摘要:

OFFICERS AND COUNCIL OF THE FARADAY DIVISION 1994-95 President Prof. J. P. Simons (Oxford) Vice- Presidents who haw held office as President Prof. A. D. Buckingham (Cambridge) Prof. R. Parsons (Southampton) Prof. P. Gray (Cambridge) Prof. N. Sheppard (Norwich) Prof. R. H. Ottewill (Bristol) Vice -Presidents Prof. A. Carrington (Southampton) Prof. M. W. Roberts (Cardiff) Prof. M. A. Chesters (Nottingham) Prof. I. W. M. Smith (Birmingham) Prof. R. N. Dixon (Bristol) Prof. F. S. Stone (Bath) Prof. M. J. Pilling (Leeds) Ordinary Members Dr. 0. C. Clary (Cambridge) Prof. D. A. Parkes (Chester) Prof. P. W. Fowler (Exeter) Dr. S. L. Price (London) Prof. A. Hamnett (Newcastle) Prof. S. K. Scott (Leeds) Prof. J. Lyklema (Wageningen) Prof. A. J. Stace (Sussex) Dr.W. Mackrodt (St. Andrews) Prof. Sir John Meurig Thomas (London) Honorary Secretary Prof. M. J. Pilling (Leeds) Honorary Treasurer Prof. F. S. Stone (Bath) Secretary Mrs. Y. A. Fish Faraday Editorial Board Prof. M. N. R. Ashfold (Bristol) (Chairman) Prof. A. R. Hillman (Leicester) Prof. J. A. Beswick (Paris) Dr. J. Holzwarth (Berlin) Dr. D. C. Clary (Cambridge) Prof. D. Langevin (Bordeaux) Dr. L. R. Fisher (Bristol) Prof. S. K. Scott (Leeds) Prof. B. E. Hayden (Southampton) Dr. R. K. Thomas (Oxford) Prof. J. S. Higgins (London) Scientific Editor Prof. A. R. Hillman Managing Editor Dr. R. A. Whitelock Production Editor Mrs. S. Shah Assistant Production Editor Dr. J. C. Thorn The Faraday Divisionof the Royal Society of Chemistry,previously The Faraday Society founded in 1903 to promote the study of Sciences lying between Chemistry Physics and Biology Faraday Discussions (ISSN 0301-7249) is published triannually by the Royal Society of Chemistry Thomas Graham House Science Park Milton Road Cambridge CB4 4WF England.1995 Annual subscription rate EC €207.00 Rest of World €217.00 USA $380.00 including air-speeded delivery Canada €217 + GST. Change of address and orders with payment in advance to The Royal Society of Chemistry Turpin Distribution Services Ltd. Blackhorse Road Letchworth Herts SG6 lHN UK. NB Turpin Distribution Services Ltd. is wholly owned by the Royal Society of Chemistry. Customers should make payments by cheque in sterling payable on a UK clearing bank or in US dollars payable on a US clearing bank. Air freight and mailing in the USA by Publications Expediting Inc. 200 Meacham Avenue Elmont NY 1103. Second class postage paid at Jamaica NY 11431. USA Postmaster send address changes to Furday Discussions Publications Expediting Inc. 200 Meacham Avenue Elmont NY 11003. All other despatches outside the UK by Bulk Airmail within Europe Accelerated Surface Post outside Europe. PRINTED IN THE UK. 0The Royal Society of Chemistry 19%. All rights reserved. No part of this publication may be reproduced stored in a retrieval system or transmitted in any form or by any means electronic mechanical photographic recording or otherwise without prior permission of the publishers.

ISSN:1359-6640    

DOI:10.1039/FD99500FX001

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Faraday Discuss. 1995,100,C3-C7 The Labile Molecule since 1947 George Porter Imperial College London We are here to celebrate our favourite science of Physical Chemistry and it is right that the celebration should be here in the Royal Institution the House of Michael Faraday where he made all his discoveries where he lectured in this theatre for over 40 years and where he lived most of his life as the Resident professor in the rooms on the floor above. I count it as the greatest privilege of my life to have been for 20 years one of Faraday’s successors as Resident professor working in his laboratories living in his rooms and lecturing in his theatre. It was in this theatre on 30th June 1903 that the Faraday Society held its founda- tion meeting for the purpose of ‘promoting the study of electrochemistry electrometal- lurgy chemical physics metallography and kindred subjects’.It was a society of physical chemists named after one who never described himself as such or even as a physicist-a species unrecognised in his time-but who liked to be called a chemist or better still a natural philosopher. The first General Discussion of the new society was held in 1907 and there were about one hundred of them before they began to be numbered as separate publications with No. 1 in 1948 and the 100th in the new series held this month. The Faraday Discussions have been the seed beds from which Physical Chemistry has developed during this century. Many of the early discussions had obvious practical applications and the early presidents were pioneers in the industrial applications of physical chem- istry Sir Joseph Swann whose electric lamp preceded Edison’s by 20 years was the first President; Lord Kelvin whose researches made possible the laying of the first trans- atlantic cable was second and Sir William Perkin who founded the synthetic dye industry and was present in this theatre in 1906 for the golden jubilee of his discovery of Mauvine was the third.Unfortunately I was myself unable to be present at that first meeting in 1903 but I was present when the society celebrated its fiftieth anniversary in this same room in 1953. The discussions of the Faraday Society over the next decade or two were to me where everything in physical chemistry happened where the very latest advances were published and discussed and where one met like-minded scientists from this country and overseas who became lifelong friends.To me there was no society like it it was part of my life. To give a balanced impersonal; account of the many themes of the Faraday Dis- cussions which have taken place during the 48 years that I have been a member even if it were possible would be exceedingly dull. Instead I offer a very personal account of the one area which has concerned me most and how for me it has been inseparable from the Faraday Discussions of the time. I am sure that many of those here could tell similar stories about the importance of Faraday Discussions to their own work. I have called this talk ‘The Labile Molecule since 1947’ because it combines the title and date of the first Faraday Discussion that I attended in Oxford 48 years ago.It was only the second in the series of published discussions whose centenary we celebrate. The pattern of the Faraday Discussions is unique; the papers are written in full and distrib- c3 C4 The Labile Molecule since 1947 uted before the meeting; the presentation lasts only five minutes and is followed by questions and discussion all of which are published. I presented my first paper on work done with Professor R. G. W. Norrish at that 1947 meeting. There was nothing remarkable about the paper but it was important to me and its presentation was to a large Oxford audience including Hinshelwood Steacie Melville Coulson M. G. Evans and other names that I ranked at least equal to the old testament prophets.And there were younger prophets in waiting like Longuet-Higgins Rex Richards and Fred Dainton. It was a daunting occasion for a second year research student from Cambridge. I rather expected to be eaten for breakfast along with the Oxford marmalade. But it was a friendly society and everybody was very kind. Our paper was about the detection and measurement of free radicals in gases by a flow technique developed in 1929 by F. Paneth who used the removal of metal mirrors as a measure of the free radicals present after thermal decompositions. It was an unspecific and difficult method for photochemical reactions even with the help of a large army searchlight as the light source but it was the most direct method available.In the whole 400 pages of this discussion there was no mention of a direct observ- ation of any free radical except for some emission spectra in flames and electrical dis- charges and indeed some were still sceptical about their very existence under normal conditions. Sir Harry Melville had no doubt about their existence but was sanguine about our chances of direct observation. In his opening address he said. The direct physical methods of measurement simply cannot reach these magnitudes far less make accurate measurements in a limited period of time for example lop3sec. This was a challenge that I heard with glee because I had that summer decided to try a new line of work which if successful would do just this. I discarded my searchlight for a flash arrangement with a large capacitor bank by courtesy of my friends in the Royal Navy.The total cost was very small or even negative since the Navy were kind enough to say that there was five shillings due to me for every crate that I returned. The idea which I called flash photolysis was very simple in principle though less so in practice. A flash of light was used to produce chemical change in a substance and a second flash of light then recorded spectroscopically the products that were present a short time afterwards. By repeating the process with increasing time intervals between the flashes a movie of the very fast chemistry of free radicals and other intermediates was recorded. The first paper (published with Professor R.G. W. Norrish) on the use of high- energy flashes in photochemistry appeared in Nature in 1949 and the second paper on the double flash pulse and probe-the flash photolysis technique-was submitted to the Royal Society in the same year (in some haste since I was to be married two days later). I read this paper the following January at an afternoon meeting of the Royal Society in Burlington House with the President the redoubtable Sir Robert Robinson in the chair. He was one of those organic chemists who had grave doubts about the existence of these labile molecules but he was gracious enough to comment after my paper ‘there does now seem to be some experimental evidence for free radicals’. The time was now opportune to attend my second discussion of the Faraday Society held in Cambridge in 1950 on ‘Spectroscopy and Molecular Structure’ where I was able to describe in more detail the first absorption spectra of several free radicals obtained by the new method.The most interesting free radical was C10 obtained by flash photolysis of a mixture of gaseous chlorine and oxygen. The audience included Gerhard Herzberg and James Franck as well as Michael Kasha; truly a contempory of mine since we were born on the same day but owing to the transatlantic time difference neither of us has been able to claim priority. To me Michael’s paper was the most memorable of this meeting. It dealt with types of molecu- lar transition that were little known singlet-triplet and n-n* transitions. It heralded the George Porter c5 coming of a new era in photochemistry but received little attention at the time least of all from photochemists.My third Faraday Discussion held in Toronto in 1953 was attended by 45 members of the society from Europe including R. P. Bell E. J. Bowen Charles Goodeve Alfred Egerton R. G. W. Norrish Harry Melville F. S. Dainton Peter Gray and John Polanyi along with E. W. R. Steacie H. S. Taylor George Kistiakowsky Linus Pauling and the Society’s secretaries F. C. Tomkins and Beatrice Kornitzer. Many of us were shipmates on the Empress of Britain for the crossing. It was the first post-war overseas meeting of the Faraday Society and a quite memorable one for all of us. The subject of this Canada meeting was ‘The Reactivity of Free Radicals’.It seemed as if the Discussion subjects were being chosen especially for me because having inter- preted the spectra of some free radicals I had now turned to the original purpose the use of these spectra to measure reactivity. The chemical reactions of the free radical ClO made a very interesting story but were of little practical importance as I had to admit to visitors to the laboratory in Cam- bridge who might become benefactors and who quite reasonably were interested in possible applications. It was 30 years later that the information gained in those days became of very practical importance. It turned out that chlorine in the stratosphere formed by photochemical decomposition of the chlorofluorocarbons used as aerosols and refrigerants were responsible for the decomposition of ozone in the stratosphere and the recently discovered ozone hole over Antarctica.It was gratifying that very shortly afterwards C10 was shown from stratospheric flights over Antarctica to be a key intermediate in this photochemical cycle of destruction. The Faraday stories run and run and the 100th Faraday discussion held this month is on ‘Atmospheric Chemistry ’. Within a few years free-radical studies were everywhere dozens of free-radical absorption spectra had been described and a new free-radical society was about to be created. We were particularly interested in the organic and particularly the aromatic free radicals and we recorded for the first time the spectra of benzyl anilino phenoxyl and phenyl and many of their derivatives.It fell to Herzberg to use flash photolysis and ultra-violet spectroscopy to discover the spectra of methylene and methyl which had been two of our original targents. About this time in 1954 Irwin Norman and I described a complementary technique for recording free-radical spectra by photolysis of low-temperature solid solutions. We now turned to the flash photolysis study of another very important class of short lived intermediates the electronically excited states of molecules. Flash photolysis was still limited by the duration of the shortest flashes available to times of a few micro- seconds whilst typical allowed electronic transitions take place in much shorter times in the nanosecond range. However the triplet states involve much slower forbidden tran- sitions and their absorption spectra had just been detected in low-temperature solids by Donald McClure.All seemed set to detect these excited species in gases and in fluid solvents at normal temperatures although it was not at all certain that their lifetimes under these conditions would be long enough for our methods. We were fortunate; the lifetimes were ten or a hundred times longer than our flashes and a whole new chemistry became available to us the chemistry of the triplet state which as soon became appar- ent was as important to a photochemist as free radical chemistry. These transient triplet-state molecules in solution formed the subject of a paper with Maurice Windsor in my fourth Faraday Discussion held in Birmingham in 1954 on the subject of ‘The Study of Fast Reactions’.Our paper was on ‘The Triplet State in Fluid Solvents’ and it described the detection and kinetic study of about a dozen aromatic molecules as well as along with Robert Livingston chlorophyll in their triplet states at room temperature where their lifetimes were very conveniently for us a few hundred microseconds. C6 The Labile Molecule since 1947 R. G. W. Norrish and I read a paper at this same discussion on the flash photolysis technique itself. This was one of the first of many discussions on techniques for fast reaction chemistry. Other methods described included the shock tube flow (and stopped flow) methods and the temperature jump and relaxation techniques described by Eigen in his paper ‘Study of Ionic Reactions in Solutions in Times as Short as seconds.” Notice that although ‘microsecond’ was now in the chemical vocabulary the ‘nanosecond’ had still to arrive.We were all getting excited about fast reactions and were already talking about very fast reactions. There was a story going around at this meeting that Manfred Eigen sought the advice of an Oxford man as to what in the English language one could call reactions that were faster than very fast. Without any hesitation Ronnie Bell replied ‘Damn fast reactions Manfred and if they get faster than that the English language will not fail you you can call them. ‘Damn fast reactions indeed!’ For chemists biologists and physical chemists this meeting began ‘the race against time’.We awaited shorter pulses in the nanosecond region to enable us to study those very labile molecules the excited singlet states of molecules. The laser which was to trans- form the technique of flash photolysis and to make it the fastest of all these fast methods appeared in 1960. It took a few years to get hold of a laser and apply it to photochem- istry but this was done by Jeff Steinfeld in Shefield by 1965 using pulses ofa few nanoseconds. In 1966 I moved from Sheffield University to the Royal Institution and a student Micheal Topp who moved with me brought the precious laser and set up our first nanosecond flash apparatus here. We had to solve several further problems before we could do nanosecond flash photolysis. We had to produce a nanosecond white light pulse and to synchronise the two pulses within a few nanoseconds delay.We were then able to detect the singlet state absorption spectra of many molecules. As new laser pulse techniques became available especially the mode-locked colliding-pulse-mode laser developed by Shank picosecond and then femtosecond flash photolysis became possible. The steps in this race against time are shown in comparison with the cosmic time scale in Fig. 1. In the femtosecond region of physical chemistry there are two areas which are advancing particularly rapidly at this moment. The first the fastest step in the course of a molecular reaction is the crossing of the transition state. This was the subject of the Faraday Discussion of 1991 on ‘Structure and Dynamics of Reactive Transition States’ at which speakers included John the second Polanyi in this field his fellow Laureate Dudley Hershbach and two pioneers of femtosecond transition state dynamics Dick Zare and Ahmed Zewail.The progress through the transition state of a dissociating NaI molecule which becomes trapped in the resonant excited state for an average of 10 vibrations was followed by Zewail using fluoresence of the product Na* and the off-resonance fluorescence of NaL*. The second rapidly developing area of femtosecond molecule dynamics is photosyn- thesis. At this point I should remind you that in 1951 the Faraday Society broadened its terms of reference to read ‘To promote the study of sciences lying between chemistry physics and biology.’ This was a wise move though it is a matter for regret to me that there have been relatively very few Faraday Discussions in these interdisciplinary fields probably because of the immense growth in the number of societies and their associated journals in the biophysico-chemical areas.Today many of the most exciting develop- ments are in the biophysical sciences. Particularly active at the present time are studies of the primary processes of photo- synthesis. These begin with the transfer of the absorbed solar excitation energy between pigment molecules in the leaf membrane followed by the transfer of electrons and protons across the membrane It turns out that both of these processes occur in picose- conds or less and it was fortunate that at the time that these systems were being dis- covered femtosecond flash photolysis was becoming available.George Porter c7 SECONDS 1ol8 Age of Earth 1015 First Man lo 12 Pyramids 10 9 10 103 1 nse FLASH PHOTOLYSIS 10 -3 OND Fig. 1 Timescale of flash photolysis compared with the cosmic scale The most recent studies of these reactions concern the reaction centre of photo-system 2 of the green leaf which is the simplest unit that still carries out both energy and electron transfer. The structure of its associated light-harvesting unit (LHC 2) is now known from the electron diffraction work of Kuhlbrandt. The structure of the reaction centre (containing six chlorophylls and two pheophytins) has not yet been fully deter-mined but much can be deduced by comparison with the structures of the reaction centres of the photosynthetic bacteria determined by Huber Mlchel and Deisenhoffer which have many close homologies with photosystem 2.The processes of energy and electron transfer in the reaction centre occur in times that lie in the femtosecond and picosecond regions respectively. The theory of energy transfer was the subject of the Spiers Memorial lecture given by Theodore Forster at the Faraday Discussion of 1959 on ‘Energy Transfer with Special Reference to Biological Systems’. Theories of electron transfer have frequently been discussed notably by Rudy Marcus in his R. A. Robinson Memorial lecture at the Faraday Discussion of 1982 on ‘Electron and Proton Transfer’. The rates predicted for these transfers are about 100 femtoseconds and a few picoseconds respectively which agree very well with those mea-sured directly by flash photolysis.There is no part of the natural world which does not come under the scrutiny of physical chemists from time to time the electrical magnetic and optical properties of matter and its internal motions rotations vibrations and electronic transitions; the structure of molecules new ones like buckminsterfullerene and the eternal complex and beautiful ones of nature; chemical changes in strange places like the stratosphere and outer space in the hottest flames and in solids near to the absolute zero. And chosen as the subject of this talk the changes in these structures that occur both in the laboratory and in nature in times so short that we are near the point where chemists biologists and most physicists must accept that they have reached the end of the timescale and the limits of certainty. Furaday Discussion 100 Celebration Paper; Presented 24th April 1995

ISSN:1359-6640    

DOI:10.1039/FD99500000C3

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Celebration of Physical Chemistry To Mark the 100th Faraday General Discussion The Royal Institution London 24 April 1995 A Celebration of Physical Chemistry 24th April 1995 Contents c1 Introduction J. P. Simons c3 The labile molecule since 1947 G. Porter c9 Tales of tortured ecstasy Probing the secrets of solid catalysts J. M. Thomas C29 Molecular simulation A view from the bond D. J. Tildesley c47 Approaching complexity S. K. Scott C61 How discoveries are made and why it matters J. C. Polanyi C67 Rearranging atoms and trapping electrons F. S. Dainton C83 List of Posters C85 List of Participants

ISSN:1359-6640    

DOI:10.1039/FD995000X007

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

FU~U~UY Discuss. 1995 100 C9-C27 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalysts John Meurig Thomas Davy Faraday Research Laboratory The Royal Institution of Great Britain 21 Albemarle Street London UK WIX 4BS and Peterhouse University of Cambridge Cambridge UK CB2 IQY 1. Introduction In view of the fact that the first part of the title of my talk has prompted some to comment on its sado-masochistic overtones I should point out that its origins lie deep in the poetry and insights of Shakespeare '. .. bitter torture shall winnow the truth from falsehood' and '. . . the torture of the mind (is) to be in restless ecstasy' In common with many others I pursue my chemistry with passion and commitment. And it seems to me to be within the bounds of poetic licence to describe one's oscillation of mood from euphoric exhilaration-when the act of probing is profitable-to satur-nine gloom-when it is not-as tortured ecstasy.Now catalysis as a phenomenon has been the subject of numerous Faraday Society Discussions. At the fiftieth held in Cambridge in 1928 T. M. Lowry in his opening address pursued a holistic approach in which he endeavoured to encompass homoge- neous heterogeneous biological and photo-catalysis. Looking back at that meeting as well as an earlier one in 1921 held in London also on catalysis we note that many of the ideas and concepts in vogue seventy years ago have fallen by the wayside broken mile- stones on a vanished road. But one of the most remarkable facts is that the sub- discipline of catalysis which was arguably the least understood seventy or so years ago-biological or enzymatic catalysis-is now almost certainly the best understood.2. Lessons from the Enzymologists and Organic Chemists Why is it that enzymologists can nowdays successfully design new biological catalysts such as artificial enzymes that rival (even surpass) the performance of the parent natural enzyme? And why is it that the inorganic catalyst scientist and engineer are a good deal less adroit in their endeavours? The principal reason is because those working with enzymes have for over thirty years been engaged in in situ structural studies of their catalysts and have consequently gained revealing insights into mechanism whereas those working with inorganic catalysis have not.In the mid 1960s D. C. Phillips et al. elucidated the structure of the enzyme lysozyme.' They identified the cavity within its structures that constitutes its active site and in particular characterized in atomic detail the stereochemistry of the site and the manner in which reactant molecules fitted snugly within it how as it were the reactant was held (Plate 1) poised for catalytic conversion. Their studies also revealed how product species and and/or inhibitors occupied the cavity of the active site (Fig. 1). The c9 c10 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst Plate 1 Packing of lysozyme with the cleft that defines the active site occupied by the reactant (substrate) sequence of the amino acid components lining the cavity active site was identified and the nature of the binding of reactants and inhibitors likewise elucidated.This in turn gave the enzyme chemist and molecular biologist equipped with the preparative and synthetic dexterity of the organic chemist a firm basis for the design of smaller (miniature) versions of naturally occurring enzymes. It became sensible to talk of biomimetic catalysts and artificial enzymes and with the later advent of site selective mutagenesis new dimensions of structural probing became possible. A particular amino acid close to or actually constituting the active site-which usually consists of a triad of amino acids-could be replaced thereby enabling the catalytic performance of the enzyme to be delicately monitored as a function of subtle change in stereochemical and electronic environment within the active site.(For an elegant account of such procedural Fig. 1 Mechanism of mode of catalytic action of lysozyme (from Phillips') J. M. Thomas c11 advances see the work of Knowles2 on the manner in which triose phosphate isomerase catalyses the interconversion of two triose phosphates known as DHAP and GAP.) Work on so-called artificial enzymes may be illustrated by reference to the proteo- lytic enzyme chymotrypsin the structure of the active site of which was elucidated by in situ X-ray crystallography several decades ago. This revealed that the key components of the catalytic reaction cavity of chymotrypsin are imidazolyl phenolic and carboxylic groups.Armed with this information a way of assembling a miniature artificial chy- motrypsin presented itself to Bender and co-worker~.~ They took a naturally occurring molecule such as P-cyclodextrin as the shallow cavity and grafted on to its rim- appropriately juxtaposed just as in the chymotrypsin itself-the three components (the triad of groups) that constitute the active site. Such a creation is shown schematically in the lower half of Fig. 2 and a scalar model is shown in Plate 2. Although there is now some doubt about the validity of the claims made in Bender and co-workers' work the notion that lies behind their overall strategy remains (Plate 2) a synthetic molecular entity of about a hundredth the size and mass of the parent enzyme can be made to function as a catalyst in such a manner as to rival the performance of the natural enzyme.Breslow4 has produced an even simpler piece of elegant analogous chemistry where the reality of the shape-selective catalysed chlorination of anisole is unmistakably E+S E-S (E -S)' EtP I t 1 catalyst substrate@) I (C -S) complex I products (C) stabilitv rate selectivity selectivity turnover artificial enzyme Fig. 2 (u) Schematic representation of the enzymatic process. The enzyme E processes the sub- strate (reactant) S into product P uiu the transition state E-S (after Lehn). (b) Strategy for the construction of an artificial chymotrypsin (miniature enzyme) based on cyclodextrin (from J. M. Thomas6) c12 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst .._-Plate 2 Model of artificial chymotrypsin.(after Render and co-workers’). (Colour graphics by R. G. Bell and J. M. Thomas.) Plate 3 Cyclodextrin-catalysed shape-selective conversion of anisole to p-chloroanisole. (Colour graphics by Dewi Lewis based on Breslow4.) J. M. Thornas C13 demonstrated (Plate 3). Only the para chloro isomer is produced in this cyclodextrin- steered reaction. 3. How Well can we Design Inorganic Catalysts? Although it seems highly unlikely that we shall ever have the benefit of an inorganic equivalent to site-selective mutagenesis great strides may be taken in designing shape- selective inorganic catalysts using molecular sieve solids typified by that shown in Plate 4.Here we see a zeolitic framework known as ZSM-5 (Zeolite Socony Mobil number 5) which is a highly siliceous aluminosilicate of general formula H '(Si -xAlx02)-nH20 where the Si/Al ratio may range from a low of ca. 10 to almost infinity. Aluminosilicate ZSM-5 is an excellent catalyst' for shape-selectively (a) converting benzene to monoethylbenzene (a precursor of styrene) by alkylation with ethene; (b) disproportionating two molecules of toluene to one of benzene and one of paruxylene; (c)isomerizing a mixture of xylenes preferentially to paraxylene; and (d) converting methanol to petrol (gasolineka mixture of predominantly benzene toluene and xylene-and water. By introducing certain key elemental substituents into the ZSM-5 framework numerous other highly selective catalytic reactions of profound petrochemical significance may be effected.6 Thus with gallium in the ZSM-5 propane and butane may be efficiently dehy- drogenated to yield benzene and toluene this being the basis of the British Petroleum cyclar process commercialized some five years ago.With zinc in the ZSM-5 framework dehydrogenation of isobutane to isobutene (known also as 2-methylpropene) is effected. This alkene first prepared in this building (and discovered by Michael Faraday) is of central importance industrially as it is the precursor of MTBE (methyl tertiary butyl Plate 4 ZSM-5 and its use as a modified catalyst for methane oxidation to methanol;* propane to benzene; and isobutane to 2-methylpropene (isobutene or isobutylene) C14 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst Fig.3 High-resolution electron micrograph of a zeolitic acid catalyst showing projected structure of pores lined with active sites. Large pores are 5.5 A in diameter [from J. M. Thomas,6 and J. M. Thomas and P. L. Gai-Boyes Nature (London),1993,364,4781. ether) a major constituent nowadays of motor car fuel since it both replaces the dele- terious lead-containing additives and facilitates7 the complete combustion of the fuel to CO rather than CO. With iron in the framework the ZSM-5 is reported to favour the oxidation of methane (with N,O as reactant) to methanol.' ZSM-5 itself (Fig. 3) was first produced (in the early 1970s) by accident. Workers at the laboratories of the Mobil Company USA decided to try a new organic molecule as a template for the production of synthetic zeolites using the general strategic prin- ciples first employed by the British scientist R.M. Barrer. More will be said about Plate 5 Graphic illustrating how microporous titanosilicalite (TS-1) converts phenol to hydro- quinone using H,O ,and propene to propene oxide (Colour graphic by Dewi Lewis.) J. M. Thomas C15 templating later; what is important to note here is that it is possible to design new catalysts having a framework structure more-or-less identical to that of ZSM-5 but implanted with a 'designed' hetero-atom so as to endow the resulting solid with alto- gether new catalytic properties. This is precisely what researchers at the Enichem Company Italy succeeded in doing when they brought forth their so-called titanosilicalite-I (TS-1).This has the same framework structure as ZSM-5 and silicalite- 1 (which is the siliceous end-member of ZSM-5 i.e. x = 0 in the above-mentioned formula) but Ti4+ is accommodated in the framework. TS-1 is a highly selective and active oxidation catalyst capable amongst other things of facilitating the epoxidation of propene and the conversion of phenol and H,O to quinol (1,4 benzene diol) an important material (as a developer) in the photographic industry (Plate 5). 4. De Novo Synthesis of New Inorganic Catalysts Plate 6 shows the protonated (acidic) form of mordenite inside the pores of which are molecules of diisopropylnaphthalene (DIIPN) a vitally important building block in the synthetic polymer industry9*'* (DIIPN is formed in high yield when propene and naph- thalene are allowed to react in the presence of an acid catalyst).Over acidic silica- alumina gel numerous other undesirable products are formed as well as DIIPN. Over the environmentally harmful AlCl again there is a multiplicity of products. Acidic mordenite however is shape-selective in that it yields predominantly 2,6 DIIPN). In Plate 6 the snugly fitting product (DIIPN) gives a clue as to the mode of action of organic templates in the synthesis of new microporous catalysts. The degree to which templates are critical in the synthesis of a particular framework varies.' ',12 Certain templates (of which ethylenediamine is an example) may be regarded simply as void-filling species that do not contribute significantly to the preferential for- mation of a desired structure.We also find that a particular framework is formed by several different template molecules a process which might be more correctly termed Plate 6 Colour graphic representation of protonated mordenite of high Si/Al ratio. Inside one of the large pores a molecule of diisopropylnaphthalene (DIIPN) is shown. As described in the text (see also Cusumano') acidic mordenite is the catalyst of choice especially on environmental grounds in the conversion of naphthalene and propene to DTIPN. (After J. M. Thomas R. G. Bell and P. A. Wright Bull. Chim. SOC.Fr. 1994 131,482.) C16 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst structure-directing rather than genuine templating.In general however all the templates suitable for a particular framework will possess similar properties-shape size basicity etc.-which direct the gel medium (usually under hydrothermal conditions) from which the crystals nucleate toward the formation of particular structural motifs. A great deal of accumulated experience supplemented nowadays by computer graphics and modelling11p13 has been acquired to enable the inorganic preparative chemist to design a wide range of zeolitic material. In particular microporous aluminium phosphates and metal-substituted (in the framework) aluminium phosphates may now be routinely pre- pared. Some of these have structures similar to those that have been characterized in the aluminosilicate zeolites (natural and synthetic).Some however are quite novel. The framework structures of three entirely microporous solids synthesized in this Laboratory or in association with collaborators in Jilin University Peoples' Republic of China are illustrated in Fig. 4. It transpires that aluminium phosphates (ALPOs) may be quite readily produced as microporous analogues of naturally occurring and synthetic zeolites. Many MeALPOs (and with Me = Co Ni Zn Mn Cu Mg and other divalent cations partially replacing the trivalent A13+ of the parent ALPO) are very good solid acid and redox catalyst^.^,'^ Thanks to the efforts of many over the past decade about a hundred distinct micro- porous structures with channel apertures falling in the range from 4 to 14 A have been prepared.Into such structures about a third of the elements of the Periodic Table may be incorporated at framework sites. And into the sites occupied by extra-framework usually exchangeable cations a further third of all the known elements may be placed. These architecturally elegant inorganic repositories have a prodigality of chemical com- position that defies precise description; millions of distinct compositions are possible. Many thousands have already been prepared. The scope for delicate variation is enor-mous. DAF-I DAF-2 (b) JDF-3 Fig. 4 Framework structures of DAF-1 (magnesium aluminium phosphate) showing (a) the wide channel containing supercages and (b) the narrower channel; DAF-2 (cobalt phosphate) contain- ing T (T = Co P) viewed along (a) [OOl] and (b) [loo] axes; and JDF-3 (zinc phosphate) contain- ing T (T = Zn P) viewed along [loo] axis.DAF stands for Davy Faraday arid JDF Jilin Davy Faraday. J. M. Thomas C17 H Plate 7 Quantitative description (of bond lengths and coordination) at the Co" active site in the CoALPO-18 solid acid catalyst for the selective dehydration of methanol. (From J. M. Thomas and G. N. Greaves Sciences 1994,265,1675.) C18 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst 5. Uniform Heterogeneous Catalysts A Vast Family of Inorganic Solids Amenable to in situ Structural Investigation ZSM-5 mordenite and all the other inorganic including solid-acid catalysts mentioned above are all examples of uniform heterogeneous ~~talysts,'~~'~ where the active sites such as bridging hydroxys of the type Si-O(H)-A1 are distributed in a spatially uniform and accessible (to reactants) fashion throughout the bulk of the solid.They have other attributes notably exceptionally large surface areas often in excess of 600 m2 g-' ninety-five percent (or more depending on crystal size) of which is inside the solid. In effect they are solid catalysts with three-dimensional surfaces permeating the entire material and accessible to all reactant (and product) molecules small enough to enter the apertures at their outer surfaces. This being so these uniform heterogeneous catalysts are amenable to the entire panoply of spectroscopic diffraction and scattering techniques refined over the years by physicists and chemists.Such catalysts present unrivalled-in the experience of inorga- nic and physical chemists-opportunities to explore the relationship between structure and catalytic performance much in the way that modern enzymologists pursue their investigations (described in Section 2). In situ X-ray crystallography is well suited to such work (as it was originally for lysozyme and is for other enzymes). But there are some serious practical obstacles that have to be surmounted when as is often the case the concentration of the active site is low. X-Ray diffraction alone even when imaginative use'7 is made of anomolous scat- tering is not sufficiently sensitive. Take for example CoALPO-18 (Plate 7) which is a good catalyst for converting methanol to light alkene~.'~.'~ The concentration of the Co" ions in the framework of the Plate 8 Experimental equipment of the combined Qu-EXAFS-XRD setup.The spectroscopy and diffraction data obtained for cordierite are shown by the two inserts relating to the two detector systems. (From J. M. Thomas et d2l) J. M. Thomas C19 parent ALP0 seldom exceeds a few percent too low to be directly addressable by X-ray diffraction. We may however use X-ray absorption thanks to the availability of syn- chrotron radiation. Moreover we have evolved methods of probing catalysts of this kind under operating conditions using combined X-ray diffraction and X-ray absorption spectroscopy (XRD and XAS) recorded with the same sample under realistic conditions of catalytic use (Plate -22 Combined studies in parallel with infrared absorption spectroscopy yield results such as those shown in Plate 7 where the changed bond lengths associated with conversion of the Co" to Co"' valence states while still retained in the framework may be quantitatively specified.Typical results obtained by the com- bined in situ XRD and XAS techniques recorded during the calcination of the CoALPO- 18 are shown in Fig. 5. There is every reason to believe that this dual in situ approach to the study of open- structure (as well as ~ther'~.~~) catalysts will provide fresh insights into the mechanisms of heterogeneous catalysis. A specific example of such new insights has recently been seen in the study of metal-ion catalysts anchored onto the interior surfaces of meso- porous supports (see below).The dual in situ XRD-XAS approach is also of great value in tracking the synthesis-the actual act of crystallization from a precursor gel-of new zeolitic cata- lysts as outlined in the next section 7900 Fig. 5 Combined Qu-FLEXAFS-XRD following the calcination of CoAPO-18. (a)Shift in the K edge of cobalt (measured in fluorescence) during initial heating in air to burn off the template and during reduction to activate the catalyst. (h)XRD patterns during initial heating showing virtually no change except for a drop in background on removal of the template. (From J. M. Thomas et a1.2l) c20 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst 6.Synchrotron-based Studies for Tracking the Synthesis of Uniform Heterogeneous Catalysts Progress in our understanding of uniform heterogeneous catalysts rests almost as much on our ability to monitor in atomic detail the formation or synthesis of such catalysts as well as the structural changes they undergo during operation. A start has recently been made by my group working closely with Professor G. N. Greaves and his colleagues at the EPSRC Daresbury Laboratory. Measurements that are experimentally demanding with conventional laboratory- based X-ray sources often become readily accessible when carried out with synchrotron radiation. This is so in the study of structural changes in solid catalysts under operation conditions.It is also true in the study of the nucleation and growth of crystals from solutions melts and gels. With photon fluxes from a typical synchrotron exit port of the order of 1013 s-' and more importantly with high-energy X-rays (up to 140 keV) capable of penetrating stainless-steel containers it is relatively straightforward to study condensed phases using appropriate high-temperature and/or high-pressure cells without serious attenuation in the intensity. Moreover taking advantage of solid-state detectors the white radiation of synchrotrons permits the rapid buildup of time-resolved diffraction patterns (under isothermal conditions) that allows the ready tracking of structural change occurring in a cell that is held in a fixed position.Fig. 6 shows the energy-dispersive X-ray diffraction (EDXRD) geometry available on station 16.4 at the 2 GeV Daresbury Synchrotron Radiation Source. The EDXRD facility is illuminated with radiation from a 6 T superconducting wiggler. The stainless-steel autoclave shown in the figure can be operated up to temperatures of ca. 250"C permitting the in situ character-ization during hydrothermal synthesis of most of the zeolitic materials. The solid-state detector is inclined at a fixed 20 angle chosen so that from the photon energies available (7-1 40 keV) representative crystallographic spacings for zeolite structures are readily identifiable. With SRS electron beam currents of around 200 mA energy-dispersed pat- terns were integrated every 2 min in order to achieve adequate statistics for phase identi- fication and the recording of reaction kinetics for the synthesis rates encountered here.To piressure gauge4 Oven -A drlu valve / Solid-state detect0r CoIIirnat0r Post Sample A Polychromati Synchrotron Radiation liquor-gel) Window-hole in oven liner Ct-;nlnee rtnnl otaiiiicaa ~LCCI autoclave Fig. 6 Schematic diagram of the experimental set-up for energy-dispersive X-ray diffraction. Syn- thesis takes place in the heated stainless-steel autoclave and the crystallization is detected with a single-element (Canberra) Ge solid-state detector. This is inclined at a fixed 28 angle of 1.4556 f0.0001" and calibrated using NBS 640b Si. The integration time for each frame is 2 min. (From F.Rey et J. M. Thomas c21 Time-resolved X-ray diffraction patterns of the catalyst CoALPO-5 as well as of its parent ALPO-5 have been collected (and publi~hed~~) and these clearly show the change from structureless diffraction patterns exhibited by the mother liquor to highly structured ones corresponding to the appearance of the ALPO-5 phase along with a co-existent chabazitic equivalent of CoALPO-5. Very recently my colleagues and I have used the combined XRD-XAS approach schematized in Fig. 7. With the capillary reaction vessel housed in an appropriately designed furnace both XRD patterns (comparable to those obtained with the set-up shown in Fig. 6) as well as richly detailed X-ray absorption spectra are obtainable (see Fig. 8).The pre-edge X-ray absorption spectra and also the XANES and EXAFS infor- mation (taken in conjunction with the XRD results) unmistakably reveal25 that prior to the onset of crystallization octahedral Co" ions in the precursor (templated) gel become tetrahedrally coordinated. 7. Mesoporous Catalysts Plate 9 summarizes progress made in the last forty years in synthesizing open-structure solids with potential as catalysts. Zeolite A produced by the Linde (Union Carbide) Company in the early 1950s has pore dimensions around 4 A (depending upon the exchangeable cation). These are of great commercial applicability-in detergency and water-softening especially-but they have essentially zero prospects as viable catalysts. DAF-1 (Davy Faraday One) already described above and discovered here in 1992 is in quite a different category and has demonstrated value-but not much staying power- as an acid catalyst for the isomerization but-1 -ene to isobutene (2-methylpropene) the material first synthesized by Faraday.MCM-41 (Mobil Catalytic Material number 41) first reported in 199226) is of even greater potential in a new generation of shape-selective catalysts and as supports for metal-based catalytic centres. Fig. 7 Schematic representation of set-up used by Sanker Rey Thomas and Greaves (in ref. 24) for combined XRD-XAS in situ measurements to follow crystallization of molecular sieve cata- lysts from liquids and gels. (See also Fig. 3 of ref. 21.) c22 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst 15.9 ~"c"7.69774 Fig.8 Stacked (a)XRD patterns and (b)X-ray absorption spectra during the crystallization of a -.--. .. .. * .. . . . . _. -_ LoALYW-3 catalyst from the liquid phase studied using the set-up shown In k'ig. 7. From the pre-edge (Is + 3d) transition the EXAFS and the shape of the XANES peaks there is no doubt that the cobalt ion originally in octahedral coordination becomes tetrahedrally bonded prior to the onset of crystallization. Mesoporous siliceous solids with channel apertures from 25 to 100 A have opened up new possibilities in heterogeneous catalysis. The large diameter (ca. 30 A) channels of the so-called MCM-41 mesoporous silicas that we and others have used for selective oxidation permit in principle the direct grafting of complete metal complexes and organometallic moieties onto the inner walls of these high-surface-area (typically >800 m2 8-l) solids.This opens routes to the preparation of novel catalysts consisting of J. M. Thomas C23 Plate 9 Zeolite A first prepared in the mid 1950s is a synthetic molecular sieve (ca. 4 A diameter apertures) of little catalytic significance. DAF-1 (Davy Faraday one) first synthesized in 1992 (diameter of apertures >7 A) as well as MCM-41 (Mobil catalytic material 41) a mesoporous solid has great potential as a catalyst. Plate 10 An example of a tethered metallic catalyst prepared by T. Maschmeyer (unpublished) at the Davy Faraday Research Laboratory (see also ref. 27). C24 Tales of’ Tortured Ecstasy Probing the Secrets of’Solid Catalyst large concentrations of accessible well-spaced and structurally well defined active sites.Maschmeyer et al. in these laboratories have described the production of titanocene- derived catalyst precursors anchored to the inner walls of the MCM-41 and its conver- sion (monitored by X-ray absorption spectroscopy) to a powerful catalyst for the epoxidation of cy~lohexene.~~ Furthermore this catalyst is also active in the oxidation of other important bulkier reactants such as pinene. X-Ray absorption spectroscopy not only helps (as does in situ FTIR) to identify key intermediates in the epoxidation catalysis but also reveals that in the activated state of the precursor Ti is four coordi- nated with no evidence for a titanyl (Ti=O) bond.During catalytic reaction it is six coordinated. With MCM-41 based catalysts it is possible not only to anchor a reactive (catalytic) centre designed to order according to the principles of organometallic chemistry but also to tether them as schematized in Plate 10. Here the active site is situated at the extremity of the tether and is free to flutter in the molecular breeze during the process of catalytic conversion. By deliberately restricting the spatial freedom in the vicinity of the active centre (see Plate 11) it should be possible to design highly stereo-selective (enantiometric) catalysts. Such possibilities are now under active investigation at the Davy Faraday Research Laboratory. One other highly convenient property of MCM-41 mesoporous solids is that they may be modified in subtle and catalytically useful ways during synthesis.Plate 12 taken from the work of my colleagues,28 shows how hetero-atoms may be incorporated into the siliceous framework of MCM-41. When titanium is anchored in the wall of such mesoporous hosts facile epoxidation (of cyclohexene as shown in Fig. 9) ensues. Plate 11 Deliberate restriction of spatial freedom at the active centre grafted onto the inner walls of a mesoporous silica offers scope for enantiomeric catalysis (Maschmeyer et al. in preparation.) J. M. Thomas C25 Plate 12 Idealised metal-containing MCM-41 structure generated by computer modelling methods (from Frey et ~1.~') (4 (d Fig. 9 Schematic representation of the preparation of the grafted catalyst (a) the support (b) anchoring reaction (c) calcination and (d) reversible reaction of the organometallic-derived Ti-MCM-41 epoxidation catalyst with water upon exposure to the atmosphere C26 Tales of Tortured Ecstasy Probing the Secrets of Solid Catalyst 8.Epilogue In summary we observe that both microporous and mesoporous crystalline inorganic solids have greatly extended the scope and nature of heterogeneous catalysis. Homoge- neous catalysts may be readily heterogenized so that the separate advantages of homo- geneous catalysts on the one hand-their high specificity-and heterogeneous catalysts on the other-their built-in ease of separation of product from reactant-may be jointly harnessed in a synergistic mode.We also note that in situ methods of probing the structure of catalysts under oper- ating conditions are rendered feasible owing to the availability of synchrotron radiation. All this enables greater insights to be gained into the mechansim of catalytic reaction insights which are often enriched by modern computational chemical techniques. Table 1 summarizes my own group’s overall strategy. In conclusion I cannot help recalling what Humphrey Davy standing at this spot said in 1803 Of modern chemistry it may be said that its origins are pleasure its growth know- ledge its objects truth beauty and utility. I hope that I have demonstrated that this admirable definition is as valid now as it was when it was first enunciated. Table 1 Overall strategy of work on heterogeneous catalysis at the Davy Faraday Research Laboratory 0 Synthesis of new micro- and meso-porous catalysts 0 Development of techniques and tools 0 Characterisation 0 Performance 0 Mechanism +computation 0 Applications (patents?) 0 Collaboration with industrial partners References I D.C. Phillips Proc. Natl. Acad. Sci. USA 1967 57 484. 2 J. R. Knowles Nature (London) 1991,350 121. 3 V. T. D. Souza K. Hanahusa T. O’Leary R. C. Gradewood and M. L. Bender Biochem. Biophys. Res. Commun. 1985 129 727. 4 R. Breslow Acc. Chem. Rex 1995 28 146. 5 P. B. Weisz Proc. 6th Int. Conyr. Catalysis Kodansha Tokyo 1990 p. 1. 6 J. M. Thomas Anyew. Chem. Int. Ed. Enyl. 1994,33,913. 7 J. M. Thomas Sri. Americun 1992 266 85. 8 K.I. Zamaraev 7op. Catal. 1996 in the press. 9 J. A. Cusumano CHEMTECH. 1992,22,482. 10 J. M. Thomas and K. I. Zamaraev Angew. Chem. 1994 106,316. 11 R. G. Bell D. W. Lewis C. R. A. Callow J. M. Thomas P. Voigl and C. M. Freeman in Zeolites and Related Microporous Muterial\ Slate of lhe Art ed. J. Weitkamp H. G. Karge H. Pfeiffer and W. Hoderich Elsevier 1994 vol. 84 p. 127. 12 D. W. Lewis C. R. A. Catlow and C. M. Freeman J. Phys. Chem. 1995,99 11 194. 13 D. W. Lewis C. R. A. Catlow and J. M. Thomas submitted. 14 J. Chen and J. M. Thomas J. (‘hem. Soc. Chem. Commun. 1994,603. 15 J. M. Thomas Angew. Chem. Inl. Ed. Engl. 1988 27 1673. 16 J. M. Thomas J. Chen and A. R. George Chem. Rr. 1992,28,991. 17 A. K. Cheetham and A. P. Wilkinson Anyew. Chem.Int. Ed. Engl. 1992,31 1559. 18 J. M. Thomas G. Sankar P. A. Wright J. Chen L. Marchese and G. N. Greaves Angew. Chem. Int. Ed. Engl. 1994 33 639. J. M. Thomas C27 19 J. W. Couves J. M. Thomas G. N. Greaves R. H. Jones D. Waller A. J. Dent and G. E. Derbyshire Nature (London) 1991,354,465. 20 G. Sankar P. A. Wright S. Natarajan J. M. Thomas G. N. Greaves A. J. Dent B. R. Dobson C. A. Ramsdale R. H. Jones J. Phys. Chem. 1993,97,9550. 21 J. M. Thomas G. N. Greaves and C. R. A. Catlow Nucl. Instruments acd Methods in Physical Research 1995 B97 1. 22 P. A. Barrett G. Sankar J. M. Thomas and C. R. A. Catlow J. Phys. Chem. Solids 1995,56 1395. 23 B. S. Clausen L. Grabek G. Steffensan P. L. Hausen and H. Topsrae Catal. Lett. 1993,20,23. 24 F.Rey G. Sankar J. M. Thomas P. A. Barrett D. W. Lewis C. R. A. Catlow S. M. Clark and G. N. Greaves Chem. Muter. 1995,7 1435; see also J. S. 0.Evans R. J. Francis D. O’Hare S. J. Price S. M. Clark J. Flaherty J. Gordon A. Nield and C. C. Tang Rev. Sci. Instrum. 1995 66 2442. 25 G. Sankar F. Rey J. M. Thomas and G. N. Greaves J. Chem. SOC. Chem. Commun. 1995 in the press. 26 C. T. Kresge M. E. Leonowicz W. J. Roth J. C. Vartuli J. S. Beck Nature (London) 1992,359 710. 27 T. Maschmeyer F. Rey G. Sankar and J. M. Thomas Nature (London) 1995,378 159. 28 F. Rey G. Sankar T. Maschmeyer J. M. Thomas R. G. Bell G. N. Greaves Top. Cataf. 1996 in the press. Faraday Discussion 100 Celebration Paper Presented 24th April 1995

ISSN:1359-6640    

DOI:10.1039/FD99500000C9

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Faraday Discuss. 1995,100 23-37 Reactions of Alkoxyl Radicals in the Atmosphere Roger Atkinson Eric S. C. Kwok Janet Arey and Sara M. Aschmann Statewide Air Pollution Research Center University of California Riverside CA 92521 USA The products of the gas-phase reactions of the OH radical with n-pentane and C2H,,]pentane have been studied at 296 2 K and 740 Torr total pres- sure of N,-0 diluent gas in the presence of NO. Direct air sampling atmospheric-pressure ionisation tandem mass spectrometry showed the for- mation of the previously predicted C,-hydroxycarbonyl product together with pentanones. The effect of varying the 0 concentration on the C,-hydroxycarbonyl and pentanones was consistent with the hydroxy-carbonyl being the product formed after isomerisation of the pen tan-2-oxyl radical.Gas chromatographic analyses of pentan-2-one and pentan-3-one and pentan-2-yl and pentan-3-yl nitrates allowed rate constants for the decomposition and isomerisation reactions of the pentan-2-oxyl and pentan- 3-0~~1 radicals relative to their reactions with 0,,to be determined. Introduction Alkoxyl (RO') radicals and hydroxy- and nitrato-substituted alkoxyl radicals are key intermediates in the tropospheric degradation reactions of alkanes and alkenes.'V2 In the troposphere alkoxyl radical reactions with NO and NO2 are minor2 and alkoxyl rad- icals react with 02,decompose unimolecularly or isomerise by a 1,5-H shift through a six-member transition state (Scheme I).' *2 decomposition tsmrisation 02 CH3CH0 + CHsCH2&-l2 / \ CX I I CH$X(OH)C H2C H&Hz Scheme I A major uncertainty in the alkane and alkene degradation reactions in the troposphere which affects the products predicted to be formed involves the relative importance of these three reaction pathways of the alkoxyl radical intermediates.1*2 Absolute rate constants are available only for the reactions of the CH,O' C2H,0' and (CH,),CHO' radicals with 0 ,2 and it is not clear whether or not the rate constants 23 Alkoxyl Radicals in the Atmosphere for the C2H,0' and (CH,),CHO' radicals can be taken as representative of those for the 3C2 primary and secondary alkoxyl radicals respectively.2 Rate constants for the alkoxyl radical decomposition reactions have generally been measured relative to those for reaction of the alkoxyl radicals with 0 or N0.2-4 M RO' + NO ---+ RON0 Furthermore apart from the studies of Dobk et ~l.,~ in which the 4-hydroxypentan- 1-yl radical formed after isomerisation of the pentan-2-oxyl radical was trapped by reac- tion with methyl radicals and Eberhard et aI.,6 in which 2-hydroxyhexan-5-one formed from the hexan-2-oxyl radical was detected and quantified as the 2,4-dinitro-phenylhydrazone derivative the occurrence of the isomerisation reaction of alkoxyl rad- icals formed from the alkanes has been inferred from the low yields of the products expected from the competing decomposition and 0 reactions.'-'' While an empirical method for assessing the relative importance of alkoxyl radical decomposition us.reac-tion with 0 has been proposed,'i2 the lack of knowledge of the alkoxyl radical isomer- isation rate constants precludes reliable assessments of the relative importance of the three alkoxyl radical reaction pathways. In this work we have investigated the products of the gas-phase reactions of the OH radical with pentane and C2H12]pentane in the presence of NO. Products of the pentane reaction were quantified by gas chromatography and the reactions of pentane and C2H ,]pentane were studied using a direct air sampling atmospheric-pressure ionisation tandem mass spectrometer. The data obtained show the formation of the &hydroxycarbonyl(s) expected from the alkoxyl radical isomerisation reaction. Addi- tionally the relative importance of the various reactions of the pentan-2-oxyl and pentan-3-oxyl radicals formed subsequent to H-atom abstraction by the OH radical from the 2-and 3-positions in pentane, has been determined.Experimenta1 Experiments were carried out in ca. 6500 and 7900 1 all-Teflon chambers with the ca. 6500 1 all-Teflon chamber being interfaced to a PE SCIEX API 111 MS/MS direct air sampling atmospheric-pressure ionisation triple quadrupole mass spectrometer. Both chambers are equipped with two parallel banks of blacklamps for irradiation and all experiments were carried out at 296 & 2 K and 740 Torr total pressure of purified air or N2-0 mixtures at ca. 5O/" relative humidity. In both chambers hydroxyl radicals were generated by the photolysis of methyl nitrite (CH,ONO) in air at wavelengths > 300 nm." CH,ONO + hv + CH,O' + NO (2) CH,O' + 0 -+ HCHO + HOz (3) HO + NO -+ OH + NO (4) NO was added to the reactant mixtures to suppress the formation of 0, and hence of NO radicals.' Analyses by API MS/MS The ion source and instrument design of the PE SCIEX API 111 MS/MS atmospheric-pressure ionisation triple quadrupole mass spectrometer have been described pre-viously.' The atmospheric-pressure ionisation source is a point-to-plane design with the point electrode being a discharge needle and the plane being part of the MS R. Atkinson et al. atmospheric-pressure-to-vacuum interface flange. Sample introduction from the Teflon reaction chamber into the ion source is through a ca. 25 mm diameter ca.75 cm length Pyrex tube at flow rates of ca. 22 1 min-' provided by a suction pump. Under positive APCI (atmospheric-pressure chemical ionisation) conditions protonated hydrates [H,O+(H,O),] (n = ca. 3 to 6 at 298 K and ca. 5% relative h~midity)'~ generated by a corona discharge in the chamber diluent gas are responsible for the protonation of anal ytes H,O+(H,O), + M + MH+(H,O) + (n -rn + 1)H20 (5) where M is the neutral analyte of interest. Ions are drawn by an electric potential from the ion source through the sampling orifice into the mass-analysing first quadrupole (Q1) or third quadrupole (43). Neutral molecules and particles are prevented from entering the orifice by a counterflow of ultra-pure nitrogen ('curtain') gas. The proto- nated molecular ([M + H] +) ions mass-analysed are the result of the declustering action of the curtain gas on the hydrated ions.The sensitivity of API-MS to the analyte M is determined by the stability of MH+(H,O) and the water vapour concentration in the ~arnp1es.l~ The lack of an [M + HI+ signal from pentane and C2H,,]pentane in the APCI mass spectra of the CH,ONO-NO-pentane (and C2H1 ,]pentane)-air reaction mixtures suggests that hydrates of [C5H12 + H]+ and [C5D,2 + HI+ were not stable under our instrumental conditions. The mass spectrometer was operated in the positive-ion mode throughout these experiments. Both MS (scanning) and MS/MS with collision activated dissociation were used to analyse the CH,ONO-NO-pentane (or ['HI ,]pentane)-0,-N mixtures.In the MS mode mass spectra (each the sum of 10 scans) of the reactants and the reacted mixtures were obtained by the first quadrupole (Ql),with the second and third quadru- poles (42 and 43 respectively) being operated in the 'total-ion' mode (rf-only mode). Each scan was acquired over the range 10-360 u using a step size of 0.2 u and a dwell time of 10 ms. The curtain gas flow (CG) orifice voltage (OR),and rod-offset voltage of the rf-quadrupole (RO) of the API-MS used during sample analyses were adjusted to minimise the fragmentation of ions in the vacuum side of the orifice.16 It was not pos- sible to eliminate fragmentation prior to 41 but the system could be optimised to favour protonated molecular ions [M + HIf and/or protonated dimer ions [M + HI'.The channel electron multiple (CEM) was operated in pulse-counting mode and optimised by the criterion of a 40% increase in signal intensity per 200 V increment in the CEM voltage. The quadrupole power supplies were adjusted so that unit resolution (50% valley definition) and 0.7 u peak width (at 70% peak height) were achieved. The proto- nated dimer ion of diisopropylmethylphosphonate (DIMP) was used to optimise the instrument parameters CG OR RO and CEM. Mass calibration and quadrupole power suppliesoptimisation were performed onionsat 29,18 1 and 361 u,corresponding to[N,H] + [DIMP + HI+ and [(DIMP) + HI' respectively. The pentanones were quantified by external calibration of the integrated signal of the ion peak at 87 u ([M + HI') to that of a standard of pentan-3-one assuming identical molar responses for pentan-2-one and pentan-3-one.Calibration of hydroxycarbonyl(s) was carried out by measuring the relative response of the [M + HI+ ions of 4-hydroxy- 4-methylpentan-2-one us. hexan-3-one and assuming the same relative response holds for C,-hydroxycarbonyl(s) us. pentan-2-one and pentan-3-one. The molar response of 4-hydroxy-4-methylpentan-2-one was cu. 0.05 of that of hexan-3-one. These calibrations should be viewed as semi-quantitiative. Two MS/MS modes of the API-MS were employed in this work. In the 'daughter ion' scan mode the precursor ion was selected by Q1 and after collision-activated disso- ciation (CAD) in 42 the fragment ions were scanned by 43. In the 'parent ion' mode the fragment ions were selected by 43 and matched to their precursors by scanning 41.Ultra-pure argon was used as the collision gas in an open rf-only Q2 collision cell. The Alkoxyl Radicals in the Atmosphere CAD spectrum of an analyte (sum of 60-200 scans) was acquired using a step size of 0.2 u and a dwell-time of 10 ms. The lower and upper limits of the mass range used for acquiring 'parent and daughter ion' spectra respectively was determined by the mol- ecular weight of the ion of interest. Optimisation of the MS/MS instrumental param- eters was performed by multiple reaction monitoring. l7 For the API-MS experiments the initial reactant concentrations (in molecule cm -3 were CH30N0 NO pentane and ['H,,]pentane (2.4-4.8) x 1013each with the initial concentrations of CH30N0 NO and pentane (or C2H1,]pentane) being equal in each experiment.Irradiations were carried out at 100% of the maximum light intensity for 5 min. In addition to experiments carried out in air (21% 02), OH radical reactions with pentane were carried out in N,-0 mixtures with low and high 0 concentrations (4 and 80% 02,respectively) to investigate the reaction mechanism (see for example Scheme I). To check on any effects of changing the 0 concentration on the API-MS analyses the relative responses of 4-hydroxy-4-methylpentan-2-one and hexan-3-one were measured at 4,21 and 80% 0,. Because the API-MS instrument as used in these experiments is insensitive to alkanes pentane and ['HI ,]pentane were analysed during the experiments by gas chro- matography with flame ionisation detection (GC-FID) as described below.For the API-MS calibrations described above 4-hydroxy-4-methylpentan-2-one, pentan-3-one and hexan-3-one were also analysed by GC-FID using the sampling procedure and GC conditions described below for pentan-2-one and pentan-3-one and pentan-2-yl and pentan-3-yl nitrate. Analyses by GC-FID Experiments with GC-FID analyses of pentane and selected products were carried out in a 7900 1 all-Teflon chamber with initial reactant concentrations (in molecule cm-3) of CH30N0 2.1 x NO (1.6-1.9) x 1014; and pentane (4.29-4.39) x 1013. Irra- diations were carried out at the maximum light intensity for 3-15 min. Experiments were carried out in air [21% 0 (155 Torr O,)] and in 'high-0,' diluent gas [74-85% 0 (590 5 40 Torr O,)].Pentane and selected products were analysed during the experiments by GC-FID. For the analyses of pentane gas samples were collected from the chamber in 100 cm3 all-glass gas-tight syringes and introduced via a 1 cm3 gas-sampling loop onto a 30 m DB-5 megabore column in a Hewlett Packard (HP) 5890 GC initially held at -25 "C and then temperature-programmed to 200°C at 8 "C min- '. For the analyses of the pentan-2-one and pentan-3-one and pentan-2-yl and pentan-3-yl nitrate products 100 cm3 gas samples were collected from the chamber onto Tenax-TA solid adsorbent with subsequent thermal desorption at 225°C onto a 30 m DB-5.625 megabore column in an HP 5710 GC initially held at 0°C and then temperature-programmed to 200°C at 8°C min-'.Gas samples were also collected onto Tenax-TA solid adsorbent for sub-sequent thermal desorption and analysis by combined gas chromatography-mass spec-trometry (GC-MS) using a 60 m DB-SMS fused silica capillary column temperature-programmed from -80 "C at 10"C min-to 250 "C in an HP 5890 GC interfaced to an HP 5970 mass-selective detector operated in the scanning mode. The NO concentrations and initial NOz concentrations were monitored during these experi- ments using a Thermo Environmental Instruments Inc. Model 42 NO-NO,-NO anal yser. Chemicals The chemicals used and their stated purities were hexan-3-one (98%) 4-hydroxy-4-methylpentan-2-one (99"/0) and pentane (99+"A,) Aldrich Chemical Company ; R.Atkinson et al. C2H12]pentane (98% D) and C2H12]pentanol (98%) Cambridge Isotope Laboratories; pentan-2-one and pentan-3-one MCB; pentan-2-yl and pentan-3-yl nitrate Fluorchem Inc.; and NO (>99.0%) Matheson Gas Products. Methyl nitrite was prepared and stored as described previously.' ' Results In the presence of NO the OH radical-initiated reactions of pentane and C2H12]pentane lead to the formation of pentanoxyl radicals by the series of reactions OH + RH +H,O + R' (6) M R' + 0 -RO,' (7) RONO RO,' + NO -c RO' + NO The expected products then arise from reaction (8a) and from the reactions of the alkoxyl radicals (as shown for example in Scheme I for the pentan-2-oxyl radical). Analyses by API MS/MS Fig. l(a) shows the positive APCI mass spectrum of an irradiated CH,ONO-NO-pentane-air mixture after ca.20% reaction of the initial pentane. As noted above no signal from pentane was observed. The C,-carbonyls (primarily pentan- 2-one and pentan-3-one) formed in the reaction appeared as a peak at 87 u [M + HI+. As shown in Fig. 2(c) for an authentic standard of pentan-2-one C,-carbonyls gave a Fig. 1 Positive APCI mass spectra of irradiated (a) CH,ONO-NO-pentane-air and (b) CH,0NO-NO-[2H,2]pentane-airmixtures Alkoxyl Radicals in the Atmosphere Fig. 2 Positive APCI mass spectra of authentic standards of (a) [2H,,]pentanol (b)4-hydroxy-4-methylpentan-2-one and (c) pentan-2-one strong protonated molecular ion under our instrumental conditions. The assignment of the 87 u peak in the reaction to pentanones was confirmed by matching its CAD spec-trum with those of authentic standards (the pentan-2-one and pentan-3-one could not be distinguished by their CAD spectra).The GC-FID analyses discussed below show that both pentan-2-one and pentan-3-one were formed. Product ion peaks of irradiated CH,ONO-NO-pentane-air mixtures were also observed at 103 101 and 85 u [Fig. l(a)]. The molecular weight of the expected 6-hydroxycarbonyl 1-hydroxypentan-4-one formed from the pentan-2-oxyl radical by reactions (9)-(13) is 102 u suggesting that the 10.3 u peak was the [M +HI+ ion of this compound. isom CH3CH(0)CH2CH2CH3-CH,CH(OH)CH,CH,CH (9) M CH,CH(OH)CH,CH,CH +0 -CH,CH(OH)CH,CH,CH,O (10) CH,CH(OH)CH2CH2CH,02 +NO -CH,CH(OH)CH,CH,CH,O +NO (11) isom CH,CH(OH)CH2CH2CH,0 -CH,C(OH)CH,CH,CH,OH (12) CH,C(OH)CH,CH,CH,OH +0 ___* CH,C(O)CH,CH,CH,OH +HO (13) As shown in Fig.2(b) the spectrum of an authentic sample of 4-hydroxy-4-methyl- pentan-2-one included ions at 117 and 99 u corresponding to [M +HI+ and R. Atkinson et al. 29 [M + H-H,O]+. The 85 u peak in the pentane reaction was therefore attributed to loss of water from the protonated molecular ion of the 6-hydroxycarbonyl. A 'parent ion' scan showed that the 101 u peak also originated from this compound and this [M-HI+ ion was attributed to loss of H from the protonated molecular ion. As described below the ion peak at 103 u may also include a contribution from the [(N2)3H30]+ cluster observed in the background spectrum without added reactants.To confirm the identification of the 6-hydroxycarbonyl the mass spectrometer was deliberately tuned to induce dimer formation and the presence of an ion peak at 205 u corresponding to the protonated 6-hydroxycarbonyl dimer provided additional evi- dence for the occurrence of this pentan-2-oxyl radical isomerisation product. Under these experimental conditions corresponding fragment ions to the 6-hydroxycarbonyl peaks at 101 and 85 u were also observed at 203 u ([M,-H]+j and 187 u ([M + H-H,O] +) respectively. Furthermore an ion peak which could correspond to the protonated dimer of pentanone and 6-hydroxycarbonyl was observed at 189 u as was an ion peak at 171 u (corresponding to the protonated heterodimer less water).Nitrogen clusters [(N,),(H,O)] and [(N,),H] (where n 2 0 and rn 2 1) and water + + clusters [(H,O),(H,O)]+ (where n 2 0) may be formed at the APCI interfa~e,'~*'~ resulting in peaks at 19 29 37 47 55 57 73,75 85 and 103 u and certain of these peaks are observed in the spectra of the background air and in the CH,ONO-NO-pentane-air reaction mixtures. Because they also appear in the CAD spectra of protonated pentanone 6-hydroxycarbonyl and their corresponding dimers the 19 29,47 75 and 85 u peaks are attributed mainly to ion fragments derived from these products. Because both fragment ions and [M + HI+ molecular ions are present at odd masses distinguishing the presence of low-molecular-weight products for example propanal ([M + H] + = 59) and acetaldehyde ([M + H] + = 49 is difficult.As shown in Fig. l(b) the reaction of the OH radical with C2H,,]pentane produced product peaks at 94,97 and 112 u with the 97 u peak corresponding to the protonated [2H,o]pentanone ([M + HI+). The protonated fully deuteriated ['H hydroxycarbonyl ion peak would occur at 113 u. However the expected rapid exchange of the D atom of the -OD group with water vapour in the chamber to form the -OH group2' leads to the protonated [2H,J6-hydroxycarbonyl peak at 112 u (for example ([CD,C(O)CD,CD,CD,OH] + H}+). Such D/H exchange in the -OD group under our experimental conditions was demonstrated for fully deuteriated [2H,,]pentanol with the [M + H]+ ion appearing at 100 u [Fig. 2(a)] instead of 101 u if D/H exchange did not occur.The peak at 112 u [Fig. l(b)] is therefore assigned to a protonated [2H,]6-hydroxycarbonyl and this assignment is further supported by the occurrence of the [M + H-H,O]+ peak at 94 u [corresponding to (CD,C(O)CD,CD,CD,] '}. Analogous to the situation for the pentane reaction [M + H-H,O]+ was the major ion from the 6-hydroxycarbonyl observed in the spec- trum from irradiated CH,0NO-NO-[2H ,]pentane-air reaction mixtures [Fig. l(a)J and an [M-D]' peak was also observed at 109 u (corresponding to [CD,C(O)CD,CD,CDOH] '}. Thus despite optimising the instrument to preserve the molecular ions of interest the majority of the 6-hydroxycarbonyl molecules prefer- entially fragmented by loss of a water molecule after their protonation. Fig. 3 shows the mass spectra obtained from the reaction of the OH radical with pentane at low (ca.4%) normal (21%) and high (ca. 80%) 0 levels with the percentage of the pentane reacted being 23 3%. As seen in Fig. 3 the ratio of the peak at 87 u (pentanones) us. that at 85 u (C,-hydroxycarbonyl) increased with increasing 0 concen-tration (note that varying the O2 concentration did not change the relative responses of hexan-3-one and 4-hydroxy-4-methyipentan-2-one in a standard mixture). From the peak areas of the hydroxycarbonyl and pentanone signals and assuming that the response for the C,-hydroxycarbonyl relative to the pentanones was identical to that of 4-hydroxy-4-methylpentan-2-one relative to hexan-3-one (ca. 0.05) the measured Alkoxyl Radicals in the Atmosphere Fig.3 Positive APCI mass spectra of irradiated CH,ONO-NO-pentane-air mixtures in N,-0 diluent gas at 740 Torr total pressure and ca. 5% relative humidity with (a) 4 (b)21 and (c) 80% 0 C,-hydroxycarbonyl/pentanones yield ratios were 4.6 2.5 and 1.1 at 4 21 and 80% 02 respectively. These yield ratios do not take into account reactions of the OH radical with the C,-hydroxycarbonyl and pentanone products. Because the C,-hydroxycarbonyl is calculated to be more reactive than the pentanones towards the OH radica1,2.2'.22 the corrected C,-hydroxycarbonyl/pentanones formation yield ratios are estimated to be ca. 6.4 ca. 3.5 and ca. 1.5 at 4 21 and 80% O, respectively. Clearly the observed change in the C,-hydroxycarbonyl and pentanones distribution with varying 0 concentration (Fig.3) is consistent with the expected reactions of the pentan-2-oxyl radical with 0,,decomposition and isomerisation and of the pentan-3- oxyl radical with 0 and decomposition (the pentan-3-oxyl radical cannot undergo isomerisation via a six-member transition state)' q2y7 [see also the Discussion below]. GC-FID Analyses GC-MS and GC-FID analyses of irradiated CH30NO-NO-pentane-air mixtures showed the formation of pentan-2-one and pentan-3-one and pentan-2-yl and pentan- 3-yl nitrate. (The pentyl nitrates did not give a detectable positive APCI signal at the concentrations present in the APCI-MS/MS experiments with the instrumental condi- tions employed.) Since these products also react with the OH radical their secondary reactions must be considered in determining their formation yields.Secondary reactions were taken into account as described by Atkinson et aZ.,23using rate constants for the OH radical reactions (in 10-l2 cm3 molecule-' s-') pentane 3.91 ;pentan-2-one 5.07; pentan-3-one 2.09; pentan-2-yl nitrate 1.85 and pentan-3-yl nitrate 1.12. The multiplicative correction factors accounting for secondary reactions with the OH R. Atkinson et al. radical which increase with the extent of reaction and with the rate constant ratio k(OH + product)/k(OH + ~entane),~~ < 1.30 for pentan-2-one < 1.12 for pentan- were 3-one <1.12 for pentan-2-yl nitrate and < 1.06 for pentan-3-yl nitrate. Plots of the amounts of pentan-2-one and pentan-3-one formed corrected to take into account secondary reactions with the OH radical against the amounts of pentane reacted are shown in Fig.4 and the formation yields obtained by least-squares analyses of these data are given in Table 1. The relative yields of pentan-2-one and pentan-3-one at 296 f 2 K and 1 atm of air measured in this work with (pentan-2-one)/(pentan-2- one + pentan-3-one) = 0.18 0.04 are in good agreement with the ratios reported by Carter et al.,7 of 0.16-0.21. Fig. 4 and Table 1 show that as expected from Scheme I the pentan-2-one and pentan-3-one formation yields increase with increasing 0 concentra-tion with the effect being most marked for pentan-2-one. In contrast the pentan-2-yl and pentan-3-yl nitrate yields were independent of the 0 concentrations (Fig.5) as Fig. 4 Amounts of (A,A) pentan-2-one and (0,0)pentan-3-one formed corrected for reaction with the OH radical vs. the amount of pentane reacted at 296 f 2 K and 740 Torr total pressure. Open symbols 155 and filled symbols 590 f 40 Torr 0,. Table 1 Measured formation yields at 296 & 2 K and 740 Torr total pressure formation yields" 0 pressure/Torr pentan-2-one pentan-3-one pentan-2-yl nitrate pentan-3-yl nitrate 155 0.054 f 0.008 0.250 k 0.030 0.061 0.012 0.047 & 0.006 590 f 40 0.149 f 0.017 0.353 f. 0.032 0.064 f 0.009 0.043 & 0.004 0.060 k O.OIOb 0.044 f. 0.006' " Indicated errors are the two least-squares standard deviations combined with the estimated overall uncertainties in the GC-FID response factors of +5% for pentane and the products.'Combined data at both 0 concentrations. Alkoxyl Radicals in the Atmosphere Fig. 5 Amounts of (0, 0)pentan-2-yl and (A A)pentan-3-yl nitrate formed corrected for reac- tion with the OH radical us. the amount of pentane reacted at 296 & 2 K and 740 Torr total pressure. Open symbols 155 and filled symbols 590 & 40 Torr 0,. expected since these nitrates are formed from the reactions of the pentan-2-yl and pentan-3-yl peroxy radicals with NO [reaction (8a)] in competition with the pathway forming the alkoxyl radical plus NO [reaction (Sb)]. The pentyl nitrate yields are dependent on the temperature (increasing with decreasing temperature) and total pres- sure (increasing with increasing total press~re),~~~~~~~ but should be independent of the 0 concentration providing that 0 and N have similar third-body efficiencies in reac- tion (8a).The formation yields of the pentan-2-yl and pentan-3-yl nitrates obtained from least- squares analyses of the entire data set are 0.060 f0.010 for pentan-2-yl nitrate and 0.044 & 0.006 for pentan-3-yl nitrate (Table 1). These formation yields can be compared with our previous yields at room temperature (299-300 K) and atmospheric pressure (735-740 Torr) of 0.071 & 0.009 and 0.046 0.006 re~pectively,~~ and 0.074 & 0.002 and 0.052 f0.004 re~pectively,~~ data where the indicated errors in the Atkinson et aZ.23*24 are two least-squares standard deviations. The agreement is reasonably good. Discussion The expected reactions of the pentan-2-oxyl and pentan-3-oxyl radicals formed after H-atom abstraction from the 2- and 3-positions in pentane are reaction with 0 [reaction (14a)] decomposition [reaction (14b)] and isomerisation [reaction (14c)l for the pentan-2-oxyl radical and 0 reaction [reaction (15a)] and decomposition [reaction (1 5b)l for the pentan-3-oxyl radical (isomerisation via a six-member transition state cannot occur).02+ CH,C(O)CH,CH,CH + HO (14a) ICH,CH(OH)CH,CH,&I (14c) CH,CH,C(O)CH,CH + HO (15a) CH,CH,CH(O)CH,CH CH,CH,CHO + CH,tH (15b) R. Atkinson et al. 33 The pentan-2-one and pentan-3-one formation yields should therefore exhibit a depen- dence on the 0 concentration as observed (Fig. 4 and Table 1). Plots of (yield)-' us.[O,] -' for pentan-2-one and pentan-3-one should be straight lines with intercepts of (yield of pentan-2-oxyl or pentan-3-oxyl radical from pentane) -and slope/intercept ratios of [(kdecomp + kisOm)/ko2] (where kisom< kdecomp for the pentan-3-oxyl radical).26 Such plots of (yield)-' us. [O,]-' are shown in Fig. 6 and the intercepts lead to frac- tional yields of the pentanoxyl radicals from the reaction of the OH radical with pentane in the presence of NO of 0.40 for the pentan-2-oxyl radical and 0.42 for the pentan-3- oxyl radical. Combining the pentan-2-one and pentan-3-one yields extrapolated to [O,]-= 0 with the measured pentan-2-yl and pentan-3-yl nitrate yields implies a yield of the pentan-2-yl radical from pentane of 0.46 and of the pentan-3-yl radical from pentane of 0.46.Since the calculated formation yield of the pentan-1-yl radical from pentane is 0.095,21 then a complete accounting of the pentanyl radicals formed from pentane is obtained. The formation yields of pentan-2-yl and pentan-3-yl radicals derived from this product study of 0.46 each are somewhat different from (though in agreement within the experimental uncertainties) the formation yields of 0.55 and 0.36 respectively calcu- lated by the Atkinson estimation technique.,' The values of [(kdecomp + kiSom)/ko2]obtained from the slope/intercept ratios in Fig. 6 are 3.3 x 10'' molecule cm-3 for the pentan-2-oxyl radical and 3.3 x 10l8 molecule ~rn-~ for the pentan-3-oxyl radical both at 296 2 K and 740 Torr total pressure and with estimated overall uncertainties of a factor of ca.2. Literature data for the rate constant ratios kisoJko2 and kdecomp/ko2for the butan- 1-oxy1,8-1* butan-2-0xyl,~.~ and hexan-2-oxyl and hexan-3-oxy16 radicals at room temperature are given in Table 2. Isomerisation by a 1,5-H shift uia a six-member transition state cannot occur for the butan-2-oxyl or pentan-3-oxyl radicals. Furthermore isomerisations by a 1,4-H shift via a five-member transition state are calculated to be 3-4 orders of magnitude slower at 296 K than the 1,5-H shift isomer is at ion^,^^ and to be slower than the decomposition reactions of the butan-2-oxyl and pentan-3-oxyl radicals.2.26 The rate constant ratio (kdecomp+ kisom)/ko2 obtained from the data for pentan-3-one formation shown in Fig.6 is therefore the rate constant ratio kdecomp/ko2for the pentan-3-oxyl radical (Table 2). The decomposition rate constant for the pentan-2-oxyl radical is expected to be a factor of ca. 2 lower than that for the pentan-3-oxyl radical (owing to the degeneracy factor of Fig. 6 (Yield)-' for the formation of 0, pentan-2-one and A pentan-3-one us. [OJ-' Alkoxyl Radicals in the Atmosphere Table 2 kisom/kol and kdecomp/ko2 for the pentan-2-oxyl and pentan-3-oxyl radicals at 296 & 2 K and atmospheric pressure together with room-temperature literature data for other alkoxyl radicals formed from alkanes alkoxyl radical kisomer/ko2u molecule cm -kdecomJkozamolecule cm -ref. 8 butan-1-oxyl 1.6 x 1019 (303 K) 9 (1.5 0.5) x 1019 (296 f2 K) 10 (1.9 f0.3) x 1019 (298 f2 K) butan-2-oxyl 3.1 x 10l8 (303 K) 8 (2.6 f0.35) x lo'* 9 (296 & 2 K) pentan-2-oxyl 3.1 x 10196 this work pentan-3-oxyl 3.3 x this work hexan-2-ox y l (1.7 f1.5) x lo2'' (3.2 1.8) x 6 (297 f3 K) (297 f3 K) (5.9 & 4.2) x (3.7 f2.7) x 6 (297 f3 K) (297 f3 K) hexan-3-oxyl (2.3 k0.9) x (5.3 &-2.0) x 10l8' 6 (297 f3 K) (297 f3 K) (5.4 f3.2) x 1019d (1.1 f0.8) x 1019d 6 (297 3 K) (297 f3 K) Indicated errors are two least-squares standard deviations.Estimated overall uncertainties are a factor of ca. 2. Derived from cited product yields from irradiated hexan-2-yl and hexan-3-yl nitrites-NO-air mixtures.6 Derived from cited product yields from irradiated HONO-hexane-air mixtures.6 2 for the pentan-3-oxyl radical) and the rate constants for the reactions of the pentan-2- oxyl and pentan-3-oxyl radicals with 0 are expected to be similar (see below) resulting in the rate constant ratio kisom/ko2for the pentan-2-oxyl radical given in Table 2.These rate constant ratios allow the C,-hydroxycarbonyl/pentanones formation yield ratios to be calculated as a function of the 0 concentration with values of 4.3 at 4% O, 1.3 at 21% 0, and 0.6 at 80% 0, in surprisingly good agreement with the values derived from the API-MS data of ca. 6.4 ca. 3.5 and ca. 1.5 respectively. Atkinson2 has postulated that the rate constants for the reactions of alkoxyl radicals with O, ko ,depend on the exothermicity of the reactions AH with k, = 1.3 x 10-"n exp(-0.32AH) cm3 molecule-' s-' at 298 K (1) where n is the number of abstractable H atoms in the alkoxyl radical and AH is in units of kcal mol-'.The calculated value of AH for the reaction of 0 with the butan-1-oxyl radical is -32.3 kcal mol- ' and those for the butan-2-oxyl pentan-2-oxyl pentan-3- oxyl hexan-2-oxyl and hexan-3-oxyl radicals are all -35.2 & 1 kcal mol 1.28-30 From eqn. (I) this leads to essentially identical rate constants k, of (8-10) x cm3 molecule-' s-at 298 K for the butan-1-oxyl butan-2-oxyl pentan-2-oxyl pentan-3- oxyl hexan-2-oxyl and hexan-3-oxyl radicals and these calculated rate constants are very similar to those measured for the ethoxyl and propan-2-oxyl radicals' (as expected since eqn. (I) was derived from the rate constants k, for the methoxyl ethoxyl and propan-2-oxyl radicals and the heats of reaction AH for the 0 reactions with the ethoxyl and propan-2-oxyl radicals are similar to those for the alkoxyl radicals con- sidered above).Hence the rate constant ratios kdefomp/k02 and kisom/k02 can be used to investigate the dependence of the isomerisation and decomposition rate constants on the structures of the alkoxyl radicals for which data are given in Table 2. R. Atkinson et al. As shown in Table 3 the rate constant for the decomposition of the butan-2-oxyl radical of kdecomp = 2.8 x lo4 s-' at ca. 300 K obtained relative to the O2 reactions,8i9 is significantly higher than the rate constant determined relative to the reaction with NO of 4.8 x lo3 s-' at 298 K,3 using a rate constant for the reaction with NO of kl = 4.2 x lo-'' cm3 molecule-' s-'.~The decomposition rate constant measured by Dobk et al.' for the pentan-2-oxyl radical of 1.0 x lo4 s-' at 298 K relative to k = 4.2 x lo-" cm3 molecule-' s-' is in reasonable agreement with the value of kdecomp = 3 x lo4 s-' obtained here for the pentan-3-oxyl radical relative to the rate constant for the reaction with 02,especially considering the degeneracy factor of ca.2 for the pentan-3-oxyl radical (Table 3). However Dobe et aL5 obtained a rate constant ratio of kiso,,,/kdecomp = 0.5 f0.1 at 301 K a factor of ca. 20-40 lower than derived from our present data (Table 3). This marked disagreement may well be due to difficulties in quantitatively trapping the 4-hydroxypentan-1-yl radical in the study of Dobe et aL5 The rate constant ratios kdecomp/ko2 given in Table 2 for the butan-2-0xy1'9~ and pentan-3-oxyl radicals are similar and are ca.3 x 10l8 molecule cm-3 (with the rate constant ratios kdecomp/ko2 for the hexan-2-oxyl and hexen-3-oxyl radicals calculated .~ from the product yields measured by Eberhard et ~1 from irradiated hexan-2-yl and hexan-3-yl ni tri te-NO-air and HONO-hexane-air mixtures ex hi bi ting significant scatter but being consistent with this value of kdecom,,/ko2). Using the rate constant k, = 9 x lo-'' cm3 molecule-' s-' calculated above this rate constant ratio kdecomp/ko2 leads to kdecomp = 3 x lo4 s-' at 296 K for the butan-2-oxyl and pentan-3-oxyl radicals (Table 3). The rate constant ratios kiso,.,,/ko for the butan- l-oxyl,8-'o pentan-2-oxyl and hexan-3-oxy16 radicals with the isomerisation reaction involving H-atom abstraction from a -CH3 group are all similar at ca.(2-5) x 1019 molecule cm-3 (Table 2). For the hexan-2-oxyl radical,6 the corresponding rate constant ratio kisomjkozis ca. (2-6) x lo2' molecule cm-3 (Table 2). Using the rate constant koz = 9 x cm3 molecule-' s-' calculated above these rate constant ratios kisom/ko2lead to isomerisation rate constants kisom of ca. (2-5) x lo5 s-' for H-atom abstraction from a -CH3 group and ca. (2-6) x lo6 s-' for H-atom abstraction from a -CH2- group both at room temperature (Table 3). The increase in the isomerisation rate constant for H-atom abstraction from a -CH2-group compared to a -CH3 group is expected owing to the lower C-H bond dissociation energy in the -CH2-groups compared with the -CH3 groups.1,2.4.26,27 However the measured increase in the rate constant kisom for H-atom Table 3 Reaction rates of selected alkoxyl radicals at ca. 298 K and 1 atm of air alkoxyl radical k,,[O,]/s -k,,,Js -* butan-1-oxyl 4.2 x 104 1.4 x 105 but an-2-oxyl 5.2 x 104 2.8 x lo4 4.8 103~~ pentan-2-oxyl 5.2 x 104 1.0 x 1044e 3.1 105 5 103d-e pentan-3-oxyl 5.2 x 104 3.3 x 104 hexan-2-oxyl 5.2 x lo4 ca. (4-30) x lo4 ca. (2-6) x lo6 hexan-3-oxyl 5.2 x 104 ca. (5-11) x 104 ca. (2-5) 105 'Calculated using eqn. (I); see text. From the rate constant ratios kdecorndkozand kisoJko2 given in Table 2 unless indicated otherwise.Ref. 3. Relative to the rate constant k for RO' + NO -+ RONO; using a value of k = 4.2 x 10-l1 cm3 molecule- s-1.3*5 From Dobe et aL5 36 Alkoxyl Radicals in the Atmosphere abstraction from a -CH2- group compared with a -CH group of a factor of ca. 10 is significantly lower than the predicted increase of a factor of ca. 100.2*4926927 Inter-estingly Atkinson and Aschmann31 also observed an increase of a factor of ca. 10 for alkoxyl radical isomerisation by H-atom abstraction from a -CH2- group compared with a -CH group in the alkoxyl radicals (CH,),C(OH)CH,C(o)(CH,) and CH,CH,C(CH,XOH)CH,C(~)(CH3) formed after OH radical reaction with 2,4-dimethylpentan-2-01 and 3,5-dimethylhexan-3-01 respectively. These experimental data then suggest that alkoxyl radical 1,5-H shift isomerisations involving H-atom abstraction from -CH2- groups in alkoxyl radicals formed from alkanes are signifi- cantly slower compared with H-atom abstraction from -CH groups than previously predicted.2349 '6,2 Conclusions Employing positive APCI-MS to study the OH radical-initiated reactions of pentane provided confirmation of the occurrence of alkoxyl radical isomerisation reactions via a 1,5-H shift through a six-member transition state in the atmospheric photo-oxidation of long-chain alkanes.Quantitative measurements of the pentan-2-one and pentan-3-one formation yields as a function of the 0 concentration lead to rate constant ratios for the decomposition and isomerisation reactions of the pentan-2-oxyl and pentan-3-oxyl radicals relative to their reactions with 02.These data for the pentanoxyl radical reac- tions are reasonably consistent with the analogous data for the butan-1-oxyl and butan- 2-oxyl radicals and the hexan-2-oxyl and hexan-3-oxyl radicals and the combined data set provides important information concerning the relative importance of decomposi- tion isomerisation and reaction with 0,of the alkoxyl radicals formed from the alkanes.The authors gratefully thank the US Environmental Protection Agency (Cooperative Agreement CR 821787-01-0) and the California Air Resources Board (Contract No. A032-067) for supporting this research and the National Science Foundation (Grant No. ATM-9015361) and the University of California Riverside for the funds to pur- chase the PE SCIEX API I11 MS/MS instrument.While this research has been sup- ported by the US Environmental Protection Agency and the California Air Resources Board it has not been subjected to review by these agencies and therefore does not necessarily reflect their views ;no official endorsement should therefore be inferred. References 1 R. Atkinson and W. P. L. Carter J. Amos. Chem. 1991 13 195. 2 R. Atkinson J. Phys. Chem. Ref Data. 1994 Monograph 2 1. 3 L. Batt Int. Rev. Phys. Chem. 1987,6 53. 4 R. Atkinson Atmos. Environ. Part A 1990,24 I. 5 S. Dobe T. Berces and F. Marta int. J. Chem. Kinel. 1986 18 329. 6 J. Eberhard C. Muller D. W. Stocker and J. A. Kerr Environ Sci. Technof.,1995,29 232. 7 W. P. L. Carter K. R. Darnall A. C. Lloyd A. M.Winer and J. N. Pitts Jr. Chem. Phys. Lett. 1976,42 22. 8 W. P. L. Carter A. C. Lloyd J. L. Sprung and J. N. Pitts Jr. Int. J. Chem. Kinel. 1979 11,45. 9 R. A. Cox K. F. Patrick and S. A. Chant Environ. Sci. Technof.,1981 15 587. 10 H. Niki P. D. Maker C. M. Savage and L. P. Breitenbach J. Phys. Chem. 1981,8S 2698. 11 R. Atkinson W. P. L. Carter A. M. Winer and J. N. Pitts Jr. J. Air Poflut. Control Assoc. 1981 31 1090. 12 T. Covey and D. J. Douglas J. Am. Soc. Mass Spertrom. 1993 4 616. 13 Z. Karpas G. A. Eiceman C. S. Harden and R. G. Ewing J. Am. SOL'.Mass Spectrom. 1993,4 507. 14 Y. K. Lau S. Ikuta and P. Kebarle J. Am. Chem. Soc. 1982 104 1462. 15 J. Sunner N. Gordon and P. Kebarle. And. Chrm. I988,60 1300. 16 D. J. Douglas and J. B. French J.Am. SOL'.Mas.s Spertrom. 1992 3. 398. R. Atkinson et al. 17 P. H. Dawson and W. F. Sun Int. J. Mass Spectrom. Ion Proc. 1983/1984,55 155. 18 R. K. Mitchum and W. A. Korfmacher Anal. Chem. 1983,55 1485A. 19 H. Kambara and I. Kanomata Anal. Chem. 1977,49,270. 20 J. R. Dunlop and F. P. Tully J. Phys. Chem. 1993,97 6457. 21 R. Atkinson Int. J. Chem. Kinet. 1987,19,799. 22 R. Atkinson J. Phys. Chem. Ref Data. 1989 Monograph 1 1. 23 R. Atkinson S. M. Aschmann W. P. L. Carter A. M. Winer and J. N. Pitts Jr. J. Phys. Chem. 1982 86,4563. 24 R. Atkinson W. P. L. Carter and A. W. Winer J. Phys. Chem. 1983,87,2012. 25 W. P. L. Carter and R. Atkinson J. Atmos. Chem. 1989,8 165. 26 W. P. L. Carter and R. Atkinson J. Atmos. Chem. 1985,3,377. 27 A.C. Baldwin J. R. Barker D. M. Golden and D. G. Hendry J. Phys. Chem. 1977,81,2483. 28 S. W. Benson Thermochemical Kinetics Wiley New York 2nd edn. 1976. 29 R. Atkinson D. L. Baulch R. A. Cox R. F. Hampson Jr. J. A. Kerr and J. Troe J. Phys. Chem. Re& Data 1992 21 1125. 30 National Institute of Standards and Technology Standard Reference Database 25 Structures and Properties Database and Estimafion Program Version 2.0 S. E. Stein Chemical Kinetics and Ther- modynamics Division NIST Gaithersburg MD 1994. 31 R. Atkinson and S. M. Aschmann Environ. Sci. Technol. 1995,29,528. Paper 5/00171D; Received 5th January 1995

ISSN:1359-6640    

DOI:10.1039/FD9950000023

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

FU~U~U~ 1995,100 C29-C45 D~SCUSS. Molecular Simulation A View from the Bond D. J. Tildesley Chemistry Department University of Southampton Southampton UK SO1 7 I BJ The last fifty years have seen a rapid development of the molecular dynamics method for the simulation of condensed phases. The method is used to test the accuracy of effective intermolecular potentials to validate approximate statistical mechanical theories and to explore systems at condi- tions and with a level of detail which would be hard to achieve by conven- tional experiment. In combination with data visualisation techniques the method acts as a powerful microscope to study structure and dynamics at the molecular level. The rapid progress in the field can be attributed to the ingenuity of its practitioners in developing new algorithms and in studying imaginative applications and to the striking increases in computing speed and memory over the period.We review the development of the technique and illustrate its use by con- sidering three problems the calculation of the surface tension of water; the structure of an alkyl cyanobiphenyl layer on graphite; and the testing of the wall-PRISM equation of polymer adsorption. Finally we consider the future of these simulation methods. 1. Introduction Traditionally science has progressed by the development of theory to explain experimen- tal observation. The experimentalist provides the facts from which a physical model is developed and the theoretician unravels the relationship between the mathematical vari- ables used to describe this model.For example the motion of the planets analysed by Kepler provided a physical model of the solar system that could be described mathe- matically using Newtonian mechanics. In the same way Rutherford’s experiments on the scattering of aparticles from gold foil indicated a basic planetary model for the atom which could then be described in precise detail by the Schrodinger and Dirac equations. Occasionally the mathematical description precedes the observation as in the case of Einstein’s general theory of relativity and Eddington’s observation of the deflection of light by the sun. Regardless of the precise order of their application theory and experi- ment have been the two essential pillars supporting the natural sciences.The second half of the twentieth century has seen the development of a third meth- odology. The digital computer developed during the second world war has enabled many of the branches of natural science to simulate complicated models that could not have been solved by standard theoretical methods.’ For example it is now possible to model the structure and dynamics of a protein molecule buried in a lipid bilayer using molecular dynamics to follow the propagation of a crack using finite difference methods to develop a circulation model for weather prediction or to solve the Euler equations of fluid flow to model a severe thunderstorm. Apart from addressing problems of enormous complexity the method allows the computational scientist to alter the underlying model at will.For instance by changing the dimensionality of a problem or the structure of the model haniiltonian it is possible C29 C30 Molecular Simulation A Viewfrom the Bond to test and extend approximate theories by studying systems that are not readily attain- able in the laboratory. In the discipline of Chemistry computers have allowed us to solve the Schrodinger equation at various levels of approximation and to obtain detailed predictions of the properties of atoms molecules and small clusters. The prediction of the energy levels in the molecule has also allowed us to understand and rationalise much experimental spec- troscopy. Energy minimisation Monte Carlo and molecular dynamics calculations using increasingly accurate force fields have played an essential part in the prediction of the properties of condensed phases such as liquids liquid crystals polymers and super- conducting materials.The combination of density functional theory and molecular dynamics in the Car-Parrinello methods allows us to simulate systems where the elec- tronic and nuclear degrees of freedom are not strongly coupled (e.y. chemical reactions) while detailed scattering calculations allow us to predict the rates of chemical reactions in the gas phase. This is a huge body of work and the UK chemistry community sup- ported by the provision of excellent central computing facilities has made important contributions in this new field over the last forty years. Since in some sense I represent Computational Chemistry at this celebration I trust that my colleagues will forgive me if I concentrate on one aspect of the discipline and limit myself to some remarks on the molecular dynamics method.2. A Brief History of the Molecular Dynamics Method It is intriguing that the statistical mechanics of Gibbs and Boltzmann had taken such a strong hold in the early part of the twentieth century that when computers became available just after the second world war they were used to calculate ensemble averages using the Metropolis Monte Carlo sampling method.2 It was a further four years before trajectory calculations were used to estimate equilibrium properties by the direct route of calculating the time average. The first molecular dynamics simulations were performed by Alder and Wainwright on the UNTVAC computer at the Livermore Laboratory in Calif~rnia.~ Simulations were performed of the hard-sphere model by locating the time collision partners and impact parameters for every collision.At each collision the new trajectories were com- puted from the conservation of energy and linear momentum. The properties of the system such as the pressure and the radial distribution function were calculated as time averages over the configurations of the fluid and the calculation was characterised by the conservation of the total energy and linear momentum. The effects of the small system size were mitigated by embedding the central simulation box in a lattice of repli- cas and using the method of periodic boundary conditions.Even these first simple simulations provided an interesting prediction that in the hard-sphere system there would be a first-order phase transition between a solid and ff uid an observation subsequently confirmed by more detailed calculations. This is a clear example of the hamiltonian surgery discussed in the last section. A simulation of this simplified model shows that it is unnecessary to include attractive forces to induce the formation of a solid from the liquid. Molecular dynamics of systems interacting through continuous potentials followed in a study of damage radiation by Gibson et aL4 and in a study of liquid argon by Rahman using central difference and predictor-corrector algorithms to solve the coupled second-order differential equations of the atomic motion.These simulations used a time-step of approximately s and solved the equations of motion for a few thousand time-steps. The total simulation length of a few tens of picoseconds was suffi- cient to calculate equilibrium properties and to estimate single particle correlation func- tions such as the velocity autocorrelation function and its integral the diffusion coefficient. D.J. Tildesley C31 By the early 1970s the power of machines had increased dramatically and super- computers were becoming available to academics outside the large government institu- tions. Although the first simulations were performed on atomic liquids and solids interacting through central forces it became possible to simulate rigid polyatomic mol- ecules by solving the Newton-Euler equations of motion."' This method could have been extended to study the dynamics of flexible molecules by finding the appropriate generalised coordinates and working with a time-dependent moment of inertia tensor.However a more direct approach treated a molecule as a set of atoms with the geometry defined using stiff intramolecular potentials' or rigid constraints. (A typical constraint force might act to keep a bond length fixed at its equilibrium value.) These constraints can be included in the equations of motion by introducing and solving for the associated Lagrangian multipliers.' The ability to simulate molecules of a completely general structure including both constraints and many-body intramolecular potentials allowed the method to be applied to the biological sciences in the simulation of the nucleic acids proteins and poly- saccharides.Although this type of simulation had become technically possible by the early 198Os there were still significant problems concerning the accuracy of the force fields and the total time over which the molecules could be simulated. The equations of motion must be solved at a time-step that covers the fastest molecular motion to s) while the problem of interest might have a characteristic timescale of between lop6 and 1 s. This need to simulate hundreds of millions of time-steps puts some prob- lems beyond the range of even the fastest available computers. An example of this type of problem would be the complete simulation of protein folding where we attempt to predict the tertiary protein structure from its amino acid sequence.It is important to stress that the production of accurate force fields and the extensions to long timescales are still the key problems facing users of the method today. The original molecular dynamics methods were performed in the microcanonical ensemble at constant N VE. During the 1980~~ the method was extended to other ensem- bles such as the canonical lo and isothermal-isoenthalpic ensemble' using the gener- alised Lagrangian method. Constant stress molecular dynamics in which the shape and size of the central simulation box changed to balance the internal and external stresses allowed for the direct study of solid phase transitions.I2 At the same time the develop- ment of methods for calculating the statistical properties of systems such as the free energy difference entropy and chemical potential' meant that simulation methods could be used to locate phase boundaries.The simulations described so far have been at equilibrium. The time correlation functions can be used to calculate the transport properties of the fluid using the Green- Kubo relationships. It also proved possible to perform non-equilibrium molecular dynamics simulations by introducing a gradient that induces an appropriate current of momentum or energy. The transport coefficients can then be measured directly as the ratio of the current to the gradient. The gradient say in velocity can be introduced mechanically by simulating a fluid between two structured walls moving in opposite direction^'^ or by using a homogeneous meth~d.'~ In the homogeneous method an additional perturbation term of the form is added to the hamiltonian.Vu is the velocity gradient tensor and qj and pi are the atomic position and momentum of atom i. This kind of hamiltonian is applied using moving or Lee-Edwards periodic boundary conditions. " It is clear that the last forty years has seen an enormous increase in the complexity and range of the systems that can be simulated using classical molecular dynamics. Of course it has been necessary to develop and refine the simulation algorithms but the C32 Molecular Simulation A View from the Bond predictive ability of the method is most directly coupled to the significant increase in computing power over the same period.’ The first statistical mechanical simulations were performed on the MANIAC machine at Los Alamos.In the mid-1940s a similar machine would weigh 30 tons contain over 18000 vacuum tubes 6000 switches and fill a large room. A machine of this type could calculate at the rate of 400 to 1000 floating point operations per second (flops). Originally the machines were programmed by hand- setting switches and wiring plug boards. In the United Kingdom the first stored- program computer (EDSAC) built by Maurice Wilkes became available in May 1949 and supercomputing in the UK had begun. EDSAC supported an enthusiastic com- munity in Cambridge. By the early 1970s the CDC 7600 at the University of London Computing Centre performed calculations at the speed of 40 megaflops.The vacuum tube had been replaced by the single transistor. In the early 1980s the Cray 1 computer became avail- able at the Rutherford Appleton Laboratory. The clock speed of 12.5 ns obtained using integrated circuit technology gave a peak performance of 170 megaflops. This per- formance was sustained by the vectorisation or the pipe-lining of instructions. The 1990s saw the introduction of massively parallel shared memory and distributed memory machines. The most recent UK supercomputer the CRAY T3D at the Uni- versity of Edinburgh consists of 256 DEC alpha processors working in parallel to produce a peak performance of 40 gigaflops. Real scientific applications may only sustain a performance of around 7 gigaflops on this machine but it is still approximately 10’ times faster than the first Electronic Numerical Integrator and Computer (ENIAC) used on the Aberdeen testing grounds for ballistic trajectories in 1945.This speed increase combined with a correspondingly sensational increase in size of computer memory and permanent storage has effected a revolution in Computational Chemistry. As always there are those who are not prepared to wait for technological develop- ments. During an evening discourse given in this very lecture theatre in April 1900 Lord Kelvin presented his paper ‘Nineteenth Century Clouds over the Dynamical Theory of Heat and Light’.’’ Unaware that a molecular vibration might not enjoy its full k Ts worth of energy he postulated that the ‘anomalous’ heat capacity ratio of a polyatomic gas might be a consequence of the failure of the law of equipartition of energy in vessels of an irregular shape.His assistant one William Anderson performed over 5000 individ-ual trajectory calculations by geometrical construction of elastic collisions in unusually shaped containers arguably performing the first molecular dynamics averaging. Perhaps the runs were too short or the containers too regular but they incorrectly concluded that the Boltzmann-Maxwell doctrine of the equipartition of kinetic energy could be disproved.19 Lord Kelvin begins the paper with a warm acknowledgement to his assist- ant. ‘I desire to express my obligations to Mr. William Anderson my secretary and assistant for his mathematical tact and skill’.It is unclear whether a request of this kind today would have the graduate students reaching for their EPSRC handbooks of good practice or their workstations 3. Molecular Dynamics at Work The classical molecular dynamics method fits the general model shown in Fig. 1. As long as there is a strong coupling between the nuclear and electronic degrees of freedom then a sensible input to a condensed phase simulation would be a force field consisting of a number of effective two- three- and four-body potentials. These can be estimated from accurate quantum mechanical calculations on small clusters or isolated molecules and in some cases from the inversion of dilute gas experimental data such as the second virial coefficient or the transport properties.20 This force field can be used in an energy minimisation calculation to estimate the zero temperature structure of the system.Ensemble averages at non-zero temperature D.J. Tildesley c33 The first model 1. density functional theory 2. SCF + CI 3. 2nd order perturbation theory of intermolecular potentials 4. 2nd virial coefficients 5. Transport properties 6. Beam spectroscopy 1. Monte Carlo classical statistical mechanics 2. molecular dynamics ensemble averaging 3. Brownian dynamics 4. Energy minimisation 1. condensed phase structure 2. mechanical properties observable macroscop 3. dielectric properties 4.transport properties 5. free energies and chemical potentials Fig. 1 Statistical mechanical simulations the first model can be obtained by using the Metropolis Monte Carlo method to sample the appropri- ate phase-space distribution. Molecular and Brownian dynamics calculations can provide additional information on the time evolution of the molecules. The outputs from the method are the measurable structural thermodynamic and transport properties of the system. The molecular dynamics method allows us (a) to adjust the intermolecular force field to reproduce experimental condensed phase data (validation); (b) to use the force field to calculate properties that are not easily observed in the laboratory (prediction); (c) to check the accuracy of statistical mechanical theories (testing); (d) to obtain a detailed picture of the structure and motion at molecular level (observation).We will illustrate some of these uses of the simulation methods by looking at three different examples. 3.1. The Surface Tension of Water We have used the molecular dynamics study2' to examine whether an effective pair- potential fitted to the bulk properties can reproduce an accurate estimate of the surface tension at the water liquid/vapour interface. The simulations use the SPC/E model of water that has been properly fitted to the configurational energy of the liquid by includ- ing the self polarisation term.22 The bond length of the molecule is fixed at 1 A and the bond angle at 109.47'.Charges of 0.42381el are fixed at the hydrogen nuclei and a cancelling negative charge is fixed on the oxygen nucleus. The repulsion-dispersion c34 Molecular Simulation A View from the Bond / / I' 0 \0-"t\ ,oA \' \' 0-4- I \ \ Fig. 2 The dimer configuration for the SPC/E model of water. In the SPC/E model the negative charge is placed on the oxygen nucleus. In similar models such as TIP4P and the Watanbe-KIein model the negative charge is displaced along the bisector of the HOH bond to the position marked A. interaction between the two water molecules is modelled by a Lennard-Jones potential centred on the oxygen atom. The total dipole moment of the model molecule is 2.351 D which is significantly higher than the gas phase value of 1.85 D.The higher dipole of the model accounts effectively for the enhancement of the gas-phase dipole by polarisation from the neighbouring molecules. Explicit polarisation is not included in this model. The dimer conformation for the SPC/E is shown in Fig. 2. There is a clear hydrogen bond induced by the interaction between the partial charges. This particular model works well in simulations of the bulk liquid. At 298 K and 0.997 g cm -the predicted configurational energy is -41.5 kJ mol- ' ( -41.5 kJ mol-I) the predicted diffusion coefficient is 5.3 x m s-' (2.4 x m s-') and the pre- dicted relative permittivity is 70 (78.3).23The experimental measurements at the same temperature and density are included in parenthesis. The water gas/liquid interface can be simulated using the geometry shown in Fig.3. A three-dimensional simulation of bulk water is performed in a cubic box using full periodic boundary conditions to produce an equilibrated sample. A number of these water boxes are stacked together to form a slab of liquid and the slab is then surrounded by a number of empty boxes. The entire sandwich is reproduced periodically through space. The articifial gas/liquid interface at the start of the simulation is a step function. Water molecules evaporate quickly from the liquid slab to form a coexisting gas phase vapour liquid vapour L -+ Fig. 3 The slab simulation geometry for studying thc gas/liquid interface D.J. Tildesley c35 80 I 70 60 fi 50 3 ‘E z 40 E W 30 r-20 10 0 250 300 350 400 450 500 550 600 650 Fig.4 The surface tension of water. The continuous line represents the experimental values from ref. 27. MD results using N = 512 and N = 1000 are shown as black and open circles respectively. Reproduced with permission of the American Institute of Physics. and a stable interface is formed with a density profile that decays monotonically from the bulk liquid density to that of the gas. At a planar interface the pressure is no longer a scalar property but a tensor with a normal component (P = P,,) and a tangential component P = 1/2(P, + Py,,).The ele- ments of the pressure tensor can be calculated from the Irving-Kirkwood definition of the stress:24 where p(z) is the molecular density profile through the interface T is the temperature riajBis the distance between site CI on molecule i and site p on moleculej while rij is the corresponding separation of the molecular centres of mass.u(riajD) is the site-site poten- tial function. For the oxygen-oxygen interaction this site-site potential includes the Lennard-Jones interaction and the repulsion between the partial charges. O(x)is the unit step function and the final product of two 8 functions means that the molecules on either side of a plane at z will contribute to the pressure at z. The tangential component of the pressure tensor can be calculated by replacing zijziajS by (xi,ixiajp+ yijyiajfl)/2. The normal component of the pressure tensor must be constant through the interface to ensure mechanical stability and this is an important test of equilibrium in these simula- tions.The tangential component changes as a function of z becoming negative around the Gibbs dividing surface. The surface tension can be calculated by integrating the difference in the elements of the pressure tensor through the interface. Here the factor of 1/2 arises from the two interfaces in the slab simulation. We have calculated the surface tension along the orthobaric curve by performing simulations with 500 and 1000 molecules for simulations of 100000 time-steps and cal- culating the surface tension from the pressure tensor profiles. The long-range inter- actions are included using the Ewald summation technique. The results are shown in Fig. 4. There is good agreement with the experimental surface tensions along the ortho- C36 Molecular Simulation A View from the Bond baric curve away from the critical point.This agreement is surprising given that the model takes no account of the effective dipole moment profile that must exist in the interface. The result has encouraged us to use the SPC/E model to calculate the orienta- tional ordering and the non-linear optical susceptibility profile for the water interface.26 This profile cannot be measured directly but the integrated non-linear susceptibilities can be studied experimentally using the sum frequency spectroscopy technique.26 This work illustrates the use of the method in the validation of a model and in the prediction of a profile that is difficult to observe experimentally.3.2. Imaging of Liquid-crystal Layers STM images of the n-alkyl-cyanobiphenyls (nCBs) adsorbed on graphite suggest that the first adsorbed layer adopts a highly ordered structure extending for thousands of ings- troms. The molecular arrangement within this layer depends on a fine balance between steric effects dipolar interactions and the strong van der Waals attraction of the surface. The most commonly observed structures consist of long antiparallel rows of mol- ecules.28 Close examination of the STM images indicates a slight interdigitation between neighbouring rows and small periodic kinks in the arrangements within these rows. Fig. 5 is a sketch of one of the commonly observed layer structures. For 8-CB these kinks are observed to occur after every four pairs of antiparallel molecules indicating a unit cell of eight molecules.The molecular model used for 8-CB was the 22-site united atom model described in ref. 29. A monolayer of adsorbed 8-CB using a system of 72 molecules was simulated with two-dimensional rhombic periodic boundary conditions. Nine separate simulations were performed each at a different combination of head-group area per molecule A, and temperature T.30 In an attempt to mimic the effect of an AFM probe we have performed three addi- tional simulations with the adsorbate sandwiched between the substrate and a smooth Lennard-Jones 9-3 half-space. This half-space positioned 7 8 above the substrate so Fig. 5 n-Octyl-cyano biphenyl adsorbed on graphite.A schematic view of the smectic layer. (Reproduced with permission of Taylor and Francis Ltd.) D.J. Tildesley c37 that it exerts a small net repulsive force on the monolayer inhibits the out-of-plane motion of the molecules. In a separate run aimed at beginning our investigation of the effect of bulk liquid crystal above the monolayer we have performed a single simulation of a bilayer of 8-CB using an initial configuration created from two planes of the energy minimised monolayer. Experimental STM and AFM images of adsorbed structures do not yield snapshots of static structures; rather as the tip is raster-scanned across the surface it records a time- averaged image of the sample below it. We can therefore hope to make better contact with experimental results by performing a similar time-averaging of our simulation con- figurations.In this spirit Plate 1 shows contour plots of (p(x y)) the simulation- averaged number density of atomic sites projected onto the surface plane. These averages are taken over the entire production phase of each run. Plate l(a) shows (p(x y)} for a high-density system A = 118.21 A2 at 298 K with planar probe above the adsorbate. When confined in this way the system remains highly ordered throughout the run. Individual molecules are clearly visible ; their locations remain virtually unchanged as the simulation progresses. This figure also displays the well defined periodic dislocations or kinks that define the eight molecule unit cell structure. If the probe is removed but the density and temperature are left unchanged Plate l(b) the system relaxes significantly.While individual molecules are still visible within the plot and the smectic strips remain intact the periodic kinks are no longer visible. At low densities and high temperatures Plate l(c) the system separates into small ordered domains coexisting with fluid regions. The hole visible on the right-hand edge of this Plate was stable throughout the production cycle of 40 ps. These results support the view that the probe helps to stabilise the first layer structure by preventing out of plane motion of the mesogens. The MD simulations described in this paper were all performed using the DL-POLY suite of routines3' developed for Collaborative Computational Project 5 of the EPSRC.This modular package was chosen since source code is supplied in full so that problem-specific in-house subroutines (such as the molecule-surface potential) can be easily included. Additionally the DL-POLY offers a high degree of platform inde- pendence since it can be run on both distributed-memory parallel computers and single- processor workstations. This work is an example of the increasingly important use of simulation technique to support experimental studies. The prediction of adsorbate structure from experiments such as the proximal probe methods neutron reflectivity sum frequency spectroscopy and second harmonic generation is model dependent. Simulation can help to explore the assumptions at the heart of these models and to build confidence in the interpretation of the observations.3.3. Polymers at a Hard Wall Theories of molecular fluids advanced significantly in the 1970s with the development of the RISM (reference interaction site model) by Chandler and co-~orkers.~~ This theory treats molecular fluids as collections of atoms correlated by interactions through the fluid [described by hay(r)]and interactions within the molecule [described by uay(r)]. The theory introduces an Ornstein-Zernike-like equation which defines a site-site direct correlation function cay(r)and uses the Percus-Yevick closure to solve the equation. The theory was extended to predict the properties of polymer melts by Schweizer and C~rro.~~ The PRISM theory results in a scalar Ornstein-Zernike-like equation with an intramolecular correlation function appropriate to a freely jointed chain.This theory gives a reasonable account of the structure of a bulk polymer melt. Theories of bulk fluids can often be extended to describe adsorption by considering a two-component mixture of atoms and allowing the radius of one of the spheres (that will C38 Molecular Simulation A View from the Bond Plate 1 A contour plot of the adsorbate density (y(x y)) averaged over the production phase of each simulation. The scale is the percentage probability of finding an atom (a)A = 118.21 A2 at T = 298 K in the presence of the planar probe; (h)A = 118.21 A2 at T = 298 K for an isolated monolayer. (c) A = 144.33 A’ at T = 350 K for an isolated monolayer.(Reproduced with the permission of Taylor and Francis Ltd.) form the wall) to become infinite.34,35 The method has been extended to the polymer RISM equation by Yethiraj and Hall.36 The wall-PRISM equation can be written as idk)= Chb(k) + Npc Lb(k)lZ;p(k) (4) D. J. Tildesley c39 where a tilde denotes a one-dimensional Fourier transform and the hat the correspond- ing three-dimensional transform. hb(r) is the total site-site correlation function describ- ing the distribution of polymer sites in the bulk field. cl)b(i") is the corresponding intramolecular distribution of sites within the polymer in the bulk fluid. y,(z) is the radial distribution function describing the distribution of sites with respect to the hard surface and C,(r) is closely related to the corresponding direct correlation function c,(z).1 In these equations N is the number of sites in the polymer and pc the density of poly- mers. The solution of the wall-PRISM equation requires a knowledge of the bulk fluid distributions hb(r) and cl),,(r) characterising the polymer and a closure relationship between cw(z)and hw(z).hb(i")can be obtained by solving the PRISM equation and the closure relationships for a slit-like pore of width H centred at the origin are This closure is analogous to the Percus-Yevick approximation for hard spheres. Three forms of the polymer intramolecular distribution function have been examined in a recent simulation study:37 (a) the freely jointed chain model where wheref'= sin(kf/kP) and f is the distance between sites in the polymer; (b) the non-overlapping freely jointed model where the overlaps between non-bonded monomers are excluded and The detailed expression for GTcan be found elsewhere;38 (c) a tabulated &b(k) calculated directly from simulation studies of polymer fluids close to a hard wall.Monte Carlo simulations using a reptation algorithm have been performed to test the wall-PRISM (These simulations could have been readily performed using the molecular-dynamics method.) A sketch of the simulation geometry is shown in Fig. 6. The polymers are free to move in the space between the hard walls and their periodic boundary conditions in the direction parallel to the wall. We have studied models for butane heptane and decane in a slit composed of hard walls.The hydrocarbon chains consist of united-atom methyl and methylene groups represented as hard spheres. The torsional potential in the model is the polynomial approximation of Ryckaert and Bellemans [utOrs(4)/kB)]/K= 11 16 + 1462 cos 4 -1578 COS~4 -368 COS~4 + 3156 COS~4 -3788 cos' 4 (9) The bond lengths and angles within the alkane chains are fixed at 1.53 A and 109.47O respectively and the simulations were performed at 300 K. The temperature simply defines the torsional distribution. C40 Molecular Simulation A Viewfrom the Bond Wall Fig. 6 A schematic view of the simulation of polymer adsorption. The structure of the polymer in the bulk phase is characterised by intermolecular and intramolecular distribution functions hb(r) and w(r’)respectively.The distribution with respect to the surface is described by the density profile p(z). Fig. 7 shows a typical monomer density profile for a fluid of heptane molecules at a high bulk monomer density pg = po3 = 1.390. The profile shows half of the simulation cell and one of the two hard walls at z = 0. The three models of the intramolecular structure factor yield different results. It is encouraging that the solution using the simu- lated h(k) is in best agreement with the simulated profile. The integral equation under- estimates the contact values and hence the pressure [BP = pb(0)]. The theory also underestimates the depth of the first minimum and the height of the following peak although they are at correct values of z.The results with the NFJC structure factor follow those of the simulated &(k) quite closely at all wall-monomer separations with a 3.5 1.5 1 05 0 2 4 6 Z Fig. 7 The monomer density profile p*(z) for athermal heptane chains between hard walls. The density profiles are plotted as a function of reduced distance from the wall z. Simulated density profiles (0)are compared with the wall-PRISM predictions with h(k)from the FJC (-..-..-) from NFJC (. . .) and from the simulation (-). The bulk monomer density in this simulation a was pz = 1.390. D.J. TildesEey C41 slightly poorer agreement at contact. The FJC chain model is clearly poorer than the other two models. The overall agreement regardless of the precise form of the intramol- ecular correlation function is slightly poorer at lower densities and higher elongations.This study is a good example of the way that simulation can be used to test simple theories of condensed-phase systems. 4. The Future There are a number of directions in which we can expect significant advances in molecu- lar simulation over the next decade. Total simulation times in molecular dynamics are limited by the time-steps of between 0.5 and 5 fs used to cover the fastest molecular motions. It seems likely that improvements in the speed of the individual processors in the massively parallel distributed memory machines will allow simulation of up to 50 x lo6 time-steps and trajectories of approximately 0.1 ps. These very long timescale molecular dynamics calculations will become important as we study phenomena such as crack propagation crystal growth and protein rearrangements at larger distance scales and longer timescales.The algorithms for performing molecular dynamics calculations in parallel such as the replicated data systolic loop and domain decomposition methods,40 are now well understood and increasingly implemented in academic and commercial simulation codes. There is still some debate about the best way of handling long-range interactions such as the Coulomb potential that decrease as slowly as r-’ where D is the dimension- ality of the system. For small systems the Ewald method is well understood and can be efficiently implemented. For large systems the Ewald methods scales as 0(N3I2)and there is much active research in developing O(N)or O(N In N) algorithms.These general approaches follow the ideas of Greengard and Rokhlin4’ in which the electrostatic inter- actions between well separated parts of the system are handled through multipole expansions. We can look forward to a simplification of these algorithms and some clear statements about the systems sizes at which they really become more effective than the Ewald summation. For some phenomena at longer times the brute force approach of increasing the length of the simulation is inappropriate. This is the case when the observable is inde- pendent of the fast degrees of freedom and they can be integrated out of the model. The solution to this problem is to use a more coarse-grained model hamiltonian.For example an atomic model of a poly(ethy1ene glycol) polymer could be replaced by a spring-bead model of a polymer in which each spherical force site represents approx- imately fifty monomer units. The increase in the mass of the force site allows a scaling of the time-step in a dynamics calculation and the fast vibrational motions associated with the chemical bonds can be avoided. Additional increases in time-step can be achieved by switching from a full molecular dynamics algorithm to the Brownian or Stokesian equa- tions of motion. Using these methods we can solve the dynamics of small colloidal aggregates with hydrodynamic forces represented by a truncated multipole expansion. Finally in some simulations we will wish to avoid an explicit representation of the particles and concentrate instead on volume elements of the fluid.The lattice Boltzmann or dissipative molecular dynamics methods are rule-based implementations of the dynamics of fluid ‘particles’. Certainly the lattice Boltzmann method will reproduce the solutions to the Navier Stokes equations in certain well defined limits and this helps to make the connection between classical molecular dynamics and the methods of compu-tational fluid dynamics. One of the major challenges in the area of mesoscale dynamics is to relate changes in the coarse-grained hamiltonian to changes in the underlying atomic model. One inter- esting advance in this field has been the bond-fluctuation model described by Binder42 C42 Molecular Simulation A View from the Bond in which an atomic model is mapped on to a lattice model in an attempt to predict the glass transition temperature of the polymer melt.There are still important advances to be made in the accurate developments of force fields.43 There is a general move towards so-called class three force fields which include a more accurate description of the intramolecular potentials. In particular they include the cross terms which might describe the effect of changes in bond lengths and bond angles on torsional potentials. In some cases there is a need to use a more accurate representation of the electrostatic potential around a molecule than can be provided by the use of distributed partial charges. An accurate representation of the potential can be obtained using the distributed multipole analysis of Stone and Price but an implementa- tion of this representation within the standard commercial and academic simulation programs is still required.Finally in some situations it will be necessary to include polarisation effects by allowing the the electrostatic field to induce an additional dipole at a neighbouring molecule. This can be achieved using a multipole expansion method.44 Alternatively it is possible to use a classical Car-Parrinello approach in which additional partial charges distributed in a molecular frame can change their mag- nitude under the influence of an external field subject to constraint of total charge neutrali ty.45 All the simulation work discussed in this review has been within the framework of the first model illustrated in Fig.1. In this case well defined intermolecular potentials are used with a classical simulation technique to make predictions. Chemistry normally involves the making and breaking of bonds and thus significant rearrangements of the electrons with respect to the nuclei. In these circumstances the effective pair potential fitted between two closed shell atoms is no longer useful and we must include our quantum mechanics at the heart of the simulation technique. In other words we must perform an electronic structure calculation for a given disposition of nuclei and use the Hellman-Feynman force to move the nuclei. This model is illustrated in Fig. 8. Perhaps the most successful approach in this area has been the Car-Parrinello simu- lation technique.46 In this method the electronic degrees of freedom are normally rep- resented by a plane wave expansion of the single-electron wave function $i.Car and The second model Quantum mechanics Molecular dynamics Density functional theory Move nuclei classically.using the local density approximation. Solve the dynamics of the Interaction of electrons with the core plane wave coefficients described by a pseudopotential. including or th og on a Iity. Properties of materials ~ Covalent bonding surface rearrangements electronic properties of condensed phases Fig. 8 The second simulation model D.J. Tildesley c43 Parrinello set up a generalised Lagrangian where the upper case variable represent the nuclei and pi is a fictitious mass used to define the electronic 'kinetic energy'.Aij is a Lagrangian multiplier used to impose orthogonality on the electronic wavefunctions. This Lagrangian produces the normal Newtonian equation of motion for the nuclei and an equation of motion for the wave- function or the plane wave coefficients. In a periodic simulation the single particle orbitals entering the lagrangian satisfy the Bloch theorem and can be expanded in plane waves where g is a reciprocal lattice vector of the molecular dynamics box and the wave vector k lies within the Brillouin zone of the reciprocal lattice of the box. c:(g) are the Fourier components of the single-particle wavefunction and are the normal dynamical variables in the calculation.The wavefunctions at a particular time-step are used to calculate the electronic density p(r). The total energy can be calculated from the Hohenberg-Kohn functional with the ion-ion interaction. The derivatives of this energy are used in the CP equations of motion. The interaction of the valence electrons with the nuclei and their core elec- trons can be handled using a transferable pseudo-potential. The orthogonality of the orbitals is maintained directly in the equations of motion using the SHAKE algorithm. Alternatively the wavefunction dynamics can be solved by minimising the electronic energy for a particular nuclear configuration using a conjugate gradient algorithm and insuring orthogonality using the Gram Schmidt scheme.This method has been widely applied in the UK by the CP consortium to study metal oxides insulator metal transitions the positions of counter-ions in zeolites the dissociation of C1 at a semiconductor surface silicon surface reconstructions and molecule-metal potential surfaces. There is also active research into hybrid quantum mechanical and classical methods. In this case a small portion or cluster is embedded in a condensed system. The cluster or region of special interest is treated quantum mechanically using self-consistent Hartree-Fock density functional or semi-empirical methods while the region outside the cluster is treated using a classical force field. One difficulty with this technique is in the treatment of the boundary between the classical and quantum regions since the defini- tion of the quantum region may require the breaking of a chemical bond.The problem of dangling bonds can be reduced by including additional link atoms at the boundary which conserve bond order and saturate the valencies of the cluster. This type of simula- tion is being used to study formamide in water. The formamide fixed in a number of different rotational conformations is treated quantum mechanically while at the same time classical Monte Carlo simulations are performed on the This type of simulation method is now available in commercial modelling software.48 c44 Molecular Simulation A View from the Bond The principal advantage of the quantum models as described is that they avoid the use of empirical pair potentials.It is important to remind ourselves that in these cases we are not solving the time-dependent Schrodinger equations. The dynamics of the nuclei are classical and these methods would not for instance provide useful informa- tion on the tunnelling of a light hydrogen atom in the active site of a protein. The methods described here allow us to generate trajectories on the ground electronic state. There are separate developments in the generalisation of the molecular dynamics method to include the possibility of transitions between electronic states driven by nuclear motion.49 Despite these recent advances in ab initio simulation there is still considerable scien- tific mileage in the idea of an effective pair potential. Classical molecular dynamics can explore phase space approximately one hundred times faster than the Car-Parrinello method and will for the present remain the method of choice for studying large systems.There are many important problems in condensed matter physical chemistry that are adequately described by classical potentials and where the switch to a meso-scale description will be the critical development. Classical molecular dynamics will continue to provide detailed validated information on microscopic structure and dynamics offer- ing a grandstand view from the bond. I would like to express my thanks to the graduate students and colleagues with whom I have worked on the areas presented in this review Jose Alejandre Gustavo Chapela Martin Callaway Douglas Cleaver and Giulio Galassi.I would like to thank Dr. V. Sokhan for reading the manuscript. References 1 W. J. Kaufmann and L. L. Meyers Supercomputing and the Transformation of Science Scientific Amer- ican Library W. H. Freeman New York 1993. 2 N. Metropolis A. W. Rosenbluth M. N. Rosenbluth A. Teller and E. Teller J. Chem. Phys. 1957 21 1087. 3 B. J. Alder and T. E. Wainwright J. Chem. Phys. 1957,27 1208. 4 J. B. Gibson A. N. Goland M. Milgram and G. H. Vinyard Phys. Rev. 1960 120 1229. 5 A. Rahman Phys. Rev. A 1964,136 A405. 6 A. Rahman and F. H. Stillinger J. Chem. Phys. 1971,55,3336. 7 D. J. Evans and S. Murad Mol. Phys. 1977,34,327. 8 B. J. Berne and G. D. Harp Adv. Chem. Phys. 1970,17,63. 9 J. P. Ryckaert G.Ciccotti and H. J. C. Berendsen J.Comput. Phys. 1977,23,327. 10 S. Nose J. Chem. Phys. 1984,81 51 1. 11 H. C. Anderson J. Chem. Phys. 1980,72,2384. 12 M. Parrinello and A. Rahman J. Appl. Phys. 1981 52 7182. 13 D. Frenkel and A. J. C. Ladd J. Chem. Phys. 1974,81,3188. 14 W. T. Ashurst and W. G. Hoover Phys. Rev. Lett. 1973,31,206. 15 W. G. Hoover D. J. Evans R. B. Hickman A. J. C. Ladd W. T. Ashurst and B. Moran Phys. Rev. A 1980,22 1690. 16 A. W. Lee and S. F. Edwards,J. Phys. C 1921,5 172. 17 J. Palfreman and D. Swade The Dream Machine Exploring the Computer Age BBC books London 1991. 18 Lord Kelvin Phil. Mag. 1901 2 I. 19 W. W. Wood Early History of Computer Simulation in Statistical Mechanics Proceedings of the Enrico Fermi Summer School Varenna 1986 pp. 3-14. 20 G.C. Maitland M. Rigby E. B. Smith and W. A. Wakeham Intermolecular Forces Clarendon Press Oxford 1981. 21 J. Alejandre D. J. Tildesley and G. A. Chapela 1995 102,4574. 22 A. R. van Buuren S-J. Marrink and H. J. C. Berendsen J. Phys. Chern. 1993,97,9206. 23 K. Watanabe and M. L. Klein Chem. Phys. 1989,131 157. 24 J. S. Rowlinson and B. Widom Molecular Theory of Capillarity Clarendon Oxford 1982. 25 V. Sokhan and D. J. Tildesley Mol. Phys. manuscript in preparation. 26 Y. R. Shen The Principles ofNonlinear Optics John Wiley 1984. 27 International Critical Tables McGraw Hill New York 1928 vol. 111. D.J. Tildesley c45 28 J. S. Foster and J. E. Frommer Nature 1988,333 542; J. K. Spong H. A. Mizes L. J. LaComb M. M. Dovek J. E. Frommer and J. S.Foster Nature 1989,338 137; W. Mizutani M. Shigeno M. Ono and K. Kajimura Appl. Phys. Lett. 1990,56 1974. 29 D. J. Cleaver and D. J. Tildesley Mol. Phys. 1994 81 781. 30 D. J. Cleaver M. J. Callaway T. Forester W. Smith and D. J. Tildesley Mol. Phys. 1995,86 613. 31 DL-POLY is a parallel molecular dynamics simulation package developed at t!ie Daresbury Labor- atory by W. Smith and T. R. Forester under the auspices of the EPSRC for the Collaborative Compu- tational Project for Computer Simulation of Condensed Phases (CCP5) and the Advanced Research Computing Group (ARCG) at the Daresbury Laboratory. 32 D. Chandler in The Liquid State of’ Matter Fluids Simple and Complex ed. E. W. Montroll and J. L. Lebowitz Studies in Statistical Mechanics Vol. VIII 275-340 North Holland Amsterdam 1982 vol.VIII pp. 275-340. 33 K. S. Schweizer and J. G.Curro J. Chem. Phys. 1988,89,3342. 34 D. Henderson F. F. Abraham and J. A. Barker Mol. Phys. 1976,31 1291. 35 Y. Zhou and G. Stell Mol. Phys. 1989,66 791. 36 A. Yethiraj and C. K. Hall J. Chem. Phys. 1991,95 3749. 37 G. Galassi and D. J. Tildesley in preparation; G. Galassi Ph.D. Thesis University of Southampton 1993. 38 K. S. Schweiser and J. G. Curro J. Chem. Phys. 1988,89,3350. 39 R. Dickman and C. K. Hall J. Chem. Phys. 1988,89,3168. 40 P. Hilbers and K. Esselink in Computer Simulation in Chemical Physics ed. M. P. Allen and D. J. Tildesley Kluwer The Netherlands 1992 pp. 473-496. 41 L. Greengard and V. Rokhlin J. Comput. Phys. 1987 73 325; H. C. Petersen D. Soelvason J.W. Perram and E. R. Smith J. Chem. Phys. 1994,101,8870. 42 K. Binder in Computer Simulation ofPolymers ed. E. Colbourne Longman 1994 pp. 91-129. 43 W. D. Cornell P. Cieplak C. I. Bayly I. R. Gould K. M. Merz Jr. D. M. Ferguson D. C. Spellmeyer T. Fox J. W. Caldwell and P. A. Kollman J. Am. Chem. Soc. 1995 117 5179. 44 G. Ruocco and M. Sampoli Mol. Phys. 1994,82 875. 45 M. Sprik in ref. 40 pp. 21 1-296. 46 D. K. Remler and P. A. Madden Mol. Phys. 1990,70,921. 47 I. Hillier personal communication. 48 See for instance The Solids Embed module available in the Catalysis program from Biosym/MSI Biosym Technologies Inc. 9685 Scranton Road San Diego CA 92121-2777 USA. 49 D. Coker in ref. 40 pp. 3 15-376. Faruday Discussion 100 Celebration Puper ; Presented 24th April 1995

ISSN:1359-6640    

DOI:10.1039/FD9950000C29

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Far~d~y 1995,100 39-54 D~SCUSS. Role of Adducts in the Atmospheric Oxidation of Dimethyl Sulfide Stephen B. Barone Andrew A. Turnipseed and A. R. RavishankaraT National Oceanic and Atmospheric Administration Aeronomy Laboratory 325 Broadway Boulder CO 80303 USA and Department of Chemistry and Biochemistry and the Cooperative Institute for Environmental Sciences University of Colorado Boulder CO 80309 USA Based on the results from recent laboratory studies and field observations we suggest that adduct formation may play a key role in the atmospheric oxidation of dimethyl sulfide (DMS). Experimental evidence for the forma- tion of the DMS * OH and CH,SO * 0 adducts are described and used in a simple ‘box model’ to calculate the ratio of [MSA]/[S0,2-] observed in the atmosphere.These calculated ratios are compared with those inferred from field measurements. Based on these comparisons we suggest that the changes with temperature in the mechanism of the OH + DMS reaction one of the primary initiation steps in atmospheric DMS oxidation are unlikely to be the source of the variation in the [MSA]/[S0,2-] ratio with temperature. We suggest that the competition between the thermal decom- position of adducts such as CH,SO and their bimolecular reactions with atmospheric species is the likely cause of the variation. Finally a few observ- ations needed to better our understanding of the DMS oxidation are identi- fied. Dimethyl sulfide (DMS CH,SCH,) emitted by oceanic phytoplankton is the major natural source of sulfur in the troposhere.It is a steady source unlike the sporadic volcanic emissions. Much effort has been focussed during the past two decades on understanding the atmospheric oxidation (see for example ref. 1) of DMS via atmo-spheric field measurements modelling studies and laboratory (direct and chamber) investigations of the kinetics and product identities. However many uncertainties still remain in quantifying the eventual fate of DMS in the atmosphere. This uncertainty is primarily due to a lack of understanding of the oxidation mechanism. The extent of branching to stable end-products and the variation of this branching with temperature and composition are unclear. Such information is essential for evaluating the role of DMS in the atmosphere and assessing its impact on Earth’s climate.The likely major end-products of DMS oxidation are species such as sulfur dioxide sulfur trioxide and methanesulfonic acid (CH,SO,H MSA). SO and MSA have been measured in chamber studies and in the The effect of DMS on the climate is critically dependent on the production of gas-phase sulfuric acid and new particles. If SO and/or SO3 is produced new particles can be produced because both will lead to gas-phase H,SO, at least part of the time. In contrast the formation of MSA is not expected to lead to new particles. There are subtle but important differ- ences even between SO2 and SO production. If SO2 is produced in DMS oxidation it t Address correspondence to this author at NOAA/ERL R/E/AL2 325 Broadway Boulder CO 80303 USA.39 Atmospheric Oxidation of Dimethyl Sulfide is expected to lead to a more diffuse distribution of H2S04 while SO3 generation should lead to immediate conversion into H2S04. In the latter case the emitted DMS will likely influence only the marine boundary layer. This expectation is based on the lifetime of SO3 being very short owing to reaction with water and that of SO being a few days. In addition if the DMS oxidation mechanism is well characterized and the variation of the yields of end-products (and intermediates) with temperature and atmospheric composition are quantified atmospheric abundances of products will provide insight into the concentrations of oxidants and temperature. This would be valuable for predict- ing the future as well as for understanding the pre-industrial atmosphere.In the latter case DMS must have been the predominant source of sulfur. Interpretation of ice-core data would also be greatly facilitated by such an understanding of the DMS oxidation mechanism. In this paper we argue that sulfur-containing molecules and free radicals are prone to forming adducts with OH and 02,and the thermal stability relative to their bimolecular reactions is responsible for the changes in the branching between MSA and sulfate in the atmosphere. This thesis rests on the simple fact that unimolecular decom- position reactions important in the atmosphere have very large activation energies and hence the decomposition rate can change by orders of magnitude for small changes in temperature.In contrast most atmospheric bimolecular reactions are not very sensitive to temperature. When the product branching depends on the competition between unimolecular reactions and bimolecular reactions it can change rapidly with tem-perature. To show the importance of these adducts we have carried out a simple modelling analysis of the DMS atmospheric oxidation based on kinetic and product studies. Field measurement data have been used to define the range of possibilities for the parameters. Thus to some extent our approach links observations from the atmosphere and ice- cores with the results of the numerous laboratory studies via a proposed DMS oxidation mechanism. Surprisingly we can draw quite a few conclusions about the DMS oxida- tion mechanism and identify future studies that will be most productive.Laboratory Studies of Adduct Formation in DMS Oxidation Laboratory studies of the reactions involved in the atmospheric oxidation have been carried out by many The rate coefficients for the initiation reactions needed to quantify the atmospheric lifetime of DMS are now reasonably well defined However the subsequent reactions which influence the end-product yields are not well understood. Here we briefly discuss the approach we have taken in our laboratory to understand the mechanism of the OH-DMS reaction and the reactions of the CH3S radical. These studies are of particular significance since both involve formation of adducts. Based on these studies we speculate on the bond strengths of other possible adducts in DMS oxidation.Experimental All measurements discussed here were carried out using the technique of pulsed laser photolysis-pulsed laser-induced fluorescence (PLP-PLIF). The PLP-PLIF apparatus and the experimental procedures used to measure rate coeficients have been described in detail else~here.'~.'' The OH and CH,S radicals were detected by LIF in an excess of other reagents. OH was monitored by exciting the (A 2C+ u' = 1) +(X 211,u" = 0) band at ca. 282.1 nm using a pulsed XeCl excimer laser pumped dye laser (0.10-1.2 mJ pulse-' cmP2 pulse width CCI. 8 ns 10 Hz). CH,S radicals were detected by exciting the (A 2A, v; = 1) + (X 2E ulj = 0) transition at 371.4 nm. OH radicals were typically generated cia the photolysis of H202 at 248 nm while CH,S radicals (when S.B. Barone A. A. Turnipseed and A. R. Ravishankara they were reactants) were generated by the 248 nm photolysis of DMS or DMDS (CH,SSCH, dimethyl disulfide). The OH (or CH,S) temporal profile was obtained by measuring the LIF signal at various delay times between the photolysis and probe lasers (5 ps-50 ms). The detection limit defined as SIN = 1 where S is the time-zero signal and N is equal to twice the standard deviation in the mean value of the measured back- ground was typically ca. 4 x lo8 molecules cm- for OH and 4 x lo9 molecule cm-3 for CH,S in 100 Torr He upon integration of 100 laser shots. All experiments were carried out by slowly flowing the gas mixture through the reactor.The concentrations of the individual stable gases were determined from pressure and flow-rate measurements or via UV absorption measurements. Status of Results in the OH + DMS Oxidation The reaction of OH with DMS is a key step for the initiation of DMS oxidation. This is also the first place in the mechanism where branching to yield different products can take place.g-' ' Recently we have focussed on the rate coeficients and products of the 9 OH reaction with DMS. The following is a short synopsis of experiments which have been recently completed in our laboratory. A complete description of the experiments and the obtained results will be published in the near future along with detailed dis- cussion of the chemical simulations involved in obtaining most of the values reported here.Along with some recent results from other laboratories these studies lay the frame- work for understanding the branching in the initiation step in the oxidation of DMS in the atmosphere. OH + DMS C2H,]DMS (k, k2) The OH + DMS reaction can proceed via several possible channels OH + DMS( +M) @ CH,S(OH)CH,( + M); AHo = -11.0 kcal mo1-' (If lr) OH + DMS + CH,SCH2 + H,O; AHo = -25.5 kcal mol-' (la) OH + DMS -+ CH + CH,SOH AHo = 0 3 kcal mol-' (1b) OH + DMS + CH,S + CH30H; AHo = -18.8 kcal mol-' (W Initially the overall rate coefficients for reaction (1) and the reaction of fully deuteriated DMS OH + [2H,]DMS -+ products (2) were measured under pseudo-first-order conditions ([DMS] 2 100[OH]) at 298 K in the absence of 0,.[Note that reaction (2) has the same possible product channels as reaction (l).] The values k1(298 K) = (4.95 & 0.35) x lopt2 and k,(298 K) = (1.75 & 0.25) x are in good agreement with other recent measure-ments.127 16.17 At low temperatures (T < 240 K) [OH] temporal profiles were characterized by an initial rapid exponential decay followed by a slower exponential decay. The slower decay is due to channels la-lc accounted for the rate coefficient kbi and is slower for C2H,]DMS than for DMS. The faster decay is due to the rapid approach to equilibrium oia the addition pathway and is controlled by the rate coefficients for addition of OH to DMS (k,) and the thermal decomposition of the adduct back to reactants (kf). By fitting the temporal profiles with a non-linear least-squares fitting routine FACSIMILE,' the parameters k, k and kbi were determined and the equilibrium constant K, 42 Atmospheric Oxidation of Dimethyl Sulfide measured from the ratio of k and k,.In the above equation AN = -1. The equilibrium constants were determined over the temperature range 217-245 K. A second-law analysis of the K us. 1/T data shown in Fig. 1 yields the following thermochemical values at 229.5 K A,, H" = -10.2 f2.0 kcal mol-'; A,, So = -28.4 f.6.4 cal mol-' K-l. Using AS" ( = -31.1 at 229.5 K) and AC; calculated from statistical mechanics a third-law calculation yields A,, H"(298 K) = -10.7 & 2.5 kcal mol- '. This leads to A Ho[2H6]DMS * OH) of -10.3 f2.5 kcal mol-' at 298 K. Equilibration of OH with C2H6]DMS has also been recently observed by Wine and co-w~rkers'~ who report A,, H" = -14.2 2.6 kcal mol-I and A,, So = -42.2 10.8 cal mol -' K -' obtained from a second-law analysis.However their measured equi- librium constants are not very different from those of the present study. There are also two contradictory ab initio studies of reaction (If). Turecek2' predicts that there is no bound DMSeOH species; however McKee2' predicts a DMSsOH bond strength of ca. 6 kcal mol-' and states that this may be underestimated by ca. 4 kcal mol-'. Our observation of the equilibration clearly supports the existence of the DMS-OH adduct and supports the calculation of McKee. The DMS-OH adduct is a bound species with a bond dissociation energy of ca.10-14 kcal mol-'. It is conceivable that the adduct could fall apart to give species other than DMS and OH. For example it may be possible to form CH,SOH + CH or CH,OH + CH,S. As shown later the possibility of the formation of CH,OH + CH,S can be neglected. Such reactions of the adduct would manifest as bimolecular reactions in our system. Products from Reaction (1). When OH was generated at 298 K in the presence of DMS production of CH,S radicals was not observed. Based on the relative detection sensitivities of our apparatus for OH and CH,S an upper limit for the branching ratio for kl was determined to be QlC = k,,/k < 0.04. Neither CH,SCH or CH were detected directly by this study but they were converted into CH,S and OH respectively by the addition of 0 and NO.CH,S and OH were detected by LIF. The results of Fig. 1 K us. 1000/T (van't Hoff plot) for reaction (1). K was measured over a range of pressures (30-100 Torr) and bath gases (He N and SF,). The error bars shown are the 95% confidence limits in K and are obtained from the data analysis. The solid line drawn is the weighted least- squares fit to the data (the second-law analysis) and the dashed line represents the fit determined from a third-law analysis (see text for details). S. B. Barone A. A. Turnipseed and A. R. Ravishankara these studies in 20 Torr of 0 at 298 K showed that k,,/k = Q1 = 0.86 f0.26 and k:' = k, + k, = (5.4_+ 1.1)x lo-' cm3 molecule-' s-'. The obtained value of mla is in excellent agreement with that measured recently by Stickel et al.,, who determined Old = 0.84 f0.15by measuring the HOD appearance from the reaction of OD radicals with DMS.Channel (lb),which appears to be small cannot be ruled out from the current experiments because OH regeneration was observed at long reaction times. However the OH regeneration in this system could be due to the reaction CH,SCH,O + 02 +CH3SCHO + HOz (3) followed by the conversion of HO to OH via reaction with NO. Reaction (3) was suggested by Butkovskaya and LeBras2 to explain their observed products in the NO, + DMS reaction in 0,. Our study indicates that under the experimental conditions used here (20 Torr 0, 298 K) channel (la)(the H atom abstraction) is the dominant channel and primarily leads to CH,S formation.0 + DMS * OH + Products (k4) Previous kinetic studies have shown that k and k increase with [02]16*24 due to the reaction of DMS -OH with 0, DMS * OH + 0 +products (4) The rate coefficients for this reaction were determined by monitoring the OH temporal profiles in the presence of DMS (or C2H6]DMS) and 0 and modelling the obtained profiles using the FACSIMILE program. A value of k = (9.7 f3.3) x lo-' cm3 molecule-' s-l was measured between 225 and 234 K and at 30 and 100 Torr of N,. Thus in the atmosphere the only two important loss processes for the adduct are reaction with 0 or thermal decomposition back to reactants. No other reactions of the adduct need to be considered because of the overwhelming abundance of 0,in the atmosphere. Our value of k does not agree with the earlier reported value of Hynes et a1.,16 who report a value of k4 = 4.2 x lo-' cm3 molecule-' s-l; however more recent work'' in their laboratory gives a value of k4 w 8 x 10-' cm3 molecule- 's-which is in good agreement with our measured value.One of the major products of this reaction was shown to be HO . At both 234 and 258 K a branching ratio for the.formation of H0 in the DMS OH + 0,reaction WHO,) = 0.50 f0.15was determined independent of temperature. This is in excellent agreement with recent values from Hynes et al. who also report a branching ratio for channel (4a) of ca. 50%.'' It is assumed that the co-product of HO is DMSO. In the atmosphere it is very likely that DMSO is either deposited in the ocean and aerosols or reacts with OH.The fate of DMSO in the atmo- sphere is not well understood. Reactions of the CH,S Radical We have shown above that in the atmosphere CH,S is a major product of the abstraction part of the OH + DMS reaction. It also has to be the dominant product in the reaction of NO with DMS in the atmosphere because it is known that the net result of this reaction is formation of CH,SCH,., The CH3S radical may also be produced by the reactions of the DMSaOH adduct. The reactions of the CH3S radical with various atmospheric reactants such as 0,and NO have been studied in various labor- at~ries.~ Recently we discovered that CH,S adds to O2 to form the CH,SOO radical.25 When CH,S was produced via DMDS or DMS photolysis in the presence of 0 at low temperatures the CH3S profiles were non-exponential and its concentration reached a 44 Atmospheric Oxidation of Dimethyl Sulfide constant value which decreases as [O,] is increased.The rate of approach to this con- stant value also increases with [O,]. These observations clearly show the occurrence of the equilibrium kt CH3S + 02+ M CH3SOO + M (3) kr CH3SOO + M ____* CH3S + 02+ M (3-1 The temporal profiles of CH3S measured at various concentrations of 0 were analysed to obtain the equilibrium constant at each temperature. From this data the following thermodynamic quantities about CH,SOO were calculated A,," H"(298 K) = -11.7 &-0.9 kcal mol- ' and A' Ho(CH3SO0,298 K) = 18.1 & 1.0 kcal mo1- ' using a calculated A,, S"(237 K) = -36.8 2.6 cal mol- ' K-'and the calculated values of the heat capacities.The adduct formed by this process must be CH3S-00 and not CH3S02 (where both oxygen atoms are bound to the sulfur) since it rapidly decomposes back to CH3S and 0 and does not appear to add another 0 molecule. Formation of the CH3S.00 adduct suggests that other weakly bound adducts such as CH,SO(OO) CH3S02(00) and HS(O0) could exist. Barnes et and Jensen et ul.,,' have observed CH3S(0)OON02which could be formed from addition of NO to CH3SO(OO). Some recent theoretical calculations also suggest that 0 could add to HS.28 There are also some indirect experimental observations that are consistent with HS(O0) formation.25 Thus it appears very likely that other S-containing free radicals could also add 0 and that the bond strengths of these adducts would be ca.10-15 kcal mol-'. In particular we assume that formation of CH,S(O)O and CH,S(O,)O are likely. Field Evidence for Changes in End-products with Temperature Field measurements of DMS and its possible oxidation products in the atmosphere have cast considerable light on key aspects of the DMS oxidation mechanism. A diurnal variation in the DMS concentration in the atmosphere has been reported in several studies all of which report a night-time maxim~m.~.~*~~ This observation suggests that the rate of daytime loss processes exceeds those at night if the strength of DMS emission is independent of time of day. Another observation with significant implications for the DMS oxidation mechanism lies in the [MSA]/[S042-] ratio measured in the atmo- sphere and in ice-cores.is assumed to be formed with unit yield in the atmo- sphere from SO or SO3). Atmospheric measurements of the MSA/S042- ratio have been made by numerous investigators. The general finding is that this ratio changes with latitude and temperature. For example Staubes and Georgii4 report the [MSA]/[S042-] ratio to be ca. 0.01 between 50"N and 30"N but increasing to ca. 0.25 between 53"s and 62"sand 0.44 at 70"s.This variation in the latitude must be accompa- nied by a temperature change in the vicinity of 300 to ca. 250 K. A more direct impres- sion of the variation in the [MSA]/[S042-] ratio with temperature can be gained from the ice-core data. For example Legrand et observed a change in the [MSA]/[S0,2-] ratio from ca.0.03 to ca. 0.18 when the surface temperature must have changed from ca. 210 to 200 K. Even though there has not been a very coherent mea- surement of the variation in the MSA/S042- ratio as a function of temperature it is clear that this ratio changes rapidly with temperature. The magnitude of this change appears to be at least a factor of four or greater in going from 300 to ca. 250 K and another factor of ca. four in going from ca. 210 to 200 K. If both MSA and non-sea salt (nss) SO,' -are primarily produced from the oxidation of DMS the oxidation mecha- nism of DMS must involve steps which reflect this temperature sensitivity. S. B. Barone A. A. Turnipseed and A. R. Ravishankara The Role of Adducts in the DMS Oxidation Mechanism A simplified DMS oxidation mechanism is shown in Fig.2. A more detailed mechanism is listed in Table 1. Fig. 2 highlights various branching points in the mechanism where weakly bound adducts play a role. We will discuss how these reactions can alter the [MSA]/[S0,2-] ratio and justify the assumptions in the mechanism. In this section CH 3s(02)00N02 Fig. 2 Schematic representation of the atmospheric oxidation mechanism of DMS. The numbers on the left indicate branching points in the proposed mechanism where weakly bound adducts may play an important role. The letters A and B indicate potential unimolecular decomposition reactions important in the oxidation mechansim. Table 1The reactions and their rate coefficients used in the box model calculations to predict the variation of the MSA/nss sulfate ratios as a function of temperature reaction rate constant/cm3 molecule-' s-' ref.OH + DMS +(abstraction) 9.6 x lO-"exp(-234/T) 16 OH + DMS +(addition in 760 air) a 16 NO + DMS +CH,SCH + HNO 1.9 x lo-' exp(-500/T) 12 CH,S + 0,+CH,SO + 0 1.98 x lo-'' exp(290/T) 15 CH,S + NO +CH,SO + NO 2.06 x lo-" exp(320/T) 15 CH,SO + 0,+CH,SO + 0 6 x lo-'' 39 CH,SO + NO +CH,SO + NO 8 x 40 CH,SO2 ACH + SO 5 x lo" expC(17.2 kcal mol-' + RT)/RT] see text CH,SO + 0,+CH,SO + 0 3 x 10-13 estimated CH,SO + NO +CH,SO + NO 4 x 10-12 estimated CH,SO 5CH + SO 5 x lOI3 exp[-(22 kcal mol-' + RT)/RT] see text CH,SO + CH,O +CH,SO,H + CHO 1.6 x 10-15 11 CH,SO + HO +CH,SO,H + 0 5 x lo-" 11 CH,S(O),OO + NO +CH,S(O),OO + NO 2.4 x lo-" estimated [T exp(-234/T) + 8.46 x lo-'' exp(7230/T) + 2.68 x lo-'' exp(7810/T)]/[1.04 x 10"T + 88.1 exp(7460/T)J.Atmospheric Oxidation of Dimethyl Suljide CH,S CH,SO CH,SO and CH,S03 are called CH,SO, while the peroxy species CH,S(O),OO refers to CH,SOO CH,S(O)OO and CH,S(O),OO. Reactions of DMS with OH and NO The results of previous studiesg*’ suggest that two reactions NO + DMS +products (6) OH + DMS +products (1) will dominate the initiation because of the large rate coeficents for these reactions and the abundances of OH and NO in the troposphere. Other initiation reactions are unlikely to be important.’ Reaction (6) has been shown to proceed through adduct formation followed by HNO elimination i.e.equivalent to an H atom abstra~tion.~~~~’ In contrast the results presented in this paper combined with those of previous investi- gations show that reaction (I) proceeds through a complex mechanism involving H abstraction and electrophilic addition.16 Work presented in the present study has estab- lished that H-atom abstraction primarily generates the CH,S radical. The reaction product of the OH addition channel (DMS-OH) has been shown to react rapidly with O2 but all the products in this reaction and their subsequent fate in the atmosphere have not fully been elu~idated.’~*’’ The effective overall branching between the addition and abstraction channels of reaction (1) has been characterized and is dependent on tem- perature.Frevious investigations3’ have suggested that NO + DMS will be of little signifi- cance in the oxidation of DMS in the remote marine boundary layer due to the pre- dicted low ambient NO concentrations in this region. However if the concentrations of NO exceed ca. six times that of OH the rate of reaction (6) will equal that of reaction (1) at 298 K. The rate coefficients for both OH and NO reactions increase with decreas- ing temperature. However the rate coefficient for the OH reaction increases slightly faster than that for NO, such that at 250 K for example the NO concentration has to be ca. 10 times that of OH for the two pathways to be equal (assuming the same number of hours of exposure to both reactants). Field measurements and modelling studies suggest that NO concentrations can easily be 10 times the diurnally averaged OH value in the marine boundary As one approaches high latitudes and winter time the [N03]nigh,-timc/[OH]daytime will only get larger.Therefore a complete quantification of the initiation step in the atmospheric oxidation of DMS requires a better knowledge of ambient NO3 concentrations in the oceanic marine boundary layer. As noted earlier the diurnal cycling of DMS concentrations has been observed in the remote marine boundary layer. If this is a common phenomenon the NO concentra-tions in the remote boundary layer are less than the field measurements indicate or more likely DMS concentrations are enhanced at night. The latter possibility could be due to meteorological differences between day and night and the height of the marine boundary layer and may not require an increase in flux of DMS.This is clearly an important issue that needs resolution. The competition between OH addition and hydrogen abstraction in reaction (1) has been suggested in several studies to be responsible for the observed temperature depen- dence of the [MSA]/[SO,’-] ratio in the atmo~phere.~*~~.~~ For the [MSA]/[SO,’-] ratio to increase as temperature decreases MSA must be a product of the addition channel and SO be a product of the H-atom abstraction. We use the rate coefficients for the addition us. abstraction given by Wine and co-workers16 and assume that the addition channel leads only to MSA and the subsequent steps leading to MSA from the DMS .OH adduct are insensitive to temperature.The temperature dependence of the OH + DMS reaction rate coefficient is such that the [MSA]/[SO,’-] ratio will not S. B. Barone A. A. Turnipseed and A. R. Ravishankara change enough with temperature (see Fig. 3). For example from 250 to 270 K this ratio increases by at most 50% while field data suggest a factor of two or more. In addition to the above arguments two points are worth noting (1) The fractional change in the [MSA]/[S0,2-] ratio for a given change in T AT will depend on T and it will become quite insensitive to AT at low T if it depended only on the changes in the branching ratios of reaction (1). Fig. 3 shows that the temperature dependence of the [MSA]/[S042-] ratio tends to decrease at lower temperatures (i.e.the slope decreases at lower temperatures). (2) It is known from chamber studies that CH,S does lead to MSA.8.26-27 Including reaction (6) decreases the temperature dependence of the [MSA]/[S042 -1 ratio even further. Fig. 3 demonstrates that the effect of NO + DMS is only to lessen the importance of the branching between OH addition cf. abstraction. Based on these arguments we propose that the cause of this temperature-dependent branching is unlikely to be due to changes in the OH reaction mechanism with temperature and has an entirely different origin. The lifetime of DMS due to reactions with OH and NO is of the order of a couple of days. OH exists during daytime while NO peaks at night. At night the concentra- tions of species such as HO and NO will be essentially zero.Therefore DMS oxidation initiated at night will probably lead to species such as CH,SCH202N02 and CH,S(O),OONO (see below). Unless these species are removed by deposition (wet or dry) they should be sources of CH,S (or CH,SO,) radicals after sunrise and undergo further reactions. Therefore it appears that even though NO initiates DMS oxidation at night and OH during daytime the oxidation to gas-phase MSA and sulfate takes place during daytime only. Product-yield Branching following the Initiation Reactions We propose that the branching in DMS oxidation arises from the oxidation of CH,S radicals involving temperature-sensitive processes such as unimolecular decomposition and reversible adduct formation with 02.To test this proposal we developed a simple oxidation mechanism of DMS based largely on available laboratory studies of CH3S 220 240 260 280 300 TiK Fig.3 The branching between MSA and SO,’-in 760 Torr air as a function of temperature for a variety of NO conditions for the simple model where only the mechanism for the OH + DMS reaction changes with T. (a)-(d) represent ambient NO concentrations of 0 3 6 and 9 x lo6 molecules cm-, respectively. The diurnally averaged OH concentration is assumed to be 1 x 1O6 molecules cm -’. Atmospheric Oxidation of Dimethyl SulJide and its higher oxides. Where kinetic data is unavailable we have estimated rate coeffi- cients by analogy with other known reactions. The reactions involved in this simplified mechanism are listed in Table 1.The ambient concentrations used in our model are listed in Table 2. We have treated the initiation chemistry in the most simple fashion possible. H-atom abstraction from OH and NO3 reactions is presumed to yield entirely CH,S whereas the addition of OH followed by the O2 reaction forms a reaction product different from SOz and MSA (probably DMSO DMSO,) OH + DMS + +products other than SO and MSA (addition) (14 OH + DMS -+ +CH,S (abstraction) (lb) We realize that it is very likely that the atmospheric reactions of the DMS -OH adduct could lead to CH,SO, MSA or SO2. This occurrence will not greatly affect our argu- ments. When these reactions are better studied one can easily incorporate them into our mechanism without major qualitative changes in the outcome.Role of Higher Oxides Our mechanism involves CH,S and of higher oxides CH,SO, reacting with O3 and NO2 in competition with their unimolecular decomposition or addition of 02.We investigate the role of higher oxides via first modelling the thermal decomposition of CH,SO and then the addition of 0 to these species. The chemistry of each CH,SO is treated analogously to that of CH,S. To calculate potential unimolecular decomposition rates of the higher oxides we have assumed an A factor of 5 x 1013 s-’ and taken the endothermicity for the decomposition as the activation energy. The simple mechanism given in Table 1 leads to a temperature sensitivity in the [MSA]/[S0,2 -J significantly steeper than that observed in the OH initiation reaction.Fig. 4 shows this temperature dependence for the conditions outlined in Tables 1 and 2. The steep temperature-dependent branching between MSA and SO2 is a result of the competition between the unimolecular decomposition of the CH,S02 radical and the reaction of CH3S02with 0 to oxidize sulfur further without breaking the C-S bond. Table 2 Assumed abundance of various species in the clean marine environment daytime species mixing ratio ref. 4 (-8) 41 3 (-11) 41 1 (-10) 42 2 (-1 1) 38 3 (-13) calculated 1 (-11) calculated 2 (-12) calculated 4 (-14) calculated The sources of the data are given in the references shown. These are ‘typical’ mixing ratios during daytime and some of them have been rounded off to one significant figure.The ratios of HO to OH and NO to NO are kept consis- tent with the above values of 0 and the known J values for NO,. S. B. Barone A. A. Turnipseed and A. R. Ravishankara t 0.1 + I I I I I 270 275 280 285 290 295 300 TIK Fig. 4 Calculated [MSA]/[S0,2-J ratios from DMS oxidation as a function of temperature. (a) All branching due to changes in initiation reaction mechanism; (b)model conditions outlined in Tables 1 and 2. The temperature dependence of the branching reflects the activation energy for the decomposition of CH3S02. We estimate the decomposition of CH3S03 to be more endothermic than that of CH3S02. This is because CH3S0 must be oxidized to CH3S03 by a species such as 03,NO or OJNO combination and these reactions have to be at least thermo-neutral to be important in the atmosphere.Other investigators have also estimated the C-S bond strength in CH3S02 to be less than in CH3S03." Because of the larger endo- thermicity one could expect it to contribute more to the temperature dependence of the branching between MSA and However we do not expect the unimolecular decomposition of CH3S03 to be a very important process in the oxidation for two reasons (1) A large amount of the CH3S02 decomposes to SO2 before it can form CH3S03 and (2) most of the CH,S03 reacts with H02 and CH20. Thus if CH3S03 is formed in the atmosphere it will very likely end up as MSA. Therefore if our assumed rate coefficients are correct it appears that the direct formation of SO3 in DMS oxida-tion as suggested by Bandy et al.is likely to be minor.35 Note,'however that the thermochemistry of CH3S0 is very poorly defined. If the C-S bond strength in CH3S03is much smaller say <20 kcal mo1-' this species could become a significant branching point in DMS oxidation. If the CH,SO radical is stable it could react with NO to form CH3S03N02; this species may be hydrolysed in the particles to give MSA in the liquid phase (see below). Also the rate coefficient for the CH3S03+ HOz reac-tion needs to be determined to estimate this branching. A systematic variation of the ambient NO concentration in our model showed a very small effect on the oxidation products of CH3S in the atmosphere. Although the rate constants for reactions of CH3S CH,SO and CH3S02 with NO are large the low NO levels in the remote troposphere prohibit these reactions from competing with 0 reactions.An order of magnitude increase in the NO concentration used in our model results in a less than 8% change in the [MSA]/[S042-] ratio at 271 K. Therefore we conclude that changes in the amount of NO in the boundary layer has the greatest effect on the initiation chemistry of DMS oxidation and little effect on the subsequent reactions for this proposed scheme of DMS oxidation until the NO level reaches > ca. 300 pptv. Molecular oxygen has been observed to add reversibly to CH3S radicals under atmospheric condition^.^^ [Reactions (5f 5r).] The rate constants for the reactions of the Atmospheric Oxidation of Dimethyl Suljide CH3SO0 radical with NO NO and 0 have been measured." In the remote atmo- sphere the CH3S + O3 reaction should dominate over all other loss processes of the CH,SOO adduct and CH,S.To predict the effect of addition of O2 to CH,S(O), we added the following reactions to the model; kr CH,S(O) + 0 + M -CH,S(O),OO + M (7f) kr CH3S(0),00 + M -CH,S(O) + 02 + M (7r) CH3S(0),00 + NO -CH,SO + NO (8) Note that CH,SO would not be expected to add 0 since the sulfur in this species is fully oxidized. Reactions (7) and (8) comprise a potentially efficient pathway for convert- ing CH3S02 to CH,S03. The equilibrium constant for the reversible addition of 0 to CH3S0 would be very sensitive to the temperature and hence this process could be another candidate in explaining the [MSA]/[SO,' -1 ratio observations in the field.Owing to the abundance of 0 in the atmosphere it is likely that the equilibrium in reaction (7) will be established rapidly and reaction (8) would be the rate-limiting step in determining the overall efficiency of this mechanism. The reaction mechanism presented up to this point is somewhat analogous to that given originally by Yin et d." and modified by Bandy et 121.~' However there are some major differences. First we can show that NO and not HO reaction followed by pho- tolysis of the peroxide is responsible for converting CH,S(O),O to CH,S(O),O. Based on simple steady-state calculations and measured abundances of NO ,36-38 the abun- dance of NO is shown to be approximately an order of magnitude greater than that of H02 in the marine boundary layer.The rate coefficients for the reaction of the larger peroxy radicals with NO is greater than those with H0 and CH302,'2*'3 and hence the rate coefficient for the reaction of CH,S(O),OO with NO must be larger than that with HO,. Therefore the majority of the CH,S(O),OO must react with NO. Even if hydroperoxides are formed since they do not usually absorb very strongly in the near- UV they are likely to be washed out of the atmosphere. We have estimated the rate constant for reaction (8) to be 2.4 x lo-'' cm3 molecule-' s-' and calculated yields for a variety of different CH,SO,-00 bond strengths. Bond strengths < 10 kcal mol-' have little effect on the outcome.However as the bond strength is increased from 10.5 kcal mol- ' the addition reactions of CH,S(O) radicals have a pronounced effect on the absolute values of the branching to MSA and SO2 and the temperature dependence of their ratio. Fig. 5 shows the effect of adduct bond strengths of 13 and 14 kcal mol-' on the temperature dependence of the [MSA)/[SO,'-] ratio. The increase in the [MSA]/[S0,2-] ratio with temperature is a result of a shift toward larger equilibrium concentrations of the CH,S0200 at low temperatures. This in turn increases the rate of reaction (8) relative to the unimolecular decomposition of CH3S02 and hence adds to the overall temperature sensitivity of the [MSA)/[S0,2 -1 ratio. Note that the incorporation of this process in the model adds to the temperature dependence in the [MSA]/[SO,'-] ratio determined by the unimolecular decomposition of the CH,S02 radical alone.Aside from influencing the temperature dependence of the branching in DMS oxida-tion the inclusion of the 0 adduct formation reaction has several other consequences. First because the NO + CH3S(0,)O0 reaction becomes the rate limiting step in the cycling of CH3S02 to CH,SO, NO will play a larger role in determining the [MSA]/[S0,2-] ratio. The branching to stable end-products in DMS oxidation will be strongly NO dependent. Secondly the increased efficiency in the conversion of CH,S02 to CH,SO makes the unimolecular decomposition of CH,SO more impor- tant. Thus branching to form SO can become more likely. However SO formation is still likely to be minor because the rate of CH,SO decomposition is likely to be slow S.B. Barone A. A. Turnipseed and A. R. Ravishankara 100.0 270 275 280 285 290 295 300 TI Fig. 5 Calculated [MSA]/[SO,’-] ratios as a function of temperature from a DMS oxidation mechanism which includes reactions (7) and (8). (a),(b)and (c) are calculated from CH,S(O,). 00 adduct bond strengths of 0 13 and 14 kcal mol-’ respectively. The other rate coefficients and ambient concentrations are listed in Tables 1 and 2. with respect to its reactions with H02 and CH,O. Lastly formation of an 0 adduct with CH3S0 is also a potentially important pathway to form CH,SO,. In our simple model all CH3S0 is converted to CH,SO,. However isomerization or other removal processes of this radical could play key roles in the oxidation of DMS.The formation of nitrate from NO addition to the peroxy-adducts; NO + CH,S(O),OO + CH,S(O),OONO (9) could be an important process in the atmosphere. Nitrates such as these have pre- viously been observed in chamber studies of DMS ~xidation.~~.’~ This observation is an indication of the likelihood of 0,-adduct reactions in the atmospheric oxidation of DMS. The formation of these peroxynitrate radicals could be especially important if they are easily hydrolysed in cloud drops. The hydrolysis of these nitrates could prove to be an important source of MSA in the atmosphere. Also if the nitrates are long-lived they could play a role in the transport of sulfur to the free troposphere.It is possible that reactions of CH,S and CH,SO with 0 and of CH,SO with NO could result directly in the cleavage of the C-S bond. These channels are exothermic and would have a significant effect on the results of the mechanism developed thus far. To evaluate this possibility we have included the following reaction in our model; CH,SO + 0 + +SO (10) Note that the above reaction merely represents the cleavage of the C-S bond to lead to SO2formation and is not meant to be an elementary reaction. Inclusion of this chem- istry significantly effects the calculated [MSA]/[S0,2 -1 ratio. Aside from increasing the absolute magnitude of SO formation including reaction (10) decreases the temperature dependence of the SO yield. Fig. 6 demonstrates the effect on the [MSA]/[S0,2-] ratio of 25% 50% and 75% channels in the reaction CH3S0 + O3 to cleave the C-S bond.Reaction (10) channels significant amounts of the sulfur to end-products by bypassing the CH,SO species and hence avoids the temperature-dependent mecha- nism for SO production via unimolecular decomposition. The effect of CH3S + O3-+ -,SO2 (1 1) Atmospheric Oxidation of Dimethyl SulJide 0.01 5 270 275 280 285 290 295 300 TIK Fig. 6 [MSA]/[S0,2-] as a function of temperature from DMS oxidation calculated with the inclusion of reaction (10). (a)-(d) reflect branching ratios of the reaction CH,SO + O3 to yield SO of 0,0.25,0.50 and 0.75 respectively.All other relevant conditions are in Tables 1 and 2. would be identical to reaction (10).However the effect of; CH3S0 + 0 + CH + SO3 + 0 (12) could be less than reactions (10) and (ll) depending on the rate coefficient for the unimolecular decomposition of CH,SO .Note that the addition of O2 to CH,SO fol- lowed by the reactions of CH,S(O)OO could mediate the effects of the C-S bond cleavage reactions involving 03. We have focussed on the potential branching reactions of the CH3S02 and CH3S03 radicals however it is important to note that the chemistry of the preceding CH,SO radical could also be a source of the branching between MSA and SO,,-in DMS oxidation. Branching at CH3S0 would most likely occur through a competition between O2 addition followed by reaction with NO and reaction (10) resulting in C-S bond cleavage.C-S bond cleavage would eventually lead to SO formation whereas 0 addition could enhance the oxidation state of the sulfur without breaking the C-S bond; CH3S0 + 0 -+CH3S(0)O0 (13) CH,S(O)OO + NO + CH3SOZ + NO2 (14) CH,SO + O3-+ SO + CH + 02. (15) Role of Heterogeneous Reactions There is a possibility that MSA is formed only via heterogeneous reactions of species such as CH3S(0),00N02. The MSA measured in the atmosphere has always been extracted from the aerosol phase. If this is indeed the case the [MSA]/[S0,2-J ratio may be related to the heterogeneous reaction probability i.e. availability of aerosols fog clouds etc. and the stability of the gas-phase species which are taken up by the con- densed medium. Both these factors are enhanced at lower temperature and hence are consistent with the observed increases in the [MSA]/[S0,2-] ratio with decreases in S.B. Barone A. A. Turnipseed and A. R. Ravishankara 53 temperature. Measurement of gas-phase MSA abundances is essential for shedding light on this problem. Summary of Modelling Studies Our simple modelling analysis of DMS atmospheric oxidation has led to several basic conclusions (1) The temperature dependence of OH addition us. H-atom abstraction in the initiation chemistry of DMS oxidation is not strong enough to rationalize completely the variations in the MSA/SO,’- ratios observed in the ice-cores and with latitude. The inclusion of NO chemistry greatly reduces the overall temperature dependence of the OH addition us.H-atom abstraction. An accurate knowledge of the ambient NO concentrations in the marine boundary layer is the key to better understanding of the initiation chemistry. (2) It appears that the abstraction pathway in the OH/NO + DMS reaction can also lead to the formation of MSA contrary to what has been assumed in some past studies but in accordance with many chamber studies. (3) It appears that the major role of NO in DMS oxidation is limited to the initiation step unless CH,SO -0 adducts are important. (4) CH,S oxidation via the formation of higher oxides that can unimolecularly decom- pose (cleave the C-S bond) or raise the oxidation state of sulfur without breaking the C-S bond can lead to a large temperature dependence of the branching between MSA and SO,.This temperature dependence will be heightened if 0 adds reversibly to some of the CH,SO species. If this were to happen NO could play a role in determining the branching between different endproducts. (5) The potential for C-S bond cleavage via other bimolecular reactions of CH,S or CH,SO lessens the calculated temperature dependence in the [MSA]/[S0,2 -3 ratio. This is especially the case if it occurs before the formation of CH,SO in the oxidation mechanism. (6) Even in the clean marine air it is unlikely that species such as CH,SO,H [CH,S(O)-OH or CH,SOOH] are formed via the involvement of HO radicals. (7) Heterogeneous reactions of species such as CH,S(O),OONO could be key steps in MSA formation.The simple modelling exercise highlights areas which need further effort (1) Measurement of the ambient levels of NO and NO in the marine boundary layer and free troposphere is essential. The observed diurnal variation in DMS may be contradictory to the measured levels of NO in the marine troposphere. (2) Determinations of the unimolecular decomposition of CH,SO radicals via experi- mental or computational investigations are needed. (3) The detection of peroxy and peroxy nitrate intermediates in the atmosphere and measurements of MSA as a functions of NO and temperature could provide valu- able insight into the key aspects of the oxidation mechanism. (4) Measurement of gas-phase MSA is needed to evaluate the role of heterogeneous processes.(5) Development of detection techniques such as photoionization mass spectrometry and chemical ionization mass spectrometry for various intermediates and end- products are needed to decipher the DMS oxidation mechanism. This work was carried out as a part of NOAA’s Climate and Global Change Pro- gramme. We thank Professor Michael McKee for assistance in calculating rotational and vibrational constants for CH,S(OH)CH . Atmospheric Oxidation of Dimethyl SulJide References 1 Dimethylsulphide Oceans Atmosphere and Climate ed. G. Restelli and G. Angeletti Kluwer Dordrecht 1992. 2 H. Berresheim J. Geophys. Res. 1987,92 13. 3 H. Berresheim M. 0. Andreae G. P. Ayers R. W. Gillett J. T. Merrill V. J. Davis and W. L. Cha- meides J.Atmos. Chem. 1990 10 341. 4 R. Staubes and H-W. Georgii Measurements of Atmospheric and Seawater DMS Concentrations in the Atlantic the Arctic and the Antarctic region Dimethylsulphide Oceans atmosphere and climate Belgi-rate Italy 1992 p.95. 5 R. W. Gillet G. P. Ayers J. P. Ivey and J. L. Gras Measurement of Dimethyl Suljide Sulfur Dioxide. Methane Sulfonic acid and Non-sea Salt Sulfate at the Cape Grim Baseline Station Dimethylsulphide Oceans Atmosphere and Climate Belgirate Italy 1992. 6 M. 0. Andreae R. J. Ferek K. Bermond K. P. Byrd R. T. Engstrom S. Hardin P. D. Houmere F. LeMarrec H. Raemdonck and R. B. Chatfield J. Geophys. Res. 1985,90 12. 7 I. Barnes K. H. Becker and N. Mihalopoulos FTlR Products Study of the Photolysis of CH,SSCH, Reactions of the CH,S Radical Dirnethylsulphide Oceans Atmosphere and Climite Belgirate Italy 1992.8 S. Hatakeyama and €1. Akimoto J. Phys. Chem. 1983,87,2387. 9 A. A. Turnipseed and A. R. Ravishankara The Atmospheric Oxidation of Dimethyl Suljide Elementary Steps in a Complex Mechanism Dimethylsulphide Oceans Atmosphere and Climate Belgirate Italy 1992 p. 185. 10 G. S. Tyndall and A. R. Ravishankara Int. J. Chem. Kinet. 1991,23,483. 11 F. Yin D. Grosjean and J. H. Seinfeld J. Atmos. Chem. 1990,11,309. 12 W. B. DeMore S. P. Sander D. M. Golden R. F. Hampson M. J. Kurylo C. J. Howard A. R. Ravishankara C. E. Kolb and M. J. Molina Chemical Kinetics and Photochemical Data for use in Stratospheric Modeling Jet Propulsion Laboratory JPL Pub. No. 92-20 1992.13 R. Atkinson D. L. Baulch R. A. Cox R. F. Hampson J. A. Kerr and J. Troe J. Phys. Chem. ReJ Data. 1992 21 1125. 14 G. L. Vaghjiani and A. R. Ravishankara J. Phys. Chem. 1989,93 1948. 15 A. A. Turnipseed S. B. Barone and A. R. Ravishankara J. Phys. Chem. 1993,97,5926. 16 A. J. Hynes P. H. Wine and D. H. Semmes J. Phys. Chem. 1986,90,4148. 17 J. P. D. Abbatt F. F. Fenter and J. G. Anderson J. Phys. Chem. 1992 % 1780. 18 A. M. Malleson H. M. Kellet R. G. Myhill and W. P. Sweetenham A.E.R.E. Harwell Publication R 13729 A.E.R.E. Publications Oxford 1990. 19 A. J. Hynes T. Pounds T. McKay J. D. Bradshaw and P. H. Wine 12th International Symposium on Gas Kinetics Reading 1992. 20 F. Turecek J. Phys. Chem. 1994,98,3701. 21 M. L. McKee J. Phys. Chem. 1993,97 10971.22 R. E. Stickel Z. Zhao and P. H. Wine Chem. Phys. Lett. 1993,212,312. 23 N. I. Butkovskaya and G. LeBras J. Phys. Chem. 1994,98,2582. 24 I. Barnes V. Bastian and K. H. Becker Int. J. Chem. Kinet. 1988,20,415. 25 A. A. Turnipseed S. B. Barone and A. R. Ravishankara J. Phys. Chem. 1992,% 7502. 26 I. Barnes V. Bastian K. H. Becker and H. Niki Chem. Phys. Lett. 1987 140,451. 27 N. R. Jensen J. Hjorth C. Lohse H. Skov and G. Restelli J. Amos. Chem. 1992 14,95. 28 D. Laakso C. E. Smith A. Goumri J-D. Rocha and P. Marshall Chem. Phys. Lett. 1994,227 377. 29 E. S. Saltzman and D. J. Cooper J. Atmos. Chem. 1988,7 191. 30 M. Legrand C. Feniet-Saigne E. S. Saltzman C. Germain N. I. Barkov and V. N. Petrov Nature (London) 1991,350 144. 31 R. Atkinson J.N. Pitts and S. M. Aschmann J. Phys. Chem. 1984,88 1584. 32 J. F. Noxon J. Geophys. Res. 1983,88 1 1017. 33 J. A. Logan J. Geophys. Res. 1983,88 10785. 34 A. J. Hynes and P. H. Wine in Biogenic Sulfur in the Environment ed. E. S. Saltzmann and W. J. Cooper American Chemical Society Washington DC 1989. 35 A. R. Bandy S. B. W. Blomquist S. M. Chen and D. C. Thornton Geophys. Res. Lett. 1992 11 1125. 36 B. A. Ridley M. A. Carroll D. D. Dunlap M. Trainer G. W. Sachse G. L. Gregory and E. P. Condon J. Geophys. Rex 1989 94 5043. 37 B. A. Ridley S. Madronich R. B. Chatfield J. G. Walega R. E. Shetter M. A. Carroll and D. D. Montzka J. Geophys. Res. 1992 97 10. 38 M. A. Carroll B. A. Ridley D. L). Montzka G. Hubler J. G. Walega R. B. Norton B. J. Huebert and F.E. Grahek J. Geophys. Res. 1992 97 10. 39 F. Domine A. R. Ravishankara and C. J. Howard J. Phys. Chem. 1992,% 2171. 40 G. S. Tyndall and A. R. Ravishankara in Biogenic. Sulfur in the Environment ed. E. S. Saltzman and W. J. Cooper American Chemical Society Washington DC 1989 p. 450. 41 B. A. Ridley and E. Robinson J. Geophys. Rex. 1992,97 10. 42 B. Heikes J. Geophys. Res. 1992 97 18. Paper 5:00197H Rcceived 9th January 1995

ISSN:1359-6640    

DOI:10.1039/FD9950000039

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Faraday Discuss. 1995 100 C47-C59 Approaching Complexity Stephen K. Scott School of Chemistry University of Leeds Leeds UK LS2 9JT In 1974 a Faraday Symposium’ was held at the Royal Institution on the subject of ‘The Physical Chemistry of Oscillatory Processes’. That timely event brought together an international group of specialists in the newly developing interdisciplinary research field of oscillations chemical waves and other ‘exotic’ phenomena. In the 21 years since the Symposium our understanding of many fundamental properties of such behaviour has deep- ened through the application of advances in pure and applied mathematics in collaboration with experimental studies based in many physical bio- logical and engineering contexts with the physical chemist being pre- eminent exploiting the natural ‘non-linearities’ in chemical kinetics.This account illustrates how these ideas have been applied and developed through two specific examples the route to chaos in the oxidation of carbon monoxide and the experimental challenge posed by Alan Turing’s predic- tions of chemical patterns. 1. Introduction Many important chemical reactions exhibit the property offeedback. Chemical feedback arises when intermediate species formed in the reaction mechanism influence their own rate of production.2p4 In positive feedback such as chain branching in combustion pro- cesses or autocatalysis in solution-phase redox reactions the intermediates accelerate their own rate of production negative feedback or inhibition also arises and is particu- larly important in regulatory biochemical systems.The characteristic signature of posi- tive feedback is clearly revealed if the overall rate of reaction (e.g. as defined by the rate of consumption of the initial reactant or of formation of the final product) is plotted as a function of the extent of reaction i.e. the percentage consumption of the initial reactant. Fig. 1 shows such plots for (a)deceleratory systems typical of simple nth-order reaction kinetics in which the maximum rate arises at the beginning of the reaction; (b) a system Q) * E C .-U cd 2 0 extent 1 0 extent I 0 extent 1 Fig. 1 Variation of reaction rate with extent of reaction; (a) deceleratory systems showing 1st (n = I) 2nd (n = 2) and approximately +-order kinetics; (b) systems with feedback showing (i) quadratic and (ii) cubic autocatalysis; (c) comparison of (ii) cubic and (iii) thermal feedback curves c47 C48 Approaching Complexity with positive feedback showing an initial acceleratory phase with the rate increasing with increasing extent of reaction until some maximum at a non-zero extent of reaction.Fig. l(c)compares the shape of the rate-extent curve for a type of chemical (isothermal) acceleration and thermal feedback. The latter arises through the self-heating accom- panying an exothermic reaction in the absence of infinitely fast heat transfer feeding back through the temperature dependence of the reaction rate constant(s). It is the char- acteristic shape of these curves that supports the various types of complex hehaviour that will be discussed here.This account will also indicate how the very recent advances in the mathematics of non-linear dynamical systems provide just the right tools for the chemical kineticist to ask the appropriate questions and in some instances to find the appropriate answers to those questions when faced with ‘complex behaviour ’. To provide a more concrete illustration of ‘complexity’ we can imagine an experi- ment based on the spontaneous reaction between hydrogen and oxygen. The H + 0 is the simplest combustion reaction the overall stoichiometry 2H + 0 -+ 2H,O hides a somewhat complex mechanism involving up to 90 elementary steps and several radical and molecular intermediates including H and 0 atoms OH and HO radicals and H,O,.Its behaviour in classic closed (static) vessels is covered in most under- graduate courses5 and gives rise to the three ignition limits sketched in Fig. 2. To the left of the limits there is a ‘slow’ reaction in which the oxidation proceeds over a timescale of minutes or longer. On the high-temperature side of the limits reaction proceeds as a millisecond timescale ignition (perhaps after some ‘induction period ’). In a flow reactor there is a similar limit structure.6 Again there is slow reaction at low ambient temperatures (the temperature of the oven in which the reactor is seated) with an ignition process at higher T,. It is the nature of this ‘ignition’ that is of interest. Under some conditions the reaction adjusts to a rate that simply matches the rate of inflow of fresh reactants.This gives rise to steady-state operation. An example of the record observed by a (coated) fine-wire thermocouple junction positioned in the reacting gas (and referenced to a similar junction fixed to the outside wall of the reactor hence measuring the self-heating due to reaction) is shown in Fig. 3(a). After some initial tran- sient behaviour the system settles to a steady signal. Assuming that we are interested only in the long-time (post-transient) response then the state of the system is essentially L 8 F- h 700 800 TdK Fig. 2 The classic pT ignition limits for H + 0 reaction in a closed vessel S. K. Scott c49 initial time t time Fig.3 Possible evolutions of combustion systems in flow reactors (a)a simple stable steady state is achieved after some initial transient from the initial conditions; (b) a complex oscillatory igni- tion response summarised by reporting a single quantity the steady-state temperature excess. Once this has been reported the future behaviour can be predicted with some confidence it will persist with the same steady-state response for as long as the experimental condi- tions are maintained. If however the experimental conditions are changed e.g. by decreasing the ambient temperature then the response may also change. The record shown in Fig. 3(b)is also collected from the H + 0 system but is clearly much less simple than the steady-state ignition process This time-dependent response is not ‘transient’ in the sense that the reaction for this set of experimental conditions will not eventually settle to a steady state.Rather this oscillatory state will persist indefinitely. This is a complex example of an oscillatory state with no obvious period. Three fundamental questions spring to mind when such a response is encountered. (a) How can we describe such a state? This question is posed in the technical sense i.e. we seek to find a way such that precise information can be compactly conveyed to other scientists without having to reproduce the whole experimental record (which in the case of a non-periodic response would have to be infinitely long anyway). Clearly reporting a single value of the temperature excess is insufficient and even a mean and a standard deviation hardly come close to conveying even the general features of Fig.3(b). C50 Approaching Complexity (b) How did such a complex response arise? This question relates to the change in response between Fig. 3(a) and (b). There may have been only a 5 K change in the ambient temperature but did the system switch suddenly from its steady-state response directly to the ‘chaos’ of aperiodicity or did the complexity develop in stages with intervening states each slightly more complex then its predecessor. Such a series of steps is known as a ‘bifurcation sequence’ in the language of non-linear dynamical systems theory. Do different reactions all have different bifurcation sequences or are there some ‘universal routes’ from simplicity to complerity? If so we might be able to recognise the signs that indicate that complexity is ready to emerge as we vary the experimental con- ditions in a particular way allowing us either to avoid or perhaps move more em- ciently towards such behaviour.(c) How can we exploit such complexity? If complexity is just a nuisance then recog- nising the early sign will perhaps allow us to move away before it develops fully. On the other hand we may deliberately seek to maximise the complexity of a systems response if we can exploit such behaviour in a way unavailable under steady-state operation. The exploitation of chaotic systems has recently emerged as a possibility with identifiable potential applications. An introductory account of ‘Controlling Chaos’ can be found in ref.7 but question (c) will not be the major focus of the present account. Questions (a) and (b) will be addressed with reference to the oxidation of carbon monoxide another exemplary combustion reaction for which the picture with regard to the global behaviour observed experimentally is somewhat clearer than for the H + 0 system (although the mechanistic understanding is perhaps less well developed). A fourth question will also be addressed through a second example. (d) Must complex systems always give rise to complex behaviour? The underlying text here is interest in whether large-scale co-operation can emerge through chemical feedback perhaps coupled to molecular transport processes in such a way as to form coherent spatial ‘patterns’ analogous to the temporal ‘patterns’ of periodic oscillations.This story involves a 40 year old theoretical prediction and the subsequent journey by many groups inspired by this prediction to realise these Turing Patterns in the chemical laboratory. 2. Complex Oscillations in the Oxidation of Carbon Monoxide The reaction between CO and 0 in the presence of small but accurately controlled concentrations of H ,is now relatively well characterised.8 At reduced pressures (10 < p/mmHg < 50) the reaction shows a second ignition limit in both closed- and continuous-flow reactors of a very similar type to that found in the H + 0 system over an ambient temperature range of approximately 750 < T,/K < 825. We will con- centrate here on the flow reactor studies as this configuration allows us to probe the behaviour on the ‘ignition’ side of the limit.’ The p-T ‘ignition diagram’ for such a system is shown in Fig.4. At ambient temperatures lower than the ignition limit there is a steady slow reaction. This is a steady-state response in which there are only small extents of reaction with the outflow composition being mainly unreacted CO and 0 with perhaps a few percent CO,. Temperature excesses are less than 10 K in this region and the behaviour is defi- nitely ‘simple’. In a given experiment it is relatively easy to vary the ambient tem- perature which in the nomenclature of non-linear dynamical systems is the primary bifurcation parameter. Other secondary parameters such as the total operating pressure the mixture composition or the mean residence time (the average time a molecule spends in the reactor) can be varied between experiments although in the present account only the pressure will be changed.On crossing the ignition limit the system displays a relatively wide region of oscil-latory ignition. Reaction is achieved by sharp ignition events in which the fuel is effec- S. K. Scott C5 1 osci 11atory . .. I I 700 750 800 ambient temperature TJK Fig. 4 The pq ignition diagram for CO + 0 in a flow reactor showing regions of slow reaction steady flame and oscillatory ignition. The behaviour in the shaded region is that of complex oscillations. tively completely consumed (the equimolar mixture is fuel-lean) and in which a large temperature excursion is detected by the fine-wire thermocouple junction there is also a weak pale-blue chemiluminescent emisson from electronically excited C02.The ignition excursions are separated by essentially quiescent periods in which the reactor is reple- nished by the inflow and the products are removed by the outflow. In fact the reaction is effectively suppressed in this intervening period by ‘inhibitor’ species most notably H,O formed as products. Over most of the oscillatory region the oscillations have a simple period-I character provided the experimental conditions are held constant and any initial transient behav- iour is allowed to die away the system settles into oscillations for which the amplitude of each peak is the same as its predecessor and the period between any two peaks is constant throughout the sequence.An example of a period-1 ignition is shown in Fig. 5(u). The amplitude and period of the oscillation vary smoothly across the region of oscillatory ignition if the ambient temperature T or pressure are varied. As the system approaches the second main boundary on the ignition diagram Fig. 4 the amplitude decreases smoothly to almost zero with the oscillatory waveform then having the appearance of a small ‘ripple’ superimposed on an almost steady non-zero temperature excess Fig. 5(b). If T is increased further so as to cross this limit into the steadyflume region the ‘ripple’ dies away and a truly steady signal is observed. The system settles onto a steady state of now high reactant conversion in which there is effectively a balance between the inflow of reactants and their complete conversion to CO in the mean residence time.The long residence times (2-16 s) typical of such reactor studies limit the maximum reaction rate to moderately low values (maximum steady-state rate = inflow concentration/residence time otherwise the reactant concentrations would become negative) so the steady-state temperature excesses associated with this ‘steady- flame’ state are typically only 10-100 K rather than those of more familiar flames. The boundary between oscillatory ignition and steady flame represents the locus of experimental conditions at which a particular qualitative change (bifurcation) in the dynamic behaviour of the system occurs.In technical terms this particular change from steady to oscillatory behaviour is called a Hopfbifurcation (and in this case it is a super-critical Hopf bifurcation as the oscillations grow from vanishingly small amplitude as the locus is crossed by decreasing T,). There is an increase in the complexity from steady to simple oscillations. Returning to the region of oscillatory ignition there is a subregion indicated in Fig. 4 of complex ignition. To investigate the sort of changes in response encountered in this subregion we can simply set the reaction system up at some ambient temperature and C52 Approaching lomplexity 300 -I! Y R Q 100 --i I! I I I I 0 5 t im e/m in 100 t n ’ 0 time Fig.5 Examples of simple oscillatory waveforms (a) period-1 relaxation ignitions with large amplitude and long period found near the ignition limit ;(b)small amplitude approximately sinus- oidal oscillations found near the boundary with the steady-flame state pressure so as to lie just outside this subregion (e.g.at lower T,) and then allow the oven temperature to increase slowly so that the system drifts through the complex ignition region. Two such ‘scans’ are shown in Fig. 6 corresponding to the paths indicated in Fig. 4. At a relatively high pressure the region of complexity is rather narrow. The response of the system on entering this region is for the simple period-1 oscillation to give way to a waveform in which there is an alternation between large- and small-amplitude peaks as indicated in Fig.6(a).Just inside the region the difference in amplitude between large and small is itself only small although the large and small character is established more clearly as is increased. The period between each individual peak is approximately the same this means that the repeating unit is twice that of the period-1 response. This qualitative change in behaviour is known as a period-doubling bifurcation and produces a period-2 response. A further increase in the ambient temperature takes the system out of the complex oscillation region. Simple period-1 behaviour is regained by a period- halving i.e. the inverse response of the way in which the increased complexity was estab- lished.At a lower operating pressure the region of complexity is somewhat more extensive. This allows ‘more room’ for a greater development of complexity as indicated by the scan in Fig. 6(b).The first stage is again a period-doubling to establish a period-2 wave- form but this then shows an additional period-doubling event leading to a period-4 and so on. There is clearly a suggestion of some very complex behaviour in Fig. 6(b),but to investigate this in a rigorous manner we really need to obtain ‘transient-free’ informa- S. K. Scott c53 T slowly increasing Fig. 6 Evolution of the oscillatory waveform during a scan of ambient temperature through the region of complex oscillations at different pressures tion. This is achieved by allowing the experimental conditions to settle out completely at various points within the complex ignition region and collecting data only once this settling-out process has happened.Examples of such asymptotic states are shown in Fig. 7(a)-(f) which are arranged in order of increasing ambient temperature. At the lowest K just outside the complex ignition region there is a period-1 response. Increasing the ambient temperature sees emergence of period-2 as described above and then at a slightly higher T, period-4 [Fig. 7(a)-(c)J. This reaction also gives strong suggestions of period-8 and period-16 but these occur over very narrow ranges of experimental conditions and it has not proved possible to stabilise the system suffi- ciently to collect entirely convincing time series of these higher complexity states.What does emerge at slightly higher ambient temperature is a modestly wide region of very complicated behaviour of the character shown in Fig. 7(d).There is no obvious repeating unit or periodicity we will return to the analysis of this response below. If the range of the aperiodic behaviour is examined in detail it is observed to contain very narrow ‘windows’ in which a periodicity can be identified two such examples are shown in Fig. 7(e)and (f)corresponding to periods 5 and 3 respectively. The emergence of a region of aperiodic responses has some interesting implications. Other evidence and experience indicates that systems that behave with this response are unpredictable in a specific context small differences in initial condition do not as is typical of periodic or steady-state behaviour decay in magnitude; instead they grow exponentially until two systems which started ‘close together’ become totally uncor-related.Thus a repeat experiment would eventually evolve in a different pattern and one system cannot be used to forecast the behaviour in another or equivalently the past behaviour of a system cannot be used to predict the future evolution beyond some ‘threshold of acceptable error’. [It is important to appreciate a distinction between two systems subjected to the same experimental condition both behaving in the same way i.e. chaotically but whilst they both have this property they will be evolving in time in c54 Approaching Complexity t ime/m in Fig.7 Different oscillatory waveforms collected post transient evolution at constant experimental conditions for different T through the region of complex oscillations. ways that are different (i.e. with a different sequence of large and small peaks) unless they have exactly identical initial conditions.] All is not completely lost there is a predictabil-ity horizon if we want complete accuracy i.e. if we want to predict the future event precisely then we are lost if the system is chaotic. However if we wish to predict to say 5% accuracy we can use the past evolution or the behaviour of a ‘repeat’ experiment to make such a prediction for a limited time (which we can predict and which may be quite long if the system is only mildly chaotic). If we accept 10% precision we can predict for longer (but not twice as long; there is a logarithmic relationship between error and time).How does this inherent unpredictability arise? For the p-T conditions surrounding the narrow region of chaotic behaviour the reaction evolves in moderately exotic ways but always with an identifiable pattern. We would as physical chemists expect the evol- ution to be determined by some reaction mechanism comprising elementary steps each with definite rate laws and Arrhenius parameters etc. The resulting reaction rate equa- S. K. Scott c55 tions even if they are not known precisely would be expected to be calling the tune so that the behaviour should in principle be predictable and regulated. So why do we have a narrow range of experimental conditions within this range of experimental conditions for which the system becomes essentially unpredictable? Does the reaction mechanism simply give up for the 'special' set of conditions? This interpretation would be uncom- fortable but is anyway undermined by the observation of the periodic windows that are embedded within the chaotic region does the mechanism keep switching in and out failing and then reviving? Logic demands a different answer with the same kinetic mechanism operating over the whole p-T region in Fig.4 but capable of rationalising unpredict a bili ty. To show that the aperiodic response of Fig. 7(d) is not random we can perform a simple analysis of the thermocouple record. Fig. 8(a) reproduces the chaotic time series and Fig.8(h) shows the result of plotting the temperature maximum of one peak against the maximum of the next peak successively through the series (i.e. we plot the maximum of peak 1 against peak 2 of peak 2 against peak 3 etc.). This next maximum map would be a simple random array of dots if the time series were random. However the form shown in Fig. 8(h) is clearly not that it has a clear structure indicating that there is a definite relationship governing the maximum in one ignition in terms of the maxima of previous peaks. The system is thus fully deterministic. So how do we square this claim with thc previous statement that the system is behaving in an essentially unpredictable manner? The answer to this is fairly involved but relates to the steepness of the slope of the map in the vicinity of its intersection with a line of unit slope drawn on the same graph.This identity line intersects with the map at a point for which the amplitude of the next maximum is equal to the amplitude of the current maximum i.e. it identifies the point at which the system would sit if every oscillation had the same amplitude which is the simple period-1 response. Because the actual map is steep in the vicinity of this point we know that the period-1 solution is unstable. This means that if we have a system that has a maximum close to this point successive iterations of the map will have maxima that move further away from the intersection point. In mathematical terms the map has a slope of magnitude greater than 1 (in fact we have slope < -1 as the gradient is negative) at the period-1 intersection point.Furthermore the steepness of the slope means that two systems that have maxima that are similar but not exactly 0 5 $ t ime/min E .-X 2 a -0 3 .-c -Q amplitude of nth maximum Fig. 8 (a)Chaotic time series from Fig. 7(d);(h)next maximum map contructed from ignition data C56 Approaching Complexity the same will have their differences amplified when they visit part of the map with such a steep slope. Similar comments apply to the intersections of higher-order maximum maps such as the next-but-one maximum map where we plot the amplitude of the nth peak against that of the n + 2th peak. For that map the intersections with a line of unit slope correspond to the oscillations that repeat every other ignition i.e.to the period-2 solutions. If the map has slope of magnitude greater than 1 in the vicinity of these intersections then the period-2 solution will also be unstable and small differences in the evolution of similar systems will be amplified in the vicinity of such steep segments of the map. Chaos or aperiodicity arises when all nth maximum maps have steep intersec- tions with the line of unit slope. The next maximum maps convey much important quantitative information in terms of the slope at the intersection point. This information allows us to quantify just how chaotic a particular system is and how long we can predict the future evolution if we are prepared to accept certain tolerance levels for the accumulating error.The maps are also important for the new techniques of stabilising and tracking periodic states from within the natural 3. Self-organisationand Pattern Formation in Non-linear Systems In 1952 the mathematician Alan Turing published a paper” entitled ‘The Chemical Basis for Morphogenesis’ in which he speculated on the evolution of chemical concen- trations in systems in which chemical feedback was coupled with molecular transport by diffusion. In ‘typical’ e.g. unreactive cases diffusive processes serve to smooth out any initial spatial dependence of concentration and the system tends towards spatial uni- formity. Even in simple reacting systems the expected tendency is for an increasingly uniform spatial distribution as diffusion operates.Turing however proposed that if there is a sufficient feedback mechanism and if the diffusivity of the feedback species is sufficiently reduced compared with that of the other participants then for some experi- mental conditions systems may actually evolve in the opposite way with the develop- ment of significant and well-organised spatial dependences of species concentrations. Such Turing Patterns became of great theoretical interest in the 1960s to 1980s not least because they were perceived as potential causes of spatial form in biology12v13 as well as chemistry (articles on Turing Patterns frequently have titles that are variations on the theme of ‘How Did the Leopard Gain its spot^').'^ The obstacles to experimental observation of Turing Patterns in chemistry labor- atories that presented themselves in the early years included the following (i) there was a general lack of known chemical reactions with clearly established feedback processes; (ii) there appeared to be no known way of selectively controlling diffusion coeficients of individual species in a reacting mixture; (iii) chemical reactions often have associated self-heating or changes in molar volume that give rise to spontaneous initiation of convective fluid motion in the reaction system which in turn readily dominates the relatively weak role played by diffusion.The last point relating to the relegation of diffusion processes when bulk fluid motion operates is particularly relevant if one wishes to sustain any potential pattern development.As the reaction must be maintained away from its state of chemical equi- librium there is a need for a continuous supply of fresh reactants and a removal of final products. This would argue for a continuous flow system much like the flow reactors described in the previous section. However such supply and removal must be provided in a way that does not disrupt the delicate diffusion-driven patterns of interest here. These ‘technical difficulties’ have been addressed by many groups over the inter- vening period and the final pieces of the jigsaw have recently been put together with the first experimental observations of Turing Patterns. The family of known reactions with chemical feedback has grown dramatically’ since the acceptance of oscillatory pheno- S.K. Scott c57 Reservoir 2 Fig. 9 Schematic representation of apparatus design and Turing Pattern obtained by Castets et al. (ref. 29). Separate reactant reservoirs are separated by a gel strip in which there are therefore natural concentration gradients from top to bottom of strip as shown the Turing Pattern is evident parallel to the reservoirs along which direction there are no natural chemical gradients. The size of the strip is indicated. mena by chemists mainly in response to the undeniable claims based around the Belousov-Zhabotinsky reaction. This is the oxidation of organic species such as malonic or (originally) citric acid by acidified bromate ions catalysed by an inorganic redox couple (typically Ce3 +/Ce4+ or the ferroin/ferrin couples) studied intially as an inorga- nic analogue of the citric acid cycle.In fact Belousov's observations made in 1951 were efrectively contemporary with Turing's work but there was stubborn resistance to accepting that oscillations in chemical systems were allowed by the Second Law of Ther- modynamics and which meant that Belousov's work was published in ob~curity'~ until taken up by Zhabotinsky17 in the early 1960's. The Belousov-Zhabotinsky system does indeed show spatial 'patterns' but these targets and spiral wave^^*'^^ are not true Turing Patterns as they rely in inhomogeneous initiation of reaction and reflect the underlying distribution of such pacemaker sites. (Spiral waves are common in natural non-linear systems including the heart,20 where they are implicated in cardiac arrhyth- mias brain tissue,12 population dynamics as well as having recently been observed in surface-catalysed reactions under ultra-high vacuum conditiom2 ') Many new oscillatory reactions have been devised based on variations on the Belousov-Zhabotinsky reaction and on related oxyhalogen reactions using an algorith- mic appoa~h~~~~~ pioneered by the groups led by Epstein and by Boissonade and De Kepper.One of these reactions the CIMA (chlorite-iodide-malonic acid) reaction has proven to be suitable for observation of Turing Patterns. (This has a pleasant resonance as this system is closely related to the so-called Landolt or iodate+reductant reaction which is commonly studied as a typical clock reaction in undergraduate laboratories Alan Turing had studied some aspects of the Landolt reaction at his school and been awarded a prize based on that The identification of a suitable reaction goes part way to meeting the demands of a Turing Pattern.The remaining obstacles were overcome by the combined ingenuity of several groups. First a whole class of new chemical have been designed which can be termed continuousTflow unstirred reactors. These allow a continuous supply of fresh reactants but in a way that the inevitable fluid flow does not influence the diffusion processes. In many cases this is achieved by conducting the actual reaction in a gel or membrane around which the reactant supplies flow. The role of I in the CIMA reaction suggested that starch or some other indicator could usefully be incorp- orated into the gel to improve the spatial resolution of spatially-resolved spectro- photometric devices commonly based now on video camera systems28 and computerised frame-grabbing data acquisition.In fact this use of I indicators turned out somewhat fortuitously to solve the remaining and perhaps most vexing problem of controlling the species diffusion coefficients. C58 Approachiny Complexity The high molecular mass of an indicator such as starch means that it will be rela- tively immobile in a gel even though typical small ions such as I- will have diffusivities essential that appropriate to aqueous media. If I or I -bind to the immobilised indica- tor their effective diffusion coefficient will be reduced in a manner depending on the complexation equilibrium constant.Thus by varying the loading of the complexing agent the diffusion coefficient of the autocatalytic species in the CIMA reaction can be effectively controlled. This observation coupled with the other advantages of gelled media provided the final piece in the Turing jigsaw and allowed the Bordeaux group led by De Kepper to make the first laboratory-designed observations of a Turing Pattern29 in work developed further by Swinney and co-workers in Texas.,' The Turing Pattern is shown in Fig. 9 and whilst not perhaps as impressive as a leopard is remarkable as it represents the culmination of combined (and competitive) science at its best. 4. Conclusions and Perspectives Chemical kinetics has both much to offer and much to gain from the interdisciplinary interaction with those involved in unravelling the secrets of non-linear systems.At this stage there are perhaps more implications than direct applications but remarkably new ideas for exploitation are already emerging at a rapid rate. We have already identified many appropriate questions to ask of a complicated response of a chemical system so that quantitative measures can be obtained. The pursuit of Turing Patterns an intellec- tual end posed by a pure mathematician has brought direct but perhaps more impor- tantly great indirect benefit to science. In addition to learning through collective efforts to this goal how to control diffusion coefficients how to understand oscillatory reactions sufficiently well to allow the design of further systems and how to suppress natural convective effects we now find ourselves with a whole array of both new theoretical challenges and new experimental reactors in which to study the effects of coupling molecular transport processes and complex chemistry.The latter has great implications for modelling of atmospheric and environmental chemistry systems as well as for biology and medical situations. References 1 Physical Chemistry of Oscillatory Phmomrna Furuduy Symp. Chrm. Soc. 1974 vol. 9. 2 S. K. Scott Oscillations Waves and Chaos in Chemical Kinrtirs Oxford University Press London 1994. 3 P. Gray and S. K. Scott Chemical 0.scillution.s and Instabilities Oxford University Press London 1990.4 S. K. Scott Chemical Chctos Oxford llniversity Press London 1991. 5 R. R. Baldwin and R. W. Walker in Essays in ChLJmisrry ed. J. A. Barnard R. D. Gillard and R. F. Hudson Academic Press New York 1972 vol. 33 pp. 1 37. 6 P. Gray. J. F. Griffiths and S. K. Scott Proc. R. Soc. London. A 1984 394 243. 7 K. Showalter Chcm. Bui. 1995 31 202. 8 P. Gray J. F. Griffiths and S. K. Scott Proc. R. Soc. London. A 1985 397 21. 9 B. Johnson and S. K. Scott J. Chcm. Soc. Ftirudrry Trans. 1990.86 3701; Chuos 1991 I 387. 10 V. Petrov E. Mihaliuk S. K. Scott and K. Showalter Phys. Roc. E 1995 51 3988. 1 I A. M. Turing Phil. Truns. R. Sot,. London B 1952 237 3395. 12 J. D. Murray Muthemuficd Biology Springer Berlin 1989. 13 Chemicul Waves unrl Purrrrns ed.R. Kapral and K. Showalter Kluwer Dordrecht 1904. 14 J. D. Murray Scirnlz/k AntLw'cun 1988 258 80 15 Oscillutions und Trtivrlin~g WUIX~S in ('hemicd Sysiems ed. R. J. Field and M. Burger Wiley New York 1985. 16 B. P. Belousov Sh. R(:/.f: Ratlitit. Mcd. 1958 145 (Russian); see also translation in ref. 15. 17 A. M. Zhabotinsky Biofiziku 1964 9 306. 18 A. T. Winfree Scirnoe 1974 175 634. 19 J. P. Keener and J. J. Tyson Plzpsicu 11 1936 21 307. nd M. C. Mackey Fuom Clocks to Chuos Princeton University Press Princeton 1988. 21 S. Nettesheim A. von Oertzen H. H. Kotermund and G. Ertl J. C'hom. Phys. 1993 98 9977. S. K. Scott c59 22 I. R. Epstein K. Kustin P. De Kepper and M. Orban Sci. Amer. 1983 248,96. 23 P.De Kepper and J. Boissonade in ref. 15 ch. 7 p. 223. 24 I. R. Epstein Chemical and Engineering News 1987,65 30. 25 A. Hodges Alan Turing the Enigma Burnett Books London 1983. 26 J. Boissonade E. Dulos and P. De Kepper in ref. 13 ch. 7. 27 Q. Ouyang and H. L. Swinney in ref. 13 ch. 8. 28 S. C. Muller T. Plesser and B. Hess Science 1985 230 661. 29 V. Castets E. Dulos J. Boissonade and P. De Kepper Phys. Rev. Lett. 1990,64 2953. 30 Q. Ouyang and H. L. Swinney Chaos 1991 I 41 1. Faraday Discussion 100 Celebration Paper; Presented 24th April 1995

ISSN:1359-6640    

DOI:10.1039/FD9950000C47

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Faraday Discuss. 1995,100,55-64 Hydrofluorocarbons and Stratospheric Ozone Timothy J. Wallington and William F. Schneider Ford Research Laboratory SR L-3083 Ford Motor Company Dearborn Michigan 48121-2053 USA Jens Sehested and Ole John Nielsen Section for Chemical Reactivity Environmental Science and Technology Department Riss National Laboratory DK -4000 Roskilde Denmark Recognition of the adverse environmental impact of chlorofluorocarbons (CFCs)' has led to an international agreement to cease their production. Hydrofluorocarbons (HFCs) are important CFC substitutes. An important question regarding HFCs is what is their impact on stratospheric ozone? While it is well known that HFCs themselves do not react with ozone ques- tions have been raised regarding the possibility that species formed during the atmospheric oxidation of HFCs could deplete stratospheric ozone.In March 1993 preliminary work was presented suggesting that CF30 radicals formed in the degradation of many HFCs could deplete stratospheric ozone.2 Later the same year it was shown that CF30 radicals react only slowly with ozone and that HFCs do not destroy ozone through a catalytic cycle involving CF30 What was not considered at that time was the possible depletion of stratospheric ozone by other species formed in the atmospheric degradation of HFCs. To place this concern into perspective it is instructive to contrast the magnitude of the natural ozone flux with that of HFC degradation products. The stratosphere con- tains ca. 3 x lo9 t of ozone in a dynamic equilibrium.Each day 3 x 10' t of ozone are destroyed by natural processes.' If we assume that ozone is distributed approximately homogeneously in a region between 10 and 50 km altitude then the volume average ozone loss rate is 4 x lo6 molecule cmP3 s-'. HFC-134a (CF3CFH2) is widely used now and other HFCs such as HFC-32 (CH,F,) may be used increasingly in the future. By the year 2020 the annual global demand for HFC-134a and HFC-32 has been esti- mated to be up to 3 x lo5 and 0.9 x lo5 t respectively.' (By comparison the global CFC production in 1986 was lo6 t.) A reasonable estimate of the future production of HFCs is 5 x lo5 t or assuming an average molecular weight of 100 5 x lo9 moles per year. HFC degradation is initiated by OH radical attack.Taking a typical activation energy for this reaction of 3.5 kcal mol-' and dividing the atmosphere into a tropo- spheric region with an average temperature of 278 K and containing 90% of the atmo- spheric mass and a stratospheric region with an average temperature of 224 K and containing 10% of the atmospheric mass," then cu. 2% of HFCs will be degraded in the stratosphere. At steady state the HFC degradation flux is ca. 0.5 molecule cm-3 s-'. Clearly the HFC degradation flux is many orders of magnitude less than the natural ozone flux so that for HFCs to impact stratospheric ozone levels requires an efficient catalytic mechanism. Three different mechanisms can be envisioned by which HFC degradation products could impact stratospheric ozone.First fluorine-containing radical species (e.y. FOX FCO or CF30,) could react directly with ozone. Secondly fluorine-containing radical species could impact ozone indirectly via coupling with radical (e.q.ClO, BrO or HO,) 55 56 Hydrofluorocarbons and Stratospheric Ozone or reservoir (e.g. HCl or CIONO,) species that are important in ozone chemistry. Thirdly non-radical HFC degradation intermediates [(e.g. FNO COF or CF,C(O)F] could interact directly or indirectly with species that are important in ozone chemistry. These possibilities are examined in turn below. Direct Impact on Ozone via Reactions involving Fluorinated Radicals To assess the potential impact of HFC degradation products on stratospheric ozone requires us to go beyond the simple estimates of HFC and ozone fluxes discussed above.A somewhat more realistic one-dimensional (1D) estimate of the natural ozone flux can be derived using the expression:". where kx+u is the bimolecular rate constant for reaction of species X with Y and [XI and [Y] are the concentrations of species X and Y. Incorpor-ating species concentration atmospheric density and temperature us. altitude profiles from the 1D radiative convective photochemical model for mid-latitudes at equinox of Brasseur and Solomon,'o along with kinetic data for the relevant reactions from DeMore et al.,' allows us to derive the ozone flux profile shown in Fig. 1. Because heterogeneous reactions are neglected this approach provides only an estimate of the natural ozone flux.However it is a reasonable basis against which any HFC-related ozone destruction can be compared. Similarly the flux of HFC degradation products can be estimated by elaborating on the approximations above. Because the mixing time for HFCs is much less than the lifetime for oxidation it is reasonable to assume that the HFCs are homogeneously mixed throughout the atmosphere (i.e. that the HFC concen- trations scale with density). Taking an annual global emission rate of all HFCs of 5 x lo5 t an average molecular weight of 100 an activation energy for attack by OH radicals of 3.5 kcal mol-' (as for HFC-l34a the most important HFC',) and the OH radical profile of Brasseur and we obtain the HFC degradation flux shown in Fig. 1. As expected from our crude estimate the flux of HFC degradation products is four to seven orders of magnitude less than the natural ozone flux.a), Fig. 1 Flux of HFC degradation (-* * HC1 (-) and 0 (-) as calculated using 1D model described in the text T. J. Wallington et al. HFCs survive for typically 1 to 40 years in the atmosphere before reacting with OH radicals. This reaction produces fluorinated alkyl radicals whose sole fate is reaction with O2to form fluorinated peroxy radicals R02 .' OH + RH +R + H20 (1) R + O2 + M + R02 + M (2) Different HFCs produce different peroxy radicals. For example CH 2F2 produces CHF202,while CF,CFH,* produces CF3CFH02. As with other peroxy radicals fluori- nated peroxy radicals survive in the atmosphere for a few minutes before reacting with NO NO2 or H02 radicals to give directly or indirectly alkoxy radicals such as CHF20 and CF3CFH0.14 The alkoxy radicals survive in the atmosphere for less than one minute before reacting with O2 or decomposing through C-C bond cleavage or both.13 The fact that the peroxy and alkoxy radicals derived from HFCs do not react rapidly with ozone but are lost rapidly by other atmospheric reactions rules out any significant interaction of these species with ozone.During the subsequent oxidation reactions of all HFCs four groups of radicals are formed that could in principle impact stratospheric ozone CF,CO, x = 1-3; CF30, x = 1 2;15 FCO, x = 1-316 and FOX,x = 0-2.'' These C,F,O radicals could interact with stratospheric ozone directly in a catalytic process X+03+XO+02 (3) XO + O/O -+ X + 0,/202 O3 + O/O 202/30 (4) -+ [X = CF,C(O)O CF30 FC(O)O FO or F] as ClO, BrO, NO, HO etc.are known to do. The available experimental data concerning the atmospheric chemistry of C,F,O radicals show that their direct participation in ozone destruction cycles is not important. Relevant kinetic data are gathered in Table 1. Values in parentheses are calculated life- times of the radicals with respect to each reaction based upon the tabulated rate con- stants and reactant concentrations at 20 km taken from Brasseur and Solomon." Table 1 Reaction rates and lifetimes of relevant fluorine-containing radicals radical(XI kX +. NO0 (life t ime)b kX +NOZ0 (lifetime)b kX + CHIO (lifetime)b c1 2.9 x lo-" exp(-260/T) CF,C( 0)W (0.024) - FC(0)O <6 x (>3.4) CF,O 1.5 x 10-14 FO2 <3.4 x 10-l6 (14) FO (605)<2 x Fd 2.8 x (1030) lo-" exp(-230/T) (0.021) a Units cm3 molecule-' s-'.Values taken from ref. 12 19 26-29. Lifetime of species (in s) towards reaction with 0, NO NO2 and CH calculated using trace-gas concentrations at 20 km taken from Brasseur and Solomon.'' ' Lifetime towards decomposition is tl s. The rate of reaction of F atoms with 0 is 4.4 x cm6 molecule-* s-'. The lifetime of F atoms at 20 km towards reaction with O2 is 3.3 x s. 'The rate constant for the reaction of F atoms with HzO is 1.4 x 10-'I. The lifetime of F atoms at 20 km with respect to reaction with H,O is 0.012 s. Hydrofluorocarbons and Stratospheric Ozone As discussed above the natural ozone flux is many orders of magnitude greater than that of HFC degradation products.For direct cycles involving CF3C0, CF30 and FCO to constitute a significant ozone-loss mechansim the rate of the chain propagat- ing reaction of the alkoxy or carboxy radical [CF,C(O)O CF30 or FC(O)O] with ozone must be orders of magnitude greater than the rate of chain termination reactions. As seen in Table 1 such is not the case. CF,C(O)O radicals are lost rapidly (within 1 s) by C-C bond cleavage and are not reformed. CF30 radicals are lost via reaction with CH4 and NO. Reaction of CF30 with CH produces CF30H which cannot reform CF30 radicals chemically or photolytically." Reaction of CF30 radicals with NO gives FNO and CF,O neither of which can reform CF30.FC(0)O radicals are lost by reac- tion with NO to give FNO and CO, which cannot reform FC(0)O radicals. The effi- cient irreversible loss mechanisms for CF,C(O)O CF30 and FC(0)O radicals together with their relative unreactivity towards ozone preclude any significant direct impact on the ozone layer associated with CF3C0, CF30x and FCO radicals. Consideration of the direct impact of FOX radicals is complicated by the rapid inter- conversion between F FO FO and FNO (see Scheme 1) F + 0 + MeFO + M FO + NO + FNO + 0 FNO + hv -+ F + NO F + O3+FO + 0 FO + NO+ F + NO F + CH,/H,O + HF + CH3/OH (10) Examination of Table 1 suggests that FOX radicals do not destroy ozone in a catalytic fashion because (i) FO and F0 radicals react extremely slowly with ozone and (ii) FOX radicals are lost efficiently by reaction of F atoms with CH and H,O.The direct impact of FOXchemistry on stratospheric ozone has been addressed quantitatively in a model- ling study by Sehested et a1." and was found to be unimportant. While the efficiency of ozone destruction varies with altitude the number of ozone molecules consumed per F atom was less than three at all altitudes." The arguments presented above rule out direct involvement of fluorinated radical . species in catalytic ozone destruction cycles of any significant length. Indirect Impact on Ozone via Reactions involving Fluorinated Radicals Indirect involvement of CxF,O radicals could occur either by reaction with radical (e.g.ClO, HO, BrO,) or reservoir (e.y.ClONO or HCl) species from the other chains. An example of such a coupling would be CF30 + C10 -+ CF + C1 + 0 (1 1) A necessary condition for significant interaction of CxFyOz radicals with species from catalytic ozone destruction cycles is that the lifetime of these species with respect to reaction with CxFyOz radicals must be significant relative to that with respect to other loss processes. In this section we will show that this condition cannot be fulfilled because the concentration of C,F,O radicals in the stratosphere will be too low. We assume initially that the CF30 and FC(O)O radical fluxes can be approx- imated by the HFC degradation flux or 0.1-1 molecule cmP3 s-' each (see Fig. 1). The actual radical fluxes will be less because neither CF,O nor FC(0)Ox are produced quantitatively from HFC degradation at all altitudes.Multiplication of the radical fluxes by their atmospheric lifetimes with respect to all reactions provides estimates of their T. J. Wallington et al. steady-state concentrations. From Table 1 the atmospheric lifetimes of CF30 and FC(0)O radicals are ca. 10-15 s at 20 km. These lifetimes are typical throughout the stratosphere. Hence we derive an estimate of 1 to 15 molecule cm-3 for the maximum steady-state concentrations of CF30 and FC(0)O radicals. The largest possible rate constant for reaction of CF30 and FC(0)O radicals with any species in the stratosphere is limited by the frequency of molecular collisions and is lo-'' cm3 molecule-' s-'.Combination of the maximum possible rate constant with our estimate for the maximum steady-state concentrations provides a range of 20 to 320 years for the lower limit of the lifetime of any species in the stratosphere with respect to reaction with FC(0)O or CF30radicals. This lifetime is many orders of magnitude greater than those of the radical and reservoir species in the ClO, HO and BrO catalytic cycles which range from a few minutes to several days. Thus no significant interaction is possible between FC(0)O or CF,O radicals and any catalytic cycle. Similar arguments apply for CF302 and FC(0)02 radicals because the lifetimes of these radicals with respect to reaction with NO is ca. 2 min. Calculation of the steady-state concentration of FOXradicals is complicated by two factors.First FOXradicals are formed uia photolysis of several HFC degradation pro- ducts. Differences in the rates of photolysis of the various degradation products must be considered. Second atmospheric FOXchemistry is rather complex (see Scheme 1) and involves the equilibrium reaction (5,-5) the existence of three different radicals (F FO and F02) and the formation and photolysis of FNO. To arrive at estimates for the steady-state concentrations of FOXradicals we need to combine information on the FOX radical lifetimes and the F atom flux from HFC degradation (labelled Ffluxin Scheme 1). 10 RH I t '\ 1 El El Scheme 1 Numbers refer to reactions in text FOXradicals are formed during the HFC degradation process following photolysis of FNO COF, HC(0)F and CF,C(O)F.The rate of photolysis of COF depends upon the altitude and its lifetime us. photolysis decreases from 64 years at 25 km to 0.6 years at 45 krn.,' Air resides for typically 3 years in the stratosphere before returning to the troposphere."' Hence photolysis of COF will be of negligible importance below 25 km. The rates of photolysis of HC(0)F and CF,C(O)F are expected to be similar to that of COF,. FNO photolyses rapidly at all altitudes to give F atoms. The atmospheric life- time of FNO is ca. 15 min in the daytime., Below 25 km photolysis of FNO is the sole source of F atoms and hence FOXradicals. The only source of FNO in this region is reaction of CF30radicals with NO. CH competes for the available CF30 radicals and produces CF,OH which is inert in the stratosphere." The data in Fig.1 can be used to estimate the F-atom flux at follows. First by analogy to HFC-l34a which is the most important HFC the degradation of the 'HFC mix' is assumed to yield CF radicals in a yield of 30% in the ~tratosphere.~~ At alti- 60 Hydrofluorocarbons and Stratospheric Ozone tudes below 25 km Ffluxis 30% of the HFC degradation flux multiplied by that fraction of CF,O radicals that react with NO. Secondly at 45 km and above we assume that all HFC degradation products photolyse to release all their F atoms. Thirdly each HFC molecule is assumed to contain four fluorine atoms. The F-atom flux from HFC degra- dation at 45 km is then four times the HFC degradation flux.Fourthly we assume that the efficiency with which F atoms are liberated at altitudes below 45 km decreases lin- early to zero at 25 km. Thus at 35 km we assume that 50% of the fluorine is liberated during the degradation and so the F-atom flux is twice (0.5 x 4) the HFC degradation flux. With these assumptions the calculated F-atom flux varies from 0.16 to 1.6 atom cm-3 s-' over the altitude range 10 to 50 km. To compute the FOXradical steady-state concentrations we use the reactions given in Scheme 1. HF is the final product in atmospheric fluorine chemistry. The strength of the H-F bond precludes reaction with all but the most energetic species in the strato- sphere O('D) atoms. While reaction of O('D) atoms with HF proceeds at essentially the gas kinetic limit (i.e.on each collision) the concentration of O('D) atoms is so low (0.1-lo00 atom cm-3 depending on altitudeI2) that this reaction has no overall signifi- cance. The Ffluxterm is evaluated as a function of altitude as discussed above. At steady state the following relationships hold [F], = Fflu~kl0; [FO], = k,[F],,/k and [FO,], = k,[F],,/(k-+ k6). k k-, k k ,k and klo are the pseudo-first-order loss rates of fluorine-containing species in the corresponding reactions given above. Results are given in Fig. 2. F02 is the most abundant FOX radical species with a maximum steady-state concentration of the order of 50-500 cm-'. The upper limit for the rate constant of reaction of F02 radicals with any atmo- spheric species is lo-'' cm3 molecule-' s-'.By comparison the largest known F02 radical bimolecular rate constant is 100 times less (for reaction with NO',). Com-bination of the maximum possible rate constant (10- lo)with the steady-state concentra- tion provides a range of 1 to 20 years for the minimum lifetime of any species in the stratosphere with respect to reaction with F02 (Fig. 3). Lifetimes with respect to reac- tion with FO radicals will be 10 to 10000 times longer. The lifetimes of the radical and reservoir species in the ClO, HO and BrO catalytic cycles due to processes unrelated to HFC chemistry range from a few minutes for radicals to a few days for reservoir species. The data in Fig. 2 and 3 show that there is no possibility of any significant Fig. 2 Concentrations of (a) F (b) FO and (c) FO radicals calculated using the 1D model described in the text.For comparison the calculated steady-state FNO concentration is shown (d). T. J. Wallington et al. Fig. 3 Minimum lifetime of any species in the stratosphere with respect to reaction with FO radicals. See text for details. coupling of F atoms or FO radicals with radical and reservoir species from these other catalytic cycles because of the very long lifetimes for reactions of F and FO with the catalytic species (i.e.because of their very low concentrations). As seen from Fig. 3 the concentration of FO radicals peaks at 30 to 35 km. At this altitude the minimum lifetime of any species with respect to reaction with FO radicals is ca. 1 year. To place this result into perspective we can consider the possibility of coupling of F0 radicals with the C10 and BrO cycles.HCl is the most persistent species in the C10 and BrO cycles and so is the most sensitive with respect to a potential reaction with FO,. HCl reacts with OH radicals to give C1 atoms. Using k(OH + HCl) = 2.6 x lo-' exp(350/T) cm3 molecule-' s-' l2 together with the tem- perature and OH radical concentration profiles discussed above the HCl lifetime with respect to OH attack is ca. 13 days in the region between 30 and 35 km. By comparing this result with the minimum lifetime of 1 year with respect to reaction with FO, it is clear that even for the most sensitive species (HCl) the impact of FO radical chemistry in the stratosphere can be at most of minimal (a few per cent) importance.This result is obtained assuming a gas kinetic reaction between FO radicals and HCl; assuming a more realistic reaction rate would drastically reduce the potential impact. In considering the arguments presented above it is relevant to note that air in the stratosphere is recy-cled back into the troposphere in a time period of typically 3-5 years. Any process taking longer than 3-5 years is unlikely to be important. Impact on Ozone via Reactions involving Non-radical Species Finally we must consider the potential impact of non-radical HFC oxidation products. The major long-lived products of the oxidation of all HFCs are FNO HF COF, HC(0)F and CF,C(O)F. None of these compounds react directly with ozone. Further none of these compounds are known to react with radical or reservoir species from other ozone-depleting catalytic cycles.As discussed above the fluorinated radical species do not have a significant impact on stratospheric ozone. For the non-radical HFC deg- Hydrofluorocarbons and Stratospheric Ozone radation products to impact ozone they must participate in reactions that involve species that are relevant in ozone chemistry. We consider the potential importance of these interactions here. First as discussed above FNO atmospheric chemistry is dominated by its rapid photolysis22 and ultimate conversion to HF. The steady-state concentration of FNO is related to that of FO radicals by the expression [FNO], = k6[F02],,/k (see Scheme l) where k6 and k are the pseudo-first-order loss rates of FO with respect to reaction with NO and FNO with respect to photolysis respectively.Assuming an altitude-independent photolytic lifetime for FNO of 15 rnin2 we derive the altitude profile for FNO given in Fig. 2. As seen from Fig. 2 the expected FNO concentration is similar to that of FO radicals. At night photolysis of FNO ceases and the FNO concentration could build up to equal the sum of the FO radical and FNO concentrations in Fig. 2. In the preceding section it was argued that because of their low concentration FO radicals will not couple significantly with radical or reservoir species from the ClO, BrO or HO cycles. The same argument applies to FNO. At the opposite extreme of reactivity the H-F bond is too robust to be efticiently cleaved either chemically or photolytically in the stratosphere.HF is too unreactive to couple with radical or reservoir species from catalytic ozone cycles. COF ,HC(0)F and CF,C(O)F represent intermediate cases. Their stratospheric life- times are determined by photolysis and cycling into the troposphere depending on alti- tude and can be of the order of several years. These lifetimes are long enough that interaction with other ozone cycles may be possible. Interaction with species from the ClO, BrO or HO catalytic ozone cycles cannot be excluded on purely thermodyna- mic grounds either as is possible with HF. Because their lifetimes are relatively long the steady-state concentrations of COF HC(0)F and CF,C(O)F are potentially rather large.For example if we assume that COF is formed quantitatively and essentially instantaneously from HFC degradation and assume that the lifetime for removal of COF from the stratosphere is 3 years then at 20 km altitude the COF steady-state concentration will be roughly 2 x lo7 molecule cmP3. We argued above that significant interaction between fluorine radical species and reservoir species from other chains could not be important because of the short lifetimes and low concentrations of the fluorine radicals. Such an argument clearly cannot be applied to the non-radical species. We can place some limits on the potential impact of COF HC(0)F and CF,C(O)F however. First any atmospherically significant reactions involving these species must involve breaking a C-F C-C or C-H bond.Because of the oxidizing nature of the atmosphere these bonds cannot be reformed once broken. Hence none of these species can interact in a catalytic fashion with other ozone-depleting chains. Secondly any reac- tions that merely convert these species into fluorine-containing radicals cannot be important. As discussed above even assuming generous values for the radical fluxes the fluorine radical species can have no significant impact on ozone. Thirdly it follows that reactions between the non-radical fluorine species and radical species from ozone-depleting chains cannot be important because such reactions cannot liberate new ozone-active species. Such reactions would only short circuit the ozone-depleting chains probably diminishing their effectiveness.We conclude that the only possibly relevant reactions are between the non-radical fluorine species and reservoir species from cata- lytic chains such as HCI and CIONOz with the regeneration of a catalytic ozone- depleting species. For reactions between non-radical fluorine species and reservoir species to be signifi- cant they must proceed at a rate at least comparable with the rate of removal of the non-radical fluorine species from the stratosphere. For example HCl is the predominant chlorine reservoir species in the stratosphere with a concentration of ca. 1 x lo9 mol-ecule cmP3. For reaction with HCl to be competitive with physical removal of non- radical fluorine species (which takes about 3 years) the rate constant must be of the T.J. Wallington et al. order of cm3 molecule-' s-'. The reservoir species from the catalytic chains are all closed-shell molecules as are COF HC(0)F and CF,C(O)F. The rate constants for reactions between two closed-shell gas-phase molecules generally lie in the range 10-30 to lo-'' cm3 molecule-' s-' at 298 K,24 clearly many orders of magnitude less than lo-'' cm3 molecule-1 s-'. The robustness of the bonds in the fluoro-organics COF, HC(0)F and CF,C(O)F as well as the low temperatures in the stratosphere makes homogeneous gas-phase reaction between these species and reservoir species extremely unlikely. Heterogeneous reactions cannot be similarly dismissed. However Hanson and Ravishankara have demonstrated that COF is unreactive on ice surfaces,25 and it seems reasonable to suppose that HC(0)F and CF,C(O)F will behave similarly.The available evidence indicates that interaction between non-radical fluorine species and reservoir species will not significantly affect stratospheric ozone. As further evidence let us suppose that some homogeneous or heterogeneous reac- tion between an HFC degradation product and HCl does liberate active chlorine. As mentioned above HCl is the most persistent of reservoir species in the chlorine cycles and HCl is converted into an active form through reaction with OH radicals. The flux of chlorine atoms generated by this process can be calculated as a function of altitude (Fig. l) and varies from 20 to 600 molecule cm- s-'. This flux is two to three orders of magnitude greater than the HFC-degradation flux depending on altitude.If one assumes that the entire HFC-degradation flux is converted to COF, and that all COF reacts with HCl to liberate C1 atoms then the Cl atoms will be liberated at most at a rate that is less than 1% faster than the background HCI + OH rate. Such a small increase in C1-atom flux would have at most a marginal impact on stratospheric ozone concentrations. Finally one might imagine other possible minor HFC degradation products such as FON02 FONO and FO,NO, that could interact with chlorine reservoir species and be reformed by gas-phase reactions thus constituting a cycle. However all such reac- tions involve FOXchemistry the efficiency of which is severely limited by the irreversible formation of HF as discussed above.Thus these processes will at most have very short cycle lengths and are unlikely to be significant. Conclusion Based upon the current understanding of the atmospheric chemistry of HFCs we can draw the following conclusions. First HFCs are much more readily oxidized in the atmosphere than are CFCs and the stratospheric burden of HFCs will be approx- imately an order of magnitude less than that of CFCs. HFC degradation products could appreciably impact stratospheric ozone only via catalytic cycles. Secondly of all the known fluorine-containing radical species only fluorine atoms themselves react at an appreciable rate with ozone and the concentration of fluorine atoms is sharply limited by the irreversible formation of HF.Fluorine radical species do not participate in direct catalytic ozone-depletion cycles of any significant length and do not impact strato- spheric ozone concentrations. Thirdly the concentrations of all fluorine radical species in the atmosphere can be estimated and will be too low to couple significantly with radicals or reservoirs from other catalytic ozone-depletion cycles. Fluorine radical species do not indirectly impact stratospheric ozone. Fourthly non-radical fluorine- containing species do not react with ozone. They may conceivably impact radicals or reservoirs from other catalytic ozone depletion cycles. However this impact is unlikely to be important because it requires either unrealistically fast homogeneous gas-phase reactions between closed-shell species or heterogeneous chemical reactions for which there is no experimental evidence.Thus the available scientific data strongly suggest that HFCs pose no threat to stratospheric ozone. Hydrofluorocarbons and Stratospheric Ozone We thank Dr. R. A. Cox (Natural Environmental Research Council UK) and Dr. Susan Solomon and Dr. A. R. Ravishankara (National Oceanic and Atmospheric Adminis- tration USA) for helpful discussions. References 1 M. Molina and F. S. Rowland Nature (London) 1974 249 810. 2 P.Biggs C. E. Canosa-Mas D. E. Shallcross R. P. Wayne C. Kelly and H. W. Sidebottom Pro-ceedings of the STEP-HALOCSIDEIAFEAS Workshop University College Dublin Ireland March 1993 p. 177. 3 0.J. Nielsen and J. Sehested Chem. Phys. Lett.1993 213,433. 4 T. J. Wallington M. D. Hurley and W. F. Schneider Chem. Phys. Lett. 1993,213,442. 5 M. M. Maricq and J. J. Szente Chem. Phys. LRtt. 1993 213,449. 6 A. R. Ravishankara A. A. Turnipseed N. R. Jensen S. Barone M. Mills C. J. Howard and S. Solomon Science 1994 263 75. 7 Ch. Fockenberg H. Saathoff and R. Zellner Chem. Phys. Lett. 1994,218,21. 8 M. McFarland and J. Kaye Photochem. Photobiol. 1992,55,911. 9 A. McCulloch Environ. Monit. Assess. 1994 31 167. 10 G. Brasseur and S. Solomon Aeronomy of the Middle Atmosphere Reidel Dordrecht The Netherlands 1986. 11 H. S. Johnston and J. Podolske Rev. Geophys. Space Phys. 1978,16,491. 12 W. B. DeMore S. P. Sander D. M. Golden R. F. Hampson M. J. Kurylo C. J. Howard A. R. Ravishankara C. E. Kolb and M.J. Molina Jet Propulsion Laboratory Publication 92-20 Pasadena CA 1992. 13 T. J. Wallington W. F. Schneider D. R. Worsnop 0.J. Nielsen J. Sehested W. J. DeBruyn and J. A. Shorter Environ. Sci. Technol. 1994 28 320. 14 P. D. Lightoot R. A. Cox J. N. Crowley M. Destriau G. D. Hayman M. E. Jenkin G. K. Moortgat and F. Zabel Atmos. Environ. A 1992,26 1805. 15 M. K. W. KO N-D. Sze J. M. Rodriguez D. K. Weistenstein C. W. Heisey R. P. Wayne P. Biggs C. E. Canosa-Mas H. W. Sidebottom and J. Treacy Geophys. Res. Lett. 1994 21 101. 16 J. S. Francisco A. N. Goldstein Z. Li Y. Zhao and I. H. Williams J. Phys. Chem. 1990,94,4791. 17 J. S. Francisco J. Chem. Phys. 1993,98 2198. 18 T. J. Wallington and W. F. Schneider Enuiron. Sci Technol. 1994,2?3 1198. 19 J.Sehested K. Sehested 0.J. Nielsen and T. J. Wallington J. Phys. Chem. 1994,98 6731. 20 A. Nolle H. Heydtmann R. Meller W. Schneider and G. K. Moortgat Geophys. Res. Lett. 1992 19 281. 21 M. K. W. KO N. D. Sze and M. J. Prather Nature (London) 1994,367,505. 22 T. J. Wallington W. F. Schneider J. J. Szente M. M. Maricq 0.J. Nielsen and J. J. Sehested J. Phys. Chem. 1995,99,984. 23 T. J. Wallington M. D. Hurley J. C. Ball and E. W. Kaiser Enuiron. Sci. Technol. 1992,26 1318. 24 F. Westley D. H. Frizzell J. T. Herron R. F. Hampson and W. G. Mallard NIST Standard Reference Database 27 Gaithersburg Maryland 1993. 25 D. R. Hanson and A. R. Ravishankara Geophys. Res. Lett. 1991,18 1699. 26 A. A. Turnipseed S. B. Barone and A. R. Ravishankara J. Phys. Chem.1994,98,4594. 27 S. B. Barone A. A. Turnipseed and A. R. Ravishankara J. Phys. Chem. 1994,98,4602. 28 T. J. Wallington and J. C. Ball Chem. Phys. Lett. 1995 234 187. 29 R. Zellner and H. Saathoff paper presented at the 12th Int. Symp. on Gas Kinetics Reading 1992. Paper 5/00196J; Received 6th January 1995

ISSN:1359-6640    

DOI:10.1039/FD9950000055

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

FU~U~UY 1995,100 C61-C65 D~SCUSS. How Discoveries are Made and Why it Matters John C. Polanyi Department of Chemistry University of Toronto Canada M5S I A1 I thought that it would be appropriate for the only North American speaker to offer an expansive title. How are discoveries made and why does it matter? Michael Faraday admonished us to stick to a single idea per presentation so I answer my question in one breath by saying that nobody knows how discoveries are made and this matters because it dictates that we not attempt to govern science as if we did know. Science has a vocabulary of supposed facts true to varying degrees and a grammer of accepted practices applicable in various circumstances. But science is more than this grammar applied to this vocabulary since it attempts not merely to say things but to say things that are true.Computer science has once again come to our aid in making this point. A program called ‘Racter’ (‘computerese’ for ‘Raconteur’) applying impecca- ble syntax to an extensive vocabulary wrote on command an essay on love of which I quote only the opening We will commence with a question does steak love lettuce? This question is impla- cably hard and inevitably difficult.. . There is a freshness of view here. It is marred however by a total lack of meaning. The computer quite evidently has not experienced love. Just as well perhaps. Michael Faraday a penniless bookbinder’s apprentice was drawn to the door of this Institution in the year 1810 out of love for science. It is a passion that many in this audience share.A decade later Faraday was living with his young wife in the attic over our heads and shortly after was lecturing to enraptured audiences in the lecture theatre where we meet. His encounter with the reality of science was triumphant; he ranks with the greatest scientists of all time. But the course of his love affair with science was as turbulent and taxing as were many of the great loves of history. Though himself gentle and pious he had quarrels with his principal scientific patron and model Sir Humphrey Davy he was accused of plagiarism and he was for a while thrown out of his church. This last was only indirectly owing to science; as an eminent scientist he had accepted an invitation to dine with the Queen one Sunday and when rebuked by his church for this frivolity he refused to repent.These dramas were merely the outward evidence of an inward collision between his ideals and the reality of science. For scientific discovery since it cannot be performed by computers is pursued by human beings with their normal strengths and weaknesses. This is not of course said to belittle science. Quite the contrary. In the greatest adventures of which science is one weaknesses contribute to strengths and vice versa. Is conservatism to give one example a weakness or a strength? It is a strength because it permits the explorer to steer a steady course in the midst of confusion but a weakness since it blocks the path to acceptance for the Michael Faradays of this world.Is oppor- tunism a strength or a weakness‘? It is a weakness when it causes the navigator to vacillate between goals and a strength when it makes him quick to move the tiller if disaster threatens. C61 C62 How Discoveries are Made and Why it Matters It is the balance between these countervailing influences that determines success or failure in the making of scientific discoveries and the optimal balance is achieved by being alert to as broad a range of life’s experiences as possible. Michael Faraday came to science with a different view. This great iconoclast was naively convinced that the best scientist was the most detached from the world. He expected others to be as dispassionate as value-free in their judgments as he supposed himself to be.All the while he was systematically placing high explosives beneath the statue of Isaac Newton. Matter he believed could have-in addition to the mass that Newton had given it-both charge and magnetism. Space could transmit the resulting forces whether it was full or empty. With exceeding modesty he proceeded on this basis to replace the Newtonian universe by the Faradayan. He did not insist on his view. He did not need to since he knew he was right. Others rather to his surprise disagreed. But not all others and not even the most influential of others. For had this been the case we would not have heard of Michael Faraday and science would not be the power in the world that it is. Faraday proceeded in this fashion for over twenty years as a scientist should and as only a great scientist can.Understandably he became dispirited. Gradually he became a recluse. He had overreached himself scientifically attempting things that would not in fact be possible for decades to come. But so sure was his insight into nature’s sym- metries that this premature science was as close as he came to being wrong. With one exception. In the matter of understanding the nature of the pursuit of which he was a master-science-he was surely at times mistaken. The scientific method is not however one that we shall fully understand until the day when human consciousness focusing on itself reveals its own nature. Meanwhile we should be content to admit that science is a skill a way of knowing not very different from other ways of knowing developed over the millennia and transmitted imperfectly in the course of a lifetime by way of emulation.Those we emulate are those who in the course of history have exhibited the skill we seek. Among those happily are some who are living. Which brings us to the Faraday Discussions of which the hundredth has recently taken place and to the role of dis- cussion in general as a means to knowing and therefore to discovery. The word ‘discussion’ comes from the Latin discussio to shake apart. Unprompted one thinks of discussion reasonably enough as the systematic weighing of arguments against counter-arguments. And it is this element in the Faraday Discussions that tends to find its way into the bound volumes on scientists’ shelves. But it is not predominantly for this that scientists come from around the world whenever a Faraday Discussion is announced.They come to be participants or witnesses at the central mystery of ‘shaking apart ’. This is not such a solemn rite as it may appear but more akin to an act of play. Kittens do the same with balls of wool. The function of this teasing is the great mystery. A paradox that has been debated off and on for over two thousand years is the way in which a problem solver proceeds purposefully and with a sense of direction toward an unseen goal. In my childhood this would have been referred to as the Indian Rope- Trick The Indian holy man was alleged to throw a rope into the sky where it remained as a result of his faith long enough for him to climb it if he hurried.A more revealing description of this mystery is to say that the knotty problem is ‘shaken apart’ by being worried this way and that until it takes on a recognisable shape. We depend upon the breadth of experience of the onlooker as a means to recognising the shape and then on his or her deductive skills in order to relate that shape back to its original form. If this is as appears likely a crucial step in determining whether discoveries are made it is not difficult to see in the words of my title ‘why it matters’. It matters for John C. Polanyi C63 two immediate reasons. In the first place this scenario illuminates the function of the group and in the second it exposes to view the crucial role of the individual. ‘Shaking apart’ is a labour that can indeed be shared for example between shaker and shook.Pattern recognition by contrast is indivisible. It takes place initially in a single prepared mind culturally attuned to the pattern that is unfolding. Meanwhile the group can be expected to continue if not redouble its efforts to shake the problem and also its purported solution apart. How are discoveries made? In these brief remarks I have been stressing the anguish of the process its subtlety its dependence on a community of truth-seekers and equally on a lonely believer who must be permitted to defy that community. If the subtlety of this process does not give pause to would-be administrators there are two short answers that should be given to the question of ‘how discoveries are made’; imperceptibly and infrequently.I comment on these in turn. Scientific discoveries occur as the result of the accumulation of scientific-evidence. Only the trained eye can see them gradually taking shape. There is no set level of evi- dence at which the bell of proof rings. It follows that governments seeking to maximise the return in terms of discoveries per research expenditure will be dependent on the judgment of scientists in applying the accountancy that good business practice requires. The only escape from this ‘peer review’ is into a never-never land of waste. Discoveries are made gradually and seldom. There is no guarantee that this gradual process once begun will continue. Lord Kelvin made the definitive statement on the prevalence of fake pregnancies in science.He was asked what in his long life in science stood out most clearly. He thought for only a moment before replying; ‘Failure’. High-gain activities tend to be high risk otherwise the world would be even more unstable than it is. But high-risk activities pose a problem for administrators whose job it is to reduce waste and therefore risk. Josef Stalin once summoned the Soviet Union’s leading film-maker to the Kremlin to inquire how many films he had made in the past year. ‘Eighteen' was the impressive response. ‘And how many of those would you regard as having fully succeeded?’ inquired the great one ominously. ‘Three’ replied the film-maker (exaggerating surely). ‘In that case next year you will make three films’ Stalin said terminating the interview.So which three? In the case of science it would in my country be those few dis- coveries that had most clearly contributed to what in previous years we called ‘socio- economic well-being’ and today more desperate or grasping we simply call ‘wealth’. Governments have however had much more success applying this type of criterion in retrospect than in prospect. In experiments extending over three-quarters of a century a number of European nations proved to their people’s satisfaction that they were quite unable to pick out in advance the businesses that would be most profitable. It is a curious fact however that if to the difficulty of predicting what businesses will yield profits one adds the further imponderable of predicting what scientific discoveries will benefit what businesses the whole undertaking appears once more to be if not feasible at least saleable.Basic science must after all be accountable and here is a method to make it seem so. The fact that it will replace real criteria of worth namely scientific achievement by what must needs be a judgment based on the seat of a bureaucrat’s pants tends to be overlooked. The problem is more serious than I have yet suggested for science today is threat- ened not only by misapplied post hoc judgments but also by the pre hoc provisions of stylistic guidelines. We may not be alone in Canada in having Centres of Excellence at our universities in which the criterion of excellence is clearly stated to count for twenty percent in the assessment.It is outweighed four to one by ill-defined (because indefinable) criteria having to do with the qualities of management team-work and relevance. And yet world-wide the lesson has been learnt where commerce is concerned that C64 How Discoveries are Made and Why it Matters the risk-taker and not the legislator is the best decision-maker. Why is that not also thought to be true in the highly competitive market for new ideas? Perhaps because governments have yet to be persuaded that the self-regulating system of the market-place for goods has relevance to that for ideas. Does science actually have a ‘bottom line’? Are international prizes (to give an example) perhaps distributed by lot? Some would be loath to admit it. Is it the case? Consider the record of twentieth century science guided so far almost exclusively by scientists on the basis of scientific merit in producing scientific advances of importance.The record is so spectacularly good as to raise fears that we cannot cope with the tidal wave of discovery. It would appear that the criteria by which the scientific community have been determining the profitability of scientific investments are real and effective. Is there nothing then in the management of science that needs improvement? Do we merely need to set our scientists free so that they can make discoveries where nature permits rather than where governments propose? No there is more that we can do. We can further improve the communication between the basic scientist and the applied the academic laboratory and the market- place for goods.Much has however already been done on these lines in recent years. So much that we must now be careful not to let the applied sector set the agenda for the basic or we shall find that we have spent the bulk of our intellectual capital in pursuit of near-term profit. The objective must be to use what we know not to restrict what we know to what we know how to use. Beyond using what we know lies a further objective; to use it wisely. This will require that we levy a tax of citizenship on knowledge. There must be an obligation on scientists (not all of them always but many from time to time) to participate in the public debate by which our future is determined. This is not because scientists are possessed of special wisdom but because they have a special form of literacy and every form of literacy carries with it obligations.Those who can read for example are required to bring warning signs to the attention of those who cannot. I said that scientists do not have a special wisdom. They do however share distinc- tive values. These values are not so much those of scientists as more broadly of scholars. Nonetheless they constitute a proud element in what we term the scientific met hod. Science requires that no nation religion ethnic group class or gender be regarded as having a monopoly on the truth. This remains an exceptionally civilised concept viewed in the light of today’s world. A further principle which flows from the first is that of civility.Since the truth is a shared concern there should be no disrespect implied in disagreement. Not only must that be acknowledged it must be made evident in the way we treat one another. John Tyndall and Michael Faraday major scientific competitors attacked one another’s views forcefully in the Philosophical Magazine of 1855. Strikingly the date-line for their conflicting letters was the same; The Royal Institution London. Conscious of their differing viewpoints Michael Faraday had invited Tyndall to join him at the Royal Institution some years earlier. There they shared a domicile and a lifelong friendship. Commenting on such altercations among colleagues Faraday wrote Each one gives views and ideas new to the rest. When science is a republic then it gains; and though I am no republican in other matters I am in that... If this were all we knew of how discoveries are made it would be enough to explain ‘why it matters’. The pursuit of science embodies a central requirement that we note with pride on the occasion of the hundredth Faraday Discussion ‘There must be dis- cussion to show how experience is to be interpreted’. John C. Polanyi C65 The quotation this time is not however from Michael Faraday but from Faraday’s contemporary John Stuart Mill writing ‘On Liberty’. What we know about how dis-coveries are made matters. It matters to science and it matters still more beyond science. Faraday Discussion 100 Celebration paper; Presented 24th April 1995 Quotations from the writings of Michael Faraday are from Faraday as a Natural Philosopher Joseph Agassi Chicago University Press 1971.

ISSN:1359-6640    

DOI:10.1039/FD9950000C61

出版商:RSC    

年代:1995

数据来源: RSC