Portal de Periódicos da CAPES

Homo Citans and Carbon Allotropes: For an Ethics of Citation

2016; Wiley; Volume: 55; Issue: 37 Linguagem: Inglês

10.1002/anie.201600655

1521-3773

Roald Hoffmann, Artem A. Kabanov, Andrey A. Golov, Davide Μ. Proserpio,

Catalysis and Oxidation Reactions

Cite we must, cite we do. We cite because we are links in a chain, using properties and methods validated by others. We also cite to negotiate the anxiety of influence. And to be fair. After outlining the reasons for citation, we use two case studies of citation amnesia in the field of hypothetical carbon allotropes to present a computer-age search tool (SACADA) in that subsubfield. Finally, we advise on good search practice, including what to do if you miss a citation. In a classic of scholarship, "On the Shoulders of Giants,"1 Robert K. Merton, a great sociologist of science, traces the involuted history of a remark by Isaac Newton, "If I have seen further it is by standing on ye shoulders of Giants." Merton's book is also a humanist romp, a deliciously humorous dissection of scholarly pretensions, including his own. Merton follows The Aphorism, as he labels this apposite expression, back to Bernard of Chartres. And he documents its passage through a menagerie of more or less illustrious Gallic, Jewish, and Anglo-Saxon writers, to Newton, and past him to Claude Bernard, Bukharin, and Freud. Each should have cited the source of that seductive simile. Some did, some made up imaginary sources. Still others just tried to pass the expression off as their own creation, deigning citation unimportant. Meanwhile, The Aphorism kept its hold. Because it packages in a physical metaphor a truth: Even when we imagine (and want others to acknowledge) that our piece of hard-won knowledge is novel, better, or deeper than that which came before, we know that in fact it depends on what others have done previously. The novelist's citations are hidden, for PhD students to disinter. The scientist (male and female) is perforce and explicitly homo citans et citatus. This essay will first take a look at the reasons why appropriate citation is essential to the well-being of our profession. It will then pass from ideals to two case studies of failures in citation in one subfield of chemistry and physics, that of hypothetical carbon allotropes.2, 3 One of these cases has managed in three decades to accumulate an intricacy that took eight centuries for The Aphorism. Fault finding is easy; we will try to move beyond it in two ways. First, by giving down-to-earth suggestions for more-effective literature searching, and even advice on what to do if, God forbid, you should be guilty of omitting a crucial citation. And second, by introducing, at least in the specific subfield we discuss, a computer-age tool for avoiding making a fool of yourself. The reasons are numerous; here is a selection. 1. The tradition of scholarship. To mix similes, if not dwarves on the shoulders of giants, we are links in a chain. Citation is natural, as old as the laziness that is most often behind the failure to give credit where credit is due. European, African, and Asian scholarly cultures have left us with a tradition. This is worth upholding, and not just in Anatevka. Take a look the citation-studded orthography of a page of the Talmud (redacted 600 CE),4 or Confucius (551–479 BC), which cites the older texts of the Shangshu (Shu-ching, Book of Documents),5 and you will see the scholarly chain displayed. Anthony Grafton, in his delicious book, "The Footnote: A Curious History," traces the evolution of referencing in European historical scholarship from the Renaissance onwards.6 2. History. Yes, there are new things in this world—a gram of buckminsterfullerene, special relativity. But everything, absolutely everything, even the molecule and theory named, has antecedents. Reserving ethical considerations for a separate category, we are unabashedly and persistently curious about how the new came about. Knowledge is received, and we respect that. 3. Utility. In our scientific papers we use measurements by others. We also use definitions, concepts, and techniques. It is inefficient to repeat the calorimetry that determined the heat released in the burning of, say, 10 g of ethanol, or to rehearse the computation of the average position of an electron in a many-electron atom. So we cite the NIST Chemistry WebBook, an electronic resource maintained by the National Institute of Standards and Technology (which in turn cites and compares critically several experimental measurements of the heat of formation of ethanol).7 Or we put in a reference to Desclaux's classic tables of atomic calculations.8 4. Avoidance of duplication. We cite so we should not, need not, repeat unnecessarily what was done before. We wish to avoid duplication. Mind you, experimental measurements and computer proofs alike need checking. As we write this, there is a report of superconductivity at 200 K in a material ever-present in the chemistry labs of yore, hydrogen sulfide. You can be sure the experiment is being repeated in several laboratories; it's hard to do so, quite a different story from cuprate superconductors. The vaunted reproducibility of scientific measurements needs to be probed; none of this hurts the scientific method.9 Is the ninth synthesis of palau′amine publishable? Of course it will be, if done by a different route than the first such synthesis or is otherwise distinguished. Is the design of a carbon allotrope, claimed to be unprecedented, publishable if there is a previous paper on the same allotrope already in the literature? This is going to be a point in the two detailed case studies we will present. But we do not evade judgment: No, unless the second paper adds value to the design, say by calculating some property. 5. Establishing credentials. We cite so that our fellow chemists see that we know the literature. This is to establish the ground on which our discovery or insight may be seen as new or an advance in understanding. Crank papers are recognizable by their lack of citations, or by obsessive self-citation. A synthesis of taxol that does not cite Robert A. Holton's first synthesis of the molecule10 is not likely to be published. Citing the other eight syntheses listed in the Wikipedia on the synthesis lists puts us into the grey area of the too little/just enough/too much citation. By the time a good graduate student has written his or her PhD thesis, they can predict 90 % of the references in a paper published on the subject of that thesis. Which is little consolation when that paper is not by themselves. 6. Priority. We also cite others (and our previous work) to establish our own research as innovative, as different from what had been done previously. Danger lurks here. Even as there is a natural tendency to doubt our own powers or originality (are we speaking about ourselves?), observation of human nature seems to point in another direction: people tend to exaggerate the quality of what they have done. We hate those papers whose authors think that the way to establish their own claims to originality is by downgrading the partial understanding that came before, citing errors, omissions, all that went wrong. The psychology is transparent: when you have little to say, you begin by pointing out what other people got wrong. When you really have something new to say, you never do this, you just launch into what you have done. Yet priority is important. It might seem that questions of priority shouldn't be of concern to "science"; if Einstein had not derived in 1905 the equation E=mc2, surely someone else would have gotten it soon thereafter.11 Or not so soon. But this is a psychologically uninformed view for two reasons. First, given the paltry financial rewards of most scientific research, the ideas we have and the molecules we make are our "mind children". As such, they are priceless. And if someone does not cite them, it feels like a violation.12 Second, the truths we glean, the molecules we make, are universal. But our world is shaped by individuals creating the new. And they do so through interacting with each other in a certain chronology. The world is changed by hazard and circumstance. If Kekulé had not given us the structure of benzene 150 years ago, organic chemistry might have gone in a different direction, emphasizing molecules and reactions different from those that have shaped our chemical experience. Establishing priority is important. Especially so when utility, attested to through commercial value, enters the picture. The rewards are all that money can buy. Recognition of priority here is through the patent system, a legal fiat to exploit an invention in exchange for revealing it. Unfortunately, we do not have the space and time here to probe the fascinating logic of priority and citation in patents, and the practice of "examining" patents, so as to find prior art. Patents aside, the meaning of discovery is sometimes complex. We direct the reader to the fascinating history of the discovery and patenting of lasers,13 and to the story of the discovery of oxygen, the subject of a play by Carl Djerassi and Roald Hoffmann.14 7. Negotiating the "anxiety of influence". This category follows hard on the previous one. The term is taken from Harold Bloom's remarkable book,15 and we are grateful to Mario Biagioli for reminding us of it.16 We are a walking congeries of influences—of our parents, our teachers, the papers we have read. And somehow out of this patchwork of influences, aided by the workings of chance, we make the new: a molecule that in fact was not on earth before, a new theory. How can we be original, when so much went into what we do? Harold Bloom saw this tension as being most explicit in poets, for their forte was their originality. This is perhaps not that different from scientists. Bloom worked out a typology of strategies used by poets, explicit and subconscious, for denying, evading, and generally finding ways around the influence of other poets. These categories make for good reading, for one can see point-by-point parallels in them to the experience of contemporary chemists. What citations do for chemists is to allow them to negotiate the anxiety of influence. We cite work that piecewise precedes ours, we name the pieces, factual or conceptual, that enter what we do. Admittedly, quite selectively, and occasionally this gets us into trouble. The web of the 50-odd citations of a typical chemical paper reveals the influences, and at the same times serves to assuage the underlying doubts of the author about the originality of the work. 8. Connecting up the world. It's wonderful to see an organic synthesis marked not only by a brilliant plan (or a brilliant salvage job once the initial plan went awry), but also the pulling into use of a variety of synthetic methodologies that are scattered here and there across the literature. And for a theoretician, what a pleasure to have a way of thinking explain two or more puzzling problems, ones that no one would think of associating. R. B. Woodward talks of this in his Cope Lecture,17 and it is what made Roald Hoffmann devote his Nobel lecture to the isolobal analogy18 and not the history of orbital symmetry control of organic reactions. 9. Fairness. Such an old-fashioned concept, some might say. And is it not moored in the hypocritical class structure of colonial gentlemen? No, fairness is important. Ultimately, we cite to be fair. To acknowledge the achievements of those who came before. Behind this is an ideal, unvoiced, a shared conception of a body of reliable scientific knowledge built by many individual contributions. And a delight, in a troubled world, that there is a place where things are approximately as they should be. Science is a microsociety, and it was Merton who delineated for us how this society differs from others. One of its fundamental obligations (and satisfactions) is that one give credit where credit is due. As Mario Biagioli suggested to us, science is also a commons; citations are part of the fees one pays to use it. The American Chemical Society has a set of Ethical Guidelines to Publication of Chemical Research.19 On the subject of citation, these read: "An author should cite those publications that have been influential in determining the nature of the reported work and that will guide the reader quickly to the earlier work that is essential for understanding the present investigation. Except in a review, citation of work that will not be referred to in the reported research should be minimized. An author is obligated to perform a literature search to find, and then cite, the original publications that describe closely related work. For critical materials used in the work, proper citation to sources should also be made when these were supplied by a nonauthor." The European Association for Chemical and Molecular Sciences, in its "Ethical Guidelines for Publication in Journals and Reviews",20 writes: "Authors have the following responsibilities: 3.1 To gather and interpret data in an honest way. Editors, referees, readers and publishers have the right to assume that submitted (and published) manuscripts do not contain scientific dishonesty and/or fraud comprising among others fictitious data, plagiarized material, reference omissions, false priority statements, "hidden" multiple publication of the same data and incorrect authorship. 3.3 To give due recognition to published work relating to their submitted manuscript by way of correct reference and citation. All sources should be disclosed, and if a significant amount of other people's material is to be used, permission must be sought by the author in accordance with copyright law." The Council of Science Editors (CSE) published in 2006 and revised in 2012 a "White Paper on Promoting Integrity in Scientific Journal Publications".21 The feeling we get on reading its 81 pages is of CSE "running scared." As far as citations go, it has a whole section on citation manipulation, but good citation practice is mentioned only two times, as far as we can see: "The reviewer should ensure that an observation or argument that has been previously reported be accompanied by a relevant citation and should immediately alert the editor when he or she becomes aware of duplicate publication. …editors should require authors to 'Cite and reference other relevant published work on which the submitted work is based.'" These are good statements, but overall, the White Paper is disappointing. Codes and guidelines serve many functions: Even as we know that human beings may violate them, we set out in them ideal (yet realistic) norms. They certainly serve as more or less legalistic standards in case of violation, but they also have a moral and exhortatory purpose. We think it is as important to show young people in our profession what is right and good as it is to draw lines, prescribing violations. We wish our journals would say more. Actually, the ethics of citation is one place where the time-honored journal peer review process does well. Scientists are people, and people are people, which means that they are lazy, even as they are decent. Roald Hoffmann has looked carefully at unrefereed papers, such as those in arXiv22 (a repository of electronic preprints not taken to by chemists, but used widely by physicists, astronomers, and others). To him, these papers are often deficient in the quality of their illustrations and the fairness and adequacy of the way they cite the literature. We posit that this is not done maliciously, just out of laziness. In a well-refereed journal, the reviewers enter the context of criticism. It is their business to find fault, often to our annoyance when it is our papers that are being reviewed. If the referees are properly chosen—it is the editors' métier to do this—then this is where reviewers shine. They tell us of work we have missed, of work to which we should have given more credit. They keep up the ethics of citation. And they also get the authors to improve their drawings. Use engenders abuse, that is the human condition. Most citation sins are of omission, but there are some of commission. We avoid explicit discussion of scientific misconduct, fraud, and plagiarism.23 This is not because we are unaware of them, and not because they are unimportant. It's just that we don't want to be distracted from showing why citation really matters, and helping people to improve their citation practice. Yet, as Harriet Zuckerman wrote to us,24 it's not a bad thing to remind scientists of these blights upon our profession, and their apparently growing incidence. If citation is our subject, it is especially important that we bring to the reader a selection of previous thought on the subject. To show, among other things, how unoriginal we are… This we will do, but it's time to move to some specific illustrations. We then return eventually to the literature on citation practice and ethics. We are going to look at hypothetical structures for arguably the most important chemical element, carbon. Diamond and graphite have been known for millennia, but it is only 100 years ago that we learned (from single-crystal X-ray diffraction) the structure and metrics of cubic diamond (1)25 and graphite (2), at roughly the same time.26 Graphite is known in hexagonal and Bernal forms, which correspond to different stacking of what are now called graphene layers. Hexagonal diamond, or lonsdaleite, was established some years later, although its existence has been recently questioned.27 It became clear early on, at least on paper, that diamond and graphite are really the first members of a family of polytypes. The small dispersion energies involved in the aggregation of graphene layers imply a variety of stacking modes: AA, AB, ABC, and so on. For diamond, strong as the bonds between what we perceive as horizontal cyclohexanoid sheets are (part of that is an illusion created by our inability to integrate layers and true tetrahedral symmetry—there are chair cyclohexane rings, but no distinct axial and equatorial bonds in cubic diamond), one can also envisage polytypes. Lonsdaleite is the first in an infinite series. SiC makes them real, with a vengeance. Humanity's incendiary proclivities have no limit. So all along we have had a variety of pyrolytic carbon materials that resist structure determination. Out of a study of one of these, in what is perhaps the first suggestion of a hypothetical carbon allotrope, came H. L. Riley's 1946 structure 3.28 Structure 3 is clearly low in density and relatively unstrained. Riley's carbon, subsequently named "polybenzene,"29 will not have a chance against diamond at high pressure. But one day, chemists will find a clever way to make it at ambient pressure. Next in the roll of real allotropes came buckminsterfullerene. It was initially a gleam in theoreticians' eyes (none of whom cite each other),30 sadly ignored, and then discovered for real in gas-phase carbon-ablation studies31 and eventually synthesized in bulk.32 Now relatively cheap, the molecule, which is thermodynamically unstable but kinetically very persistent, has become a wonderful nexus of chemistry and physics. And of course, there is a family of larger and equally persistent fullerenes, as well as nanotubes. With no disrespect towards fullerenes, let us restrict ourselves to 3-dimensional infinite networks of elemental carbon. One of us (Roald Hoffmann)33 and Ivan V. Stankevich34 separately began to think about carbon allotropes in the early 1980s. We show here two structures the Hoffmann group came up with: 433 (with Tim Hughbanks, Miklos Kertesz, and Peter Bird) and 535 (with Mike Bucknum; further work on 3,4-connected nets with Ken Merz, Sandy Balaban).36 Where available, we provide bold three-letter symbols for nets (as suggested by M. O'Keeffe)37, here ths and tfi, respectively. tfi stands for three-four-#i, and there are also tfa, tfb and so on. Both ths and tfi have been observed in many coordination networks. The nets in structures 1, 2, 3 are called dia, hcb, pbz, respectively. As plane-wave-based calculations of extended systems became easy to do, the flood gates opened, so to speak. Computational chemistry and physics at an intermediate level has always been easier than experiment. We have counted several hundred papers suggesting "new" carbon allotropes. We put "new" in quotations, because, as we will show, many are repeats. And the titles of the papers that describe them don't stop with "new" or "novel"; enhancers such as "superhard", "remarkably stable", and "viable" abound. Let us not get into what is at work here, namely hype.38 We introduce next the first of two detailed case studies of citation amnesia in the field of carbon allotropes. We do so with trepidation, since the terminology of this subfield of solid-state chemistry quickly grows arcane. Given the sad compartmentalization of our molecular science, the extension in three dimensions of what on close examination are no more than the simplest organic building blocks, coupled with the nomenclature of networks, builds up in no time the kind of complexity that makes our mind cloud over. We know that this is so, and apologize to the reader for subjecting him or her to the detail necessary to establish our case (or Merton's in "On the Shoulders of Giants"). You will be forgiven if you skim over the next section. In August 2014, while working on building a database on carbon allotropes, DMP came across a paper in Phys. Rev. Lett. proposing a three-dimensional elemental carbon kagome lattice (CKL) and the structurally related interpenetrated graphene network (IGN).39 The authors did not report the coordinates—unfortunately this tendency is quite common in the literature of hypothetical carbon allotropes, which makes the results quite difficult to reproduce—but only the unit cell and some bond lengths. Nonetheless it was evident that the reported 4-connected net was the same as one called hcp-C3, which was reported in 1999 by P. A. Schultz, K. Leung and E. B. Stechel from Sandia National Laboratories.40 This net with small 3-rings was later observed in the zeolite nitridophosphate-1 and called NPO,41 hence the net name npo in the RCSR database (6). Digging deeper, we found that npo was reported in 1992 as net 36 in Figure 2 by M. O'Keeffe in his enumeration of uninodal nets with 3-rings,42 which refers back to J. V. Smith in 1979 (net 94 in his Figure 7),43 both reported as already described by A. F. Wells in 1977 (Figs 9.15a and 9.16 of the remarkable Wells book).44 The story just begins here. In 2003, npo is described as a hexagonal sphere packing 4/3/h3 by Sowa, Koch, and Fischer in their ongoing research on the complete derivation of all homogeneous sphere packing, which was started in the seventies.45 More recently, npo was re-examined as hcp-C3 in 2012.46 Furthermore, the structurally related IGN (see above) can be traced to a net called 3,4-bik-Cmcm47 and had been reported as ZGM-12.48, 49 Indeed the work on CKL and IGN focuses on properties not studied in the previous report on hcp-C3 and ZGM-12, but nonetheless, previous reports describing these structures should have been cited. But no one is safe in this field, for many networks were known before their re-discovery as carbon allotropes; in fact in the cited 1999 paper on hcp-C3, a body-centered tetragonal allotrope with 4-rings was reported as bct-C4. Electronic bibliographic searching of the literature was harder then, and so P. A. Schultz, K. Leung, and E. B. Stechel also missed that the same net was reported earlier by R. H. Baughman and D. S. Galvao,50 there called 8-tetra(2,2)tubulane, and was mentioned a few years later again by the same group, now calling it R2.51 What's in a name? The same net has also been called "rectangulated carbon",52 with proper reference to the Baughman works; simply "D",53 with no references except to A. F. Wells seminal works (see below); and (2,2) I4/mmm(2),54 with no specific reference. After a few years of silence, an important experimental study on cold-compressed graphite55 stirred the theoretical community to action, and the structure reappeared twice in 2010: in March as bct-C4, a "viable sp3 carbon",56 with reference to the 1999 Schultz paper, and in October as plain "bct-carbon" with the laconic sentence "This structure appears to be similar to that found in previous studies" referring to the 1999 Schultz and 2004 Strong papers.57 A veritable torrent of papers followed, recomputing the same structures (together with other hypothetical ones) and calculating all kind of properties,58-65, 66-68 without citing the older references. The earlier work of Baughman was acknowledged, together with Umemoto 2012, only in two papers.69 Three other publications cited the 1999 Schultz paper.70 Only one group, that of E. A. Belenkov, has carefully collected all references and atomic coordinates in a book71 and several papers.72 They used the nomenclature LA3 for bct-C4 and TB for npo. Moving to more chemical literature, it is easy to find that the network of bct-C4 is known as crb (7; the boron framework of CrB4 and related compounds).73, 74 In 1988, it was proposed as a tetragonal carbon net by J. K. Burdett and E. Canadell.75 But the net was known much earlier to the great structural chemist A. F. Wells; one finds it in his 1954 second paper of the series "The Geometrical Basis of Crystal Chemistry." There it is called Net 7 and illustrated in his Figure 6.76 In 1971, W. Fischer in his search for tetragonal sphere packing shows crb as 4/4/t5 in his Figure 4.77 And the zeolite expert J. V. Smith called it net 3.78 In 1977, Wells reported crb as (4.65)-a in his book44, together with npo, and in the paragraph dedicated to nets with point symbol79 m.n5 writes: "The nets 3.65 and 4.65 are particularly closely related, since they consist of "cylindrical" tunnels on the walls of which the plane 6-gon net is inscribed, three tunnels being linked by 3-gons in 3.65 and four by 4-gons in 4.65. The latter represents the arrangement of B atoms in CrB4 [our crb] and 3.65 [our npo] correspond to the positions of the centers of the spheres in the open sphere packing 42 (hexagonal variant) of Heesch and Laves…"80 Both nets are drawn as projections in Figure 9.15 and as stereopictures of handmade models in the Wells reference Figures 9.16 and 9.17. The latter 42 sphere packing npo is also illustrated in the cited 1954 Wells paper as a packing of tetrahedra in Figure 17. Like npo, crb was found as the underlying net for a body-centered tetragonal tectosilicate with the zeolite name BCT. Some final comments. As should be crystal-clear to anyone in the field, if you think up a new net, you'd be well advised to look for it in Wells.44, 81 And/or in more recent collections like RCSR [Ref. 37]. More on nets could be found in Ref. 82; for the use of net collections in the search of allotropes see Ref. 83.82, 83 The story we have elaborated also resembles one told a few years ago84 about another hypothetical carbon allotrope, this time a 3-connected one called srs. We also note, sadly, that the original older papers are much less cited/remembered that the two that stirred the field in 2010. The citations collected to date are: 23 for 1993 Baughman et al.50 and 25 for 1999 Schultz et al.,40 while 2010 Umemoto et al.56 has 124 citations and 2010 Zhou et al.57 has 60. This is not Merton's "obliteration by incorporation", it's something else. To summarize: Not quite Bernard of Chartres, but Heesch and Laves in 1933 (3.65-npo) and Wells in 1954 (4.65-crb). There is another case that we wish to put before you, but let us return first to citation practice. We have no pretensions to being original in delineating the reasons why we cite. The practice of citation itself is very old, as we've mentioned. Science came to citation late, its practices borrowed from established modes of scholarly argument in literature, religious dispute, and legal practice. Once, there were guides to good citation practice. Here is what Jacques Barzun and Henry Graff write in the 5th Edition of "The Modern Researcher" (the first edition, in 1957, had the same words): "Though the researcher is never entirely free from the necessity of accounting for his words through footnotes, it is not the writer who determines the number and fullness of these notes, but the subject at hand and the presumable audience. To the extent that footnotes communicate a part of the meaning and attest reliability, they are as important as any other part of the work. Hence an author should develop judgment about when and what to footnote. All quotations that are more than passing phrases or anonymous remarks require a footnote. So do all novel or startling assertions and all distinct elements in a demonstration or argument. Beyond this. A good rule is to write a note whenever you think an alert person might feel curiosity about the source of your remarks."85 Barzun and Graff's book, directed primarily at humanities scholarship, has a full, readable chapter on "The Rules of Citing." From which the above quotation is drawn. More directly aimed at scientists is E. Bright Wilson's 1952 advice in his "An Introduction to Scientific Research": "Ample references are important in order to enable the reader to obtain the immediate historical background of the problem and any previous attempts to solve it. References should also be given to more complete descriptions of the apparatus used or to descriptions of the apparatus or method from which the present one was evolved. Any outside data, facts, equations, or arguments employed should be supported by references. Finally, papers reaching similar or opposed conclusions should be listed… …In the whole matter of credit to others, including a proper perspective of the background in the introduction, references throughout the text, and credit at the end, a generous attitude is the most effective one from a purely selfish viewpoint. Scientists form one group which is practically never deceived by men who push themselves forward on