半導(dǎo)體(芯片）發(fā)展史

什么是半導(dǎo)體？

半導(dǎo)體是導(dǎo)電性介于導(dǎo)體和絕緣體中間的一類物質(zhì)。與導(dǎo)體和絕緣體相比，半導(dǎo)體材料的發(fā)現(xiàn)是最晚的，直到20世紀(jì)30年代，當(dāng)材料的提純技術(shù)改進(jìn)以后，半導(dǎo)體的存在才真正被學(xué)術(shù)界認(rèn)可。

半導(dǎo)體主要由四個(gè)組成部分組成：集成電路，光電器件，分立器件，傳感器，由于集成電路又占了器件80%以上的份額，因此通常將半導(dǎo)體和集成電路等價(jià)。集成電路按照產(chǎn)品種類又主要分為四大類：微處理器，存儲(chǔ)器，邏輯器件，模擬器件。通常我們統(tǒng)稱他們?yōu)樾酒?/p>

提起芯片，很多人可能見過，就是一塊黑色類似于小盒子的東西，它是由晶體管組成的。

什么是晶體管呢？

嚴(yán)格意義上講，晶體管泛指一切以半導(dǎo)體材料為基礎(chǔ)的單一元件，包括各種半導(dǎo)體材料制成的二極管、三極管、場效應(yīng)管、可控硅等。晶體管有時(shí)多指晶體三極管。

所以我們可以知道：由半導(dǎo)體材料制造出了晶體管，由晶體管組成了芯片。

2. 晶體管的誕生

晶體管的發(fā)明，最早可以追溯到1929年，當(dāng)時(shí)工程師利蓮費(fèi)爾德就已經(jīng)取得一種晶體管的專利。但是，限于當(dāng)時(shí)的技術(shù)水平，制造晶體管的材料達(dá)不到足夠的純度，而使其無法制造出來。

1947年12月，美國貝爾實(shí)驗(yàn)室的肖克利、巴丁和布拉頓組成的研究小組，研制出一種點(diǎn)接觸型的鍺晶體管。1956年，肖克利、巴丁、布拉頓三人，因發(fā)明晶體管同時(shí)榮獲諾貝爾物理學(xué)獎(jiǎng)。肖克利也被譽(yù)為晶體管之父。

芯片有如此強(qiáng)大的功能，為什么晶體管可以勝任呢？

我們知道，對于數(shù)字電路來講，邏輯是其精髓所在，所有的功能歸根結(jié)底，都可以說是邏輯功能。而邏輯的基本構(gòu)成元素是邏輯0和邏輯1。

而晶體管恰好具備這種功能--通過電信號(hào)來控制自身開合，以開關(guān)的斷開和閉合來代表0和1。

3. 歷史中的八卦

當(dāng)然，一味的講原理或者講歷史就沒意思了，我們來挖掘半導(dǎo)體發(fā)展過程中的八卦。

前面有提到過，被譽(yù)為“晶體管之父”的肖克利，出生于倫敦，三歲時(shí)隨父母漂洋過海來到加州。受父母對他科學(xué)思想的灌輸，考入MIT，隨后獲得固體物理學(xué)博士，留校任教。后來，被位于新澤西州的貝爾實(shí)驗(yàn)室副主任凱利來麻省“挖墻角”，將肖克利挖走了。當(dāng)晶體管發(fā)明成功后，肖克利并不滿足，他依然進(jìn)行著不斷地嘗試，希望發(fā)明性能更好的晶體管，并將其商品化。

與此同時(shí)，高純硅的工業(yè)提煉技術(shù)已成熟，用硅晶片生產(chǎn)的晶體管收音機(jī)也問世。在貝爾實(shí)驗(yàn)室工作的肖克利坐不住了，他看到了未來的商機(jī)，而現(xiàn)在只能看著貝爾實(shí)驗(yàn)室拿他的發(fā)明賺錢，并且晶體管的性能不穩(wěn)定，有損個(gè)人聲譽(yù)。

最后，矛盾爆發(fā)了!當(dāng)然最終還是因?yàn)槔妗?/p>

硅谷誕生

1955年，肖克利回到了自己的家鄉(xiāng)圣克拉拉（Santa Clara）谷，并得到了貝克曼的支持，創(chuàng)辦了自己的公司。

圣克拉拉位于舊金山灣區(qū)南部，與圣何塞（San Jose）[4] 、森尼韋爾（Sunnyvale）。地理位置優(yōu)越環(huán)境優(yōu)美；氣候清新宜人；交通便利。從此，這一片狹長的山谷舉世聞名。

肖克利在創(chuàng)辦了自己的公司后，依靠自身的威望，很快招到了一批學(xué)識(shí)淵博，技術(shù)過硬的人才。此時(shí)，我們仿佛看到一家半導(dǎo)體商業(yè)巨頭正要崛起，屹立于世界之巔，但是，總有意外發(fā)生。

硅谷八叛將

肖克利雖然是一個(gè)聰明絕頂?shù)奶觳?，但卻不是一個(gè)好的管理者。在公司發(fā)展方向上，幾乎由他一人掌控，專橫獨(dú)裁，而且他不知道自身的缺陷，也不接受同事的合理化建議，最終導(dǎo)致公司在很長一段時(shí)間里都沒有產(chǎn)品做出來。在同事關(guān)系上，他忽略了最重要的兩點(diǎn)--尊重與信任，肖克利通過各種辦法，牢牢的將技術(shù)專利掌握在自己的手中，這種自私自利的管理方式，終將會(huì)讓公司走向沒落。

1957年9月18日，以諾伊斯為首的八位年輕人憤然提出離職，肖克利得知后勃然大怒，罵其曰“八叛逆”（Traitorous Eight）。

硅谷八叛將分別為：諾依斯（N. Noyce）、摩爾（R.Moore）、布蘭克（J.Blank）、克萊爾（E.Kliner）、赫爾尼（J.Hoerni）、拉斯特（J.Last）、羅伯茨（S.Boberts）和格里尼克（V.Grinich）

傳奇的仙童

很快，這八個(gè)人就拿到了一筆風(fēng)險(xiǎn)投資，投資人是具有遠(yuǎn)見卓識(shí)的謝爾曼 菲爾柴爾德，成立了仙童半導(dǎo)體（Fairchild），公司便是以投資人命名。

公司由謝爾曼 菲爾柴爾德控股，管理人為諾伊斯。在新型的管理模式下，仙童半導(dǎo)體快速發(fā)展，不到半年的時(shí)間里就已經(jīng)開始盈利。

與此同時(shí)，仙童的兩項(xiàng)發(fā)名專利，更使其立于世界的半導(dǎo)體之巔。其一是平面工藝--一種制造半導(dǎo)體電路的工藝方法，發(fā)明人為約翰·霍爾尼（Jean Hoerni）。

另一個(gè)發(fā)明專利便是集成電路。顧名思義，集成電路就是用一定的工藝，把一個(gè)電路中所需的晶體管、電阻、電容和電感等元件及布線互連一起，制作在一小塊或幾小塊半導(dǎo)體晶片或介質(zhì)基片上。

1958-1959年，來自仙童的羅伯特·諾伊斯（Robert Noyce）發(fā)明了硅集成電路。事實(shí)上，在早些時(shí)候，來自德州儀器的杰克·基爾比（Jack Kilby）發(fā)明了鍺集成電路。由于兩人在同一年獨(dú)立且不知情的情況下分別發(fā)明了集成電路，所以兩人共享集成電路發(fā)明者的榮譽(yù)。

現(xiàn)在，在我們眼里看來，把多個(gè)電路集成到一起而減少面積是個(gè)自然而然的事情，然后這個(gè)簡單的想法，卻改變了我們的世界。很多偉大的發(fā)明，往往源自一個(gè)很簡單的想法。也許，即使沒有這兩位，依然有人會(huì)想到這個(gè)點(diǎn)子，但是歷史只會(huì)記住最先吃螃蟹的人。

此時(shí)的仙童半導(dǎo)體公司風(fēng)光無限，而半導(dǎo)體行業(yè)在那時(shí)宛若一個(gè)巨大的金礦，任憑仙童肆意挖掘。而仙童的股權(quán)大部分都在投資人謝爾曼菲爾柴爾德的手里，與此同時(shí)，仙童半導(dǎo)體公司的利潤被不斷轉(zhuǎn)移到東海岸，去支持Fairchild攝影器材公司，此時(shí)仙童的員工開始坐不住了，開始了新一輪的離職創(chuàng)業(yè)潮。

4. 花開遍地

1968年，諾依斯（N. Noyce）和摩爾（R.Moore）從仙童離職后創(chuàng)辦了我們所熟知的英特爾（Intel），這里的摩爾就是我們所熟知的摩爾定律的提出者。

1969年，杰里·桑德斯（J. Sanders）當(dāng)時(shí)在仙童擔(dān)任銷售部的主任，帶著7位仙童員工創(chuàng)辦AMD。

還有許多我們所熟知的公司，比如美國國家半導(dǎo)體（現(xiàn)已被TI收購），Altera（現(xiàn)已被英特爾收購）等的創(chuàng)始人都出自仙童半導(dǎo)體公司。

正如江湖流傳的蘋果公司喬布斯形象比喻的那樣：“仙童半導(dǎo)體公司就像個(gè)成熟了的蒲公英，你一吹它，這種創(chuàng)業(yè)精神的種子就隨風(fēng)四處飄揚(yáng)了?！?/p>

隨后，英特爾， TI，三星等巨頭開始在世界的舞臺(tái)大放異彩！

在處理器（CPU）領(lǐng)域，英特爾的發(fā)展史代表了處理器的發(fā)展史。1971年，英特爾推出了它的第一款處理器：4004，這是一款4位的處理器，僅包含2300個(gè)晶體管，現(xiàn)在來看，這款處理器簡直就是個(gè)小弱，但它的誕生意義重大，實(shí)現(xiàn)了從0到1的突破。1978年，英特爾推出了一款16位的處理器:i8086。1979年，英特爾又推出了8088,8088是第一個(gè)成功應(yīng)用于個(gè)人電腦的CPU。1982到1989年，期間又陸續(xù)推出了80286，80386，80486微處理器。1993年奔騰處理器橫空出世，2005年酷睿走進(jìn)大眾的視野，酷睿i3，i5，i7成為PC的主流。

英特爾，AMD主要做PC,服務(wù)器的芯片。對于現(xiàn)在炙手可熱的智能手機(jī)，處理器的競爭則更為激烈。蘋果，三星，高通一直占領(lǐng)著高端機(jī)的市場，國內(nèi)近幾年半導(dǎo)體業(yè)強(qiáng)勢崛起，海思，展訊可謂國內(nèi)的代表。

5. 展望未來

縱觀過去的半個(gè)世紀(jì)，半導(dǎo)體的迅猛發(fā)展為我們的科技爆炸提供了基礎(chǔ)。當(dāng)10nm的芯片已經(jīng)商用，7nm,5nm制程已經(jīng)接近極限的情況下，摩爾定律似乎已經(jīng)開始走向終點(diǎn)，半導(dǎo)體下一個(gè)轉(zhuǎn)折點(diǎn)在哪里，我們還不能確定，但目前量子技術(shù)的發(fā)展似乎給我們指明了方向。

隨著量子通信，量子糾纏等新鮮詞語開始出現(xiàn)在我們的視野中，量子芯片也橫空出世，與傳統(tǒng)芯片用0和1進(jìn)行運(yùn)算處理不同，量子芯片具備多個(gè)量子位，而不是再只有0和1兩種邏輯狀態(tài)，這樣使得芯片的運(yùn)算能力成指數(shù)增長，從而突破摩爾定律的限制。

回顧半導(dǎo)體發(fā)展的輝煌歷史，也在一定程度上代表了人類的文明史。如果說機(jī)械的發(fā)展解放了人類的勞動(dòng)力，那么半導(dǎo)體的發(fā)展則解放了人類的計(jì)算力。而且半導(dǎo)體的發(fā)展勢頭絕不會(huì)就此停歇，必將隨著科技的發(fā)展大放異彩，對我們每個(gè)人來講，未來的半導(dǎo)體，未來的世界，值得我們期待。

附：摩爾定律：集成電路芯片上所集成的晶體管數(shù)目，每隔18個(gè)月就翻一倍。History of semiconductor (chip)

What is a semiconductor?

Semiconductors are substances whoseelectrical conductivity is intermediate between conductors and insulators.Compared with conductors and insulators, the discovery of semiconductormaterials was the latest. It was not until the 1930s, when the purificationtechnology of materials was improved, that the existence of semiconductors wastruly recognized by the academic community.

Semiconductors are mainly composed of fourcomponents: integrated circuits, photoelectric devices, discrete devices andsensors. Since integrated circuits account for more than 80% of devices,semiconductors and integrated circuits are usually regarded as equivalent.Integrated circuits are divided into four main categories according to the typeof products: microprocessor, memory, logic devices, simulator parts. We usuallyrefer to them as chips.

If you think of a chip, many of you haveprobably seen it, it's a black piece of stuff that looks like a little box, andit's made of transistors.

What is a transistor?

Strictly speaking, transistor refers to allsemiconductor materials based on a single element, including a variety ofsemiconductor materials made of diode, transistor, field effect tube, siliconcontrolled, etc. Transistors sometimes refer to crystal triodes.

So we know that transistors are made ofsemiconductor materials, and transistors are made of chips.

2. The Birth of the transistor

The invention of the transistor can betraced back to 1929, when engineer Lillienfeld obtained a patent for atransistor. However, given the state of the art at the time, the materials usedto make transistors were not pure enough to make them.

In December 1947, a research team ofShockley, Bardeen and Bratton at Bell Laboratories developed a point-contactgermanium transistor. In 1956, Shockley, Bardeen and Bratton shared the NobelPrize in physics for their invention of the transistor. Shockley is also knownas the father of the transistor.

Chips have so much power, why can transistorsdo it?

As we know, logic is the essence of digitalcircuits, and all functions can be said to be logic functions after all. Thebasic building blocks of logic are logic 0 and logic 1.

Transistors do just that -- they turnthemselves on and off with electrical signals, turning on and off to representzeros and ones.

3. Gossip from history

Of course, blindly talking about principlesor history is not interesting, let's dig the gossip in the process ofsemiconductor development.

Shockley, known as the "father of thetransistor," was born in London and moved to California with his parentswhen he was three. Educated in science by his parents, he went to MIT, where heearned a doctorate in solid-state physics and stayed on the faculty. He waslater lured away by Kelly, deputy director of Bell Laboratories in New Jersey,who came to Massachusetts. When the transistor was invented, Shockley was notsatisfied. He kept trying to invent better transistors and commercialize them.

At the same time, the industrial refiningof high-purity silicon has matured, and transistor radios made from siliconwafers have been developed. Shockley couldn't sit still at Bell LABS. He saw afuture business opportunity. Now he could only watch Bell LABS make money fromhis invention, and the transistor's performance was unstable and damaging tohis reputation.

Finally, the conflict erupted! Ultimately,of course, it's about profit.

Born in silicon valley

In 1955, Shockley returned to his nativeSanta Clara Valley and enlisted Beckman's support to start his own company.

Santa Clara is located in the southern SanFrancisco Bay Area, along with San Jose [4] and Sunnyvale. Superiorgeographical location and beautiful environment; The climate is fresh andpleasant; The transportation is convenient. Since then, this narrow valley hasbeen known all over the world.

After starting his own company, Shockleyrelied on his prestige to quickly recruit a group of knowledgeable and skilledpeople. At this time, we seem to see a semiconductor business giant is about torise to the top of the world, but there are always accidents.

Silicon Valley's eight defectors

Shockley, though a brilliant genius, wasnot a good manager. In the direction of the company, he was almost in controlof himself, dictatorial, and he did not know his own shortcomings, and he didnot accept the reasonable suggestions of his colleagues, which eventually ledto the company without products for a long time. In terms of the relationshipbetween colleagues, he neglected the most important two points -- respect andtrust. Shockley managed to keep the technology patents firmly in his own handsby various means. This self-serving management style will eventually bring downthe company.

On Sept. 18, 1957, the Eight young men, ledby Noyce, angrily resigned, and Shockley was furious, calling them the"Traitorous Eight.

The Silicon Valley eight will be: N. Noyce,R.Moore, J.Blank, E.K. Liner, J.Hoerni, J.Laster, S.Roberts and V.Grinich.

The legendary fairy

Soon, the eight men raised venture capitalfrom the visionary Sherman Fairchild, who founded Fairchild, after whom thecompany was named.

The company is controlled by ShermanFairchild and managed by Noyes. Under the new management mode, FairchildSemiconductor developed rapidly and became profitable in less than half a year.

At the same time, Fairchild's two patentsput it at the top of the world's semiconductor industry. One is the planarprocess, a process for making semiconductor circuits invented by Jean Hoerni.

Another invention patented was theintegrated circuit. As the name implies, integrated circuit is to use a certainprocess, the transistor, resistance, capacitance and inductance and othercomponents needed in a circuit and wiring interconnection together, made in asmall piece or a few small pieces of semiconductor chip or dielectric substrate.

In 1958-1959, Robert Noyce from Fairchildinvented the silicon integrated circuit. In fact, in the early days, Jack Kilbyfrom Texas Instruments invented germanium integrated circuits. Both men sharethe credit for inventing integrated circuits, since they independently andunknowingly invented them in the same year.

Now, we think of it as natural to integratemultiple circuits to reduce the area, and that simple idea has changed ourworld. Many great inventions come from a very simple idea. Maybe someone wouldhave thought of it without these two, but history will only remember the first.

Fairchild was in the wind, and thesemiconductor industry was like a huge gold mine for Fairchild to dig.Fairchild's equity was largely in the hands of investor Sherman Fairchild, andas profits from Fairchild Semiconductor continued to be diverted to the EastCoast to support Fairchild Photographic Equipment, Fairchild's employees beganto lose their seats and a new wave of new businesses began to leave.

4. Flowers are everywhere

In 1968, N. Noyce and R.Moore, the man we knowas Moore's Law, left Fairchild to start the company we know as Intel.

AMD was founded in 1969 by Jerry Sanders,then director of sales at Fairchild, with seven Fairchild employees.

Many of our most familiar companies, suchas National Semiconductor (now acquired by TI) and Altera (now acquired byIntel), were founded by Fairchild.

As Apple's Steve Jobs famously put it:"Fairchild is like a ripe dandelion. When you blow on it, the seeds ofentrepreneurship flutter everywhere."

Subsequently, Intel, TI, Samsung and othergiants began to shine on the world stage!

In the world of processors (cpus), Intel'shistory represents the history of processors. In 1971, Intel introduced itsfirst processor, the 4004. It was a 4-bit processor that contained only 2,300transistors. Today, the processor was a mere twit, but it was a significantbreakthrough from zero to one. In 1978, Intel introduced a 16-bit processor:the I8086. In 1979, Intel introduced the 8088, the first CPU to be successfullyused in personal computers. From 1982 to 1989, the 80286,80386,80486microprocessors were successively introduced. In 1993, the Pentium processorwas born, and in 2005, Core came into the public's vision. Core I3, I5, and I7became the mainstream of PC.

Intel and AMD make chips for PCS andservers. There is even more competition for processors in today's hotsmartphones. Apple, Samsung and Qualcomm have been occupying the high-endcomputer market. In recent years, the domestic semiconductor industry hasemerged strongly. Hesis and Spreadtrum are the representatives of China.

5. Look to the future

Throughout the past half century, the rapiddevelopment of semiconductors has provided the basis for our technologicalexplosion. With 10nm chips already commercially available and 7nm and 5nmprocesses approaching their limits, Moore's Law seems to be coming to an end.We are not sure where the next turning point will be for semiconductors, butthe development of quantum technology seems to point us in the right direction.

As quantum communication, quantumentanglement and other new words began to appear in our field of vision,quantum chip also was born, and the traditional chip processing using 0 and 1,quantum chip with multiple qubits, rather than only two logic state of 0 s and1 s, which makes chips into exponential growth of computing power, so as tobreakthrough the limitation of Moore's law.

Looking back on the glorious history ofsemiconductor development, it also represents the history of human civilizationto a certain extent. If the development of machinery has liberated human labor,the development of semiconductors has liberated human computing power. And thedevelopment momentum of semiconductor will never stop here, will shine with thedevelopment of science and technology, for each of us, the future ofsemiconductor, the future of the world, we are worth looking forward to.

P.S. Moore's Law: The number of transistorson an integrated circuit chip doubles every 18 months