Fabrication of NBM-LEDs - IMPROVED PERFORMANCE OF INGAN-GAN LIGHT-EMITTING DIODE

CHAPTER 4 IMPROVED PERFORMANCE OF INGAN-GAN LIGHT-EMITTING DIODE

4.2.3 Fabrication of NBM-LEDs

Fig. 4.3 shows the fabrication processes of the NBM-LEDs. To fabricate the NBM-LED, PR-LED wafer was bonded to a host substrate using a glue layer.[9] The bonding process was performed at 200 °C for 1 hour. After bonding, the sapphire substrate was lifted off by irradiating it with KrF pulsed excimer laser (248 nm) to expose the u-GaN surface. To fabricate the n-bowl structure, photoresist reflow and dry etching method were used.[12] A photoresist layer was spun onto the u-GaN surface and patterned by standard photolithography into cylindrical disks. When baked at the temperature higher than 200 °C, they turned to hemispherical shapes because of the surface tension. The performed shapes were then transferred to u-GaN using ICP process. Fig.

4.4 shows the SEM image of the 4-3 n-bowl structure. A mirror system (Ni/Ag/Ni) was then introduced on the etched surface to reflect the downward photons.[13] Sample was subsequently bonded to Si substrate with an adhesive layer.[9] The host substrate and glue layer were subsequently removed. For baseline comparison the performance of the PR-LED, sapphire was chosen as the substrate of NBM-LED. With the same sapphire substrate, we could investigate the influence of n-bowl structure on the performance of the LED chips. The samples described herein were only cut into chips without encapsulation.

4.3 Results and Discussions

Fig. 4.5 shows the current-voltage (I–V) characteristics of LEDs. It was found that the

forward voltages of NBM-LEDs were in the range of 3.4-3.6 V (at 20 mA), which were similar to that of PR-LED, indicating that transfer method did not change much of LED performance. The little difference might due to the uniformity of the original wafer. The effects of the injection current on the luminous intensity are depicted in Fig. 4.6. The intensities of three NBM-LEDs were all higher than that of PR-LED. This is because the n-bowl mirror structure in NBM-LEDs not only reflected the downward photons to the front side, but also redirected the photons which were originally emitted out of the escape cone, back into the escape cone.[9]

Fig. 4.6 also indicates that the light intensity of NBM-LEDs increased with the decrease of n-bowl dimension. As shown in Table 4.1, the light intensity of 3-3 NBM-LED was 176.0 mcd,

which was 2.33 times higher than that of the PR-LED, 1.95 times higher than that of the 25-4 NBM-LEDs, and 1.43 times higher than that of the 4-3 NBM-LED. This is because Ni/Ag/Ni n-bowl system acting as a concave mirror. The scattering ability of concave mirror increases with

the curvature of n-bowl. According to the optical lever principle, the variation of the reflection angle of photons is two times than that of the surface angle.[14] It means that the surface of n-bowl with steep slope (large curvature) can cause more changes in the photon paths. As

mentioned by Oder et al.,[12] the bowl structure is a part of the sphere (Fig. 4.7). If the sphere has radius R, then R can be deduced from the measured values of bowl radius r and height h, and gives the relation as

The radius of curvature is listed in Table 4.1.

The light output versus injection current curves for the LEDs were also shown in Fig. 4.8.

LEDs were measured in an integrating sphere and were not encapsulated during the measurements. It was found that the 3-3 NBM-LED achieved an output power of 6.21 mW at 20

mA, which was 43% larger than the PR-LED, 19% larger than the 25-4 NBM-LED, and 4%

larger than the 4-3 NBM-LED, as shown in Table 4.1. The improvement of the light output was also due to the reflection of the downward photons to the front side and redirection of photons from the roughened u-GaN surface and the n-bowl mirror on sapphire substrate.

The radiation patterns of the LEDs are shown in Fig. 4.9. Their view angles (half-center brightness, which is the angle for 50% of full luminosity) were shown in Table 4.1. It was found that the view angle decreased with the diameter of n-bowl. The viewing angle of PR-LED was 130°. As the diameter of n-bowl decreased from 25 to 3 μm, the view angle decreased from 120°

to 118°. This is because the surface of n-bowl can change the photon path. These changes can be explained by the basic geometric optics, as shown in Fig. 4.10. For a spherical mirror,[15] a simple equation relates the source distance S, the imaging distance S’, the focal length f, and the radius of curvature R is as follow:

1 1 2 1

S+S =R = f (4.2)

From the measured values listed in Table 4.1, the focal lengths of the 3-3, 4-3, and 25-4 n-bowls could be calculated as 1.73, 2.25, and 31.12 μm, respectively. The actual source

distances S from active layers to three different n-bowl structures were 3.78, 3.93, and 4.15 μm.

If the source is at the focal point (S=f), the imaging distance will be infinite (S’=∞). In other words, photons were redirected to the vertical direction as shown in Fig. 4.11.

The positions of light sources of 3-3 and 4-3 NBM-LEDs were quite near their focal points.

As a result, photons from the active layers might be redirected toward the vertical direction and benefit to the vertical intensity. On the other hand, the location of the light source of 25-4 NBM-LED was far from the focal point. As a result, the redirect ability of 25-4 NBM-LED is not as good as that of 3-3 and 4-3 NBM-LEDs.

These improvements were similar to our previous finding of RM-LED (LED with a roughened KOH-etched GaN surface and an Ag mirror on the backside of sapphire substrate).[9]

The output power enhancement of 3-3 NBM-LED was 1.43 times, which was less than that of RM-LED (1.49 times). This observation suggested that scattering ability of the n-bowl structure was not as good as that of KOH-etched GaN surface. On the other hand, the vertical intensity enhancement of 3-3 NBM-LED was 2.33 times, which was higher than that of RM-LED (2 times). This is because the n-bowl structures had good focusing ability.

4.4 Summary

NBM-LEDs with roughened p-GaN surface and n-bowl mirror structure were fabricated by wafer bonding and laser lift-off technology. The forward voltages of these NBM-LEDs were close to PR-LED, indicating that transfer method did not change much of LED structure. The performance of NBM-LEDs was better than that of PR-LED. The luminance intensity of 3-3 NBM-LED at 20 mA was 176.0 mcd, which was 2.33, 1.95, and 1.43 times higher than PR-LED, and 25-4 and 4-3 NBM-LEDs. The 3-3 NBM-LED achieved an output power of 6.21 mW, which was 43% larger than the PR-LED, 19% larger than the 25-4 NBM-LED, and 4% larger than the 4-3 NBM-LED. Besides, the view angle decreased with the diameter of n-bowl. The viewing angle of PR-LED was 130°. As the diameter of n-bowl decreased to 3 μm, the view angle decreased to 118°. This is because the n-bowl mirror structure acting as a concave mirror. It not only reflected the downward photons to the front side, but also redirected the photons which were originally emitted out of the escape cone, back into the escape cone.

Figure 4.1 Schematic illustration of LEDs: (a) PR-LED, (b) NBM-LED, and (c) the n-bowl mirror array of the 25-4 NBM-LED. (d) is the cross-sectional SEM image of the 4-3 n-bowl structure.

Table 4.1 The parameters and performances of LEDs.

Figure 4.2 Fabrication process of the PR-LED.

Figure 4.3 Experiment flowchart.

Figure 4.4 SEM image of the 4-3 n-bowl structure.

0 50 100 150 200

0 1 2 3 4 5 6 7 8

Voltage (V)

C urre nt (mA)

3-3 NBM-LED 4-3 NBM-LED 25-4 NBM-LED PR-LED

Figure 4.5 Current-voltage characteristic of the LEDs.

0 10 20 30 40 50

0 50 100 150 200 250 300 350

In te nsi ty (m cd )

Current (mA)

3-3 DSM-LED 4-3 DSM-LED 25-4 DSM-LED PR-LED

Figure 4.6 The effects of injection current on the luminous intensity of the LEDs.

Figure 4.7 Illustration of the physical parameters of the bowl. The bowl has radius r, height h, and R is the radius of the complete sphere.

0 10 20 30 40 50

0 5 10 15

3-3 DSM-LED 4-3 DSM-LED 25-4 DSM-LED PR-LED

Po we r (mW)

Current (mA)

Figure 4.8 The effects of injection current on the light output of the LEDs.

-100 -80 -60 -40 -20 0 20 40 60 80 100 0.0

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

3-3 NBM-LED 4-3 NBM-LED 25-4 NBM-LED PR-LED

Ill umi na nc e (a .u.)

Angle (degree)

Figure 4.9 The radiation patterns of LEDs.

Figure 4.10 Illustration of the physical parameters of the concave mirror.

Figure 4.11 Schematic illustration of the n-bowl structure.

CHAPTER 5 INGAN-GAN LIGHT-EMITTING DIODE PERFORMANCE IMPROVED BY ROUGHENING INDIUM TIN OXIDE WINDOW

LAYER VIA NATURAL LITHOGRAPHY

5.1 Introduction

High-brightness GaN-based light-emitting diodes (LEDs) have attracted intense research for their various applications in mobile phones, full-color displays, mini projector and lightings in the past few years. Internal quantum efficiency and light extraction efficiency are two principal parameters for improving the efficiency of LEDs; both of which can be improved through enhancing crystal quality and modifying the LED structure.[1-3] The internal quantum efficiency of GaN LEDs generally exceeds 70%, which is much higher than the efficiency of conventional light sources such as incandescent lamps and fluorescent lamps.[4] However, the light extraction efficiency of GaN-based LEDs is limited by the large difference in refractive index between the GaN film and the surrounding air. According to Snell’s law, the critical angle for light traveling from GaN (n=2.5) to air (n=1.0) is 23°. Only the light within the critical angle will cross the air, while the other experiences total internal reflection until it is absorbed. Many methods have been developed to solve this problem, which focus mostly on sidewall obliquing,[5] n-GaN morphology changing,[6] and sapphire patterning of the LED.[7] p-Side treatment, however, is still hard to perform because of the small thickness of p-GaN. In addition, Fresnel loss is another problem for light extraction. When photons pass through the GaN and air interface, there will be a reflection loss of about 18.4% caused by the large refractive index difference between the two

mediums.[8] Inserting an intermediate layer, like indium tin oxide (ITO), can moderate this phenomenon. This ITO layer not only serves as a good window layer, but also makes the p-side treatment possible. In this study, a roughened ITO surface was introduced to LED by natural lithography, in which a photoresist layer was used as a mask for dry etching.

5.2 Experiments

5.2.1 Illustration and Abbreviation of LEDs

Two kinds of LEDs were investigated in this study. As shown in Fig. 5.1, samples designated as “CV-LED” were conventional LEDs without any surface treatment, while samples designated as “IR-LED” were LEDs with ITO roughened surface. Two types of IR-LEDs were investigated. They were denoted as “IR1.6-LED” (ITO roughened surface produced from a 1.6-μm photoresist mask) and “IR1.9-LED” (ITO roughened surface produced from a 1.9-μm photoresist mask).

5.2.2 Fabrication of the CV- and IR-LEDs

InGaN-GaN films were grown by metalorganic chemical vapor deposition (MOCVD) on a sapphire substrate. The LED structure contained a 0.3-μm-thick p-type Mg-doped GaN grown at 950°C, an InGaN-GaN multiquantum well (MQW) with six pairs of InGaN (3 nm)/GaN (9 nm) at 800°C, a 2-μm-thick n-type Si-doped GaN at 1050°C, a 2-μm-thick undoped-GaN layer film at 1050°C, and a buffer layer at 550°C on the sapphire substrate. The Si-doped n⁺-InGaN layer was employed to form the ohmic contact between ITO and p-GaN.[9] A 2.8-μm ITO film, which was much thicker than that of conventional LED (0.3 μm), was then deposited on the top of the LED

structure. This extra thickness of the ITO layer was employed to create the roughened surface.

The experiment flowchart is shown in Fig. 5.2. To perform ITO roughening, photoresist layers (AZ-1518) of different thicknesses (1.6 and 1.9 μm) were coated as the dry etching mask.

The inductively coupled plasma (ICP) etching conditions used in this study were Cl2/BCl3/CH4/He (4:6:1:2) mixture gas, 5 mTorr pressure, 400 W ICP power, and 100 W rf power. During etching, the surface morphology of the photoresist layer changed due to the formation of giant folds.[10] This undulation was subsequently transferred to ITO in the same ICP run which etched the entire photoresist and most of the ITO away until only 0.3 μm of ITO remains. The device mesa with a chip size of 300 × 300 μm² was then defined by ICP to remove Mg-doped GaN layer and MQW until the Si-doped GaN layer was exposed. Samples were then post-annealed at 600°C to reconstruct the ICP damaged layer. Finally, Cr/Au was deposited onto the ITO layer and n-GaN layer as electrodes. For baseline comparison, all samples were prepared from the same epitaxial wafer. The samples described herein were only cut into chips without encapsulation.

5.3 Results and discussions

The dry-etching process is a critical step in the fabrication of nitride-based LEDs. During the ICP etching process, the ICP-damaged layer is a well-known and inevitable problem.[11]

Therefore, the variation in contact between the ICP-etched ITO surface and Cr/Au pad is an important issue. The specific contact resistance (ρc) was examined using the transmission line model (TLM).[12] Five CV-LED samples covered with 2-μm ITO were utilized to study the effect of post-annealing temperature on the ρ_c (ICP-etched ITO surface). Sample A was untreated, while samples B-E underwent ICP treatment for 10 min. Then, samples C-E were post-annealed

at 200°C, 400°C and 600°C for 30 min, respectively. Sample B did not undergo any annealing.

TLM gives two relations as follow:

2 ^SH 2 ^SK ^t ^SH

where R_T is the impedance between the two adjacent contact pads, R_C is the contact resistance, R_SH is the sheet resistance of the semiconductor layer outside the contact region, W is the width of the contact area, L is the distance between two adjacent pads, RSK is the modified sheet resistance under the contact, and Lt is the transfer length. The derived specific contact resistances are shown in Table 5.1. It was found that specific contact resistances of the samples after ICP treatments were smaller than the untreated one (sample A). This was due to the creation of oxygen vacancies during ion bombardment.[13] The ρc increased slightly with post-annealing temperature and reached the maximum value of 7.2 × 10^-7 Ω·cm² at 400°C. This was probably due to the In₂O₃ aggregation and SnO formation at low temperature, which led to inhomogeneous distribution of Sn, and reduction in carrier concentration.[14] When annealing temperature reached 600°C, crystallites grew and the grain boundary reduced. The tin oxide in the matrix also changed from SnO to SnO₂ state.[14] Thus, the ρ_c reduced to about 1.8 × 10^-7 Ω·cm². From this measurement, it appears that the pad/ITO contact was not deteriorated when the ITO surface was after ICP etching.

The ρc of the interface was even reduced to its minimum when the sample was treated at the post-annealing temperature of the regular device process (600°C).

The surface morphologies of ITO films were examined by scanning electron microscopy (SEM) as shown in Fig. 5.3(a)-(c). The SEM image was taken at an angle of 45° from the perpendicular direction of the surface. A wave-like structure formed on IR-LED was clearly observed. This was caused by the formation of giant folds.[10] During ion bombardment, the

photoresist layer turned into two distinct layers, which comprised a highly cross-linked polymer top layer modified by the high-energy ions, and a bottom layer of normal resist not affected by ions. As the temperature of the substrate rose, the volatile components evaporating from the bottom layer contributed to the bubble formation. When the tension of the bubble skin reached the limit, it peeled off, and giant folds were formed at the inner area of the initial bubble.[10] The surface morphology of the 1.9-μm-thick photoresist after 4 min etching is shown in Fig. 5.3(d).

This undulation was subsequently transferred to ITO in the same ICP run; and hence, gave the wave-like surface as shown in Fig. 5.3.

The surface of ITO films were also measured by atomic force microscopy (AFM) to determine the degree of texturing as shown in Fig. 5.4. The root mean square (rms) roughness of CV-LED was only 9.19 nm, while that of IR1.6-LED was 122.3 nm. When the photoresist thickness reached 1.9 μm thick, the rms roughness increased to 162.5 nm due to the folds being filled by the melted photoresist.[10] During ICP etching, the substrate temperature rose as a result of thermal accumulation, causing the bottom resist to melt and fill the folds. If the photoresist was thick, there would be much more resist melted and folds would deforminto high undulation structures,[10] which yielded the rougher ITO surface after ICP etching.

I-V characteristics of LEDs are shown in Fig. 5.5. The forward voltages of IR-LEDs were

nearly the same and lay within the range of 3.50-3.55 V (at 20 mA), which were similar to those of CV-LED. These results agreed with the TLM measurement mentioned above. Owing to the creation of oxygen vacancies during ion bombardment, ρ_c of the sample surfaces still remained small after ICP treatment, hence showing similar I-V characteristics.

The effects of the injection current on luminous intensity are depicted in Fig. 5.6. The emission peak wavelength of the LEDs was 461 nm. The intensity of the two IR-LEDs was higher than that of the CV-LED. The luminance intensity of IR1.9-LED was 71.6 mcd (at 20 mA),

which was 1.5 times higher than that of the CV-LED, and 1.2 times higher than that of the IR1.6-LED. The light output powers were also shown in Fig. 5.7. LEDs were measured in an integrating sphere. It was found that the IR1.9-LED achieved an output power of 5.75 mW, which was 1.27 times higher than that of the CV-LED, and 1.07 times higher than that of the IR1.6-LED. The improvements in light intensity and output power were predictable because the roughened ITO surface provided the photons multiple opportunities to escape from the LED surface, and redirected the photons.[6]

Fig. 5.8 shows the radiation patterns of the IR-LED with various roughnesses of the ITO layers. The view angle (half-center brightness, which is the angle for 50% of full luminosity) of CV-LED was 138.0°, which was slightly larger than 137.3° and 137.2° of the IR1.6- and IR1.9-LED, respectively. This was because the cone-like surface of the undulation changes the path of photons to the vertical direction of the LED. It was found that the view angle maintained nearly the same with increasing thickness of the PR mask. That is the view angle didn’t change with increasing roughness of ITO layers. This was because the shapes of the cones on surface were the same. As shown in Fig. 5.4, although the surface roughness of the IR1.9-LED was higher than IR1.6-LED, but the slope of the sidewall of the cones didn’t change much. The only difference of the two surfaces is the height of cones.

5.4 Summary

In summary, an investigation of the relationship between ICP etching conditions (photoresist thickness and post-annealing temperature) and LED performance (specific contact resistance, forward voltage, light intensity and output power) has led to the development of a simple, effective natural lithography process for preparing ITO-textured surfaces useful for fabricating

high-brightness GaN-based LEDs. In this lithography process, photoresist layers (AZ-1518) of different thicknesses (1.6 and 1.9 μm) were used as a mask for ICP dry etching. During etching, surface of the photoresist deformed because of the thermal accumulation, and this undulation was subsequently transferred to ITO surface. It was found that the forward voltages of IR-LEDs were similar to those of the CV-LED. The light intensity and output power of IR-LEDs were better than those of the CV-LED. The luminance intensity of the IR1.9-LED was 71.6 mcd, which was 1.5 times higher than that of the CV-LED, and 1.2 times higher than that of the IR1.6-LED. The IR1.9-LED achieved an output power of 5.75 mW, which was 1.27 times higher than that of the CV-LED, and 1.07 times higher than that of the IR1.6-LED. This is because the roughened ITO

在文檔中表面微結構對氮化鎵發光二極體光電特性之影響 (頁 74-0)