Antarctic Dreams: A Literary Essay About AMANDA - Francis Halzen

Francis Halzen, laureate of the Nobel Prize in Physics in 2026.

Francis Halzen is a particle physicist at the University of Wisconsin–Madison and one of the founders of Antarctic neutrino astronomy. In Antarctic Dreams, a literary essay about the AMANDA project, he writes as both scientist and project leader, tracing the idea of using Antarctic ice as a neutrino telescope and recounting the technical, logistical, and human difficulties behind AMANDA's first detection.

First published in The Sciences in 1999 and later collected in The Best American Science Writing 2000, the essay captures the moment when a decade of theory, drilling, and engineering finally produced a high-energy neutrino event.

The English text below is followed by a Chinese full-text translation.

English Full Text

Antarctic Dreams

By Francis Halzen

A literary essay about AMANDA. Originally published in The Sciences, Vol. 39, No. 2 (1999), page 19. Archived by the Exploratorium.

Big science, as often as not, hinges on small moments. Once the grants have been secured and the politics navigated, the ground broken and the visionary promises made, when banks of computers flicker to life and fingers curl above keyboards, ready to flash the first discoveries via E-mail, a decade’s work can still come crashing down in the final hour—a castle erected in the thin air of theory, too weak to withstand true gravity.

Last August, when Dennis S. Peacock came to visit my office at the University of Wisconsin-Madison, he must have had something like that in the back of his mind. As the head of the Antarctic sciences section for the National Science Foundation (NSF), Peacock had helped funnel some $10 million into building the world’s largest neutrino telescope deep in the Antarctic ice. Would we have anything to show for the investment? Or would our project go the way of its predecessor, partly built in seawater off the coast of Hawai’i and then abandoned after nearly twenty years of work?

Our telescope—also known as the Antarctic Muon and Neutrino Detector Array, or AMANDA—had been detecting neutrinos for some time, but we had yet to finish analyzing our first data. Or so I thought. Instead, when Peacock and I arrived at the desk of my graduate student, Rellen R. Hardtke, she had a surprise waiting for me. Two of her colleagues had spent the night finishing the analysis, she explained with a smile. Then she calmly called up an image on her computer screen: the first precision map of a high-energy neutrino event ever recorded.

Hardtke’s screen showed a faint blue line streaking diagonally across columns of black dots. Most of the dots, each of which represented a photomultiplier sunk in the ice, were small and black. But a few, clustered along the line, were blue or green or red, and two, near the beginning of the line, were bright orange and very large. At five in the morning on October 12, 1997, the diagram told us, a neutrino—one of nature’s smallest and most elusive elementary particles—had entered the earth in the middle of the Pacific Ocean, between Midway Island and the Aleutians, hurtled straight through the planet, and collided head-on with a proton on the underside of the Antarctic ice. Two kilometers beneath the surface, our grid of photomultipliers had picked up a subatomic spark from that collision as it flew upward through the ice and flared past them for about a microsecond.

The end result, abstracted on a computer screen, would have seemed unremarkable to most people. But to us it was the sole trace of a particle that had traversed vast distances to reach us, perhaps a remnant of one of the most spectacular events in the universe. More to the point, it was the first concrete proof that AMANDA worked—that it could help map the distant depths of space, thereby perhaps resolving some of the most heated controversies in physics. Later, when I E-mailed the diagram to our collaborators, one of them wrote back: "This is why I’ve spent five years of my life on this project."

"NEUTRINOS, THEY ARE VERY SMALL," John Updike wrote, famously. "They have no charge and have no mass / And do not interact at all." As it turns out, Updike was wrong on two counts, but he got the spirit right, anyway. Neutrinos are so small and slippery that they pass through the Earth (and stars and cities and most everything else) like a bullet through a rainstorm. Unfazed by magnetic fields or the strong nuclear force, they have to make a direct hit on a proton to be stopped at all—a highly unlikely event. At the same time, neutrinos are about as plentiful in the universe as the photons that constitute light: 3 × 10^16 of them pass through our bodies every second.

That combination of factors makes neutrinos the best focus for deep-space astronomy. Most photons cannot reach us from the most distant points in the universe (the most energetic ones cannot even make it from the edge of the galaxy: they crash into microwave background radiation along the way). And though radio waves routinely travel as far as neutrinos, they are emitted even by rather mundane astronomical objects—the moon, for instance. Neutrinos, on the other hand, not only travel long distances, they are easy to categorize: low-energy neutrinos are generated by the sun, by cosmic-ray collisions in the upper atmosphere, and by other nearby phenomena; high-energy neutrinos only reach the earth from distant, supremely violent events—gamma-ray bursts, for instance, or black holes at the center of new galaxies. By focusing on high-energy neutrinos alone, a telescope can naturally filter out all but the most interesting things in the sky.

But there is no free lunch. If neutrinos can fly through planets without stopping, they hardly brake for your average telescope mirror. In fact, neutrinos are so hard to detect that for decades they existed only in theory. The Swiss theoretical physicist Wolfgang Pauli "invented" them in 1930 to balance out the energy apparently lost when radioactive matter decays. ("I have done a terrible thing," he told the German astronomer Walter Baade. "I have postulated a particle that cannot be detected.") It was not until twenty-six years later, when the physicists Frederick Reines and Clyde L. Cowan built a neutrino detector near the Savannah River nuclear plant in South Carolina, that the existence of the particle was confirmed.

IN THE PAST FOUR DECADES neutrino detectors of increasing scale and sensitivity have cropped up in odd spots around the globe: in an iron mine in Minnesota; underneath a mountain in Italy’s Apennine range; in an abandoned railway tunnel on the outskirts of Osaka, Japan. None of them are really telescopes, however: rather than tracking high-energy neutrinos to map deep space, as AMANDA does, they simply detect low-energy neutrinos from the sun.

One might think that one or two solar neutrino detectors would be enough. But a single mystery has continued to tantalize investigators. According to standard astrophysical theory, nuclear fusion inside the sun ought to spawn a stable number of neutrinos, and so physicists ought to detect a predictable number of them. Instead, month after month, in detector after detector, no more than half of the expected neutrinos are counted. Are the detectors faulty, or does solar astrophysics need some revision? Neither one, most physicists now say. The flux of solar neutrinos is too weak because the particles transform themselves en route to the earth. What were once electron neutrinos become their particulate cousins: muon neutrinos and tau neutrinos, which most neutrino detectors cannot detect.

The fact that neutrinos come in three "flavors" is old and undisputed. But the idea that they oscillate between those three flavors entails a fundamental rethinking of their nature—and perhaps of the universe itself. According to the standard model of elementary particle physics, neutrinos have no mass at all. Yet according to quantum theory, only particles with mass can oscillate between one flavor and another. By recent estimates, there are about 100 million times as many neutrinos in the universe as there are protons and neutrons combined. Even if the mass of each neutrino is no more than a tenth of an electron volt, their collective mass would be as great as that of all the visible matter in the universe.

"Neutrino oscillations have been discovered at least four times and undiscovered at least twice," notes the particle physicist Donald H. Perkins of the University of Oxford. Last June, however, physicists in Takayama, Japan, announced results that have silenced most of the remaining doubters. Their Super-Kamiokande detector, built in a working zinc mine a kilometer underground, incorporates more than 13,000 photomultipliers to survey 50,000 tons of water for telltale flashes of light. (A photomultiplier looks like a lightbulb and works like a lightbulb in reverse: light goes in and electricity comes out. But what a lightbulb! The photomultipliers in Super-K amplify signals by a factor of 100 million.)

In principle, Super-K ought to detect equal numbers of low-energy neutrinos radiating in from all sides from the collisions of cosmic rays with the atmosphere. But in the past three years, the detector found that fewer neutrinos were coming in from the far side of the earth than from the near side. The only explanation for the discrepancy, the physicists concluded, was that some muon neutrinos, which Super-K can detect, must have changed into undetectable tau neutrinos as they passed through the earth.

The Super-K data are so precise, so elegantly conceived, that they seem to prove once and for all that neutrinos have mass—thereby punching a hole in one of the standard model’s tires. Whether fixing that hole will require a simple patch or a complete overhaul remains to be seen: Super-K can only measure the difference in mass between two neutrinos—a quantity somewhere between 0.1 and 0.01 electron volt. Based on that figure and other data, however, many physicists now estimate that the three types of neutrinos have a combined mass of around 0.1 electron volt.

What Super-K does not do—what no working detector has ever done—is pay attention to neutrinos from beyond our galaxy. Enter AMANDA. Because high-energy neutrinos are 1012 times easier to detect than solar neutrinos, our telescope can afford to trade sensitivity for size. The same volume of water surveyed by 13,000 photomultipliers in Super-K is surveyed by only around ten in AMANDA; our photomultipliers are less than half as large; and ours do not detect solar neutrinos at all. But AMANDA, when complete, will watch over thousands of times more water than Super-K. As a result, it will track neutrinos across as far as a kilometer, whereas Super-K can track them across no more than fifty meters.

When a high-energy neutrino collides with a proton in the ice, it creates a muon—a particle closely related to the electron but more than 200 times more massive—that continues along the neutrino’s upward path, streaming photons along its sides like a bottle rocket. The result is a hurtling cone of blue Cherenkov light—light of the same kind emitted by nuclear reactors. From the timing of the cone’s reception by our grid of photomultipliers, the muon’s direction can be reconstructed and the direction of the incoming neutrino inferred.

Why are we looking for neutrinos that have passed through the earth? Muons generated by cosmic rays also bombard the earth constantly from every direction, and some of them can travel through miles of ice before petering out. But no muon can cross the entire planet—as much as 8,000 miles of dense iron, magma and rock. Hence by pointing its photomultipliers downward, AMANDA uses the earth itself to screen out all upward-traveling muons except the ones thought to come from high-energy neutrinos.

AS EARLY AS THE 1960s, physicists had dreamed that radio antennas, operating near gigahertz frequencies, might listen in on the electric charges sparked by neutrinos crashing into ice. Thirty-five years later, however, working through the theory behind that idea, my colleagues and I showed that the radio signal created by the neutrinos was too weak to be of any use. It was then that I hit upon the obvious alternative: Why not try to detect the flash from a neutrino collision, rather than its noise?

I suspect that others must have contemplated the same idea and given up on it. Had I not been completely ignorant of what was then known about the optical properties of natural ice, I would probably have done the same. Instead, I sent off a flurry of E-mail messages to my friend John G. Learned, then the spokesperson for the Deep Underwater Muon and Neutrino Detector (DUMAND). Like AMANDA after it, DUMAND was designed to detect high energy muons—though in ocean water off the coast of Hawai’i rather than in Antarctic ice. But though Learned’s group had already deployed a test string of photomultipliers with some success, the project was eventually abandoned. (The DUMAND concept lives on, albeit at a smaller scale, in a detector now operating in the depths of Lake Baikal in Russia.)

Learned immediately appreciated the advantages of an Antarctic neutrino telescope. For starters, sinking the photomultipliers into ice would enable investigators to walk around on top of the experiment, as well as to keep all the fragile electronics at the surface. As a result, the neutrino signals could be identified with off-the-shelf electronics. Better yet, the ice would be geologically stable (Antarctica almost never has earthquakes) and completely dark. DUMAND did not have it so easy. Although the water off the Hawaiian coast is exceptionally clear and deep, Learned’s group had to contend with waves, storms, background light from bioluminescent organisms and the radioactive decay of sea salt. Most important, NSF was already operating a research station at the geographical South Pole, with an infrastructure to rival that of a national particle physics laboratory. AMANDA, in other words, would be much cheaper and easier to build than DUMAND.

AS I WRITE, it is ten degrees below zero Fahrenheit outside my office in Madison, and I am dressed just warmly enough to be slightly cold. In most places, I think, people have made an art of underdressing in winter. But not in Antarctica. Faced with temperatures that regularly dip to negative fifty degrees Fahrenheit, even on a summer’s day, our drillers and engineers wear outfits akin to the space suits worn by astronauts on the moon. They live in a comfortable base camp with a wonderful professional chef, and the few times they expose themselves to the elements are when they relax themselves in the pools of hot water created by the AMANDA drills.

In the heroic early days of AMANDA, before a heated, portable hut was built for each drilling site, there were some tough stretches. Teams of ten people sometimes worked up to twenty-four hours without a break—often without gloves when assembling delicate components. But on the whole, the work has been astonishingly unadventurous.

The true challenges have been technical and logistical. Typically we fly 100,000 pounds of cargo and forty people from Christchurch, New Zealand, to the South Pole each summer—enough to fill twenty Hercules C-130 transport planes. It is a massive undertaking, but one accomplished with the utmost efficiency: where Antarctic research is concerned, there is no margin for excess baggage of any kind—be it fiber-optic cable, canisters of fuel or theorists with no real business on the ice.

Once in Antarctica, the operation is orchestrated by Bruce R. Koci, our mechanical engineer and drillmeister. Much of the project is entirely novel, making improvisation the rule, and Koci is a genius at it. Early on, for instance, we found that our hot-water drill was inadequate: it needed eleven days to drill to a depth of 800 meters. So Koci and his colleagues designed a new one. As sleek as a rocket, it dives into the ice gushing 190-degree water from its nose. In its first incarnation, it traveled a thousand meters in four days. These days it goes twice that fast. Mapping AMANDA’s geometry is like manufacturing an optical telescope in a dark room. Yet Koci’s drill, guided by gravity alone, deviates from the vertical by less than a meter over a depth of two kilometers.

As the drill descends, it leaves a hole about fifty centimeters in diameter, filled with hot water. (Because the hot water is continuously circulated in the hole, and because the ice around it acts like a giant thermos bottle, the water remains liquid for a few days.) Once the drill is removed, the AMANDA crew, often assisted by drillers and volunteers from other scientific missions, attaches a 600-pound weight to a fiber-optic master cable and then drops it into the hole. For the next ten to twenty hours, photomultipliers are attached to the sinking cable with carabiners of the kind used by rock climbers and plugged in at predetermined positions like beads on a rosary. Pressure and temperature meters, lasers, radio receivers, pulsing or steady light-emitting devices and other devices are also attached. (On one occasion a pair of television cameras was sent into the hole; their images can be viewed on our home page at <amanda.berkeley.edu>.)

IT WOULD HAVE BEEN EASY to build a more conventional neutrino telescope than AMANDA. We would have covered a square kilometer with spark chambers, shielding them from cosmic radiation with lead plates a few inches thick. The end result would have detected neutrinos beautifully and cost about $10 billion—a thousand times as much as we could afford. Instead, we resigned ourselves to using the simplest instruments possible to detect neutrinos across the greatest possible volume of water at the least possible cost. We would build a telescope that barely works.

Unfortunately, such a design depends, to some extent, on nature’s cooperation. So it was that our initial euphoria, on Christmas Eve, turned quickly to perplexity. We knew that some downward-traveling muons, created by cosmic-ray events at the surface, would reach our photomultipliers—even 800 to 1,000 meters beneath the surface. But we detected a hundred times as many as we expected. And though we had expected that bubbles, at that depth, would scatter the Cherenkov light to some degree, what we saw instead was a nearly meaningless blur.

Everything glaciologists had told us about Antarctic ice, it seemed, was wrong. To begin with, the ice down there was far more transparent than anyone had expected. Condensed from snow that fell 10,000 years ago, at the end of the last ice age, it could transmit a streak of blue light as far as one hundred meters, not just the eight meters that had been predicted. (The discrepancy seemed to arise from the fact that glaciologists carried out their tests with distilled water, which was much less pure than Antarctic ice.) In fact, because our photomultipliers operate at wavelengths where neither atomic nor molecular excitations absorb photons, the ice was almost infinitely transparent. That implied we could detect a few more upward-traveling muons than we had hoped, but it also explained the excess downward-traveling muons.

The bubbles were the real problem, however. A kilometer beneath the surface, we had been told, there would be just a few bubbles, and those would be no more than a micron in diameter. Instead, bubbles were everywhere, and they were fifty times larger than predicted. (We later found a paper on bubbles in ice cores that corroborated our results. But I suspect that some glaciologists still do not believe us.) After a good deal of data analysis and modeling, we predicted that the bubbles would disappear below 1,400 meters. But though we were eventually proved right, we had lost a year by then and still had to go back to drill some new holes.

THIS PAST SUMMER, the second phase of AMANDA was finally calibrated and the third phase was begun. Our telescope is now made up of 420 photomultipliers on thirteen separate strands, sunk between 1,500 and 2,000 meters beneath the Antarctic ice. The photomultipliers are working without a hitch, the telemetry data from the drill and in situ laser measurements agree on their architecture, and the telescope is detecting neutrinos. If the project as a whole has been a roller-coaster ride—one so exciting that we often failed to notice when we were hitting bottom—we are now, undeniably, at a peak. After years of being consumed by engineering problems, we can now at least concentrate on physics. Within days of calibrating the telescope, the intercontinental E-mails that buzz between us changed topics from scattering angle and photomultiplier noise to dark matter, gamma-ray bursts and neutrino oscillations.

Although we have yet to mine three-quarters of our data, we have already begun to follow some veins of interesting science. Theorists have speculated for a decade or so, for instance, that the dark matter in the universe is made up of supersymmetric particles known as neutralinos. According to that theory, most neutralinos are trapped at the centers of stars and planets, where standard instruments cannot detect them. When neutralinos collide at the center of the earth, however, they ought to generate high-energy neutrinos, which AMANDA can detect.

Early on, an investigator in our Stockholm group noted that a muon had skittered up one of AMANDA’s strings, illuminating the photomultipliers in sequence like a string of Christmas lights. Of all the neutrinos that had streamed through AMANDA from known cosmic sources, we calculated that only three neutrinos should have aligned randomly with the center of the earth. Yet when we looked at our data, we tentatively identified nine such events. Unfortunately, when we went back over the numbers, the discrepancy between the number of predicted and observed events largely evaporated.

Neutrino oscillation should prove an equally intriguing and elusive quarry. If neutrinos, contrary to current assumptions, are relatively massive—say, between ten and fourteen electron volts—they should oscillate fairly quickly. In that case, even short-range tests, such as the 450-mile neutrino beam that investigators at Fermi National Accelerator Laboratory in Batavia, Illinois, are planning to aim at a neutrino detector in Minnesota, should confirm oscillations. (As it happens, the Fermilab beam passes right under my office in Madison.) If neutrinos have almost no mass, it may take a detector as large as ours, observing neutrinos that come from as far as 10 billion light-years away, to see oscillations. Some of us have even been dreaming about sending a beam from Fermilab to the South Pole: a 6,000-mile trip that may be just long enough to reveal oscillations, if neutrinos have the mass that the Super-K data suggest.

In the next five years we plan to sink seventy more strings of photomultipliers, enabling the telescope to keep watch over a full cubic kilometer of ice. Even then, in some ways, our telescope will be a pretty crude instrument. Compared to a standard optical telescope, its resolution is a joke: as of this writing, it is accurate only to within two degrees of arc—roughly four times the diameter of the full moon—though we hope to cut that margin in half. AMANDA’S size alone will give it tremendous power, of course, enabling it to detect gamma-ray bursts and super-massive black holes several times a year. And yet, spectacular though they are, such cosmic light shows should prove no more than opening acts for the main event—whatever that may be.

Chinese Full Text

南极之梦

弗朗西斯·哈尔岑（Francis Halzen）

关于 AMANDA 的文学性随笔。原载 The Sciences，第 39 卷第 2 期（1999 年），第 19 页。由 Exploratorium 存档。

大科学常常取决于一些微小的时刻。当经费已经落实、政治已经疏通、场地已经破土、愿景已经许下，当成排的计算机终于亮起，手指悬在键盘上方，准备通过电子邮件传出第一批发现时，十年的工作仍可能在最后一刻轰然坍塌——就像一座建在理论稀薄空气中的城堡，承受不住真实的重力。

去年八月，丹尼斯·S·皮科克（Dennis S. Peacock）来到威斯康星大学麦迪逊分校，走进我的办公室时，他心里大概也想着这些。作为美国国家科学基金会（NSF）南极科学部门的负责人，皮科克曾帮助把约一千万美元投入这项计划，用于在南极冰层深处建造世界上最大的中微子望远镜。这笔投资究竟能不能带来成果？或者，我们的项目会不会步它的前身——DUMAND——的后尘？那个项目部分建在夏威夷近岸海水中，在近二十年的工作之后最终被放弃。

我们的望远镜，全称是“南极 μ 子和中微子探测阵列”（Antarctic Muon and Neutrino Detector Array，AMANDA），其实已经探测中微子有一段时间了，只是当时我们还没有完成第一批数据的分析。至少我是这么以为的。可是，当皮科克和我走到我的研究生雷伦·R·哈德克（Rellen R. Hardtke）的桌前时，她给了我一个惊喜。她微笑着解释说，两位同事熬了一个通宵，把分析做完了。接着，她平静地调出电脑屏幕上的一幅图像：有史以来第一张精确记录高能中微子事例的图。

哈德克的屏幕上，一条淡蓝色的线斜斜掠过由黑点组成的阵列。大多数黑点都很小、呈黑色，它们各自代表一个沉入冰中的光电倍增管；但有几个点沿着这条线聚集，呈现蓝色、绿色或红色，其中两个靠近线的起点，又亮又大，呈鲜橙色。图像告诉我们，1997 年 10 月 12 日凌晨五点，一个中微子——自然界最小、最难捉摸的基本粒子之一——从太平洋中部、中途岛与阿留申群岛之间进入地球，笔直穿过整个星球，在南极冰层下表面与一个质子迎头相撞。在地表以下两公里处，我们的光电倍增管阵列捕捉到了这次碰撞产生的亚原子火花；它向上穿过冰层，在探测器旁闪烁了大约一微秒。

抽象地呈现在电脑屏幕上时，这个最终结果对大多数人来说也许毫不起眼。但对我们而言，它是一个穿越浩瀚距离抵达地球的粒子留下的唯一痕迹，也许来自宇宙中最壮观的某个事件的残余。更重要的是，它是 AMANDA 真正有效的第一个具体证据——它能够帮助我们绘制遥远宇宙深处的图景，并可能解决物理学中一些最激烈的争论。后来，我把这张图发给合作者，其中一个人回复说：“这就是我为这个项目花了五年的原因。”

“中微子，它们非常小。”约翰·厄普代克（John Updike）写下了那句著名的诗。“它们不带电荷，也没有质量／而且完全不发生相互作用。”事实证明，厄普代克有两处说错了，但他还是抓住了要义。中微子又小又滑，像一颗子弹穿过暴风雨那样穿过地球、恒星、城市，以及几乎所有东西。它们不受磁场和强核力影响；要想让它们停下来，必须让它直接撞上一个质子——而这是极不可能发生的事。同时，中微子在宇宙中的数量几乎与构成光的光子一样多：每秒钟有 3 × 10¹⁶ 个中微子穿过我们的身体。

这些因素共同使中微子成为深空天文学最好的研究对象。大多数光子无法从宇宙最遥远的角落抵达地球（能量最高的光子甚至无法从银河系边缘赶来，它们途中会撞上微波背景辐射）。射电波虽然通常能传播到和中微子一样远，但连月球这样相当普通的天体也会发出射电波。中微子则不同：它们不仅能传播极远的距离，而且容易分类。低能中微子来自太阳、大气上层的宇宙线碰撞，以及其他邻近现象；高能中微子只会从遥远而极端暴烈的事件中抵达地球，例如伽马射线暴，或新生星系中心的黑洞。只要专注于高能中微子，望远镜就能自然过滤掉天空中几乎所有无趣的东西。

但天下没有免费的午餐。既然中微子能毫无阻碍地穿过行星，它们也不会为了寻常的望远镜镜面停下来。事实上，中微子极难探测，以至于在几十年里，它们只存在于理论之中。1930 年，瑞士理论物理学家沃尔夫冈·泡利（Wolfgang Pauli）“发明”了中微子，为的是解释放射性物质衰变时看似消失的能量。他对德国天文学家瓦尔特·巴德（Walter Baade）说：“我做了一件可怕的事：我提出了一种无法被探测到的粒子。”直到二十六年后，物理学家弗雷德里克·莱因斯（Frederick Reines）和克莱德·L·考恩（Clyde L. Cowan）在南卡罗来纳州萨凡纳河核电站附近建造了一台中微子探测器，这种粒子的存在才得到确认。

在过去四十年里，规模和灵敏度越来越高的中微子探测器出现在地球上一些意想不到的角落：明尼苏达州的铁矿、意大利亚平宁山脉的山腹之下、日本大阪郊外一条废弃的铁路隧道。不过，它们其实都不是真正的望远镜：它们没有像 AMANDA 那样追踪高能中微子以绘制深空图景，而只是探测来自太阳的低能中微子。

有人也许会想，一两台太阳中微子探测器应该就够了。但有一个谜团始终困扰着研究者。按照标准天体物理理论，太阳内部的核聚变应当产生数量稳定的中微子，因此物理学家也应当探测到可以预言的数量。可是，月复一月，探测器换了一台又一台，人们数到的中微子从不超过预期的一半。是探测器出了问题，还是太阳天体物理学需要修正？现在大多数物理学家说，两者都不是。太阳中微子通量之所以显得太弱，是因为这些粒子在前往地球的途中发生了转变：电子中微子变成了它们的“表亲”——μ 子中微子和 τ 子中微子，而大多数探测器探测不到这两种中微子。

中微子具有三种“味道”，这是早已确立、无可争议的事实。但认为它们会在三种味道之间振荡，就意味着必须从根本上重新思考中微子的本性，甚至重新思考宇宙本身。按照基本粒子物理学的标准模型，中微子完全没有质量。可是按照量子理论，只有具有质量的粒子才能在两种味道之间振荡。根据近期估计，宇宙中中微子的数量约为质子和中子总数的一亿倍。即使每个中微子的质量不超过十分之一电子伏特，它们的质量总和也会与宇宙中所有可见物质相当。

“中微子振荡至少被发现过四次，也至少两次被‘发现又被否定’。”牛津大学粒子物理学家唐纳德·H·珀金斯（Donald H. Perkins）这样写道。不过去年六月，日本飞驒市的物理学家宣布的结果，让大多数残存的怀疑者闭上了嘴。他们的超级神冈探测器建在地下一公里处的一座仍在运作的锌矿中，使用超过 1.3 万个光电倍增管监测 5 万吨水中提示性的闪光。（光电倍增管看起来像灯泡，工作原理却像反过来的灯泡：光进入，电出来。可这是多么了不得的“灯泡”！超级神冈中的光电倍增管能把信号放大一亿倍。）

从原理上说，超级神冈应当从四面八方探测到数量相等的低能中微子，它们来自宇宙线与大气层的碰撞。但在过去三年里，探测器发现，从地球远侧来的中微子少于从近侧来的中微子。物理学家得出结论：造成这个差异的唯一解释，是一些 μ 子中微子——超级神冈能探测到的中微子——在穿过地球时变成了探测不到的 τ 子中微子。

超级神冈的数据如此精确、设计如此优雅，以至于它似乎一劳永逸地证明了中微子具有质量，从而在标准模型的一只轮胎上扎出了孔。至于修补这个孔需要一个简单补丁，还是一次彻底大修，仍有待观察：超级神冈只能测量两种中微子之间的质量差，这个量介于 0.1 到 0.01 电子伏特之间。不过，根据这一数字和其他数据，许多物理学家现在估计，三类中微子的总质量约为 0.1 电子伏特。

超级神冈做不到的——也是目前任何一台实际运行的探测器都做不到的——是关注来自银河系之外的中微子。于是，AMANDA 登场了。因为高能中微子比太阳中微子容易探测 10¹² 倍，我们的望远镜可以用灵敏度换取规模。超级神冈用 1.3 万个光电倍增管监测的水体，在 AMANDA 中只需要约十个光电倍增管；我们的光电倍增管尺寸不到它们的一半，而且完全不探测太阳中微子。但是，AMANDA 完成之后，将监测比超级神冈多数千倍的水。因此，它能追踪中微子远至一公里，而超级神冈最多只能追踪五十米。

当一个高能中微子与冰中的质子相撞时，会产生一个 μ 子——一种与电子关系密切、但质量超过电子 200 倍的粒子。它沿中微子向上的路径继续前进，像一枚瓶式火箭那样沿两侧洒出光子。结果是一道急速掠过的蓝色切伦科夫光锥——这种光与核反应堆发出的光同类。根据我们的光电倍增管阵列接收到光锥的时间，可以重建 μ 子的方向，进而推断入射中微子的方向。

我们为什么要寻找穿过地球而来的中微子？宇宙线产生的 μ 子也在从四面八方不断轰击地球，其中一些能够在几英里的冰层中穿行才逐渐消失。但没有 μ 子能穿越整个行星——那可能相当于 8000 英里致密的铁、岩浆和岩石。因此，AMANDA 把光电倍增管朝向下方，利用地球本身屏蔽所有向上行进的 μ 子，只留下那些被认为来自高能中微子的 μ 子。

早在 20 世纪 60 年代，物理学家就曾设想，工作在吉赫兹频率附近的射电天线也许能“听见”中微子撞入冰中时激发的电荷。三十五年后，我的同事们和我推演了这一想法背后的理论，结果发现中微子产生的射电信号太弱，没有实用价值。就在那时，我想到一个显而易见的替代方案：为什么不试着探测中微子碰撞发出的闪光，而不是它的噪声？

我猜想，别人一定也想到过这个主意，只是后来放弃了。如果当时我完全了解关于天然冰的光学性质的已知结论，我大概也会这么做。相反，我给我的朋友约翰·G·利尔尼斯（John G. Learned）连发了一连串电子邮件。他当时是“深海水 μ 子和中微子探测器”（DUMAND）的发言人。与后来的 AMANDA 一样，DUMAND 也设计用于探测高能 μ 子，只不过它使用的是夏威夷近岸海水，而不是南极冰。利尔尼斯的团队虽然已经较为成功地部署过一串试验性光电倍增管，但该项目最终还是被放弃了。（DUMAND 概念依然存在，只是规模较小，体现在如今在俄罗斯贝加尔湖深处运行的一台探测器中。）

利尔尼斯立刻意识到南极中微子望远镜的优势。首先，把光电倍增管沉入冰中，研究者可以站在实验装置上方走动，并把所有精密易损的电子设备留在地表；这样一来，中微子信号可以用现成的电子学设备识别。更好的是，冰在地质上非常稳定（南极几乎没有地震），而且完全黑暗。DUMAND 没有这么好的条件。虽然夏威夷近岸海水异常清澈、深邃，但利尔尼斯的团队必须对付海浪、风暴、生物发光带来的背景光，以及海盐放射性衰变的影响。最重要的是，NSF 已经在南极点运行着一座研究站，其基础设施足以与国家粒子物理实验室媲美。换句话说，AMANDA 会比 DUMAND 便宜得多，也更容易建造。

写下这些文字时，麦迪逊我的办公室外是华氏零下十度，而我只穿得刚好略感寒冷。我想，在大多数地方，人们已经把冬季穿得少当成一门艺术。但在南极不行。即使在夏季，气温也常常降到华氏零下五十度，我们的钻探人员和工程师必须穿上类似宇航员登月穿的太空服。他们住在舒适的大本营里，还有一位很棒的专业厨师；他们真正暴露在户外恶劣环境中的时候，反而是跳进 AMANDA 钻探产生的热水池里放松。

在 AMANDA 的英雄早期，每个钻探点还没有加热的可移动小屋时，有过几段艰苦的日子。十人小组有时连续工作二十四小时不休息，装配精密部件时常常连手套都不戴。但总体而言，这项工作出奇地不算冒险。

真正的挑战来自技术和后勤。通常，每个夏季我们要把 10 万磅货物和四十个人从新西兰克赖斯特彻奇运到南极点——足以装满二十架 C-130“大力神”运输机。这是一项庞大工程，却必须以极高效率完成：在南极研究中，任何多余的行李都没有容身之地，无论是光缆、燃料罐，还是根本没有必要上冰的理论家。

到了南极，整个行动由布鲁斯·R·科奇（Bruce R. Koci）统筹，他是我们的机械工程师，也是钻探大师。项目中有许多东西完全前所未有，因此随机应变成了常态，而科奇正是这方面的大才。例如，早期我们发现热水钻不够好：钻到 800 米深处需要十一天。于是科奇和同事设计了一台新钻机。它像火箭一样流线，从钻头喷出华氏 190 度的热水扎进冰里。第一代钻机四天能钻一千米；现在的速度已经是原来的两倍。给 AMANDA 绘制几何形态，就像在暗室里制造光学望远镜；而科奇的钻机仅靠重力导向，在两公里深度内偏离垂直线不到一米。

钻机下探时，会留下一个直径约五十厘米、充满热水的孔。（热水在孔中不断循环，周围冰层又像一个巨大的保温瓶，因此孔里的水能维持几天液态。）钻机取出后，AMANDA 团队——常有其他科学任务的钻探人员和志愿者协助——把一个 600 磅的重物系在光纤主缆上，放入孔中。接下来的十到二十小时里，人们用攀岩者使用的那种登山扣，把光电倍增管装到不断下沉的缆绳上，并像念珠一样插到预定位置。压力计、温度计、激光器、无线电接收器、脉冲或稳定发光装置，以及其他设备也会一并安装。（有一次，人们把两台电视摄像机送进孔中；图像可以在我们的主页 <amanda.berkeley.edu> 上查看。）

其实，要建造一台比 AMANDA 更常规的中微子望远镜很容易。我们本可以用火花室覆盖一平方公里，并用几英寸厚的铅板屏蔽宇宙辐射。这样造出来的设备会非常漂亮地探测中微子，但成本约为 100 亿美元——是我们负担得起的一千倍。于是，我们退而求其次，用尽可能简单的仪器，在尽可能大的水体中，以尽可能低的成本探测中微子。我们要造一台“勉强能用”的望远镜。

不幸的是，这种设计在一定程度上依赖大自然的配合。于是，圣诞夜的狂喜很快变成了困惑。我们知道，宇宙线事件在地表产生的某些向下运动的 μ 子会到达我们的光电倍增管——哪怕是在地下 800 到 1000 米。但我们探测到的数量是预期的一百倍。而且，我们原本以为在那个深度，气泡会让切伦科夫光发生一定散射；结果看到的却是一片几乎毫无意义的模糊。

冰河学家告诉我们的关于南极冰的一切，似乎都错了。首先，那里的冰远比任何人预期的透明。这些冰由末次冰期结束时、一万年前落下的雪压缩而成，能让一道蓝色光传播一百米，而不是预言中的八米。（这个差异似乎源于这样的事实：冰河学家用蒸馏水做实验，而蒸馏水远不如南极冰纯净。）事实上，由于我们的光电倍增管工作在原子或分子激发都不吸收光子的波长上，这里的冰几乎无限透明。这意味着我们能探测到的向上运动的 μ 子比预期更多，但同时也解释了向下运动的 μ 子为何过多。

真正的问题却是气泡。我们被告知，在地下一公里处，气泡应该很少，直径不超过一微米。实际上，气泡到处都是，而且比预测大五十倍。（后来我们找到一篇关于冰芯气泡的论文，支持了我们的结果。但我怀疑有些冰河学家仍然不相信我们。）经过大量数据分析和建模，我们预测气泡会在 1400 米以下消失。虽然我们最终被证明是对的，但那时已经损失了一年，还必须回去重新钻一些孔。

刚过去的这个夏天，AMANDA 第二阶段终于完成校准，第三阶段也开始了。我们的望远镜现在由十三根独立缆绳上的 420 个光电倍增管组成，沉放在南极冰层下 1500 至 2000 米之间。光电倍增管运转毫无故障，钻探遥测数据与现场激光测量在结构上相互吻合，望远镜正在探测中微子。如果说整个项目像一趟过山车——它刺激得让我们常常没有察觉自己何时落到了谷底——那么现在，毫无疑问，我们正处在峰顶。多年被工程问题占据之后，我们至少可以把注意力转回物理。望远镜校准后的几天里，我们在各大洲之间往来的电子邮件，很快从散射角和光电倍增管噪声，转向暗物质、伽马射线暴和中微子振荡。

尽管我们还有四分之三的数据没有挖掘，但已经顺着几条有趣的科学脉络展开工作。例如，理论家大约十年来一直推测，宇宙中的暗物质由被称为中性微子的超对称粒子构成。按照这种理论，大多数中性微子会被困在恒星和行星中心，常规仪器无法探测到它们。不过，当中性微子在地球中心相互碰撞时，应当会产生高能中微子，而 AMANDA 正能探测它们。

早期，我们斯德哥尔摩小组的一位研究者注意到，一个 μ 子沿 AMANDA 的一根缆绳向上掠过，像一串圣诞灯那样依次点亮光电倍增管。根据已知宇宙源穿过 AMANDA 的所有中微子计算，只有三个中微子应当随机地与地心对齐。然而，我们粗略地从数据中辨认出九个这样的事例。不幸的是，重新核对数字后，预测事例数与观测事例数之间的差异大多消失了。

中微子振荡应当同样迷人，也同样难以捕捉。如果中微子不像目前假设的那样几乎没有质量，而是相对较重——比如在 10 到 14 电子伏特之间——它们就会较快振荡。在这种情况下，即便是短程实验也应该能证实振荡，例如伊利诺伊州巴达维亚费米国家加速器实验室的研究人员正计划把一束中微子射向明尼苏达州的一台中微子探测器。（碰巧，费米实验室的中微子束正好从我在麦迪逊的办公室下方穿过。）但如果中微子几乎没有质量，也许就需要像我们这样大的探测器，观测来自一百亿光年之外的中微子，才能看到振荡。我们中有些人甚至梦想着把一束中微子从费米实验室送到南极点：这段 6000 英里的旅程，也许刚好长到足以揭示振荡——如果中微子质量真如超级神冈数据所暗示的话。

未来五年，我们计划再沉放七十多串光电倍增管，让望远镜监测整整一立方公里的冰。即便如此，在某些方面，我们的望远镜仍会是一件相当粗糙的仪器。与标准光学望远镜相比，它的分辨率实在好笑：截至写作时，它只能精确到两度弧——大约是满月直径的四倍——不过我们希望把这个误差再减少一半。当然，AMANDA 的规模本身就会带来巨大威力，使它能每年探测到几次伽马射线暴和超大质量黑洞。然而，无论这些宇宙光秀多么壮观，它们都只是主要节目的开场表演——至于那主要节目是什么，谁也不知道。