前言:

LED屬於單點發光的點光源,因為封裝關係,光源的角度都朝前, 不像傳統燈具的光是四面八方投射,LED的發光效率測試沒採取傳統燈具的積分球去累計光流明,而是用自己的方法,這兩種測試方法無法直接對比,我統計分析後,覺得應該要除以2到3

日亞化工的LED自稱是249 lm/W,但是測試條件是放20mA/mm,如果用標準的350mA/mm則降到145 lm/W,略高於Bridgelux,但是日亞用的是藍寶石基板

TSMC成立的風險投資公司有投資Bridgelux,我估計應該還小於5%,所以TSMC要使用Bridgelux的技術來生產LED的話,可能障礙不小代價很高

主文:

藍光LED源於美國沒成功,日本日亞化工解決磊晶與藍寶石基板插排及P GaN 問題而大發利市,美國CREE使用SiC基板長出品質更好的LED用於Power LED,台灣夾縫求生辛苦了十多年,長一些相對較低價的藍光LED,至今各LED廠機台總數早已經破千台,但是EPS其實都不高,這產業很奇怪,掌握通路的封裝廠的獲利遠高於原料磊晶廠,封裝不論在技術成次或者設備資本都遠低於磊晶廠,卻吃走大部分利潤

白光LED其實是用藍光LED再塗上稀疏的螢光粉,螢光粉吸收部分的藍光放出黃光,再和其他的藍光混和變成白光

現今量產的 白光LED的發光效率,如果不是吹牛,應該落在30-60 lm/W,省電燈泡發光效率落在50-60 lm/W,長燈管日光燈落在70-95 lm/W,效率最高但是演色性不好的高壓鈉燈落在100-120 lm/W

Bridgelux不但解決了矽晶圓的插排問題(比藍寶石大很多),而且發光效率也直追傳統的藍寶石基板,更何況現在製作power LED需要用雷射去掉藍寶石基板,而矽基板只要傳統低成本的晶背研磨與蝕刻,甚至可將控制器直接整合,這技術確實偉大,而且降低成本到原來的25%,這根本是秋風掃落葉,沒得比,新技術一定是新機台,國內的1000多台磊晶設備等著被淘汰(當然至少撐2年,等Bridgelux進入量產就沒救了)

2003年我就是看破老闆的經營理念以及博士的政治鬥爭,才瀟灑的離開這個行業(這家公司還是我救的), 回家經營自己的part time公司,國內多數的經營者,其實沒比我老婆好多少,我會一年花個幾十到上百萬去探索非常艱難的未來性產品,每次都要跟老婆嘔氣爭取,她認為出國去玩花再多都值得,花在研發是浪費錢,哀,暑假又要留我顧公司了

國內的所謂研發其實都是在做產品Push to the limit來提升品質降低成本,開創性的研發幾乎是零,所以台灣產業的困境是,很難找到可以投資的項目,沒辦法,只想做代工的就只能繼續打工

Bridgelux Demonstrates 135 lm/W GaN-on-Silicon LED
LIGHTimes Staff

March 7, 2011…Bridgelux Inc. of Livermore, California USA, announced today its first major lab-based development, a demonstration of 135 Lumens per Watt GaN-on-silicon-based LED Technology. GaN-on-silicon LEDs, widely seen as the holy grail of LED production efficiency, have been an elusive goal for the many companies that have pursued them. Bridgelux says that its demonstration represents the industry’s first commercial grade performance for a silicon substrate-based LED. The LED had a 350mA operating current and it required just 2.9 volts and under 3.25 volts at 1 amp. The 4730K CCT LED measures 1.5mm by 1.5mm.

The LED was produced on a standard 8-inch silicon wafer. Bridgelux contends that growing GaN on larger, low-cost silicon wafers that are compatible with modern semiconductor manufacturing can deliver a 75% improvement in cost over current approaches, which commonly use 2- to 4-inch sapphire wafers. Bridgelux points out that when scaling up production, large diameter sapphire and silicon carbide substrates are costly, difficult to process, and sometimes not widely available. This has kept production costs high, inhibiting widespread adoption of LED lighting in homes and commercial buildings.

Additionally, the Bridgelux devices are demonstrating a low forward voltage and superior thermal resistance which make them ideally suited for high-performance, illumination-grade applications. Optimization of the epitaxy process on 8-inch Si wafers will make LED manufacturing more compatible with existing automated semiconductor lines. Bridgelux notes however that 『A number of the processes [used to produce conventional LEDs] … would need to be modified to eliminate some materials that are currently used, and which are incompatible with silicon substrate manufacturing.』

Bridgelux sees the move to silicon substrates as a revolutionary step for the LED industry. Over the past 5 years, Bridgelux CTO Dr. Steve Lester has shepherded a quietly-dedicated GaN-on-Silicon R&D team. Concurrently, industry-wide research and development of GaN growth on silicon has increased, both for production electronic devices and in the labs for optical technology. Bridgelux says that its GaN on silicon performance levels today are comparable to state-of-the-art sapphire-based LEDs available 12-24 months ago. Over the course of the next two to three years, the company anticipates the delivery of its first commercially viable GaN-on-Silicon products that meet performance requirements of LED lighting. “Bridgelux’s achievement is a significant reflection of the strength of our leadership in Silicon materials and epitaxial process technology,』 said Bill Watkins, Bridgelux CEO. “The significantly reduced cost-structures enabled by Silicon-based LED technology will continue to deliver dramatic reductions in the up-front capital investment required for solid state lighting. In as little as two to three years, even the most price-sensitive markets, such as commercial and office lighting, residential applications, and retrofit lamps will seamlessly and rapidly convert to solid state lighting.”