XN EQUATIOX OF STATE; FOR ETHY L ESE GAS*

BY LOUI S J. GILLESPIE
I n the present paper the equation of state of Beattie and Bridgeman' i +
utilized to smooth and to correlate, as far as is possible, the pressure-volurrrr,-
temperature data for ethylene gas. The available data are those of Amagat?
which extend over a large range of pressure and temperature into the liquid
phase; the 24.95' isotherm of Masson and Dolleyj3 and about four points on
each of four isotherms by hlathias, Crommelin and Watts.' Data on thf:
normal volume of ethylene at oo and I atmosphere are collected and reviewved
by Blanchard and Pi ~keri ng.~
KO equation of state has apparently been presented for ethylene, with the
exception of virial pv expansions. Aside from the general usefulness of
equations of state, especially for gases of considerable chemical activity, an
equation for ethylene is especially desirable in view of the work on mixtures
of ethylene and argon by Masson and Dolley, which have permitted the cal-
culation of partial molal free energies of the constituents of these binary
mixtures.6 The present work was in fact undertaken with a view toward a
more complete thermodynamic investigation of these data than has been
hitherto possible.
Interpolation of the Isothermals
For each temperature the pv products of Xmagat were graphically
smoothed as a function of the density, using an appropriate deviation function
and large scale coordinate paper, and values of the pressure were interpolated
for even values of the density in moles per liter (0.5, 1.0~ 1.5, etc.). The
transfer from Aniagat units of density to moles per liter was effected through
the normal density given by Batuccas' and chosen by Blanchard and Picker-
ing,; namely 1.2604 grams per liter.
By a similar procedure, the data of hIasson and Dolley and of the Leiden
laboratory were interpolated to the same even density values, thus per-
mitting a comparison of data. I n the former case the transfer of density
units required a knowledge of the ratio of'the volume at 24.95' to that at oo,
both volunies at I atmosphere. From Xmagat's data this ratio, V,,/Vo, was
calculated to be 1.0936, from >\lasson and Dolley's data, 1.0932. The mean
* Contribution from the Research Laboratory of Physical Chemistry, Massachusetts
I nstitute of Technology, S o . 216.
J . A. Beattie and Oscar C. Bridgeman: J . Am. Chem. Soc., 49, 1665 (1927).
Ann. Chim. Phys., 16) 29, 68 (1893).
Ppoc. Roy. SOC., 103-4, j24 (1923).
4 Cinquihme C0ngrt.s International du Froid, Rome; Premiere commission interna-
5 Sci. Paper U. S. Bureau of Standards, So. j29 11926).
7 J . Chim. phys., 16, 322 (1918).
tionale de 1'Institut Internationale du Froid, Rapports et Communications. Leiden, 1928.
Gibson and Sosnick: J . .im. Chem. Soc., 49, 2172 (1927).
A S EQU.4TION OF STATE FOR ETHYLEKE GAS
355
value 1.0934 was chosen. At a subsequent period it was found that the
higher ratio was yielded by the equation of state finally derived, but it was
found that the effect of the difference, 0.0002, was practically negligible in
all relations under consideration.
I t was observed in this isothermal interpolation that the pv data at 24.95'
do not approach RT at low pressures in a perfectly smooth way around I
atmosphere. The data are smoother at all higher pressures. These data
were already smoothed once before publication,* but probably without the
aid of a value of RT, which in fact could not be obtained without knowledge
FI G. I
Typical plot of an unsmoothed isometric ( 2. 5 *) showing the isothermally smoothed
points due to hmagat (circles), Masson and Dolley (square), and Mathias, Crommelin and
Watts (crosses). The plot shows the value of the deviation,-p - 34 - 0 . 2 7 j 16t" rn a
function of the temperature Centigrade. The line is furnished by the equation of state
finally selected. The circles centered on this line are drawn with radii equal to O . j % of
the calculated pressures.
of the ratio 1125,/T'0. h like situation obtained with reference to the Leiden
isotherms; no way being discovered of smoothing these together with RT
values without assuming in this case rather large experimental errors.
The interpolated values of the pressures, which are hereafter described
as the observed pressures, were now grouped as isometrics for the next step.
Representation of the Isometrics
For each density, the pressure was plotted as a function of the absolute
temperature. Fig. I shows a typical plot, that for 2 . 5 moles per liter. The
line, shown for comparison only, was calculated from the final equation. It
liter
8 See Fig. 2 , ref. So. 3, where irregularities in the original data are obviously not en-
tirely removed in the smoothing.
356 LOUIS J. GILLESPIE
is clear that the results of Masson and Dolley, of the Leiden Laboratory, and
of Amagat differ too much to permit utilization of all data in the final smooth-
ing. Only the Amagat data are extensive enough to determine an equation
of state; such an equation must be based on Aniagat's data alone. It is also
clear from the line drawn that the equation does not include a certain trmd
in the Amagat data. This trend, evidenced by an inflection of the isometric.,
suggests at first a discordance between the high temperature (137', 198.;")
and the low temperature (0-100') data. Exclusion of the high temperature
data would however leave d2p/dt2 positive, whereas this is as a rule negative
(above the critical volume)Q. It seemed best therefore to include all tempera-
tures and to assume such a curvature of the isometrics as would keep them
within about 0.5 percent of the measured pressures. Following the pro-
cedure of Beattie and Bridgemanlo a value of c was found which would secure
this, at least from a density of zero to about 7 moles per liter, with some
difficulty as regards the highest temperature-pressure corner of the field.
The other constants of their equation were then determined as described by
them.
The constants so determined are given in Table I.
TABLE 1
Constants in Beattie-Bridgeman Equation of State for Ethylen?.
R A0 a B o b C Mol n-t
o 08206 6 152 o 04964 o 12156 o 03597 2 2 68 IO^ 28 031
The equation is p = RT(I - e).(v + B)/v2 - A/v2
Where A = A, (I - a/v)
B = Bo (I - b/v)
E = c/vT3
v = volume in liters of a mole
p = pressure in international atmospheres
T = 273.13 + t"C.
The agreement of the equation with the observed pressures of Amagat is
exhibited in Table 11, which lists the observed pressures and the deviation
A p (observed minus calculated pressure) in atmospheres. The average
deviation (taken without regard to sign) over the entire range considered is
0.4j per cent. This range goes to a density 8 mols per liter, slightly higher
than the critical density. The agreement is fair. The equation holds very
well indeed up to 7 mols per liter, the average deviation being 0.36 per cent.
Table I11 shows the magnitude of the disagreement, already noted by
Masson and Dolley, between their results and those of Amagat. There is
agreement only at low densities. It is however precisely at such low densi-
9 Onnes and Keesom: Encyklopadie der mathematischen Wissenschaften, Art. V IO,
page 756 (1912); also Communications from the Phys. Lab. Un i v . of Leiden, 11, Supple-
ment 23, p. 142, Leiden (1g12).
'OProc. Am. Acad. Arts Sei., 63, 229 (1928).
AN EQCATIOS OF STATE FOR ETHYLENE GAS
3 57
3 58 LOCIS J . GILLESPIE
ties (1.1 to 2 3) that the Leiden data differ, and at all temperatures. from the
equation, as is shown in Table Is, for which the unsmoothed Leiden data
were utilized. The data deviate more strongly than those of Masson and
Dolley and in the opposite direction. If the percentage deviation is plotted
as a function of the density, it is found not to approach zero asymptotically
at zero pressure. This is not attributable to a defect in the equation. Even
though the purity of Amagats ethylene was not so great as it is now possible
to obtain, the fact is important that the equation furnishes numbers, smooth
in two dimensions, which must be considered to approach correctness, in
smooth fashion, as the pressure approaches zero.
There exist relations in which the disagreement between Masson and
Dolley and Amagat, which reaches 4 6 5 up to a density of 8> is not so im-
portant as might appear from Table I11 alone. Compare the discrepancy of
37c at a density of 4.5 with the error of 8 jTCj which is the percentage deviation
of the perfect gas law from the observed value. At a density of 6, the perfect
gas law is in error by 1 2 2 %.
TABLE 111
Comparison of Calculated and Observed (Masson and Dolley) Isotherms at
24 95O
obs - calc,
calc.
Ap, = IO0
Density p calc. Ah 5 Density p cnlc A P, 6
- I - 0. 02 7 66.7 4. 6
0 . 5 11.41 - 0 . 2 7. 5 68. 7 4. 2
I 21. 2; +O 8 7 1 . 2 3. 1
2 36.77 0. 6 9
77. 8 - 0. 3
2 . 5 42.71 1. 0 9. 5 82. 4 - 2.5
3 47.63 1. 5 I O 87. 9 - 4 7
4 54. 9 2 . j I 1
102.8 - 8.7
4. 5 j 7 . 6 3. 1 1 . ; 112.3 - 10.0
5 . 5 61. 7 4. 0 12. j 136.6 - 10. 5
6 63. 3 4. 6 I 3 151.6 - 9. 0
6. 5 65. 0 4. 6
TABLE Is
Percentage deviations betu-een Leiden experimental values and the values
Temp./Serial order
I 2 3 4 5
- I . 36C 1. 3 1. 4 I . 6 2 . o
$0. I O0 0. 9 I . o 1 . 2 1 . 5 1. 8
20. 18~ I . o I . o 1 . 2 1. 4
1 . 5 29.66 +0. 3 8. j 74.1 I . 6
3 . 5 51.7 1. 9 1 0 . 5 94.7 - 6. 9
5 59.8 3. 7 I 2 123 5 -IO. j
calculated. The calculated pressure is always greater.
__
I O . 170 1.1 I . o 1. 3 I . 6
~
__
Density at 20. 18~ I . I 414 1.3356 1.6474 2.2644 ~
AN EQCATI ON OF STATE FOR ETHYLESE GAS
359
Calculation of the Normal Density
The equation does not necessarily furnish upon calculation the same value
of the density at oo and I atmosphere as that which was used to obtain i t)
but may give a better value.
Table V shows the summary of Blanchard and Pickering (5) with the
addition of the value calculated from the equation.
TABLE V
Normal Density of Ethylene, in grams per liter
Source Density Source Density
Leduc I . 2605 Batuecas I. 2604
Stahrfoss I . 2 6 1 0 Equation 1.2599
Blanchard and Pickering selected the value of Batuecas. This selection is
supported by the value here found.
Calculation of Critical Data
I n the isothermal smoothing of the pu values under the critical tempera-
ture it was noticed that the calculated pressures became almost a zero func-
tion of the density and indeed, without great care in the smoothing, t'he
pressures actually decreased with increasing density. Khen the equation
was derived i t was thought interesting to see whether it would exhibit this
trend. It was found that the equation gave a 7. j" isotherm with an inflec-
tion at about 7 moles per liter. By a succession of trials i t was found that an
inflection occurred at as high a temperature as 8.5", but not at 8. j 5' , when
the pressures were calculated with a precision of 0.0027c. The critical tem-
perature is therefore given by the equation as 8.5', to the nearest 0.1'.
Pickering" selects the value 9. i o, a difference of 0. 47~ on the absolute tempera-
ture, which is of course the quantity calculated by the equation. Since this
selection, Masson and Dolley (3) obtained 9.35', the lowest recent experi-
mental value.
The critical pressure calculated is 49.19 atmospheres against 50.9 selected
by Pickering, a difference of 3.4yc. The critical density calculated is 6.4
to 6. 5, selected value 7. 9 in moles per liter. Here the errors seem to accumu-
late, making a positive disagreement.
Equations Tvhich represent the measured pressures are not generally ex-
pected to furnish correct critical constants (nor do those which are derived
from critical constants represent the measured pressures), so that the success
of the equation for ethylene in furnishing at least the critical temperature is
surprising. The agreement may of course be accidental.
l 1 Sci. Papers, U. S. Bureau of Standards, ?To. 541 (1926).
360
LOUI S J. GI LLESPI E
summary
The constants in the equation of state of Beattie and Bridgeman have
been determined for ethylene gas from the data of hmagat. Using atmos-
pheres, liters per mole, T = 273.13 + t " C, R = 0.08206, they are A, = 6.152,
a = 0.04964, Bo = 0.12156, b = o.o3j97$ and c =22.68 I O?.
The representation of Amagat's data is good up to a density of 7 molcs
per liter and fair to 8, slightly above the critical density; the average devia-
tions being o 36 and o 45 per cent respectively.
The equation does not represent closely the isotherm of Masson and Dol-
ley, except at low pressures, in accordance with their statement that this iso-
therm does not agree with interpolations from Amagat's data
The critical temperature calculated from the equation agrees to o 4y0 with
that observed, the difference being 1.2"; the critical pressure calculated is
3.4Yc in error.
The normal density calculated from the equation, 1. 2 599, supports the
value of Batuecas as against that of Stahrfoss, being lower than either.