Crank 1975 Diffusion

THE MATHEMATICS OF DIFFUSION
1. CRANK

FORMERLY OF BRUNEL UNIVERSITY UXBRIDGE

SECOND EDITION

PRESS . OXFORD

Oxford University Press, Walton Street, Oxford OX2 6DP Oxford New York Toronto

Delhi Bombay Calcutta Madras Karachi Kuala Lumpur Singapore Hong Kong Tokyo Nairobi Dar es Salaam Cape Town Melbourne Auckland Madrid

and associated companies in

Berlin Ibadan

Oxford is a trade mark of Oxford University Press

Published in the United States

by Oxford University Press Inc., New York

© Oxford University Press 1975

First edition 1956 Second edition 1975 Reprinted 1976

first published in paperback 1979 Reprinted 1983, 1985, 1986, 1989, 1990, 1992, 1993

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press

This book is sold subject to the condition that it shall not, by way

of trade or otherwise, be lent, re-sold, hired out or otherwise circulated without the publisher'S prior consent in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed on the subsequent purchaser

British Library Cataloguing in Publication Data

Crank, John

The mathematics of diffusion.-2nd Jd. 1. Diffusion=Mathematicat models 1. Title

531'.1137 QC185 79--40632 ISBN O-J9--853411-fJ

Printed in Great Britain by Redwood Books, Trowbridge, Wiltshire

PREFACE TO SECOND EDITION

IN preparing this second edition I have tried to incorporate as much new material as possible but to preserve the character of the original volume. The book contains a collection of mathematical solutions of the differential equations of diffusion and methods of obtaining them. They are discussed a background of some of the experimental and practical situations they are relevant. Little mention is made of molecular mechanisms,

I have made only fleeting excursions into the realms of irreversible . These I hope are self-explanatory. A number of general of the subject are already available, but very few mathematical of the equations of non-equilibrium thermodynamics have been for practical systems.

the last 15~20 years the widespread occurrence of concentrationdiffusion has stimulated the development of new analytical and solutions. The time-lag method of measuring diffusion coefficients

been intensively investigated and extended. Similarly, a lot of has been devoted to moving-boundary problems since the first published. These and other matters have now been included by revision of several chapters. Also, the chapter dealing with the solution of the diffusion equations has been completely rewritten up to date. It seems unbelievable now that most of the calcula-

first edition were carried out on desk calculating machines.

new chapters have been added. In one are assembled some of ~""'''''''''''' models of non-Fickian or anomalous diffusion occurring systems in the glassy state. The other attempts a ' ... ",,, • .,.,,, of diffusion in heterogeneous media, both laminates and

.•.. A succession of improved solutions are described to the problem in a medium in which are embedded discrete particles with properties.

the temptation to lengthen appreciably the earlier chapters. edition of Carslaw and Jaeger's book Conduction of heat in a wealth of solutions of the heat-flow equations for constant Many of them are directly applicable to diffusion problems, that some non-mathematicians have difficulty in making pn'm"'r~··, .... n . For them I have included a brief 'translator's

solutions have been added, however, some of them in the they arose, that is the measurement of diffusion coefficients. . to express my appreciation to the Vice Chancellor and . University for so readily agreeing to my application for which I could not have undertaken the preparation of

viii

PREFACE

1. The diffusion equations

2. Methods of solution when the diffusion coefficient is constant

3. Infinite and semi-infinite media . .•. Diffusion in a plane sheet

to Miss D. Eldridge who, by patient and skilful typing, transformed an almost illegible manuscript into a very clear typescript for the printer. I should also like to thank the following who readily gave permission to use material from various publications: Professor R. M. Barrer, Mr. M. B. Coyle, Dr. P. V. Danckwerts, Dr. L D. Hall, Dr. P. S. H. Henry, Professor J. C. Jaeger, Dr. G. S. Park, Dr. R. H. Stokes, Dr. C. Wagner, and the publishers of the following journals, British Journal of Applied Physics, Journal of Chemical Physics, Journal of Metals, Journal of SCientific Instruments, Philosophical Magazine, Proceedings of the Physical Society, Transactions of the Faraday Society. Finally, it is a pleasure to thank those members of the staff of the Clarendon Press who have been concerned with the production of this book for the kindness and consideration they have shown to me.

Maidenhead December 1955

CONTENTS

in a sphere

-dependent diffusion: methods of solution

1 11 28 44 69 89

104 137

calculated results for variable diffusion coefficients 160

definition and measurement of diffusion coefficients 203

254

in heterogeneous media 266

286

and chemical reaction

ultaneous diffusion of heat and moisture

326 352 375 399 407 411

1

THE DIFFUSION EQUATIONS

The diffusion process

FUSION is the process by which matter is transported from one part of a .:iiiitid,f'm to another as a result of random molecular motions. It is usually by the classical experiment in which a tall cylindrical vessel has part filled with iodine solution, for example, and a column of clear

poured on top, carefully and slowly, so that no convection currents up. At first the coloured part is separated from the clear by a sharp, boundary. Later it is found that the upper part becomes coloured, getting fainter towards the top, while the lower part becomes corless intensely coloured. After sufficient time the whole solution uniformly coloured. There is evidently therefore a transfer of iodine from the lower to the upper part of the vessel taking place in the convection currents. The iodine is said to have diffused into the

possible to watch individual molecules of iodine, and this can be by replacing them by particles small enough to share the motions but just large enough to be visible under the microscope,

found that the motion of each molecule is a random one. In a each molecule of iodine behaves independently of the others, meets, and each is constantly undergoing collision with solas a result of which collisions it moves sometimes towards a sometimes of lower, concentration, having no preferred towards one or the other. The motion of a single molecule

in terms ofthe familiar 'random walk' picture, and whilst it is the mean-square distance travelled in a given interval of to say in what direction a given molecule will move in

random molecular motions, in which no molecule has a . of motion, has to be reconciled with the fact that a transfer from the region of higher to that of lower concentration Consider any horizontal section in the solution and of volume one just below and one just above the

not possible to say which way any particular iodine .. a given interval of time, it can be said that on the raction of the molecules in the lower element of volume . from below, and the same fraction of molecules in the

2



3

upper element will cross the section from above, in a given time. Thus, simply because there are more iodine molecules in the lower element than in the upper one, there is a net transfer from the lower to the upper side of the section as a result of random molecular motions.

concentration of diffusing substance is C. Let AaCD and A'B'CD' be the faces perpendicular to the axis of x as in Fig.!.!. Then the rate at which diffusing substance enters the element through the face ABCD in the plane t - dx is given by

1.2. Basic hypothesis of mathematical theory

Transfer of heat by conduction is also due to random molecular motions and there is an obvious analogy between the two processes. This was recognized by Fick (1855), who first put diffusion on a quantitative basis by adopting the mathematical equation of heat conduction derived some years earlier by Fourier (1822). The mathematical theory of diffusion in isotropic substances is therefore based on the hypothesis that the rate of transfer of diffusing substance through unit area of a section is proportional to the' concentration gradient measured normal to the section, i.e.

( 8Fx )

4dydz Fx-ax-dx ,

F; is the rate of transfer through unit area of the corresponding plane P. Similarly the rate of loss of diffusing substance through the face is given by

F = -D8C;8x, where F is the rate of transfer per unit area of section, C the concentration diffusing substance, x the space coordinate measured normal to the o .... ,.u,JJ1. and D is called the diffusion coefficient. In some cases, e.g. diffusion in solutions, D can reasonably be taken as constant, while in others, e.g. in high polymers, it depends very markedly on concentration. If F, amount of material diffusing, and C, the concentration, are both ",V, ..... ""ON" in terms of the same unit of quantity, e.g. gram or gram molecules, then it clear from (1.1) that D is independent ofthis unit and has dimensions " ........... " (time)" I, e.g. crrr' s -1. The negative sign in eqn (1.1) arises because UUIH.""VJ occurs in the direction opposite to that of increasing concentration.

It must be emphasized that the statement expressed mathematically b (Ll jis in general consistent only for an isotropic medium, whose and diffusion properties in the neighbourhood of any point are the relative to all directions. Because of this symmetry, the flow of diffusing stance at any poin t is along the normal to the surface of constan t through the point. As will be seen later in § 1.4 (p. 5), this need not be in an anisotropic medium for which the diffusion properties depend on direction in which they are measured.

0'

p.

aF. 4dydz (r;+ ~ dx)

B' o x

FIG. 1.1. Element of volume.

to the rate of increase of diffusing substance in the element faces is thus equal to

-8 dx dy dz 8~x.

other faces we obtain

and

8Fz - 8 dx dy dzTz.

of diffusing substance in the element

1.3. Differential equation of diffusion

The fundamental differential equation of diffusion in an isotropic med is derived from eqn (1.1) as follows.

Consider an element of volume in the form of a rectangular whose sides are parallel to the axes of coordinates and are 2 dx, 2 dy, 2 dz. Let the centre of the element be at P(x, y, z),

8C 8 dx dy dzat,

ac 8F" 8Fy 8F. --+-+-+- = O.

8t 8x 8y 8z

(1.2)


5

4


If the diffusion coefficient is constant, Fx, Fy, F. are given by (1.1), and (1.2) becomes

we obtain the equation for diffusion in a cylinder,

dT = f(t) dt,

ec 1 { a ( ac) a (D ac) a ( ac)}

at = -;: ar rDar + ae -;: ao + az rD at '

(1.7)

reducing simply to

in terms of the cylindrical coordinates r, e, z. The corresponding equation a sphere in terms of spherical polar coordinates r, e, 4> is obtained by

ac a2c -=D~ at ax2

if diffusion is one-dimensional i.e. if there is a gradient of concentration only along the x-axis. Expressions (1.1) and (1.4) are usually referred to as Fick's first and second laws of diffusion, since they were first formulated by Fick (1855) by direct analogy with the equations of heat conduction.

In many systems, e.g. the interdiffusion of metals or the diffusion of organic vapours in high-polymer substances, D depends on the concentration diffusing substance C. In this case, and also when the medium is not homogeneous so that D varies from point to point, eqn (1.2) becomes

sc a ( ac) a ( ac) a ( ac)

at = ax D ax + ay D oy + oz D oz '

where D may be a function of x, y, z, and C.

If D depends on the time during which diffusion has been taking but not on any of the other variables, i.e.

x = r sin e cos 4>, y = r sin 0 sin 4>,

z = r cos 0,

considering an element of volume of a sphere of sides dr, r dO, r sin 0 d4>.

oC = _!_ {~(Dr2 ac) +_1_ !_ (D sin 0 ac) +____E_ aZc}.

at rZ or or sin e ao ae sin? e a4>2

11 .... '.U.', .... forms of (1.7) and (1.8) for purely radial diffusion, e.g. in a long where end effects are negligible or in a spherically symmetrical given in Chapters 5 and 6, where some solutions ofthe differential are to be found. All these diffusion equations can be expressed in nomenclature of vector analysis as

(1.8)

D = f(t),

then on introducing a new time-scale T such that ,

ae = div (D grad C). at

the diffusion equation becomes ec »c a2c ec or = axl + oyl + azz '

which is the same as (1.3) for a constant diffusion coefficient equal to unity. 1.3.1. Diffusion in a cylinder and sphere Other forms of the above equations follow by transformation of

ordinates, or by considering elements of volume of different shape. Thus putting

media have differen t diffusion properties in different directions. examples are crystals, textile fibres, and polymer films in have a preferential direction of orientation. For such always true, as was stated in § 1.2 (p. 2) for isotropic media, of flow of diffusing substance at any point is normal to the concentration through the point. This means that (1.1) in general by the assumptions

x = r cos e, y = r sin e,

or by considering an element of volume of a cylinder of sides dr, r dO,

ac oc ae

-Fy = D21 ax +Dnay+D13az '

ac ac ac

-Fz = D31 ax +D3Z ay +D3Jaz

(1.9)

6



7

Certain properties deduced by Carslaw and Jaeger (1959, p. 46) indicate the physical significance of the ellipsoid and also of the principal axes of diffusion. Thus it can be shown that the square of the radius vector of the ellipsoid in any direction is inversely proportional to the diffusion coefficient normal to the surfaces of constant concentration at points where their normals are in that direction. Hence the diffusion coefficient Dn at right angles surfaces whose normals have direction cosines 1, m, n relative to the princiaxes of diffusion is given by

so that Fx, for example, depends not only on oCjox but also on oCjoy oCjoz. The Ds have the significance that D13 oe/oz, for example, is the tribution to the rate of transfer in the x-direction due to the component concentration gradient in the z-direction. Substituting from (1.9) for Fs in (1.2) we obtain

ec o2e o2e e-c ale at = Dll ox2 + D22 ay2 + D33 az2 + (D23 + Dd oy oz

a2e a2e

+(D31 +D13) ozox +(D12+D21) ox oy'

(1.16)

if the Ds are taken as constant. The extension to non-constant Ds is 0 from (1.5). A transformation to rectangular coordinates ~, 11, ( can be which reduces (1.10) to

ae a2c alc alc at = Dl ae +Dz 0112 +D3 oe .

This is the same transformation as that by which the ellipsoid DllX2 +D22y2 +D33ZZ +(D23 +D32)yz+(D31 +D13)ZX +(D12 +D21)XY = constant

and Jaeger further show that if there is symmetry about the planes and IJ = 0, then the general relationships (1.9) for the Fs reduce to

,run'ITl{"{Jm"" of measurements in anisotropic media

in the majority of experiments designed to measure a diffusion the flow is arranged to be one-dimensional, it is worth while to measurements are affected by anisotropy. If the diffusion is :lJlJ'>lU,lJ<lJ in the sense that a concentration gradient exists only along of x, it is clear from (1.10), since both e and aCjax are everywhere of y and z, that the diffusion is governed by the simple equation

ec 02C at = o., ox2'

(1.18)

is reduced to

Dl~2 +D2112 +D3(2 = constant.

The new axes may be called the principal axes of diffusion and D1, D2, D3 principal diffusion coefficients. If we make the further transformation

where D may be chosen arbitrarily, (1.11) becomes

ec (a2C 02C 02C)

at = D a~i + alJi + an .

diffusion coefficient measured. If the direction of diffusion is of a principal axis, then Dll is equal to one or other of the coefficients D l' D2, or D3• Otherwise the coefficient to D1, D2, D3, by (1.16) is measured. This would be

,,"'''Ufo''''. by an observation of the rate of flow through a plane ,',cut so that its normal has direction cosines (1, m, n) relative of diffusion of the crystal. Similar remarks apply to a in which there is both uniplanar and undirectional molecules are arranged with their long axes lying mainly of the sheet and all parallel to one direction in that plane. of diffusion of such a sheet will be normal to the plane

. . perpendicular to the preferred direction of orientation . ..• if a concentration gradient exists in one direction only,

This has the same form as eqn (1.3) for isotropic media, and hence problems in anisotropic media can be reduced to corresponding nr''''''''TT1~ isotropic media. Whether or not this can be done in a given case depends the boundary conditions. Thus it is possible when the medium is . . when it is bounded by planes perpendicular to the principal axes of so that the boundary conditions are of the familiar form e = ~ = 0, (= 1, t > 0, for 'example, and similarly for 11 and" The diffusion into an anisotropic cylinder which has its axis along ~ and is by planes perpendicular to ~ reduces to the corresponding problem isotropic cylinder provided D2 = D3•

8



9

it is clear from (1.9) and (1.15) that the diffusion flow is not along this direction unless it coincides with a principal axis of diffusion.

between P and C such as C = Sp, where S is the solubility. There is no analogue to S in heat flow.

(b) Prescribed heat flux corresponds to flux of diffusant and we have

1.4.2. Conversion of heat flow to diffusion solutions

Carslaw and Jaeger (1959) and other books contain a wealth of solutions of the heat-conduction equation. There is general awareness among scientists and engineers that the phenomena of heat flow and diffusion are basically the same. Nevertheless, many non-mathematicians experience difficulty in making the changes of notation needed to transcribe from one set of solutions to the other. In this section we examine in detail the correspondence between the physical parameters, the variables, and the equations and boundary conditions which occur in heat-flow and diffusion problems. We take the one-dimensional case with constant properties as an illustration.

(i) The equations. Diffusion theory is based on Fick's two equations (1.1) and (1.4), where C is the concentration of diffusant expressed, say, in mass per unit volume, and D is the diffusion coefficient. The two corresponding eq tions in heat flow are

F = - K o8/8x,

8B (K) o2B

at = cp 8x2'

- K 8e/ox = F(t), in heat,

- D 8C/ox = F(t), in diffusion,

(1.22a) (1.22b)

F is in general a known function of time but may be constant.

A heat-insulated surface corresponds to an impermeable surface and is case of (1.22a, b) with F = 0, i.e.

oC/ox = oB/ax = O.

is referred to as a 'radiation boundary condition' in heat flow means that the heat flux across unit area of the surface is proportional between the surface temperature 8. and the temperature Bo medium, i.e. is given by H(8. - Bo). But the rate of heat loss from of a surface is - K 8e / an in the direction of the normal n, measured

the surface, so that the boundary condition is

where 8 is temperature, K is the heat conductivity, p is density, and c heat, so that pc is the heat capacity per unit volume. The space coordinate

x and t is time. In (1.19), F is the amount of heat flowing in the direction of increasing per unit time through unit area of a section which is normal to direction of x.

oB/on + h(B. - Bo) = 0,

H/K. Sometimes this is referred to as Newton's law of cooling. to surface evaporation in diffusion, and we have

(ii) Variables and parameters. In order that the two sets of equations correspond we may identify concentration C with temperature B, and D = Kin (1.1) and (1.19) and D = K/(cp) in (1.4) and (1.20). The two l .... fT~,lh'" mean cp = 1. This is a consequence of our having identified C with B. 'diffusing substance' in heat flow is heat not temperature. The factor cp needed to convert temperature to the amount of heat per unit volume; concentration is, by definition, the amount of diffusing substance per volume and so no conversion factor is needed, i.e. cp = 1. It is usual to K/(cp) = k, the heat diffusivity. All three of K, k, cp appear in u .. <,,-,.IU equations, and to convert to diffusion terms we take

=H/D.

is perpendicular to the x-direction, as an example, 8Cjon = along the direction of x increasing but aC/on = - oC/ox if fl;u;:;mjl(. Thus for a slab between x = 0 and x = 1 we have

ac/8x+IX(C.-Co) = 0, X = 1,

D = K = k and cp = 1. (iii) Boundary conditions. (a) Prescribed surface temperature to prescribed concentration just within the surface, say of a plane

If the surface is in contact with a vapour at pressure p, there is some

-oc/ox+a(Cs-Co) = 0, x = O.

~~~",,~,f"~ of heat is always at a uniform temperature and so is respect to a well-stirred fluid. Thus, a boundary condition contact with a perfect conductor also describes diffusion

well-stirred solution or vapour. Conservation of heat from a volume Vand uniform temperature Os, a function of condition on the surface x = 0 of a medium, the fluid

cp V o8Jot = K 8B/ax,

x = 0,

10


where c, p have their usual meaning for the fluid. Correspondingly for UUIUO,IUJ.J.! we have

VOcJat = DoC/ox,

x = O.

2

In addition, es will be the temperature just within the surface but the ', concentrations may be related by some isotherm equation. Care is neededs with algebraic signs as in the previous section.

(f) The conservation principle applied at the interface between two of different properties leads immediately to boundary conditions

METHODS OF SOLUTION WHEN THE DIFFUSION COEFFICIENT IS CONSTANT

L solutions of the diffusion equation can be obtained for a variety of and boundary conditions provided the diffusion coefficient is constant. ... solution usually has one oftwo standard forms. Either it is comprised

. of error functions or related integrals, in wllich case it is most suit-

·_",,~,.~i<· .. evaluation at small times, i.e. in the early stages of diffusion, .. the form of a trigonometrical series which converges most satis.. for large values of time. When diffusion occurs in a cylinder the

series is replaced by a series of Bessel functions. Of the three of solution described in this chapter, the first two illustrate the Ml':JlUJ",""'l'A; of the two standard types of solution. The third, emLaplace transform, is essentially an operator method by which solution may be obtained. It is the most powerful of the three, for more complicated problems. The methods are presented here possible. The fuller treatments necessary to make the discussion rigorous are to be found in works on heat conduction, e.g.

Jaeger (1959).

in heat flow and

in diffusion, where the suffices 1 and 2 denote the two media and P and are constant.

(g) If heat is produced in a medium, e.g. as a result of an exothermic tion, at a rate A per unit volume, this must be added to the right side of (1.20), which can be written

oe o2e

cp ot = K ox2 + A.

The diffusion equation (1.4) must be similarly modified if the diffusing stance is created or removed as diffusion proceeds. We identify A as the of creation per unit volume and put cp = 1, K = D as usual.

(h) Problems in heat flow may involve moving boundaries on which changes occur, accompanied by the absorption or liberation of latent In the first case, the relevant feature is that latent heat is removed taneously from the heat-conduction process, in which it takes no further The diffusion counterpart is the immobilizing of diffusing molecules fixed sites or in holes. The velocity of the transformation boundary at is related to the difference between the rate of heat arriving and leaving it conduction by a condition

reflection and superposition

A 2

C = texp (-x /4Dt), t

(2.1)

-Kl oedox+Kzoe2/ox = LpdX/dt,

where L is the latent heat per unit mass and p the density, assuming no changes accompany the transformation. The condition in a diffusion follows by writing e = C, D = K, p = 1 as usual, and L becomes the of the immobilizing sites in unit volume for trapped diffusing molecules.

Useful collections of mathematical solutions of the diffusion equations to be found in books by Barrer (1951), Jost (1952), and Jacobs (1967). solutions are of particular interest to biologists and biophysicists.

constant, is a solution of

(2.2)

for diffusion in one dimension when D is constant. is symmetrical with respect to x = 0, tends to zero as x positively or negatively for t > 0, and for t = 0 it except at x = 0, where it becomes infinite. The total . •..•.. M diffusing in a cylinder of infinite length and unit

12

METHODS OF SOLUTION WHEN THE

DIFFUSION COEFFICIENT IS CONSTANT

13

1·0

dx = 2(Dt)t d~,

the problem just considered, half the diffusing substance moves in the direction of positive x and the other half along negative x. If, however, we have a semi-infinite cylinder extending over the region x > 0 and with an impermeable boundary at x = 0, all the diffusion occurs in the direction of positive x. We can consider the solution for negative x to be reflected in the plane .. 0 and superposed on the original distribution in the region x > O. Since •.•. original solution was symmetrical about x = 0 the concentration distribufor the semi-infinite cylinder is given by

cross-section is given by

M = f~oo Cdx,

and if the concentration distribution is that of expression (2.1) we see, writing

that

M

C = (-)texp (-x2j4Dt). nDt·

(2.7)

M = 2ADt f~oo exp(-~2)d~ = 2A(nD)t.

Expression (2.5) shows that the amount of substance diffusing constant and equal to the amount originally deposited in the plane x = Thus, on substituting for A from (2.5) in eqn (2.1), we obtain

C M 2

= 2(nDt)t exp (-x /4Dt),

and this is therefore the solution which describes the spreading by UlI.IU1loJIU of an amount of substance M deposited at time t = 0 in the plane x = Fig. 2.1 shows typical distributions at three successive times.

of reflection and superposition is mathematically sound, for

at x = 0 means the adding of two solutions of the diffusion equathis equation is linear the sum of the two solutions is itself a soluwe see that (2.7) satisfies the condition that the total amount of substance remains constant at M. Furthermore, the condition to be at the impermeable boundary is

ac;ax = 0,

x = 0,

(2.8)

is the mathematical condition for zero flow across a boundary. is zero at x = ° in the original solution (2.6), it is clearly still zero and superposition.

1·25

initial distributions

•.. have considered only cases in which all the diffusing substance is initially in a plane. More frequently in practice, however, the occupies a finite region and we have an initial state such

0·75 ~

V 0·5

x < 0,

C = 0,

x > 0,

t = O.

(2.9)

x

distribution, for example, when a long column of clear a long column of solution, or when two long metal bars are end to end. The solution to such a problem is readily the extended distribution to be composed of an of line sources and by superposing the corresponding solutions. With reference to Fig. 2.2, consider the in an element of width t5~ to be a line source of strength ...... (2.6) the concentration at point P, distance ~ from the

0·25

FIG. 2.1. Concentration-distance curves for an instantaneous plane source. Numbers on are values of Dt.

2.2.2. Reflection at a boundary

Expression (2.6) can be used to build up solutions of other pro linear flow by introducing the concept of reflection at a boundary.

Co o~ 2

2(nDt)t exp ( - ~ j4Dt),

solution due to the initial distribution (2.9) is given by

14



15

summing over successive elements 6~, i.e. by

1·0r---~

C foo C foo

C(x, t) = 2( ; )t exp (- ~2/4Dt) d~ = -T exp (-112) dl1,

11: t x 11: x/2.,f(Dt)

where 11 = ~/2.j(Dt).

Ic

-----.,.,.".--IC{j

2

x

FIG. 2.2. Extended initial distribution.

Concentration-distance curve for an extended source of infinite extent.

A standard mathematical function, of which extensive tables are is the error function, usually written as erf z, where

way, we can study the diffusion of a substance initially confined - h < x < + h as in Fig. 2.4. Here the integration is from instead of from x to 00 as in (2.10), leading immediately to the

This function has the properties erf(-z) = -erfz, and hence, since

erf(O) = 0,

erf( 00) = 1,

C = !Co{erf 2~~;t) +erf 2~~;tJ

distribution at successive times is shown in Fig. 2.4. It is ~S1{stc~m can be cut in half by a plane at x = 0 without affecting

(2.15)

= 1 - erf z = erfe z,

where erfe is referred to as the error-function complement, the (2.10) of the diffusion problem is usually written in the form

1 f X

C(x, t) = lCO eric 2.j(Dt)'

Convenient tables of the error function are those of the Works Association (1941) and shorter tables are to be found, for example, in Thomson and Comrie (1944). Table 2.1, taken from Carslaw and J (1959), is sufficient for many practical purposes. The form of the "'Vl.l"'''lU tion distribution is shown in Fig. 2.3. It is clear from (2.14) that C = x = 0 for all t > O.

The error function therefore enters into the solution of a diffusion as a consequence of summing the effect of a series of line sources, yielding an exponential type of distribution.

o x/h

curves for an extended source of limited extent. Numbers on curves are values of (Dt/h2)t.

16



17

the distribution, which is symmetrical about x = O. Therefore """QQ,,...";; in which case it is necessary to change the sign of the concentration when

(2.15) also gives the distribution in a semi-infinite system. Such a system . reflected at the boundary. A further example of the use of this method is

realized in practice in the classical experiment in which a cylinder contains by Jost (1952). For more complicated problems, however, the reflection

layer of sol ution having on top of it an infinitely-long column ofw ater.i superposition method soon becomes unwieldy and results are more

clear. In practice, this means that concentration changes do not reach obtained by other methods.

top of the column during the time of the experiment.

oC;ox = 0,

x = I.

ME~tlI(]'d of separation of variables

method of obtaining a solution of a partial differential equation that the variables are separable. Thus we may attempt to find a of (2.2) by putting

2.2.4. Finite systems

If the column of water, referred to above, is of finite length I, the .... Vl.1U.UUJ that the concentration tends to zero as x approaches infinity is to be by the condition that there is no flow of diffusing substance through the surface, i.e.

C = X(x)T(t),

(2.18)

and T are functions of x and t respectively. Substitution in (2.2)

We have seen that this condition is satisfied if the concentration curve considered to be reflected at the boundary and the reflected curve on the original one. In the finite system we are considering now, the reflected at x = 1 is reflected again at x = 0, and then at x = I, and so on, result of each successive reflection being superposed on the original (2.15). Since the original solution is the sum of two error functions, UII;; I.OUllllUJC expression for the concentration in the finite system is an infinite series functions or error-function complements so that

{ x-h x+h 2/-h-x

C = !Co erfc 2J(Dt) -erfc 2J(Dt) -t-erfc 2J(Dt)

21+h-x 21-h+x 21+h+x

- erfc 2J(Dt) + erfc 2J(Dt) , erfc 2J(Dt)

41-h-x 41-X+h}

+ erfc 2J( erfc J + ...

Dt) 2 (Dt)

1 00 {h+2nl-x h-2nl+x}

= "lCo L erf 2J( ) --erf J .

n= - 00 Dt 2 (Dt)

A solution of this kind can be obtained for most problems in diffusion by of the Laplace transform, to be discussed in § 2.4 (p. 19), or otherwise. . solutions are most useful for calculating the concentration the early stages of diffusion, for then the series converges rapidly and t three terms give sufficient accuracy for most practical purposes.

In all cases the successive terms in the series can be regarded as from successive reflections at the boundaries. The nature of the depends on the condition to be satisfied. For the impermeable already considered a simple reflection ensures that iJC/ox = 0 as req Another boundary condition which occurs frequently is of the type C

1 dT D d2X

Tdt = X dx2'

(2.19)

. have on the left-hand side an expression depending on t only, t-hand side depends on x only. Both sides therefore must same constant which, for the sake of the subsequent algebra, taken as -AlD. We have, therefore, two ordinary differential

1 dT T dt

(2.20)

(2.21)

T -;.'Dr =e ,

(2.22)

x = A sin AX + B cos AX,

(2.23)

of (2.2) of the form

= (A sin AX + B COsAX) exp ( - A 2 Dt),

(2.24)

18



19

The boundary conditions (2.27) demand that

2.4. Method of the Laplace transform

. The Laplace transformation is a mathematical device which is useful foi' the solution of various problems in mathematical physics. Application Mthe Laplace transform to the diffusion equation removes the time variable, an ordinary differential equation the solution of which yields the of the concentration as a function of the space variables x, y, z. where Am, Bm, and Am are determined by the initial and boundary "VJliUiUVIJ"·"'.,.,,,,~.·,~ then interpreted, according to certain rules, to give an expression for

for any particular problem. Thus if we are interested in diffusion out of tion in terms of x, y, z and time, satisfying the initial and

plane sheet of thickness " through which the diffusing substance is condition. Historically the method may be regarded as derived

uniformly distributed and the surfaces of which are kept at zero operational methods introduced by Heaviside. Full accounts ofthe

the conditions are transform and its application have been given by Carslaw and

(1941), Churchill (1944), and others. Shorter accounts by Jaeger Tranter (1951) are also available. Here we shall deal only with its to the diffusion equation, the aim being to describe rather than to

procedure.

of many problems in diffusion by this method calls for no beyond ordinary calculus. No attempt is made here to explain to the more difficult problems for which the theory of func,I,OUlJll-"'''''' variable must be used, though solutions to problems of this in later chapters. The fuller accounts should be consulted for of such solutions.

of the Laplace transform

) to be a known function of t for positive values of t, Then f(p) of f(t) is defined as

f(p) = {'J e-P'l'(t)dt, (2.32)

number sufficiently large to make the integral (2.32) converge.

,",VIIUp""'" number whose real part is sufficiently large, but in the it suffices to think of it simply as a real positive number. = e21, p must exceed 2. Unless it is necessary to emphasize of p, just as f is a function of t, we shall usually denote the of j by f.

of common functions are readily constructed by carrytegration in (2.32) as in the following examples:

f(t) = 1, f(p) = 100 e - pI dt = lip, (2.33)

where A and B are constants of integration. Since (2.2) is a linear eq the most general solution is obtained by summing solutions of type (2. so that we have

m=l

00

C = L. (Am sin AmX + Bm COsAmX) exp ( - A;,Dt),

C =0,

x = 0,

x = t,

t > O.

fOO foo 1

f(p) = e-pt e'" dt = e-(P-a)1 dt = ~ ,

o 0 p-a

(2.34)

C = Co, 0 < x < l, t = 0

Am = mnf],

and hence the initial condition (2.26) becomes

00

Co = L Am sin (mnxll), 0 < x < I.

I

By multiplying both sides of (2.29) by sin (pnx/t) and integrating from 0 to using the relationships

f' . pnx . mnx d {O,

sm-Slll-- x =

o 1 1 tl,

m #p,

,m = p,

we find that terms for which m is even vanish, and

Am = 4Colmn, m = 1, 3, 5, ....

The final solution is therefore

C - 4Co ~ 1 {D(2 1)2 2/121 . (2n+ l)nx

- ~ L.. --. - exp - n + n t f sm ,

n n=O 2n+ 1 {

where 2n + 1 has been substituted for m for convenience so that n takes 0,1,2, .... This trigonometrical-series type of solution converges satistacton for moderate and large times, and it is then used for numerical evaluation preference to the error-function type of solution discussed earlier in § (p. 16).

In (2.29) the initial distribution is expressed as a sum of sine This reveals the physical significance of the trigonometrical series in (2. each term of which corresponds to a term in the Fourier series (2.29) which the initial distribution can be represented.

foo w

f (p) = e - pr sin wt dt = 2 2 .

o P +w

(235)

20



21

x = o.

the Laplace transform reduces the partial differential equation (2.36) ordinary differential equation (2.42). The solution of (2.42) satisfying and for which C remains finite as x approaches infinity is

C = Co e-QX,

p

(2.44)

Extensive tables or dictionaries of Laplace transforms are available, ~~¥~"'~"''_ ..• in the works referred to above. A short table of transforms occurring quently in diffusion problems is reproduced from Carslaw and J book (1959) in Table 2.2.

2.4.2. Semi-infinite medium

As an example of the application of the Laplace transform, consider problem of diffusion in a semi-infinite medium, x > 0, when the boundary kept at a constant concentration Co, the initial concentration being throughout the medium. We need a solution of

OC a2c at = D ox2'

satisfying the boundary condition

= pfI). Reference to Table 2.2, item 8, shows that the function whose is given by (2.44) is

(2.45)

erfcz = l-erfz.

(2.46)

C = Co, and the initial condition

x = 0,

t » 0,

to verify that (2.45) satisfies (2.36), (2.37), and (2.38) and that it is required solution of the diffusion problem.

C= 0,

x> 0,

t = O.

sheet

just considered the transform solution could be interpreted

reference to the table of transforms. Consider now, as an slightly more difficult problem in which this is not so, a plane 21, whose surface are maintained at constant concentration concentration of diffusing substance throughout the sheet

sheet occupy the region -/ .:;;; x .:;;; I, so that there is symmetry the boundary conditions may be written

On multiplying both sides of (2.36) by e - pI and integrating with respect from 0 to 00 we obtain

fOO »c 1 foo ec

e-PI-2 dt-- e-P'-dt = O.

o ox D 0 at

If we assume thatthe orders of differentiation and integration can be . changed, and this can be justified for the functions in which we are n', ,t""'",ct-A; then

C = Co, x = I, t ~ 0,

(2.47) (2.48)

aC/aX = 0, x = 0, t ~ o.

fOO ec foo

e-P'-dt = [Ce-Pt]O'+p Ce-ptdt = pC,

o at 0

since the term in the square bracket vanishes at t = 0 by virtue of the' condition (2.38) and at t = 00 through the exponential factor. Thus reduces to

the condition that there is no diffusion across the central It is often more convenient to use this condition and to the sheet, 0 .:;;; x .:;;; I, instead of using the condition

02C _

D ox2 = pc.

By treating the boundary condition (2.37) in the same way we obtain C = foo Coe-ptdt = Co,

o p

for the Laplace transform Care

d2C 2-

dx2 -q C = 0,

0< x < 1,

(2.49)

dC/dx = 0,

x = 0,

(2.50)

C = Co/p,

x = l,

(2.51)

This is a series of error functions such as we obtained by the JllCUll,IU reflection and superposition, Successive terms are in fact the \;UI.I\;CUlidUI at depths 1- x, I + x, 3/- x, 31 + x, ... in the semi-infinite medium. The converges quite rapidly for all except large values of Dt/F. For ClI.i:lllI1J1C. have for the concentration at the centre of the sheet (x = 0) when Dt/12

,

22


where q2 = p/ D as before. The solution of these is C = Co cosh qx.

p cosh ql

There are two methods of dealing with this transform solution, leading to two standard types of solution we have already met. We shall first solution useful for small values of the time.

(i) Expansion in negative exponentials. We express the hyperbolic in (2.52) in terms of negative exponentials and expand in a series by binomial theorem. Thus we obtain from (2.52),

C ( q" + - qX) C 00

C = 0 e e = _Q{e-q(I-X)+e-q(I+X)} L (-It e-2nql

p eq'(1 +e 2ql) p n~O

C 00 C 00

=_Q L (_lte-q{(2"+1)I-xl+~ L (_1)"e-Q(2n+l)I+::CI.

p n~O P "=0

Thus, using item 8 of the table of transforms (Table 2.2), we obtain

00 (2n+l)l-x 00 (2n+l)l+x

C = Co n~o (_1)n erfc 2.j(Dt) + CO "~O (-1)" erfc 2.j(Dt)

qco = 0·9590-0·0678+0·0008 = 0·8920, and when Dt/P = 0·25

qco = 0·3146-0·0001 = 0·3145.

(ii) Expression in partial fractions. It can be shown that if a has the form

_ f(p) Y = g(p) ,

where f(p) and g(p) are polynomials in p which have no common degree of f(p) being lower than that of g(p), and if

g(p) = (p-a1)(p-a2)· .. (p-an),

where aI' a2, ... , an are constants which may be real or complex but


23

g'(ar) denotes the value of dg(p)/dp when p = ar• A proof of this by (1949) is reproduced in the Appendix to this chapter. It is derived by (2.57) in partial fractions. Since the hyperbolic functions cosh z z can be represented by the following infinite products (see, e.g. . (1909, p. 275»

cosh z = (1 + ~:) ( 1 + 3~~2) ( 1 + 5~~2) ,

sinh z = z( 1 + ::) ( 1 + 2~:2) ( 1 + 3~:2) ,

these functions such as in (2.52) may still be regarded as being •. (2.57) except that now f(p) and g(p) have an infinite number of . aI' a2"" are the zeros of g(p), i.e. solutions of the equation, if these are all different it is plausible to assume that (2.59) still = 00. The justification of this assumption involves the theory of a complex variable in order to carry out a contour integration found in the fuller accounts of the subject. There is, in fact, a '.11""'",''''''''' argument by which the use of (2.59) with n = 00, can ,UHIU"lUl1 problems in a finite region only. It must not be applied .....1 ....... 1f.'1", for the semi-infinite region. The above refers to al, The extension of (2.59) to cases in which g(p) has repeated

its factors is square, is given in the Appendix. Its application of factors is still justifiable.

consider the application of (2.59) to (2.52). First the zeros of . must be found. Clearly, p = 0 is a zero, and the other zeros of q for which cosh ql = 0, i.e,

(2.60)

(2.61)

(2n+ l)ni

q = ± 21 ,n = 0, 1, 2, ...

(2.62)

D(2n+l)2n2

p = - 4/2 ,n = 0, 1,2, ....

we need

(2.63)

= d: (p cosh ql) = cosh ql + !ql sinh qt.

(2.64)

24



25

CICo = 1-0·1080 = 0·8920,

OCr t > 0, (2.73)
Cl.ra-P,C, = 0, x, = a.,
x.
,oc. , x, = b., t > 0, (2.74)
CI..,r-+f3.c. = 0,
xr For the zero p = 0, g'{p) = 1. For the other zeros, given by (2.62) and cosh ql = 0, and

. . h(2n+ l)ni .. (2n+ l)n .( 1)"

smh ql = sin 2 = 1 sin 2 = 1 - ,

so that for these zeros, by substituting in (2.64) we obtain

, (2n + 1 )n{ -1)" + 1

g (p) = 4

Hence finally by inserting the zeros into (2.59) we obtain

4C 00 (- 1 t (2n + 1 )nx

C = Co-_o L --exp{ -D(2n+1)2n2tI412} cos 21 .

n "=02n+1

This is the trigonometrical-series type of solution obtained previously the method of separation of the variables. The series converges rapidly large values of t. Thus for the concentration at the centre of the sheet (x when Dtl12 = 1,

(J.. and p, are constants, either of which may be zero (so that the surface concentration and no flow of heat at the surface are with initial conditions

c.(x., t) = C.(x,),

t = 0,

a. < x, < b..

(2.75)

of (2.70) in the region defined by (2.71) with

t = 0,

(2.76)

' ... ac

--pc = 0

Or'

X.

Xr = an

t > 0, r = 1,2,3,

(2.77)

and when Dtlll = 0·25

qco = 1-0·6872+0·0017 = 0·3145.

t > 0, r = 1,2,3,

(2.78)

Consider the equation of diffusion

02C fPc alc 1 oc -+-+---.oxI ox~ ox~ - D ot'

c = C1(Xl, t)Cl(X2, t)C3(X3, t). (2.79) in (2.70) gives

02Cl f)2C3 1 (OCl OCl OC3)

1 ox~ +ClCl OX~ -15 C2C3Tt+C3CC8t+C1C28t" = 0,

(2.79)

2.S. Solutions in two and three dimensions

2.S.1. Solutions expressed as the product of the solutions of simpler

,

(2.80) the initial and boundary conditions (2.76), (2.77), and

in the rectangular parallelepiped

JUUIUU'U is (2.76), namely that the initial condition must be of the initial conditions for the one- variable problems wand Jaeger (1959) give solutions for a rectangular

parallelepipeds, cylinders and some examples of isotherms

F or certain types of initial and boundary conditions, the solution of the product of the solutions ofthe three one-variable problems, and be written down immediately if these are known. The following given by Carslaw and Jaeger (1959, p. 33).

Suppose c,(x" t), i = 1,2,3, is the solution of

derived a much more general relationship between the two-dimensional problems for an infinite cylinder of and the transient solutions of the corresponding nrobtems in finite cylinders. The boundary conditions

26



27

on the end faces of the cylinder may be of the type describing ,.. ....... C".t_"',«l!~~ .. 'I1F.nu concentration, constant flux, or evaporation, and heat may be generated rate independent of time and the axial coordinate.

An example of Goldenberg's relationships is afforded by the "~""~r:>w"-~

slab, bounded by the planes z = 0 and z = 1, with heat produced in cylindrical region R of arbitrary cross-section for time t > 0, at the rate A per unit time per unit volume.

The slab is initially at zero temperature with its faces maintained at temperature for t > O. The corresponding two-dimensional nrl"1.hl .. "" obtained when the thickness 1 of the slab is infinite and the independent of z. Goldenberg (1963) shows that the solution for the finite slab is related to f(x, y, t) for the infinite slab by the .... A~}L .... " ... u

4 ~ sin {(2n+ 1)1I:zjl} {-d'f J: J,' -d'fd}

V = - £... e +u e t,

11:.=0 2n+l 0

where b = Dn2(2n + 1)2//2 and D is the diffusion coefficient.

The same relationship is valid for a hollow cylinder of arbitrary section and for the region external to a cylinder of arbitrary ,·rr.""_,,prn; Similar relationships hold in other situations discussed by Goldenberg.

APPENDIX TO CHAPTER 2

p = a, in this gives

= Ar(ar-at) ... (ar-ar-d(ar-ar+l) ... (ar-an) (r = 1,2, ... ,n). (4) for Ar from (4) in (2) gives

(5)

(6)

differentiating by the ordinary rule for differentiation of a product,

n

g'(p) = L: (p - al) ... (p - ar- 1)(P - ar+ 1) ... (p -an)' (7)

r= 1

a, in this, gives

g'(a,) = (ar- a1) ... (ar - a.: l)(ar - ar+ d ... (ar- a.), in (5) gives a further form for y(p) namely

(8)

2.6. Other solutions

Langford (1967) obtained new solutions of the one-dimensional equation for temperature and heat flux both prescribed at the same boundary. They take the form of series of polynomial and quasi-p .... 'u"". solutions for plane sheets, cylinders, and spheres. They include as cases some of the old or classical solutions. They also have phase change problems with boundaries moving at a constant velocity.

_( ) _~. f(a,)

y p - i.... ,.

r= 1 (p - ar)g (ar)

3 of Table 2.2 to successive terms of (9) we obtain immedi-

(9)

(10)

only to the case in which g(p) has no repeated zeros, but it

. for the case of repeated factors. Thus (10) implies

factor p - a, ofthe denominator of y(p) there corresponds a

To deduce the function y(t) whose Laplace transform y(p) is given by

_() f(p)

y p = g(p)'

we first put y(p) into partial fractions in the usual way by assuming

f{p) = ± ~ = ~+~+~+ ... +~.

g(p) r=lP-ar p-al p-a2 p-a3 p-a.

f(ar) arr --e g'(ar)

(11)

generalization is that, to each squared factor (P_b)l

of y(p) there corresponds a term

~-"--J tebr+ [~{(P-b)2f(P)}J e" (12)

p=b dp g(p) p=b

multiple factor (p- c)" of the denominator of y(p)

Then

n

f(p) == L A,(p-ad .. ·(p-ar-l}(p-ar+d···(p-a.),

r= 1

(13)

INFINITE AND SEMI-INFINITE MEDIA

29

3

a point source on an infinite plane surface is thus given by

M 2

C = -~exp (-r /4Dt).

4nDt

(3.4)


3.1. Introduction

IN this and the following three chapters solutions of the diffusion are presented for different initial and boundary conditions. In nearly all the diffusion coefficient is taken as constant. In many cases the solutions readily evaluated numerically with the help of tables of standard matical functions. Where this is not so, and where numerical evalua tedious, as many graphical and tabulated solutions as space permits are

c = M iexp(-r2/4Dt). 8(nDt)

the appropriate solution for a point source with respect space variables, solutions may be obtained for line, surface, sources. Thus for surface diffusion in the x, y plane due to a line the j-axis we have

(3.5)

00 M M 2

. _- exp { _(x2 + y2)/4Dt} dy = t exp (- x /4Dt), (3.6)

..• 00 4nDt 2(rcDt)

-

M is the amount of diffusing substance deposited initially per

of the line source. This is the same as expression (2.6) of Chapter 2 source of strength M per unit area in an infinite volume. The result for a line source of strength M per unit length in an , obtained by integrating (3.5), is

M

C = _~ exp (- r2/4Dt), (3.7)

4nDt

expression as (3.4) for a point source on an infinite plane M has a different significance in the two cases. Results for a are derived by Carslaw and Jaeger (1959, p. 255). The • cylindrical sources are likely to be of practical interest. If the is initially distributed uniformly through a sphere of C at radius r, and time t is given by

3.2. Instantaneous sources

Under this heading are included all problems in which an diffusing substance is deposited within a certain restricted region at t = 0 and left to diffuse throughout the surrounding medium. For it may be located initially at a point, or in a plane, or within a sphere, we have an instantaneous point, plane, or spherical source as the case be.

The solution for an instantaneous plane source in an infinite medi already been given in Chapter 2, eqn (2.6). The corresponding solution instantaneous point source on an infinite plane surface is obtained same way by recognizing that

. {erf 2)~;t) +erf 2~7;t)}

J(~t) [exp {-(a-r)2/4Dt} -exp {-(a+r)2/4Dt}], (3.8)

····uniform concentration in the sphere initially. Expression written in terms of the total amount of diffusing substance

is a solution of

which is the equation for diffusion in two dimensions when the coefficient is constant. The arbitrary constant A is expressed in terms the total amount of substance diffusing, by performing the integration

. M = J~co J~co C dx dy = 4nDA,

the concentration C being expressed in this problem as the diffusing substance per unit area of surface. The concentration at a

(3.9)

......•... result for a cylinder of radius a may be written in the form

exp ( - r2/4Dt) s: exp ( - r'2 /4Dt)I o( ;~t) r' dr', (3.10)

where lois the modified Bessel function of the first kind of order Tables of 10 are available. The integral in (3.10) has to be evaluated cally except on the axis r = 0, where (3.10) becomes

C = Co{1-exp(-a2/4Dt)}.

These expressions may be applied, for example, to the diffusion of a __ ' __ 00··.·'.·,· or cylinder of solute into a large volume of solvent. Curves showing the centration distribution at successive times are given in Figs. 3.1 and 3.2.

FIG. 3.1. Concentration distributions for a spherical source. Numbers on curves are (Dt/al)i.

30


31


2 ria

expression (3.10) and the curves of Fig. 3.2 also apply to a circular of radius a, on an infinite plane surface, if Co denotes the uniform of the diffusing substance over the region 0 < r < a initially. I"·_¥~."""p solution given by Rideal and Tadayon (1954) is

C = «c; {'" J1{ua)JO(ur) exp (-Dtu2)du, (3.10a)

and J 1 are Bessel functions of the first kind and of order zero and

.... ectivelv, Tables of J 0 and J 1 are available. Rideal and Tadayon an expression for the total amount of diffusing substance Q rethe disc after time t, which is

f'" J2(ua)

Q = 2na2Co _1 - exp ( - Dtu1) duo

o u

(3.10b)

of t, (3.10b) becomes

Q = na2co{ 1-~(~~r},

(3.l0c)

(3.10d)

extended source in an infinite medium, where the diffusing y occupies the semi-infinite region x < 0 or is confined -h < x < h, have been considered (see eqns (2.14) and (2.15) The solution to the corresponding problem in which the region . at zero concentration and [x] > h at a uniform concentration

3

4

1 {h-X h+x }

C = zCo erfc 2.J(Dt) -t-erfc 2.J(Dt) .

(3.12)

Philip (1973) have obtained an exact explicit solution for 2 with a and b positive constants and for an instantaneous at concentration b and the initial concentration uniform from the source. The analysis holds for infinite and semiphysical applications are discussed.

.s
G"
0·4
0·2

0'°0
3 4 a continuous source, from which diffusing substance is at a certain rate, is deduced from the solution for the '''u,",u •• " source by integrating with respect to time t. is liberated continuously from a point in an incp per second, the concentration at a point distant r

FIG. 3.2. Concentration distributions for a cylindrical' source. Numbers on curves are lDt/a1)t.

32



33

from the source at time t is obtained by integrating (3.5) and is

1 II dt'

C = 8( D)~ ¢(t')exp {-r2/4D(t-t')}~~I~"

n 0 (t - t )T

If ¢ is constant and equal to q, then

C = ~q~erfc--r-.

4nDr 2J(Dt)

Solutions for other continuous sources are obtained similarly and are given by Carslaw and Jaeger (1959, p. 261).

,~Jr!:triSf('rm Cases of practical interest are given below. Here MI is used through" denote the total amount of diffusing substance which has entered the

3.3. Semi-infinite media

The solution for a plane source deposited initially at the surface, x of a semi-infinite medium was given in Chapter 2, eqn (2.7), and that for initial distribution C = Co, 0 < x < h, C = 0, x > h, was seen to be by eqn (2.15) for x positive.

The problem ofthe semi-infinite medium whose surface is maintained constant concentration Co, and throughout which the {'(.,.nN,,,tr,, initially zero, was handled by the method of the Laplace Chapter 2, p. 21 (see eqn (2.45». Other results of practical importance may be obtained in the same way are given below.

(i) The concentration is Co throughout, initially, and the maintained at a constant concentration C i -

C-C1 X

CO-C1 = erf2J(Di)'

The special case of zero surface concentration is immediately obvious. rate of loss of diffusing substance from the semi-infinite medium surface concentration is zero, is given by

(DOC) = DCo

ax :<=0 J(nDt),

so that the total amount M I of diffusing substance which has left the at time t is given by integrating (3.14) with respect to t and is

Cx=o = kt, where k is a constant.

C = kt[(l+~) erfc~x x_exp {-x2/4Dt}]

2Dt 2J(Dt) .J(nDt)

k ·z ~ x

= 4 t 1 errc 2J(Dt)'

(3.16)

Me = !kt( ~t) 1.

iZerfc is defined and tabulated in Table 2.1, so that values of C down immediately. The effect of an increasing surface is shown in Fig. 3.3.

(3.17)

6

M, = 2Co(~tr

The same expression with Co replaced by C 1 gives the total amount up by the medium in time t if the initial concentration Co is zero. If the concentration is zero throughout" the semi-infinite medium, and the concentration varies with time, solutions are still obtainable by the

1·6 2'0

for variable and constant surface concentrations in a semi-infinite medium.

34


(iii) Cx=o = kt», where k is a constant.

C = kf'(!n+ 1)(4t)tn i"erfc----;'-. 2y (Dt)

MJ =!_ I(Dt) (4t)tn f'(!n + 1).

2nY f'(!n+!)

The function r'erfc is defined and tabulated up to n = 6 in Table 2.1. is the gamma function defined and tabulated, for example, by M Thomson and Comrie (1944). If n is even, so that!n = N, an integer,

f'(!n+l) = f'(N+l) = NL If n is odd, so that !n = M - 1, then 1(!n+ 1) = f'(M +1) = 1.3.5 .... (2M,-3)(2M -1)n1-j2M.

Other properties of gamma functions are

f'(n + 1) = nf'(n);

A polynomial in t1- may sometimes be a useful way of representing a surface concentration empirically. In such a case the complete for the concentration at any point is the sum of a number of terms of (3.20) corresponding to successive terms in the polynomial.

(v) These solutions can be extended to cover other initial and conditions by using the fact that for a linear differential equation the two solutions is itself a solution. For example, if

Cx=o = Co+kt and the semi-infinite medium is initially at zero concentration the solution is given by

C = Co erfc 2 JX( )+4kt i2erfc----;'-,

y Dt 2y (Dt)


35

the first term on the right-hand side of (3.26) is the solution satisfying

C = Co, C = 0,

second term satisfies

C = kt, C= 0,

x = o}. x>o

(3.27)

C = kt1-{exp (-xZj4Dt)-2 ~n( 1- erfc----;'-}

y Dt) 2y(Dt)

= k(nt)! ierfc 2.J~Dt)'

M, = !kt(nD)t.

The function ierfc is defined and tabulated in Table 2.1. In this case M directly proportional to t and so the rate of uptake of diffusing constant.

(iv) Cx= 0 = kttn, where k is a constant and n is any positive integer, even' odd.

(3.28)

the solution to the problem of the semi-infinite medium in surface concentration is given by F(t) and in which the initial is f(x), is given by

C = C1 +C2'

a solution of the diffusion equation which satisfies

(3.29)

C1 = F(t), solution satisfying C2 = f(x),

t = O}, x=o

(3.30)

t = a}.

x=o

(3.31 )

1 and Cz so defined, the diffusion equation and the initial and . are satisfied. Consider, as an example, the problem of a semi-infinite medium having a uniform initial concentra-

o .. ~t".,,~ concentration decreasing according to (3.25), with k

k '2 l' X

C = Co+4 t I errc 2.J(Dt)'

(3.32)

by adding to (3.26) the solution satisfying

t = a},

t > °

x » 0,

(3.33)

C= 0,

. {x/2.J(Dt)}.

x = 0,

,){]r.n rUll1 condition

boundary condition relates to the rate of transfer of the surface of the medium. Thus, if a stream of dry of a solid containing moisture, loss of moisture . Similarly if the solid is initially dry and the

36



37

ac

-D ax = IX(Co-Cs), x = 0,

0·4

],6

2·0

air contains water vapour, the solid takes up moisture. In each case the of exchange of moisture at any instant depends on the relative the air and the moisture concentration in the surface ofthe solid. The ·LmlO1e~!&':·:--·:"':'· reasonable assumption is that the rate of exchange is directly to the difference between the actual concentration C. in the surface at

time and the concentration Co which would be in equilibrium with

vapour pressure in the atmosphere remote from the surface. MaUJLt;l1.lCl this means that the boundary condition at the surface is

where IX is a constant of proportionality.

If the concentration in a semi-infinite medium is initially C 2 and the surface exchange is determined by (3.34), the solution is

go-=- ~: = erfc 2.j~Dt} - exp (hx + h2 Dt) erfc h-:_;~[)t)+ h.j(Dt)},

where h = IX/D. The special cases of zero concentration in the initially (Cz = 0), and evaporation into an atmosphere of zero humidity (Co = 0), are immediately obvious from (3.35). The rate at which total amount M, of diffusing substance in the semi-infinite medium per cross-sectional area changes is given by

0·0

'nN'n,,.o.i,,n distribution for a surface evaporation condition in a semi-infinite medium. Numbers on curves are values of h(Dt)!·

evaluation for large h.j(Dt) is made easier by using the asymp-

dMt (ac)

~d = - D--;- = o:(Co-C.),

t vX x=o

and, on substituting for Cs the value obtained from (3.35) by putting x after integration with respect to t we obtain for the total quantity of substance having crossed unit area of the surface,

2 1 (1 1 1 .3 )

exp (z ) erfc z = t ---2 3 +'-22 5 .. ··

n z z . z

(3.39)

u, = (CO~C2) {eXP(h2Dt)erfch.J(Dt)-1+ :th.J(Dt)} If Co is greater than C 2 this amount is taken up by the medium; if less than Cz this amount is lost by evaporation from the surface. pression (3.35) can be written in terms of any two of the dimensionl parameters

relationship

shows that the solution ofthe problem of diffusion into a ",ClIIIIII' having zero initial concentration and the surface of constant, involves only the single dimensionless para-

x

2J(Dt)'

(3.40)

x

2.j(Dt)' h.j(Dt), or hx.

In Fig. 3.4 the ratio (C-C2)/(CO-C2) is plotted as a function of for various values of hJ(Dt). In order to evaluate hMt/(Co - C2) from only one dimensionless parameter h.j(Dt) is needed. The readily evaluated from standard functions and is shown

:.'::"

> of penetration of any given concentration is proportional to of time ;

for any point to reach a given concentration is prosquare of its distance from the surface and varies diffusion coefficient;

of diffusing substance entering the medium through unit varies as the square root of time.

FIG. 3.5. Sorption curve for a surface evaporation condition in a semi-infinite

INFINITE AND SEMI"-INFINITE MEDIA


39

38

k is the ratio of the uniform concentration in the region x < 0 to that

o when final equilibrium is attained. The condition (3.42) expresses that there is no accumulation of diffusing substance at the boundary. to this problem is easily obtained by combining solutions for the medium so as to satisfy the initial and boundary conditions. We of the type

x

Cl = At +Bt erf 2.J(Dlt)'

[x]

C2 = A2 + B2 erf 2..J(D2t),

x> 0,

(3.43)

x < 0,

(3.44)

known to satisfy the diffusion equations in the two regions. By constants At, B, , A2, B2 to satisfy the initial conditions and we obtain

_ Co { t x}

Cl - 1 +k(D2/Dl)t 1 +k(D2/D1) erf 2..J(Dtt) ,

kc; [x]

C2 = 1 +k(D2/Ddt erfc 2..J(D2t)"

(3.45)

(3.46)

2

3 h(DI)1 4

5

6

a typical concentration distribution for the case where k = 1. Graphs for other cases are shown by Jost (1952) (1951). We may note that, as diffusion proceeds, the conthe interface, x = 0, remain constant at the values

These fundamental properties hold in general in semi-infinite media, vided the initial concentration is uniform and the surface ,... ... 'r..,.·p·nh·"· remains constant. They also hold for point and line sources on faces or in infinite media, and also for the case of diffusion in an medium where the diffusing substance is initially confined to the x < O. Clearly they do not apply to cases where parameters other x/2..J(Dt) are involved, such as the width of an extended source or the change of surface concentration, etc.

(3.47)

3.4. The infinite composite medium

Here we consider diffusion in systems in which two media are Suppose the region x > 0 is of one substance in which the diffusion is Dt, and in the region x < 0 the diffusion coefficient is D2. In the case, the initial conditions are that the region x > 0 is at a centration Co, and in x < 0 the concentration is zero initially. If c t for the concentration in x > 0 and C2 in x < 0 the boundary ""J.I<"' ....... the interface x = 0 may be written

C2/Cl = k, x = 0,

Dl Gct/ox = D2 GCz/ox, x = 0,

2

distribution in a composite medium. Dlt = 1, D2 = 4DI, k = !.

4

DIFFUSION IN A PLANE SHEET

45

just within the surface is C2, so that

BC/ox +h(C - Cz) = 0,

x = l,


C-C1 hx

Cl-C1 l+hl'

(4.3a)

4.1. Introduction

I N this chapter we consider various cases of one-dimensional diffusion medium bounded by two parallel planes, e.g. the planes at x = 0, x These will apply in practice to diffusion into a plane sheet of material so that effectively all the diffusing substance enters through the plane faces a negligible amount through the edges.

(4.4a)

conditions are

4.2. Steady state

Consider the case of diffusion through a plane sheet or membran thickness 1 and diffusion coefficient D, whose surfaces, x = 0, x = maintained at constant concentrations Cl, Cz respectively. After a steady state is reached in which the concentration remains constant points of the sheet. The diffusion equation in one dimension then

d2Cjdx2 = 0,

provided the diffusion coefficient D is constant. On integrating with to x we have

(4.3b)

Dh1h2(CI - Cz)

F= .

hi +hz+h1hzl

solutions for a funnel-shaped region are easily obtained

(4.4b)

dC/dx = constant, and by a further integration we have, on introducing the condinoi x = 0, x = I,

constant

systems, the surface concentrations C l' C z may not only the gas or vapour pressures PI' pz on the two sides of

The rate of transfer in the steady state is then sometimes

C-C1 X

CZ-C1 t'

Both (4.2) and (4.3) show that the concentration changes linearly from Cz through the sheet. Also, the rate of transfer of diffusing substance' same across all sections of the membrane and is given by

F = P(PI - Pl)fl,

(4.5)

F = -DdCjdx = D(C1-Cz)/I.

If the thickness 1 and the surface concen trations C l' C 2 are known, be deduced from an _observed value of F by using (4.4). C"~~~ •• ;_ rangements for measuring D in this and other ways have been Newns (1950).

If the surface x = 0 is maintained at a constant concentration

at x = 1 there is evaporation into an atmosphere for which the eq

P is referred to as the permeability constant. Here P is pIe, as em 3 gas at some standard temperature and pressecond through 1 ern? of the surface of a membrane 1 em difference across the membrane is 1 em of mercury.

constant is a much less fundamental constant than the which is expressed in units such as ern" s" 1, particularly use different units and even different definitions

coefficient is constant, and if the sorption isotherm is is a linear relationship between the external vapour •. corresponding equilibrium concentration within the

C = Sp,


47

46


membrane, then eqns (4.4) and (4.5) are equivalent, but not otherwise. linear isotherm may be written

P = DS.

Fl1 n, n,

~+~+ ... +- = (RI +R2+ ... -:l-Rn)F,

DI D2 o,

= (tiD], etc. may be termed formally the resistance to diffusion Thus the resistance to diffusion of the whole membrane is sum of the resistances of the separate layers, assuming that there to diffusion between them. This subject is treated more

§ 12.2(i) (p. 266).

(4.12)

where C is the concentration within the material of the membrane in eq brium with an external vapour pressure p, and S is the solubility. Since P1 and C2, P2 in (4.4) and (4.5) are connected by (4.6) it follows that, with regard to units,

4.2.2. Concentration-dependent diffusion coefficient

If the diffusion coefficient varies with concentration it is clear that simple value of D deduced from a measurement of the steady rate of some kind of mean value over the range of concentration involved. if D is a function of C (4.1) is to be replaced by

presented here can be obtained either by the method of the variables or by the Laplace transform as described in the results are quoted by Barrer (1951), Carslaw and Jaeger 967), Jost (1952) and others. The emphasis here is on numeri-

d ( dC)

dx D dx = 0,

and hence the relationship

still holds, as of course it must in the steady state. Integrating h .. tu"',,~,, and Cz, the two surface concentrations, we have

C = f(x),

x = 0, x = I,

a < x < I,

t ): 0, t ): 0,

t = 0,

(4.13) (4.14) (4.15)

F = -D dCjdx = constant

.. form of a trigonometrical series is

···.··C. )x 2~C2cosnn-C1 . nnx (D 2 2/12)

. I-+-.t.... sm-exp - n n t

.... '. ... 1 n1 n I

(' nnx'

exp (- Dn2n2t/12) Jo f(x') sin "T' dx'.

(4.16)

where

1 IC'

DJ=C_C DdC,

I 2 C2

.eommon occurrence f(x) is either zero or constant so is readily evaluated. Very often the problem is plane of the sheet, and the formulae are then taken as x = 0 and the surfaces at x = ± I.

and this is the mean value deduced from a measurement of F. It from (4.9) that if D depends on C the concentration no longer linearly on distance through the membrane. Concentration for D depending on C in a number of ways are given in Chapter 9.

Surface concentrations equal

and desorption by a membrane. If the region ... a uniform concentration Co. and the surfaces are ·ntlr"t,·r ", C1, the solution (4.16) becomes

(2n+ l)nx

exp {-D(2n+ 1)2n2t/4f2} cos 21 . (4.17)

of diffusing substance which has entered the

4.2.3. Composite membrane

If we have a composite membrane composed of n sheets of II , 12, .•• , in, and diffusion coefficients D 1 , D z , ... , D n' the fall in concern through the whole membrane is the sum of the falls through the sheets. Since the rate of transfer F is the same across each section,

48


sheet at time t, and Moo the corresponding quantity after infinite time,

Mr _ ~ 8 { 2 2 2

M - 1- c: (2 1)2 2 exp - D(2n+ 1) 11: t/41 }.

00 n=O n+ 11:

The corresponding solutions useful for small times are

C-Co ~ ( l)n s: (2n+1)I-x

-----"- = t... - erIC -'----,::--;-,...:--

C1 - Co n=O 2J(Dt)

~ ( 1)n r: (2n+ 1)I+x

+ t... - erIC ,

n=O 2J(Dt)

and

:t = 2(~:)t{1I:-t+2 f. (-I)nierfc jnl }.

00 n=1 v (Dt)

The modifications to these expressions for Co = 0 or C1 = 0 are 0 Jason and Peters (1973) analyse the bimodal diffusion of water in muscle by combining two expressions of the type (4.18), one for each and each having its own diffusion coefficient.

Eqn (4.18) can be solved graphically for Dt/12 and hence D obtained measured sorption or desorption time curves. An alternative Talbot and Kitchener (1956) is the approximate formula

q rq2

(J = +In x+-+~

p 2p3

where () = 1I:2Dt/412, X = in2Mt/Mro and p = 1+x8+x24, q = ~X8+ and, r = 1 +9x8 + 25x24.

The solution is correct to four significant figures when M] Moo desorption. The solution (4.18) also applies to diffusion along a rod or tube of length I, with one end and its surface sealed and the maintained at a constant concentration.

Talbot and Kitchener also obtained a solution for a slightly .... p .. UJJ!!!

If d 1 and d 2 are the diameters of the two ends, the degree of taper is

by tt = (d2 -d1)/d1, where d1 is the closed end, and may be negative. For small /1, e.g, around 0·01, we have effectively radial and a solution is

M, 6(1 + tt)2 00 IZ; + tt2

- = " exp (- DIZ2t/12)

Moo 3+311+tt2/;'ilZ;(IZ;+fL2+/1) n'

where the IZn are the- roots of tan IZ = -IZ/ fL, and, to the first

a; = (n-t)2n2 +2tt. The solution becomes, with this .

Mr ro 8

M = L: (2 _1)2 2{1+(kn-A()ttexp(-(2n-l)2())},

00 n= 1 n 11:


49

= n2Dt/4f2, A = 8/1£2, kn = I-tA/(2n- nz, i.e.

kl = -0·21585, k2 = 0·86491,

k3 = 0·95137.

x = 11l:2M/M co as before, () must satisfy

x = {I +(k1 -A()tt} e-6 +M 1 +(k2 -AB)tt} e-96 + ls-{1 +(k3 -A()tt} e-2S8+ ....

(4.18) is applied to the tapered tube to calculate (), the apsay (Jo, obtained can be corrected by using

(J = Oo{ 1- /.(1 +q~-s/Oo } = Bo(l-Ktt)

+~k2X8 + /Sk3X24, p and q are defined above, and KP. is the Talbot and Kitchener (1956) quote

" = 0.807 -0·60X8 +(0·212-0·25x8)/Oo

and discuss in more detail the capillary tube method (Ander- 1949) for measuring diffusion coefficients.

expressions (4.17), (4.18), (4.19), (4.20) can be written in

T = Dt/12,

(4.21)

){ = xfl,

for all values of D, I, t, and x can be obtained from values covering these two parameters. Graphs of are shown for various times in Fig. 4.1. These are reof nomenclature from Carslaw and Jaeger's book ... vu ....... "'" values of (C-CO)/(C1-CO) and of Mr/Moo are Values of M/M <Xl have also been tabulated by McKay numerical values for the concentration at the centre of ..... given by Olson and Schulz (1942). The curve labelled in Fig. 4.6 shows how M,/M 00 varies with the square of thickness 2a when the concentration at each surface

nUUlIl[JrI. Surface concentrations different

through a membrane. If one face x = 0 of a memconcentration C1 and the other x = I at C2, and at a uniform concentration Co, there is a finite the steady-state condition previously discussed


0' 0'6 I

~ 0·4

r,J

I ~

x/I FIG. 4.1. Conce~tration distributions a t various times in the sheet -I < x < [with initial form concentration Co and surface concentration C1. Numbers on curves are values of

~n § 4.2 (p. 44) is set up. During this time the concentration changes mg to

C C (C C x 2 ~ C 2 cos nn - C nnx

= 1 + 2 - 1)-1 +- f.... 1 sin ~ exp ( - Dn27t2tl

7tn=1 n I

4Co ~ 1 . (2m + l)nx

+---;-",7"o2m+18m 1 exp{-D(2m+l)2n2tI12}.

As t a~proache8 i?finity the terms involving the exponentials vanish have simplythe hnear concentration distribution obtained in § 4.2. (l?34) examined the errors introduced by assuming a linear 5"'"'''''' exist across the membrane during the whole course of diffusion.

If M, ~eno~es the total amount of diffusing substance which sheet during time t, and Moo the corresponding amount during infinite' then

-

MI 8 co 1

Moo = 1- n2 n~o (2n+1)2 exp {-D(2n+1)2n2tIF}.

In this case M <Xl = I {1< C I + C 2) - Co} and the total content of the


51

t is given by Mt + ICo· The expression (4.23) is similar to (4.18) and evaluated from the curve labelled zero fractional uptake in Fig. 4.6, proviso that in (4.23) ( signifies the whole thickness of the membrane 18) it denotes the half-thickness.

at which the gas or other diffusing substance emerges from unit face x = I of the membrane is given by - D(iJC;ox)x=i which is from (4.22). By integrating then with respect to t, we obtain

amount of diffusing substance Qt which has passed through the in time t.

t 21 ~ C 1 cos nn - C 2 2 2 2

D(C1-C2)-+2i... 2 {l- exp(-Dn n til)}

1 n 1 n

4Col ~ 1 { 2 2 12}

+~2 c: (2 12 1- exp(-D(2m+1) n tl ) .

n m=O m+)

lU .. JU"",,, experimental arrangement both Co and C2 are zero, i.e. is initially at zero concentration and the concentration at the which diffusing substance emerges is maintained effectively at tration. In this case we find

(4.24)

Q, Dt 1 2 I<Xl (-It .r.. 2 2 2

- = ----- ~-exp(-vn n til) IC 1 12 6 n2 1 n2 '

00, approaches the line

(4.24a)

DC ( [2 )

QI = _{-I t - 6D .

(4.25)

UlL'''l,",',",'''. L on the t-axis given by L = [2/6D

(4.26)

(1920), Barrer (1951) has used (4.26) as the basis of a "~"'lU"'15 the diffusion constant, the permeability constant, and . gas by analysing stationary and non-stationary flow through from an observation of the intercept, L, D is deduced by '''t'·''f1lu_.,t''1~P flow rate the permeability constant P is deduced S follows from (4.7). The intercept L is referred to as the

1 as a function DtlF is shown for the case Co = C2 =0 the accuracy of plotting the steady state is achieved

•....... approximately

of solution useful for small times, usually attributed to and Alpert 1954) is easily derived by using Laplace 2.4.3(i). Still keeping Co = C2 = 0, we find the rate of

0·2

r.S ~

cl

0·1

IN A PLANE SHEET

0·0

FIG. 4.2. Approach to steady-state flow through a plane sheet.

0·2 0·3 0·4 Dt//2

0·5

0·6

flow F(t) per unit area of the surface x = I to be

F(t) = - (DaJC) = 2CI f (D)t exp {-(2m+ 1)2Pj(4Dt)}

x x~1 1 nt

This series converges most rapidly for small t. Rogers et al. take only leading term and obtain essentially

In (ttF) = In {2Cl(~r}- :;t'

From the slope and intercept of this line experimental data yield C1, and hence solubility. They discuss the advantages of their compared with the use of the time lag given by eqn (4.26).

Other applications of Holstein's solution are discussed in § 10.6 and § • (pp. 216 and 224). Jenkins, Nelson, and Spirer (1970) examined the general problems of deducing both the diffusion coefficient and the bility coefficient from experimental data and mathematical solutions the outflow volume is finite so that the concentration varies with time outgoing face. The necessary solution is given by Carslaw and Jaeger Jenkins et al. tabulate some useful calculated data and also consider concentrations at the ingoing face. In particular, they examined the tion, frequently made in time-lag measurements of the Daynes


53

is established after a period of about 3 times the time lag. They that the time lag is underestimated by about 4 per cent by making

Dibenedetto (1965) also obtained solutions for finite outflow "v,a"".1\. and Kubin (1967) allowed the concentrations on both sides lllU'.u •• v to vary with time.

surface concentration

to the general problem of diffusion in the region 0 < x < I at concentrations 4JI(t) and 4J2(t) and the initial concentragiven by Carslaw and Jaeger (1959, p. 102). For empirical values and f(x), three integrals arise which have to be evaluated numerically. In certain cases, however, where the surface

can be represented by a mathematical expression, the solution simplified.

of practical interest is that of a sheet in which the concentration and each surface of which approaches an equilibrium con, exponentially, i.e.

(4.28)

a surface concentration which is changed rapidly but not a situation which usually arises when an instantaneous in an experiment. For the sheet whose surfaces are at

00 (-1)" exp (-D(2n+ Wn2t/412) cos (2n+ l)nx, (4.29)

"~O (2n+l){4pP-Dn2(2n+1)2} 21

equal to any of the values D(2n + 1)2n2/412• The sorption. curve showing the total amount M, of diffusing substance ..... function of time t, is obtained by integrating (4.29) with the limits -I and + 1 and is

8 00 exp {-(2n+ 1?n2Dt/412}

- n2 "~O (2n+ 1)2[1-(2n+ If{Dn2/(4PI2))]'

curves for different values of the parameter pt2/D . fly!. When f3 = co, the surface concentration rises in.. '. 0 and the curve of Fig. 4.3 has the characteristic initial

(4.30)


5

.·.·rel¢rr·ed to as the 'quadratic-profile' solution. The constant k is

uu,nHu,,,· differ from unity by more than O(hb). Least-squares or

criteria can again be used but we choose k = 1. Crank and Parker (1 their approximate solutions and compared them with the analytical solution which is

~ 2h cos a,X

V = L.. 2 2) ---b

.=1 h+b(h +a. cos 0:.

{sinhO::,y (2) ~ (-1)'rexp(-r2n1t). } ( ..

x . h + 2n exp -as t L.. 2 2 2 sin my ,

sm as ,=1 r n +a. .

DIFFUSION IN A CYLINDER

a long circular cylinder in which diffusion is everywhere radial. is then a function of radius r and time t only, and the diffusion

where as> (s = 1,2, ..... ) are the positive roots of

a tan ab = h, for a range of values of hand b. They concluded that the ' .... "' •• "U'LII solution which is usually used as an approximation for a thin sheet or in fact, poor for a thin sheet with a high surface evaporation. But it good approximation for a thick sheet with low surface loss. Both the ratic profile' and the 'partial separation of variables' yield much approximations over all. Concentration profiles across the sheet these general conclusions.

As hb -.. 0, the roots of (4.82) approach

ba, ~ ~(hb), ba"+ 1 ~ nn, n ): 1.

Also the constant terms outside the curly brackets in (4.81) tend unity for s = 1 and the other terms either stay of the same order of or decrease, as s increases. It is easy to see that as'hb -+ 0 the tion (4.81) approaches the approximate solution given by (4.73), ( (4.79). Crank and Parker show, however, that the approximate readily be extended to cover some non-linear situations.

ac _ ! ~(rDac)

at - r ar ar'

(5.1)

is a hollow cylinder whose inner and outer radii are a and and if the diffusion coefficient is constant, the equation decondition is

~(rdC) =0

dr dr '

2nDt(C2-C1)

QI = In (bla) .

in a concentration-dependent system, the mean value of "' ...... ticient obtained from (5.5) is (f~~ DdC)/(C2-C1) as for (4.11». The concentration distribution defined by (5.4)

is for the plane sheet. Typical distributions are shown in C2 = 0, b]« = 2, 5, 10.

(5.5)

a < r < b.

(5.2)

C = A+Blnr,

(5.3)

are constants to be determined from the boundary conditions If the surface r = a is kept at a constant concentration C1,

C 1 In (blr) + C21n (ria)

c= .

In (b/a)

(5.4)

.••.•... diffusing substance QI which diffuses through unit length of

. .. t is given by

70

DIFFUSION IN A CYLINDER·


71

0·60

FlO. 5.1. Steady-state concentration distributions through cylinder wall. Numbers on values of bla.

Another steady-state problem leading to an interesting result is the hollow cylinder whose surface r = a is kept at a constant C1, and at r = b there is evaporation into an atmosphere for equilibrium concentration just within the surface is C2. The condition, with the constant of proportionality denoted by h, is

5.2. Effect of thickness of cylinder wall on steady-state rate of flow.

r = b,

(S.6a)

and we find

Cd1 +hb In (blr)} +hbC2 ln (ria)

C= .

1 + hb In (ria)

The outward rate of diffusion per unit length of the cylinder is Ql' hb

ah1 C 1 {I +bh21n (blr)} + bh2C2{ 1 +ah 1 In (ria)}

ah, + bh , + abh, h21n (bla) ,

(S.7a)

2nDtabhlh2(C 2 - C 1)

Qt = ah, + bh2 + abh, h2 In (bla)

on diffusion in regions bounded by surfaces of the cylinsystem and in which the flow is not necessarily radial are and Jaeger (1959, p. 214).

(S.8a)

By differentiating this expression with respect to b, it is easily seen ." ah > 1 the rate of diffusion decreases steadily as b increases from' . ah < 1 the rate first increases and later decreases, passing through a when b = l/h. This is due to the two opposing changes associated increase in b. On the one hand, the rate of evaporation is increased the increase in area of the surface, r = b, as b increases, but on the the gradient of concentration through the cylinder decreases as b is

In certain circumstances, therefore, the rate of diffusion through the pipe may be increased by making the wall thicker (Porter and This is illustrated in Fig. S.2 for ah = }.

the method of separating the variables described see that

C = u exp ( - Da2t)

(S.9)

72


is a solution of(5.1) for D constant provided u is a function ofr only,

d1u 1 du 2

-+--+0:: U = 0,

dr2 r dr


73

(1959, p. 196). Finally the solution satisfying (5.11) and (S.12) is

2 00 J (ro:: ) fa

C = 2 I exp(-Do::;t)~ rf(r)Jo(ro::o) dr.

a n=[ J1(ao::n) a

(S.18)

which is Bessel's equation of order zero. Solutions of (5.10) may be in terms of Bessel functions, suitably chosen so that the initial and '-'V'''U~<<l conditions are satisfied. Thus if the initial concentration distribution fer) and the surface r = a is maintained at zero concentration, a (5.1) is wanted satisfying

C = 0, r = a, t ? 0,

, solutions for both large and smaIl times can be obtained by . Laplace transform.

concentration constant: initial distribution fer)

cylinder of radius a the conditions are

C=Co, r=a, t?O,

C = fer), 0 < r < a, t = o.

The boundary condition (5.11) is satisfied by

00

C = L AnJ o«(;(or) exp (- Do::;t),

n= I

provided the O::nS are roots of

where J o(x) is the Bessel function of the first kind of order zero. (S.14) are tabulated in tables of Bessel functions. For this function finite at r = O. The initial condition (S.12) becomes

00

fer) = L AnJ o(ro::n),

0=1

it being assumed that fer) can be expanded in a series of Bessel order zero. The Ans are determined by multiplying both sides of (5.1 r J O(O::nr) and integrating from 0 to a using the results,

s: rJ o(o::r)J 0(f3r) dr = 0,

when 0:: and fJ are different roots of (5.14), and

where J 1 (x) is the Bessel function of the first order and 0:: is a root of The derivation of the relationships (S.16) and (S.17), and of ""T'rp~·n"T expressions which hold when 0:: is a root not of (5.14) but of equations which commonly arise in diffusion problems, is given by

-

C = fer), 0 < r < a, t = 0,

(5.19)

(5.20)

C = Co{l-~ f 2_ Jo(ro::J eXP(-DO::;t)}

a n= 10::0 J 1 (ao::")

2 00 J o(rO::n) f f() ( ) d

+2 L exp (-Do::;t)-2-- r r JO ro::. r,

a • = 1 J 1 (ao::.)

are the positive roots of (5.14).

is initially uniform throughout the cylinder fer) = C 1

(5.21)

C-C1 _ 1-~ f exp(-Da;t)Jo(ro::n).

CO-c1 - an=. o::.J.(ao::n)

the quantity of diffusing substance whic~ has e~tere~ Of. left in time t and Moo the corresponding quantity after infinite tune,

u, ~ 4 (D 2)

- = 1 - L... """"2'2 exp - O::n t .

Moo n= I a an

~uuU""o solution useful for small times is

at a-r (a-r)(Dta)t. a-r

- - erfc -~+ l lerfc-J

·"'----C~1 - rt 2J(Dt) 4ar' 2 (Dt)

(9a2-7r2-2ar)Dt·2 f' ~+

+ 32afr! 1 erIC 2J(Dt) ... ,

ria is not small. The case of ria small is discussed by (1944). They give a series solution involving modifi~ of order n ±i- The necessary functions are tabulated ill numerical calculation is straightforward.

(5.22)

(S.23)

(S.24)

Also for small times we have



75

:G?L

':)J&I.'"

74

M, 4 (Dt)! Dt 1 (Dt)*

Moo = nt ~ - a2 - 3nt a2 + ....

Clearly these solutions are not as valuable as the corresponding ones the plane sheet. In practice the range in t over which they are convenient evaluation is less than in the plane case.

le surface concentration

concentration in the cylinder is zero and that at the surface is solution is

2D 00 2 anJ o(ri.1.n) I' 2

C = - L exp ( - Dan t) ( ) exp (DanA}$(A) dA,

a n= 1 J 1 aa; 0

(5.26)

.. a.s are the roots of (5.14).

the plane sheet, a case of practical interest is when

$(t) = Co{1- exp (- j3t)},

(5.27)

a surface concentration which approaches a steady value Co, . The solution (5.26) then becomes

(j3r2 j D)t} (R) 213 ~ J o(ran) exp ( - Da;t)

o{(j3aljD)t} exp -,A + aD/;;\ IlnJ1(aan) (1l;-j3jD)' (5.28)

curve is given by

2Jd(j3a2jD)t}exp(-j3t} 4 00 exp(-DIl;t)

(j3aljD)tJo{(j3al/D)t} + al n~l a;{o:;/(j3/D)-1}" (5.29)

uptake curves for different values of the parameter j3al/D.

FIG. 5.3. Concentration distributions at various times with initial concentration C1 and concentration Co. Numbers on curves are values of Dtia".

The solutions for the cylinder can be written in terms ofthe two . less parameters Dt/a2 and ria. Curves showing (C-C1)j(CO-Cl .: function of ria for different values of Dtla? drawn by Carslaw and (1959, p. 200) are reproduced in Fig. 5.3. The curve of Fig. 5.7 for tional uptake shows how Mr/M 00 depends on Dtla2 when the ,",V,,""Jl"U at the surface of the cylinder remains constant.

4

5

sorption curves for surface concentration given by Co(l- e-P'j. Numbers on curves are values of pa2 / D.

Crank 1975 Diffusion

Încărcat de

Informații document

Descriere originală:

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Crank 1975 Diffusion

Încărcat de

Drepturi de autor:

Formate disponibile

THE MATHEMATICS OF DIFFUSION

S-ar putea să vă placă și

Crank 1975 Diffusion

Încărcat de

Informații document

Descriere originală:

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Crank 1975 Diffusion

Încărcat de

Drepturi de autor:

Formate disponibile

﻿THE MATHEMATICS OF DIFFUSION

S-ar putea să vă placă și

THE MATHEMATICS OF DIFFUSION