1. Introduction

In recent years, lanthanide-doped fluorides have received much attention in materials research due to their unique properties, including large anti-Stokes shifts, sharp emission peaks, low biotoxicity, and simple synthesis conditions [1–5]. Most lanthanide-doped fluorides are low phonon energy systems such as lithium yttrium fluoride (LiYF₄), sodium yttrium fluoride (NaYF₄), and potassium manganese fluoride (KMnF₃), which are doped with lanthanide ions enabling outstanding up-conversion and down-shifting emissions under near-infrared excitation [6–10]. Furthermore, lanthanide ions have ladder-like energy levels, which can provide various range light bands, that cause major advantages in optic applications by using the system of lanthanide-doped fluoride [11].

As the heaviest alkali metal apart from radioactive francium, Cesium (Cs) will significantly reduce the vibrational frequency of fluoride lattice compared to Li/Na/K-based fluorides [12–14]. Moreover, the tight cubic structure and ordered cation arrangement will significantly reduce the non-radiation relaxation process and thus improve the luminescence environment of the lanthanides. Unfortunately, Cs-based fluorides have an Achilles’ heel in that they are usually susceptible to deliquescence in air and cannot withstand high-power laser irradiation. Limited by the deliquescence properties of Cs-based fluorides, it is challenging to provide an environment suitable for light emission. In the past few years, considerable progress has been made in the protection methods of luminescent materials in special environments, especially in the field of optical fiber sensors and biosensors. Wang's team improved the silicon rubber encapsulated optical fiber sensor and proposed a feasible scheme to improve the feasibility and durability of the sensor by packaging design [15,16]. Theoretical studies on the temperature- and stress-related effects of polymeric optical fibers are abundant and provide important insights into the design of Cs-based fluoride encapsulation to follow [17–19]. The functionalization of biosensors and the protection measures under the detection of the surrounding environment have raised concerns, which also provide important references for solving Cs-based fluoride deliquescence [20]. More encouragingly, the doping of Yb³⁺/Er³⁺/Ho³⁺/Tm³⁺ ions in Cs₃YF₆ has been investigated by Wang’s group, not only the outstanding luminescence properties but also the unique anti-deliquescence approach has attracted the attention of many scholars. Wang’s creative preparation of Cs₃YF₆ embedded glass (Cs₃YF₆@glass) by in situ glass crystallization strategy, is a positive beginning and opens the minds of later researchers [14]. Generally, the investigation of the deliquescence of perovskite films is mainly to adjust the proportion of water vapor and dry inert gas to determine the crystal phase change [21]. We designed the soaking experiment in different polar solvents to illustrate the effect of deliquescence on Cs-based fluoride and verified it by XRD patterns, which is a pioneer try for deliquescence effect research.

It is well known that temperature is the fundamental physical parameter for evaluating various systems, such as the immune system, meteorology, and microelectronic integrated systems. Meanwhile, high-temperature drying treatment is also an important tool to deal with perovskite sample deliquescence. Lanthanide-based luminescence intensity ratio (LIR) thermometry is promising for temperature evaluation in complex environments [22–27]. Typically, two well-characterized emission peaks in a series of variable temperature spectra are selected for comparison to obtaining LIR values, which means that the temperature information carried by many other emission peaks is ignored. Nowadays, many researchers realize that using more emission peaks to compare with each other helps to obtain more comprehensive temperature information about the system [28–31]. As for Cs-based fluorides, Shuai et al. first introduced Cs₃YF₆:Yb³⁺/Er³⁺ to the LIR temperature sensing field and obtained high sensitivity during 303∼573 K [32]. To the best of our knowledge, there is no research on exploring the temperature sensing properties of up-conversion luminescence in Cs₃ErF₆ phosphor, and introducing the multimode LIR thermometers, which would be a promising host material for the next generation of noncontact-temperature sensing.

Here in this work, we describe the hazards of deliquescence to Cs₃ErF₆ and the approach to protecting Cs₃ErF₆ away from the deliquescence effect. Meanwhile, this work also realizes dual-mode LIR thermometry at 333K-573 K. In detail, the Cs₃ErF₆ has excellent visible and near-infrared emission due to its perovskite-like structure. However, the luminescence of Cs₃ErF₆ was greatly affected by deliquescence due to the properties of Cs-based materials, and for investigating the deliquescence phenomenon, we designed a series of soaking experiments to verify the effect of deliquescence. In addition, we also offered a silicone rubber sheet encapsulation solution isolating the vapor from the sample to prevent deliquescence. As for LIR thermometry, we conducted 15 minutes of heat treatment at 573 K to remove the influence of deliquescence and measured the temperature dependence spectra during the cooling process. Furthermore, we designed two different modes to determine the temperature of samples. The first mode relies on two Stark emissions at the red-light band named “rapid mode” by virtue of the main emission in the visible range and the close wavelength, which can get spectra rapidly. The second mode of LIR thermometry is conventional strategy, interrelated with non-TCL emission, which can always obtain the highest relative sensitivity of LIR thermometry. Based on the reason above, we call the second mode “ultra-sensitive mode”, then gains absolute sensitivity 0.621%K^-1 and relative sensitivity 7.362%K^-1 respectively. The results of the two modes provide a more complete picture of the variation in the ambient temperature of the sample and have the potential for cross-validation.

2. Experimental

2.1 Chemicals and materials

The materials of cesium carbonate (Cs₂CO₃, 99.9%) were obtained from Shanghai Macklin Biochemical Co., Ltd., ammonium fluoride (NH₄F, A.R.) was obtained from Sinopharm Chemical Reagent Co., Ltd., yttrium oxide (Y₂O₃, 99.99%) and erbium oxide (Er₂O₃, 99.99%) were purchased from Aladdin industrial corporation. Ethanol absolute was purchased from Jingdongtianzheng Precision Chemical Reagent Factory (China, Tianjin). Cyclohexane (CYH, ACS) was purchased from J&K Scientific LTD. All chemicals and materials were purchased from commercial sources. Deionized water was used throughout the experiment. All chemicals were used without any purification unless otherwise stated.

2.2. Synthesis of Cs₃YF₆: x% Er (x = 20,40,60,80,100) particles

All the samples were synthesized by solid-state reaction. Initially, Cs₂CO₃, NH₄F, Y₂O₃, and Er₂O₃ powders were mixed with stoichiometric molar ratios and ground in the mortar for 15 min. Then transfer the weighed materials to an agate mortar and grind them thoroughly to make the raw materials mixed evenly. The mixture was then put into corundum crucibles and sintered at 600 ◦C for 3 h in air. After the samples are cooled to room temperature, ground them again to obtain the resultant material. Finally, a series of powder samples of Cs₃YF₆: x%Er (x = 20,40,60,80,100) were made and kept samples into sealed plastic bags.

2.3 Characterization

XRD analysis was performed using a SmartLab SE (Rigaku, Japan) equipped with Cu Kα radiation (λ = 0.15405 nm) in the 2θ range from 10 to 80°. SEM images were acquired using a Regulus 8100 microscope with a field emission gun operated at an accelerating voltage of 15 kV and working distance of 12.5 mm. The luminescence spectra were measured using an Andor Shamrock SR750 fluorescence spectrometer with a 980/808 nm diode (DS321312-112 and DS3-51412-0309 BWT Beijing Ltd.), which was coupled to a fiber laser. The signals of the samples were collected using a CCD detector combined with a monochromator, and the luminescence spectra were recorded using an Andor SR-500i spectrometer (Andor Technology Co., Belfast, UK). Temperature sensing was performed in an iron sample cell heated using resistance wire elements.

3. Results and discussion

A series of different concentrations of Er-doped Cs₃YF₆ are synthesized via solid-state reaction. Figure 1(a) shows the XRD patterns of Cs₃YF₆:x%Er³⁺ (x = 20,40,60,80,100), it can be seen that the samples have cubic phase structure, matching well to the standard pattern of Cs₃YF₆ (PDF#23-1059), as well as the main diffraction peak at 25.9 degrees has been slightly shifted due to the mismatch of the radius of Er^3 +(R_Er= 0.881 Å) to the radius of Y^3 +(R_Y= 0.893 Å). Furthermore, the Cs₃ErF₆ crystal structure is displayed in Fig. 1(b), which belongs to the perovskite-like structure and has Fm-3 m space group. As shown in Fig. 1 (b), Cs⁺ can occupy two different sites, Cs1 turns itself into the center of octahedral sites with six-fold coordinated by fluorine ions. The other type of Cs⁺ is combined with twelve F ions to form the tetrakaidecahedron structure [CsF₁₂]. For Er³⁺ in Cs₃ErF₆ crystal structure, there is obvious that Er is in an octahedral structure [ErF₆] similar to Cs1 sites. This is a material with typical perovskite-like structure, which means that there is only one site in the crystal suitable for lanthanide occupation, and it also indicates that this is an ordered crystal conducive to luminescence.

 

Fig. 1. (a) X-ray diffraction patterns of Cs₃YF₆: x%Er (x = 20,40,60,80,100) particles. (b) Perovskite-like structure of up-conversion material Cs₃ErF₆. (c) SEM patterns of Cs₃ErF₆ and EDS result inside the image (elements Cs, F, and Er are holding in their position). (d) The illustration shows the EDS spectrum of Cs₃YF₆ and the atomic weight analysis.

Download Full Size | PDF

After discussing the crystallographic structure, it is necessary to continue with the microscopic characterization of realistic samples. As shown in Fig. 1(c), the prepared sample appeared to be irregular particles with a grain size of about 50 µm. In order to know the distribution of various elements inside the Cs₃ErF₆, elemental mapping images are provided to confirm the location of Cs, Er and F properly in Fig. 1(c inside). The EDS spectrum of Cs₃ErF₆ clearly shows Cs, Er and F characteristic peaks. In addition, the weight percentage of each element present in the composition obtained from the peak intensities is also shown in the inset of Fig. 1(d). From another view to confirm element distribution, Fig. 2(a) is the XPS survey scanning spectrum of the sample, in which most of the main peaks have been identified. It is clear that the peaks of Cs, F and Er elements are all presented, further indicating that the obtained samples are successfully prepared. Besides, in the detail scanning XPS spectrum, the F (1s) peak is located at about 691.88 eV while Cs (3d) twin peaks are found to be centered at 732.13 and 746.08 eV, which could be distributed to Cs (3d_3/2) and Cs (3d_5/2) core levels, as shown in Fig. 2(b,c). Ultimately, the Er (4d_5/2) is confirmed at peaks with binding energy at 176.53 eV in Fig. 2(d). Another property that needs to be evaluated is the thermal stability of the material to prevent decomposition during prolonged heating. The result of the thermogravimetric analysis is shown in Supplement 1, Fig. S1, that there is no decomposition effect before 800 °C heating. The mass of the sample is stable before heating to 800 °C, which assures that the temperature measurement range of the thermometry is within the thermal stability range of the sample.

 

Fig. 2. (a) The XPS analysis survey spectrum of the Cs₃ErF₆ sample and the high-resolution spectrum of Cs, F and Er, elements are shown from (b) to (d).

Download Full Size | PDF

The up-conversion spectra of the Cs₃YF₆: x%Er (x = 20,40,60,80,100) are measured in the range of 500–700 nm at room temperature, as shown in Fig. 3(a). Under 980 nm laser excitation, all the samples exhibit typical Er³⁺ 534 nm (²H_11/2 → ⁴I_15/2), 544 nm (⁴S_3/2 → ⁴I_15/2) green and 655 nm (⁴F_9/2 → ⁴I_15/2) red emissions. In the visible range of the emission spectra, the red-light band dominates the entire emission intensity due to the fact that doping of Er³⁺ heavily leads to severe cross-relaxation (⁴I_11/2+ ⁴F_7/2→⁴F_9/2), and this cross-relaxation process significantly increases the electron population in ⁴F_9/2 energy level (Fig. 3(c)). In addition, two Stark characteristic emission peaks (655 nm and 673 nm) in the red-light band appear due to the Er³⁺ energy level Stark splitting strongly caused by crystal field effects. As the doping Er³⁺ concentration elevates, the integral luminescence intensity gradually enhances except for the Cs₃YF₆ sample doping with 80% Er³⁺. The most reliable explanation is that concentration quenching plays a decisive role, during the up-conversion exciting process in 80% Er³⁺ doping sample. After investigating the up-conversion emission, the downshifting process also needs to be described. Figure 3(b) displays that the downshifting emission intensity shows a clear linear upward trend with the rise of the doped Er³⁺ concentration from 20% to 100%. The downshifting band of Er³⁺ centered at 1535 nm is assigned to ⁴I_13/2 → ⁴I_15/2 transitions (Fig. 3(c)). The effect of pump power on the up-conversion emission intensity is investigated using different excitation powers ranging from 0.5 to 0.75W. With the increase of pump power, the up-conversion emission intensity increases rapidly. The relationship between up-conversion intensity and excitation pump power can be applied as follow:

 

Fig. 3. (a) UCL spectra and (b) downshifting spectra of Cs₃YF₆: x%Er (x = 20,40,60,80,100) particles excited by 980 nm. (c) Schematic energy level diagram of the Cs₃ErF₆ and energy transfer pathways and the possible radiative transitions (d) Pump power dependence of the UCL intensity of emission peaks at 534 nm and 655 nm.

Download Full Size | PDF

It is well known that Cs-based fluorides have a distinct defect of being susceptible to deliquescence, but the effect of deliquescence on the photoluminescence of Cs-based fluorides has not been specifically reported. As revealed in Fig. 4(a), the UCL spectra of Cs₃YF₆ were completely soaked in ethylene glycol (EG) and deionized water (DI water), and the spectral results confirmed that there was a significant luminescence quenching of the samples soaked in water, where the response of red and green emissions to the quenching effect was different. To visually reflect the quenching effect, we compared the red and green emission soaking in EG and DI water in illustration Fig. 4(a). Compared to red light, which retained 81.1% of its emitted intensity, green emission lost 63.4% intensity. Furthermore, the change of the red-green ratio column will lead to the color of the sample shifting, and the change of CIE coordinates can support this opinion in Fig. 4(b). Predictably, the structure of the samples may have changed in these solvents, and for this reason, we performed drying operations after Cs₃YF₆ soaking in solvents with different polarities. The polarity of the above three solvents from small to large is cyclohexane, ethylene glycol, and deionized water. Moreover, the normalized XRD patterns of the samples soaked in different solvents are shown in Supplement 1, Fig. S2, and it is obvious that as the polarity of the solvent increases, the more destructive the structure of the sample becomes.

 

Fig. 4. (a) UCL spectra on Cs₃ErF₆ soaked in ethylene glycol and deionized water, quenching statistics of red and green light in two solutions. (b) CIE color coordinates for the sample of Cs₃YF₆ after soaking in glycol and deionized water. (c), (d) the process of encapsulating Cs₃ErF₆ with silicone rubber sheets and UCL spectra after encapsulation.

Download Full Size | PDF

Interestingly, it is the first time found that the Cs₃YF₆: Yb³⁺/Er³⁺ sample in soaking in deionized water and cyclohexane displays a completely different color from red to green under 980 nm exaction in Supplement 1, Fig. S3(d). Figure S3(a) in Supplement 1, depicted the XRD peaks of Cs₃YF₆: Yb³⁺/Er³⁺, which were consistent with the standard peak position of cubic Cs₃YF₆ (PDF# 22-1059). When the sample was soaked in water, a miracle happened, the originally green light-dominated up-conversion spectra became a red light-dominated spectrum unexpectedly in Supplement 1, Fig. S3(b). This sudden color shift effect offered potential for disposable anti-counterfeiting applications. As shown in Supplement 1, Fig. S3(c), the red light intensity of the sample in aqueous solution decreased by 23.3%, while the green emission sharply decreased by 93.6% of the light intensity, which reversed the red-green ratio. The above conclusions are consistent with those obtained in Cs₃ErF₆, which indicates that the crystal structure is eroded by highly polar solvent water, leading to the overall quenching of luminescence. Meanwhile, the green emission from high energy levels ²H_11/2 and ⁴S_3/2 is more susceptible to change in the crystal field environment thus causing the enormous loss of green light.

There are many methods to cope with sample deliquescence, but for the research of Cs-based fluorides, the current effective method is to grow Cs₃YF₆ in glass to avoid deliquescence effects. Based on the idea of isolating the contact between vapor and the sample surface, we propose encapsulating the Cs₃ErF₆ particles in two silicone rubber sheets. Compared with the growth of specific crystalline phases in glass, the method of encapsulation in silicone rubber sheets is simpler and faster, and there is no concern for damage to the crystal structure caused by high temperature required for glass growth [33]. A schematic of the entire encapsulation process has been shown in Fig. 4(c). The choice of silicone rubber sheet is mainly based on the following three reasons: 1) silicone rubber sheet has self-healing ability, which will only have adhesion with similar substances in contact and will not react with other materials; 2) silicone rubber sheet has high thermal stability and will not be deformed by long-term laser exposure; 3) silicone rubber sheet is a flexible material that is more conducive to preservation and other extensive applications. The encapsulated sample can still emit high-intensity UCL spectra in Fig. 4(d), and the shape and position of the emission peak have not changed, which confirms that the scheme of packaging and preservation with silicone rubber sheet is feasible and also shows the pictures of the encapsulated real sample as illustrated in Fig. 4(d).

Temperature as a physical parameter requires timeliness and accuracy. Thanks to LIR technology development, long-distance and non-contact temperature sensing can be realized in many extreme environments. Figure 5(a) shows the detective system of LIR thermometry, which can precisely control the temperature in the range of 333-573 K and real-time monitoring of sample temperature with high accuracy (0.1 K). In the previous work, materials with perovskite structure deliquescent in a short time can return to their original properties under high-temperature drying. In order to ensure that the sample is in a completely dry state, the sample shall be drying pretreatment by heating 573 K for 15 min before obtaining the temperature-dependent spectra. At the same time, to minimize the possible deliquescence, we measure the temperature-dependent spectrum during the cooling process rather than the heating process. By using the equipment above, the temperature-dependent emission spectra of Cs₃ErF₆ excited under 980 nm laser is shown in Fig. 5(b). As the temperature rises from 333 K to 573 K, the emission intensity at 655 nm, 673 nm decreases dramatically, whereas the intensity of 534 nm, 544 nm green-band emission decreases slightly. The position and peak shape of each emission band of the sample did not change as the temperature increased. Meanwhile, the integrated intensity of the main emission peaks is shown in Fig. 5(c,d). Utilizing the different responses of Cs₃ErF₆ visible peaks to temperature, two LIR modes between Stark energy level (⁴F_9/2(1), ⁴F_9/2(2)) and non-TCL level are used as temperature-detecting signals.

 

Fig. 5. (a) Schematic of the method to gain temperature-dependent UCL spectra and (b) the temperature-dependent UCL spectra of Cs₃ErF₆ from 333 K to 573 K. Temperature evolution of the luminescence intensity of (c) 655 nm/673 nm, (d) 534 nm/673 nm, respectively, Temperature-dependent LIR values of “rapid mode” (e) and “ultra-sensitive mode” (f), sensitivity as a function of temperature for rapid, ultra-sensitive mode are shown in (g) and (h), respectively.

Download Full Size | PDF

Many extreme environmental changes are rapid and dramatic, such as the heating of high voltage cables where a quick determination of internal situations is required, so temperature is time-sensitive as an important parameter. For this reason, this paper selects the strongest luminous intensity and single red-light band to evaluate the environment temperature in which the Cs₃ErF₆ sample stands. Detecting only one band of spectra is a rapid and convenient operation, and the strong signal response reduces the low signal-to-noise ratio associated with fast measurement spectra, so the name “rapid mode” is used to describe this method. The energy gap between these Stark levels is small, so when the temperature increases, the upper level can be thermally populated from the lower level, indicating that the two levels are thermally coupled, which allows us to use the relative intensity between these two peaks to determine the temperature by the LIR technique. As is shown in Fig. 5(c), the integrated intensity of 655 nm and 673 nm decrease during the heating process; in detail, 655 nm as the upper level can be repopulated by 673 nm lower level electrons during the heating process, which causes the emission intensity of 655 nm to decline more slowly than 673 nm. Since the two Stark energy levels are two closely spaced energy levels, the overlap of fluorescence peaks from two separate thermally coupled energy levels and the influence of excitation sources will become more prominent. Therefore, the LIR formula for Stark energy levels will have some changes compared with the formula for TCL energy levels. On the basis of previous studies, LIR can be expressed as follows [34–38]

The corresponding “rapid mode” sensitivities S_a and S_r of phosphor Cs₃ErF₆ under 980 nm excitation are displayed in Fig. 5(g). In the temperature range of our measurement, S_a, S_r initially increased and then decreased quickly during the temperature raising. According to our fitting results, the maximum absolute sensitivity S_amax= 0.0654%K^-1 is found at 300 K, while the maximum relative sensitivity S_rmax= 0.0102%K^-1 occurs at 300 K. In order to investigate the recyclability and temperature resolution in the temperature range (313-373 K) most susceptible to deliquescence, 50 tests are conducted at 313(RT), 343 K and 373 K, and the temperature resolution (δT) of the phosphor is calculated by the following formula [22,44]

To reveal the other thermometry strategy based on non-TCL energy level, the 534 nm and 673 nm integral intensity statistics chart is shown in Fig. 5(d). It is clear that the green emission intensity loss at 534 nm does not vary as drastically with temperature as the red emission at 673 nm. For this conventional strategy, the LIR value will conform to the following empirical equation instead of formula (2) with non-thermal coupling properties [45]

Unlike the phenomenon exhibited in Stark energy level, the sensitivity of non-TCL energy level temperature measurement is always decreasing with the increase of temperature (seen in Fig. 5(h)). It is noteworthy that the maximum value of S_a is 7.362%K⁻¹(at 333 K), while the maximum value of S_r is 0.621% K⁻¹ (at 333 K). Putting the ultra-sensitive modes of thermometers involved in this paper into Table S1 and comparing with the same type of thermometers, our synthesized samples have great advantages in both relative sensitivity and absolute sensitivity. So the outstanding advantage of this thermometry mode is a high sensitivity, for which the “ultra-sensitive mode” name is used. As far as we know, this kind of thermometer requiring extremely high sensitivity is usually popular in the field of cell biology, but it is still difficult to achieve clinical application due to the size of the particles, but we believe that it will be realized someday with continuous improvement of the synthesis process.

4. Conclusion

In summary, this work is the first to investigate the up-conversion and down-shifting photoluminescence properties under 980 nm excitation in perovskite-like Cs₃ErF₆ phosphors and discusses in detail the enormous impact of the deliquescence effect on the Cs₃ErF₆ material, while we innovatively propose our solution to prevent deliquescence - silicone rubber sheet encapsulation. The silicone rubber sheet encapsulation method protects the high luminescence characteristics and simplifies the operation process, which is a promising method to prevent sample deliquescence. Based on the unique luminescent properties, we have designed a dual-mode non-contact LIR thermometer. We also take into account the effects of deliquescence and carry out high-temperature pretreatment and measurement of the cooling process to minimize the effects of deliquescence. The “rapid mode” can judge the approximate temperature of an object only by using the spectrum of the red-light band. If a thermometer with higher sensitivity is required in some cases, we can simply expand the spectral measurement range and switch to the “ultra-sensitive mode” to obtain an ultra-sensitive thermometer with relative sensitivity of 0.621% K⁻¹ and absolute sensitivity 7.362%K⁻¹. It is believed that with deeper understanding of Cs-base fluoride, more and more interesting applications and solutions to deliquescence will make Cs-base fluoride a superior host material rivaling NaYF₄.