1. Introduction

Since the first research was reported in the 1960s [1], studies on visible spectrum light-emitting diodes (LEDs) have progressed rapidly. As a practical optoelectronic component that lacked decisive technologies, the LED was first commercially applied only in indicators and digital numeric readouts; it was not widely used for illumination until the high-luminosity blue LED was developed [2–4]. The advent of blue LEDs was a key step in making solid-state lighting a potential candidate for replacing traditional light sources such as incandescent lamps and compact fluorescent lamps.

Numerous manufacturers worldwide fabricate white LEDs as well as LEDs of other colors. Various methods are used to realize a white LED; the two most commonly used approaches are to use a blue LED to pump a yellow phosphor [5,6] or mixed red and green phosphors [7–11], or to combine three or more LEDs with distinct emission spectra spaced across the visible spectrum. In the latter method, red, blue, and green LEDs, and sometimes amber LEDs, are usually chosen; thus, it is called the RGB or RGBY system. The RGB system is flexible regarding the available colors, but it requires complex circuits for calibration to obtain the desired result. Further, it suffers a critical problem with thermal effects. During continuous operation, the elevated temperature greatly decreases the light output of the amber and red LEDs and slightly changes that of the blue LEDs, shifting the color and reducing the efficacy. The former method uses a combination of blue or near-ultraviolet (near-UV) LEDs and phosphors to generate white light. It requires no individual control, since the only type of LED used is blue, and the performance is excellent and stable. Because of its efficacy and simplicity of design, the first method is adopted more often than RGB white LEDs. To apply this method, a gallium-nitride-based blue LED chip is usually adopted and coated with a cerium-doped yttrium aluminum garnet (Y₃Al₅O₁₂:Ce³⁺, or YAG:Ce) phosphor [12,13]. Therefore, products made by this method are called phosphor-converted LEDs. Blue LEDs have been extensively used for integration with materials that possess wavelength-converting properties to generate dichromatic light that humans perceive as white light [14,15]. The rapid development of both technologies, that is, III-V nitride compound semiconductors and wavelength-converting phosphors, has given white light LEDs more powerful advantages. For example, their high energy conversion efficiency reduces the power loss and increases the emitted luminosity, and they contain no mercury; both points are considered as energy saving and environmentally friendly. In addition to their small volume, fast response, and variety of colors, these properties make them feasible for various applications. Further, they have additional desirable properties such as saturated color, long lifetime, and reliability [16]. Possessing these strong advantages, LEDs are deservedly regarded as potentially the most powerful light source in next-generation lighting.

Although white LEDs possess powerful advantages, their characteristics are not easy to control. Owing to unavoidable fluctuations during fabrication, the colors of light they produce are not uniform, even for products in the same batch. The products must be categorized for different uses with respect to their color. The most commonly used metric in the description of their color is the correlated color temperature (CCT) [17] which is defined by the comparison between the hue of the light source and that of an ideal black-body radiator that radiates light with a similar color at a specific temperature. Applying remote phosphor technology offers an opportunity to calibrate the color shift of the final products or to improve the performance of LEDs by designing the structure of the phosphor layer [18–20]. Researchers have also developed CCT-variable LED devices that can overcome the problem of uncontrollable CCT and expand their applications. The variation is obtained by adjusting the relative intensity of individual LED chips that emit different colors to produce white light of various chromaticities [21–23]. In general, this method can yield white light of precisely the desired chromaticity. However, individually controlling each LED chip to match a specific chromaticity requires complex electronic circuits and algorithms and may be costly to integrate with other devices. In this study, we introduce an alternative method of realizing a CCT-tunable white light whose chromaticity can be accurately controlled. The concept uses a liquid crystal (LC) cell that can act as a light regulator, in combination with blue LEDs and a remote phosphor, to generate CCT-variable white light. The method requires no complex circuit and is simple to implement. Further, CCT variation during operation is achieved quite intuitively.

2. Experiments

Royal blue (Luxeon Rebel, Philips) surface-mount device (SMD) LEDs were chosen as the light source in this experiment. The SMD LEDs were arranged in a 3 × 3 array and welded onto a printed circuit board for further area allocation. The chosen yellow phosphor, YAG:Ce (from LWB GmbH), was carefully blended with silicone gel (Sylgard 184, Dow Corning). The well-mixed phosphor/silicone mixture was spin-coated onto a cleaned glass substrate at a low spin rate to form a uniform layer. After the silicone polymerized, the phosphor layer was then segmented into 3 × 3 square arrays to match the arrangement of the LEDs.

The modulating device, an LC cell, was constructed mainly from two pieces of indium tin oxide (ITO)-coated glass with LCs between them. The detailed fabrication processes were as follows: two pieces of well-cleaned ITO glass were spin-coated with polyimide (PI), which acted as alignment layers, and a subsequent series of post treatments, including soft baking at 80 °C for 10 min and hard baking at 220 °C for 1 h, was executed immediately to avoid gradual corruption. The subsequent rubbing procedure, which can ensure an orderly arrangement of the LC molecules in a particular direction, was performed using a rolled-up velvet cloth with high-speed rotation to scratch and form uniform nanogrooves on the top of the PI layers. Silica spacers 4 μm in diameter were then evenly sprayed on top of the PI layer to maintain a uniform separation between the two pieces of glass. In the combination process, the paired glasses were placed face to face with their rubbing directions perpendicular to each other and fixed with UV glue. Nematic LCs (ZCE-5099LA, Chisso) were injected through the assistance of capillary force, and the final seal with UV glue was executed immediately after the cell was completely filled with LCs. The structure of an as-prepared LC cell is shown schematically in Fig. 1(a).

 

Fig. 1 (a) Schematic structure of the LC cell. (b) Arrangement of the LEDs, phosphor, and LC cell.

Download Full Size | PDF

The LC cell was divided into nine regions consisting of 3 × 3 arrays that match the area arrangement of the phosphor pattern. Polarizers with the same polarization direction were attached to four of the nine regions, as shown in Fig. 1(b). Another four polarizers were attached to the opposite side of the cell, and their polarization directions were set to be perpendicular to those of their counterparts on the other side of the LC cell. These polarizer-covered areas were defined as active areas.

The parts mentioned above were arranged as follows: the LC cell was placed at the top, and the phosphor pattern was located beneath it. Further, the light source, i.e., the LED array, was located at the bottom. Figure 1(b) shows a schematic illustration of the arrangement. Because LEDs generate heat during emission, some preventive means should be considered to maintain good functioning of the LC during the experiment. The phase transition temperature of the LC used in this study is about 163 °C. To avoid direct heat transfer, the LC cell was placed separately above the LEDs. Before the measurement, the board that carried the LEDs was placed upon a copper plate to dissipate the excess heat generated by the LEDs. The entire device was covered with a dome-shaped diffuser for good light mixing.

3. Results and discussions

The emission spectra of the LED and phosphor used in this experiment are shown in Fig. 2. The spectra were acquired by a spectrometer (SD1200, OTO Photonics), and the fluorescence spectrum of the phosphor was measured under irradiation by the blue LEDs.

 

Fig. 2 Emission spectra of royal blue LEDs (blue line with solid circles) and yellow phosphor (red line with triangles).

Download Full Size | PDF

To mix the spectra well, a dome-shaped diffuser was used. According to former studies, the dome-shaped diffuser possesses higher spatial-intensity and spatial-spectral uniformity owing to interior reflection [24]. A well-designed dome-shaped diffuser can provide an essentially omnidirectional emission pattern, which is desirable for general lighting applications, even from a directional-emission light source. To analyze the light-mixing properties, the uniformity of the intensity and CCT were determined by analyzing 80 points from the top of the light source [see Fig. 3(a) and (d)] with the LC cell and phosphor pattern installed.

 

Fig. 3 Relative positions of 80 points measured from the top of light source (a) without and (d) with dome-shaped diffuser. (b) Intensities and (c) CCT distributions of the corresponding points from (a); (e) measured intensities and (f) CCTs from (d).

Download Full Size | PDF

Figure 3(a) shows the measured positions of the light source without the dome-shaped diffuser, and their relative intensities and CCTs are mapped in Fig. 3(b) and (c), respectively, which reveal the inevitably non-uniform distribution. When the dome-shaped diffuser was applied, the intensity [Fig. 3(e)] and CCT [Fig. 3(f)] became fairly uniform. The angular distribution of the intensity in both setups (with and without the dome-shaped diffuser) are plotted in Fig. 4.

 

Fig. 4 Angular distribution of intensity for setup with (solid line) and without (dotted line) dome-shaped diffuser installed.

Download Full Size | PDF

Figure 4 shows that the intensity distribution is widely dispersed into a large angular region when the dome-shaped diffuser is used. Comparing to the original intensity (without the dome-shaped diffuser) that decreased by about 20% around 30°, the intensity with the dome-shaped diffuser retained more than 80% within 60°. Although the result did not reach the standard of “omnidirectional emission” defined by ENERGY STAR, the dome-shaped diffuser did provide fairly good diffusion for further measurements.

When a driving voltage was applied to the LC cell, the transmittance of the active areas gradually changed (the change started at 3 V and ended at 8 V). The active areas were the only regions that dominated the variation of transmitted blue light for light mixing. As the transmittance of the active area changed under the applied voltage, the resulting spectrum of the mixed light was altered as well. The spectra of the mixed light under various applied voltages are shown in Fig. 5.

 

Fig. 5 Spectra of mixed light under various applied voltages.

Download Full Size | PDF

Figure 5 shows that the intensity in the short-wavelength region gradually decreases as the applied voltage increases. The reduction resulted from the change in the transmittance in the active area. In the experimental setting, the source light from the blue LEDs passed evenly through both active areas and phosphor-patterned regions, which did not overlap. In the phosphor regions, most of the blue light was absorbed and stimulated the phosphor to emit yellow light, whereas a small amount of blue light was transmitted and penetrated the phosphor layer. The only light that passed through the active areas was blue light emitted from the LEDs. The intensity of the blue light passing through these areas was moderated by adjusting the applied voltage. This light was then mixed with the yellowish white light from the phosphor regions by the dome-shaped diffuser, which generated uniform light with a tunable spectrum in the short-wavelength region. The corresponding 1931 Commission Internationale de L'éclairage (CIE) chromaticity coordinates [25] of each spectrum in Fig. 5 were calculated and plotted in Fig. 6. Each dot in Fig. 6 corresponds to a different voltage applied to the LC cell from 3 to 8 V at 0.5 V intervals.

 

Fig. 6 CIE chromaticity diagram of mixed light obtained under different applied voltages.

Download Full Size | PDF

Figure 6 shows that the variation track is a straight segment. The variation was smooth, and it ranged from about 4500 to 6900 K, which ran across the black-body locus (BBL). According to the figure, the point located precisely on the BBL corresponds to an applied voltage of 4.5 V. With the applied voltage which can be finely adjusted, the light could be precisely tuned to an expected CCT along the variation track.

The use of an LC cell as a light regulator for transmittance adjustment is simple and precise. Its adjustment is linear, smooth, continuous, and reproducible, properties that are still problematic for current tunable LEDs. However, the operation of an LC cell requires two polarizers, which will unavoidably reduce the transmittance of the light. In this study, we lowered the proportion of the active area to diminish the light loss, and the result showed that the insertion of the LC cell reduced the efficacy by about 20%, with the CCT variation remained in a useful range. In the above experiment, the CCT variation track ran across the ANSI white quadrangles, a standard description of chromaticity for LEDs and solid-state lighting, with a certain slope. This slope was related to the CCT of the LEDs and phosphor, and the variation track was part of the line that linked both CCT points. Therefore, if both the LEDs and phosphor had been appropriately chosen, the CCT variation track would span the region of the ANSI white quadrangles near the BBL. To achieve this goal, the slope of the track should be mitigated; that is, LEDs and a phosphor with appropriate CCTs should be chosen. In the following experiment, cyan LEDs (Luxeon Rebel, Philips), which possess a higher position in the chromaticity diagram, were substituted for the royal blue LEDs. Further, the phosphor was modified as well. It was slightly doped with red phosphor (Ca₂Si₅N₈:Eu²⁺, Dott Technology) and therefore emits chrome yellow light. The spectra and CCT variation of the mixed light after these adjustments are shown in Fig. 7.

 

Fig. 7 (a) Spectra and (b) CIE chromaticity diagram of modified mixed light under different applied voltages.

Download Full Size | PDF

Figure 7(a) shows the spectra of the mixed light under various applied voltages. The trend of the CCT variation is similar to that in the first experiment. The spectrum shows a gradually quenched peak in the short-wavelength region due to activation of the LC cell in the active areas. The variation in the chromaticity diagram is shown in Fig. 7(b). The CCT variation begins at about 6100 K and ends at 3800 K, corresponding to an applied voltage of 3 V to 8 V. For the modified LEDs and phosphors, the entire variation track falls within the region of the ANSI white quadrangles, and its slope is commensurate with the BBL. The result demonstrated that the CCT variation could be easily and precisely modulated by appropriately adjusting the CCT of the LEDs and phosphor. In other words, with a better choice of blue LEDs in combination with suitable phosphors, a white LED with a highly useful variation range could be fabricated.

4. Conclusion

In conclusion, LEDs have shown their powerful potential as a replacement for traditional light sources, but some problems remain to be overcome. White LED lamps with tunable chromaticity are a good alternative solution, and we proposed a concept for achieving this goal by introducing an LC cell as a light regulator. By controlling the voltage applied to the LC cell, the intensity of the blue light in the mixed white light could be accurately manipulated, thereby generating CCT-tunable white light. We demonstrated that the CCT variation track was linear, and its slope could be regulated by choosing LEDs and a phosphor with an appropriate CCT. A modified LED exhibited a CCT variation track near the BBL and spanning the ANSI white quadrangles. This method greatly simplifies the complexity of the electronic circuit and algorithms that traditional CCT-variable LEDs require; additionally, by applying LC technologies, many aspects of the performance of the integrated LEDs can be improved; for example, doping fluorescent quantum dots into the LC cell is expected to increases the color rendering index of white LEDs. The method also provides a flexible concept for manipulating the CCT of LEDs for further applications.