Upconversion luminescence in CaSc2O4 doped with Er3+ and Yb3+

Upconversion luminescence in CaSc2O4 doped with Er3+ and Yb3+

Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎ Contents lists available at ScienceDirect Journal of Luminescence journal homepage: www.elsevier.com/locate...

2MB Sizes 11 Downloads 177 Views

Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin

Full Length Article

Upconversion luminescence in CaSc2O4 doped with Er3 þ and Yb3 þ Angela Ştefan a,b, Octavian Toma a,n, Şerban Georgescu a a b

National Institute for Laser, Plasma and Radiation Physics, 409 Atomiştilor Street, 077125 Măgurele-Ilfov, Romania University of Bucharest, Faculty of Physics, 405 Atomiştilor Street, 077125 Măgurele-Ilfov, Romania

art ic l e i nf o

a b s t r a c t

Article history: Received 25 February 2016 Received in revised form 19 April 2016 Accepted 20 April 2016

Upconversion luminescence in ceramic samples of CaSc2O4:Er:Yb is investigated. The main upconversion mechanisms, as well as the influence of various cross-relaxation processes on the luminescence emission are discussed. The optimum Yb3 þ concentration was found at 5%. The red/green intensity ratio is found to be mainly influenced by the energy back-transfer process ((2H11/2, 4S3/2) (Er3 þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3 þ )). A simple mathematical model based on rate equations is developed and used for the discussion of the experimental results. The maximum upconversion efficiency is 0.94% for the CaSc2O4:Er(1%):Yb(5%) sample. & 2016 Elsevier B.V. All rights reserved.

Keywords: CaSc2O4 Er3 þ Yb3 þ Upconversion

1. Introduction CaSc2O4 is considered a promising host for efficient upconversion/downconversion ([1] and references therein) due to its lowenergy phonons (540 cm  1 [2]), short distances between positions that can be occupied by the dopants and high solubility of ytterbium ions. Intense visible upconversion luminescence was obtained in CaSc2O4 codoped with Er3 þ and Yb3 þ [3], Tm3 þ and Yb3 þ [1], and with Ho3 þ and Yb3 þ [4,5]. Cooperative downconversion and near-infrared luminescence were observed in CaSc2O4 doped with Tm3 þ and Yb3 þ [6]. Efficient phosphors were obtained doping CaSc2O4 with Ce3 þ [7], Tb3 þ [8], and Eu3 þ [9]. In a series of recent papers, we adapted the Judd–Ofelt methods to rare-earth-doped ceramic samples of CaSc2O4 which scatter the transmitted light [10–12]. The absorption spectra were calibrated using either the measured lifetime of an emitting level (considered purely radiative [10–12] due to the large energy gap beneath this level), or using the line strength of a magnetic-dipole transition (practically independent of the host [11]). Thus, we succeeded to calculate the radiative lifetimes of the main emitting levels of Ho3 þ , Tm3 þ , and Er3 þ in CaSc2O4, allowing the estimation of the quantum efficiencies of these levels. Concerning the upconversion luminescence of CaSc2O4 doped with Er3 þ and Yb3 þ , there is only one report [3]: the CaSc2O4:Er:Yb nanopowders were prepared by the combustion method. The optimal concentrations of the dopants Yb3 þ and Er3 þ were 6.0 mol% and 1.0 mol%, respectively. n

Corresponding author. Tel.: þ 40 214574550. E-mail address: [email protected]flpr.ro (O. Toma).

In this paper, we perform a systematic analysis of the upconversion properties of ceramic CaSc2O4:Er:Yb phosphors synthesized by solid-state reaction. To our best knowledge, our analysis is the first to take into account the multiple sites that can be occupied by the doping ions; a re-distribution of the doping ions on these sites is put into evidence using the luminescence spectra. These phenomena are carefully handled by taking into account the entire luminescence spectrum (with contributions from all luminescent centers) when measuring the kinetics of the emitting levels. The change of the color of the Er3 þ visible emission, function of the Yb3 þ concentration, is illustrated with CIE diagrams. As rare-earth-doped CaSc2O4 is considered an efficient host for upconversion, we estimate its efficiency when doped with Er3 þ and co-doped with Yb3 þ ; as far as we know, such an estimation was not performed until now for any of the CaSc2O4 – based phosphors. A mathematical model is developed for the interpretation of the experimental results.

2. Experiment Three series of CaSc2O4:Er(x%):Yb(y%) ceramic samples were synthesized by solid state reaction: (i) x¼ 0.05, 0.1, 0.5, 1, and 2, y¼0; (ii) x ¼0.5, y¼2, 5, 8, and 10; (iii) x¼ 1, y¼2, 5, 8, and 10. Details concerning the synthesis can be found in Ref. [13]. The luminescence spectra in the UV–vis domain were measured with a Horiba Jobin-Yvon (1000M series II) monochromator equipped with an S-20 photomultiplier (EMI 9658B). The luminescence of the CaSc2O4:Er:Yb ceramic samples was excited in visible with a tunable system composed of a 150-W Xe-lamp monochromator illuminator Oriel APEX coupled to an Oriel

http://dx.doi.org/10.1016/j.jlumin.2016.04.042 0022-2313/& 2016 Elsevier B.V. All rights reserved.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

Cornerstone 260 monochromator (referred to as Oriel pump system). The upconversion luminescence was excited at 980 nm with a LIMO70-F200-DL980-FP-A laser diode. The signal from the detector was processed with a lock-in amplifier (SR830 from Stanford Research Systems). For decay measurements, the luminescence was excited with a Quantel Rainbow OPO. In the UV–vis spectral range, the luminescence was analyzed using the Horiba 1000 M monochromator equipped with a Hamamatsu H8259-02 photon counting head, while in the IR a 1-m Horiba Jobin-Yvon FHR 1000 monochromator with a S1 photomultiplier (EMI 9684B) was used. The kinetics curves were analyzed, for the UV–vis domain, with the SR430 Multi-Channel Scaler (MCS) from Stanford Research Systems; for the IR domain, the kinetics curves were recorded on a Tektronix 2024C oscilloscope. For upconversion efficiency measurements an integration sphere with four ports (IS 200-4, Thorlabs) was used. The IR radiation from the laser diode was focused on the ceramic sample placed at the opposite port of the integrating sphere. The emitted radiation was collected with a fiber and transported to the entrance slit of the 1-m Jarrell-Ash monochromator equipped with an S20 photomultiplier for the visible range and an S1 photomultiplier for the IR range. A calibrated lamp (ORIEL 63358) was used to correct the spectral sensitivity of the apparatus. All the measurements were performed at room temperature.

3. Results and discussion 3.1. Host properties The CaSc2O4 crystal has the calcium ferrite structure [14], space 3þ group Pnam, D16 positions 2h , with three cationic positions: two Sc with six-fold coordination and one Ca2 þ position with eight-fold coordination. All these positions have Cs point symmetry. The ionic radius of Sc3 þ is 0.745 Å while the ionic radius of Ca2 þ is 1.12 Å [15]. Small trivalent rare-earth ions such as Yb3 þ (0.868 Å in sixfold coordination) and Tm3 þ (0.88 Å, six-fold coordination) enter preferentially the Sc3 þ positions [2,16] while larger trivalent ions such as Ce3 þ (1.143 Å, eight-fold coordination [15]) and Eu3 þ (1.066 Å, eight-fold coordination [15]) enter the Ca2 þ position [7,9]. Er3 þ (ionic radius 0.89 Å, six-fold coordination or 1.003 Å, eight-fold coordination [15]) and Ho3 þ (ionic radii 0.901 Å in sixfold coordination and in eight-fold 1.015 Å [15]) could enter both Sc3 þ and Ca2 þ positions. In order that Er3 þ ions enter Ca2 þ positions, charge compensation is necessary as for CNGG (calcium niobium gallium garnet) [17] or CaWO4 [18]. The charge compensation could be done by random impurities or by vacancies. Our measurements concerning the kinetics of (2H11/2, 4S3/2) levels in CaSc2O4 put into evidence the multisite structure of this material. For example, in CaSc2O4:Er(0.1%) ceramics, our measurements have evidenced three lifetimes in the decay of 4S3/2 level at 10 K (55 ms, 190 ms, and 218 ms). Since a detailed analysis of the multisite structure of Er3 þ in CaSc2O4 is beyond the aim of this paper, we consider only the fact that monochromatic excitation as the second harmonic of the Nd: YAG laser or the emission lines of the Argon laser can favor one center in respect to the others. In upconversion experiments, the FWHM of the pump diode is  3– 4 nm (non-selective) and all the Er3 þ centers participate to the upconversion luminescence. Because the efficiency of the back transfer Er3 þ -Yb3 þ is estimated from the values of the Er3 þ lifetimes, special attention was paid to decay experiments in order to include the contribution of all Er3 þ centers. Peng et al. [3], using the combustion synthesis method to prepare CaSc2O4:Er:Yb, obtained optimal Er3 þ and Yb3 þ concentrations of 1% and, respectively, 6%; therefore, to enable a comparison with the

results in [3], we synthesized a series of CaSc2O4:Er(1%):Yb(y%) ceramic samples with y¼0, 2, 5, 8, and 10. Another series of samples, with 0% Yb3 þ concentration, was synthesized for the study of the Er3 þ cross-relaxation. Taking into account the strength of this cross-relaxation, we also synthesized a series with lower erbium concentration (0.5%) and Yb3 þ concentrations of 0, 2, 5, 8, and 10. 3.2. Upconversion mechanisms Under pumping at 980 nm, the ceramic CaSc2O4 samples doped with Er3 þ and Yb3 þ show bright green ((2H11/2, 4S3/2)4 I15/2, wavelength range 500–600 nm) and red (4F9/2-4I15/2, 625–725 nm) luminescence. The luminescence spectra of CaSc2O4: Er(0.5%):Yb(y%) samples (y¼0, 2, 5, 8, 10) for 124 mW incident pump power are given in Fig. 1. Besides, in Fig. 2, other two upconversion-pumped luminescence transitions are shown: 4G11/2-4I15/2 (UV, at  380 nm), 2H9/ 4 2- I15/2 (violet,  405 nm), for the CaSc2O4:Er(0.5%):Yb(5%) sample, for the same incident pump intensity, but these luminescence bands are much less intense than the green and red ones. For pumping at 980 nm, the energy levels 4F9/2, 4S3/2 and 2H11/2 are populated by two-photon processes, while 2H9/2 and 4G11/2 – by three-photon processes, as shown in Fig. 3, where the emission intensity of the various emitting levels vs. incident pump power is plotted in double logarithmic scale [19]. Since the slopes in Fig. 3 are close to 2 (respectively, 3), we can suppose that saturation effects are negligible in this pump power range (56–125 mW). A similar behavior was obtained for the other CaSc2O4:Er:Yb samples. The small difference between the slopes of the thermalized levels 2H11/2 and 4S3/2 (Δn ¼0.06) in Fig. 3 can be explained by a local increase of temperature due to the pump absorption. It can

Fig. 1. Visible upconversion spectra of CaSc2O4:Er(0.5%):Yb(y%) for y¼0, 2, 5, 8, 10. Incident pump power P¼ 124 mW.

Fig. 2. UV-violet upconversion spectrum for the CaSc2O4:Er(0.5%):Yb(5%) sample.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

be shown that, in the three-level approximation (2H11/2, 4S3/2 and 4 I15/2) [20], the local temperature increase of the sample, Δt, due to the laser irradiation, when the incident power increases from P1 to P2, is related to the difference between the slopes Δn, by the approximate relation [21]: Δt 

  kB T 21 Δn ln P 2 =P 1 ΔE

ð1Þ

where T1 is the absolute temperature of the sample corresponding to the incident power P1 and kB is the Boltzmann constant. For Δn ¼0.06, ln(P2/P1)¼0.684, T1 ¼300 K and ΔE ¼800 cm  1 [22], a temperature increase ΔtE 3 °C results, when the incident power increases from 56 mW up to 125 mW. This temperature variation is very small and does not influence the results of the luminescence measurements.

3

The slope for 4G11/2 luminescence is 2.99 while for 2H9/2 is 2.76; for both levels, three-photon processes are necessary. An examination of the energy level schemes of Er3 þ and Yb3 þ (Fig. 4) corroborated with the values of the slopes, suggests that the level 4 G11/2 is populated by the energy transfer process (5). Indeed, the slope of 4G11/2 (2.99) differs from the slope of thermalized levels (2H11/2, 4S3/2) – which is around 2 – with 0.99, i.e. one supplementary photon is necessary to promote the system from (2H11/2, 4 S3/2) to 4G11/2. Similarly, the slope of 4F9/2 is 1.84 which differs from the slope of 2H9/2 (2.76) by 0.92. Therefore, the level 2H9/2 is populated mainly by energy transfer from 4F9/2 (energy transfer process (4)). Since the gap between the Er3 þ levels 4G11/2 and 2H9/ 1 ), we expect that the multiphonon 2 is rather small (  1400 cm 4 transition from G11/2 could have a contribution to the population of 2H9/2 level [21,23]. In the absence of Yb3 þ , for pumping at 980 nm (transition 4 I15/2-4I11/2), both excited-state absorption (4I11/2-4F7/2) and energy-transfer upconversion – ETU ((4I11/2, 4I11/2)-(4F7/2, 4I15/2)) contribute to the population of 4F7/2 level and, by rapid multiphonon transition, to the populations of the green emitting levels (2H11/2, 4S3/2). In this case, the 4F9/2 level is populated mainly by multiphonon transitions from 4S3/2. In the presence of Yb3 þ , the main energy transfer processes from Yb3 þ to Er3 þ are (see Fig. 4): (1) (4I15/2 (Er3 þ ), 2F5/2 (Yb3 þ ))-(4I11/2 (Er3 þ ), 2F7/2 (Yb3 þ )), followed by multiphonon transition to 4I13/2 (Er3 þ ); (2) (4I11/2 (Er3þ ), 2F5/2 (Yb3þ ))-(4F7/2 (Er3þ ), 2F7/2 (Yb3 þ )) followed by rapid multiphonon transition to (2H11/2, 4S3/2) (Er3 þ ); (3) (4I13/2 (Er3þ ), 2F5/2 (Yb3 þ ))-(4F9/2 (Er3þ ), 2F7/2 (Yb3 þ )) (phononassisted transition); (4) (4F9/2 (Er3 þ ), 2F5/2 (Yb3 þ ))-(2H9/2 (Er3 þ ), 2F7/2 (Yb3 þ )); (5) ((2H11/2, 4S3/2) (Er3þ ), 2F5/2 (Yb3 þ ))-(2K13/2 (Er3þ ), 2F7/2 (Yb3þ )) followed by rapid multiphonon transitions to 4G9/2, 4G11/2, 2H9/2 (Er3 þ ).

Fig. 3. Double logarithmic plot of luminescence intensity vs. incident pump (980 nm) power for the CaSc2O4:Er(0.5%):Yb(5%) sample. Up triangles: 4F9/2-4I15/2; down triangles: 4S3/2-4I15/2; diamonds: 2H11/2-4I15/2; circles: 2H9/2-4I15/2, squares 4G11/2-4I15/2. (incident power 56–125 mW, no saturation).

Besides, there are two Er3 þ -Yb3 þ back-transfer processes: a. (4I11/2(Er3 þ ), 2F7/2(Yb3 þ ))-(4I15/2(Er3 þ ), 2F5/2(Yb3 þ )) and b. ((2H11/2, 4S3/2) (Er3 þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3þ )).

Fig. 4. Energy level scheme of Er3 þ and Yb3 þ . The main energy transfer processes between Er3 þ and Yb3 þ are shown. 1 to 5: energy transfer Yb3 þ -Er3 þ ; a, b: back transfer Er3 þ -Yb3 þ . CR1, CR2: cross-relaxation processes in the Er3 þ system.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

3.3. Upconversion luminescence Inside the Er3 þ system, the cross-relaxation processes involving the ground level 4I15/2 are shown with gray arrows in Fig. 4: CR1 and CR2 (((2H11/2, 4S3/2), 4I15/2)-(4I9/2, 4I13/2)) [24,25]. They cause a decrease (see Table 1) of the effective lifetime (area under normalized decay curve, measured with the wide-band experimental setup described below) of the green-emitting levels. For the samples co-doped with Yb3 þ , the intensities and the intensity ratios of green and red luminescence change with Yb3 þ concentration (see Fig. 1). The dependence of the total luminescence intensity, green light intensity and red light intensity on the Yb3 þ concentration is given in Fig. 5 for both CaSc2O4:Er:Yb series. We note that for both series, the optimum Yb3 þ concentration is 5%, close to results obtained previously [3]. No notable difference was observed between the maximum intensities corresponding to 0.5 at% and 1 at% Er3 þ concentrations. The dependence of red/ green ratio (RGR) on ytterbium concentration, for both CaSc2O4:Er: Yb series, for three incident pump powers (24 mW, 124 mW and 497 mW) is given in Fig. 6. RGR increases with Yb3 þ concentration, more rapidly for CaSc2O4:Er(0.5%):Yb series. The results are approximately the same for 24 mW and 124 mW incident pump powers. In this incident power range, the saturation effects are negligible (see the double logarithm dependence in Fig. 3). For higher power (497 mW), RGR is lower. Thus, for CaSc2O4:Er(0.5%): Yb(10%), and P ¼24 mW, RGR E23, while, for CaSc2O4:Er(1%):Yb (10%) and the same incident power, RGR E 10. At incident pump power P ¼497 mW, for the Er(0.5%):Yb(10%) sample, RGR E15, while for the Er(1%):Yb(10%) sample, RGR E5.5. This can be explained by a more rapid saturation of the population of 4I13/2 Table 1 Effective lifetimes of (2H11/2, 4S3/2) and 4F9/2 Er3 þ levels in CaSc2O4:Er ceramic samples. Er3 þ concentration (at%)

τeff(2H11/2, 4S3/2) (ms)

τeff(4F9/2) (ms)

0.05 0.1 0.5 1.0 2.0

119.0 110.5 48.2 33.3 20.4

64.4 63.0 60.3 53.1 45.1

Fig. 5. Intensity of the visible upconversion emission vs. Yb3 þ concentration for CaSc2O4:Er:Yb ceramic samples. Solid symbols: CaSc2O4:Er(0.5%):Yb samples; open symbols: CaSc2O4:Er(1%):Yb samples. Incident pump power: 124 mW. Squares: total emission; triangles: green ((2H11/2, 4S3/2)-4I15/2) emission; circles: red (4F9/2-4I15/2) emission.

(Er3 þ ) level involved in the population of the 4F9/2 level (process (3) in Fig. 4) [19,26,27] than that of the 4I11/2 (Er3 þ ) level involved in the population of the (2H11/2, 4S3/2) levels (process (2) in Fig. 4). The changes in the RGR, function of Yb3 þ concentration, are illustrated with the CIE diagrams given in Fig. 7. The color range of the emitted visible light is broader for the sample series doped with a smaller Er3 þ (0.5%) concentration. As discussed before, in order to include the contribution of all Er3 þ centers in CaSc2O4 to the (2H11/2, 4S3/2) or 4F9/2 decays, a wideband experimental setup should be used. The pumping light should be non-selective; the OPO emission (FWHM  3 nm) satisfies this condition. Besides, the emitted light should be detected with very wide band in order to include all the emission spectrum. Since the resolution of our monochromators (1 m length) is high even for the largest slits (2 mm), we put the monochromator's grating in zero order and placed before the entrance slit several filter combinations. For measuring the kinetics of (2H11/2, 4S3/2) Er3 þ levels, the transmission spectrum of the filters centered at 555 nm has FWHM¼44 nm. For excitation, OPO emission at 488 nm (Er3 þ transition 4I15/2-4F7/2) was selected. Keeping constant the Er3 þ concentration and increasing the 3þ Yb concentration, the kinetics of the (2H11/2, 4S3/2) Er3 þ levels becomes more rapid due to the back-transfer process ((2H11/2, 4S3/ 3þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3 þ )) (process (b) in 2) (Er Fig. 4). The dependence on the Yb3 þ concentration of the effective lifetime of (2H11/2, 4S3/2) for both series (CaSc2O4:Er(0.5%)):Yb and (CaSc2O4:Er(1%):Yb) is given in Fig. 8. For lower Yb3 þ concentrations (up to 5%) the effective lifetimes of the samples doped with 0.5% Er3 þ are longer; for higher Yb3 þ concentrations, the effective lifetimes of the samples doped with 1% Er3 þ become longer. For lower Yb3 þ concentrations, the kinetics of (2H11/2, 4S3/2) is dominated by the cross-relaxation inside the Er3 þ system; in this case, we expect longer lifetimes for the samples with lower Er3 þ concentrations. For higher Yb3 þ concentrations, the back-transfer process (b) dominates. The shape of the luminescence spectra changes with both Er3 þ and Yb3 þ concentrations. As an illustration, the luminescence spectra corresponding to the Er3 þ transition (2H11/2, 4S3/2)-4I15/2 for CaSc2O4:Er:Yb(0%) series are given in Fig. 9. The pumping is performed with the Oriel pump system, at 488 nm, with large slits (non-selective pumping). For easier comparison, all the spectra in

Fig. 6. Red/green intensity ratio (RGR) vs. Yb3 þ concentration for the CaSc2O4:Er: Yb ceramic samples. Solid symbols: CaSc2O4:Er(0.5%):Yb samples; open symbols CaSc2O4:Er(1%):Yb samples. Squares: P¼ 24 mW; circles: P ¼124 mW; triangles: P¼ 497 mW.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5

Fig. 7. CIE diagrams for CaSc2O4:Er(0.5%):Yb and CaSc2O4:Er(1%):Yb series. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

Fig. 10. Luminescence spectra of CaSc2O4:Er(0.5%):Yb(y%), transition (2H11/2, 4 S3/2)-4I15/2. Pumping at 488 nm (Oriel pump system, large slits). The Yb3 þ concentrations are shown. All the spectra have normalized areas.

Fig. 8. Effective lifetime of (2H11/2, 4S3/2) vs. Yb3 þ concentration. Solid symbols: CaSc2O4:Er(0.5%):Yb; open symbols: CaSc2O4:Er(1%):Yb.

the Er3 þ concentration (0.5%), the shape of the luminescence spectrum changes with the Yb3 þ concentration (Fig. 10). The modifications of the luminescence spectra with Er3 þ and Yb3 þ concentrations can suggest a re-distribution of the dopant ions on the available cationic sites. This could explain also the longer lifetimes (see Fig. 8) measured for CaSc2O4:Er(1%):Yb samples doped with 8 and 10% Yb3 þ . The efficiency of the back-transfer process (b) is given by

ηEr-Yb ¼ 1  τErYb =τEr

ð2Þ

levels ( H11/2, S3/2) in where τEr and τErYb are the lifetimes of Er absence and in presence of Yb3 þ . The dependence of the backtransfer (b) efficiency on Yb3 þ concentration is given in Fig. 11. This process is more efficient for the CaSc2O4:Er(0.5%):Yb samples. This is related to the RGR (Fig. 6) [27]. The back-transfer process (b) decreases the population of (2H11/2, 4S3/2) Er3 þ levels and increases the populations of both 4I13/2 (Er3 þ ) and 2F5/2 (Yb3 þ ), favoring the population of the 4F9/2 Er3 þ level (process (3) in Fig. 4). In fact, the RGR is higher for CaSc2O4:Er(0.5%):Yb samples (Fig. 6). Another Er3 þ -Yb3 þ back-transfer process is (4I11/2(Er3 þ ), 2F7/ 3þ ))-(4I15/2(Er3 þ ), 2F5/2(Yb3 þ )) (process (a) in Fig. 4). This 2(Yb process should be very efficient since the Er3 þ transition 4I11/ 4 3þ absorption 2F7/2-2F5/2. 2- I15/2 is resonant with the Yb Unfortunately, due to the superposition of the spectrum of the Er3 þ transition 4I11/2-4I15/2 with the spectrum of Yb3 þ 3þ

Fig. 9. Luminescence spectra of CaSc2O4:Er(x%):Yb(0%), Er3 þ transition (2H11/2, 4 S3/2)-4I15/2. Pumping at 488 nm (Oriel pump system, large slits). The Er3 þ concentrations are shown. All the spectra have normalized areas.

Fig. 9 have normalized area. An inspection of Fig. 9 shows clearly the changes in the luminescence spectra with the increase of the Er3 þ concentration: the intensity of the luminescence lines at 563.5 nm and 566 nm increases, while the intensity of the line at 543.5 nm decreases, the 551 nm line is split etc. Keeping constant

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

2

4

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

Fig. 11. The efficiency of the back-transfer process (b) vs. Yb3 þ concentration. Solid symbols: CaSc2O4:Er(0.5%):Yb; open symbols: CaSc2O4:Er(1%):Yb.

(2F5/2-2F7/2) transition, it is difficult to put into evidence the decrease of the lifetime of 4I11/2 level in presence of Yb3 þ . Also, to measure the decay of 4I11/2 on the transition 4I11/2-4I13/2 (at 2.8 mm) is beyond our present experimental possibilities. The 4F9/2 kinetics was excited at 640 nm with the Quantel OPO (non-selective excitation). In order to measure the contribution of all Er3 þ luminescent centers, the combination of filters used to select the red luminescence gives a transmission profile centered at 659 nm with FWHM ¼ 30 nm. The kinetics of 4F9/2 level changes slowly with both Er3 þ and 3þ Yb concentrations. For the CaSc2O4:Er series, the lifetime varies from 64 ms (sample doped with 0.05% Er3 þ ) down to 45 ms (2% Er3 þ ) – see Table 1. For the series CaSc2O4:Er(0.5%):Yb, the 4F9/2 lifetime decreases from 60 ms (0% Yb3 þ ) down to 47 ms (10% Yb3 þ ). Similarly, the lifetime decreases from 53 ms down to 50 ms, for CaSc2O4:Er(1%):Yb series. The dependence of the measured lifetime of 4F9/2 level on Yb3 þ concentration is given in Fig. 12. For low ytterbium concentrations, the lifetimes are larger for the samples doped with 0.5% Er3 þ ; for higher Yb3 þ concentrations, the lifetimes of samples doped with 1% Er3 þ become larger. The same behavior was observed for the lifetimes of (2H11/2, 4S3/2) Er3 þ levels (see Fig. 6). As far as we know, no cross-relaxation process was proposed to explain the reduction of the 4F9/2 lifetime with Er3 þ concentration. Since the energy levels of Er3 þ in CaSc2O4 are not measured, we used the energy levels scheme of Er3 þ in Sc2O3 (Er3 þ substitutes Sc3 þ ) [28] for an estimation of the resonances between the Er3 þ energy levels. The analysis of this energy levels scheme shows that a possible cross-relaxation process (4F9/2, 4I15/2)-(4I13/2, 4I11/2) has a deficit of 638 cm  1 in Sc2O3. Such deficit in CaSc2O4:Er is not known. The reduction of the 4F9/2 lifetime with increasing Yb3 þ concentration could be explained by a mechanism inverse to (3) (see Fig. 4): (4F9/2 (Er3 þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3 þ )).

Fig. 12. Measured effective lifetime of 4F9/2 vs. Yb3 þ concentration. Full symbols: CaSc2O4:Er(0.5%):Yb; open symbols: CaSc2O4:Er(1%):Yb.

CaSc2O4:Er:Yb is dN5 dt

¼

1 T 5 þ wCR N 0 þ W BTb n0



N 5 þW up2 N2 n1

¼  NT 44 þ β54 NT 55 þ W up3 N 1 n1   dN2 ¼  T12 þ W BTa n0 N 2 W up2 N 2 n1 þ W up1 N 0 n1 þ wCR N 0 N 5 þ β43 NT 44 þ β53 NT 55 dt   dN 1 ¼  NT 11 W up3 N1 n1 þ βT515 þ wCR N 0 þ W BTb n0 N 5 þ β41 NT 44 þ β21 NT 22 dt   dn1 ¼  τ11 þ W up1 N 0 n1 W up2 N 2 n1 W up3 N1 n1 þW BTa n0 N 2 þ W BTb n0 N 5 þRp n0 dt

dN 4 dt

ð3Þ 3þ

levels where N0, N1, N2, N4, and N5 are the populations of the Er 4 I15/2, 4I13/2, 4I11/2, 4F9/2, and (2H11/2, 4S3/2), and n0 and n1 are the populations of the Yb3 þ levels 2F7/2 and 2F5/2 and T1, T2, T4, T5 are the lifetimes of the respective Er3 þ levels; τ1 is the lifetime of Yb3 þ excited level 2F5/2. The population N3 of 4I9/2 was not considered due to the very rapid decay to 4I11/2. Also, the population of 4 F7/2 was omitted, due to the very rapid decay on (2H11/2, 4S3/2). wCR represents the cross-relaxation processes ((2H11/2, 4S3/2), 4I15/ 4 4 2)-( I9/2, I13/2) [24,25] inside the erbium system. Wup1, Wup2 and Wup3 represents the upconversion processes (1), (2) and respectively (3) (see Fig. 4). The back-transfer processes included in the rate equations are (a), represented by WBTa and (b), represented by WBTb. βij are the branching ratios for the transitions from the Er3 þ level i to the Er3 þ level j, including the nonradiative rates. The values of the radiative branching ratios can be found in Ref. [12]. In stationary regime ðdN i =dt ¼ 0; dn1 =dt ¼ 0Þ and for low pump intensities ðN i o o N 0 ¼ N t ; n1 o o n0 ¼ nt Þ, relatively simple expressions can be derived for stationary populations of the emitting levels. For the green emitting levels (2H11/2, 4S3/2) we obtained: N5 

W up1 W up2 T 2 T 5 N0 n21 ð1 þwCR T 5 N 0 þW BTb T 5 n0 Þð1 þW BTa T 2 n0 Þ

ð4Þ

Large upconversion (Wup1, Wup2), low cross-relaxation (wCR) and low back-transfer (WBTa, WBTb) rates favor green upconversion luminescence. We note that the slope in log-log plot (Fig. 3) for green luminescence is close to 2, involving n1 proportional to Rp. For the red emitting level 4F9/2 we have:

3.4. Rate equations model Neglecting upconversion on Er3 þ levels higher than 4F7/2, the rate equation system describing the population dynamics in the



N4 

" # β54 W 2up2 T 2 T 5 n1 W up1 T 4 N 0 n21 β21 W up3 T 1   þ 1 þ W BTa T 2 n0 ð1 þ wCR T 5 N 0 þ W BTb T 5 n0 Þ2 1 þ W up3 T 1 n1 1  β41

ð5Þ

where the first term in the square parenthesis is the contribution

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

of the decay from the upper levels (2H11/2, 4S3/2) and the second – the contribution of the upconversion process from 4I13/2, represented by Wup3. In usual pump conditions, for low erbium concentration and high ytterbium concentration, and low phonon energy of the host, the second term dominates. Then, we can further approximate Eq. (5): N4 

β21 W up1 W up3 T 1 T 4 N 0 n21

   ð1 þ W BTa T 2 n0 Þ 1 þ W up3 T 1 n1 1  β41

ð6Þ

The factor containing n1 in the denominator of Eq. (6) could explain the slope (1.84 o 2) obtained for the red luminescence in Fig. 3. The ratio N4/N5 (proportional to the RGR) can be calculated using Eqs. (4) and (6): N4 T 1 T 4 W up3 ð1 þwCR T 5 N 0 þW BTb T 5 n0 Þ  β21 1 þ W up3 T 1 n1 N5 T 2 T 5 W up2

ð7Þ

taking also into account that β41 o o 1 [12]. We note that, in the frame of our approximations, the back-transfer process (a), represented by WBTa, does not influence the RGR. Only the backtransfer process (b) influences RGR. A similar result was obtained for zirconia doped with Er3 þ and Yb3 þ [27]. According to Eq. (7), the ratio N4/N5 should increase with N0 (erbium concentration) and n0 (ytterbium concentration). An inspection of Fig. 6 shows indeed a monotonous increase of RGR with ytterbium concentration for both CaSc2O4:Er:Yb series. Concerning the influence of the cross-relaxation inside the Er3 þ system (wCR), RGR is larger for the sample with higher erbium concentration (1%) only for 2% Yb3 þ . For higher Yb3 þ concentrations, RGR is larger for the samples with 0.5% Er3 þ . In fact, the RGR follows the concentration dependence of 1=T 05 ¼ 1=T 5 þ wCR N 0 þ W BTb n0 ; T 05 being the effective lifetime of (2H11/2, 4S3/2) Er3 þ levels in the rate equation picture (see Fig. 8). 3.5. Emission efficiency For upconversion efficiency measurements an integrating sphere with four ports (IS 200-4, Thorlabs) was used. The IR radiation from the laser diode was focused on the ceramic sample (a 12 mm diameter pellet) placed at the opposite port of the integrating sphere. In comparison with powder samples, the use of ceramic samples greatly simplifies the experiments. An undoped CaSc2O4 sample was used for the estimation of the incident pump power: its spectrum (Iu) was recorded in the spectral range 950–1025 nm that contains the laser diode's emission spectrum. The absorbed pump power was found by subtracting the integral of the spectrum of the doped sample (Id – recorded in the same spectral range) from the integral of the spectrum of the undoped sample. The efficiency of the upconversion emission was calculated as the ratio of the luminescence integral intensity and the absorbed pump power: 21025 nm 3 750 nm 1025 nm Z Z Z       4 ð8Þ ηeff ¼ I λ dλ= I u λ dλ  I d λ dλ5 500 nm

950 nm

950 nm

where the integral at the numerator is the area of the visible emission spectrum including green and red emission bands. The maximum value of the upconversion efficiency was obtained for a CaSc2O4:Er(1%):Yb(5%) ceramic sample: ηef f ¼ 0:94 % for an incident power density of approximately 80 W/cm2. This value is close to the upconversion efficiency of Y2O3:Er:Yb [29]. The fraction of absorbed pump light is 44%. We checked our experimental setup measuring the upconversion efficiency of the commercial NaYF4:Er:Yb (PTIR 550/F). Prior to measurements, the

7

NaYF4:Er:Yb powder was pressed into a 12 mm diameter pellet. We obtained 3.5% upconversion efficiency. The quantum efficiency of the upconversion emission (the number of emitted photons per absorbed photon) can be calculated as 21025 nm 3 750 nm 1025 nm Z Z Z       4 ð9Þ ηqe ¼ λI λ dλ= λI u λ dλ  λ I d λ dλ 5 500 nm

950 nm

950 nm

In this case, we obtained ηqe ¼ 0:65%.

4. Conclusion Ceramic samples of CaSc2O4:Er:Yb with various doping concentrations were synthesized by solid-state reaction. The changes in the luminescence spectra when Er3 þ and Yb3 þ concentrations change indicate a re-distribution of the doping ions on the three different cationic positions. The upconversion mechanisms contributing to the population of Er3 þ emitting levels were discussed. For incident pump powers lower than 124 mW, the saturation effects are negligible. For both CaSc2O4:Er:Yb series, the maximal values of efficiency were obtained for 5% Yb3 þ concentration. The red/green intensities ratio presents a steeper increase with the Yb3 þ concentration for the sample series doped with 0.5% Er3 þ due to the strong back-transfer (Er3 þ -Yb3 þ ) process ((2H11/ 4 3þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3 þ )), more 2, S3/2) (Er efficient for the sample series doped with 0.5% Er3 þ . To illustrate the changes of the red/green ratio with Yb3 þ concentration, the CIE diagrams are given. The decays of (2H11/2, 4S3/2) and 4F9/2 Er3 þ levels were measured with a wide-band detection setup in order to take into account the contribution of all luminescent centers. Their lifetimes decrease with both Er3 þ and Yb3 þ concentrations. The rate equations model describes qualitatively the experimental results. According to the rate equation model, only the back-transfer process ((2H11/2, 4S3/2) (Er3 þ ), 2F7/2 (Yb3 þ ))-(4I13/2 (Er3 þ ), 2F5/2 (Yb3 þ )) influences the red/green intensities ratio; the back-transfer process (4I11/2 (Er3 þ ), 2F7/2 (Yb3 þ ))-(4I15/2 (Er3 þ ), 2 F5/2 (Yb3 þ )) has no notable influence. The estimated upconversion efficiency is 0.94% for an incident power density of 80 W/cm2 for a CaSc2O4:Er(1%):Yb(5%); the corresponding quantum efficiency is 0.65%.

Acknowledgment This work was supported by the Romanian Executive Agency for Higher Education, Research, Development and Innovation Funding (UEFISCDI), in the frame of the Project IDEI 82/06.10.2011.

References [1] J. Li, J. Zhang, Z. Hao, X. Zhang, J. Zhao, Y. Luo, J. Appl. Phys. 113 (2013) 223507. [2] J. Li, J. Zhang, Z. Hao, X. Zhang, J. Zhao, Y. Luo, Appl. Phys. Lett. 101 (2012) 121905. [3] W. Peng, S. Zou, G. Liu, Q. Xiao, J. Meng, R. Zhang, J. Rare Earths 29 (2011) 330. [4] J. Li, J. Zhang, Z. Hao, X. Zhang, J. Zhao, Y. Luo, Chem. Phys. Chem. 14 (2013) 4114. [5] J. Li, J. Zhang, Z. Hao, L. Chen, X. Zhang, Y. Luo, Chem. Phys. Chem. 16 (2015) 1366. [6] J. Li, J. Zhang, X. Zhang, Z. Hao, Y. Luo, J. Alloy. Compd. 583 (2014) 96. [7] Y. Shimomura, T. Kurushima, N. Kijima, J. Electrochem. Soc. 154 (2007) J234. [8] Z.D. Hao, J.H. Zhang, X. Zhang, S.Z. Lü, X.J. Wang, J. Electrochem. Soc. 156 (3) (2009) H193. [9] Z. Hao, J. Zhang, X. Zhang, X. Wang, Opt. Mater. 33 (2011) 355. [10] S. Georgescu, A. Ştefan, O. Toma, A.M. Voiculescu, J. Lumin. 162 (2015) 174.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i

8

A. Ştefan et al. / Journal of Luminescence ∎ (∎∎∎∎) ∎∎∎–∎∎∎

[11] S. Georgescu, A. Ştefan, A.M. Voiculescu, O. Toma, J. Lumin. 166 (2015) 130. [12] S. Georgescu, A. Ştefan, O. Toma, J. Lumin. 167 (2015) 186. [13] S. Georgescu, A. Ştefan, A.M. Voiculescu, O. Toma, C. Matei, R. Birjega, J. Lumin. 154 (2014) 142. [14] J.R. Carter, R.S. Feigelson, J. Am. Ceram. Soc. 47 (1964) 141. [15] R.D. Shannon, Acta Crystallogr. A 32 (1976) 751. [16] R. Gaume, B. Viana, J. Derouet, D. Vivien, Opt. Mater. 22 (2003) 107. [17] Yu.K. Voron'ko, A.V. Malov, K.N. Nishchev, P.A. Ryabochkina, A.A. Sobol, S. N. Ushakov, Opt. Spectrosc. 102 (2007) 722. [18] F. Cornacchia, A. Toncelli, M. Tonelli, E. Favilla, K.A. Subbotin, V.A. Smirnov, D. A. Lis, E.V. Zharikov, J. Appl. Phys. 101 (2007) 123113. [19] M. Pollnau, D.R. Gamelin, S.R. Lüthi, H.U. Güdel, M.P. Hehlen, Phys. Rev. B 61 (2000) 3337. [20] S.A. Wade, S.F. Collins, G.W. Baxter, J. Appl. Phys. 94 (2003) 4743. [21] S. Georgescu, A.M. Voiculescu, C. Matei, A.G. Stefan, O. Toma, Phys. B 413 (2013) 55.

[22] W. Ryba-Romanowski, G. Dominiak-Dzik, P. Solarz, Mol. Phys. 101 (2003) 1067. [23] R. Yan, Y. Li, Adv. Funct. Mater. 15 (2005) 763. [24] L.G. van Uitert, L.F. Johnson, J. Chem. Phys. 44 (1966) 3514. [25] R.V. Bakaradze, G.M. Zverev, G.Ya Kolodnii, G.P. Kuznetsova, A.M. Onischenko, Solid State Phys. 9 (1967) 939. [26] G.Y. Chen, Y.G. Zhang, G. Somesfalean, Z.G. Zhang, Q. Sun, F.P. Wang, Appl. Phys. Lett. 89 (2006) 163105. [27] G. Chen, G. Somesfalean, Y. Liu, Z. Zhang, Q. Sun, F. Wang, Phys. Rev. 75 (2007) 195204. [28] A. Lupei, V. Lupei, C. Gheorghe, A. Ikesue, J. Lumin. 128 (2008) 918. [29] R.H. Page, K.I. Schaffers, P.A. Waide, J.B. Tassano, S.A. Payne, W.F. Krupke, W. K. Bischel, J. Opt. Soc. Am. 15 (1998) 996.

Please cite this article as: A. Ştefan, et al., J. Lumin. (2016), http://dx.doi.org/10.1016/j.jlumin.2016.04.042i