Temperature dependence and temporal dynamics of Mn 2+ upconversion luminescence sensitized by Yb 3+ in codoped LaMgAl 11 O 19

A detailed spectroscopic study of the upconversion properties in Mn 2+ -Yb 3+ codoped LaMgAl 11 O 19 is presented. Pulsed and continuous-wave infrared excitation in the 2 F 7 / 2 → 2 F 5 / 2 Yb 3+ absorption peaks induces broad Mn 2+ green emission at 19450 cm −1 , which is assigned to the 4 T 1 → 6 A 1 transition in tetrahedral Mn 2+ and sharp peaks associated with Yb 3+ -pairs luminescence. Both emissions have very different temporal evolution and can be separated by time-resolved spectroscopy. Among the different concentrations under investigation, the 2%Mn 2+ -5%Yb 3+ codoped system presents the highest upconversion efﬁciency. The corresponding emission remains visible to the naked eye up to 650 K. The time dependence of the Mn 2+ luminescence upon Yb 3+ excitation shows distinct behaviors for different doping concentrations. The temporal evolution of the intensity for the diluted system doped with 2%Mn 2+ and 5%Yb 3+ together with the pure manganese compound doped with 1%Yb 3+ , as well as the temperature dependence of the upconversion emission intensity and lifetime are relevant to identify the underlying upconversion mechanisms. We show that the main processes responsible for upconversion in this doubly transition-metal rare-earth doped oxide are both ground-state absorption (cid:1) GSA (cid:2) /excited-state absorption and GSA/energy-transfer upconversion. An analysis of these processes yielding highly efﬁcient luminescence is discussed on the basis of crystal structure and dopants


I. INTRODUCTION
2][3] The most relevant UC mechanisms are ground-state absorption/excited-state absorption ͑GSA/ ESA͒, energy-transfer upconversion ͑GSA/ETU͒, cooperative luminescence, cooperative sensitization and photon avalanche processes. 4,5The majority of UC studies investigated up to date have involved combinations of rare-earth ͑RE͒ ions.In particular, Yb 3+ has been extensively used as a sensitizer due to the relatively high oscillator strength of the 2 F 7/2 → 2 F 5/2 transition.This transition is located in the near infrared ͑NIR͒, just in the range of inexpensive laser diodes and is resonant with the energy levels of some other trivalent lanthanide ions.Opposite to f-f transitions in RE ions, d-d transitions of transition-metal ͑TM͒ ions are much more sensitive to the local environment and therefore, associated UC properties can be tuned by applying hydrostatic pressure, 6 and eventually by changing the crystal composition ͑internal or chemical pressure͒.In this sense, the combination of lanthanide and TM ions extends the UC luminescent tuning capability through changes in both energy resonances and emitting state energies.
The first demonstration of UC luminescence in TM-RE mixed systems was observed in Yb 3+ -doped CsMnCl 3 and RbMnCl 3 , and in Cr 3+ -Yb 3+ codoped Y 3 Ga 5 O 12 , at low temperatures. 7,8A GSA/ESA process between Mn 2+ -Yb 3+ dimer states was established as the main mechanism involved in the red UC luminescence in Yb 3+ -doped RbMnCl 3 single crystal. 9This host incorporates trivalent impurities as Mn 2+ -Yb 3+ pairs with charge compensation vacancies.The use of manganese hosts ensures clustering since an Yb 3+ ion has always some Mn 2+ ion as a near neighbor.The same model was proposed to explain both Mn 2+ and Yb 3+ pairs UC emission in Yb 3+ -doped CsMnBr 3 single crystals. 10Evidence of GSA/ETU between Mn 2+ -Yb 3+ dimers was found in Yb 3+ -doped CsMnCl 3 . 11All these compounds show UC luminescence only at low temperature.It is well known that TM-concentrated compounds exhibit luminescence quenching as temperature increases.This phenomenon was explained by thermally activated excitation migration and subsequent transfer to nonradiative traps. 12Therefore, the temperature-induced luminescence quenching is inherent to concentrated manganese systems and occurs independently of the excitation mechanism via NIR or directly into Mn 2+ in the visible region.
Another difficulty in the UC luminescence investigation involves finding suitable host lattices able to accommodate both Mn 2+ and Yb 3+ at well-defined sites without charge compensation.LaMgAl 11 O 19 ͑LMA͒ meets all these requirements.LMA crystallizes in the magnetoplumbite-type structure ͑hexagonal, P6 3 / mmc space group͒. 13It consists of spinel blocks separated by an intermediate layer containing three oxygen ions, one lanthanum, and one aluminum per spinel block.Mg 2+ ions are accommodated into the spinel blocks.LMA doped with Mn 2+ is a well-known green phosphor, which can be excited by vacuum-ultraviolet radiation and used in plasma display panels. 14In Mn 2+ -Yb 3+ codoped LMA, Mn 2+ ions occupy the tetrahedral Mg 2+ sites whereas Yb 3+ replaces La 3+ .Recently, we have shown intense green UC luminescence in LMA codoped with 1%Mn 2+ -1%Yb 3+ upon Yb 3+ 2 F 7/2 → 2 F 5/2 excitation, which persists up to 500 K. 15 Here, we present an extended spectroscopic investigation to clarify the effects of Mn 2+ and Yb 3+ concentrations and temperature on UC luminescence and lifetime.Experimental results are reported in order to clarify the UC mechanisms in Mn 2+ -Yb 3+ codoped LMA yielding Mn 2+ luminescence after Yb 3+ excitation and determine optimum doping concentrations to achieve the highest UC efficiency.We have selected three different Mn 2+ and Yb 3+ doping levels in order to cover all the aspects of the UC processes; LMA: 1%Mn 2+ , 1%Yb 3+ in which UC luminescence at high temperature was observed for the first time, LMA: 2%Mn 2+ , 5%Yb 3+ in which twice the Mn 2+ concentration is used for Yb 3+ in order to increase the UC efficiency and pure manganese lattice doped with 1%Yb 3+ in order to ensure that every Yb 3+ ion has always one Mn 2+ ion as a near neighbor.

II. EXPERIMENTAL SECTION
Microcrystalline powders of LMA codoped with different nominal molar concentrations, 1%Mn 2+ -1%Yb 3+ and 2%Mn 2+ -5%Yb 3+ as well as 1%Yb 3+ -doped LaMnAl 11 O 19 ͑LMnA͒ were prepared by the precipitation method described elsewhere. 16Briefly, stoichiometric amounts of , and Yb͑NO 3 ͒ 3 •5H 2 O were dissolved in distilled water under stirring.Then aqueous NH 3 was added until a pH of 8.5 was reached and precipitation as hydroxides occurred.The hydroxides were centrifuged, washed with distilled water, and heated at 90 °C for 20 h.A final treatment at 700 °C for 2 h and 1500 °C for 1 h was carried out.
X-ray diffraction ͑XRD͒ measurements were carried out in a Thermo ARL X´TRA powder diffractometer, operating in the Bragg-Brentano geometry and equipped with a Cuanode x-ray source ͑K␣, = 0.15418 nm͒, using a Peltier Si͑Li͒ cooled solid-state detector.The powder patterns were collected in the 15°-85°2 range at a scan rate of 0.1°/ min and a time of exposure of 10 s.Polycrystalline samples were ground in a mortar and then put in a low-background sample holder for data collection.Phase identification was performed with the PDF-4+ 2007 database provided by the International Centre for Diffraction Data.Rietveld structural refinement was further performed on all samples using MAUD program. 17n 2+ photoluminescence and excitation spectra were measured upon excitation with a Xe lamp.Yb 3+ emission, and Mn 2+ and Yb 3+ -pairs UC luminescence spectra were obtained by exciting with a continuous-wave laser diode emitting at 975 nm.The emission was detected by a Hamamatsu ͑R928͒ photomultiplier and all the samples were measured under identical conditions and detection geometry.An optical parametric oscillator system ͑Vibrant II, OPOTEK͒ was used for lifetime and time-resolved spectroscopy experiments upon excitation with 10 ns laser pulses at selected wavelengths.Two different setups were used for light intensity detection: a TRIAX monochromator ͑TRIAX 320͒ equipped with an intensified CCD ͑HORIBA iCCD-3553͒ and a single monochromator ͑CHROMEX 500IS/SM͒ with a Hamamatsu ͑R928͒ photomultiplier or an extended IR ͑R7102͒ photomultiplier together with a multichannel scaler ͑Stanford Research SR-430͒.Temperature-dependent measurements in the range 15-300 K were carried out in a closed-cycle helium cryostat ͑Air Products CS202E͒ whereas hightemperature experiments ͑300-650 K͒ were performed with a Leitz microscope heating stage.All spectra were corrected for instrumental response and therefore given in photons versus wavenumber. 18

III. RESULTS AND DISCUSSION
A. Crystal structure XRD patterns of 2%Mn 2+ -5%Yb 3+ codoped LMA and 1%Yb 3+ -doped LMnA are shown in Fig. 1.Both patterns indicate a fairly good crystallization of synthesized powders in the PbFe 12 O 19 magnetoplumbite structure ͑P6 3 / mmc space group͒. 19For the structural refinement the "vacancy model" published by Iyi et al. 20 on a lanthanum hexaaluminate compound was exploited.
The Rietveld refinement of the LMA: 2%Mn 2+ , 5%Yb 3+ XRD pattern is consistent with the magnetoplumbite structure with lattice parameters a = 5.5602͑5͒ Å and c = 22.055͑4͒ Å but also reveals small traces ͑4%, volume fraction͒ of Al 2 O 3 ͑R-3c space group͒.The divalent ion occupies exclusively the 4f site ͑0.5 occupancy factor͒ together with an Al 3+ ion and both ions are tetrahedrally coordinated by oxygens with a C 3v point symmetry. 13For the LMnA powder containing Mn 2+ ion instead of Mg 2+ we obtain similar results; 97% of LaMnAl 11 O 19 magnetoplumbite structure and small traces of Al 2 O 3 .Mn 2+ ions also occupy only the 4f site ͑0.5 occupancy factor͒. 21 FIG. 1. ͑Color online͒ XRD patterns of ͑a͒ LMA: 2%Mn 2+ , 5%Yb 3+ and ͑b͒ LMnA: 1%Yb 3+ .Colored dots show the experimental data and black lines the Rietveld refinements.The final R w statistical indexes for the refinements are 13.8% for LMA: 2%Mn 2+ -5%Yb 3+ and 12.8% for LMnA: 1%Yb 3+ .creased cell parameters observed for LMnA in comparison with LMA reflect the bigger ionic radius of Mn 2+ with respect to Mg 2+ .The unit cells of the LMA and LMnA crystal structures are shown in Fig. 2.There are two different sites for the lanthanides in both LMA and LMnA phases.In the LMA sample the principal site ͑0.49occupancy͒ has D 3h symmetry and a 12-fold coordination by oxygens whereas the satellite La site ͑0.115 occupancy͒ has a lower point symmetry ͑C 2v ͒. 13 In the case of the LMnA compound, the satellite La site has the same point symmetry as in LMA, but with lower occupancy factor ͑0.05͒, whereas the principal La site is slightly shifted from the theoretical position ͑0.667, 0.333, and 0.25͒, with concomitant change in the point symmetry ͑from D 3h to C s ͒.In addition while the oxygen atoms show the same crystallographic sites occupation for both phases, light differences in the sublattice of Al 3+ ions are detected.In particular, the sites with Wyckoff notation 12k, 4f, 2a, and 4e are populated with almost the same occupancy by Al 3+ ions in both LMA and LMnA phases but only in the case of the former compound, one more 12k site is available for these ions ͑occupancy 0.0556͒.From our structural refinement the following distances are obtained: La͑1͒-Mg = 5.844͑1͒ Å and La͑2͒-Mg= 5.728͑3͒ Å for LMA and La͑1͒-Mn= 5.795͑2͒ Å and La͑2͒-Mn= 5.724͑4͒ Å for LMnA.The shortest La-La distance is 11.337͑5͒ Å and 11.324͑4͒ Å for LMA and LMnA, respectively.These distances are relevant for Mn 2+ -Yb 3+ interaction and the subsequent energy-transfer process.

B. Concentration dependence
Figure 3 compares the RT luminescence spectra of codoped LMA for different Mn 2+ and Yb 3+ concentrations upon visible and IR excitation.The green broad band emission is assigned to the Mn 2+ 4 T 1 → 6 A 1 transition in tetrahedral coordination and the high-energy peaks observed in the UC spectra are due to Yb 3+ -Yb 3+ cooperative luminescence. 15No shift in energy or change in bandwidth is observed comparing the Mn 2+ luminescence after direct and UC excitation.Considering the Mn 2+ sensitivity to surrounding, this indicates that both direct and UC luminescence are originated from Mn 2+ ions having the same environment, what is an indication of the homogenous distribution of Mn 2+ and Yb 3+ impurities in LMA.The green Mn 2+ emission in LMA: 2%Mn 2+ , 5%Yb 3+ is centered at 19430 cm −1 ͑515 nm͒ while the same band appears about 50 cm −1 toward higher energy for LMnA: 1%Yb 3+ according to expectations due to the smaller ionic radius of Mg 2+ compared to Mn 2+ .
It is worth noting that Mn 2+ emission is obtained for all doping concentrations upon excitation in both 23580 cm −1 ͑424 nm͒ and 10250 cm −1 ͑976 nm͒ bands even above RT.For direct Mn 2+ excitation ͓Fig.3͑a͔͒, the RT luminescence intensity ratio of the pure manganese compound ͑LMnA͒ to the Mn 2+ -doped LMA was found to be approximately 1:5 while the intensity for 1%Mn 2+ and 2%Mn 2+ compounds is roughly the same.This drastic reduction in emission intensity is ascribed to thermally activated energy migration and partial trapping into nonluminescent impurities as commonly occurs in concentrated manganese systems.Considering crystalline quality and emission efficiency, LMA codoped with 2%Mn 2+ and 5%Yb 3+ doping levels provides an optimum efficient UC system.Figure 3͑b͒ shows that UC intensity in 2%Mn 2+ -5%Yb 3+ codoped LMA is about an order of magnitude higher than the UC in 1%Mn 2+ -1%Yb 3+ codoped LMA.An increase in Mn 2+ concentration beyond 2% leads to a progressive UC decrease.3. ͑Color online͒ RT luminescence of Mn 2+ -Yb 3+ codoped LMA upon excitation at ͑a͒ 23580 cm −1 and ͑b͒ 10250 cm −1 for the following concentrations: 1%Mn 2+ -1%Yb 3+ ͑black-dotted line͒, 2%Mn 2+ -5%Yb 3+ ͑red-thick line͒ and pure manganese doped with 1%Yb 3+ ͑blue-thin line͒.‫ء‬ is a laser artifact appearing at twice the excitation frequency.Measurements were carried out under the same experimental conditions.

C. Temporal evolution
The temporal evolution of the RT Mn 2+ 4 T 1 → 6 A 1 green luminescence was recorded after direct Mn 2+ excitation into the 4 A 1 , 4 E levels at 23580 cm −1 ͑424 nm͒ and IR excitation into the 2 F 7/2 → 2 F 5/2 Yb 3+ transition at 10205 cm −1 ͑980 nm͒ with 10 ns short pulses for both LMA: 2%Mn 2+ , 5%Yb 3+ and LMnA: 1%Yb 3+ samples ͑Fig.4͒.The timedependence intensity of the Mn 2+ emission after visible excitation shows a single exponential decay with lifetimes of = 6.2 ms for LMA: 2%Mn 2+ , 5%Yb 3+ , and = 3.8 ms for LMnA: 1%Yb 3+ ͓Figs.4͑a͒ and 4͑b͔͒.However, the time dependence of the UC Mn 2+ luminescence after IR excitation presents a different behavior in both systems.A single exponential decay ͑ = 5.2 ms͒ is observed in LMA: 2%Mn 2+ , 5%Yb 3+ ͓Fig.4͑c͔͒.This lifetime is significantly faster than the one obtained by direct Mn 2+ excitation ͑ = 6.2 ms͒ as it was previously observed in Yb 3+ -doped RbMnCl 3 ͑Ref.9͒ and is probably related to an additional increase in emission probability due to the Mn 2+ -Yb 3+ interaction within the UCefficient clusters.In the case of LMnA: 1%Yb 3+ , a luminescence intensity rise before decaying is clearly detected ͓Fig.4͑d͔͒.These differences are crucial to identify the UC mechanism involved in these systems as will be discussed later.
Figures 5͑a͒ and 5͑b͒ show the temporal evolution of the Yb 3+ -Yb 3+ cooperative luminescence in LMA: 2%Mn 2+ , 5%Yb 3+ after IR excitation at 10205 cm −1 at 15 K and RT, respectively.The decay can be described by a single exponential function with lifetimes of = 470 s at 15 K and = 220 s at RT.
In LMA: 2%Mn 2+ , 5%Yb 3+ a lifetime of 5.2 ms has been obtained for the Mn 2+ UC emission ͓Fig.4͑c͔͒ while the RT lifetime of Yb 3+ -Yb 3+ cooperative luminescence is 220 s ͓Fig.5͑b͔͒.Since both lifetimes differ by an order of magnitude, we can separate both UC emissions by time-resolved spectroscopy.Figure 6 shows both the Mn 2+ green emission ͑broad band centered at 19430 cm −1 ͒ and the Yb 3+ -pairs luminescence ͑higher energy peaks above 19500 cm −1 ͒ upon IR excitation at 10205 cm −1 taken at different delay times after the excitation pulse.It can be seen that the Yb 3+ -Yb 3+ cooperative emission intensity decreases much faster than the Mn 2+ UC luminescence intensity according to their lifetimes.The figure clearly illustrates this phenomenon showing the rapid decay of Yb 3+ -pairs luminescence on a millisecond time scale while the Mn 2+ UC emission remains almost constant for this time.

D. Temperature dependence
Figure 7 shows the temperature dependence of the normalized emission intensity ͑I / I max ͒ and the lifetime ͑͒ of the Mn 2+ 4 T 1 → 6 A 1 transition in LMA: 2%Mn 2+ , 5%Yb 3+ and LMnA: 1%Yb 3+ upon IR excitation at 10205 cm −1 in the range 15-600 K.It must be noted that for LMA: 2%Mn 2+ , 5%Yb 3+ the UC luminescence intensity remains almost constant up to RT.Although above this temperature the quenching process becomes important and the UC luminescence intensity starts to decrease, it still persists up to 600 K.The associated UC lifetime follows a similar trend; it diminishes slightly from 10 to 300 K and steeply decreases above this FIG.4. RT temporal evolution of the normalized 4 T 1 → 6 A 1 Mn 2+ green luminescence intensity after pulsed excitation at 23580 cm −1 in ͑a͒ LMA: 2%Mn 2+ , 5%Yb 3+ and ͑b͒ LMnA: 1%Yb 3+ and temporal behavior of the Mn 2+ UC emission after pulsed excitation at 10205 cm −1 in ͑c͒ LMA: 2%Mn 2+ , 5%Yb 3+ at RT and ͑d͒ LMnA: 1%Yb 3+ at both 15 K and RT.The insets show the same data in semilogarithmic scale.Continuous lines are the results of different fitting procedures ͑see text͒.temperature.Both the intensity and the lifetime have a similar behavior pointing out the nonradiative character of the UC luminescence decrease.The quenching process of the Mn 2+ UC luminescence is rather different in LMnA: 1%Yb 3+ , as is shown in Fig. 7.As it usually happens in concentrated materials, the Mn 2+ emission is thermally quenched at much lower temperatures ͑T Ͻ 100 K͒ than in Mn 2+ diluted samples.This is an intrinsic behavior of concentrated manganese and is usually accompanied by a similar lifetime decrease with temperature. 12,22However, since in this case lifetime remains pretty constant until RT, the reduction in the intensity due to migration and subsequent quenching is unlikely.Instead, the photoluminescence dwindle above 100 K, which is induced either by direct excitation into the Mn 2+ 4 A 1 4 E state or via UC in LMnA: 1%Yb 3+ , must be ascribed to a decrease in the effective Mn 2+ excitation with temperature.As we show below, this puzzling I͑T͒ behavior between 0 and 300 K can be phenomenologically described through a thermal activation processes for Mn 2+ excitation.
In order to model the Mn 2+ luminescence in LMA: 2%Mn 2+ , 5%Yb 3+ let us assume that the radiative lifetime, R , does not change significantly with temperature.According to a model proposed by Mott, 23 thermally activated nonradiative processes can be described by an activation energy, E a , with a pre-exponential frequency factor, p. Consequently, the lifetime and intensity dependences on temperature have been fitted according to Eqs. ͑1͒ and ͑2͒.In both cases the fitted parameters are E a = 0.26 eV and p = 180ϫ 10 3 s −1 for LMA: 2%Mn 2+ , 5%Yb 3+ .
It must be noted that this simple model explains why I͑T͒ and ͑T͒ behave with temperature in the same manner.The activation energy ͑E a = 0.26 eV͒ and the Mn 2+ dilution both suggest that the nonradiative process is associated with multiphonon relaxation within ͑MnO 4 ͒ units.
In this analysis we assume that the concentration of excited Mn 2+ via UC or by direct excitation 6 A 1 → 4 A 1 4 E is approximately independent on temperature.Nevertheless, in the case of LMnA: 1%Yb 3+ , Eq. ͑2͒ must account for the fraction of excited Mn 2+ as a function of temperature: E R ͑T͒ = ͓Mn͔͑T͒ / ͓Mn͔͑0͒ with ͓Mn͔ being the concentration of excited Mn 2+ .Hence, Eq. ͑2͒ transforms to On the assumption that the concentration of up-converted Mn 2+ does not change with temperature ͑E R =1͒ then we obtain the same variation for I͑T͒ and ͑T͒ as Eqs.͑1͒ and ͑2͒.However, this is not the case if E R varies with temperature.The I͑T͒ reduction above 100 K observed in LMnA: 1%Yb 3+ keeping ͑T͒ constant up to RT can be explained through this model if we consider that Mn 2+ excitation is thermally deactivated: E R = ͑1−␣e −⌬/K B T ͒, where ⌬ is the activation energy and ␣ is the pre-exponential factor.Experimental I͑T͒ data ͑blue triangles͒ behaves in this way for values of ⌬ = 5.6 meV and ␣ = 1.2.Although the present model makes it compatible the reduction in intensity and the lifetime constancy, this thermal behavior is unusual for Mn 2+ concentrated systems in which I͑T͒ correlates with ͑T͒.The microscopic origin of this puzzling phenomenon is unclear and deserves further investigation.

E. UC mechanism
The RT Mn 2+ 4 T 1 → 6 A 1 UC luminescence intensity versus the excitation power density at 10250 cm −1 for LMA: 2%Mn 2+ , 5%Yb 3+ is plotted on a double-logarithmic scale in Fig. 8.This system presents a quadratic power dependence below 1.5 W cm −2 which is the typical behavior of a twophoton excitation process.The slope decreases from 2 to 1.5 at higher excitation power densities as explained elsewhere. 24,25Since excitation occurs into 2 F 5/2 Yb 3+ states and emission takes place from 4 T 1 Mn 2+ states, the active UC mechanism in LMA: 2%Mn 2+ , 5%Yb 3+ and LMnA: 1%Yb 3+ must involve participation of both Yb 3+ and Mn 2+ ions.Since Mn 2+ has no intermediate resonant states with Yb 3+ , then GSA/ESA and GSA/ETU mechanisms in pure ions must be ruled out in these systems.A simple model based on exchange-coupled Mn 2+ -Yb 3+ dimer states has been previously proposed to explain the experimental behavior in Yb 3+ -doped RbMnCl 3 , CsMnBr 3 , and CsMnCl 3 , 7,9 ͑see above͒.
The UC emission in LMA: 2%Mn 2+ , 5%Yb 3+ ͓Fig.4͑c͔͒ shows an immediate decay after the laser pulse with no rise due to energy transfer longer than 80 ns ͑low-detection ͑Color online͒ Temperature dependence of the LMA: 2%Mn 2+ , 5%Yb 3+ Mn 2+ UC luminescence intensity ͑red circles͒ and Mn 2+ UC emission lifetime ͑open squares͒ and temperature dependence of the LMnA: 1%Yb 3+ Mn 2+ UC luminescence intensity ͑blue triangles͒ and UC emission lifetime ͑black squares͒ after 10205 cm −1 excitation.Fits according to Eqs. ͑1͒-͑3͒ ͑see text͒ are also shown.limit͒.According to the LMA crystal structure ͑Fig.2͒, the Mn 2+ -O-OЈ-Yb 3+ dimers in LMA: 2%Mn 2+ , 5%Yb 3+ exhibits a superexchange pathway with shortest Mn 2+ -Yb 3+ distances of 5.728 Å. 26,27 Taking into consideration a GSA/ ESA-type mechanism in Mn 2+ -Yb 3+ dimers, the GSA step is predominantly a single-ion process in Yb 3+ but the ESA step requires the mixing of Mn 2+ and Yb 3+ states.The rise of the emission intensity after the excitation pulse in LMnA: 1%Yb 3+ ͓Fig.4͑d͔͒ evidences the contribution of some nonradiative energy-transfer process after the pulse to the UC mechanism.The involved processes can be GSA/ETU in a dimer or cooperative sensitization.Again, the Mn 2+ -Yb 3+ distance in LMnA ͑5.724 Å͒ is short enough to allow dimer formation, hence we propose the GSA/ETU as the actual mechanism.The Mn 2+ lifetime is shorter in LMnA: 1%Yb 3+ than in LMA: 2%Mn 2+ , 5%Yb 3+ , this evidences a larger exchange contribution for the pure manganese system.The fact that the intensity rise does not start from zero is noteworthy.This implies that there must be an additional excitation within the duration of the laser pulse governed by a GSA/ ESA mechanism.Considering that the lifetime of the 2 F 7/2 -4 T 1 dimer state is approximately the same as the Mn 2+ lifetime after direct excitation, it is possible to estimate the GSA/ESA and GSA/ETU contribution to the total intensity. 11,28At 15 K about 23% of the total upconversion luminescence is due to excitation within the pulse ͑GSA/ ESA͒ while contributions of 33% and 67% have been obtained at RT for GSA/ESA and GSA/ETU mechanisms, respectively.
Figure 9 shows a schematic representation of an exchange-coupled Mn 2+ -Yb 3+ dimer and an Yb 3+ nearby monomer along with the proposed mechanisms for the UC luminescence; GSA/ESA for LMA: 2%Mn 2+ , 5%Yb 3+ and both GSA/ESA and GSA/ETU for LMnA: 1%Yb 3+ .The dimer ground state and intermediate state are dominantly localized on Yb 3+ .The higher excited states are mainly localized on Mn 2+ ions. 9he temporal evolution of the LMnA: 1%Yb 3+ can be simulated considering GSA/ESA and GSA/ETU mechanisms according to the five levels system shown in Fig. 10 involv-ing an Yb 3+ ion and a Mn 2+ -Yb 3+ dimer.Within this model the coupled differential rate equations describing the population of each level, N i can be written as where lifetime; 4.4 ms at 15 K and 3.8 ms at RT.By fitting the time-dependent evolution of the UC luminescence in LMnA: 1%Yb 3+ to this set of equations ͓Fig.4͑d͔͒, we obtain the following energy-transfer rates, W ETU , 2310 s −1 at RT and 3125 s −1 at 15 K, respectively.Nevertheless, the variation in the UC intensity depicted in Fig. 4͑d͒ can be rationalized in terms of the transition rates of Mn 2+ and Yb 3+ .In fact, the rate equation for N 4 ͑t Ն 0͒ can be solved analytically on the assumption that Yb 3+ populations, N 1 and N 3 , are similar ͑N 1 Ϸ N 3 ͒, both decaying as N 3 = N 3 0 e −w 3 t , where w 3 is a mean Yb 3+ transition rate.This approximation yields a timedependent Mn 2+ population as N 4 = N 4 0 ͑e −kt − ␣e −2w 3 t ͒ The parameter k = k 4 + k 43 , governing the temporal decay of Fig. 4͑d͒, is directly related to the Mn 2+ transition rate whereas parameter w 3 controls the intensity rise immediately after t = 0.The parameters ␣ and N 4 0 determines the Mn 2+ population at t = 0 due to GSA/ESA.This equation properly describes the I͑t͒ behavior of Fig. 4͑d͒ with fit parameters: k = 238 s −1 and w 3 = 1043 s −1 at RT and k = 213 s −1 and w 3 = 2295 s −1 at 15 K. Their reciprocal values are similar to the Mn 2+ and Yb 3+ lifetimes, respectively, what supports the proposed UC scenario of Mn 2+ pumping via two Yb 3+ ions, hence the involvement of GSA/ETU in LMnA: 1%Yb 3+ .

IV. CONCLUSIONS
In this paper, the spectroscopic properties of LMA doped with different Mn 2+ and Yb 3+ concentrations indicate that these systems are very efficient for UC luminescence in Mn 2+ .LMA crystallizes with the magnetoplumbite crystal structure where Mn 2+ ions are tetrahedrally coordinated by oxygen atoms.UC Mn 2+ luminescence and Yb 3+ -pairs emis-sion have been detected in all samples.We demonstrate that UC luminescence takes place via GSA/ESA mechanism for LMA: 2%Mn 2+ , 5%Yb 3+ whereas an additional GSA/ETU contribution participates in LMnA: 1%Yb 3+ .This conclusion is based on time-resolved spectroscopy and impurity concentration dependent experiments.In both LMA and LMnA systems the luminescence via UC is observed up to RT and above 500 K in LMA: 2%Mn 2+ , 5%Yb 3+ , the latter UC emission being an order of magnitude stronger than in any other.It remains very intense up to RT but at higher temperatures both the emission intensity and lifetime decrease in a similar manner pointing out the nonradiative character of the quenching.A multiphonon relaxation process within MnO 4 is likely.In LMnA: 1%Yb 3+ we detected an unusual decrease in the Mn 2+ luminescence above 100 K with a constant lifetime which is not ascribed to Mn 2+ energy migration and subsequent quenching but to a dwindle of Mn 2+ pumping efficiency.This puzzling behavior has not been observed so far and its unclear origin yet deserves to be clarified.The experimental results and the theoretical rate equations model confirm the different proposed UC mechanisms involving Mn 2+ -Yb 3+ dimer formation.
For this phase the refined cell parameters are: a = 5.5818͑5͒ and c = 22.070͑3͒ Å.The in-
k i represents the decay rate of each level N i , G, and E are the power-dependent GSA and ESA rate constants, respectively, and W ETU is the two-center energy-transfer process parameter.k 1 and k 3 are known from the experimentally measured Yb 3+ lifetime; 940 s at 15 K and 370 s at RT, which are significantly longer than those obtained for LMA: 2%Mn 2+ , 5%Yb 3+ , while k 4 is the inverse of Mn 2+ emission ͑Color online͒ Energy states in a Mn 2+ -Yb 3+ dimer notation with the Mn 2+ emission and the proposed UC mechanisms.