CHA P T E R

t5

Diffusivityand the Mechanisms

of MassTransport

In Chapter I we beganby statingNewton'slaw of viscosity,and in Chapter

8 we beganwith Fourier'slaw of heat conduction. In this chapterwe present
Fick's law of diffusion,which describesthe movementof one species,say A,
through a binary mixture of I and -B becauseof the concentrationgradient
of A.
The movementof a chemicalspeciesfrom a region of high concentration
to a region of low concentrationcan be observedwith the naked eye by
dropping a small crystal of potassiumpermanganateinto a beaker of waterThe KMnOn beginsto dissolvein the water, and very near the crystal there
is a dark purple concentratedsolution of KMnOn. Becauseof the concenthe KMnOn diffusesaway from the crystal.
tration gradientthat is established,
The progressof the diffusioncan then be followed by observingthe growth of
the purple region dark purple where the KMnOn concentrationis high and
light purple where it is low.
Diffusion is more complicatedthan viscousflow or heatconductionbecause
here for the first time we have to deal with mixtures. In a diffusing mixture
the velocities of the individual speciesare different, and there are several
useful ways of averagingthe velocitiesof the speciesto get a local velocity
for the mixture.' It is necessarytg choose such a local velocity before the
495

Diffu sivity a n d th e M echani sms of Mass Transport

rates of diffusion can be defined. Hence it is appropriatethat in

$16.1we
discuss in some detail the definitions of concentrations,velocities, and
fluxes. Although the discussionin $16.1involvesno new physicalprinciples,
a masteryof thesedefinitionsis essentialfor study of the later topics.
In 916.2 Fick's law of diffusion is stated, and the binary diffisivity gou
is thereby defined. In $16.3 the temperatureand pressuredependenceo1
9uu are discussed. In $16.4 and $16.5the diffusivitiesof binary gas and
liquid mixtures are consideredfrom a molecularpoint of view.
The presentlack of diffusivity data on most mixtures of interestmakes it
necessary
to useestimatedor extrapolatedvaluesof I uo in most calculations.
one should, of course, endeavor to use measured values, since they are
usually (but not always)more reliable.
We have indicated that diffusion of A in a binary system of A and B
occurs becauseof a concentration gradient of l. This phenomenon is
sometimes called ordinary diffusion to distinguish it from pressure clilfusion
(motion of I resulting from a pressuregradient), thermal difusion (motion
of I resulting from a thermal gradient), andforcetr difusion (motion of I
becauseof unequal external forces on A and B). These additional effects,
and also diffusion in multicomponent systems,are discussedbriefly in
Ch a pt er I 8.
Note, ln this and the following chapterswe adopt a fixed rule on notation.
when considering two-componentslstems, we label the speciesA and B.
when dealing with multicomponent
s)istems,we label specificspecies1,2,3,
etc., or in generaldiscussions
we may use a dummy index such as i,7, or ft to
stand for the various speciesin turn. Formulas that are limited to binary
systemsare thus distinguishedby the appearanceof A and B as subscripts.

C oncentrati ons ,

V el oc i ti es , and Ma

In a diffusingmixture the vario

velocities. Let a, denote the ve
stationarycoordinateaxes.2Ther
oDerageuelocity o is definedas

Note that pu is the local rateat wh

placed perpendicularto the veloc
measureby meansof a pitot tubea
chaptersfor pure fluids. Simila
aelocity a* as

l)'

Note that cu* is the local rateat wl

placedperpendicularto the veloci
of theseaveragevelocities.Still ot
such as the t'olumeaterage telocit
In flow systemsone is freque
specieswith respect to u or u*
coordinates. This leadsto the def
at - 0: dif f usionve

V,E LOC | TIE S .

$ t6. 1 DE F |Nt T t ON SO F C ON C E N T R AT T ON S
A ND M A S S F L U X ES 1
In a multicomponentsystemthe concentrationsof the various speciesmay
be expressedin numerousways. we limit our discussionto four: the mass
concentration,
p,, which is the mass of speciesi per unit volume of solution;
the molar concentrarion,
ci: pil Mt, which is the number of moresof species
i per unit volume of solution; the ntass
fraction, at, : pil p, which is the mass
concentrationof speciesi divided by the total mass density of the solution;
and the molefraction, rr: c,fc, which is the molar concentrationof species
i divided by the total molar densityof the solution. (By the word "solution"
we mean a one-phasegaseous,liquid, or solid mixture.) In Table l6.l_l a
summary is given of these concentrationunits and their interrelation for
binary systems.
I R. B. Bird, Theory of
Diffusion, in Aduances in Chemical Engineering, vol. I, Academic
_
Press, New York (1956), pp. 170, et seq.

0t -

a* : diffusion vr

Thesediffusion velocitiesindicate
local motion of the fluid stream.
The notation for velocitiesin
16.l-2, togetherwith someuseful
It i s i mpor t ant t hat t he r eadert l
various kinds of velocities.There
representits behavior pictorially.
when liquid I evaporatesand diff
filled with vapor -8. As I evapo
there is not a clear dividing line b
as sugges t edin Fig. l6. l- la; ins

'?B y "vcl oc i ty " w e do not mean the ,

rather we mean the sum of the velociti
volume element divided by the number <

I
: h e M ech anis m s o f M a s s T r a n s p or t

Co n c entrati ons, V el oci ti es, and Mass Fl uxes

t is appropriatethat in 916.1we
' concentrations,velocities, and
/olvesno new physical principles,
' r s t ud) o f th e l at er t opic s .
l, and the binary diffusivity 2on
ure and pressuredependenceof
.e diffusivities of binary gas and
cular point of view.
rostmixturesof interestmakes it
iluesof 9uu in most calculations.
neasuredvalues, since they are

In a diffusingmixture the various chemicalspeciesare moving at different

velocities. Let a, denote the velocity of the ith species with respect to
stationarycoordinateaxes.2Then, for a mixture of n species,the local nrass
at)erageuelocity o is defined as

in a binary system of A and B

'n t o l A. Th i s phenom enon i s
nguish it from pressure difusion
.dient),thermal difusion (motion
nd forced difusion (motion of ,4
nd B). These additional effects,
'stems,are discussedbriefly in
v ea dopta fi xe dr ule on not at io n .
we label the speciesA and B.
, w e lab e lsp e c if ics pec ies1, 2, 3 ,
dummyindex such as i,7, or k to
mulasthat are limited to binary
rranceof A and B as subscriots.

2 p,,,
n

\s^

a t i o nsof th e var iouss pec iesm a y

rur discussionto four: the mass
es i per unit volume of solution;
s the number of moles of species
ion,ao : prlp, which is the mass
tal massdensityof the solution;
e molar concentrationof species
rlution. (By the word "solution"
rlid mixture.) In Table 16.1 I a
rnits and their interrelation for
lhemical Engineering, Yol. I, Academic

(16.1-1)

i=l

Note that pu is the local rate at which masspassesthrough a unit crosssection

placedperpendicularto the velocity u. This is the local velocity one would
measureby meansof a pitot tube and correspondsto u asusedin the preceding
chapters for pure fluids. Similarly, one may define a local molor auerage
uelocity u* as

2r ru

(16.r-2)

T" .
i-l

Note that cu* is the local rate at which molespassthrough a unit crosssection
placedperpendicularto the velocity u*. we shall find extensiveuse for both
of theseaveragevelocities. Still other averagevelocitiesare sometimesused,
such as the rolume areroget'elocityur. (SeeProblem 16.K.)
In flow systemsone is frequently interestedin the velocity of a given
s p c c i esw i th respectto u or u* rather than w i th respect to st at ionar y
coordinates. This leadsto the definition of the "diffusion velocities":
at - o:

O N S , V EL OCIT I E S ,

4gl

ui -

o* :

di ffusi onvel oci tyof i w i th respectto o

( 16. 1- 3)

diffusion velocity of i with respectto u*

(16.1-4)

Thesediffusion velocitiesindicatethe motion of componenti relativeto the

local motion of the fluid stream.
The notation for velocitiesin binary systemsis summarized in Table
16.l-2, togetherwith some useful relationsamong the various velocities.
It is important that the readerthoroughly understandthe meaning of the
various kinds of velocities. Therefore,let us cxamine a specificsystemand
representits behavior pictorially. In Fig. l6.l-l we considerwhat happens
when liquid I evaporatesand diffusesupward through a long tube initially
filled with vapor B. As I evaporates,it pushesgas B upward. However,
there is not a clear dividing line betweenpure ,4 vapor and pure ,B vapor,
as suggestedin Fig. 16.l-la; instead, the upward displacementof B is
'?By "velocity" we do not mean the velocity of an individual molecule of species i;
rather we mcan the sum of the velocities of the molecules of species i within a small
volume element divided by the number of such molecules.

TABLE I 6./ - I

P : pA + p./r :
PA :

mass density ol solution (g/cm3)

r.\

NOTATION FO R CO NCEI I TRATI O NS

I N B I N A R YS Y S T E M S
(A)

c tM s, : mass concentrationof A (g of Alcm3 of solution)

(B)

o.

:U

@1

: mass fraction of I

It
3a

3t

(C)

a\

ll

P
a

c : cA * cu : molar density of solution (g-moles/cm3)

(D)

J!-a-!

,l l l l
E

,"

Pi

molar concentrationof I (g-molesof Af cm} of solution)

io:

mo lef r ac t ionof I

'o:;:
At -

p
'

(E)

*i

!<i

5'n\

(F)

r0t

:u

: number-mean molecular weight of mixture

(G)

tn
o

r_ 4+rB:I

( H)

rr1 M1

xB MR:

( ua

(r)

z
a

(1)

oalaS:l
an

:------+-:1
M;
Mrt
M

i4

(,()

ra : ---A !

-:+j
M,
M,

r,M,

(L)

(,th

@A:

rnMn

* r3M j

o.

z:

()3

ad

rl
5

!4

.. x

U7

M^

Ja
co

(M)

'!
a

. ^'t
-,

. -l

-a
l rl l

::J

't
a'\-

td.
J

M,B
da,

u*l
*^*"(ft.

( N)

do,^:#:#r"

E
o
z

(o)

\
Oa
9O
ex

3
<
o-

>.

qd
a-

TABLE I 6. I - 2
NOTATION FOR VELOCITIES
IN BINARYSYSTEMS

z,/ : velocity of speciesA relative to stationary coordinates

Da - u : diffusion velocity of speciesI relative to u
uA - u* : diffusion velocity of speciesA relative to u*

(A)
\

(B)

u* : oa(ua - D*) + an(us - a*)

u* - ,; : ro(un - u) + ro@o - u)

(C)

(D
(C)

.9

o^q
^alaE
qi

\d

.96

i,

o;

Y=

v:

o P x1 I,i

I
E
E
I
I
I
t

498

oooa
'6
9, : . P+

: mass averagevelocity : (llilGdU

ts -

<V

JE'z'

* paon) -- @AuA* apas (D)

u' : molar averagevelocity : (1lc)(cnua * caus) : nAoA ! tuos (E)
U

av
lttl

=a

fi
[:
(
ri

?Id

9tovdi
yo2ho=oo

F: P

;>
sqorlrug ?p JrsPfl

---

i-l
v a{.J

) NS IN B I NA RY S Y S T E M S

rn (g/cm3)

.^_Qs

\U

'

>l>

(A)

(g of Alcms of solution)

(B)

o
o

o
qa

(c)

*1

(D)

(E)

ll

*i

ll
+*

il

.-i

5'5\

i l tl
l<

v.e

ts,E

^F \

:U

Ft

ua

__

@t:

,tMt

* rrtMn

*a

ll

iz
!1?

(K)

,:\

z
a

Ur

.r ,M,

(1)

(up
+_ _ _ a
Mt 'M B - M
j

'\

-l-

a
q

a1 * a o : l

'\

< t<

(c)

frt of mixture

\a

Els l'.

:,++

*i

''1
' \<

z
i

-'

Xo
l tl l
.*

ll

.V

a-

Itrl
l
I
g-molesof Af cm3 of solution)

al

3-

ion (g-moles/cm3)

r q- - -

sv
1t q

'q

-f
,-,

T1
q

.l

.{\
++

":

llll

(M)

-!'l

oe
l l rl

tl

II

.i

-'l

P-J
al:
-{
' {-ll > > t<

llll
'!i

. -l

o
:
o
z

M, M -dr,
.
" -A - k AMA + xB lt4u y
d, , ,

\
A

T\

!r<

O6

-X
8<

^a

o. >c,l <d

llll

s '3

2
N BINARYSYSTEMS
:o stationarycoordinaies
relativeto u
relativeto o*

jz

l ttl

76

!f

e'z

itr'Z'
(A)
(B)

-^

(c)

.9

\i

'-ooaw) 9 , l 6
^ a a C^ q

Gi

'aua * psos) : @AuA* apuo (D)

cAaA + caun) : roua * xpop (E)

o'P xr

.98

R+

Y:

ir

;.;

":

5-q

.:d
F: P
Ytr 4- c dT

y ( J 2baEd,o

(n
(c)

t
i

9-:-:s qol l !ug op c l s r l T

'-

6AY

IJ ^.F + r5

qa6A

r8 6 5

E.i P r

iq

( !GL G
d^

o50E
tr aotr uo
5-5A(n

Es
o tr

'oEt"
<

E
o

cE

cP
'- :'- ;- cd

Er

6 6 o "a l

q
X
J

-LLI!IJ.II.

X >,
=<. -

c E

o' X
J.r'

El
o
3+

{lrro cJuJJiJirJJ(U 'sixnU orll Fuotutssuollsl3u

4ln

X.-I
=-

=I
and the Mechanisms

Diffusivity

500

of Mass Transport

accompaniedby an intermingling of the two vapors, as shown in Fig. 16.1-lb.

Thus at any point in the column I is moving upward more rapidly than the
over-all averagemotion and B is moving less rapidly, becauseof diffusion.
In Fig. l6.l-lb the various velocity vectors are shown for the situation that
The readershoul dveri fy
ra : t , u*: 12, oa - o * :3 , a n d M ,t:5 M s .
vectors.
the other velocity

dii
o
ll
I
a
I
s

.\'
tl

C oncentrati ons ,

V el oc i ti es , and Mas

and the mass and molar fluxesre

ir* : pr(a
Jn* : cr(u

The notation for fluxes in binary

together with some relations an
emphasizedthat the definitionof a
and referenceframe have been give
Mathematically speaking,any c
adequate for all diffusion problem
have appeared in the literature.
has been us_edin engineeringbec
desirableto refer to a coordinates
Jland, Jr* are the usual mdtsures(
lating the equations of changef<
Jo andj r* are seldom usedbut are i
We use N, in most subsequent
dis

Exam ple l6. l- 1. Relat io

a. How are the fluxesJ;* and N; r

b. Show that the sum of the fluxes
Solution. rz. First we combineth
l0) to get

Velocityvectors

Jr* : ct(a

Fig, l6.l-1.

I ncorrect

Correct

(a)

(b)

: ci u t

C o n c en tr a tio n d istr ib u tio n a t a sPe cifictime i n a di ffusi ngsystem.

Now that concentrations and velocities have been discussed,we are in a

position to define mass and molar fluxes. The mass(or molar) flux of species
i is a vector quantity denoting the mass (or moles) of speciesi that passes
through a unit area per unit time. The motion may be referred to stationary
coordinates,to the local massaveragevelocity u, or to the local molar average
velocity u*. Thus the mass and molar fluxes relatiue to stationary coordinates
are
fli:

Ptot

mass

(16.1-5)

Nn:

cPi

molar

(16.1-6)

The mass and molar fluxes relatioe to the m(tss-auerogeoelocity o ate

i t:

p n (u n - a )

J t:c t(a i -u )

mass

(16.1-:1)

molar

(16.1-8)

Application of the definitions of N,. ar

Ji * :r

From this result it can be seenthat the

ence between the molar flux N, and
species i due to the local molar flux
similarly interpreted.
D. Summation of Eq. 16.1 12 from

,i

which shows that the sum of the nolar

vglocity is zqro- gr any mixture. Thus
and JB* are of equal magnitude and a
J.t*

An application of this result is given ir

r e Mech anis m s o f M a s s T r a n s p o rt

vapors,as shownin Fig. 16.l-lb.

Lgupward more rapidly than the
rssrapidly, becauseof diffusion.
are shown for the situation that
5Mu. The reader should verify

Co ncentrati ons,

Y el oci ti es, and Mass Fl uxes

S OI

and the mass and molar fluxes relatiue to the molar-auerapeuelocitv u* are
jo* - pr(uo- a*)
mass
(16.1-e)
Jo* :

cr(uo- u*)
molar
(16.1-10)
The notation for fluxes in binary systemsis summarizedin Table l6.l-3,
together with some relations among the various fluxes. It should be
emphasizedthat the definition of a massflux is not completeuntil both unirs
and referenceframe have been given.
Mathematically speaking, any one of the foregoing six flux notations is
adequatefor all diffusion problems,but each has certain advantagesand all
have appearedin the literature. The flux N, (and to a resserextent n,)
h a s been used i n engi neeri ngbecausei n processcai cul ati onsit is usually
desirableto refer to a coordinatesystemfixed in the equipment. The fluxes
frand' Jr* are the usual measuresof diffusion ratesand aie usefulin formulating the equations of change for multicomponent systems. The fluxes
J, andj r* are seldomusedbut are includedherefor the sakeof completeness.
We use N, in most subsequentdiscussions.
Example l6.l-1. Relations Among

the Molar

Fluxes

a. How are the fluxes Jo* and N, related in an n-component svstem?

b. Show that the sum of the fluxes Jr* is zero.
Solution. a. First we combine the definitionsof u* and Jr* (Eqs. l6.l_2 and
l0) to get
Ji *
Correct

ct(ur-

u*)

: ciui- '1 t rrr,

c j Zt'

(b)
e cif icti me in a dif f u s i n gs y s t e m .

Application

of the definitions of No and

-^,

( 16. 1- l)
l

then gives
,L

rve been discussed,we are in a

e mass(or molar) flux of species
moles) of speciesi that passes
)n may be referred td stationary
/ t),or to the local molar average
relatiue to stationary coordinates

Ji *:N ;-"i IN i

(r6.1-12)
j'

From this result it can be seenthat the molar diffusion flux Jo* is simply the difference between the molar flux N, and the rate of bodily transport rtult flow) of
speciesi due to the local molar flux of the mixture. Equation 16.1-10 mav be
similarly interpreted.
b. Summation of Eq. 16.l-12 from i : I to i : /, grves
1l

rss

(16.1-5)

I"r,*: o

rlar

(16.1-6)

which showsthat the sum of the molar diffusionfluxesrqlgtiyg_te$e ,rlq qypr-?g"

4l
velocityis zqro i4 any mixture. Thus in a binary mixture the diffusioniuxes Jr*
and JB* are of equalmagnitudeand are oppositelydirected:

,ss-auerage
uelocity u are

mass

(16.r-7)

molar

(16.1-8)

Jo* : -1n*
An application of this result is given in Problem 16.I.

(r6.1-1
3)

( 16. 1- 14)

502

Diffu sivity a n d th e M e chani sms of Mass Transport

0 1 6. 2 F lc K ' S LA W OF D IF F U SION
In E q. l. l- 2 t he v i s c o s i typ i s d e fi n e da s th e p roporti onal i tyfactorbctrvcen
mo m ent um f lux an d v e l o c i tyg ra d i e n t(N e w to n ' s l aw of vi scosi ty). In E q.
8 .1 - 6 t he r her m al c o n d u c ti v i tyk i s d e fi n c d a s the proporti onal i tyfactor
g ra d i en t (F o u ri e r' sl aw of heatconducti on).
b e tweenheatf lux an d te m p e ra tu re
T ABL E I6 .2 _ I
E Q U I V A L E N T F ORM S OF F ICK' S F IRST L AW OF B IN A R Y D IFFU S ION
Flux

Y at,t

nA
N/

Y",t

it

Ya tn

J,t*

Y' t

j t

vr_l

J,t*

c\un -

F o r m of Fi ck's Fi rst Law

G r a d ie n t

ll3)

n ., N,, ia

J,,t* :

i::

o t,( n .,

n73) :

r _ ,( N., + N 1)
- p !/

-p!1 4rrV ot .,
:

-c?,r,,Y :.,

(A )

(D)

- c? ,tttVa :^

which is the molar flux of I rt

and the vector 4* : -clJtn\
from the diffusion superimpos
bul k fl ow and dif f usiont er m si
speciesI (becauseI is diffus
,B is diffusin
spccies,B (becar.rse
The units of the massdiffusi
the ki nem at icviscosit yr andt he
The way in which thesethreeqr
following equations for the fl
:
onc-dimensionalsystems

(B)

' (c )

a sVo ,,

Fi ck's Law of D i ffus i on

i, , u:

- 9. , o*
dq

O,rl

,l

r,,,:

-v

u
-

(pt' r)

aa

lc2\

-\;)r'r,''.r/,V'.r

(E)
4u :

/p2\

Vr,tn

J .rr:

-\* .fl r,,l

Vr',1

c (u, { -

ut t l :

t.,,,Y " t.1

c 'J ,,,
f,

1 'V t ;

(F)
(G)

-{t ,li

!y't.r in a binary systcm in an

Now we define the mass diffusiviry 9,tr:
a n a logousf as hiont
(16.r l )
,,,* : -c ,/,r,,v 1 .,
written in terms of the molar diflusion
This is Fick'sfrst law of tlilfu.sion,r
fl u x J , r *. T his equ a ti o ns ta te sth a t s p e c i e s;1 d i ffuses(movesrel ati vcto the
m o l e fra c ti ol tof A ,j ust as heatfl ow s
mi xtur e) in t he dir e c ti o no f d e c re a s i n g
g mperature.
b y conduc t ionin t h e d i re c ti o no f d e c re a s i n te
A num ber of ot h e r ma th e ma ti c a l l ye q u i v a l e ntstatementsof Fi ck' s fi rst
i n th e l i te ra tu re ,a n d s o me of thcm are summari zcdi n
l a w hav e- appc ar e d
Tablc 16.2-1 for referenceonly. The diffusivity .?jB is identicul in all
th e s eequat ions .O f s p e c i a il mp o rta n c ei n th e fo l low i ngchaptersi s thc form
of Fick's first law in terms of N,r, the molar flux relative to stationarycoo rd inat es :
(16.2 2)
N,r : /,r(N; * Nrr) - c(/.r,,Vr.,
Th i s equat ion s ho w s th a t th e d i ffu s i o n fl u x N ., rel ati veto stati onarycoordinates is the resultafil of two L-ectorquanlities: the vcctor ;r,r(N-, * Nn),
I T e m p e r a t u r e g r a d ie n ts, p r e ssu r e g r a d ie n ts, a n d e xte r nal forces al so contri bute to the
dif f u s i o n f l u x , a l t h o u g h th e ir e ffe ctsa r e u su a lly m in o r . M ore compl ete expressi onsfor thc
dif f usi o nf l u x a r eg i v e ni n $ 1 8 .4[A. F ick, An n , d e r Ph ysik,9 4 , 5 9-86 (l l ]-55).1

- 0";

( PC, T)

4A

Thcsccquat ionsst at e.r espect

a gradicntin m assconcent r at
a gradient in m om ent umcon
becauseof a gradientin energy
in two- and three-dimension
quantit y wit h nine com ponen
component s.
In Tables 16. 2- 2, 3, and 4 s

D IFFU
EXPER IM EN TAL
Gas Pair

co2-N2o
co2-co
co2-Nr

Ar-O,
H2-SF6
Hr-CH1

a This table is abstractedfrom

Molecular Theory of Gasesand Lit,
given are for 1 atm pressure.

the Mec hanis ms o f M a s s T r a n s po r t

t h e pr o p o rti o n al
it y f ac t orbet w c e n
le w t on' sl a w o l v is c os it y ) . I n Eq .
n e d as th e p ro por t ionalit yf ac to r
( F ourier'sl a w o f heatc onduc ti o n ).
- LAW O F BI NA R Y D I F F U S I O N
Fo rm of Fic k ' s F i r s t L a w
, ( .nt *

n 1) :

-p !

, ( N , r + N 1) :

-1yVtu 1

-c!1 ,, 1,Y.r,,

(.4)

- c9 nrrV t : . ,

( D)

i .r,:

-f| ,tn4

'rar:

d
-v ; (pr,)
aa

t .2\

,,t ,, ,, V .r .,

lt[^M

l 12

l:^
-,
l L ,,,,Y ' u. ,
\,4t v r, | t v r I

(E)

c'J , ,,

' " ' V .,

a',t''rt

( F)

(G)

: 9, . r i n a b inar y s y s t c ntin a n
( 16. 2 l )
,V r -,
n i n t erms o f th c m olar dif f Lrs i o n
; I d if f u se s(mov esr elat iv et o th c
le lr ac t i o no l ,1 ,jus t as heatf lo w s
1 gt emPe ra tu rc .
riv a lentsta te m ent sof F ic k ' s fi rs t
s om cof th e m a r e s r r m m ar iz e di n
iffusivity.(r,,,, is ilentical in all
the f ollo wi n gch apt c r sis t hc f o rm
rlar f lux re l a ti vct o s t at ionar yco - c 9. r, ,Vr,,

O.rl
dy ' "

4 u : -u rt f c C" rl

ll

u) :

503

which is the molar flux of I resulting from the bulk motion of the fluid,
and the vector Jr* : -c9,ttYr,,
which is the molar flux of I resulting
fr om the di ffusi onsuperi mposed
on the bul k fl ow . Thus i n Fig. 16. 1- 1bt he
bulk flow and diffusion terms in Eq. 16.2-2 are in the same direction for
species,4 (because,4 is diffusing with the current) and are opposed for
species,B (becauseB is diffusingagainstthc current).
The units of the massdiffusivity9,r, arecm2sec I or ft2hr 1. Note that
the ki nemati cvi scosi tyl and the thermaldi ffusi vi tya al sohavethe sam eunit s.
The way in which thesethree quantitiesare analogouscan be seenfrom the
following equations for the fluxes of mass, momentum, and energy in
systems
:
o ne-di mensi onal

(B)

' (c)

? , , , Yu,.

F ick's Law of D i ffusi on

( 16. 2 -2 )

u x N-, re l a ti vct o s t at ionar yc o ntities: the vector L(N_.r + Nr),

d exte rnal f orc es a l s o c o n t r i b u t e t o th e
n o r . More c omplc t e e x p r e s s i o n sf o r th c
: i k, 9 4,59 86(1855 ) . 1

(Fi ck' s l aw l for constantp)

(Newton's law for constantp)

( 16. 2- 3)
(16.24)

(F o u rie r'laswf o r c o n s t an
t ) O e. z - s )
pe

Theseequationsstate,respectively,
that (a) masstransport occursbecauseof
a gradientin massconcentration,(6) momentum transport occursbecauseof
a gradient in momentum concentration,and (c) cnergy transport occurs
bccauseof a gradi enti n energyconcentrati on.Theseanal ogiesdo not apply
in two- and three-dimensionalproblems, however, becauser is a tensor
quantity with nine components,whercasj , and q are vectors with three
components.
In Tables 16.2-2,3, and 4 some valuesof f)u, are given for a few gas,
TABLE I6.2-2
E X P E R I M E N T ADLI F F U S I V I T I E
OSF S O M ED I L U T EG A S PAIR Sa
g
Temperature
tn
('
(cm2
K)
Gas Pair
sec-r)

co2-N2o
co2-co
co2-Nr

Ar-O,
H2-SF6
H2-CH1

273.2
273.2
273.2
288.2
298.2
293.2
298.2
298.2

0.096
0 . 139
0 . 144
0 . 158
0 . 165
0.20
0.420
0.726

a This tableis abstracted

from J. O. Hirschfelder,
C. F. Curtiss,and R. B. Bird,
MolecularTheoryof GasesandLiquids,Wiley, New York (1954),p. 579. The values
givenare for I atm pressure.

Diffusivity and the Mechanisms of Mass Transport

Temperature

and Pressure Depet

TAELE t6.2-3

DI
EXPERIMENTAL

IN THE LIQUID STATEN

EXPERIMENTAL
DIFFUSIVITIES

B
Chlorobenzene

Bromobenzene

r('c)
10.01

39.97

Ethanol

Water

Water

n-Butanol

9ou x lO5
(cmzsec-r)

25

30

0.0332
0.2u2
o.st22
0.t617
0.9652

1.007
1.069
1.146
1.226
1.291

0.0332
0.2@2
0.5122
0.7617
0.9652

1.584
1.691
1.806
1.902
1.996

0.05
0.275
0.50
0.70
0.95

I .13
0.41
0.90
1.40
2.20

0.13I
0.222
0.358
0.454
0.524

| .24
0.920
0.560
0.437
0.267

a This tableis abstractedfrom a reviewarticleby P. A. Johnsonand A. L. Babb,

Chem. Reus.,56,387-453(1956); in this
Liquid Diffusionin Non-Electrolytes,
for liquid systemsis given,
diffusioncoefficients
articlea summaryof experimentaf
as well as a surveyof methodsof measurement.Another excellentreviewarticle
and Interpretationof Diffusion Coefficients
is that of L. J. Gosting,Measurement
of Proteins,Aduancesin Protein Chentistry,Vol. XI, AcademicPress,New York
( l 956).
liquid, and solid systems. Diffusivities of gasesat low density are almost
composition independent,increasewith the temperature,and vary inversely
with pressure. Liquid and solid diffusivities are strongly concentrationdependentand generallyincreasewith temperature. In the next three sections
we summarize the available means for estimating diffusivities'
EEP E N D E N C E
E D PR ES SU RD
s 16. 3 T E M P E R AT U RAN
O F M A S S D IF F U SIV IT Y
The massdiffusivityguBfor a binary systemis a function of temPerature,
pressure,and composition, whereasthe viscosity p and thermal conductivity

System

He in SiO,
He in pyrex
H, in SiO2
H, in Ni

Bi in Pb
Hg in Pb
Sb in Ag
A1 in Cu
Cd in Cu

'Values taken from R. M. Ban

New York (1941),pp. 141,222,an

k for a pure fluid are functions

available on I ,s for most bine
range and accuracy. The availa
and are based more on theory tl
For binary gas mixtures at lo'
the pressure,increaseswith incre
of composition for a givengas
of 9oo at low pressureJhas bee
theory and corresponding-state

POet

(p"np"u)%(T"eT,tf/'

cm2sec-l,p I
in which 9uol:f
mental data gave the following'
For nonpolar gas-pairs:
;
For HrO with a nonpolargas:

I J. C. Slatteryand R. B. Bird, A.I.

h e Me ch anis m s o f M a s s T r a n s p o r t

Temperature

and Pressure

Dependence

of Mass Diffusivity

TABLE I6.2-4

-3
{' THE LIQUID STATEA

IN THESOLIDSTATE"
DIFFUSIVITIES
EXPERIMENTAL

gnu
ilA

x IOE
(cm2 sec-r)

0.0332
0.2642
0.s122
0.7617
0.9652

1.007
1.069
t . 146
1.226
1.291

0.0332
0.2&2
0.5122
0.7617
0.9652

1. 584
1. 691
1. 806
1.902
1. 996

0.05
0.27s
0.50
0.70
0.95
0 .1 3 1
o.222
0 .3 5 8
0.454
0.524

1. r 3

0.4r
0.90
1. 40
2.20
1. 24
0.920
0.560
0.437
0.267

;leby P. A. Johnsonand A. L. Babb,

Reus.,56, 387453 (1956); in this
is given,
for liquid systems
oefficients
nt. Another excellentreview article
telpretationof Diffusion Coefficients
/ol. XI. AcademicPress.New York

I sasesat low densitv are almost

le"temperature,and vary inversely
rit ies are stro ngly c onc ent r a ti o n lerature. In the next three sections
imatingdiffusivities.
D E P EN DEN CE
/stemis a function of temperature,
scositypr and thermal conductivity

Diffusivity,9nu
(cm2sec-l)

r('c)

System
He in SiO,
He in pyrex

20
20
500

2 . 4 - 5 . 5 x 10 - 1 0
4 . 5 x l 0 11
2 x 10-8

H, in SiO,
H, in Ni

500
85
165

x1 0 8
0.6-2.1
1.16 x 10 8
10.5 x l0-8

20
20
20
20
20

Bi in Pb
Hg in Pb
Sb in Ag
Al in Cu
Cd in Cu

l.l
2.5
3.5
1.3
2.7

x
x
x
x
x

l 0 rG
l 0 15
10-21
10-30
10-15

aValuestaken from R. M. Barrer, Difusion in and throughSolids,Macmillan,

New York (1941),pp. l4l, 222, and 275.
k for a pure fluid are functions only of temperature and pressure. The data
available on 9AB for most binary mixtures are, moreover, quite limited in
range and accuracy. The available correlations of 9uo are of limited scope
and are based more on theory than on experiment.
For binary gas mixtures at low pressure,9uo is inversely proportional to
the pressure,increaseswith increasingtemperature,and is almost independent
of composition for a given gas-pair. The following equation for estimation
of 9oo at low pressureshas been developedl from a combination of kinetic
theory and corresponding-statesarguments:

PQar:

(r n)%,
(p,np o)%
"
".q.r
" (h.

" ( J-I-T_T;)'

(r6.3-r)

i,)"

in which 9,to[:] cm2 sec-l, I [:] atm, and 7 [:] " K. Analysis of experimental data gave the following values of the constants a and b:
F or nonpolar gas-pairs:
a :2.745 x 10-4
b : 1.823
For HrO with a nonpolar gas:
a:3.640 x 10-a
b :2.334
1J. C. SlatteryandR. B. Bird,A.I.Ch.E.Journal,4,137-142
(1958).

506

Diffu sivity

a n d th e M e chani sms of Mass Transport

Equation 16.3-1 agrees with experimentaldata at atmospheric pressure

within about 8 per cent.r,2 If gasesA and B are nonpolar and their LennardJonesparametersare available,the kinetic theory method given in $16.4will
usually give better accuracy.
At high pressures9uo no longer varies inverselywith the pressure. Very
little is known about this pressuredependence,except in the limiting caseof
self-diffusion (interdiffusion of identical molecules), which can be closely

E x a m p l e 1 6 .3 - 1 . Esti m a ti o r

Estimate 2 nu for the systemarg

Solution.
The propertiesneed

I (argon)
B (oxygen)
(p"eprn)'A: (4

Therefore

1.0

- 7 " =, r.0-

(T"AT,i%' : (l:

b--

0.9

G--

0.8

| \tA

\r ^ . *)

0.7
pD

Temperature and Pressure Depe

06

tr\b
a\:l

\l

: 2.

T"A TrR l

: 9.

fpof o.s

Substitutionof thesenumericalva

0.4

0.3
0.2

0. 0) aaB: ( 9. 0
9 ta : o'18
The measuredvalue (Table16.2-2

1.o

0 .1
0
0I

: (/ ,

methodbasedon kinetictheory.
0.15 0.2

0.3 0.4 0.5 0.6 0.8 1.0

p,=

i.5

2.0

3.0 4.0

p/pc

Fig. 1 6 . 3 - 1 . G e n e r a l ize dch a r t fo r se lf- d iffu sivitie o

s f d e n segases[J. C . S l attery, M.S .thesi s,
Univ e r s i t y o f W i s c o n s in ( 1 9 5 5 ) ]; th is ch a r t is b a se do n ve ry few data.

approachedexperimentallyby usingisotopictracers.The pressuredependence

of the self-diffusivity,?nt, may be estimated from Fig. 16.3-l; this
plot is based on the Enskog kinetic theory of dense gasesand on some
fragmentarydata.r The ordinate of the plot is p9"nl@?ur)o, the ratioof
the pressure-diffusivity product at pressure p and temperature 7 to the
pressure-diffusivityproduct at the same temperature but at low pressure.
This ratio is plotted as a function of the reducedtemperaturaT,: TfT"and
the reduced pressurep,: plp"- The chart is analogousto Figs. 1.3-2 and
8.2-2; its use is illustratedbelow.
In the absenceof other information, it has been suggested
that Fig. 16.3-l
may be used to estimateI np if p and T are replacedby the pseudocritical
"
"
valuesp"' andT"' givenin Eqs. 1.3-3
and 4. Very few dataareavailableto
test this procedure,and thereforeit has to be regardedas provisional.
'zDetailed comparisons of this and other estimation methods with experimental data
are given by R. C. Reid and T. K. Sherwood, The Properties of Gases and Liquids, McGrawHill, N e w Y o r k ( 1 9 5 8 ) , p p .2 6 7 - 2 7 6 .

E x a m p l e 1 6 .3 - 2 . Esti m a ti o t

Estimate g AB for a mixture of I

ethane at 200Opsia and 104"F (li
(pQ ns)o at 293" K is 0.163 atm cn
Solution. First correct the give
using Eq. 16.3-1 with the nonpola
(Pe 7)o at

"

The critical properties from Table

I, : 190
T, : 305'

methane:
ethane:

The pseudocritical properties acco

n
t' :\i
rc

uri P c t:'

,_|
n

r,-S -r
t;

i :t

*; t

ri

....l.-F

the Mechanisms

of Mass Transport

tal data at atmospheric pressure

.Bare nonpolar and their Lennard) theory method given in $16.4will

Temperature and Pressure Dependenceof Mass Diffusivity

507

E x a m p l e 1 6 . 3 - 1 . E s t i m a t i o n o f M a s s D i f f u s i v i t y a t L o w D e n si ty
Estimate g oo for the system argon-oxygen at 293.2' K and I atm total pressure.
Solution. The propertiesneededfor Eq. 16.3-l (seeTable B-l) are

i inverselywith the pressure. Very

ence,exceptin the limiting case of
molecules),which can be closely

MTCPC
I (argon)
B (oxygen)
Therefore

ln_

151.2
154.4

39.94
32.00

(p"eprn)% : (48.0 x 49.7)%:

48.0
49.7

13.36

1 T " o T " p ) %,: ( 1 5 1 . 2 x 1 5 4 . 4 ) %z: 6 6 . 0 8

h-

=D--

I \' ,
lt
l \rr
/ 1
:l _* _l :0.2372
l _+ _l
' 32.00/
M nl
\M,t
\39.e4
293.2
/
T
\,
/
\r.82 3
x l 0 {l :l
al -:l :2.745
x154'
4/
\!T" tT,n/
\V l 5l .2
: 9.01 x lO-a

1.4

=q
t.2

Substitution
of thesenumericalvaluesinto Eq. 16.3-l gives

(1.0)eau : (9.01x l 0-4X 13.36X 66.08)(0.23 72)

g
: 0' l 89 cm2sec-l
'q'a
The measuredvalue(Table 16.2-2)is 0.20 cm2sec-l. See Example 16.4-l for a

t.n

methodbasedon kinetictheorv.

r 0.8 1.0

1. 5

3.0 4.0

tp

^c

r o f de nsegas es[ J . C . S l a t t e r y , M . S . t h e sis,

d on very few data.

ctracers. The pressuredependence

stimated from Fig. 16.3-1; this
cry of dense gases and on some
tlot is p?uol(p9uo)r, the ratio of
ure p and temperature I to the
temperature but at iow pressure.
:ducedtemperature Z, : TlT" and
rt is analogousto Figs. 1.3-2 and,
rasbeensuggested
that Fig. 16.3-l
are replacedby the pseudocritical
4. Very few data are available to
be regardedas provisional.
nationmethodswith experimentaldata
of Gases
andLiquids,McGraw'roperties

Example 16.3-2. Estimation

o f M a s s D i f f u s i v i t y a t H i g h D e n si ty

Estimate g n" for a mixture of 80 mole per cent methane and 20 mole per cent
ethane at 200o psia and 104" F (136 atm and 313" K). The experimentalvalue of
(pQ ap)o at 293' K is 0.163 atm cm2 sec-r.
Sof ution. First correct the given value of (p97)o to the desired temperature,
using Eq. 16.3-1 with the nonpolar constants:
/313\ 1.823

(pea)o at 313"K : 0.163t2e31

: 0.184 atm cm2 sec-r

The critical properties from Table B-l in the Appendix are

T" : 190.7"K,
r, : 305.4'K,

methane:
ethane:

p, : 45.8atm
p" : 48.2atm

The pseudocritical propertiesaccordingto Eqs. 1.3-4 and 5 are

T

p; :> ni pc;:

(0.80x 45.8)+ (0.20x 48.2)

t:I

: 46.3atm
n

Tc' :> ri Tci :

i :1

(0.80x 190.7)+ (0.20x 305.4)

:213.6" K

Diffu sivity a n d th e M e chani sms of Mass Transport

a'

ztts:

r'47

From Fig. 16.3-l at thesereducedconditionswe obtainpgtnl(poeio

Then
(p e a )o p e n o
0 .1 84.-_-

e,tn:

o f Ordi nary

D i frus i on i n t

The moleculesreaching anY Pla

last collision at a distancea frol

The pseudo-reduced
pressureand temperatureare
p
136
' :_ :2 .e 4
46.3
P"'
T
313

Theory

= 0.7:.

To determine the diffusivitY:

in the y-direction under a conc
when the mixture movesat a fini
Z and total molar concentratio

(0 . 7 3 )
fpr*1o: 136

:9.9 x 10-4cmzsec-l
An observedvalue of 9tn:8.4
x 10-a cm2sec-lhas been reportedat these
conditions.sIf (pOn)o is predictedby the Chapman-Enskogtheory
givenin $16.4,
9ru is predictedto be 8.8 x l0-a cm2sec 1,which is unexpectedly
good agreement.
$ t6. 4 T HE O Ry O F OR D T N AR YD tF F U S| ONt N GA S E S
A T LO W D E N S IT Y
The mass diffusivity I .ru for binary mixtures of nonpolar gasesis predictable within about 5 per cent by kinetic theory. Again, as in $$1.4and
8.3,we begin with a simplifiedderivationto illustratethe mechanismsinvolved
arld then present the more accurate results of the Chapman-Enskog theory.
Considera large body of gas containingtwo molecularspeciesA and A*,
both specieshaving the same mass ,??_.1
and the same size and shape. Such a
is
pair
closely approximated by two heavy moleculescontaining different
isotopes, such as U235Faand U238Fu. We wish to determine the mass
diffusivity9r,o. in terms of the molecularpropertieson the assumptionthat
the moleculesare rigid spheresof diameterr{r.
As before, we shall use the following resultsof kinetic theory for a purer
rigid-spheregasat low densityin which thc temperature,pressure,and velocity
gradientsare small:
il :
7 :

/8" '"-'"
J'
Tm

f,nu

.l
--=^ - \J
l^

ljr

mc a n m o l e c u l arspeed
re l a ti v eto u o r u*

(16.4-1)

wall collision frequency

per unit area in
stationarygas

(16.4-2)

mean free path

(16.+3)

Fi g. l 6.,Ll .

Mol ec ul artrans P o

E qs. 16. 4 1, 2, 3, and 4 r em ai

molar flux, N.t, of species,4
counting the moleculesof I th
y-directionand subtractingthe
Thus

Ntu : l.pr.r,

In writing the last two terms

ro(y) is essentiallylinear for :
of this ass
further consequence

r,

A- n

3 V. J. Berry, Jr. and R. C. Koeller, Diffusion in CompressedBinary GaseousSystems,

Preprint45, A.I.Ch.E.40th National Meeting,1959.
t The usc of the expressionsfor a pure gas is permissiblehere becausethe mechanical
propertiesof ,4 and l+ are the same.

frtl , - ,

,olo*,

he Mechanismsof MassTransport
are

T heory

of Ordi nary

D i ffusi on i n Gases at Low D ensi ty

509

The moleculesreachingany plane in the gas have, on the average,

had their
last collision at a distancea from the plane, rvhere

.94

o:34
1 .47
i we obtainpgaal(po,")o .. 0.73.
0 . 184
:136 (0 .7 3 )

(16.4-4)

To determinethe diffusivity g,r.r*, we consider the motion of

speciesr
in the r7-directionunder a concentrationgradient clr.rlrty(seeFig.
16.4-l),
when the mixture movesat a finite velocityur* throughout. The te"mperature
z and total molar concentrationc are assumedconstant. we assume
that
Mole-fractionprofilex t)
n(

^-l

lsec-r has been reported

at these
rman-Enskog
theorygivenin 916.4,
ch is unexpectedly
good agreemenr.
O N I N GASES
<tures:-f nonpolar gasesis pretheory. Again, as in $$1.4and
lustratethemechanisms
involved
cf the Chapman-Enskogtheory.
wo molecularspeciesA and A*,
he samesizeand shape. Such a
' moleculescontaining different
e wish to determine the mass
'opertleson the assumptionthat

that are of sp ec i es
A ts x^l
^ll-a

Fi g. 15.'l -1. Mol ecurartransport of speci esA from praneat (u

-

Lltsof kinetic theory for a puret

nperature,pressure,and velocity
rlecularspeed
o u o r u*

(16.4-r)

sion frequency
lrea in
F gAS

(16.4-2)

e path

(16.4-3)

Eqs. 16.4 1,2,3, and 4 remai n val i d i n thi s nonequi ri bri umsit uat ion.
The
molar flux, Nr, of speciesl. acrossany plane of constant y is found
by
counting the moleculesof I that cross unit area of the plane in the positive
y-directionand subtractingthe number that crossin the negativer/-direction.
Thus
N ,l u : l o{rr,,r,* l u + f,,rc nal ,-, -

nissiblehere becausethe mechanical

1nr.rul ,,* " 1

( 16. 4- 5)

In writing the last two terms, it is assumedthat the concentrationprofile

r:o(y) is essentiallylinear for a distanceof severalmean free paths.
As a
further consequence
of this assumptionwe may write

)ompressed
Binary GaseousSystems,

o) to praneat y .

x-tl u-o:rtl r-:A --

2^fu,
ray

:
x.r lr +n
' .r 1,*:,

(16.+6)
(16.4-7)

510

Diffu sivity a n d th e M e chani sms of Mass Transport

By com biningE qs . 1 6 .4 -5 ,6 , a n d 7 a n d n o ti n g th at cu* :

Ta b l e l6. l- 3, E q. M ) , w e g e t

N,4o: x,t(N,tr* N,.r.r)

-

N ,r * N ,* (see

lrO^k

(16.+8)

This equation correspondsto the y-component of Fick's law (Eq. 16.2-2)

with the following approximatevalue for 9uo:
q 18:

(16.4-e)

+ilA

Evaluation of a and l, from Eqs. 16.4-l and 3 and use of the ideal-gaslaw
n : cRT : llrl sives

(16.4-10)
which represcntsthe mass diffusivity of a mixture of two speciesof rigid
spheresof identical mass and diameter. The calculation of 9;r' for rigid
spheresof unequal mass and diameter is considerablymore difficult; the
correspondingresult2is

?,,,::(q)'(+
* +)'' , T"',,
,,,
r \ 7r'

\ z nt 1

t nt R'

(16.4-11)

p\ u ' t r o n ) -

Equations 16.4-10and 1l predict that the massdiffusivityvariesinversely

with pressure; this predictionagreeswell with experimentaldata up to about
l0 atm for many gas mixtures. (SeeFig. 16.3-1.) The predictedtemperature
dependence
is too weak,just as it was for viscosityand thermal conductivity,
but it is qualitativelycorrect.
The foregoing developmentis illustrative only; for accurateresults,the
Chapman-Enskogkinetic theory should be used. The Chapman-Enskog
formulas for viscosityand thermal conductivitywere given in $$1.aand 8.3;
the correspondingformula for c9,r,, for the gaseousstate at low density
i s:3 'a

c 9,tt:

2 .2 6 4 6x 1 0 -s

^lwil

(16.4-12)

o ,a p 2 { )e,-aB

'?A similar result is given by R. D. Present,Kinetic Theory of Gases,McGraw-Hill,

New York (19 58 ),p. 55 .
3 S. Chapman and T. G. Cowling, MathematicalTheoryof Non-UnifurntGases,Second
Editidn,CambridgeUniversityPress(1951),Chaptersl0 and 14.
a J. O. Hirschfelder,C. F. Curtiss,and R. B. Bird, MolecularTheoryof GasesandLiquids,
Wiley,New York (1954),p. 539.

Theory

o f Ordi nary

D i ffus i on i n (

If we approximate c by the ideal

9m : 0'0018

in which 9uof:l
cm2sec r, c
o,tnl.:lAngstrdm units, and (
temperature and of the intermo
I and one of B. lt is convenie
Lennard-Jonesfunction:

v'qn(r): 4

Table B-2 gives C)",_rofor thir

rTf e,ro. From these resultsone
the 2.0 power of T at low temp
high temperatures.(CompareEq
unity at all temperaturesand a
obtained.
The Lennard-Jonesparame
determined directly from accura
of temperature; with somewha
of mixturesof A and.Bcouldbe
rare; therefore,one usuallyhas
fairly satisfactoryestimates
cant
pairs by combining the Lenn
empi ri cally: 6
oA B '

e,lB t

In this way it is possibleto pred

deviationof about 6 per centby l
-B or within about l0 per cent if
are estimatedfrom boiling-poin
For isot opic pair s, d11. :
the intermolecular force fieldsf<

5By nonreacting
we meanheretha

S uch a col l i s i on, of c ours e,rul es out t

pairs that react only catalytically are:
o J. O. Hirschfelder, R. B. Bird, ant
? R. C. Reid and T. K. Sherwood,
New York (1958), pp. 27+5.

-.-z

I the Mec hanis m s o f M a s s T r a n s p o r t

Theory

noting that cu* :

If we approximate c by the ideal-gaslaw c : plRT, t"hisbecomes

1 -^ d r,
,
' "3
, ,)du
--cu y'" "

N,, * Nr. (see

)''4
P Ct,t-

06.4_s)

(r6.4-10)

a mixture of two speciesof rigid

The calculation of Iro for rigid
; considerablymore difficult; the

(16.4-rr)

he massdiffusivityvariesinversely
vith experimentaldata up to about
6.3-1.) The predictedtemperature
riscosityand thermal conductivity,
ve only; for accurateresults, the
be used. The Chapman-Enskog
t iv i ty we reg i ve n in ggl. 4 and 8 .3 ;
the gaseousstate at low density

RilA
oa62 {)s,,1p

in Gases at Low Density

9ta :0.0018583

lnd 3 and use of the ideal-gas law

Diffusion

(16.,t-8)

ponent of Fick's law (Eq. 16.2-2)

9.qn:

nA

of Ordinary

(16.4-r2)

Kinetic Theory of Gases,McGraw-Hill,

I Theoryof Non-Unifornr Gases,Second
tersl0 and 14.
l, MolecularTheoryof Gasesand Liquids,

"'( M"

.h)

P6.qnzQs,a.n

(16.4-r3)

i n w hi ch- ?uol :l
cm2sec-l , c [:] g-mol escm-3, ?n[:] ' K , p [ : ] at m ,
oanl:lAngstrcim units, and Qr.no is a dimensionlessfunction of the
temperature and of the intermolecular potential field for one molecule of
A and one of ,8. It is convenient to approximate this potential field by the
Lennard-Jonesfunction :

o,^,1(T)"v,,n?):
(TI]

(16.4-r4)

Table B-2 gives Q",.r' for this potential energy model as a function of
rc.Tfeuo. From these results one can compute thatQuu increasesroughly as
the 2.0 power of T at low temperaturesand as the 1.65power of I at very
high temperatures.(CompareEq. 16.3-1.)For rigid spheres,Qr.uuwould be
unity at all temperaturesand a result analogousto Eq. 16.4_ll would be
obtained.
The Lennard-Jones parameters ouo and ero could, in principle, be
determined directly from accurate measurementsof guu over a wide range
of temperature; with somewhatgreaterdifficulty,other measuredproperties
of mixtures of A and.B could be used. Suitable data are, hovrever,extremely
rare; therefore, one usually has to estimate o"o and euu. It turns out that
fairly satisfactoryestimatescan be made for nonpolar, nonreactingbmolecule
pairs by combining the Lennard-Jonesparameters of speciesA and B
e m pi ri cal l v:6
6Ar]:
'uo

$(o, * orr)

: J
'n, n

(16.4-1s)
(16.4-16)

In this way it is possibleto predict measuredvaluesof 9uo within an average

deviationof about 6 per cent by useof viscositydata on the pure speciesI and
B or within about l0 per cent if the Lennard-Jonesparametersfor A and B
are estimatedfrom boiling-pointdata by meansof Eqs. 1.4-14 and 15.?
For i sotopi c pai rs, o1t.: d1 : ou. and A A r: a: 1*; t hat is,
the intermolecular force fields for the various pafts A-A, A*-A*, and A-A*
5 By nonreacting we mean here that A and
B, in a bimolecular collision, do not react.
Such a collision, of course, rules out the presence of any catalytic agent, so that moleculepairs that react only catalytically are nonreacting pairs in the present discussion.
6J. O. Hirschfelder, R. B. Bird, and E. L. Spotz,
Chem. Reus.,4,205-231 (1949).
7 R. C. Reid and T. K. Sherwood, The
Properties of Gases and Liquids, McGraw-Hill,
New York (1958), pp. 27,f"5.

=..|-F

the Mechanisms of Mass Transport

A may be obtained from viscosity

ge,Eq. 16.4-12 simplifies to

T heori es of Ordi nary

D i ffusi on i n Li qui ds

513

This gives r.Tlea} :293.21118.5:2.47, and from Table B-2 we then obtain
ae,as : 1.003. Substitutionof the foregoingvaluesinto Eq. 16.4-13gives

t-----:-

r'!fU".

oe'Qs,'c

06.4-17)

; derived earlier. (Compare Eq.

howsthat the self-diffusivtty9uu.
isotopic gas-pairsat low density:
{ c)^

(16.,t-18)

6 Qu

functions of xTf en. (See Table

lhe self-difusiuity I *1. are of the
, density. The relation between l
imple becausev : pl p may vary
.tio pl p0uu lies between 0.2 and
pQ_Ea: yl0aB is known as the
quantity repeatedly in problems
:ncountered the Prandtl number
on in flow systems.
derived for monatomic nonpolar
ior polyatomic nonpolar gasesas
e used to predict Onu for interby using combining laws different

r of Mass Diffusivity
sity
gon (l) and oxygen(B) at 293.2'K
iollowingconstants:
^

A;

-3 :

1 2 4' K

:0.0018583@
"/tzrl.zx1-

l\

+, lz^oo/

: 0.188cm2sec-l
This comparesfavorablywith the measuredvalueof 0.20cm2sec-rgivenin Table
16.2-2.
S OR D TN A R YD | FFU S | ONtN L| QU TDS
$t6.5 TH E OR TE OF
In the absenceof rigorous theory for diffusion in liquids, there are two
rough theories that have been useful in making order of nragnitudeof calculations: the hydrodynamicaltheory and the Eyring theory. (We have already
given some results of the theories of Eyring and co-workers in connection
with the calculation of viscosity and thermal conductivity of liquids.)
The hydrodynamicaltheory takes as its startjng point the Nernst-Einstein
equation,l which states that the diffusivity of a single particle or solute
molecule of I through a stationary medium ,B is

9,sn:
"rfr

(16.s-1)

in which u,afF,1 is the "mobility" of the particle of I (that is, the steadystate velocity attained by the particle under the action of a unit force). A
relation between force and velocity for a rigid sphere moving in "creeping
flow" (that is, Re ( l) may be obtained from hydrodynamics. If the
possibility of "slip" at the sphere-fluidinterface is taken into account, thenz

F,t:6tltnu^^^\ffffi)

(16.s-2)

A; ? : tt:'r
K

of argon with oxygenmay be esti.33): 3.426A

il8. 5" K
rd,loc.cit.,$8.6b.

in which po is the viscosity of the pure solvent, R, is the radius of the

diffusing particle, and Bno is the coeficient of sliding friction. It has been
pointed out3 that there are two limiting casesof interest:
1 F. Daniels and R. A. Alberty, Physical Chentistry, Wiley, New York (1955), p. 650.
See also Eq. 18.4-14a.
2 H. Lamb, Hydrodynanrics, Dover, New York,
$337.
3 R. Fi.irth, in Handbuch der physikalischen und technischen Mechanik, Barth, Leipzig
( 1 931), V ol . 7, p. 635.

5l {

Diffu sivity a n d th e M e ch a n ism s of Mass Transport

a. If there is no tendencyfor the fluid to slip at the surfaceof the diffusing

particle(flen: co), then Eq. 16.5-2becomesStokes'slaw. (See$2.6.)
Ft:

6n\sutRt

(16.5-3)

and substitution into Eq. 16.5-l gives

9z.n\n

(16.5-4)
6nR,1

rT

which is usually called the Stokes-Einsteinequation. Equation 16.5-4 has

been shown to be fairly good for describing the diffusion of large spherical
particles or large spherical molecules,under which conditions the solvent
to the diffusing speciesas a continuum-4
appears
b. If th... is no tendencyfor the fluid to stick at the surfaceof the diffusing
particle(0en:0), Eq. 16.5-2becomes
Ft :

4t[tnutR,t

(16.s-5)

and Eq. 16.5-l becomes

9ta\t

xT

(16.5-6)

4nRa

If the moleculesare all alike (that is, self-difusion) and if they can be assumed
to be arrangedin a cubic lattice with all moleculesjust touching,then2Ro
may be set equal to (lolfr)% and

z-EsLt,t:
lfII'
xI

(16.5-7)

2n\Vu/

It has been shown5 that Eq. 16.5-7 predicts the self-diffusiondata for a
number of liquids within about f 12 per cent; that comparison includes
liquid metals,and molten sulfur.
associatedsubstances,
polar substances,
It is thus seenthat the simple hydrodynamicalapproachgivesexpressions
for the diffusion coefficient for spherical molecules in dilute solution and
also for the coefficientof self-diffusion. The theory further predicts that there
should be a variation of 9uB with the size of the diffusing species' The
hydrodynamic theory suggeststhat the shapeof the diffusing speciesmay well
be important, since the friction factor increasesby a factor of about 2 as the
length-to-width ratio of a body goes from I to 10'
The El,ring rate theoryt attempts to explain the transport phenomena on
the basisof a simple model for the liquid state. It is assumedin this theory
that there is some unimolecularrate processin terms of which the diffusion
{ W. Sutherland,
(1905).
Phil.Mag.,9'781-785
5J. C. M. Li andP. Chang,
(1955).
518-520
Phys.,23'
"/.Chem.
6S. Glasstone,
McGraw-Hill,
K. J. Laidler,and H. Eyring,Theoryof RateProcesses,
IX.
Chapter
NewYork(1941),

Theori es of Ordi nary

D i ffus i on i n Li qui ds

processcan be described,and it is further

is some configuration that can be iden
Eyring theory of reaction rates is applied
On the assumption of a cubic lattice c
following relation between the self-diffus
viscosity:
gupe
(
_
-\
r <f

This relation differs by a factor of 2n fr'

the self-diffusion coefficientaccordingto 1
16.5-7 has been shown to fit the experir
Recently Li and ChangThave suggesteda
in order to account for the discrepancy.
Becauseof the approximate nature of t
empirical relationshave been proposed,8o
Wilkee has developed a correlation foI
of the Stokes-Einsteinequation. His rt
following approximate analytical relation
cient in cm2 sec-l for small concentratio
?t a : 'l '4 x l0-

Here Vl is the molar volume of the solut

normal boiling point, pris the viscosityof
"associationparameter"for the solvent.d
o
in K. Recbmmendedvaluesof tpoare
for ethanol, and 1.0 for benzene,ethe"
solvents. This equation is good only for
solutes; for such solutionsit is usuallyg
If the reader has by now concludedthat
of densegas and liquid diffusivities,he i
for experimental measurements,both for
ment of future theories.

Est im at io
E xam ple 16. 5- 1.
for a Binary Liqu
Estimateg no for a dilute solutionof TN
l 5' c.

? Li and Chang, loc. cit.

8 See, for example, R. C. Reid and T. K. Sher
McGraw-Hill, New York (1958), pp. 28+298.
45,218-224
'0C. R. Wilke, Chem, Eng. Prog.,
10C. R. Wilke and P. Chang, A.LCh.E. Journ

515

I Mechanisms of Mass Transport

T h e o r ies of Ordi nary

ip at the surfaceof the diffusing

s Stokes'slaw. (See$2.6.)

processcan be described,and it is further assumedthat in this processthere

is some configuration that can be identified as the acliuated state- The
Eyring theory of reaction rates is applied to this elementaryprocess.
On the assumptionof a cubic lattice configuration,this theory gives the
following relation between the self-diffusioncoefficientand the coefficientof
viscositv:

).A

(16.5-3)

(16.s-4)
A

equation. Equation 16.5-4 has

the diffusion of large sphericol
r which conditions the solvent
rm .4
ck at the surfaceof the diffusins

(l

(16.5-5)
(16.5-6)

,sion)and if they can be assumed

:leculesjust touching, then 2R n

)-

(16.5-7)

ts the self-diffusion data for a

ent; that comparison includes
d metals,and molten sulfur.
rical approachgivesexpressions
roleculesin dilute solution and
:heoryfurther predict-sthat there
e of the diffusing species. The
of the diffusing speciesmay well
.sesby a factor of about 2 as the
to 10.
.in the transport phenomena on
rte. It is assumedin this theory
in terms of which the diffusion
(19s5).
L8-520
tory of Rate Processes, McGraw-Hill,

D i ffusi on i n Li qui ds

+:(il)"

(16.5-8)

9 . t x : 7 ' 4 x l0 -. (oa 4 M u \ " 7

ffi

( 16.5-e)

This relation differs by a factor of 2zr from Eq. 16.5-7, the expressionfor
the self-diffusioncoefficientaccordingto the hydrodynamictheory; but Eq.
16.5-7 has been shown to fit the experimentaldata better than Eq. I6.5-8.
RecentlyLi and Chang?have suggesteda modificationof the Eyring theory
in order to account for the discrepancy.
Becauseof the approximatenature of the foregoingtheories,a number of
empiricalrelationshavebeenproposed,sone of which we mention briefly.
Wilkee has developeda correlation for diffusion coefficientson the basis
of the Stokes-Einsteinequation. His results may be summarizedby the
following approximateanalyticalrelation,l0which gives the diffusion coefficient in cm2sec l for small concentrationsof I in B:

Here Va is the molar volume of the soluteI in cm3 g-mole"r as liquid at its
rp, is an
normal boiling point, pris the viscosityof the solutionin centipoises,
"associationparameter"for the solventB, and Zis the absolutetemperature
o
in K. Recbmmendedvaluesof tyo are 2.6 for water, 1.9 for methanol, 1.5
for ethanol, and 1.0 for benzene,ether, heptane,and other unassociated
solvents. This equation is good only for dilute solutionsof nondissociating
solutes; for such solutionsit is usually good within -f 10 per cent.
If the readerhas by now concludedthat little is known about the prediction
of densegas and liquid diffusivities,he is correct. There is an urgent need
both for their own value and for developfor experimentalmeasurements,
ment of future theories.
E sti mati onof MassD i ffusi vi ty
E xampl e 16.5-1.
for a B i nary Li qui d Mi xture
in benzeneal
Estimategoufor a dilute solutionof TNT (2,4,6-trinitrotoluene)
1 5 ' C.
? Li and Chang, loc. cit.
8 See, for example, R. C. Reid and T. K. Sherwood, The Properties of Gasesand Liquids,
McGraw-Hill, New York (1958), pp. 28+298.
(1949).
'0C. R. Wilke, Chem. Eng. Prog., 45,218-224
10C. R. Wilke and P. Chang, A.l.Ch.E. Journal, l,26+270, 1955.

--Z

d the Mechanismsof Mass Transport

Pr o bl ems

rd Chang, taking TNT as comPonent

red data are
Insideredas pure benzene)
-lt

P R OB LE MS
1 6 .A1 P redi cti on of Mass D i ffusi vi ty for a Gas-P ai r at Low D ensi ty
Predict 9n" for the methane-ethane system at 104" F and I atm, using the ChapmanEnskog theory. Compare your result with the experimental value given in Example 16.3-2.
A nsw er: 0. 166 c m! s ec -I
l6 .Br

78.11)'/,(213
+ 15)
0.705(140)0.t,
iec-1
e d val u e of 1. 39 x 1 0 5 c m 2 s e c - l .

5 t7

P redi cti on of Mass D i ffusi vi ty at Low D ensi ty from cri ti cal

a . The S l attery equati on, E q. 16.3-1.

b . The chapman-E nskog theoreti cal equati on, E q. 16.4-13, usi ng cri ti cal pres s uresand
te m peraturesto esti mate the Lennard-Jonesparameters. (S eeE qs. 1.4-l t and 13.)
A nsw ars: a. 0.1 12 c m2 s ec 1
b. 0.1 56 c mz s ec -l
I6 .C1 C orrecti on of Mass D i ffusi vi ty for Temperature

DI SC US S T O N
e diffusion to occur?
o n?
1 . l6 .l -l

at Low D ensi ty

A v al ue of 9uo:0.151
cm, sec'1 has been reportedr for the system C Or-a i r at 293. K
a n d I atm. E xtrapol ate gnoto 1500" K by the fol l ow i ng methods:

a. Eq. 16.3 1.
b. Eq. 16.4-11.
c. Eg. 16.4-13.

wit h t he a v e r a g ev e l o c i t y d e fin e d

a n d Jr*. Com pare t h e s i g n i f i c a n c eof th e

properti es

Predict 9 uo for the situation of Problem l6.,4, by the following two methods :

Answersin cm2sec-r: a. 2.96

b . 1 .7 5
c. 2.49
Obserued:r 2.45 cm2 sec r

1 6 .D 1 P redi cti on of Mass D i ffusi vi ty for a Gas Mi xture

at H i gh D ensi ty

Pr edi ct 9 .o for an equi mol ar mi xture of C O, and N , at 288.2. K and 40 atm:

ocity in reporting data on rates ofdiffusion ?
wton's law of viscosity and Fourier's law
: set h ree relat ions a n a l o g o u s ?
the flux N, that apply to (a) equimolar
4 through stagnant .B (N" : 6;.

tnd of c9 uo for gases and liquids at I atm.

liffusivity-concentration curve for ethanol-

2-3?
the statementconcerningthe temperature
predictingthe intermolecularforce field
?
lby the Stokes-Einstein
equation? Verify
rd p for gaseswith that for liquids.
d value of 9"" from Eq. 16.4-13would
er"fx1
rat the mole fraction gradient Vo, is noF
Explain.
f Schroeder
as quoted by R. C. Reid arfd
ids,McGraw-Hill,New York (1958),p. $1.

a. Use the value of 9no at 1 atm from Table 16.2-2.

b. Use the chapman_Enskog theory to predict (p9 n),, .
Answers; a. 0.0033 cm, sec_r
b. 0.0030 c mz s ec r
l6 .El

E sti mati on of Mass D i ffusi vi ty for a B i nary Li qui d Mi xture

Esti mate gnofor aceti caci di ndi l uteaqueoussol uti on at12.5" C byusi ngE q.

16.5-9.
T h e densi ty of aceti c aci d at i ts normal boi l i ng poi nt i s 0.937 g cm- 3. (A n ex peri mental
value of 0.91 * 0.04 x l0 5 cmz sec-r has been reported.r)
I5 .F r

correcti on of Mass D i ffusi vi ty for Temperature for a B i nary Li qu i d Mi x ture

T h e val ue of 9uo for a di l ute sol uti on of methanol i n w ater at 15" c i s l .2g x l 0 5
cm z sec r. E sti mate ?ru for the same sol uti on at 100. C , usi ng thc Wi l ke-C hang equuti on.
A nsner: 6.7 x l 0 5 c m2 s ec -r
l6 .Gr

correcti on of Mass D i ffusi vi ty for Temperature for a D ense Gas Mi x ture

T h e observed val ue of guufor a mi xture of 80 mol e per cent methane a nd 20 mol e
p e r centethaneatl 04'Fand2000psi ai s8.4
x l 0 acm2sec r. predi ct 2uufo r the s ame
m ixture at 2000 psi a and 171" F:
a . Use E q. 16.3-1 and Fi g. 16.3-1.
b . U se E q. 16.+13 and Fi g. 16.3-1.
A nsw er: a. 1.19 x l 0 3 c m2 s ec r
1 T s. M. K l i banova. V . V . P omerantsev,
and D . A . Frank-K amenetski i , J.Tec h. phy s .
(U.S.S.R.), 12, 14-30 (1942), as quoted by C. R. Wilke and C. y. Lee, Ind. En,q. Chem.,
4 7 , l25r (1955).
2 International Critical Tables,
McGraw-Hill, New york (1929) Vol. V, p. 70.

5! 5

DSffu sivity a n d th e M echani sms of Mass Transport

P robl ems

Solution. Use the equation of Wilke and Chang, taking TNT as comPonent
I and benzeneas component B. The required data are
p : 0.705 cp (for solution considered as pure benzene)
l,t : t40 cc g-mole- I for TNTll
T n : l'O for benzene
MR :78' 11 f or bc nz c nc
Sub stitu tionin E q. 16. 5- 9giv es

?,,t: 7.4x to- .' t " J.t;l/i;:Tl.",

:

1 .3 8 x 1 0 5 cm 2 se c I

PR
16.A r

P red i c ti on of Mas s D i ffus i v i ty I

Predict 9no for the methane-ethane sy

Enskog theory. Compare your result with
I6.8r

P redic ti on of Mas s D i ffus i v i ty a

Predict 9 unfor the situation of Problem

a. The S l attery equati on, E q. 16.3-1.

b. The Chapman-Enskog theoretical ec
temperatures to estimate the Lennard-Jor

T hi s r e s u l t m a y b e co m p a r e d with a m e a su r e d va lu e of 1.39 x l 0 5 cm2sec-l .

l 6.C r
EUEST IONS

F OR DISCUS S ION

l. W h a t i s d i f f u s i o n ? Wh a t fa cto r s m a y ca u sc d iffu sion to occur'l

2 . W h e n i s m a s s . fr a ctio n cq u a l to m o le fr a ctio n ?
3 . C o m p a r e t h e m ass a ve r a g e ve lo city u in Eq . l6 .l- l
in $ 2 . 2 .

C orre c ti on

of Mas s D i ffus i v i ty

A value of I no : 0. I 5 I cm'! sec-r has t

and I atm. Ex trapol ate 9ruto 1500"K l
a. E g. 16.1-1.
b. E q. 16.4-11.
c. E q. 16.4-13.

w i th the averagevel oci ty defi ned

4 . C o m p a r e t h e s i g n ifica n ceo f th e flu xe s N, a n d Jr *. C ompare the si gni fi canceof the

f luxe s n , a n d 7 , .
5 . W h y i s i t n c c e s s ar yto sp e cilya r e fe r e n ce ve lo cityin r e porti ngdataonratesofdi ffusi on?
6 . C o m p a r c F i c k ' s la w o f d iffu sio n with Ne wto n ' s law of vi scosi ty and Fouri er's l aw
of t h e r m a l c o n d u c t i v ity. T o wh a t e xte n t a r e th e se th r e e rel ati ons anal ogous?
7 . G i v c t h e s p c c i al fo r m s o f F ick' s la w fo r th e flu x N r that appl y to (a) equi mol ar
counterdiffusion (N, : -N"), (b) diffusion of ,4 through stagnant ,B (N, : 0).
8 . W h a t a r e t h c d im e n sio n s o f 9 u u , r , a n d a ?
9 . C o m p a r e t h e o rd e r o f m a g n itu d e o f 9 n u a n d o f c?uo for gasesand l i qui ds at I atm.
1 0 . W h a t i s t h e d i f f er e n ce in b e h a vio r o fth e d iffu sivity- concentrati oncurveforethanol wat e r a n d t h a t f o r r r- b u ta n o l- wa te rin T a b le 1 6 .2 3 2
1 1 . B y m c a n s o f a sm a ll ca lcu la tio n , ve r ify th e sta lement concerni ng the temperature
dep e n d e n c eo f 9 r r j u s t a fte r Eq . 1 6 .4 - 1 4 .
12. What empirical mcthod is available for prcdicting the intermolecular force field
be t w e c n a p a i r o f d i ssim ila r n o n p o la r m o le cu le s?
13. What bulk physical properties are related by the Stokes-Einstein equation? Verify
t ha t t h e e q u a t i o n i s dim cn sio n a lly co n siste n t.
14. Compare the relationship between ?nuand 7ufor gaseswith that for liquids.
1 5 . W h a t p e r c e n ta g e o f e r r o r in th e p r e d icte d va lu e of 9uu from E q. 16.21-13w oul d
res u l t f r o m a 5 p e r ce n t e r r o r in o u o ? ln T ? ln e n u lr ?
16. In binary diffusion, does the statement that the mole fraction gradient Vc, is noi
zero imply that the molar flux N, is not zero? Explain.
1r Estimated by the atomic-volume method of Schroeder as quoted by R. C. Reid ay'd
T. K . S h e r w o o d , T h e Pr o p e r tie so f Ga se sa n d L iq u id s,M cGraw -H i l l ,N ew Y ork(1958),p. l l .

I6,D r

P redic ti on of Mas s D i ffus i v i ty I

Predict 9

for an equimolar mixture <

-ro
a. Use the value of 9"o at 1 atm from
b. Use the Chapman-Enskog theory to

l 6.E r

E sti m ati on of Mas s D i ffus i v i ty '

Estimate gno for acetic acid in dilute :

The density of acetic acid at its normal
value of 0.91 + 0.04 x 10-5 cm2 sec-r ha
l 6.Fr

C orre c ti on

of Mas s D i ffus i v i ty I

The val ue of 9u" for a di l ute s ol uti or

cmz sec-r. Estimate g tn for the same solr
I6.Gr

C orrec ti on of Mas s D i ffus i v i ty

The observed value of guo for a mixt
per cent ethane at 104" F and 2000 psia is
mi xture at 20 00 ps i a and 171' F:
a. U se E q. 16.3-l and Fi g. 16.3-1.
D . U se E q. 16.4-13 and Fi g. 16.3-1.

I Ts. M. Klibanova. V. V. Pomerantse

(U..t..t.R.), 12, l4-3O (1942), as quoted t

47, 1253(19s5).
2 Inlernational

Critical Tables, McGraw