вход по аккаунту


I. The protium-deuterium ratio and the atomic weight of hydrogen. II. A new twin-ring differential surface tensiometer

код для вставкиСкачать
University Library
Manuscript Theses
Unpublished theses submitted for the Master's and
Do ct or’s degrees and deposited in the Northwestern University
Library are open for inspection, but are to be used only with
due regard to the rights of the authors.
references may be noted, but passages may be copied only with
the permission of the authors, and proper credit must be given
in subsequent written or published work.
Extensive copying
or publication of the thesis in whole or in part requires
also the consent of the Dean of the Graduate School of
Northwestern University.
This thesis by
has been used by the following persons, whose signatures
attest their acceptance of the above restrictions.
A library which borrows this thesis for use by
its patrons is expected to secure the signature of each user.
for the degree
JUNE 1940
ProQuest Number: 10102026
All rights reserved
The quality o f this reproduction is d e p e n d e n t upon th e quality o f th e c o p y subm itted.
In th e unlikely e v e n t th a t th e author did not send a c o m p le te manuscript
an d th e re are missing p ag es, these will b e n o te d . Also, if m aterial had to b e rem o ved ,
a n o te will in dicate th e d eletio n.
ProQuest 10102026
Published by ProQuest LLC (2016). Copyright o f th e Dissertation is held by th e Author.
All rights reserved.
This work is p ro te c te d against unauthorized copying under Title 17, United States C o d e
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106 - 1346
The author wishes to express his indebtedness
to Dr. Malcolm Dole for the valuable assist­
ance and suggestions which have been so
generously offered during the course of
these investigations.
The example furnished
by his devotion to science and his unfailing
optimism has provided a constant source of
enc ouragement •
|Part Is The Protium-Deuterium Ratio and the Atomic
Weight of Hydrogen.
. . . •
Protium-Deuterium Ratio
Relative Atomic Weight of Oxygen in Air and Water. 11
Atomic Weights of Hydrogen and Oxygen . . . . . . 13
Exchange Reaction; H® + HDO = HD + HgO........... 1 4
General Procedure
.................... . . . . . 18
Preparation of Deuterium-Free W a t e r ..................18
Preparation of Water Containing Deuterium-Free
Hydrogen and Atmospheric Oxygen
......... 20
Preparation of Deuterium-Free Water Containing
Normal Oxygen
............................ 21
Exploratory Experiments
........... . . . . . . 2 3
Equilibration Experiments..........
Difference in Density Between Waters Containing
Air and Water O x y g e n ............................ . 2 6
Purification of Water Samples, and Density
Calculation of the Protium-Deuterium Ratio........... 3 2
The Relative Atomic Weight of Atmospheric and
Water Oxygen..............
The Atomic Weight of Hydrogen
.................... 36
Summary . .
• • • • • • • 3 7
Part II: A New Twin-Ring Differential Surface Tensiometer.
Introduction • • • • • • • • • • • • • • . • • • • • . • • 3 9
The Relation Between Surface Tension and Con­
The Ring M e t h o d .........................
. . . 51
General Procedure • • • • • ............
Developments Leading to the Adoption of the
Ring Method • • • . . • • • • • • • • • • • • . • • 6 0
The Ring and Support............................ 61
Auxiliary Apparatus . . . . ................ . 6 2
Table of Contents (continued)
Cleaning Procedure and Preparation of Solution. . . .
Measurement Procedure. . . . . . .......... • . . .
Elimination of the Apparent Change of Surface
Tension with Time. . .
Discussion of R e s u l t s . ..............
The concentration of the heavier isotope of hydrogen,
deuterium, in normal water has been determined by a number of
investigators who have obtained an almost equal number of
diverse values.
In view of these discrepancies which lie out­
side the experimental error of the measurements, it was con­
sidered advantageous to devise an entirely independent method
which would eliminate the uncertainties inherent in previous
An exact knowledge of the protium-deuterium ratio Is
essential for an accurate value of the chemical atomic weight of
Until this ratio has been definitely established to
limits far within the present range from 5000 to 7000, a precise
statement of the atomic weight of hydrogen to more than four
significant figures is meaningless.
Of the many previous measurements only a few are free from
obvious error In method or technique.
An agreement between the
value obtained in this independent Investigation and one or more
of these few should attest to the accuracy of this figure.
In addition to the increase in accuracy which has been
obtained by the elimination of dubious procedures, a greater
sensitivity In the actual physical measurements to such an
extent that our density values are concise to one-tenth of a
part in a million contributes to the reliability of the results.
Until it was found that the isotopic composition of air
and water oxygen differed, the discovery of the presence of
[more than one isotope of oxygen had had no effect upon the
firmly embedded use of the value of 16.0000, the atomic weight
of oxygen, as the standard for atomic weights.
The justification
for this standard becomes questionable if the atomic weights of
samples from different sources vary.
literature show varying values from
The published data in the
parts per million
for the difference in density of water containing air oxygen
and water containing normal water oxygen.
In order to establish
this important constant of nature, a reinvestigation of the
density of water made from atmospheric oxygen as compared with
that of normal water was simultaneously made.
The Protium-Dauterium Ratio.
Following the discovery of the hydrogen isotope of atomic
weight two, numerous attempts were made to obtain an accurate
value for the concentration of this isotope, which can be con­
veniently expressed as the ratio between the concentrations of
protium, the isotope of mass one, and deuterium, the isotope
of mass two.
Although the presence of more than one isotope of hydrogen
was indicated by Allison
from results of his magneto-optical
method, the first definite and quantitative proof was given by
Urey, Brickwedde, and Murphy
in 1932#
They evaporated large quantities of liquid hydrogen at the
triple point and collected the gas which was evolved from the
last fraction of the last cc.
spectra of H
This was examined for the atomic
and H
In a hydrogen discharge tube.
Faint lines
occurring in the calculated positions permitted an estimation
of the abundance ratio, H/D, as 4000.
No evidence for H
was found.
This ratio agreed favorably with previous predictions of
BIrge and Menzel
who estimated it to be 4500 in order to
account for the discrepancies between the chemical atomic weight
of hydrogen and Aston t
mass-spectrographic value measured in 1927.
(1) F. Allison, J •Ind.Eng.Chem.,4,8 (1932); J.Chem.Ed.10,7 (1933).
(2) H.G.Urey, F.G.Brickwedde, and G-.M.Murphy, Phys.Reff. ,40,13
(1932); ibid, 3£, 164 (1932); ibid, 39, 864 (1932).
(3) R.T.Birge and D.II.Menzel, Phys.Rev., 37, 1669 (1931).
(4) F.W.Aston, Proc.Roy.Soc., A115, 487 (1927).
However, widely divergent values were obtained by a series
of investigations which followed, showing a much lower concens
tration of deuterium. Thus, Bleakney found 30,000 as the ratio
by a mass-spectrographic method which was confirmed by Hardy,
Barker, and Dennison
who obtained 35,000 in a study of the
infrared spectra of HC1 and DC1 and by Tate and Smith
Likewise, a rough estimate by Rank
using a high dis­
persion prism spectrograph agreed with this order of magnitude.
In opposition to these, Stern and Volraer
had previously
stated that deuterium could not exist In a greater concentration
than that represented by the value 100,000.
was given by Menzel
A similar figure
and Unsold
for the cosmic concentration
of H .
The reason for these large deviations became apparent when
Washburn and Urey
discovered that the isotopes could be sep­
arated by electrolytic fractionation.
Since electrolytic
hydrogen had been used by the various investigators, none of
them had been dealing with hydrogen in its natural state.
Employing this principle, Lewis and Macdonald
made a more
accurate estimation basing their calculations on the fractionation
of hydrogen isotopes in electrolytic cells.
Their results dis-
C5) W.Bleakney, Phys.Rev., 41, 32 (1932).
(6 ) J.E.Hardy,E.F.Barker,and P.M.Dennison,Phys.Rev.,42,279 (1932).
(7) J.T.Tate and P.T.Smith, Phys.Rev.,43, 672 (1932).
(8 ) D.H.Rank, Phys.Rev., 42, 446 (193277
(9) O.Stern and M.Volmer, Ann.der Physik.,59, 225 (1919).
(10) D.H.Menzel, Astron.Soc.Pac. 44, No.257, 41 (1932).
(11) Unsold, Naturwiss., 20, 936 71932).
(12) E.W.Washburn and H.C.Urey, Proc.Nat.Acad.Sci.,18, 496 (1932).
(13) G-.N.Lewis and R.T.Macdonald, J.Chem.Physics, 1, 341 (1932).
credited the ratio of 30,000 as follows.
It was found that the
electrolysis of some cell residues of a concentration of 3000,
as determined from the density which was 1.000034, to a volume
two-thirds of that of the cell residue increased the deuterium
concentration hy fifty per cent.
Prom this the fractionation
factor was calculated from which the total amount of
had "been concentrated was found.
This corresponded to a greater
amount of deuterium than had been added to the cells in the course
of their history on the basis of the 30,000 ratio.
either this was too high or an accumulation of heavy oxygen had
The latter possibility was eliminated by a simple
experiment in which a sample of the water obtained in electrolysis
was distilled over hot iron wool which converted part of the
water to hydrogen.
The hydrogen was combined with the oxygen
of CuO and the density of this water found to be the same as
that of the original sample.
A stream of ordinary hydrogen was
passed over the hot Pe^Ga giving water containing all of the
oxygen of the original sample, but having a density equal to
that of ordinary water.
Since no change in the oxygen isotopic concentration was
discovered, the only alternative was to assume a higher deuterium
This value was calculated to be 6500 from the
increase in density of water which was concentrated in an
electrolytic cell, the efficiency of which was known.
objection to their method lies in their failure to correct for
the fractionation of the oxygen isotopes during electrolysis.
Bleakney, himself, added proof to the incorrectness of his
former ratio when he and Gould
determined the abundance of
deuterium in rain water by the same mass-spectrographic method
to find 5000.
A redetermination using electrolytic hydrogen
showed electrolytic fractionation to be the actual cause of the
previous error since they again found a low concentration of
deuterium in a ratio of 25,000.
Bleakney and Gould used the
novel method of decomposing their rain water by passing it
repeatedly over iron turnings at 510°C.
It is interesting to
note that as late as 1938 the International Committee on Atomic
apparently considered this as the most valid ratio
which had been determined, and used it in their revision of the
atomic weight of hydrogen.
A limit of 7000 for the Ii/D ratio was set by Applebey and
on the basis of electrolyses and the assumption of a
reasonable value of the electrolytic separation factor.
making many electrolyses In which the initial and final electrolyte
volumes and deuterium concentrations were known, a calculation
of ex., the separation coefficient or factor, was made.
using these data and assuming successive deuterium concentrations
of 4500, 5500, 6500 and 7500, the separation factors were cal­
culated and plotted against the mean D s 0 concentration in per
Above a concentration ratio of 7000 the separation factor
was found to be negative, allowing them to place this as the
limit for the minimum concentration of deuterium.
(14) W.Bleakney and A.J.Gould, Phys.Rev., 44, 265 (1933).
(15) G.P.Baxter, O.Honigschmid and P.LeBeau,J.Am.Chem.Soc.,60.
737 (1938).
(16) M.P.Applebey and G.Ogden, J.Ghem.Soc., 163 (1936).
If this limit is accepted, the results of later experi­
ments -which give lower concentrations become questionable.
Another ratio was obtained by means of a calculation
similar to that of Lewis and Macdonald by Edwards, Bell and
Their ratio of 6200 differed but slightly from
Lewis 1 6500.
These methods which have been summarized were based upon
either mass-spectrographic measurements or upon miscellaneous
calculations as described.
A third general type of determination
having greater accuracy and sensitivity has displaced the other
two to a large extent.
In essence, this so-called density
comparison method consists of obtaining water containing a
minimum amount of deuterium by subjecting a water sample to
repeated fractional electrolysis, fractional distillation, or a
combination of both of these.
To be more specific, when water
is electrolysed, the hydrogen coming off first at the cathode
contains less deuterium than does the water.
If this hydrogen
is combined with oxygen, the water which is produced may be
subjected to successive electrolyses.
Prom the difference in
density between the final product which should contain almost
pure protium and the original sample, the deuterium concentration
may be found.
If it were not for the complicating separation of the
joxygen isotopes during this electrolysis, the problem would be
greatly simplified, because the decrease in density of the final
(17) A.J.Edwards, R.P.Bell, and J.H.Wolfenden, Nature 155, 793
water would represent the change due solely to the removal of
all deuterium.
It is in the method of maintaining the oxygen
composition constant or of restoring it to its initial state,
that the methods of various investigations vary, and, also,
therein lie the sources of the main discrepancies between
A tabulation and criticism of the independent results
which have been obtained by this general method will eliminate
the obviously incorrect.
Such a tabulation is made in Table I,
Table I
The Difference in Density Between DeuteriumFree Water and Various Natural Waters
Source of Water
London, England
- 12.0 *
h /d
Crabtree, and
Melbourne rain water
and Brun
R jukan, Norway
C olumbus , Ohi o
Morita and
Osaka, Japan
Lake Mendota
7 020
Hall and Jones
Gabbard and Dole
1 /IorIta and
Titani2 5
Lake Michigan
Osaka, Japan
Tronstad and Brun
Rjukan, Norway
(18) E.H.Ingold, C.K.Ingold. H.Whitaker, and R.Whytlaw-Gray,
Nature, 134, 661 (1934).
(19) W.N. Christiansen, R.W.Crabtree, and T.H.Laby, ibid,135,870 (1955)
(20) L.Tronstad, J.Nordhagen and J.Brun,ibid, 156, 515 (1935).
Of these, the first three are rendered questionable by the
omission in each publication of evidence for the correction for
the electrolytic fractionation of the oxygen isotopes.
possibility that the results of Christiansen, Crabtree, and
Laby indicating a low concentration of deuterium, might be due
to burning their deuterium-free hydrogen with atmospheric
oxygen thus obtaining too small a decrease in density is ex­
cluded by their statement that the evolved gases were burned, as
suggested by Morita and Titani.
The method of extrapolation which Johnston used is susceptible
to error, for, as Gabbard and Dole
showed, an extrapolation of
his data can give a value of -15.9 with equal mechanical
accuracy as the value -18.3 which he gave.
There is no apparent error in the work of Morita and Titani,
but in that of Hall and Jones there is the possibility that
their normalization of the oxygen in the final product by
equilibration with previously equilibrated carbon dioxide may
have been incomplete because of the slowness with which equi­
librium is reached in the reaction.
Evidence that the difference
between their results and those of Gabbard and Dole is dLie
only to the difference in equilibration of oxygen is given from
H.L. Johnston, J. Am.Chem.Soc., 57_, 484, 2737 (1935).
N.Morita and T.Titani, Bull .Chem. Soc. Japan, 11, 403 (1936).
N.F.Hall and T.O.Jones, J.Am.Chem.Soc., 58, 1915 (1936).
J.L.Gabbard and M.Dole, ibid, 59, 181 (1937).
II.Morita and Titani, Bull.Chem.Soc.Japan, 13, 515 (1938).
L.Tronstad and L.Brun, Trans.Faraday Soc.,54, 766 (1938).
is defined as the excess density of the water in question
over that of normal water in parts per million. A
negative sign for Y means that the water is lighter than
normal water.
the agreement of the values -8.9 and -9*0 V which were found
for the difference in density between water containing deu­
terium-free hydrogen and atmospheric oxygen and normal water.
The equilibration of Hall and Jones brought this down to -16.5 *
whereas Gabbard and Dole found -15.4 Y by subtracting an
average value for the difference in density between water con­
taining atmospheric oxygen and that containing aqueous oxygen,
jboth having the same hydrogen.
In Gabbard and Dole*s work the only probable error lies in
the determination of the value 6.4V ; the difference in density
between water containing atmospheric oxygen and that containing
aqueous oxygen both having the same Hs .
The method for doing
will be discussed later.
Morita and Titani consider their second value to be more
accurate than the first, but suggest that the isotopic comp­
osition of different waters varies too much to permit the
establishment of a universal value.
The last figure due to Tronstad and Brun is obviously
incorrect since their normal water appears to be that con­
taining oxygen having the same composition as atmospheric
oxygen, not water oxygen.
This elimination leaves three independently determined
ratios which may be considered to be free from serious errors,
although the uncertainty inherent In the method exists in each.
These ratios are 6550, 7020, and 6320 which were determined by
Hall and Jones, Gabbard
and Dole, and Morita and Titani,
The Relative Atomio Weight of Oxygen in Air and Water:
In 1935 Dole,
during an investigation of the deuterium
content of benzene and of Nevada hot springs v/ater, found that
water which was made from atmospheric oxygen was 4*6
than water containing aqueous oxygen with apparently the same
hydrogen. A repetition of this experiment
to eliminate
uncertainties in the initial determination increased this value
+ 0 * 6 p.p.m. corresponding to a greater atomic weight of
atmospheric oxygen of 0*000108 + 0 . 0 0 0 0 1 atomic weight units.
His methodconsisted essentially of "burning air with tank
hydrogen over a
copper catalyst using an excess ofhydrogen to
prevent fractionation of the oxygen isotopes.
This water and
normal Lake Michigan water were "both electrolysed and the
oxygen of each was combined with tank hydrogen.
The same
fraction of the water was electrolysed in each case with the
assumption that whatever fractionation of the oxygen isotopes
occurred would be the same in each electrolysis.
A calculation
showed that the difference due to the greater absolute electro­
lytic separation in the case of air ox ygen would be only
This so-called Dole effect has been
0 .1
confirmedby other
the results of which are tabulated In Table II.
Considering that the techniques were different and the
water samples were from widely scattered sources in each case,
the agreement Is very good.
However, the four dependent upon
(27) M.Dole, J.Am.Chem. Soc., 57, 2731 (1935).
(28) M.Dole, J.Chem. Physics, 4, 268 (1936).
Table II
The Difference in Density Between Water
Containing Atmospheric Oxygen and Vifater
Containing Normal Oxygen
Green and Voskuyl
Hall and Johnston
6.0 V
Air Oo + tank Hs
Cambridge,Mass. Air 03 + electro­
lytic H-g
C olumbus,Ohio
Air 03 + tank Hs
Exchange on a hot
Pt filament with
Gorr. for water
vapor-liquid dif­
Osaka, Japan
Air 0S + electro­
lytic Hs
Hall and Jones
Morita and Titani
Smith and Ma the son
Equilibration with
the electrolytic method are subject to uncertainties inherent
in the electrolytic separation of isotopes, while the equili­
bration methods involve isotopic exchange equilibria which are
functions of the temperature and to which corrections must be
It was In the endeavor to substantiate these assumptions
and corrections that the following independent method was
devised which is free from all the above errors.
C.H.Greene and R.J.Voskuyl, J.Am.Chem.Soc., 58, 693 (1936).
W.H.Eall and H.L.Johnston, ibid, 58, 1920 (1936).
T.O.Jones and N.F.Eall, ibid, 59, 259 (1937).
E.R.Smith and H.Math©son, J.Res.Nat.Bur.Standards, 17,
625 (1936).
The Atomic Weights of Hydrogen and Oxygen.
Since the adoption of a standard table of atomic weights
with oxygen as the standard, the value of 16 has been retained
for oxygen with occasional changes in the number of significant
Thus, we have the revisions with their respective
dates ; 3 3 16.0 (1894), 16 (1895), 16.00 (1896-1899), 16.000 (19001902), 16.00 (1903-1922), 16.000 (1925), 16.0000 (1931-1940).
Until the discovery of the presence of more than one iso34=
tope of oxygen by Giauque and Johnston
in 1929, it was
assumed that the arbitrary value represented the weight of one
atomic species.
The presence of more than one isotope meant
that the chemical atomic weight represents the weighted average
of the isotopic atomic weights, and is, thus, dependent upon
the relative abundance of the isotopes.
As long as the isotopic composition of oxygen was constant
for oxygen from all sources, this discovery introduced no error.
However, the discovery of the Dole effect, brought with it the
question of which oxygen was to be considered the standard.
If aqueous oxygen is taken as the standard with an atomic weight
of 16.000000 then air oxygen has an atomic weight of 16.000108,
according to his results.
In discussing this subject Dole
indicates the need for a revised atomic weight standard based
upon a pure isotope but different from the physicistrs standard,
the oxygen isotope of weight sixteen, which cannot be isolated.
(33) Lange’s Handbook of Chemistry, 3rd Edition, 1939.
(34) W.F.Giauque and H.W.Johnston, Nature, 125. 318,831 (1929).
(35) W.F.Giauque and H.Yif. Johnston, J.Am.Chem. Soc., 51, 1436,
3528 (1929).
The atomic weight of hydrogen has likewise undergone
revision since 1894
with a change from 1.008 (1894-1925) to
1.0078 (1931-1937), 1.0081 (1938-1939), and 1.0080 (1940).
As was mentioned on page
, the abundance ration h/d, was
taken to be 5000 in the revision of this figure by the Com15
in 1938.
According to the three most accurate ratios
which have been singled out, the concentration of deuteriimi is
less than is represented by this figure.
Therefore, the chemical
atomic weight of hydrogen is lower than the then accepted
The calculation of the committee was made on the basis
of 1.00785 for the atomic weight of H
and 2.01363 for H •
From these values, the mean atomic weight becomes 1.00805, or
The effect of changing the ratio from 5000 to 7000,
which is in the vicinity of the three ratios, decreases the
mean atomic weight to 1.00799, or 1.0080.
evident that an accurate value of the
h /d
From this it is
ratio is necessary
before more accurate methods for atomic weight determinations
are justified.
The Exchange Reaction, Ha + EDO = ED 4- Ha0.
Exchanges between the hydrogen isotopes have been observed
in a great many reactions involving both inorganic and organic
For a complete discussion of this subject reference
is made to reviews by Urey and Teal
and by Jones and Sherman.
A study of the exchange reaction, Ha + HDO - HD + Ha0,
(36) H.C.Urey and G.K.Teal, Rev.of Modern Phys.,7.* 72 (1935).
(37) T.Jones and A.Sherman, J.Chem.Physics, 5_, 375 (1937).
Indicated that this might be the solution to the problem of
removing deuterium from water without changing the oxygen
isotopic content.
Previous work with this reaction or with its
3 8-48
had proved that, although the exchange equilibrium
was reached very slowly, the rate was greatly accelerated by
some metals, of which platinum and, in particular platinum
hiack, were most active.
Thus, Horiuti and Polanyi
in a study of the reverse re­
action were able to reduce the deuterium content of a sample
of hydrogen containing 1.08^ deuterium to
.6 6 ^ deuterium by
shaking for one hour a quartz flask containing the gas and
acidified water at room temperature, whereas Oliphant
observed the exchange after six weeks when no catalyst was
The catalytic action of platinum black was found to be
greatest when the catalyst was dry,
still less when under water.
less when moist, and
Grease or mercury were found to
poison the catalyst at room temperature.
When the reaction
was carried out In solution, the presence of acid accelerated
the exchange whereas the presence of base decreased the rate.
(4 4 )
M.L.Oliphant, Nature, 132, 675 (1933).
J.Koriuti and M.Polanyi, Nature, 152, 819,931 (1933).
J.Horiuti and M.Polanyi, Nature, 155, 139 (1934).
Eley and M.Polanyi, Trans.Far.Soc*, 52, 1388 (1936)
A.Farkas and L.Farkas, J.Chem.Physics, 2_, 468 (1934).
A.Farkas and L.Farkas, Trans .Far .Soc.,30, 1071 (1934)
A.Farkas and L.Farkas, ibid., 32, 932 Tl936).
A.Farkas and L.Farkas, ibid., 55, 678 (1937).
K.F.Bonhoeff er and K.W.Hummell, Naturwiss.,22, 45 (1934).
K.F.Bonheoff er and K.W.Rummell, Zeit.fur Elektrochem. ,40,
469 (1934).
(48) R.K. Crist and G. A. Dal in, J* Chem. Physics, 2_, 735 (1934).
The equilibrium constant for the reverse reaction,
K = JeQ LHDQ) ^
first determined by Bonhoeffer and Rummel
[HD] [H2a
who made use of the catalytic action of Pt. At room temp­
erature K was found to be 3.1 + 0*1#
Jones and Sherman investigated the same reaction over a
range of temperatxires from room temperature to 1000°K.
III Includes their values of K and the values of K 1 for the
reverse reaction, which is equal to l/K, at the various temp­
Table III
Equilibrium Constants for the Exchange
Reaction and Its Reverse
2 *63s
1* 056
A plot of K* against T showed that at 450°G, K 1 = 0.825
and at 400°C, K 1 = 0.800*
Therefore, it seemed practical to
attempt the removal of deuterium from water vapor at this temp­
erature range using an excess of hydrogen.
This was necessarily
the upper limit because of the restriction set by Pyrex apparatus.
Using this value of K 1 at 450° a preliminary calculation
was made to determine the optimum ratio of hydrogen pressure
and water vapor pressure.
This calculation indicated that with
the pressure of H s 0 = 340 mm. and the pressure of Eg = 420 mm.
seven complete evaporations and condensations of a water sample
would be necessary to remove the deuterium to a concentration
of less than
The experimental procedures and apparatus involved in this
equilibration will be described in detail.
General Procedure:
The general procedure consisted of removing the deuterium
from a sample of Lake Michigan or Atlantic Ocean water by
application of the exchange reaction previously described.
Deuterium-free water was prepared by fractional electrolysis,
then electrolysed to produce deuterium-free hydrogen.
hydrogen was purified and passed through a sample of water at
The mixture of the gas and water vapor was passed over
a platinized asbestos catalyst at 450°C. where the exchange
The water vapor was condensed and the hydrogen was
recovered by burning with air over hot copper oxide.
process was repeated until a minimum in density of the water
sample was obtained.
The density of the final product con­
taining no deuterium and the same oxygen as the original sample
was compared with the density of normal water.
h /d
Prom this the
ratio was calculated.
By burning deuterium-free hydrogen in excess with oxygen
of the air water was obtained whose density could be compared
with that of normal water.
From the appropriate combination
of these data the relative atomic weight of atmospheric oxygen
was obtained.
Preparation of Deuterium-free Water:
The deuterium-free water, which served as a source of the
deuterium-free hydrogen was prepared by fractional electrolysis
-of water using the cells of Gabbard and Dole
, which were
constructed according to the design of Taylor, Eyring and
A bank of six Pyrex cells, each. 25 x 4 cm., and of 240 ml.
capacity, was immersed in a cooling bath.
The electrodes,
each 25 x 3 cm., were of nickel plate since it has been showed
that the separation factor for the hydrogen isotopes is greatest
on this metal.
A D.G. current of 8-10 amperes was furnished
by a mercury-arc rectifier operating from the 220 A.C. line.
Enough sodium hydroxide was added to the water to bring
the concentration to 4$ for the first two electrolyses.
the third and following electrolyses sodium peroxide was added
in proportionate amount in order to avoid the addition of
The hydrogen and oxygen which were evolved were passed
through a chain consisting of, first, a spray trap; second,
a mercury safety trap; third, a concentrated sulfuric acid
bubbler to remove sodium hydroxide; fourth, a cotton packed
tower to remove acid and water spray; fifth, a sand trap to
prevent striking back of the flame into the apparatus; and
sixth, a Pyrex jet of about 0.2 mm. diameter.
The gases were
iburned at this jet and the water vapor was condensed in a wide
Ibore condenser of 15 mm. inner dian&er.
In the first electrolysis one-third of the water was
electrolysed, whereas in subsequent electrolyses one-half or
less was electrolysed.
The data for the total amount of
^(49) H.S.Taylor, PI.Eyring, andA.A.Frost,
823 (1933).
P. P.M .
deuterium-free water which was prepared in the course of this
research is given in Table IV.
Table IV
Preparation of Deuterium-Free Water
Total electrolyte
added to cells
1st electrolysis
172,000 ml.
53,990 ml. 104,100 ml
2nd electrolysis
3rd electrolysis
5, 550
Preparation of Water Containing Deuterium-free Hydrogen and
Atmospheric Oxygen:
The complete elimination of deuterium from the water during
the electrolyses described above was proved by combining the
hydrogen from each successive electrolysis with atmospheric
oxygen over hot copper oxide, the hydrogen always being in
excess, and then by measuring the density of the resulting water.
When a minimum value of the density had been attained the water
was assumed to be deuterium-free. Data obtained In this work
jare given in Table V and are plotted in Figure 1, where the
^rapidity with which deuterium is eliminated by electrolytic
fractionation (separation factor of 5.2) is readily seen.
jlthe plot the diameter of the circles is twice the experimental
error of the density determination.
The minimum value is compared with that of Gabbard and
jDole and of Hall and Jones.
Considering the excellent agree­
ment between these independent measurements of the Jf value of
FIG. 2
deuterium-free water containing atmospheric oxygen, there seems
to be little reason to doubt the reliability of the final value
of -8.9 Y.
Table V
Data for the Densities of V/ater Made from
Hydrogen on Successive Electrolytic Frac­
tionation and Atmospheric Oxygen.
Humber of the electro­
lytic fractionation
(G-abbardand Dole)
fHall andJones)33
—8.9 + 0.1
-9.0 + 0.3
-8.9 + 0.3
Preparation of Deuterium-Free Water Containing Normal Oxygen:
The apparatus for the equilibration of normal water with
deuterium-free hydrogen is illustrated in Figure 2.
The deu­
terium-free hydrogen was produced by five electrolytic cells,
so constructed that the hydrogen and oxygen could not mix, as
is evident from Figure 3, in which the dimensions of the cells
are given.
The electrodes of nickel foil, 2 x 15 x 0.015 cm.,
were spot welded to nickel wire to which electrical connection
was made.
The hydrogen evolved from these cells by a current
of 10 amperes was freed of oxygen by passage through copper at
The water formed due to the oxygen was removed by
condensation then by passage through calcium chloride.
hydrogen then passed into the 500 ml. sample flask which was
immersed in a cottonseed oil bath maintained at 85-88°0.
Isample flask was especially designed with a ground glass stopScock for admission of the sample and with an entrance tube for
I-*-2.5 C M -*"1
6 C M -FIG. 3
the hydrogen ground Into the side.
The water vapor and
hydrogen emerged from this directly into the catalyst chamber.
The catalyst consisted of platinized asbestos which was pre­
pared by heating asbestos which had been previously soaked in
chloroplatinic acid to effect the decomposition to metallic
Earlier experiments had demonstrated the comparative
ineffectiveness of glass beads coated with a layer of platinum
black, presumably because of the smaller active surface.
platinized asbestos was contained in a Pyrex tube, 60 x 2.5 cm.,
and was surrounded by an electric furnace maintained at 450°C.
This temperature was frequently confirmed by direct measurement
with a quartz high-pressure mercury thermometer.
The electric
furnace was conveniently and quickly made by winding 30 feet of
No. 24 chromel wire around two spirally grooved alundum cores
placed end to end.
The cores were enclosed by a 1 3/4 inch
steam pipe casing, the ends of which were smoothed over with an
asbestos paste.
The water vapor emerging from the catalyst chamber was con­
densed in a water condenser, then in a trap Immersed in a dry iceacetone bath at -70 to -30°C.
The sample was weighed before and
after each equilibration to guarantee the absence of either the
escape of water vapor or the entrance of atmospheric oxygen into
the system.
Except for the exploratory experiments the maximum
loss in any series of eight equilibrations was 0.6# for a 300 g.
The hydrogen was burned with an excess of air; the
water vapor was condensed to be re—electrolysed m
of deuterium-free water.
the preparation
The Exploratory Experimenta:
In order to verify experimentally the preliminary cal­
culations which were made to determine the number of successive
equilibrations necessary to reduce the deuterium content to a
negligible amount, and which are described on page
, an
exploratory set of experiments was conducted in which the
density of the sample was measured after each equilibration.
In these and all following experiments the deuterium-free
hydrogen was conserved by using hydrogen from the second elec­
trolysis in the first three equilibrations, that from the third
electrolysis in the fourth and fifth, and that from the fourth
electrolysis in all succeeding equilibrations.
The final result of this exploratory set can be given no
weight because, first, a portion of the sample was lost during
the density measurements; second, the temperature measurements
at which floating equilibrium occurred were made with a Beckmann
Instead of a platinum thermometer; third, no extreme precautions
were taken in the purification of the samples, being distilled
only from alkaline permanganate In Pyrex apparatus; and fourth,
the hydrogen in the final equilibration may not have been
deuterium-free since no check was made upon it.
A 150 ml. sample of normal Lake Michigan water was taken
and six equilibrations made.
The data for the densities measured
after each equilibration are collected in Table VI and are
plotted in Figure 4.
From this it is evident that excellent
verification of the preliminary calculation was obtained*
: © = SAMPLE
Table VI
Data for the Densities of the Water Ob­
tained in the Exploratory Test of the
Exchange Fractionation
Equilibration Number
Equilibration Experiments:
Following this work, three new sets of experiments were
made to determine the concentration of deuterium in Lake
Michigan water*
One sample of 300 ml* was taken directly from
Lake Michigan while one of 300 ml* and one of 100 ml* were
taken from the laboratory distilled water supply*
showed that the laboratory distilled water has the same density
as that in the lake within +0.1 V.
on each sample*
Six equilibrations were made
After the sixth equilibration the water was
purified and the density determined by the improved method
which will be described shortly*
Repeated exchange equilibrations
were then made until the density of the resulting water had been
reduced to a minimum value (within the experimental
error) as
becomes apparent from Table VII and Figure 4.
Following these equilibrations a blank was run using
nitrogen instead of hydrogen in order to make certain that no
factor other than the removal of deuterium, such as a possible
change in the oxygen isotopic ratio caused the decrease in the
Nitrogen was passed through the apparatus in exactly
Table VII
Densities of Waters After Equilibration
with Deuterium-Free Hydrogen
Equilibration Number
Sanple Number
Sample 1 = 300 ml. of water from laboratory distilled
water supply.
Sample 2 = 300 ml. of water directly from Lake Michigan.
Sample 3 = 100 ml. of laboratory distilled water.
the same manner as was the hydrogen.
Two evaporations and
condensations of normal water were.made, then the density was
compared with that of normal water.
No difference in density
could be detected*
A 100 ml. sample of water taken from the Atlantic Ocean
45 miles east of Cape Ann, Massachusetts, was subjected to the
same number of equilibrations as was the Lake Michigan water.
Before equilibration the sample was completely distilled from
alkaline KMn04 over copper oxide at 450°C. to remove inorganic
salts and organic material which would contaminate the apparatus
and possibly poison the catalyst.
Table VIII gives the results
of density measurements made on this sample.
Table VIII
Data for the Density of Deuterium-Free
Atlantic Ocean Water
Equilibration Number
Normal ocean
Prom the last value of V In Table VIII it is evident that
normal ocean water Is 1.7 p.p.m. heavier than Lake Michigan
water, and that ocean water from which all deuterium has been
removed is 14.0 p.p.m. lighter than normal lake water.
indicates that the difference between normal ocean water and
water with no deuterium is equal to: 1 4 . 0 + 1 . 7 = 15.7 p.p.m.
Another equilibration might have been made upon the ocean water
since the difference in density between the water after equili­
brations six and seven is slightly outside the experimental
error, but by analogy to the other experiments seven equili­
brations should have been sufficient to remove all the deuterium.
Difference in Density between Water Containing Air and Yifater
The difference in concentration between the oxygen isotopes
in air and water can be calculated from the difference between
the densities of water containing air oxygen and no deuterium
CTable V) and water containing normal Lake Michigan oxygen and
no deuterium (Table VII).
VII is -15.5, from which A
The average of the values of Table
= -15.5 - (-8.9) = -6.6
Purification of Water Samples and Density Measurements:
Exploratory Experiments:
Of the three general methods for the measurement of density
(by pycknometer, falling drop, and the submerged float, either
at constant pressure or constant temperature), the totally submerged float at constant pressure was chosen.
(50) T.W.Richards and
36, 1 (1914)
(51) TTW. Richards and
(x) Reference is also
using the density
The previous work
J.W.Shipley, J. Am.Chem.Soc. ,54, 599 (1912) ;
G.W.Harris, ibid, 58, 1000 (1916).
made to almost all previous isotope work
5 0*51
using this method has proved its dependability and sensitivity,
A 100 ml, sample taken after each equilibration was dis­
tilled from alkaline KMnQ4 in Pyrex apparatus.
The first 25 ml.
was returned to the equilibration apparatus; the next 50 ml.
was retained for the density determination after which it, too,
was returned to the equilibration flask; the remainder was
completely distilled and returned to the flask.
The sample was placed in a Pyrex tube, 3 x 33 cm., with a
quartz float 7.5 cm. long, whose temperature of floating equi­
librium in pure water was around 24.85°C.
The tube was then
heated to 35°C. and outgassed by evacuation to 15-20 mm. pressure
with constant shaking to prevent excessive bumping.
The tube
was then placed in the constant temperature bath for thirty
minutes so that temperature equilibrium might be reached.
rate of rise or fall of the float was then measured by means of
a cathetometer, accurate to 0.01 mm., and a stop-watch.
change in height of the float was observed over a three minute
interval at constant temperature.
After the rate at two or
three temperatures had been observed, a plot of the change in
height for a three minute interval against the temperature gave
the temperature at which there would be no motion of the float.
The difference hetween the temperature of floating equilibrium
for the sample and for the standard water (which was purified by
jthe same method), multiplied by the temperature coefficient of
the density of water between 24.80 and 25.00°C, 0.000255, gave
the approximate difference in density considering the coefficient
of expansion of the quartz float to be negligible.
The constant temperature bath, for the construction and
design of which acknowledgment is gratefully made to Mr. J. L.
Gabbard, consisted of a glass jar, 30 x 32 cm., enclosed in an
asbestos-wool insulated wooden box which was provided with
vertical slots, three cm. in width, in the center of each side
to permit observation of the float.
The temperature was con­
trolled by means of cooling coils and intermittent heaters whose
action was governed by a toluene-mercury expansion regulator
and a vacuum-tube relay.
Although this relay has proved to be uneconomical to use
because of the overload on the plate current leading to frequent
replacement, it provided satisfactory control.
A description
of the design, which is due to Thiessen and Frost,
will be
omitted in favor of a discussion in Part II of this thesis of a
cold-anode tube circuit which was later found to be more
Since a Beckmann thermometer which can be read only to
+0.001° was used, the accuracy of the measurements was limited
to +0.6
which was sufficiently sensitive considering the other
limitations of this exploratory experiment which were sum­
marized on page
Equilibration Experiments: The technique for purifying the
sample and the measurement of its density was worked out
jcompletely by Dr. R. L. Slobod.
Reference is made to this
C52) G. W. Thiessen and L. J. Frost, J.Chem.Ed., 12, 72 (1S35).
detailed account which, was follovjed with no alteration and with
the same apparatus#
The description of the constant temp­
erature "bath, platinum resistance thermometer, temperature
control, and the purification apparatus is so thorough that
additional discussion would he superfluous*
It is sufficient in summarizing the advantages of this
improved technique to state that the accuracy of the actual
density measurement was better than 0.03 Y due to the accurate
platinum thermometer and the constancy of the temperature.
reproducibility of results was obtained by extreme precautions
during purification of the sample.
A 40 ml. sample was first
distilled from alkaline KMn04 over hot CuO at red heat.
second distillation from KMn04 acid with HaP04 followed In
which the first 10 ml. was discarded and the next 15 ml. was
retained for the density measurement.
The float used in these
measurements was a small Pyrex float of 1 ml. volume.
The extent of the Isotopic fractionation during the second
distillation was reduced by a special steam jacketed "condenser"
which prevented the refluxing of the water vapor.
A series of
density determinations with various fractions of waters of dif­
ferent Isotopic composition led Dr. Slobod to conclude that
the fractionation was greatest for the water having the highest
content, whereas the difference in hydrogen isotope content
caused but slight difference.
It was also concluded that the
15 ml. sample which was taken for the density measurement was
such that its density represented that of the original sample
.(53) R.L.Slobod, Ph.D. Dissertation, Northwestern University,
1939, pages 23-47.
due to the fact that the decrease in density of the first
part of the distillate was compensated exactly by the increase
in the density of the last part of the distillate within this
If this were true, no error would be present due to
difference in isotope fractionation of the samples*
This is contrary to results published after the completion
of this experimental work, by Greene and Voskuyl.
obtained evidence for a greater fractionation and subsequent
decrease in density of normal water, than for water containing
no deuterium.
If this is the case, the results of this in­
vestigation are too positive by as much as 0*25 Y*
The data for the isotopic fractionation during distillation
obtained by Slobod was reconsidered by Dole and Slobod.
theoretical calculation using the Rayleigh distillation formula
assuming one theoretical plate showed that the 15 cc. distillate
in the case of normal water was 0.58 Y. lighter than the original
sample, whereas the difference for deuterium-free water was
0.41 y.
The difference of 0.17 Y Indicates that the differences
in density which were obtained in this investigation are 0.17 Y
too positive.
Therefore, the deuterium-free water containing
normal oxygen should be 15.5 +0. 17 = 15.7 Y lighter than
normal water.
The error of 0.17
in the h/d ratio of 100.
corresponds to a difference
This is twice the experimental error,
but Is not sufficient to alter the conclusions which will be
made later regarding the chemical atomic weight of hydrogen (p.3£).
(54) C.H.Greene and R.J.Voskuyl, J.Am.Chem.Soc•,31, 1342 (1939).
(55) M.Dole and R.L.Slobod, J.Am.Chem.Soc.,62, 471 (1940)•
From this it is evident that the possible density errors
in the distillation and purification of water samples are the
limiting factors of the experiment.
The method of calculating the difference in density be­
tween a sample and standard water from the difference in the
resistance of the platinum thermometer at the temperature of
floating equilibrium for each sample is also given in detail
by Dr. Slobod.
Calculation of the Protlum-Deuterium Ratio.
Because the difference in d e n s i t y between normal water
and water containing no deuterium and normal oxygen is due
only to the difference In deuterium content, It is possible to
compute the concentration of deuterium in fiormal water.
order to do this It is also necessary to know the density of
pure deuterium oxide, Ds 0.
The most recently published values
for this, 1.10764 and 1.10763, were determined by Stokland,
Tronstad, and Ronaess
and by Johnston
respectively, and
have replaced the previously accepted value of 1.10790 given
by Selwood, Taylor, Hippie and Bleakney.
The first relation for the mol fraction of Ds 0 in terms
of these values, derived by Lewis and Luten
and later modified
by Baker and LaMer,
was found to give results In error by as
much as 0.5^: N(Ds0) = 9.377>4S - l.OLfiS ; where A S = 1 - dS5 .
measured the densities of H30-D3G mixtures of
known content and found that perfect solutions were formed
within the limit of his density measurements, i.e.,+ l x l O
Employing this concept he derived the equation which with
modification was used in this work.
(56) K.Stokland,E.Ronaess, and L.Tronstad,Trans.Far.Soc.,35,
312 (1939).
(57) I-T.L.Johnston,J.Am.Chem.Soc•, 61, 878 (1939).
(58) P.W.Selwood, H.S.Taylor, J.A.Hippie and W.Bleakney,
J.Am.Chem.Soc•, 57, 642 (1935).
(59) G.N.Lewis and D.B*Luten, J.Am.Chem.Soc., 55, 5061 (1933).
(60) W.N.Baker and V.K.LaMer, J.Chem.Physics, 3, 406 (1935).
(61) L.G. Longs worth, J.Am.Chem.Soc., 89, 1483 71937).
F op the ideal solution, Hs0 + Ds0:
1) V & = NiVi + Nsvs
Va = volume of a mol of solution
v = molal volume of a pure component
Subscripts x and s refer to H30 and D30
2) V = (NiMx + Hglvls )/d: V = actual molal volume of solution*
d = observed density*
A combination of equations (l) and (2) gives the relation
for the mol fraction of D30.
3) NCDa O) =
* 1 where 00 = Mi/Ms Cl-di/ds) = 9.235
& ~ Cm3 d^/da-Mi )/M3 (l*"d^/d3 )
/3= 0*0309
substituted the corrected values for the density
of D20 to obtain oc = 9.257 and (5 = 0*033.
Since the
h /d
ratio is given by the ratio of the mol
fractions of H3 0 and D30, the value 697o/l is obtained after
insertion of 15.5 x 10
7000 + 50 for the
h /d
for A S .
ratio in Lake Michigan water.
Ocean water for which V = -15*7 the
+ 50.
In round numbers this is
h /d
In Atlantic
ratio is 6880 or 6900
Table IX compares these results with the three previously
chosen as most reliable, recalculated from Swiftrs equation.
From this table it is evident that the results of this
investigation for both Atlantic Ocean water and Lake Michigan
water gives a smaller amount of deuterium in water than the
other investigators.
Indeed the method of analysis of the water
is such that it should give a minimum value; that Is, It is known
(62) E.Swift, J.Am.Chem.Soc., 61, 198 (1939).
Table IX
Density'Differences Between Deuterium-Free
Water and Normal Water From Various Investi­
Source of Water
This investigation
Lake Michigan
Atlantic Ocean
Lake Michigan
Lake Mendota
Osaka, Japan
Gabbard and Dole
Hall and Jones
Morita and Titani
697 0
that there is at least as much heavy hydrogen in water as was
found because the chief sources of error all tend to make the
h /d
ratio larger.
Thus, if all the deuterium was not eliminated
in the preparation of the deuterium-free water, if any oxygen
of the air leaked into the exchange apparatus, if the deuterium
free water became contaminated in any way with ordinary water,
the density of the deuterlum-free water would be greater than
it should be and the amount of deuterium would be correspondingly
too small.
However, the reduction of the density of normal
water to a constant minimum value and the fact that no water
was gained or lost in the exchange experiments seem to rule
out the first two possibilities.
Five ml. of ordinary water mixed with 100 ml. of the water
of V value equal to -15.5 would produce an error of 1 y, but
it is estimated that the amount of ordinary water gained by
the light water in the course of a density determination was
0.1 ml. or less since the distilling flask, condenser, receiver,
and density measurement tube were always dried out either In an
oven at 130°C. or by heating directly with a flame prior to
Introducing the light water.
The Pyrex float was the only
object covered, with a film of ordinary water that came into con­
tact with the sample under investigation.
It is firmly believed that this method, eliminating as it
does all uncertainty in regard to the isotopic composition of
the oxygen yields a reliable result, probably more reliable
than those of previous investigations.
ference was found between the
Since no great dif­
h /d
ratio for Lake Michigan water
and Atlantic Ocean water, the conclusion of Tronstad and Brun
that a variation in concentration of deuterium in waters from
different sources is the conclusion to be drawn from the
varying results of different investigators is contested.
factors contribute to this decision; first, possible sources
of error in previous investigations have been pointed out with
the elimination of some results; second, it has been showed
previously by Dole
that slight differences In the density of
water from various natural sources are due to variations in
oxygen Isotope ratios*
bility of this
h /d
An additional indication of the relia­
ratio is given In the following section.
The Relative Atomic Weight of Atmospheric and Water Oxygen.
It has already been indicated that the difference between
the density of water containing atmospheric oxygen and water
containing normal aqueous oxygen is 6*6
(see page ?6), the
water containing air oxygen being the heavier#
If this is
included in Table II, the average of the results excluding
(63) M.Dole, J.Chem.Physics, 4, 778 (1936).
that of Smith and Matheson, is 6.4 y'.
When this average is
added to the value for the difference in density between normal
water and water containing deuterium-free hydrogen and atmos­
pheric oxygen, 8.9 V
(Table V), the value 15.3 V is obtained
for the difference in density between normal water and deuteriumfree water having identical oxygen isotope ratios.
the data of Table VII are consistent with the data of Table II
and Table V, which is the additional criterion of the accuracy
of the ¥ value from which the
h /d
ratio and subsequently the
chemical atomic weight of hydrogen were calculated.
The value 6*6 V corresponds to a greater atomic weight of
atmospheric oxygen of 0*000119 + 0*000002 atomic weight units.
This value is easily reached from the consideration that for
1 ml* of water the difference in weight amounts to 0*0000066 g.
For the molecular volume of water, 18 ml*, the difference is
0.000119 g. corresponding to a greater atomic weight of air
oxygen of this numerical value.
The Atomic Weight of Hydrogen:
The chemical atomic weight of hydrogen was recalculated
substituting 6970 for the value of 5000 for the
ratio by
the same method as that of the International Committee on Atomic
As has been demonstrated on page 14, this alteration
reduces the atomic weight from 1*0081 to 1.0080.
As a direct result of the evidence given by this research
for the Incorrectness of the chemical atomic weight of hydrogen
and as a result of the suggestion made by Swartout and Dole
(64) J . A . Swartout and M.Dole, J.Am.Chem.Soc..61. 2025 (1939)
in their di sous si on of the problem, the Committee on Atomic
Weights of the International Union of Chemistry
has very
recently altered the chemical atomic weight of hydrogen from
1.0081 to 1.0080.
1. Water containing normal Lake Michigan oxygen and pure
protium has been prepared by an exchange reaction in the gas
phase: Hs + HOD = HD + HOH.
The density of this water has been
found to be 15.5 + 0.1 p.p.m. lighter than normal Lake Michigan
From this the ratio of hydrogen atoms to deuterium
atoms was calculated to be 6970 + 50.
2. Water containing normal Atlantic Ocean oxygen and pure
protium was prepared by the same method.
The difference in
density between this and normal Atlantic Ocean water was found
to be 15.7 + 0.2 p.p.m., corresponding to a
h /d
ratio of 6900
+ 50.
3. Deuterium-free hydrogen was burned with atmospheric oxygen
and the density of the water produced was compared with that of
water containing normal Lake Michigan oxygen and pure protium.
The atmospheric oxygen water was found to be 6.6 p.p.m.
heavier, which corresponds to a greater atomic weight of oxygen
in air of 0*000119 + 0.000002 atomic weight units.
4. The chemical atomic weight of hydrogen as calculated from
the new
h /d
ratio is 1.0080 instead of the value of 1.0081
j(65) G. P. Baxter, M. Guichard, 0. Hftnigschmid and R. Whytlaw
Gray, J.Am.Chem.Soc.,62. 669 (1940j.
adopted by the International Committee on Atomic Weights in
In accord with this result which was published by J. A.
Swartout and M.Dole, J.Am.Chem.Soc., 61, 2025 (1939), the
Committee on Atomic Weights of the International Union of
Chemistasyin presenting the 1940 table of atomic weights has
altered the chemical atomic weight of hydrogen from 1.0081 to
Until very recently the measurement of surface tension
was not performed with a high degree of accuracy in the labor­
Because of the experimental difficulties which were
involved the slight differences between the surface tension
of pure water and the surface tension of dilute salt solutions,
that is of concentrations below 0*01 N, could not be determined.
Therefore, it was assumed that the surface tension of these
solutions followed the theory which had been applied to more
concentrated solutions in such a way that the surface tension
Increased linearly with concentration as will be discussed in
detail In the next section*
However, very accurate measurements of Jones and Ray by
the capillary rise method Indicated that the slope of the
surface tension - concentration curve was negative up to about
0*001 N then positive from here to higher concentrations.
This Jones-Ray effect could be explained on a theoretical
basis as was showed by Dole who derived an equation for the
variation of surface tension with concentration which agreed
fairly well with the experimental data*
This apparent contradiction to the well developed theory
of Debye-Wagner-Onsager-Samaras was rendered questionable by
a theoretical consideration of Langmuir who proposed that this
lowering in surface tension was not real but, rather, an
apparent effect due to the elimination of the zeta-potential
on the glass wall of the capillary by the salt in the very
dilute solution.
In order to verify if possible one or the other of these
conflicting sides of the question, this research was undertaken
with its immediate object the development of a method of
measuring relative surface tension which did not involve the
use of glass walls as in the capillary rise method.
Of the
many other methods for the measurement of this physical
property, the du Nolly circular ring method appeared to be the
most promising.
With this as a basis it was proposed to develop
a twin ring method which would equal the oapillarimeter method
of Jones and Ray in accuracy, and by means of which the surface
tension of salt solutions over a range of concentration from
1 N to 0.0001 N could be studied.
N o te ?
myi <
. -^-v* c-1 * rn -t
yy] n 4- > 1 r> r t
a o (=> pr> oy-'
S ^
T ) I"'*’’'1 • p O T ’ t r i ^
t T 1"'1+■
TVr> s
V )* * r
T,/ro 1
f V l ^
<-n 1 "t” V* O T *
Tin 1
' ,ir^
4-Vi o
H o ir o l
n f*
o ^
The Relation Between Surface Tension and Concentration.
From the point of view of the surface tension of aqueous
solutions, soluble substances may be divided into two classes.
The first is composed of caplllary-inaetive materials which
raise the surface tension above that of pure water and which
have a lower concentration in the surface than in the bulk of
the solution.
The second is composed of capillary-active
materials whose concentration is greater at the surface, or
which are positively adsorbed, thus lowering the surface tension#
To the first group belong most of the inorganic salts, whereas
to the second belong chiefly the water soluble organic com­
pounds •
The relation between the concentration of these substances
and the surface tension of the solution has been the subject
of extensive research which has been briefly summarized by
An enlargement and modernization of this will be
given below.
The early work on salt solutions showed that the surface
tension increased as a linear function of the concentration.
An attempt to express this mathematically was given in Quincke fs
law, <rc/cr^ = 1 + kc, where
and <r0 are the surface tensions of
the solution of concentration, c, and of pure water.
It was found that ( 0 ^ - ^ )/ ^ c was not equal to k as this
(66) R.C. Brown, Rep. on Progress in Physics, 4, 40 (1937).
1(67) R.C. Brown, ibid, 5, 1 (1938).
relationship requires, but decreased with increasing concen­
tration below 0.2 N, descended to a flat minimum, and then
rose for solutions above 2 N.
Heydweiller and coworkers,
and ICleine
were the first to make accurate
measurements on dilute solutions below 0.2 N.
They used the
capillary rise method, observing the magnitude of rise very
In his paper (to which reference is made) Heydweiller70
summarises the work prior to 1910.
Because of the very slight difference between the surface
tension of dilute salt solutions and that of pure water, very
little experimental work has been done on solutions below
0.2 N.
using a modification of the du Notiy ring
method, made accurate measurements at 0°G of salt solutions
ranging from 0.015 to 0.18 N.
He used a very sensitive torsion
balance from which was suspended a straight platinum wire so
arranged that it lay parallel to the liquid surface.
A sub­
merged float was attached to the wire to extend the range of
the balance.
By this means an accuracy of 0.05$ was obtained.
At the lower concentrations the surface tension-concentration
curves for uni-univalent chlorides such as. potassium, sodium,
and lithium were identical, each showing a downward curvature
at low concentrations.
This curvature was in agreement with
previous results of Heydweiller and with the theory of Lenard
A.Gradenwitz, Diss.Breslau, 1902; Physik. Z. 3, 329 (1902).
A.Kleine, Diss.Munster, 1908.
A.Heydweiller, Ann.Physik. (4) 33, 145 (1910).
G.Schwenker, Ann.Physik. (5) 11, 525 (1931).
P.Lenard, Ann.Physik. (4) £7., 463 (1915).
proposing that the surface layer of water is negatively charged
at a water-air interface but that there is a positively charged
layer at lower levels.
Recognizing that the atoms or molecules of dissolved sub­
stance are surrounded by fields of force which are stronger
than those around the solvent molecules, Langmuir
that the surface layer should consist of a single layer of
solvent molecules from which the solute molecules are excluded.
This is due to the improbability of the solute molecules dis­
placing the solvent molecules in the surface as a result of
the inequality of forces.
This conception of a layer of pure
solvent was claimed to explain the requirement of the Gibb*s
equation, that surface tension increase linearly with concen­
tration, or that the deficiency of solute in the surface is
proportional to the concentration.
The interionic attraction theory of Debye and Hiickel was
applied to the theory of surface tension by Wagner
and by
Onsager and Samaras,
who gave an equation for the surface
tension of uni-univalent salt solutions as a function of the
^ 79.5X7 , .
-B~5T" lo®
1.143 x 10"13 (DT)
where D is the dielectric constant of the solution and T is
the absolute temperature.
This equation requires that the
(73) I. Langmuir, J.Am.Chem.Soc., 39. 1896 (1917).
(74) C. Wagner, Phys.Z., 25. 474 (1924).
(75) L. Onsager and N.N.T.Samaras, J.Chem.Physics, 2, 528 (1934).
surface tension increase gradually from the value for pure
water as a uni-univalent salt is added, that the curve obtained
by plotting relative surface tension against concentration be
the same for all uni-univalent salt solution, that a maximum
should be reached at a concentration of 0.54 N, that the
limiting slope at zero concentration should be plus infinity,
and that this curve should have a negative slope in the very
dilute regions.
The treatment which resulted In this equation was criticized by Ariyama
on the basis that it was too complex.
According to the Debye-Hfickel theory any ion in solution is
surrounded by an ion atmosphere having on the average a charge
opposite to that of the ion.
To simplify the treatment
Ariyama considered that this Ionic atmosphere could be replaced
by a sphere of radius a = l/m, with the charge distributed
evenly over the surface.
Then, any ion which lies at a depth
less than a below the surface of the liquid Is acted upon by
an incomplete sphere of influence which gives the ion a dif­
ferent potential energy from ions in the bulk of the solution.
By combining this consideration with the Boltzmann equation for
the relation between the concentration at any two points and
the energy of the particles at these points an expression for
the concentration of the ions on the surface was found.
a comparison of this expression,
oc - - c E3/kT4D
(76) K •Ariyama, Bull.Chem.Soc.Japan, 11. 687 (1956).
with the Gibb1s adsorption equation, <* =
pression for (dG")
"fdje]" ' 9 the ex­
was found to he.
4D #
Xn these equations ©^ is the excess concentration of solute on
the surface, E is the ionic charge, and IT is Avagadro*s number.
The last relationship requires that the surface tension of
dilute solutions should increase with the concentration at the
same rate for all salts regardless of the valence type*
Objections are present for these theories since no maximum
has been found in the surface tension curve, slight variations
have been found between different salts and especially between
valence types*
Theories have been developed using similar
reasoning by Oka
and by Belton
which also require that the
surface tension increase with increasing salt concentration.
An experimental contradiction to these theories was
obtained by Jones and Ray
who developed the capillary rise
method to an accuracy sufficient to measure the surface tension
of salt solutions down to 0*0001 N*
The extent of accuracy
which was required is indicated by fact that the surface tension
of a 1 N salt solution is only 3$ higher than that of pure water*
They used a U-shaped silica capillarimeter one. arm of which was
a microscopically graduated capillary tube, the other was a
tube of about 4*3 cm* diameter ground to a true cylinder*
(77) S.Oka, Proc.Phys.Math.Soc*Japan, (3), 14, 649 (1932)*
(78) J.W.Belton, Trans .Far. Soc* .55* 1449 (193*7).
(79) G-.Jones and W. A* Ray, J.Am.Chem.Soc. * 59* 187 (1937)*
Prom the weight of water necessary to bring the level in the
capillary to & given mark, from the weight of solution necessary
to bring the level to the same mark, and from the difference
in height between the levels in the capillary tube and large
tube the relative surface tension was calculated.
precautions were taken in the cleaning of apparatus, preparation
of water and of solutions.
The water and solutions were allowed
to stand in tall towers so designed that the liquid could be
withdrawn from beneath the surface after capillary active
material had risen to the surface.
The precision of the
measurements of densities, heights of liquids and weights per­
mitted an overall accuracy of 0.002$.
The relative surface tension of potassium chloride,
potassium sulfate, and cesium nitrate solutions of concen­
trations ranging from 0.0001 N to 3, 1, and 0.1 N were measured
at 25.00°C.
Contrary to predictions, the slope of the surface
tension - concentration curve was negative from zero concen­
tration to about 0*001 N -where a minimum in surface tension,
0.02$ below that of pure water was reached.
Prom here to
higher concentrations the surface tension increased becoming
equal to that of water at about 0.006 N.
The curves for the
three salts were very similar except that for potassium sulfate
the minimum was reached at 0.002 N.
Later results on a greater
variety of salts and valence types have agreed with the original
At higher concentrations, within the range of
previous investigations the relative surface tension agreed
(80) G.Jones and W.A.Ray, Private Communication.
with that of other workers#
For example, the curves were
approximately linear with a small positive slope hut with a
slight downward curvature in the dilute range below 0.1 N in
agreement with Heydweiller#
(crc -
The numerical values of
<roc were very similar to those of Schwenker*71 and
of Heydweiller.
To explain this negative slope which is greatly in
opposition to the plus infinity required by the QnsagerSamaras equation, Jones and Ray assumed that it was erroneous
to consider only the electrical forces between ions.
interionic attraction causing an increase in surface tension
is of slight importance in the dilute range but its contribution
increases linearly becoming effective above the minimum point,
0.001 N.
Below this range the electric forces between ions
and the dipolar water molecules are considered to be the
decisive factors. In accordance with the theory of Bernal and
these forces establish an ordered arrangement of the
When salt is added in very great dilution the
arrangement of the water molecules is disturbed around the
ions which results in a repulsion due to the water molecules
forcing the salt molecules to the surface#
In agreement with
the Gibbfs theory, this positive adsorption decreases the
surface tension#
Above 0.001 N the attractive forces between ions is greater
than between ions and water molecules causing a negative ad­
Since this is similar to the treatment of Onsager
(81) J#X)*Bernal and R.H.Fowler, J#Chem.Phys*, 1., 515 (1933).
and Samaras, Jones and Eay conclude that the previous theory
was merely oversimplified, whereas Ariyama has considered it
too© complex*
Considering that the data of Jones and Ray represented
true relative surface tension and assuming the validity of
the Gibbfs equation Dole
explain this anomaly*
derived an equation which would
His fundamental assumptions were;
first, to conform with the concepts of Langmuir73 and also of
the forces of the first adsorption layer predominate
over the weak long-range mirror image forces postulated by
Onsager and Samaras; second, that negative ions are adsorbed
at the interface at definite and limited spots; third, that
neither positive nor negative ions may be adsorbed on any
other location of the first adsorption layer.
He objected to
the treatment of Jones and Ray on the ground that ”the inter­
action between ions and the water dipoles produces a repulsion
from the surface and therefore negative adsorption”.
The assumption of a number of ”active spots” was not
entirely without foundation because a definite structure for
water has been proved by Stewart,
Bernal and Fowler,
by Debye.
That negative ions would be adsorbed to water
molecules on the surface whose proton ends point inward con-
M.Dole, J.Am.Chem.Soc*, 60* 904 (1938).
Harkins, Davies and Clark, ibid, 39* 541 (1917).
G.W.Stewart, Phys.Rev*, 35, 1426 Tl930); 37, 9 (1931).
P.Debye, Chem.Rev., 19, TTl (1936).
forms with ideas of Frumkin
although contrary to those of
Negative ions will he adsorbed on these active spots
causing the surface tension to decrease until a minimum is
reached when all of the spots have been filled.
From here to
higher concentrations they will be repelled by the long-range
mirror image forces since no more negative ions may be
This minimum should occur at the same concentration
for all salts, as has been shown by Jones and Ray to be true.
A mathematical development was given using Langmuir*s concept
of a layer of pure solvent with the addition of active spots.
The equation for the variation of surface tension with change
in concentration was as follows:
<Te - <T0 = RTk ( v__ + v+ )c -
( v— + v+ )ln \l + kv-c)
In this equation, k is t/lOGO, where t is the thickness of
the first adsorption layer; v- and v+ are the number of negative
and positive ions into which each molecule of solute dissociates,
a is the number of surface adsorbing spots per and is
given by a = Cjakv., where cm is the concentration at the minimum
of surface tension; W- is the adsorption potential and is
given by, W- = RT in ( kv-ci )
, where cx is the eoncenz kc-iV.
a(e— I— -i)
tration at which the surface tension again equals that of water
after passing through the minimum.
(86) A#Frumkin, Z.Physik.Chem., 111, 190 (1924).
(87) K.Ariyama, Bull.Chem.Soc. Japan, 12. 109 (1937).
When the relative surface tension was calculated from this
equation, a curve was obtained which agreed fairly well with
the data of Jones and Ray up to 0*02 N.
The data of Jones and Ray and the theory of Dole were
rendered questionable by a mathematical presentation of
which claimed that the experimental data did not
represent true relative surface tension and therefore, Dolefs
theory could not apply*
The essence of Langmuirfs argument was
that, when the surface tension of water was measured in the
capillarimeter of Jones and Ray, a film of water adhered to
the glass wall as result of the zeta-potential at the waterglass Interface.
This film effectively diminished the diameter
of the capillary causing the water to rise higher than It
theoretically should have risen.
When the dilute salt solution
was added the zeta-potential was neutralized, thus removing the
film and enlarging the effective diameter of the tube, giving
an apparent decrease in surface tension.
If this is correct,
then Jones and Ray should have calculated the surface tension
from the relation, cr= l/2^e>gh(r - Ar), instead of from the
Y = l/2 ^ghr. A r was calculated by Langmuir from the
X / J3
= (kT/e) (ttd/&( pgh - <r/v)
, and from the
previous equation.
The combination of these two gives,
i /s
= 2.16 x l(Ts/h
• The difference between the value for
the surface tension obtained by Jones and Ray and the true
value for water will be 0.0107 b.1^ .
This would account for a
(88) I . Langmuir, Science, 88 a 430 (1938).
lowering in surface tension of 0.G35 dynes per cm. whereas
Jones and Ray found a maximum lowering of 0.015 dynes per cm*
This calculation hy Langmuir, although admittedly approximate,
gives a result of the same order of magnitude as the lowering
found by Jones and Ray*
Langmuir suggested that his theory
could be tested experimentally by the addition of thorium
nitrate in a concentration of 10"s molar which would neutralize
the zeta-potential of water and would produce the same apparent
lowering as the 0*001 N salt solutions of Jones and Ray.
This has been done by Jones
who found a lowering in surface
tension which would seem to confirm Langmuir*s hypothesis*
The Ring Method for the Measurement of Surface Tension*
Although the determination of the surface tension of
liquids by measuring the force necessary to pull a circular
wire ring from the surface is usually referred to as the
du Ifofty method, work is recorded in the literature which far
antedates that of P. Lecomte du Nouy.
The credit for devising this method is probably due to
who worked with various sizes of rings up to twenty
cm. in diameter.
Among other observations it was noted that
the pull varied with the radius, of the wire and with the radius
of the ring; also, that a maximum pull was reached which
diminished before the ring broke.
These results were confirmed
(89) G.Jones, Private communication*
(90) P*Lecomte du Noiiy, J.Gen.Physiol. , 1, 521 (1919)*
(91) C. Sondhauss, Ann.der Physik., 8, 266 (1878).
by a number of later workers,
the work prior to 1892 being
well summarized by Cantor.
The first marked success in the attainment of any degree
of accuracy was obtained by Weinberg.
He used a copper ring
which was supported at three equally spaced points from a
second ring which was suspended from one beam of a balance.
In order to magnify the motion of the balance, a mirror was
attached to the beam giving a sensitivity of 0.05 mg.
weight on the opposite end of the beam was ingeniously changed
by allowing water to flow out of a cylinder in which a wire
attached to that arm of the balance was suspended.
this accuracy in actual weighing his results were not repro­
ducible to within 5 mg* due to an apparent decrease in the
surface tension with time amounting to 20-30 mg. in five minutes.
A scraping of the surface with nickeled bars was found to in­
crease the apparent surface tension until a maximum was reached.
This undoubtedly was due to the presence of organic impurities
on the surface which were eliminated by scraping, as shall be
seen later*
The study of this drift with time has been extensive.
Bigelow and Washburn
claimed that in the case of organic
solutions a falling surface tension curve indicates the ad­
sorption of the organic solution at the liquid-vapor interface
and at the liquid-glass interface.
Their experiments indicated
(92) M.Cantor, Ann.der Physik., 47. 499 (1892).
(93) T*P.Hall, Phil.Mag., 36, 385 (1893).
(94) P. Lenard, R.v.DallwItz-Wegener and E.Zachmann, Ann.der
Physik., 74, 381 (1924).
(95) A.W.Fahrenwald, J*0pt.Soc., 6^, 722 (1922)*
(96) B. Weinberg, Z.Physik.Chem., 10* 34 (1892)*
(97) S*L.Bigelow and E.R.Washburn, J.Phys.Chem., 32* 321 (1928)*
the absolute necessity for cleanliness and the great effect
that organic impurities have*
The importance of the pre­
vention of evaporation was also proved, since it was found that
the cooling which resulted greatly raised the surface tension*
In cleaning their apparatus, the glassware was treated with hot
alkaline potassium permanganate solution followed by hydro­
chloric acid, hot water, and then the removal of all moisture
by drying in an oven while dry, filtered air was drawn through
it* du Ho&y
had previously cleaned his apparatus by boiling
in a saturated solution of potassium dichrornate in sulfuric
acid, followed by a rinsing with grease free water.
platinum rings were washed and flamed before each measurement*
Especially emphatic were his warnings against the use of
either alcohol or ether for the drying of glassware or of the
ring* In contrast to this, Lottermoser and Stoll
their ring with hot water then with benzene before flaming to
a red heat*
Although these previous investigations appear to indicate
that the change of surface tension with time is due to the
presence of impurities, Lottermoser and Giese
sought to
explain this drift on the basis of concentration changes within
the solutions*
They found that if the ring was removed and
flamed after each measurement, there was a rapid rise in surface
(98) P*Lecomte du Noily, Biochem.Z*, 155, 113 (1925)*
(99) A.Lottermoser and P*Stoll, Kolloid*Z*, 65, 49 (1933)*
(100) A.Lottermoser and E.Giese, ibid, 75, 155 (1935).
tension from which they concluded that the surface tension is
a function of the number of measurements*
When a platinum
gauze was dipped into a solution of sodium lauryl sulphonate
then withdrawn, the surface tension increased.
Prom this it
was concluded that the platinum adsorbed capillary active
material from the body of the solution as a result of which
the concentration in the solution decreased causing an elevation
of the sufface tension*
It Is more logical however, that the
platinum gauze removed the capillary active material from the
surface thus cleaning the surface.
This point of view Is
confirmed by their observation that upon standing the surface
tension fell*
This was due to the diffusion of the capillary
active material from the interior to the surface.
That the observed change was not an actual change In sur­
face tension was indicated by a theoretical consideration of
Bond and Puls.
They developed a theory for the actual
change in surface tension with time based upon the time re­
quired for the molecules of the dissolved substance to diffuse
from the interior to the surface.
The relation for the time
necessary for the surface tension to pass one half of the way
/ 3 ©
to the final value was given by, T = - —
/D R T , where
or = the static surface tension for a bulk concentration c,
D = coefficient of diffusion, R = gas constant, and T = the
(101) W.N.Bond and H.O.Puls, Phil.Mag. 24, 864 (1937)
absolute temperature.
An equation was also derived to give
T from an observation of the surface tension at a known age of
the surface.
Measurements were made on liquids by the liquid
disc method of Bond,
, and T was calculated from these results
and an estimated age of the surface.
For pure liquids and
salt solutions such as, sodium chloride, T was found to be in
the vicinity of 10“9 seconds.
In the case of organic sub­
stances T approached the magnitude of seconds. This confirmed
measurements of Adam and Shute
who found that the change in
surface tension in the case of aqueous solutions of paraffin
chain salts was very slow, amounting to days.
The effect of changing the material of which the ring was
made was investigated by M e t z and Lambert
and by Ferguson.
As long as metals, i.e., copper, silver, or platinum, were used
the effect of contact angle was negligible.
However, when the
rings were covered with solids such as hydrocarbons, alcohols,
aldehydes, or acids, there was a definite relationship between
contact angle and the pull for which an equation was developed.
According to their results there may be a size of wire for
which the pull is independent of contact angle, as well as a
contact angle for which the pull is independent of the wire size.
The maximum pull which has been observed has been explained
on a theoretical basis by Lenard.94
This theory for the pull
Cl02) W.N.Bond, Proc.Phys.Soc., £7, 549 (1935).
(103) N.K.Adams and H.L.Shute, Trans.Far.Soc. , 34, 758 (1938;.
fl04) A.H.Nietz and R.II.Lambert, J •Phys.Ghem. , 55, 1460 (1929).
Ferguson, Trans.Far.Soc •, 17, 370,(192l).
on a straight wire demands the presence of two maximum pulls
separated by a minimum, both of which have been observed.
first maximum, which is that usually observed, corresponds to
a state of equilibrium whereas the second is observed only if
the measuring wire has been thoroughly cleaned and is well wet
by the liquid.
Lenard succeeded in deriving an equation for
the force on the wire at the maximum pull and also an equation
for the shape of the meniscus of the liquid surface.
It is
interesting to note that in their work the horizontal wire was
suspended from a balance while the liquid level was raised to
meet it, then was lowered gradually.
The limit of accuracy was
1 mg.
The development of the theory of the ring method did not
keep pace with the technique, poorly developed though this was
for a long time.
106 107
Until the exhaustive work of Harkins and
standardized the ring method, results obtained
by it deviated as much as
from results obtained by the
capillary rise, drop-weight, or other methods.
The outstanding
reason for this discrepancy was, as Harkins et al, pointed out,
the incorrectness of the simple equation; o' =
; where <T =
surface tension in dynes per cm., P = total pull on the ring in
dynes, and R = mean radius of the ring.
In order to give a
correct result a factor, F, must be inserted.
By the use of
rings of various sizes and of wire of different dimensions,
(106) W.D.Harkins, T.F.Young, and L.H.Cheng, Science 6£, 333 (1926).
(107) W.D. Harkins and H.F.Jordan, J.Am.Chem.Soc., 52. 1751 (1930).
this factor was found to be a function of R/r and of R /V,
where r = radius of the wire and V = volume of the liquid
raised by the ring.
A table was compiled giving F for rings
having a great range of values of R/r and R
Of almost equal importance as this standardization was
the enumeration and clarification of the experimental pre­
cautions which must be taken to obtain reproducible results.
Most of these had been observed previously, but each investi­
gator apparently resorted to his own ingenuity to rediscover
Those given by Harkins are listed below.
1) The pull was found to change with the deviation of the plane
of the ring from the horizontal.
This deviation in pull is
given by the relation, A p = -ka , where a » angular deviation
from the horizontal, k = 0.36, and A p = change in pull in
A deviation of less than one half a degree from the
horizontal may be detected by sighting in two directions at
right angles, between the ring and its mirror image in the liquid.
2) The ring must lie in one plane.
Therefore, it must be free
from any irregularities or bends.
3) The ring must be perfectly round if measurements of absolute
surface tension are to be made.
4) The vessel containing the liquid must be large enough to
prevent a curvature of the surface which will affect the shape
of the surface of the liquid raised by the ring.
The liquid
surface should be great enough that the surface between the
ring and the edge of the vessel is plane.
5) The surface must "be free from wave motion caused by vi­
6) The motion separating the ring and the liquid must be slow
and regular.
There should be no horizontal motion of the ring.
7) Evaporation from the surface of the liquid with its sub­
sequent cooling effect should be prevented.
The liquid and
vapor should be buried in a thermostat.
8) It is advantageous to renew the liquid surface by overflowing
or sweeping with a bar of glass at intervals.
9) An additional precaution has been noted by Dorsey.
internal diameter of the ring must be no less than that of the
tube in which the liquid will rise to a height which is equal
to the elevation of the ring when the pull is at a maximum.
similarly observed that the radius for the
measurement of the surface tension of water should be greater
than 0.75 cm.
In his experimental work Harkins raised his balance and
the ring away from the liquid by a smoothly operating mechanical
He objected to the alternative of lowering the
liquid because ripples migjit be set up.
balance must be sLow
The raising of the
so that no momentum is imparted to the
C108) H.E.Dorsey, Science, 69, 187 (1929).
(109) F.H. MacDougall, ibid, 62. 290 (1925).
The theoretical treatment of the determination of absolute
surface tension by the ring method including a mathematical
analysis of the form of the surface raised by the ring has been
given by Freud and Freud*
Since this does not apply to the
determination of relative surface tension, a reference to the
work should be sufficient*
(110) B*B.Freud and H.Z.Freud, J*Am*Chem*Soc.,52, 1772 (1930)
General Procedure:
Relative surface tension was measured by an adaptation
of the du Noiiy ring method#
In brief, two, approximately
identical platinum-iridium rings were suspended from opposite
arms of a balance into pure water and into the solution whose
surface tension was to be determined, respectively#
The two
liquids were lowered simultaneously by a leveling device
until one of the rings broke free from the surface#
were then added to that side of the balance, the liquid levels
were raised until the rings made contact, and the process was
repeated until the opposite ring broke free#
Repetition of
this with gradually diminishing differences in weight resulted
in an "end point" at which a shift of 0*1 mg* from one side
of the balance to the other would cause the opposite ring to
break free*
From the difference in pull between the two rings
for the solution and water, the difference in pull between the
two rings with water on both sides, and the total pull of the
ring used for both solution and water, the relative surface
{tension of the solution was found*
Extreme precautions were taken in the cleaning of all
•apparatus and in the preparation of water and solutions so
|that the least possible amount of grease or organic matter
]would be present*
Developments Leading to the Adoption of the Ring Method*.
The object of the investigation was to measure the surface
tension of very dilute salt solutions by a method lacking the
limitation of the capillary rise method, that is, the necessity
of using glass capillaries with the accompanying glass-liquid
interface, yet having an accuracy as great as that obtained
by Jones and Ray.
Preliminary experimentation with the drop-weight method
j the accuracy of which was previously probably second to that
|of the capillary rise method gave no encouragement that the
J|accuracy could be increased to the extent desired.
This experi­
mentation involved an attempt to increase the size of the
drops by driving a current of air to oppose their fall and an
attempt to use concentric tubes in order to increase the drop
An indication that the ring method could be used was obi
jtained by the construction of various shaped rings from copper
The force necessary to pull these from a water surface
■was roughly estimated using a Joly spring balance. Prom a coni
jsideration of the amount of this force and the practicality of
the use of the various sizes of rings, one was chosen which
liseemed suitable.
The Ring and Support.
The ring adopted was in reality two concentric rings of
!3 cm. and 6 cm. diameter, connected by three equally spaced
bars (Pig. 5). This was attached to a small ring 5 mm. in
idiameter by three supports, also equally spaced about the cirilcumference of the outer ring.
All parts of the ring were of
Hplatinum-10^ iridium wire 1 mm. in diameter, having a total
F IG . 6
weight of about 12.5 g.
The rings were suspended from the balance arms by two
nickel wires of 1 mm. diameter.
In order to prevent the
jrotation of the rings about a vertical axis, the ends of the
!wires were bent into a hook and filed to a sharp V which is
|visible in Pig. 5. This precaution was necessary because the
!pull was found to vary depending upon the rotation of the ring.
This was caused by a slight deviation of the plane of the rings
from horizontal or by the possibility that the small ring was
jjnot located exactly on the axis of the ring.
It Is very essential that the shape of the ring be altered
Iin no way between measurements; that It be placed upon the hook
[in exactly the same position each time, and that it assume
|exactly the same position about the vertic&i axis. In order
jfor the last to be true care must be taken to prevent a bending
lor twisting of the nickel supporting wire.
The rings were cleaned by heating to a bright red heat
with a gas burner, while they were suspended from the supporting
No strain should be put upon the ring during this heating
(because of the increased possibility of altering the shape.
Therefore, it is better to allow them to hang freely than to
hold in tongs while heating.
[The Auxiliary Apparatus.
(Pig.6 )
The water and the solutions were contained in two dishes
I|l2 cm. in diameter which were placed inside a glass jar within
|ja constant temperature bath.
The dishes were made by cutting
toff the bottom of a 750 ml. Erlenmeyer flask after a tube had
been sealed into the bottom.
As recommended and devised by
jHarkins, Young, and Cheng
arrangement was made to fill
|these dishes through this tube from the exterior of the bath.
jIn order to renew the surface the dishes were overflowed by
jaddition of solution through the same Inlet. The excess which
Ioverflowed could be withdrawn by suction through the appropriate
|exit tube*
The level of the liquid was changed by the leveling arrange-
Each dish was connected to a bulb containing mercury,
|jwhich in turn was connected to a leveling bulb.
An exterior
ijmercury leveling device altered the level in the inner leveling
[bulb which then altered the level in the two bulbs and as a
|result the level of the liquid in the dishes.
Thus, the level
!of the liquids could be raised or lowered at approximately the
|same rate without the liquids coming in contact with each other.
Since they did come in contact with the mercury, it was necessary
that this be clean and free from organic material.
The cleaning
j!operation will be described later.
The dishes, jars, and mercury bulbs were supported in the
!jconstant temperature bath by a framework of brass consisting
Iof two rectangular brass plates with openings to support the
jars as is indicated in the illustration, and of four
jithreaded brass rods.
7 /l 6
These rods were rigidly attached to the
wooden framework upon which the balance rested.
II the
balance, dishes, leveling bulbs, and the liquids were free
[from any vibration due to the bath mechanism. The framework
which supported the balance and the brass framework was con-
[structed solidly of 2 x 4 In. timber.
This and the large glass
R elay
C ircuit
.Fig. 7
tub rested upon a one ton concrete pier -which was cushioned
from the floor with rubber*
Although this rubber freed the
apparatus from exterior vibrations, it was essential not to
jjar the pier itself*
The use of the exterior leveling device
connected to the inner leveling bulb by a flexible rubber tube
;prevented jarring during the changing of the liquid level.
The temperature of the bath was maintained at 25*00°C
|by the use of cooling coils and intermittent heaters whose
|action was governed by a mercury-toluene regulator and a cold;cathode tube type relay.
This relay for the design of which
11indebtedness is expressed to Dr. A. A. Frost and Dr* P. G.
|JSmith has proved to be more economical to operate than the
|previously employed vacuum tube relay.
The diagram for this
is given in Fig* 7 where the tube is a three element, 0A4-G
When the thermo-regulator circuit is closed due to the
contact made by the expansion of the toluene and mercury, the
potential across the small ring anode and the disc cathode is
sufficient to start the discharge of the argon contained in
the tube.
This initiates the discharge between the verticil
[anode and the cathode, as a result of which a current flows
jithrough the induction coil of the relay which breaks the
jjheater circuit. A resistance of 200 ohms and a 1 mf. con;i
Ijdenser are in parallel with the coil of the relay. A resistance
megohms is placed in series with regulator to reduce
the possibility of arcing across the regulator contacts.
A variable voltage control, or Variac, was used in the
heater circuit to control the rate of heating.
The heaters,
themselves, consisted of two spirally wound nichrome wire im|mersion heaters connected in series*
Through the use of this temperature control the outer
hath which was constantly stirred hy an electrical stirrer
|could he maintained constant to G.01°C, which was the limit of
accuracy of the thermometer employed.
The constancy of the
temperature in the inner air hath was even greater than this*
•The stirring motor was attached to a wall bracket entirely
independent of the remainder of the apparatus, in order to
|prevent the transmission of vibrations*
Cleaning Procedure and the Preparation of Solutions*
The water was prepared hy redistilling the laboratory dis­
tilled water from alkaline potassium permanganate through a
modified Kraus still*
The distillate was collected in glass
stoppered flasks which had been cleaned with a boiling
mixture of concentrated nitric and concentrated sulfuric acids*
A cover was provided to prevent the contamination of the water
during collection by dust from the air*
The water and solu­
tions were stored in Pyrex columns whose use was advocated by
Jones and Ray.
These were 1 meter long by 7 cm. in diameter,
and were provided with an ungreased stopcock located 15 cm*
from the bottom for the removal of liquid from beneath the
A petri-dish was ground onto the top to provide an
jair tight cap. The entire column was frequently cleaned with
the cleaning-mixture whenever drainage became uneven* The
water and solutions were allowed to stand at least twelve
ihours in these columns to permit capillary active material to
jrise to the surface*
Water and solutions were added directly
ifrom these columns to the dishes in the apparatus, thus
| eliminating unnecessary intermediate receptacles*
In the preparation of the solutions a master 1 N
:potassium chloride solution was made using the density data of
jJones and Hay*
In one set of measurements the potassium
|chloride was taken directly from the C.P* stock, heated to
jredness for one hour, cooled, and then weighed out and dis|solved in the corresponding weight of pure water*
For the
jother measurements the potassium chloride was recrystallized
|then heated to redness*
Since the presence of inorganic
impurities affects the surface tension only slightly the
recrystallization should not greatly alter the results.
solutions down to
hy weight*
N were made from this
N solution
Below 0.001 N dilution to the proper volume using
volumetric flasks was sufficiently accurate.
The procedure necessary to cleanse the apparatus preceding
each day*s measurements was worked out in detail and carefully
Boiling concentrated nitric and sulfuric acids in a
mixture were added to the dishes within the constant temp­
erature bath.
After standing overnight, this mixture was
iremoved by suction and the dishes were rinsed with distilled
The temperature bath was emptied of water, following
:which, steam was passed Into the apparatus through the liquid
admittance tubes for one hour.
The steam was generated by
heating distilled water which had been previously refluxed
over alkaline potassium permanganate for six hours in Pyrex
!vessels provided -with stoppers wrapped with pure tin foil to
prevent contamination by the rubber* This steaming out was
!chiefly to remove acid vapors from the apparatus*
this steaming process, the bath was refilled and the dishes
were rinsed with the pure redistilled water until the pH of
the water which was removed was the same as that of the water
in the column*
This ranged from 5.8 to 6.2*
This measure­
ment of pH was made chiefly to determine whether all of the
icleaning acid had been removed from the apparatus*
As has
|been recognized before, a slight difference in acidity has
little or no effect upon surface tension, although greater
amounts of acid lower the surface tension.
A Cameron pH
meter accurate to
units was used for these measurements.
The mercury in the leveling bulbs which came in contact
with the liquids was cleaned by first filtering through a dry
Then a concentrated solution of sodium hydroxide was
poured over the mercury and air was run through it for twelve
After this, it was washed with distilled water, con­
centrated sulfuric acid was added, and air was bubbled through
for another twelve hours. The acid was washed out with disi
|tilled water using the air to again agitate the mercury.
IMeasurement Procedure.
The dishes were filled via the admittance tubes until the
dishes overflowed.
After standing for twenty to thirty minutes
to allow temperature equilibrium to be attained, the dishes
|(111) K.Ariyama, Bull.Chem.Soc.Japan, 12., 32 (1937).
were overflowed again to renew the surface and the levels were
The rings were flamed and suspended from the balance
jabove the liquid surfaces.
These were balanced while dry
the addition or removal of weights, then the liquid level was
raised until contact was made with the rings.
The beam of the
(balance was released, then the level of the liquids was lowered
|slowly over approximately a five minute interval until one ring
broke free.
Weights were added to this side as has been pre­
viously described in order to obtain the difference in pull
jbetween the rings.
Elimination of the Apparent Change of Surface Tension with Time:
The greatest difficulty in the experimental measurements
was an apparent drift in surface tension with time which has
been encountered by other investigators.
This phenomenon can
be accredited to a number of Gauses including, first, the
jadherence of impurities to the ring which breaks free, thus
changing the contact angle, second, accumulation of grease or
organic material on the surface of the liquid and on the rings,
third, a so-called aging of the surface, and fourth, changes
inherent in the ring itself.
The last possibility was eliminated by using rings of
-various materials. This drift was found with smooth platinum
jirings, with rings coated with paraffin, dicetyl, stearic acid,
and with Pyrex rings of approximately the same dimensions as
the platinum rings.
However, rings which had been covered with
jplatinum black by electrolytic deposition from chloroplatinic
jjacid solution showed no drift until in service for at least two
This indicated that the much greater surface area of
|the platinized ring might he the factor correcting the drift.
The third possibility has been proved to be such a rapid
process for salt solution and pure liquids that an aging could
jnot account for the slow drift which was observed.
The temporary correction of the drift by the use of
|platinized rings and the fact that the drift was always toward
a diminution of the surface tension pointed to the first two
as the causes.
In order to determine the direction of drift,
the pull on each ring was measured separately.
A f,chainomaticH
jdevice was attached to the balance which permitted a gradual
increase of weight to the side opposite from the ring while the
!liquid level was lowered.
The chainomatic attachment was made
'by suspending a fine gold plated chain from one side of the
balance beam and from a pointer which was attached to a movable
icord running between two small brass pulleys before a vertical
graduated scale. A dial was used to turn the lower pulley
through a rubber friction drive, thus lowering or raising the
chain. The arbitrarily chosen scale was calibrated by the
jaddition of weights to the opposite side of the balance. The
pull at the point of detachment of the ring from the surface
instead of the maximum pull was measured over a period of time.
jiAlthough the result was only accurate to +5 mg. an indication
jof a decrease in surface tension was provided.
Both of the first two possibilities can account for the
jdrift by assuming that traces of impurities diffuse to the
surface where they are spread out actually altering the surface
jtens ion or merely adhering to the ring thereby changing Its
contact angle.
The latter was proved by the fact that If the
ring was removed and flamed, the next measurement would return
|to the original value. When a platinized ring was used the
surface area of adsorption was so large that a longer time
elapsed before sufficient impurities had been adsorbed to alter
the ring.
The correction of this difficulty lay in the removal of
impurities on the apparatus by the rigorous cleansing proI
cedure which has been described. Despite this It was frequently
necessary to flame the rings after every fifth detachment from
the surface.
A scraping of the surface using glass or brass
!scrapers coated with paraffin or dicetyl as recommended by
Young and Boyd proved to be of no aid.
This may have been
due to the difficulty in manipulating the scrapers within the
As developed, the method is now accurate to a pull of
+ 0 . 0 0 0 1 g. or + 0 . 0 0 1 dyne and gives a reproducibility with
different samples of the s ame concentration to approximately
this limit.
Besides the Increased sensitivity, the method is
jcharacterized by the fortunate coincidence that unless the
japparatus is scrupulously clean and the solutions are free
from grease, no measurements can be obtained because of the
continual drift.
Although this increases the experimental
(difficulty, it insures that, when a result has been obtained,
jthe pull on the ring is that due to the solution In question
and not to one contaminated with organic material which alters
jthe surface tension.
i (112)
T.F.Young and E. Boyd, Private Communication.
The difference In pull between the two rings when each was
dipped in pure water was determined before a measurement with
solution on one side and water on the other was made.
measurement of pure water was repeated each time to insure
that the rings had not changed and that the apparatus had been
thoroughly cleaned.
In the course of the measurements the
balance point for pure water did shift twice but remained
constant following each shift.
This shift was probably due
to a mechanical change in the ring or supporting rod.
example of the data for a single measurement with pure water
on each side is given in Table X to indicate the procedure.
Table X
Example of Data for Measurement with
Water on Both Sides
Weight on Right Pan
-0.0082 g.
Ring Breaking
Balance point = 0.00806 + 0.0001 g.
The constancy of the balance point for pure water over a
period of time can be obtained from Table XI in which the
date of measurement, the balance point, and the pH of the
water before measurement are given.
(x) A negative sign indicates that the weight is placed on
the left pan of the balance.
(z) R and L designate right and left ring respectively.
Table XI
The Constancy of the Balance Point for Pure Water
Weight on Right Side of Balance
0.0161 + 0 . 0 0 0 1 g.
0.0160° ~
0 *0161q
Q.OI6 O5
0. Q161r> Average * 0.0161o
-0* 00796
-0.OO8 O5
-0 .00806
"0.Q08Qfi Average = -0.0080s
-0.0180k Average = -0.01815
pH of Water
Thus, over a duration of one month each of the first two
series remained constant within the limit of error of the
The two changes were so pronounced that the
possibility of error due to the non-recognition of a change
in the balance point for water is eliminated.
When solutions were used, the water was always removed
from the left dish, which was then rinsed thoroughly and
jSfilled with the solution to be measured.
Since It was neces-
jjsary to know the total pull on the ring which was used for
both water and solution, that is, the left ring, this was
determined using the chainomatic device.
For the first series
of balance points for water, the force necessary to pull the
left ring from water was 5*699 + 0.005 g. for the second
series this was 5.770 + 0.005 g. and for the third, 5.706
± 0.005 g.
This measurement is sufficiently accurate for the calcu­
lation of the relative surface tension, since the absolute
accuracy of its value need not be so great as that of the
measurement of the differences between the forces necessary
to pull out the two rings.
The experimental data for potassium chloride solutions
which were prepared from the un-recrystallized stock are given
in Table XII.
Table XII
Measurements for Potassium Chloride Solutions
C one entrati on
1 .0 0 0 0 N
0 .1 0 0 0
0 .0 2 0 0 0
0 .0 1 0 0 0
0 .0 0 2 0 0 0
0 .0 0 1 0 0 0
0 .0 0 1 0 0 0
0 .0 0 0 1 0 0
Wt.on Right Arm
of Balance
0.1473 g.
—0 .00935
Balance Point
for Water
0.016 35
-0 .QO8 O5
0.016 0 6
-O.GO8 O5
-A g.
0.0167 o
The apparent relative surface tension of the solutions was
calculated from thdsedata assuming that the F factor (see page 56)
for the ring is the same for both solution and for water.
dilute solutions this is a reasonable assumption since F is
a -
a function of R/r which is constant, and R /V, where V is
the volume of the solution lifted by the ring.
This volume is a function of the density of the liquid and
the weight of liquid lifted by the ring.
Since volume equals
weight/density, an Increase in weight caused hy an increase
| in surface tension when a solution is substituted for water
will be partly compensated by an increase in density.
case the surface tension is less for the solution than for
| water, the decrease in weight and Increase In density will
| decrease the volume, thus increasing F.
For dilute solutions
these changes would be expected to be very small.
The possi­
bility of a change in the angle at which the liquid makes
contact with the ring is the greatest argument against the
j assumption to be made below that the observed pulls represent
true surface tension.
If the total pull necessary to detach the ring used for
both water and solution form water is assumed to represent
<To> the surface tension of pure water, then <TC, the surface
tension of the solution, will be represented by the value for
<r 0 in terms of grams plus the difference between the balance
points for solution on one side and water on the other and
water on both sides.
Values of
which were obtained
in this manner are Included in Table XIII and are plotted In
as a function of the concentration.
The scale of plot (b) In Fig.
Is ten times that of fa),
and that of (c) is 100 times that of fa).
The values below
0.001 N are compared with those of Jones and Ray in Fig. 9, in
which the dotted line represents their data.
It is evident
from this that the trend of the results of Jones and Ray has
been confirmed.
Like their data, those of this investigation
give a curve which has a negative slope until a concentration
K Cl
^ — ^ - tT
---- -
N ormality
Rg . 8
Table XIII
Apparent Relative Surface Tension of
Aqueous Potassium Chloride Solutions
1*0000 N
, & S
Total Weight to
_(from Table XII) Detach Ring
from Water
0.1309s+0.0001 g. 5.699+0.005 g.
0 *0017 o
-0 *0017 o
of 0.001 N has been reached.
& e/fro
0.9997 o
The maximum decrease which
occurs at this point is 0.03$, whereas in their case it was
The concentration at which the relative surface
tension again becomes equal to unity is 0.012 N which is
exactly twice their value of 0.006 N.
Above this concen­
tration their values are slightly higher until a concentration
of 0.10 N has been reached.
Although there are these dis­
crepancies between the numerical values of the relative sur­
face tension, it is significant that a lowering of surface
tension has been found by two entirely independent methods.
A comparison of values of C^c *" ^o^/^^o
Jones and Ray is made in Table XIV.
those of
These values multiplied
by 100 are plotted against concentration in Fig. 10.
and Schwenck
worked at higher concentrations
than the significant ones and Schwenck*s were at 0°C. a
comparison with their results is meaningless.
£ I
_l ..
a: u
Table XIV
1*0000 IT
0 .0 0 1 0 0 0
0 .0 0 0 1 0 0
C^c - cr0)/G cr'o
Jones and Ray
This Investigati
- 0.6
Although the data of this Investigation do not fall upon
| as smooth a curve as that of Jones and Ray In Fig. 9, those
in Fig. 10 give a more regular curve.
The value obtained at
| 0.0005 N is apparently incorrect since it falls upon neither
the curve in Fig. 9 nor Fig. 10.
Measurements which have been made upon 1 IT potassium
chloride solutions prepared from the salt which had been recrystallized once or twice, indicate that the twin ring
method may be in error at higher concentration unless a
correction factor is applied.
Two solutions were made up
independently by weight from once recrystallized salt and
one was prepared from twice recrystallized salt.
In order to
determine whether the change which was found would be present
in the more dilute solutions, a
from one of* these 1 N solutions.
IT solution was made
No change outside of the
experimental error was found for this solution.
The data
are given in Table XV.
The differences among the values for the 1.0000 N solu­
tions may also be due to temperature changes resulting from
Table XV
Surface Tensions of Solutions Pre­
pared from Recrystallized Potassium
_ _ ,.
Solution 2
1.0000 N
A g
Total Weight to
Detach Ring from
5.770 g.
0 .9 9 9 7 ^
Solution 4
(Twic e re cry st •)
Solution 3
1# 0000
0 *0010000
unequal rates of evaporation from the solution and from pure
The apparatus was so constructed that there was free
access of vapors between the two liquids.
Therefore, for the
concentrated solution the change in concentration due to
distillation may have been effective as well*
To check this
a more nearly air tight apparatus would be necessary.
the more dilute solutions the difference in vapor pressure
from that of water is too minute to produce a noticeable
change during the time of measurement.
Although the variation
in the values for the relative surface tension in Table XV is
part in
this is far beyond the experimental
The question arises from a consideration of the data for
the relative surface tensions of potassium chloride solutions
as to whether the published data of Jones and Ray represent
true surface tension, whether the theory of Langmuir is
valid, or whether some other explanation must be given to
account for the phenomenon.
As was suggested by Langmuir for the capillary rise
method, an experiment was conducted to determine whether a
film might he present on the ring as he claimed to be the
case for the glass capillary*
1 0 ”s
molar thorium nitrate
solution was added to the left dish and the measurement was
made as for a salt solution*
for the
If the apparent lowering found
N potassium chloride solution was due to the
removal of a film of water, the same effect should be found for
the thorium nitrate solution, assuming that the zeta-potential
for a platinum-water interface is the same as for a glasswater Interface.
This assumption is reasonable since it has
been shown that the zeta-potentials for quartz and platinum
colloids are of the same order of magnitude, -0.044 v, in
However, the balance point was found to be 0.0003o g.
lower than that of pure water.
This corresponds to a relative
surface tension of 0.99996 which is just outside of the ex­
perimental error.
This lowering, slight though it is, may be
explained by reasoning similar to Langmuir 1s.
If a film is
present on the wire in pure water which is rigidly bound to
the wire, it will add to the weight of the ring.
The removal
of the film by elimination of the zeta-potential would produce
an apparent lowering in surface tension.
The magnitude of
this may be calculated assuming Langmuir*s equations to be valid.
At a height of 1 cm. above the plane surface, which was
approximately the height to which the ring rose before breaking,
(113) F.Freundlich, f,Colloid and Capillary Chemistry p.257,
transl. by H.S.Hatfield, E. P.Button and Co., New York.
the thickness of the film would be given by,
A r = 2.16 x 10"6 /h1//s = 2.16 x
The volume of water adhering to the ring is,
A V sb LirCCr + A r ) S - r*),
where L = total length of wire and r m radius of the wire,
32.8 and 0.05 cm. respectively.
in the equation gives
The insertion of these values
.0 0 0 2 s ml. for A V.
Therefore, the
decrease in balance point caused by this effect would be
Q.0002s g. which is the same order of magnitude as the observed
decrease for the thorium nitrate.
If the lowering which was observed for thorium nitrate is
due to a film, this value should be subtracted from the
lowerings observed for potassium chloride solutions more con­
centrated than 0.001 N.
the corrected values.
The dotted line in Fig.
(a) shows
Below 0.001 N the correction was made
proportionate to the observed lowering, which is not neces­
sarily true.
The fractional elimination of the zeta-potential
should be known for an accurate correction of these.
An attempt was made to correlate Dolefs theory with the
}observed data. In order to bring the minimum decrease in
[surface tension into coincidence with the experimental, the
ithickness of the surface adsorption layer, t, must be taken
las 4.9 A, which is lower than the previously chosen values of
7-8 A.
liShen the concentration at which the relative surface
|tension again becomes equal to unity after passing through
|the minimum is taken as 0 . 0 1 2 N, the energy of adsorption is
-5900 cal.
Values for o'o/a'o were calculated from these
figures and compared •with, the observed, in Figure 9.
this it is apparent that the initial decrease required by
the theory is much too rapid, and that the minimum is too
Although the evidence points toward the reality of the
i lower surface tension of very dilute solutions, this does not
cons titute proof*
It is probable that the Langmuir theory
is correct as far as it goes, but that it fails to include
the other contributing factors*
It is possible that the
initial slope may be due to a change In the contact angle
| between the ring and solution.
A slight change in this would
produce a decided change in the observed balance point*
1* A new method for the measurement of relative surface
tension has been developed.
This involves the measurement of
the difference between the forces required to detach two twin
rings from the surface of water as a standard and the liquid
in question.
The technique involving the pur ification of
materials, cleansing of apparatus, and the experimental pro­
cedure has been perfected to an extent such that the relative
surface tension may be measured to an accuracy of 0.005$.
2. The drift in surface tension with time which has been
previously observed is accredited to the accumulation of
capillary-active impurities on the surface.
Extreme pre­
caution to obtain absolute cleanliness was found to eliminate
the drift.
3. The apparent relative surface tensions of potassium
chloride solutions of concentrations over the range
to 0.000100 K have been measured.
The plot of apparent
relative surface tension against concentration gives a curve
similar in shape to that of Jones and Ray.
4. The failure of the Langmuir and Dole theories to account
satisfactorily for the experimental results is discussed.
John Arthur Swart out
March 17, 1916 at Madison* Wisconsin*
Kenmore High School, Kenmore, New York, 1929-1933
B.A., University of Buffalo, Buffalo, New York,
Graduate Assistant in Chemistry, Northwestern
University, 1937-1938, Summer 1938.
Research Assistant in Chemistry, Northwestern
University, 1938-1939.
University Fellow in Chemistry, Northwestern
University, 1939-1940.
Sigma XI
Phi Lambda Upsilon
1) ”The Protium-Deuterium Ratio and the Atomic
Weight of Hydrogen”, John A. Swart out and
Malcolm Bole, J.Am.Chem.Soc., 61, 2025 (1939).
2) "A New Method for the Measurement of Relative
.Surface Tension”, Malcolm Bole and John A.
Swartout, Presented at the National Meeting
of the American Chemical Society in Cincinnati,
Ohio, on April 10, 1940.
Без категории
Размер файла
4 225 Кб
Пожаловаться на содержимое документа