Three summers ago, much attention was paid to a search for new superheavy nuclei at the Lawrence Berkeley National Laboratory’s 88-inch cyclotron. In June 1999, the LBNL heavy-element search team announced the discovery of elements 116 and 118 (see Physics Today, Physics Today 0031-9228 52
At a meeting of LBNL employees in June of this year, director Charles Shank announced that the laboratory had recently disciplined one of the members of the team for “scientific misconduct.” A yearlong internal investigation had convinced the laboratory’s directorate that the evidence for the creation of element 118 and its decay sequence through element 116 in the 1999 experiment had, in fact, been surreptitiously fabricated by one of the experimenters.
“I’m proud of the intensity and professionalism of the [internal review that got] to the bottom of this,” Shank told the LBNL staff. “There is nothing more important for a laboratory than its scientific integrity.” Although Shank did not reveal details of the investigation or the name of the accused at the employees’ meeting, the press quickly jumped on the story and identified him as Victor Ninov, the first author of the 1999 paper.
Last month, the laboratory finally made the report of the internal committee that carried out the formal investigation of Ninov’s alleged misconduct available to journalists who invoked California’s Public Records Act. Ninov, who was placed on indefinite paid leave six months before being finally dismissed in May, emphatically denies the accusations. He has initiated a grievance proceeding that will be adjudicated by an arbitrator chosen by him and the lab.
LBNL had, in fact, already retracted the element-118 discovery claim in July 2001, after various experimental groups around the world, including the Berkeley team itself, were unable to reproduce the 1999 result (see Physics Today, Physics Today 0031-9228 54
At that point, there was as yet no serious suspicion of malfeasance. But the retraction—in an LBNL press release—was ominously unconventional. The customary retraction was sent for publication to the editors of Physical Review Letters in July 2001. But the editors declined to publish it, noting that one of the original authors—Ninov—had removed his name from the retraction. “I asked that my name be removed,” Ninov told us, “because the retraction had been written and sent to the editors behind my back. That’s when I became an object of suspicion.”
A retraction of sorts did finally appear in the journal a year later (15 July 2002) as an “Editorial Note,” with an introductory explanation pointing out that “all but one of the authors of the original Letter have asked us to publish the following retraction.” 2
Island of stability
What had attracted so much attention to the 1999 Berkeley paper was the giant step it appeared to represent in the quest for the “island of stability,” a conjectured region of relatively long-lived nuclei at atomic numbers somewhere around Z = 120. The highest-Z nucleus known at the time was element 114, discovered just a few months earlier at Russia’s Joint Institute for Nuclear Research, in Dubna. Conventional extrapolations from the observed production rates for elements 110–114 predicted that element 118 could not be produced at a useful rate at any existing nuclear physics facility.
But Warsaw University theorist Robert Smolańczuk, then a visiting scholar at LBNL, thought otherwise. He predicted that the cross section for producing element 118 with a low-energy krypton (Z = 36) beam hitting a lead (Z = 82) target might be perhaps 105 times higher than what the usual extrapolation predicts.
So the Berkeley team directed an intense beam of 450-MeV 86Kr19+ ions from the 88-inch cyclotron at an array of very thin 208 Pb targets mounted on an innovative detector system called the Berkeley gas-filled separator (BGS). (See figure 1.) The system was configured to capture, with high efficiency, any ion of element 118 in a downstream silicon detector plane that would record precisely when and where it came to rest, as well as the times and energies of the telltale sequence of alpha particles emitted in its chain of decays to lower Z.
The Berkeley team’s 1999 paper claimed to have found three atoms of element 118 in 10 days of running. The reported evidence, reproduced in figure 2, was the observation of 17 of the 18 alphas from the three decay chains from 293118 down to 269106 (seaborgium).
These three neat alpha-decay sequences, so impressive when they were first reported, are now exhibit A against Ninov. The LBNL formal investigation committee has concluded that these sequences were largely fabricated by him. And Ninov’s coauthors sadly agree. “After all this digging, we now know how and when he did it,” says BGS team leader Kenneth Gregorich. “But we’ve given up trying to figure out why.”
Suspicion of deliberate fabrication did not arise before the BGS team undertook a confirmatory run in April 2001 after experiments in Germany, Japan, and France had failed to find evidence of element 118. During the new run, coworkers recall, Ninov announced at a team meeting that his analysis program, which was supposed to search the raw data files for evidence of alphas correlated in time and space, had found a new 118 decay sequence. But several of his colleagues, using their own computer programs, simply couldn’t find the new event in the data file. In the 1999 run, Ninov—as the team’s acknowledged expert—was the only one working directly from the raw data files. Two years later, that had changed, and the discrepancy between what Ninov was claiming and what the others could find quickly became a grave concern.
At that point, a working group chaired by Darleane Hoffman, a senior member of the BGS team, examined both the 1999 and 2001 data files, and found that none of them contained a record of any 118 decay sequence. Now, it was clear, the issue had to be examined by experts from outside the team. In June 2001, Lee Schroeder, director of LBNL’s nuclear-science division, convened a technical-review committee headed by Gerald Lynch, a veteran Berkeley high-energy experimenter. Four months of detailed digging by the Lynch panel into the records of the BGS group’s experiments produced a report that was then passed on for evaluation to Stewart Loken, former head of LBNL’s computing division.
A formal investigation
After interviewing Ninov and other principals in the light of the Lynch technical report, Loken recommended that the laboratory proceed to a formal investigation of Ninov’s conduct under the provisions of LBNL’s stated policy on integrity in research. Appointed in November, the formal-investigation committee was chaired by Caltech physicist Rochus Vogt. Two of its Berkeley members, Andrew Sessler and George Trilling, are past presidents of the American Physical Society.
Not all of the allegations the Vogt committee had to consider originated in Berkeley. From GSI, the German heavy-ion research laboratory in Darmstadt, came evidence that Ninov may have been fabricating alpha-decay chains since 1994. Last December, Sigurd Hofmann and coworkers at GSI reported a reanalysis of their 1994–96 experiments on elements 110, 111, and 112. Ninov, who got his PhD at GSI in 1992, was a major participant in those experiments before he came to Berkeley in 1997. The reanalysis, undertaken to check consistency with new GSI data, revealed that two alpha-decay chains reported in the original papers were not to be found in the data. 3
“We have excluded the possibility that this was simply a mistake,” Hofmann told us. “Our new paper doesn’t explicitly name Ninov, but he was, in fact, in charge of our data analysis at that time. To mistrust a close collaborator and good friend of many years was unthinkable. I can’t understand why he would have done it. The genuine data were so good. In fact, despite the revelation of the fabrications, we have not had to retract any of our substantive conclusions for elements 110 to 112.”
Ninov said to us: “Once you’re bookmarked as a fabricator, the avalanche begins. If you’re trying to frame someone, you also go after his past.” Ninov submitted a written statement challenging the Lynch panel’s findings, and he appeared once before the Vogt committee. Although he declined an invitation for a second meeting, he did respond in writing—through his lawyers—to questions from the committee.
Clear and convincing evidence
In March, the Vogt committee reported its findings to Schroeder and Shank. “We find clear and convincing evidence that the data in 1999, upon which the reported discovery was based, were fabricated,” states the report. It expresses “regret that our findings revealed intentional fabrication … instead of honest error or honest differences in interpretation.”
Responding to Ninov’s suggestion that some unspecified malefactor may have removed genuine events from the data at a later time, the committee wrote that “there is no evidence to suggest that the raw data files that exist now are in any way different from those that were produced in the course of the experiment.” Nor, in the committee’s view, could the spurious events simply have resulted from bugs in the analysis program. “There is clear evidence,” the Vogt committee continues, “that Dr. Ninov [carried] out this fabrication. … If anyone else had done [it], Dr. Ninov would almost surely have detected it.”
Ninov claims that “the report makes it clear that the only way I could prove my innocence is by pointing a finger at someone else. But that I refuse to do.”
Although the Vogt report accuses no one else in the BGS group of complicity in the data fabrication, it does close with a rather harsh criticism of the team’s procedures in the 1999 experiment: “[We] find it incredible that no one in the group, other than Ninov, examined the original data to confirm the purported discovery of element 118.”
Deliberate fraud is, happily, rare in physics research. But mistakes and unconscious bias are not rare. “The message is that experimenters must exercise vigilance, not so much against fabrication, but against honest error and misjudgement,” says LBNL Deputy Director Pier Oddone. “And against these, the first line of defense is vigorous, independent checking within an experimental group.”
One member of the BGS team, Albert Ghiorso, has been discovering transuranic elements for half a century. “When you’re dealing with a complicated apparatus like the gas-filled separator, you tend to trust the group’s best guy to do it right,” he told us. “That was our first mistake. Ninov was really good at everything: theory, experiment, computers, building things. And he was very personable. We didn’t want to believe the accusations—until the evidence was overwhelming. It’s a tragedy.”