求中英文对照的文章-GPS 或 航位推算 或 地图匹配

最新问答

• 小精灵926

  三篇论文造就的未来科技
  - 相对论不只是抽象的理论，还可以被用来微晶片。工程师从爱因斯坦的想法中，持续做出各种创新的工具。
  
  作者╱吉布斯 ( W. Wayt Gibbs )
  译者╱林世昀
  
  ATOMIC SPIN-OFFS FOR THE 21ST CENTURY
  - A new generation of technologies aims to put Einstein's theories to work in computers, hospitals—even submarines
  
  By W. Wayt Gibbs
  
  1905年，爱因斯坦26岁，正为了完成他探讨分子大小的博士论文而奋斗。为了维持家计，他在瑞士专利局工作，专门分析其他人的发明。或许你会想，由於日常工作的启发，爱因斯坦应该会想办法把余暇时所发展的理论，做成实际的应用。可惜，在他发表他那非凡的研究生涯中最著名的五篇论文那一年，几乎看不到他有这种倾向。不过他对物质、能量和时间所主张的新看法，终究还是启发了新型机器的发明，促进了人类的和医疗技术。
  其实爱因斯坦也不是蔑视工程学，只是工程并非他的强项而已：他自己的发明，包括不含机械式运动零件的冰箱，以及防漏帮浦，从来没有进入量产阶段。不过没有关系，经过整个20世纪，其他人基於爱因斯坦的性想法，也建立了许多令人印象深刻的技术，其中最有名的想法，就是光以波包的形式传播，所有的光子总是遵守同一个速限c，而且能量和物质可以相互转换，以数学语言来表示，就是E = mc2（见36页〈天天遇见爱因斯坦〉）。
  到了21世纪，工程师开始用新的方式来开发那些著名原理的用途，其中或以完全创新的电脑最值得一提。他们也在为一些爱因斯坦较不为人知的理论寻找实际应用。比如说，奈米技术正在一种装置，这个装置利用分子的随机运动，来加速DNA的分析；而分子随机运动的正确解释，就是1905年由爱因斯坦首先提出的。世界上还有许多实验室，正在创因斯坦於1925年的经典思考实验中所预见的各种物质特殊状态。这些同调的极冷原子聚，是类似雷射光的物质，可以用在可携式原子钟、航行用的超精确陀螺仪，以及描绘矿脉和油田的重力感测器上。
  本文将检视三种由研究实验室中脱颖而出、最新也最令人兴奋的爱因斯坦副产品；当然更多这类的创新发明，在未来的几年到几十年间，也将源源不绝而来。虽然距离这位物理大师著手发展更好的数学工具来描述宇宙的那天，已经将近一个世纪了，聪明的发明家运用爱因斯坦的理论来巧妙的装置，却还方兴未艾。
  In 1905 Albert Einstein was 26 and struggling to finish his doctoral dissertation on the size of molecules. To pay the bills, he worked at the Swiss patent office, analyzing the inventions of others. You would think his day job would have inspired Einstein to contemplate practical uses for the theories he was developing in his spare time. Yet he showed little inkling that year, as he published five of the most remarkable papers of his extraordinary career, that the new views of matter, energy and time he was urging would eventually inspire novel kinds of machines to advance human industry and health.
  It isn't that Einstein disdained engineering. It just wasn't his strong suit: his own inventions, including a refrigerator with no mechanical moving parts and a leak-proof pump, never advanced to mass production. No matter; over the course of the 20th century, others built an impressive range of technologies [see “Everyday Einstein,” by Philip Yam, on page 50] on Einstein's radical notions that light comes in individual packets, that those photons always obey a universal speed limit c, and that energy and matter can be interconverted: E = mc2, in mathematical shorthand.
  In the 21st century, engineers have begun to exploit those famous principles in new ways, perhaps most notably in designs for radically innovative computers. They are also finding practical applications for some of Einstein's lesser-known theories. Nanotechnologists, for example, are making devices that could speed up DNA analysis by harnessing the random motion of molecules, a phenomenon first correctly explained by Einstein in 1905. And laboratories around the world are creating exotic forms of matter that Einstein envisioned in 1925 in one of his classic “thought experiments.” These coherent swarms of ultracold atoms—the matter cousins to laser beams—could find use in portable atomic clocks, superprecise gyroscopes for navigation, and gravity sensors for mapping mineral lodes and oil fields.
  This article examines three of the newest and most exciting Einsteinian spin-offs emerging from research labs; more such innovations will certainly follow in the years and decades to come. Although nearly a century has passed since the master physicist began fashioning ter mathematical tools to describe the universe, there seems no end to the useful gadgets that clever inventors can make with them.
  
  相对论与自旋 Taking Relativity for a Spin
  1905年，爱因斯坦研究狭义相对论时所用的唯一一部计算机，就是装在他脑袋里的那部。在许多方面，那部生化机器要比任何计算机厉害得多了。当然，至今还没有任何半导体微处理器，可以和人脑的密度与能源效率相提并论：人脑大约有一公斤重，其中有1000兆个处理单元，可是使用的功率和产生的热量，却比Pentium 4微处理器还要小。
  的确，对半导体来说，在追求以同样的单位成本更高效能微晶片的路途上，热与能源消耗是当今最难以克服的障碍。在未来的20年内，我们熟知的以矽为原料的数位微处理器，将会碰到经济与物理的根本极限。晶片商除了转而利用不同物理原理的设计，如狭义相对论之外，也许没有什麼别的选择了。
  表面上，这似乎是个古怪的结合。通常我们只会把狭义相对论和高速运动联想在一起。在这个理论中，爱因斯坦抛弃了绝对时间和绝对静止的观念。他断言，唯一的常数是c，也就是光在穿越真空时所拥有的速度。这个定律，为任何高速运动（相对於观察者）的物体带来了奇怪的结果。比如说，该物体的长度会变短，而它所感受到的时间似乎要比观察者的慢。如果物体穿越静电场，它会觉得有一部份的场是磁场。话说回来，这些所谓的相对论效应都很微小，除非物体的速度和c相比很显著，而c大概是每秒三亿公尺。
  在这个标准之下，即使是「行动」电脑也不算动得很快。不过电脑里的却够快。今年稍早美国加州圣巴巴拉分校由奥沙隆（David D. Awschalom）的研究，展示了一种利用相对论的方式，他们让半导体中快速运动的，表演出令人印象深刻的新把戏。
  这项研究还处於早期阶段，大概类似40几年前造出第一个半导体逻辑闸时的状况。不过假如工程师有办法把几百万个相对论性逻辑闸整合在一小块矽晶片上，其成果可能就是执行速度比当今机种快很多，而功率消耗与热辐射却少得多的微处理器；奥沙隆目前正在和英特尔与惠普的研究合作研发这种晶片。
  更引人注目的是，相对论性晶片可运用比目前所有电脑用的二进位运算更复杂的逻辑。原则上，这些新型机器甚至可以自己调整它们的布线方式，而且几乎在瞬间就可变成专为手中工作所量身订做的电路。举例来说，想像这种行动电话吧，它能重新配置它的无线电收发器，来使用世界上各种网络，而且只要按一个钮，就能重新设定它的微处理器，把一种语言翻译成另外一种。
  诸如此类的晶片，在现有的微处理器工厂中的可能性非常高。因为秘方并不是新材料，而是近代物理：由相对论和量子力学所描述的行为。
  THE ONLY COMPUTER that Einstein used to work out his special theory of relativity in 1905 was the one inside his skull. In many ways, that biochemical machine was far more capable than any electronic computer. Certainly no semiconductor microprocessor yet built can rival the density and energy efficiency of the human brain, which packs roughly a million billion processing elements into a one-kilogram package that uses less power and generates less heat than a Pentium 4 microprocessor.
  Indeed, heat and energy consumption today stand as the most formidable obstacles to the semiconductor industry as it seeks to produce ever more powerful microchips at the same unit cost. Within the next 20 years, the advance of digital silicon processors as we know them will hit fundamental economic and physical limits. Chipmakers will have little choice but to move to designs that exploit different principles of physics—those of special relativity, for example.
  On its face, that seems an odd combination. Special relativity is all about high-velocity motion. In the theory, Einstein discards the concepts of absolute time and absolute rest. The only constant, he asserts, is c, the speed at which light travels through empty space. That law has strange consequences for any object as it accelerates (relative to the observer). The object's length shortens, for example, and it seems to experience time more slowly than the observer does. If the object moves through a static electric field, it perceives the field as partially magnetic. These so-called relativistic effects are all minuscule, however, unless the object accelerates to a significant fraction of c, which is about 300 million meters per second.
  Even “mobile” computers don't move very fast by that standard. But the electrons inside them do. And earlier this year a group of physicists led by David D. Awschalom of the University of California at Santa Barbara demonstrated a way to exploit relativity to make the fast-moving electrons in semiconductors perform impressive new tricks.
  The work is at an early stage, roughly analogous to the construction of the first semiconductor logic gate some 40 years ago. But if engineers can figure out how to integrate millions of relativistic gates on a small silicon chip—and Awschalom is working with research groups at Intel and Hewlett-Packard to do just that—the result could be processors that run much faster than current models do, while consuming far less power and radiating far less heat.
  Even more dramatically, relativistic chips could employ logic that is more sophisticated than the binary operations all computers now use. In principle, these new machines could even modify the way they are wired, adapting almost instantaneously into a circuit customized for the task at hand. Imagine a cell phone, for example, that can reconfigure its transceiver to use any network in the world and that at the push of a button can reprogram its processor to translate speech from one language to another.
  Chips such as these could most likely be made in existing microprocessor factories. The secret ingredient is not some new material, but modern physics—behaviors described by the theories of relativity and quantum mechanics.
  
  磁的吸引力 The Magnetic Attraction
  一般传统的半导体微晶片运作的基础，是19世纪的「古典」电磁理论。矽晶圆用离子轰击，而在其上形成微小的岛，各自具有过量或是不足的。在这些岛之间布置的微电极上加电压，就能推拉进出这些区域、开关逻辑闸，并且调控经过其间的电流。
  把大量的邻近撞开，是很不精确的：有些会凌乱地弹开，而浪费能量；同时也会产生许多碰撞而发热。10多年来，已有物理学家实验过另一类更精密的方式：以磁力代替电场来纵。
  美国爱荷华的物理学家弗拉提（Micheal E. Flatté）解释道，这个点子行得通，是因为「和外界的交互作用，就好像它随身带著一根小小的磁铁棒一样。」磁铁有S极N极。而就和地球绕著连接两极的轴自转一样，每颗也都具有磁指向，就是物理学家称为「自旋」的量子性质。这些粒子并非真的在旋转，不过它们的表现确实很像是个小陀螺仪。把磁力加在上，的两极会开始进动——它的转轴本身会绕圈圈。把磁场拿掉，的自旋就定住了（见左页〈磁的魔术〉）。弗拉提说：「利用这种效应，把自旋从指向往上的状态进动到往下，你就可以把所带的资讯位元从1变成0。」
  学以改变数量与能量的方式，在电路中移动资讯，而刚萌芽的自旋学（spintronics）则将编在自旋的指向中，并且用各种扭转自旋的方式来做逻辑运算（2002年9月号〈前途无量的自旋〉）。摩托罗拉从今年开始量产一种自旋记忆晶片，叫做MRAM（magnetic RAM，磁性随机存取记忆体）。和一般的电脑记忆体不同，MRAM晶片在电源中断时不会损失；电源再度打开前，的自旋会一直保持在它的指向上。
  自旋元件很容易用电池驱动，因为自旋反转作所消耗的能量极其微小，而且晶片在两次运算之间可以停止电源。改变一个的自旋实际上并不会增加粒子的动能，因此电路几乎不会发热。而且整个程序进行得极快：实验用的装置，只消在几皮秒（10-12秒）以内，就能让转头。
  不过直到最近，所有的自旋元件都得用铁磁金属才行，这和当前微晶片的技术并不协调。奥沙隆说：「很难想像你怎麼能在晶片上的几百万个位置添造小磁铁，而且还能各自地控制它们；不是不可能啦，只是很困难。运用现有价值几兆元的闸极技术会比较好，也就是用电场而非磁场，来纵自旋。」
  CONVENTIONAL SEMICONDUCTOR microchips operate based on “classical” 19th-century theories of electromagnetism. Silicon wafers are zapped with ions, which form tiny islands with either an excess or a dearth of electrons. Voltages, applied to microscopic electrodes built up around these islands, push and pull electrons in and out of these regions, opening and closing the logic gates and regulating the flow of electric current through them.
  Shoving large numbers of electrons around is imprecise—some shoot out in random directions, wasting energy—and it creates lots of collisions, which produce heat. For more than a decade now, physicists have been experimenting with a subtler alternative: using magnetic forces, rather than electric fields, to manipulate the electrons.
  This can work, explains physicist Michael E. Flatté of the University of Iowa, because “an electron acts as if it carries around with it a little bar magnet.” Magnets have north and south poles, and just as the earth spins around the axis that connects its poles, an electron, too, has a magnetic orientation, a quantum property that physicists call “spin.” The particles don't actually rotate, but they do behave like little gyroscopes. Apply magnetic force to an electron, and its poles will precess—the axis itself rotates in a circle. Remove the field, and the electron holds its spin steady [see box on opposite page]. “By using this effect to precess the spin from pointing up to pointing down, you can change the bit of information carried by that electron from a 1 to a 0,” Flatté says.
  Whereas electronics move information around by changing the number and energy of electrons in a circuit, the nascent field of spintronics encodes data in the orientation of electrons and performs logical operations by twisting their spins this way and that [see “Spintronics,” by David D. Awschalom, Michael E. Flatté and Nitin Samarth; Scientific American, June 2002]. This year Motorola began mass-producing spintronic memory chips, called MRAM (for magnetic RAM). Unlike conventional computer memories, the MRAM chips do not lose their data if power is interrupted; the electron spins simply hold their position until power returns.
  Spintronic devices are easy on batteries, because spin-flipping operations consume very little power and the chips can shut off ween operations. Changing an electron's spin adds virtually no kinetic energy to the particle, so the circuits produce almost no heat. And the process is exceedingly fast: experimental devices have turned electrons on their heads in a few picoseconds (trillionths of a second).
  Until recently, however, all spintronic devices have required ferromagnetic metals, which do not mesh well with current microchip production techniques. “It's difficult to imagine how you could build little magnets at millions of places on a chip and control each one individually—not impossible, but difficult,” Awschalom says. “It would be much nicer to use the trillions of dollars' worth of electronics gating technology that already exists and to use electric fields, not magnetic fields, to play with spins.”
  
  突破0与1的 From Bits to Phits
  现在要进入爱因斯坦与他的奇怪想法了：对一个高速运动的来说，部份电场看起来会变成截然不同的磁场。在今年1月发表的研究工作中，奥沙隆的研究就展示了，若将两层成份稍有不同的半导体叠起来，晶片的应变会造成一个内部的电场。当通过半导体时，这电场的高低分布就像围栏一样把驱赶在一起。他解释：「由於相对论的关系，由正在穿越的看来，这电场会有一部份像是磁场。」於是的自旋开始像摇晃的陀螺般进动。
  「我们可以用两种方式来控制。」奥沙隆继续说，「一种方式是改变电压，这会影响到穿越的速度。它们跑得越快，看到的有效磁场越大，」而自旋就进动得越快。第二种手段是利用应变在各方向上的不同性。他说：「我们也可以仔细地设计用来规范路径的线路形状和方向。」
  在1月的论文中，该研究描述了如何运用雷射光脉冲来排列入射的指向，以造出自旋位元，以及如何测量它们的自旋。「下一步是在同一个装置中造出它们，把它们四处移动，并且侦测到它们全部。那是相当重要的一步，而我们现在已经办到了，」奥沙隆说，「这个装置和目前电脑中的CMOS晶片使用一样的电压。当撞击到半导体应变的部份时，自旋会在瞬间极化。然后我们就可以同调地上下翻转它们的自旋。」这用的是开关闸极的方式。
  「同调」是这里的关键字，因为它提出了自旋晶片最有趣的可能性：超越只有0与1两个数字的位元，而达到相位元（phase digit, phit）的境界，而有更大范围的数值可取。的相位就是它自旋所指向的方向。把它想成罗盘的指针好了：假如微晶片可以分辨一自旋分别指向东、西、南、北方向的，那麼每个相位元就可以是0或1、或2或3。
  奥沙隆指出：「相位读得越精确，你就可以把储存的密度增加得越夸张。至於增加50还是一万倍，端看你测量那个角度有多精确。」感谢几十年来侦测原子核自旋的磁共振造影技术的发展，「我们确实知道怎麼把这些角度量得很准。」他补上一句。
  即使如此，弗拉提警告说：「一个完整可用的自旋电晶体尚未发展成功。」电晶体是不可或缺的，因为它能放大信号，使信号在微处理器中原封不动地穿过一长串逻辑闸。不过虽然依据自旋学设计的电晶体目前还不存在，却显然会在不久的将来诞生，研究者也正热切地计画著用他们来做些什麼事。
  去年，德国柏林德鲁得固态研究所的柯克（Reinhold Koch）与他的研究，发表了一个运用自旋逻辑元件的设计，它能在软体的控制下改变自己的功能。在某个时刻它可以是布林（Boolean）运算的AND闸，几奈秒（10-9秒）之后，它又可以转变成一个OR闸、NOR闸或NAND闸。
  能在一瞬间重新布线的电脑，的确威力强大。柯克最近设计了一个完整的加法器（电脑逻辑单元里最普通的一种），其中只用到了四个自旋逻辑元件，而非通常所需要的16个电晶体。自旋版的加法器可以节省85%的能源以及75%的空间，执行的速度却和当今最顶级的微晶片设计一样快。
  工程师距离能得心应手地运用相对论来当成自旋电路的设计工具，还差很远。不过在现有道路上充满障碍的情况下，爱因斯坦的理论却也能为电脑另辟蹊径。奥沙隆说：「对这里的物理有一项有趣的观点是，元件越小，工作得越好。」
  ENTER EINSTEIN and his curious notion that an electric field can look distinctly magnetic to a high-speed electron. In work published this past January, Awschalom's group showed that layering two semiconductors of slightly different composition on top of one another strains the chip in ways that set up an internal electric field. The field has high and low spots that act like a corral to herd electrons as they pass through the semiconductor. “And because of relativity, that electric field looks like a partially magnetic field to the passing electrons,” he notes. The electrons' spins start to precess like wobbly gyroscopes.
  “We can control the electrons in two ways,” Awschalom continues. “One way is to change the voltage, which affects the speed at which the electrons travel. The faster they move, the larger the effective magnetic field seems to them” and the faster their spins precess. The second trick exploits the fact that the strain varies with direction. “We can also operate on electrons by carefully designing the shape and direction of the wire that sets their path,” he says.
  In the January paper, the group described using pulses of laser light to align the orientation of incoming electrons—thus creating the spintronic bits—as well as to measure their spins. “The next step is to create them, move them around and detect them all in one electric device. That's a substantial step, but we've done that now,” Awschalom reports. “The device uses the same small voltages currently used in CMOS computer chips. Electrons instantaneously polarize their spins when they hit the strained part of the semiconductor. We can then flip their spins back and forth coherently” by turning gate electrodes on or off.
  “Coherently” is the key word here, because it raises the intriguing possibility of spintronic chips that go beyond bits—the binary digits 0 and 1—to “phits,” or phase digits, which can take on a wider range of values. The phase of an electron is simply the direction in which its spin points. Think of it as the needle of a compass: if a microchip can distinguish groups of electrons with north-, south-, east- and west-pointing spins, then each phit could be a 0 or a 1—or a 2 or a 3.
  “The more precisely you can read the phase, the more dramatically you can increase the density of data storage,” Awschalom points out. “Whether it increases by a factor of 50 or by 10,000 depends on how precisely you can read that angle.” Thanks to decades of wo

  浏览 424赞 145时间 2022-09-19