Liquid-liquid extraction

Basic principles
In liquid-liquid extraction, a soluble component (the solute) moves
from one liquid phase to another. The two liquid phases must be
either immiscible, or partially miscible.
usually isothermal and isobaric
can be done at low temperature (good for thermally fragile
solutes, such as large organic molecules or biomolecules)
can be very difficult to achieve good contact between poorly
miscible liquids (low stage efficiency)
extracting solvent is usually recycled, often by distillation
(expensive and energy-intensive)
can be single stage (mixer-settler) or multistage (cascade)
Extraction equipment
Batch:
mixer-settler
column:
separatory funnel
rotating-disk contacter
a. agitator; b. stator disk
single-stage:
Continuous:
Mixer-settler column
Design

Mixer-settlers, both as stand-alone and as in-
column type, are offered for special
applications. As implied by the name, the
mixer-settler-column is a series of mixer-
settlers in the form of a column. It consists of a
number of stages installed one on top of the
other, each hydraulically separated, and each
with a mixing and settling zone (see below).
This design enables the elimination of some of
the main disadvantages of conventional mixer-
settlers, whilst maintaining stage-wise phase
contact.
The mechanical design of the mixer-settler-
column is comparable to the agitated ECR Khni
column.
Key characteristics
For long residence times: >15 min
Extraction controlled by residence time
Reactive extraction systems
Long phase separation
For extraction controlled by pH (stage-
wise pH adjustment)
For batch extraction
Stage-wise phase contact
Agitated column
Packed extraction column
The ECP packed column is based on current state-of-the-art extraction research. The column consists of a packed bed, distributors for the two liquid phases and, optionally, a pulsator.
Packing
The special Sulzer extraction packing reduces the back-mixing of the continuous phase, thereby providing nearly plug flow conditions and a narrow droplet size distribution. These ensure the high throughput and high efficiency of your process. To further increase the separation performance, additional dual flow perforated plates can be inserted between the packing elements.
Liquid distributors
In order to create an even liquid flow velocity profile at either end of the packed bed, both liquid phases are distributed over the whole cross sectional area, by suitable distributors. The design of the dispersed phase distributor requires special attention in order to achieve a narrow drop size distribution. When operated without external energy input (unpulsed), the packed extraction column is a static apparatus without any moving parts.
Main benefits
High specific throughput facilitation:
Small column diameters
Revamp of existing columns to increase capacity
Use in cases of difficult physical properties:
Low density difference < 50 kg/m3
Low interfacial tension: < 2 mN/m
Tendency to form emulsions
Reliable scale-up
Stream labeling
1
N
F, x
A,0
S, y
A,N+1
E, y
A,1
R, x
A,N

Usually specified:
y
A,N+1
, x
A,0
, F
D
/F
S
and x
A,N
.
Feed (F) contains solute A (x
A
) dissolved in
diluent D (x
D
= 1 x
A
).
Solvent (S) extracts A (y
A
), creating the product
extract stream (E). The depleted feed becomes
the product raffinate stream (R).
Equilibrium (no longer VLE!) is defined by the
distribution ratio, K
d
:


K
d
= y
A
/x
A

Note that y
A
does not refer to gas composition.

F R
diluent flow rate
= F
D
= constant
S E
solvent flow rate
= F
S
= constant
feed

mixture

extract

raffinate

mixer settler
solvent

McCabe-Thiele analysis:
Counter-current extraction with immiscible liquids
N = 3


(X
0
,Y
1
)

(X
N
,Y
N+1
)



1

2

3
X
0
(F
D
/F
S
)
max
gives F
S,min
for N = .
Can also use Kremser eqns, if solutions
are dilute and equil. line is straight.
For dilute solutions,



y =
R
E
x +(y
1

R
E
x
0
)


Y =
F
D
F
S
X +(Y
1

F
D
F
S
X
0
)
Equation of the operating line:
(analogous to operating line for
stripper column).

N =
ln 1
mE
R

y
N+1
y
0
y
1
y
0

+
mE
R

ln
R
mE
( )
Cross-flow cascade
Increase overall efficiency by introducing
fresh extracting solvent at each stage.
Each stage has its own mass balance and
operating line
Uses much more solvent than counter-
current cascade (requires much more
solvent recovery)
A mixer-settler is just one cross-flow stage.

From Separation Process Engineering, Third Edition by Phillip C. Wankat
(ISBN: 0131382276) Copyright 2012 Pearson Education, Inc. All rights reserved.
Figure 13-8 Cross-flow cascade
N = 3



y
j
=
R
E
j
x
i
+(y
j ,in
+
R
E
j
x
j 1
)
From mass balance around stage j:
(x
0
,y
1,in
)

(x
1
,y
2,in
)

(x
2
,y
3,in
)


x
3

(x
1
,y
1
)


(x
2
,y
2
)


(x
3
,y
3
)

Dilute fractional extraction
A common situation:
the feed contains two important
solutes (A, B), and we want to
separate them from each other.

Choose two solvents:
A prefers solvent 1 (extract)
B prefers solvent 2 (raffinate)
K
d,A
= y
A
/x
A
> 1
K
d,B
= y
B
/x
B
< 1

1
N
F
z
A
z
B
solvent 1
y
A,N+1
= 0

y
B,N+1
= 0

solvent 2
x
A,0
= 0

x
B,0
= 0

extract
y
A,1
y
B,1
raffinate
x
A,N
x
B,N
E R


E


R
a
b
s
o
r
b
i
n
g


s
e
c
t
i
o
n

s
t
r
i
p
p
i
n
g


s
e
c
t
i
o
n

One operating line for each solute i, in
each section of the column (i.e., 4 total).
McCabe-Thiele analysis: dilute fractional extraction

x
A,N



5

3
y
A,1

2
1
6


N
F
= 4,
feed stage



If y
A,1
and x
A,N
are specified, and N
F
is
known, use M-T diagram to obtain N, then
use trial-and-error to find x
B,0
and x
B,N+1
If y
A,1
and x
B,N
are specified, vary N
F
(trial-and-error)
until N is the same for both solutes.

Operating lines intersect at feed
composition (not shown, may be
very large).


y
i
=
R
E
x
i
+(y
i,1

R
E
x
i,0
)
Top operating lines (absorbing section):

y
i
=
R
E
x
i
+(y
i ,1

R
E
x
i ,0
)
Bottom operating lines (stripping section):
Equilibrium data is different for each
solute (use separate McCabe-Thiele
diagrams!)
Center-cut extraction
When there are 3 solutes: A, B and C,

and B is desired
(A and C may be > 1 component each)

solvent 1
solvent 2 solvent 1
+ A
solvent 2
+ B + C
solvent 3
solvent 2
solvent 3
+ B
solvent 2
+ C
F
z
A
, z
B
, z
C
Requires two columns:
column 1 separates A from B+C
column 2 separates B from C
Requires three extracting solvents:

A prefers solvent 1 over solvent 2
B, C prefer solvent 2 over solvent 1

B prefers solvent 3 over solvent 2
C prefers solvent 2 over solvent 3
Partially miscible solvents
There are two liquid phases
Each phase is a ternary (3-component)
mixture of solute A, diluent D and
solvent S
Ternary equilibrium diagrams have 3
axes: usually, mole or mass fractions of
A, D, and S
Literature data is commonly presently
on an equilateral triangle diagram (note
NO origin)

From Separation Process Engineering, Third Edition by Phillip C. Wankat
(ISBN: 0131382276) 2012 Pearson Education, Inc. All rights reserved.
Figure 13-14 Effect of temperature on equilibrium of
methylcyclohexane-toluene-ammonia system from
Fenske et al., AIChE Journal, 1,335 (1955), 1955, AIChE
Each axis is bounded 0 x 1
Miscibility boundary = equilibrium line
(depends on T, P)
Consider the point M:
water content (x
A
) is ?
ethylene glycol content (x
B
) is ?
furfural content (x
C
) is ?
0.19
0.20
0.61
Reading ternary phase diagrams
Read the mole/mass fraction of each
component on the axis for that component,
using the lines parallel to the edge opposite the
corner corresponding to the pure component.
A 2-component mixture of furfural and water is partially miscible over the composition
range from about 8 % furfural to 95 % furfural. Separation by extraction requires a
furfural/water ratio in this range (otherwise single phase).
The mixture M lies inside the miscibility
boundary, and will spontaneously separate
into two phases. Their compositions (E and
R) are given by the tie-line through M.
The compositions of E and R converge at the plait point, P (i.e., no separation).
region of partial miscibility A-C
check: x
A
+ x
B
+ x
C
= 1



Right-triangle phase diagrams
Raffinate (diluent-rich): x
A
+ x
B
+ x
C
= 1
Extract (solvent-rich): y
A
+ y
B
+ y
C
= 1

We need to specify only two of the
compositions in order to describe each
liquid phase completely .
From Separation Process Engineering, Third Edition by Phillip C. Wankat
(ISBN: 0131382276) Copyright 2012 Pearson Education, Inc. All rights
reserved.
Figure 13-12 Equilibrium for water-chloroform-acetone at 25C and 1 atm
This can be shown on a right-triangle phase
diagram, which is easy to plot and read.
raffinate compositions are
represented by coordinates (x
A
, x
B
)
extract compositions are
represented by coordinates (y
A
, y
B
)

More tie-lines can be obtained by trial-and-
error, using the conjugate line.
Ex.: find the tie-line that passes through M.
Vertical axis corresponds to both x
A
and y
A.

Horizontal axis corresponds to both x
B
and y
B

Q: Where does pure C appear on this diagram?

Obtaining the conjugate line







Each point on the conjugate line is composed of
- one coordinate from the extract side of the
equilibrium line
- one coordinate from the raffinate side of the
equilibrium line
On this graph, which component is the diluent? which is the solute?
Hunter-Nash analysis of mixer-settler
F

M

E

R

mixer settler
S

Overview of solution using RT
diagram:
1. Plot F and S and join with a
line.
2. Find mixing point, M, which
is co-linear with F and S.
3. Find tie-line through M; find
E and R at either end (co-
linear with M).
4. Find flow rates of E and R.
mixing line
tie-line
F
S

M
Flow rates of E and R are related
by mass balance.
Compositions of E and R are also
related by equilibrium.
Why does F appear on or
near the hypotenuse?
Why does S appear at or
near the origin?
E

coord.:
(y
D
,y
A
)

R
coord.:
(x
D
,x
A
)
Co-linearity
Why are F, S and M co-linear on the Hunter-Nash diagram?
F

M

mixer
S

TMB: F + S = M
CMB
A
: Fx
A,F
+ Sx
A,S
= Mx
A,M
= (F + S)x
A,M
CMB
D
: Fx
D,F
+ Sx
D,S
= Mx
D,M
= (F + S)x
D,M

solve for coordinates of M: (x
A,M
, x
D,M
)


x
A,M
=
Fx
A,F
+Sx
A,S
F +S

x
D,M
=
Fx
D,F
+Sx
D,S
F +S

F
S
=
x
A,M
x
A,S
x
A,F
x
A,M
=
x
D,M
x
D,S
x
D,F
x
D,M
slope from
M to S
slope from
F to M
F (x
D,F
, x
A,F
)
S (x
D,S
,x
A,S
)

M (x
D,M
, x
A,M
)
Therefore F, S and M are co-linear. To locate
M on the FS line: calculate either x
A,M
or x
D,M
.

x
A,M
x
A,S
x
D,M
x
D,S
=
x
A,F
x
A,M
x
D,F
x
D,M
rearrange
CMB
A
CMB
D
The lever-arm rule
S

M

F

Another way to locate M:

MF

MS

F
M
=
x
A,M
x
A,S
x
A,F
x
A,S
=
MS
FS
Fx
A,F
+ Sx
A,S
= Mx
A,M

Fx
A,F
+ (M F)x
A,S
= Mx
A,M

F(x
A,F
- x
A,S
) = M(x
A,M
- x
A,S
)

R
M
=
x
A,M
y
A,E
x
A,R
y
A,E
=
ME
RE
M = R + E
Your choice! Use mass balances, or
measure distances and use lever-arm rule.
similar triangles
E

M

R

To calculate flow rates E and R:

EM

MR
similar triangles
Hunter-Nash analysis of cross-flow cascade
1
F = R
0
R
1
S
1
E
1
2
S
2
E
1
R
2
F


S
E
2


R
2

Treat each stage as a mixer-settler.

each R
i
, S
i
pair creates a mixing line
find each E
i
, R
i
pair using a tie-line
E
1


R
1


M
1

M
2
Hunter-Nash analysis of counter-current cascade
F

M

E
N
R
1
mixer separator
(column)
S

Overview of solution using RT
diagram:
1. Plot F and S and join with a line.
2. Find mixing point, M, which is co-
linear with F and S.
3. Plot specified x
A,1
on raffinate
side of equilibrium line to find R
1
.
4. Extrapolate R
1
M line to find E
N
.
5. Find flow rates of E and R.
mixing line
NOT a
tie-line
F
S
E and R are both points on the
equilibrium line. But they are not
related by the same tie-line.
E
N


M

R
1

x
A,1


Stage-by-stage analysis
1
N
F = R
N+1

x
A,N+1
S = E
0

y
A,0
E
N

y
A,N+1
R
1
x
A,1

R
2
E
1
stage 1 TMB: E
0
+ R
2
= E
1
+ R
1


E
0
R
1
= E
1
R
2
= E
2
R
3
etc.

constant difference in flow rates of passing streams
= E
j
R
j+1
= constant
stage 1 CMB
A
: E
0
y
A,0
+ R
2
x
A,2
= E
1
y
A,1
+ R
1
x
A,1


E
0
y
A,0
R
1
x
A,1
= E
1
y
A,1
R
2
x
A,2
= etc.

constant difference in compositions of passing streams
net flow of A: x
A,

= E
j
y
A,j
R
j+1
x
A,j+1

net flow of D: x
D,

= E
j
y
D,j
R
j+1
x
D,j+1

The difference point
does not necessarily lie inside the RT graph.

All pairs of passing streams E
j
, R
j+1
are co-linear with .

Using the -point to step off stages on Hunter-Nash diagram:
using the specified location of R
1
(as x
A,1
), can find E
1
(use tie-line);
given the location of E
1
, can find R
2
(use );
given the location of R
2
, can find E
2
(use tie-line);
given the location of E
2
, can find R
3
(use );
and so on, until desired separation is achieved.

First, need to locate . It may be on either side of the Hunter-Nash diagram.

x
A,
=
E
0
y
A,0
R
1
x
A,1

x
D,
=
E
0
y
D,0
R
1
x
D,1

Define a difference point, , with coordinates (x

A,
, x
D,
):

Procedure:

1. Plot F (= R
N+1
), S = (E
0
). Locate M.

2. Plot R
1
and locate E
N
.

3. Extend the lines joining E
0
-R
1
,
and E
N
-R
N+1
, to find at the
intersection point.
Finding the -point
last mixing line
E
N


M

R
1

first mixing line

S = E
0

F = R
N+1

4. All intermediate mixing lines
must pass through .
Stepping off stages on the H-N diagram
Procedure:

1. Use R
1
and conjugate line to find E
1

S = E
0

F = R
N+1

Stop when you reach or pass E
N
.
N = 3

E
N

E
1

E
2


R
1


R
2


R
3

2. Use E
1
and D-point to find R
2
3. Use R
2
and conjugate line to find E
2
4. Use E
2
and D-point to find R
3
3. Use R
3
and conjugate line to find E
3
equilibrium line ends at P
Using McCabe-Thiele diagram instead of Hunter-Nash
M-T diagram can be used with much greater accuracy than H-N diagram
Need to transfer ternary equilibrium data from RT diagram
Need to obtain the operating line
Transferring equilibrium data from RT diagram
y
A
x
A
A
D
0
0
1
1






P


raffinate
compositions
extract
compositions
Each tie-line represents a pair of equilibrium streams
extract composition represented by y
A
raffinate composition represented by x
A
Each (x
A
, y
A
) pair is a point on the M-T equilibrium line






P
1
x
A
y
A
0
0 1
x
A
y
A
0
0
1
1






P
Obtaining the M-T operating line
1
N
F = R
N+1

x
A,N+1
S = E
0

y
A,0
E
N

y
A,N+1
R
1
x
A,1

A
D
x
1

E
N
y
0
y
N
R
N+1
E
0


x
N+1
Mixing lines represent passing streams.
All mixing lines lie between the limits:
(x
1
, y
0
) and (x
N+1
, y
N
)
Note: passing streams are (x
j+1
, y
j
) instead of
(x
j
, y
j+1
) as in distillation, simply due to our
labeling convention (feed enters at stage N).

(x
1
, y
0
)

(x
N+1
, y
N
)

R
1
M


WAIT! In general, operating line is
not straight.
Plot arbitrary intermediate mixing
lines to obtain more points.


Choice of extracting solvent flow rate
As S increases, separation improves, but extract becomes more dilute
As S decreases, N must increase to maintain desired separation
S
min
achieves the desired separation with N =
A
D S

F

M


M
min

M
max
as M moves towards S, (S/F) increases
(lever-arm rule)
when M reaches the equilibrium line, all
feed dissolves in extracting solvent (M
max
)
as M moves towards F, (S/F) decreases
before reaching the equilibrium line, there is
usually a pinch point (M
min
)
It is not easy to locate this pinch point on a McCabe-Thiele diagram, since the
operating line curvature changes as S changes.
On a Hunter-Nash diagram, D
min
(corresponding to M
min
) occurs when a mixing
line and a tie-line coincide.
E
N,min

S
F
R
1

Minimum solvent flow rate

min

M
min
1. Plot S = E
0
, F = R
N+1
, R
1

2. Join S and F
3. Extend SR
1
mixing line
4. Locate several tie-lines
5. Extend tie-lines to the SR
1

mixing line
6. Tie-line with furthest intersection
from S locates D
min
7. Mixing line from D
min
through F
locates E
N,min
8. Connecting R
1
and E
N,min

completes the mass balance
9. M
min
is located at the intersection
of SF and R
1
E
N,min





10. (S/F)
min
= (FM
min
)/(SM
min
)
On H-N diagram whose tie-lines have negative slopes:
Rule-of-thumb: (S/F)
act
~ 1.5 (S/F)
min
Minimum solvent flow rate
Strategy:
1. Plot S = E
0
, F = R
N+1
, R
1

2. Join S and F
3. Extend SR
1
mixing line
4. Locate several tie-lines
5. Extend tie-lines to SR
1

mixing line
6. Find tie-line which gives
closest intersection to S;
this locates D
min
7. Draw mixing line from D
min

through F to locate E
N,min
8. Connect R
1
and E
N,min
to
complete mass balance

S
F
R
1

min
E
N,min

M
min
9. M
min
is at the intersection of SF and R
1
E
N,min
10. (S/F)
min
= (FM
min
)/(SM
min
)
On H-N diagram whose tie-lines have positive slopes:
Two feed counter-current column
1
N
F
1
= R
N+1
E
0
= S R
1
E
N
F
2
E R


E


R
Feed balance: F
1
+ F
2
= F
T
S

F
2
F
1
F
T
M

E
N
R
1
mixer 1 mixer 2 separator
Overall balance:
hypothetical mixed feedstream FT is co-linear with F
1
, F
2

Stage-by-stage analysis:
mass balance changes where F
2
enters the column
upper and lower sections have different sets of operating
lines different D-points


Hunter-Nash analysis of 2-feed column
Overall balance:

1. Plot F
1
and F
2
. Locate F
T
(co-
linear with F
1
and F
2
).

2. Plot S . Locate M (co-linear with
S and F
T
).

3. Plot R
1
. Locate E
N
(co-linear with
R
1
and M).

1. Calculate flow rates R
1
and E
N
.
E
N


M

R
1


S = E
0

F
T


F
2

F
1


Stage-by-stage analysis
Balance around top of column:
R
1
E
0
= R
j+1
E
j
= D
1
R
1
, E
0
, D
1
are co-
linear


Note: D
1
and D
2
may be on different sides of the phase diagram.
1
N
F
2
F
1
= R
N+1
E
0
= S R
1
E
N
E R
j


E


R
k
Balance around bottom of column:
E
N
R
N+1
= E
k
R
k+1
= D
2
R
N+1
, E
N
, D
2
are co-linear

Overall balance:
F
2
+ R
N+1
+ E
0
= E
N
+ R
1
F
2
= (E
N
R
N+1
) + (R
1
E
0
) = D
1
+ D
2


F
2
, D
1
, D
2
are co-linear feed-line
D
2
is located at the intersection of two mixing lines:
R
N+1
, E
N
, D
2
and F
2
, D
1
, D
2

Need another line to locate D
1
:
TMB: F
T
= F
1
+ F
2
= E
N
+ (R
1
E
0
) = E
N
+ D
1
F
T
, E
N
, D
1
are co-linear
D
1
is located at the intersection of two mixing lines:
R
1
, E
0
, D
1
and F
T
, E
N
, D
2

3. Step off stages, initially using D

1
to
generate the first mixing lines

Using the feed-line

feed line

E
N

S
F
T
R
1

F
2

F
1
1. Locate D
1
at intersection of R
1
E
0

and E
N
F
T


2. Locate D
2
at intersection of F
2
D
1

and E
N
R
N+1
5. When the tie-line crosses the feed
line, the next mixing line will be
generated using D
2

4. Identify the optimum feed stage
when the mixing line crosses the feed
line, F
2
D
1
D
2


E
1

E
2


R
2

Countercurrent liquid-liquid extraction with reflux
1
N
F
1
= R
N+1
x
A,N+1
E
0
= S R
1
E
N

y
A,N

E R
In a conventional liquid-liquid extraction column:
y
A,N
is related by equilibrium to x
A,N


x
A,N
depends on x
A,N+1
dilute feed gives dilute extract
highest y
A,N
obtained with S S
min
, but this requires very large N
How to increase y
A,N
?
need to increase x
A,N+1

make R
N+1
an reflux stream
1
N
F

R
N+1
reflux
E
0
R
1
E
N
E R


E


R
P
E

product extract
makeup solvent
Turning extract into raffinate :
extract is mostly solvent
raffinate is mostly diluent
Q

recovered
solvent
S
R
solvent
separator
extract reflux
(no benefit to raffinate reflux)
We need to remove solvent,
e.g., distillation, stripping
Analogy to distillation reflux
1
V
1
L
0
D

Saturated liquid reflux stream is
obtained by condensing V
1
(vapor
stream rich in A) to give L
0
(liquid
stream rich in A)

External reflux ratio = L
0
/D

Internal reflux ratio = L/V
N
R
N+1
reflux
E
N
P
E

product extract
Q

recovered
solvent
S
R
solvent
separator
Extract reflux stream is obtained by removing
solvent from E
N
(extract stream rich in A and
solvent) to give R
N+1
(raffinate stream rich in A
and depleted in solvent)

External reflux ratio = R
N+1
/P
E

Internal reflux ratio = R
N+1
/E
N

Stage-by-stage balances
Similar to 2-feed liq-liq extraction column:
- two D-points (mass balance above and below feed stage)
- if F, E
0
, R
1
and R
N+1
are specified, same stage-by-stage analysis
But R
N+1
is an internal stream, usually not specified.

Usually specified:
F, x
A,F
, x
D,F
plot F

y
A,0
, y
D,0
plot E
0
x
A,1
plot R
1
on satd raffinate curve
x
A,PE
, x
D,PE
plot P
E
(same location as R
N+1
and Q, different flow
rates)
y
A,SR
, y
D,SR
plot S
R
R
N+1
/P
E


F
T
= F + R
N+1
cant locate F
T
(or E
N
) because we dont know R
N+1
Mass balance: solvent separator
P
E

Q

S
R
solvent
separator
R
N+1
E
N
E
N
is co-linear with Q and S
R
.
E
N
also lies on satd extract line.

E
N
S
R

S
R
P
E
=
R
N+1
P
E
+1+
S
R
P
E

E
N
S
R
=
R
N+1
S
R
+
P
E
S
R
+1
E
N
= Q + S
R




S
R
P
E
=
R
N+1
P
E

+1
E
N
S
R

1

R
N+1
S
R
=
E
N
S
R

P
E
S
R
1
Obtain E
N
/S
R
from lever-arm rule.
We will also need R
N+1
/S
R
:
= R
N+1
+ P
E
+ S
R


dont know
A
D

E
0
F

S
R


R
1

E
N

P
E
, Q, R
N+1


Finding the D-points
D
2
= E
N
- R
N+1

D
2
x
A,D2
= E
N
y
A,N
- R
N+1
x
A,N+1




R
1

Locate D
1
at the intersection
of two mixing lines:
D
1
= E
0
- R
1

F = D
1
+ D
2


x
A, 2
=
E
N
y
A,N
R
N+1
x
A,N+1
E
N
R
N+1
We dont know the individual
flow rates E
N
, R
N+1
, but we know
E
N
/S
R
and R
N+1
/S
R
. We can
calculate x
A,D2
and thereby locate
D
2
on the E
N
R
N+1
line.
Proceed to step off stages.

D
1

F


P
E
, Q, R
N+1


D
2

E
0


E
N


S
R