Talk:Fick's laws of diffusion/Archive 1

Dimensions for D
Could someone explain how we obtain the dimensions of [length2 time-1] for the diffusion coefficient/diffusivity D?


 * Well assuming x is in [length], t is in [time] and $$\phi$$ on the both sides is in the same units, then we don't have other choices. abakharev 01:39, 23 March 2006 (UTC)


 * D is velocity of flow, dimension length1time-1, divided by gradient of concentration, dimension length-1. --Vaughan Pratt (talk) 07:02, 25 July 2008 (UTC)

Article name
Shouldn't the article be called Fick's laws of diffusion ? MP  (talk) 09:20, 25 September 2006 (UTC)
 * I was just thinking the same, although it's only minor maybe we should change itAzo bob (talk) 10:38, 31 March 2008 (UTC)
 * I agree as well. I think I know how to do it (it is actually MOVING a page that is done to accomplish this). I haven't done this before though, so I want to read up on it some more . Help:Moving a page --Lacomj (talk) 22:06, 17 December 2008 (UTC)
 * Done. Lone Skeptic (talk) 03:15, 11 December 2009 (UTC)

variant phi characters
This is picky, but some sources distinguish between the two ways of drawing phi, i.e. ɸ and φ. It's the same character, but different equations pick one or the other for historical reasons. I suggest picking one and sticking with it. Ojcit 22:59, 2 October 2006 (UTC)
 * I prefer $$\phi$$, to ensure that the symbol in text is identical to the symbol in equations (typeset with latex). Berland 11:21, 29 January 2007 (UTC)

History Section
Can somebody please improve the history section? It's not particularly important but still, for people who need to know the background information, there's a lot left to be desired.--Gabycs 21:12, 14 March 2007 (UTC)

A Biological Perspective
How does it combine with Graham's law to show "the exchange rate of a gas across a fluid membrane?" Graham's law deals with different molar masses, does that mean you just have to worry about the different molar masses of the gas with the fluid?

In a shake flask of microorganisms the demand for oxygen exceeds the supply; if you measure the oxygen in the liquid phase when the microorganisms are trying to grow exponentially, the oxygen concentration is observed to be zero. Does Fick's Law term of (P2-P1) change the partial pressure of oxygen only so as to raise the oxygen concentration only, when the experimenter mixes in some extra oxygen into the gas phase? Richard8081 21:46, 14 August 2007 (UTC)

Vandalism removed
Removed something about his cod piece. —Preceding unsigned comment added by 80.202.72.196 (talk) 21:08, 15 October 2007 (UTC)

Fick's law in biology
For the transport of CO2 across cell walls of leaves Fick's law takes the form dq/dt = -(c2 - c1)/R where c2 < c1 and R is referred to as the resistance and is used for uniform flow, i.e., the flow is constant in time. It would appear that the area of the membrane and its thickness would have to be constant. dc/dt=d(q/V)/dt=(dq/dt)/V=-Δc/RAΔx=-(1/RA)Δc/Δx so Fick's law is of the form,

dc/dt = - K Δc/Δx

which is not the diffusion equation. The flow goes down a concentration gradient. Also by the ideal gas law P=cR'T so dq/dt = -Δc/R = -ΔP/(R*R'T) = - K'ΔP so we have,

dq/dt = - K' ΔP

It might be better to redo Fick's law to make the constants independent of the area and thickness of the membrane so that they are characteristic of the material of the membrane and not the dimensions.

I will restore some of the changes I made. —Preceding unsigned comment added by Jbergquist (talk • contribs) 11:19, 21 October 2007 (UTC)

I have been reviewing the derivation of the diffusion equations for temperature and concentration. Both use a constitutive equation and a continuity equation stated in vector form. For thermal conductivity heat transfer takes place with no net exchange of carriers which would involve convection. Diffusion takes place at constant temperature and pressure.

The biological usage is that of Fick's first law. It differs from conduction which the resistance causes Joule heating while the passage of a gas through a barrier is associated with the Joule-Thomson effect which causes cooling. The matter requires further study. As stated the "flow law" may require some modification such as an A² term as in Poiseuille's law which also involves a connection between a flow and pressure differences. --Jbergquist 00:26, 22 October 2007 (UTC)

The term "rate of diffusion" can be confusing. The reference cited uses flux = P A Δc where the flux would contain an extra A. P is the permeability of the membrane and is more easily determined than a conductivity K. One should probably use K if one wants to express the law in terms of a constant which is not dependent on the amount of material or its geometry. So one would arrive at R = d/KA² as the formula for the resistance. --Jbergquist 23:33, 22 October 2007 (UTC)

The reason why one might want to use permeability is because it follows the same rules of combination as conductance since P=KA/d. Hence doubling the thickness of a membrane halves the permeability and doubling the area would double the permeability. --Jbergquist 00:11, 23 October 2007 (UTC)


 * I think I'd prefer to see this kind of information in a separate article, with a link from this general article. I.e., this is more of a special case.  --Lacomj (talk) 22:19, 17 December 2008 (UTC)

Ideas for improvement
Here are a few ideas for improvement of this article that I'd be able to help with. I appreciate any comments. Some are more radical than others. --Lacomj (talk) 22:45, 17 December 2008 (UTC)
 * 1) Discussion (brief) of generalized transport equations (i.e., compare with heat, momentum, charge transport).
 * 2) Derivation of Fick's 2nd Law
 * 3) Expand section on solutions to Fick's 2nd law.  Erf solution is most common, but there are a few other important ones.
 * 4) An example solution to a 2D case analogous to the 1D case? Lack of clear example in the literature.71.37.148.130 (talk) 22:31, 16 October 2011 (UTC)maehlen
 * 5) Brief discussion of more advanced topics including:
 * 6) multicomponent diffusion (not simply binary diffusion)
 * 7) Onsager's equations & thermodynamics
 * 8) Random walks, stoichastics, and Green's function representations
 * 9) Diffusion isotropy
 * 10) Move the sections on the diffusivity (temperature and pressure dependence) to a new article on Diffusivity-- which can be expanded considerably to include:
 * 11) Intrinsic diffusivity
 * 12) Tracer diffusivity

An idea for improvement
Fickian diffusion is an ideal case and real cases are different. Why the real case is different from fickian diffusion can be discussed


 * sounds like a good idea to me, although anomalous diffusion can get quite complex, so it'd need to be a fairly general statement. I did touch a bit on this in the revised history section, where I mention that there is such a thing as "non-Fickian" diffusion.  It can be expended a bit perhaps, or maybe this should instead be the emphasis for a separate page?  --Lacomj (talk) 00:24, 24 December 2008 (UTC)

– You don't need to turn to anomalous (eg single file) diffusion to find examples of non-Fickian behaviour. There are simple examples in ternary mixtures (see eg the books and publications of Krishna) where transport goes against the direction of the concentration gradient. Most would consider this to be non-Fickian behaviour, since Fick's law is really a phenomenological law relating diffusive flux to concentration gradients. At best, it shows that the Fickian diffusion coefficients are highly dependent on fluid composition (much more so than the Maxwell-Stefan diffusion coefficients). Snowjeep (talk) 17:00, 16 October 2009 (UTC)

Re-location of text on diffusion coefficient to mass diffusivity
I believe it might be appropriate to re-locate the sections that deal exclusively with the diffusion coefficient (=mass diffusivity) to its own article (mass diffusivity). Any opinions, better ideas? Cheers. Stan J. Klimas (talk) 22:43, 5 March 2009 (UTC)

Correction within notes necessary
Hello,

the notes are not correct. For Ficks original article it has to read: "Poggendorffs Annalen der Physik" short: "Pogg. Ann."

Unfortunately the notes are hidden in some "ref.list" for which I do not know to get access. Perhaps somebody elso knows and can do this correction?

there is also the lonk to the abstract: http://onlinelibrary.wiley.com/doi/10.1002/andp.18551700105/abstract. I will try to get it into the page.

Best regards, Denkgenau (talk) 06:48, 3 May 2011 (UTC)
 * I corrected this. Next time you could just edit the whole article and do a cntrl-f search for the ref definition. Thank you for your effort. --Rosentod (talk) 09:07, 6 May 2011 (UTC)

Suggestions for improvement
1. The reasoning behind Fick's 1st Law is pretty evident; can someone provide an explanation or brief derivation of Fick's 2nd Law?

~Consider Fick's 1st law acting either side of a given point. Then, as I believe Darken put it: "What goes in but doesn't come out, stays there", hence Fick's 2nd! —Preceding unsigned comment added by 143.167.129.202 (talk) 10:14, 9 October 2007 (UTC)

2. I've seen another form of Fick's 2nd Law that goes:

$$\frac{\partial \phi}{\partial t} = \frac{\partial}{\partial x} (D \frac{\partial \phi}{\partial x})$$

What's the difference between this form and the form listed in the article?

Answer : $$D \frac{\partial^2 \phi}{\partial^2 x}$$ is actually a simplified form of $$\frac{\partial}{\partial x} (D \frac{\partial \phi}{\partial x})$$. The latter takes into account the fact that $$D$$ may not be spatially constant, while the first doesn't.

As an example, in an $$A_{x}B_{1-x}$$ alloy $$D$$ can be a function of x, and thus change with the spatial coordinates if the A and B concentrations are not constants along x, y or z.

132.168.109.116 (talk) 09:25, 6 July 2016 (UTC)

3. Including the del-operator forms of the equations would be helpful.


 * 1) Imagine a plane in space, and two small volumes A and B straddling the plane (one on each side). In a unit of time, a particle in volume A has the same probability to cross the plane into B as the particle in B has to cross into A. However, if there are more particles in box A than B, then on-net, more particles will flow from A to B.

Let P be the probability of any particle crossing the boundary, let NA and NB be the number of particles in volumes A and B respectively. Then the expected net number of particles to cross the boundary in a unit of time, Q, is the expected number of particles that leave A minus the expected number of particles that leave B:

$$Q = P*NB - P*NA$$

Dividing by the volume V of the box,

$$q2 = -P\frac{NA-NB}{V}$$

As the volume of the box goes to zero, NA/V-NB/V is spatial derivative of the concentration dc/dx. As the area of the boundary and the time interval correspondingly go to zero, q2 becomes the flux q. This results in Fick's Law:

$$q = -P\frac{dc}{dx}$$

This easily generalizes to higher dimensions since you could do the same thing for a plane in each of the x,y, and z directions:

$$q = -P\frac{dc}{dx} - P\frac{dc}{dy} - P\frac{dc}{dz} = -P \nabla c.$$

If, for some reason, the particles like to flow more in one direction than another (say they are molecules in a material that permits flow more easily in the z-direction than the x or y directions), then you would have a different probability in each direction:

$$q = -P_x\frac{dc}{dx} - P_y\frac{dc}{dy} - P_z\frac{dc}{dz}$$

If you hadn't oriented your coordinate system exactly along the high and low probability directions, then when you change coordinates you would get cross terms:

$$q = -(P_{xx}\frac{dc}{dx} + P_{xy}\frac{dc}{dx} + P_{xz}\frac{dc}{dx} + P_{yx}\frac{dc}{dy} + P_{yy}\frac{dc}{dy} + P_{yz}\frac{dc}{dy} + P_{zx}\frac{dc}{dz} + P_{zy}\frac{dc}{dz} + P_{zz}\frac{dc}{dz}) $$

This can be expressed more concisely as

$$q = -P \nabla c$$

if we understand P to be a matrix (the diffusivity tensor).


 * 2) This is a combination of Fick's law with the conservation of mass law, which says that the net flux through the boundary of any volume must equal the change in concentration in the volume. Mathematically,

$$\frac{d}{dt}\int_V c dV = \int_{\partial V} q \cdot \vec{dA}$$

Using the divergence theorem,

$$\frac{d}{dt}\int_V c = \int_{V} \nabla \cdot q$$

Since this holds for any arbitrary volume, we can take away the integral,

$$\frac{dc}{dt} = \nabla \cdot q = -\nabla \cdot (P \nabla c)$$

67.9.148.47 (talk) 07:37, 15 December 2008 (UTC)

4. There's an example of a 1D solution. It would be useful to include which equation does this solution solve (D constant). —Preceding unsigned comment added by 193.2.69.149 (talk) 14:15, 26 May 2009 (UTC)

5. Isn't the discussion of diffusive flux using the equation

$$J_i = -\frac{D c_i}{RT} \frac{\partial \mu_i}{\partial x}$$

a bit misleading? This makes it look like the rate of non-equilibrium diffusion is inversely proportional to temperature, T, when actually the opposite is the case. Isn't this equation derived assuming the instructive but special case of quasi-equilibrium of diffusion and an exactly matched force field countering the diffusion? In this case the stationary gradient is made shallower at "equilibrium" by the thermal disorder of increased temperature. But, unless the unusual assumptions are clearly explained (and is this worth the space?), this special case would give a reader exactly the opposite impression of the more general case where non-equilibrium diffusional flux is actually increased by higher temperatures. — Preceding unsigned comment added by 99.155.157.47 (talk) 01:28, 2 September 2012 (UTC)


 * Please take into account that the chemical potential $$\mu$$ depends both on the concentration and temperature. For the perfect gas $$\mu=RT\ln c +\mu_0$$. To get the standard Fick law this factor RT should be deleted. This is the simplest explanation of the RT in the denominator. This denominator does not mean that rate of diffusion decreases with temperature growth. If you like a more formal consideration then you have to calculate the variational derivative of S/R (dimensionless entropy). This is the driving force for the diffusion. It is $$-\nabla (\mu/RT)$$.-Agor153 (talk) 13:12, 2 September 2012 (UTC)

Error in section "Example solution in one dimension: diffusion length" (?)
I tried to apply the formulae in the section "Example solution in one dimension: diffusion length" and met two problems.

First, I am not sure the approximation by the Taylor series is correct. If x → ∞, we should have n(x,t) → 0 whereas here n(x,t) → -∞.

Second, I am puzzled by the sentence : "If, in its turn, the diffusion space is infinite (lasting both through the layer with n\left(x,0\right) = 0, x >0 and that with n\left(x,0\right) = n_0, x \le 0 ), then the solution is amended only with coefficient ½ in front of n0 (this might seem obvious, as the diffusion now occurs in both directions)"

If the solution is only amended with 1/2, then there should be a discontinuity when x → 0 as n(x,t)→n_0/2 and n(0,t)=n_0. I wonder if in this case the diffusion length should be changed.

As I am not a physicist nor a mathematician I do not know which precize corrections should be made but hope theses remarks will help to improve the understanding of the page. — Preceding unsigned comment added by 129.199.31.253 (talk) 09:58, 14 March 2014 (UTC)

Steady-state?
The first sentence under the heading "Fick's first law" reads:

Fick's first law relates the diffusive flux to the concentration under the assumption of steady state.

That Fick's first law would require the assumption of steady state does not makes sense to me - shouldn't Fick's law be assumed valid also under time dependent processes? (It may, of course, not always be valid but it doesn't make sense to me to use steady-state as a part of its definition).

This statement becomes really troublesome as Fick's first law later in the article is used to derive Fick's second law, whose essence is a non steady-state description (dc/dt non-zero).

Cheers, Martin — Preceding unsigned comment added by 85.235.1.28 (talk) 14:08, 26 June 2014 (UTC)

Derivation of first Fick's law contains wrong definition for D
Derivation of first Fick's law states that $$D = \frac{\Delta x^2}{2\cdot\Delta t}$$. Well, this is obviously wrong because you can choose arbitrary $$\Delta x$$ and $$\Delta t$$ so the value of D will be arbitrary. I think, better proof of first Fick's law is given in "Statistical Mechanics: Entropy, Order Parameters, and Complexity" by James P. Sethna - there first Fick's law is actually derived from second Fick's law. — Preceding unsigned comment added by 85.140.208.170 (talk) 14:40, 25 May 2015 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on Fick's laws of diffusion. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20090205030323/http://www.uni-leipzig.de:80/diffusion/journal/pdf/volume2/diff_fund_2(2005)1.pdf to http://www.uni-leipzig.de/diffusion/journal/pdf/volume2/diff_fund_2(2005)1.pdf

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 08:07, 31 December 2016 (UTC)

Derivation
The derivation of Fick's first law, as presented in the article, is not correct.
 * The number of particles at position x at time t is zero: N(x,t)=0. The number may only differ from zero if it's the number of particles in a volume.
 * It's unclear what is meant by the 'lenght scale' $$\Delta x$$ and the 'time scale' $$\Delta t$$. Madyno (talk) 21:57, 25 February 2019 (UTC)