In recent years, high power and high wavelength stability GaN-based blue laser diodes (LDs) have drawn great attention due to their potential applications in laser display and lightening.^[1–3] Instead of watt-class single stripe laser diodes, integrating multiple stripes into one single chip to constitute an LD array is an effective way to achieve high output power.^[4–7] Usually, the LD chips are packaged in a TO-can to realize electrical pumping and thermal dissipation. However, for an LD array, the temperature in each stripe can vary because the stripes are not thermally independent.^[8,9] Careful consideration should be taken to optimize the arrangement of the stripes to achieve a uniform temperature distribution among the stripes. Furthermore, there is an inherent disadvantage for conventional TO package. Due to the asymmetric structure of the TO-can, the heat dissipation of the area close to the rear cavity facet is more effective than that of the area close to the front cavity facet. The temperature within the chip is not uniform along the cavity direction.^[10] The nonuniformity of temperature among the stripes and along the cavity direction may deteriorate the wavelength stability, output power of LDs, and the device reliability.^[5,11,12]

In this paper, the thermal characteristics of a GaN-based LD mini-array (with 5 laser stripes) in a TO-9 package is analyzed by using finite element method (FEM). Based on the analysis results, the arrangement of stripes is studied and optimized numerically to enhance the temperature uniformity. The influences of different submount structures are compared, and the temperature uniformity can be significantly improved if a trapezoid AlN submount is used.

The following heat conduction equation is numerically solved for the analysis of thermal characteristics of the GaN based LD mini-array in a TO-9 package.

Three-dimensional TO-9 packaged LD mini-array model is built for thermal simulation. Figure 1 shows the schematic structure of the TO-9 packaged LD. The GaN-based LD mini-array chip is mounted on an AlN submount. The AlN submount is then attached on a copper heat sink. AuSn solder is used to bond the LD mini-array chip, the AlN submount, and the copper heat sink. Figure 2 shows the schematic structure of the LD mini-array chip. The width of the LD mini-array chip is 500 μm and the cavity length is 1200 μm. In the middle of the chip is a 100 μm wide emitting area, which consists of five laser stripes (stripe 1–5). The structural parameters rw1, rw2, rw3, rw4, and rw5 represent the width of each stripe; s1, s2, s3, and s4 represent the spacing between the stripes.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Schematic diagram of a GaN-based LD mini-array in a TO-9 package.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) Schematic diagram of a GaN-based LD mini-array chip with five stripes.

Table 1 shows the material parameters used in the thermal simulation. We suppose that the wall plug efficiency (WPE) of the LD mini-array is 30%, and the input power is set to 10 W, which means that the optical power is 3 W and the thermal power is 7 W. In GaN based LDs, except for the active region, the heat generation in the P-cladding layer is also large,^[14] so we assume that the thermal power is uniformly distributed in the active layer and the P-cladding layer of each stripe. The stem in the TO-9 package is supposed to be seated on a thermoelectric cooler (TEC), and the temperature is fixed to 300 K.

Table 1.

Table 1.

Table 1. Material parameters in the simulation at 300 K. . Component in the packaged structure Material Thermal conductivity/W·mK−1 1 stem copper 397 2 heat sink copper 397 3 solder AuSn 57 4 submount AlN 170 5 substrate GaN 130 6 N-cladding layer AlGaN 20 7 N-waveguide layer InGaN 20 8 active layer InGaN/GaN MQWs 70 9 P-waveguide layer InGaN 70 10 P-EBL AlGaN 70 11 P-cladding layer AlGaN 20 12 P-contact layer GaN 20 13 insulating layer SiO2 1.4 14 anode Au 317 15 cathode Au 317

Table 1. Material parameters in the simulation at 300 K. .

By using the parameters mentioned above, we calculated the temperature distribution of a GaN based LD mini-array in a TO-9 package, as shown in Fig. 3. In the middle of the chip is a 100 μm wide emitting area, in which five 10 μm wide laser stripes are uniformly spaced. We extracted the temperature data among the stripes and along the cavity direction (along the x-axis and along the z-axis), as shown in the inset of Fig. 3. It is found that among the stripes, higher temperatures locate in the middle stripes, while along the cavity direction, temperatures near the front cavity facet is higher than that near the rear cavity facet. The close spacing causes serious thermal crosstalk between the stripes, which leads to a nonuniform temperature distribution among the stripes. The rear cavity facet is closer to the stem (seated on a TEC), which also leads to a nonuniform temperature distribution along the cavity direction.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Temperature distributions of GaN-based LD mini-array in a TO-9 package. The inset is the extracted temperature data along the x-axis and the z-axis.

To improve the temperature uniformity of the LD mini-array chip, the temperature distribution among the stripes and along the cavity direction is separately optimized by reasonable arrangement of the stripes and the structure of AlN submount. Details are discussed in the following sections.

3.1. Optimization of the temperature distribution among the stripes

Figure 4 shows the schematic diagram of the emitting area, and the structural parameters rw1–rw5 and s1–s4 are signed in the diagram. Four sets of structural parameters are listed in Table 2, namely P1, P2, P3, and P4, and figure 5 is the corresponding temperature among the stripes (x-axis direction) extracted from the simulation results when using different sets of structural parameters. The parameters of P1 represent the conventional structure in which five 10 μm stripes are equally spaced in the emitting area. The temperatures of the middle stripes are higher than those of the stripes in both sides, and the temperature varies by more than 5 K. Firstly, we enlarge the spacing s2 and s3 from 22.5 μm to 26 μm, and reduce the spacing s1 and s4 from 22.5 μm to 19 μm, as shown by P2 in Table 2. Compared with using structural parameters of P1, the temperatures of the stripes in the middle decrease and the temperatures of the side ones increase, and the temperature difference is reduced to 3 K. Then we continue to increase s2 and s3 to 28 μm and reduce s1 and s3 to 17 μm, as shown by P3 in Table 2. The temperatures of stripe 2, stripe 3, and stripe 4 become similar, but the temperatures of stripe 1 and stripe 5 are still low, and the temperature difference is reduced to 2.5 K. It is difficult to achieve a uniform temperature distribution among stripes via adjusting the spacing only. At last, based on the parameters of P2, we decrease the widths of stripe 2, stripe 3, and stripe 4, and increase the widths of stripe 1 and stripe 5, as shown by P4 in Table 2. Parameters rw3, rw4, and rw2 are reduced to 8.5 μm, 8.25 μm, and 8.25 μm, and rw1 and rw5 are increased to 12.5 μm, while the total width of the stripes is still 50 μm (keep the same heat power generated at the emitting area). The temperatures of all stripes become similar and the temperature difference is reduced to less than 0.4 K.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Schematic diagram of emitting area in the GaN based LD mini-array chip.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Temperature distribution among the stripes (x-axis direction) with different structural parameters: (a) P1, P2, and P3, (b) P2 and P4.

Table 2.

Table 2.

Table 2. Structural parameters in μm. . rw1 s1 rw2 s2 rw3 s3 rw4 s4 rw5 P1 10 22.5 10 22.5 10 22.5 10 22.5 10 P2 10 19 10 26 10 26 10 19 10 P3 10 17 10 28 10 28 10 17 10 P4 12.5 19 8.25 26 8.5 26 8.25 19 12.5

Table 2. Structural parameters in μm. .

Compared with using structural parameters of P1, adopting structural parameters of P4, as shown in Fig. 6, can significantly improve the temperature uniformity of the stripes, while the highest temperature decreases from 367.7 K to 364.4 K.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Temperature distributions of GaN-based LD mini-array in a TO-9 package using structural parameters of P4. The inset is the extracted temperature data along the x-axis and the z-axis.

3.2. Optimization of the temperature distribution along the cavity direction

As mentioned above, the temperature along the cavity direction (z-axis direction) is also nonuniform. As shown in Fig. 6, the temperature near the front cavity facet is higher than that near the rear cavity facet. We have changed the structure of AlN submount to improve the temperature distribution. The idea is to enlarge the volume of the AlN submount near the front cavity facet to dissipate more heat, while reduce the volume of the AlN submount near the rear cavity facet to improve the temperature uniformity near the rear facet.

Figures 7 and 8 show four different AlN submount structures and their corresponding temperature distributions along the cavity direction, respectively. The thicknesses of the AlN submount used are all set to 300 μm,^[15] but the shapes vary. Figure 7(a) shows 900 μm × 1400 μm rectangular AlN submount, which is conventionally used in the TO-9 package. In our simulation results, the temperature difference along the z-axis can be 7.6 K. In Fig. 7(b), triangular AlN submount is used in simulation model. Since the upper surface size of the copper heat sink in TO-9 package is 1850 μm × 1750 μm, we set the width in front to 1650 μm and the length to 1800 μm (50 μm reserved for AuSn solder). The overall temperature is reduced due to the increase in the volume of AlN submount, and the temperature difference along the z-axis is reduced to 5.5 K, which is still large. In Fig. 7(c), we cut off the volume of the triangular AlN submount near the rear cavity facet, and the length of the AlN submount is reduced to 1400 μm. The shape of the AlN submount becomes a trapezoid column and the temperature difference is reduced to 4.5 K. In Fig. 7(d), the volume behind is further reduced until the length of the AlN submount is 1210 μm. The overall temperature slightly increases but the highest temperature is still lower than that using the conventional rectangular AlN submount, and the temperature difference along the z-axis is only 0.5 K.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. (color online) Schematic diagram of different AlN submount structures. (a) Rectangular AlN submount. (b) Triangular AlN submount. (c) and (d) Trapezoid AlN submount.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. (color online) Temperature distribution along the cavity direction (z-axis direction) with different AlN submount structures. Curves a, b, c, and d correspond to Figs. 7(a), 7(b), 7(c), and 7(d), respectively.

3.3. Temperature distribution under different input powers

By optimizing the structural parameters and using a trapezoid AlN submount, the temperature difference among the stripes and along the cavity direction are less than 0.4 K and 0.5 K, respectively, as shown in Fig. 9. The temperature distribution of the LD mini-array operating under different input powers is also simulated by the model described above, as shown in Fig. 10. The input power is set to be 3.33 W, 6.67 W, 10 W, and 13.33 W, respectively. Although the overall temperature on the ridge is increased with the increase of the thermal power, the temperature difference among the stripes is within 0.7 K, while the temperature difference along the cavity direction is within 1 K.

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. (color online) Temperature distribution of GaN-based LD mini-array in a TO-9 package using structural parameters of P4 and a trapezoid AlN submount. The inset is the extracted temperature data along the x-axis and the z-axis.

Fig. 10.

	Figure Option ViewDownloadNew Window
	Fig. 10. (color online) Temperature distribution under different thermal power (a) among stripes; (b) along the cavity direction.

We have calculated the temperature pattern of a GaN-based LD mini-array with 5 stripes in a TO-9 package and optimized the temperature distribution by reasonable arrangement of the stripes and using the trapezoid AlN submount. From the analyses it is suggested that the temperature distribution among the stripes can be adjusted by the widths of stripes and the spacing between stripes, and the temperature distribution along the cavity direction can be adjusted by optimizing the structure of AlN submount. When the input power of LD mini-array is 10 W, by using the optimal structural parameters and a trapezoid AlN submount, the temperature variation among stripes is less than 0.4 K and along the cavity direction is less than 0.5 K. Meanwhile, comparing with the conventional structure, the highest temperature decreases from 366.7 K to 364.2 K. Regardless of the variation of the input power, the temperature fluctuations among stripes and along cavity direction are always within controllable range. As a result, the lasing wavelength uniformity and the reliability of the GaN based LD mini-array can be improved.