I spent much of the summer of 2002 looking into allegations of scientific misconduct against Jan Hendrik Schön, who had until then been a golden boy of organic semiconductor research at Bell Labs. I was part of a five-person "blue ribbon" committee headed by Mac Beasley of Stanford, which also included Supriyo Datta of Purdue, Herwig Kogelnik of Bell Labs, and Nobel-prizewinner Herb Kroemer of UCSB. In the end, we found the evidence compelling that many of Hendrik's results were fictitious or deceptive, putting a rather definitive asterisk on his meteoric career. The investigation was at times dispiriting, but in the end highly satisfying, in that we were able to help purge the scientific literature and community of some highly erroneous results. I certainly don't regret the time and effort I put into the investigation.
As described in Eugenie Samuel Reich's recent book, Plastic Fantastic (Amazon, B&N), however, I was involved in allegations earlier, in the fall of 2001. Specifically, when Hendrik moved from organic crystals to molecular-gate length transistors, his claims began to violate known laws of transistor physics, drawing a great deal of critical attention. My colleagues and I at Agere Systems, who had recently been spun out from Bell Labs, had many serious questions about the results, as did researchers within Bell Labs and elsewhere. I arranged a special seminar by Hendrik, at which Ashraf Alam noticed that some data were unreasonably smooth. Further investigation (described in technical detail in our final committee report, p. E-39) showed that there it was almost inconceivable that the data could come from real samples: it had to have been made up (not merely massaged so it looked better). This wasn't the only problem, just the hardest one to explain away as sloppiness or ignorance.
At this point I started heavy discussions with managers at Bell Labs, as described in Reich's book. Why did I not make it public, as others did with the copied data that started the formal investigation in April? Two main reasons: First, I thought there was a small chance I was wrong, and that a public allegation would still have lasting consequences and be hard to withdraw, so it would be best to check first if there was an innocent explanation. Second, the consequences for Bell Labs looked to be so severe that I thought the managers (whom I knew personally from my days at Bell Labs) deserved a chance to do some damage control (I didn't intend to help in any cover-up, of course).
I don't regret the decision to first raise the issue privately. What I regret was my response when Hendrik concocted an "explanation" for the smoothness, claiming that the histogram bars were so wide as to obscure data that were less smooth. At that point I emailed the managers saying that I considered the issue "CLOSED." That was a mistake. In reality, I still felt strongly that there was something horribly wrong, although there was a tiny patch of daylight in an otherwise airtight case for intentional fraud. (And as discovered by a Bell Labs manager months later, Hendrik's explanation was itself provably fraudulent.) I had seriously underestimated how willing the Bell managers would be to threaten the goose that had laid so many golden eggs for them. By removing my charge from the table, I gave them the excuse to delay a serious inquiry for another five months, during which Hendrik's false claims would continue to divert good researchers as well as job offers and prizes. That is my regret.
In 2006 I gave an hour-plus-long talk on this subject at Cornell's Center for Nanoscale Systems. Video is here.