journal ot

MEMBRANE
SCIENCE
ELSEVIER Journal of Membrane Science 100 (1995) 33-38

On liquid diffusion
Adolph Fick
Ziirich, Switzerland

A few years ago Graham published an extensive stratum between the horizontal planes at x and x + dx
investigation on the diffusion of salts in water, in which (in which the concentration isy) there will pass, during
he more especially compared the diffusibility of differ- an element of time dt, into the immediate superjacent
ent salts. It appears to me a matter of regret, however, stratum, bounded by the horizontal planes x + dx and
that in such an exceedingly valuable and extensive x + 2dx (in which the concentration y + dy/dx dx pre-
investigation, the development of a fundamental law, vails), a quantity of salt = - Q. k. dy/dx dt, in which
for the operation of diffusion in a single element of Q signifies the surface of the stratum, and k a constant
space, was neglected, and I have therefore endeavoured dependent upon the nature of the substances. It is evi-
to supply this omission. dent that a volume of water equal to that of the salt
It was quite natural to suppose, that this law for the passes simultaneously out of the upper stratum into the
diffusion of a salt in its solvent must be identical with lower.
that, according to which the diffusion of heat in a con- Exactly according to the model of Fourier's mathe-
ducting body takes place; upon this law Fourier matical development for a current of heat, we can obtain
founded his celebrated theory of heat, and it is the same from this fundamental law for the diffusion-current, the
which Ohm applied with such extraordinary success, differential equation
to the diffusion of electricity in a conductor. According
to this law, the transfer of salt and water occurring in a 6y6t - [62y (~-d-f
dQ 6y_)
unit of time, between two elements of space filled with
differently concentrated solutions of the same salt, must when the section Q of the vessel in which the current
be, caeteris paribus, directly proportional to the differ- takes place is a function of its height above the bottom.
ence of concentration, and inversely proportional to the If the section be constant (i.e. the vessel cylindrical or
distance of the elements from one another. prismatic), the differential equation becomes simpli-
In mathematical language this may be thus fied to
expressed: - - In a volume of salt solution, let the con- 82y
centration in each horizontal elementary stratum be 6y6t - k 3x 2 (2)
constant and = y, a function of the height x of this
stratum above any other stratum which may be assumed Several methods for the experimental confirmation
as the primary horizontal plane; the limitation being of this differential equation, and consequently of the
made, that the function y must diminish as x increases, fundamental law above advanced, presented them-
that is, each higher stratum must be less concentrated, selves. In the first place, by integration of equation (2)
and therefore lighter, than all the subjacent ones, the expression y =f(x, t) could be obtained, and the
because it is only under this condition, that the diffusion calculated value ofy compared with its observed value.
will not be interfered with by gravity; then from the I have, however, entirely omitted this method, because

0376-7388/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved

SSDI0376-7388 (94)00230-4
34 A. Fick/ Journalof MembraneScience100 (1995)33-38

even in those cases in which the integral has a finite showed it to be sufficiently accurate. The quotation of
form, the numerical calculation of a sufficient number the numerical results of one experiment may suffice
of values would have been extremely troublesome, here.
whilst other unequivocal proofs were possible. For the Depth of Stratum below the Surface.
same reason I also here omit to develop the particular 10 millims. 32.2 54.4 76.6 98.9 121.0 143.2 165.4
integrals of equation (2) for special cases of diffusion- 187.6 209.8 220.9.
currents. Specific Gravity* of Stratum.
The experimental proofs just alluded to, consist in 1.009 1.032 1.053 1.073 1.093 1.115 1.135 1.152 1.170
the investigation of cases in which the diffusion-current 1.187 1.196.
has become stationary, in which a so-called dynamic That the degrees of concentration in the lower layers
equilibrium has been produced, i.e. when the diffusion- decrease a little more slowly than in the upper ones, is
current no longer alters the concentration in the spaces easily explained by the consideration, that the station-
through which it passes, or in other words, in each ary condition had not been perfectly attained.
moment expels from each space-unit as much salt as A second case of dynamic equilibrium was also
enters that unit in the same time. In this case the ana- observed, by replacing the cylindrical vessel in the
lytical condition is therefore dy/dt = 0. Such cases can above-described arrangement, by a funnel-shaped one
be always produced, if by any means the concentration with the apex downwards. As the section was now no
in two strata be maintained constant. This is most easily longer constant, the condition for the dynamic equilib-
attained by cementing the lower end of the vessel filled rium was deduced from the more general equation ( 1 )
with solution, and in which the diffusion-current takes in the form
place, into a reservoir of salt, so that the section at the
62y 1 dQ 8y
lower end is always maintained in a state of perfect O= ~--~ -I- ~ • -~- • ~x (4)
saturation by immediate contact with solid salt; the
whole being then sunk in a relatively infinitely large For a perfect cone with circular base (the funnel-
reservoir of pure water, the section at the upper end, shaped vessel), we have Q = 7r- a2x 2, if the origin be
which passes into pure water, always maintains a con- placed in the apex of the cone, and we call a the tangent
centration = 0. Now, for a cylindrical vessel, the con- of half the angle of aperture. By the substitution of this
dition dy/dt= 0 becomes by virtue of equation (2), value, equation (4) becomes
d2y d2y 2 dy
O= dx----
5 (3)
0 = ~ - - ' ~ + - . - - , x dx
The integral of this equation y = ax + b contains the the integral of which y + c' = - c/x. The two constants
following proposition: - "If, in a cylindrical vessel, c and c' are to be so determined, that for a certain x
dynamic equilibrium shall be produced, the differences (where the cone is cut off and rests upon the salt res-
of concentration of any two pairs of strata must be ervoir), y is equal to perfect saturation; and for a certain
proportional to the distances of the strata in the two other value of x which corresponds to the base of the
pairs," or in other words, the decrease of concentration funnel, y becomes = 0. In an experiment made with
must diminish from below upwards as the ordinates of this apparatus, the following numerical values were
a straight line. Experiment fully confirms this propo- obtained: -
sition. Depth of Stratum below the Surface.
For the determination of the decrease of concentra- 27.7 millims. 55.5 72.1 88.8 105.4 122.1 138.7 155.4.
tion in the cylindrical vessel conducting the diffusion- Observed excess of Specific Gravity above unity (pro-
current, I sunk into the stratum to be estimated a glass portional to the concentration).
bulb suspended from the beam of a balance, and cal- 0.000 millim. 0.008 0.019 0.030 0.040 0.055 0.075
culated the specific gravity from the weight which O. 105.
required to be placed in the other scale-pan to balance
the glass bulb. This method creates little confidence at * The excess of which over unity is proportionalto the concentra-
first sight, nevertheless preliminary experiments tion.
 A. Fick / Journal of Membrane Science 100 (1995) 33-38 35

T e m p e r a t u r e Centigrade k, calculated from the k, calculated f r o m the k, calculated f r o m the

during the operation quantity w h i c h passed quantity w h i c h passed quantity w h i c h passed
t h r o u g h the longest tube through the medium tube through the shortest tube

Not determined 11.71 12.36 11.08

15.8°-14.8 ° 9.67 9.70 9.30
15.5°-16.0 ° - 9.57 -
16.0°-16.5 ° - 9.94 -
17.5-18.5 ° 10.79 - -
18.0°-19.0 ° 10.71 11.08 10.50
20 ° ! 1.14 - 11.02
19.0°-22.0 ° 11.44 11.33 -
20.0°-21.0° 11.89 - 11.12

Calculated excess of Specific Gravity above unity. same time, must be inversely proportional to the length
0.006 millim. 0.015 0.023 0.031 0.043 0.057 0.078 of the tubes; and that it these amounts be divided by
0.107. the time and multiplied by the length (expressed in
The determination of the constants for different salts millimetres), the same value, viz. the above-defined k,
could now be proceeded with, - a determination which must be obtained for all three tubes. I here annex a short
measures their diffusibility, and is expressed by k in table of the best experiments with common salt, to
equation (1). It may be defined as the quantity of salt which I have hitherto confined my attention (see table).
which, during a unit of time, passes through the sec- If we take into consideration the unavoidable sources
tional-unit, out of one stratum into the next adjacent of error, a nearer identity of numbers could scarcely be
one, when the rapidity of the diminution of concentra- expected. The table shows, as might be already
tion (dy/dx) is equal to unity. Let the sectional-unit be expected from Graham's experiments, that the value k
the section of our tubes, i.e. the surface of a circle of 1 increases with increase of temperature; probably, how-
millim, radius. The diminution of concentration (dy/ ever, this dependence upon temperature is not a simple
dx) will then be equal to unity, if, through a column of one. On the relations ofk to other values expressing the
liquid assumed permanently constant, and the height essential properties of bodies, as, for instance, to atomic
of which is equal to the linear-unit 1 millim., it has for weight, nothing can be said until extensive series of
its result a difference of concentration of both terminal experiments with different substances have been made.
surfaces of such a character, that the one possesses a It is now of importance to make the views thus
concentration corresponding to absolute saturation, the obtained furnish an insight into the diffusion of salt
other a concentration = 0. Let one day be a unit of time. solutions through porous partitions. Briicke* first made
Three tubes of different lengths were now arranged an attempt to give a mechanical explanation to these
exactly as the cylinder and funnel above described, and phaenomena. He went upon the assumption that the
after the establishment of the stationary condition, the substance of the partition attracted the particles of water
amount of salt was determined which diffused out of more strongly than the particles of salt. This assumption
the terminal section of the tube in measured spaces of gained important support from the experiments of Lud-
time, and which therefore also, in the same time, passed wig* and Cloetta tt, who found that the solution of salt,
through any single section of the cylindrical tube. In imbibed by an animal membrane, was richer in water,
this case the whole of the values required for the deter- and poorer in salt, than the solution in which the
mination of k were known, and the rapidity of the dim- membrane was immersed. BriJcke's theory, which was
inution of concentration (dy/dx) was especially only indeed obscurely intimated, can be more clearly
determinable, inasmuch as it produced permanently,
throughout the entire length of the tube, the difference
* P o g g e n d o r f f ' s Annalen, vol. lviii, p. 77.
between perfect saturation below and 0 above. t Ibid. vol. lxxviii, p. 307.
It is easy to perceive, that if our law be correct, the , t Experiments on Diffusion through M e m b r a n e s with t w o Salts.
amounts of salt, passed through the three tubes in the Ziirich, 1851.
36 A. Fick / Journal of Membrane Science 100 (1995) 33-38

determined by the help of our fundamental law as fol- surface of the membrane. This arrangement would have
lows. the effect of producing a diffusion-current in the ele-
Let us imagine a cylindrical pore of the radius p, in mentary stratum, which, according to our laws, would
a membrane which is immersed in a saturated solution deliver a quantity of salt 2~-. k . f ( p - r ) / h dr down-
of a salt, and let us assume with Brticke, a stronger wards, and an equal volume of water upwards, if we
attraction between water and the molecules of the understand by k, as above, the diffusion-constant for
membrane, than between the latter and the salt mole- the combination of salt and water in question, and by
cules; then the density of the solution in each concentric h the thickness of the membrane, consequently the
cylindrical film of radius r, of the porous partition will length of the pore. No regard has been paid, however,
evidently be constant, and = f ( p - r ) . Respecting the to the retardation which the flowing molecules suffer
nature of this function, it can only be said that its value from the attraction of the substance of the membrane
for r = p, which is equal to O, and that from this point along the sides of the pores; it is certain, however, that
it must increase (probably very rapidly) with the the total quantity of salt, passed through the entire pore,
decrease of r; but if necessarily under all circumstances could not be greater than
for r = O, it must be increased up to the density of the p
circumambient solution, is left undetermined, and
appears to me, in fact, for very narrow pores very
2 7r -~
k ff(p-r)dr
o
improbable. In a word, the density in the pore will
increase from the wall towards the centre, and in the But the passage of the water to the other side requires
cylindrical film of the radius r there can certainly occur a separate consideration. We have seen, namely, that
no higher concentration than f ( p - r ) (but possibly a at the upper extremity of the cylindrical stratum-unit,
lower one). The radius p of the pores must be imagined with the internal radius r, no higher concentration could
so small, that a filtration of liquid, by virtue of their take place than f ( P - r ) , which is certainly less than
cohesion, together with the attraction of the membrane perfect saturation, and, in fact, becomes proportionally
itself, is prevented, even under high pressure; and that smaller the greater r is assumed to be. If, as we suppose,
also equalization of the differences of pressure when a relatively inexhaustible volume of saturated solution
liquid is present on both sides of the porous membrane, (obtained by the addition of crystals) be present on the
if possible at all, is only so in a relatively long space of upper side of the membrane, then, at the upper extrem-
time; further, that currents of mixture, by virtue of the ity of our elementary stratum, there must take place a
variations of specific gravity within the pore, cannot sudden increase in the concentration, f r o m f ( p - r) to
Occur. perfect saturation. If we assume that this is the case for
Let us now suppose that a membrane of this descrip- the first moment, there will now be, according to the
tion separates a saturated solution of salt from pure general principles of diffusion, from the elementary
water, and further, that the former is above, and the stratum a relatively (in comparison with the amount
latter is under this horizontal membrane. In any cylin- which a continual transition of density requires) infi-
drical pore of the said membrane, whose radius may be nite quantity of water required, and an equally infinite
again = p, let us suppose a concentric elementary stra- amount of salt forced in. The latter will be inevitably
tum, bounded by two cylindrical films infinitely near hindered by the nature of the membrane, and the excess
to each other, the radii of which are respectively r and of salt forced against the pore must in some way glide
r + dr. At the upper extremity, the highest possible con- off laterally; on the other hand, more water, than the
centration, f ( p - r ) , of the layer will immediately be arrangement of the densities in our elementary stratum
produced. On the other hand, the lower extremity, by requires, can easily, to a certain extent, be drawn
immediate contact with a relatively infinite quantity of through towards the denser solution, so that in the pore,
pure water, will be constantly maintained at a concen- the particles of water move upwards with a greater
tration = 0; and when a stationary condition is brought velocity than the particles of salt move downwards.
about, the degrees of concentration, within the assumed The excess of water now spreads out on all sides, into
elementary stratum, must increase from 0 t o f ( p - r ) , the saturated solution (as the mouths of the pores must
in proportion to the height above the lower boundary lie at a certain distance from each other), partly by
 A. Fick / Journal of Membrane Science 100 (1995) 33-38 37

diffusion, partly by mixing streams proceeding from tion. This conclusion is confirmed by experiment. Two
difference of specific gravity; until a stationary condi- membranes, which differ only in the diameter of their
tion has been in such a way produced, that a conical pores, cannot, it is true, be obtained; nevertheless we
space increasing upwards is supported upon the upper are safely warranted in assuming, that a transparent film
annular section of the elementary stratum, in which of collodion possesses pores so very much narrower,
space the concentration f ( p - r ) increases to perfect than an animal membrane formed of interlacing fibres,
saturation, and which determines a diffusion-current of that all other differences between the two tissues influ-
such a strength, that thereby exactly as much water is encing diffusion, disappear in this difference of the size
passed upwards, as in the same time can diffuse itself of pores; in fact, an animal membrane was found to
into the reservoir of saturated solution from the upper possess a very much smaller endosmotic equivalent
end of the space, without changing the concentration. than a collodion film; for whilst, by the use of pig's
Then the above-mentioned space would evidently be bladder, only from 4 to 6 times as much water as com-
immediately lengthened upwards, (and thereby the mon salt passed through, there diffused through a col-
intensity of the diffusion-current be diminished) so lodion film, during the time that an imponderable trace
soon as more water passed through, and therefore the of chloride of sodium passed, a considerable quantity
concentration at the upper end of the space continues of water, probably many thousand times greater than
to vary; and if, on the contrary, less water passed that of salt.
upwards, sudden transition of concentration must Let us now suppose, that, instead of pure water, a
immediately occur in certain places, which sudden solution of a certain concentration c of the same salt
changes determine, as it were, a diffusion-current of which is present on the upper side of the membrane in
infinite strength, and therefore again immediately draw concentrated solution, be placed on the under side of
the requisite quantity of water. Now, this quantity of that membrane, then the following considerations pres-
water, which, caeteris paribus, during the unit of time, ent themselves. All the elementary strata, from the wall
of the pore, to an imaginary cylindrical film of the
can diffuse itself into the saturated solution without
radius r, t h a t f ( p - r ) =c, can only be filled from the
sensibly altering the concentration at that point,
top to the bottom with solution of the there highest
depends upon the easy mobility of the particles of the
possible concentration, and cannot therefore give origin
solution. The space in which the compensation takes
to a usual double diffusion-current. Water will, how-
place must therefore also be, caeterisparibus, so much
ever, pass over from the lighter to the denser solution,
shorter, and consequently the stream of water towards
because a force of suction comes into play on each side
the saturated solution stronger, the more easily mobile
of the membrane, proportional to the difference of con-
the particles of the solution are. I was unable to verify
centration, consequently a stronger force at the upper
this theoretical conclusion by direct experiment. I
side corresponding to the saturated solution. On the
expected that the excess of water passing through the other hand, all those strata lying nearer to the axis,
membrane, above the quantity of salt passing in the behave exactly in the above manner, with the single
contrary direction, must be found much smaller, when difference, that in them the concentration, instead of
the mobility of the upper solution was impeded by the increasing from 0, now increases from below upwards
addition of chalk magma. No such diminution of this from c to the respective maxima, and consequently an
excess, however, could be remarked in the experiment. absolutely weaker diffusion-current is produced. It
A second conclusion from the hypothesis just devel- therefore follows in this case, that in places where, in
oped can be thus expressed: the excess of the diffusing the first instance, salt passed to the one side and water
water, above the salt passing in the opposite direction, to the other, only a single diffusion of water takes place,
is smaller, the wider the pores of the separating parti- consequently the proportion between water and salt
tion. Or if, according to Folly's* method, we define, as (the endosmotic equivalent) must be greater here than
the endosmotic equivalent, the quotient obtained by there. If, for instance, c > ( p - r) (which is possible in
dividing the amount of water diffused by the quantity the case of very fine pores), no salt whatever could
of salt contemporaneously passed, the endosmotic
equivalent is smaller the larger the pores of the parti- * Poggendorff's Annalen, vol. lxxviii, p. 361.
38 A. Fick / Journal of Membrane Science 100 (1995) 33-38

pass through; the endosmotic equivalent would be = oo, as much greater as c is smaller, therefore the endos-
or the current one-sided. motic equivalent must also decrease very rapidly with
This conclusion is also confirmed by experiment. c; and we might expect that, for moderately small val-
Even the above-cited experiments of Ludwig and ues of c, the endosmotic equivalent would be found
Cloetta intimate this to be the case. I have myself smaller than unity (because a volume of salt weighs
obtained still more decided results. A membrane which more than an equal volume of water). Ludwig's exper-
gave an equivalent between 5 and 6, when it separated iments with common salt are perfectly in accordance
a saturated solution from pure water, yielded an equiv- with this conclusion; in some cases in his experiments,
alent = 11.05, in another case even = 17.05, when the the equivalent sunk from 4 to 1 when the concentration
pure water was replaced by a solution of common salt of the upper solution was decreased from saturation to
of 22 per cent. Lastly, if we analyse the case in which, 2 per cent. My own experiments in the same direction
instead of a saturated solution, a dilute one be placed with common salt gave a purely negative result, as the
above, and again pure water below. Let us again express equivalent was not perceptibly decreased.
the concentration of the upper solution by c. Evidently The comparison of the experiments adduced above
in an axial cylinder whose radius r is so great, as exactly with the hypothesis developed on the foundation of the
to m a k e f ( p - r) = c, the concentration from the lower diffusion law, shows, though not absolutely, that the
to the upper end, will now regularly increase from 0 to truth of this hypothesis may be determined; and it is in
c, and a usual diffusion-current take place, which will fact highly probable that, with or without modification,
pass a volume of salt downwards, exactly equal to the such an hypothesis may serve as the foundation of a
volume of water propelled upwards. Only in those lay- subsequent theory of these very dark phenomena.
ers situated nearer to the wall, can the concentration at
the upper end never rise to c, and there will therefore A. Fick (1829--1901), On Liquid Diffusion, The London, Edin-
burgh, and Dublin Philosophical Magazine and Journal of Sci-
pass through these layers alone, according to the pre-
ence, Vol. X (1855) 30-39. Abstracted by the author from the
vious considerations, more water than salt. The radius German original: [)ber Diffusion, Poggendorff's Annalen der
r, of the above-mentioned axial cylinder, is evidently Physik und Chemie, 94 (1855) 59-86.