. Zhuang, S. Ye, Y. Teng and G. Lin, Appl. Phys. Lett., 2009, 95, 141101.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra25800f

---