Misconduct in Science: Physical Science Cases
Misconduct in Science: PHYSICAL SCIENCE CASES
In the year 2002, two cases of scientific misconduct by physicists received prominent attention. One involved a young scientist at Bell Laboratories named Jan Hendrick Schön, and the other a researcher at Lawrence Berkeley National Laboratory (LBNL) named Victor Ninov. This was a surprising development because nearly all cases that had arisen prior had been in biology, biomedicine, and related fields. The questions arose: Why had the physical sciences previously seemed immune to this kind of misbehavior, and what had suddenly changed?
Before responding to these questions, it is important to consider the scope of misconduct and some charges of historical significance. Misconduct is a narrower concept than ethics in science. There are many ethical issues having to do with conflict of interest, not properly sharing credit, not hyping results or prospects in grant applications, covering up misconduct, reprisals against whistleblowers or malicious allegations of misconduct, violation of due process in handling misconduct cases, treating graduates students fairly, and so on, that are not part of scientific misconduct in the strict sense. During the 1980s and 1990s, after considerable debate, scientific misconduct was carefully defined as fabrication, falsification, or plagiarism (FFP) of results. It is this FFP definition that is most appropriate to bring to bear on considering misconduct in science because, without a well-crafted understanding, many activities can unfairly be called misconduct when they should more properly be called moral weaknesses or improper behavior. This is not to downplay the importance of a host of ethical issues, but simply to be clear in discussion.
Until the two physics cases arose, the fabrication and falsification type of misconduct seemed to be confined to biology and related sciences. A considerable number of such cases surfaced during the 1980s and 1990s. From those cases a pattern emerged of preconditions for such misconduct. First the scientists who commit misconduct are under career pressure. Of course all scientists are almost always under career pressure, but the point is they engage in misconduct for motives more subtle than simple monetary gain. Second scientists do not purposely insert falsehoods into the scientific record, but rather fabricate or falsify data, giving a result they believe to be true without taking the time to do the science properly. In other words, this kind of misconduct is always a violation of the scientific method, never purposely a corruption of the body of scientific knowledge. Such is almost certainly the case because even corrupt scientists believe that science is self-correcting, and a wrong result will eventually be found out. Finally misconduct occurs in fields where reproducibility is not very precise. This last point explains why the physical sciences seemed immune to such behavior while biology did not. If two organisms as identical as they can be made, for example, two transgenic mice, are exposed to the same carcinogen under the same conditions, they are not expected to produce the same tumor, at the same time, in the same place. This is an example of what is known as biological variability. Experiments in biology are not as precisely reproducible as those in physics generally are supposed to be, so a biologist disposed to cheat does not fear that someone else repeating the same experiment will find it out quickly. The two physics cases that arose in 2002 pose a severe test of this pattern.
Dr. Victor Ninov was a leader of the Berkeley Gas Filled Separator (BGS) group at LBNL. Ninov had joined LBNL in 1997 after a stint at the rival GSI, German acronym for the Laboratory for Heavy Ion Research, in Darmstadt. The BGS is a device designed to sort through the debris of nuclear collisions between a stationary target, and particles that are accelerated in the LBNL 88-inch cyclotron. The Berkeley laboratory has a distinguished history of discovering heavy, radioactive elements by this means. However, although even heavier elements were believed to be possible, it was widely thought that this so called cold fusion method of producing new elements had pretty well run its course and entirely new approaches would be needed. This would not have come as good news to Dr. Ninov and the BGS group.
The possibility of a reprieve from this situation arose when a theory published by Robert Smolańczuk predicted a highly enhanced probability of creating superheavy element 118 if projectiles consisting of an isotope of krypton were fired with the right energy into a lead target. The signature of such an event would be a chain of subsequent events, in which the original nucleus shed alpha particles at times and with energies predicted by the theory. This was just the kind of experiment the BGS was designed to do. In May 1999, a paper was submitted for publication, and a few days later a press release was issued by LBNL, both announcing that three instances of decay chains characteristic of element 118 had been observed.
By international agreement, new elements are not official until their discovery has been reproduced. The GSI in Germany and a research group in Japan immediately undertook to reproduce the new result, but both failed. The BGS group did a new series of experiments, and in 2001, produced a fourth signature decay chain. But by now, suspicions had been aroused. A series of investigations ensued determining that the data for all four significant decay chains had been fabricated, and that Ninov was the only person in a position to have done it. The entire BGS group was criticized for not checking the raw data more carefully in what would have been a major scientific discovery, but Ninov alone was found guilty of scientific misconduct. Furthermore the investigations uncovered that in the earlier discovery of element 112 at the GSI in Darmstadt, a discovery that was real and that had been reproduced, data had nevertheless been fabricated, and Ninov had been a member of the group at the time. Ninov was fired by LBNL.
The other physics case involved Jan Hendrick Schön, a young superstar who had recently arrived at Bell Laboratories in Murray Hill, New Jersey, after completing his Ph.D. at the University of Konstanz in Germany. Schön, a postdoctoral member in the research group of a well known and highly respected physicist named Bertram Batlogg, did experiments in which an intense electric field drew electrons to the interface between a semiconducting material and an insulating layer. Such devices are known as MOSFETs (metal-oxide semiconductor field effect transistors) and, using conventional semiconductors such as silicon, they had been the mainstay of the electronics industry for years. The Batlogg group's work involved substituting exotic materials such as organic crystals for the silicon, and using the field effect to alter their properties. Schön's results seemed truly spectacular. In a period of only two years, together with a total of some twenty collaborators, he turned out eighty research papers announcing remarkable breakthroughs that many others had attempted but failed to achieve.
Then questions arose. In some cases, the data just looked too good to be true. In other cases, completely independent curves had identical noise, little glitches in the data that are inevitable in any real experiment, but that should be random, meaning no two experimental curves should be identical to one another. These anomalies and others were reported to the management of Bell Laboratories, which, in May 2002, announced that it had appointed a committee, headed by Malcolm Beasley of Stanford University, to investigate. It also announced that the committee's report would be made public. By contrast, the report of the committee that investigated the Ninov case at LBNL is regarded as a confidential personnel matter, and has not been released to the public.
The Beasley committee, whose report was issued at the end of September 2002, chose to investigate some twenty-four specific allegations, and found that Schön had committed misconduct in at least sixteen of them. They also decided that Schön alone, and none of his collaborators, was responsible. The insulating layer in the MOSFET was the key to the whole affair. The process by which the insulating layer is laid down on the semiconductor is called sputtering. Schön, who started his collaboration with Batlogg when he was still a graduate student, had tried his hand at sputtering an insulating layer on to one of the group's exotic samples in a very modest apparatus at his university, in Konstanz. The insulating layer proved to be much more robust than those that others were able to make. It allowed stronger electric fields to be applied, producing results that no one else could achieve. Because sputtering involves complex processes that are not well understood or controlled, it seemed believable to Schön's collaborators that for unknown reasons, the apparatus in Konstanz could make better insulating layers than could be made anywhere else. Thus it was believable that Schön could get experimental results no one else could produce. People and samples shuttled back and forth between Murray Hill and Konstanz, but all the sputtering was done in the magic machine at Konstanz and Schön alone made nearly all the measurements. The results were, literally, too good to be true. When the Beasley report came out, Schön was immediately fired by Bell Laboratories.
The two physics cases of 2002 can be analyzed in light of the pattern, described above, that had emerged from previous cases of scientific misconduct. The three necessary (but certainly not sufficient) factors that seemed to be present whenever misconduct occurred were career pressure, belief in knowing the answer before the experiment was performed, and the expectation that the experiment was not easily and precisely reproducible. All three factors were unmistakably present in the Schön case. The atmosphere at a place like Bell Laboratories puts great pressure on scientists to succeed. The effects that Schön and his collaborators reported were widely believed to be possible, even though no one else had managed to obtain them yet. In fact, in an addendum to the Beasley report, Schön admits that he made some mistakes, but says he still believes all the effects he reported were real. And finally, the field is notorious for its lack of reproducibility. The problem lies not only in the sputtering, but also in the difficulties of preparing good samples of the exotic materials involved. If an experiment fails to reproduce a given result, it does not necessarily show the result was mistaken, it just means the experiment was performed on an imperfect sample Thus a failure to reproduce has no significance at all.
The Ninov case is more subtle, and requires some speculation as to cause. Certainly Ninov and the BGS group were under pressure to produce something new because their measurement technique seemed to have run its course, giving them less leverage to get expensive beam time on the 88-inch cyclotron and perhaps even threatening the continued existence of the group itself. The theory by Smolańczuk gave the group new hope, and quite possibly, Ninov came to believe in it because he needed to. The question of reproducibility appears to pose a contradiction, though, because the field is one in which results must be reproduced before they are official. Ninov seems to have turned the irreproducibility factor upside down. If he believed that element 118 existed, he also must have believed that its discovery could be reproduced, and, when that happened, that he and his group would get credit for the original discovery. This, of course, is exactly what occurred in the discovery of element 112, an experiment he had also been involved in; data had been faked, but the discovery turned out to be real.
These cases demonstrate that the physical sciences never were immune to FFP misconduct, and that nothing has suddenly changed. The necessary factors may line up less often than in some other areas of science, but when they do, misconduct can follow just as in other fields.
DAVID L. GOODSTEIN
Beasley, Malcolm R., et al. (2002). "Report of the Investigation Committee on the Possibility of Scientific Misconduct in the work of Hendrik Schon and Coauthors." Unpublished.
Broad, William, and Nicholas Wade. (1982). Betrayers of the Truth. New York: Simon and Schuster.
Goodstein, David. (1994). "Pariah Science—Whatever Happened to Cold Fusion?" American Scholar 63(4): 527–541.
Goodstein, David. (2001). "In the Case of Robert Andrews Millikan." American Scientist 89(1): 54–60.
Goodstein, David, and James Woodward. (1996). "Conduct, Misconduct and the Structure of Science." American Scientist 84(5): 479–490.
Vogt, R., et al. (2002). "Report, Committee on the Formal Investigation of Alleged Scientific Misconduct by LBNL Staff Scientist Dr. Victor Ninov." Unpublished.