Co-doped crystals and ceramics that contain a sensitizer (e.g., ${\mathrm{Yb}}^{3+}$ ions) and several activators (e.g., ${\mathrm{Er}}^{3+}$, ${\mathrm{Ho}}^{3+}$, ${\mathrm{Tm}}^{3+}$, and ${\mathrm{Tb}}^{3+}$ ions) are commonly used to realize broadband upconversion [1–4]. The upconversion process consists of various subprocesses, such as excited-state absorption, energy transfer, cooperative sensitization, and the photo avalanche. Additionally, these processes are associated with relatively low efficiencies.

In 2010, Wang and Tanner reported on a new type of broadband anti-Stokes emission (BASE) from a single rare-earth-doped material [5], and some related studies followed [6,7]. This BASE has a long buildup time, a long quenching time, and a high efficiency in vacuum. In 2014, Wang et al. used ${\mathrm{Yb}}^{3+}$-doped ${\mathrm{ZrO}}_2$ to realize BASE with a power efficiency of 16%. These experimental results verified the potential of this material to improve the conversion efficiency of Si solar cells owing to the fact that more infrared light could be converted to visible light [8]. However, to the best of our knowledge, all reported materials that can generate this kind of BASE were powders (in particular, nanopowders) or ceramics, and no reports concerning bulk crystals have yet appeared. Accordingly, in this Letter, we present an approach for obtaining high-efficiency BASE at room temperature and atmospheric pressure from a bulk crystal.

When ${\mathrm{Yb}}^{3+}$-doped materials, such as Yb:YAG, Yb:FOV (${\mathrm{Yb}}^{3+}$-doped oxyfluoride vitroceramics), and $\mathrm{Yb}\text{:}{\mathrm{NaYF}}_4$, are excited by a 940 or 976 nm laser diode (LD), downconversion around 1.0 μm can be observed, as well as upconversion at the blue–green region. This upconversion can be ascribed to ${\mathrm{Yb}}^{3+}$ ion cooperative luminescence and the presence of rare earth impurity ions [9,10]. Aside from this familiar type of luminescence, we found that ${\mathrm{Yb}}^{3+}$-doped crystals with random cracks are able to generate bright BASE for excitation powers above a certain threshold. To investigate this mechanism, a 940 nm laser with a power of 23 W was reimaged onto YAG crystals doped with 20 at. % ${\mathrm{Yb}}^{3+}$ ions by a pair of plano–convex lenses; this high-intensity laser caused the crystals to crack. After this procedure, samples with cracks were excited by the LD at 940 nm. When the excitation light was focused on a suitable position of the crystals and adjusted to a suitable power, the samples were able to generate bright BASE. Importantly, this BASE cannot be observed before treating the crystals in this manner.

In this experiment, a laser diode at 940 nm is used for the excitation; the output laser is focused on the crystal, and the diameter of the beam spot on the crystal is 100 μm. The emission light from one of these samples before and after reaching the excitation power threshold is shown in Figs. 1(a)–1(c). Under an excitation power of 1.4 W, the CCD camera can only capture the excitation light that is scattered by the crystal. On the other hand, the bright yellowish broadband emission can be observed under an excitation power of 1.42 W, and its intensity becomes maximal after a few seconds. The spectrum of this BASE is illustrated in Fig. 1(d), where one can see that the highest emission peak is located at 568 nm and that the full width at half-maximum (FWHM) is about 112 nm.

 

Fig. 1. Broadband anti-Stokes emission (BASE) from a Yb:YAG crystal with random cracks. The crystal excited by a laser diode (LD) at 940 nm with a power of (a) 1.4 W, (b) 1.42 W, and (c) 1.42 W after 20 s. (d) Anti-stokes emission spectrum of the crystal with random cracks (red line) under an excitation power of 1.42 W, along with the spectrum of sunlight at noon (blue line). The inset depicts the sample that was used in this test. (e) Infrared emission spectra of a crystal with cracks (blue line), as well as a normal crystal (red line) under an excitation power of 1.42 W. (f) Emission intensity (red line) and emission spectrum (inset) as functions of the excitation power. (g) Buildup time ($\sim 2.89\mathrm{s}$, red line) and quenching time ($\sim 170\mathrm{ms}$, blue line) of the emission under an excitation power of 1.8 W. (h) Emission light separated by a dispersion prism. (i) Pattern of the dispersion light on the viewing screen.

Download Full Size | PDF

A dispersion prism was used to separate the emission light, and the pattern captured on the viewing screen is shown in Figs. 1(h) and 1(i). Its spectral characteristics are similar to those of sunlight at noon, except for a small redshift. In fact, all samples with cracks demonstrate the same spectral characteristics under excitation at 940 or 976 nm; their emission spectrum is not sensitive to the temperature of the crystal in our experiments. As a continued broadband emission, it differs from the upconversion associated with energy transfer excited-state absorption or cooperative sensitization, while it is identical to the charge-transfer luminescence [11].

The emission intensity and emission spectrum as functions of excitation power are shown in Fig. 1(f), which demonstrates that there exists a certain threshold ($P_{\mathrm{th}}\approx 1.41\mathrm{W}$). Moreover, the power dependence of this BASE can be described in terms of the power law $I_{em}\sim P^N$ [12], with an order parameter $N\approx 3$, when the excitation power falls above 1.41 W. All these features indicate that there may be an energy avalanche effect that occurs during the BASE process. Under an excitation power of 1.8 W, the power of the emission light in a $0.8\pi$ solid angle reaches 67 mW with an optical-optical efficiency of 2.7%. This means that the total optical-optical efficiency in a $4\pi$ solid angle may reach 13.5%.

Several possible mechanisms of the BASE from ${\mathrm{Yb}}^{3+}$-doped materials have been presented in former studies. Most researchers believe that this BASE is a kind of charge-transfer luminescence that involves some complicated processes [6,7,13]. The charge-transfer luminescence of Yb:YAG crystal is relative to the oxygen ion in YAG (${\mathrm{Y}}_3{\mathrm{Al}}_5{\mathrm{O}}_{12}$) and the ${\mathrm{Yb}}^{3+}$ ion. If the absorbed energy is high enough, an electron will escape from the oxygen ion to the ${\mathrm{Yb}}^{3+}$ ion. After the transfer, the ${\mathrm{O}}^{2-}$ ion is changed to the ${\mathrm{O}}^{-}$ ion, and the ${\mathrm{Yb}}^{3+}$ ion is changed to the ${\mathrm{Yb}}^{2+}$ ion. However, this status is not stable. When this electron goes back to the oxygen ion, it will generate the charge-transfer luminescence. On the basis of the tests above and, by considering the effects of energy migration, energy avalanche, and charge transfer luminescence, we present a theoretical model to interpret the underlying physics of this BASE from bulk crystals.

Our model explains the observed BASE using a three-step process. During step one [Fig. 2(a)], photons of excitation light are absorbed by ${\mathrm{Yb}}^{3+}$ ions, which causes electrons to jump from the ground state to the excited state. When the electrons return to the ground state, a portion of this energy is transformed into radiation, whereas another portion is transferred to the adjacent ${\mathrm{Yb}}^{3+}$ ions; in turn, it causes the electrons of these adjacent ${\mathrm{Yb}}^{3+}$ ions to jump to the excited state. This nonradiation effect between ions is referred to as energy migration and allows for energy to be stored in the ${\mathrm{Yb}}^{3+}$ ions. During step two [Fig. 2(b)], because of the energy migration, many electrons migrate to the excited state as the excitation power increases; this energy accumulation can easily trigger an avalanche effect. When the excitation power reaches a certain threshold, many electrons in the excited state will return to the ground state in a very short time, and a large amount of energy will be rapidly released. This high amount of energy leads to the transition of electrons from the adjacent oxygen ion to the ${\mathrm{Yb}}^{3+}$ ion (from the valence band to the charge-transfer state). During step three [Fig. 2(c)], the electrons fall to the ${}^2{\mathrm{F}}_{5/2}$ or ${}^2{\mathrm{F}}_{7/2}$ energy state, release their energy in the form of photons, and then travel back to the valence band via a nonradiation relaxation.

 

Fig. 2. Mechanism of the broadband anti-Stokes emission (BASE). (a) Energy migration process. (b) Energy avalanche process. (c) Charge-transfer process.

Download Full Size | PDF

The charge-transfer state is a broadband energy state; it is much higher than the energy state ${}^2{\mathrm{F}}_{5/2}$ of the ${\mathrm{Yb}}^{3+}$ ions, which enables some electrons at the charge-transfer state to jump to the conduction band of the YAG crystal. According to our analysis, as a result of the energy migration and energy avalanche effects, the charge-transfer behavior can occur under infrared laser excitation, which has already been observed in the context of photocurrent detection [14]. This theoretical model also indicates that the energy migration in the buildup process and the relaxation in the emission process of the BASE would lead to a long buildup time and a long quenching time, which is consistent with the test results in our experiment [Fig. 1(g)]. However, it still remains to be explained why the bright BASE can only be observed from Yb:YAG crystals with random cracks.

Usually, when the electrons return to the ground state from the excited state, most of their energy will be released as radiation, and only a small portion of it will take part in the energy migration process. However, the significant fluorescence peak at the wavelength of 1031 nm cannot be found when the Yb:YAG crystal with random cracks is excited by a laser; this means that almost all of the energy is transferred to the energy migration process, as shown in Fig. 1(e). Though changes in ${\mathrm{Yb}}^{3+}$ ion concentration or the crystal lattice structure can affect the conversion efficiency of energy migration, none of these effects was found to occur in the samples (surface or inside) according to the test results of X-ray diffraction (XRD) and X-ray photoelectron spectroscopy (XPS). As shown in Fig. 3, the positions and relative intensities of diffraction peaks of samples after the laser treatment are in good agreement with the YAG crystal (PDF#88-2048). The diffraction peaks are sharp, which means there is no glassy phase in these samples. In general, the crystal structure is unchanged after the laser treatment. Besides, according to the XPS test, the ratios of ${\mathrm{Yb}}^{3+}$ ion concentration to ${\mathrm{Y}}^{3+}$ ion concentration in a normal and a broken Yb:YAG crystal are detected to be $16:84$ and $19:81$, respectively. We believe that the fluctuation of the ratio is caused by the measurement error, and there is no significant change of ${\mathrm{Yb}}^{3+}$ ion concentration after the laser treatment.

 

Fig. 3. XRD pattern of samples after the laser treatment and standard PDF card of the YAG crystal.

Download Full Size | PDF

We found that if the treatment power is too high, the surface of the Yb:YAG crystals will be broken into a powder, as shown in Fig. 4(a). This variety of crystal differs from the variety of crystal with random cracks [Fig. 4(b)], as it cannot generate the BASE. For this reason, we believe that the conversion efficiency of the energy migration is related to the presence of cracks in the crystal.

 

Fig. 4. (a) Yb:YAG crystal mounted in a copper block. The surface was broken into a powder by using a high-power laser. (b) Yb:YAG crystal with random cracks mounted in a copper block. (c) Total internal reflection of excitation light in the Yb:YAG crystal with random cracks. (d) Yb:YAG crystals after treatment. (e) One small piece of the Yb:YAG crystal after treatment.

Download Full Size | PDF

As shown in Fig. 4(c), the cracks filled with air separate the Yb:YAG crystal into several parts, and each part of the crystal can be seen as an anomalous polyhedron. According to the theory of total internal reflection, when light is emitted from the crystal (optically denser medium) to the air (optically thinner medium), it may be totally reflected on the surface and return to the crystal again. In particular, certain areas of the crystal may experience a strong total internal reflection effect. The incident light will be reflected multiple times on different surfaces before it escapes from the crystal. Because there is an overlap between the emission spectrum and the absorption spectrum of Yb:YAG crystals, it can either emit or absorb the light at $\sim 940\mathrm{nm}$. Thus, the multi-reflection of the incident light at 940 nm in the crystal will induce the stimulated radiation of ${\mathrm{Yb}}^{3+}$ ions at 940 nm and reduce its spontaneous radiation at 1031 nm. Finally, almost all the energy will be used for the stimulated radiation, and this radiation will be re-absorbed by other ${\mathrm{Yb}}^{3+}$ ions, ultimately leading to the energy migration process mentioned above. As a result of the effects of multi-reflection, stimulated radiation, and re-absorption, the conversion efficiency of the energy migration process becomes significantly increased. Though this hypothesis is difficult to be verified by experiments directly, some test results of our experiment still give indirect evidences to the validity of it, and this explains why the BASE can only be observed from crystals that contain random cracks and why the efficiency of the BASE is so high.

It is easy to produce crystals with random cracks by using a high-power infrared laser. All the samples obtained after treatment [Fig. 4(d)] were able to generate bright BASE; this is true even of the small piece shown in Fig. 4(e). Therefore, this procedure represents a simple way of obtaining bright BASE from a bulk crystal. Nevertheless, some performance characteristics of the BASE, such as the threshold and conversion efficiency, depend on the shape of the crystal, which cannot be determined in advance using this method. According to our analysis, the key to generating high-efficiency BASE is enhancing the multi-reflection effect, thereby taking full advantage of the excitation light. Thus, in further studies, we plan to design a special shape for the crystal that can allow the excitation light to be reflected as much as possible on every surface, and test the hypothesis with these samples.

In conclusion, the BASE obtained from Yb:YAG crystals is a variety of charge-transfer luminescence that results from the effects of the energy migration and energy avalanche processes. In our experiment, we found that it can also be obtained from other ${\mathrm{Yb}}^{3+}$-doped crystals, such as Yb:KYW or Yb:KGW, and that the basic precondition for generating this kind of emission in bulk crystals is the presence of random cracks in the crystal.