Effective Heat Dissipation from Color-Converting ... - ACS Publications

Dec 9, 2015 - application in a remote phosphor configuration of high-power white light emitting diodes (WLEDs) .... International Commission on Illumi...
0 downloads 12 Views 7MB Size
Download PDF
Effective Heat Dissipation from ColorConverting Plates in High-Power White Light Emitting Diodes by Transparent Graphene Wrapping Eden Kim,†,⊥ Hyeon Woo Shim,§,⊥ Sanjith Unithrattil,†,⊥ Yoon Hwa Kim,†,⊥ Hojin Choi,§ Ki-Jin Ahn,§ Joon Seop Kwak,∥ Sungmin Kim,‡,§ Hyeonseok Yoon,*,‡,§ and Won Bin Im*,† †

School of Materials Science and Engineering and Optoelectronics Convergence Research Center, ‡Alan G. MacDiarmid Energy Research Institute, School of Polymer Science and Engineering, and §Department of Polymer Engineering, Graduate School, Chonnam National University, Gwangju 61186, South Korea ∥ Department of Printed Electronics Engineering, Sunchon National University, Jeonnam 57922, South Korea S Supporting Information *

ABSTRACT: We have developed a hybrid phosphor-in-glass plate (PGP) for application in a remote phosphor configuration of high-power white light emitting diodes (WLEDs), in which single-layer graphene was used to modulate the thermal characteristics of the PGP. The degradation of luminescence in the PGP following an increase in temperature could be prevented by applying single-layer graphene. First, it was observed that the emission intensity of the PGP was enhanced by about 20% with graphene wrapping. Notably, the surface temperature of the graphene-wrapped PGP (G-PGP) was found to be higher than that of the bare PGP, implying that the graphene layer effectively acted as a heat dissipation medium on the PGP surface to reduce the thermal quenching of the constituent phosphors. Moreover, these experimental observations were clearly verified through a two-dimensional cellular automata simulation technique and the underlying mechanisms were analyzed. As a result, the proposed G-PGP was found to be efficient in maintaining the luminescence properties of the WLED, and is a promising development in high power WLED applications. This research could be further extended to generate a new class of optical or optoelectronic materials with possible uses in a variety of applications. KEYWORDS: LEDs, graphene, heat dissipation, cellular automata

W

conformal LED packaging, unfortunately, a large portion of light, isotropically emitted by the phosphor, is reabsorbed by the LED chip. In addition, the separation between the LED chip (which is the source of heat) and the phosphor significantly reduces the temperature of the latter and, as a result, delivers maximum efficiency. Thus, the light extraction efficiency of the remote phosphor configurations is superior to that of conventional conformal LEDs. The remote phosphor configurations studied over the past few years have been approached by employing different materials, designs, and auxiliary components.7 Prominent among these configurations were phosphor glass ceramics, luminescent glasses, and phosphor-in-glass plates (PGPs). Glass ceramics, which require a rigorous synthesis process, offer very

ith the advent of high-efficiency blue light-emittingdiodes (LEDs) and broadband emitting phosphors, phosphor-converted white LEDs (WLEDs) have received much attention. Even though new WLED configurations and new packaging techniques have been developed with the goal of eventually replacing the conventional incandescent light in many applications including general illumination sources,1−4 only a few of the common drawbacks of these configurations have been rectified, and many longstanding problems remain. Among these newly proposed configurations, the remote phosphor configuration of the WLED is the most promising.5 In particular, the aging of the WLEDs, determined by the durability of the packaging materials, was reported to be slowed down by more than a factor of 2, and the efficiency of the devices was increased by more than 20% compared to that of WLEDs in which conventional packaging techniques are employed.6 This increase in the efficiency is due to the large separation between the LED chip and the phosphor; in a conventional and © XXXX American Chemical Society

Received: June 24, 2015 Accepted: December 9, 2015

A

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (a) High resolution transmission electron microscope image of the single graphene layer showing the interface, (b) atomic force microscope (AFM) images of the deposited graphene layer illustrating the single-layer features and wrinkles, (c) Raman map of the 2D-to-G peak intensity ratio of the graphene on a Si/SiO2 wafer, (d) transmittance spectra of the chemical vapor deposited graphene layer, and (e) schematic diagram showing the transfer of graphene from Cu foils to the phosphor-in-glass plates (PGP) by a poly(methyl methacrylate)assisted transfer method. (f, g) Large-area AFM topography images of the PGP surface before (f) and after (g) graphene coating, and (h) the contact angle of the PGP before and after coating.

limited flexibility in tuning the emission properties. Moreover, many of the commercial phosphors cannot be incorporated into glass ceramics because of the nonexistence of the corresponding glass compositions. Luminescent glass, which has a lower conversion efficiency compared to ceramic phosphors, proved to be an unfavorable candidate. On the other hand, PGPs that are formed by enclosing ceramic phosphors in a glass matrix offer a wider extent of tunability and flexibility in the choice of the component phosphor. The fabrication of the PGPs requires the least sophisticated techniques and offers a protective environment for chemically unstable phosphors inside the glass matrix.8 LEDs operated under high power conditions generate an enormous amount of heat. Although the color converting plate has very low thermal conductivity and is separated from the LED chip, which is the source of heat, prolonged operations under a high forward bias current cause heat to accumulate in the plate. This heat then causes phonon relaxation of the excited electrons and can substantially reduce the efficiency of the constituent phosphors of the color converting plates and thereby decrease the overall efficiency.9,10 In addition, for colorconverting plates in which multiphosphors are employed, the thermal quenching of the luminescence varies among the

constituent phosphors and the disproportional variation in the efficiency of the constituent phosphor causes a shift in the overall emission color of the device. The fact that each phosphor has a different characteristic temperature dependence of luminescence makes such a device unfit for general illumination purposes. Multiphosphor techniques can only be undertaken if the operating temperature of the phosphors is kept within the limits of the predetermined window, where all the component phosphors have similar temperature dependence. Owing to its unique properties, graphene has received tremendous attention in recent years.11 Optoelectronic applications of graphene have mostly been limited to transparent electrode materials;12−15 however, several articles have described the use of graphene for better thermal management in the interior of LED chips.16−18 In this work, we propose wrapping a PGP, but not the LED chip itself, with a single-layer of graphene to circumvent the adverse temperature dependence in emissive performance, and in turn to achieve long-term reliability for prolonged operation of the LED. We also provide an in-depth scientific insight into how wrapping graphene affects the main characteristics of the PGP and determines its performance. Graphene with high thermal B

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Figure 2. (a, b) Electroluminescence (EL) spectra of the white light emitting diode (WLED) employing (a) bare-PGP (B-PGP) and (b) graphene-wrapped-PGP (G-PGP). (c) EL of G-PGP in comparison with B-PGP, (d) transmittance of G-PGP and B-PGP, and (f) International Commission on Illumination (CIE) color coordinates of the WLED with G-PGP and B-PGP under an operating current of 350 mA.

conductivity19 effectively drives all the heat accumulated at the center of the PGP to the periphery of the plate. This reduces the thermal energy reaching the bulk of the PGP. The heat that spreads over the surface of the plate is finally dissipated through convection. As part of our study, the mechanisms underlying the process of heat transfer and temperature distribution were calculated using a new cellular automata simulation technique. To the best of our knowledge, this is the first attempt to systematically analyze the effect of graphene wrapping on PGP.20

substrate transfer process can cause cracks, folds, tears, or residues on the graphene layer. The quality and stacked layer number of the graphene was further examined through largearea Raman spectroscopic analysis. Figure 1c shows a Raman mapping image of the 2D-to-G peak intensity ratio of the graphene. The Raman image demonstrates that the graphene is composed entirely of a single-layer. During the substrate transfer process, as seen in Figure 1b, the graphene was partly wrinkled, which would result in minor fluctuations in the intensity ratio. The thermal conductivity of the graphene was measured by an optical method using Raman spectroscopy.21 The graphene deposited on the Si/SiO2 substrate had a thermal conductivity of 978 W m−1 K−1 (assuming a temperature coefficient of −1.62 × 10−2 cm−1/K);22,23 this was somewhat higher than the thermal conductivity of the supported singlelayer graphene reported previously.24−26 Lastly, the transmittance of the graphene layer of between 400 and 800 nm was above 95%, as shown in Figure 1d. The high transmittance in the visible range is one of the advantages of graphene over other high thermal conductivity materials.27 An overall process for the graphene wrapping of the PGP is illustrated in Figure 1e. Both sides of the PGP were wrapped with two pieces of the CVD graphene, employing a poly(methyl methacrylate) (PMMA)-assisted transfer method to move the graphene (grown on copper) to the PGP. Largearea AFM topography images of the PGP surface before and after graphene wrapping are shown in Figure 1f and g, respectively. The coating of the graphene has reduced the

RESULTS AND DISCUSSION Single-Layer Graphene Wrapping. In order to completely wrap the PGP in graphene, single-layer graphene of excellent quality was initially prepared on a copper foil by chemical vapor deposition (CVD). A representative highresolution transmission electron microscopy (HRTEM) image, directly showing the edge of the as-grown graphene, is presented in Figure 1a. The HRTEM image clearly illustrates the single-layer of graphene. The atomic force microscopy (AFM) images and Raman spectra of the graphene layer were measured after transferring the graphene layer from the copper foil to a silicon wafer. The AFM image shown in Figure 1b depicts the surface topology of the graphene. The graphene showed mild roughness and height fluctuation with a rootmean-square (RMS) roughness of about 0.7 nm (see Figure 1b inset). A linear texture was observed on parts of the AFM image, indicating a source of mild roughness. Commonly, the C

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano Table 1. Comparison of Variation in Optical Parameters of WLEDs Employing B-PGP and G-PGPa B-PGP bias current (mA) 100 150 200 250 300 350 a

CIE (x, y) 0.411, 0.422, 0.428, 0.432, 0.435, 0.437,

G-PGP

color rendering index, Ra

CCT (K)

luminous efficacy (lm/W)

92.3 93.9 92.8 91.5 90.6 89.9

4258 3732 3522 3414 3345 3290

5.06 3.91 3.30 2.88 2.54 2.31

0.405 0.415 0.420 0.425 0.428 0.430

CIE (x, y) 0.407, 0.418, 0.424, 0.429, 0.432, 0.434,

0.404 0.414 0.420 0.424 0.427 0.430

color rendering index, Ra

CCT (K)

luminous efficacy (lm/W)

93.4 92.1 91.0 90.1 89.5 89.0

3775 3598 3510 3460 3414 3391

5.01 4.23 3.66 3.22 2.88 2.88

CCT: Correlated color temperature.

Figure 3. (a) Top and cross-sectional views of the WLED configuration, (b) relative difference in emission intensity of the configurations as a function of bias current, (c) thermal imaging of the LEDs under operation after different operating times, and (d) temperature on both surfaces (green rectangular domains) of the PGPs measured using a thermocouple as a function of the time of operation.

graphene is more hydrophilic than previously assumed and acts as a translucent barrier, transmitting some of the watersubstrate interactions.30−33 Of course, the contact angle is also influenced by other factors, such as surface roughness and contaminants. Finally, the increased contact angle indicates that the PGP surface becomes more hydrophobic to repel water molecules in air. The wrapped graphene may also render a physical barrier to prevent the direct contamination of PGP by volatile organic compounds.31 PGP Luminescence. The electroluminescence (EL) spectra of the configuration, using as-prepared PGP with three distinguishable emission bands, are shown in Figure 2a. The blue emission band at around 450 nm originates from the emission of the InGaN LED chips, while those at around 544

surface roughness of the PGP. As estimated from the AFM images, the bare PGP (B-PGP) showed an RMS roughness of approximately 4.4 nm due to surface irregularity, while the graphene-wrapped PGP (G-PGP) showed an RMS roughness of approximately 2.5 nm. Little difference in appearance was observed between the B-PGP and the G-PGP. The insets of Figure 1f and g show digital photographs of the PGP before and after graphene wrapping, respectively. Figure 1h describes the effect of graphene wrapping on the contact angle of the water on the PGP surface. It was observed that the contact angle of water on G-PGP is 9−12% greater than that of water on B-PGP. Although the contact angle did increase, the value was somewhat lower than expected (mostly more than 90° in the literature),28,29 which supports the latest findings that D

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Figure 4. Layout of the actual WLED configuration used in this study, (b) cellular modeling of analysis domain (the dimensions of graphene, PGP, and air were calibrated to 10−9, 10−3, and 10−3 m/unit-cell, respectively, by adjusting the specific mass), and (c) heat transfer between adjacent cells in the domain. (d, e) Simulated heat transfer behavior in PGP before (d) and after (e) graphene-wrapping at four different elapsed times. Histograms showing the calculated temperature at different sites of the analysis domain, in which the temperature was normalized by dividing it by the maximum temperature: (f−i) air at the PGP bottom edge (f), the PGP bottom center (g), the PGP bottom edge (h), and the PGP top edge (i).

intensity distribution of the component peaks was observed in the B-PGP system as compared to the G-PGP based system at a higher temperature. This resulted in a shift in the color coordinate of the B-PGP based configuration to a much wider extent than that of G-PGP, as shown in Figure 2e. Due to the larger Stokes shift, the decline in emission intensity of the constituent red phosphor is likely to be faster than that of its green counterpart. Thus, when the temperature increases, the red component of the WLED emission spectra decreases and the color coordinate shifts from those that existed at the lower temperature. Figure 3a shows the arrangement of G-PGP on top of the six LED chips arranged circumferentially inside the jig. A PGP with a diameter of 25 mm was mounted at the top of the jig (which has an internal diameter of 23 mm) by placing it inside a groove machined at a depth of 2 mm so that the periphery of the PGP maintains contact with the ringlike surface of the metallic jig. The separation between the LED chips and the PGP was around 12 mm. Though the design provides a relatively low luminous efficiency, it is ideal for investigating the parameters under consideration and to make the mathematical model more reliable. Measures like the introduction of a diffuse reflector cup, and thereby reducing trapped optical modes, can further enhance the efficiency. Figure 3b shows the difference in the emission intensity of the configuration, normalized with respect to the value at 50 mA, employing B-PGP and G-PGP with standard deviation error bars. This figure clearly demonstrates that, as the forward bias current increases, the difference

and 612 nm originate from the constituent phosphors, (SrCa)Ga2S4:Eu2+ (SCGS:Eu2+) and CaAlSiN3:Eu2+ (CASN:Eu2+), respectively. This corresponds to the transition from the 4f65d excited state to the 4f 7 ground state of a Eu2+ ion. Figure 2b shows the corresponding EL spectra of the configuration using G-PGP. For the configuration using BPGP, the relative intensity of the component peaks reveals a significant variation as the operating current was increased. This trend continued until the highest operating current was reached. On the other hand, the configuration using G-PGP attained a more or less stable value after the initial start. In addition, the long-term EL stability of the G-PGP was excellent (see Figure S3 in the Supporting Information). This is due to the difference in the thermal quenching between the constituent phosphors, SCGS:Eu2+ and CaASN:Eu2+, which is a fundamental drawback of phosphor-converted WLEDs. The outcome of this difference is demonstrated in Figure 2c and the corresponding values are summarized in Table 1. Although the decline in luminous efficacy was observed in both configurations, the configuration using G-PGP attained saturation and remained stable for further increases in temperature. The variation in the transmittance of the PGP on the coating with graphene is shown in Figure 2d. The transmittance in the region from 400 to 620 nm, where three of the emission bands were located, was not significantly altered by the graphene wrapping. This indicates that the single-layer graphene layer had no qualitative influence on the luminescence of the PGP at lower temperature. A noticeable difference in the relative E

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

and that the heat spreading over the PGP surface then dissipates effectively into the neighboring air via convection (Figure 4e). On the other hand, in the absence of the graphene layer, the heat propagation occurs directly in the PGP and thus the temperature of the PGP increases gradually. The heat transfer coefficient of PGP is also approximately 3 orders of magnitude less than that of graphene. Eventually, a temperature gradient along the optical axis of the plate, with the center of the bottom surface having the highest temperature, will be developed (Figure 4d). This temperature gradient will then conduct heat from the bottom surface to the top surface. Thus, the B-PGP has a nonuniform temperature distribution and its average temperature also reaches a higher value than that of the G-PGP. Specifically, Figure 4f−i exhibits the calculated temperature of four representative sites in the analysis domain. The temperature of the site “1” which is a surrounding air domain near the lower surface of the PGP is calculated to be 1.1−2.3 times higher for G-PGP than for B-PGP (Figure 4f). The higher air temperature in the vicinity of B-PGP supports the idea that graphene facilitates heat dissipation through conduction followed by convection. At sites 2, 3, and 4, inside the PGP, the temperatures were always higher for B-PGP than for G-PGP; the ratio of the B-PGP temperature to the G-PGP temperature maintained a large value of 7.7 in the case of site 2 even at stage 4 (Figure 4g−i). These predictions closely match the experimental findings obtained from thermal imaging using the IR camera and conclusively demonstrate that the wrapped graphene contributes to the enhancement of the PGP EL characteristics.

between B-PGP and G-PGP becomes significant, and the emission intensity of the configurations normalized for the value at 50 mA shows about 20% higher yield for G-PGP than for the uncoated counterpart, B-PGP. The graphene wrapping of the PGP provides a thermal conducting layer, which disperses the generated thermal energy, owing to its higher thermal conductivity. This was evaluated using thermal imaging with an infrared (IR) camera that demonstrated the distribution of heat on the surface of the PGP as shown in Figure 3c. A dramatic rise in surface temperature was observed during the operation of the WLED. The surface temperature of the PGP increased to 100 °C after only 10 min and then up to a maximum of 160 °C after 30 min. The thermal image of the BPGP employed in the reference configurations shows the concentric distribution of the temperature zones. On the other hand, interestingly, the configurations in which the G-PGP was employed show a wider radial temperature gradient zone. Moreover, the increase in the area over which heat has spread makes the convection process more effective. The temperatures of the top and bottom of the PGP were recorded more precisely by using a thermocouple arranged on either surface of the PGP. Figure 3d plots the temperatures as a function of time. The G-PGP showed 3.6−7.8% higher temperatures at both surfaces compared with the B-PGP, and the temperature of the G-PGP bottom was 66.7−71.3% higher than that of the G-PGP top. First, the higher temperature at the bottom indicates that the convection process occurs more effectively at the bottom surface, in which the graphene layer of the G-PGP as an efficient heat spreader could contribute to the enhanced heat convection. The elevated temperature on the top of the GPGP additionally suggests that the heat spreading by the graphene layer extends over the entire surface of the G-PGP, and is not only confined to the bottom. Importantly, a major difference between the two color converting plates is whether graphene is wrapped or not, indicating that graphene played a critical role in maintaining the luminescent characteristics of the PGPs under high power operations. Cellular Automata Simulation. In order to better understand the mechanism of the heat transfer from the center of the PGP to the periphery, a simulation model was created based on two-dimensional cellular automata. The analysis model consists of three domains including air, the graphene layer, and the PGP. It was assumed that each domain consisted of rectangular unit cells and that heat is provided by a laser beam, as shown in Figure 4a and b. Each cell has physical properties including specific mass, specific heat, thermal conductivity, and heat transfer coefficient. The thickness of the graphene layer was taken as one unit cell size, for which the physical parameters were calibrated to reflect the real dimensions of constituent materials. Heat flows between the adjacent cells, from a cell with a higher temperature to one with a lower temperature, as depicted in Figure 4c. The twodimensional cellular automata simulation made it possible to directly visualize the heat propagation in and out of the B-PGP and G-PGP. Figure 4d and e shows the changes in temperature distribution throughout the analysis domain. A remarkably large difference in temperature distribution was observed before and after graphene wrapping. The thermal conductivity of graphene is several orders of magnitude greater than that of the PGP or that of air. Because of the low thermal conductivities of air and the PGP, the heat transfer into the air and the PGP would be negligible. Thus, it is reasonable to assume that the heat propagation rapidly occurs along the graphene via conduction

CONCLUSIONS The simulated temperature profiles in G-PGP are in good agreement with the experimental results observed using thermal imaging. Thus, it is believed that the graphene layer acts as an excellent heat spreader along the surface of the PGP and in turn protects the core of the plate through heat dissipation. This not only maintains the efficiency of the system, but also keeps intact the emission color properties of the configuration at high power operating conditions by preventing the differential thermal quenching of the constituent phosphors in the glass plate. In addition, the graphene layer modification might protect the PGP from humidity and volatile organic contaminants, thereby enabling the utilization of environmentally sensitive phosphors. It is also expected that the cellular automata algorithm developed in this work can be used in the future for the investigation of heat transfer in graphene multilayers and graphene-heat sink structures. METHODS Characterization of Single-Layer Graphene. Raman spectra were collected using a JASCO NRS 5100 spectrometer with a 532.13 nm argon ion laser as an excitation source while the graphene was placed on a Si/SiO2 (200 nm) wafer substrate. Ultraviolet (UV)− visible spectroscopy was conducted using a Mecasys Optizen 2120UV spectrophotometer after the CVD graphene had been transferred to a glass slide to measure the transmittance. HRTEM was performed using a JEOL JEM-3010 microscope. AFM was carried out using a Park Systems XE-100 microscope in noncontact tapping mode with the graphene deposited on a silicon wafer. Contact angles were obtained using a SurfaceTech GSA-X goniometer, in which the contact angles were automatically measured using the embedded software. Measurement of EL and Heat Flow. The PGPs were mounted on a blue LED module with six 450 nm LED chips placed circumferentially on the base of the jig. The walls of the cylindrical F

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano enclosure were parallel to the optical axis of the light source. The PGP was mounted on the edge of the holder, with a separation from the LED chips of about 5 mm, making it a forward emitting configuration. The luminescence measurement was then performed with an integrating sphere in the 2π configuration, for different values of forward bias current. The heat transfer mechanisms of the coated and uncoated PGPs were studied using thermographic images obtained with a FLIR T425 infrared camera. Two-Dimensional Cellular Automata Simulation. The overall calculation algorithm is described as follows. By repeating this calculation on all the cells, the temperature of the individual cells with time can be estimated.

that of PGP (a few millimeters). It is difficult to set the dimensions of unit cells to 1 nm because of the relatively large sizes of the PGP and surrounding air. Therefore, the specific masses of the individual components were adjusted, as summarized in Table S2 (see the Supporting Information), to compensate for the difference between their actual dimensions and their assumed dimensions, The specific masses of air and PGP were increased by 6 orders of magnitude relative to that of graphene. Consequently, in terms of the specific masses, the unit cell of the PGP was calibrated to be comparable to a micrometer scale, while that of graphene, a nanometer scale. Lastly, the thickness of PGP was set to be more than 10 unit cells. The simulation was developed using Embarcadero C++ Builder 2010 and was run on the Microsoft Windows 7 operating system; key parameters such as domain sizes, heat flux, and physical properties were controllable.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.5b06734. Experimental details, thermal conductivity measurement, simulation details, qualitative analysis of the PGP, EL spectra of G-PGP vs B-PGP, and so forth (PDF)

AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions ⊥

E.K., H.W.S., S.U., and Y.H.K. contributed equally.

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS This research was supported by the Basic Science Research Programs (NRF-2014R1A1A1002909, NRF2015R1A2A2A01007166) through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Science and Technology. This work was also supported by the Strategic Key-Material Development and the Materials and Components Research and Development bodies, funded by the Ministry of Knowledge Economy (MKE, Korea). The authors also appreciate the financial support from the Joint Research Project.

The representative physical properties of individual components are presented in Table S1 (see the Supporting Information). Most of the values were taken from the literature. The specific mass of graphene was assumed to be equal to that of graphite. The specific heat of graphene is known to be slightly higher than that of graphite and diamond around room temperature. It has been reported that suspended graphene has a thermal conductivity of up to 5000 W m−1 K−1. However, the thermal conductivity of the supported graphene is lower than that of the suspended graphene. The thermal conductivity of the supported graphene used in this work was measured via a Raman spectroscopy based optical method, and calculated to be either 978 or 1835 W m−1 K−1 depending on whether the temperature coefficient of graphene is taken as −1.62 × 10−2 or as −3.00 × 10−2 cm−1/K (see the Supporting Information). An intermediate value of 1000 W m−1 K−1 was chosen for the simulation. Additionally, it has been reported that the heat transfer coefficient between the graphene and substrates (SiO2 and SiC) ranges from 104 to 108 W m−2 K−1.34−36 The phosphor, which is encapsulated in glass matrices, assumes a value within the same range. The thermal properties of graphene are influenced by the atomic defects, grain boundaries, folding, and impurities that can be formed during the CVD and transfer processes. The surface roughness of the PGP also induces partially incomplete contact between the PGP and graphene, which would increase the thermal resistance between them. Considering the above, the thermal property values of graphene were adopted at a mild level. In fact, the calculated temperature of individual cells would be inaccurate since the physical properties vary gradually with temperature and pressure. Thus, the temperature of each cell was normalized per cycle to have a value of either 0 or 1, where 0 and 1 represent the lowest temperature and the highest temperature of the cells, respectively. As a result, the relative temperature distribution over the domains can be quickly calculated. Another concern is that the thickness of graphene (∼1 nm) is 6 orders of magnitude smaller than

REFERENCES (1) Suehiro, T.; Xie, R.-J.; Hirosaki, N. Facile Synthesis of (Sr,Ca)2Si5N8:Eu2+-Based Red-Emitting Phosphor for Solid-State Lighting. Ind. Eng. Chem. Res. 2013, 52, 7453−7456. (2) Bilir, G.; Ozen, G.; Collins, J.; Di Bartolo, B. Sensors and Displays: Principles, Materials, and Processing. ECS Trans. 2013, 50, 1−7. (3) Im, W. B.; George, N.; Kurzman, J.; Brinkley, S.; Mikhailovsky, A.; Hu, J.; Chmelka, B. F.; DenBaars, S. P.; Seshadri, R. Efficient and Color-Tunable Oxyfluoride Solid Solution Phosphors for Solid-State White Lighting. Adv. Mater. 2011, 23, 2300−2305. (4) Setlur, A. A.; Lyons, R. J.; Murphy, J. E.; Prasanth Kumar, N.; Satya Kishore, M. Blue Light-Emitting Diode Phosphors Based upon Oxide, Oxyhalide, and Halide Hosts. ECS J. Solid State Sci. Technol. 2013, 2, R3059−R3070. (5) Luo, H.; Kim, J. K.; Schubert, E. F.; Cho, J.; Sone, C.; Park, Y. Analysis of High-Power Packages for Phosphor-Based White-LightEmitting Diodes. Appl. Phys. Lett. 2005, 86, 243505. (6) Kim, J. K.; Luo, H.; Schubert, E. F.; Cho, J.; Sone, C.; Park, Y. Strongly Enhanced Phosphor Efficiency in GaInN White LightG

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano Emitting Diodes Using Remote Phosphor Configuration and Diffuse Reflector Cup. Jpn. J. Appl. Phys. 2005, 44, L649. (7) Chen, D.; Xiang, W.; Liang, X.; Zhong, J.; Yu, H.; Ding, M.; Lu, H.; Ji, Z. Advances in Transparent Glass-Ceramic Phosphors for White Light-Emitting Diodes−A Review. J. Eur. Ceram. Soc. 2015, 35, 859− 869. (8) Lee, J. S.; Unithrattil, S.; Kim, S.; Lee, I. J.; Lee, H.; Im, W. B. Robust Moisture and Thermally Stable Phosphor Glass Plate for Highly Unstable Sulfide Phosphors in High-Power White LightEmitting Diodes. Opt. Lett. 2013, 38, 3298−3300. (9) Lin, C.; Liu, R.-S. Thermal Effects in (Oxy)nitride Phosphors. J. Sol. State Light 2014, 1, 1−16. (10) Im, W. B.; Fellows, N. N.; DenBaars, S. P.; Seshadri, R.; Kim, Y. I. LaSr2AlO5, a Versatile Host Compound for Ce3+-Based Yellow Phosphors: Structural Tuning of Optical Properties and Use in SolidState White Lighting. Chem. Mater. 2009, 21, 2957−2966. (11) Chang, H.; Wu, H. Graphene-Based Nanomaterials: Synthesis, Properties, and Optical and Optoelectronic Applications. Adv. Funct. Mater. 2013, 23, 1984−1997. (12) Lee, C. − H.; Kim, Y. − J.; Hong, Y. J.; Jeon, S. − R.; Bae, S.; Hong, B. H.; Yi, G. − C. Flexible Inorganic Nanostructure LightEmitting Diodes Fabricated on Graphene Films. Adv. Mater. 2011, 23, 4614−4619. (13) Seo, J. − T.; Han, J.; Lim, T.; Lee, K.-H.; Hwang, J.; Yang, H.; Ju, S. Fully Transparent Quantum Dot Light-Emitting Diode Integrated with Graphene Anode and Cathode. ACS Nano 2014, 8, 12476−12482. (14) Kim, J. H.; Ko, E.; Hwang, J.; Pham, X. − H.; Lee, J. H.; Lee, S. H.; Tran, V. − K.; Kim, J. − H.; Park, J. − G.; Choo, J. B.; Han, K. N.; Seong, G. H. Large-Scale Plasma Patterning of Transparent Graphene Electrode on Flexible Substrates. Langmuir 2015, 31, 2914−2921. (15) Lee, J. M.; Choung, J. W.; Yi, J.; Lee, D. H.; Samal, M.; Yi, D. K.; Lee, C. − H.; Yi, G.-C.; Paik, U.; Rogers, J. A.; Park, W. I. Vertical Pillar-Superlattice Array and Graphene Hybrid Light Emitting Diodes. Nano Lett. 2010, 10, 2783−2788. (16) Yan, Z.; Liu, G.; Khan, J. M.; Balandin, A. A. Graphene Quilts for Thermal Management of High-Power GaN Transistors. Nat. Commun. 2012, 3, 827. (17) Han, N.; Cuong, T. V.; Han, M.; Ryu, B. D.; Chandramohan, S.; Park, J. B.; Kang, J. H.; Park, Y.-J.; Ko, K. B.; Kim, H. Y.; Kim, H. K.; Ryu, J. H.; Katharria, Y. S.; Choi, C.-J.; Hong, C.-H. Improved Heat Dissipation in Gallium Nitride Light-Emitting Diodes with Embedded Graphene Oxide Pattern. Nat. Commun. 2013, 4, 1452. (18) Shim, J. − P.; Kim, D.; Choe, M.; Lee, T.; Park, S. − J.; Lee, D. − S. A Self-Assembled Ag Nanoparticle Agglomeration Process on Graphene for Enhanced Light Output in GaN-Based LEDs. Nanotechnology 2012, 23, 255201. (19) Xu, X.; Pereira, L. F. C.; Wang, Y.; Wu, J.; Zhang, K.; Zhao, X.; Bae, S.; Tinh Bui, C.; Xie, R.; Thong, J. T. L.; Hong, B. H.; Loh, K. P.; Donadio, D.; Li, B.; Ö zyilmaz, B. Length-Dependent Thermal Conductivity in Suspended Single-Layer Graphene. Nat. Commun. 2014, 5, 3689. (20) Bai, G.; Tsang, M. − K.; Hao, J. Tuning the Luminescence of Phosphors: Beyond Conventional Chemical Method. Adv. Opt. Mater. 2015, 3, 431−462. (21) Balandin, A. A.; Ghosh, S.; Bao, W.; Calizo, I.; Teweldebrhan, D.; Miao, F.; Lau, C. N. Superior Thermal Conductivity of SingleLayer Graphene. Nano Lett. 2008, 8, 902−907. (22) Calizo, I.; Balandin, A. A.; Bao, W.; Miao, F.; Lau, C. N. Temperature Dependence of the Raman Spectra of Graphene and Graphene Multilayers. Nano Lett. 2007, 7, 2645−2649. (23) Nguyen, K. T.; Abdula, D.; Tsai, C. − L.; Shim, M. Temperature and Gate Voltage Dependent Raman Spectra of Single-Layer Graphene. ACS Nano 2011, 5, 5273−5279. (24) Seol, J. H.; Jo, I.; Moore, A. L.; Lindsay, L.; Aitken, Z. H.; Pettes, M. T.; Li, X.; Yao, Z.; Huang, R.; Broido, D.; Mingo, N.; Ruoff, R. S.; Shi, L. Two-Dimensional Phonon Transport in Supported Graphene. Science 2010, 328, 213−216.

(25) Chen, J.; Zhang, G.; Li, B. Substrate Coupling Suppresses Size Dependence of Thermal Conductivity in Supported Graphene. Nanoscale 2013, 5, 532−536. (26) Cai, W.; Moore, A. L.; Zhu, Y.; Li, X.; Chen, S.; Shi, L.; Ruoff, R. S. Thermal Transport in Suspended and Supported Monolayer Graphene Grown by Chemical Vapor Deposition. Nano Lett. 2010, 10, 1645−1651. (27) Chen, Z.; Qin, X.; Zhang, Q.; Li, Y.; Wang, H. Enhanced Fluorescence and Heat Dissipation of Calcium Titanate Red Phosphor Based on Silver Coating. J. Colloid Interface Sci. 2015, 459, 44−52. (28) Shin, Y. J.; Wang, Y.; Huang, H.; Kalon, G.; Wee, A. T. S.; Shen, Z.; Bhatia, C. S.; Yang, H. Surface-Energy Engineering of Graphene. Langmuir 2010, 26, 3798−3802. (29) Kim, K.-S.; Lee, H.-J.; Lee, C.; Lee, S.-K.; Jang, H.; Ahn, J.-H.; Kim, J.-H.; Lee, H.-J. Chemical Vapor Deposition-Grown Graphene: The Thinnest Solid Lubricant. ACS Nano 2011, 5, 5107−5114. (30) Xu, K.; Heath, J. R. Wetting: Contact with What? Nat. Mater. 2013, 12, 872−873. (31) Shih, C. J.; Strano, M. S.; Blankschtein, D. Wetting Translucency of Graphene. Nat. Mater. 2013, 12, 866−869. (32) Rafiee, J.; Mi, X.; Gullapalli, H.; Thomas, A. V.; Yavari, F.; Shi, Y.; Ajayan, P. M.; Koratkar, N. A. Wetting Transparency of Graphene. Nat. Mater. 2012, 11, 217−222. (33) Li, Z.; Wang, Y.; Kozbial, A.; Shenoy, G.; Zhou, F.; McGinley, R.; Ireland, P.; Morganstein, B.; Kunkel, A.; Surwade, S. P.; Li, L.; Liu, H. Effect of Airborne Contaminants on the Wettability of Supported Graphene and Graphite. Nat. Mater. 2013, 12, 925−931. (34) Freitag, M.; Steiner, M.; Martin, Y.; Perebeinos, V.; Chen, Z.; Tsang, J. C.; Avouris, P. Energy Dissipation in Graphene Field-Effect Transistors. Nano Lett. 2009, 9, 1883−1888. (35) Chen, Z.; Jang, W.; Bao, W.; Lau, C.; Dames, C. Thermal Contact Resistance between Graphene and Silicon Dioxide. Appl. Phys. Lett. 2009, 95, 161910. (36) Persson, B. N. J.; Ueba, H. Heat Transfer between Weakly Coupled Systems: Graphene on a-SiO2. Europhys. Lett. 2010, 91, 56001.

H

DOI: 10.1021/acsnano.5b06734 ACS Nano XXXX, XXX, XXX−XXX