Near-infrared quantum cutting and energy transfer mechanism in Lu₂O₃: Tm³⁺/Yb³⁺ phosphor for high-efficiency photovoltaics

Abstract

Near-infrared downconversion phenomenon has been demonstrated in Lu₂O₃: Tm³⁺/Yb³⁺ phosphor upon direct excitation of Tm^3+:1G₄ level at 463 nm. The efficient energy transfer from Tm^{3+: 1}G₄ → Yb: ²F_5/2 has been elucidated by the excitation spectra, the visible and NIR spectra as well as the decay curves of Tm: ¹G₄ state. The mechanism of downconversion in Lu₂O₃:Tm³⁺/Yb³⁺ has been discussed in detail. According to analysis of the dependence of the initial transfer rate over Yb³⁺ ion concentration, it could be included that energy transfer (ET) from Tm³⁺ to Yb³⁺ is a single-step ET process instead of a cooperative one. By varying the Yb³⁺ concentrations, we obtain the Lu₂O₃: 0.2%Tm³⁺/30%Yb³⁺ sample with theoretical quantum efficiency as high as 148.2%. Because the excited state of Yb³⁺ just above the band edge of crystalline silicon, it suggested that Lu₂O₃: Tm³⁺/Yb³⁺ sample will be beneficial to improve the conversion efficiency of c-Si solar cells.

1 Introduction

The long-run demand to exploit renewable, sustainable and clean energy sources continues to stimulate new approaches to manufacture of efficient and low-cost photovoltaic devices [1, 2]. At present, photovoltaic devices fabricated from silicon wafers dominating the marketplace possess a quite low efficiency of only 18% [3]. This is attributed to the large spectrum mismatch between solar radiation spectrum and the response spectrum of solar cells [4]. To circumvent this difficulty, either improvements in structure and composition of solar cells or modification of solar radiation spectrum has been suggested [5–8]. The spectral distribution of sunlight spectrum covers from 300 to 2500 nm at Air Mass 1.5 global (AM 1.5G), but only a small fraction around 800–1100 nm could be effectively utilized by silicon solar cells with 1.05 eV bandgap [9–11]. This is due to the higher reflection and absorption by the antireflection coatings optimized for longer wavelengths of solar cells [12]. To take advantage of high-energy photons with short wavelength, downconversion (DC) is considered as the most promising candidate for converting the incident high-energy photons into efficient near-infrared (NIR) photons to realize spectrum modification so far [13–15]. DC has been realized in RE³⁺/Yb³⁺ couples, such as Er³⁺/Yb³⁺, Pr³⁺/Yb³⁺, Ce³⁺/Yb³⁺, Ho³⁺/Yb³⁺ and Tm³⁺/Yb³⁺ [16–20]. Yb³⁺ is suitable as an acceptor and emitter for the reason that the Yb³⁺ ion has an only excited state approximately 10,000 cm⁻¹ just above the band edge of crystalline silicon [21]. Herein, we mainly focused on Tm³⁺/Yb³⁺ couples which could convert blue photons around 463 nm into NIR photons around 1000 nm by efficient energy transfer from Tm³⁺ to Yb³⁺. The wavelength at 1000 nm could be effectively utilized by solar cells.

The cubic rare earth sesquioxide Lu₂O₃ is selected as a host because of its chemically and environmentally stable properties [22]. On the contrary, the use of fluoride materials is restricted by their toxicity owing to the fluorine-containing species unlike the oxides [23]. It could also achieve a high rare-earth doping level in Lu₂O₃ [24]. Besides, Lu₂O₃ possesses a larger thermal conductivity (12.5 W/mk) [25] than YAG(10.7 W/mk) [26] making it more desirable for device manufacture. Recently, it has been reported that Lu₂O₃ is an excellent host for phosphors with efficient upconversion luminescence properties between Tm³⁺/Yb³⁺, Er³⁺/Yb³⁺, Ho³⁺/Yb³⁺ couples [22, 27]. Moreover, strong downconversion luminescence in Tb³⁺ and Yb³⁺ codoped Lu₂O₃ has also been investigated [28]. However, little researches refer to downconversion between Tm³⁺/Yb³⁺ combinations in Lu₂O₃ host and the transfer mechanism has not been demonstrated yet.

In this study, we report the efficient NIR DC in Lu₂O₃: Tm³⁺/Yb³⁺ phosphor and analyze the energy transfer(ET) mechanism in Tm³⁺/Yb³⁺ couples. We systematically demonstrate the DC process between Tm³⁺/Yb³⁺ couples is a single-step ET process instead of a cooperative ET process assumed in other materials [29]. Visible and NIR emission spectra, decay time and energy transfer efficiency have been investigated in detail with varying Yb³⁺ concentration. Besides, the theoretical quantum efficiency as high as 148.2% has been obtained, well over the efficiency limit (29%) estimated by Shockley and Queisser [30]. Therefore, results show that Lu₂O₃ doped with Tm³⁺/Yb³⁺ offers the opportunity for application in solar cells.

2 Experimental

2.1 Sample preparation

The series of samples investigated in this work with the general formula Lu_{2−0.2%−x%}O₃: 0.2%Tm³⁺, x%Yb³⁺ (x = 0, 1, 5, 10, 20, 30) were prepared by a solid-state reaction. The oxides Lu₂O₃(4 N), Yb₂O₃(4 N), and Tm₂O₃(4 N) were employed as the raw materials, which were mixed homogeneously by an agate mortar for 30 min, placed in a crucible with a lid, then sintered at 1500 °C for 4 h.

2.2 Measurements and characterization

Powder X-ray diffraction (XRD) data was collected using Cu-Kα radiation (λ = 1.54056 Å) on a Bruker D8 advance diffractometer equipped with a linear position-sensitive detector (PSD-50m, M. Braun), operating at 40 kV and 40 mA with a step size of 0.01° (2θ) in the range of 10°–80°. The steady state excitation and emission spectra under direct excitation were measured using an FLS920 spectrometer (Edinburgh Instruments, U.K.). In energy level lifetime measurements, an OPO was used as an excitation source, and the signals were detected using a Tektronix digital oscilloscope (TDS 3052). The lifetimes were calculated by integrating the area under corresponding lifetime curves with the normalized initial intensity.

3 Results and discussion

The structures of Lu₂O₃ samples with nominal compositions 0.2% Tm³⁺/ x% Yb³⁺ (x = 0, 1, 5, 10, 20, 30) were examined by typical powder X-ray diffraction pattern, as shown in Fig. 1. The positions and relative intensity of diffraction peaks for the powder samples can be well indexed to the standard cards of JPCD cards 12–0728. According to Fig. 1, there is no other phase in all XRD patterns. It demonstrates that Tm³⁺ and Yb³⁺ substitutions for Lu³⁺ sites have no effect on the phase structure.

Fig. 1

The standard XRD data of Lu₂O₃ and the XRD patterns of the samples doped with different concentrations of Yb³⁺

In the UV region, high energy photons might be absorbed by either the Lu₂O₃ host or the rare-earth ions through charge transfer and f–d transitions. The possibility of the occurrence of DC is then clarified experimentally. To demonstrate the existence of DC from Tm: ¹G₄ to Yb: ²F_5/2 in Lu₂O₃: Tm³⁺/Yb³⁺ host, excitation spectra (PLE) was measured to be a direct evidence of DC as shown in Fig. 2. The intense excitation band centered at 463 nm is ascribed to ¹G₄ → ³H₆ transition when monitoring the Tm^{3+: 1}G₄ → ³F₄, ¹G₄ → ³H₅, ¹G₄ → ³H₄ emission, respectively. Similarly, when monitoring the Yb^{3+: 2}F_5/2 → ²F_7/2 infrared emission at 980 nm, the excitation band could also be observed. This indicates the presence of direct ET from Tm³⁺ to Yb³⁺. The weak excitation intensity for monitoring 980 nm emission of Yb³⁺ and 1200 nm emission of Tm^{3+: 1}G₄ → ³H₄ is because of lack of signal at infrared wavelength region for InGaAs detector used in present work.

Fig. 2

Excitation spectra of Tm^{3+: 1}G₄ → ³F_4, ¹G₄ → ³H_S, ¹G₄ → ³H₄ emission in Lu₂O₃: 0.2%Tm³⁺/5%Yb³⁺ sample

As another direct evidence of Tm³⁺ → Yb³⁺ ET, the emission spectra covering from visible and NIR range of Lu₂O₃ samples with varying Yb³⁺ concentrations are depicted in Fig. 3. The emission peaked at 654 nm is originated from Tm^{3+: 1}G₄ → ³F₄ transition. The group of emission lines at around 800 nm is corresponding to Tm^{3+: 1}G₄→³H₅ transition. In the NIR region (900–1300 nm), the strong emission band peaked at 1024 nm is observed for the Tm³⁺ and Yb³⁺ codoped samples, which is responsible for Yb^{3+: 2}F_5/2 → ²F_7/2 transition under 463 nm Tm³⁺ excitation. The appearance of Yb³⁺ emission upon Tm^{3+: 1}G₄ excitation demonstrates the effective energy transfer from Tm³⁺ to Yb³⁺. The emission band in the range of 1100–1300 nm is assigned to Tm^{3+: 1}G₄ → ³H₄ transition.

Fig. 3

Comparison of visible and NIR emission spectra of Lu₂O₃ samples with different Yb³⁺ doping levels and fixed 0.2%Tm³⁺ concentration under 463 nm excitation

As depicted in Fig. 3, it can be concluded that the intensities of ¹G₄ → ³F₄, ¹G₄ → ³H₅, ¹G₄ → ³H₄ transition of Tm³⁺ have shown a markedly fast decrease with Yb³⁺ concentration increasing. Meanwhile, the NIR emission of Yb³⁺ increases rapidly when raising Yb³⁺ doping levels. This is attributed to the energy transfer process described as Tm^{3+: 1}G₄ → Yb^{3+: 2}F_5/2 to convert blue photons into NIR photons. It is noticed that when the concentration of Yb³⁺ is over 5%, the intensity of Yb³⁺ emission decreases gradually, which is due to concentration quenching between Yb³⁺ ions. Therefore, 5 mol% denotes the optimal Yb³⁺ doped concentration in this series samples.

The PLE and PL spectra prove the energy transfer from Tm³⁺ to Yb³⁺ does exist in Lu₂O₃ host. For better understanding the energy transfer mechanism, the schematic thulium–ytterbium energy level diagram under 463 nm excitation with involved ET processes is depicted in Fig. 4. Upon 463 nm excitation, the Tm³⁺ ion in the ground state ³H₄ is populated to the upper ¹G₄ level. The Tm³⁺ ions in the ¹G₄ level may undergo two possible energy transfer routes: (1) Cooperative ET, as indicated in Fig. 4(a). As the excited energy level of Tm³⁺ donor ion is located at approximately twice the energy of that of Yb³⁺, the DC mechanism was supposed to be a second-order cooperative ET process before [31]. A Tm³⁺ ion in the ¹G₄ level transfers energy simultaneously to two nearby Yb³⁺ ions in the ground state and excites them to ²F_5/2 level, resulting in two emitting photons of Yb³⁺ with a wavelength around 1000 nm. (2)Single-step energy transfer, as shown in Fig. 4(b). A Tm³⁺ ion at ¹G₄ state transfers energy to one Yb³⁺ nearby through a phonon-assisted ET process, leading to one NIR emitting photon of Yb³⁺. Meanwhile, it populates the ³H₅ state of Tm³⁺ and results in mid-NIR emission from ³F₄ state through fast nonradiative relaxation from ³H₅. It is known that cooperative ET is hard to occur because of the less probability of a second order ET comparing with a first-order one. In order to determine the real energy transfer process in Lu₂O₃: Tm³⁺/Yb³⁺ system, the dependence of energy transfer rate upon Yb³⁺ ion concentration has been discussed. For the single-step energy transfer situation, the transfer rate of a Tm³⁺ ion at site 0 with a Yb³⁺ surrounding at site i can be described as.

Fig. 4

Schematic thulium-ytterbium energy level diagram under 463 nm excitation including possible ET processes

$${{W}_{0i}}=xf({{R}_{0i}}){{X}_{0}}$$

(1)

x denotes the concentration of the acceptor Yb³⁺ and thus could be representative of the probability of a Yb³⁺ at site i. X₀ is the radiative decay rate of Tm^3+:1G₄ state in Lu₂O₃. As a consequence, for all Tm³⁺ ions with Yb³⁺ neighboring at distances R_0i, the transfer rate for single-step energy transfer can be written as

$${{W}_{0}}_{S}=x\underset{i}{\mathop \sum }\,f({{R}_{0i}}){{X}_{0}}$$

(2)

It leads to

$${{W}_{0}}_{S}\propto \text{x}$$

(3)

In this way, the energy transfer rate of single-step energy transfer is proportional to the acceptor concentration.

For the cooperative ET situation, one Tm³⁺ transfers its energy simultaneously to two Yb³⁺ adjacent at site i and j (i < j). According to Eq. (1), the transfer rate could be obtained as

$${{W}_{0}}\left( i,j \right)={{x}^{2}}f({{R}_{oi}})f({{R}_{oj}}){{X}_{0}}^{2}$$

(4)

Taking account into all the pairs of Yb³⁺ ions, the energy transfer rate of cooperative ET is

$${{W}_{0}}_{COOP}={{x}^{2}}\underset{i<j}{\mathop \sum }\,f({{R}_{0i}})f({{R}_{0j}}){{X}_{0}}^{2}$$

(5)

Thus, it could be deduced that

$${{W}_{0}}_{COOP}\propto {{x}^{2}}~$$

(6)

That is to say, the energy transfer rate of cooperative ET is proportional to the square of acceptor concentration.

The energy transfer rates are obtained from the decay curves of Tm^{3+: 1}G₄ level in Lu₂O₃: 0.2%Tm³⁺/ x%Yb³⁺ (x = 0, 1, 5, 10, 20, 30) depicted in Fig. 5. A rapid decline in decay curves has been observed due to the existence of extra decay pathways with Yb³⁺ concentration increasing. Considerable energy transfer of Tm^{3+: 1}G₄ → Yb^{3+: 2}F_5/2 accelerates the depopulation of Tm^{3+: 1}G₄ state and leads to a variety of ET rates with different Yb³⁺ doping concentrations. All decay curves exhibit nonexponential characteristics. This result suggests that the radiative decay of Tm³⁺ might be accompanied by both multiphonon assisted relaxation and ET to adjacent Yb³⁺ ions. We could calculate the lifetime by integrating the area under the corresponding decay curves with the normalized initial intensity at different Yb³⁺ doping levels. The lifetime results are presented in Table 1. The decay rate can be generally expressed by W = W_r + W_nr + W_ET, where W_r, W_nr, and W_ET are the radiative decay, non-radiative decay, and the ET rates. Accordingly, the energy transfer rate W_ET can be defined as

Fig. 5

Decay curves of Tm^3+:1G₄ state in Lu₂O₃: 0.2%Tm³⁺/x%Yb³⁺ (x = 0, 1, 5, 10, 20, 30) under 463 nm excitation

$${{W}_{ET}}=\frac{1}{\tau }-\frac{1}{{{\tau }_{0}}}$$

(7)

where ${{\tau }_{0}}$ and $\tau$ denote the average lifetimes of Tm^{3+: 1}G₄ state in the absence and presence of energy transfer acceptors (Yb³⁺ ions). Consequently, we obtained the Yb³⁺ ion concentration dependence of the energy transfer rates and plotted in a double-logarithmic diagram, as shown in Fig. 6. When the concentration of Yb³⁺ ions changes from 1 to 30%, the slope is found to be around 1. It indicates that the Tm³⁺-Yb³⁺ ET system is dominated by the single-step ET process rather than a cooperative ET process according to the analysis above. The possibility is quite low for Tm³⁺→Yb³⁺ energy transfer to achieve DC with one-to-two photon emission in Lu₂O₃ material.

Fig. 6

Plot (log–log) of the energy transfer rate versus the Yb³⁺ concentration in Lu₂O₃: 0.2%Tm³⁺/x%Yb³⁺ (x = 0, 1, 5, 10, 20, 30)

Based on the decay curves shown in Fig. 5, the energy transfer efficiency (η _ETE ) and theoretical quantum efficiency (η _TQE ) can be determined. The η _ETE is defined as the ratio of Tm³⁺ depopulated by energy transfer to Yb³⁺ over the total number of excited Tm³⁺ ions. Hence, the η _ETE could be expressed as a function of Yb³⁺ concentration by lifetimes

$${{\eta }_{ETE,x\%Yb}}=1-\frac{\tau }{{{\tau }_{0}}}$$

(8)

where $\tau$ denotes the decay time of Tm^{3+: 1}G₄ level with various concentrations of Yb³⁺, the same as Eq. (7). For the samples doped with x mol% Yb³⁺ ions (x = 1, 5, 10, 20, 30), the ET from Tm³⁺ to Yb³⁺ occurs by single-step ET process. In this way, one infrared photons is emitted by Yb³⁺ ions per absorbed one blue photon (around 463 nm) by Tm³⁺ ions in theory. Correspondingly, theoretical quantum efficiency (η _TQE ) is defined as the ratio for the number of emitted photons over the number of absorbed photons. The relation between the energy transfer efficiency and the theoretical quantum efficiency is linear and can be expressed as

$${{\eta }_{TQE,x\%Yb}}={{\eta }_{Tm}}\left( 1-{{\eta }_{ETE,x\%Yb}} \right)+{{\eta }_{Yb}}{{\eta }_{ETE,x\%Yb}}+{{\eta }_{Tm}}{{\eta }_{ETE,x\%Yb}}$$

(9)

where ${{\eta }_{Tm}}$ and ${{\eta }_{Yb}}$ correspond to the quantum efficiencies of Tm³⁺ and Yb³⁺ ions, respectively. By assuming that all excited ions decay radiatively, their values are set to 1. This means nonradiative processes such as phonon-assisted relaxation, cross-relaxation between Tm³⁺ and Yb³⁺ as well as energy back transfer from Yb³⁺ to Tm³⁺ are all eliminated. The assumption can reach the upper limit of the theoretical quantum efficiency, the Eq. (9) thus can be simplified as

$${{\eta }_{TQE,x\%Yb}}=1+{{\eta }_{ETE,x\%Yb}}$$

(10)

The ${{\eta }_{ETE}}$ and ${{\eta }_{TQE}}$ are listed in Table 1 based on the formulas mentioned above. The estimated ${{\eta }_{ETE}}$ exhibits an obvious increase from 3.9 to 48.2% with increasing Yb³⁺ concentration. It is worth noticed that ${{\eta }_{TQE}}$ reaches 148.2% for the Lu₂O₃: 0.2%Tm³⁺/30% Yb³⁺ sample.

4 Conclusion

In this study, we demonstrate the efficient NIR downconversion in Tm³⁺ and Yb³⁺ codoped Lu₂O₃ samples. By measuring the excitation and emission spectra as well as the lifetime of ¹G₄ level, effective energy transfer from Tm^{3+: 1}G₄ to Yb^{3+: 2}F_5/2 has been proved under 463 nm excitation. The dependence of the initial transfer rate upon Yb³⁺ concentration has been discussed to determine the downconversion process. It demonstrates that the ET from Tm³⁺ to Yb³⁺ occurs by single-step ET process. With raising Yb³⁺ doping level, a continuous increasing in theoretical quantum efficiency has been obtained by carefully calculation. The estimated maximum theoretical quantum efficiency reaches 148.2% in the Lu₂O₃: 0.2%Tm³⁺/30%Yb³⁺ sample. As the main emission peak of Yb³⁺ in Lu₂O₃ around 1000 nm matches better with the optimal spectral response of the c-Si solar cell, it indicates that Lu₂O₃: Tm³⁺/Yb³⁺ is promising DC material for application in solar cells.

Table 1 Decay time, energy transfer efficiency (η _ETE ) and theoretical quantum efficiency (η _TQE ) as functions of Yb³⁺ concentration in Lu₂O₃: 0.2%Tm³⁺/x%Yb³⁺ samples