|
 |
 |
|
|
|
 |
|
|
4. GaN-based blue light emitting device development by Nakamura
4.1. Selection of GaN
Shuji Nakamura began development of a blue light emitting semiconductor device at Nichia Chemical Industries, Ltd., in 1988. In order to undertake this research, he had to obtain the consent of the president of Nichia Chemical Industries through direct negotiation, something that was extraordinary for a researcher in a Japanese company to do. Material selection was an important issue. Since many researchers were pursuing ZnSe-based blue LED research and would patent their manufacturing technology processes, Nakamura decided it would be difficult to develop a novel technology. Therefore, he selected GaN as the focus of his research, which was only being pursued by a minority of researchers. Nakamura had the confidence and engineering creativity to solve the technical problems stemming from GaN-based research.
4.2. GaN thin film formation with superior uniformity
Nakamura began his experiment to make GaN thin film using standard equipment with a well-known substrate material, sapphire, and a well-known method, MOCVD. However, GaN thin film with good uniformity had yet to be produced. Nakamura thought that the gas flow scheme was the key issue to improving film uniformity. This approach is different from that taken by Akasaki and Amano, who introduced the low-temperature buffer layer. Nakamura thought over many kinds of gas flow schemes that were different from the conventional layer flow. In the conventional gas flow, material gases, including Ga compound and nitrogen, were supplied in parallel to the substrate surface as the layer flow. This conventional scheme was considered to be rational but had achieved only inferior uniformity and less than stoichiometric nitrogen content. Nakamura repeatedly modified the MOCVD equipment by himself to change the gas flow mode, and performed experiments over 500 times in a short period.
Finally, he found the most appropriate method where the main material gases, including Ga compound, flow in parallel to the substrate, and nitrogen and hydrogen are supplied vertically from the upper side to the substrate surface. In 1991, he succeeded in fabricating a GaN thin film with superior uniformity using this method, which he named the "two-flow method"10). The electron mobility calculated from the Hall effect measurement with this GaN thin film was 200 cm2/V s, which was much larger than the conventional result of 90 cm2/Vs. In an additional experiment, in which the low temperature buffer layer was introduced as well as the two-flow method, the resultant GaN thin film showed 500 cm2/Vs, which is a satisfactory figure for light emitting device fabrication. Moreover, in 1992, he achieved high quality InGaN ternary compound thin film by the two-flow method. This film was necessary to optimize the emitted light wavelength and to improve the emission efficiency11).
Nakamura reexamined the material for the low-temperature buffer layer and found that not only AlN but also GaN itself was suitable12). In the case of GaN, the material gas change is not needed between the buffer layer and the active layer. This is practically important and beneficial for mass production technology.
4.3. P-type GaN layer fabrication
In order to fabricate the p-type GaN layer, Nakamura investigated another more practical method than that with electron irradiation studied by Akasaki and Amano. Heat treatment, using a specimen doped with magnesium as the acceptor impurity, had been studied but had proven unsuccessful. Nakamura reexamined the heat treatment effect and found that a hydrogen-free atmosphere was essential to activate the doped magnesium and get the p-type layer. He clarified the mechanism of p-type formation, proposing a model based on the connection between hydrogen atoms and acceptor impurities13).
Using this method, in 1992, the specific resistance of heat-treated GaN thin film was improved to 2 Ohm-cm - smaller than usual by more than a factor of 100,000 - and the superior p-type layer was successfully obtained where the hole mobility by Hall measurement was 10 cm2/Vs14).
4.4. Development of blue LED
Using the above-mentioned technologies, Nakamura fabricated p-n homo-junction blue LEDs with 0.18 % light emitting efficiency and the double heterostructure type with 0.22 % light emitting efficiency in 1992. The latter one was improved to 2.7 % light emitting efficiency using the above-mentioned InGaN thin film as the light emitting layer, in 199315). Based on these results, Nichia Chemical Industries delivered the world's first blue LEDs in 1993. Moreover, the quantum-well structure was adopted to develop the high brightness blue LED with 9.2 % light emitting efficiency16).
Nakamura investigated the InGaN light emitting layer in detail. Before that time, to achieve a long light emission lifespan, the maximum allowable number of crystal defects was considered to be under 103cm-2. Although there were many crystal defects - about 1010cm-2 - in the InGaN film of LED, the lifespan of this LED was longer than 100,000 hours. The reason for this long lifespan with that level of defects is as follows. The distribution of indium atoms doped in GaN film is not uniform but fluctuates slightly, and consequently the potential for electrons in the crystal is not uniform but varies locally. Injected electrons are trapped in this localized potential and recombined with holes to emit the light without being captured by the crystal defect region. This kind of localized potential was a thoroughly new phenomenon not found in any previous crystal material and found first in this ternary mixed crystal InGaN. The research project then started intentionally to design and to make this kind of localization in the crystal.
4.5. Development of blue LD
Based on these technologies for GaN-based blue LEDs, Nakamura succeeded in achieving high power pulse oscillation of the GaN-based blue LD with an InGaN multi-quantum well structure and clad layers. This was reported in the January 1996 issue of Japanese Journal of Applied Physics17). This paper attracted remarkable attention from researchers all over the world, and the number of resulting citations was nearly 160 by 1997.
The key step for achieveing this LD oscillation was introducing InGaN twenty cycles multi-quantum well structure for the light emitting layer and AlGaN film for blocking layer18).
In the case of LDs, compared to LEDs, the number of defects had to be decreased still more to obtain a long lifespan, since the number of injected electrons in LDs is larger. Nakamura adopted the Epitaxial Lateral Overgrowth (ELOG) method19). In 1998, he succeeded in decreasing the defects from 1010cm-2 to 107cm-2 and achieved 290 hours of continuous oscillation at room temperature, from which the oscillating lifespan was estimated to be 10 thousand hours. In 1999, Nichia Chemical Industries delivered the world's first blue LD20).
|
|
|
