See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/242086759

Einstein, 'parachor' and molecular volume: Some history and a suggestion

Article  in  Current Science · June 2008

CITATIONS READS
15 2,888

1 author:

Saraswathi N. Balasubrahmanyam
Indian Institute of Science
62 PUBLICATIONS   216 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Saraswathi N. Balasubrahmanyam on 13 November 2014.

The user has requested enhancement of the downloaded file.

HISTORICAL NOTE

Einstein, ‘parachor’ and molecular volume: Some history and a

suggestion
S. N. Balasubrahmanyam

Einstein’s first paper P = P∞ – c1c2φ (r). closer to a popular potential used today if
he had written for a pair of molecules:
In 2002, the Royal Society commemo- In this case, c is a characteristic constant
rated the centenary of young Einstein’s of the molecule and φ (r) is a function of P = P∞ – (∑α ∑β c1αc2β)φαβ (rαβ ),
first scientific paper by publishing J. N. the distance [between] the molecules,
Murrell’s analytical comments1 on ‘an taken as independent of the nature of the which is a sum of atom–atom potentials.
extraordinarily advanced paper for a re- molecule’. This model does introduce non-central
cent graduate’. At the end of his com- Murrell comments: ‘We can see im- terms [through the variability of r] into
ments, Murrell makes an interesting mediately the limitations of this model the intermolecular potential. . .5.
connection between certain additive con- because there are no angular variables, to ‘Einstein defines an integral (K′) of the
stants calculated by Einstein and ‘para- allow for electrostatic contributions to potential taken over a surface layer and
chor equivalents’2 that were to be descri- the potential, for example’. However, then comes to the formula:
bed some two decades after Einstein’s there have been many subsequent models
paper. that do no better, employing, for exam- ∑aca = V [(γ – T(dγ/dT))/K′]1/2.
As a part of his examination of the na- ple, a Lennard-Jones functional form for
ture of intermolecular forces, Einstein, the potential4. Indeed, the analysis of gas Einstein goes on to say, “Since it is pos-
started his paper titled, ‘Folgerungen aus imperfections has commonly ignored sible to calculate the unit of the boiling
den Capillaritätserscheinungen’3, with an angular variables by considering only temperature (for most materials) accord-
analysis of the thermodynamics of liquid spherically averaged potentials. . . . ing to R. Schiff’s observations, we now
surfaces: ‘. . . experiments reveal that ‘The total potential is a sum of all pair have many tools to determine ca. I took
since surface tension γ varies linearly interactions, and if all the molecules are all the information from a book about
with temperature, the energy needed to the same, Einstein could write General Chemistry by W. Ostwald”6. He
build a surface unit is independent of does not specify further this information,
temperature. Hence the specific heat of P = P∞ – 1
2
c2∑a ∑bφ (r). but the assumption is that it is values of
the surface is zero. The energy of the the surface tension γ, the molecular vol-
surface can therefore be considered as a Einstein also assumed that the potential ume V and the boiling point T that go
potential energy, and this energy can be of the molecular forces is the same, as if into the equation. As K′ is unknown, the
expressed by the function: the matter were to be evenly distributed unit for ca was chosen arbitrarily. He
in space, and says, “This, however, is an used a least squares method for optimiz-
γT(dγ /dT). assumption which we can expect to be ing the ca parameters, and from a series
only approximately true”. of hydrocarbons [and oxygenated com-
‘The double sum is replaced by a dou- pounds], he obtained the values:
This function is a better unit with which
ble integral over all space, with a multi-
to analyse “stoichiometric aspects” than cH = –1.6; cC = 55.0, cO = 46.8,
plying factor that is the number of mole-
γ ’.
cules per unit volume, or the reciprocal
I now give excerpts from Murrell’s using data on 17 compounds in the least
of this, the molecular volume V’ [empha-
paper (co-authored with N. Grobert): squares analysis.’
sis mine]. Murrell continues, ‘This would
‘Einstein is interested in the possibility Table 1 provides illustrative values for
be a very poor approximation because
that the surface energy can be expressed two examples taken from Murrell and
molecules are kept apart at short dis-
by additive contributions from the atoms Grobert’s paper.
tances by a repulsive branch of the po-
making up the molecules in the surface. Einstein then examined the data for a
tential, and any model in which the
This was not the original idea, because series of 24 halogenated compounds (e.g.
distribution of matter is assumed to be
atom-additive contributions to γ /M, where chlorobenzene, ethyl bromide, allyl io-
uniform would greatly overestimate the
M is the molecular mass had been derived
attractive forces.
earlier and regularities seen in the sur-
‘To introduce stoichiometric proper-
face tensions of liquids from molecules
ties Einstein expresses the constant c as a
related within a homologous series’. Table 1. Illustrative summed ca values
sum of constants ca, one for each of the
In Einstein’s own words: ‘I started obtained by Einstein
atoms a in the molecule:
from the simple idea of attractive forces
amongst the molecules, and I test the Fitted Observed
consequences experimentally. I took P = P∞ – (∑α c1α)(∑β c2β)φ12(r). Compound ∑aca ∑aca
gravitational forces as analogues. There- Formic acid HCHO 140 145
fore, the relative potential of two mole- ‘This does not introduce angular vari- Carvol C10H14O 587 574
cules is: ables either. He could have been much

1650 CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008

 HISTORICAL NOTE
dide, etc.) and obtained the ‘less accu- their constituent elements. Isomeric γ = γ0(1 – T/Tc)n,
rate’ values for the halogen parameters: compounds of similar constitution, e.g.
methyl propionate, ethyl acetate and pro- where n is a universal constant, and γ0
cCl = 60; cBr = 152; cI = 198. pyl formate (all with empirical formula depends on the critical constants of the
C4H8O2) have almost identical molar liquid’. This equation has been verified
He then comments: ‘I have the impres- volumes at their boiling temperatures. In for a number of organic compounds and
sion that there are greater deviations various homologous series there is a con- n was found to be very close to 1.2 (refs
(from the theoretical values) for sub- stant difference of ~22 ml for each CH2 2b, 12). The Eötvös equation, a statement
stances exhibiting a relatively large mo- group. Kopp’s original list of volume on the nature of the (inverse) co-
lecular mass and a small molecular equivalents which features different val- variation of density and surface tension
volume’. In a concluding statement he ues for the constitutive factors for oxy- of liquids with temperature8d, may be
says, ‘It is noteworthy that the constants gen in different chemical combinations, written as:
ca in general increase with increasing has seen additions and modifications
atomic weight, but not always propor- from the later work of Lossen and Le γ (M/ρ)2/3 = k(tc – t),
tionally. . . the introduction of the func- Bas8(b). Examples are given in Table 2.
tion φ (r), which should be independent where ρ is the density of the liquid and
of the nature of the atoms, and the substi- M its molecular weight. Katayama8e pro-
tution of the sums by integrals are both The ‘parachor’ concept posed a modification by which ρ is re-
approximations. In fact, our theory does placed by ρ – ρ′, where ρ′ is the density of
not apply for substances exhibiting small Fast forward to 1923: Everyone may ex- the vapour at the same temperature:
molecular volumes. . .’. pect both density and surface tension of
Murrell sees a connection to the para- liquids to change inversely (lessen) with γ (M/ρ – ρ′)2/3 = k(Tc – T)
chor in Einstein’s work as a young scien- temperature, maintaining proportionality
= kTc(1 – (T/Tc)),
tist presenting his very first paper: ‘. . . of some kind. In 1923, Macleod10 proposed,
it was an extraordinarily advanced paper on entirely empirical grounds, what has
the absolute temperature difference Tc – T
for a recent graduate . . . The idea behind always been described as a remarkably
being the same as tc – t. This equation
a stoichiometric analysis of surface ten- simple expression for the estimation of
holds quite accurately for normal liquids
sion goes back to the work of Schiff, and the surface tension11:
at temperatures much nearer to the criti-
this was well discussed in Ostwald’s cal value, than does the original form of
book, but to suppose that one could ob- γ = K ( ρL – ρ V ) 4,
Eötvös. By combining the van der Waals
tain some information about intermo- and the Katayama expressions, and as-
lecular potentials by such an analysis is where ρL and ρV are respectively, the
suming n to be 1.2, as found experimen-
probably Einstein’s own idea. It is inter- density of a liquid and of its vapour in
tally, it follows that
esting that the stoichiometric analysis of equilibrium with it at a given tempera-
surface tension and its interpretation ture, γ the surface tension at the same
temperature and K a constant characteris- γ 1/4/(ρ – ρ′) = C,
through molecular structure became very
popular from the work on the parachor, a tic of the liquid. The equation is obeyed
quantity introduced by Sugden and oth- with considerable accuracy by many liq- where C is a constant for each substance.
ers in the 1920’s. . . The parachor was uids over a wide range of temperature up [This relationship is a rewritten form of
subjected to a large number of analyses to the critical point8c. It has been the Macleod equation.]
and interpretations in subsequent years, described as a fundamental discovery ‘The remarkable constancy of the
but the subject died a rapid death in the that the surface tension of a liquid at any quantity γ 1/4/(ρ – ρ′) over a large tem-
1940s’. Not quite, as we shall see! temperature is in a constant ratio to the perature range would, however, appear to
fourth power of the difference of the imply something more fundamental’.
orthobaric densities of liquid and vapour Could that ‘something’ be chemical
Earlier idea of the additivity of measured at the same temperature. To structure?
molecular volumes quote freely from the classic treatise of Sugden decomposed C into a molecu-
Samuel Glasstone8: ‘According to J. van lar weight component (M) and a new
Einstein’s effort to express the constant c der Waals, the surface tension of a liquid component P, which he called ‘para-
for a molecule (which implies molar vol- at temperature T should be related to the chor’2,8f:
ume) as a sum of constants ca, one for critical temperature Tc by an equation of
each of the atoms a in the molecule, has the form: γ 1/4 = [P](ρL – ρV)/M.
a form of precedence from about the
middle of the 19th century. Kopp7,8a had
Table 2. Volume equivalents of elements considered by Bayliss
demonstrated that molecular (or molar)
volume, i.e. the volume occupied by a Atom Kopp Le Bas Atom Kopp Le Bas
mole of liquid shows marked additivity.
He noted that molar volumes of organic Hydrogen 5.5 3.7 Iodine 37.5 37.0
Carbon 11.0 14.8 Oxygen (–O–) 7.8 7.4
compounds at their respective boiling
Chlorine 22.8 22.1 Oxygen (O=) 12.2 12.0
points9 were, to a great extent, additive
Bromine 27.8 27.0 Sulphur 22.6
functions of the volume equivalents of

CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008 1651

HISTORICAL NOTE
He rearranged this expression into a form In 1953, Quayle published a review fundamental question of what exactly is
where, since ρV was much smaller than titled, ‘The parachors of organic com- the dimensionality of [P]. In spite of the
ρL away from the critical point, the dif- pounds’, comprehensively surveying the possibility that attributing the dimen-
ference (ρV – ρL) could be replaced by d, field14a. He paid much attention to his sionality of volume (L3) to the function
the density of the liquid at the tempera- own and earlier attempts to improve the could be fundamentally untenable16g,
ture at which γ was measured: parachor equivalents in order to obtain there were important works premised ac-
calculated values that would fit the ex- tually on [P] representing volume.
[P] = γ1/4 d–1M = ∑[Pi]. perimental values better. He constructed
extensive tables comparing, most impor-
This expression was seen as implying tantly, the values of Sugden, of Mumford ‘Parachor’ represents molecular
that parachor is a number that represents and Phillips15 (who had expressed dissat- volume
the molar volume Vm of a substance when isfaction with some of Sugden’s values
its temperature is such that its surface for the parachor equivalents) and of his In a 1937 paper, Bayliss17 described re-
tension is unity. From data on over 600 own group, paying particular attention to sults purportedly showing that Sugden’s
substances of variegated types, Sugden the purity of materials tested for recor- original interpretation that parachor is a
evolved ‘parachor equivalents’ (Pi in the ding new experimental data14b. Efforts in function of molecular volume may be
equation above) for atomic and structural these directions have continued into later applied with marked success even while
units (e.g. double bonds). That parachor, times11,14b. he declared that, ‘Parachor suffers from a
together with certain constitutive factors, There were many attempts to derive lack of a clear physical interpretation. . .
is largely additive was demonstrated at a the idea of parachor from theoretical first attempts at a theoretical treatment have
time when trying to discover additivities principles based on statistical thermody- not succeeded in supplying an interpreta-
among parameters in molecular proper- namics even in the early period16. As- tion. . . and without a considerable
ties was near centre stage. [Discovering a sessments of the manner of the co- advance in our knowledge of the theory
new additivity principle appeared espe- variation of surface tension of a liquid of the liquid state, it would seem impos-
cially significant after the discovery of and the difference between its equilib- sible to obtain one on purely theoretical
the additivity, for example, of molar heat rium liquid and vapour densities gener- grounds’.
capacities8g or molar magnetic suscepti- ally started with the common-sense Bayliss recalculated parachor constants
bilities; Pascal8h.] Following earlier work proposition that the contacts of a ‘parti- (or ‘equivalents’ as they were also called)
on molar volumes, Sugden2b also showed cle’ at an interface would, on the aver- by fitting the data on the n-paraffins
that the parachor was actually 0.77 times age, be less numerous than those of one from published tables18, by least squares
the molar volume at the critical tempera- in the bulk. Next considered were procedure using the linear expression
ture [[P]/Vc = 0.77], presumably on the changes in the potential energy that may P(CnH2n+2) = nP(CH2) + 2P(H). The other
basis that it actually had the dimensions accompany the creation of an additional parachor constants necessary for drawing
of volume. [Getman7b comments that unit of surface under reversible condi- comparisons were those of the halogens.
parachor was a refinement of Kopp’s tions. Functions of fluid characteristics, These were calculated from the data on
original finding of the additivity of molar such as critical volume (Vc), temperature n-alkyl halides19 by the relation P(X) =
volumes.] (Tc), surface free-energy (φ0) and charac- P(CnH2+1X) – nP(CH2) – P(H). The recal-
During the next three decades parachor teristic length (r0), etc. were taken in culated atomic constants as compared by
studies were seen as useful in some early various combinations and raised to frac- Bayliss with previously recorded values
applications to the solution of problems tional powers in the derived expressions are given in Table 3; the only signifi-
of chemical constitution13. Noteworthy (one example: [P] = 0.324Tc1/4Vc7/8; Fer- cantly wide differences are seen in the
among these is a ‘straight line mixture guson and Kennedy16a). Attempts were values for carbon and hydrogen.
law’ by Hammick and Andrew13, to hold also made to give the power index a Bayliss proceeded on the following
for solutions in ‘normal’ liquids enabling value other than 4 or ¼ respectively, in formal bases: ‘The closeness of approach
parachors of substances that either melted the Macleod and Sugden expressions, ei- of two atoms is governed by their bond
high or decomposed before melting to be ther for theoretical reasons or for better radii (denoted hereafter by r) if they are
calculated. A ‘mean parachor’ [P]m was fit with experimental results. chemically bound, and by the vaguer
calculated from the equation: It would not be unfair to say that most ‘packing radii’ (r′) if they are not. . . In a
or all such efforts have foundered on the tetrahedral molecule, the central qua-
[P]m = Mm γ1/4/dm,

in a manner reminiscent of the calcula- Table 3. Comparison of parachor equivalents

tion of molar refraction of a solute8i Atom Sugden Mumford and Phillips Vogel Desreux Bayliss’ paper
and its parachor evaluated using the rela-
tion: CH2 39.0 40.0 40.3 39.9 39.92
C 4.8 9.2 11.5 8.3 9.1
[P]m = x[P]solute + (1 – x)[P]solvent. H 17.1 15.4 14.4 15.8 15.4
F 25.7 25.5 – 26.1 26.4
The method was thought to be of great Cl 54.3 55 – 55.0 54.6
Br 68.0 69 – 68.5 68.5
practical importance, but its legitimacy
I 91.0 90 – 90.0 90.3
appears questionable now.

1652 CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008

 HISTORICAL NOTE
drivalent atoms touch only those atoms lated from the Pauling and Huggins bond ters. These instruments, priced within the
to which they are chemically bound, and radii21, whereas the corresponding budgets of university chemistry depart-
since the space they occupy is determi- tetraalkyl compounds have been found to ments, were easily operable by the wet-
ned by their bond radii only, it will be behave normally22. This behaviour is re- lab organic chemist trying to make struc-
assumed that their volume is that of a flected in the parachors, since the atomic tural assignments. While these could be
sphere whose radius is r. Adopting the parachors of silicon, germanium and tin among the basic reasons why parachor
well established value of r = 0.77 Å for are lower in their tetrahalides than in was superseded, soon to follow were the
the carbon atom20, and assuming that the their tetraethyls. Bond radii calculated powers of such structure-determination
atomic parachor of carbon is a measure for these elements from the parachors of tools as single-crystal X-ray crystallogra-
of its atomic volume, one obtains the re- their tetrachlorides and their tetraethyls phy, multiple irradiation pulse NMR
lation that one parachor unit = 0.210 Å3’. are given in Table 5, and are compared techniques and associated software.
This relation was used to calculate the with the bond radii obtained from similar
bond radii of other quadrivalent atoms compounds by the electron diffraction
whose atomic parachors were known. method21,22. A simplification has been Exner’s stricture
The results, shown in Table 4, include made in making the central atom of the
the Pauling and Huggins bond radii for tetrahedral molecule wholly responsible In 1966–67, Otto Exner of Prague publi-
comparison. for changes in both the parachor and the shed a series of papers24 in which he
‘The agreement between the two sets bond distance. The case of lead has also examined questions relating to the ‘Addi-
of bond radii is good [with the value for been included in Table 5, since the bond tivity of physical properties’. In a pre-
carbon recalculated on the basis of data radius calculated from the parachor is in liminary communication in Nature25, he
of authors other than Sugden – Table 3], better agreement with the value observed argued that taking the fourth root of the
but becomes better when some individual in lead tetramethyl22 than with the Pauling surface tension so weakens its contribu-
cases are considered in detail. . . and Huggins radius20’. tion that any additivity found merely re-
‘The volume of a univalent atom is de- Bayliss’ results seem persuasive but flects the extent to which molecular
termined largely by its packing radius, then the Md–1 component, dominant in volume (the Md–1 term) remains additive,
not as easily defined as its bond radius. . . parachor (see Exner’s stricture below), molecular weight being strictly additive.
If one were to adopt a naive atomic would make it behave as though it repre- He then advanced statistical reasons
model of a segment of a sphere of radius sented a volume. However, it would why, ‘It can be concluded that the con-
r' cut by a plane surface at a distance r appear that the idea that the parachor is a ception of the parachor has no material
from the centre (Figure 1), one sees that function of molecular volume had be- significance and its liquidation can be
values of r and r' cannot both be derived come in a way entrenched before the suggested’.
from parachor data alone; but using the publication of Bayliss’ paper, even The essence of Exner’s arguments is
relation one parachor unit = 0.210 Å3 though statements like ‘[parachor values best presented by directly reproducing
and by assuming the Pauling and Hug- for the same group] indicate some variation two graphs (Figure 2) from the last of his
gins values of r, it is possible to deter- in the value for different compounds’, papers in the series24c. A comparison of
mine the packing radii of hydrogen and surfaced in many publications23. the two panels in Figure 2 shows that the
the halogens from their atomic para- Interest in parachor determination as additive scheme is fulfilled more accu-
chors. . .’ [the full method pertinent to an aid to the understanding of chemical rately for the molar volume than for the
this is not described here]. Bayliss’ re- constitution waned in the decade of the parachor. Multiplication by γ 1/4 actually
sults are gathered in Table 4. 1950s of the last century, which saw the deteriorates the convergence of the best
‘[In] the chlorides and fluorides of cer- appearance of Beckman’s ultraviolet and line-fits. [This puts paid to such state-
tain quadrivalent atoms, the observed Perkin–Elmer’s infrared spectrophotome- ments as, ‘Parachor, which is practically
bond distances are less than those calcu-
Table 4. Bayliss’ comparison of parachors and bond radii

Atomic parachor Bond radius (Å)

Atom Sugden Bayliss Calculated Pauling and Huggins

C 4.8 9.1 0.77 0.77

Si 25.0 26.2 1.10 1.17
c
Ti 45.3 44.1 1.30 1.22
a
Ge 37.4 36.0 1.22 1.22
Sn 57.9 55.4 1.41 1.40
Pb 76.2 75.6 1.56 1.46
b d
Te – 56.7 1.42 1.32
e
1.37
a
Sidgwick, N. V. and Laubengayer, A. W., J. Am. Chem. Soc., 1932, 54, 948–952.
b
Singh, B. and Krishen, R., J. Indian Chem. Soc., 1935, 12, 711.
c
Calculated from Wierl, Ann. Phys., 1931, 8, 521–564, assuming that r for Cl = 0.99 Å.
d
Figure 1. Atomic radii from parachor Tetrahedral radius.
e
data (r) and electron diffraction data (r′). Covalent bond radius.

CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008 1653

HISTORICAL NOTE
forgotten these days (italics mine), is es- purpose of calculating ‘surface ten- situations may be encountered in the oil
sentially a molecular volume with the sions’31 (even of substances that may not and gas industry, where either there are
fourth root of the surface tension as a melt!) from parachor values calculated no surface tension data or when the
correcting factor’26.] The scattered posi- from known parachor equivalents. Many number of variables is too large to make
tioning of plots in the left panel clearly
implies that the parachor equivalent of Table 5. Packing radii from atomic parachors and from calculations
the –CH2– varies with the straight-chain
system of which it is a part. One may ex- Atom H F Cl Br I
pect that parachor equivalents of other Packing radius (Å)
groups will also vary with the chemical From atomic
system of which they are a part. parachors 1.03 1.16 1.44 1.54 1.67
a c c d
It appears to me that Einstein’s ap- From the literature or 1.1 and 1.0 1.12 1.44 and 1.47 1.56 1.71
b
proach was nothing, if not quite rational. r = ½(X–X) 1.07
We see him anticipating that the atom a
Stuart, H. A., XIV of Molekülstruktur, Struktur und Eigenschaften der Materie in
values of c should be expected to vary
Einzeldarstellungen, Springer, Berlin, 1934, pp. 47–48.
with the system or series being studied. b
Pauling, L. and Carpenter, D. C., J. Am. Chem. Soc., 1936, 58, 1274–1278.
To this extent even Exner’s stricture c
From ½(F–F) in CF4; from ½(Cl–Cl) in CCl4 and CHCl3; electron diffraction: Pauling, L.
seems to have been anticipated. The and Brockway, L. O., J. Am. Chem. Soc., 1935, 57, 2684–2692.
d
‘mystery’ as to why the addition of a ½(Br–Br) as calculated from C–Br distance of 1.91 Å reported in ref. 22.
–CH2– decreases the density in some
series and increases it in others (Figure Table 6. Parachors and bond radii for tetrahedral molecules
2) is solved when it is noted that hetero-
Bond radius (Å) from
atoms in the group 7–12 control the
average distance between molecules in a Atom Compound Parachor Parachor Electron diffraction
way different from what may be the case
a
with the less polar (or polarizable) Si Tetraethyl 31.0 1.16 1.16
systems 1–5. This may be due to the Tetrachloride 23.8 1.06 1.03
a
Ge Tetraethyl 38.5 1.25 1.21
manifestation of incipient liquid crystal-
Tetrachloride 34.8 1.20 1.11
like properties in the materials of the a
Sn Tetraethyl 60.1 1.45 1.41
former group. Tetrachloride 54.4 1.40 1.30
a

I presumed most people had become Pb Tetraethyl 75.6 1.56 1.52

a

aware of and accepted Exner’s position a

based, as it is, on telling arguments. But, Values obtained from tetramethyls.
surprisingly, none of the papers on trying
to get at a theoretical justification of
Macleod’s empirical formula from first
principles27 or for seeing a meaning in
the concept of parachor or even attempts
to obtain better parachor additivity equi-
valents11a, published much after Exner’s
‘exposure’ that was as early as in 1967,
makes a reference to his paper. Nor have
there been any rebuttals, as far as I am
aware. Certain relatively modern text-
books or courses treat the subject exten-
sively28, without mentioning pitfalls that
may lurk in the concept. Continued also
are instances of the use of parachor as
one of the molecular descriptors29a in
correlating the properties of substances
through methods of quantitative structure
property/activity relationships (QSPR/
QSAR)29b. It is possible that Exner’s
comments in a paper titled, ‘How to get
wrong results from good experimental
data: A survey of incorrect applications Figure 2. (Left panel) Test of additive character of parachor: Dependence of specific
parachor γ d on the reciprocal of the molecular weight. (Right panel) Test of additive
1/4 –1
of regression’30 could be relevant to
some examples of the application of character of molar voslume: Dependence of specific volume at 20°C on the reciprocal of
the molecular weight for (1) n-Paraffins; (2) 1-Olefins; (3) 1-Acetylenes; (4) Alkyl ethyl
QSPR/QSAR methods. ethers; (5) Aldehydes, (6) α,ω-Dinitriles; (7) 4-Alkylpyridines; (8) Alkyl benzyl ketones;
Remarkably, the parachor function has (9) 1-Nitroparaffins; (10) α, ω-Dichloroparaffins; (11) Alkyl nicotinates and (12) α, ω-
24c
been extensively used for the inverse Dibromoparaffins. (Figure reproduced from Exner .)

1654 CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008

 HISTORICAL NOTE
practical a parametric study of the sur- [P] = 2.841 × 1019T 1/4 f (y)(σ 3/σl/2)’. We should not fail to note here that
face tension of complex mixtures. Inte- Satherley and Schiffrin go by Sugden’s
rest also lay in predicting the surface The implication here is that the SPT pre- definition of ‘parachor’ and so did Len-
tension of hydrocarbon mixtures contain- dicts a simple relationship between the nard-Jones and Corner16d. Going by Ex-
ing dissolved gases at high pressures be- molecular parachor and the hard sphere ner’s stricture the contribution from the
cause capillary action may become the diameter. Seemingly, a physical meaning surface tension component becomes weak
controlling property in oil recovery and could be given to the past empirical ob- in the SPT definition of parachor. It ap-
surface tension data may be needed for servation that the parachor is related to pears what the authors did was to compare,
the solution of problems with equipment molecular dimensions. The authors con- in effect, two definitions of molecular
design for operations like separation of tinue: ‘However, there is no fundamental dimensions, one based on statistical me-
phases and mass transfer. It is possible reason for the choice of a function such chanics (hard sphere model) and the other
that the approximate values of surface as the parachor to describe the surface on calculations based on van der Waals
tension obtainable from calculations tension of a liquid. It just happens that molecular volume – a slope close to 1.0
based on parachor equivalents (e.g. of with this particular choice of function of cannot be regarded as a surprise.
the –CH2– group) were sufficient for the the surface tension [and] the packing frac-
calculation, for example, of phase equi- tion function f (y) is fairly independent of
libria of fluids confined inside porous the chemical nature of the substance and The present position
media (e.g. oil shale), a situation natural hence, a fairly simple universal relation-
to petroleum fluids. Theoretical appro- ship with the hard sphere diameter [can A molecule is an assemblage of moving
aches to derive parachor or parachor-like be arrived at]’. particles (nuclei and electrons) which are
functions, especially those based on the The authors note, ‘The expression for held together by electrostatic and magnetic
employment of additional parameters of parachor given by Lennard-Jones and forces. It cannot be thought of as having
liquid properties or statistical mechanics Corner16d based on statistical thermody- a definitive boundary, but thinking of it
have been continued even lately32. namic considerations and using a lattice as a solid figure with a ‘volume’ has its
In an interesting combination of the model for the liquid was of the form: uses, such as in the study of intermole-
ideas of statistical thermodynamics and cular forces (e.g. those involved in surface
statistical mechanics, Satherley and [P] = const ε1/4σ 5/2, tension, especially at gas–liquid interfaces).
Schiffrin33 have analysed parachor from At this point it is interesting to recall
the point of view of what is called the where ε is the energy parameter in the that Lennard-Jones writing in 1940 won-
scaled particle (SPT; hard sphere appro- Lennard-Jones potential4. ‘The hard ders, ‘The additive nature of the parachor
ximation; Pierotti34) in attempting to sphere diameter dependence predicted has been found of great value to the
arrive at a fundamental understanding ‘in is the same as that obtained with the chemist. It is not obvious why this func-
view of its widespread practical use of SPT’. tion . . . should be additive when more
equations using parachor as a parameter’. The authors compared the values of complicated molecules are built up. If
The SPT predicts that the surface tension the hard sphere diameter calculated from this involved only the attractive term one
is given by the parachors of a large number and might be able to relate it to the numbers
variety of substances by plotting them of electrons in the molecules, using such
γ = σRT(2 + y)/4V (l – y)2, against those predicted from the van der general relations as the Slater–Kirkwood
Waals molecular volume Vw, calculated formula [italics mine]’. With the modern
where σ is the hard sphere diameter of from functional group additivity contri- methods of quantum chemistry, it looks
the molecules in a pure liquid, V the molar butions. The correlation of Vw with the possible to arrive at a definition of ‘mo-
volume, T absolute temperature, R gas con- hard sphere diameter σ was based on the lecular volume’, perhaps as the volume
stant, and y the packing fraction given by relation: enclosed within a contour surface (or iso-
cline) of an agreed electron density. From
y = πσ 3N/6V, (π /6)Nσ 3 = –10 + 1.13Vw. this it seems possible to obtain values of
specific atomic (plus structural?) contribu-
where N is the Avogadro constant. where σ is the hard sphere diameter. The tions. Can such a ‘volume equivalent’ be
The authors state: ‘From these two expected straight-line plot was obtained related to Einstein’s c parameters?
equations the parachor can be easily but with a slope of 1.07 instead of [ex- Recent methods of estimating molecu-
calculated: actly] 1.0. This was taken as a systematic lar volumes, based on theoretical princi-
deviation in the values of the calculated ples, have actually attempted the first
[P] = Mγ 1/4/ρl = (RN3π3/4 × 63)1/4T1/4 hard sphere diameters but since this was part of the enterprise trying to get at
/f (y)(σ3/σl/2), small ‘it is quite remarkable that a com- numbers that represent a volume en-
paratively simple hard sphere liquid closed within an isocline of electron den-
with theory can give such an excellent corre- sity. They can be seen as falling into two
lation with molecular parameters derived categories: van der Waals (referred to as
f (y) = [y(2 + y)]1/4/(y(1 – y)1/2)’. from entirely different considerations. vdW in the paper) and semi-empirical.
This result supports the idea that the mo- Rellick and Bechtel35 have compared the
‘Expressing the parameter as defined lecular parachor is really only a function results of calculating molecular volumes
above in the same units as employed by of molecular properties encompassed in of small molecules and proteins by the
Sugden one gets: the hard sphere diameter.’ two methods. They note, at first, that

CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008 1655

HISTORICAL NOTE
what may be taken as falling under ‘vdW from the regions of highest density to lo- c. Would the semi-empirical molecular
methods’ suffers from such inconsisten- cations an infinite distance away from volume calculations show progressive
cies as using arbitrary variations in vdW the centre of the molecule. Zero values diminution in the contribution from the
radii36 from calculation to calculation for were assigned to the points seen as hav- –CH2– group in the higher homologues
the atoms involved. However, a practical ing insignificant electron density. These accompanying changes that must be ex-
method for calculating molecular volumes values were then tested to determine pected in the additivity equivalents as the
could be to use interatomic distances de- whether they lie within a certain limit. If chain lengthens?
termined by X-ray crystallography in they did so, a location in a parallel array d. One expects that efforts to obtain
conjunction with atomic radii published corresponding to the point is assigned a parachor-like functions, using deep-
by Pauling or Bondi37 (refer also to Bay- value of one. If they did not meet, they based theories, for better technical appli-
liss’ work described above). The authors are assigned a value of zero. After this cations mentioned earlier in connection
detail certain simplifying assumptions process had been completed for all the with uses of parachor in the practical cal-
needed for what may be described as a points in the density array, the extent of culations would continue. These would
general method for calculating volumes inclusion of the points was varied to de- pertain mostly to the use of numbers
of congregations of interpenetrating termine what per cent inclusion most (generally referred to as ‘parachors’)
spheres of different radii. The principle, closely approximated the volume of useful in the field of chemical engineer-
outlined in a website38, is as follows: molecules, determined experimentally ing processes (miscibility, mass transfer,
Calculating the total volume of the (X-ray + vdW). One then had a three- etc.).
spheres requires summing the volumes of dimensional array of points, ones and ze- The subject of parachor should remain
each sphere (Figure 3), subtracting out ros, representing a map of the areas of in the syllabi of chemistry teaching,
the pairwise intersection volumes, since significant electron density. The simplest since its story is illustrative of the way
each was counted once for each ball it is possible method for calculating the vol- science moves – and moves on.
inside. The intersection volume of the ume at this stage is to determine the total
three balls must then be added back be- number of grid points that have electron
References and notes
cause, although it was added three times densities greater than or equal to that of
initially, it was also subtracted once in the limiting value, then to multiply this
Many of the references given are available
each of the three pairwise intersections39. number by V0, the volume of one element
only in special libraries, libraries of large re-
For their calculation of volumes of of the cubic grid. At sufficiently fine search institutions or through inter-library
small molecules (and small globular pro- values of the grid spacing, the calculated loans. However, abstracts of most of the pa-
teins), Rellick and Bechtel plotted elec- volume was taken as having been accu- pers can be viewed through search engines
tron-density contours employing semi- rately represented. In this approximation, (e.g. Google or ScienceDirect). Most journals
empirical methods using a popular soft- the electron density is tested at different require subscriptions for downloading full
ware package. The electron density at points along each of the three coordinate texts of papers. The present author has
each point of a 3D grid covering the axes in the cubic lattice. On the basis of downloaded and perused most of the refer-
molecule was next calculated. The grid their results, Rellick and Bechtel state ences from websites (not all of which are
mentioned) during the latter part of 2007 and
size and spacing between grid points that molecular volumes derived using
the early part of 2008.
were varied as were the orientation of the van der Waals radii are generally 30%
molecule within the grid. No assump- lower than such experimentally deter-
1. Murrell, J. N. and Grobert, N., Notes and
tions were found necessary regarding the mined volumes for small molecules. Records of the Royal Society of London,
values of the radii of individual atoms or They conclude that semi-empirical tech- 2002, 56, 89–94.
groups of atoms. The electron density niques are more reliable, less arbitrary, 2. (a) Sugden, S., J. Chem. Soc., 1924, 125,
value for each grid point was then used and are more accurate for the determina- 1177–1189; (b) Sugden, S., The Parachor
to calculate the volume of each molecule tion of molecular volumes. and Valency, Routledge, London, 1930.
as a function of the percentage of the to- 3. Einstein, A., Ann. Phys., 1901, 4, 513–
tal calculated electron density. 523 (English translation, in The Collec-
More particularly, the density values Conclusions and outlook for the ted Papers of Albert Einstein, Supple-
mentary Volumes, Princeton University
output by the semi-empirical calculations future
Press, Princeton, New Jersey, 1989, vol.
represent a continuum approaching zero
2, pp. 1–11; There is an analysis of this
a. Of great historic importance, possibly, paper in vol. 2, p. 3.
would be a plotting of Einstein’s c values 4. For a definition, see http://en.wikipedia.
against the values for the elements based org/wiki/Lennard-Jones_potential and al-
on semi-empirical molecular volume cal- lied websites.
culations. Would one obtain a straight 5. The necessity to use an angular term in
line? It may be necessary to recalculate calculating bulk properties from poten-
the Einstein’s c values using the latest tials may be attempted to be overcome
using an ‘acentric factor’, usually sym-
relevant physical data.
bolized by the Greek letter ω. It roughly
b. Would the semi-empirical molecular
measures the deviation of the intermo-
volume calculations also predict different lecular potential function from that of
additivity equivalents for the –CH2– group simple fluids (non-associated, ‘spherical
in different series? Would these follow particles’, etc.). Expressed in terms of its
Figure 3. Three overlapping spheres. the trends/changes seen in Exner’s plot? defining vapour pressure relation, the

1656 CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008

 HISTORICAL NOTE
acentric factor is a measure of vapour 12. Ferguson, A., Trans. Faraday Soc., 23. Noyes, W. A. and Singh, B., J. Am.
pressure deviation from simple fluid be- 1923, 19, 407–413. Chem. Soc., 1936, 58, 802–803.
haviour ascribable to size–shape, polarity, 13. Sugden, S., J. Chem. Soc., 1927, 1786– 24. (a) Exner, O., Collect. Czech. Chem.
association and quantum effects. Pitzer 1798; Sugden, S. and Wilkins, H., J. Commun., 1966, 31, 3222–3251; (b)
showed that the acentric factor can be Chem. Soc., 1929, 1291–1298; Ham- 1967, 32, 1–23; (c) 1967, 32, 24–55.
used to correlate real fluid properties mick, D. L. and Andrew, L. W., J. Chem. 25. Exner, O., Nature, 1962, 196, 890–891.
(Danon, F. and Pitzer, K. S., J. Chem. Soc., 1929, 754–759; Hunten, K. W. and 26. Cole, A. R. H., Hist. Rec. Austr. Sci.,
Phys., 1962, 36, 425–430 and previous Maass, O., J. Am. Chem. Soc., 1929, 51, 1996, 11, 191–211; http://www.asap.uni-
papers by Pitzer going back to 1939). 153–165; Hammick, D. L., Randal, C. melb.edu.au/bsparcs/aasmemoirs/bayliss.
6. Reference 1 says that earlier work by A., Sidgwick, N. V. and Sutton, L. E., J. htm#cole.
Schiff (Schiff, R., Liebigs Ann. Chim., Chem. Soc., 1930, 1876–1887; Ray, S. 27. Boudh-Hir, M.-E. and Mansoori, G. A.,
1884, 223, 47–106; Schiff, R., Berichte, K., J. Indian Chem. Soc., 1936, 13, 484– J. Phys. Chem., 1990, 94, 8362–8364
1885, 18, 1601–1606) and Ostwald (Ost- 485; Ray, S. K., J. Indian Chem. Soc., (parts of this paper, and others by the
wald, W., Lehrbuch der allgemeinen 1938, 15, 43–46 and several previous same school, were found not easy to fol-
chimie, vol. 1 (Stoichiometrie), Engel- papers; Mann, F. G., Purdie, D. and low partly because of apparent inconsis-
man, Leipzig, 1891, 2nd edn) were use- Wells, A. F., J. Chem. Soc., 1936, 1503– tencies in the notations used); Escobedo,
ful to Einstein in developing his own 1513 and earlier papers of this school; J. and Mansoori, G. A., AIChE J., 1996,
ideas. Gibling, T. W., J. Chem. Soc., 1941, 42, 1425–1433; extension of the work to
7. Getman, F. H., Outlines of Theoretical 299–304; Bhagwat, W. V. and Toshni- mixtures: Escobedo, J. and Mansoori, G.
Chemistry (ed. Daniels, F.), Wiley, New wal, P. M., J. Indian Chem. Soc., 1942, 19, A., AIChE J, 1998, 44, 2324–2332; full
York, 1937, 5th edn, p. 86 ff.; This work 225–230; Bhagwat, W. V. and Shukla, R. text of both papers available at: http://
quotes from the following papers by P., J. Indian Chem. Soc., 1950, 27, 283– tigger.uic.edu/labs/trl/SurfaceTensionPre
Kopp, H. F. M., Liebigs Ann. Chim., 284 (also, papers by Bhagwat et al., vol. dictionOfLiquidMixtures.pdf
1842, 41, 79; 1855, 96, 153, 303. These 29 and prior volumes of the J. Indian 28. For example, Negi, A. S. and Anand, S.
original papers were not available to the Chem. Soc.). C., A Textbook of Physical Chemistry,
present author. 14. (a) Quayle, O. R., Chem. Rev., 1953, 53, Wiley-Eastern, New York, 1986 (re-
8. See Glasstone’s classic treatise: Glass- 439–591 (see http://chemistry.emory. viewed in Nature, 1987, 326, 214.
tone, S., Textbook of Physical Chemistry, edu); (b) Reid, R. C., Prausnitz, J. M. and 29. (a) Tichy, M., Int. J. Quantum Chem.,
Macmillan, London, 1948, 2nd edn; (a) Poling, B. E., The Properties of Gases 1979, 16, 509–515; Gómez-Jeria, J. S.
p. 524; (b) p. 440; (c) p. 495; (d) p. 491 and Liquids, McGraw Hill, New York, and Ojeda-Vergara, M., J. Chilean Chem.
ff. (original reference: Eötvös, R. von, 1987. Soc., 2003, 48, N4; (b) Katritzky, A. R.,
Ann. Phys., 1886, 27, 456); (e) p. 494 15. (a) Mumford, S. A. and Phillips, J. W. Lobanov, V. S. and Karelson, M., Chem.
(original reference: Katayama, M., To- C., J. Chem. Soc., 1928, 155–162; (b) Soc. Rev., 1995, 24, 279–287; Murugan,
hoku Univ. Sci. Rep., 1916, 4, 373); (f) p. 1929, 2112–2133; (c) J. Am. Chem. Soc., R., Grendze, M., Toomey Jr, J. E.,
526; (g) p. 415; (h) p. 614 (original ref- 1930, 52, 5295–5297. Katritzky, A. R., Karelson, M., Lobanov,
erence: Pascal, P., Compt. Rendu, 1909, 16. (a) Ferguson, A., Nature, 1930, 125, V. and Rachwal, P., CHEMTECH, 1994,
149, 342, 508; review: Melville, H. W., 597–598 (This author who refers to the 24, 17–23.
Morton, R. A., Angus, W. R. and Thomp- parachor as ‘a healthy baby’ also says it 30. Exner, O., J. Phys. Org. Chem., 1997,
son, H. W., Ann. Rep. Prog. Chem., has proved, as a weapon of attack on the 11, 797–813.
1941, 38, 7–64; (i) p. 528; (j) p. 496 for problems of chemical constitution, much 31. See, for example, http://www.pirika.com/
the complexities caused by ‘association’ more powerful than the molecular chem/TCPEE/ST/MS.htm. An Internet
e.g. hydrogen-bonding; (k) p. 238, 531, volume.); (b) Ferguson, A. and Kennedy, site http://www.pirika.com/chem/TCPEE/
477. Original papers of Katayama and S. J., Trans. Faraday Soc., 1936, 32, ST/MS.htm has a Java applet by which
Pascal were not available to the present 1474–1481; (c) Fowler, R. H., Proc. R. one can estimate the surface tension by
author. Soc. London, Ser A, 1937, 159, 229–246; the ‘McLeod–Sugden empirical relation’
9. The molar volume of a liquid, defined as (d) Lennard-Jones, J. E. and Corner, J., for any small molecule. You ‘draw the
the volume occupied by a mole of liquid Trans. Faraday Soc., 1940, 36, 1156– molecule’ and enter the density, pre-
under atmospheric pressure, is obtained 1162; (e) Telang, M. S., Curr. Sci., 1943, dicted by the applet using what its origi-
by dividing its molecular weight by its 12, 19; (f) Telang, M. S., J. Am. Chem. nator calls ‘a neural network method’
density. The molar volume at the boiling Soc., 1949, 71, 1883–1885. and get an estimate of the surface ten-
point is obtained by extrapolation from 17. Bayliss, N. S., J. Am. Chem. Soc., 1937, sion.
measurements of densities at lower tem- 59, 444–447. 32. Weinaug, C. F. and Katz, D. L., Ind.
peratures. The choice of values at boiling 18. British Association for the Advancement Eng. Chem., 1943, 35, 239–246; Harla-
points as the temperatures at which the of Science Report, 1932, p. 284. Au- cher, E. A., A four parameter extension
molar volumes of different liquids are to thors’ names not available. of the theorem of corresponding states,
be compared is known, post Kopp, to 19. References for these data are (a) 2b, p. Ph D thesis, The Pennsylvania State
have a theoretical basis in the ‘Law of 181; (b) 15b; (c) Vogel, A. I., J. Chem. University, 1968; quoted by Harlacher,
Corresponding States’8b. Soc., 1934, 333–341; (d) Desreaux, Bull. E. A. and Braun, W. G., Ind. Eng. Chem.
10. Macleod, D. B., Trans. Faraday Soc., Soc. Chim. Belg., 1935, 44, 249–287. Process Des. Dev., 1970, 9, 479–283;
1923, 19, 38–42. 20. Pauling, L. and Huggins, M., Z. Krist., Kalback, W. M. and Starling, K. E.,
11. Notts, T. A., Wilding, W. V., Oscarson, 1934, 87, 205–238. Proc. Oklahoma Acad. Sci., 1976, 56,
J. L. and Rowley, R. L., J. Chem. Eng. 21. Brockway, L. O. and Wall, F. T., J. Am. 125–132 (http://digital.library.okstate.edu/
Data, 2001, 46, 1007–1012; see also da- Chem. Soc., 1934, 56, 2373–2379; oas/oas_htm_files/v56/p125_132nf.html);
tabase maintained by the American Institute Brockway, L. O., J. Am. Chem. Soc., Edalat, M., Bozar-Jomehri, R. B. and
of Chemical Engineers in the Design Insti- 1935, 57, 958–959. Mansoori, G. A., Oil Gas J., 1993, 39–
tute for Physical Property Data at http:// 22. Brockway, L. O. and Jenkins, H. O., J. 40; Riazi, M. R. and Mansoori, G. A.,
www.aiche.org/TechnicalSocieties/DIPPR Am. Chem. Soc., 1936, 58, 2036–2044. Oil Gas J., 1993, 108–111; Miqueu, C.,

CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008 1657

 HISTORICAL NOTE
Broseta, D., Satherley, J., Mendiboure, 33. Satherley, J. and Schiffrin, D. J., J. 1942; Bondi, A., J. Phys. Chem., 1964,
B., Lachaise, J. and Graciaa, A., Fluid Chem. Phys., 1992, 97, 2168–2169. 68, 441–450.
Phase Equilibria, 2000, 172, 169–182 34. Pierotti, R. A., Chem. Rev., 1976, 76, 717– 38. http://cnx.org/content/m11616/latest/#Volu
(these authors quote Danesh, A. S., 726. meCalculation
Dandekar, A. Y., Todd, A. C. and Sarkar, 35. Rellick, L. M. and Becktel, W. J., Bio- 39. In the general case, with n balls, all of
R., SPE 22710, Dallas, TX, 1991. Origi- polymers, 1997, 42, 191–202 (http://www. which may overlap, intersections of odd
nal of this paper is not available to the ncbi.nlm.nih.gov/sites/entrez?cmd=Retriev numbers of balls are added, and intersec-
present author); Broseta, D., Meleán, Y. e&db=PubMed&list_uids=9234998&dopt tions of even numbers of balls subtracted
and Miqueu, C., Fluid Phase Equilibria, =AbstractPlus; see also http://www. to calculate the total volume.
2005, 233, 86–95; Espósito, R. O., Ta- ccl.net/cca/documents/molecular-modeling/
vares, F. W. and Castier, M., Braz. J. node5.html). S. N. Balasubrahmanyam formerly at the
Chem. Eng., 2005, 22, 93–104 36. For definition see http://en.wikipedia. Department of Organic Chemistry,
(http://www.abeq.org.br/bjche); Ayirala, org/wiki/Van_der_Waals_radius Indian Institute of Science, Bangalore
S. C. and Rao, D. N., J. Phys.: Condens. 37. Pauling, L., The Nature of the Chemical 560 012, India.
Matter, 2004, 16, S2177. Bond, Cornell University Press, NY, e-mail: [email protected]

Edited and published by P. Balaram, Current Science Association, Bangalore 560 080.
Typeset by WINTECS Typesetters (Ph: 2332 7311), Bangalore 560 021 and Printed at Printek Printers, Bangalore (Ph: 2328 7763)

1658 CURRENT SCIENCE, VOL. 94, NO. 12, 25 JUNE 2008

View publication stats