TH I S WE E K

NEW S
65 PAG E 1 3 8 3 1383
64 Official Dry spell for U.K.
63 reprieve for environmental
62 smallpox research
61
60
59
58 right course of action. “Until one completes
57 RESEARCH INTEGRITY an investigation, it’s premature to make any
56 decisions about the papers,” he says.
55
54
Pioneering Physics Papers Under Until last week, most physicists viewed
Schön and his collaborators with something
53 between envy and awe. Schön joined Bell
52
51
Suspicion for Data Manipulation Labs as a postdoc in 1998 to work with Bat-
logg and Kloc, setting out to study the way
50 Recent discoveries at Bell Laboratories—the Angeles, and director of the California electrical charges conduct through organic
49 research arm of Lucent Technologies in Mur- NanoSystems Institute: “It’s hard to under- crystals. They soon propelled Bell Labs be-
48 ray Hill, New Jersey—said to be of Nobel stand. I know these people. Most of them are yond all competition in the nascent field of
47 quality suddenly became mired in questions good, careful scientists.” “It’s a little over- organic transistor research.
46 last week. Outside researchers presented evi- whelming,” adds Lydia Sohn, a Princeton In a series of groundbreaking papers—
45 dence to Bell Labs management on 10 May University physicist who helped bring some most of which are not directly implicated in

Downloaded from www.sciencemag.org on June 13, 2011

44 suggesting possible manipulation of data in- of the discrepancies to light. “It’s just disturb- the current inquiry—the researchers showed
43 volving five papers published in Science, ing, and disappointing, and sad.” The noise that they could use devices called field ef-
42 Nature, and Applied Physics Letters over 2 pattern is particularly disturbing, says Charles fect transistors (FETs) to inject large num-
41 years. In response, Bell Labs officials said Lieber, a chemist and nanoscience expert at bers of electrical charges into organic mate-
40 that they are forming a committee of inde- Harvard University in Cambridge, Mas- rials. By changing the concentration of
39 pendent researchers to investigate. Their con- sachusetts: “It’s virtually impossible for me to charges, they could tune the electronic prop-
38 clusions may not be known for months, but believe that some of this wasn’t made up.” erties of the materials to behave in any num-
37 scientists who have seen the data are already Schön himself acknowledges that the ber of ways—as an insulator or semiconduc-
36 saying that the potential fallout from the in- similar noise pattern is “difficult to explain.” tor, a metal or superconductor—exhibiting a
35 vestigation could be devastating. But others affiliated with Bell Labs suggest malleability that had never been seen before.
34 The Bell Labs papers describe a series of privately that a systematic artifact in the The group also reported that organic
33 different experiments with organic conduc- measurement equipment might account for FETs displayed superconductivity at a tem-
32 tors, but portions of the figures perature higher than had ever been seen in an
31 seem almost identical, according organic material, revealed quantum signa-
30 to the physicists who raised sus- tures never before seen in organics, and
29 picions. Particularly puzzling, could be made to act as lasers and novel su-
28 they say, is the fact that two perconducting switches. Physicist Art
27 graphs show a pattern of “noise” Ramirez of Los Alamos National Laboratory
26 that looks identical, although it Image not in New Mexico, praising the work in an in-
25 should vary randomly. terview prior to the recent revelations, says
24 Bell Labs physicist Jan Hen-
available for “the string of papers is really outrageous” in
23 drik Schön is lead author on the online use. its success. “I don’t know of anything like
22 papers in question and the only it.” Heath says he was equally impressed: “I
21 author whose name appears on saw Batlogg talk about [the team’s results] a
20 all five. Among his most fre- year ago at a meeting in Venice. I was blown
19 quent co-authors are two col- out of my chair. I thought, ‘These guys are
18 leagues from Murray Hill, going to Stockholm.’ ”
17 Bertram Batlogg—a former In happier times. Bell Labs scientist Jan Hendrik Schön (left) The astounding results prompted groups
16 Bell Labs physicist who has with co-authors Christian Kloc and Bertram Batlogg. around the world to attempt to replicate the
15 since moved to the Swiss Feder- work. But to date, although other re-
14 al Institute of Technology in Zurich—and the similar noise trace, and that in the other searchers have made some progress, no one
13 Bell Labs chemist Christian Kloc. Schön cases, computer files containing similar data has reported duplicating any of the high-
12 told Science he stands behind his data and could have been mixed up. profile results. That troubled many in the
11 says it’s not surprising that experiments with Still, Lieber and others say the concerns community, says Cornell University physi-
10 similar devices produce similar-looking are so serious that the authors should imme- cist Paul McEuen, the first to notice the ap-
9 data. “We are trying as hard as we can to re- diately withdraw the papers in question. parent duplication of data.
8 peat those measurements,” Schön says. “I “They should be retracted until they can be Some physicists grew more concerned
7 am convinced they will show I haven’t done duplicated,” Lieber says. But Cherry Mur- last fall when Schön published a pair of pa-
6 anything wrong.” Co-authors on the five pa- ray, who heads physical sciences research at pers on a different topic in Nature and
5 pers either declined public comment or Bell Labs, says the company won’t take any Science with Bell Labs colleagues Hong
4 could not be reached. action until the external review committee Meng and Zhenan Bao. In the first, pub-
CREDIT: BELL LABS

3 Many scientists have reacted with dis- reaches its conclusion. “We are not rushing lished in the 18 October 2001 issue of Na-
2 belief. “I’m shocked,” says James Heath, a to judgment,” Murray adds. Science’s editor- ture, the researchers reported making a novel
1 chemist at the University of California, Los in-chief, Donald Kennedy, says that’s the type of transistor in which the key charge-

1376 24 MAY 2002 VOL 296 SCIENCE www.sciencemag.org

 FOCUS
65 LEAD STORY 1384 1387 1394
64 Disputes Last chance to Profile: NCI’s
63 over early fix science on new chief
62 life the station?
61
60
59
58 conducting layer was composed of a single alter the ability of charges to conduct er uncomfortable coincidence. In this case,
57 layer of an organic conductor. In the Science through a simple circuit of two FETs. The six data traces in a figure in the 3 November
56 paper, published in the 7 December issue devices in the 11 February 2000 Science pa- 2000 issue of Science appear virtually identi-
55 (p. 2138), they reported diluting that charge- per reportedly contain different materials in cal to ones in the 4 December 2000 issue of
54 conducting layer with nonconducting insu- the key charge-conducting channel in each Applied Physics Letters (APL). But, whereas
53 lating molecules, allowing them to track the FET and a different physical geometry, both the Science paper tracked the conductivity of
52 electrical conductivity in a transistor through of which should cause these devices to con- a light-emitting organic material known as
51 a single molecule. Together, the results re- duct current differently from devices de- α-6T in a FET, the APL paper followed the
50 ceived international press attention as a tri- scribed in the other papers, says McEuen. conductivity in a non–light-emitting FET
49 umph of molecular-scale electronics. But But when McEuen resized the figures and made with an organic compound called pery-
48 McEuen says the papers puzzled researchers overlaid the data, he found that the seeming- lene. Moreover, whereas the FETs in the
47 because, despite the novel architecture of the ly uninteresting background data on the α-6T figure are “n-channel” devices, which
46 devices, they seemed to conduct current in a right portion of the figures looked similar. conduct negatively charged electrons, those
45 manner similar to traditional FETs. “The noise looks almost identical, bumps in the perylene figure are “p-channel” de-

Downloaded from www.sciencemag.org on June 13, 2011

44 Last month, a more troubling aspect came and all,” McEuen says. “This is very confus- vices, which conduct positive charges. Ac-
43 to light: Researchers noted that fig- cording to McEuen, most physi-
42 ures describing the conductivity in cists believe that should cause the
41 the two papers appeared identical, devices to conduct current in a
40 even though the measurements were slightly different manner. “They
39 supposedly done at temperatures dif- overlap, noise and all,” says
38 ferent enough to affect the results. Science 287, 1022 (11 February 2000) McEuen. “They are identical,” ex-
37 According to Princeton’s Sohn, sev- cept that the labels on the axes re-
36 eral Bell Labs researchers pointed ferring to the voltages applied to
35 out the identical f igures to her, the devices have an opposite sign,
34 McEuen, and others. “Collectively, he adds.
33 people at Bell [Labs] were nervous,” Nature 413, 713 (18 October 2001)
Taken together, the three exam-
32 says Sohn, although she declines to ples are deeply troubling, says Leo
31 identify who tipped her off. Word of Kouwenhoven, a physicist at Delft
30 the duplicate f igures began to University of Technology in the
29 spread. And late last month, Lieber Netherlands. “I think that it is very
28 and Harvard physicist Charles Mar- Science 294, 2138 (7 December 2001) worrisome,” Kouwenhoven says.
27 cus contacted manuscript editors at “I can imagine you switch one fig-
26 Science and Nature informing them Striking resemblance. Published data from studies of different de- ure by mistake. It’s hard to imag-
25 of the apparent problem. vices revealed a similarity in recorded “noise.” Schön says the bottom ine how you switch 10 figures.”
24 A few days later, even before figure was sent to Science by mistake (see correction, p. 1400). Schön says that because his pa-
23 he had heard from Science, Schön pers report the conductivity in
22 e-mailed Science associate editor Ian Os- ing and disturbing. They should be vaguely FETs, “I would expect them to be very simi-
21 borne to say there had been a mix-up and similar, maybe roughly similar. But certain- lar.” He declines to comment on other specif-
20 that the wrong figure had mistakenly been ly the noise shouldn’t be the same,” McEuen ic issues. Bell Labs’ Murray declines to com-
19 incorporated in the Science paper. Schön says. “This knocked me for a loop.” ment on specifics as well, but adds: “I am
18 also sent along a new figure, which appears He quickly got another shock. McEuen very concerned. … This deserves a full and
17 as a correction on page 1400 of this issue. noticed that the same 11 February 2000 complete investigation.”
16 But Sohn says the mix-up explanation Science and 18 October 2001 Nature papers A five-member committee headed by
15 just didn’t sit well with her or McEuen. contained another similar figure, which also Stanford University physicist Malcolm
14 “Paul said, ‘Lydia, I’m just going to look at closely resembled a figure in a third paper, Beasley began an investigation last Friday.
13 the data, the figures,’ ” says Sohn. And on from the 28 April 2000 issue of Science. All Beasley says he cannot estimate how long it
12 Thursday, 9 May, McEuen stayed up much three papers describe different organic con- will take or whether it will be broadened to
11 of the night looking through Schön’s ductors. Yet, if one ignores the labels, sever- look at data presented in other papers.
10 Science and Nature papers and found what al of the data traces appear very similar. Schön has been the first author on 17 papers
9 he calls two “disturbing” coincidences. “There is no physical reason why they in Science and Nature alone in the last 2.5
8 The first involves the same duplicate fig- should be so similar,” McEuen says. years and a co-author on dozens elsewhere.
7 ures that prompted the heads-up from The next day, 10 May, McEuen says he Beasley says: “We are hoping for something
6 Lieber and Marcus. McEuen noticed a close and Sohn were concerned enough that they by the end of summer.”
5 resemblance with yet another figure, this contacted Schön, Batlogg, managers at Bell McEuen, for one, believes Bell Labs is
4 one in the 11 February 2000 issue of Labs, and manuscript editors at Science and taking the proper first step. “Beasley has
3 Science (see figures above). The figures Nature. He says that all involved expressed great stature in the community. … Every-
2 show how changes in an electrical voltage deep concern. body wants to get to the truth.”
1 applied to a control electrode called a gate A couple of days later, Sohn found anoth- –ROBERT F. SERVICE

www.sciencemag.org SCIENCE VOL 296 24 MAY 2002 1377