1. Introduction

Lanthanide-doped upconversion nanocrystals have attracted much research interest in the recent decades. Under excitation of near-infrared laser, i.e. at 976 nm, upconversion nanocrystals exhibit intense visible emissions that have been widely applied in biolabeling and bioimaging [1, 2]. The energy transfer upconversion (ETU) has been considered as the most efficient approach to achieve upconversion [3]. Ytterbium ions (Yb³⁺) are used as sensitizers to absorb energy from pump laser and transfer to activators such as Er³⁺, Tm³⁺, and Ho³⁺ by ETU [4–7]. Some high energy levels of other lanthanide ions, i.e. Tb³⁺ and Eu³⁺ cannot be excited by ETU from Yb³⁺. Energy transfer and migration from Tm³⁺ and Er³⁺ were extensively used to sensitize other type of lanthanide activators, e.g. Tb³⁺ and Eu³⁺ [8–10]. Energy transfer also occurs between low energy levels (i.e. less than 3000 cm⁻¹) in lanthanide ions except for Gd³⁺ and Y³⁺ which have no excited states in this region. The energy transfer processes sufficiently change the population of low-lying levels and further affect the population of high energy levels. In other words, the energy transfers among low-lying levels determine the upconversion processes. Unfortunately, there is no clear investigation of sensitization processes from other activators such as Tm³⁺, especially the energy transfers between low energy levels.

Temperature sensing is one of the potential applications of upconversion phosphors [11–15]. Ishiwada et al. reported that the visible emissions from Tm³⁺ and Tb³⁺ can be affected by temperature via energy transfer processes from the ¹G₄ level of Tm³⁺ to the ⁵D₄ level of Tb³⁺ [16]. Although energy level positions are similar, the strongest emission peaks from the ¹G₄ and ⁵D₄ level are different that they are about 474 nm (blue light) and 544 nm (green light), respectively [17–19]. Owing to the energy transfer process, the color of emission lights can be tuned between green and blue with various thermal conditions under the excitation by 355 nm UV light [16]. If visible emissions from upconversion phosphors can be widely tuned by energy transfer processes between activators, it is meaningful not only for sensing but also for the investigation of energy transfer processes.

Upconversion processes are strongly dependent on the power density of pump laser [20]. Zhao et al. reported that emission ratios in NaYF₄:Tm³⁺,Yb³⁺ nanocrystals changed when the power density of pump laser increased [21]. If energy transfer processes between two types of lanthanide activators exist, the populations of certain energy levels will have dramatic changes under excitation at various power density. As a result, visible upconversion emissions from the phosphors can be consecutively tuned over a wide range based on the population changing. This type of upconversion phosphors is promising for sensors to indicate power density variation.

Although nanocrystals have many applications, particle size, surface defects and adsorbed ligands determine quantum yield of nanocrystals [22–24]. The decrement of quantum yield is due to the quenching effect caused on the surface. Therefore, microcrystals that have low surface-to-volume ratio are preferred for the investigation on energy transfer to exclude the influence of surface [24].

Here, we report the energy transfer between Tm³⁺ and Tb³⁺ dependent on the power density of pump laser in NaYF₄: Tb³⁺,Tm³⁺,Yb³⁺ microcrystals. Hexagonal phase NaYF₄ is known as one of the most efficient host materials for upconversion [4,25]. The Tb³⁺,Tm³⁺,Yb³⁺-doped NaYF₄ microcrystals were synthesized via a hydrothermal method with similar shape and surface-to-volume ratio. The samples showed intense upconversion emissions. The color of upconversion emissions were tuned from green to blue via increasing the power density of 976-nm pump laser. When the power density was low, the energy transfer from ³F₄ (Tm³⁺) to ⁷F₀ (Tb³⁺) was sufficient. The population on levels of Tm³⁺ were inhibited. Only emissions from Tb³⁺ were observed. When the power density of pump laser was increased, the number of Tb³⁺ ions on the ground state decreased, consequently, the energy transfer from ³F₄ (Tm³⁺) to ⁷F₀ (Tb³⁺) was relatively reduced. The high energy levels of Tm³⁺ were populated and emitted through radiative transitions which are dominant in emission spectrum. Energy transfer efficiency from Tm³⁺ to Tb³⁺ increased when power density of pump laser became higher. The cross-relaxation between Tm³⁺ and Tb³⁺, which determines the populations of energy levels for upconversion emissions, depends on the laser power density. This type of upconversion phosphors can be used for power density sensors in the future.

2. Experiments

The samples were synthesized via a hydrothermal method as previously reported [19]. Typically, 0.5 mmol of RE(NO₃)₃ aqueous solution (RE = 59 mol% Y, 20 mol% Tb, 1 mol% Tm, and 20 mol% Yb), and 0.5 mmol of EDTA were well mixed under vigorous stirring for 1 h. The rare earth complex (RE-complex) was formed. Then 6 mmol of NaF aqueous solution was added into the RE-complex and kept stirring for 1 h. The mixture was transferred into a 25 mL autoclave and kept at 180 °C for 12 h. The resulting products were collected by centrifugation, washed several times by distilled water, and finally obtained by drying at 60 °C. The procedure of synthesizing other samples were similar, except concentrations of lanthanide ions. The samples doped with 1%Tm³⁺/20%Yb³⁺, 10%Tb³⁺/1%Tm³⁺/20%Yb³⁺, 20%Tb³⁺/1%Tm³⁺/20%Yb³⁺, and 20%Tb³⁺/20%Yb³⁺ were denoted as TmYb, 10TbTmYb, 20TbTmYb, 20TbYb, respectively.

Crystal structures were characterized by a X-ray diffraction (XRD) on a LabX XRD-6100 X-ray diffractometer with a Cu Kα radiation source (λ=1.5405 Å) operated at 40 kV and 30 mA. The scan was performed in the arrange from 2θ = 10° to 80° with a scan speed of 0.02°/s in steps of 0.02°. Morphologies were observed on a field emission scanning electron microscope (FE-SEM, JEOL-7000F). Upconversion and near-infrared emission spectra were measured using a monochromator equipped with a photomultiplier tube (Hamamatsu). Absorption spectrum in the near-infrared region was measured on a spectrometer (Perkin Elmer, Lambda 900). A fiber coupled 976 nm laser diode (LD) with tunable power up to 154 mW was used as a pump source. Emission decay curves were recorded by a 200 MHz digital oscilloscope for temporal investigation.

3. Results and discussion

3.1. Crystal structure and morphology

Crystal structures of the as-prepared samples were characterized by a X-ray powder diffraction. Figure 1 shows the XRD patterns of samples TmYb, 10TbTmYb, 20TbTmYb, and 20TbYb. The obtained patterns can be indexed to hexagonal phase NaYF₄ (JCPDS 16-0334). Although Tb³⁺ and Yb³⁺ concentrations were high, up to 20%, no diffraction peaks of impurities were detected. It indicates that hexagonal phase NaYF₄ was the main phase. Sharp diffraction peaks indicate that samples were crystallized well.

 

Fig. 1 XRD patterns of Tb³⁺-Tm³⁺-Yb³⁺ doped NaYF₄ microcrystals synthesized by a hydrothermal method: (a) TmYb, (b) 10TbTmYb, (c) 20TbTmYb, and (d) 20TbYb. Black sticks correspond to the diffraction peaks of hexagonal phase NaYF₄ (JCPDS 16-0334).

Download Full Size | PDF

Morphologies of samples were all in the shape of hexagonal rod, as shown in Fig. 2. The averaged lengths of the rods were 12.5, 11.1, 9.6, and 7.0 µm and the averaged diameter of the rods were 6.0, 2.8, 4.0, and 2.6 µm for TmYb, 10TbTmYb, 20TbTmYb, and 20TbYb samples, respectively. The surface-to-volume ratios of the as-prepared samples were about 0.0008, 0.0016, 0.0012, and 0.0018 nm⁻¹. In the case of nanocrystals with the size of about 6 nm, the surface-to-volume ratio was about ∼ 0.9 nm⁻¹ that was three orders of magnitude larger than those of microcrystals [23]. We have reported that surface adsorbed ligands had large quenching effect on nanocrystals [23,24]. To reduce the influence of surface quenching, microcrystals are more suitable owing to their low surface-to-volume ratio. The aspect ratio of crystal rods is another crucial factor. Gao et al. reported that the aspect ratio of NaYF₄: Er³⁺,Yb³⁺ rods affected upconversion emission properties [26]. The aspect ratios were 2.1, 4.0, 2.4, and 2.7 for TmYb, 10TbTmYb, 20TbTmYb, 20TbYb samples, respectively. Since the as-prepared samples have the same shape, similar aspect and surface-to-volume ratio, the influence caused by crystal size and shape can be neglected.

 

Fig. 2 SEM images of as-prepared Tb³⁺,Tm³⁺,Yb³⁺ doped NaYF₄ microcrystals: (a) TmYb, (b) 10TbTmYb, (c) 20TbTmYb, and (d) 20TbYb.

Download Full Size | PDF

3.2. Emission properties

The samples showed intense visible and near-infrared emissions excited by a 976-nm laser, as shown in Fig. 3(a). The spectra were normalized according to the intensity of emission at 544 and 1220 nm in the visible and near-infrared region, respectively. In the TmYb sample, the emissions centered at 346, 362, 452, 476, 648, 690, and 803 nm are assigned to the radiative transitions of ¹I₆→³F₄, ¹D₂→³H₆, ¹D₂→³F₄, ¹G₄→³H₆, ¹G₄→³F₄, ³F_2,3→³H₆, and ³H₄→³H₆ of Tm³⁺ ions, respectively [27,28].

 

Fig. 3 (a) Emission spectra of as-prepared NaYF₄ phosphors with various dopant (sample TmYb, 10TbTmYb, 20TbTmYb, and 20TbYb) in the visible and near-infrared region excited by 976 nm laser. (b) Energy level diagram and possible upconversion processes among Tb³⁺, Tm³⁺, and Yb³⁺.

Download Full Size | PDF

In the Tb³⁺-Yb³⁺ co-doped sample, the emissions centered at 381, 417, 438, 490, 544, 586, and 621 nm are assigned to the radiative transitions ⁵D₃→⁷F_J (J=6, 5, and 4) and ⁵D₄→⁷F_J (J=6,5,4, and 3) of Tb³⁺ ions, respectively [29,30]. In the 20TbTmYb and 10TbTmYb samples, emissions generated from both Tm³⁺ and Tb³⁺ ions were observed. However, the intensities of emissions at 346, 360, and 452 nm from the ¹I₆ ¹D₂, and ¹G₄ levels of Tm³⁺ were greatly reduced compared with the TmYb sample. When the Tb³⁺ concentration increased to 20%, the intensity ratio of emission at 476 nm to that at 544 nm clearly showed a decreasing tendency. The different spectral contribution implies the existence of energy transfer between Tm³⁺ and Tb³⁺. The emissions centered at 1170, 1220, 1470, and 1639 nm in the near-infrared region are assigned to the transition of ³F_2,3→³F₄, ³H₅→³H₆, ³H₄→³F₄ and ³F₄→³H₆ of Tm³⁺, respectively. The emission band at 1550 nm is caused by the slight amount of Er³⁺ impurity in rare materials. The tail near 1100 nm in the Tb³⁺-doped sample is caused by the strong pump laser and the spontaneous emissions from Yb³⁺. The TmYb sample shows intense near-infrared emissions while they were greatly quenched in Tb³⁺-Tm³⁺-Yb³⁺ samples. It implies that the Tm³⁺ might transfer energy to Tb³⁺.

Figure 3(b) schematically depicts possible upconversion and energy transfer processes among Tb³⁺, Tm³⁺, and Yb³⁺. In this system, Yb³⁺ ions act as sensitizers to absorb energy from the 976-nm laser irradiation, then transfer to the activators, Tb³⁺ and Tm³⁺. Electrons on the ground state of Tm³⁺ are excited to the ³H₅ level by phonon-assisted energy transfer [31], then non-radiatively relaxed to the ³F₄ level. The electrons on the ³F₄ level are excited to the ³F_2,3 level via energy transfer from Yb³⁺, relaxed to the ³H₄ level and then pumped to the ¹G₄ level. Owing to the large energy mismatch between the ¹D₂ and ¹G₄ level of Tm³⁺ (about 3500 cm⁻¹) [32], the ¹D₂ level cannot be directly excited from the ¹G₄ level by ETU process. It is ascribed to cross-relaxation processes between Tm³⁺ pairs. Four possible cross-relaxation approaches were reported as ¹G₄+³H₄→¹D₂+³F₄, ¹G₄+³H₄→³F₄+¹D₂, ³F₂+³H₄→³H₆+¹D₂, and ³F₃+³F₃→³H₆+¹D₂ [28,33,34]. Population of the ³F_2,3 level is strongly quenched by multi-phonon relaxation process because of the small energy gap to the nearest low-lying level (about 1500 cm⁻¹) [32]. Thus, the latter two approaches related to the ³F_2,3 level cannot be taken into account [28]. We attribute the ³F_2,3 level is populated by the former two cross-relaxation processes. The ¹I₆ level is excited by energy transfer from Yb³⁺. Thus, emissions from the ¹I₆ (346 nm), ¹D₂ (360 and 452 nm), and ¹G₄ level (474 and 649 nm) are generated by a five-, four-, and three-photon upconversion process, respectively [27]. Cooperative sensitizing upconversion process by two Yb³⁺ ions is the main mechanism to populate the ⁵D₄ level of Tb³⁺. Through an excited state absorption or energy transfer from Yb³⁺, the ⁵D₃ level was further populated [30,35].

Some energy levels of Tm³⁺ and Tb³⁺ have similar positions, i.e. ¹I₆-⁵H₄, ¹D₂-⁵G₅, ¹G₄-⁵D₄, and ³F₄-⁷F₀. The energy transfer between Tm³⁺ and Tb³⁺ will occur effectively, which is confirmed from the spectral difference in Fig. 3.

Owing to the effective energy transfer from the ¹I₆ and ¹D₂ levels, emission intensity in the 10TbTmYb sample is very weak, and that in the 20TbTmYb sample is hardly to be observed, as shown in Fig. 3(a). The relative intensity of the emission generated from the ³F₄ level of Tm³⁺ decreased in the Tb³⁺-Tm³⁺-Yb³⁺ tri-doped samples, which indicates the energy transfer from Tm³⁺ to Tb³⁺. Because of the narrow energy gap to the nearest low-lying levels, nonradiative relaxation is the main depopulating process for the ⁷F₀₋₅ levels. There are no suitable levels in Tb³⁺ which match the ³F_2,3, and ³H₄ level of Tm³⁺. Thus, the decrements of relative intensity were not caused by resonant energy transfer, but by the cross-relaxation between Tb³⁺ and Tm³⁺. This will be discussed in the next subsection.

We attempted to clarify the energy transfer processes by changing concentration of acceptor Tb³⁺ ions. It is difficult to clearly demonstrate energy transfer processes between Tm³⁺ and Tb³⁺, because Tm³⁺ and Tb³⁺ both can be pumped by the ETU from Yb³⁺. As well known, population of electrons on excited states in upconversion processes is proportional to the power density of pump light [20]. If the power density is low, the population of electrons on high levels of Tm³⁺ and Tb³⁺ will be relatively low. In this case, the change in electronic population caused by energy transfer on certain levels will be more obvious.

Two approaches were utilized to adjust power density of pump laser: controlling spot area with fixed power and changing power with fixed spot area. First, we fixed power of 976-nm laser about 154 mW, and then changed the distance between sample and the laser. Figure 4 shows upconversion emission spectra measured with various power densities. Spectra of the TmYb sample and all Tb³⁺-doped samples were normalized according to the intensity of emission at 476 (¹G₄, Tm³⁺), and 544 nm (⁵D₄, Tb³⁺), respectively. Electronic population on the ¹D₂ and ¹I₆ level of Tm³⁺ and the ⁵D₃ level of Tb³⁺ became large when laser power density increased. The emission intensity is enhanced when the power density of pumping laser increased from 3 to 122 W/cm² as shown in Fig. 4(a) and 4(d). The emissions from Tm³⁺ and Tb³⁺ were observed in the 10TbTmYb and 20TbTmYb samples, as shown in Figs. 4(b) and 4(c). However, the intensity ratios of emission at 476 nm (Tm³⁺) to 544 nm (Tb³⁺) were dependent on power density of pump laser. The intensity ratio was used to illustrate the contribution of Tm³⁺ and Tb³⁺. When the power density was 3 W/cm², emissions of Tm³⁺ can be easily observed in the TmYb sample as shown in Fig. 4(a). However, the intensity of emission at 476 nm in the 20TbTmYb sample was significantly reduced, and only emissions from Tb³⁺ were observed under the same excitation conditions. It implies that the energy transfer from Tm³⁺ to Tb³⁺ is efficient in the case of low power density.

 

Fig. 4 Upconversion emission spectra of samples (a)TmYb, (b) 10TbTmYb, (c) 20TbTmYb, and (d) 20TbYb excited by 976-nm laser with various power density. The chromaticity diagrams of each sample were inset correspondingly. The black vertical arrows indicate the power density increasing from 3 to 122 W/cm².

Download Full Size | PDF

The chromaticity diagrams of samples are inset in Fig. 4. The intensity of blue emissions from the TmYb and 20TbYb sample increased with increasing the power density. Chromaticity coordinates of the TmYb sample shifts slightly within the blue region and the chromaticity coordinates of 20TbYb sample shifts towards white region. In the 10TbTmYb and 20TbTmYb samples, the contribution of blue emissions from Tm³⁺ was greatly enhanced with increasing power density. The coordinates of the 20TbTmYb sample will shift from the green towards white region and finally to the blue region with the power density increased from about 3 to 122 W/cm² as shown in Fig. 4(c). Under similar excitation conditions, however, the coordinates of the 10TbTmYb sample transit from the blue to cyan region. It illustrates that excessive Tb³⁺ could effectively quench the electronic populations of Tm³⁺ excited states.

Figure 5 shows the emissions in the near-infrared region depends on power density. These spectra were normalized according to the peak intensity of emission at 1220 nm. The intensities of emission from ³F_2,3, ³H₄, and ³F₄ in the TmYb sample were increased with increasing power density. No emission from Tb³⁺ was observed in the 20TbYb sample, except weak emission band from Er³⁺ impurity. In the discussions, we can ignore the influence from Er³⁺ impurity. Emission peak from Er³⁺ was observed in samples doped with Tb³⁺, as shown in Figs. 3(a) and 5. However, its concentration was too low to affect upconversion processes in Tm³⁺ and Tb³⁺. As shown in Fig. 4, no upconversion emissions from Er³⁺ were observed. It seems that only the ⁴I_13/2 level of Er³⁺ was populated. In the next section we propose energy transfer process from the ³H₄ level of Tm³⁺ to the ⁷F₀ level of Tb³⁺ which affects their upconversion processes. The ⁴I_13/2 level of Er³⁺ is higher than the ³H₄ level of Tm³⁺ and the ⁷F₀ level of Tb³⁺ which means energy transfer from Er³⁺ to Tm³⁺ and Tb³⁺ should be more sufficient than energy back transfer to Er³⁺. Thus, the tiny amount of Er³⁺ impurity did not seriously affect energy transfer process between Tm³⁺ and Tb³⁺. Although intensities were weak, emissions from the ³F_2,3, ³H₄, and ³F₄ level in the 10TbTmYb sample could be observed when the power density was high. In the 20TbTmYb sample, however, they cannot be detected even under high power density pumping. Only the transition ³H₅→³H₆ with weak intensity was recorded. It indicates the energy transfer from Tm³⁺ to Tb³⁺ was more sufficient in samples with higher Tb³⁺ concentration.

 

Fig. 5 Emission spectra of samples (a) TmYb, (b) 10TbTmYb, (c) 20TbTmYb, and (d) 20TbYb in near-infrared region excited by 976-nm laser. The black vertical arrows indicate the power density increasing from 3 to 122 W/cm².

Download Full Size | PDF

3.3. Energy transfer between Tm³⁺ and Tb³⁺

Some cross-relaxation approaches are proposed to explain the power density dependent color tuning feature of the as-prepared sample as follows,

Besides cross-relaxation between Tm³⁺ and Tb³⁺, self-quenching processes between Tm³⁺ pairs and Tb³⁺ pairs also affect electronic populations, which are described as follows,

Figure 6 schematically shows the cross-relaxation and laser power density dependent energy transfer processes. The existence of energy transfers between Tm³⁺ and Tb³⁺ is inferred according to the above subsection. The electrons on a certain level will be depopulated via several approaches: radiative transition and nonradiative relaxation to low-lying levels, being pumped to higher levels, cross-relaxation between ion pairs, and phonon-assisted energy transfer processes. The difference in spectral profiles can be used to investigate and gain insight into the dynamics of electronic populations of energy levels. If the emissions from an excited state cannot be observed, it implies that electrons are depopulated by nonradiative relaxation, cross-relexation, or energy transfer processes.

 

Fig. 6 (a) Energy level diagram, upconversion processes and possible energy transfer between Tb³⁺, Tm³⁺, and Yb³⁺. (b) Schematic illustration of energy transfer between Tm³⁺ and Tb³⁺ under low and high power density pumping.

Download Full Size | PDF

In Fig. 4, the TmYb and 20TbYb samples show the upconvesion emissions from Tm³⁺ and Tb³⁺ are independent of the power density of pump laser, whereas the 10TbTmYb and 20TbTmYb samples show only Tb³⁺ emissions when the pump power density is low. The simple speculation that all the elections on Tm³⁺ were depopulated via energy transfer to Tb³⁺ by ¹I₆→⁵D₃, ¹D₂→⁵D₃, and ¹G₄→⁵D₄, as reported in [36], cannot perfectly explain this phenomenon. This is because it is impossible to excite only high energy levels without populating low energy levels such as ³F_2,3, ³H₄, ³H₅, and ³F₄ of Tm³⁺, in upconverson processes. The reported energy transfer processes probably have no influence on the low lying levels to change their electronic population. It other words, it is possible to observe the emissions from these levels, for example, at 696, 803, and 1220 nm. This is inconsistent with the experimental results that no emissions from Tm³⁺ were observed when the pump power density was lower than 5 W/cm².

The only possibility is that the upconversion processes in Tm³⁺ were inhibited. For low pump power density, the population of the ³F₄ level was too low to excite the ³F_2,3 levels and other high energy levels. Energy transfer from ³F₄ (Tm³⁺) to ⁷F₀ (Tb³⁺) is the main process to reduce the population of ³F₄ [37]. It is well known that energy transfer probability depends on the distance between sensitizer and activator and the overlap of emission spectrum of sensitizer and absorption spectrum of activator [38]. Figure 7 shows the overlap of emission spectrum of the TmYb sample and absorption spectrum of TbYb sample in the near-infrared region (1600–2400 nm). The emission band centered at 1639 nm is assigned to the transition ³F₄→³H₆ of Tm³⁺. The absorption bands centered at 1667, 1723, 1932, and 2264 nm are assigned to the transition from the ⁷F₆ to ⁷F₀₋₃ level of Tb³⁺, respectively. The spectral overlap (1600–1800 nm) implies the possibility of energy transfer between the ³F₄ and ⁷F₀ levels. The cross-relaxation process between Tm³⁺ pairs (³H₄+³H₆→³F₄+³F₄) concentrates electrons from the low energy levels of Tm³⁺ on the ³F₄ level. According to the multiphonon relaxation, the electrons on the ⁷F₀ level will soon relax to the ground state because of the high multiphonon relaxation rate caused by the narrow energy gaps to the nearest low-lying levels. The energy transfer process from the ³F₄ to the ⁷F₀ levels was enhanced. As a result, the ³F₄ level was greatly depopulated.

 

Fig. 7 Emission spectrum in the 1600–1700 nm range in TmYb excited by 976-nm laser, and absorption spectrum of 20TbYb in 1600–2400 nm range.

Download Full Size | PDF

The electrons on the ⁵D₃ level of Tb³⁺ will be quenched to the ⁵D₄ level by cross-relaxation ➀ and self-quenching ⓑ. The electrons on the ⁵D₄ level will be quenched by cross-relaxation ➁ to the ⁷F₀ level, then fast relax to the ground state by multi-phonon relaxation processes. For Tm³⁺ ions, the ¹I₆, ¹D₂, and ¹G₄ levels are depopulated and transfer the energy to the ⁵H₄, ⁵G₅, and ⁵D₄ levels of Tb³⁺ by phonon-assisted energy transfer. The electrons on the ³F_2,3 and ³H₄ level are quenched to the ³F₄ level by the cross-relaxation process ➂ and ➃ and self-quenching ⓐ, then depopulated by the energy transfer process, ³F₄→⁷F₀.

Since the measurements were carried out by recording the scattered emission lights, the comparison of peak intensities among samples cannot give reliable information. The integrated intensity ratios were used to analyze the change of emissions from different energy levels. Figure 8(a) shows the ratios of emission at 476 nm relative to that of 544 nm with varying power density of pump laser.

 

Fig. 8 Dependence of integrated intensity ratio of (a) emission at 476 nm versus 544 nm, (b) Tm³⁺ versus Tb³⁺ emissions, (c) emissions from the ¹I₆ level versus those from the ¹G₄ level, (d) emissions from the ¹D₂ level versus those from the ¹G₄ level, (e) emissions from the ¹G₄ level versus those from the ³H₄ level, and (f) emissions from the ⁵D₃ level versus those from the ⁵D₄ level on power density of pump laser.

Download Full Size | PDF

The high ratio shows that the contribution of Tm³⁺ was larger in the 10TbTmYb sample. Increasing tendency of the ratios in both samples implies the contribution of Tm³⁺ increases with power density. The ratio of Tm³⁺ to Tb³⁺ emission in Fig. 8(b) confirms that the contribution of Tm³⁺ in 10TbTmYb is higher and it increases with power density in both samples. As power density became higher, more ions were pumped to excited states. If the population on the ground states becomes less, the cross-relaxation processes ➀ – ➃ between Tm³⁺ and Tb³⁺ will be insufficient. Thus, contribution of emissions from Tm³⁺ becomes higher. The excessive amount of Tb³⁺ in 20TbTmYb guarantees the numbers of ions on the ground state for the cross-relaxation and the sufficient energy transfer from Tm³⁺ to Tb³⁺. Most of the electronic populations on the Tm³⁺ levels are quenched by Tb³⁺. As a result, the ratio of Tm³⁺ to Tb³⁺ emission is low.

When the power density increases, the quenching effect from Tb³⁺ was inhibited, and the contribution of Tm³⁺ emission increased significantly. The quenching effect was investigated by the intensity ratios of emissions from different energy levels in Tm³⁺, as shown in Figs. 8(b) and 8(c). Similar to the above discussion, the ratio of ¹I₆ and ¹D₂ emissions to ¹G₄ emissions decreased as increasing the Tb³⁺ concentration from 0 to 20%. The energy transfer from the ¹I₆ and ¹D₂ levels of Tm³⁺ to the ⁵H₄ and ⁵G₅ levels of Tb³⁺ effectively quenched their populations. As the power density was increased, the upconversion process in Tm³⁺ was enhanced. The ratio of emissions from the ¹I₆ and ¹D₂ levels increased in the Tb³⁺-doped samples, but the slope values became lower. However, the ratios of emissions from the ¹G₄ level versus those from the ³H₄ level in Tb³⁺-doped samples are large. It implies the increment of emissions from the ¹G₄ level or the decrement of emissions from the ³H₄ level. The population on the ¹G₄ level will be quenched by energy transfer to ⁵D₄ level of Tb³⁺. Thus, the cross-relaxation process ➃ greatly reduces the population on the ³H₄ level.

Figure 8(f) shows the ⁵D₃ to ⁵D₄ emission intensity ratio. As power density increases, contribution of emissions from the ⁵D₃ level gradually increased. In the low power density case, the higher slope value in 20TbTmYb implies that the Tm³⁺ energy transfer contribute to the population of the ⁵D₃ level. When power density was high, the ratio was saturated in the 20TbYb but shows decreasing tendency in 20TbTmYb. Self-quenching ⓑ from Tb³⁺ occurs in both samples. Owing to the cross-relaxation processes ➀ and ➁, however, the population on the ⁵D₃ and ⁵D₄ level will decrease. The decreasing tendency shows that ⁵D₃ level was quenched more effectively.

The upconversion process can be confirmed by fitting required photon number based on the dependence of integrated emission intensity on the power of pump laser [20,39]. The number of required photons to populate a certain energy level can be obtained, according to the relation I_f = Pⁿ, where I_f is the emission intensity, P is the pump power, and n is the number of photons. The n values for the emissions at 345 (¹I₆→³F₄ in Tm³⁺), 452 (¹D₂→³F₄ in Tm³⁺), 476 (¹G₄→³H₆ in Tm³⁺), and 803 nm (³H₄→³H₆ in Tm³⁺) in TmYb were 4.3 ± 0.3, 3.9 ± 0.3, 2.9 ± 0.2, and 1.9 ± 0.1, respectively. It confirms these levels were populated by five-, four-, three-, and two-photon upconversion processes [40].

The n values of the emission at 381 (⁵D₃→⁷F₆ in Tb³⁺), and 544 nm (⁵D₄→⁷F₆ in Tb³⁺) in 20TbYb were about 2.7 ± 0.2 and 2.0 ± 0.1. It shows that the ⁵D₃ and ⁵D₄ level were populated by three- and two-photon upconversion processes. The n values for the Tm³⁺ emissions in the 10TbTmYb and 20TbTmYb samples were similar to those in TmYb sample, as shown in the inset values in Fig. 9. It implies that energy transfer back to the ¹I₆, ¹D₂, and ¹G₄ level of Tm³⁺ has little influence. The ³H₄ level could be populated by electrons relaxed from the ³F_2,3 level excited by cross-relation process ➁ involving three photons. Thus, the n value of 803 nm emission in the Tb³⁺-doped sample slightly increased. Moreover, the n values of Tb³⁺ emissions were increased with increasing the molar ratio of Tm³⁺ to Tb³⁺. It was 3.3 ± 0.3 and 3.4 ± 0.2 for emissions at 381 nm and 2.3 ± 0.2 and 3.2 ± 0.3 for emissions at 544 nm in 20TbTmYb and 10TbTmYb, respectively. It indicates that the ⁵D₃ level requires more than three photons to be excited and the ⁵D₄ level needs more than two photons to be excited. The energy transfers from Tm³⁺ to Tb³⁺ are confirmed.

 

Fig. 9 Log-Log plot of upconversion emission intensity versus excitation power density for the (a) ¹I₆→³F₄ (345 nm), (b) ¹D₂→³F₄ (452 nm), (c) ¹G₄→³H₆ (476 nm), and (d) ³H₄→³H₆ (803 nm) transition of Tm³⁺, the (e) ⁵D₃→⁷F₆ (381 nm) and (f) ⁵D₄→⁷F₆ (544 nm) transition of Tb³⁺ in TmYb, 10TbTmYb, 20TbTmYb, and 20TbYb samples.

Download Full Size | PDF

The fluorescence lifetimes were obtained by fitting the measured emission decay curves with double-exponential function. Averaged lifetimes (τ_av) were estimated using the following expression [41,42],

 

Fig. 10 Fluorescent lifetime of the emissions at (a) 345 nm, (b) 452 nm, (c) 476 nm, and (d) 803 nm, (e) 381 nm and (f) 544 nm versus excitation power density in TmYb, 10TbTmYb, 20TbTmYb, and 20TbYb samples.

Download Full Size | PDF

The analysis of fluorescence lifetime confirms the presence of energy transfer and cross-relaxation processes between Tb³⁺ and Tm³⁺, which are inferred according to the analysis in Fig. 8. The Tm³⁺-Tb³⁺-Yb³⁺ tri-doped samples were considered as doping Tb³⁺ into Tm³⁺-Yb³⁺ dual-doped sample and doping Tm³⁺ into Tb³⁺-Yb³⁺ dual-doped sample. Although complicated energy transfer processes occurred among the three types of lanthanide ions, we discuss the quenching effect of Tb³⁺ to the Tm³⁺-Yb³⁺ system and Tm³⁺ to the Tb³⁺-Yb³⁺ system, respectively.

Energy transfer efficiency from Tb³⁺ to Tm³⁺ and backward from Tm³⁺ to Tb³⁺ could be calculated based on the obtained fluorescent lifetime by,

Figure 11 shows that the estimated energy transfer efficiencies in 10TbTmYb and 20TbTmYb samples increase with increasing power density of pump laser. The energy transfer efficiency from the ³H₄ to ⁷F₀ level was around 70%. When the energy level position of Tm³⁺ enhanced from low to high (³H₄, ¹G₄, ¹D₂, and ¹I₆), the transfer efficiencies were reduced. The electrons on the ³H₄ level was sufficiently depopulated via cross-relaxation processes ➂ and ➃. Therefore, the upconversion processes were inhibited. The efficiencies from the ⁵D₃ and ⁵D₄ levels, which are approximately 20–30%, demonstrate the presence of cross-relaxation process ➀ and ➁. The average lifetimes of the ⁵D₄ and ³H₄ level are invariant around 3700 and 650 µs, respectively. The ratio of Tm³⁺ ions on the ground state should be relatively larger. Consequently, cross-relaxation processes ➀ and ➁ are effective, the electrons on the ⁵D₃ and ⁵D₄ level will be relaxed to the ⁷F₀ level, and then finally nonradiatively relaxed to the ground state. This process helps to increase the number of Tb³⁺ on the ground state, which will enhance the energy transfer and cross-relaxation processes from Tm³⁺ to Tb³⁺. Therefore, the energy transfer efficiency from Tm³⁺ to Tb³⁺ and Tb³⁺ to Tm³⁺ of all levels increase with increasing power density of pump laser.

 

Fig. 11 Energy transfer efficiency between Tm³⁺ and Tb³⁺ calculated from fluorescent lifetime of emissions generated from various energy levels in (a) 10TbTmYb and (b) 20TbTmYb sample.

Download Full Size | PDF

Quantum yield Φ is given by the radiative and nonradiative transition rate as,

Owing to the strong quenching effect caused by the cross-relaxation between Tm³⁺ and Tb³⁺, the radiative rate is determined by crystal field on dopant ions; the nonradiative transition rate is decided by phonon energy of host, temperature, and energy gap to the nearest low-lying level. Although they will be affected after inducing the third type of lanthanide ions, the strong influence of quenching effect caused by the cross-relaxation between Tm³⁺ and Tb³⁺ can dominate. The W_CR term is proportional to the energy transfer efficiency, which increases with increasing power density as shown in Fig. 11. Therefore, it is predicted that quantum yield in tri-doped sample will be much lower than that in dual-doped samples and will be reduced when the power density of pump laser becomes higher.

4. Conclusions

In conclusion, Tb³⁺,Tm³⁺,Yb³⁺ tri-doped NaYF₄ microcrystal rods were synthesized via a hydrothermal method. The as-prepared samples showed intense visible upconversion emissions. The energy transfer process between Tm³⁺ and Tb³⁺ depends on the power density of pump laser. Upconversion emissions of samples can be tuned by the laser power density dependent energy transfer process. When the power density was low, energy transfer from the ³F₄ (Tm³⁺) to ⁷F₀ (Tb³⁺) level was sufficient. As a result, upconversion processes in Tm³⁺ were inhibited. Because of the low population on the ³F₄ level, green color from Tb³⁺ was observed. When the power density became high, the energy transfer processes between low energy levels of Tm³⁺ and Tb³⁺ were insufficient, the blue upconversion emissions of Tm³⁺ were significantly enhanced. These phosphors with tunable visible upconversion emissions dependent on power density of pump laser would be utilized for novel laser power density sensors in the future.