r x y - ethz.ch · 1 n r t c 0 1 1 0 n1 j1 c1v d c c v where v0is the volume average velocity. 1 0...
TRANSCRIPT
reactionchemicalby
destroyedorproduced
1speciesofmass
outthatusmin
in1species
offluxmass
zyx
inngaccumulati
1speciesofmass
zyxr
yxnyxn
zxnzxn
zynzyn
zyxct
zzz1zz1
yyy1yy1
xxx1xx1
1
Divide by the volume x y z:
1z1y1x11 r
z
n
y
n
x
n
t
c
GENERALIZED MASS BALANCESEquations that describe both steady and non-steady state
Mass balance for species 1 in volume Δx Δy Δz
1
5‐2
Mass balance for species “1” for different coordinate systems:
2
5‐3
111 rnt
c
011
0111 vccDvcjn
where v0 is the volume average velocity.
10
1121 r)vc(cD
t
c
(1)
Express contributions of diffusion and convection to the speciesmass flux:
Mass balance for species “1” in vector notation:
3
5‐4
Dividing by x y z yields the continuity equation:
Overall (total) mass balance, considering all species
zzzzz
yyyyy
xxxxx
yxvyxv
zxvzxv
zyvzyv
zyxt
where vx, vy, vz are the components of the mass average velocity.
Note: The overall mass balance has no reaction term since no total mass is generated or destroyed.
zyx vz
vy
vxt
v
t
or (2)
4
5‐5
Overall (total) mass balances for different coordinate systems:
5
5‐6
Simplify the species mass balance (1) with the help of the continuity equation (2)!Problem: Different reference velocities used.
For constant density, the mass and volume average velocities are the same and
00 v)v()v(0t
Divide by ρ, multiply by c1 and note that 011
001 vccv)vc(
100
1100
10
1 cvvcsocvvcvc0 →
so eqn. (1) becomes 112
101 rcDcv
t
c
6
5‐7
Mass balance for species “1” with diffusive and convective terms for constant density systems :
1
01
010
r1 c
sinr
vc
r
v
r
cv
t
c
121
2
221
212
2r
c
sinr
1csin
sinr
1
r
cr
rr
1D
7
5‐8
ExamplesFast diffusion through a stagnant film
1. Step:Select appropriate mass balance
1z11
r11 r
z
nn
r
1nr
rr
1
t
c
or for const. density:
12
12
2
12
2
1
10z
10
10r
1
rz
cc
r
1
r
cr
rr
1D
z
cv
c
r
v
r
cv
t
c
Remember few weeks back….
8
y1l
5‐9
2. Step: Simplify mass balance
1z11
r11 r
z
nn
r1
nrrr
1t
c
=0
steady state
=0
no reaction
=0
no flow inr-direction
=0
symmetry
gives: 0z
n z1
or:
21
210
zz
cD
zc
v
Then solve as before:
9
5‐10
Fast diffusion into a semi-infinite slab
Remember again:1. Step:Select appropriate mass balance
Here: capillary 1z11
r11 r
z
nn
r1
nrrr
1t
c
10
2. Step: Simplify mass balance
• No flow in x-direction• No flow in y-direction• No chemical reaction
0z
n
t
c z,11
Or: 21
210
z1
z
cD
z
cv
t
c
Then solve as before (Ex. 3.2.4).
5‐11
The flux near a spinning disk
A solvent flow approaches a spinning disk made out of a sparingly soluble solute, as shown below. Calculate the diffusion-controlled rate at which the disk slowly dissolves at steady state.
1. Step:Select appropriate mass balance
12
12
2
12
2
1
10z
10
10r
1
rz
cc
r
1
r
cr
rr
1D
z
cv
c
r
v
r
cv
t
c
11
5‐12
2. Step: Simplify mass balance
12
12
2
12
2
110z
10
10r
1 rz
cc
r
1
r
cr
rr
1D
z
cv
c
r
v
r
cv
t
c
=0
steadystate
=0
noreaction
=0 =0
angularsymmetry
angularsymmetry
seeflow pattern
=0 =0
seeflow pattern
21
210
z z
cD
z
cv
12
5‐13
Velocity over a spinning disk From: Levich, “Physiochemical Hydrodynamics”, Prentice-Hall, 1962.
ϴ
The velocity profile suggests that there is no gradient in radial direction! This is true for sufficiently large disks where edge effects are negligible.
This special flow behavior is taken advantage of in the application of homogeneous coatings, e.g. on wafers.
13
5‐14
3. Step: Boundary conditions
B.C.: z=0 c1=c1(sat)z= c1=0
bdrds)s(vD/1expacz
0
r
0 z1
Where a and b are integration constants calculated from the boundary conditions, so
The solution is:
drds)s(vD/1exp
drds)s(vD/1exp1
)sat(cc
0
r
0 z
z
0
r
0 z
1
1
14
5‐15
If we know the velocity vz we can get c1. From the literature (Levich, 1962) we know
22123z s51.0v
and inserting in the above eqns. gives:
duuexp
duuexp
)sat(cc
03
03
1
1
where
1
2/1
6/13/1D82.1z
is the kinematic viscosity of the liquid and is the angular velocity of the disk.
The diffusion flux is:
)sat(cD
62.0zc
Dj 16/1
2/13/2
0z1
0z1
15
5‐16
This equation can also be written as:
)sat(cScRed
D62.0j
)sat(cD
d
d
D62.0j
1
chaptersgminupcotheinMore)tcoefficientransfermass(k
3/12/11
1
3/12/12
1
There is an excellent agreement when plotting j1 d / D c1(sat) vs. Re1/2 justifying the strong assumption made in the analysis.
The fact that this device gives uniform mass transfer over the entire surface area makes it popular in CVD and electrochemistry.
16
5‐17
Example : Dissolving Pill
In pharmacy it is important to know HOW FAST a medicine can permeate the body to act. Estimate the time it takes to start a steady-state dissolution of a drug pill. Is dissolution diffusion controlled? This can be checked by gentle stirring. If it dissolves faster than in the absence of stirring then it is diffusion-controlled!!
1. Step:Select appropriate mass balance
1
21
2
22
12
122
10
10
10r
1
rc
sinr
1
csin
sinr
1
r
cr
rr
1
D
c
sinr
vc
r
v
r
cv
t
c
17
5‐18
2. Step: Simplify mass balance
121
2
221
212
2
10
10
10r
1
rc
sinr
1csin
sinr
1rc
rrr
1D
csinr
vcr
vrc
vt
c
=0
noreaction
=0
symmetry
=0
symmetry
=0=0symmetry
=0
“Sparingly” soluble→ stagnant
surroundings
r
cr
rr
D
t
c12
2
1
18
5‐19
3. Step: Boundary conditions
t=0 r c1=0t>0 r=R0 c1=c1(sat)
r= c1=0 (meaning that the drug is consumed at thegut wall)
This is easily solved by introducing a variable =c1r and the equation reduces to that describing unsteady diffusion through a semi-infinite slab, so
tD4
Rrerf1
r
R
)sat(cc 00
1
1
19
5‐20
Thus the flux for a sparingly soluble pill:
stateunsteady
0
0
1111
tD
R1
R
)sat(cD
r
cDjn
0Rr
Now you can calculate how long you need to reach steady-state by calculating when the term
11.0tD
R0
Say that the pill is 3 mm and the D=10-5 cm2/s (typical for D in liquids) then t80 hours.
This is clearly wrong. We know that usually a pill acts within 10-20 minutes. So where is the mistake? Revisit the approximations and think looking at slide 2-24…..
20
5‐21
Example: Diffusion through a Polymer Membrane
A diaphragm-cell separates olephins (ethylene) from aliphatic hydrocarbons. Upper compartment: vacuum, lower: ethylene.
Measure the ethylene concentration as a function of time. Find the diffusivity.
1. Step:Select appropriate mass balance
12
12
2
12
2
1
10z
10
10r
1
rz
cc
r
1
r
cr
rr
1D
z
cv
c
r
v
r
cv
t
c
21
5‐22
Under which other assumptions this eqn. becomes this?
2. Step: Simplify mass balance
21
21
zc
Dt
c
3. Step: Boundary conditions for the polymer film:
t=0 z c1=0t>0 z=0 c1=H p0
z=L c1=H pL0
where L is the film (membrane) thickness and H the partition coefficient relating ethylene conc’n(partial pressure) in the gas to ethylene conc’n in the membrane.
22
Now the solution to this equation is (J. Crank “The mathematics of diffusion”, 2nd ed., 1975, p.50 eqn. 4.22):
5‐23
2
22
1n0
1
L
tnDexp
n
L/znsin2
L
z1
pH
c
This is the concentration profile c1(z,t). We need to bring this result into a convenient form for experimental measurements.
Make a mass balance or better, a mole balance as we are dealing with gases. Start from the ideal gas law:
dtdp
TRV
dtdN1
(1)
moles enteringupper compartment at
z=LLz
1Lz1 z
cDAjA
(2)
23
Substitute (1) into (2) and integrate for p assuming that at t=0 p=0 at upper compartment.
2
22
1n22
20
L
tnDexp1
n
ncosLH2tDH
LV
pTRAp
5‐24
At long times the exponential term 0
6
LHtDH
LV
pTRAp
2
known
0
Thus the intercept is related to Henry’s constant while the slope is related to DH, the permeability of ethylene through the membrane.
24
From another point of view now: As you can obtain H from the intercept, you can estimate D from the slope!
This is an elegant problem combining verification (extraction) of both equilibrium (H) and transport (D) properties!
The challenge of an engineer is to QUANTITATIVELY describe the phenomenon, e.g. connect the math to the physical problem.