1. Introduction

White light-emitting diodes (WLEDs) have recently received considerable attention as the next generation solid-state lighting source owning to their overwhelming merits of lower power consumption, higher energy efficiency, longer operate lifetimes, and eco-friendly constituents, compared to the traditional incandescent and mercury-containing fluorescent lightings [1–4]. For the purpose of overcoming the mechanical, chemical, and thermal stable issues, a new type of WLEDs using Ce:YAG transparent ceramics instead of the traditional encapsulation which involves packing yellow phosphors with organic resins, has been proposed [5, 6]. Nevertheless, the correlated color temperature (CCT) is high, and the color rendering index (CRI) is poor due to the red light deficiency [7]. Dual-layered phosphors of different colors are integrated or red-emitting ions are introduced into the crystal to enhance the red spectrum, ultimately to improve the CCT and CRI [8–12]. However, these strategies are facing some problems. For example, the cost of production sharply increased once an extra red-emitting coating/layer is used, and emission intensity decreases because severe energy loss happens between the activators in a co-doped or triple-doped compound during the energy migration. Consequently, it is an important topic to strike a balance between the cost, the luminous efficiency, and the light quality of WLEDs.

Since the luminescent property is known to be related to the environment around the Ce³⁺, a promotion of light quality can be realized by tailoring the material compositions and band structures [13]. One of the tailoring ways is to distort the cubic symmetry around the RE (rare earth elements) via increasing or decreasing the ionic radii or introducing impurity atoms in the crystal structure, as has been shown in several papers [14–22]. The luminescence properties were improved, including an enhancement of emission intensity, or a modification of band structure. Among which, Gd³⁺ has been reported to successfully induce a red shift of the emission spectrum of Ce³⁺ via altering the crystal structures in YAG:Ce powders, Mg₃(Y, Gd)Ge₃O₁₂:Ce powders and (Lu, Gd)₃(Ga,Al)₅O₁₂ single crystals [13, 21, 23]. However, they concentrated either on the powder materials or on their scintillation properties, but did not pay any attention to the effect of Gd³⁺ on the white emission properties of transparent ceramic-based phosphors excited by a blue LED.

To make clear whether and how the Gd³⁺ ions would improve the lighting quality, (Y, Gd)₃Al₅O₁₂: Ce transparent ceramics were fabricated via a solid-state reaction method under vacuum in this study. The phase composition and luminescence performance of the (Y, Gd)₃Al₅O₁₂: Ce transparent ceramics were investigated. Actually, the Gd³⁺ additive was found critical to the color tuning. With proper content of Gd³⁺ admixing, WLEDs with higher CRI and lower CCT compared with that of the YAG: Ce can be obtained.

2. Experimental procedure

(Y_0.998-xGd_xCe_0.002)₃Al₅O₁₂ (x = 0~0.8) transparent ceramics (Abbreviated as (Y, Gd)AG: Ce, herein after) were fabricated via solid-state reaction and vacuum sintering. The starting materials, Y₂O₃ (99.995%), Al₂O₃ (99.99%), CeO₂ (99.999%), and Gd₂O₃ (99.99%) were used directly without any further treatment. Sintering aids (0.5 wt% TEOS and 0.05 wt% MgO) and the raw materials were weighed stoichiometrically and thoroughly mixed by ball milling in anhydrous ethanol. After that, the slurries were dried in the oven to remove the residual ethanol. The dried powders were grounded and sieved through 200-mesh screen to obtain homogeneous submicron-sized powders. Prior to sintering, the powders were uniaxially pressed followed by a cold-isostatic press at ~200 MPa. Then the green bodies of Φ20 mm were sintered at 1680~1720°C for 6 h under vacuum and further annealed at 1400°C for 6 h in air to remove oxygen vacancy. Finally, transparent ceramics were double-surface polished to a thickness of 0.5 mm for characterization.

Phase compositions of the final transparent ceramics were characterized in the 2 θ range from 10° to 80° with a step of 0.02°, using X-ray diffraction (XRD) on a Japan Rigaku D/MAX 2200PC diffractometer with Cu Kα radiation (λ = 1.54056 Å). Transmittance and absorption spectra of the ceramics were measured on a UV-Vis-NIR spectrometer (Carry-5000, Varian company, America). Blue LED chips with emission wavelength of ~450 nm were used for photoluminescence characterization. The LED chips were operated at 27 V, 100 mA. Luminous efficiency, Commission Internationale de l’Eclairage (CIE) color coordinates, CCT and CRI of the LED-devices were characterized with a UV-Vis-near IR spectrophoto colorimeter with an integrating sphere under a forward-bias current of 20 mA at RT (PMS-80, Everfine Corp., China). The conditions of excitation and detection systems were fixed, the samples were kept the same thickness, shape and surface status.

3. Results and discussion

For lighting applications, transparent ceramics with high optical quality are significant to the performance of the LED devices. The Gd³⁺ additive of no more than 80 mol% (x = 0.8) was chosen to obtain pure garnet phase, because the Gd₃Al₅O₁₂ is an unstable incongruent compound, and only with small ions stabilization (such as Y³⁺), can pure garnet phase be obtained, according to the previous studies [24, 25]. Figure 1 shows the photographs of the (Y, Gd)AG: Ce transparent ceramics, through which the characters written on the paper can be clearly read, indicating the high optical quality of the samples. It also can be seen in Fig. 1 that the color of the ceramics varies from yellow to yellow-orange with increasing Gd³⁺ concentration. This may be caused by the change in absorption spectra of the ceramics, which will be discussed later.

 

Fig. 1 Photographs of the (Y, Gd)AG: Ce transparent ceramics (0.5 mm thick).

Download Full Size | PDF

XRD patterns of the (Y, Gd)AG: Ce ceramics are shown in Fig. 2(a). The phases are pure and no secondary phases are found with various Gd³⁺ concentrations. The YAG: Ce ceramic (x = 0) can be well indexed as cubic garnet structure of YAG (PDF No. 33-0040). Magnified XRD patterns in the region between 33 and 34 degrees are shown in the right of Fig. 2(a), from which a distinctive shift of the peaks centered at 2θ = 33.8° to the lower 2θ value is found with increasing Gd³⁺ concentration from x = 0 to 0.8. This can be ascribed to the larger ionic radius of Gd³⁺ (1.05 Å) than Y³⁺ ions (1.02 Å). Figure 2(b) shows the lattice constants of the ceramics using XRD results. Clearly, the cell parameter increases linearly with increasing Gd³⁺ concentration, which obeys Vegard’s law, implying that Gd³⁺ ions incorporate into the YAG: Ce lattice totally and homogeneous solid solutions have been formed.

 

Fig. 2 (a) XRD patterns, (b) lattice constants of the (Y, Gd)AG: Ce ceramics as a function of Gd³⁺ concentration. The right of (a) shows the magnified XRD patterns of the characteristic peak of (4 2 0) crystal plane.

Download Full Size | PDF

The in-line transmittance spectra of the (Y, Gd)AG: Ce transparent ceramics are illustrated in Fig. 3. The ceramics have high optical quality with transmittance higher than 75% in the wavelength range from 550 to 1100 nm, despite the variation of Gd³⁺ concentration. The low transparency at around 345 and 455 nm can be ascribed to the intrinsic absorption of Ce³⁺ ions that caused by 4f-5d₂ and 4f-5d₁ transitions, respectively [26]. Figure 3 shows that the differences of transmittance in the long wavelength range are tiny and negligible with Gd³⁺ substitution. However, the wavelength and transmittance at the bands characteristic of Ce³⁺ absorption vary a lot as Gd³⁺ ions are introduced. This phenomenon suggests that the corresponding electron transitions of Ce³⁺ are disturbed, rather than that the optical quality of the ceramic samples is influenced.

 

Fig. 3 In-line transmittance spectra of the (Y, Gd)AG: Ce transparent ceramics (0.5 mm thick).

Download Full Size | PDF

To confirm that the wavelength and transmittance differences at bands centered around 345 and 455 nm are caused by the absorption behavior of Ce³⁺, UV-Vis absorption spectra of the (Y, Gd)AG: Ce transparent ceramics are recorded, shown in Fig. 4. There are two main absorption bands in the UV-Vis spectral range, corresponding to the 4f-5d₂ and 4f-5d₁ transitions, respectively. This is consistent with what was shown in the transmittance spectra in Fig. 3. It should be noted that the absorption intensity of 4f-5d₁ is far stronger than that of 4f-5d₂ transition, which indicates that the 4f-5d₁ has a strong ability to absorb the incident photons. Due to the 4f-5d₁ transition is critical to excitation and emission of the phosphors when pumped by a 460 nm LED, this material shows a great potential to realize high output. Another phenomenon one can see from the absorption spectra is that the Ce:YAG transparent ceramic shows strong absorption at the two characteristic bands without Gd³⁺ incorporation, while with 10 mol% Gd³⁺ introduced into the YAG lattice, the absorption intensities of both absorption bands decrease drastically. This phenomenon must be attributed to the decrease of transition probability. With more and more Y³⁺ replaced by Gd³⁺, the relative absorption intensities change little compared to that of the 10% Gd³⁺-admixed one. But the 4f-5d₁ absorption bands shift towards longer wavelength whereas those of the 4f-5d₂ transitions shift towards shorter wavelength as Gd³⁺ increases, as shown in Fig. 4. The spectral shifts of the 4f-5d₂ and 4f-5d₁ absorption bands in two opposite directions are also observed in Fig. 3, indicating that the splitting of the Ce³⁺ 5d₁ and 5d₂ states enlarges with increasing Gd³⁺ incorporation, caused by the intensified crystal field, which is consistent with that of many similar research works [13, 27]. The red shift in the absorption bands at around 455 nm explains the changes of ceramic color, and further illustrates that the 5d₁ state becomes lower towards the top of the valence band. On the contrary, it is inferred from the blue shift of the absorption bands that the 5d₂ state becomes higher.

 

Fig. 4 UV-Vis absorption spectra of the (Y, Gd)AG: Ce transparent ceramics.

Download Full Size | PDF

Strictly speaking, the absorption intensity of the 4f-5d₁ band decreases monotonously as Gd³⁺ concentration increases from x = 0 to 0.5, but it is observed to increase with further increased Gd³⁺ concentration from x = 0.6 to 0.8, even though the variations are subtle. This phenomenon is also observed at the 4f-5d₂ band. The explanation for the increase of absorption intensity should be ascribed to the deterioration of the ceramics’ optical quality when Gd³⁺ concentration is higher than 0.5. The deterioration is unambiguously reflected in the transmittance spectra in Fig. 3. Although this is not consistent with the behavior reported in the case of (Y, Gd)AG: Ce phosphor that the absorption intensity increases proportionally as Gd³⁺ increases [27], the impact of Gd³⁺ on the energy states can be proved to be great.

To be mentioned, the luminescence of the Ce:YAG phosphors is ascribed to the stimulated transition from the ground state (4f) to the excited state (5d) of Ce³⁺, and then by the radiative transition from 5d to 4f [28, 29]. Consequently, the effect of Gd³⁺ on the luminescence properties of the (Y, Gd)AG: Ce transparent ceramics should be comprehensively discussed.

Figure 5 shows the PL spectra of the (Y, Gd)AG: Ce transparent ceramics. The ceramics are excited by 440~460 nm blue-LEDs, where the LEDs match the absorption characteristic of Ce³⁺ well. The 4f-5d transition of Ce³⁺ emission shows asymmetric broad band configurations, which consist of doublet sub-emissions from 5d₁→ ²F_7/2 and 5d₁→ ²F_5/2 transitions, because the ground state of Ce³⁺ includes ²F_7/2 and ²F_5/2 sublevels after taking the spin-orbit interaction into consideration [26, 30]. As was discussed previously, partial of the blue light can be absorbed by the Ce³⁺ centers due to 4f-5d₁ transitions. The electrons are stimulated to the excited state, and simultaneously, they relax from the excited state to the ground state via radiation transition, emitting yellow and yellow-orange light. The remaining blue light penetrates through the transparent ceramics, appropriative as the other component of the light. Both the intensity of the blue light penetrating through the ceramics and that of the converted yellow or yellow-orange light are critical to the chromaticity of the LED devices. With balanced blue and yellow/yellow-orange light component, highly qualified white light can be obtained.

 

Fig. 5 (a) PL spectra and (b) Normalized PL spectra of the (Y, Gd)AG: Ce ceramic-phosphors with various Gd³⁺ substitution for Y³⁺.

Download Full Size | PDF

Table 1 shows the luminous parameters of the LED devices under blue-LED excitation using the structure depicted in the inset of Fig. 6. Without admixing with Gd³⁺, the YAG: 0.002 Ce shows a color coordinates of (0.4130, 0.5051), which is far from the ideal white light illumination with color coordinates of (0.33, 0.33). The high concentration of Ce³⁺ leads to an intense absorption of blue light and high conversion efficiency of the phosphor from blue to yellow light. As a result, the color coordinates are located in the yellow region. Fortunately, the blue light penetrating through the ceramics is found to increase, while the emitted yellow/yellow-orange light shows monotonous decrease with increasing Gd³⁺ content, as can be seen in Fig. 5(a). Therefore, with proper content of Gd³⁺ admixing, the color coordinates could be tailored to the ideal white light illumination. One can conclude from Table 1 that the color coordinates change dramatically from (0.4130, 0.5051) to (0.2755, 0.1847) as Gd³⁺ increases from 0 to 80%. Color coordinates of the samples are plotted on the CIE-1931 chromaticity diagram [31], as shown in Fig. 6, The luminescence color of the ceramic phosphors changes from yellow to near white and further to blue with successive increased Gd³⁺ content. Among which, with 10% Gd³⁺ incorporation, the color coordinates (0.3286, 0.3388) are extremely close to the ideal white illumination (0.33, 0.33).

Table 1. Luminous parameters of the WLEDs using (Y, Gd)AG: Ce ceramic phosphors.

View Table

 

Fig. 6 CIE-1931 chromaticity diagram of the (Y, Gd)AG: Ce ceramic phosphors. The inset shows the schematic diagram of the WLED device, and the luminescent photographs of the WLEDs at an operating current of 100 mA.

Download Full Size | PDF

On the other hand, Table 1 presents a decrease of luminous efficacy as a function of Gd³⁺ content. Emission intensity shown in Fig. 5(a) also reveals a decrease. These phenomena are unambiguously relative to but never solely decided by the reduction of absorption intensity of the Ce³⁺ centers in the (Y, Gd)AG garnet ceramics. As has been depicted previously in Fig. 3 and Fig. 4, the absorption of blue light decreased with increasing Gd³⁺ content, leading to a decrease of emission intensity. However, even though the absorption intensity fluctuates with more and more Y³⁺ replaced by Gd³⁺, the emission intensity decreases monotonously. Thus, there should be other interpretations for the decrease of luminous efficacy and emission intensity. According to Shao and Li [15, 32], it is the lattice parameter expansion that causes the decrease of luminescence. As more Gd³⁺ incorporated into the YAG lattice, the lattice is enlarged and would further produce a softer host lattice, the phonon energy increases accompanied by an increase of non-radiative energy transfer. As a result, the luminescence decreases. In addition to this, the splitting of the 5d state of Ce³⁺ might also contribute to the decrease. As the electrons are excited from 4f to 5d₁ energy level, certain amount of electrons would be excited to the 5d₂ state according to the Fermi-Dirac distribution [27]. It has been discussed in the former paragraph that the 5d₂ energy level shifts towards the bottom of the conduction band, whereas the 5d₁ energy level shifts towards the top of the valence band as Gd³⁺ increases, the energy gap between the two energy states is getting large. Thus the nonradiative transition from 5d₂ to 5d₁ energy level becomes less efficient, causing the population of electrons in the 5d₁ state to decrease compared to that of the ceramics with less Gd³⁺ incorporation. Therefore, in the present work, the decrease of the emission intensity might be caused by combinational cases discussed above.

Figure 5(b) showing the PL spectra with normalized intensity exhibits a red shift of the emission band from approximately 550 to 575 nm as a function of Gd³⁺ concentration. Many studies have attributed the red shift of the emission spectra to the intensified crystal field. Pan and associates hold the viewpoint that the red shift of (Y, Gd)AG: Ce emission as a function of Gd³⁺ concentration is due to the difference in ionic radius between Gd³⁺ and Y³⁺ ions, based on a strong crystal field splitting due to the lattice expansion [33]. The present work also shows a increase of lattice constant as Gd³⁺ concentration increases, however, the interpretation above is not consistent with that commonly reported in the case of expansion of a host lattice, which consider that the expansion of a host lattice would cause decrease of the crystal field splitting, and blue-shift of the 5d-4f emission [34]. Most recently, Chen and associates gave the intrinsic mechanism that causes the emission spectra’s red shift from the viewpoint of compression deformation of the electron cloud [27]. They hold the opinion that when Y³⁺ is substituted by Gd³⁺ with larger ionic radius, Gd³⁺ is subjected to a strongly compressive effect, which compels the Y 4d and Gd 5d orbital to hybridize with the O 2p orbital to form a molecular orbital. The band gap becomes narrow as more Gd³⁺ is taking place of the Y³⁺ site in the YAG crystal lattice, because of the orbital hybridization. Theoretical calculations on band structure and density of states also revealed that the d orbital hybridizes with O 2p orbital intensively, and the band gaps of the samples are getting narrower with Gd³⁺ incorporation. On the other hand, quantitative calculation of the energy states is carried by Chen’s group [27], which shows that Stokes shift (SS) takes place and increases with an increase of Gd³⁺ concentration. In conclusion, in the Gd-admixed YAG cases, the red shift of the emission bands is attributed to compression of band gaps, crystal-field splitting, and the Stokes shift. The diagram of Ce³⁺ energy levels, crystal-field splitting of 5d orbital, and Stokes shift in the (Y, Gd)AG: Ce ceramics is plotted in Fig. 7. The similar diagram was also proposed separately by Chen via theoretical calculations [27], confirming the consistent between our experimental results and the theoretical calculation results.

 

Fig. 7 The possible energy level diagram for Ce³⁺ in the (Y, Gd)AG: Ce transparent ceramics.

Download Full Size | PDF

The red shift of emission bands would enhance the red component, thus further benefit the CRI of the LED devices. As listed in Table 1, the CRI increases monotonously at first with increasing Gd³⁺ admixing. The highest CRI of the device is obtained when the Gd³⁺ content is 40%, afterwards it decreases. This is because with proper Gd³⁺ incorporation, the red shift of the emission bands enhances the red light component, thus the CRI increases. With further increased Gd³⁺, however, the intensity of blue light penetrating through the ceramics increases a lot, and the green light components are reduced, since the whole emission bands shift towards longer wavelength. The combined reasons lead to a falling of CRI.

Meanwhile, Table 1 tells a top the luminous efficacy of the samples is 42 lm/W when Gd³⁺ was added, which is much lower than that of the reported YAG: Ce phosphors [35, 36]. It is well acknowledged that the luminous efficacy is closely related to the internal quantum efficiency (IQE) and packaging efficiency (PE) [37]. IQE of the YAG: Ce transparent ceramics is about 82%, twice or three times higher than that of the YAG: Ce nanophosphors (~38%) or thin film phosphors (~25%) [38]. Consequently, the low luminous efficacy in the present case must be due to the low PE, since few technological steps have been taken to prevent scatter loss during the measurement. Well packed devices using the present ceramics are believed to have excellent luminous efficacy. In addition, the luminous efficacy decreases continuously with increasing Gd³⁺ content (Table 1). The results are consistent with the PL intensity shown in Fig. 5(a). It is noted that CRI of the sample with 10% Gd³⁺ incorporation is relatively high compared to that of the sample without Gd³⁺ incorporation. Considering the color coordinates, CRI, the luminous efficacy, 10% Gd³⁺-admixed YAG: Ce transparent ceramic is a promising candidate for WLEDs.

4. Conclusions

The Ce-doped garnet transparent ceramic phosphors with high CRI were designed for WLEDs by incorporating Gd³⁺ into the YAG lattice. Gd³⁺ ions are found critical to both the energy band of the garnets and the energy levels of Ce³⁺ in the (Y, Gd)AG: Ce ceramics. With increasing Gd³⁺ substituting for Y³⁺ in the garnet, the band structure of the (Gd, Y)AG: Ce ceramics are compressed. The 5d₂ energy level of Ce³⁺ shifts towards the bottom of the conduction band, while the 5d₁ energy state shifts towards the top of the valence band, with increased Gd³⁺ content, leading to shift of emission bands. The YAG: Ce ceramic phosphor shows color coordinates of (0.4130, 0.5051). With 10% Gd³⁺ admixing, the color coordinates are succeeded to be tailored to (0.3286, 0.3388), extremely close to the ideal white light illumination. CRI of the ceramic phosphors is also improved by Gd³⁺ incorporation, indicating that with balanced composition with respect to the Gd³⁺ admixture into the YAG structure, the ceramics are promising candidates for warm WLED devices for indoor illumination. Such an approach concerning the band structure and energy level engineering is of significant importance and might be applied to further optimize the performance of other compounds for lighting.