Solute Transport
Part 4. General force balances on ions and water in membranes
matrix of porous medium
ion ion ion
ion
water fills all space in the pores between the ions/solutes
ion ion ion
water water water
water
pores
feed membrane permeate
Donnan equilibrium Steric Partitioning
advection diffus ion diffus ion advection electrom igration electro-
migration
c+
c
advection
acid-bas e equilibrium
NH4+ NH3+ H+ H2CO3 HCO3+ H+
surface complexation
R R + H+
mem brane potential Passage =
feed conc.
permeate conc.
Donnan potential
pores
Fig. 7.7:Illustration of the terminology used to describe transport of solvent and solutes in a porous medium. Solutes, such as ions, flow through the pores in the porous medium (also called structure, or matrix), and all space around the solutes which is not occupied by the structure, is filled with solvent (fluid), often water. The porous medium can be a membrane, ion absorbent material, or a porous electrode. When water is the solvent, an ion will be hydrated. A bare ion together with its shell of tightly bound hydration water as a whole is called an ion. The solvent (for instance water) is βfreeβ and flows through all the space that is not occupied by the (hydrated) ions and other solutes, or occupied by the porous medium.
196 Solute Transport
of the ion with all other phases, and these depend on the velocity difference between the ion and that phase, resulting in
βπ ππ
π π₯β² +βοΈ
π
ππ
-π π£β²
πβπ£β²
π
=0 (7.64)
where the summation over π is over all phases and species, including the solvent and the porous structure, and where ππ
-π is a measure of the friction of a mole of ionsπwith all of phaseπ present. In Eq. (7.64) we introduce the notationβ² for movement following a path through a tortuous pore. Velocitiesπ£β²
π are βrealβ interstitial velocities following this path.
When phaseπis a species of which the concentration can change, such as for any of the ion types, it is relevant to make the frictional term proportional to the amount (concentration) of that species. Thus for ion-ion friction, acting between ion typeπandπ, in the force balance for ion typeπ, Eq. (7.64), ππ-πis made a linear function of the concentration of the other type of ion, π. Thus for ion-ion friction, in the force balance for speciesπ, ππ
-π is replaced by π½π
-πππ, where π½π
-π is a constant, andππ is the local concentration of the other solute. (In doing so, there will ultimately be one ion-ion friction coefficient for each ion-ion pair, i.e., π½π
-π =π½π
-π.) However, when the other phase βis simply thereβ, the use of ππ
-π as a friction factor between ionic speciesπand that other phase, suffices. This applies for the friction with the porous medium, such as a membrane, and for friction with the fluid, because it surrounds all ions and solutes. There is no βconcentrationβ to speak of for these other phases. Indeed for friction of an ion with these two phases, there is no need to consider a proportionality with the βconcentration of the porous mediumβ or with βthe concentration of waterβ. The porous medium is just there, while each ion is always completely enveloped by water molecules.
Of course, for friction of solutes with the porous medium, if we have an open structure with large pores, ππ
-mwill be much lower than for a tight structure with small pores, but for a given porous structure, ππ
-mcan be taken as a constant that describes the ion-matrix friction.
From this point onward, we first leave out ion-ion friction. Including that the porous medium has zero velocity, combination of Eqs. (7.63) and (7.64) results in
βπlnππ
π π₯β²
βπ§π
π π
π π₯β²
βπ π
exc,π
π π₯β²
βππ
π πtot
π π₯β²
=βππ
-F π£β²
Fβπ£β²
π
+ ππ
-mπ£β²
π (7.65)
where we include the excess term that describes volumetric interactions, and the insertion pressure of a solute that has volume. These last two terms on the left would be zero when we assume the ions to be point charges.
From a force balance to the Nernst-Planck equation 197
Effect of temperature on transport. We continue here a discussion of temperature effects set up for equilibrium (EDL theory) on p. 39. As explained above, the driving force on (a mole of) ions is minus the gradient of the chemical potential, and in that analysis we should multiply equations given above with the factorRTto obtain a driving force with the correct unit of N/mol, for instance in Eq. (7.65). Then the right side is also multiplied byRT and the resulting friction factorsπ π ππ
-Fandπ π ππ
-mdo not have a specific relation toRT. Below we replace ππ
-Fwith 1/π·π and thus π·π linearly increases withπ π. Thus, in a typical NP equation, such as Eq. (7.67) below, diffusion coefficients linearly depend on temperature, T, and the origin of this effect is in a temperature dependence of driving forces, not in a an intrinsic change to its mobility. In addition, there will be such an intrinsic effect because the viscosity of solvent decreases with temperature and thus mobility will go up. So a T-dependence of the diffusion coefficient combines an intrinsic increase in solute mobility, with theRT-term.
In the next step we multiply each side byππand we make the replacementπ½β²
π =π£β²
πππ, because the flux of an ion through a poreπ½β²
π is equal to its own velocityπ£β²
πtimes its concentrationππ. We make the replacement ππ
-F=1/π·πfor the ion-fluid friction, resulting in
βπ ππ
π π₯β²
βπ§πππ
π π
π π₯β²
βππ
π πexc,π
π π₯β²
βππππ
π πtot
π π₯β²
=β 1 π·π
πππ£β²
Fβπ½β²
π
+ ππ
-mπ½β²
π (7.66)
which we rewrite to an expression explicit in fluxπ½β²
π that is π½β²
π =πΎ
f,ππππ£β²
FβπΎ
f,ππ·π π ππ
π π₯β² +π§πππ
π π
π π₯β² +ππ
π πexc,π
π π₯β² +ππππ
π πtot
π π₯β²
(7.67) where (Spiegler and Kedem, 1966, Eq. 49)
πΎf,π=1/(1+ ππ
-m/ππ
-F) . (7.68)
is a factor which describes the importance of ion-matrix friction relative to ion-fluid friction, withπΎf,π β 1 for very low ion-matrix friction, andπΎf,π β 0 when ion-matrix friction is very large.
In the next step, Eq. (7.67) βin which fluxes and velocitiesπ½β²
π andπ£β²
Fare defined inside a tortuous poreβ is converted to an expression with fluxes and velocities defined on the macroscopic scale of the porous material,π£
Fandπ½π, which is the common usage of fluxes and velocities in this book. To this end, we implement thatπ½π =π½β²
πΒ·π/πwherepis porosity andπis tortuosity. The latter describes that the path through a pore is longer than straight
198 Solute Transport
across a material, because it is constantly at an ever-changing angle to a direct line that goes from one to the other side of a material. If there is a certain fluxπ½β²
π along that curvy path, it is less by a factorπ(withπlarger than 1) if we project on the direction straight through the material. The factorpfor porosity enters because fluxπ½β²
π is per unit pore and withπ½π we aim to describe the superficial flux (which is per unit geometrical, i.e., βtotalβ area).
These conversions fromπ½β²
π toπ½π also apply to the conversion fromπ£
F
β²toπ£
F. But for the fluid, also a factor 1βπenters in the conversion, because inside the pores, a volume fraction π is blocked out for the fluid.xii Thus the interstitial fluid velocity in the pore, along the pore direction,π£β²
F, relates to the superficial fluid velocity straight across the material,π£
F, by π£F=π£β²
FΒ· (1βπ) Β·π/π. The final implementation is that the coordinate axis straight through the material is shorter than the curved path, and thusπ π₯=π π₯β²/π. If we include all of this in Eq. (7.67), we obtain the intermediate result that
π½π π π
=πΎ
f,ππππ£
F
π
π (1βπ) β1 π
πΎf,ππ·π π ππ
π π₯β² +π§πππ
π π
π π₯β² +ππ
π πexc,π
π π₯β² +ππππ
π πtot
π π₯β²
. (7.69) Multiplying all sides by π/π, and making use of the variableπ =π/π2, we obtain the most general NP equationxiii,xiv
π½π = πΎ
f,π
1βπ
πππ£Fβπ πΎf,ππ·π π ππ
π π₯ +π§πππ
π π
π π₯ +ππ
π πexc,π
π π₯
+ππππ
π πtot
π π₯
(7.70) where the fraction of the pores occupied by solutes isπand the porosity (fraction of total volume that is pores) isp. Fluxes are defined per cross-section of the total material (pores plus matrix), but concentrations are defined per volume of pore phase. Thus, a balance of forces acting on solutes inside a porous medium results in a general NP-equation that includes a groupπΎ
f,π/(1βπ) β πΎβ²
c,π andπ πΎ
f,π β πΎβ²
d,π, whereπΎβ²
c,π> πΎβ²
d,π. These latter πΎβ²-values have similarities to the convective and diffusive hindrance functions derived on the basis of hydrodynamic theories for transport of spherical particles through cylindrical pores, where also the term for convection is larger than for diffusion, as used for instance in SEDE theory (see next box).
If we neglect all possible solute volume effects, Eq. (7.70) simplifies to the extended NP
xiiThe volume fractionπused here is different fromπin Ch. 4 where it relates to the volume fraction of the immobile structure. Here it relates to the fraction of the pores (that themselves have a porosityp) that contains solutes.
xiiiOften this termπ2
is replaced byπand then called the tortuosityfactor.
xivIn this book we generally present flux expressions as if there is only one direction,π₯, to make the equations more insightful. The reader familiar with flow in multiple directions will understand that a gradientπ/π π₯can be generalized, for instance by using the notationβ.
From a force balance to the Nernst-Planck equation 199
equation
π½π=πΎ
f,ππππ£
Fβπ πΎ
f,ππ·π π ππ
π π₯ +π§πππ
π π
π π₯
(7.71)
which is an equation we will use extensively throughout this book. When solutes have no direct friction with a porous medium, thenπΎf,π=1 and Eq. (7.71) simplifies to Eq. (7.2).
The complete transport model based on Eq. (7.71), partition functions such as Eq. (2.1), and solvent transport equations given by Eq. (8.2), is called the two-fluid model (TFM), and in the context of membrane transport is called the solution-friction (SF) model.
The DSP model for ion transport in membranes. When the model explained in this section for multicomponent transport of charged and uncharged species through a membrane (which is typically also charged) is combined with a description for the Donnan equilibrium at the membrane-solution interfaces, and also includes contributions to the partitioning of ions at these interfaces such as related to ion volume, as discussed in Β§2.2, Β§2.8, and Β§4.2, we end up with a model called the Donnan Steric (Partitioning) Pore model, or DSP model. The state-of-the-art version is called the extended DSP-model. This ext-DSP model also includes how membrane charge is a function of local ion concentrations (especially pH and absorbing ions such as Ca2+) and also includes how ions react with one another in the pore space, as will be addressed in Ch. 10. For instance, the bicarbonate ion, HCO3β, can react with a proton to form carbonic acid, H2CO3, and this reaction is included in the ext-DSP model. And many similar reactions can be included at the same time as well. Also the formation of ion pairs, for instance because Mg2+ reacts with Clβ to MgCl+, can be included in the ext-DSP model, see Kimaniet al.,J. Chem. Phys. 154124501 (2021), and references therein.
A different line of theory that was specifically developed for nanofiltration, is the SEDE model, which is a very detailed extension to the SF model, as explained in detail in Lanteriet al.,J. Colloid Interface Sci. 331, 148β155 (2009). For a comparison of this SEDE model with the SF model of this section, seehere.
Finally we explain the theory to describe ion fluxes in five different cases. The first case we consider is to include ion-ion friction in addition to the other frictional forces. Leaving out all effects related to solute volume, we then have an implicit relationship for ion flux,