1. Introduction

Frequency upconversion (UC) from infrared (IR) to visible/ultraviolet (UV) by materials doped with trivalent rare-earth (RE) ions, such as Tm³⁺, Er³⁺, Ho³⁺, Nd³+, and Pr³⁺, has attracted intensive attention for more than 30 years [1,2]. Its potential applications include detection of infrared radiation, optical storage, color display, UC lasers, biosensor, and so on [3-7]. Most of these applications have been achieved with the help of fluorides due to their low phonon energy and high quantum efficiencies as luminescent hosts [8]. Among RE ions, Tm³⁺ has attracted increasing attention because it has metastable levels (Fig. 1) suitable for emitting blue and UV UC luminescence [9-11]. Especially, UV UC luminescence from Yb³⁺-sensitized Tm³⁺ in fluorides has become one hot topic of optical functional materials in recent years [12]. Our group has reported strong UV UC emissions from Yb³⁺/Tm³⁺-codoped fluoride film and attributed the UV enhancement to the decrease of Judd-Ofelt parameter Ω₂, which reflects the symmetry of the crystal field [13]. Wang et al. have reported intense UV UC luminescence from Yb³⁺/Tm³⁺-codoped YF₃ nanocrystals embedded glass ceramic [14]. Chen et al. demonstrated six-photon and five-photon UV UC luminescence in Yb³⁺/Tm³⁺-codoped ZBLAN fluoride glass [12]. Despite the fact that UV UC emissions have been widely investigated and some novel spectral phenomena have been reported, the study of new approaches to obtain efficient UV luminescence as well as the mechanism of the enhancement is still a challenge for the development of short-wavelength solid-state lasers [12].

 

Fig. 1. Energy level diagrams of Yb³⁺ ions and Tm³⁺ ions and UC emission mechanism in NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals.

Download Full Size | PDF

In this study, intense UV UC luminescence was presented in hexagonal NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals under 980 nm excitation. The ³P₂→³H₆ and ³P₂→³F₄ emissions were observed for the first time. In comparison with the strong emissions from ¹D₂ and ¹I₆, the emissions from ¹G₄ and ³H₄ almost vanished.

2. Experimental

Y₂O₃, Yb₂O₃, and Tm₂O₃ (purity ≥ 99.999 %) were supplied by Shanghai Chemical Reagent Company. EDTA, NaF, and HNO₃ were supplied by Beijing Chemical Reagent Company, and were of analytical grade. Deionized water was used to prepare solutions. Ln₂O₃ (Ln=Y,Yb, and Tm) were dissolved in dilute HNO₃ by heating to prepare the stock solution of Ln(NO₃)₃.

In a typical synthesis, 1 mL of 0.5 M Ln(NO₃)₃ aqueous solution and 0.5 mmol of EDTA were dispensed into 20 mL of deionized water and magnetically stirred for 1 h, forming a chelated Ln-EDTA complex. Then 16 mL of 0.5 M NaF aqueous solution was added to the solution. After vigorous stirring for 1 h, the mixture was transferred into a 50-mL Teflon-lined stainless steel autoclave. The autoclave was sealed and maintained in an oven at 160 °C for 18 h, and then cooled down slowly to room temperature. Subsequently, the suspension was centrifuged at 8000 rpm for 10 min. The resultant product was then washed thoroughly and dried in vacuum at 80 °C. To improve the crystallinity of the nanocrystalline powder, the product was annealed at 400 °C for 2 h.

The crystal structure was analyzed by a Rigaku RU-200b X-ray powder diffractometer (XRD) using a nickel-filtered Cu-Ka radiation (λ=1.4518Å). The size and morphology were investigated by scanning electron microscope (SEM, KYKY 1000B). UC luminescence spectra were recorded with a Hitachi F-4500 fluorescence spectrophotometer (2.5 nm for spectral resolution (FWHM) of the spectrophotometer and 400 V for PMT voltage) at room temperature.

Figure 2(a) shows the XRD pattern of the microcrystals. All the diffraction peaks can be indexed to the pure hexagonal NaYF₄ (JCPDS 16-0334). No other impurity peaks were detected. The corresponding SEM image (Fig. 2(b)) shows that the NaYF₄:Yb³⁺/Tm³⁺ microcrystals are hexagonal pillars.

 

Fig. 2. XRD pattern (a) and SEM image (b) of the NaYF₄:Yb³⁺/Tm³⁺ microcrystals.

Download Full Size | PDF

3. Results

3.1. Intense UV and weak visible emissions

Under 980 nm excitation (~320 W/cm²), unusually intense UV and weak visible emissions were observed in NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals, as shown in Fig. 3. These emissions come from the following transitions of Tm³⁺ ions: ³P₂→³H₆ (~264 nm) [11], ¹I₆→³H₆ (~291 nm), ³P₂→³F₄ (~309 nm), ¹I₆→³F₄ (~347 nm), ¹D₂→³H₆ (~363 nm), ¹D₂→³F₄ (~454 nm), ¹G₄ ³H₆ (~475 nm), ¹D₂→³H₅ (~508 nm), ¹D₂→³H₄ (~574 nm), ¹G₄→³F₄ (~642 nm), ³F₃→³H₆ (~689 nm), and ³H₄→³H₆ (~795 nm). In comparison with usual UC emission spectra of Tm³⁺ [13,14], the ¹G₄→³H₆ and ³H₄→³H₆ emissions almost vanished. Although Noginov et al. have predicted that the back energy transfer of ¹G₄ (Tm³⁺)+²F_7/2 (Yb³⁺)→³H₄ (Tm³⁺)+²F_5/2 (Yb³⁺) could lead the ¹G₄ level to be quenched [15], this phenomenon has never been observed in experiments so far.

 

Fig. 3. UC luminescence spectrum of NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals under 980 nm excitation (320 W/cm²). Inset: magnification of the spectrum in the range of 430–850 nm.

Download Full Size | PDF

3.2. Unusual ³P₂→³H₆ (~264 nm) and ³P₂→³F₄ (~309 nm) emissions

This is the first report on the ³P₂→³H₆ and ³P₂→³F₄ emissions in Yb³⁺/Tm³⁺-codoped systems under 980 nm excitation. In our previous observations [8,13], the ³P₂ level usually relaxed to ¹I₆ level and emitted 291 and 347 nm UV UC fluorescence. Radiative transitions from the ³P₂ level are difficult to occur due to the nearby ³P_0,1 and ¹I₆ levels offer the routes for rapidly nonradiative relaxation. The nonradiative relaxation of ³P₂→³P₁→³P₀→¹I₆ [11] is greatly influenced by the phonon energy and competes with the radiative transitions of ³P₂→³H₆ and ³P₂→³F₄. Presumably, two mechanisms might be responsible for 264 and 309 nm emissions. (1) Low phonon energy helps to decrease the nonradiative transition of ³P₂→³P₁→³P₀→¹I₆ [16], resulting in the ³P₂→³H₆ and ³P₂→³F₄ emissions. However, the low phonon energy is unfavorable to phonon-assisted energy transfer from Yb³⁺ to Tm³⁺, leading the population of the ³P₂ level to decrease. Therefore, 264 and 309 nm emissions can be observed only when the doped host has appropriate phonon energy, on which no final conclusion has yet been reached. (2) The efficient population of the ³P₂ level causes the 264 and 309 nm emissions. Here, the luminescence intensity of ¹I₆→³H₆ (~291 nm) is ~6 times stronger than that of ¹G₄→³H₆ (~475 nm), indicating that the population of ³P₂ is high enough to populate efficiently the ¹I₆. Corresponding, the enhanced population of ³P₂ causes the 264 and 309 nm emissions, which will be described later.

3.3. Power dependence of UC luminescence.

For nearly any UC mechanism, the emission intensity is proportional to the n-th power of the excitation intensity, and the integer n is the number of photons absorbed per upconverted photon emitted [17]. In the present case, the results are different. Figure 4 shows the double logarithmic plots of the emission intensity as a function of excitation power for the ³H₄→³H₆, ¹G₄→³H₆, ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions. The corresponding slopes (n) of the dependencies are listed in Table 1.

In the initial stage (0.09-0.14 W), n=1.66, 2.80, 3.74, and 4.57 for the ³H₄→³H₆, ¹G₄→³H₆, ¹D₂→³H₆, and ¹I₆→³H₆ emissions, respectively. This means that the populations of the states ¹I₆, ¹D₂, ¹G₄, and ³H₄ come from five-photon, four-photon, three-photon, and two-photon UC processes, respectively. However, with the increase of excitation power, the values of n changed significantly. In the range from 0.15 to 0.26 W, n=1.23, 1.86, 2.87, 3.14, and 3.17 for the ³H₄→³H₆, ¹G₄→³H₆, ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions, respectively. In the range from 0.26 to 0.31 W, the UC luminescence deviates from a linear relationship. In the final stage (0.31-0.39 W), n=0.78, 1.78, 3.16, 3.69, and 3.87 for the ³H₄→³H₆, ¹G₄→³H₆, ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions, respectively. Obviously, when the excitation power is higher than 0.15 W, the slopes (n) of the ³H₄→³H₆ and ¹G₄→ ³H₆ emission at low laser power are larger than those at high laser power, while the slopes (n) of the ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions at low laser power are smaller than those at high laser power.

 

Fig. 4. Plots (log-log) of emission intensity versus excitation power.

Download Full Size | PDF

Table 1. Experimental slopes (n) of the dependencies log(emission intensity) versus log(excitation power) for different emissions in NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals under 980 nm excitation.

View Table

4. Mechanism analysis

4.1. Analysis on the nearly vanished radiative transitions from ¹G₄ and ³H₄

A careful analysis of the luminescence dynamics of NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals can reveal the physical mechanism behind the nearly vanished radiative transitions from ¹G₄ and ³H₄. Figure 1 shows the detailed UC processes of populations, relaxations, and emissions in the microcrystals. According to our previous report [12], the pump light excites only the Yb³⁺ ions, and three successive energy transfers from Yb³⁺ to Tm³⁺ populate ³H₅, ³F₂, and ¹G₄ levels. The ¹D₂ level of Tm³⁺ cannot be populated by the fourth photon from Yb³⁺ via energy transfer to the ¹G₄ due to the large energy mismatch (about 3500 cm^-1) between them [13]. The cross-relaxation between Tm³⁺ ions may alternatively play an important role in populating ¹D₂ level. Usually, there are primarily two cross-relaxation processes in populating the ¹D₂ level: ³F₂+³H₄→³H⁶+¹D₂ [14] and ¹G₄+³H₄→³F₄+¹D₂ [12]. They will become stronger with increasing Tm³⁺ concentration and lead the populations of the ¹G₄ and ³H₄ to decrease. To further verify the mechanism related to the cross-relaxation processes, the dependence of the UC luminescence on the Tm³⁺ concentration was studied (Fig. 5). From the bottom to the top in Fig. 5, the doped concentrations of Tm³⁺ ions are 0.01mol%, 0.1mol%, 0.5 mol%, 1 mol%, and 1.5 mol%, respectively. We can see that the UC emissions from the ¹I₆ and ¹D₂ get strong and strong with Tm³⁺ concentration increasing, and the UC emissions from the ¹G₄ and ³H₄ get weaker. When the Tm³⁺ concentration reaches to 1.5 mol%, the ¹G₄ and ³H₄ emissions nearly vanished.

The populated ¹D₂ level radiatively relaxes to the ground-state and inter-states, which cause 363, 454, 508, and 574 nm emissions, as shown in the inset of Fig. 3. On the other hand, the ¹D₂ state may be excited to the ³P₂ state via another energy transfer from excited Yb³⁺, producing 264 and 309 nm emissions, simultaneously. Before this work, there has not been being any experimental evidence to verify which of ³P_0,1,2 and ¹I₆ levels is the terminal level from the ¹D₂ level for the five-photon UC process of Tm³⁺ ions. The UC emissions of ³P₂→³H₆ (~264 nm) and ³P₂→³F₄ (~309 nm) provide evidence for the energy transfer from Yb³⁺ to Tm³⁺ populating ³P₂. Of course, the UC emissions from the ¹I₆ are still dominating in the five-photon UC processes due to the rapidly nonradiative relaxation of ³P₂→³P₁→³P₀→¹I₆. For a relative high doping concentration, the microcrystals therefore emitted intense 347 and 291 nm fluorescence under the NIR excitation. Furthermore, since the photon (980 nm) exactly matches the energy separation between ¹D₂ and ³P₂, the single-photon absorption of ¹D₂→³P₂ might be exist.

 

Fig. 5. Dependence of UC luminescence spectra (normalized to ¹D₂→³F₄ transition) on Tm³⁺ concentration (20% Yb³⁺): (a) 0.01%, (b) 0.1%, (c) 0.5%, (d) 1%, and (e) 1.5%.

Download Full Size | PDF

4.2. Competition between linear decay and UC processes

For any UC mechanism, although the n-photon process corresponds to a slope approximately equal to n, the dependence of UC luminescence intensity on pump power is also expected to decrease in slope. Pollnau et al. attributed the decrease of n to the competition between linear decay and UC processes for the depletion of the intermediate excited states [17]. They reported that the UC luminescence intensity for an n-photon energy transfer process was proportional to the n-th power of the pump power (Pⁿ) in the limit of infinitely small UC rates, while the intensity was proportional to the pump power (P ¹) in the limit of infinitely large UC rates. Therefore, the intensity of an UC luminescence excited by the sequential absorption of n photons has a dependence of P^β on absorbed pump power P, with 1<β<n [17].

Although the population of the state ³H₄ comes from a two-photon process, the slope of the ³H₄→³H₆ emission was estimated to be 1.23 of pump power in the range from 0.15 to 0.26 W. For the ³H₄→³H₆ emission, the intermediate excited state is ³F₄. Therefore, The UC becomes the dominant depletion mechanism for ³F₄ in NaYF₄:Yb³⁺/Tm³⁺ microcrystals [18]. For the ¹G₄→³H₆ emission (a three-photon process), n=1.86 with a large discrepancy from n=3 is observed. It is reasonable to ascribe this to the increase of the nonradiative relaxation from the ³H₄ state together with the reduction of n for the ³H₄ state [18]. Similarly, n=2.87 with a large discrepancy from n=4 for the ¹D₂→³H₆ emission, n=3.14 with a large discrepancy from n=5 for the ¹I₆→³H₆ (~291 nm) emission, and n=3.17 with a large discrepancy from n=5 for the ³P₂→³H₆ (~264 nm) emission are also observed.

4.3. Cross-relaxation processes and thermal effect at high laser power

In most cases, the values of n at lower excitation power are higher than those at higher excitation power due to typical saturation phenomenon and thermal quenching. (1) The typical saturation phenomenon is caused by the population exhaustion at the ground state; (2) The high non-radiative rates at high excitation densities lead to an increase of the temperature in the internal sample, inducing a thermal effect and which will lead to the quenching of UC luminescence [19,20]. However, for the ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions, the values of n in the range from 0.15 to 0.26 W are smaller than those in the range from 0.31 to 0.39 W. For the ³H₄→³H₆ and ¹G₄→³H₆ emissions, although the values of n in the range from 0.15 to 0.26 W are higher than those in the range from 0.31 to 0.39 W, the log-log plots of UC luminescence versus excitation power are not typical saturation curves at all. The typical saturation curves should gradually bend downward.

We suggested that two mechanisms might cause this anomalous phenomenon. First, under different excitation power, different UC mechanisms are dominant [21]. When the excitation power is high enough, the ¹G₄ and ¹D₂ are populated effectively, which are benefit for the cross-relaxation of ¹G₄+³H₄→³F₄+¹D₂ between the two neighboring Tm³⁺ ions [20]. This cross-relaxation process can result in the increase of n for the emissions from ¹D₂, ¹I₆, and ³P₂. Figure 6 shows the UC luminescence spectra under different excitation power density. When the excitation power increase, the effective populations of the ¹G₄ and the ³H₄ intensively enhance the cross-relaxation mentioned above.

 

Fig. 6. UC luminescence spectra (normalized to ¹D₂→³F₄ transition) of NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals under different excitation power density.

Download Full Size | PDF

Second, this anomalous phenomenon we observed must be closely related to the effects of phonon density on the nonradiative transitions of Tm³⁺ levels and the phonon-assisted energy transfers between Tm³⁺ and Yb³⁺ ions. The observed values of n are results of competition between nonradiative relaxation (depopulation) and phonon-assisted energy transfer (population). It is well-known that the nonradiative relaxation rate of the excited state can be written as [20]

where, W_NR(0) is the nonradiative relaxation rate at 0 K,ħω is the phonon energy, and <n> is the phonon density of states, which strongly depends on phonon energy and temperature [20]

where, k is Boltzmann’s constant and T temperature. The phonon density of states increases with increasing temperature. The increased phonon density of states not only results in the increased nonradiative relaxation rate and quenched UC luminescence, but also benefits the phonon-assisted energy transfers from Yb³⁺ to Tm³⁺. Therefore, we suggest that the complicated dependent relationships were mainly caused by phonon-assisted energy transfers and nonradiative relaxation. At lower excitation power (0.15 – 0.26 W), the thermal effect can be negligible, and the log-log plots of UC luminescence versus excitation power are straight lines. With the increase of excitation power (0.26 – 0.31 W), the plots gradually bend downward, indicating that the nonradiative relaxations increase faster than the phononassisted energy transfers. When the excitation power is high enough (0.31 – 0.39 W), the slopes (n) of the ¹D₂→³H₆, ¹I₆→³H₆, and ³P₂→³H₆ emissions again increase, indicating that the energy transfers increase faster than the nonradiative relaxations. Of course, the underlying upconversion dynamics are complex and require further study.

5. Conclusions

Under 980 nm excitation, novel UC luminescent properties were presented in hexagonal NaYF₄:Yb³⁺(20%)/Tm³⁺(1.5%) microcrystals. The ³P₂→³H₆ (~264 nm) and ³P₂→³F₄ (~309 nm) emissions were observed for the first time. In comparison with usual UC emission spectrum of Tm³⁺, the ¹G₄→³H₆ and ³H₄→³H₆ emissions almost vanished. The dependence of the Tm³⁺ concentration on the UC luminescence indicated that the unusual phenomenon was caused by the efficient cross-relaxation of ¹G₄+³H₄→³F₄+¹D₂ (Tm³⁺) under high Tm³⁺ concentration.

In addition, the number of laser photons absorbed in one UC excitation process, n, changed in different excitation power range, which was theoretically explained considering different UC excitation mechanisms together with thermal effect under higher excitation power.