Inside a pore, a gas molecule may hit other molecules or the pore wall. The Knudsen number tells you which one controls, and that decides whether you use , the Knudsen diffusivity or a combination.
Learning objectives
- Calculate the mean free path and the Knudsen number, and classify the regime as bulk, transition or Knudsen.
- Calculate the Knudsen diffusivity and explain why it does not depend on pressure or on the partner gas.
- Combine bulk and Knudsen resistances for a given flux ratio, and correct for porosity and tortuosity.
- Predict how pressure and pore size move a pore from one regime to another.
Background
Catalyst pellets, adsorbents, membranes and porous solids being dried all carry gas through pores that may be only nanometres wide. Compare the mean free path with the pore diameter (Perry’s Eq. 5-261 and Table 5-16, p. 5-55):
Below , molecule-molecule collisions control and the ordinary bulk applies (Fuller or Table 2-141), corrected for the solid by Eq. 5-249, . Above , wall collisions control, and the Knudsen diffusivity applies (Eq. 5-250):
( in m²/s, in m, in K, in kg/kmol)
has no pressure term, while . So raising the pressure pushes a pore from Knudsen toward bulk behavior. In between, add the resistances, with (Eq. 5-252):
Perry’s writes Eq. 5-252 with the effective values and ; applying after combining gives the same . For equimolar counterdiffusion () this reduces to (Eq. 5-253). Table 5-16 labels that case ; since requires , it means . Perry’s gives a normal tortuosity range of 2 to 6 for silica gel, alumina and other porous solids, and 5 to 65 for activated carbon (p. 5-55). In pure Knudsen flow with no reaction, Graham’s law fixes the flux ratio: .
Simulator
Diffusion in Porous Solids
Bulk, transition and Knudsen regimes | Perry’s Chemical Engineers’ Handbook, 9th ed.
Experiment controls
Live calculationsPore and gas conditions
Diffusivity reference
Flux relationship
Porous solid
Perry’s typical tortuosity: silica gel/alumina 2 to 6; activated carbon 5 to 65. The slider explores τ = 1 to 10.
Current diffusion state
Diffusivity vs pore diameter
Diffusivity vs pressure
Inside one idealized pore (schematic animation)
Gold: tracked molecule. Blue/grey: pore gas. Red flashes: gas-gas encounters. Wall flashes: wall reflections.
Schematic, not to scale. The animation only switches the mix of gas-gas and wall encounters with NKn; collision rates, speeds and spacing are illustrative, not a molecular simulation.
Self-checks
Checks use the default inputs (illustrative H₂ diffusing in N₂, 300 K, 1 atm, 100 nm pores, ε = 0.4, τ = 4, EMCD) whatever the current controls, plus exact scaling identities. Expected values come from an independent implementation. None is a homework answer.
Select Run tests to calculate the checks.
Governing equations
Perry’s 9th ed., p. 5-55 (Eq. 5-261 and Table 5-16). SI used internally.
λ = (3.2 μ / P) √(RT / (2π Mgas))
P in Pa; μ in Pa s; Mgas in kg/mol; λ in m.
NKn = λ / dpore
Bulk: below 0.01. Transition: 0.01 to 10. Knudsen: above 10.
DK = 48.5 dpore √(T / MA)
d in m, MA in g/mol, DK in m²/s. Assumes straight cylindrical pores: 48.5 = (1/3)√(8R/π) in these units, from DK = (dpore/3)√(8RT/(πMA)).
DAB = Dref(T/Tref)1.75 (1 atm/P)
Approximate low-density gas relation.
1/D = (1 − α xA)/DAB + 1/DK
α = 1 + NB/NA (NA ≠ NB). Perry’s writes it with Deff and DK,eff; because both carry the same εp/τ, the intrinsic pore form shown here is equivalent.
1/D = 1/DAB + 1/DK
Equimolar counterdiffusion, α = 0.
Deff = εp D / τ
Same geometric correction applied to molecular and Knudsen resistances.
Dbulk,model = DAB / (1 − α xA)
What Eq. 5-252 gives with the wall resistance removed (DK → ∞). Compare at the same α, xA, porosity and tortuosity. At α = 0 or xA = 0 it is DAB.
Open this simulator on its own page.
Try this
- Sweep the pore diameter from 2 nm to 10 µm at 1 atm. Where do the regime boundaries fall, and which diffusivity dominates in each?
- Fix the pore size and change the pressure from 0.02 atm to 20 atm. Why does the flux plot flatten at one end?
- Switch the flux condition from equimolar to Graham’s law to stagnant B at the same composition. How much does the combined change, and when would using the wrong matter?
Credits and sources
- Perry’s Chemical Engineers’ Handbook, 9th ed.: Table 5-16 (Eqs. 5-249 to 5-253), Eq. 5-261 and text, p. 5-55.
- Secondary: Geankoplis, 3rd ed., Sec. 7.6 (diffusion of gases in porous solids and capillaries).
- Simulator: Dr. Ray’s Learning Center. Drafted with AI assistance, then checked against Perry’s.
Course: Transport Phenomena. Previous: Pseudo-Steady Evaporation. Next: The Semi-Infinite Solid.