Energy transfer in Eu 3+ doped scheelites: use as thermographic phosphor

: In this paper the luminescence of the scheelite-based CaGd 2(1-x) Eu 2x (WO 4 ) 4 solid solutions is investigated as a function of the Eu content and temperature. All phosphors show intense red luminescence due to the 5 D 0 – 7 F 2 transition in Eu 3+ , along with other transitions from the 5 D 1 and 5 D 0 excited states. For high Eu 3+ concentrations the intensity ratio of the emission originating from the 5 D 1 and 5 D 0 levels has a non-conventional temperature dependence, which could be explained by a phonon-assisted cross-relaxation process. It is demonstrated that this intensity ratio can be used as a measure of temperature with high spatial resolution, allowing the use of these scheelites as thermographic phosphor. The main disadvantage of many thermographic phosphors, a decreasing signal for increasing temperature, is absent.


Introduction
Scheelites are A 1-BO 4 compounds (A = alkali, alkaline-earth or rare-earth element, B = Mo, W and some other tetrahedrally coordinated cations; denotes a cation vacancy) in which broad possibilities of isovalent and heterovalent cation substitutions result in a range of compounds with interesting optical properties, good stability and simple preparation [1][2][3].Upon heterovalent substitution at the A sublattice, the charge balance can be maintained by the introduction of cation vacancies, giving rise to compositions characterized by a (A + A'):(BO 4 + B'O 4 ) ratio different from 1:1.This paper describes the temperature dependent luminescence properties of the CaGd 2(1-x) Eu 2x (WO 4 ) 4 (x = 0 to 1, = vacancy) solid solutions and discusses their potential application as a thermographic phosphor.
The technological applications of luminescent materials are very diverse, going from lighting to display and imaging applications [4][5][6][7][8][9][10].In some cases, the luminescence has a very specific temperature dependence, so that it can be used as temperature sensor [11][12][13][14][15][16].This was first mentioned by Neubert in 1937 and the first application dates back to the 1950s, when a phosphor was painted on the wing surfaces of a wind-tunnel model in order to probe the temperature of different parts [17][18][19].In principle, many different response modes are possible, but the three most important ones are the relative intensities of emission peaks, changes in decay time and shifts in peak position [20][21][22].For the presently investigated scheelites, the focus will be on the fluorescence intensity ratio response mode.Kusama et al. were among the first to discuss this technique for Y 2 O 2 S:Eu 3+ , but the intensity ratios of the Eu 3+ transitions were insufficiently dependent on the temperature to allow its use as thermographic phosphor [23].Until now, most research focused on Dy 3+ or Sm 3+ as more suitable thermographic rare earth activators, using the temperature dependence of the intensity ratio of different emission lines [24][25][26][27][28].However, in this paper it will be shown that in the studied scheelite materials, the varying intensity ratio of the emission of the 5 D 0 and 5 D 1 energy level of Eu 3+ is sensitive enough for use as temperature sensor.The mechanism behind the temperature dependence is explained by a phonon-assisted cross-relaxation process, and not by thermalisation like in most cases [4].To demonstrate the feasibility and prospects of the concept, a patterned resistive heater was coated with CaEu 2 (WO 4 ) 4 and illuminated with monochromatic light at 465 nm.The temperature map obtained by using the 5 D 0 to 5 D 1 intensity ratio showed high spatial resolution, in addition to an excellent correspondence with the temperature map recorded by more common infrared thermography.Structure Analysis: The phase purity of the obtained CaGd 2(1-x) Eu 2x (WO 4 ) 4 (x = 0, 0.01, 0.05, 0.1, 0.25, 0.5, 1) solid solutions was checked with X-ray powder diffraction patterns collected on a Thermo ARL X'TRA powder diffractometer (CuKα radiation, λ = 1.5418Å, Bragg-Brentano geometry, Peltier-cooled CCD detector).

Synthesis of the CaGd
Spectral characterization: Luminescence emission and excitation spectra were obtained with a FS920 spectrometer (Edinburgh Instruments), using a 450W xenon light source, double excitation monochromator and a R928P photomultiplier.All emission and excitation spectra were corrected for the sensitivity of the spectrometer and the intensity of the excitation.Decay times were collected using a pulsed dye laser (λ exc = 385 nm) based on optical pumping by a nitrogen laser (λ exc = 337 nm, pulse duration < 1ns, 1Hz repetition frequency) in combination with an intensified CCD detector (Andor Instruments) coupled to a 0.5 m monochromator.All measurements were taken at room temperature unless mentioned otherwise, and the temperature dependent measurements were performed using an Oxford Optistat CF cryostat.
Thermometry: The resistive heater was made by depositing a thin Al layer by electron beam evaporation onto a AF45 glass substrate (Präzisions Glas & Optik, 5x5 cm 2 ), through a mask.Some irregularities were introduced in the heater to obtain temperature gradients upon application of a voltage over the outer contacts.Finally, the patterned heater was covered with a thin film of Al 2 O 3 (100nm) by e-beam evaporation as a diffusion barrier.To deposit the thermographic phosphor, 0.06 g of CaEu 2 (WO 4 ) 4 was mixed with 0.18 g of Silres® (Wacker Chemie), painted on the patterned resistive heater and cured for two hours at 200°C.The material was excited with a Melles Griot argon ion laser with an excitation wavelength of 465 nm.The laser spot was expanded to approximately 3 cm in diameter.Photos were taken with a Nikon D3200 digital camera combined with a 535FS10-M52 green filter (Andover Corporation, transmission peak at 537 nm, 10 nm FWHM) or a red interference filter (peak wavelength of 625 nm) for selecting the green and red emission respectively.To obtain the reference temperature plot a FLIR A35sc (48°) infrared camera was used (320 x 256 resolution, temperature range from −40 °C to 550 °C with an accuracy of ± 5 °C).

Luminescence emission and excitation spectra
In Fig. 1(a) the photoluminescence excitation spectrum is shown for CaGd 1.8 Eu 0.2 (WO 4 ) 4 upon monitoring the red emission at 612 nm, originating from the 4f 6 ( 5 D 0 ) -4f 6 ( 7 F 2 ) transition within Eu 3+ .The excitation spectrum consists of the intraconfigurational 4f 6 -4f 6  transitions of Eu 3+ in the 300 to 500 nm region and a broad band in the region from 250 to 350 nm.The most intense Eu 3+ 4f-4f excitation peak is due to the 7 F 0 -5 L 6 (395 nm) transition.The small peak at 313 nm corresponds to the 8 S 7/2 -6 P 7/2 transition within the 4f 7 configuration of Gd 3+ .As the 6 P 7/2 level is the lowest excited level of Gd 3+ , no luminescence is expected from Gd 3+ due to re-absorption and energy transfer to Eu 3+ .The presence of the 313 nm peak in the excitation spectrum of the Eu 3+ emission points to energy transfer between the Gd 3+ and Eu 3+ ions.Apart from that, Gd 3+ has only limited effect on the optical characteristics, which is common in Eu 3+ -Gd 3+ co-doped samples [30,31].The excitation spectrum is representative for all studied samples, with composition CaGd 2(1-x) Eu 2x (WO 4 ) 4 (x = 0.01, 0.05, 0.1, 0.25, 0.5, 1).The broader band peaking at 285 nm is ascribed to a combination of a charge transfer (CT) transition from the 2p orbital of oxygen to the 3d orbital of tungsten in the (WO 4 ) 2-group and an O 2--Eu 3+ charge transfer band [32][33][34].A typical emission spectrum is shown in Fig. 1(b).The most intense emission peak in the region from 600 nm to 620 nm is due to the forced electric dipole transition 5 D 0 -7 F 2 .A magnetic dipole emission is found at 590 nm, originating from the 5 D 0 -7 F 1 transition [35].The emission wavelengths of these 4f-4f transitions are only moderately influenced by the environment of the lanthanide ions since the partially filled 4f shell is well shielded by the filled 5s and 5p orbitals [36,37].No additional peaks or shifts in peak positions are observed when varying the europium concentration, only the relative intensities change.For the remainder of this work we will describe the emission originating from the 5 D J level to the 7 F J' ground state multiplet as " 5 D J emission".In Fig. 2 the dependence of the emission intensity of the 5 D 0 -7 F 2 transition on the concentration of Eu 3+ in CaGd 2(1-x) Eu 2x (WO 4 ) 4 is illustrated, which can be regarded as a measure for the external quantum efficiency.As can be seen, the intensity reaches a maximum for x = 0.5 and levels off for higher concentrations.Note that the concentration quenching is fairly limited, given the high emission intensity for the fully doped CaEu 2 (WO 4 ) 4 .

Decay measurements
The luminescence lifetime of the emission from the different 5 D J levels was found to be independent of the final level 7 F J' .At 75 K all samples show a similar mono-exponential decay for the 5 D 0 emission (λ exc = 385 nm), with a decay constant of 550 µs (Fig. 3), which is in line with previously reported decay times for similar materials [38][39][40].For low dopant concentration, the decay profile shows almost no temperature dependence in the 75 K to 475 K range.For the highly doped samples, the temperature dependence is much more pronounced.The sample with full Eu substitution (i.e.CaEu 2 (WO 4 ) 4 ) has a mono-exponential decay profile, but the decay time decreases gradually from 470 µs at low temperature to 230 µs at 475 K.For the intermediate case of CaGdEu(WO 4 ) 4 the decay starts deviating from a monoexponential profile upon increasing temperature, and an accurate fit can be obtained by combining the low temperature decay component (τ 1 = 550 µs) on the one hand and the changing decay component (τ 2 = 550 µs to 230 µs) of CaEu 2 (WO 4 ) 4 at all temperatures on the other hand, in the following way: The fraction of the temperature dependent component in the decay is then defined by: and remains constant over the whole temperature range (see Table 1).Taking into account the limited size difference of Gd 3+ (r = 1.053Å) and Eu 3+ (r = 1.066Å), we expect both cations to be uniformly distributed in the crystal.For low Eu concentrations this will result in a relatively large average distance between different Eu ions, leading to a low probability of Eu -Eu energy transfer.As the Gd 3+ ion is optically silent in the visible part of the spectrum, we can consider the Eu 3+ ion as an isolated center in the Gd 3+ dominated host for these low concentrations.For high Eu concentration, energy transfer is likely to occur between the different europium ions and this is obviously the case for the fully substituted CaEu 2 (WO 4 ) 4 .Interestingly, we can consistently fit the thermal behavior of the decay for different dopant concentrations with only two types of Eu ions (i.e.isolated ones and those showing energy transfer), both characterized by their own specific thermal behavior.

Thermal quenching
The luminescence spectrum was monitored as a function of temperature in the range from 75 K to 475 K to evaluate the thermal quenching behavior.For the samples with low Eu concentration, the thermal quenching profile of the emission from the 5 D J levels is quasiidentical for all transitions.The integrated luminescence intensities upon excitation at 465 nm are shown in Fig. 4(a) for CaGd 1.8 Eu 0.2 (WO 4 ) 4 , with each intensity normalized to the value at 75 K.At a temperature of 300 K, the emission has dropped to 25% of the initial intensity at liquid nitrogen temperature, suggesting that at room temperature a significant fraction of the emission is already quenched.This is somewhat in contrast to the luminescence decay behavior ( §3.3), where hardly any decrease of the decay constants was observed upon increasing the temperature.In principle, thermal quenching is accompanied by a faster luminescence decay, due to an increase in non-radiative depopulation of the excited states.This discrepancy could not be explained yet, but might be related to the temperature dependent absorption strength for the excitation light.
Similar to the decay profiles, the temperature dependence of the luminescence is completely different for higher Eu concentrations.The 5 D 0 emission has similar quenching behavior compared to the samples with low Eu concentration, but for the 5 D 1 emission a strong absolute increase in intensity is observed for temperatures above room temperature.This is illustrated in Fig. 4(b) for CaEu 2 (WO 4 ) 4 .For higher Eu concentration the emission output and the decay profiles show a marked temperature dependence and it is therefore expected that energy transfer mechanisms play a role.
In Fig. 5, the decay of the 5 D 1 emission of CaEu 2 (WO 4 ) 4 is depicted and apparently it consists of both a temperature independent component of the order of a microsecond and a temperature dependent slow component.The luminescence decay for Eu 3+ is commonly much faster for the 5 D 1 emission than for the 5 D 0 emission, due to non-radiative depopulation of the higher excited state, e.g. by phonon relaxation or cross-relaxation processes.For higher dopant concentration -and thus for higher probability of energy transfer -the emission from 5 D 1 is often even not visible at all.Here we observe a fast component in the decay for 5 D 1 emission with a decay constant of 13 µs for CaGd 1.8 Eu 0.2 (WO 4 ) 4 and 1.8 µs for CaEu 2 (WO 4 ) 4 , which is indeed considerably faster than the 5 D 0 emission.In addition, an unexpectedly slow component with a decay constant of hundreds of µs is observed as well.By comparing this slow component of the 5 D 1 emission with the decay of the 5 D 0 emission at the respective temperatures, a clear correspondence is observed.The fraction of the slow component (defined by Eq. ( 2)) in the decay of the 5 D 1 emission varies from f 2 = 0.3 at 75 K to f 2 = 0.8 at 475 K.This points to an increased feeding of the 5 D 1 level from the 5 D 0 level upon increasing temperature for CaEu 2 (WO 4 ) 4 , by an energy transfer process.
The same decay analysis was performed for CaGd 1.8 Eu 0.2 (WO 4 ) 4 (not shown).Although some transfer from 5 D 0 to 5 D 1 indeed occurs for these samples, the contribution is too low to result in a considerable increase of the 5 D 1 emission beyond the normal decay, which can also be observed in Fig. 4(a).Both the thermal quenching and the decay profiles point at a concentration dependent cross-relaxation process, rather than a simple thermalisation process where the energy difference to populate the 5 D 1 level from the 5 D 0 level is fully covered by phonons.As will be substantiated in §3.5, the increase in the 5 D 1 emission can be explained by a thermally assisted cross-relaxation process with the involvement of two europium ions initially in the excited 5 D 0 and the 7 F 2 state: ( ) ( ) ( ) ( ) This process, depicted in Fig. 6., favors the 5 D 1 emission of ion 2 (green arrow) at the expense of the 5 D 0 emission (red arrow), due to the transfer of the 7 F 2 -7 F 0 energy difference from ion 1 (green dotted arrow).As this is not a resonant process, phonons are still needed to overcome the energy difference between 7 F 2 -7 F 0 (1000 cm −1 +/− 100 cm −1 ) and 5 D 0 -5 D 1 (1885 cm −1 +/− 100 cm −1 ).The energy differences between the 2S + 1 L J manifolds were derived from the emission spectrum.For higher Eu concentrations, this cross-relaxation process is more likely to occur due to a reduced average distance between the Eu ions.Note that a similar temperature dependence was observed for the emission of Eu 3+ in GdVO 4 by Nikolić et al, where the thermal quenching was explained by thermalisation from the lowest to the highest excited level, leading to a thermal barrier of about 1500cm −1 [41].However this thermalisation cannot account for the observed concentration dependence in our case.
The involvement of the phonons in Eq. (3) explains for the observed temperature dependence.This thermally assisted cross-relaxation should indeed result in a slow component in the decay of the 5 D 1 emission, comparable with this for the 5 D 0 emission, as the energy transfer can only occur as long as the 5 D 0 level remains populated.Note that the same temperature dependency for the decay constant of the 5 D 0 emission (Table 1) and the slow component in the 5 D 1 emission (Fig. 5) is found.

Use of CaEu 2 (WO 4 ) 4 as thermographic phosphor
Because of the strongly changing emission characteristics upon increasing temperature for CaEu 2 (WO 4 ) 4 , this material is suitable as thermographic phosphor.Figure 7(a) shows the ratio of the 5 D 1 to 5 D 0 emission intensity as a function of temperature.The intensities are obtained by integrating over the wavelength ranges 535 to 545 nm ( 5 D 1 -7 F 1 ) and 585 to 600 nm ( 5 D 0 -7 F 1 ) respectively, upon steady state excitation at 395 nm.However, one is not limited to this excitation wavelength and specific 5 D J -7 F J' transitions, as the cross-relaxation mechanism only depends on the emitting 5 D J levels.
Due to the non-resonant character of the cross-relaxation process, the 5 D 0 and 5 D 1 energy levels need to be thermally coupled for the process to occur.In case of thermal coupling, it can be shown that the relative population of both levels can be described by a Boltzmann distribution: with B depending on the degeneracy, emission cross-section and angular frequency of the respective levels, k the Boltzmann constant, T the absolute temperature and ΔE the energy difference between the involved states [42][43][44].The uncertainty of the experimental data in Fig. 7(a) is on the order of a few degrees Celsius.Measurements were found to be reproducible upon consecutive heating cycles and did not lead to degradation of the material.The Arrhenius plot of the ratio R is given in the inset of Fig. 7(a), and the fit was obtained by varying both B and ΔE.The fitted value of 1068 cm −1 for ΔE should be compared with the energy of the phonons needed for the process.This energy is determined by the mismatch between the 7 F 2 -7 F 0 (1000 cm −1 +/− 100 cm −1 ) and 5 D 0 -5 D 1 (1885 cm −1 +/− 100 cm −1 ) energy difference.The variation on each energy difference arises from the multiplet splitting of 7 F 2 and 5 D 1 .This value agrees reasonably well with the fitted value of 1068 cm −1 , so the use of the Boltzmann equation is validated.The one to one correspondence between the data and the fit makes CaEu 2 (WO 4 ) 4 applicable as thermographic phosphor from 300 K to at least 500 K.For a quantitative comparison of thermographic phosphors, the relative sensitivity S rel is the most appropriate parameter and is defined as follows [45,46]: The resulting curve is given in Fig. 7(b).The sensitivity varies from 0.014 K −1 at 300 K to 0.0047 K −1 at 475 K.These values are slightly lower than the relative sensitivities for other Eu 3+ doped thermographic materials.For example Eu 3+ doped (Y 0.75 Gd 0.25 ) 2 O 3 has a maximum relative sensitivity of 0.025K −1 [47].On the other hand, the scheelite thermographic phosphor has a much larger temperature operation range compared to most self-referencing materials [48].Compared to the performance of Dy 3+ , Sm 3+ and Nd 3+ doped materials or up-conversion nano-particles, the obtained maximum sensitivity is similar or even higher [27,[47][48][49][50][51][52][53].
To prove its usefulness in thermal imaging, CaEu 2 (WO 4 ) 4 phosphor powder was mixed in a 1:3 weight ratio with SILRES® binder and coated onto a patterned, purposely made resistive heater (Fig. 8) to obtain an inhomogeneous temperature distribution.The device was heated to a maximum local temperature of about 150°C.Then, the coated heater was illuminated with the 465 nm emission of an Ar laser to excite Eu 3+ via the 7 F 0 -5 D 2 transition.In order to map the temperature gradients of the heated surface, pictures were taken with a Nikon D3200 digital reflex camera combined with a green and a red narrow band filter, collecting the relevant fractions of the green 5 D 1 and the red 5 D 0 emission separately.Images were read out in the NEF data format, and divided pointwise.Using Eq. ( 4), a thermal map was obtained (Fig. 8, middle).As a reference, images were made with a thermal imaging infrared camera (Fig. 8, right).For both types of measurements, a horizontal and a vertical temperature profile was extracted from the corresponding thermal maps (Fig. 9) at the positions indicated on the photograph of the coated heater (Fig. 8, left).It can be seen that the thermographic scheelite phosphor is capable of reproducing the temperature gradients in a very accurate way, which proves the applicability of the material in the postulated temperature range.A major advantage specific for this response mode is that there is no need for an absolute intensity measurement, eliminating errors introduced by fluctuations in excitation intensity and light collection efficiency [45].Together with the obtained high spatial resolution and the non-destructiveness of the method, this makes the material a very promising candidate for thermometry.Indeed, after use, the phosphor coating can be easily removed using an appropriate solvent such as acetone.In addition, the involved emission peaks are very well separated, making the detection relatively easy.The main drawback for the material is the low 5 D 1 emission intensity compared to the 5 D 0 emission intensity.Nevertheless, the crossrelaxation process can still be optimized, for example by minimizing the defects in the host material.This reduces the non-radiative decay paths from 5 D 0 to 7 F J , which can then increase the probability of populating the 5 D 1 level.In addition, other host materials could be chosen with a different distance between the europium ions or with phonon energies having an improved match with the energy needed for the cross-relaxation process.

Conclusion
In this work, we investigated the temperature dependence of both the luminescence and the decay pathways of CaGd 2(1-x) Eu 2x (WO 4 ) 4 scheelites in a systematic way.All phosphors emit intense red light dominated by the 5 D 0 -7 F 2 transition at 612 nm, and have similar excitation and emission spectra.The temperature dependence of the 5 D J emission intensities was however found to be strongly dependent on the europium concentration.For high Eu concentration strong cross-relaxation occurred upon increasing temperature due to energy transfer between nearby Eu ions.This thermally assisted cross-relaxation manifests itself in an increasing emission from the 5 D 1 level, characterized by the same decay time as the 5 D 0 emission, upon increasing temperature.Because of this specific temperature dependence, CaEu 2 (WO 4 ) 4 was proven to be suitable for intensity ratio based temperature measurements.The ratio of the 5 D 1 to 5 D 0 emission intensity follows a Boltzmann distribution equation.Consequently, an accurate estimate of the temperature can be made in the range from 300 K to at least 500 K by measuring the relative emission intensity of both peaks.This temperature range is large compared to other ratiometric systems and it was experimentally shown that the phosphor was able to reproduce temperature gradients with high accuracy and spatial resolution, confirming the applicability of the material for thermometry and thermography.A disadvantage of the present phosphor is the relatively low integrated emission intensity from the 5 D 1 emission compared to the dominating 5 D 0 emission, which lengthens the acquisition time for the thermal imaging.Future work will therefore focus on the optimization of the cross-relaxation process and the total emission intensity of the 5 D 1 state, by optimizing the host composition and reduction of the non-radiative decay paths.

Fig. 1 .
Fig. 1. a) Excitation spectrum of CaGd 1.8 Eu 0.2 (WO 4 ) 4 upon monitoring the emission at 612 nm.The excitation peak at 313 nm related to Gd is indicated by (*).b) Emission spectrum of CaGd 1.8 Eu 0.2 (WO 4 ) 4 upon excitation at 395 nm.The electronic transitions for the main excitation and emission peaks are indicated.

Table 1 .
Decay constant and fraction of the variable decay time component of the5

Fig. 5 .
Fig. 5. Decay of the 5 D 1 emission (λ exc = 385 nm) at 75 K and 475 K for CaEu 2 (WO 4 ) 4 .The fast decay component remains constant and equals 1.8µs.The slow decay component equals the decay constant of the 5 D 0 emission at the respective temperatures.

Fig. 7 .
Fig. 7. a) Ratio (R) of the integrated intensities of the 5 D 1 to 5 D 0 emission for CaEu 2 (WO 4 ) 4 and the corresponding Arrhenius plot (inset).b) Calculated relative sensitivity S rel as a function of temperature.

Fig. 8 .
Fig. 8. Patterned resistive heater with cross-sections for the vertical and horizontal profile indicated by the black lines (left).Temperature plot of the patterned resistive heater, imaged with the use of the thermographic phosphor (middle) and with an infrared camera (right).