Journal of Physical and Chemical Reference Data David R. Lide, Jr, Editor ‘The Journal of Physicaland Chemical Reference Data 's published quartery by the American Chemical Society and the American insttute of Physi forthe National Bureau of Standards. The objective of ine Journal isto provide ciicaly evaluated physical and ‘chemical property dat, fly documented as tothe original sources and the etter used for evaluation, Citeal reviews of measurement techniques, whose ‘aim Is 19 assess the accuracy of avallabie dala In & ‘iven technical aroa, are also Included. One of tho Principal sources forthe Journal isthe National Standard Reterence Data System (NSROS), which Sescribed more fuly below. Tha Journal notin tended as a publication outet fr orginal experimental measurements such as are normally reported tn the primary research terature,nortorrovow atiles of ' descriptive of primatly theoretical nature, ‘Supplements tothe Journal are publshed at requ: larintervals and are not Included in subscriptions to the Journal. They contain compilations which are too lengthy fora journal format. ‘The Editor welcomes appropriate contributions from indviguas or groups for consiceration by the Eaton Board, Potential contributors are invited to subret an Cutie ofthe nature and scope ofthe proposed omplation, wih cera for evaluation of the data and other pertinent factors to: Davia A Lido, Jr, Etor ‘Phys. Cham Ret, Data ‘National Bureau of Standards Washington, 0.C. 20284 ‘The National Standard Reference Data System was estabienedin 1963 asa means of coordinating ‘ona national sealethe production andaissemination ‘ferticaly evaluated reference datainthe physical Sciences. Under the Standard Reference Data Act (Public Law 90-396) the National Bureau of Standards ofthe US. Depanment of Commerce has the primary ‘esponaolty in the Federal Goverment fo providing Fellable scenic and tectricalretorence data. The Ofice of Standard Rsterence Data of NBS coord natos a complexof data evaluation contra, located Inuniverety,industial. and gther Government abor: Aories as weilas within the National Bureau o! Stan ards, whicn are engagedin the complaion and crtical {evaluation of numercal data on physical and chemical Properties retrieved from the word scenic itera tre. The panicipants in ths NAS-sponsored program together with similar groups under pivato or oor Government support which are pursuing the same fends, comprise the National Standard Feterence Bata’ Systom, ‘The primary focus of the NSRDS is on well-defined physical and chemical properties of wel-characterized ‘atorals or systems. An effort is made to assess the Accuracy of data reported inthe primary research erature and to prepare compliations of cicaly evaluated data which wil serve a reliable and con Yeniot reference sources for the scientiie and {echnical community. Editorial Board Term ending 31 December 1974 John B. Wachtman, Je, S.A. Goudsmit, Daniel R. Stull Teem ending 31 December 1975 Howard J. White, J., Richard K, Cook, Everett R. Johnson Term ending 31 December 1976 David Garvin, Gilbert J. Perlow, Robert L. Pigford ‘Management Board David R. Lide, Jr, Richard L. Kenyon, Robert H. Marks Eéiroval Staff at NBS: Herman M. Welsman, Julian M. Ives, Betty L. Oberholtzer Production Staff at AIP: Mitchell Koch, Managing Editor Information for Contributors All mathematical expressions and formulas, insotar as possible, should be typed, ‘wth unavalable Greek letters o” symbols insertod realy in ink. Distinguish between Capitals and smal laters. Avoid complicated superscripts and subscripts, Avo! repetition of a complicated expression by representing by a symbol. ientty the margin handuriten Greek letters and unusual symbols Use laetional exponents Instead of root signs, and use the solids () for fractions whereever is use wil ‘save vertical space Literature references are to be numbered consecutively in brackets. They are to be placed at the end of the paper. A footnote appears afer the frst reference staling: Figures in brackets inaate erature reforences at the end of ths paper Literature tations of technical journals ehoula be Inthe lorm indicated inthe Ste Manual ofthe Amencan Insitute of Physics (once $250 prepa from American Institute of Physics, Dept. BN, 335 East 45 St, New York, NY. 10017). ‘Tables may bo typewritten on sheels ftom the running text. Each table must have «2 caption that wil make the data In ho table ineligible without reference 10 the ‘ext Avoid complicated column headings. Figures should, whenever practicable, be planned for reproduction in one-column wiath (8.7 om). Line drawings should be india ink on whe paper or tracing cath ‘The orignal crawing ora high-qualty glossy print should be submitted. To facta mailing and handling, orginal swings should not exceed 22 by 28 om in size, Letiorng, preferably srawn with a mechanical set, should be of a size so that vwinen reduced the smaliostlowor-cago letters wil not be less than about 1.6mm ‘Avoid gross dispaiies in lettering size on a drawing. Consul the Style Manual of the American insite of Physies for futher discussion of figures. Type figure Captions on a separate sheet of paper. Number figures consecutively, Identy each ‘tiginal drawing or photograph withthe figure number and authore names, writen where they wil not be reproduced, Subscriptions, renewals, address changes, single-copy and reprint orders, land orders for supplements shouldbe addressed to: Subscriation Service Depart ‘mont. American Chemical Socly, 1159 Siteonth Sireet, NW, Washington, D.C £20056. Members of AlP momber andatliatedsacioties requesting Member subscrp tion rates should renew subscriations and direct orders to American Insitute of Physics, Dept. SIF, 335 East 45 Street, New York, NY. 10017. Allow at toast ‘8k weeks advance notice. For address changes please send both old and now addresses, and, if possible, inciudo an adorese-stencl imprint frm the maling ‘wrapper of a recent Issue. Prices for reprints and supplemonts are tated an the last page. Subscription Prices. U.S.A, Canada, (vot inc. supplements) ‘and Mexico Etsewnere Members (afiated and s20 $23 member eocilie) All oters $60 $63 ‘The Journal of Physical and Chemical Reference Data appears in Sect. 1 of Currant Physics Microform (CPM) along with 18 olher primary journals pubished by the ‘American Insitute of Physics. CPM Sect, 11s avaiable on an annual subsergtion for $1600. Copyright and Permission to Copy though this Journal ls copyeghisg, ihe making of a single copy of an article by 8 cients for his own use is considored “air use." Permission fs granted to quote from this journal in scene articles and to reproduce tgures and tables, with proper Cation ofthe source, provided the authors consent Is ablainea and noliteaton Is sont tothe Publications Division of American Insitute of Prysies. Reproduction of whole articles for commercial purposes of in mule copies requires consent ‘ofthe author and specitc permission from the Publications Divison of American Insitute of Physics. The right of the U.S. Government to umestricted copying for its own use of copyrighted matonal onginaling ns laboralories or under ts cntracts is speciicaly recognizedJournal of Physical and Chemical Reference Data Volume 3, 1974 Supplement No. 1 Thermal Conductivity of the Elements: A Comprehensive Review C. Y. Ho R. W. Powell P. E. Liley Thermophysical Properties Research Center Purdue University West Lafayette, Indiana 47906 o Published by the American Chemical Society and the American Institute of Physics for the National Bureau of StandardsCopyright © 1875 by the U.S. Secretary of Commerce on behalf ofthe United States. This copyright will be assigned to the American Institute of Physics and the American Chemical Society, t0 whom a requests garding reproduction should be addressed. Library of Congress Catalog Card Number 75-4440 International Standard Book Numbers 0-88318.2165 (Paper cover) ‘American Institute of Physics, Inc. 335 East 45 Stra Now York, New York 10017 Printed in the United States of AmericaForeword ‘The Journal of Physical and Chemical Reference Data is published jointly by the ‘American Insitute of Physics and the American Chemical Society for the National Bureau of Standards. ts objective is to provide critically evaluated physieal and chemical property data, fully documented to the original sources and the eriteria sed for evaluation. One of the principal sources of material for the journal isthe National Standard Reference Data System (NSRDS), 2 program coordinated by NBS for the purpose of promoting the compilation and critical evaluation of property data. The regular isues of the Journal of Physical and Chemical Reference Data are published quarterly and contain compilations and critical data reviews of moderate Tength. Longer monographs, volumes of collected tables, and other material unsuited to a periodical format are published separately as Supplements to the Journal. This monograph, “Thermal Conductivity of the Elements: A Compre- hensive Review,” by C. ¥. Ho, RW. Powell, and P. B. Lie, is presented as Supplement No. 1 to Volume 3 of the Journal of Physical and Chemical Reference Data Dovid R, Lide, Jr Editor Journal of Physical and Chemical Reference Data 4. Phys. Chom, Raf. Date, Vel 3, Suppl. 1, 1974 mThermal conductivity of the element: a comprehensive review C. Y. Ho, R. W. Powell, and P. E. Liley Thermophysical Properties Research Center, Purdue University, West Lafayette, Indiana 47906 ‘This work presents and discusses the availeble data and information on the thermal conductivity of the elements and contains the recommended reference values resulting from critical evaluation, analysis and synthesis of the fvailable ‘ata and information. It also gives estimated values, at least for normal tempers ture, for all those elements for which no thermal conductivity information is available. Experimental thermal conductivity data ate available in the world Iiterature for 82 clements and estimated values for four other elements. Estimated values for the remaining 19 elements are given here, although only rough estimates are given for the trans-plutonium elements. Thus, this work provides recommended oF estimated thermal conductivity values for all the 10S elements ‘This work is published in two companion versions: this comprehensive volume and an abridged version. In addition 10 the recommended and estimated thermal conductivity values for elements, the comprehensive volume presents the original data, specimen characterization, and measurement information for the 5200 sets of raw data compiled; contains a detailed discussion for every element reviewing the individoal pieces of available data and information together with the considerations involved in arriving at the final assessment and recommendations, land with the theoretical guidelines or semi-empirical correlations on which the critical evaluation, analysis, and synthesis are based; and includes also the complete bibliographic citations for the 1658 references. The abridged version contains oxy the recommended and estimated thermal conductivity values. Key words: Conductivity; critical evaluation; data analysis; data compil: data synthesis; elements; gases; liquid metals liquids, metallic elements; metals molten metals; most probable values; nonmetallic elements; recommended refer cence values; reference data; solids; standard reference data; thermal conductivity. 4. Phys. Chem. Ref Dato, Vol. 3, Suppl. 1, 1974 wWPreface 4. Phys. Chem. Ref, Date, Vol. 3, Suppl 1, 1974 Thermal Conductivity of the Elements: A Comprehensive Review is the result ‘of many years of effort by the Thermophysical Properties Reseach Center (TPRC) 188 part of an overall program to generate tables of numerical data for science and technology. The property, thermal conductivity, was selected as a priority task because ofits sientific and technical importance and because of TPRC's extensive bibliographic coverage of the literature an this property “This volume can serve many purposes. It provides engineering and design data for those elements such as tungsten, molybdenum, carbon (graphite), aluminum, copper, germanium, silicon, liquid sodium, mercury, etc. which are used in nearly ppure form as engineering materials. It provides reliable data for those elements that ean be used as reference materials to check apparatus for thermal conductivity measurements or as standards in comparative thermal conductivity measurements. 1 provides data agains: which theoreticians can test their theories. Furthermore, the knowledge of the thermal conductivity of the elements is essential for the estimation end prediction of this property for the more complex engineering. alloys. Since precise measurement of thermal conductivity is very difficult, a capability for the estimation and prediction of this property within confidence levels acceptable to technological requirements would be very valuable. A knowl- ‘edge of the thermal conductivity of the elements is an essential requirement for the development of such a capability Thermal Conductivity of the Elements: A Comprehensive Review has been pub- lished in two companion versions. This comprehensive volume makes it possible for serious students of the subject 10 have access 10 the original data without having to duplicate the laborious and costly process of literature search and data extraction, It is quite appropriate at this point to mention that only original sources have been used for the critique of the data and that all cited documents fare available at TPRC in standard microfiche format, Also, for the active researchers in the field, a detailed discussion is presented for each element reviewing the available experimental data and the considerations by which the authors arrived at their inal assessment and recommended values. Since the comprebensive version is voluminous and perhaps somewhat cumber- some for frequent use as a reference, an abridged version has been published (Journal of Physical and Chemical Reference Data, 1, 279, 1972) which gives the recommended values with minimal discussion, Reprints of the abridged version should provide inexpensive, personal, desk-top reference sources for designers, ‘engineers, students and scientists who have a need for ready access to these data. ‘The input of data to Thermal Conductivity of the Elements: A Comprehensive Review has a cut-off date of January 1971; works published subsequent to this date have not been considered. However, TPRC monitors and retrieves the world feature on a continuing basis and our state of knowledge on the thermal con- acconeeneoner zo = & Wy/4*Auonpu09 youuoys, peonpay @ 3 80 o B14 4. Phys. Cham, Raf Dote, Vol. 3, Sepp. 1, 1974THERMAL CONDUCTIVITY OF THE ELEMENTS In this work, at low temperatures only one recommended curve for one particular sample has been generated and this usually relates to the lowest value of for the purest sample for which a thermal conductivity measurement has been made. For generating other curves for other samples ‘equations (7) and (8) and the recommended constants of table 1 may be used. It often happens that electrical resistivity investigations have included purer samples yield- ing much lower values for , but to use these values seems unwise at present as some doubt exists as to the validity of this simple treatment for samples of much greater purity, especially for transition metals. There is some evidence [600, 1193, 1275] that clectron-clectron scattering may hecome important for exceedingly high: purity samples and necessitate one additional term »T such that k= [ol 4p $77}. (20) Also, for many of the elements available data are in- sufficient to determine the constants m,n, and @” of equation (8). Tante 1. Constants for low-temperature thermal conductivity calelations wsing equations (7) and (8) Element mie a"x10" Aluminum 2.62 0.0879 Cadmium (to caxis) 5,00 0.0168, (to caxis) 5.00 0.0968 Apolyerystalline) 5.00 0-068 Chromium 2.20 0.592 Cobate 2.20 0.540 Copper 2.68 0.0825 Galtaen (| to eaxs) 2.78 | 2.00 | 208 (10 bass) 2.78 | 2.00 | 0.806 (1 to caxis) 27a | 200 | 6.37 Gold 2.46 | 2.00 | 0.460 Indium 3.00 | 2.00 | 3.50 idium 440 | 3.00 | 0.000272 Tron 2.20 | 200 | ost Lead 350 | 3.00 | 4.2 Lithium 22s | 2.00 | 0.77 Magnesia 210 | 200 | 0.627 Molybdenum | 320 | 2.60 | 0.00967 Nickel 260 | 2.00 | 0.192 Niobium 2.00 | 2.00 } 6.21 Osmium 5.80 | 3.00 | 000000379 Palladium 240] 2.00 | 1s Potassium 210 | 200 | 18.0 Rhenium 3.30 | 2.20 | 0.0656 Rhodium 3.00 | 2.80 | 00132 Rathenfum 5.60 | 2.60 | 0-co0onser Silver 2.75 | 2.20 | 0.0730 Sodium 213 | 200 | 289 Tantalum 24 | 2.00 | 139 Thalia 2:60 | 2.00 | 262 Thorium zo | 27 | 17 Titanium 200 | 230 | 0.188 Tungsten 2.80 | 240 | 0.0539 Zine 3.40 | 3.00 | 0.0750 Zirconium 2.40 | 200 | 99 \ us A further complication may arise with metallic samples of very high purity in that boundary scattering can be- ‘come important and render the thermal conductivity at very low temperatures dependent on the size of sample or. on that of the individual erystallites of which it posed. Since the 1930's, see for instance the work of Casimir [243], size dependence has been known for the thermal conductivity of nonmetallic erystals, but measure- ments by Olsen and Wyder [1053] and by Boughton and Yaqub (185] have more recently directed attention to the influence of exystal size on the electronic thermal con ductivity of a metal of sufficiently high purity and perfec tion, Isotopic content is another factor that has been, shown to influence the thermal conductivity at low tem= peratures; see for instance the work on an isotopically enriched germanium by Geballe and Hull [480], on tellurium by Oskotskii et al. [1057], and on solid helium by Berman, etal. (141] (see also (148]) As the temperature rises from the liquid-helium tem. perature region, the value of the Lorenz function falls quite appreciably to a minimum, but near the Debye temperature it again tends asymptotically towards the ‘theoretical value (see, eg. Wilson [1569], Makinson [875]). For some metals, including the transition metals, definitely higher values of the Lorenz function may be altained, but the excess seldom exceeds about 30 percent. It follows that in the region from about normal to high temperatures the Lorenz function is generally reasonably close to the theoretical value, and for a particular metal follows a fairly predictable departure curve, Thermal conductivity values can then be calculated from the derived, assumed, or experimentally determined Lorenz function values as « function of temperature and from the measured electrical resistivity data. Considerable use of the Lorenz relationship has therefore been made in this work, both when analyzing thermal conductivity data in the above-normal temperature region and when attempting to make estimations or extrapolations in this range. For elements such as gallium and yttrium, whose trans. port properties are strongly anisotropic, uncertainties are associated with the derivation of values from single crystal data that would apply to a polycrystalline sample. Consider an orthorhombic crystal, such as that oi gallium, for which ke, ky» and ky are the thermal con ductivity values for the three main crystal axes a, b, ane , and hy is the thermal conductivity of the polyerystal By considering the conductivities to be additive, Voig [1487] showed that 1 k= (he thy the). ay If however the thermal resistivities are considered to by additive, which Hall, Legvold, and Spedding [577] re garded to be preferable in the cate of rods of yttrium, thet 1 li. 1 1 Sa + 4) by Bk by ke az 4. Phys. Chom. Re Dote, Vol. 3, Suppl 1, 197.+6 es 6 Tp hke + hk a For gallium at 300 K, kz = 0.406, ky = 0.883, and k, = 0.159 W emt K-!, Hence the values of fy according to equations (11) and (12) are respectively 0.483 and 0.308 W emt K+, and differ by some 25 percent from the mean valve of 0.398 W enc K-*, A more recent treat ment, in which Hashin and Shtrikman [592] used a varia- ional method, shows that for the case where hy < ky < hy hy (Ake? + Bkcka + Bkeky + ) TOE! -} Skcks + Sky} haky ? 14) which leads to extreme values of O44 and 0.377 W ‘em K~" for hy in the case of gallium at 300 K. The treat- iment embraces a narrower (++8%) range of values and ‘gives a mean of 0.410 W cnr K* which happens to be only about 1 percent greater than k, In this instance the value of a has been taken as representing approximately the thermal conductivity of polyerystalline gallium, but it is clear that more attention could well be devoted both experimentally and theoretically to this problem. Electrical conductivity would behave similarly and this property is likely to be measurable with greater accuracy, although high accuracy would not be so necessary with the large differences indicated for gallium. A practical difficulty could however arise in this instance from the ease with which gallium solidifies in the single crystal form, and the difficulty experienced so far in preparing truly poly crystalline samples of this metal. >h> In this work, the average of the values given by equa- tions (11) and (12) has been adopted as the estimated value for a polycrystalline sample of an element of large anisotropy if experimental data are not available. Tn connection with the thermal conductivity of molten metals, reference will frequently be made to estimated values that are due to Grosse [546, 548]. These values hhave been derived from the melting to the critical points using the equation k = Lye? with derived values for the clectrical conductivity, o, and usually assuming the theo- retical Lorenz number, Ly, to hold throughout the range. To derive an expression for the electrical conductivity, Grosse has proposed an equation of the form of a simple ‘equilateral hyperbola (545] (¢ $b) (T+) =, as) where the reduced electrical conductivity of = or/ay, the reduced temperature T= (T — 1))/(Te — T)), oy is the electrical conductivity of the molten metal at the melt- ing point, and cr is the electrical conductivity at a tem- perature T between 7), the melting point, and T., the critical temperature. The quantities a and B are constants. ALT both ¢ and k are assumed to be zero, Since these predictions were made, increasing uncer- tainty has developed as to the Lorenz function of molten 4. Phys, Chom, Raf. Dato, Vol. 3, Suppl 1, 1974 HO, POWELL, AND LILEY metals and its variation with temperature. Previous work, for instance of Powell [1120], had indicated the Lorenz function to approximate the theoretical value, as was assumed by Grosse (546, 548], but according to the work of Filippov [435] on tin and lead and some other recent measurements (884, 1593] the Lorenz function continues to decrease with increase in temperature to values that are well below Ly. This uncertainty needs resolving and, pend- ing confirmation and theoretical support for the lower values, values closer to those of Grosse have provisionally been adopted in the present work. 2.2. Data Evaluation, Correlation, Analysis, and Synthesis Data analysis and synthesis involve critical evaluation of the validity and accuracy of available data and related information, resolution and reconciliation of disagreement in conflicting data, correlation of data in terms of various controlling parameters (sometimes in reduced forms ‘ple of corresponding states), curve fi 1 or empirical equations, comparison of result ing data with theoretical predictions or with results derived from semitheoretical relationships or from general- ized empirical correlations, etc. Besides critical evaluation and analysis of the existing data, thermodynamic, kinetic, or statistical mechanical principles and semiempirical tech- niques are employed to fill gaps and to extrapolate existing data so that the resulting recommended values are inter- ly consistent and cover as wide a range of each of the controlling parameters as possible. In the critical evaluation of the validity and uncertainty of a particular set of data, say, the thermal conductivity ‘of a solid, the temperature dependence of the thermal con- ductivity was examined and any unusual dependence or anomaly carefully investigated, the experimental technique reviewed to see whether the actual boundary conditions in the experiment agreed with those assumed in the theory and all the stray heat flows and losses were prevented or ized and evaluated, the reduction of data examined to see whether all the necessary corrections had been applied, and the estimation of uncertainties checked to ensure that all the possible sources of errors had been considered, For a steady-state absolute measurement of the thermal conductivity of a solid specimen, for example, the sources of errors may include the uncertainty in the measurements ‘of specimen dimensions and of the distances between points of temperature measurements; the uncertainty in determining the necessary correction to the thermal con Auctvity value due to the effect of thermal expansion; the uncertainty in determining the power input to the spec rmen heater; the uncertainty in determining the heat gains or losses to or from the specimen due to direct radiation interchange or to conduction through the surrounding insulation, along the electric leads, and along the thermo: couple wires and the ceramic insulating tubings or beads; the uncertainty in temperature measurements due to poorTHERMAL CONDUCTIVITY OF THE ELEMENTS thermocouple calibration, poor thermocouple contact, poor sensitivity of the measuring circuits, and temperature drift; the uncertainty due to the effect of thermal contact. ance; the uncertainty for measurements at elevated tem- peratures due to thermocouple contamination, specimen oxidation, and reaction of specimen with apparatus com- ponents, etc. In a comparative measurement, additional ‘uncertainties may come from the conductivity mismatch between the specimen and the reference sample(s), from the additional interfacial thermal contact resistance, and from the additional uncertainty in the conductivity of the reference sample (especially if the conductivity values of the “reference” sample are blindly taken from the litera ture). For a nonsteady-state measurement, larg tainty may result if the density and specific heat values are taken from the literature and not directly measured on the specimen for which the thermal diffusivity data are obtained. The above-mentioned and other possible sources, of errors have been carefully considered in critical evalua tion of experimental data in this work. Many authors have included detailed error estimates in their published papers, and from these it is possible to evaluate the uncertainty for a particular method. However, experience has shown that the uncertainty estimates of most authors are unreliable. In many cases the difference Detween the results of two sets of data is much larger than that given by the sum of their stated uncertainties. Cases feven occur where measurements reported to be accurate to within 1 or 2 percent differ from each other by more than 100 percent. In these cases either the actual error ‘must greatly exceed its estimated value, oF the author was tunaware of the sources of systematic errors, or there must be essential unrecorded sample differences. Therefore, al- though one has to assume that the author's error estimate bears some relationship to the truth, the exact functional relationship depends on the author himself and on other factors, e.g, was he a new investigator or an experienced worker? was the paper a rushed project to meet # com ference or contract deadline? etc. Besides evaluating and analyzing individual data sets, correlation of data in terms of various controlling param- eters is a valuable technique that is frequently used in data analysis. These parameters may include purity, come position, residual electrical resistivity or electrical resis Livity ratio (if a metal), density or porosity, hardness, crystal axis orientation, degree of cold working, degree of heat treatment, etc. Applying the principle of correspond- ing states, reduced property values may be correlated with reduced temperature and other reduced parameters. Certain properties of the elements may also be correlated with the ‘atomic numbers of the elements in the periodic system; ‘examples are critical temperatures, critical pressures, criti- cal volumes, and atomic volumes at 0K, Wherever appro- pripate, such correlation techniques have been applied to the thermal conductivity of the elements in the present work. Several properties of the same material can also be cross-correlated. For example, thermal conductivity, specific Lz heat, and density can be correlated with thermal diffusivity, and viscosity and specific heat of a gas can be correlated With thermal conductivity through the Chapman-Enskog theory or through the experimental Prandtl number. For 4 fluid, the property of the saturated liquid ean also be correlated with that of the saturated vapor. For meaningful data correlation, the information on specimen characterization is very important especially for solid specimens. A full description of a solid specimen should include, whenever applicable, the following: purity or chemical composition, carrier concentration; type and concentration of lattice defects; type of crystal, crystal axis orientation for a single crystal; microstructure, grain size, preferred grain orientation, pore size and shape and orientation, inhomogeneity, and additional phases for a polycrystalline specimen; specimen shape and dimensions, ‘method and procedure of fabrication; thermal history and cold work history, heat treatment, mechanical, irradiative, and other treatments; manufacturer and supplier, stock number, and catalog number; test environment, degree of vacuum or pressure, heat flow direction, strength and orientation of an applied magnetic field: pertinent physi- cal properties such as density, porosity, hardness, electrical resistivity (residual, ratio, and temperature variations), Lorenz function, transition temperature, etc.; and refer- fence material and its property values for a comparative method of measurement. Data (no matter how accurate) con poorly characterized materials can hardly be analyzed or used for data correlation. It has been found in this and other studies that the specimen purity or composition reported by the author is often unreliable. This is because in many cases the stated purity or composition is the result of ladle analysis which the author obtained from the company who supplied the specimen and it ean at best represent only the nominal purity or composition (the actual purity or composition varies from sample to sample). In other cases there is a strong tendency for only certain elements to be covered by a particular chemical analysis, which could miss other quite important constituents. Besides specimen characterization, a full description of experimental details should, of course, be given by the author in order that his data can be meaningfully evalu ated and fully utilized. Sometimes, as an initial method of evaluating the quality of a paper, consideration may be given to the amount of experimental details reported in the paper. Lack of experimental details could lead to the results being given less weight, However, it should be emphasized again that the real enemy of reliable data is the systematic measurement error, whose existence in his rmeasurement is usually unknown to the author. As the author was unaware of it, he provided litle information about it in his paper. Therefore, in data evaluation one should not only try to find out what was wrong with @ measurement according to the author's description, but also try to detect with reasonable suspicion any possible sources of systematic error unknown to the author in his measurement. 1k is apparent that any data that can be considered 4. Phys. Chem, Ref Dote, Vol. 3, Suppl. 1, 1974His Ho, reliable must be free of systematic error. In practice, for example, if several sets of data that agree with one another were produced by different authors using different abso- lute experimental methods, these data can usually be considered reliable. However, if they were produced by the same experimental method, the agreement may be spurious and there still exists a probability that they may all suffer from a common, but unknown, systematic error. In estimating the degree of uncertainty of our recom- ‘mended values for the various ranges of temperature, it is clear from the above discussion that only for the thermal conductivity of the few much studied materials has it been possible to place close error limits that can be considered reliable. For the less well studied materials, wider limits of uncertainty are generally given; these are based also ‘on other factors and considerations such as general know- edge of the worker, the accuracy of measurements of other materials using the same or similar apparatus, ete. The estimated uncertainty also takes into consideration the behavior of the material itself. For a well-behaved mate- rial narrower limits are given when the temperature dependence of its thermal conductivity is predictable from theoretical considerations or from empirical correlations For an illbehaved material or material with ph and/or magnetic transformations, such as the rare earths, the estimated uncertainties are greater. For the recom- ‘mended values of the thermal conductivity of fluids, the ty estimation is sometimes based on the degree ‘of agreement of our values with those proposed by other experimental or analysis specialists, coupled with a more personal opinion of the experimental accuracy of the existing stateof-the-art of the measurement techniques ‘At other times, the scatter around the recommended values ‘of those experimental data considered reliable forms the basis of uncertainty e 2.3. Summary Graphs With a view to bringing out any similarities or difer- ‘ences between the recommended values for the elements of a particular group of the periodic table, these values for all the elements of each group have been plotted in figures 3 to 14, which show some of the generalizations for the property of thermal conductivity that were men- tioned at the beginning of this section. These figures may prove helpful when making estimations to temporatures not covered in the sections which follow. In figure 15 the thermal conductivity of each element at 300 K is plotted against the atomic number of 1 clement. A fairly definite pattern can be traced, and this hhas been of assistance in deriving estimated values for certain elements for which no information is available. ‘These include actinium, francium, and the trans-plutonium clements, The thermal conductivities of the elements at 300 K and at the Debye temperature are further tabulated in a periodic table appearing on page I-10. This table shows more clearly the variations of the thermal conduc- tivity values ofthe elements within each group. an Chom, Ref, Dato, Vol. 3, Suppl 1, 1974 POWELL, AND LILEY Figure 15a is a plot of the thermal conductivities of a amber of solid nonmetallic elements at 300 K against the melting temperatures of the elements. A reasonable curve can be dravn through the points to indicate roughly the thermal conductivity at 300 K as a function of the melting temperature. This information was used to obtain an estimated value for the thermal conductivity of astatine as shown in the figure. In the other estimations, estimated thermal conductivity values for barium, calcium, europium, polonium, pro- tactinium, and strontium have been based on electrical resistivity data, and those for radon are based mainly on a generalized correlation by Owens and Thodos (1058). ‘The value for radium comes from collected data by Samsonov [1239], and is attributed to Chirkin [274], Dut no detail is given. 3. Specific Considerations Concerning the Body of Data ‘This compilation contains a large number of graphs and tables of thermal conductivity as # function of tempera ture, The conventions used in this presentation and special comments on the interpretation and use of the data are given below. ‘The thermal conductivities of the elements are pre- sented alphabetically by the names of the elements, but it should be noted that where information is given for iflerent forms of a particular element, these follow that element. Thus, entries for amorphous carbon, diamond, and for several types of graphite come in the entry for carbon, and those for deuterium and tritium are found after the entry for hydrogen. For most of the nonmetallic clements which are liquid or gaseous at normal tempera- ture and pressure (N:T.P.) and for iodine, thermal con- ductivity values are given for the solid, saturated liquid, saturated vapor, and gas. For the other elements, values are given only for the solid state or for both the solid and liquid states. Tn all figures containing experimental data, a data set consisting of a single point is denoted by a number en- closed by a square and a curve that connects a set of data points is denoted by a ringed number. ‘These numbers correspond to those given in the accompanying tables on specimen characterization and measurement information. When several sets of data are too close together to be distinguishable, some of the data sets, though listed in the table, are omitted from the figure for the sake of clarity. For those omitted curves, one is referred to the TPRC Data Series, Volumes 1 and 2 [1421, 1422], in which imerical data tables are also given. The much heavier curves drawn in the figures represent the proposed values of the thermal conductivity. These heavy curves may be continuous, shortdashed, or long-dashed. Heavy con- tinuous (solid) curves properly labeled represent recom- mended reference values or provisional values. Accompany- ing sections of short-dashed curves represent values in theHs ‘THERMAL CONDUCTIVITY OF THE ELEMENTS T | t I T THERMAL CONDUCTIVITY OF ELEMENTS OF GROUP O TPR 4 Phys. Cham. Ra Dot, Val. 3, Suppl 1, 197HO, POWELL, AND LILEY 20 a ELEMENTS OF GROUPS IA AND IB | THERMAL CONDUCTIVITY OF - : ~ ge eown NS -wenen «0 oooemn 2 182 MALIGNO WHEEL 4. Phys. Chem, Ref, ta, Vol. 3, Sepp 1, 1974‘THERMAL CONDUCTIVITY, W om" et ee cr sinetwlo 2 ne cele Ol ie ol nea lei whan os THERMAL CONDUCTIVITY OF THE ELEMENTS bar THERMAL CONDUCTIVITY OF ELEMENTS OF GROUPS IIA AND IB st 4. Phys, Cham. Ref. Date, Vol. 3, Seppl 1, 197412 HO, POWELL, AND LILEY THERMAL CONDUCTIVITY OF ELEMENTS OF GROUPS IIA AND 1B re em alstee ac) THERMAL CONDUCTIVITY, Won K 4 Phys. Chom, Raf, Det, Vol 3, Suppl 1, 1974THERMAL CONDUCTIVITY, W cnf~ ‘THERMAL CONDUCTIVITY OF THE ELEMENTS 123 Tt TT oppo TTT t T THERMAL CONDUCTIVITY OF ELEMENTS OF GROUPS IZA AND WB i | L nw bao o% wow seo es, ww Rao Od pw hae od 7 4. Phys, Chem. Ref. Date, Vol. 3, Suppl 1, 19744 HO, POWELL, AND LILEY ea 7 ope eS] THERMAL CONDUCTIVITY OF ELEMENTS OF GROUPS VA AND VB oben 2% THERMAL CONDUCTIVITY, Werk” 4. Phys. Chom, Raf. Date, Vol. 3, Suppl. 1, 1974‘THERMAL CONDUCTIVITY, Wem" K* see 0% THERMAL CONDUCTIVITY OF THE ELEMENTS 125 THERMAL CONDUCTIVITY OF ELEMENTS OF GROUPS WIA AND WIB | 4. Phys. Chem. Ref Dota, Vol 3, Suppl. 1, 19741-26 HO, POWELL, AND LILEY THERMAL CONDUCTIVITY OF — ELEMENTS OF GROUPS VITA AND WITB H HH | o san 0% THERMAL CONDUCTIVITY, W em" K* waee eh nv usano- w wtanad ' 2 345680 2 345680 2 345680 2 3 45 TPRC TEMPERATURE, K EIGURE 10 4. Phys. Chem. Ref, Dota, Vel. 3, Suppl: 1, 1978‘THERMAL CONDUCTIVITY, Went” wo seo eg ee eere a cee oc) THERMAL CONDUCTIVITY OF THE ELEMENTS THERMAL CON! Lar DUCTIVITY OF ELEMENTS OF GROUP VI . 7% Chom, Ref, Dot, Vol 3, Sop. 1, 1974128 HO, POWELL, AND LILEY & ‘5 ie 4 y 4 z [ g - a 2 zg z : z | ° oman iron! : ran i wena . Lo Jo imc po t oN nn Ht =F ape A [AR aalele | fat jes ee fT al | 1 E | | | a iil iid sty uP tt bib TPRC ‘TEMPERATURE , K FIGURE 12 4. Phy. Chem. Re. Det, Vel. 3, Suppl 1, 1974‘THERMAL CONDUCTIVITY OF THE ELEMENTS 29 THERMAL CONDUCTIVITY , Wom "K* 1 / (| CONDUCTIVITY OF ‘Ty ELEMENTS OF | | #/ THE RARE EARTH GROUP |] uf boiy iy | | 4 L vi A LL! rit ' 2 3 #56 80 2s 456 a 2 3 ese 88 a TPR ‘TEMPERATURE. , K FIGURE 2a 4. Phys. Chem. Ref Date, Vol 3, Supp. 1, 1974130 HO, POWELL, AND LILEY THERMAL CONDUCTIVITY OF = ELEMENTS OF THE ACTINIDE GROUP aK! THERMAL CONDUCTIVITY , Wer 41 Phys. Chom, Raf, Date, Vol 3, Suppl. 1, 1974THERMAL CONDUCTIVITY, Wen" K" THERMAL CONDUCTIVITY OF THE ELEMENTS Lat (7 Betesn (0 eson on) ——— or ae 7 Rea wet aor N a (proud S068, Ne THERMAL, im 4 CONDUCTIVITY OF 7 GRAPHITES 4. Phys, Chem. Raf Date, Vol. 3, Suppl. 1, 1974HO, POWELL, AND LILEY L-s2 ST SunOS 7pm 83 YG UY AN od OY 14 HG LL TUL ut 01M aA 3 oN mss opr yada N33 n ‘sigs eg is levlzeiay ov ee ZeiSeleel e162, ‘8G 9G 1G 20 0G Ob 9b by Zh Ob OF OF bE ZOE MEIN SINOL M ‘ALIALLONGNOD WWHSHL ye 4 Phys. Chem. Raf Dot, Vol. 3, Suppl 1, 1974THERMAL CONDUCTIVITY, Wen! k7 THERMAL CONDUCTIVITY OF THE ELEMENTS 183 ] ize lea By | THERMAL CONDUCTIVITY OF ~T 2) NONMETALLIC ELEMENTS ~~} —_ isl AT 300 K | 16) 8 6 - 4 3 2 15;— a 16 8 6 = 4} Ae — | 10% 15. 2 3 TPRC MELTING TEMPERATURE, K EIGURE 15a 4. Phys Chom. Ref, Date, Vol. 3, Seppl 1, 1974baa temperature ranges where no experimental data are avai able, In some instances, notably for nonmetals and sen ‘metals at low temperatures where the specimen cannot be uniquely characterized to correspond exactly with the thermal conductivity values, values considered as typical are represented by heavy long-dashed curves. For all the elements, logarithmic plotting of thermal conductivity agairst temperature is adopted in order that, details are clearly shown for the lower temperature region. In the cases of elements which become superconductors fat low temperatuces these figures contain the available data for both the normal and superconducting states, but so far all recommendations have been limited to the normal state, Corresponding linear plots are also given for those solid elements which have received considerable attention at temperatures above normal. In addition, for each of the liquid and gaseous elements one or more departure plots ace included which show the deviations of the available data from the recommended values, In the figures, the melting point (MLP.), phase transition point (T.P. , supe:eondueting transition point [T.P.(s..)}, critical temperature (C-T.), Curie temperature, Néel tem- perature, etc, of she elements have been indicated. Some of these transitior. points are also mentioned in the text. The inclusion of these transition points is intended to caution the reader against the existence of such transi tions so that one must be extremely cautious in attempting to extrapolate the thermal conductivity values across any such transition temperature, since at such temperature the thermal conductivity generally exhibits sharp discontinui- ties. No attempt has been made to critically evaluate these transition temperatures, and they should not be considered as recommended values. Some of the given values, how- ever, are the defining fixed points or secondary reference points of the International Practical Temperature Seale of 1968 (IPTS-68) such as tho indicated melting. points of gold, silver, tin, ine, aluminum, antimony, bismuth cadmium, cobalt, copper, indium, iridium, lead, mercury nickel, palladium, platinum, rhodium, and tungsten, and the boiling point ef mercury. Inthe departure plots for the nonmetallic elements which are liquid or gaseous at N.TP., Percent departure Experimental value — Recommended value Experiment nded value 392, (16) Recommended value By the above definition, departures are positive if the experimental data are grester than the recommended values and viee versa. ‘The tables on specimen characterization and measure- ‘ment information give for esch set of data the following information: the publication reference number, author's name (or names) method used for the measurement, temperature range covered by the deta, substance name and specimen desig nation, as well as detailed description and characterization cof the specimen and information on measurement condi year of publication, experimental 4. Phy Cham, Raf Dota, Vo. 3, Suppl 1, 1974 HO, POWELL, AND LILEY tions that are contained in the original paper. Whenever available, information on the electrical resistivity has also been included. In these tables the code designations used for the experimental methods for thermal conductivity determinations are as follows: € Comparative method E Direct electrical heating method F L Forbes’ bar method Longitudinal heat flow method P-Periodie or transient heat flow method R_— Radial heat flow method T — Thermoclectrical method For a comprehensive yet concise review of all these methods, the reader ig referred to the text in [1421]. In the tables of recommended, provisional, or typical thermal conductivities, the values are presented with uni- form but step-inereating the temperature increases, For those elements which are solid at N-T.P, and for mercury, the values are presented such that temperatures with uniform increments in both kelvin and Celsius are accommodated, In other words, those values given for temperatures 123.2 K, 173.2 K, 223.2 K, 273.2 K, 323.2K, ... are for ~150° C, 100° C, —50° C, 0° C, 50° » The 2” has been dropped for temperatures above 3000 K. In the tables the third and occasionally the fourth y values, but this is only for internal comparison and for tabular smoothness and should not be considered indica tive of the degree of accuracy. The accuracy of the recom- ‘mended or provisional values for each element in different tomperature ranges is given in the discussion, The aster- isked values in the tables are interpolated, extrapolated, for estimated, but more factually they sre in the tempera- where no experimental data are available. The thermal conductivity is zero at absolute zero temperature, ive at the point (T = 0, k = 0). This is a theoretical consequence based upon the premise that the specific heat is zero at absolute zero temperature according to the third law of thermodynamics. ‘The compiled 5200 sets of experimental thermal con- ductivity data were published over a period of 110 years from 1861 to 1970, It is realized that many different tem- perature scales were used for these data. However, in thermal conductivity measurements, the thermal conduc- tivity values are determined by the measured differences in temperature and not by the absolute magnitude of temperature. Furthermore the thermal conductivity is only a weak function of temperature, Therefore, the effect of using different temperature scales on the reported thermal conductivity values is practically negligible. Consequently, no attempt has been made to convert the original data to ‘a common scale. For the recommended values, the tem- peratures are based on the IPTS-68, In the Thitteenth General Conference of Weights and Measures held in October 1967 in Paris, the unit “watt per metre-kelvin” (symbol: W mt K-*) was adopted as erements in temperamre as significant figure are given for the thermal conduct ture rangTHERMAL CONDUCTIVITY OF THE ELEMENTS the ST unit for thermal conductivity. In this work, the unit “W emt K" js used which isa slight modification ofthe SI unit, Table 1a gives conversion factors which may be used to convert the thermal conductivity values in Weem-* K+ presented in this work to values in the SI unit or in any of the several other units listed. For a solid slement at moderate and high temperatures the true thermal conductivity values for different well annealed highpurity (99.99%) samples at each tem- perature shoull be close, and therefore a set of recom mended thermal conductivity values can be given for a wellannesled high-purity element. At low temperatures, however, the thermal conductivity values for different samples with small differences in impurity and/or imper- fection differ greatly, and a set of recommended or pro visional values applics only to a sample with a particular amount of impurity and imperfection, Thus, the low. temperature thermal conductivity of a solid element may be represented by a family (or families, for a non-cubie crystal) of curves, each of which is recommended for a sample of a perticular amount of impurity and imperfec- tion, and hence having a particular residual electrical resistivity for a metal, as shown in figure 1. In this work, such a family of recommended curves for specimens with different hypothetical impurities and imperfections has not been generated. Instead, a single, well-defined curve is drawn to ling with the recommended curve for moderate and high temparatures s0 as to complete the functions for the full range of temperature, The recommended low- temperature velues in the table, which are for the purest form of each element for which a measurement has been made, are of course only applicable to that particular characterized sample whose residual electrical resistivity has clearly been specified. Consequently, this recommended curve should not be interpreted as a uniqae function for the low temperature region, but itis only applicable to a sample of specified conditions. For samples with different amounts of impurities and imperfections, ie, having dif- ferent residual electrical resistivities for a metallic clement, one may similarly derive low-temperature thermal conduc tivity curves following the same guidelines and procedures 4s used in this work or may exercise proper selectivity and discretion based on the extensive information reported for each data set in the accompanying table for specimen characterization and measurement information mentioned before, the residual electrical resistivity vis used for the characterization of « metallic sample to Correspond with the recommended low-temperature thermal conductivity values, At temperatures around 4 K or below, fo >> po and hence p, may be written as ot ke ary po which is determined experimentally as the residual 3s electrical resistivity resulting from electron-dfect scatter- ing. If however p, is calculated from an equation similar to (17) but using a measured value of the thermal con- ductivity, this value is F and not k.. Denoting the value so caleulated by p.', then Le fe 14h It ean be seen from equation (18) that if ky =O then py’ = io This is uspally true, and the experimental p, is then the value given to correspond with the recommended values. It happens occasionally, however, that the calculated ix! > por implying that ky <0. As negative values for ky are impossible, the measured p, is concluded to he in error, and in this case the calculated value 9,’ has been given as corresponding with the recommended k values. pe (as) Regarding those elements which are liquid or gaseous at NT-P,, the provision of recommended values of the thermal conductivity at the critical point takes no account of anomalies in the immediate vicinity of this point. While evidence seems to be accumulating that a 19) in thermal conductivity to very large, if not infinite, values oes occur in the immediate vicinity of the critical point, the temperature span of any such departure is very short, and in the preparation of the present tables this factor has Ibeen disregarded, The values recommended here for the critical point are thus obtained through arbitrary extrap- olations of the saturated liquid and vapor curves with no considerations being given to such anomalies. This approach was considered justified by the very meager and indefinite investigations which have been concerned with such an effect, The present approach has been taken so that interpolation of the recommended critical-point values, with those tabulated for lower temperatures will enable intermediate temperature values to be obtained which will be accurate except for the small temperature region where anomalies may occur. Furthermore, the values at the critical point are needed for data correlation using the principle of the corresponding states, Likewise, the error estimates refer to possible errors in estimating such values, Should recent studies on anomalies prove to be confirmed, the present values might he regarded as “pseudo-critical” values of thermal conductivities. While the merit of our present approach could be questioned by some, it might bbe added that the above defined “critical” thermal con- ductivities have been found to give consistent values when, comparing “critical” thermal conductivities of families of substances, At the present time, similar treatments of “true” critical values present serious difficulties. ‘The recommended values for the various gases, which le ranges of temperature, are only for a pressure of one atmosphere. The pressure dependence of thermal conductivity is not yet included in this work. cover very wi 4. Phys, Chem. Raf Date, Vol 3, Suppl. 1, 19741 v 0X1] e011] 01x T2H09"8) -TXe) uoisoxuey “ay ZN, : i é éTHERMAL CONDUCTIVITY OF THE ELEMENTS 4. Thermal Cond a conductivity of actinium, However, very rough estimation, of its room-temperature thermal conductivity may he made ‘on the basis of similarities between actinium and the other elements of the same group. The thermal conductivity values at 300 K of the other three members scandl yltrium, and lanchanum of Group III B are 0.158, 0.172, of the Elements and 0.135 W cm~ K-, respectively. The extrapolation to atomic number 89 of a curve fitted to these points plotted in a large working graph of thermal conductivity versus atomic number similar to figure 15 gives a value of 0.115, W em K> for actinium at 300 K, This derived value is, probably good to -£50 percent, Aluminum of data available for the thermal conductivity of aluminum as listed in table 3 and shown partly in Figures 16, 17, and 17a, At low temperatures, most of the thermal conductivity maxima conform well to a straight line (in a log-log graph) for which the slope is —2.62, and most of the experimental data at temperatures below 1.5 Tq can be fitted by equatior (7) using constants m = 2.62, n = 2.00, a” = 4.79 X 10-* as given in table 1 and using appro- priate values for the parameter 8. The heavy eurve shown in figure 16 for which the cal- culated p, equals 0,000594 a em and = 0.0243 fits the data of Fenton, Rogers, and Woods [425] (curve 64), whose experimental p, of 0.000568 af em gives an experi- mental L of 2.337 X 10+ V K-? which is 4.3. percent below the theoretical value, Therefore po = 0.000594 0 cm (instead of 0.000568 x em) is used to correspond to this recommended curve. To derive recommended values at higher temperatures, the curve which fits the data of Fenton, et al. [425] (curve (64) to their upper limit of 25 K continues to decrease smoothly towards the data of Powers, Schwartz, and Johnston [1154] (curve 1) at a little above 100 K. It continues in an intermediate position between their data ‘and the lower curve of Flynn [446] (curve 110) and, after 4 minimum at sbout 240 K, rises to a gentle maximum around 360 K. Through weight given to the recent dete minations by Duggin [372] (curves 106 and 107) the recommended curve to the melting, point has been lowered by about 1 percent at 800 K from TPRC’s earlier recom- mendation [1126], and now lies above the data of Duggin [372] and Flynn [446] and below those of Powell, Tye, and Woodman [1143] (curves 48 and 49). It is interesting to note that it was not until the publ tion of the data of Powell, Tye, and Woodman [1143] (curve 50), which show a pronounced minimum of thermal conductivity in the subnormaltemperature region, that the existence of such a minimum was fully recognized. Although theoretical investigations of the electronic thermal conductivity of metals by Bloch [169a], Wilson (1568, 1569], Makinson [875], Kroll [789a], Umeda and Yamamoto [14424], and Sondheimer [13424-13426] all indicate a minimum in the theoretical curve of thermal fe around 6/4, where @ is the Debye temperature, it has often been stated that such a minimum has never been observed experi mentally [see, ef 719a, 1612, 287a]. On the premise of nonexistence of such @ minimum of thermal condue- tivity, several subsequent theoretical in been devoted to the elimination of this mi theoretical curve. Thus, by modifying Bloch theory in his recalculations, Ziman (1612a] has greatly reduced the dip in the theoretical conductivity curve to about 9 percent from Bloch’s original 40 percent of the high-temperature limiting value, and furthermore the minimum is shifted to a higher temperature of hetwoen 0.4 6 and 0.5 6. He was disappointed with the remaining, “discrepancy” that the conductivity mi ly eliminated, though, he said, he was working in the right direction to remove Collins and Ziman (287a] have pursued this idea further and their results show that with sufficient ine crease in the proportion of electron-phonon Umklapp scattering to normal scattering in their model used, the minimum eventually disappears. Klemens [770a] has modified Sondheimer’s method by solving the Bloch equation numerically. He obtained a shallower minimum, around which his result for thermal conductivity exceeds Sondheimer’s by 11 percent. A further modification of Sondheimer’s method was made by Kasuya [719a]. His result is, in turn, about 20 percent greater than that of Klemens, and shows that the conductivity minimum is completely removed. In spite of the efforts which successfully remove the ‘minimum from the theoretical thermal conductivity eurve, such @ minimum exists in reality. In fact, long before the publication of the data of Powell, et al. [1143], Lees [830] hhas reported a thermal conductivity minimum for a sample ‘of 99 percent aluminum (curve 108) more than half a century azo, and the thermal conductivity curve of Powers, Ziegler, and Johnston [1156] (curve 109) published in 1951 for an aluminum alloy with 98.17 percent AI (by difference) has also a minimum; both appear to have been conductivity, which occurs at a temperat yum was not ent 42 Phys. Chom. Ref, Data, Vi 3, Suppl 1, 197438 overlooked. Subsequent to the publication by Powell, etal, the thermal conductivity minimum for aluminum has been reported by Flynn [446] (curve 110) and by Moore, McElroy, and Barisoni [987] (curve 81). All the con- ductivity minima cecur at different temperatures, ranging from about 0.35 0 to 0.6 6, and that the lower the conductivity minimum, the lower is the corresponding temperature, ‘The present recommended ‘curve shows a mirimum at about 240 K, which is about 0.62 6 (6 = 390 K for aluminum at room temperature) It should be poisted out that most of the theories men- tioned above have been developed for monovalent metal to which aluminum does, of course, not belong, However, a minimum does exist also in the thermal conductivity of sodium, a monovslent metal whose earlier experimental thermal conductivity has often been compared with theory and has ironically been the basis for the assumption of nonexistence of the conductivity minimum, licate a general trend ‘The few experimental values for the thermal conduc- tivity of molten aluminum differ considerably. Only Powell, ‘Tye, and Metcalf [1140] (curve 51) included measure iments of the electrical conductivity, and it is interesting to note that their values of the Loren function are close to the theoretical valie, being respectively only about 0.5 percent and 1.5 percent lower at 978 and 1223 K. Mention of this is made because Grosse [546] has assumed the theoretical value of the Lorenz function to hold from the relting point to the critical point when deriving values for the thermal conductivity of aluminum over the entire Tiquid range (curve 111). ‘The electrical conductivity values assumed by Grosse [516] for liquid aluminum near the melting point were those of Roll and Motz [1217]. These values were some 6 percent higher than those of Powel, etal. [1140] which accounts for much of the difference hetween the two almost parallel thermal conductivity curves. The present recom- mended curve is the heavy and partly short-dashed line shown between them in figure 17a, but biased toward the experimentally dezrmined thermal conductivity values. [At about 950 K this line passes close to the values of Konno [778] (curve 13) which are the first such measure- ments made on liquid aluminum. ‘These recommended values indicate that on passing from the solid to the liquid state the thermal conductivity of aluminum decreases by a factor of about 2.3. ‘An observation regarding the effect of heat treatment fon the electrical conductivity of molten aluminum was reported by Koronenko, Yatsenko, Rubinshtein, and Privalov [779], who found that preheating to 1300 K for ‘one hour in vacuum led to lower resistivity values. They believed that oxides became partially dissociated and rose to the surface of the melt. Following this treatment their electrical resistivity values at 973 and 1273 K were respec- tively 7 and 7.6 percent less than those of Powell, et al. [1140], whose electrical resistivity determinations have been made in air, whereas their thermal conductivity values were determined ander vacuum conditions, but without 41. Phys. Chem. Re. eto, Vol. 3, Suppl 1, 1974 HO, POWELL, AND LILEY any preheating. ‘The findings of Kononenko, ot al. [779] suggest that further measurements on molten aluminum are required for both thermal and electrical conductivities. ‘The recommended values for the solid are thought to be accurate to within = 5 percent below room temperature and +2 to 23 percent above. The values below 150 K are applicable only to aluminum having residual electrical resistivity of 0.000594 49 em. For liquid aluminum near the melting point the values are probably good to within “£8 percent. Above 1273 K the values are provisional. ‘Tarue 2, Recommended thermal conductivity of aluminum (Temperature, T, Ki Thermal Conductivity, by W em" KY) Solid rhe} rtel rftelor|e Oe |poe| los: lors.2 ase 1 Jaalo & lz800° fits 2 |arslao fe iavs.2 [L.1s* 3 fa jo &. js000 fi-13* 4 |isr hoo lsors fri 5 fisa ha3.2 fe. [3200 ht 6 fas so fe lors fk 7 fox hae lssoo ft 8 sr boo lssz3 fh 9 ps9 hese fps fs600 10 35 ps0 fe. fisoo —fr.ose sors fi.ose ni f226fats.2 2.36 hiars.2 fr hoa 2 bia foe.2 2.37 fisoo h.oze por foo fe'sz fisrs.2 fh jo.907" 14 iso 523.2 [2.39 iso b lo. 986° 1s firs [s50_—fo.ao fisrs.2 woe fso73_ fo, 16 fies fars.2 f2'40 fiz00 i'ma {500 o. 1s fsa fio f-aohizzs.2 rue larz3_ fo. 20 jiuz_|as.2 {2.87 a0 faze [S000 _o. 25} 75.2500 2.36 his7s.2 [rise fers fo ao | 49.5 [575.2 [2.88 fi900_ frase. 3500 35 | a.a loo fia fisva.2 frase |sr73__o. 40 | 24.0 [673.2 [2.26 lo000 fra [600 |o. 45 | 17-7 {roo e's faors.2 [uitse lezrs_o. so | 13.8 /773.2 pao farvs.2 fase |6s00 fo. js00 2.18 |2200 frase for lazs.2 [2-12 fza7a.2 f1.ise {7000 1900 fo 10 [2100 [n-as* fram Joaa.se [2.08 Josra.2 |x15* [7500 1 fpso0” hase frzzs | froa0 | lazra. | i500 {The recommended values are for vwell-ennealed high-purity aluminum, and those below 150 K are applicable only toa specimen hhaving residual electrical resistivity of 0.000594 401 em. Above 1273 K the values are provisional. ‘Estimated or extrapolated.Lae THERMAL CONDUCTIVITY OF THE ELEMENTS to HO A ALIALLONGNOO WWEBHL 41. Phys, Chem. Ref. Dato, Vol 3, Suppl. 1, 1974140 HO, POWELL, AND LILEY __ THERMAL REN OF | | ALUMIN! ‘THERMAL CONDUCTIVITY, Wen K 4 Phys. Chom, Raf, Dato, Vol 3, Suppl 1, 1974‘THERMAL CONDUCTIVITY, W oni! K* THERMAL CONDUCTIVITY OF THE ELEMENTS Mar ‘guia! ‘TEMPERATURE. K FIGURE Io 4. Phys. Chem. Ref. Dato, Vol. 3, Supp. 1, 1974Ho, POWELL, AND LILEY La rw ser ae corner oraz ooretv orem “iy wow 5 01> 30 wrmoea v oy pases ~proo ‘eorroury 1099 tamaremTy sq porTean ust “antsy oF NORE 10N "179 ‘oy paw “9°D “TAPE ‘4 ‘ew209 pun “x “WOSTeCRI of 'a ‘eu209 pu “a “woe oot a a eo att or 6 ® toot not oN "0x HO NOLLVNWOANI INSHTUASVEN ANY NOLLYZIUALOVEVEO NENTOGAS ~ RANIKATY 40 KUALLONGNOO TVNUSHLL “t ATEVL 4. Phys. Chom, Ref, Date, Vol. 3, Suppl 1, 1978a3 THERMAL CONDUCTIVITY OF THE ELEMENTS “qeprope 96 vu s.0ELZ0N ow ow rw ew rv ow e96T -V'O ‘GO}WeHE PUR “A ‘MoEsTOPMERT SES ez seer “wy “arms w res BE 9 SOT OTH MopeR PEE “Te ‘OOM u ov were 1 weer moe a were T1s6t or rw pees Tse au ‘soqnem “Va ‘@mompUy OL ot ‘woneaiisog (a) ~ — ee (oysormmy een povouey —“duox “pio PHD, (@onsr09) NOLLVHYOUNI LNINSINSYSNW GNY NOLLVZTISLOVUYHO NaWIOGES ~ WANNA ZO ALIATLONGNOD TWRUSHL “8 wTEvL (Chom. Ref. Data, Vel. 3, Suppl 1, 1974 4. PhysHO, POWELL, AND LILEY 44 weiveee as curs 8 as mee 1 ae 64 ww ote rw ete ow 1 2 ° 1 “0.00 paw or 78 jut we 03 THe aaabp | x HH ssoqssera po ay ‘208 wormiBivea ‘eum poe “s00m: “(quoouad 128194) wonteodno our ee s}s0%9 Poe ‘eoomroprondy “(2030 eo) eoRatee ceed eer ele (ayo (ons t09) NOILYAINOSNI ININSNSVEN ANY NOLLVZRIELOVUVHLO NaWIDAdS ~ WONINOTY 40 ALIALLONGNOD TWAWANL * aTEV 41. Phys, Chem. Ref, Det, Vol 3, Suppl. 1, 1978THERMAL CONDUCTIVITY OF THE ELEMENTS owe 1 eee a oy see ow Go ras a4 as mw oT eds ey eds a za a en 26 ey -T see E "tORIO pun“ ‘wospmmy go ga 14s eT Stet E “toO PER “L ‘uowpMMY zo yg 145 a ws fonres Syaponpooo Teerzam) ‘320% g 9F~Jo wMNIEA Be PoeREwoUN uo] TWO GT UOWOES steed os6r ore — = woneaiisea (3) syremoe por ‘evornogioods “{qusozal yam) woRtweduog somods of PE sag oe ox “oN poveuexy “ano Pan a EO (onar809) NOLEVIUOANT INGHATUNSVAN GNY NOLLVZREALOVUVHO NSWIOMAS ~ RNNINATY 40 AUALLONGNOO TVA“ aTEVI 4. Phy. Chem. Raf Date, Vol 3; Suppl: 1, 1974HO, POWELL, AND LILEY 46 ants www ON , “porniousrunysumas » Aq pounsvous ‘uot wo 2°8t X Sip IO 6°z sus 9 oper ‘a ‘umug yeet ve “porpo juqsues © Aq pomsvour ‘Suoq wo ¥°TL x wp WO GE Bree a per a ‘uems ¥ecl oe (os ouese ounrecodwios Ptmsvon) “sTejoneu of Jo areMMAxo YwuLNG FuLTEIOS A ‘ourego Ayationpuoo Teuzoyy ‘Teron outfesnduicn av pow faddoo “oa Pp Be 90° zee 9 or “ACD ‘Poo OIF ze 10 ‘most poe 1 seer na “Koxgzone ar aH 1 over sew on eR *¥ ‘nopomecuyy pe 1 oer ‘a's “iosteg "e "TepmE OTT GL sees eeu 91 oF ‘0 wos; Sur2uex spon onouFeur Torre wy pommevout rec) 01940 sos Sees pa oeon reel poze seoo4 cee ere, 1 0"9 wosy SuySuex sproy onoutes zernoRpuedsad ur posnsvodr Wounds ee a -sema3on 91 $0"0 mo AuySuex sprou opowel zemorpuadsed uy pasmevons wu sort *E “uas10 paw “y,“wonpamay “2 ‘woaT0 paw “y ‘wospaneey Cer ‘ums NW Trae [err “vues 1408 £9 00% 20 IO ono wr pazmsvons woustoads eAoge 2G “Wi “ON ums ‘oqms Sumonpucosodrs uF passa nomads oAoKe UL wartinnd amo peo “euoneoutaeds “392568 eto) sos woenedg) SOE POD sng (oom (Gonur#00) NOLLYHMOANT INGMTUASVAN ANY NOLLVZNESLOVUVHO NEWIOMAS ~ WONINN'TY 40 KUIALLONGNOO IVHWHL “© TAYE 4. Phys, Chem, Ref, Date, Vol. 9, Suppl. 1, 1974Lar THERMAL CONDUCTIVITY OF THE ELEMENTS xR OF UHOGE ION ow soot s0r toe vor coor vee oor ae-sos sa cares Sty. 0) 4 60"F0 =n eEs0) \Fporteaa wmuicode one ous OME. ES"O ast seo “wansones yor 400T ower ers “ero ‘omens pat 400 oy gree 10°9 ‘omamaseins prot 480 crv 10 “areupsomes “ory ‘mS pow aa can “0 “onperp pee“ “0 9 ‘enpa pee “a “ene “9 ‘ememo pew “a “ae +o ‘ames ee “x “quo 26-0 "8 fine ton © nome aso i pom “onano prqoows 035 OMe ep POON 74 EOIN 8 od = Sl (spomey po (onay100) NOLLVMOSNE INGWSUNSTNN CNV NOLLVZNULLOVUVHO NaWIOSES ~ WANDVOTY 40 KLALLONGXOO TWHCHAHES, “€ wav, of Dato, Vol. 3, Suppl: 1, 1974 3 PhysHO, POWELL, AND LILEY has “wot zonsor-roe-emwepoq4 OM O BuNpLOO—N AARETEOX WoIsTOATS mOSy PATENTED osoa-¥ 008, soot ‘omy uf UMOYE Ye “AV ‘998000 gpg TTT “stp ox, ozuceet soot wa ‘muse ony ort “1-H, ‘ensue pow “32-1070 Pm “30 10°0 “uN E00 "206-0 “WS HOO “3H 980 ‘0H SEO Se uae sor mE soon a SuOmod STE eoT ‘2030 wo enor 2 00 £0 08 18 89 9 01°5 Sp £09 pee bet "aoe Pe tuna pease Y 6¢ rarer 1 fot W9 ‘Sor oss sor “uoutoade ase are ae see 9p 5055 9 00-0 sos rm wtma cut OT “22 1000 PaR “He 0‘0-100"0 “9 :0°0-£00"0 “2a £000 “AO £000 ero-e8e set ‘em "mama us G0r - ‘wonwaineg Gi) 7 ‘suremoy por “suopeotoodg ‘(imoos0d ito) woreodI0g, ene SO Pon rao (eouny ee (ons 09) NOLLVICUOANT INGMAUNSVANE ANY NOLLVZRIALOVEVHO NAWIOAAS ~ MANINATY JO ALIALLONGNOO IVUSHL “t adv 4. Phys. Chem, Raf, Dota Vol. 3, Suppl 1, 1974THERMAL CONDUCTIVITY OF THE ELEMENTS Americium No information is available in the literature for the thermal conductivity of americium. Based upon the un- published measurements of M. B. Brodsky on the electrical resistivity of americium at temperatures from 4 to 60 K, ‘Meaden [914] has made an estimation of its thermal con- ductivity at low temperature. The electrical resistivity values of Brodsky [94] are 29, 30, 33, 39, and 45 yo cm at 4, 10, 20, 40, and 60 K. Using the theoretical Lorenz number, Mesden [914] obtained & = 0.008 W em* K* at 10 K. However, he believed that this value was far too ow and the actual thermal conductivity of americium at 10 K might well be two or three times greater. ‘The reason- 1g is based on the fact that the above residual electrical resistivity contains an impurity-induced spin-disorder term fity should be used if the electronic thermal conductivity is sought. At the same time magnon, and phonon conductions are to be expected, leading to an increase in the Lorenz function which might far exeeed 2443 X 10 V? K*. fand only the truc impurity resist ‘The actinide clements are chemically very similar to the lanthanide clements and their electronic structures are, as a whole, also closely similar. The elements thorium, protactinium, and uranium also partly resemble the IV B to VIB transition metals (which may partially explain their higher thermal conductivities than those of ‘the lanthanides), but the transuranium elements do not. The transuranium elements are chemically and electronically similar to the lanthanides only. For instance, available evidence shows that americium and eurium are, respec- tively, quite equivalent to europium and gadolinium in knovn chemical and physical properties. Since the first toro transuranium elements, neptunium and plotonium, hhave their respective thermal conductivity values of 0.063, and 0.0674 W em Kat 300 K and since most of the lanthanides have their Unerunal conductivity values between 0.10 and 0.14 W em-* K~! at 300 K, it seems reasonable to estimate that the thermal conductivity of americium at 300 K is of the order of O.1 W cmr* K+, This estimated value is probably good to +50 percent, Antimony ‘The available information on the thermal conductivity of antimony is all for the metallic erystalline form. Since. antimony crystallizes in the rhombohedral form, a single crystal of the metal has three prineipal axes. Rausch [1184] has reported thermal and electrical conductivity values at 79.5 K and 91.2 K for flow directions parallel to each axis (curves 13-18). The extreme differences in thermal con: ductivity were only about 3 percent at 91.2 K and 6 per- cent at 79.5 K, yet the corresponding electrical conduc tivities diffezed by nearly 50 percent. This would suggest the presence of a considerable phonon component of thermal conductivity. For the trigonal axis direction Red'ko, Bresler, and Shalyt [1189] (curve 30) find the phonon and electron components of the thermal con- ductivity to be comparable in the region of 90 K. These ‘workers also find both components to increase to maxima below 10 K, the total thermal conductivity having a maximum value of 7.62 W em! K+ at 742 K. Only magnetic fields of moderate strength were required to suppress the electronic component, and this enabled the two components to be measured. Long, Grenier, and Reynolds [358] (curve 26) have previously used this ‘method to determine the phonon component, ky, for another single crystal in which the heat flow was perpendicular to the trigonal axis, the total thermal conductivity then being calculated ftom the relation k = ky + LeI'/p, where Ly is the theoretical Lorenz function and p is the measured clectrical resistivity. This yielded a maximum of 9.74 W cm K-* at 3.04 K. Subsequently, further measurements to lower temperatures have been made on this sample. The inclusion of proposed values for the principal erystal direc- tions shoul however avait more extensive experimental information. ‘Antimony is a metal for which the low-temperature thermal conductivity is predominantly due to phonon con- tribution and therefore the values do not follow the corzela- tion edopted by Cerairliyan and Touloukian [248-250]. Hence for polycrystalline antimony a smooth curve has been drawn for the range 2 to 90 K through the date of White and Woods [1151] (curve 6) for their sample of highest thermal conductivity and having p. = 0.054 no cm, This curve has been extended to high temperatures as a curve of gradually decreasing slope to the 850 K value of Dutchak and Panasyuk [383] (curve 23). The courte followed is a little above, but in fair agreement with the data of Rausch [1184] (curve 13-18), Gebihoff and Neumeier [481] (curve 2), Eucken and Neumann [406] (carve 11), and Smith [1327] (curve 12), but is entirely at variance with the data of Konno [778] (curve 3). These last constitute the only set of measurements in the range 380 to 850 K, but have been considered doubtful. There is dearly «strong need for further determinations of the thermal conductivity of pure antimony from room temperature upwards, ‘The thermal conductivity of molten antimony has been measured by Konno [778] (curve 8) and by Dutchak and Panasyuk [383, 384] (curves 23 and 24), whilst Mardykin and Filippov [884] (curve 31) have derived values for the range 1170 to 1430 K from thermal diffusivity observa- tions for a sample of very low purity, 96.8 percent anti- mony. The proposed curve has tentatively been drawn to fit the later of the two determinations by Dutchak and 4 Phys, Chom. Raf, Dato, Vol 3, Sopp. 1, 19741-50 Panasyuk 383] (curve 23) as these had indicated the thermal conductiviry to increase by 55 percent on passing from the solid to the liquid phase, a change in fair agree- ment with that of the electrical conductivity. The curve is also expected to lie above that of Mardykin and Filippov ‘owing to the low parity of their sample. ‘The tabulated values are for well-annealed igh-purity Taste 4. Provision HO, POWELL, AND LILEY ‘metallic antimony. Those above 100 K are provisional values and are considered accurate to within =+15 per cent of the true values at moderate temperatures and +25 percent near the melting point and above, Values below 100 K are merely typical values for the lowe temperature thermal conductivity of high-purity poly: crystalline antimony. thermal conductivity of antimony (Temperature, 7, K: Thermal Conductivity, by W em"tK-4) Solid Polycrystalline {The values for wellannealed high Liquid 0.996 0.202 0283, 0.267 | 1073.2 0.255 0.267 0.270 ome 0.2808 0.244 0.243 0.235 0.236 0219 0.213 0.199 0.195 0.186 0.183, 0.116 ot 0.170 0.168 ost 0.167 0.167 0.167 purity antimony, and thoee blow 100 K are merely typieat values, *Estrapolated. 4. Phys. Chom, ete, Val. 3, Sopp. 1, 1974THERMAL CONDUCTIVITY OF THE ELEMENTS > "aunvuadnaL eo dO ALIAL IT ANOWLLNV Ect JLONGNOD TWWHSHL + 11 MOM ALIALLONNOO TWHUSHL 42. Phys, Chem. Ra Dota, Vol 3, Soppl. 1, 1974HO, POWELL, AND LILEY 52 ‘omsiy oy unoys 20M « 22°19 50 dsmy w ye pamsvou AUS wuoode oAogE 94L, os 1 ase [WA ‘wos Gert ST ors a [EA ‘worn us 1 wer [HA oon as jie) eer [WA twee pert 4ST ta see 1 ater SFA oo ETT OT ly set SHA ‘oer set cay Ss yo Or € 20 0 pm uDurjoods BuyTEAIS Ao veat ——*0 “awounoy pu -y “ueyons 907 TT YO Or = 902 PuE 80° Seton 3 £2 of uimpow GEA UoUuoods ou TeyRSIOAIou yest “0 Yumeumay por “y ‘ujonx SOF OT feet “0 Tuuwumay pu -y ‘um|INE 90y =syoancod 28 Ayaponpuoo BORON? Foor °0 Twmwumoys poe v ‘um{OnE 90y "48 seer W'S “SPoOM PUT HD ‘omMA TosE £95 60'S SRT HS SPOON PUY HD ‘mW Toro 145 ScI-0's-Tacer W'S “SPO PUE'-D ‘oN Teer 9 Te eee ooter SCT Baaqea804, soe-ser Teta ete Teter ‘woneamieoa Bi sauasusy pur “suopeomrondg * Guaosod y5p2m) wopodwuog ‘wousodg Bun paen aK Ghouny og “ON proswex “duo (pA, eum NOLLYIHOANT LNUMIWASVANE GAY NOLLVZTUALOVEVHO NUNIOAS ~ ANOWLLNY 40 KUALLOAUNOD TRUST “S STEVE 2. Phys. Chom, Ref. Date, Vol. 3, Suppl 1, 197883 THERMAL CONDUCTIVITY OF THE ELEMENTS ‘muaoods osoqe of Jo unt zoqoay Breuot & 996T “ AN Mma poH eK SeRIOING, 9 user “aca moeueg pom “Tex “yoGoINL 1 sure 1 ‘a “wpmmy por “5 ameaINes “a"V"9 ‘4 poryddns Auowyae 3 305 .-w10 yx op0s 18 topuod moay posvoxd woUIDEdS fuo-est = TeteE “a “NOFoUMON PEE “D “ROMTHOD aaa ‘vonwaiised (a) ‘mammioy pur ‘suorEo{Toeds *(ywozedseBTOM) worsOdNOD Tuompodg uy pen OR (none parse “duo “PM (onsns00) NOILVHWOINE LNAICEUASVEN ONY NOLLYZIUELOVUVHO NAWIOS ~ ANONLINY 40 XLIALLONGNOD TYNUALL. “@ STAVE 41. Phys, Chem. Rat Date, Vol 3, Suppl. 1, 1974154 Ho, POWELL, AND LILEY Argon ‘The thermal conductivity of argon in each physical state is discussed separately below. Solid [Available data on the thermal conductivity of solid argon includes the work of Dobbs and Jones [365], White and Woods [1352], Berné, Boato, et. al. [148 170], and Krupskii et al. [792 791] while some calculations and correlations have appeared [696, 751, 183, 131, 752]. Most of the above results have only been presented in ‘xaphical form. Comparison of the available data revesls reasonable agreement at temperatures above about 10 K and severe disagreement at temperatures lower than 10 K. In the region from 5 to 8 X an order of magnitude difference exists between the [1852] and [148] data. Such differences are probably produccd by structure variations caused by different impurity cortent, although further high precision experimentation is ne:ded to confirm this supposition. The disagreement at the higher temperatures has also been ascribed by Krupskii as being due to this factor. ‘The provisional values were obtained from a large scale plot of the available information and were not generated {or temperatures below 8 K due to the experimental uncer tainty. From 8 to 20K the uncertainty may be as much as fifty percent, the uncertainty gradually decreasing with increasing temperature to ten percent at the highest tet: peratures tabulated, Due to the almost complete absence ‘of tabulated experimental data, apart from the Krupskii results, no departure plot is presented. Further experi- rental studies aze required to resolve the large discrep: fancies between the results of Krupskii and others. While the general trend of the Krupskii results appears to be reasonable, the adoption of those results as reference values would not enable any estimates of thermal conductivity maximum or values for temperatures below the maximum in thermal conductivity to be made using the White and ‘Woods values. It was felt that, at this time, the retention of the former choice [1420, 605, 844] of the White and Woods values as the basis, long with other results where appropriate, would provide a means of tabulation of values over a larger temperature range than would otherwise be possible, even at possibly reduced accuracy. Further experimental studies seem urgently needed to resolve this difference. Liquid Five experimental works were located on the thermal conductivity of liquid argon. Keyes [748] made measure- ments in a coaxialeylinder apparatus near saturation con ditions at three temperatures from 87 K to 112 K, The measurements of Uhlir [1447] were made in a coaxial cylinder apparatus covering temperatures from 86 to 150 K and pressures up to 96 atm. with an uncertainty reported to be from 0.5 to 2.5 percent. Other measurements for the Tiquid and gaseous phases were carried out in a coaxial cylinder apparatus with an accuracy of two percent, by 2. Phys Ch Ret, Dato, Vol 3, Suppl. 1, 1974 Ziebland-Burton [229, 1612], from 93 to 151 K for the liquid phase with pressures to 120 atm. Bailey and Kellner [105] also used a concentric cylinder apparatus from 90 to 300 K and pressures to 500 kg/em!, while Ikenberry and Rice [654] made measurements from 91 to 150 K for the compressed liquid. In the initial analysis, values for saturated vapor pres: sures were obtained from graphical extrapolation of the data of both Uhlir and Ziebland-Burton, no correction being made to the values of Keyes. The three sets of data for the saturated liquid were given equal weight and were fitted to a quadratic equation represented by & (Wm K-) = 0.216149 — 9.714328 - 10+ F — 1.070133 - 10 7, In arriving at this formula, values at the critical point vere excluded, This equation, considered valid in the temperature range from 80 to 140 K, fitted the above data with a mean deviation of 0.6 percent and a maximum of 19 percent, Recommended values below M40 K were ‘generated from the above formula. Subsequent to the initial analysis, the Ikenberry and Rice and Bailey and Keliner data were examined. Values for saturated vapor pressure were obtained, where required, by graphical extrapolation. No significant discord with the above formula was noted except for the 29 and 49 ke em? isobars of Bailey. These isobars were also omitted from Bailey's own evaluation in his figure 8. Vasserman and Rabinovich [1481] tabulated thermal conductivities for integral temperatures and pressures from a considera- tion of the [229, 748, 447, 1611, 654] data. Approximate values for the saturated state can be obtained from their tables and agroe with those recommended here to wi 5 percent for temperatures from 85 to 140 K. correlation favors the Ziobland-Burton data. ‘The values recommended below 140 K are considered to be correct within 2 percent. Values above 140 K were obtained from a largesseale graph. The experimental difi- culties inerease considerably in this temperature region and the recommended values at 145, 150 K and the critical point aze probably uncertain by as much as five, ten and twenty-five percent, respectively. As shown by Liley [845], values of the thermal conductivity of argon in the saturated states near the critical point ean be correlated with the enthalpy of vaporization. However, possible critical anom- alies have not been considered. Both the Bailey paper cited above and a further paper (104] show that any anomaly rust occur above about 150 K. The reliability of existing data is not felt, at this time, suficient to enable a detailed study of this region. Saturated Vapor No experimental data were found for the thermal con- ductivity of saturated argon vapor. The only information 1s of Owens and Thodos [1058],THERMAL CONDUCTIVITY OF THE ELEMENTS about 140 K the estimates are in fair agreement, the Uhlir values being intermediate, Above 140 K a wide variation in estimates exists, The more recent Vasserman and Rabinovich [1481] tables present correlated values for integral temperatures and pressures and not for the turated states. ‘The values were plotted on a large scale graph in which the Owens and Thodos values were adjusted to agree with the atmospheric pressure value at 88 K. The increase necessary at 88 K was linearly reduced for higher tem- peratures to zero at the critical temperature. Values ob- tained in this way were in excellent agreement with the Ublir values up to 125 K. Above 125 K they were lower than the other estimates. ‘The provisional values were deduced from the plot of the Owens and Thodos estimates. Based upon the agree: ment of these with other estimates and upon the uncer- tainty in the saturated liguid values, they should be accurate to about 2.5 percent below 125 K, fifteen percent at 185 K and twenty-five percent at and above 145K. ‘These error estimates agree roughly with the deviations of the few values obtainable for saturated conditions from [1481]. However, their values are invariably higher than those presented here. Bailey and Kellner [105] measured thermal conductivity-temperature isobars for 19.5 and 39.5 kg em From plotting their tabulated data, it was possible to draw a curve for the saturated vapor through ‘Tanvx 6, Recommended thermal conductivity of argont (Temperature, T, Ki Thermal Conductivity, k, W em? K-9) Soli Saturated vapor pr} mao} or | axa | or | exto as | 1.270" as | 12s | a5 | o.osse 8 oo 90} 1.201 | 90 | 0.050 9 46 9 | vise | 95 | c.06ee 10 sz | 100 | 1.02 | 100 | 0.060" w az | 10s | ro21 | 105 | 0.072" rn 22 | m0 | 0.963 | 10 | oor 16 1 | us | 0.903 | 15 | 0.082 1B 16 | 120 | 842 | 120 | 0-089 20 13.6 | 125 | 0.700 | 125 | 0.095 2 9.9 | 130 | os | 130 | 0.103" 30 78 | 1s | 0655 | 135 | 0-109" 35 6s | uo | 0592 | 140 | 0120" 0 36 | us | osis | 145 | 0.110" 45 s.1 | 150 | 0.404 | 150 | 0.19% 50 46 | asi | o.2seg | tsi | 0.25% © 38 70 33 0 3.0 Values forthe solid and saturated vapor are provisional ‘Estimated o extrapolated, {Pseudo-critical value. Taste 6, Recommended thermal conductiv of argon Continued (Temperature, T, K; Thermal Conductivity, Wem! K-!) Gas (Ar Tat) T J axio| 7 | exo! fo 0 ° 6 6 100 |o.06s2 | 50. | o. nio foconis | 460 | 0 129 [0.0779 | azo |. 130 [o:o009. | 400. | 0 140 |o-0a98 | 480. | o. 150 Jo.0957 | 00 | 0.2638 | 850 | 0.281 | 2000 | 0.652* 160 fo.i016 | 10 | 0.268 | 860 | 0.387 | 2100 | 0.717* 170 |o.1076 | 520 | 0.272 | a70| 0.390 | 2200 | 0.741 eo fo.r1s1 | 530 | 0.276 | a80 | 0.398. | 2300 | 0.766 190 fo.11a8 | sao | 0.280 | 890 | 0.396 | 2400 | 0. 750° 200 0.1244 | 550 | 0.283 | 900] 0.398 | 2500 | 0.215 210 |0.1300 | 560 | 0.287 | 910 | 0.401 220 (0.1355 | 570 | 0.290 | 920 | 0.404 230 j0.1409 | 580 | 0.298 | 930 | 0.407 240 (0.1462 | 590 | 0.297 | 910 | 0.410 0.301 | 950 | 0.413 0.203 | 960 | 0-416 0,308 | 970 | 0.118, o.s1 ) 980 | 0.421 0.315 | 950 | 0-326 0.819 | 1000 | 0.427 0.322 | 1050 | 0.481 0.225 | 1100 | 0.454 0.329 | 1150 | 0.468 0.333, | 1200 | 0.481 50 0.2013 | 700 | 0.236 | 1250 | 0.495, 260 [0.2059 | 710 | 0.889 | 1300 | 0.508 ‘0 0.2103 | 720 | 0.343. | 1350 | 0.521 ‘0 [0.2167 | 730 | 0.346 | i400 | 0:525° 390 40.2190 | 740 | 0.349 | 1450 | 0-548 “Eaimated or extrapolated. the isobaric values at saturation temperatures. The values 0 obtained are higher than those of [1481] or suggested. here. Experimental measurements have een reported for the thermal conduetivity of gaseous argon for temperatures between about 90 and 1173 K and many correlations and calculations have appeared, the more recent extending to temperatures well above 15 000 K. While our earlier tables recomended values to 10.000 K [843, 1420, 608, 844), the present work has limited the selection to an upper 4. Phys. Chom, Rf. Dot