Multiple Temperature-Sensing Behavior of Green and Red Upconversion Emissions from Stark Sublevels of Er3+

Upconversion luminescence properties from the emissions of Stark sublevels of Er3+ were investigated in Er3+-Yb3+-Mo6+-codoped TiO2 phosphors in this study. According to the energy levels split from Er3+, green and red emissions from the transitions of four coupled energy levels, 2H11/2(I)/2H11/2(II), 4S3/2(I)/4S3/2(II), 4F9/2(I)/4F9/2(II), and 2H11/2(I) + 2H11/2(II)/4S3/2(I) + 4S3/2(II), were observed under 976 nm laser diode excitation. By utilizing the fluorescence intensity ratio (FIR) technique, temperature-dependent upconversion emissions from these four coupled energy levels were analyzed at length. The optical temperature-sensing behaviors of sensing sensitivity, measurement error, and operating temperature for the four coupled energy levels are discussed, all of which are closely related to the energy gap of the coupled energy levels, FIR value, and luminescence intensity. Experimental results suggest that Er3+-Yb3+-Mo6+-codoped TiO2 phosphor with four pairs of energy levels coupled by Stark sublevels provides a new and effective route to realize multiple optical temperature-sensing through a wide range of temperatures in an independent system.


Introduction
Optical temperature-sensing devices have been widely researched to promote their application in electrical power stations, oil refineries, coal mines, and fire detection, as they have been shown to overcome the interference of strong electromagnetic noise, hazardous sparks, or corrosive environments inaccessible to traditional temperature-measurement methods such as thermocouple detectors [1][2][3][4][5]. Sensors built based on the fluorescence intensity ratio (FIR) technique have attracted particular attention due to their ability to reduce dependence on measurement conditions and improve accuracy and resolution. FIR functions independent of fluorescence loss or fluctuations in excitation intensity can be applied to fluorescence systems in which two closely spaced energy levels with separations of the order of thermal energy are involved, following a Boltzmann-type population distribution [1,6,7]. Optical temperature sensors using the FIR technique are mainly focused on fluoride and oxides matrixes [8][9][10][11][12][13][14]. The fluoride matrixes possesses higher fluorescence efficiency and lower excitation power; however, the maximum operating temperature is usually low. On the contrary, the oxides matrices can operate at high temperature, although the fluorescence intensity is lower.
Upconversion emissions of rare earth ion-doped materials are typically utilized to realize FIR measurement because of the large amount of coupled energy levels in many rare earth ions and the easily accessible upconversion luminescence with near-infrared radiation from low-cost, commercially available diodes. Xu et al. [8], for example, reported the FIR of Ho 3+ using two blue emissions from coupled energy levels of 5 G 6 / 5 F 1 and 5 F 2,3 / 3 K 8 and found that Ho 3+ -Yb 3+ -codoped CaWO 4 possessed higher absolute sensitivity due to a larger energy gap between the thermally coupled 5 G 6 / 5 F 1 and 5 F 2,3 / 3 K 8 levels of Ho 3+ ions. The paired energy levels of 3 F 2 and 3 F 3 in Tm 3+ ions have also been used to investigate temperature-dependent red upconversion emissions and corresponding FIR properties [9]. The FIR properties of green upconversion emissions ascribed to paired energy levels of 2 H 11/2 and 4 S 3/2 in Er 3+ -doped materials, in particular, have been quite widely studied [10][11][12][13][14].
In addition to the intrinsic thermally coupled energy levels of rare earth ions, the pair energy levels of Stark sublevels can also be thermally coupled and used to investigate FIR versus temperature characteristics [15][16][17][18]. Baxter et al. [17], for example, used the coupled energy levels of 2 F 5/2(a) and 2 F 5/2(b) by Stark split of 2 F 5/2 levels in Yb 3+ ions to study FIR properties of Yb 3+ -doped silica fiber. Feng et al. [18] investigated the FIR properties of Er 3+ -doped fluoride glass using coupled Stark sublevels of 4 S 3/2(1) and 4 S 3/2 (2) in Er 3+ ions.
In this study, four thermally coupled energy levels of Er 3+ ions based on the Stark sublevels were simultaneously observed in Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphors. FIR properties of the four coupled energy levels from green and red emissions in Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphors were studied as a function of temperature in the range of 307-673 K. The effects of the energy gap of thermally coupled energy levels, FIR value, and upconversion emission intensity on the sensitivity and accuracy of the optical temperature sensor are discussed in an effort to explore potential developments in optical temperature-sensor technology based on different FIR routes in an independent system.

Experimental Section
The sol-gel method was used to prepare Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphors. The rare earth nitrates Er(NO 3 ) 3¨5 H 2 O (99.99%) and Yb(NO 3 ) 3¨5 H 2 O (99.99%) were purchased from Aladdin. Other chemicals including Iso-Propanol (i-PrOH), n-butyl titanate (Ti(OBu) 4 ), acetylacetone (AcAc), and concentrated nitric acid (HNO 3 ) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). All chemicals are of analytical reagent and were used without any further purification. i-PrOH was first added as a solvent to modified titanium(IV) n-butoxide by facilitating a chelating reaction between Ti(OBu) 4 and AcAc under agitation for 1 h at room temperature. Next, a mixture of deionized water, i-PrOH, and HNO 3 was slowly added into the solution. The mixed solution was stirred for 6 h to form a clear and stable sol. The molar ratios of Ti(OBu) 4  The codoped sols were dried at 373 K for 8 h to remove the solvent. The xerogels were then heated at a rate of 4 K/min and maintained at the sintering temperature of 1073 K for 1 h, then cooled to room temperature in the furnace. The sintered 2 mol % Er 3+ -20 mol % Yb 3+ -2 mol % Mo 6+ -codoped TiO 2 phosphors were finally milled into powders for structural analysis and spectral measurement.
The phase structures of Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphor samples were analyzed by SHIMADZU XRD-6000 X-ray diffractormeter (XRD) with Cu-Kα radiation. A homemade temperature control system, which was composed of a small stove and an intelligent digital-display-type temperature control instrument, was used to adjust sample temperature from 307 to 673 K, at measurement and control accuracy of about˘0.5 K. Temperature-dependent upconversion emissions from each sample were focused onto a Jobin Yvon iHr550 monochromator and detected with a CR131 photomultiplier tube by 976 nm laser diode (LD) excitation. The LD pump current varied from 0 to 2 A, and the spectral resolution of the experimental set-up was 0.1 nm. Figure 1 shows XRD patterns of the Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphor samples. The XRD pattern observed was characteristic of the anatase phase of TiO 2 (JCPDS No. 21-1272) and the face-centered cubic phase of Yb 2 Ti 2 O 7 (JCPDS No. 17-0454) referenced below. There was no diffraction peak of Mo compounds, and the main diffraction peak shifted toward small angles, indicating Mo 6+ stochastically located at the interstitial sites of the matrix lattice as a solution element.
Sensors 2015, 15, page-page 3 peak of Mo compounds, and the main diffraction peak shifted toward small angles, indicating Mo 6+ stochastically located at the interstitial sites of the matrix lattice as a solution element.  Figure 2 shows the upconversion emission spectra of Er 3+ -Yb 3+ -Mo 6+ -codoped TiO2 under different pump currents. Green and red upconversion emissions were observed in the wavelengths of 500-540 nm, 540-580 nm, and 620-710 nm, corresponding to 2 H11/2 → 4 I15/2, 4 S3/2 → 4 I15/2, and 4 F9/2 → 4 I15/2 transitions of Er 3+ ions, respectively. Each transition ( 2 H11/2 → 4 I15/2, 4 S3/2 → 4 I15/2, and 4 F9/2 → 4 I15/2) was divided into two emission peaks, which indicated 2 H11/2, 4 S3/2, and 4 F9/2 levels of Er 3+ split into three coupled Stark sublevels of 2 H11/2(I)·(HI) and 2 H11/2(II)·(HII), 4 S3/2(I)·(SI) and 4 S3/2(II)·(SII), and 4 F9/2(I)·(FI) and 4 F9/2(II)·(FII), respectively, due to the effect of crystal field environment on Er 3+ ions. As the LD pump current increased from 0.8 to 2.0 A, the position and number of upconversion emission peaks did not change, whereas the intensity of green and red emissions markedly increased due to the increase in excitation power.  The inset in Figure 2 shows the upconversion emission intensity ratios of H I /H II , S I /S II , F I /F II , and (H I + H II )/(S I + S II ) versus the pump current. All intensity ratios of H I /H II , S I /S II , F I /F II and (H I + H II )/(S I + S II ) increased alongside the pump current, implying that the nonradiative processes of Er 3+ in Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphor can partially transform pump energy into heat energy, therefore elevating the phosphor temperature. The temperature variation induced by increasing the pump current caused changes in the intensity ratio [19]; this suggests that the temperature-dependent intensity ratio for the four coupled energy levels of H I /H II , S I /S II , F I /F II , and (H I + H II )/(S I + S II ) can be utilized for optical temperature sensing. Figure 3 shows a schematic energy level diagram of the Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphors under 976 nm LD excitation. The upconversion mechanism of Er 3+ after the addition of Mo 6+ was reported in a previous study on the sensitization of the Yb 3+ -MoO 4 2´d imer to Er 3+ [20][21][22]. The inset in Figure 2 shows the upconversion emission intensity ratios of HI/HII, SI/SII, FI/FII, and (HI + HII)/(SI + SII) versus the pump current. All intensity ratios of HI/HII, SI/SII, FI/FII and (HI + HII)/(SI + SII) increased alongside the pump current, implying that the nonradiative processes of Er 3+ in Er 3+ -Yb 3+ -Mo 6+ -codoped TiO2 phosphor can partially transform pump energy into heat energy, therefore elevating the phosphor temperature. The temperature variation induced by increasing the pump current caused changes in the intensity ratio [19]; this suggests that the temperature-dependent intensity ratio for the four coupled energy levels of HI/HII, SI/SII, FI/FII, and (HI + HII)/(SI + SII) can be utilized for optical temperature sensing. Figure 3 shows a schematic energy level diagram of the Er 3+ -Yb 3+ -Mo 6+ -codoped TiO2 phosphors under 976 nm LD excitation. The upconversion mechanism of Er 3+ after the addition of Mo 6+ was reported in a previous study on the sensitization of the Yb 3+ -MoO4 2− dimer to Er 3+ [20][21][22]. Through a cooperative sensitization process in the Yb 3+ -MoO4 2− dimer, two excited Yb 3+ ions nonradiatively transfer their energy to MoO4 2− . This process is followed by a high excited state energy transfer (HESET) to the 4 F7/2 level of Er 3+ ions. After nonradiative relaxations from 4 F7/2 to the Stark sublevels of HI, HII, SI and SII, green upconversion emissions are produced by transitions of HI/HII/SI/SII → 4 I15/2. The nonradiative relaxation from SII to FI and FII levels and subsequent transitions of FI/FII → 4 I15/2 generate red emissions. In order to distinguish the effects of temperature from the pump current on the intensity ratio (Figure 2), the upconversion emission properties of Er 3+ -Yb 3+ -Mo 6+ -codoped TiO2 were measured under different temperatures. Figure 4 shows the upconversion emissions spectra of Er 3+ -Yb 3+ -Mo 6+codoped TiO2 at measured temperatures between 307 and 673 K. Changes in temperature had no influence on the bands of green and red emissions from 2 H11/2/ 4 S3/2 → 4 I15/2 and 4 F9/2 → 4 I15/2 transitions of Er 3+ between 500 to 580 nm and 620 to 700 nm, respectively; the intensity varied with temperature, however. The inset in Figure 4 shows the intensity of green and red emissions and the intensity ratio of green to red emissions as a function of temperature. The intensity of red emissions decreased with increasing temperature, in accordance with the classical theory of thermal quenching. Temperature-dependent intensity of the red emissions can be expressed as follows [23]: In order to distinguish the effects of temperature from the pump current on the intensity ratio (Figure 2), the upconversion emission properties of Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 were measured under different temperatures. Figure 4 shows the upconversion emissions spectra of Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 at measured temperatures between 307 and 673 K. Changes in temperature had no influence on the bands of green and red emissions from 2 H 11/2 / 4 S 3/2 Ñ 4 I 15/2 and 4 F 9/2 Ñ 4 I 15/2 transitions of Er 3+ between 500 to 580 nm and 620 to 700 nm, respectively; the intensity varied with temperature, however. The inset in Figure 4 shows the intensity of green and red emissions and the intensity ratio of green to red emissions as a function of temperature. The intensity of red emissions decreased with increasing temperature, in accordance with the classical theory of thermal quenching. Temperature-dependent intensity of the red emissions can be expressed as follows [23]: where T is the absolute temperature, and I(T) and I(0) are the fluorescence intensities at temperatures of T and 0 K, respectively; ∆E 1 is the activation energy, k is the Boltzmann constant, and A is a constant. The temperature-dependent intensity of red emissions fits well to Equation (1), where ∆E 1 (FI+FII) = 0.074 eV. Conversely, the intensity of green emissions increased with increasing temperature, which does not satisfy the classical theory of thermal quenching, likely due to the increased Yb 3+ absorption cross-section at elevated temperatures [22,24]. A general theoretical description of the green upconversion emission can be given by [22]: where B is a constant, and hν is the phonon energy participating in the multiphonon-assisted excitation. The dependence of green upconversion emissions on temperature fits well to Equation (2). The I green /I red value increased with temperature, causing the color to turn from red to green with elevated temperature. where T is the absolute temperature, and I(T) and I(0) are the fluorescence intensities at temperatures of T and 0 K, respectively; ΔE′ is the activation energy, k is the Boltzmann constant, and A is a constant. The temperature-dependent intensity of red emissions fits well to Equation (1), where ΔE′(FI+FII) = 0.074 eV. Conversely, the intensity of green emissions increased with increasing temperature, which does not satisfy the classical theory of thermal quenching, likely due to the increased Yb 3+ absorption crosssection at elevated temperatures [22,24]. A general theoretical description of the green upconversion emission can be given by [22]: where B is a constant, and hν is the phonon energy participating in the multiphonon-assisted excitation. The dependence of green upconversion emissions on temperature fits well to Equation (2). The Igreen/Ired value increased with temperature, causing the color to turn from red to green with elevated temperature. According to previous research [1], the relative population of two "thermally coupled" energy levels with separation of the order of thermal energy follows a Boltzmann-type population distribution, causing variation in the transitions of two closely spaced levels at elevated temperature if pumped through a continuous light source. After populations are thermalized at two closely spaced levels, the FIR of upconversion emissions (R) related to the transitions of both levels can be written as follows: where Iupper, Ilower, Nupper, and Nlower are the fluorescence intensity and number of ions for the upper and lower thermalizing energy levels, respectively; ΔE is the energy gap between two coupled levels, and C is a constant relative to the degeneracy, emission cross-section, and angular frequency of corresponding transitions. Equation (3) suggests that FIR is related to the energy gap ΔE and temperature T. Figure 5 shows FIR plots of (HI + HII)/(SI + SII), HI/HII, SI/SII, and FI/FII as a function of  (1) and (2).
According to previous research [1], the relative population of two "thermally coupled" energy levels with separation of the order of thermal energy follows a Boltzmann-type population distribution, causing variation in the transitions of two closely spaced levels at elevated temperature if pumped through a continuous light source. After populations are thermalized at two closely spaced levels, the FIR of upconversion emissions (R) related to the transitions of both levels can be written as follows: where I upper , I lower , N upper , and N lower are the fluorescence intensity and number of ions for the upper and lower thermalizing energy levels, respectively; ∆E is the energy gap between two coupled levels, and C is a constant relative to the degeneracy, emission cross-section, and angular frequency of corresponding transitions. Equation (3) suggests that FIR is related to the energy gap ∆E and temperature T. Figure 5 shows FIR plots of (H I + H II )/(S I + S II ), H I /H II , S I /S II , and F I /F II as a function of inverse absolute temperature from 307 to 673 K. The inset shows corresponding upconversion emission intensity and the intensity ratio relative to temperature. The experimental data fits well to Equation (3). Energy gaps ∆E of the four coupled energy levels of (H I + H II )/(S I + S II ), H I /H II , S I /S II , and F I /F II are calculated in Table 1. The decreased intensity of two red emissions with elevated temperature, shown in the inset of Figure 5d, can also be fitted to Equation (1). The activation energy of F I and F II levels is calculated as ∆E 1 FI = 0.069 eV and ∆E 1 FII = 0.080 eV, which is consistent with the average activation energy of (F I + F II ) level (∆E 1 (FI+FII) = 0.074 eV) shown in Figure 4.
Sensors 2015, 15, page-page 6 inverse absolute temperature from 307 to 673 K. The inset shows corresponding upconversion emission intensity and the intensity ratio relative to temperature. The experimental data fits well to Equation (3). Energy gaps ΔE of the four coupled energy levels of (HI + HII)/(SI + SII), HI/HII, SI/SII, and FI/FII are calculated in Table 1. The decreased intensity of two red emissions with elevated temperature, shown in the inset of Figure 5d, can also be fitted to Equation (1). The activation energy of FI and FII levels is calculated as ΔE′FI = 0.069 eV and ΔE′FII = 0.080 eV, which is consistent with the average activation energy of (FI + FII) level (ΔE′(FI+FII) = 0.074 eV) shown in Figure 4.   Table 1. Energy gap of coupled energy levels ΔE, pre-exponential factor C, maximum sensitivity Smax, temperature of maximum sensitivity Tmax and upconversion emission intensity for the four coupled energy levels of (HI + HII)/(SI + SII), HI/HII, SI/SII and FI/FII. For optical temperature-sensing applications, it is crucial to know the rate at which the FIR changes with temperature, known as the absolute sensitivity Sa, which is expressed as follows [1]:

Coupled Energy Levels (HI + HII)/(SI + SII) HI/HII SI/SII FI/FII
Equation (4) makes clear that the appropriate selection of two thermally coupled energy levels with a suitable energy difference ΔE is very important. Larger ΔE benefits absolute sensitivity and accurate measurement of emission intensity, due to the decrease of fluorescence peak overlap originating from the two individual thermally coupled energy levels. Knowing this, the absolute sensitivity Sa when using coupled energy levels of (HI + HII)/(SI + SII) (with the largest possible ΔE = 0.0558 eV) is higher than those using the other three coupled levels, as shown in Table 1. The energy gap ΔE must be not too large, though, or thermalization no longer occurs.
Considering practical applications, it is extremely useful to be aware of variations in sensitivity with temperature. Relative sensitivity Sr is expressed [25]: dT kT   (5) Compared to absolute sensitivity Sa, relative sensitivity Sr is dependent on not only energy gap ΔE, but also the intensity ratio FIR. Equation (3) indicates that larger FIR causes larger C. Thus, larger ΔE and FIR (or C) contribute to higher Sr. Table 1 also shows pre-exponential factor C values for the four pair energy levels (HI + HII)/(SI + SII), HI/HII, SI/SII, and FI/FII. The coupled energy levels of (HI + HII)/(SI + SII) processed larger relative sensitivity Sr than those of HI/HII, FI/FII, or SI/SII. Sr as a function  Table 1. Energy gap of coupled energy levels ∆E, pre-exponential factor C, maximum sensitivity S max , temperature of maximum sensitivity T max and upconversion emission intensity for the four coupled energy levels of (H I + H II )/(S I + S II ), H I /H II , S I /S II and F I /F II . For optical temperature-sensing applications, it is crucial to know the rate at which the FIR changes with temperature, known as the absolute sensitivity S a , which is expressed as follows [1]:

Coupled Energy Levels (H I + H II )/(S I + S II ) H I /H II S I /S II F I /F II
Equation (4) makes clear that the appropriate selection of two thermally coupled energy levels with a suitable energy difference ∆E is very important. Larger ∆E benefits absolute sensitivity and accurate measurement of emission intensity, due to the decrease of fluorescence peak overlap originating from the two individual thermally coupled energy levels. Knowing this, the absolute sensitivity S a when using coupled energy levels of (H I + H II )/(S I + S II ) (with the largest possible ∆E = 0.0558 eV) is higher than those using the other three coupled levels, as shown in Table 1. The energy gap ∆E must be not too large, though, or thermalization no longer occurs.
Considering practical applications, it is extremely useful to be aware of variations in sensitivity with temperature. Relative sensitivity S r is expressed [25]: Compared to absolute sensitivity S a , relative sensitivity S r is dependent on not only energy gap ∆E, but also the intensity ratio FIR. Equation (3) indicates that larger FIR causes larger C. Thus, larger ∆E and FIR (or C) contribute to higher S r . Table 1 also shows pre-exponential factor C values for the four pair energy levels (H I + H II )/(S I + S II ), H I /H II , S I /S II , and F I /F II . The coupled energy levels of (H I + H II )/(S I + S II ) processed larger relative sensitivity S r than those of H I /H II , F I /F II , or S I /S II . S r as a function of temperature for the four coupled energy levels calculated by Equation (5) is shown in Figure 6, in accordance with the above results in the measured temperature range 307-673 K.
Sensors 2015, 15, page-page 8 of temperature for the four coupled energy levels calculated by Equation (5) is shown in Figure 6, in accordance with the above results in the measured temperature range 307-673 K. Maximum sensitivity Smax and temperature Tmax, at which the sensor has maximum sensitivity Smax, are of utmost importance because these two parameters indicate the highest sensitivity properties and optimum operating temperature range of optical thermal sensors. According to Equation (5) (7) Equation (6) indicates that a larger pre-exponential factor C and smaller energy difference ΔE of coupled energy levels help to increase Smax. Equation (7) shows that Tmax is relative to the energy difference ΔE, in which the sensor with a larger ΔE has a higher Tmax. Smax and Tmax for the four coupled energy levels are shown in Table 1. The highest Tmax was found for (HI + HII)/(SI + SII) coupled energy levels used for thermal sensing, due to a larger ΔE. The relatively larger C and smallest ΔE in FI/FII coupled energy levels used for thermal sensing resulted in the highest sensitivity Smax.
Temperature measurement error can be calculated using the relation [8,26]: Larger Sr and smaller ΔR imply better accuracy. As shown in Figure 6, larger Sr at a higher temperature for coupled energy levels of (HI + HII)/(SI + SII) led to a better accuracy in the high temperature range. Likewise, better accuracy can be expected in the low temperature range using HI/HII, SI/SII and FI/FII coupled energy levels for thermal sensing.
The separation of two coupled energy levels ΔE should be large enough to avoid overlap of the two fluorescence emissions and to produce efficient luminescence for feasible and accurate intensity measurement. The efficient luminescence of Er 3+ -doped materials also contributes to the ready detection of luminescence and ΔR accuracy, where only low excitation power is needed. Table 1  Maximum sensitivity S max and temperature T max , at which the sensor has maximum sensitivity S max , are of utmost importance because these two parameters indicate the highest sensitivity properties and optimum operating temperature range of optical thermal sensors. According to Equation (5), S max and T max can be calculated by dS r {dT " 0 as follows: Equation (6) indicates that a larger pre-exponential factor C and smaller energy difference ∆E of coupled energy levels help to increase S max . Equation (7) shows that T max is relative to the energy difference ∆E, in which the sensor with a larger ∆E has a higher T max . S max and T max for the four coupled energy levels are shown in Table 1. The highest T max was found for (H I + H II )/ (S I + S II ) coupled energy levels used for thermal sensing, due to a larger ∆E. The relatively larger C and smallest ∆E in F I /F II coupled energy levels used for thermal sensing resulted in the highest sensitivity S max .
Temperature measurement error can be calculated using the relation [8,26]: Larger S r and smaller ∆R imply better accuracy. As shown in Figure 6, larger S r at a higher temperature for coupled energy levels of (H I + H II )/(S I + S II ) led to a better accuracy in the high temperature range. Likewise, better accuracy can be expected in the low temperature range using H I /H II , S I /S II and F I /F II coupled energy levels for thermal sensing.
The separation of two coupled energy levels ∆E should be large enough to avoid overlap of the two fluorescence emissions and to produce efficient luminescence for feasible and accurate intensity measurement. The efficient luminescence of Er 3+ -doped materials also contributes to the ready detection of luminescence and ∆R accuracy, where only low excitation power is needed. Table 1 shows where (H I + H II )/(S I + S II ) coupled energy levels had the highest accuracy of all samples, due to a larger ∆E and the strongest luminescence intensity; conversely, S I /S II coupled energy levels had the lowest accuracy, evidenced by a smaller ∆E and the lowest luminescence intensity, which are altogether consistent with the results shown in Figure 5.

Conclusions
The green and red upconversion emissions by transitions of Er 3+ Stark sublevels were observed in Er 3+ -Yb 3+ -Mo 6+ -codoped TiO 2 phosphors in this study. There are four coupled energy levels of Er 3+ ions due to the effect of the crystal field environment on Er 3+ , each of which was utilized to study temperature-dependent upconversion emission properties. Based on the FIR technique, the optical temperature-sensing behaviors of sensing sensitivity, measurement error, and operating temperature for the four coupled energy levels were discussed in detail, with all closely related to the energy gap of the coupled energy levels, FIR value, and luminescence intensity. High sensitivity and negligible error are obtainable through the use of different coupled energy levels for optical sensing, throughout a wide range of temperature in an independent system. The utilization of coupled energy levels by Stark split is a new and effective method in the realization of multiple optical temperature measurement.