Research Article www.acsami.org
Cite This: ACS Appl. Mater. Interfaces 2017, 9, 43790−43798
Luminescence Anisotropy and Thermal Effect of Magnetic and Electric Dipole Transitions of Cr3+ Ions in Yb:YAG Transparent Ceramic Fei Tang,† Honggang Ye,† Zhicheng Su,† Yitian Bao,† Wang Guo,‡ and Shijie Xu*,† †
Department of Physics, and Shenzhen Institute of Research and Innovation (HKU-SIRI), The University of Hong Kong, Pokfulam Road, Hong Kong, P. R. China ‡ Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002, P. R. China S Supporting Information *
ABSTRACT: In this article, we present an in-depth optical study on luminescence spectral features and the thermal effect of the magnetic dipole (MD) transitions (e.g., the R lines of 2E → 4A2) and the associated electric dipole transitions (e.g., phononinduced sidebands of the R lines) of Cr3+ ions in ytterbium−yttrium aluminum garnet polycrystalline transparent ceramic. The doubly split R lines predominately due to the doublet splitting of the 2E level of the Cr3+ ion in an octahedral crystal field are found to show a very large anisotropy in both emission intensity and thermal broadening. The large departure from the intensity equality between them could be interpreted in terms of large difference in coupling strength with phonons for the doubly split states of the 2 E level. For the large anisotropy in thermal broadening, very different effective Debye temperatures for the two split states may be responsible for it. Besides the 2E excited state, the higher excited states, for example, 4T1 and 4T2 of the Cr3+ ion, also exhibit a very large inequality in coupling strength with phonons at room temperature. By examining the Stokes phonon sidebands of the MD R lines at low temperatures with the existing ion−phonon coupling theory, we reveal that they indeed carry fundamental information of phonons. For example, their broad background primarily reflects Debye density of states of acoustic phonons. These new results significantly enrich our existing understanding on interesting but challenging luminescence mechanisms of ion−phonon coupling systems. KEYWORDS: trivalent Cr ion-activated transparent ceramic, magnetic/electronic dipole transitions, ion−phonon coupling, luminescence, thermal effect
1. INTRODUCTION Inspired by the successful applications of yttrium aluminum garnet (YAG)-based single crystals doped with various active ions in lasers,1−3 their polycrystalline counterparts, namely, transparent ceramics (TCs), are alternatively extensively being investigated with the rapid development of high-temperature sintering technology.2−9 This kind of novel material possesses a full-dense microstructure with numerous grains bonded together via grain boundaries.10 Such unique microstructures endow it not only with outstanding optical and thermal properties but also with better mechanical properties with respect to single crystals.8 Like single crystals, TCs formed via the solid-state reaction sintering process can also be optically activated by doping various active ions, such as rare-earth (RE) and transition-metal ions. Currently, RE-doped YAG single crystals and TCs are the most extensively studied and widely used for high-power lasers. 9 Neodymium-doped YAG and ytterbium−YAG (Yb:YAG) lasers are the two application examples of REdoped YAG single crystals and TCs. Unlike RE ions having a stable 4f electronic configuration, transition-metal ions often show changeable luminescent behaviors in insulating crystals © 2017 American Chemical Society
and TCs, depending sturdily on the surrounding lattice environment.11−16 For example, the transition-metal Cr3+ ion with a 3d3 electronic configuration,16 which has been widely recognized as one of the most significant active ions since 1960s when inventors developed the first solid-state ruby laser,17 may show a +4 chemical state and thus exhibits a 3d2 configuration in the YAG crystal.18,19 However, it has been shown that Cr3+ rather than Cr4+ acts as an efficient luminescent center in some host lattices such as ZnAl2O4, MgAl2O4, and MgO crystals.20,21 On the other hand, the coupling of transition-metal ions with lattice vibrations (phonons) in these crystals is relatively stronger than that of RE ions in the same crystals.22 In general, stronger impurity− phonon coupling results in more noticeable phonon sidebands in luminescence and absorption spectra. Very recently, there has been a renewed interest in the control and manipulation of magnetic dipole (MD) transitions of transition-metal ions in various crystal fields.16,23 For transition-metal ions such as Cr Received: September 16, 2017 Accepted: November 23, 2017 Published: November 23, 2017 43790
DOI: 10.1021/acsami.7b14061 ACS Appl. Mater. Interfaces 2017, 9, 43790−43798
Research Article
ACS Applied Materials & Interfaces ions in crystals, they often simultaneously show a strong MD transition and broadened electric dipole (ED) transitions.24−26 The former transition leads to the so-called zero-phonon line (ZPL), whereas the latter ED transitions contribute to the broadened phonon sidebands.25,26 Clearly, investigating the MD and ED transitions of Cr3+ ions in YAG TC materials is of both fundamental and technological significance. In this article, we present a comprehensive study of luminescence properties of Cr3+ ions in Yb:YAG TC that was prepared by a high-temperature vacuum sintering method. It is found that the double sharp MD lines of Cr3+ ions exhibit a large anisotropy in both luminescence intensity and thermal broadening. Our study reveals that the large fluorescence anisotropy may stem from very dissimilar coupling strengths of the relevant impurity levels with host phonons.
Figure 1. SEM image of the surface-polished TC sample prepared by high-vacuum high-temperature sintering technology.
the TC sample taken under the conditions of natural light. Obviously, the prepared ceramic exhibits light green color under the illumination of natural light. However, it emits intense red light when it was illuminated with a 405 nm laser beam, as illustrated by the photograph in Figure 2b. Note that the strong white background in the photograph is due to the mixture of emitted red light and scattered blue laser by the TC sample. As reported and argued in the literature, the intense red light emission can be ascribed to the R lines of 2E → 4A2 of the Cr3+ ion in the YAG host crystal. It has been well-argued from crystal chemistry considerations that in garnet crystals including the YAG crystal, Cr3+ ions preferentially occupy the octahedral 16(a) sites rather than the tetrahedral 24(d) sites.27,28 To determine the major occupation state of the Cr3+ ions in the studied YAG TC sample, a theoretical calculation was performed via applying the Rietveld refinement method on the experimental XRD data of the TC sample. As seen in Figure 2c, theoretical results are in excellent agreement with the measured XRD patterns. The crystalline structure of the TC sample was thus determined and can be uniquely characterized in a cubic space group of Oh10-Ia3d. The determined unit cell parameters were tabulated in Table 1. It can be concluded that incorporating Yb ions and Cr ions into the YAG lattice results in a slightly enlarged unit cell. Also, some Al3+ ions surrounded by six nearest O2− ions are most likely replaced by Cr3+ ions. For the Cr3+ ion occupying Al3+-octahedral center site in the YAG host crystal, the dependence of its energy levels on crystal-field strength may be illustrated by using a Tanabe− Sugano energy diagram.29 In the crystal-field approximation, the effect of the electric field produced by surrounding ions on the energies of the impurity levels may be calculated by the use of four parameters: spin−orbit coupling ζ, crystal-field Dq, and two electrostatic interaction parameters, B and C defined by Racah in terms of the Slater integrals.30 Crystal-field parameters for Cr3+ in the YAG host crystal adopted by Wood et al.20 were relisted here: ζ = 170 cm−1, Dq = 1640 cm−1, B = 650 cm−1, and C = 3250 cm−1. Note that the spin−orbit coupling in the Cr3+ ion is relatively much smaller, and thus, it contributes only in a very minor way to the energy level locations of the Cr3+ ion. In fact, the Tanabe−Sugano energy matrices of the Cr3+ ion are valid, provided the spin−orbit interaction is neglected entirely. According to the calculations of Tanabe and Sugano, the 2E level of the Cr3+ ion located approximately 9B + 3C above the ground state 4A2. The separation between the ground level 4A2 and the first quartet state 4T2 was just 10Dq, whereas the second quartet 4T1 was higher by ∼12B. Figure 3a shows the calculated splittings of 2E and 4A2 levels of the Cr3+ ion under
2. EXPERIMENTAL SECTION 2.1. Material Preparation. As mentioned earlier, Cr3+-activated Yb:YAG TC was prepared through a solid-state reaction process by means of high-temperature vacuum sintering technology. Commercially available powders of Al2O3, Y2O3, Yb2O3, and Cr2O3 were chosen as reaction precursors. Tetraethyl orthosilicate was used as the sintering additive, while oleic acid served as the dispersion agent. On the basis of the formula of (Y0.99Yb0.01)3(Al0.997Cr0.003)5O12, stoichiometric amounts of the precursors were thoroughly mixed in ethanol for 20 h by a planetary milling machine. The obtained slurry was dried, ground, and then sieved through a 200-mesh screen. Ceramic green body was formed via cold-isostatically pressing the powder into 15 mm disk at ∼2 MPa, followed by a de-binding process at 800 °C. Then, a high-temperature sintering process was carried out under a vacuum of 5 × 10−7 Torr at 1750 °C for 10 h. Finally, TC was obtained after mirror-polishing both surfaces. There is no need of additional thermal annealing on the sintered ceramics. 2.2. Structural Characterization. The crystalline structure of the ceramic was determined via X-ray powder diffraction (XRD, Type D8 ADVANCE ECO, Bruker, UK) with Cu Kα radiation (λ = 1.54 Å) as the radiation source. The current and cathode voltage of the radiation source were set to be 40 mA and 40 kV, respectively. Coupled 2θ/θ was selected as the scan type with continuous power spectral density as the fast scan mode. The measured scanning range in this study was set from 2θ = 10° to 120° with an increment of 0.02°/step and a delay time of 1 s. The obtained XRD data of the as-sintered TC were utilized as an initial base for the Rietveld refinement using the GSAS package. The morphology of the sample was observed using a field emission scanning electron microscope (JOEL JSM6700-F). 2.3. Variable-Temperature PL Measurements. High-resolution photoluminescence (PL) spectra of the ceramic were obtained at temperatures ranging from 4 to 300 K by means of a home-assembled variable-temperature laser spectroscopy system, in which a 405 nm laser light was employed as the excitation source. Variable-temperature PL measurements were realized through mounting the ceramic sample on the cold finger of a closed cycle cryogen-free refrigerator (Janis Research), and a temperature controller (Lake Shore 325) was employed to control the sample temperature. The light signal emitted from the sample was collected by a pair of lenses and then guided into a 1200 g/mm grating monochromator (Spex 750M) with a photomultiplier detector (Hamamatsu R928) for dispersion and detection. To improve the signal-to-noise ratio, a standard optical chopper plus lock-in amplifier was adopted. The major parts of the PL system were controlled with a desktop computer.
3. RESULTS AND DISCUSSION Figure 1 shows the scanning electron microscopy (SEM) image of the surface-polished Cr3+-activated Yb:YAG TC sample. Fulldense microstructures without any pores or other second phase can be clearly seen from the image, which indicates good quality of the prepared TC. Figure 2a shows a photograph of 43791
DOI: 10.1021/acsami.7b14061 ACS Appl. Mater. Interfaces 2017, 9, 43790−43798
Research Article
ACS Applied Materials & Interfaces
Figure 2. Photographs of the as-prepared TC under the illumination of natural light (a) and a 405 nm laser light (b). Measured and simulated XRD patterns of the sample (c). The inset shows the crystal structure of the sample. 2
E, 4T2, and 4T1 with respect to the ground state 4A2, as depicted in the configurational coordinate diagram in Figure 3b. The parabolas in the figure represent the potential profiles of Cr3+ electronic orbits coupled with a host lattice vibration (phonon). The difference between the equilibrium positions of parabolas may be uniquely characterized by the Huang−Rhys factor.35 A larger equilibrium position difference implies greater lattice relaxation during relevant vibronic transition and a larger Huang−Rhys factor, which is the most important parameter in theoretical calculations of absorption and emission of the electron−phonon coupling system.36−38 For the 4T2 and 4T1 states, two major photoluminescence excitation (PLE) bands due to the optical transitions from 4A2 to 4T1 and 4T2, respectively, were observed at room temperature, as shown in Figure 3c. The monitored wavelength was set at the R1 line for the PLE measurements. A generalized multimode Brownian oscillator (MBO) model, which is developed to calculate nonlinear optical responses of a two electronic-level system with some primary nuclear coordinates coupled linearly to it and to a harmonic phonon bath,39,40 was adopted to compute the two broad PLE bands following the pioneering application of the MBO model in the calculation of luminescence spectrum of the electron−phonon coupling system in solid.41 The calculated curves with the MBO model are illustrated by green solid lines in Figure 3c. Obviously, they are in good agreement with the experimental spectra. The parameters used in the calculations are EZPL = 1.8931(2.3846) eV, S = 3.1(7.6), γ = 140(50) cm−1, h̵ωP = 547 cm−1, and T = 300 K for the PLE band located at lower (higher) energy side. Here, EZPL represents the so-called ZPL location, S the Huang−Rhys factor, γ the coupling constant between the primary phonon mode and the continuous phonon bath, and h̵ωP = 547 cm−1
Table 1. Unit Cell Parameters Obtained from the Rietveld Structural Refinement of XRD Dataa samples
a = b = c (Å)
α = β = γ (deg)
V (Å3)
31
12.01159 ± 0.000034 12.015681 ± 0.000775 12.019558 ± 0.000031
90 90 90
1733.01 1734.78 1736.46
YAG Yb:YAG Cr,Yb:YAG a
Herein, YAG denotes Y3Al5O12.
the combined action of crystal field and spin−orbit coupling.32 From a theoretical point of view, four possible transitions may occur. However, only double R lines were resolved in the PL spectrum, as shown in Figure 3c. We thus sketched two downward arrows in Figure 3a to show R1 and R2 transitions. At 300 K, the PL spectrum is dominated by R1 and R2 lines located at 688.75 and 687.79 nm, respectively. At 4 K, the R1 and R2 lines peak at 687.31 and 686.38 nm, respectively. The R line splitting still remains as 20 cm−1. These data are in good agreement with the ones reported previously.28,33 The experimental R line splitting is smaller by about 9 cm−1 than the theoretical splitting of the 2E level of Cr3+ in the cubic crystal field. The discrepancy has not yet been well-understood by now. The complicated crystal structure of YAG and inaccurately known local lattice distortion may be important factors for the obvious discrepancy,20,27 although from a macroscopic viewpoint the overall symmetry is that of a cubic nature.34 Nevertheless, it is still concluded that the observed double R lines are predominantly due to the doublet splitting of the 2E level under the action of the crystal field. In addition to the 2E excited state, there are another two higher excited states, namely, 4T2 and 4T1, for Cr3+. Evidently, an order from a smaller displacement difference is obtained for 43792
DOI: 10.1021/acsami.7b14061 ACS Appl. Mater. Interfaces 2017, 9, 43790−43798
Research Article
ACS Applied Materials & Interfaces
Figure 3. (a) Energy level splitting diagram induced by the combined effect of both distorted crystal field and spin−orbit coupling. Two downward arrows are sketched to point out R1 and R2 lines. (b) Configurational coordinate diagram for Cr3+ ions in the YAG host and possible optical transitions. (c) Room-temperature PL and PLE spectra of Cr3+-activated Yb:YAG TC. The inset shows the ZPL spectrum enlarged in the wavelength range of 686−692 nm, where the R1 and R2 lines can be well-resolved.
Figure 4. Measured 4 K PL spectrum (thin solid line) of the TC sample. The thicker solid line is a theoretical curve jointly with eqs 2 and 3. The inset shows a whole 4 K PL spectrum, where the relative intensity of R1 and R2 as well as the Stokes phonon sidebands can be seen.
the characteristic energy of primary phonon mode. 42 Interestingly, good agreement between the theory and the experiment reveals that there is a much larger Huang−Rhys factor for the absorptive transition from 4A2 to 4T1 with respect to the transition from 4A2 to 4T2. As argued earlier, a relatively larger lattice relaxation may occur upon the optical transition
from 4A2 to 4T1. It is thus expected that a relatively much larger Huang−Rhys factor is attached to the 4A2 → 4T1 transition because of S ∝ Δ2 in the strong electron−phonon coupling cases.43 Besides the sharp R lines and the two broad PLE bands, actually, both Stokes and anti-Stokes phonon sidebands were 43793
DOI: 10.1021/acsami.7b14061 ACS Appl. Mater. Interfaces 2017, 9, 43790−43798
Research Article
ACS Applied Materials & Interfaces
Figure 5. Four PL spectra (solid squares) of the R lines measured at (a) 4, (b) 100, (c) 200, and (d) 300 K. The decoupled curves with two standard Lorentzian lineshape functions and the cumulative curves are plotted in solid lines.
phonon population of the κ(ωij) mode. When the temperature approaches zero, eq 1 becomes π W id = ∑ ωijρ(ωij)|⟨j|C|i⟩|2 Mν 2ℏ j < i (2)
also observed in the long- and short-wavelength sides of the split R lines at room temperature, as shown in Figure 3c. These sidebands of the R lines are outcomes, resulting from interactions between the Cr3+ ion and surrounding phonons. They carry important information of lattice vibrations, especially low-frequency phonon modes. Figure 4 shows the R lines and the phonon sidebands at 4 K. The inset shows the overall spectrum of the R lines and the phonon sidebands at 4 K, whereas the main figure depicts a closer spectrum of the phonon sidebands. Being consistent with the spectral data previously reported by Wall et al.,33 at 4 K only highly structured Stokes phonon sidebands due to the simultaneous phonon emission were observed. Only when the temperature is beyond 100 K, the antiStokes sidebands become observable because of the thermal excitation and simultaneous absorption of phonons. It has been well-established that being a paramagnetic impurity, Cr3+ ions in various crystals including the studied YAG crystal often show a sharp MD ZPL luminescence and broadened ED phonon-assisted luminescence.24−26 For the phonon sidebands, they have been extensively theoretically treated in terms of lattice relaxations.35−38,43−46 For the ith level of an impurity ion in the host crystal, the phonon relaxation probability due to one-phonon or “direct” processes may be written as46 W id =
The most important conclusion of eq 2 could be that at extremely low temperatures the transition probability is proportional to ωijρ(ωij) if the transition matrix element takes a constant. In other words, the detailed phonon density of states might be reflected in the low-temperature phonon sidebands. On the other hand, another well-known result in the multiphonon transition theory is that the total transition probability of the pth phonon mode at extremely low temperatures can be proved as47
Wij = |Mij|2
⎫ ⎪
∑ ωijρ(ωij)|⟨j|C|i⟩|2 N(ωij)⎬⎪ j>i
⎭
(3)
where |Mij| is the transition matrix element, S is the dimensionless Huang−Rhys factor, and n is the order number of multiphonon transition. Equation 2 gives the transition probability of continuous or quasicontinuous phonons owing to the direct one-phonon process, whereas eq 3 accounts for the transition probability of an individual phonon mode owing to the multiphonon relaxation process. Therefore, the total transition probability at extremely low temperatures is the sum of two transition probabilities. Assuming the constant transition matrix element and Debye-type phonon density of states (e.g., ρ(ω) ∝ ω2), we employ jointly eqs 2 and 3 to make a fit to the 4 K PL spectrum shown in Figure 4. The theoretical fitting curve is in reasonable agreement with the experimental PL spectrum considering the complexity of situations (e.g., extreme difficulty in the theoretical treatment of impurity− phonon coupling problem and complicated phonon modes and distribution of the YAG crystal). In spite of an obvious derivation between the theoretical curve and the experimental
⎧ π ⎪ ⎨∑ ωijρ(ωij)|⟨j|C|i⟩|2 [N (ωij) + 1] Mν 2ℏ ⎪ ⎩ j
