42
The Journal of Physical Chemistry, Vol. 83, No.
1, 1979
D. L. Baulch and D. C. Montague
Some Aspects of the Evaluation of Kinetic Data D. L. Baulch and D. C. Montague” Department of Physical Chemistry, University of Leeds, Leeds, England (Received Ju/y 10, 1978)
The need for reliable reaction rate data for the analysis of complex gas-phase reaction systems has stimulated the production of tables of critically evaluated rate constants. The procedures and criteria used in such evaluations are outlined and some of their limitations discussed. The part played by theory in guiding the evaluator is discussed and illustrated with examples from decomposition-combination reactions. Attention is drawn to the particular problems associated with evaluation of the temperature dependence of rate constants.
Origins and Current Work The need for critically evaluated reaction rate data has been demonstrated most dramatically in the Climatic Impact Assessment Program mounted to evaluate the effects of supersonic aircraft on the ozone layer and, more recently, in attempts to model the effects of halogenated hydrocarbons on the chemistry of the stratosphere. However the need was appreciated much earlier being prompted by a number of factors prominent among which were (i) the rapid expansion of the scientific literature in the 1960’s, (ii) the then growing significance of combustion problems, particularly pollution, which was realized to originate largely from kinetic rather than equilibrium effects, and (iii) developments in digital computing which enabled complex chemical systems to be modeled numerically. In the United Kingdom in 1966 a working party of the Science Research Council reporting on high temperature processes recognized (i) the importance of kinetics data to this field, (ii) that kinetics data of widely variable quality was being used in industry and elsewhere, and (iii) there was a need to bring some order to this and to the proliferating kinetics literature. It was recommended that a group be established to compile and evaluate rate data and publish their findings. The work began in the Department of Physical Chemistry, University of Leeds, in late 1967. About this time similar work was initiated a t Birmingham (Dr. J. A. Kerr), in the USSR (Professor V. N. Kondratiev), and the National Bureau of Standards established the Chemical Kinetics Information Center under the direction of Dr. D. Garvin. As an international forum CODATA established a Task Group on Chemical Kinetics to encourage such work and to try to identify areas where good data were required. The output from these programs, together with other useful sources of kinetics data, has been listed in a recent article by Hampson and Garvin.’ Most of these evaluation groups are still active. The CODATA task group is currently preparing a set of recommended rate data and photochemical data for modeling atmospheric reactions. The Chemical Kinetics Information Center has now established a most comprehensive collection of bibliography, computer filed, from which it abstracts and publishes periodically bibliographies on specific areas in kinetics; it has also initiated, and continues to participate in, evaluation work. The Leeds group has a fourth volume in their series “Evaluated Kinetics Data for High Temperature Reactions” in preparation for publication dealing with reactions of halogenated species, *Address correspondence to this author a t the Department of Chemistry, University of California, Irvine, Calif. 92717.
OG22-3654/79/2083-0G42$Gl .OO/O
and work has commenced on a fifth volume treating reactions of atoms and small radicals with hydrocarbons. Current work in Birmingham is concentrated on compilation of bibliographies rather than evaluation. However, there appears to be little Russian work of this kind in progress. Individuals continue to contribute evaluations, reviews, and compilations. The advent of the Journal of Physical and Chemical Reference Data specializing in work of this kind has probably been a most important influence on standards and the wider distribution of critically evaluated material.
Methods and Criteria Evaluation. It is important t o distinguish between compilation and evaluation. The former merely lists the available data for a particular reaction, while in the latter this material is critically assessed and an opinion expressed as to its reliability. There is, however, a continuous range of possibilities between these two extremes, and all are useful. T o some extent this range reflects the variety of users of the material, from the “connoisseur” who may be more knowledgeable than the evaluator, to the user from another discipline with only a limited knowledge of the field, but in need of an “instant rate constant” to use in attacking an immediate problem. The degree of evaluation in a particular review also reflects the amount of effort that the evaluator is able or willing to spend on the task. Truly critical evaluation requires the services of experienced scientists and is expensive and time consuming. To satisfy the needs of this whole range of customers we consider the following items essential in a published evaluation: (i) a comprehensive bibliography (ii) thermodynamic data for the reaction (iii) sufficient information on individual rate measurements to allow correlation of results with the experimental methods and conditions (iv) a record of the measured rate constants and a visual display of the data, usually an Arrhenius plot (v) wherever possible a recommended rate expression, prominently displayed, conditions of temperature (and pressure) over which it is valid, and realistic error limits reflecting the evaluators confidence in the recommendations rather than the scatter of the experimental points (vi) a brief discussion justifying the recommendations It is useful also to record reviews, other evaluations, theoretical predictions, papers on peripheral but relevant work, and major papers in which particular values for a rate constant have been used. Thoroughly done such work should allow future evaluators to build on it without the 0 1979 American Chemical Society
Evaluation of Kinetic Date
The Journal of Physical Chemistry, Vol. 83,
No. 1, 1979 43
need for extensive consultation of the earlier literature. closely. We therefore discuss some of these comparisons in more detail if only to emphasize how crude such criteria T h e Literature Quantity and Quality. The basis of a are a t present. successful evaluation is an adequate bibliography. A number of useful compilations and reviews are a ~ a i l a b l e , ~ , ~ Comparison with Other Measurements. Comparison the work of earlier evaluators can be consulted and, most with other measurements, using the same or different valuable of all, the NBS collection of papers can be methods, may reveal differences well outside the quoted error limits. Quoted error limits are usually derived from searched. These sources are of most value for retrospective searching but since the status of a reaction, or group of the extent of scakter of the primary data which is then reactions, can change dramatically in a time which is averaged to give the finally quoted value for the measured comparable with publication times, some systematic quantity. They are thus estimates of the precision of the work, and the differences revealed by comparison with scanning of the current literature is essential and preprints from authors may be particularly helpful. other work reflect differences in accuracy, Le., systematic errors undetected by the experimenters. In deciding which As well as papers containing new measurements of of the measuremlents is to be preferred it is essential that gas-phase reaction rate constants it is necessary to collect the decision be based, as far as possible, on scientific related material from such fields as spectroscopy, thercriteria. The evaluator must look again a t the methods modynamics, theoretical chemistry, and liquid phase kithat have been wed, whether they have been successfully netics. If the whole field of gas kinetics of neutral species applied to closely related systems, and scrutinize the is to be covered then in our experience this amounts to chemistry of the system more closely. However, even then collecting and filing approximately 1000 articles per anit is sometimes difficult to find good reasons for accepting num, perhaps 50% of which contain new rate data. Thus certain data and rejecting others. Where there are only assembling bibliographies is riot a trivial task. perhaps two or three measurements to choose from, all The information which a n evaluator requires from a apparently well based scientifically, but differing subpaper reporting new data depends on the nature of the stantially, the evaluator will probably make no recomevaluation task but generally it is desirable that (i) there mendation, or recommend some value within the range, should be sufficient experimental detail to allow the quality but with wide error limits. Where there is good agreement of the work to be evaluated, (ii) the results should be between a number of measurements with only one or two reported in siifficient detail for them to be reanalyzed and well outside the quoted error limits the evaluator will reinterpreted by others, (iii) there should be an estimate probably accept the “majority verdict” but may temper of the reproducibilit,y and, if possible, the accuracy of the this judgement with an increase on the error limits in his result. These three principles are the basis of a set of recommendation. detailed guidelines, formulated by the CODATA Task Modern developments in detection and characterization Group,4 for the reporting of experimentally determined of gaseous atomrs and radicals now permit much lower numerical data. They are aimed a t authors and referees concentrations of these species to be detected, and even who bear the responsibility of ensuring that worthwhile very rapid reactions can now be followed under unamresults are not devalued by inadequate presentation. The biguously pseudo-first-order conditions with such methods referee is in many respects an evaluator and in this role as flash-photolysis and fast-flow discharge systems. This can have a substantial influence on the durability of reshould be reflected in a considerable improvement in ported data. accuracy of results for those reactions over the regions of Unfortunately the publishing policy of journals is often temperature and pressure accessible to these methods. For in conflict with the needs of the evaluator. For example fast bimolecular gas-phase reactions over even a relatively presentation of rate constant measurements in the form small temperature range (a few hundred degrees) the of a small Arrhenius diagram or as an Arrhenius expression evaluator has hitherto rarely been able to define rate are inadequate substitutes for tables of results which show constants to better than A2570 but with these new exthe experimental scatter and the conditions under which perimental developments it should be possible to improve each experiment was performed. This kind of problem can significantly on this, a t temperatures close to ambient a t be alleviated if authors will put their detailed results in least. It will be interesting to see whether this promise is national depositories which are now available. fulfilled. Procedures. The factors which the evaluator considers Agreement wkh other measurements under the same in assessing published material have been outlined by conditions is probably the prime criterion used in practice Hampson and Garvin.2 These procedures fall into two by the evaluator but there are a t least two other comcategories, first the assessment of the technique and data parisons commonly used, namely, compatability with (i) processing used by the experimenter and, secondly, thermodynamic data and (ii) values of rate constants a t comparison of the results with material external to the other temperatures. study, e.g., results of others, theoretical predictions, compatability with thermodynamic data. Thermodynamic Data. Such data may be used in several ways including setting limits to activation energies In examining the experimental methods the details listed (and hence rate constants) for decomposition reactions, in the CODATA guidelines are needed. The experience trying to decide between possible channels for reactions of the evaluator both in kinetics and in evaluation plays capable of yielding a variety of products, and using the an important role a t this stage, but a certain amount of familiar relationship K , = k f / k , to calculate one of the help is available from the literature, for example, quantities in thk formula knowing the other two. The last Cvetanovie et al. have discussed the accuracy of gas-phase of these is potentially the most useful; the relationship kinetics measurements and error e ~ t i m a t i o nthere , ~ are appears to be valid even up to fairly high temperatures brief but useful reviews of techniques,6 and the critical (-2000 K). assessment of other authors working on the same or a closely related problem is most valuable. However it is the However there are problems in using thermodynamic comparison with other results etc. which usually reveals data foremost among which are their limited accuracy. For any deficiency in a technique, and which forces the gas-phase reactions which have been studied over a wide evaluator to return to the method and scrutinize it more temperature range the best gas-phase kinetics data ap-
44
The Journal of Physical Chemistry, Vol. 83, No. I , 1979
D. L. Baulch and D. C. Montague
proach an accuracy of f3070 and for a number of reactions, over a restricted temperature range, the error limits are much less than this. If equilibrium constants are to be used at this level of error then free energy changes for the reactions need to be known to approximately 1 kJ at ambient temperature. For atoms, many stable compounds, and a number of diatomic radicals free energies of formation are known to the required accuracy, but this is not 'so for the remaining majority of chemical species, particularly polyatomic radicals. The best semiempirical methods for estimating thermodynamic data cannot approach this level even in favorable cases. Nevertheless thermodynamic data which allow no better than an order of magnitude estimate of rate constants are often useful to the evaluator in setting "reasonable" limits to possible values for rate constants. In favorable cases the accuracy of the thermodynamic data is much superior to that from kinetics and the relationship K , = h f / k , may be used (i) to check the consistency of independently measured values of hf and k,, and (ii) to calculate kf or k , knowing the other. Its value in checking existing data is exemplified in its recent application to the system H HCl * H2 + C1
Figure 1. Arrhenius plot for the reaction CH, f HO
where the lack of compatability of measured values of kf and k , with the known K , has led to a closer examination of the experimental methods used and has revealed that early measurements of the rate of the forward reaction were in error due to reactions a t the walls removing hydrogen atoms.6B7 Moreover even if both the kinetic and thermodynamic data are accurately known a common difficulty which still arises is the fact that the kinetics data on hf and k, often cover different temperature ranges. Worst in this respect are reactions for which the forward and reverse reactions have very different activation energies so that their rates come into the time range of existing experimental methods at different temperatures, e.g., dissociation/recombination reactions. Similarly in generating values of, e.g., h,, from known values hf the temperature range for h, will be limited to that for which h, is known. Often this is not the range of most practical value. Temperature Dependence of Rate Constants. The problems involved in extrapolating and interpolating rate measurements with respect to temperature arise from inadequacies in both theory and existing data. Experimental rate measurements on a particular reaction often cover two distinct temperature ranges: (i) the 300-600-K region where photolytic and discharge-flow techniques are commonly used, and (ii) the range accessible to shock tube measurements, often >1500 K. As well, for some reactions data from flame studies are available a t -1000 K, but often there is little intermediate data to assist the evaluator in deciding how to relate the high temperature and low temperature results. It is therefore important that existing high and low temperature techniques should be "stretched" to cover the whole temperature range. It is usual to use an Arrhenius expression k = A exp( B I T )to bridge the gap, despite the fact that the high and low temperature sets of results often appear to show significantly different values of the constants A and B. Indeed an Arrhenius expression that passes through both the high and low temperature data has sometimes been used t o judge the validity of results a t intermediate temperatures. Such a procedure has obvious dangers since there are now experimental data for a number of bimolecular reactions which are better fitted by a non-Arrhenius
form.' An example is shown in Figure 1. Curvature is most marked a t high temperatures, corresponding to increased importance of the preexponential term. This poses difficulties in selecting rate constants for modeling combustion systems operating in these high temperature regions, since the experimental data often show the greatest scatter at high temperatures, and a criterion frequently used for selecting the best of the high temperature data is compatibility with the usually more accurate low temperature data via an Arrhenius expression. This points to the need for more precise experimental work a t high temperatures and better theoretical guidance. I t also becomes incumbent on the evaluator to give more regard to high temperature measurements, not giving undue weight to low temperature results as has often been the case in the past. However clear evidence for curvature in these plots is the exception rather than the rule. For most bimolecular reactions the available data can be represented within experimental error by a simple Arrhenius expression. This is probably as much a reflection on the quality and quantity of the data as on the form of temperature dependence, since there are very few reactions for which there are accurate data over a sufficiently wide temperature range (-2000 deg) that show curvature unambiguously. It is mainly for this reason that the evaluator will continue to use the simplest (Arrhenius) form fitting the data. The theory of the temperature dependence of decomposition/combination reactions is better developed and can often be usefully used by the evaluator to assess the quality of the existing data. Frequently results are available for reactions studied at both the high and low pressure limits. In the first-order region experimental activation energies, E", can be checked against the reaction thermochemistry. Agreement is usually good for well understood systems, although E" is sometimes found to be slightly less than the endothermicity. Theories have been put forward to explain this for diatomic molecule^,^ and sometimes it merely reflects the fact that the rate constant measurements have been made a t pressures below the high pressure limit. Experimental preexponential factors are
+
TI
0.0
10
K
20
30
lo3 T-' I K-'
I
-
CH, 4- H,O. Points and dashed lines are experimental results. The solid line is our evaluated fit to the data.
Evaluation of Kinetic Data
less easily checked, although methods such as the transition state approach of Benson and co-workers1° provide theoretical estimates that often can act as a useful guide. In the second-order region the comparison of activation energies with the thermochemistry is not as direct and must involve theory. The evaluator has to decide whether the value of (E" - Eo) is reasonable, or, since E" values are not always available, whether the difference between Eo and the endothermicity is acceptable. A commonly used guide in dealing with this problem is the relationship Eo N Eo - (s - 1)RT,whlere Eo is the threshold energy for the reaction =Moo, and s is the Kassel parameter given by s = U,/RT (where U, is the energy in the vibrational modes). Nonzero values of (s - 1) imply a temperature dependent activation energy and hence a curved Arrhenius plot.1' For a number of reactions of this type the data conform to these expressions reasonably well. In other cases the data seem widely a t variance with theory and can be rejected, but there itr also a third more difficult category, where the departurle from theory is significant but not sufficient to reject the data with confidence. In no instance are there data covering a sufficiently wide temperature range to test adequately the predicted curvature of the Arrhenius plot. Cuirvature is most marked a t high temperatures under which conditions the reaction rate is becoming very fast and complications from competitive or consequential reactions more likely, all leading to less precision and greater difficulty in performing the measurements. Additional uncertainty can also be introduced by the effects of weak collisions, leading to observed rate constants that are lower than the "true" (strong collision) bimolecular limit rate constants. It is often particularly difficult to assess the magnitude of such effects and they have therefore been omitted from consideration for the present from the following examples, where we illustrate the application of the theoretical expressions outlined above to specific decomposition reactions, to show their value to the evaluator, and some of their limitations. (i) Nitric Acid. An evaluation of the rate data available in 1973 for nitric acid decomposition suggested a value of E o / R = 15400K a t 1000 K.12 The difference between this value and AHoo/R== 24000K required s = 9.6, a value in excess of the total number of vibrational oscillators. The data therefore seemed to be a t fault, a conclusion subsequently confirmed by more recent measurement~l~ giving E o / R = 23000K corresponding to s = 3, which is close enough to the value of U,/R = 4 to be acceptable. (ii) Hydrazine. For hydrazine decomposition we have suggested a value of Eo/R = 20600K based on data ranging around 1300 K.12 Combined with AHoo/R = 27900K this gives s I- 6.5 in good agreement with U,/RT = 7. This is also one of the few examples where the high temperature values for the rate constants appear to be lower than those predicted by linear extrapolation of the low temperature data on an Arrhenius plot. However a curved In h vs. 1/T plot based on s = 7 does not adequately €it both low and high temperature data, a much higher value of s being required (see Figure 2). In the present state of theory it is difficult for the evaluator to know whether this difference is significant, and that would still be the case even if a more detailed calculation were to show that a discrepancy of this magnitude still existed. (iii) Amn'bonia. The data of Michel and Wagner,14 apparently reliable when judged purely in terms of experimental technique, gives E o / R = 40000K a t T = 2500 K, whereas AHoo/R = 51200. This difference requires s = 5.5, a number approaching the total number of fun-
The Journal of Physical Chemistry, Vol. 83, No. 1, 1979
45
10,ot-
7 ; l 9.0 o r 0
E
m
E
," 8 0 0
Y I
-
m
7,o -
60-
5.0 -
LO 0
I
I
05
10
lo3
T-'/ K-'
-
Figure 2. Arrhenius plot for the reaction NH , , -t M 2NH, -t M. The solid line is based on s = 7. Points and dashed lines are experimental results.
damental vibrational modes in the molecule, whereas U,/RT = 3. Here the absolute difference between the two values of s is not very great but relative to the total number of modes it is large and, for this reason, it seems likely that the experimental values are in error probably because of the participation of rapid secondary reactions following the relatively slow initial decomposition step. Later work by Henrici,15 in which an induction period was detected, provided further evidence of this complexity, and yielded a v,slue of E o / R = 45800K and hence s = 3. However, these results are only available in thesis form and the evaluator is reluctant to adopt them without further experimental confirmation or refereeing of the work, nor is there sufficient confidence in theory to base predictions of E o / R solely on it. (iv) N2F4and N02C2.16The decomposition of both of these molecules has been studied a t relatively low temperatures and in both cases large values of (AH," -Eo)/R have been observed. For N2F4decomposition below 500 K and N02Cl below 1000 K it is necessary to invoke approximately 10 and 7 effective oscillators respectively to explain such large (moo - E o ) / Rvalues, whereas values of U,/RT indicate values of approximately 3 and 2.5. The thermochemistry of both systems is not very precise, but it appears that here theory indicates unambiguously that the experimental data are suspect. In the case of NOzCl it seems also that the decomposition a t high pressures (up to 300 atm) is more complex than has been appreciated, since the present data give a value for E" identical with the value for Eo! Theory as a Guide Despite the imprecision of current theory the evaluator must take some account of it and, as can be seen from the examples, usually does so in two ways. (i) It can be used as a guide to "reasonable" limits for rate parameters: for a particular reaction. Simple rules of
46
N. Cohen and K.
The Journal of Physical Chemistry, Vol. 83, No. 1, 1979
the type used in the earlier examples are of great value. However they cannot be taken too far. The theories are often semiempirical relating known rate data to structure of reactants and thence extrapolating, via structural considerations, to give rate constants of related reactions. It is difficult to assess how closely the related reactions must be for these extrapolations to remain valid, and inevitably there are borderline cases which can only be resolved by further experimental work. Even where the theory indicates strongly that the experimental results are in error, the evaluator is reluctant to adopt as an alternative purely theoretical values, which generally he is unlikely to trust to better than an order of magnitude. (ii) Theory is also used as a means of extending existing data to regions of temperature and pressure in which data are sparse. Strictly this falls outside the usual terms of reference of the .evaluator but often such data are needed and the evaluator can help to supply them. Some of the difficulties in such extrapolation with respect to temperature are evident from the examples given earlier. Where a linear relationship between In h and 1/T can be assumed, good data in any part of the temperature range can be used as a basis. However the recently found curvature in such plots for bimolecular reactions and the known, but not well characterized, curvature for low pressure unimolecular reactions now presents substantial problems for the evaluator in deriving data for high temperatures. Theoretical guidance is urgently needed, but any theories must be thoroughly tested before they can be accepted, and currently there are insufficient good quality data to do that. For this reason the emphasis still must be on good experimentation to give rate data over a wide range of conditions. Finally we draw attention to one other important area in which the evaluator and experimentalist needs assistance from the theoretician. The form in which theory is presented is often difficult and very time consuming for the evaluator to understand. The growth in complexity and variety of the language of theory presents a very real
Westberg
and increasing barrier to the evaluator which few of us have the energy to surmount.
References and Notes (1) CIAP Monograph 1, "The Natural Stratosphere of 1974", Department of Transportation DOT-TST-75-51, Sept 1975; R. F. Hampson and D. Garvin, Natl. Bur. Stand. Tech. Note, No. 866 (1975). (2) R. F. Hampson and D. Garvin, J . Phys. Chem., 81, 2317 (1977). (3) "A Catalog of Compilation and Data Evaluation Activities in Chemical Kinetics, Photochemistry and Radiation Chemistry", CODA TA Bull., No. 3 (1971). Report of the CODATA Task Group on Data for Chemical Kinetics. (4) "The Presentation of Chemical Kinetics Data in the Primary Lierature", CODATA Bull., No. 13 (1974). Report of the CODATA Task Group on Data for Chemical Kinetics. (5) R. J. Cvetanovie, R. P. Overend, and G. Paraskevopoulos, Int. J . Chem. Kinet., Suppl. 7, 249 (1975): R. J. Cvetanovie and D. L. Singleton, Int. J . Chem. Kinet., 9, 481, 1007 (1977). (6) R. T. Watson, J . Phys. Chem. Ref. Data, 6, 871 (1977). (7) J. E. Spencer and 6. P. Glass, J . Phys. Chem., 79, 2329 (1975). (8) W. C. Gardiner, Acc. Chem. Res., IO, 326 (1977). (9) D. G. Truhlar and N. C. Blais, J . Am. Chem. Soc., 99, 8108 (1977). (10) S. W. Benson, "Thermochemical Kinetics", 2nd ed, Wiley, New York, 1976. (11) J. Troe, Symp. (Int.) Combust., [Proc.], 15th, 1974, 667 (1975). (12) D. L. Baulch, D. D. Drysdale, D. G. Horne, and A. C. Lloyd, "Evaluated Kinetic Data for High Temperature Reactions", Vol. 2, "Homogeneous Gas Phase Reactions of the Hz-Nz-Oz System", Butterworths, London, 1973. (13) K. Gianzer and J. T r k , Ber. Bunsenges. Phys. Chem., 78, 71 (1974). (14) K. W. Michel and H. Gg. Wagner, Symp. (Int.) Combust., [Proc.], loth, 1964, 353 (1965). (15) H. Henrici, PhD. Thesis, Gottingen, 1966. (16) D. L. Baulch, J. Duxbury, S. Grant, and D. C. Montague, "Evaluated Kinetic Data for High Temperature Reactions", Vol. 4, "Homogeneous Gas Phase Reactions of Halogenated Species", to be published as a supplement of J . Phys. Chem. Ref. Data, in press: M. L. Dutton, D. L. Bunker, and H. H. Harris, J . Phys. Chem., 76, 2614 (1972).
Discussion H. M. FREY(University of Reading). I agree that good referencing is essential if evaluators of rate data are to do a good job. However journal policy often makes it difficult to ensure that enough data is presented in the paper. One should always insist that whether or not a graph (Arrhenius plot etc.) is presented, the rate constants on which it is based should be
reported.
Evaluation and Compilation of Chemical Kinetic Data' Norman Cohen" and Karl Westberg Aerophysics Laboratory, The Ivan A. Gefting Laboratories, The Aerospace Corporation, El Segundo, California 90245 (Received Ju/y 24, 1978) Publication costs assisted by The Aerospace Corporation
A new program for the critical evaluation of chemical kinetic data and for the compilation of data sheets is described. A sample data sheet is given. Its format is still somewhat preliminary and will be modified if there is a consensus for doing so. These data sheets are designed for use by nonkineticists as well as kineticists. The initial efforts in this program have led to some general conclusions concerning the information content in journal articles, the usefulness of certain kinetic techniques, and the meaningfulness of rate data obtained in some studies. The field of chemical kinetics has matured rapidly in the past decade or two. The growing body of data and derived rate coefficients that have been published in journals, technical reports, and symposia has generated a need for bibliographies and reviews to aid the practicing kineticist and also the engineers who are not kineticists but need to use kinetic data. Kinetic data compilations have been published in several different formats: (1) bibliographies of references per0022-3654/79/2083-0046$01 .OO/O
taining to a given reaction; (2) noncritical tables of published rate coefficients; (3) critical reviews in tabular or data sheet format; and (4)in-depth critical monographic reviews of individual reactions or groups of reactions. For the experienced kineticist, all four of these formats can be useful in different cases. For the engineer and computer modeler, only the third type is useful, and even then most of the published compilations leave something to be desired. 0 1979
American Chemical Society
