1. Introduction

The first decade of the 21st century has witnessed a great change in the lighting industry, as high brightness light-emitting diodes have been encroaching (LEDs) the market share of traditional light sources [1,2]. The current commercial white LEDs (WLEDs) consist of InGaN/GaN multiple quantum wells (MQWs) blue chips and a combination of yellow phosphor blended with organic resins. Among various types of phosphors, cerium doped on yttrium aluminum garnet (Ce:YAG) proves to be the most suitable one for white light generation. However, as the work environment of high power LEDs are always of both high irradiation and high temperature, the chemical interaction on the interface of phosphor powders and gel would cause severe carbonization and therefore deteriorate the luminous efficiency as well as induce perceivable color shifting [3,4]. To improve the long-term performance, researchers have developed alternative materials, among which the polycrystalline Ce:YAG transparent ceramics are considered as an promising substitution for its excellent optical property and high thermal conductivity [5–8]. These ceramic-based LEDs nonetheless share the same deficiency in chromaticity as do traditional LEDs, i.e. high correlated color temperature (CCT) and low color rendering index (CRI), both as the result of lack of red components in the emission spectrum. A practical solution is to induce red-shift in emission spectrum by substituting Y³⁺ with larger Gd³⁺ ions, which has been studied in the literatures [7–10]. Nishiura et al. has reported that such modification could induce red-shift to the degree of 10-30 nm in the photoluminescence with the resultant changing in chromaticity of the total light emission. For this Gd³⁺ doped solution, room remains to be further improved, especially for the luminous efficiency, of which the data has not been reported in the literature yet.

In this letter, we report our recent experimental results of transparent Ce:YAG ceramic phosphors in various Gd³⁺ doping concentrations. Ceramic samples with good performance were obtained by solid-state reaction [11–13]. In-line light transmittances of the ceramic samples are above 81% and few pores or impurity phase is observed. We have investigated the emission wavelength shifting and Quantum yield (QY), as well as, when combined with InGaN blue chips, the chromaticity of the white light emission, upon the Gd³⁺ doping concentration. Results suggest an optimum Gd³⁺ doping concentration of ~10%, on which the luminous efficiency (LE) of WLED lamp reaches the maximum of 128.15 lm/W at forward current of 20 mA, with a decent CRI of 78.1.

2. Experiment

2.1 Material fabrication

We employed high-pure α-Al₂O₃ (99.99%;Sumitomo Chemical Co. Ltd, Osaka, Japan), Y₂O₃ (99.99%; Alfa Aesar, WardHill, MA), Gd₂O₃(99.99%; Alfa Aesar) and CeO₂(99.99%; Alfa Aesar) as the precursor material and the 0.5 wt% Tetraethoxysilane (TEOS, 99.999 + %, Alfa Aesar, United States) and 0.5 wt% oleic acid (99%, Alfa Aesar,United States) as sintering aid and dispersant agency, respectively. Such precursor materials were weighted according to the chemical composition of (Gd_xCe_0.06%Y_99.94%-x)₃Al₅O₁₂ (x = 0, 0.05, 0.10, 0.15, 0.20). In this work, Ce³⁺ concentration was kept on 0.06% to ensure a proper and uniform thickness over samples. Mix all the materials in ethanol in planetary-milling machine for 24h, and set the weight ratio of m_p:m_s:m_m on 1:1:3. (p, s, and m denote powder, solvent and mill-ball respectively) Dry the as-obtained dispersive slurry, which was then grounded and sieved in sequence through 100-mesh and 200-mesh screens to obtain target powders with the size falling in between 75 and 150 microns. These powders were uniaxially pressed into Φ16 mm disks at about 8 MPa and then were de-bindered at 750 °C for about 10 h in oxygen, followed by the cold-isostatic-presses at 200 MPa. Then we sintered the green compacts under 1.5 × 10-6 torr vacuum condition at 1800 °C for 10 h. After annealing and mechanical polishing, we obtained the high quality transparent (Ce,Gd):YAG ceramics.

2.2 Properties characterization

We performed the morphology observation through a JSM-6700F field emission scanning electron microscope (FE-SEM, JEOL, Japan), obtained the in-line transmittance by using an UV/Vis/NIR spectrophotometer (Lambda-900, PerkinElmer, USA), and recorded the photoluminescence (PL) with a spectrometer (FLS920, Edinburgh instrument, United Kingdom) in conjunction with an xenon lamp as the excitation source.

The QY, a quantitative measure of the light conversion ability of a fluorescent body, is defined as the ratio of the numbers of emitted photons out of the fluorescent body (n_em) to the numbers of photons it absorbs (n_abs) during a unit time period. It was measured by a spectrometer specifically designed for the measurement of fluorescent materials (FLS920), including an integrating sphere, in the center of which are the ceramics samples located, and the light is extracted through a fiber and collected by a spectrometer. The whole system was calibrated via a standard tungsten lump. We measured the chromaticity parameters such as CIE color coordinates, CRI, CCT and luminous efficiency by using the integrated optical and electrical measuring system for LEDs (Hangzhou Everfine Photo-electricity Information Co, Ltd. China), the forward current of the blue chip was fixed at 20 mA.

3. Results and discussion

We prepared 5 samples in this work, illustrated in Fig. 1, and each with different Gd³⁺ concentrations, from left to right are of 0%, 5%, 10%, 15% and 20%, respectively. Photographs of transparent Ce:YAG ceramics plates doped with various Gd3 + concentrations are shown in Fig. 1(a). All the samples exhibit excellent transparency. As the Gd³⁺ concentration increases, the color of ceramics samples gradually turns from yellowish green to pure yellow. Provided with the high hardness of ceramics, we are able to reinforce WLED with robust structures by letting the ceramics serve as the encapsulations. As illustrated in Fig. 1(b),we mounted the InGaN/GaN LED chips on a metal-backed printed circuit board (MBPCB), and then covered them with the ceramics sample, which in this way severs simultaneously as light-converter and encapsulation. In this structure, given the high thermal conductivity of ceramics, the heat generating in the chips dissipate in two paths: one is towards the beneath MBPCB; the other towards the ceramics encapsulation, rather than sole downwards path in those traditional structures. A light source features such ceramics LED is shown in Fig. 1(c).

 

Fig. 1 (a) Photographs of transparent Ce:YAG ceramics doped with various Gd³⁺ concentrations (from left to right: 0 at%,5 ta%, 10 at%, 15 at%, 20 at%); (b) WLED encapsulated by a piece of Gd³⁺ doped Ce:YAG transparent ceramics; (c) The light source based on such ceramics LED.

Download Full Size | PDF

The results of FE-SEM are shown in Fig. 2. Under the microscope, the ceramics exhibit a beehive-like structure, in which clear boundaries distinguish neighboring grains. Except for the last sample with 20% Gd³⁺ contents, there are no obvious pores observed inside or outside grains of the rest four samples. The grain sizes of all samples ranges from 10 to 50 μm and no abnormal grain is found.

 

Fig. 2 FE-SEM images of Ce:YAG transparent ceramics samples doped with various Gd³⁺ concentrations: (a) 0 at%, (b) 5 at%, (c) 10 at%, (d) 15 at%, and (e) 20 at%.

Download Full Size | PDF

The plot of in-line transmittance of the samples with different Gd³⁺ concentrations is shown in Fig. 3(a). The transmittance of the Gd³⁺-free ceramics sample reaches 82.3% and that of Gd³⁺ doped ones are about 81% in the wavelength range of 530-780 nm. Three absorption peaks, located on 225 nm, 340 nm and 460 nm, originate from the 4f-5d transition of Ce³⁺, which has also been demonstrated by in the literatures [14–17]. Compared with the ceramics sample without Gd³⁺ doping, and for Gd³⁺ doped ones, those absorption peaks located on 275 nm region correspond to the transition from the ⁸S_7/2->⁶I_J in Gd³⁺ ions, which has been reported in the literature [18].

 

Fig. 3 (a) The in-line transmittance, (b) the photoluminescence spectra of the Ce:YAG transparent ceramics samples doped with various Gd³⁺ concentrations.

Download Full Size | PDF

Figure 3(b) shows PLs of all the samples excited by a Xenon lamp at room temperature. The broad Ce³⁺ emission band in the YAG lattice originates from the 4f-5d electronic transition which results from the crystal field splitting on the 5d level of Ce³⁺. The gap exists between the ionic sizes of Gd³⁺ and Y³⁺ leads to lattice expansion and thus enhances the crystal field spitting, which reveals itself in the emission as red-shift, that is, by replacing Y³⁺ with Gd³⁺, the peak red-shifts from 534 to 565 nm with the increasing Gd³⁺ concentration [9,19]. Such red-shift in spectrum has a great influence on luminescent properties of WLED, which will be discussed in the subsequent sections.

In the phosphor-converted LEDs, 3 main factors are bearing on the LE, QY, Stokes Loss, and the luminous efficacy of radiation (LER). It is firstly worth discussing the effect of doping concentration on the QY of samples. The plot of QY vs. Gd³⁺ concentration is illustrated in Fig. 4(a). The QYs of the sample with concentration lower than 20% are close to each other, above 90%, despite a slight decrease with the increasing concentration, whereas the QY of 20% sample is distorted to 84.7%. With the help of the FE-SEM image of these samples, we consider this distortion as an aftermath of the structural imperfection, that is, pores amid the grains could serve as distraction centers that absorb photons, thus reducing the QY. On the other hand, the high QY on those samples with lower concentration benefit from the excellent micro structures. The Stokes loss is another critical issue, which is the energy loss during a high energy photon (e.g. photon of blue light) being converted to a lower one (e.g. photon of yellow light). In this consideration, even though the QY of ceramic is 100%, there remains inevitable energy loss on generating the yellow light. The Stokes Loss ratio η can be determined via Eq. (1).

 

Fig. 4 (a) The measured QY and LE upon Gd³⁺ concentrations; (b) The theoretical LER and Stokes loss data upon Gd³⁺ concentrations.

Download Full Size | PDF

In Eq. (1), E(Δ) is relative intensity of different wavelength. In the case of phosphor-converted LED, the Stokes loss relies closely on the energy difference between the yellow photons and the blue photons. The increasing Gd³⁺ concentration induces the red-shift on the phosphor peak, which consequently enlarges the energy deviation from the blue light and enhances Stokes loss. Data are listed in the Table 1 and plotted in Fig. 4(b). However, this adjustment in the spectral shape raises the LER, of which the definition is in Eq. (2).

Table 1. Results of QY, Stokes Loss and LER

View Table

In Eq. (2), K_m is a constant equal to 683lm/W, V(Δ) is visual acuity function of human eyes. Table 1 lists the data upon the Gd³⁺ concentrations. The net LE would be affected by all the 3 issues. Increasing the Gd³⁺ concentration obtains high-LER spectral, but with a high Stocks loss and lower QY samples, which in turn would increase the energy consumption, yielding even lower net LE. In Fig. 4, the LE decreases monotonically from 133.2 to 117.7 lm/W, which is consistent with results in the literature [9].

As is shown in Fig. 5(a), the chromaticity coordinate of blue emission from the chip is located in the blue region of CIE 1931 color diagram, while a series of points corresponding to the phosphor emission on different Gd³⁺ concentrations is located in the yellow region, among which the shift associates with the doping-induced red-shift of the fluorescence spectra. The color coordinates of net white light of all sample are all located in the vicinity of the white point: (0.33, 0.33), right on the lines connecting the blue point and the related yellow points. The CCT decreases from 6869 to 4537 K as the Gd³⁺ concentration increases from 0 to 20%. The CRI increases from 69.0 to 78.1 when the Gd³⁺ concentration increases from 0% to 10% and then drops to 65.3 as the Gd³⁺ concentration goes on increasing to 20%, as plotted in Fig. 5(b). It contradicts to previous results in the literature, which identified monotonically increasing CRI [7]. According to Lin’s theory, a light source with high color rendering capacity should have spectral components that match the whole CRI-sensitive regions. Hence, in this case, the decreasing in CRI could be ascribed to the discrepancy, or the intensity loss between the CRI-sensitive region and the spectral peak, i.e., the over-doped Gd³⁺ induces too much red-shift that the yellow peak has again moved away from the CRI-sensitive region [20]. On this consideration, choosing a proper Gd³⁺ doping concentration matters a lot in the light source design.

 

Fig. 5 (a) The chromaticity coordinates of emission from blue chips, ceramics and net white light on CIE 1931 color diagram; (b) CCT and CRI data of the net white light from LEDs based on Ce:YAG ceramics doped with various Gd³⁺ concentrations.

Download Full Size | PDF

4. Conclusion

In summary, doping Gd³⁺ ions into Ce: YAG transparent ceramics can induce red-shift in the fluorescent peak, while maintaining high in-line transmittance within visible spectral region. Its utility in the solid-state lighting enhances the color rendering capability for adding red portion into the spectra. We investigate in this work the connection between chromaticity parameters and the doping concentration, and demonstrate that the optimized concentration should be around 10%, which ensures the highest CRI as well as a decent efficiency.