In this study, the thin-film vertical-type AlGaInP LEDs on Cu substrates were fabricated. By performing the epitaxial lift-off (ELO) process, the LED device can be transferred from GaAs to Cu substrate. Then the GaAs substrate was separated and the ELO-LED was completed. To overcome the drawback of crack formation in the epilayer during the ELO process, various patterned Cu substrates were designed. Moreover, the finite element method was used to simulate the stress distribution in the LED sample during the ELO process. From the simulation results, an optimum structure of patterned Cu substrate was obtained since its maximum stress can be confined to the chip edges and the stress was decreased significantly during the ELO process, resulting in an apparent reduction of crack generation after separating the GaAs substrate. This optimum patterned Cu substrate was employed for the fabrication of ELO-LED. In addition, the chemical etching process was also used to etch the GaAs substrate, and this device transferred to Cu substrate was denoted as CE-LED. Based on the measurements of device performances, the forward voltages (@350 mA) of the CE-LED and ELO-LED were measured to be 2.20 and 2.29 V, while the output powers (@350 mA) of these two devices were 49.9 and 48.2 mW, respectively. Furthermore, the surface temperatures (@350 mA) of these two samples were 46.9-48.3 and 45.2-47.0 °C, respectively. Obviously, the device characteristics of the ELO-LED are very similar to those of the CE-LED. It confirms that the design of patterned Cu substrate is very helpful to obtain the thin-film vertical-type AlGaInP LEDs. Additionally, via the ELO process, the separated GaAs substrate can be reused for production cost down.
© 2015 Optical Society of America
At present, the AlGaInP-based epitaxial materials are widely used in high-brightness LEDs, and their excellent device performances have attracted great interest for many applications consisting of display, traffic signal, and general lighting [1, 2]. Generally, to achieve high brightness and high efficiency, the development of large-sized LED device with high current level is required. It is well known that the GaAs substrate is usually applied for the growth of AlGaInP epilayer due to the consideration of lattice match . However, because the absorption of visible light by the GaAs substrate, it would limit the light extraction from the active region of LED . In addition, the GaAs substrate possesses poor thermal conductivity, which leads to the deterioration in heat dissipation of device .
To avoid the disadvantage in heat dissipation and enhance the light extraction, the transfer of AlGaInP epitaxial structure from GaAs to highly thermal conductive substrates including Si, Cu, and Al is a promising method [5, 6]. There are two common techniques to transfer the AlGaInP epitaxial structure to other substrates. For the first method, the chemical etching of GaAs is used to remove the substrate, and then the epilayer is transferred to the heat-sink substrate . The second method is applied for the so-called epitaxial lift-off (ELO) technique . In the ELO process, the AlAs sacrificial layer, GaAs buffer layer, and the GaInP etching stop layer are needed to grow between the GaAs substrate and LED epitaxial structure. Next, the AlAs layer is laterally etched by HF solution for the separation of LED structure and GaAs substrate. Additionally, the GaAs substrate can be retained and reused after the ELO process .
However, to successfully separate the epitaxial structure and GaAs substrate, the soakage of LED sample into HF etchant for a long time is required. This would degrade the epilayer quality and device performance of AlGaInP LEDs. In our previous research about the ELO technique , several hydrophilic substances such as acetone (ACE), isopropanol (IPA), and methanol (MA) were added into HF etchant, and the separation rate for GaAs substrate and epitaxial structure can be enhanced remarkably because of the efficient venting of bubbles generated in the ELO process, especially for the use of HF:ACE etchant (i.e., ACE added into HF etchant). Although the drawback of long-term separation for the GaAs substrate is solved, the problem of stress created in the sample during the ELO process still exists. In fact, when the ELO process was performed, the AlInGaP epilayer was easily affected by the stress induced by the substrate transferring, which would lead to serious crack formation in the epilayer and severe deterioration of device performance.
In this study, the vertical-type AlGaInP LEDs were fabricated on Cu substrates with highly thermal conductivity. We used the ELO process with the HF:ACE etchant to separate the GaAs substrate and epitaxial structure. Then the AlGaInP LED can be successfully transferred to Cu substrate, and this device was denoted as ELO-LED. To solve the problem of crack generation in the epilayer during the ELO process, various patterned Cu substrates were designed (i.e., various through-hole arrays from Cu substrates to AlAs sacrificial layers were designed in the LEDs). Up to now, there are very few researches for the modification of pattern shape on Cu substrate. In addition, the finite element method (FEM)  was employed to simulate the stress distributions in these ELO-LEDs, and an optimum patterned Cu substrate can be obtained for the fabrication of ELO-LED. On the other hand, the chemical etching of GaAs substrate was also used to prepare the vertical-type AlGaInP LED on Cu substrate, and this device denoted as CE-LED was chosen as the contrasted sample. The device performances consisting of optoelectronic and thermal characteristics of the CE-LED and ELO-LED were analyzed and compared, as discussed later.
The epitaxial structures of AlGaInP LEDs were grown on the GaAs substrates by metalorganic chemical vapor deposition (MOCVD). Firstly, a 0.08-μm-thick GaAs buffer layer was prepared on the substrate. Then, for the ELO process, the AlAs sacrificial layer with a thickness of 0.5 μm was deposited on the GaAs buffer layer. Subsequently, a 0.2-μm-thick GaInP etching stop layer and the LED epilayers with total thickness of 14 μm were grown in sequence, as shown in Fig. 1(a). The epitaxial structures consist of a n+-GaAs ohmic contact layer, a n-GaInP layer, a n-AlGaInP cladding layer, the multiple quantum well (MQW), a p-AlGaInP cladding layer, and a p-GaP ohmic contact layer. To improve the optoelectronic performances of AlGaInP LED, the LED epitaxial structures were transferred to the Cu substrate. Therefore, after the epitaxial growth of AlGaInP LED, the AuBe/Au (40/80 nm) layer and the 80-μm-thick Cu metallic layer were deposited sequentially. The AuBe/Au is not only employed as the mirror layer, but also the seed layer for Cu electroplating. This layer was deposited by E-beam evaporation and annealed at 450 °C for 3 min. Then the electroplated Cu layer was used as a metallic substrate.
In this study, there are two methods applied for the substrate transferring process. For the fabrication of CE-LED, the GaAs substrate was chemically etched by using NH4OH:H2O2 (1:9) solution, as displayed in Fig. 1(b). As the GaAs substrate was absolutely etched, the etching stop layer was further etched by using H3PO4:HCl. For the device process, a mesa pattern of LED sample was defined with the size of 1 mm2. The Ni/AuGe/Ni/Au (2.5/40/5/300 nm) was prepared on the n+-GaAs contact layer by E-beam evaporation as the front grid electrode. Finally, the NH4OH:H2O2:H2O was used to remove the n+-GaAs contact layer, and the fabrication of vertical-type AlGaInP LED on Cu substrate was completed. For the preparation of ELO-LED, the ELO process was employed to separate the epitaxial structure and GaAs substrate, and the GaAs substrate can be reused. To form the etching through hole array, the mesa pattern for the LED sample should be fabricated before the Cu electroplating and ELO processes. Next, the AlAs sacrificial layer was etched using the HF:ACE etchant, resulting in the separation of the GaAs substrate and epitaxial structure, as shown in Fig. 1(c). The volume ratio of HF:ACE solution was kept at 1:1. It is worth noting that an extremely high lateral etching rate of the AlAs sacrificial layer can be achieved via the employment of the HF:ACE solution, as proposed in our previous research . After separating the GaAs substrate, the LED epitaxial structure can be successfully transferred to the Cu substrate. Then the front electrode process (Ni/AuGe/Ni/Au) was performed for the ELO-LED. Figure 1(d) exhibits the schematic diagram of the completed vertical-type AlGaInP LED on Cu substrate by using these two methods. Finally, these LED chips were encapsulated by using epoxy, which can improve the light extraction.
The spatial distribution image of the light emission in the LED was obtained by a CCD camera with an optical microscope. The current-voltage (I-V) characteristics of LEDs were measured by using a semiconductor parameter analyzer, and the output powers were analyzed with a calibrated integrating sphere. The optical property of the LED epitaxial structure was determined by the electroluminescence (EL) measurement. The LED device morphologies were observed by means of the optical microscopy (OM) and scanning electron microscopy (SEM). The surface temperature distribution of LED sample was measured by thermal infrared imaging at thermal equilibrium under a driven current of 350 mA.
3. Simulation method
Figure 2(a) shows the sample structure applied for the simulation by using FEM. In the sample structure, the AlAs sacrificial layer (3 μm)/AlGaInP epilayer (14 μm) was fabricated on a 400-μm-thick GaAs substrate. Then the electroplated Cu metal layer (80 μm) was employed as the permanent substrate for the AlGaInP epilayer. Additionally, the parameters of all materials for the simulation are displayed in Table 1 [11–13].
To realize the effect of stress from the Cu electroplating on the epitaxial layer, we observe the change of stress via the simulation. In this simulation, five samples fabricated on various patterned Cu substrates were designed, and these five patterns for Cu substrates were denoted as the structures A-E, as shown in Figs. 2(b)-2(f). The stress originated from the Cu substrate was varied by modifying the pattern shape. These structures were designed to simulate the stress variation of the samples when the ELO process was performed. For the setting of boundary conditions, the stress in the sidewall of GaAs substrate was set to zero. Meanwhile, the stress in the sidewall of Cu substrate was assumed to suffer the compressive stress of 28.3 Mpa, which was referred to the actual experimental result measured by stress measurement system (FLX2320-S, Toho Technology). In Figs. 2(b)-2(f), the compressive stress of 28.3 Mpa formed in the edges and corners of the structures A-E were marked with red color. Moreover, the other edges and corners (marked with green color) in these structures were connected to those of other Cu patterns in the wafer, and their stresses were set to zero. To simulate the stress variation in the sample during the ELO process, various etching amounts of 10%, 30%, 50%, 70%, and 90% in the AlAs sacrificial layer were assumed, as displayed in Fig. 2(g).
4. Result and discussion
Figure 3(a) shows the schematic diagram of AlGaInP LED on typical patterned Cu substrate with the structure B for the ELO process. After soaking the device into HF:ACE solution for 1.5-2 hours, the LED epitaxial structure was separated from the GaAs substrate and transferred to the Cu substrate. Figure 3(b) displays the light emission image of this vertical-type AlGaInP LED on Cu substrate with the structure B at a driven current of 20 mA. After the separation process, the high leakage current and non-uniform light emission often occurred in the LED chip because of the crack formation in the epilayer. Actually, the leakage current (@-5 V) of the ELO-LED fabricated on typical patterned Cu substrate with the structure B was as high as 0.2-0.8 mA (not shown here). As shown in Fig. 3(c), the L-shaped cracks formed on the epilayer surface of LED chip can be observed by OM, and these cracks were extended to the center of chip. Further observation of the cracks on the chip was performed by SEM, as exhibited in Fig. 3(d). It shows the surface image of AlGaInP LED on Cu substrate with the structure B after the ELO process. The cracks and voids were found in the upper right corner of LED chip, as marked by the red circle in Fig. 3(d). Then the cracks and voids in this region with higher magnification were clearly observed via the focused-ion-beam technique, as shown in Fig. 3(e). In the magnified image, it can be seen that the cracks were extended along the specific directions (marked by the green arrows). It can be speculated that the cracks were generated from the stress between the epilayer and Cu substrate. In the ELO process, the etching solution would permeate to the epilayer through the cracks and then the voids were formed. In our previous study , we have confirmed that the etching rate of the AlAs sacrificial layer was enhanced efficiently when the hydrophilic substance was added into HF etching solution (especially for the HF:ACE solution), and the void formation can be expected to avoid. However, the decrease of the stress between the epilayer and Cu substrate is really essential to reduce the crack formation. Therefore, we would determine the optimum pattern structure of Cu substrate to overcome the stress issue.
Figure 4(a) exhibits the FEM simulation result of stress distribution for the GaAs substrate/AlAs/AlGaInP/Cu substrate (with the structure A) after the ELO process, and the etching amount of the AlAs sacrificial layer was set at 10%. In this case, because the Cu substrate was bent upward by the tensile stress, there was a smallest deformation created in the Cu substrate. Further observation revealed that the maximum stresses were generated at both the interfaces of AlGaInP/AlAs and AlAs/GaAs substrate. Figure 4(b) displays the simulation results of stress distributions for these five AlGaInP LEDs on various patterned Cu substrates (with the structures A, B, C, D, and E), and the stress distributions on these samples with various etching amounts of the AlAs layer (10%, 50%, and 90%) were clearly presented. At an etching amount of 10%, it can be found that the LEDs on Cu substrates with the structures A and B both possessed maximum stress in four corners of the chip. Moreover, with increasing the etching amount, the location with maximum stress gradually shifted to the chip center. As shown in Figs. 3(b) and 3(c), we observed that the cracks on the sample extended from the corner to the center, which was in good agreement with the simulation results. For the LED on Cu substrate with the structure C, the maximum stress was formed in the edges of LED chip at the initial stage of ELO process. When the AlAs layer was gradually etched, the location with maximum stress also moved toward the chip center. On the other hand, for the LEDs on Cu substrates with the structures D and E, the stress distribution at the initial stage of ELO process (with an etching amount of 10%) was similar to that with the structure C, i.e., the maximum stress was created in the edges of LED chip. However, with increasing the etching amount to 50% and 90%, the maximum stress was still formed in the edges of chip. This indicates that the maximum stress can be confined to the chip edges during the ELO process through the designs of patterned Cu substrates with the structures D and E. Figure 4(c) shows the simulated maximum stresses in these five samples as a function of etching amount of the AlAs layer from 10% to 90%. The simulation results reveal that maximum stress of the sample on Cu substrate with the structure A is largest than that with the structures B-E. After designing the patterned Cu substrates with the structures D and E, maximum stresses of these two samples were reduced significantly, resulting in an efficient decrease of the crack formation in the AlGaInP epilayer.
On the other hand, according to the results of etching tests, the lateral etching rates of the AlAs layers in the samples fabricated on patterned Cu substrates with the structures D and E were determined to be 14.6 and 10.7 μm/min, respectively. As shown in Figs. 2(e) and 2(f), it can be found that the total area of etching through-hole in the sample on Cu substrate with the structure D is larger than that with the structure E. Obviously, the sample on Cu substrate with the structure D possesses both relatively higher injection rate of etching solution into the AlAs layer and venting rate of etching reactants during the ELO process. As a result, the lateral etching rate of the AlAs layer in the sample on Cu substrate with the structure D is higher than that with the structure E. In other words, to remove the GaAs substrate, the soaking time in the etching solution for the sample on Cu substrate with the structure D is relatively shorter than that with the structure E. The shorter soaking time, i.e. the higher lateral etching rate of the AlAs layer, would avoid excessive etch to the LED epitaxial structure. Therefore, although the LED on Cu substrate with the structure E had smaller stress than that with the structure D, the patterned Cu substrate with the structure D (shown in Fig. 4(d)) was still chosen as the optimum structure for further fabrication of ELO-LED due to its higher lateral etching rate of the AlAs layer.
Figures 5(a) and 5(b) show the spatial distribution images of light output of the ELO-LEDs fabricated on Cu substrates with the structures B and D, respectively, at an injection current of 50 mA. In Fig. 5(a), it can be seen that the emitted light in the sample on Cu substrate with the structure B was highly non-uniform, resulting from the generation of a large number of cracks. The non-uniform light emission was apparently observed in the corners of the sample, revealing that the cracks were easily formed in the corners. However, when the ELO-LED was prepared on Cu substrate with the structure D, the emitted light in the entire chip became very uniform, as shown in Fig. 5(b). It indicates that the AlGaInP epilayer in this sample is almost crack-free.
Then, the CE-LED and ELO-LED were both fabricated to compare their optoelectronic performances and thermal properties, as discussed in Figs. 6 and 7, respectively. It should be noted again that the patterned Cu substrate with the structure D was employed to prepare the ELO-LED. Figure 6(a) shows the photographs of these two devices before and after operating. We can observe that the light emissions were both uniformly distributed in these two devices. It reveals that there is almost no crack formation in the AlGaInP epilayer of the ELO-LED. Figure 6(b) exhibits the I-V characteristics measured under the forward bias for the CE-LED and ELO-LED. The forward voltages (@20 mA) of CE-LED and ELO-LED were measured to be 1.80 and 1.82 V, respectively. With increasing the driven current to 350 mA, the forward voltages of these two devices were increased to 2.20 and 2.29 V, respectively. Besides, the I-V characteristics measured under the reverse bias were presented in the inset of Fig. 6(b). The results indicate that the leakage currents (@-5 V) of CE-LED and ELO-LED were 8 and 11 nA, respectively. As mentioned above, the leakage current (@-5 V) of the ELO-LED fabricated on Cu substrate with the structure B was 0.2-0.8 mA. This implies that the leakage current of the ELO-LED can be significantly reduced via the employment of patterned Cu substrate with the structure D because of an efficient decrease in the crack formation. The output powers and wall plug efficiencies as a function of injection current for the CE-LED and ELO-LED are displayed in Fig. 6(c). The output powers (@350 mA) of CE-LED and ELO-LED were 49.9 and 48.2 mW, respectively. Meanwhile, the wall plug efficiencies (@350 mA) of these two samples were determined to be 6.43% and 6.13%, respectively. Based on the above measurement results consisting of forward voltage, leakage current, output power, and wall plug efficiency, it can be found that there is a very slight difference in the optoelectronic characteristics between the CE-LED and ELO-LED. This is ascribed to the design of patterned Cu substrate with the structure D for the ELO-LED fabrication.
Figure 7(a) shows the peak emission wavelength (determined by EL measurement) as a function of injection current for the CE-LED and ELO-LED. The results indicate that the peak position of the emission for these two LEDs both shift toward longer wavelength with increasing the injection current, resulting from the heat formation in the device. Moreover, the peak positions of emission wavelength at various injection currents and the red-shift tendency for these two devices are both close to each other. To further confirm the thermal characteristics of these two LEDs, the samples were measured by thermal infrared imaging. Figures 7(b) and 7(c) exhibit the surface temperature distributions of the CE-LED and ELO-LED at an injection current of 350 mA. The surface temperatures of the CE-LED and ELO-LED were characterized to be 46.9-48.3 and 45.2-47.0 °C, whereas the temperature differences were 1.4 and 1.8 °C, respectively. According to the results of Figs. 7(a)-7(c), the thermal characteristics of the ELO-LED are very similar to those of the CE-LED. This proves again that there is almost no crack formation in the epilayer of the ELO-LED with the aid of optimum patterned Cu substrate (with the structure D), leading to the uniform surface temperature distribution. Obviously, through the FEM simulation method, the optimum patterned Cu substrate can be designed for the fabrication of ELO-LED. The main disadvantage of crack formation in the epilayer of this LED during the ELO process is successfully solved, and the device performances of the ELO-LED were significantly improved. Most importantly, the GaAs substrate can be removed and reused by employing the ELO technique, indicating that the pattern design of Cu substrate will benefit the application of AlGaInP-based LEDs.
Actually, we have successfully performed the ELO process to reuse the GaAs substrate at least five times. It indicates that the ELO process proposed in this research can be applied for the multiple reuses of GaAs substrate, and this is very beneficial for the cost reduction of optoelectronic industries. Furthermore, in the future, the use of large-area substrate is the main trend of optoelectronic applications. When the ELO process is performed on the sample with the etching through-hole arrays, the separation time of a large-area substrate is the same as that of a small-area substrate. However, the substrate thickness is increased with increasing its area, which will lead to an increment in the removal time of substrate as the CE process is carried out. Thus, in comparison to the CE process, the substrate reuse obtained from the ELO process is indeed an excellent advantage, especially for the multiple reuses and the employment of large-area substrate.
In this study, the vertical-type AlGaInP LEDs on Cu substrates were fabricated by using two methods, i.e. the chemical etching process for the removal of GaAs substrate and the ELO technique for the separation of GaAs substrate from the LED device. Then the AlGaInP LEDs can be transferred to Cu substrates. To solve the problem of crack formation in the epilayer of AlGaInP LED during the ELO process, five patterns for Cu substrates were designed, and the FEM simulation method was used to confirm the feasibility of these Cu substrates. As the typical Cu substrate (with the structure B) was used, the maximum stress was formed in the corners of LED chip at the initial stage of ELO process. Further increasing the etching amount of the AlAs sacrificial layer, the location of maximum stress shifted to the chip center gradually, resulting in severe crack generation in the epilayer. For the simulations of other patterned Cu substrates (with the structures D and E), we observed that the maximum stress can be confined to the chip edges during the ELO process, even though the etching amount of the AlAs layer reached as high as 90%. Moreover, the maximum stress values in the samples fabricated on these two patterned Cu substrates (with the structures D and E) were much lower than that on the typical Cu substrate (with the structure B). For the consideration of lateral etching rate of the AlAs layer, the patterned Cu substrate with the structure D was chosen for the fabrication of ELO-LED. After the measurements of device performances, it can be found that the optoelectronic and thermal characteristics of the ELO-LED were very similar to those of the CE-LED. Therefore, via the pattern design of Cu substrate with the structure D, the performances of ELO-LED can be enhanced significantly. In addition, by performing the ELO process, the separated GaAs substrate can be reused, which is beneficial for the optoelectronic applications.
This work was supported by the Ministry of Science and Technology (MOST), Southern Taiwan Science Park and HsinChu Science Park, Taiwan, R.O.C. under Grant nos. MOST 104-3113-E-005-001, 102CE06, 101A08 and 103A13, respectively. Authors also thank Epistar company for LED epiwafers supporting.
References and links
1. M. G. Craford, “LED’s challenge the incandescent,” IEEE Circ. Dev. Mag. 8(5), 24–29 (1992). [CrossRef]
2. T. Gessmann and E. F. Schubert, “High-efficiency AlGaInP light-emitting diodes for solid-state lighting applications,” J. Appl. Phys. 95(5), 2203–2216 (2004). [CrossRef]
3. K. Streubel, N. Linder, R. Wirth, and A. Jaeger, “High brightness AlGaInP light-emitting diodes,” IEEE J. Sel. Top. Quantum Electron. 8(2), 321–332 (2002). [CrossRef]
4. M. R. Krames, M. Ochiai-Holcomb, G. E. Hofler, C. Carter-Coman, E. I. Chen, I.-H. Tan, P. Grillot, N. F. Gardner, H. C. Chui, J.-W. Huang, S. A. Stockman, F. A. Kish, M. G. Craford, T. S. Tan, C. P. Kocot, M. Hueschen, J. Posselt, B. Loh, G. Sasser, and D. Collins, “High-power truncated-inverted-pyramid (AlxGa1-x)0.5In0.5P/GaP light-emitting diodes exhibiting >50% external quantum efficiency,” Appl. Phys. Lett. 75(16), 2365–2367 (1999). [CrossRef]
5. W. C. Peng and Y. C. S. Wu, “High-power AlGaInP light-emitting diodes with metal substrates fabricated by wafer bonding,” Appl. Phys. Lett. 84(11), 1841–1843 (2004). [CrossRef]
6. R. H. Horng, C. E. Lee, C. Y. Kung, S. H. Huang, and D. S. Wuu, “High-power AlGaInP light-emitting diodes with patterned copper substrates by electroplating,” Jpn. J. Appl. Phys. 43(4B4B), L576–L578 (2004). [CrossRef]
7. R. H. Horng, M. C. Tseng, F. L. Wu, C. H. Li, C. H. Wu, and M. D. Yang, “Thin film solar cells fabricated using cross-shaped pattern epilayer lift-off technology for substrate recycling applications,” IEEE Trans. Electron. Dev. 59(3), 666–672 (2012). [CrossRef]
8. J. Yoon, S. Jo, I. S. Chun, I. Jung, H. S. Kim, M. Meitl, E. Menard, X. Li, J. J. Coleman, U. Paik, and J. A. Rogers, “GaAs photovoltaics and optoelectronics using releasable multilayer epitaxial assemblies,” Nature 465(7296), 329–333 (2010). [CrossRef] [PubMed]
9. F. L. Wu, S. L. Ou, R. H. Horng, and Y. C. Kao, “Improvement in separation rate of epitaxial lift-off by hydrophilic solvent for GaAs solar cell applications,” Sol. Energy Mater. Sol. Cells 122, 233–240 (2014). [CrossRef]
10. S. Liu, K. Wang, Z. Gan, X. Luo, X. Song, Z. Chen, Z. Liu, H. Yan, W. Wei, and P. Wang, “Several Modeling Issues in LED, 3D-SiP, and Nano Interconnects,” in 10th. Int. Conf. on Thermal, Mechanical and Multiphysics Simulation and Experiments in Micro-Electronics and Micro-Systems, (EuroSimE, 2009), pp. 1–10. [CrossRef]
11. M. Levinshtein, S. Rumyantsev, and M. Shur, “Handbook Series on Semiconductor Parameters,” 2, Ternary And Quaternary III–V Compounds, (World Scientific, 1999), Chap. 1.
12. B. Weiss, V. Gröger, G. Khatibi, A. Kotas, P. Zimprich, R. Stickler, and B. Zagar, “Characterization of mechanical and thermal properties of thin Cu foils and wires,” Sens. Actuators A Phys. 99(1–2), 172–182 (2002).
13. T. L. Chou, C. F. Huang, C. N. Han, S. Y. Yang, and K. N. Chiang, “Fabrication process simulation and reliability improvement of high-brightness LEDs,” Microelectron. Reliab. 49(9–11), 1244–1249 (2009). [CrossRef]