Cup-shaped copper heat spreader in multi-chip high-power LEDs application

In this study, cup-shaped copper sheets were developed to improve heat dispassion for high-power light emitting diodes (LEDs) array module (3 × 3, 4 × 4, and 5 × 5) using an electroplating technique. The cupshaped copper sheets were directly contacted with sapphire to enhance the heat dissipation of the chip itself. The lateral emitting light extraction and heat dissipation of high-power LEDs were enhanced and efficient. The surface temperature was not only decreasing but also uniform for each LED chip with the cup-shaped copper heat spreader adoption. The high thermal transmitting performance of cup-shaped copper heat spreader allows thermal resistance reducing 0.7, 0.6, and 0.7 K/W of 3 × 3, 4 × 4, and 5 × 5 LED array module, respectively. In addition, the light output power was increased of 14, 13, and 12% with 3 × 3, 4 × 4, and 5 × 5 LEDs array module using cup-shaped copper sheet at high current injection. High heat dissipation performance and light extraction were obtained by cup-shaped copper sheet with copper bulk and silver mirror. ©2012 Optical Society of America OCIS codes: (250.0250) Optoelectronics; (230.3670) Light-emitting diodes. References and links 1. K. C. Chen, R. W. Chuang, Y. K. Su, C. L. Lin, H. C. Hsu, J. Q. Huang, and K. F. Yang, “High thermal dissipation of ultra high power light-emitting diodes by copper electroplating,” in Proceedings of 57th IEEE Electronic Components and Technology Conference (IEEE, 2007), pp. 734–736. 2. L. Yuan, S. Liu, M. Chen, and X. Luo, “Thermal analysis of high power LED array packaging with microchannel cooler,” in Proceedings of 7th Electronics Packaging Technology Conference (IEEE, 2006), pp. 1–5. 3. W. Y. Lin, D. S. Wuu, K. F. Pan, S. H. Huang, C. E. Lee, W. K. Wang, S. C. Hsu, Y. Y. Su, S. Y. Huang, and R. H. Horng, “High-power GaN-mirror-Cu light-emitting diodes for vertical current injection using laser liftoff and electroplating techniques,” IEEE Photon. Technol. Lett. 17(9), 1809–1811 (2005). 4. T. Tamura, T. Setomoto, and T. Taguchi, “Illumination characteristics of lighting array using 10 candela-class white LEDs under AC 100 V operation,” J. Lumin. 87–89, 1180–1182 (2000). 5. S. F. Wong, P. Malatkar, C. Rick, V. Kulkarni, and I. Chin, “Vibration testing and analysis of ball grid array package solder joints,” in Proceedings of 57th IEEE Electronic Components and Technology Conference (IEEE, 2007), pp. 373–380. 6. R. H. Horng, C. C. Chiang, H. Y. Hsiao, X. Zheng, D. S. Wuu, and H. I. Lin, “Improved thermal management of GaN/sapphire light-emitting diodes embedded in reflective heat spreaders,” Appl. Phys. Lett. 93(11), 111907 (2008). 7. R. H. Horng, H. Y. Hsiao, C. C. Chiang, D. S. Wuu, Y. L. Tsai, and H. I. Lin, “Novel device design for highpower InGaN/sapphire LEDs using copper heat spreader with reflector,” IEEE J. Sel. Top. Quantum Electron. 15(4), 1281–1286 (2009). 8. R. H. Horng, W. K. Wang, S. Y. Huang, and D. S. Wuu, “Effect of resonant cavity in wafer-bonded green InGaN LED with dielectric and silver mirrors,” IEEE Photon. Technol. Lett. 18(3), 457–459 (2006). #165584 $15.00 USD Received 28 Mar 2012; revised 29 Jun 2012; accepted 30 Jun 2012; published 10 Jul 2012 (C) 2012 OSA 10 September 2012 / Vol. 20, No. S5 / OPTICS EXPRESS A597 9. L. Kim, J. H. Choi, S. H. Jang, and M. W. Shin, “Thermal analysis of LED array system with heat pipe,” Thermochim. Acta 455(1–2), 21–25 (2006). 10. L. Kim and M. W. Shin, “Thermal resistance measurement of LED package with multichips,” IEEE Trans. Compon. Packag. Tech. 30(4), 632–636 (2007). 11. N. Narendran and Y. Gu, “Life of LED-based white light sources,” IEEE Photon. Technol. Lett. 1(1), 167–171 (2005). 12. G. Faekas, Q. V. V. Vader, A. Poppe, and G. Bognar, “Thermal investigation of high power optical device by transient testing,” in Proceedings of 9th International Workshop on Thermal Investigations of ICs and Systems (IEEE, 2003), pp. 213–218. 13. R. H. Horng, J. S. Hong, Y. L. Tsai, D. S. Wuu, C. M. Chen, and C. J. Chen, “Optimized thermal management from a chip to a heat sink for high-power GaN-based light-emitting diodes,” IEEE Trans. Electron. Dev. 57(9), 2203–2207 (2010). 14. S. Nakamura, M. Senoh, N. Iwasa, and S. Nagahama, “High‐power InGaN single‐quantum‐well‐structure blue and violet light‐emitting diodes,” Appl. Phys. Lett. 67(13), 1868 (1995). 15. B. Zhang, T. Egawa, H. Ishikawa, Y. Liu, and T. Jimbo, “Thermal stability of InGaN multiple-quantum-well light-emitting diodes on an AlN/sapphire template,” J. Appl. Phys. 95(6), 3170 (2004).


Introduction
Recently, the applications of high-power LEDs (HP LEDs) have attracted more attentions due to the efficiency increasing of GaN light emitting diodes (LEDs) [1][2][3].Nevertheless, wallplug efficiency for HP GaN-LEDs with sapphire substrate is about 40% at 350 mA injection, meaning that there is 60% power energy transferred into heat.Moreover, it is necessary to package the multi-chip HP LEDs into the light engine with die bonding.Traditionally, the multi-chips LED package is always used in pick and place technology and put one die on the metal-core printed circuit board (MCPCB) per one time [4,5].There are many disadvantages for this kind of package.For example, the chip contacting heat sink (MCPCB) is by die paste (it is silver paste) and the contact area is only the chip area.This will limit thermal dissipation due to the low thermal dissipation of sapphire substrate and silver paste.If the thermal paste cannot offer thermal conductivity, the heat will be accumulated in the HP LEDs during the operation, which results in the thermal running and reducing the output power and life of multi-chip HP LEDs.Moreover, the distance between the chips cannot be near each other.One reason is the limitation of the pick and place machine.The other reason is the light emitting from the sides of the LED is easily absorbed by the surrounding chips.Theoretically, the light can be reflected in a normal direction if the distance between the chips is distant.Nevertheless, the light emitted from the sides of the LED remains low due to the low reflectivity of MCPCB.In our previous study, the sapphire-based GaN LED with a cupshaped copper (Cu) heat spreader was useful to enhance thermal extraction from sapphirebased LEDs.This occurs to avoid directly attaching the bare chip to MCPCB as well as the cup-shaped Cu enhances the light refection to the top surface.In this paper, the embedded reflective heat spreaders technology for the multi-chips HP LEDs with 3 × 3, 4 × 4, and 5 × 5 arrays will be developed.The output power, surface temperature, and thermal resistance for multi-chip LED arrays with and without embedded reflective heat spreaders are discussed in this paper.

Experimental
The InGaN LEDs (λp = 455 nm) were grown by metal organic chemical vapor deposition on (0001) sapphire substrates.The structure consisted of a 30 nm-thick low-temperature GaN buffer layer, a 1.5 µm thick undoped GaN layer, a 3 µm-thick Si-doped GaN layer, InGaN/GaN multiple quantum well (MQW) stack, and a 0.3 µm-thick Mg-doped GaN layer.
For the device process, one LED chip was defined with the size of 45 mil × 45 mil. Figure 1 shows the schematic diagram of the LED array module package fabrication processes.The LED array modules fabrication with copper electroplating technique and selfaligned lithography were adopted.We used a schematic fabrication process as follows.First (Fig. 1(a)), photoresist (PR) is coated on glass substrate, and the chips are placed on the glass, then soft baked at 80°C for 45 min.Second, shown in Fig. 1(b), the Ag/Cr/Au mirror films are deposited onto the PR area and backside of the sapphire substrate using thermal evaporation.Third, shown in Fig. 1(c), the copper cup-shaped is fabricated using electroplating.Finally, shown in Fig. 1(d), removal of the acrylic PR was performed by immersion in acetone and a cup-shaped copper heat spreader covered with the Ag reflector was obtained successfully [6][7][8].The 3 × 3, 4 × 4, and 5 × 5 LEDs array modules are fabricated with the heat, electrical, and optical discussed.Figure 2 shows the products of LED arrays with cup-shaped copper heat spreader mounted on MCPCBs by AgSnCu solder with and without current injection.
Current-voltage (I-V) characteristics for LEDs were measured at room temperature using an Agilent 4155B semiconductor parameter analyzer.The light output power of an LED was measured by integrating sphere detectors (CAS 140B, Instrument Systems).Thermal infrared image was recorded after the device in thermal equilibrium to evaluate the thermal properties of LEDs.A T3Ster Master system was used to measure the total thermal resistance of an LED based on thermal transient analysis.

Results and discussion
Thermal resistance analysis is a straightforward method to evaluate the thermal performance of LEDs.Cumulative structure functions are graphic representations of one-dimensional equivalent thermal resistance-capacity networks of the measured system.Thus, capacity and thermal resistance values that belong to the heat conduction path from the LED junction-todie attach to the heat sink and are directly indicated [9,10].Figure 3 shows the measurement results of thermal resistance of 3 × 3, 4 × 4, and 5 × 5 LEDs arrays under each chip being injected with 350 mA.From our measurements, the forward voltages (@ 350 mA) of one LED chip with original structure and cup-shaped copper heat spreader were 3.72 and 3.7 V, respectively.For the 3 × 3, 4 × 4 and 5 × 5 LED arrays with original structure, the forward voltages (@ 350 mA) were 11.3, 14.8, and 18.5 V, respectively.Further measurements for LED arrays with cup-shaped copper heat spreader, the forward voltages (@ 350 mA) were measured to be 11.1, 14.6 and 18 V for 3 × 3, 4 × 4 and 5 × 5 LED arrays, respectively.The thermal resistances of 3 × 3 LED arrays with and without cup-shaped copper were 1.8 and 2.5 K/W, respectively, (Fig. 3(a)).Thermal resistance was decreased 0.7 K/W for the 3 × 3 LED arrays with cup-shaped copper heat spreader.Our results have indicated that the copper spreader is useful in thermal resistance reduction of array LEDs.Moreover, thermal resistances of 4 × 4 and 5 × 5 LEDs array with cup-shaped copper heat spreader were 1.4 and 0.6 K/W, respectively, (Figs.3(b) and 3(c)).There were also approximately 0.6 and 0.7 K/W improvements as compared with the 4 × 4 and 5 × 5 LEDs arrays without cup-shaped copper heat spreaders.Moreover, it is noteworthy that thermal resistance decreased as LED arrays increased.The more LEDs arrays present, the better the heat dissipation will be because of the large thermal spreader area.4 is the surface temperature distribution of 3 × 3 LEDs (@ 350 mA) with and without the copper heat spreader.The highest temperatures were 64.2 °C and 87.8 °C of LED chips with and without the cup-shaped copper heat spreader at same injection current, respectively.Obviously, the LEDs with cup-shaped copper heat spreaders provide better thermal dissipation.Moreover, the temperature distribution of LED arrays without cupshaped copper heat spreaders present non-uniform and the heat accumulates surrounding each chip (Fig. 4(a)).Normally, the thermal dissipation path of the LED arrays (resulting from junction temperature) is through the sapphire substrate, solder between LED chips, and MCPCB substrate for the LED arrays for the original package structure.The bottleneck for thermal dissipation is the sapphire substrate and the solder.Moreover, the thickness of solder cannot be easily controlled the same for each chip during the die bonding.Thus, heat will be concentrated in the sapphire substrate.The surface temperature distribution becomes nonuniform.Conversely, the surface temperature distribution of the LED arrays with cup-shaped copper heat spreaders is uniform (Fig. 4(b)).The heat generating from LEDs can be lateral thermal spreading to the copper and dissipated to MCPCB via solder because of the perfect thermal conduction and large heat capacity of copper.This contributes to the reliability improvement for LED array modules operating under high current injection [9,11].Figure 5 shows the output power and light efficiency of LED arrays with and without cupshaped copper heat spreaders as a function of total injection current.The output power of 3 × 3 LED array modules with cup-shaped copper heat spreaders (4.1 W @ 1.05 A) was higher than original package structure (3.6 W @ 1.05 A) as shown in Fig. 5(a).A power efficiency of 33.9% for 3 × 3 LED arrays at 1.05 A was obtained with the cup-shaped copper heat spreader.In general lighting applications, a large area and high current injection for multichips LED arrays is required.Thus, the 4 × 4 and 5 × 5 LED arrays were developed.Figures 5(b) and 5(c) show the output power and power efficiency.The power efficiency is decreased with the current injection increased.However, the power efficiency of 4 × 4 and 5 × 5 LED arrays remained at 35 and 34.6% at 1.4 and 1.75 A with current injection, respectively.The obvious power enhancement was attributed to the cup-shaped copper heat spreader with high thermal conductivities and light reflection.The light emitted from the edge of the LEDs is reflected via the Ag mirror.This results in the improvement of output power and power efficiency [7].
Determination for the thermal resistance (R th ) of LED chip can be calculated by using the equation of R th = ∆T/(P e -P op ) [12], where ∆T is the difference between junction temperature and room temperature defined as 25 °C, P e is the input power, and P op is the output power.We can apply the R th , P e and P op to obtain the ∆T values of LED chips.For the 3 × 3 LED arrays with original structure and cup-shaped copper heat spreader, the P e values (@ 350 mA) were about 11.9 and 11.7 W, respectively by using P e = 3 × 0.35 × V f .The P op values were measured to be as 3.6 and 4.1 W for the LED chips with original structure and cup-shaped copper heat spreader, while the R th were 2.5 and 1.8 K/W (as shown in Fig. 3(a)), respectively.The corresponding ∆T values were calculated as 20.8 and 13.7 °C, respectively.Furthermore, the junction temperatures of LED chips with original structure and cup-shaped copper heat spreader were determined to be 45.8 and 38.7 °C, respectively.It was found that the temperatures and temperature difference obtained by calculating were all lower than those by measuring, as shown in Fig. 4. As we known, the R th measurement is transient because of the fast heat dissipation (around microseconds) through the T3Ster cooling system.However, the surface temperature is measured by thermal equilibrium, resulting in the thermal accumulation.Therefore, it leads to the higher surface temperature measured by thermal infrared image than the junction temperature obtained by R th .Actually, the similar results in our previous research [13] were observed for single chip LED.Nevertheless, the R th and surface temperature has the same tendency.
As mentioned above, the total thermal resistance of LEDs with copper substrate is effectively reduced.This contributes to lowering the junction temperature of the LEDs.The output power of a LED strongly relates to carrier confinement in the quantum wells.As chip temperature increases, carrier confinement in quantum wells becomes less efficient, leading to deterioration in output power [14,15].In this study, the LED arrays with copper substrate effectively suppress as the junction temperature increases.

Conclusion
The direct-electroformed copper heat spreader is proposed to fabricate array LED modules.The improvement of thermal management and optical performance are reported.The heat generation of array LEDs was efficiently extracted with cup-shaped copper heat spreader.The additional optical property allows the realization of reflected LED lateral emission in an inexpensive and mass-production method with an Ag mirror for cup-shaped copper heat spreader.The fine behavior of thermal characteristics of surface temperature and thermal resistance is demonstrated with the cup-shaped copper heat spreader adoption.The cupshaped copper package structure is a useful method to fabricate large LED array modules in general lighting.

Figure
Figure4is the surface temperature distribution of 3 × 3 LEDs (@ 350 mA) with and without the copper heat spreader.The highest temperatures were 64.2 °C and 87.8 °C of LED chips with and without the cup-shaped copper heat spreader at same injection current, respectively.Obviously, the LEDs with cup-shaped copper heat spreaders provide better thermal dissipation.Moreover, the temperature distribution of LED arrays without cupshaped copper heat spreaders present non-uniform and the heat accumulates surrounding each chip (Fig.4(a)).Normally, the thermal dissipation path of the LED arrays (resulting from junction temperature) is through the sapphire substrate, solder between LED chips, and MCPCB substrate for the LED arrays for the original package structure.The bottleneck for thermal dissipation is the sapphire substrate and the solder.Moreover, the thickness of solder cannot be easily controlled the same for each chip during the die bonding.Thus, heat will be concentrated in the sapphire substrate.The surface temperature distribution becomes nonuniform.Conversely, the surface temperature distribution of the LED arrays with cup-shaped copper heat spreaders is uniform (Fig.4(b)).The heat generating from LEDs can be lateral thermal spreading to the copper and dissipated to MCPCB via solder because of the perfect thermal conduction and large heat capacity of copper.This contributes to the reliability improvement for LED array modules operating under high current injection[9,11].

Fig. 4 .
Fig. 4. Surface temperature distribution at injection current of 350mA with (a) array LEDs on Al MCPCB and (b) array LEDs with copper heat spreader on Al MCPCB.