Groves appointed a committee of men thoroughly experienced by now in Manhattan Project troubleshooting: W. K. Lewis, Eger Murphree and Richard Tolman. They visited the Philadelphia Navy Yard on June 1 and turned in their conclusions on June 3.2097 They thought Oppenheimer’s estimate of 12 kilograms a day of 1 percent U235 optimistic but emphasized the possibility—with 300 columns instead of 100—of producing 30 kilograms per day of 0.95 percent U235.
Groves thought bigger than that. He had a power plant with 238,000 kilowatts rated capacity coming on line within weeks in the K-25 area at Oak Ridge that K-25 would not be ready to draw on until the end of the year. It was designed to generate electricity to run the barrier diffusers but it made electricity by making steam. The steam could serve a thermal-diffusion plant that would enrich uranium for the Alpha and Beta calutrons until such a time as K-25 needed electricity. Then the permanent K-25 installation could be phased in gradually and the temporary thermal-diffusion plant phased out.
The proposal cleared the Military Policy Committee on June 12, 1944. On June 18 Groves contracted with the engineering firm of H. K. Ferguson to build a 2,100-column thermal-diffusion plant beside the power plant on the Clinch River in ninety days or less. That extraordinary deadline allowed no time for design. Ferguson would assemble the operation from twenty-one identical copies—“Chinese copies,” Groves called them—of Philip Abelson’s 100-column unit in the Philadelphia Navy Yard.2098
The general must have appreciated the fortuity of his decision when he learned the following month of the plutonium crisis at Los Alamos. But the thermal-diffusion plant was not immediately Oak Ridge’s savior. Ferguson managed to build a capacious 500-foot barn of black metal siding and began operating the first rack of columns in sixty-nine days, by September 16, but steam leaked out almost as fast as it could be blown in and couplings needed extensive repair and even partial redesign. The gaseousdiffusion plant, K-25, was more than half completed but no barrier tubes shipped from Houdaille-Hershey yet met even minimum standards. The Alpha calutrons smeared uranium all over the insides of their vacuum tanks, catching no more than 4 percent of the U235; that valuable fraction, reprocessed and fed into the Beta calutrons, reached the Beta collectors in turn at only 5 percent efficiency. Five percent of 4 percent is two thousandths. A speck of U235 stuck to an operator’s coveralls was well worth searching out with a Geiger counter and retrieving delicately with tweezers. No essence was ever expressed more expensively from the substance of the world with the possible exception of the human soul.
* * *
In the Pacific the island war advanced. As the Army under General Douglas MacArthur pushed up from Australia across New Guinea toward the Philippines, the Marines under Admiral Chester Nimitz island-hopped from Guadalcanal to Bougainville in the Solomons, north across the equator to Tarawa in the Gilberts, farther north to Kwajalein and Eniwetok in the Marshalls. That brought them, by the summer of 1944, within striking distance of the Japanese inner defense perimeter to the west. Its nearest bastions were the Marianas, a chain of volcanic islands at the right corner of a roughly equilateral triangle of which the Philippine main island of Luzon was the left corner and the Japanese main island of Honshu the apex. The United States wanted the Marianas as primary bases for further advance: Guam for the Navy; Saipan and Tinian for the new B-29 Superfortresses that the Army Air Force had begun deploying temporarily at great risk and expense in China’s Szechwan province, ferrying aviation fuel and bombs over the Himalayas to support their mission, which was the high-altitude precision bombing of Japan. By contrast, only fifteen hundred miles of open water separated Saipan and Tinian from Tokyo and the islands could be supplied securely by sea.
Nimitz named the Marianas campaign Operation Forager; it began in mid-June with heavy bombing of the island airfields. Then 535 ships carrying 127,571 troops sailed from Eniwetok, the largest force of men and ships yet assembled for a Pacific naval operation. “We are through with flat atolls now,” Holland Smith, the Marine commanding general, briefed his officers. “We learned to pulverize atolls, but now we are up against mountains and caves where Japs can dig in. A week from now there will be a lot of dead Marines.”2099
Intelligence estimates put 15,000 to 17,000 Japanese troops on Saipan, 10,000 on smaller Tinian three miles to the south. The marines invaded Saipan first, on the morning of June 15, and won a long but shallow beachhead onto which, by afternoon, amphtracs had delivered 20,000 men. Time correspondent Robert Sherrod was among them dodging shells from Japanese artillery inland; he had seen action before on the Aleutian island of Attu and on Tarawa and knew the Japanese as America had come to know them:
Nowhere have I seen the nature of the Jap better illustrated than it was near the airstrip at dusk. I had been digging a foxhole for the night when one man shouted: “There is a Jap under those logs!” The command post security officer was dubious, but he handed concussion grenades to a man and told him to blast the Jap out. Then a sharp ping of the Jap bullet whistled out of the hole and from under the logs a skinny little fellow—not much over 5 ft. tall—jumped out waving a bayonet.2100
An American tossed a grenade and it knocked the Jap down. He struggled up, pointed his bayonet into his stomach and tried to cut himself open in approved hara-kiri fashion. The disemboweling never came off. Someone shot the Jap with a carbine. But, like all Japs, he took a lot of killing. Even after four bullets had thudded into his body he rose to one knee. Then the American shot him through the head and the Jap was dead.
While the marines advanced into Saipan, fighting off the harrowing Japanese frontal assaults they learned to call banzai charges, 155-millimeter Long Toms brought ashore and set up in the southern sector of the island began softening up Tinian.2101 That smaller island of thirty-eight square miles, ten miles long and shaped much like Manhattan, was far less rugged than Saipan. Its highest elevation, Mount Lasso, rose only 564 feet above sea level; its lowlands were planted in sugar cane; it had roads and a railway to recommend it to tank operations. To the disadvantage of amphibious assault the island was a raised platform protected on all sides by steep cliffs 500 to 600 feet high—The Rock, the marines would come to call it. It had two major beaches, one near Tinian Town on the southwest coast and the other, which the marines named Yellow, on the east coast at the island’s waist. Navy frogmen explored both by night and found them heavily mined and defended.
Two other smaller beaches on the northwest coast hardly deserved the name; one was 60 yards long and the other 150 yards. The United States had made no division-strength landing across any beach less than twice the length of those two toeholds combined in the entire course of the war. The Japanese on Tinian accordingly defended them with nothing more than a few mines and two 25-man blockhouses. The marines coded them White 1 and White 2 and chose them for their assault.
The invasion of Tinian began on July 24, two weeks after Saipan had been secured. Because of the larger island’s proximity the marines could deploy shore-to-shore rather than ship-to-shore, embarking in LST’s and smaller craft directly from Saipan. A feint at Tinian Town beach decoyed the Japanese defenses and the invaders achieved complete tactical surprise, rushing ashore and pushing inland as fast as possible to escape the dangerously narrow landings. By the end of the day, when the advance halted to organize a solid defense against the Japanese troops rushing up the island from Tinian Town, most of the tanks had been brought ashore, four howitzer batteries were in place and a spare battalion was even at hand. The defenders had killed fifteen marines and wounded fewer than two hundred; the American perimeter extended inland more than two miles.
With the coming of darkness the Japanese began a mortar barrage. Near midnight their artillery arrived and they added it in. The marines answered with their howitzers. To watch for the expected Japanese counterattack they illuminated the area with flares. The attack started at 0300 hours, Japanese soldiers rushing the American lines head-on in the naked light of the flares. Against strong Marine defenses
challenge quickly became slaughter.
The marines needed only four days to advance down the island. They encountered tanks and infantry and in the mild terrain easily destroyed them. They took Tinian Town on July 31, that night shattered a last banzai charge from the south and the next day, August 1, 1944, declared the island secure. More than 6,000 Japanese combatants died compared to 300 Americans. Another 1,500 marines were wounded. Soon the Seabees would arrive to begin bulldozing airfields.
Saipan before had been bloodier: 13,000 U.S. casualties, 3,000 marines killed, 30,000 Japanese defenders dead. But a more grotesque slaughter had engulfed the island’s population of civilians. Believing as propaganda had prepared them that the Americans would visit upon them rape, torture, castration and murder, 22,000 Japanese civilians had made their way to two sea cliffs 80 and 1,000 feet high above jagged rocks and, despite appeals from Japanese-speaking American interpreters and even fellow islanders, had flung themselves, whole families at a time, to their deaths. The surf ran red with their blood; so many broken bodies floated in the water that Navy craft overrode them to rescue. Not all the dead had volunteered their sacrifice; many had been rallied, pushed or shot by Japanese soldiers.
The mass suicide on Saipan—a Jonestown of its day—instructed Americans further in the nature of the Jap. Not only soldiers but also civilians, ordinary men and women and children, chose death before surrender. On their home islands the Japanese were 100 million strong, and they would take a lot of killing.
* * *
“The view was stupendous, and the wind was bitter cold,” Leona Marshall recalls of a day at Hanford, Washington, in September 1944 when she, Enrico Fermi and Crawford Greenewalt climbed giddily to the top of a twelve-story tower to survey the secret reservation.2102 They could see the Columbia River running deep and blue in both directions out of sight over the horizon; they could see the gray desert and the distant hazy mountains. By then construction was more than two-thirds completed and nearer at hand they overlooked a city of industrial buildings and barracks and three massive blockhouses, the three plutonium production reactors sited on the river’s western shore. The number of construction workers had peaked at 42,400 the previous June. Marshall was working now at Hanford; Fermi and Greenewalt had traveled out to monitor the start-up of the B pile, the first one finished. The day the construction teams left it, September 13, Fermi had inserted the first aluminum-canned uranium slug to begin the loading, the Pope conferring his blessing as he had on the piles at Chicago and Oak Ridge.
Slug canning had almost come to a crisis. Two years of trial-and-error effort had not produced canning technology adequate to seal the uranium slugs, which quickly oxidized upon exposure to air or water, away from corrosion. Only in August had the crucial step been devised, by a young research chemist who had followed the problem from Du Pont in Wilmington to Chicago and then to Hanford: putting aside elaborate dips and baths he tried soaking the bare slugs in molten solder, lowering the aluminum cans into the solder with tongs and canning the slugs submerged. The melting point of the aluminum was not much higher than the melting point of the solder, but with careful temperature control the canning technique worked.
Greenewalt then pushed production around the clock. Slugs accumulated in the reactor building faster than the loading crews could use them and Marshall and Fermi observed them there on one of their inspections:
Enrico and I went to the reactor building . . . to watch the loading. The slugs were brought to the floor in solid wooden blocks in which holes were drilled, each of a size to contain a slug, and the wooden blocks were stacked much as had been the slug-containing graphite bricks in CP-1. Idly I teased Fermi saying it looked like a chain-reacting pile. Fermi turned white, gasped, and reached for his slide rule. But after a couple of seconds he relaxed, realizing that under no circumstances could natural uranium and natural wood in any configuration cause a chain reaction.2103
Tuesday evening, September 26, 1944, the largest atomic pile yet assembled on earth was ready. It had reached dry criticality—the smaller loading at which it would have gone critical without cooling water if its operators had not restrained it with control rods—the previous Friday. Now the Columbia circulated through its 1,500 loaded aluminum tubes. “We arrived in the control room as the du Pont brass began to assemble,” Marshall remembers.2104 “The operators were all in place, well-rehearsed, with their start-up manuals on their desks.” Some of the observers had celebrated with good whiskey; their exhalations braced the air. Marshall and Fermi strolled the room checking readings. The operators withdrew the control rods in stages just as Fermi had once directed for CP-1; once again he calculated the neutron flux on his six-inch slide rule. Gradually gauges showed the cooling water warmed, flowing in at 50°F and out at 140° “And there it was, the first plutonium-production reactor operating smoothly and steadily and quietly. . . . Even in the control room one could hear the steady roaring sound of the high-pressure water rushing through the cooling tubes.”
The pile went critical a few minutes past midnight; by 2 A.M. it was operating at a higher level of power than any previous chain reaction. For the space of an hour all was well. Then Marshall remembers the operating engineers whispering to each other, adjusting control rods, whispering more urgently. “Something was wrong. The pile reactivity was steadily decreasing with time; the control rods had to be withdrawn continuously from the pile to hold it at 100 megawatts. The time came when the rods were completely withdrawn. The reactor power began to drop, down and down.”2105
Early Wednesday evening B pile died. Marshall and Fermi had slept by then and returned.2106 They talked over the mystery with the engineers, who first suspected a leaking tube or boron in the river water somehow plating out on the cladding. Fermi chose to remain open-minded. The charts, which seemed to show a straight-line failure, might be hiding the shallow curve of an exponential decline in reactivity, which would mean a fission product undetected in previous piles was poisoning the reaction.
Early Thursday morning the pile came back to life. By 7 A.M. it was running well above critical again. But twelve hours later it began another decline.
Princeton theoretician John A. Wheeler had counseled Crawford Greenewalt on pile physics since Du Pont first joined the project. He was stationed at Hanford now and he followed the second failure of the pile closely. He had been “concerned for months,” he writes, “about fission product poisons.” B pile’s heavy breathing convinced him such a poisoning had occurred. The mechanism would be compound: “A non-[neutron-]absorbing mother fission product of some hours’ half-life decays to a daughter dangerous to neutrons. This poison itself decays with a half-life of some hours into a third nuclear species, non-absorbing and possibly even stable.”2107 So the pile would chain-react, making the mother product; the mother product would decay to the daughter; as the volume of daughter product increased, absorbing neutrons, the pile would decline; when sufficient daughter product was present, enough neutrons would be absorbed to starve the chain reaction and the pile would shut down. Then the daughter product would decay to a non-absorbing third element; as it decayed the pile would stir; eventually too little daughter product would remain to inhibit the chain reaction and the pile would go critical again.
Fermi had left for the night; Wheeler on watch calculated the likely half-lives based on the blooming and fading of the pile. By morning he thought he needed two radioactivities with half-lives totaling about fifteen hours:
If this explanation made sense, then an inspection of the chart of nuclei showed that the mother had to be 6.68 hr [iodine]135 and the daughter 9.13 hr [xenon]135. Within an hour Fermi arrived with detailed reactivity data which checked this assignment. Within three hours two additional conclusions were clear. (a) The cross section for absorption of thermal neutrons by Xe135 was roughly 150 times that of the most absorptive nucleus previously known, [cadmium]113. (b) Almost every Xe135 nucleus formed in a high flux reactor would take a neutron out of circulation. Xenon
had thrust itself in as an unexpected and unwanted extra control rod. To override this poison more reactivity was needed.2108
Greenewalt called Samuel Allison in Chicago on Friday afternoon. Allison passed the bad news to Walter Zinn at Argonne, the laboratory in the forest south of Chicago where CP-1 was meant to be housed and where several piles now operated. Zinn had just shut down CP-3, a shielded sixfoot tank filled with 6.5 tons of heavy water in which 121 aluminum-clad uranium rods were suspended. Disbelieving, Zinn started the 300-kilowatt reactor up again and ran it at full power for twelve hours. It was primarily a research instrument and it had never been run so long at full power before. He found the xenon effect. Laborious calculations at Hanford over the next three days confirmed it.
Groves received the news acidly. He had ordered Compton to run CP-3 at full power full time to look for just such trouble. Ever the optimist, Compton apologized in the name of pure science: the mistake was regrettable but it had led to “a fundamentally new discovery regarding neutron properties of matter.”2109 He meant xenon’s consuming appetite for neutrons. Groves would have preferred to blaze trails less flamboyantly.
If Du Pont had built the Hanford production reactors to Eugene Wigner’s original specifications, which were elegantly economical, all three piles would have required complete rebuilding now. Fortunately Wheeler had fretted about fission-product poisoning. After the massive wooden shield blocks that formed the front and rear faces of the piles had been pressed, a year previously, he had advised the chemical company to increase the count of uranium channels for a margin of safety. Wigner’s 1,500 channels were arranged cylindrically; the corners of the cubical graphite stacks could accommodate another 504. That necessitated drilling out the shield blocks, which delayed construction and added millions to the cost. Du Pont had accepted the delay and drilled the extra channels. They were in place now when they were needed, although not yet connected to the water supply.