Editing Nuclear weapon design (section)

==Explosive testing==
Nuclear weapons are in large part designed by trial and error. The trial often involves test explosion of a prototype.

In a nuclear explosion, a large number of discrete events, with various probabilities, aggregate into short-lived, chaotic energy flows inside the device casing. Complex mathematical models are required to approximate the processes, and in the 1950s there were no computers powerful enough to run them properly. Even today's computers and simulation software are not adequate.<ref>Walter Goad, [https://fas.org/irp/ops/ci/goad.html Declaration for the Wen Ho Lee case] {{webarchive |url=https://web.archive.org/web/20160308031512/https://fas.org/irp/ops/ci/goad.html |date=March 8, 2016}}, May 17, 2000. Goad began thermonuclear weapon design work at Los Alamos in 1950. In his Declaration, he mentions "basic scientific problems of computability which cannot be solved by more computing power alone. These are typified by the problem of long range predictions of weather and climate, and extend to predictions of nuclear weapons behavior. This accounts for the fact that, after the enormous investment of effort over many years, weapons codes can still not be relied on for significantly new designs."</ref>

It was easy enough to design reliable weapons for the stockpile. If the prototype worked, it could be weaponized and mass-produced.{{Citation needed|date=June 2021}}

It was much more difficult to understand how it worked or why it failed. Designers gathered as much data as possible during the explosion, before the device destroyed itself, and used the data to calibrate their models, often by inserting [[wikt:fudge factor|fudge factors]] into equations to make the simulations match experimental results. They also analyzed the weapon debris in fallout to see how much of a potential nuclear reaction had taken place.{{Citation needed|date=June 2021}}

{{Anchor|Light pipes}}

===Light pipes===
An important tool for test analysis was the diagnostic light pipe. A probe inside a test device could transmit information by heating a plate of metal to incandescence, an event that could be recorded by instruments located at the far end of a long, very straight pipe.{{Citation needed|date=June 2021}}

The picture below shows the Shrimp device, detonated on March 1, 1954, at Bikini, as the [[Castle Bravo]] test. Its 15-megaton explosion was the largest ever by the United States. The silhouette of a man is shown for scale. The device is supported from below, at the ends. The pipes going into the shot cab ceiling, which appear to be supports, are actually diagnostic light pipes. The eight pipes at the right end (1) sent information about the detonation of the primary. Two in the middle (2) marked the time when X-rays from the primary reached the radiation channel around the secondary. The last two pipes (3) noted the time radiation reached the far end of the radiation channel, the difference between (2) and (3) being the radiation transit time for the channel.<ref>Chuck Hansen, ''The Swords of Armageddon'', Volume IV, pp. 211–212, 284.</ref>
[[File:Castle Bravo Shrimp composite.png|600 px|centre]]

From the shot cab, the pipes turned horizontally and traveled {{convert|7500|ft|km|abbr=on|order=flip}} along a causeway built on the Bikini reef to a remote-controlled data collection bunker on Namu Island.{{Citation needed|date=June 2021}}

While x-rays would normally travel at the speed of light through a low-density material like the plastic foam channel filler between (2) and (3), the intensity of radiation from the exploding primary creates a relatively opaque radiation front in the channel filler, which acts like a slow-moving logjam to retard the passage of [[radiant energy]]. While the secondary is being compressed via radiation-induced ablation, neutrons from the primary catch up with the x-rays, penetrate into the secondary, and start breeding tritium via the third reaction noted in the first section above. This [[lithium-6|<sup>6</sup>Li]] + n reaction is exothermic, producing 5 MeV per event. The spark plug has not yet been compressed and thus remains subcritical, so no significant fission or fusion takes place as a result. If enough neutrons arrive before implosion of the secondary is complete, though, the crucial temperature differential between the outer and inner parts of the secondary can be degraded, potentially causing the secondary to fail to ignite. The first Livermore-designed thermonuclear weapon, the Morgenstern device, failed in this manner when it was tested as [[Castle Koon]] on April 7, 1954. The primary ignited, but the secondary, preheated by the primary's neutron wave, suffered what was termed as an ''inefficient detonation'';<ref name="swordsIV">{{cite book |author-link=Chuck Hansen |first=Chuck |last=Hansen |title=Swords of Armageddon |volume=IV |date=1995 |url=https://www.uscoldwar.com/ |access-date=2016-05-20 |url-status=live |archive-url=https://web.archive.org/web/20161230020259/http://www.uscoldwar.com/ |archive-date=2016-12-30}}</ref>{{rp|165}} thus, a weapon with a predicted one-megaton yield produced only 110 kilotons, of which merely 10 kt were attributed to fusion.<ref name="swordsIII">{{cite book |author-link=Chuck Hansen |first=Chuck |last=Hansen |title=Swords of Armageddon |volume=III |date=1995 |url=https://www.uscoldwar.com/ |access-date=2016-05-20 |url-status=live |archive-url=https://web.archive.org/web/20161230020259/http://www.uscoldwar.com/ |archive-date=2016-12-30}}</ref>{{rp|316}}

These timing effects, and any problems they cause, are measured by light-pipe data. The mathematical simulations which they calibrate are called radiation flow hydrodynamics codes, or channel codes. They are used to predict the effect of future design modifications.{{Citation needed|date=June 2021}}

It is not clear from the public record how successful the Shrimp light pipes were. The unmanned data bunker was far enough back to remain outside the mile-wide crater, but the 15-megaton blast, two and a half times as powerful as expected, breached the bunker by blowing its 20-ton door off the hinges and across the inside of the bunker. (The nearest people were {{convert|20|mi|km|order=flip}} farther away, in a bunker that survived intact.)<ref>Dr. John C. Clark, as told to Robert Cahn, "We Were Trapped by Radioactive Fallout", ''The Saturday Evening Post'', July 20, 1957, pp. 17–19, 69–71.</ref>

===Fallout analysis===
{{See also|Nuclear forensics}}
The most interesting data from Castle Bravo came from radio-chemical analysis of weapon debris in fallout. Because of a shortage of enriched lithium-6, 60% of the lithium in the Shrimp secondary was ordinary lithium-7, which doesn't breed tritium as easily as lithium-6 does. But it does breed lithium-6 as the product of an (n, 2n) reaction (one neutron in, two neutrons out), a known fact, but with unknown probability. The probability turned out to be high.{{Citation needed|date=June 2021}}

Fallout analysis revealed to designers that, with the (n, 2n) reaction, the Shrimp secondary effectively had two and half times as much lithium-6 as expected. The tritium, the fusion yield, the neutrons, and the fission yield were all increased accordingly.<ref>{{cite book |first=Richard |last=Rhodes |title=Dark Sun; the Making of the Hydrogen Bomb |url-access=limited |publisher=Simon and Schuster |year=1995 |page=[https://archive.org/details/darksunmakinghyd00rhod/page/n568 541] |isbn=9780684804002 |url=https://archive.org/details/darksunmakinghyd00rhod}}</ref>

As noted above, Bravo's fallout analysis also told the outside world, for the first time, that thermonuclear bombs are more fission devices than fusion devices. A Japanese fishing boat, ''[[Daigo Fukuryū Maru]]'', steamed home with enough fallout on her decks to allow scientists in Japan and elsewhere to determine, and announce, that most of the fallout had come from the fission of U-238 by fusion-produced 14 MeV neutrons.{{Citation needed|date=June 2021}}

===Underground testing===
{{Main|Underground nuclear weapons testing}}
[[File:Nevada Test Site craters.jpg|thumb|right|Subsidence Craters at Yucca Flat, Nevada Test Site.]]
The global alarm over radioactive fallout, which began with the Castle Bravo event, eventually drove nuclear testing literally underground. The last U.S. above-ground test took place at [[Johnston Island]] on November 4, 1962. During the next three decades, until September 23, 1992, the United States conducted an average of 2.4 underground nuclear explosions per month, all but a few at the [[Nevada Test Site]] (NTS) northwest of Las Vegas.{{Citation needed|date=June 2021}}

The [[Yucca Flat]] section of the NTS is covered with subsidence craters resulting from the collapse of terrain over radioactive caverns created by nuclear explosions (see photo).

After the 1974 [[Threshold Test Ban Treaty]] (TTBT), which limited underground explosions to 150 kilotons or less, warheads like the half-megaton W88 had to be tested at less than full yield. Since the primary must be detonated at full yield in order to generate data about the implosion of the secondary, the reduction in yield had to come from the secondary. Replacing much of the lithium-6 deuteride fusion fuel with lithium-7 hydride limited the tritium available for fusion, and thus the overall yield, without changing the dynamics of the implosion. The functioning of the device could be evaluated using light pipes, other sensing devices, and analysis of trapped weapon debris. The full yield of the stockpiled weapon could be calculated by extrapolation.{{Citation needed|date=June 2021|reason=W88 was test full yield before ban}}