‘Thermodynamic Properties of Hellum 4 from 2 to 1500 K at Pressures to 10% Pa Robert D. McCarty CrroseniesDisision, Intute for Base Stondords, Notional Bureau of Stondarde, Boulder, Caleredo 80302 “Tebular values of deny, itera energy enthalpy entopy est capacity and sped of sound or liquid and gaseous hellom are presented for temperatees from 2 t0 1500 Kelvin at preatres fam 110% 10'10 1.0 10% pascal Disgrams of temperature vs entropy ae alia given. The properties pre seated ae ealeulated from an equation of sate which was Sted to experimental P-P-T and wher thermodynamic deta fm the worlds Hteratare The equation of sate was tiled to thee tan theee separate regions of pressure and temperature. The regional equations are forced te jin smeothy at the preconceived boundaries, Extensive comparisons between the equation of slate and experimental data have been made, and evition plots are presented. A particularly careful determination of the ‘cond vir coefiient over hell temperature range 2-130 kelvin presented. Te Joue-Thamson Inversion curve hasbeen calculated and comparisons ade with other sources. Equations forthe den- sty of the satrated guid and vapor ae Included es well as an equation which represents the 1958 Ilion vapor premure temperature scale Key words: Creal pont ertcally evaluated dat; embalpys entropy: equation of state: heiom: Imeral energy: Joule Thomson eoeficient: lambda line: meng line; P-Y-P; specie heat vapor pressure; speed of sound viral coeicient Contents fh vsutse 94 0.2. Specific Hest inthe Critinal Regians 948 2 Survey ofthe Liteatufes ccc 925 9.3. Speed of Sound in the Critical 3. Vapor Pressure.. 925 Region. 980 4. 10. Thermodynamic Properiea Near the Lambda ne ce 950 5 928 11. The Melting Curve. = 951 6. The Equation of State. 933 12. Joule-Thomson Inversion Curvesenewnense 952 6.1. Estimation of the Coefficients for the 18, Referenees sme 933 Equation of Stat... 934 Appendix A. TemperatureEstropy Diagrams. 985 62, Estimation of the Coeficients for Appendix B. Symbols... 958 Region I of the Mul ‘Appendix C. Conversion” Factors and. Fixed Equation of States... 935 Points. 958 63, Estimation of the Coeficients for Appendix D. Thermodynamic Properties of Co- Region I ofthe Muli-Funtion Eau: existing Liquid and Gaseous tion of State. os Phases of Helium. 959 6A, Estimation ofthe Coefficients for pendix E, Isobarie Tables of the Thermody. Region IIL_of the Multi-Function samc Properties of Helium4...... 963 Equation of State 995 oe eee Poe 2 cael “Thermodynamic Properties Rela- ws ot carey cen 8. Evaluation of the Equation of State and Br ee ee eee ee Lo Differences Between Experimental Meas- 8A. Comparison of Experimental and Cab: iors ieee Tosaperaiare are 1055) culated P-V-TDat. 936 Helium Seale CA 8.2, Compansons of Experimental” and Deviations Between Equations Gj), @) and Galealated C, Daa oa oper eben oon 83. Comparison of Experimental and 2.23). a Calculated Speed of Sound 942 4, The Second Vial Goefficient from 2 to 10 8.4, Comparison with Other Correlations... 943 K. The Solid Line is from Equation (8)... 929 8.5. Discussion of Error... 943 The Second Virial Coefficient from 10 to 9. Thermodynamic Properties in the Critical 48 K. The Solid Line is from Equation (8)... 929 Region.. 345 6. ‘The Second Virial Coeficient from 48 to 9.1, PY-Tin the Critical Region cus 946 200 K. The Solid Line isfromEquation(@).. 930 7. The Second Virial Coefficient from 200 to ‘ra ae ant i eon etl Sa ee 1500 K. The Solid Line is from Equation Scien stm en ina ar Bhevrnnn - 930 923 4. Ph Chem, Re. Dot Vl. 2. No. 4 1973one 10, nL. 2 8 rs 6, 16. 1, 18, 19, 2. 2, Second Virial Coefficient from 2 to 1500K. Schematic p-T Diagram for Helium Show: ing Regions Repovemed by Diferent Equations... Comparison of Densities from Wiebe, et al. [77] with Equation of State. Comparison of Densities from Canfield, et al [9] with Equation of State Comparison of Compressibil from Briggs, et al. (8) with Equation of State. ‘Comparison of Densities from White, etal [76] with Equation of State... ; ‘Comparison of Densities from Sul with Equation of State... Comparison of Densi Canfield {30} with Equation of State. Comparison of Densities from Lounasmaa 50] wialve Eguastou of Sete acne Comparison of Densities from Edeskuty and Sherman [17] with the Equation of State. Comparison of Densities from’ Giassiord and Smith 25] with the Equation of State. Comparison of Densities from Grilly and Mill {20} with he Equation uf Statesenn Comparison of Densities from Kierstead [47] with the Equation of State . Comparison of Pressures from Keller (411 with the Equation of State... Comparison of C's from Lounasmes (50) with Values Caleulated using Equation m 28. Frank {16} with Those Caleulated from the Equation of State Camparison of Speed of Sound Data from Gammon and Dovslin [24] wth Vales Cal colated fom the Equation of State Dilforoncco Between Correlation ofthe ‘Thermodynamic Properties of Helium by Mann (531 and this work. Estimates of Uncertainty in Densities Cal- culated using the Equation of Stat. Estimates of Uncertainty in Entsipies Calculated using the Eauatin of State... Endnotes of Unestainy iw Gr's Caeulved using the Equation of State A TemperatureDensity Plot of Roach’s Experimental P-Y-T Data for Helium inthe Near Critical Region... Ceinthe Near Critical Region of Hive |. Cyn the Near Critical Region of Helium... 2; Speed uf Suund for Helis fn the Peas Critical Region.. . Pressure Coefficient, (OP/97)p, for Helium ‘Across the Lambda Line Speed of Sound for Liqui 4. Phys. Cham Rel eto, Vo 2, No 4, 1973 932 933 936 937 937 97 938 339 940 oa oa a2 943 948 94s 946 947 43 949 949 950 9st 981 35, Joule‘Thomson Inversion Curve Calculated From Equation of State. 386, TemperatureEntropy Disgram for Helium for Temperatures from 2 to 12 K. 3, Temperature-Entropy Diagram (or Helium Between 10 and 110 K.. 956 38. Temperature-Entropy Diagram for Helivm Retween 100 and 1500 K. 987 ‘a. List of Tables nae 1, Constraints on Equation (1), 96 2, Constraints on Equation @. 926 3, Coefficients for Equation (1). 926 4. Coefficients for Equation (2) 926 5. Coefficients for Raquations (#) and) 928 6. Coefficients for Equation (8)... 932 7. Second Viral Coeficients for Helium... 933 8. Coefficients for Equation (17) Reg 934 9 Coefficients for Equation (17) (Negion I... 934 LL, Coefficients for Equations (1) and G2)... 12, Coufficionte for Equation (23). 13, C, for Helium Along the & Curve... 4, Coefficients of Equations for Py and ps... 950 15, P-o-T of the Lambda Line... 950 16, Parameters for Equation (8) 952 17, Melting Line for Helium. 982 18, Joule-Thomson Inversion Curve... 92 29, Comparison of Joule"Tlvaoun Coeffcien at 1 Atmosphere. 1, Introduction ‘The investigation ofthe properties of helium has been tne of the more popular endeavors in science since it ‘was discovered in 1868, A large part of this interest can be attributed to the spectacular superfuid propenies of helium. This work does not include the temperature ‘and pressure range at which the superfluid properdes ‘of helium occur. In recent years the application of helium as a reftig ‘erant and other technical uses have accelerated interest in the thermodynamic properties of helium, Although many investigators have published data for the thermo- dynamic properties of helium, in most cases these in- vestigauons have been I temperature and pressure w! interest to the individual. In no case has there been tabulations published which cover the range of tempera- ture and pressure of this work, In addition, where pub lished data do overlap, there is usually a substantial disagreement which, in many cases, makes the transition, foun oe data set tw anther quite eificult Fo the aa For these reasons this laboratory undertook the job of critically analyzing the existing thermodynamic. di for helium and ereating from this analysis « new set of data which covers a more extensive range of pressureTHERMODYNAMIC PPOPERTIES OF HELIUM 4 and temperature than was previously available in Single document. 2. Survey of the Literature A search of the world’s scientific literature began with ‘2 computerized search of some 40,000 articles at the Cryogenic Data Center of the National Bureau of Standards at Boulder, Colorado. This search produced 684 references. The literature search was periadically up dated so that current data were continually assimilated. ‘A bibliography by Barieau (2]' contains all the ref. erences he could find with experimental P-/—T data for helium 4. The bibliography contains 103 references. In addition to these P-V-T data there are many references in the literature which contain other useful experimental dintn ech ae cpecific heat, epaed of eanind, and many others. All of the articles which contained appropriate experimental data were considered for possible con- tribution to the correlation of the thermodynamic proper- hes ot nem. Over the past few years, a number of other corre tions of properties of helium have been published. Most af thoss hav heer af limited seape and anly ane [SAY hae included the two-phase region. Since Mann’s correla- tion many new experimental data for helium have been Published. Other more recent correlations are [72] and {05}; both of these correlations are restricted to tempers: tures of O°C and above. The agreement between the above correlations and this one is discussed in detail in ection 8.4. Many of the references in the literature cont ‘equations of state; however, none of the equations of state were adequate for the scope of this correlation, In the last ten years or $0, the continued increase in usage of the high speed digital computer has made the equation of state approach to thermodynamic properties ‘more and more attractive. Many equations of state have been proposed in the literature. In every case the author hhas found these equations do not represent the existing data, or theory, adequately at one point or more on the P-VAT surface. However, this is not to say that the single equation of state presentation is not useful. In the majority of cases to accept these not affect his particular calculator an alternative is presented, A single equation of state is presented, with three different sets of adjustable param- eters. The three sets of parameters were determined from data of limited temperature and pressure ranges and the three ranges were joined smoothly at their ‘common boundaries by various means. The tabulations given here were calculated from the set of three equations. Tn the ease of helinm, for s number of reasons, itis difficult to assess the general overall adequacy of an ‘equation of state. First, the entire normal two-phase 25 ascous-liquid coexistence region takes place in a span of about three Kelvin. This amplifies by at least an order of magnitude the importance of temperature measure- ‘ment errors, Second, these temperature measurements are made in an extremely dificult portion of the temper: ture seale, Le, 2 to 5 K. Third, when determining on equation of state for @ fluid, it is important to have accurate data from a single source covering both the easeous and liquid region from the triple point to about twice critical temperature. For helium there is only one such set of dat such source of data for this region it is make extensive comparisons, In the regions of pressure and temperature where there isa small overlap of these data with others, there is about a 1.5 percent disagree- wy at high density which fo dioastrowe f the determination of an equation of state. ‘In general, the three regional equations of state repre sent the existing experimental P-V-T data to within the ‘accuracy of the data in their respective regions with the exception of the critical region. The critical region of hhelium has been investigated quite extensively and is treated here completely independently of die balance of the P-V-T surface. mecnt ind 3. Vapor Pressure In recent years most of the experimental work at temperatures between 2 and 5 K has used the 1958 hhelium temperature scale {7}, and when older data which are based on other temperature scales are. com pared and correlated with the newer data conversion to the 1958 scale has served to establish a common temper: ature scale. Some af the moat recent experimental workers have used other temperature scales but these are almost always given in terms of “corrections” to the 1958 scale. Therefore, it seemed worthwhile to create an equation or equations to reproduce the 1958 helium vapor pressure temperature scale. These equ tions are: nea Sane, w where T is in kelvin and pressure is in micrometers of Hg (1 um He= 133.322 P,). The range of validity for 4 (1) is 2.172 to 5.1994 K. For the temperature range of 05 to 2.172 K the equation dare, e was determined. Equation @) is in the same units as eq). Equation (1) was fit to 611 smoothed points from [7], taken every 0.005 K betwoon 2.172 and 5.1904 K. In addition to the 611 data points, six constraints were imposed on the least squares estimates of the param- eters. These constraints are listed in table 1. The coefficients (4;) are given in table 3. 1 Phy. Cham Ba to, Vo. 2, No.4, 1973,‘Tame 1 Contains equation) > 31000 wm He at p 110000 urs 4a5 > neo T= Sams ajar = 92960 wm Heke T= 217 rar = caw) nr law ‘iar = 26% r= 52 Taste? Constants on equation 2) P= sro ym gat T= DIRK aan = 92340 ym HgK-* at T= 207 Tante’, Coeficiente for equation () ROBERT D. MECARTY asus 4 Coeficiente for equation 2) B= sioase4i70 x 10" By = ~3:3075130856 x 10° B= 1.3860673801 x 10¢ BS 49ssa0T4s6 x 10¢ By —Sszssesud x 10. B, = s.zsniz¥ans x 108 = canons > 10+ By = ~Ssa0e567 x 10° Bu = 190s01506 x 10 By = —Lensaea5s9 x 108 Bh = Seas deviation in pressure for either equation (1) oF (2) is 0.02 percent. The maximum dé ah = SSE fe = Yanotsta x 10: 4s = —Lewormises > 10° 4 = “Vises x toe i. -Sisaussome % 10. te = 16621856508 x 108 4 = Sasa x 108 f= Soenasaxisn x 108 ‘ho = as9sa04i08 x 10¢ Equation @) was ft to 383 points taken every 0.005 K from 0.5 to 2.172 K [7 In addition tothe 333 data points, two constraints were imposed on the least squares estimate of the parameters (see table 2). The coeficiente (B) for equation (2) are given in table 4. The maximum K DIFFERENCE, ve tion in temperature predicted by either equation (1) or @) is 0.0001 K (see figure 1), Deviations are from the 1958 helium vapor pressure temperature swale ay defied a [7h ‘Temperature scales both practical and thermodynamic hhave always been a fascinating subject to scientists and engineers. The 1958 helium vapor pressure temperature seale (hereafter referred to as the Tis scale) i no excep tion, Cataland and Plumb [10] heve been engaged in a program of tecoustie thermometry which includes the liquid Whelan temperature range from about 2 Ko Ad ‘ther paper by Rogers, et al. 70] makes an interesting comparison of various experimental temperature deter rinations with those calculated from the Tss. Figure 2 is from Rogers, et al. (70). Two things are evident in 2 PRESSURE, MICRONS Froune 4 Phe Cham. Rel, Dato, Vl 2. No 4.1973 We Devinionsbetwees the 188 bela temperature ule snd equations (1) and (2) aeTHERMODYNAMIC PROPERTIES OF HELIUM 4 7 ‘©. NSL, Consort Volume Gos Thermometer = ==NBS, Accustic TEMPERATURE, K icine 2. Differences between experimental measarements of temperature andthe 1958 heli seal figure 2. First, the Tys scale assigns temperatures lower than the thermodynamie temperature and, second, the temperature determinations have a seatter uf about five millidegrees. On the basis of the above referenced work it was decided to adjust the Tss scale by T= Top + 0.001 + 0.002 Tye, ® which raises the temperatures assigned to the normal boiling point and the critical point of helium by about nine millidegrees and 11 milldegrees, respectively. [After equation (8) was decided upon, more recent work by Cetas and Swenson (11,12, 13], using a paramagnetic salt experiment to determine absolute temperatures, suggests that the adjustments from equation (3) are too large by about 4 or 5 millidegrees for temperatures approaching 8. ‘The measurements by Cetas and Swenson [11, 12, 13] have a greater precision than those in figure 2 by at naet a feotor of 2. Kieretead 4S, 46] reporte expert mental P-V-T measurements in the eritieal region of hhelium which tend to confirm the Ts scale discrepancies reported by Cetas and Swenson [11, 12, 13}. Since the difference between the adjustments to the Tss scale siven by equation (3) and those proposed by Cetas and ‘Swenson (11, 12, 13] would have a significant effect only at temperaturer in the near critical region and since the equation of state is admittedly not accurate in the critical region no further adjustments to the temper- atures were made. One of the fixed points of the Ts scale is the critical point. The value of pressure assigned to the critical point is 1718 mm of mercury reported by Kamerlingh Onnee [36]. Kamerlingh Onnes states th tion of that value (1718 mm) was made by observing the the determina disappearance of the meniscus and that this could be determined to no better than one centimeter of mercury. Some of the more recent messuremients of the critical pressure (19] and {67} show that 1718 mm of mercury is too high and that a better value of the critical pressure ‘of helium is 1706 mm of mercury. Tf one takes as a best estimate the critical pressure of helium to be 1706.1 mm Hg and solves equations (3) and (1) for a temperature at this pressure, a critical temperature of 5.2014 results. Before leaving the helium vapor pressure temperature scale there is one other disagreement between Tus and recent experiments that bears mentioning. Kierstead [46] made measurements of the slope of the critical isochore for helium 4 and found the slope (@P/37) at Pe and 7. to be 1289.2 torr/K as compared to 1267.0 torr/K as qaven by Tyg. Since the 54 scale as based on an analytic representation of the behavior of the vapor pressure of helium this disagreement is not unexpected. Fer vorenelrtio repens pressure see equation (8 fon of the halite weper section 9. 4. Density ofthe Saturated Liquid and Gaseous During the course of the correlation of the P-V-T data for helium, equations for the density of the satu- rated liguid and vapor phases of helium were developed as aids in the analysis. These equations have not been used to produce any of the numbers which appear in the numerical tabulations. They are presented here as a eonvenience for those who might have use for such equations. They do represent the experimental densities to which they were fit to within the precision of those densities. ‘The analysis of the experimental density data for the saturated liquid and gaseous phases of helium 4 included ‘2 number of soureor, references [19}, [20}, (21), 22), {23}, (42), (4], [54], and [67]. Most of these data are for the saturated liquid and are of limited temperature range. The data of Roach [67] and of Edwards [19, 20, 21) are mainly concerned with the critical region. Kerr's [42, 43, 44] experiments were mainly concerned with the region of the lambda transitions, G42] dues contain dats up wo 44 K, however, while 1 generally agrees with the more recent data of el Hadi (22, 23] it contains considerably more scatter. The paper by Kerr and Taylor [44] containe neve data only for temperatures of 2.8 K and below. The old Leiden data by Mathias, et al [54] were not in agreement with the newer data, The data by el Hadi (22, 23] cover the widest range of temperature and seemed to be the best suited for the formulation of useful saturation equ tions. Therefore, the el Hadi (22.23] data were used in the leact squares determination of the following two equations, The density of the saturated vapor is given by pa pet FS = TIT ” ths Chom, Ra, Dot, Vol 2, No 4, 1973928 ROBERT D, McCARTY and that of the saturated liquid pet 3 Sy TIT)", 6 where the density is in glem*, and the temperature is in kelvin, The parameters for equations (4) and (6) are siven in table 5, The saturated liquid data below the TaRLE S. Coeficens for equations (4) and() 056 elem 5 esaniwssszz xiv | 0.00868 lem? = SaerTan6 10-1 s.= Laasi9nted s.= —B3somees7 a, = — Loveizas6te lambda temperature were omitted from the least squares fit. The average deviation for the liquid equa- tion was 0.027 percent with a maximum of 0.1 percent fan der 0 JX Sires Ltrs 65) / \ ne i rm patina 8 PERCENT DEVIATION , p TEMPERATURE , K Ficume 3. Devistons between equations (), 6) and (17) and the ‘sven tn by el ad (22, 23) of the vapor equation is from 2.2 K to the critical point and the range of the liquid equation is from the lambda point to the critical point. Ifthe best possible accuracy sired between 5 and 8.2 K. equations (31) and (82) should be used. 4 Phys Cham Re. Data, Vel 2, No 4 1979 5. The Second jal Coefficient ‘The virial eauation of state for sases is often aiven in form of RT(1 + Bp + Cp*+ Dp+. © Although this is one of the earliest proposed equations of state for gases, it is still extensively used and the virial coefficients (B, C, D, ete.) aro of epecial interest to theoreticians. The second virial coefficient, B, is the easiest to obtain, both from experimental data and from theory. Therefore, the experimentally derived B is of particular interest to theoreticians for investigation of the intermolecular potential, Most of the existing experimental P-V—T' date for helm have been published in the bterature in 1s0- thermal form, and in many eases these articles include rial coefficients derived from the experimental iso- therme. Unfortunately, the value of « virial confhoiont derived from a given set of data is highly dependent ‘upon # number of factors related to the method used in ‘obtaining the virial coefficient from the data, Since the ‘method slmost always varies from one investigator tc another and since the details of the method are often omitted, itis inappropriate, in most cases, to compare ‘ne author's visial cueffcieuts direedly with anuthet’s, ‘There seemed to be enough low pressure isothermal P-V-T data in the literature to warrant a careful invest sation of the second virial coeficient. Each experimental isotherm found in the literature was treated in exactly the same way. Least squares estimates of 4, B, and C, were determined from the equation PV=A+ Bp+ Cpt o ‘The estimating procedure began by least squares fitting equation (7) to a few of the lowest density data points and repeating the fitting procedure many times, esta re time adding anather date pnint of creasing density until the virial C hecame statistically significant. As soon as the virial C hecame significant, the last few data points (highest density) were removed and a final fit made. The point at which the third virial began to contribute to a detectable degree is dependent ‘upon the random scatter of the data, Theoretically, it would contrilte a al ith error present. However, if the contribution (of C) less than the random error in the data, it must not be included in the estimating equation because it will tend to compensate for random error by shifting the curve towards one or more of the data points with large random error. This, in turn, will influence the value of the B, because of the strong covariance which exists between the B and C. A total of 148 isotherms were analyzed in this manner. The partial results of the snalysis are chown in part in figures 4,5, 6, and 7. Even when each experimental source of data is compared on ite eww nv seseinnTHERMODYNAMIC PROPERTIES OF HELIUM 4 920 5 8 8 sessom 942) o Kistemaker and Keesom (1946) ® 29 ‘y= Keller (1955) ‘@- Keesom and Wolstra (1940) a) Peo Le) TEMPERATURE, K ‘Fiount 4 The second vr coeficent rm 2t 10 K. The sol line i from equation) 02, © Zelmanov (1944) ‘8+ Van Agr ond onnes (1925) ‘8 Beenakker et.al (959) ‘= Boks and Onnes (1924) = Nijhoff et.al. (1927) '* Dobrovoiskit and Golubev (1966)_| 2+ White et.ol, (i960) ‘== Keesom (1942) ‘SECOND VIRIAL,£/mot 30 40 ‘TEMPERATURE, K roume §. The second vr coefficient rom 10 $8 K. The solid ine i frm equation) 4 Phy. Cham Ra Dat, Vo 2, Ne 4, 1973930 ROBERT D. McCARTY a 3 & 2 8 ©: Boks and Onnes (1824) # - Perring (3g) B og dobro od Chup 2 Haba on Ste (968) 2: Gontld eto (956) White etal (1960) © Helborn and oft (1324) Keesom (1342), ‘5° Gammon and Dousiin a i370) 50 To) 150 200 TEMPERATURE, K FLOUNE 6. The second vial coeicos from 48 to 200K. The si ine fom equation ~ Boks ond Onnes (1924) Holborn and Otto (1924) | 4 Holborn ond Otto (1825) Keesom (1942) aoeem ana Van Santen (1033) Miller eto (1961) Yntema ond Schneider (1950) + = Gammon and Dousiin (1970) os | 1 1 1 °°%00 400600. 800.1000 a 1400 TEMPERATURE, K Fioune 7. The second ial coeficet fom 80040120 K. The sil line i frm eguton SECOND VIRIAL, 4/mol 4. Phy. Chem Re, Dat, Vo. 2, Ne. 4 1978THERMODYNAMIC PROPERTIES OF HELIUM 4 oat ‘4 common base, small differences in the data tend to influence the virial coeficients. Even though the results ‘of the analysis were less than expected from the stand: point of the second virial coefficient, other information which came out of the analysis proved to be very valu: able, The estimates of the variance of the B's from each source of data provided an absolute and a relative est- mate of the experimental random error in the data. In equation (2), appears as a parameter to be estimated by the least squares procedure, which corresponds in the virial equation to RT, Using a commonly accepted value of R it was then possible to place confidence limits ‘on the temperature of the Isotherm. In some cases the temperature assigned to the isotherm by the original source was outside the confidence limit by a wide mar- in, indicating same aystematie error in the ment of the P-Y-T data, In a few of the ca systematic error was detected, the error could be as signed to the temperature measurement and therefore corrected. If this was not possible, the data were climinated from further consideration unless they hap- pened to be the only data available for that temperature and pressure range. ‘On the basis of the information gained from the iso- thermal analysis just described, particular sets of P-V-T data were chosen for the final estimation of B. ‘The data of Canfield, et al. [9] were chosen for the ten perature range of 133 t0 278 K. The data of Sullivan (71), were chosen for the temperature range of 70 t0 120K. ‘The data of White. et al. (761 were chosen for the tem: perature range of 20 to 70K, and finally, the data of Keller [41] were used for temperatures from 2.154 to 3.957 K. In the case of Canfield, et al, (9] and Sullivan [71], care was taken to exclude dats above densities ‘where higher order virals began to contsibute. In addition to the P-V-T data mentioned above, two “other kinds of data were fitted to equation (®) simultane- ously with the P-V-T data, The speed of sound data by Plumb and Cataland [66], which cover the temperature range between 2.323 and 20.051 K, were inserted into the ‘Av as well as the second virial eoemetent data of Ynvema and Schneider (79] which cover the temperature range of 273 to 1473 K. ‘The function chosen to represent the temberature dependence of the second virial coefficient is BU T)= S60, @ where 7 is in kelvin and B is in liters/mole. The tech nique of simultaneously fiting different kinds of data requires only that the different kinds of data being used are mathematically related and that the resulting equa: tions are linear and separable in the parameters being cetimated. For the second vital data equation (8) ws used. For the P-V-T data equation (9) was used. prat—10-$ ner, —@) where P is in atmospheres, kelvin, ‘To introduce the speed of sound into the least squares ‘equations, the virial expansion is again truncated after x order virial contribution is present the following relationships are vali liters/mole and T is in Wea Wp(+aP), (0) the speed of sound squared; W}— (Cr! Colreo AF is the idealeas speed of sound: a= 1/RT (2B-+4/3TdB|aT'+ sST+d°B/dT*); and P isthe pres sure. Since the second virial coefficient has the same value in either the pressure or density form of the vii equation of state, it is now possible to write equation (00) a» shore HP is Substituting for B from equation (8) into equation (11) results in G19) -¥ wfer$ asim +g 1512) (0.5-40)] t7H*-™], 2) where the units of P, T and R are the same as in equa tion (9). The right hand side of equations (8), (9), and (12) are now all linear in the by. In least squares estimating procedures a general equation is sometimes waitten a5 as) ‘The left hand side of equations (), (9), and (12) become the y in equation (13) and the term multiplying the by in equations (8), (9), and (12) becomes the X; in equation (1), Por more detail vt simultauevuy dave Qing see Hust and MeCarty [35] Since the least squares ing procedure mini- izes the sum of the squares of the deviations in the dependent variable (the y in equation (13)) with respect to the parameters being estimated (the by in equation (03) it is important that the variance in y be as nearly a eustant ao pyssible. This Is especlally irae In dhe exe of simultaneous data fiting because the independent variable differs in magnitude from one kind of data to the next; therefore the variance ig likely to differ more than when using only one kind of data. Fortunately, a transformation via a weighting factor will compensate for this to a large degree. Applying the weighting factor to equation (13) at the wil dita points gives WbXn, a4) where022 ROBERT D. MeCARTY as) ‘and 8 an estimate of the standard deviation in the rnth y. An estimate of &y, may be had from the precision of the experiment, If there are large random errors in fone or more of the independent variables of the X; in ‘equations (13) and (14), or ifthe random error is smal in 1X; but a small error in the X; causes a large error inthe Y, Le., ay/AX1 is large, then these random errors should ‘also be taken into account, This is easily done by ‘modifying the weighting factor to include ¥, miner 06) where the dz, , is estimated again from the precision of the experimental measurements and the (@0)921,0) is estimated by obtaining a preliminary set of parameters and using them to form the proper partial derivatives. Ideally, ifthe estimates are exact the variance of the ting weighted lnast ecptaree 64 would he ane, and would be distributed statistically as x*. An approximate rule of thumb is that if the variance ofthe fit is between 04 and 10, the weighting is probably quite good. Com- hnng the above procedures and the referenced data the estimated parameters for equation (8) which are given in table 6 were obtained. The number of terms in ‘equation (8) was determined by successively adding terms to the expansion until the statistical significance ‘of one of more of the coeficients was lost. ‘The error bars shown in figures 4,5,6 and 7 are est- mated uncertainties for 4 95 percent confidenve. The 80. oo ° second virial coefficients of Gammon and Douslin [24] shown in figures 6 and 7 were derived from speed of ound measurements. These data appeared in the literature after the parameters for equation (8) were determined. The agreement between Gammon and Douelin’s {24} virile and those from equation (8) ix ‘g00d, ranging from slightly less than 1 percent difference at the highest temperature to slightly more than 2 per- ‘cent at the lowest temperature, all well within the ‘uncertainties assigned to equation (8) Tawi, Conciente for uation = —Sa81s7IGoH x 10-7 b= 1 meng 1 by = Ladsoeabt 2 10-* be = Tarasenna x 10-* 2 = Susanne 10-+ be = —95stopasa0g > 10-+ b= soars by = ~S:smas B= 7 Steoonase The numberof sgnieantBigures ned for these and all ther tubuitins Is no way inditv af the aceurcy ofthe properties borer ven ie necessary o avoid emors duet compuaions found of Even more recently Kerry {9} reported second vinal coefficients of ~2.35, ~43.6 and ~79.0 cm'[mol for the temperatures of 20.2746, 6.6989, and 4.2240 K, respectively. Using equation @) values of 2.0. 44.0. ‘and — 80.3 were calculated for the above temperatures, respectively. Figure 8 shows the shape and relative size of the second virial for helium over the entire range of tempera- 005 SECOND VIRIAL, /mol $5 8 9 0 mSCOSC«D a a ee mo woo ‘TEMPERATURE, K Loum Second vil coeBicent from 2 to 1500 K 4. Phe chor. Rel Dat, Vol 2, Ne 4, 1973‘THERMODYNAMIC PROPERTIES OF HEUUM 4 oa ture from 2 to 1500 K; table 7 gives values of the second virials calculated from ea. ‘Tanue 7. Second vil efficent for hellams Tesp | 8x10 [Temp [axi0 | Temp | Bxim P| Ai | [Armee | TE? | ee 2 [ame | 2 |-osr | am | za s foie | a [our | a | aie 5 | “om | a | 356 | sso | ants 6 | =| a | sa | 40 | ans 7 | sas) 2 | sz | wo | 0 5 | “anos | 3 | e60 | sso | i950 w | san | x | ti | ao | ion n | Wis | | 202 | 60 | 103s B | Tine | @ | war | 30 | tote we | -inz | 2 | m.06 | 00 | rooe ws | 925} a | us | 90 | om i | <2 | 9 | a | 100 | sat 1 | =s9u | 100 | 1200 | tam | eee w | Wes | 125 | 1229 | 190 | m0 wy | W336 | as | i23e | goo | a6 am | -225| is | i230 | 1500 | aie 6. The Equation of State When the task of correlating the thermodynamic properties of helium was undertaken, one ofthe ultimate als was the production of a single equation of state which would adequately predict the existing experi ‘mental data from 2 to 1500 K with pressures as high as possible. As is always the case when trying to fit a large surface such as a P-V-T surface, part of the DINsITY, mist ‘TEMPERATURE, surface is compromised to allow even @ reasonable fit of another part of the surface. It was therefore decided after a long and tedious period of experime tion with, and development of, various mathematical rodels for the equation of state of helium that further cefte eloug the siuale equation uf state ayproseh would likely be fruitless. ‘The alternative to a single equation approach is obvi= ously a multiple equation approach which presents a problem of matching two or more surfaces at common boundary. Since the equation of state does do a reason- able job of representing the entire P-7—P surface, it ‘seemed reasonable to assume that it would do a hetter job for just part of the surface. Maintaining the mathe- ‘matical form of the equation of state from region to region aleo hae come practical advantegee. The matching at the boundaries is greatly simplified and the computer programming is minimal as one only needs to switch coeficients to go from one region to another. The form ‘of the equation of state used for helium is P= pRTT1+ Bib. Tp) +3 mygrT09-00 +S momen proce an sZagoorenn ageor Sn where P isin atmoscheres. pix in moles ner Ficume 9. Schematic pT dagrem for helium showing repone represented by diferent easton 4. Phys. Chem Rel. Det, Vo. 2, Na. 4, 1973934 ROBERT D. McCARTY kelvin, R= 0,0820658 I-atm/mol-K and B(b;.T) is given by equation @). ‘The P-V-T surface was divided in three regions ws shown in figure 9. ‘The thermodynamic properties for helium tabulated in appendices D and F have heen calculated using the ‘multisurface approach. All properties in region I were calculated using equation (17) together with the coeffi cients in table 8. Region I is defined to include tempers tures from 2 to 10 K with densities to the saturated liquid density for temperatures from the lambda point to the critical point, for temperatures between the critical temperature and 10 K, the limiting density isthe ertical density. All properties in region II were calculated using equa- tion (17) together with the coeficients in tables 8 and 9. Rezion I is defined to include temperatures from the lambda line to 10 K with densities from the boundary of region Ito the solid line. The transition between regions 1 and IL was accomplished along a line of constant temperature in the following manner, Prop (p.T)= Prop (Pramas 7) Region IT Region T a8) [Per (p.T)—Prop | Region I Region IL ‘This formulation insures continuity between the two regions; however, to obtain the best results in region II ‘equation (18) must actually be used in the least squat estimation of the coefficients for region Il. More vi be said about the estimation of the coefficients later this section, All the properties in region III were calculated using equation (17) together with the coefficients in table 10. Region III is defined to include temperatures from 15 to 1500 K with limiting densities determined by a maximum pressure of 1000 atm. In the temperature range of 10 to 15 K either regions Lor Il may be extrapolated upward or region IIT may be extrapolated downward in temperature); however, 10 insure a smooth transition in the tabulations a weighted average of the two extrapolations was used. The actual lone ware made ae fllowe: Prop (p.7)=Prop (p,T)05.=TV5 Region I or I +Prop Region II (p, TY (P-15.)15.] a) To insure a smooth transition between the various resions. equations (18. 19) were used for all oroverties tabulated in Appendices D and E, 4: Phys. Chom, Ret, Dota, Vo. 2 No.4, 1878 ‘TaBts 8 Coeficens fr equation (17) Resion D inom 14792568008 %10- ‘n= 3.7910190855 x 10-" Inga 3.250185877 fham 1380654009 x 10-+ hom 19618739286 ag 2 5085812088 10-2 p= LOSTTBTTISS mm 230560008 10 rin =~9 5726461065 10-* a= 49060810%10-" =~ 2.616004577«10-* = 84221685545 10 Slsoacnsa2 10 a. ogoestee9 x 10-* 1, 5096862619 > 10-7 689045 10-" astoagrt7 agora 10-+ “6235736710 ‘TaBts 9 Coeficiens for equation (17) (Reon I) fny= 2OWS0LITT10-* fnem 1.2aooseH8 10-2 sgn 20272909585 x 10-* sng 1721796521 10-" ime 8.S25460556510-> = 25169060255 %10-+ oem 3.2877709285 x 10-4 ses = 1058758078 % 10 nam 7996601200 % 10-3 em .9 3486656 10-* om 1 0Teseo0ss 10°" fv 2.688494827 «10 term 3.371685 10" ‘mem LA1S9016970310"" fem 14725690701 x10 mama LOMGISOISH 10% | mam as20854818 10-7 ‘Tame 10. Coeficiente fr equation (17) (Regan I) 3600775292 %10-* noe 2264769615 %10-" (= 18798505957 10-1 43590488810" = 408651108010 es=Sloasnsers 10+ fem 112765049 x 10-+ a= 1 2899888207 10-+ 7040s75125 Nam 14898145605 10-7 tum 903667HNDx10-" | Ram 83768074851 > 10-7 f= Loeeiseoeesi0-+ | y= LoTsa01218%10- fom 1712990566 10 io Sesazae 5 10-* fom 2ansioosas 10-1 | aym Sages 10-> —9.56a39100% 0-4 | nym LABAIB5568x 10-* 2SSM5592T1 10°" | tem 41TRITOILIX 10-7 | Vsex10- 46.1. Estimation of the Coefficients for the Equation of State In the ease of helium there exists in the world liters ture a large amount of experimental data which are nplieable to the develooment of an eavation of state. A bibliography of references to helium 4 experimental P-V-T data was published by Barieau (2). The list contains 168 references. "All of these data were considered; bomerer, waly @ selected set was used in the estimation of the coef cients. Many of these data are quite old and are not of sufcient precision to be of value in estimating the coef ‘ents of an equation of state, For the temperature rangeTHERMODYNAMIC PROPERTIES OF HELIUM 4 935 of 60 to 300 K the existing experimental P-V—T data seem to be adequate in precision, accuracy, and ‘quantity. For temperatures below 60 K additional experi ‘mental data would be highly desirable forthe determina tion of an equation of state. Such an experimental pro- ‘gram should include ae wide a temperature and pressure snge as possible, particularly below 20 K. In addition to the P-V-T needed, higher order data such as specific heat and speed of sound data would be extremely valu ble. Using modem mulidata least squares estimation techniques, the inclusion of data such as specific heat data together with the traditional P-V~T data is neces ary if the roeulting equation of etate ie to be uoed for calculating higher order thermodynamic properties. The other most important factor in determining an equation of state isthe availability of accurate data from a single source over a wide enough pressure and temperature range. As is the ease for helium, data coming from several different laboratories often vary systematically from one eource to another although each individual source is internally consistent. The discontinuities ‘caused by such systematic disagreement are very detrimental to the results of least squares estimation techniques. 6.2. Estimation of the Coefficients for Region | of the ‘Muli-Function Equation of State ‘The coefficients for equation (17) given in table 8 were determined by least squares estimation using several different kinds of data simultaneously. Selected P-V-T data from Lounasmaa (90), el Hadi, et al. [22, 23}, Roach {67} and Sullivan [71] were included in the fit together with Cy data from Lounssmaa [50]. The exper mental data weed in the least annerae Bt of region T include temperatures and densities much higher than the limiting densities and temperatures of the bound- aries of region I. Data for temperatures to 70 K and. ‘densities to Pe were included in the fit. The data out ‘of region I included in the least squares fit were sven a very low weight. However, their inclusion in the densities and temperatures higher thar boundaries of the region. In addition to the experimental data_mentioned above, the condition of isothermal thermodynamic ‘equilibrium of the coexisting liquid and vapor phases was inserted into the fit at 30 temperatures between 2.2 and 5.1 K, Aleo;the entropy of waporiaation wae inaludod in the fit at these temperatures. The thermodynami relationships for these two conditions are given in equa: tions (20) and (22): those at the Cr-Gy=0= [3 (Bp) ado a0 where G is the Gibbs energy, Sis the entropy and the subseripts | and refer to the liquid and gaseous ‘phases, respectively. The entropy of vaporization, neces sary for the implementation of equation (21), was cal culated using equations (2), (@) (), and (). ‘The actual ftting of equation (17) was accomplished by using many different. thermodynamic properties simultaneously. However, for exch property the second virial B(p, 7), as discussed in section 5, was preserved, and the fting was performed on a AP of AP=P—pRT(1+B (bi. TP) smainder of equation (17), or aome function of AP. For an example of simultaneous use of diferent kinds ‘of data in least squares techniques see section 9. ‘As a result of including the thermodynamic relaton- ships of equations @20) and (21) in the actual last tqsares cetimation procedure, the derived. thermo- fynamie properties in the Liquid state (P Pusine) may be calculated by the integration of the quation of state across the two-phase reson assuring mathematical conunuity. Usually when an equation af state is ured to calculate the derived properties in the ‘compressed Liquid region (7' Pau) some tltemnate method of erasing the tweephaxe regan ix fused which always creates discontinuity inthe caleu: lations at Ts for all p> pe. The least squares ft as described above was also ‘constrained atthe enitical point to aP) _(aP (BG), @ and 10 the state point of Pe=2.2449 atmospheres, pe= 17.3987 moles/liter, and T-—5.2014 kelvis 62. Estimation of the Cootficiants for of the ‘Multi-Function Equation of Stote ‘The coefficients for equation (17) in table 9 were Astorminad hy Inset nation aing cover different kinds of input data simultaneously. Selected P-V-T data from Lounasmaa (50), el Hadi (221, Roach [67], Glassford and Smith [25] Kierstad [47], Edeskuty and Sherman (11), and Grilly and Mills(26] were included in the ft together with C, data from Lounasmea {50} In addition tothe experimental P-V—T and Cy data,values of P-V-P, (3Plip)ry (BPI3T yy and Cy saloulated Seon, the equation for region IIT were included in the least squares treatment of region Il. About 20 points for each ‘of the above properties were used for both 10 and 15 K. TThe eritical point constraints of section 6.1 were also included in the fit of this region. The accuracy of the resulting P-V~T surface is discussed in a later section. Estimation of the Coefficients for Region Il of the ‘Multi-Function Equation of State ‘The coefficients for equation (17) in table 10 were determined by least squares estimation using several 4. Phys. Chem Rel Dot, Vol 2, Ne 4, 1973998 ROBERT D. McCARTY different kinds of data simultancously. Selected P-V-T data from Lounasmaa [50], Glassford and Smith [25], Canfield, et al. (9}, Sullivan (T1], Wiebe, et al. (77) Hall and Canfield (30), Grily and Mills [28], were included in the fit together with C, data from Lounasmae (50) and Dugdale and Franck (16) ‘The eritical point constraints of section 6.2 were also included in the fit of this region. The accuracy of the resulting P-V-T surface is discussed in a later section, 7. Derived Thermodynamic Properties Relationships All of the tabulated property valuce appearing in in appendices C and D have been calculated using equ tion (17) as explained in section 6, and the following relationships: 5, nin 22) «FE oe alin] tears f'qan, 0 7.511 J/mol K, and HP. phere. The above values of $, and H®, are based on the key values for thermodynamics reported in CODATA Bulletin [14]. The S*, and HY, given here differ by 0.443. J/mol K and ~"'58,980 J/mol, respectively from the earlier tables of McCarty [55, 56] and Mann [53 ene f ath me +a Gale "-lOG) = In all cases the input variable pair was pressure (P) and temperature (7). Thie requires an iterative colation ‘of equation (17). In the course of preparing the tabula tions it was leared that in solving. equation (17) for density, for all pressures at temperatures below 15K, it is important to begin the iteration with a reasonable guess, For 7' Pray the staring value for denshy shouhd be p> pat. For T>Te, the starting density of 1.5 pe will work for all ressures, The reason for this is that a whole set of wanted sclutions exiat inthe Tiquidwanor eoexiatence region and another set exist at densities slightly larger than the solid densities at temperatures to 20 K. ‘The transition between the various rogions of equation (17) was handled the same way with the derived proper- 2. Ph Chom Rt, Bote, Vo. 2, No 4 1972 ties as it was with the P-V~T. For a detailed explanation, of this see section 6, 8. Evaluation of the Equation of stare and Tables ‘There are numerous tests which ean be equation of state. One of the most important the equation reproduces experimental data. Others are not quite so direct, but nevertheless they are important ‘Each subsection im this section will desl with a particular property or problem relating tothe validity of the equa- tion of state or the tabular date, 12.1, Lompanton ot Expenmental and Calculated P-V-T Data ‘No attempt will be made to compare the equation of state wit all of te existing experimental data, Com parisons with the data used in the fit will be made, and in some cases data which were not used in the ft but which are of particular interest. Wiebe, et al, [77] reported helium P-V—T data for six isotherms from —70 to 200°C for pressures to 1000 ‘atmospheres, The 50, 100, and 200°C isotherms were lused im estimation of parameters tor equation (Li). A ‘comparison of their data for the higher temperature isotherms with values from the region III equation of 100 ite Pow PERCENT DENSITY DEVIATON, T 1 DENSITY, melt Ficome 10, Comparison of denies feTHERMODYNAMIC PROPERTIES OF HEUUM 4 937 ‘eee eee eee — Te 273Isk aa] 8 a nb "8 cl wn) gts | Paap Pet «100 eb | PERCENT DENSITY ps) at Temperatures below 15_K. The present tabulations Should be ‘sed in_ preference 10" the esi ones. A correlation of the thermophysical properties of helium by McCarty (56] is based on this work, and the thermedynamic.propertce in thet publication are identical with those tabulated here with the exception of units and a different reference state fr entropy and enthalpy (ee section 7. The earlier publication was {ended to make tabultion availble as sn Pow: sible and does not contain any of the details of the correlation given here, the eavier publication does Der of properties sul where one ity, viscosity, diclectric constant, and. special parameters used in heat transfer calculations. The Additional properties tabulated. in the easier publ Cation are of interest to a limited. audience and in ome cases ate highly uncertain and are therefore not sppropritetoinchide hee. 8.5. Discussion of Errors ‘There are a few seneral comments which may be made about the accuracy of the helium tables. First it seems that for temperatures above 15 K the maximum error in any of the properties should not be much more ‘than 1 percent, except perhaps the Cy inthe transposed critical region (locus of maxima of C, along isobars for pressures grester than critical pressure) which is still fuident hetween 16 and 90K Tn the range af tempers. ture from 30 10 70 K there are no reliable experimental data above about 30 atmospheres, which makes the estimate of 1 percent in this range a speculation; however at 30 K there are P=p-T' and Ce-p-T data to nearly 3000 atmospheres, and these data agree well with the equation of state, For temperatures above ‘about 600 K there are no data for preseures beyond the second virial coefficient range, but the contribution of higher order virials becomes 20 small that the 1 percent estimate seems realistic. Belov about 10 K one begins to encounter uncera 4. Phys. Cham Rel ete, Vl 2, Me 4, 1973PERCENT DEWATION FIGURE 25. ferences between correlation ofthe thermynamie roperis of heli ‘by Mann 3] and his work ties greater than 1 percent in most of the properties, especially inthe high density fuid. The region of greatest Incertainty in the tabulated properties is for tempéra- tures below 7K and densities greater than critical density. The estimates of errors shown in figures 26, 21, and°28 are based on many different factors but Prlarlly ws the uncerealases of ae experiitenal date ‘used in determining the equation of state and the agree- ‘ment between those data and the equation of state. Tn general the estimated error of Cy i the same as Ge, except in the critical region and the transposed critical region where the ealeulation of Cy depends hheavily on the compressibility, and the compressibility sm thhs region 1s uncertain by several percent. Ineretore, 4. Phy. Chem Rel. Det, Vo. 2, No.4, 1973 the uncertainty of Cy in the region outlined above is sareater, but unfortunately itis dificult to estimate how much greater. Tn the two regions of uncertainty of greater than 10 Percent in figure 28, the incorrect functional form of the ‘equation of state is at fault, In the critical region the ‘calculated Cy ts 10 simall because the equation of stave is analytic at the critical point, and in the dense fluid region along the solid line, the equation of state predicts 2. Cx which is ton Tarze becanse the influence af the ‘uptum in Cy as the lambda line is approached has car- ried on up the solid line to a region where experimental evidence indicates no such upturn exists. The Cy calculated via the equation of state in nese two ressone