Introduction

Persistent luminescence, also called afterglow, is a special optical phenomenon in which a material continues emitting for an appreciable time, from minutes to hours, after the stoppage of excitation1, 2. Persistent phosphors emitting in the visible spectral region have been extensively studied, and some visible persistent phosphors, such as green-emitting SrAl2O4:Eu2+,Dy3+ (ref. 3) and blue-emitting Sr2MgSi2O7:Eu2+,Dy3+ (ref. 4), have achieved commercial success and are widely used as night-vision materials in a variety of areas. In recent years, persistent luminescence in wavelengths beyond the visible spectral region, that is, in the near-infrared (NIR; ~700–900 nm) and the short-wave infrared (SWIR; 900–1700 nm) spectral regions, has received considerable attention because longer-wavelength persistent luminescence has many promising advanced applications, ranging from infrared night-vision surveillance to biomedical imaging. For example, for infrared night-vision surveillance, NIR light and SWIR light are invisible to the naked eye (need to be detected by NIR or SWIR detection devices), and SWIR light has high penetrating power in haze, fog, smoke and dust5. For biomedical imaging, NIR light and SWIR light correspond to the first ‘biological window’ (~650–950 nm) and the second ‘biological window’ (~1000–1400 nm)6, 7, respectively, producing low optical scattering and absorption in tissues (especially for SWIR light) and thus enabling deep-tissue imaging8, 9, 10. More importantly, the ‘self-illuminating’ feature of persistent luminescence allows the imaging to be conducted in an excitation-free and, hence, autofluorescence-free manner, enabling a high signal-to-noise ratio and exceptional imaging sensitivity11. As a result of extensive research, an array of Cr3+-activated gallate-based NIR persistent phosphors (for example, Zn3Ga2Ge2O10:Cr3+ and LiGa5O8:Cr3+)12, 13, 14, 15, 16, emitting at 690–750 nm with a very-long afterglow of >100 h, were recently discovered and synthesized and quickly found exciting applications in biomedical imaging17, 18. In contrast, however, reliable persistent phosphors emitting in the SWIR spectral region are lacking. A few phosphors were found to emit short-duration SWIR afterglow but were accompanied by dominant visible persistent luminescence from the dopants or the host compositions19, 20.

Trivalent ytterbium (Yb3+) has only two manifolds in the 4f shell, that is, a 2F7/2 ground state and a 2F5/2 excited state, with an energy difference between the states of ~10 000 cm−1. Typical emission for Yb3+-activated phosphors features a fine-structure spectral shape owing to the Stark splitting of the two Yb3+ manifolds, as shown schematically in the inset of Figure 1a. This simple and unique energy-level structure endows Yb3+ with several fascinating optical properties, including no absorption in the visible range, strong absorption near 980 nm (well-suited for InGaAs diode laser emission), and intense emission at 950–1150 nm (2F5/22F7/2 transition; in the SWIR range)21. Accordingly, the Yb3+ ion has been extensively studied over the past two decades for use in a variety of important applications, for instance, as the emitter in InGaAs diode-pumped solid-state SWIR lasers22, 23, as the sensitizer (by absorbing 980 nm photons) to activate other lanthanide ions (for example, Er3+ ion) for up-conversion luminescence24, 25, and as the spectral converter in Ln3+-Yb3+ doubly doped quantum cutting phosphors for high-efficiency Si photovoltaics (the emission of Yb3+ at ~1000 nm perfectly matches the maximum spectral response of Si solar cells)26, 27. In addition to the unique energy-level structure, Yb3+ also has a unique 4f13 electronic configuration, which can easily gain one electron to reach the more stabilized 4f14 configuration of the full shell. This tendency of reduction to Yb2+ enables the Yb3+ ion in some hosts, such as oxides28, fluorides28, oxysulfides29 and nitrides30, to receive an electron from the host’s anion under high-energy external excitation (for example, ultraviolet (UV) light), forming a charge transfer state (CTS). The CTS of Yb3+ can transfer the excitation energy to the 2F5/2 emitting state via a non-radiative relaxation process (Figure 1a), giving SWIR photoluminescence at around 1000 nm.

Figure 1
figure 1

Energy diagram of the Yb3+ ion and photoluminescence spectra of the MgGeO3:Yb3+ phosphor. (a) Configuration coordinate diagram of the Yb3+ charge transfer state (CTS) and Yb3+ 4f ground (2F7/2) and excited (2F5/2) states. The inset is the Stark energy level diagram of Yb3+. (b) Normalized photoluminescence excitation and emission spectra of MgGeO3:Yb3+ phosphor at room temperature. The emission spectrum is acquired under 300-nm light excitation, and the excitation spectrum is obtained by monitoring the 1019-nm emission.

Here we report on a new function of the Yb3+ ion—as an emitter for SWIR persistent luminescence. We have developed the first real SWIR persistent phosphor, Yb3+-doped MgGeO3, which exhibits a very long SWIR persistent luminescence at around 1000 nm for more than 100 h after UV light excitation. The MgGeO3:Yb3+ phosphor is a promising UV-to-SWIR spectral convertor as well as a superb storage phosphor, exhibiting a photostimulated persistent luminescence capability.

Materials and methods

Materials synthesis

The MgGeO3:Yb3+ phosphor was fabricated using a high-temperature solid-state reaction technique. Stoichiometric amounts of MgO and GeO2 powders and 0.05 mol% Yb2O3 powder (the corresponding Yb3+ concentration is 0.1 mol%) were thoroughly mixed and ground in an agate mortar. The mixed powder was pressed into preformed discs (~15 mm in diameter and ~1 mm in thickness) using a 16-T dry pressing machine. The discs were then sintered at 1250 °C in air for 2 h to form solid ceramic discs.

Characterization methods

The crystal structure of the as-synthesized MgGeO3:Yb3+ phosphor was characterized on a PANalytical X’Pert PRO powder X-ray diffractometer (PANalytical Inc., Westborough, MA, USA) with Cu Kα1 radiation (λ=1.5406 Å). The spectral properties, including the photoluminescence emission and excitation spectra, persistent luminescence decay curves, persistent luminescence excitation spectra and photostimulated persistent luminescence (PSPL) properties, of the phosphor were analyzed using a Horiba FluoroLog-3 spectrofluorometer (Horiba Scientific, Edison, NJ, USA) equipped with a 450-W xenon arc lamp, a photomultiplier tube (measurement range, 240–850 nm) and a InGaAs detector (measurement range, 800–1600 nm). The thermoluminescence spectra were recorded using a homemade thermoluminescence setup (temperature range, 20–280 °C; heating rate, 4 °C per second). Before all the spectral measurements, the samples were heat-treated in a furnace at 450 °C to completely empty the energy traps in the material. A 4-W, 254-nm UV lamp was also used to excite the samples. A white light emitting diode (LED) flashlight (with an output of ~1 mW cm2) and a 100-mW 808-nm NIR laser were used as the illumination sources in the PSPL measurements.

Results and discussion

X-ray diffraction measurements show that the as-synthesized MgGeO3:0.1 mol% Yb3+ phosphor is ilmenite structured MgGeO3 (Supplementary Fig. S1). Although the ionic radius of the Yb3+ ion (0.087 nm for six coordination) is larger than that of the Mg2+ ion (0.072 nm for six coordination), the Yb3+ ion most probably occupies the Mg2+ site in the MgGeO3 lattice. The substitution of Mg2+ with Yb3+ distorts the lattice and requires secondary impurities to offer local charge compensation. Therefore, the Yb3+ concentration in MgGeO3 is limited to be very low (<0.5 mol%) (Supplementary Fig. S1).

Figure 1b shows the normalized photoluminescence emission and excitation spectra of MgGeO3:Yb3+ phosphor at room temperature. Under the excitation of 300 nm light, the material exhibits an intense photoluminescence at 930–1150 nm, featuring a fine spectral structure with two main peaks at 980 and 1019 nm and several weak peaks at 1020–1100 nm, which correspond to different optical transitions between the Stark levels of the 2F5/2 and 2F7/2 states. The excitation spectrum (monitoring 1019 nm emission) covers only the UV spectral region from 250 to 400 nm (250 nm is the measurement limit of our spectrofluorometer). The high-energy side (<260 nm) of the excitation spectrum may be assigned to the fundamental absorption edge of the host, while the excitation shoulder on the low-energy side (300–400 nm) is attributed to the absorption of CTS28, which can efficiently transfer excitation energy to the 2F5/2 emitting states, followed by the 2F5/22F7/2 transition. Figure 1b also shows that there is no excitation or emission from either the Yb3+ emitter or the MgGeO3 host in the visible–NIR spectral region (~400–900 nm), indicating that the MgGeO3:Yb3+ phosphor is spectrally transparent to the visible and NIR light and that the MgGeO3:Yb3+ phosphor is a promising UV-to-SWIR spectral convertor.

When the UV excitation ceased, very long-lasting persistent luminescence in the 930–1150 nm range was observed in the MgGeO3:Yb3+ phosphor. Figure 2a shows the persistent luminescence decay curve of the MgGeO3:Yb3+ phosphor monitoring at 1019 nm after the sample was irradiated with a 254-nm UV lamp for 15 min. The intensity of the persistent emission decreases quickly in the first half an hour and then slowly over time. After 6 h of decay, the intensity of the persistent emission is still considerably high. The inset of Figure 2a shows a persistent luminescence emission spectrum acquired at 30 min after the removal of the excitation. The persistent luminescence emission spectrum shares an identical profile with the photoluminescence emission spectrum (Figure 1b), indicating that the SWIR persistent luminescence originates from the Yb3+ emitting centers. Remarkably, the Yb3+ persistent luminescence can be recorded even after 100 h of decay (Supplementary Fig. S2), showing that the SWIR persistence time of the MgGeO3:Yb3+ phosphor is longer than 100 h. We also studied the effects of excitation duration on the charging capability and the persistent luminescence performance of the phosphor. Within 15 min of excitation using a 254-nm UV lamp, there is a direct relationship between the persistent luminescence intensity and the excitation duration; over 15 min of excitation, the material appears to be fully charged (Supplementary Fig. S3).

Figure 2
figure 2

Persistent luminescence properties of MgGeO3:Yb3+ phosphor at room temperature. (a) Persistent luminescence decay curve monitored at 1019 nm after irradiation by a 254-nm UV lamp for 15 min. The inset shows the persistent luminescence emission spectrum recorded at 30 min after stopping the irradiation. (b) Persistent luminescence excitation spectrum obtained by plotting the persistent luminescence intensity (I30s) monitored at 1019 nm as a function of the excitation wavelengths over the 260–450 nm spectral range. The sample was irradiated for 5 min at each measured wavelength using a xenon arc lamp. The dash-line curve shows the photoluminescence excitation spectrum monitored at 1019 nm.

We also determined the effectiveness of different excitation energies in producing SWIR persistent luminescence in MgGeO3:Yb3+ by studying the relationship between the intensity of persistent luminescence and the excitation wavelength. In our experiments, we excited the material using different wavelengths between 260 and 450 nm in 5 nm steps, and acquired the persistent luminescence decay curve (monitoring at 1019 nm emission) after each excitation (Supplementary Fig. S4). The intensity of the persistent emission at a decay time of 30 s (I30s; indicated by the vertical line in Supplementary Fig. S4) was used as the reference point to plot the persistent luminescence excitation spectrum, as shown in Figure 2b. For comparison, the photoluminescence excitation spectrum in Figure 1b is re-displayed in Figure 2b. The excitation spectrum for persistent luminescence is different from that for photoluminescence. The SWIR persistent luminescence in MgGeO3:Yb3+ can only be achieved by UV irradiation with wavelengths shorter than ~320 nm (see inset of Figure 2b; showing that the onset wavelength for achieving persistent luminescence is at ~320 nm) and the effectiveness increases sharply as the excitation moves to shorter wavelengths. The irradiation with wavelengths between 320 and 400 nm, which can effectively produce SWIR photoluminescence via the excitation of the CTS of Yb3+ (Figure 1a), does not produce persistent luminescence in MgGeO3:Yb3+ because the Yb3+ persistent luminescence in MgGeO3 has a different excitation mechanism from Yb3+ photoluminescence.

Under ambient conditions, we observed that UV irradiation can cause the body color of MgGeO3:Yb3+ discs to change from white to beige, as the digital picture shown in Figure 3. This coloration phenomenon is long-lived, taking several days for the discs to restore their original white body color, and the coloration can repeatedly occur upon UV irradiation. The occurrence of coloration in MgGeO3:Yb3+ can be ascribed to the creation of a photochromic center31, 32, which may arise from the trapping of photogenerated electrons by lattice defects, for example, oxygen vacancies, in the host (note: oxygen vacancies were identified as effective electron trapping centers in the MgGeO3 host33. Moreover, undoped MgGeO3 also exhibits weak coloration phenomenon). Our previous study on LiGa5O8:Cr3+ NIR persistent phosphor13, which also exhibits similar coloration phenomenon, has revealed that the photochromic centers in persistent phosphors may serve as deep energy traps for photogenerated electrons. To understand the nature of photochromic centers and the distribution of energy traps in MgGeO3:Yb3+, we conducted thermoluminescence measurements at 25–230 °C by monitoring the 1019-nm emission on MgGeO3:Yb3+ samples undergoing different decay times from 10 min to 100 h after excitation, as the thermoluminescence curves displayed in Figure 4a. For the case of 10 min short decay, the thermoluminescence curve exhibits two broad bands with band maxima at ~100 °C (low-temperature band) and ~200 °C (high-temperature band). With the increase of decay time, the intensity of the low-temperature band decreases sharply, and the lower-temperature side of the band decreases, resulting in the band maximum moving to higher temperature. After 12 h of decay, most of the low-temperature band disappears. For the high-temperature band, however, while its intensity gradually decreases with decay time, the position of the band maximum almost remains unchanged. After 100 h of decay, the high-temperature band is still significant, meaning that a considerable amount of trapped electrons were still not released after 100 h of natural decay. The thermoluminescence results are therefore consistent with the persistent luminescence emission measurements shown in Supplementary Fig. S2.

Figure 3
figure 3

UV-irradiation-induced coloration of the MgGeO3:Yb3+ phosphor. Digital picture of an MgGeO3:Yb3+ phosphor disc. The right side of the disc was irradiated by a 254-nm UV lamp for 5 min, while the left side was covered by a piece of paper. UV irradiation causes the body color to change from white to beige.

Figure 4
figure 4

Thermoluminescence curves and trap depths of the MgGeO3:Yb3+ phosphor. (a) Thermoluminescence curves monitored at 1019 nm emission over 20–230 °C for samples undergoing different decay times from 10 min to 100 h. The samples were pre-irradiated by a 254-nm UV lamp for 15 min. The red curves at the low temperature side are fit using an initial rise analysis method (see text for further details). (b) Plot of the trap depths as a function of decay times. The trap depths were obtained from the fittings in a.

Because the spectral shape of a thermoluminescence curve is correlated to trap distribution34, the change of the thermoluminescence curve profiles with decay time in Figure 4a shows the emptying process of traps in material during the persistent luminescence process, that is, the shallow traps are emptied first, followed by the emptying of the deep traps. Based on the thermoluminescence curves in Figure 4a, we determined the trap depths corresponding to different decay times in MgGeO3:Yb3+ using an initial rise analysis method35, in which the shallowest trap occupied at each decay time was evaluated by fitting the low-temperature sides of the thermoluminescence curves using I(T)=Cexp(−ΔE/k(T+273)), where I is the thermoluminescence intensity, T is the temperature, ΔE is the trap depth, C is a fitting constant and k is the Boltzmann constant. The fittings are shown as the red curves in Figure 4a (the fittings are also drawn in ln(I) versus 1/T coordinates, as shown in Supplementary Fig. S5). The obtained trap depths (ΔE) are plotted as a function of decay times, as the ball curve shown in Figure 4b. As the decay time increases from 10 min to 6 h, the trap depth increases from ~0.66 to~0.78 eV. The gradual deepening of the trap depth along with the decay time indicates the presence of a continuous trap distribution in MgGeO3:Yb3+ (ref. 35). When the decay time is over 6 h, the trap depth remains almost unchanged at ~0.78 eV. This constant trap depth is a strong indicator of the presence of deep traps corresponding to the photochromic centers13, whose ground state is located at ~0.78 eV below the conduction band of MgGeO3. Therefore, there are two types of traps in MgGeO3:Yb3+, that is, the shallow traps with continuous trap distribution and the high-energy photochromic-center-type deep traps.

To further understand the photochromic centers and their role in persistent luminescence, we conducted thermoluminescence measurements after photostimulation on 120-h-decayed MgGeO3:Yb3+ samples, as shown in Figure 5a. The photostimulation technique has been routinely used to investigate the photochromic centers in storage phosphors36 and was recently used by us to study the photochromism in LiGa5O8:Cr3+ NIR persistent phosphor13. The solid-line curve and dash-line curve in Figure 5a show the thermoluminescence spectra of a 120-h-decayed MgGeO3:Yb3+ sample with and without white LED flashlight illumination (for 120 s), respectively (the emission spectrum of the flashlight is shown in Supplementary Fig. S6). The white LED illumination significantly modifies the thermoluminescence spectrum profile: the intensity of the high-temperature band decreases, and part of the low-temperature band reappears. Because low-energy white light cannot directly fill the energy traps (including the shallow traps) in MgGeO3:Yb3+ (Figure 2b), the changes in the thermoluminescence curve profiles indicate that the white LED illumination causes the remaining trapped electrons (due to UV pre-irradiation) in the material to redistribute: some electrons in the deep traps are photo-liberated, while some shallow traps are refilled.

Figure 5
figure 5

Photostimulation of the MgGeO3:Yb3+ phosphor. (a) Thermoluminescence curves monitored at 1019-nm emission over 20–250 °C on 120-h-decayed samples with (solid-line curve) and without (dash-line curve) white LED stimulation. The samples were pre-irradiated by a 254-nm UV lamp for 15 min. The white LED stimulation time was 120 s. (b) PSPL decay curves monitored at 1019 nm. The brown curve was acquired on a 120-h-decayed sample (pre-irradiated by a 254-nm UV lamp for 15 min), while the gray curve was recorded on a fresh sample (without UV pre-irradiation). The inset is the PSPL emission spectrum of the 120-h-decayed sample, which was recorded at 1 min after stopping the stimulation.

The photostimulation-induced refilling of the shallow traps is expected to enhance the persistent luminescence intensity of the decayed MgGeO3:Yb3+ phosphor at room temperature13. When a 120-h-decayed MgGeO3:Yb3+ disc was illuminated with a white LED flashlight for 120 s, the persistent luminescence was rejuvenated and the persistent luminescence intensity was significantly enhanced, as the persistent luminescence decay curve (brown curve) shown in Figure 5b (note that this phenomenon also occurs with red light (>590 nm) or NIR light (for example, 808 nm) illumination, as shown in Supplementary Fig. S7). This low-energy light illumination-induced persistent luminescence has been called PSPL13. The inset of Figure 5b shows the PSPL emission spectrum of the 120-h-decayed MgGeO3:Yb3+ sample recorded at 1 min after stopping the stimulation. The PSPL emission spectrum has an identical profile to the photoluminescence emission spectrum (Figure 1b) and the persistent luminescence emission spectrum (inset of Figure 2b). In PSPL, the essential excitation source is UV pre-irradiation. The low-energy white light (or red light or NIR light) illumination just triggers the liberation of the trapped electrons in deep traps, leading to subsequent electron transfer to shallow traps for enhanced and prolonged persistent luminescence. To show the critical role of UV pre-irradiation in the generation of the enhanced SWIR PSPL signal, we also performed the photostimulation experiment on a fresh MgGeO3:Yb3+ sample (that is, without UV pre-irradiation), as shown in the gray curve in Figure 5b. As expected, no persistent luminescence signal was detected in the fresh sample. The PSPL phenomenon observed in MgGeO3:Yb3+ phosphor has promising implications for biomedical imaging. For example, the PSPL capability enables the MgGeO3:Yb3+ phosphor (in the form of nanoparticles) pre-irradiated by UV light in vitro to be activated in vivo by high-tissue-penetrating low-energy light, giving rise to an enhanced SWIR PSPL signal that can greatly expand the imaging duration17.

On the basis of the above results, we propose a model to interpret the SWIR persistent luminescence and the SWIR PSPL mechanisms in MgGeO3:Yb3+, as schematically shown in Figure 6. In Figure 6, the band energy of the MgGeO3 host37, the energy level of the CTS of Yb3+ and the depth of the deep traps obtained in this study are drawn to scale. To simplify the discussion, we assign the shallow traps with continuous trap distribution and the filled deep traps (that is, the photochromic centers) as TRAP-1 and TRAP-2, respectively. Under high-energy UV excitation (for example, 254 nm), the traps (TRAP-1 and TRAP-2) are directly filled via electron excitation from the top of the valence band (process 1). This excitation transition is allowed in principle because the traps are oxygen-vacancy type33, whose electronic composition is generally identical to that of the conduction-band bottom (the oxygen-vacancy type traps are further testified in the Pr3+-activated MgGeO3 SWIR persistent phosphor, as shown in the results in Supplementary Fig. S8). In the persistent luminescence process, the trapped electrons recombine with the CTS (process 2). The excited state of Yb3+ is subsequently populated via an intersystem crossing process (process 3), followed by the persistent luminescence emission from 2F5/2 to 2F7/2 (process 4). For the PSPL process (right panel in Figure 6), after an initial decay at room temperature, TRAP-1 is thermally emptied. With the stimulation of low-energy light (for example, white light), some trapped electrons in TRAP-2 are liberated and promoted to the conduction band, resulting in the refilling of the depleted TRAP-1. As a result, the persistent luminescence signal is enhanced, that is, PSPL in the SWIR occurs.

Figure 6
figure 6

A schematic representation of SWIR persistent luminescence and SWIR PSPL mechanisms in MgGeO3:Yb3+. The band energy of the MgGeO3 host, the energy level of the CTS of Yb3+ and the depth of the deep traps are drawn to scale. The straight-line arrows and curved-line arrows represent optical transitions and electron transfer processes (see text for further details), respectively.

Finally, besides the MgGeO3:Yb3+ phosphor, SWIR persistent luminescence was also obtained in Yb3+-doped gallate-based host materials, such as LiGa5O8, Zn3Ga2GeO8 and La3Ga5GeO14, as shown in Supplementary Fig. S9 (these gallates are also excellent hosts for Cr3+ NIR persistent luminescence12, 13, 14, 15).

Conclusions

We found a new function for the Yb3+ ion—as an emitter for SWIR persistent luminescence—and developed the first real SWIR persistent phosphor, MgGeO3:Yb3+. The MgGeO3:Yb3+ phosphor exhibits SWIR persistent luminescence at around 1000 nm, with a long persistence time of more than 100 h. UV-light pre-irradiated MgGeO3:Yb3+ samples also exhibit a photostimulated persistent luminescence capability after short stimulation by low-energy white or red light. The Yb3+-activated SWIR persistent phosphors are expected to have applications in biomedical imaging, night-vision surveillance and Si photovoltaics.