Membranes II

Model Equations
Simple well-mixed module, solution-diffusion
model, reverse osmosis, gas permeation,
pervaporation

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Outline

1. Simple well-mixed module

2. Solution-diffusion model equations

3. Reverse osmosis & water desalination

4. Gas permeation & process design considerations

5. Pervaporation & process design considerations

6. Comparison

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Simple well-mixed module
Assumptions
• No chemical reactions
• No pressure drop
∞ • Well-mixed in both sides

∞
Nomenclature
• n:ሶ molar flow rate
• yi: molar fraction of species i

Control volume Total mass balance :

Single species mass balance :

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Simple well-mixed module
Assumptions
• No chemical reactions
• No pressure drop
∞ • Well-mixed in both sides

∞
Nomenclature
• n:ሶ molar flow rate
• yi: molar fraction of species i

Control volume Total mass balance :

• J is the total molar flux of permeating species:
• A is the membrane section area where mass transfer occurs

Single species mass balance :

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Solution-diffusion model

Porous membranes (pore diameter > 10Å) Dense membranes (pore diameter < 10Å)

The pore-flow model is applicable. The The solution-diffusion model is applicable.

different compounds are transported by a The different compounds first dissolve in the
pressure-driven convective flow through tiny membrane matrix and then diffuse through the
membrane pores membrane under a concentration gradient
=> Separation achieved due to differences in => Separation achieved due to differences in
the steric hindrances between the compounds the solubility of each compound into the
molecules and the membrane material membrane material and their diffusivity
through the membrane

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Solution-diffusion model
p
Assumptions
• Isothermal process
pF = pR Discontinuity in the pressure
• Constant pressure within the
membrane, equal to the
retentate pressure
• Components dissolved in pP
dense membranes act as if
they were a liquid x
δ

The flux happens due to a concentration gradient. Assuming the validity of Fick’s Law within the
membrane, one finds :
Integration over the
membrane thickness

where wi is the molar fraction of species i dissolved in the membrane

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Solution-diffusion model
Since dissolved components in dense membranes act as if they were liquid, the fugacity of species i inside of
the membranes is given by :

Where :
• is the pure liquid fugacity of species i
• is the Poynting factor of species i
• is the vapor pressure of species i

By replacing wi,R and wi,P in the expression of the permeation flux using the definitions of the fugacities, one
can obtain the final expression of the solution-diffusion model :

Apply the iso-fugacity equilibrium condition at the fluid/membrane interface to obtain the
permeating flux of species i as a function of the retentate and permeate concentrations
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Reverse osmosis
Osmosis (reverse or normal) refers to the use of a permselective membrane (membrane that is freely
permeable to water but much less permeable to salt) in order to separate a salt solution from pure water.
More generally, to derive the equations, we have to consider the separation of liquids.
Solution-diffusion model :
Assumptions
• Local equilibrium at interfaces
• No mass transfer resistance in
the retentate and permeate Fugacities of the gases at the membrane interface and local
streams (bulk concentrations equilibrium condition:
are equal to membrane
interface concentrations)

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Reverse osmosis
The equation is simplified by assuming that the ratio between the activity coefficient is the same at the
retentate or at the permeate side:

: Sorption coefficient of
species i in the membrane
: diffusion coefficient of species i in the
membrane

: permeability of species i
in the membrane (extent
at which a species
dissolves and diffuses
through a membrane)
: permeance of species i in
the membrane

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

 Application: water desalination
At t=0s, we have side (1) that has salt water and side (2) that has only
water. If both p1 and p2 are atmospheric pressure, then we have :

p1 p2

The osmotic pressure Δπ is the pressure difference necessary to

t = 0s have a null water flux through the membrane, due to the
Δπ
concentration differences in the two sides of the membrane :
t = ∞s

Salt + H2O H2O

(1) (2)
Jw • p1 – p2 < Δπ : Normal osmosis, water flows from the pure water
side to the sea-water side
• p1 – p2 = Δπ : Osmotic equilibrium, no flux since the osmotic
pressure is counterbalanced by the pressure difference
• p1 – p2 > Δπ : Reverse osmosis, water flows from the sea water
side to the pure water side : water desalination
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Gas permeation
Gas permeation corresponds to the separation of gases. Typically, a high pressure gas mixture at pR is
fed to one side of the membrane while the permeate is removed on the other side at a lower pressure pP
Solution-diffusion model :
Assumptions
• Local equilibrium at interfaces
• No mass transfer resistance in
the retentate and permeate Fugacities of the gases at the membrane interface and local
streams (bulk concentrations equilibrium condition:
are equal to membrane
interface concentrations)

: sorption coefficient of species i in the membrane

: diffusion coefficient of species i in the membrane permeability of species i
in the membrane
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Diffusion coefficient, sorption coefficient and permeability
In polymeric membranes, Di, Ki(g) and Qi typically depend on the molecular volume:

Diffusion coeff. decreases Sorption coeff. increases Permeability is proportional to the

with molar volume (larger with molar volume (larger product of Di and Ki. For natural rubber
molecules interact more molecules are usually membranes sorption dominates, for
with the polymer chain) more condensable) glass membranes diffusion dominates.

Rubbery polymer: Qi
increasing with Vm

Glassy polymer: Qi
decreasing with Vm

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Gas permeation
Assuming the equality of sorption coefficients respectively on the retentate and permeate sides, and
considering that the Poynting correction factor is close to 1 for systems at relatively low pressure, we
have:

Where ; : Permeance of species i in the membrane

Selectivity: Pressure ratio:

Permeate concentrations from the permeating fluxes :

For a binary mixture (A+B), the equation becomes quadratic with respect to y1,P and can be solved:

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Simple well-mixed module, binary gas mixture

∞ Assumptions
• No chemical reactions
• No pressure drop
∞ • Well-mixed in both sides

For a binary mixture A-B ,we do not write the subscripts for the compounds anymore, but consider that
yA,P=yP and yA,R=yR. The complete model equations are :

Known parameters (3) :

Unknown parameters (8) :

Equations (6)

2 degrees of freedom

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Design considerations gas separation
We define characteristic dimensionless parameters for a membrane separation :

Selectivity: Pressure ratio: Stage cut:

Recovery: Purity:

Given the operating conditions (pressures, retentate stream properties, flow patterns), the membrane
technology (permeability, selectivity, thickness) and the upstream conditions (feed stream properties),
the process performance (recovery, purity and required membrane area) can be calculated.
Two degrees of freedom : in the design of a membrane
process, two parameters can be fixed (e.g: Re and Pu)
For the simple well-mixed module :

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Design considerations : operating conditions

Let : Selectivity limited region

Pressure ratio limited region

(φ >> α) selectivity limited (α >> φ) pressure-ratio limited

16
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Design considerations : membrane technology

Selectivity and permeability (in Polymeric Membrane

Materials) have a pareto-optimum, obtained from
empirical observations.

For materials close to the

Upper Bound, it is not
possible to increase
simultaneously the
selectivity and the
permeance.
=> Trade-off between
increasing the purity or
log Pi  log k  m log i , j
the stage-cut
(P is obtained in [barrer] )

Robenson, L. M., The Upper Bound Revisited, 2008.

17
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Design considerations : process performance
The recovery and purity maps on the A – β shows
the limitations of a one stage membrane process :

• Recovery and purity behave oppositely against

the membrane area. The recovery increases
while the purity decreases due to the
permeation of the least permeant component
along the module.
=> High permeate purity is only achievable at
low recovery rate, which is not economical

• An increase in selectivity (decrease in

permeability) leads to an increase in purity
(decrease in recovery).

• For large selectivity and pressure ratio, the

permeate purity is almost independent from the
membrane area

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Applications

19
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Pervaporation
Pervaporation can be considered as an intermediate separation process between gas separation and
reverse osmosis processes : the retentate is in liquid phase and the permeate is in gas phase. The
feed pressure is sufficient to maintain liquid state on the retentate side even as the composition is
changing. The permeate pressure is fixed to a value below the saturation pressure of the permeating
mixture, often applying vacuum.
The isothermal assumption will be assumed for simplicity, but in reality an energy balance equation should
be coupled with material balances to account for the heat of vaporization.

Solution-diffusion model :
Assumptions
• Local equilibrium at interfaces
• No mass transfer resistance in
the retentate and permeate Fugacities of the gases at the membrane interface and local
streams (bulk concentrations equilibrium condition:
are equal to membrane
interface concentrations)

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Pervaporation
The solution-diffusion model equation for pervaporation becomes:

: gas sorption coefficient of species i in the membrane

: liquid sorption coefficient of species i in the membrane
: diffusion coefficient of species i in the membrane

: Henry constant solubility for species i (liquid-vapor equilibrium property)

: gas permeability of species i in the membrane

Considering that the Poynting correction factor is close to 1 for systems at relatively low pressure,
we have:

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Pervaporation: Separation Factor
We can define the vapor pressures of the liquid retentate (pi,R) and of the permeate (pi,P) :

The separation factor indicates how well was the component i separated compared to component j.
To note that we have :
In pervaporation, the separation
factor is a product of two terms: One
represents the contribution of
relative volatility, and the other the
contribution of the membrane.

Now we have three factors: the

contribution of relative volatility, the
selectivity of the membrane and the
vapor pressures of feed and
permeate (operating parameters).

22
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Separation Factor: Limiting Scenarios

1. High selectivity scenario α >> (pR/pP) : 3. Low pressure ratio, low selectivity:
Species i is preferred and dilute :

All these effects are dominated by the

fact that α is not constant, but usually
a strong function of yR:
α = α (yR)
2. High pressure ratio scenario (pR/pP) >> α Hence it is possible to have both
:If strong vacuum is applied: α>>(pR/pP) and (pR/pP)>>α for the
same membrane and ratio pR/pP,
simply by varying yR.

23
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Pervaporation: Combined Effects

Hydrophilic membrane material:

• High Henry‘s constant for water,
low Henry‘s constant for Ethanol
• Strong swelling if yH2O,R >> yEtOH,R
• High permeance and rather low
selectivity if yH2O,R >> y EtOH,R
• No swelling if yEtOH,R >> yH2O,R
• Low permeance and rather high
selectivity if yEtOH,R >> yH2O,R
This leads typically to y(x) curves as
the ones shown in the figure (for
different permeate pressures)

24
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Design considerations for pervaporation processes

• The minor component should permeate

(always favorable in membrane separation)
• To benefit from high selectivities, one should start from highly concentrated feeds
(hence hybrid processes are favorable, e.g. distillation + membrane separation)
• Working at elevated temperatures facilitates the process
(the latent heat of vaporization has to be provided with the feed)
• The pressure ratio is usually increased by applying vacuum on the permeate side
(e.g. by cooling condensation of the permeate stream)
• Mass transfer resistance through the surface boundary layer has to be considered
(especially if high purity of the retentate stream is desired)

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Dehydration of Ethanol
• Distillation of 60% EtOH-water feed
• Products are the azeotrope (95.6%
EtOH, top) and water (bottom)
• Pervaporation of the azeotrope yields a
retentate of > 99.5% EtOH (product)
• The permeate (25% EtOH) is condensed
and recycled to the distillation column
• Similar setups for dehydration of other
azeotrope forming organic solvents

Other pervaporation processes :

• Removing VOC‘s from water using hydrophobic membranes (e.g.
silicone rubber)
• Separation of organic solvent mixtures in cases where distillation is
difficult (Methanol / MTBE), (Benzene / Cyclohexane)
26
Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations
Comparison of osmosis, gas permeation and pervaporation

Solution-diffusion model equations

pP,sat

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations

Comparison of osmosis, gas permeation and pervaporation

Separation Processes Laboratory - Prof. Mazzotti - Rate Controlled Separations