Learning Objectives
After completing this chapter, you will be able to:
Develop differential models for problems in mass transfer where the concentration varies along only one spatial dimension.
Derive differential equations for the species concentration by use of the conservation law followed by Fick’s law in Cartesian coordinates.
Derive the model equations in cylindrical and spherical coordinates and understand the differences arising due to geometry.
State the equation in vector form and be able to use it in any coordinate system.
Solve some simpler problems, especially those involving mass transfer with reaction, and examine the property of the solution.
This chapter illustrates the development of differential models by examining simple examples where the transport occurs in only one spatial direction. This simplifies both the model and the mathematics of deriving the solution. By focusing on such simple problems, we build our understanding of the model development. In Chapter 5, we will generalize this understanding to 3-D problems where we derive the complete differential equation for the concentration field and where the particular cases studied here can be obtained as a simplification of the general model. However, the model development is better understood by studying particular cases individually, as is done in this chapter.
To begin, we consider problems posed in Cartesian geometry. Here the concentration is assumed to be a function of only one coordinate, say x. We start off with a simple diffusion problem, followed by diffusion with reaction, transient diffusion, and diffusion with convection. Model development is illustrated in a step-by-step manner, followed by solution of some illustrative problems.
Problems posed in cylindrical and spherical coordinates where the concentration varies only in the radial direction are examined next. One distinguishing feature in contrast to the case involving Cartesian coordinates is that the area for diffusion varies as a function of the radial coordinates; hence the resulting equations appear to be slightly different from those in Cartesian coordinates. Illustrative problems are then solved for problems posed in these geometries. This chapter also indicates that the equations for all three geometries can be unified by using the divergence and the Laplacian operator from vector calculus. This approach provides the vector representation of the model, which is general and useful in any chosen coordinate system.
The first set of problems we will cover are those concerned with mass transfer in a slab of finite thickness in the x-direction with rather large height and width in the y- and z-directions. This makes the transport one-dimensional. First we examine a problem with no generation, no accumulation, and no externally imposed flow across the slab. We can progressively include these terms in the analysis.
Steady state diffusion across a slab is the simplest problem in mass transport analysis. It is so simple that a linear profile across the slab is often taken for granted in steady state analysis. Nevertheless, the steps and the assumptions involved in arriving the result should be examined and the problem should be analyzed step by step since the approach is similar to that used in other cases and in other coordinate systems.
An illustrative problem statement is as follows: A membrane or a medium of thickness L has concentrations CA0 on one side and CAL on the other side. This can, for example, be accomplished by circulating a gas containing the solute A at the appropriate concentration levels on both sides of the membrane. The diffusion takes place along the x-direction only and the concentration profile in the slab and the quantity of species transported across the slab are to be computed. The schematic of the problem is shown in Figure 2.1.
We start with the conservation law, which reduces for this case as follows since there is no generation nor accumulation:
in – out = 0
Input at x is equal to the flux times the area evaluated at x and can be represented as NAxA. Here A is the area of cross-section in the direction perpendicular to diffusion (perpendicular to x here). Similarly, the output is NAxA but evaluated at x + Δx.
The conservation principle therefore gives
If one divides the expression by Δx and takes the limit as Δx tends to zero, the following differential equation for the flux is obtained:
Note: This expression is general and holds for all three geometries examined in this chapter when there is no generation due to chemical reaction and a steady state exists.
The areas for diffusion at x and x + Δx are the same and hence this term cancels out.
So far we have dealt with the combined flux to keep the analysis general. In addition, the area term did not matter and canceled out. For some other cases (a cylinder and a sphere, for example), the area is a function of the position. These cases are examined in Sections 2.2 and 2.3.
Although the problem examined here is a diffusion-dominated process and there is no forced flow, convective flow can arise due to diffusion itself. The effect of such convection is examined in Chapters 5 and 6. For the time being, let us assume that the convection flux is negligible and the combined flux has only the contribution resulting from the diffusion flux. With this assumption we have
NAx ≈ JAx
which is sometimes referred to as a low flux approximation. In other words, any convective flux (resulting from the diffusion process itself or due to any superimposed external flow) is neglected, and the conservation model reduces to
Note: The effect of diffusion-induced convection (examined in Chapter 6) is sometimes referred to the high flux model. The effect of a superimposed (external) flow in the direction of diffusion (the so-called blowing problem) is discussed in Section 2.1.4.
This is as far as we can proceed just by use of the conservation law alone. A model for diffusion flux is needed, and we use Fick’s law as the model:
where DA is the diffusion coefficient of A in the medium or the membrane. Substituting Fick’s law into the flux conservation equation, Equation 2.4, the following equation is obtained for the concentration profile:
We assume now that the diffusion coefficient is not a function of concentration, which turns out to be good assumption for many cases (e.g., diffusion of gases such as oxygen in polymeric membranes). This permits us to take DA out of the derivative sign. It cancels out as well, leading to the following expression:
Integrating twice gives us a general expression for the concentration profile:
CA = B1 + B2x
Here B1 and B2 are two integration constants that are evaluated to satisfy the boundary conditions imposed at the two ends.
Using the values of CA0 at x = 0 and CAL at x = L, we can find the integration constants and eliminate them from the previous equation. The final expression for the concentration profile is
The concentration profile is simply a straight line connecting the concentration values at the two ends. The flux from the slope of this straight line is therefore
The rate at which the solute is transported across the media, , can now be calculated as JAx times A and is equal to
We now have both the concentration profile and the quantity transported across the slab, and the solution is complete. Let us apply the results to examine a simple problem dealing with membrane transport, which is shown in Example 2.1.
Light gases such as helium (He) has reasonable solubility in materials such as glass, so there may be leakage of this gas from a helium-containing glass shell that is used for a helium–cadmium laser in a copying machine. Calculate the helium leakage rate across a glass shell of thickness 1.5 mm with pure helium on one side and air containing negligible helium on the other side. Assume that the glass shell is at a constant temperature of 115 °C and the pressure inside the shell is 460 Pa at the given instant of time. (Problem adapted from Mills, 1995.)
The diffusion coefficient of He (m2/s) in the solid glass is reported as (Mills, 1995)
DA = 1.40 × 10–8 exp(–3280/T)
The solubility parameter has to be known as well and is given in a part of the solution.
Solution
Equation 2.7 is applicable for the flux. However, the concentrations of He are in the glass near the wall, not on the gas near the wall. Hence the concentration jump condition (introduced in Section 1.4) must to be used. We need to know the solubility relationship, which is the additional data needed to solve the problem.
The solubility parameter reported by Mills (1995) is
CAs = KCAG
where K is reported as
K = 3.0 × 10–5 T – 0.0012
For the given conditions, T = 115 + 273 = 388 kelvin, so K equals 0.0104.
On the gas side we have pure helium, whose concentration is calculated as
Using the K value we find CA0 = 0.0104 × 0.1426 = 1.15 × 103 mol/m3, which is the concentration in the glass shell on the helium-exposed side.
On the air side, the helium concentration in air is zero. The glass-side concentration of helium CAL is also equal to zero.
The diffusion coefficient has a value of 2.98 × 10–12 m2/s at 388 K using the equation and the data given earlier.
The flux can be calculated as
which is the leakage rate per unit surface area of the glass.
Note that thermodynamic relation was needed in addition to the diffusion flux expression to account for the concentration jump at the gas–glass interface. A similar approach is used to model the transport rate in gas separation membranes as shown in Section 27.1.
Now we extend the model by assuming that the diffusing solute undergoes a chemical reaction in the slab. The problem statement is as follows: Consider a material in the form of a long slab with a chemical reaction taking place at a rate of RA per unit volume of the slab. An example would be a porous catalyst cast in the form of a long slab. A second example would be oxygen diffusing into a pool of liquid containing some biomass or a liquid-phase reactant that consumes the diffusing oxygen. Many other examples can be given.
Assume that the mass transfer of species A is only in the x-direction. In this section, we develop a differential equation for the concentration variation of A in the system as a function of x and use this in the flux expression to obtain the rate of mass transfer into the slab.
We proceed in a similar manner as in the earlier transfer problem, but include the chemical reaction term in the conservation law. This law should now read as follows:
in – out + net generation = 0
The input and output terms are the same as before and the additional term due to generation is equal to RA(A Δx). Hence
[NAxA]x – [NAxA]x+Δx + RAA Δx = 0
Further simplification is done by taking the diffusion flux alone (NAx ≈ JAx)—that is, by ignoring any effects due to diffusion-induced convection. Also A is constant and does not vary with x. We next divide through by Δx and take the limit as Δx → 0 to obtain a differential equation for the variation of the diffusion flux in the system:
This is the basic model for the system based on the conservation principle. The model can be completed if a constitutive (transport) law for diffusion of a species A in a porous solid is available or developed. A Fick’s law type of model is now used here to complete the problem:
where DeA is the (effective) diffusion coefficient of A in the porous medium.
Note: The rate of diffusion is modified due to the porous matrix. Specifically, a pore structure–dependent effective diffusivity is used in these cases in lieu of fluid-phase diffusivity. This topic is discussed in further detail in Section 7.6. We use the simplified notation DA for the effective diffusivity DeA in the remainder of this chapter.
Using a Fick’s law type of model for pore diffusion as shown in the mass conservation equation and assuming constant diffusivity (Equation 2.8), we get
This equation is known as the diffusion-reaction equation and is widely applied in the analysis of transport in catalysts. Note that the second derivative term on the left side of the equation is the x-dependent part of the Laplacian equation. Hence the equation can be compacted as follows:
Further information is needed on the kinetics of the reaction. In general, RA is a function of the concentration of species A and temperature. It is also often a function of the concentration of other species in the system (e.g., bimolecular reactions, reactions with strong product inhibition). These dependencies are given by a kinetic model for the process. In Examples 2.2 and 2.3, we examine the (general) solution to the problem for simple forms for RA— namely, for a zero-order reaction and a first-order reaction.
Derive a general expression for the concentration profile for diffusion accompanied by a zero-order reaction in a slab geometry.
Solution
For a zero-order reaction, the rate is given as
RA = –k0
where k0 is the volumetric rate of consumption of reactant A (units of mol/m3 s). Equation 2.9 now takes the following form:
The resulting differential equation can be integrated twice to obtain
which is the general solution for the concentration profile. The constants of integration B1 and B2 can be found if the boundary conditions are specified. The boundary conditions are problem specific and depend on what is happening at the two ends of the system.
A common application of this model is in trickling bed filters used in waste water treatment. It is also used for modeling oxygen transport in tissues, where the oxygen metabolism follows a zero-order rate law for most cases.
Example 2.3 shows the application to a first-order reaction.
Derive a general expression for the concentration profile for diffusion accompanied by a first-order reaction in a porous catalyst in a slab geometry.
Solution
In this case the rate is given as
RA = –k1CA
where k1 is the rate constant for a first-order reaction in the units of s–1. Equation 2.9 now takes the following form:
This equation can be solved by treating it as a linear second-order differential equation. The general solution can be expressed in two equivalent forms:
with A1 and A2 being integration constants or
where B1 and B2 now represent the integration constants.
Note: The B1 and B2 will be a linear combination of A1 and A2, and vice versa.
A common application is in constructing a diffusion model for a catalyst pore—an application that is discussed in Chapter 18. At this point, the model development and the nature of the resulting differential equation should be noted. The general solution with two integration constants is also noteworthy since it can be applied to many different boundary conditions. Various types of boundary conditions of common occurrence are summarized in Chapter 5.
In Example 2.4, we apply Equation 2.13 to a specific problem. We also illustrate the use of two common types of boundary conditions.
A pool of liquid is 10 cm deep and a gas A dissolves and reacts in the liquid. The solubility of the oxygen is such that the interfacial concentration is equal to 2 mol/m3 in the liquid and the diffusivity of the dissolved gas is 2 × 10–9 m2/s.
Assume a first-order reaction with a rate constant of 10–6 s–1 in the liquid. Find the concentration profile in the liquid. Express the solution in dimensionless form. Also find the flux of oxygen into the pool of liquid.
Solution
The schematic of the problem and the anticipated concentration profile of oxygen in the pool of liquid are shown in Figure 2.2. Equation 2.13 is applicable for the concentration profile. Also, Equation 2.13 (cosh and sinh form) is more convenient here (rather than the exponential form), and the constants of integration must be evaluated using the boundary conditions. You will find that taking x = 0 at the bottom will simplify the math in the cosh and sinh form of the solution. The location x = L (10 cm here) is then the position of the gas–liquid interface. We need to evaluate the two constants by specifying appropriate boundary conditions.
Since no oxygen diffuses into the bottom of the tank, we take dCA/dx = 0 at x = 0, the bottom of the tank. This type of boundary condition is generally known the boundary condition of the second kind, also known as the Neumann boundary condition.
The procedure to get one of the constants of integration is as follows. First, differentiate Equation 2.13 to get the derivative of the concentration:
Then set this expression equal to zero at x = 0. We find A2 = 0.
The solution now reduces to
Now use the boundary condition at x = L to get A1. The concentration is set equal to CAs, the interfacial concentration of dissolved oxygen at x = L. This is known as the boundary condition of the first kind, also referred to as the Dirichlet condition. Using this information and some minor algebra yields
Both constants are evaluated and the solution for the concentration profile is completed. Backsubstituting of A1 leads to the following result:
The dimensionless version of the solution requires the use of dimensionless concentration cA defined as CA/CAs and a dimensionless distance parameter, ξ = x/L. In addition, a dimensionless parameter called the Thiele modulus for diffusion-reaction systems is needed, which is defined here:
With these definitions, Equation 2.14 can be written in dimensionless form:
The use of dimensionless quantities is a common practice in representation and solution to engineering models, since it makes the mathematical manipulations far easier and allows the final results to take a more compact form.
The rate of absorption of oxygen is calculated using Fick’s law at the gas–liquid interface:
where A is the exposed surface area for oxygen transfer.
Note the plus sign used in Fick’s law for the flux; it is appropriate because we are looking at the flux in a direction of decreasing x. Substituting for the derivative of the concentration leads to
This result has wide applicability in reaction engineering.
Substituting the values, the following results are obtained:
oxygen transfer rate per unit area using Equation 2.16 = 8.74 × 10–8 mol/m2 s
Consider the problem examined in Section 2.1.2, but now assume the slab is initially at a uniform concentration of CAi. At t > 0, one end of the slab (say, x = 0) is exposed to a different concentration of CA0. The other end is still at CAi. The problem is how to predict the temporal evolution of the concentration profile in the system and how use this information to get the instantaneous rate of mass transfer. Here we derive the governing differential equation for the transient diffusion process.
The conservation law should now include the accumulation term:
input – output = accumulation
The input and output terms are the same as before and the additional term due to accumulation is equal to the time derivative of moles of A in the control volume, CAA Δx. Hence:
The final equation with the approximation of NA = JA and the use of Fick’s law for JA leads to the following partial differential equation for the transient problem:
This equation is sometimes referred to as Fick’s second law, but in reality is a combination of the conservation law and Fick’s first law. Constant diffusivity is also implied. The equation needs two boundary conditions and one initial condition since it is now a partial differential equation (PDE). The solution method is described in detail in Section 8.2, but Example 2.5 shows an illustrative result. The example also shows how to represent the problem in dimensionless form.
A membrane, cast as a slab, has zero concentration at time zero. Suddenly, after time zero, one surface is exposed to a gas such that the concentration in the membrane at this point is CA0. The other side of the membrane is kept at zero concentration.
State the governing equation and the boundary and initial conditions to be used. Also sketch the anticipated temporal evolution of the concentration profile in slab.
Solution
The governing equation is the transient model given by Equation 2.17, Fick’s second law. Before stating the required conditions it is useful to write the equation in a dimensionless form. This can be done by using the following variables:
Notice that L2/DA has the same units as time, so we can use it to scale the actual time. Thus we can define a dimensionless time as
The differential equation reduces to a simpler form with these variables:
Note that there are no free parameters, which is a characteristic of a well-scaled problem. You should verify the dimensionless version starting from the original version using the chain rule.
The boundary conditions to be used are as follows: At x = 0, CA = CA0, which in dimensionless form is cA(ξ = 0) = 1.
At x = L, CA = 0, which in dimensionless form is cA(ξ = 1) = 0
Both of these relations hold for all values of τ > 0.
The initial condition is at τ = 0, cA = 0 for all ξ in the domain to 1, since the slab is maintained at zero concentration at the begining of the process.
The problem specification is now complete. This problem can be solved analytically by a modified method of separation of variables or by numerical methods (e.g., MATLAB PDEPE solver). These methods are elaborated in a later chapter. At this point you should focus on problem formulation and dimensionless representation. The computed result using MATLAB PDEPE solver is shown in Figure 2.3. Note that for large time, a linear profile is obtained that is the steady state profile for diffusion across a slab.
L2/DA is the characteristic time scale for diffusion. In other words, it is the time scale for any disturbance to be smoothed out by diffusion, and one can expect that a steady state profile will be reached around this time. This parameter is called the diffusion time. In this particular case, the steady state is reached around 0.3 × L2/DA.
We now examine a simple problem where the convective flow is superimposed on diffusion. Specifically, the case involving steady state and no chemical reaction is analyzed. Consider again the problem discussed in Section 2.1.1, but with the addition of a superimposed flow across the system with a velocity of vx. This information is used to derive the differential equation and to find the flux across the system. The problem is illustrated in Figure 2.4.
The application of the conservation law shows that the combined flux is the same at x (in) and at x + Δx (out), since this is steady state and no reaction is taking place. Therefore Equation 2.3 applies here as well; it is repeated here for ease of reference:
The combined flux has now contribution from both diffusion and convection:
where vx is the blowing or transpiration velocity. Substituting into the flux equation and assuming constant DA and vx, we obtain the following second-order differential equation:
This is the simplified version of the general 3-D convection-diffusion equation that will be discussed in more detail in Chapter 5. This general equation is stated here to help you better appreciate the vector representation of mass transfer models:
DA∇2CA – v · ∇CA = 0
The general solution to Equation 2.19 can be written as follows:
where A1 and A2 are two integration constants. (You may wish to verify this by direct substitution.)
The boundary conditions of the first kind are useful for this case since the concentrations are specified at the ends. At x = 0, the concentration is CA0; at x = L, the concentration is CAL. Using these conditions, we can obtain the constants of integration and complete the solution. The details are left as an exercise. In this scenario, the profiles are no longer linear, in contrast to the result obtained in the pure diffusion case.
More useful than the concentration profile is the (combined) flux across the system. This can be calculated from the concentration profile as
or equivalently as
The result after some standard algebraic manipulations is as follows (which you should verify):
The expression can also be written as
where the bracketed term on the right side can be identified as the factor by which the blowing enhances the rate of mass transfer. This is illustrated in Example 2.6.
A porous plug 10 cm long has hydrogen at a partial pressure of 0.1 atm on one side and air at 1 atm on the other side at 293 K. Find the transport rate of hydrogen if the system is stagnant, if there is a superimposed gas flow at a rate of 0.1 cm/s in the same direction as that for diffusion, and if the flow is in the opposite direction to diffusion.
Solution
The parameter vxL/DA is the key dimensionless group in Equation 2.22 showing the effect of the blowing velocity. This parameter is known as the Peclet number:
It can be interpreted as the ratio of the convection rate to the diffusion rate. For no blowing, the pure diffusion model applies:
Equation 2.22 indicates that the pure diffusion flux should be corrected by the following factor:
The correction factor is more compactly expressed using the Peclet number as
For numerical calculations, the value of the diffusion coefficient of hydrogen in air is needed. We will use a value of 0.7 cm2/s here. Hence the Peclet number is Pe = 0.1 * 10/0.7 = 1.4286 and the correction factor for blowing calculated using Equation 2.23 is 1.8788.
The concentration of hydrogen at one end CA0 is calculated as ph/RgT , or 4.03 mol/m3.
Flux in the absence of blowing is DA(CA0 – 0)/L, which is found to be 0.0028 mol/m2 s. The flux in the presence of blowing is 0.0028 × 1.8788 = 0.0053 mol/m2 s.
If the flow is in the opposite direction, vx is negative (suction condition) and hence Pe = –1.4286. The correction factor is now 0.45 (less than 1), so the flux is 0.0013 mol/m2 s. Flux is reduced due to the blowing counter to the direction of diffusion.
We have now examined four prototypical problems in Cartesian coordinates: (1) diffusion alone, (2) diffusion balanced by reaction, (3) transient diffusion with accumulation, and (4) diffusion with superimposed flow. We now study similar problems posed in cylindrical coordinates but again with transport taking place only in the radial direction.
In this section, we repeat the analysis for the case of a cylindrical geometry to show some nuances that arise primarily due to the nature of the coordinate system. Specifically, we look at problems where the concentration is a function of only the radial position. The analysis is similar to that for a slab geometry, but the point to bear in mind is that the area over which diffusion takes place is a function of the radial position. We repeat some preliminary steps in the slab analysis and show the differences for clarity.
Steady state diffusion is considered first, followed by diffusion with reaction and then transient diffusion.
The differential control volume, shown in Figure 2.5, is a ring element located at r and r + Δr. We start with the conservation law for this control volume, which reduces for the case of no generation and accumulation to
in – out = 0
Input at r is equal to the flux times the area evaluated at r and can be represented as NArA. Similarly, the output is NArA but is evaluated at r + Δr. The conservation principle therefore gives
Next, we note that the area across which diffusion is taking place is equal to 2πrL. (This is different at r and r + Δr.) The conservation equation with the area A term substituted is
Dividing by Δr and taking the limit leads to
The combined flux is taken as diffusion flux for further analysis. Using Fick’s law to close the problem for JAr
leads to
For a constant diffusivity case, the following differential equation is obtained:
This suggests that
Now a second integration can be done and a general expression for the concentration profile can be obtained:
CA = A1 ln r + A2
The constants A1 and A2 can be found by fitting the boundary conditions.
Consider the diffusion across a cylindrical shell with radius of Ri at the inner surface and Ro at the outer surface. If we set the concentrations as CAi and CAo at these points, we can evaluate the constants and obtain the following equation for the concentration profile in this shell:
The concentration profile is now logarithmic in r, rather than a linear profile across the shell.
The rate of transport across the system can be computed as
or as
Both expressions will lead to the same result:
The expression for moles transported is often written in a format similar to that used in the Cartesian case:
where t
is the thickness for diffusion (equal to Ro – Ri here) and Am is the mean area for diffusion. Comparing the two expressions for , we find the mean area for diffusion is the logarithmic mean area of the inner and outer cylinder:
In our next scenario, we include the effect of reaction. This problem is representative of diffusion with reaction in a catalyst, which now takes the shape of a long cylinder or a concentric annular cylinder. We follow the method shown in Section 2.1.2, with the only addition being the reaction term, which is equal to control volume times the rate of reaction per unit volume, or (2πLr Δr)RA. The final equation with the same assumption of constant DA can be derived as
Note that the differential term on the left side of the equation is the r-dependent part of the Laplacian operator. Also note the following two forms for this part:
or equivalently
We now examine the solution for simple cases for RA. However, we also note that in general a numerical solution is needed for reactions with nonlinear dependency on concentration.
For a zero-order reaction RA = –k0, the governing equation is
This model holds for both a solid cylinder and an annular ring cylinder. The general solution is obtained easily by performing integration twice. The final solution is, however, different due to the change in the location where the boundary conditions are applied.
The geometry considered further here is a solid cylinder with a radius R. The domain of the solution is therefore 0 < r < R.
The boundary condition of constant surface concentration CAs is imposed at r = R, while the condition dCA/dr = 0 at r = 0 is imposed at the center. The final result for the concentration profile is as follows (which you should verify by doing some minor algebra):
The final solution for a concentric cylinder is somewhat different since the domain of the solution is Ri < r < Ro; it is left as an exercise. An application of this geometry is in oxygen transport in tissues, a problem of importance in biomedical engineering that is discussed in Section 23.2.
The solution for a first-order case where the rate is a linear function of concentration is somewhat complicated. The rate is given as
RA = –k1CA
Use of this rate form in Equation 2.30 leads to
The solution of this differential equation is stated in terms of the Bessel functions. The general solution is
The I0 and K0 components are the modified Bessel functions of the first and second kind, respectively.
This solution may be applied to diffusion with reaction in a catalyst having the shape of a long solid cylinder or long concentric cylinder (discussed further in Section 23.2). The constants of integration depend both on the geometry (solid or concentric cylinder) and on the imposed boundary conditions at the end points of the geometry. The solid cylinder case is described in Example 2.7.
A porous catalyst is cast in the form of a long cylinder and is exposed to a gas concentration of CAs at the surface. State the boundary conditions and derive an expression for the concentration profile.
Solution
The general solution shown in Equation 2.33, applies but the concentration should be finite at r = 0. Note that this can be taken as one of the boundary conditions. Since the K0 function goes to infinity at r = 0, we have to set A2 = 0 so that the concentration does not becomes unbounded. The solution reduces to
The second boundary condition at r = R is applied. Here CA is set as CAs, the prescribed value. A1 can be estimated and substituted back to get the final result for the concentration profile. The result is compactly represented in dimensionless form as
where cA is a scaled or dimensionless concentration CA/CAs, and ø is a dimensionless regrouping of the variables. You should verify that ø is defined as
This dimensionless parameter is referred to as the Thiele modulus. Finally, ξ = r/R is the dimensionless radial coordinate.
In the case of transient diffusion in a cylinder, we balance the in – out term with the accumulation. Details are not shown but you should be able to derive the following PDE for transient diffusion in a long cylinder with diffusion in only the radial direction:
Solution methods for such PDEs are examined in detail in Chapter 8.
In this section we show examples of problems posed in spherical coordinates. Diffusion in a spherical shell, diffusion with reaction in a porous spherical catalyst, and transient diffusion from a sphere are the three examples explored here, as they are prototypical problems encountered in many applications.
The control volume in spherical coordinates is a spherical shell contained between r and r + Δr, as shown in Figure 2.6. The key point to note is that the area for diffusion is 4πr2 at any location r. Similarly, the volume contained in the differential control volume is equal to 4πr2 Δr. With these modifications the governing differential equations can be set up for various cases, following a similar procedure as the two earlier cases.
For the diffusion-only case, we obtain
This suggests that
Now a second integration can be done and a general expression for the concentration profile can be obtained:
This concentration profile is no longer linear. The constants A1 and A2 can be found by fitting the boundary conditions at r = Ri and at r = Ro. Further details are straightforward and not shown.
The rate of transport across the shell is of more interest and can be computed as
Here Fick’s law is used at r = Ri and multiplied by the area for diffusion at this radius. Alternatively, we can use Fick’s law at r = Ro:
Both expressions will lead to the same result. The final result can be put in a form similar to the Cartesian geometry solution, using a mean area parameter for the diffusion area. The result is
where t
is the diffusion length equal to Ro – Ri and Am is the area for diffusion. The latter can be shown to be equal to the “geometric” mean area of the inner and outer shell:
Am = 4πRiRo
In this section, a chemical reaction is assumed to take place together with diffusion. Again we consider concentration to vary as a function of r only. A shell balance approach leads to the following equation for diffusion with reaction in a spherical geometry:
Again, the leading differential term on the left side of Equation 2.36 is found to be the Laplacian operator
or equivalently
We find the diffusion-reaction problem for all the geometries can be generalized as
A constant value for the diffusion coefficient is implied in this model.
Details and the implications of the solution for calculation of the rate of reaction in a porous catalyst are examined in Chapter 18.
For now, it is useful to present the final solution for a spherical catalyst that is exposed to a surface concentration CAs. Solutions are presented in dimensionless form, with their detailed derivation left as exercises.
The solution for a zero-order reaction in dimensionless form is
where ξ = r/R and .
The solution should be used only if . If , a concentration-depleted layer (a region of negative concentrations, which is unrealistic) develops near ξ = 0, the center of the sphere. A modified analysis is needed in such a case. Section 18.2.3 takes this issue up in more detail.
The solution for a first-order reaction can be shown to be
where ξ = r/R and .
Transient diffusion in a sphere in considered here briefly, as it has important applications in many areas. For example, this case may be used to find the rate of drug release from a capsule. A shell balance approach leads to the following partial differential equation for transient diffusion in a spherical geometry:
Here we illustrate the case in which the sphere has a uniform concentration CAi, which provides the initial condition. The sphere is exposed to a constant surface concentration CA0 at r = R after time zero. This provides one of the boundary conditions. The concentration is symmetric at r = 0 and hence dCA/dr = 0 at this point, which provides the second boundary condition. The problem formulation is complete and analytical solutions or numerical solutions are used to solve for the temporal evolution of the concentration profile and the flux at the surface. These details are examined in Chapter 8.
For illustrative purposes, a sketch of dimensionless concentration versus position is shown in Figure 2.7. It is intended to help you get a qualitative feel for the result.
Figure 2.7 uses dimensionless variables—specifically, dimensionless distance, ξ = r/R; dimensionless time, τ = tDA/R2; and dimensionless concentration (more of a ratio of concentration differences), defined as:
This concentration is actually a dimensionless concentration difference. Thus the initial value of cA is 1, and the final value is equal to the surface concentration of 0. Key points to note from the solution shown in Figure 2.7 are the following: At the initial time, the concentration drop is confined to the region near the surface, so the initial profile is rather steep. A nearly steady state profile is achieved at τ = 0.3. In terms of actual time, the value is about 0.3R2/DA. The final steady state (dimensionless) concentration should be 0.
Diffusion in a slab of finite thickness with the ends held at different concentrations is the simplest problem in mass transfer. For constant diffusivity, the concentration profile is linear and the flux can be calculated from the slope of this linear profile.
An important application of slab diffusion is to calculate the transport rate of a gaseous species across a membrane. Note that the solubility of the gas in the membrane is also needed to solve such a problem, because the concentrations given in Section 2.1 are in the membrane phase, rather than in the external fluid phase.
The concentration profile for diffusion in long cylindrical shell is not a linear function of r, unlike the case in slab geometry. Instead, this profile is a logarithmic function. The difference arises due to the geometry or curvature effect—that is, due to the area of diffusion changing in the direction of diffusion. The curvature effect can be neglected for thin shells and the problem may be approximated as though a slab model applies.
Likewise, the concentration profile for diffusion across a spherical shell is an inverse function of r due to the curvature effect.
Diffusion with reaction is a well-studied and important problem in reaction engineering. The governing equation is the diffusion reaction equation given by Equation 2.9 for a slab, Equation 2.30 for a cylinder, and Equation 2.36 for a sphere. The equations can be compacted using vector notation and with the Laplacian operator of the concentration. The vector form applies to three-dimensional problems in any coordinate system.
The solution of the equation for diffusion with reaction for a zero-order reaction is the quadratic function for a slab. It can be shown to be a combination of a quadratic function and a logarithmic function in cylindrical coordinates (with r-variation only), but is a combination of r and 1/r in spherical coordinates.
Diffusion with a first-order reaction is amenable to analytical solution for the three common geometries considered in this chapter. Simple solutions in terms of exponential or hyperbolic functions are obtained for a slab. By comparison, the solutions in cylindrical coordinates are stated in terms of modified Bessel functions, while those for spherical coordinates are stated in terms of spherical Bessel functions.
Transient diffusion is governed by a PDE (partial differential equation). An initial condition is also required. Results are usually shown using a dimensionless time. The definitions of time constant and dimensionless time are worth remembering.
Convection imposed in the same direction as diffusion is one of the simplest problems in convective mass transfer. The effect of convection can be represented by a correction factor (blowing factor), which depends on a dimensionless number called the Peclet number.
2.1 Equation 2.6 shows that the concentration profile for diffusion in a slab is linear. State the assumptions that have gone into this result.
2.2 When can the concentration profile for diffusion in a slab geometry be nonlinear?
2.3 State the differential equation for concentration in a slab geometry for diffusion with a second-order reaction.
2.4 State the two commonly observed boundary conditions in diffusion-reaction problems. State which condition applies when.
2.5 What is the definition of the Thiele modulus for a first-order reaction and what is its significance?
2.6 How is the Thiele modulus defined for a zero-order reaction?
2.7 What is the definition of diffusion time and what is its significance?
2.8 What is the definition of the Peclet number and what is its significance?
2.9 Define blowing factor and state its use.
2.10 A catalyst is cast in the shape of a long concentric cylinder. A first-order reaction is taking place here. What is the general solution for the concentration profile?
2.11 State the solution to diffusion with a first-order reaction in a spherical catalyst particle.
2.1 Membrane for oxygen separation. Data for gas permeation in polymers have been reported by Seader et al. (2010). Table 2.1 shows these data for low-density polyethylene. Note that both diffusivity and solubility are needed to calculate the rate of diffusion.
Gases |
H2 |
O2 |
N2 |
CH4 |
D × 1010 m2/s |
0.474 |
0.46 |
0.32 |
0.193 |
H × 106m3 STP/m3 · Pa |
1.58 |
0.472 |
0.228 |
1.13 |
The solubility parameter H here is defined as the concentration in the solid (membrane) divided by the partial pressure of the gas at equilibrium. The concentration in the solid is defined as m3 gas in STP perm3 solid (qA) rather than in mol/m3 Thus the Henry’s law constant is defined as qA = HApA.
Find the rate of oxygen and nitrogen transport rate (in mol/m2 s) for the previously described membrane when it is exposed to air with 150 psia and 78 °F on one side and atmospheric pressure on the other side. If a rate ratio of at least 5 is needed for good oxygen separation, is this a good membrane to enrich oxygen from air? The membrane thickness is 5 μ m.
2.2 Slab diffusion with diffusivity varying with concentration. A membrane is set up such that the diffusion coefficient varies as a function of concentration according to the relation
D = D0(1 + αCA)
The membrane is of thickness L with the concentrations of CA0 and CAL as the end point values. Find an expression for the concentration profile and the flux across the system. Compare this expression with the flux value using the flux based on a constant value of the diffusion coefficient based on the average concentration—that is, the diffusivity based on (CA0 + CAL)/2.
2.3 Hydrogen leakage from a steel tank. A spherical steel tank of 1 L capacity and a 2-mm wall thickness is used to store hydrogen at 673 K. The initial pressure in the tank is 5 bar, and the hydrogen partial pressure outside the tank is zero. Hydrogen has finite solubility in steel and leaks out of the tank. Find the rate of leakage at the start of the process where the pressure in the tank is 5 bar.
The solubility and diffusivity of hydrogen are functions of temperature and given by the following relations:
where CA is the concentration of dissolved hydrogen, mol/m3, in equilibrium with a pressure of pA, with pressure unit in bars.
D = 1.65 × 10–6 exp(–4630/T)
with T in K and D in m2/s.
Also calculate the leakage rate if the tank pressure is 10 bar.
Note that the solubility relation is nonlinear. This is often observed for gases such as hydrogen, which dissociates at the metal surface and is present as H-atoms in the solid matrix rather than as H2 molecules in the gas phase. Hence doubling the pressure will not double the transport rate.
2.4 Mean area for diffusion in a cylindrical shell. Consider the diffusion across a shell of long cylinder of inner radius Ri and outer radius R0. Often the results are expressed in the form similar to that for a slab:
where t
is the thickness, which is equal to R0 – Ri
The area Am is some representative average area for mass transfer. Verify that the area should be log mean of the inside area 2πRiL and the outer surface area 2πRoL.
Also show that if the membrane thickness is small, either of the areas (inner or outer) can be used as an engineering approximation. The slab model is sufficient for such cases and the actual logarithmic profile is close to a linear profile.
2.5 Mean area for diffusion in a spherical shell. Repeat problem 4 for diffusion across a spherical shell. Show that the area should be the geometric mean of the inside area and the outer surface area .
2.6 General convection-diffusion equation. Write out in detail the convection-diffusion equation for a 3-D problem in Cartesian coordinates. Show that Equation 2.19 will be obtained after making suitable simplifications.
Verify the general solution given by Equation 2.20 for the 1-D case shown in the text. Find the solution for a unit concentration at x = 0 and zero concentration at x = L.
2.7 Convection augmentation of mass transfer in a slab. Plot the augmentation factor due to convection (the blowing parameter) as a function of the Peclet number. Find the limiting values of this factor for large values of Pe. Plot the factor if the velocity is opposite to the direction of diffusion. What would be flux under large velocity values?
2.8 Convection effects in a cylinder. Consider a porous membrane cast as the shell of a long cylinder of inner radius Ri and outer radius R0. Concentrations are maintained at different values CAi and CAo at each of these surfaces, and diffusion is taking place. What is the number of moles transported across the system? This provides the base value in the absence of convection.
Now to increase the transport rate, a gas at a flow rate of Q is forced across the system in the radial direction, and convection is also taking place in addition to diffusion. Set up the model for this system and derive the governing differential equation. Solve the equation and find the augmentation in the rate of mass transfer due to the flow.
2.9 Effectiveness factor of a catalyst. A catalyst is cast in the form a thin slab of thickness 2L. Species A diffuses and undergoes a first-order reaction. Set up a 1-D differential model. Show that the solution can be expressed in dimensionless form using a Thiele parameter, ø, defined as
Verify the following solution:
where cA is CA/CAs and ξ = x/L.
The concentration variation along position causes the rate of reaction to vary along with the position. For design purposes, an average rate of reaction, is often interest. This is defined as
Derive an expression for this concentration variation.
The average reaction rate is often scaled by the maximum reaction rate and the ratio is called the effectiveness factor of the catalyst. It is defined as
Show that the following expression can be derived for the effectiveness factor:
Show that this simplifies to η = 1/ø for ø > 3.
Calculate the effectiveness factor for a catalyst slab with thickness L = 3 mm having an effective diffusion coefficient of 2 × 10–5 m2/s. The rate constant has a value of 0.2 s–1.
2.10 Concentration profiles in a cylindrical catalyst. A porous catalyst is used for CO oxidation and the process is modeled as a first-order reaction with a rate constant of 0.2 s–1. The effective diffusion coefficient for pore diffusion was estimated as 4 × 10–6 m2/s. The catalyst is exposed to a bulk gas at 600 K and 1 bar pressure with a CO mole fraction of 2%. External mass transfer coefficient is assumed to be not rate limiting, so a type I condition can be used. The catalyst is in the form of a long cylinder of radius 5 mm.
What is the solution for the concentration profile? Write the solution for concentration in terms of a Thiele parameter defined as . Plot the CO concentration within the catalyst as a function of radial position. Find the concentration value at the center of the catalyst.
2.11 Zero-order reaction in catalyst of three shapes. Derive an expression for the concentration profile for a zero-order reaction taking place in (a) a rectangular slab catalyst, (b) a long cylinder, and (b) a sphere.
2.12 Transport of oxygen in a tissue. Krogh (1919), in a Nobel Prize–winning paper, modeled oxygen transport in tissues by assuming the tissue to be an annular cylinder where a blood vessel of radius Ri is surrounded by a tissue region with an outer radius of Ro.
Oxygen diffuses into the tissues and undergoes a zero-order reaction. What is the governing differential equation? State the boundary conditions under the following assumptions:
r = Ri, the inner radius, is in contact with blood and the oxygen concentration at this point is fixed as the concentration in the blood.
Oxygen does not diffuse past r = Ro, the outer radius of the tissue. Solve the concentration profile in the tissue. If the oxygen concentration at the outer edge of tissue is nearly zero, what is the concentration in the blood?
2.13 First-order reaction in a spherical catalyst. A spherical catalyst is 6 mm in diameter and is maintained at a concentration of 20, 000 g · mol/m3 at the surface. The effective diffusion coefficient is 0.02 cm2/s and the rate constant is 0.2 s–1. What is the center concentration?
An average rate of reaction can be defined as
where Vp is the volume of the sphere equal to 4πR3/3. What is the average rate of reaction?
The ratio of the average rate of reaction to the value based on the surface concentration is called the effectiveness factor of the catalyst. Derive an expression for the effectiveness factor. Calculate the numerical value for the given data.
2.14 Effect of particle size. In problem 13, if the catalyst size is reduced by half what would be the effectiveness factor? If an effectiveness factor of 0.9 is needed, what should be the diameter of the catalyst?
2.15 Drug release from a capsule. A drug capsule is a 0.6-cm-diameter sphere and has an active component that diffuses out the capsule. If the diffusion coefficient is 3 × 10–6 cm2/s, estimate the approximate time for which the active component will be effective.
3.144.48.3