Driving-force output

[jan]

Hi all!

I recently got the impression that many MICRESS users do not use the .driv output or do not know exactly what it is for. From my point of view it is one of the most important outputs, so I just want to explain what it is and what I personally use it for. I will explain it for the case of concentration coupling, for other types of coupling it should be similar.

The driving-force output is a field output which gives you the value of the chemical driving-force deltaG in the interface region. To be exact it is just the "bulk" of the interface, i.e. not the cells which have zero fraction in one phase but still are in the interface list. Therfore the interface region apperars somewhat broader in the .intf output than in the .driv output. In tripel or higher junctions you cannot get all driving-forces for the pair-wise interactions but only the 'highest' one, e.g. for 1/2 if you have a triple juction built up from grain 0, 1 and 2.
You can transfer deltaG to a undercooling by dividing with the transformation entropy deltaS, which you either have to input in linearized calculations or which you get written in the .log file for thermodynamic coupling.
Because the chemical driving force is a function of the local concentrations and the local concentrations are in local equilibrium conditions a function of curvature you can see the curvature effects in the .driv output. In fact the observed deltaT (= deltaG/deltaS) is always the sum of a kinetic and a curvature contribution:
deltaT = deltaTKin + deltaTCurve

So in sum there are a lot of informations from the .driv output:

- if the interface mobility is high, deltaTKin is very small and you see directly the curvature undercooling or driving-force. (If you do not use averaging or the driving-force along the interface gradient, deltaTKin will be only zero for the average but there will be still opposite values in the inner and outer side of the interface)

- If the mobility is low, deltaTKin maybe much higher than deltaTCurve, then you will see mainly the kinetic undercooling

- If your interface is not curved in some regions of your simulation, you always will see only the kinetic contribution. As a general rule for diffusion-coupled growth the kinetic undercooling should be in the order of or less than the curvature undercooling, but off-course that depends also on your individual simulation. But especially if the driving-force values are steadily increasing with simulation time, this is a good sign that your mobility value is too small. If you are using a temperature-dependent mobility, then you will barely be able to ajust this mobility curve reasonably without monitoring the .driv output!

- If you choose a too high mobility your interface may get instable. This instability can be first seen as a freckling of the .driv output in the corresponding interface which indicates strongly fluctuating values. In many cases if you get whatever instability, it can be first seen in the .driv output!


So in summary the .driv output is essential for adjusting interface kinetics and for getting numerically stable simulations. If you found out other features or applications please let us know!

Bernd

---

original message from Bernd

[jan]

Hi Bernd,

thanks for the nice explanation.
Now we know how to change the interface mobilty in order to get into the
diffusion-controlled phase transformation limit.

But how do we see if our mobilty is too high? Are there also hints from the *.driv output?

Nils

---

original message from Nils Warnken

[jan]

Well,

as I already explained above a too high mobility can lead to instabilities, which are easily visible as 'freckles' in the .driv output. Another hint is the driving force at non-curved regions of the interface. If here the values are very small (compared to typical curvatures) you may like to afford a lower mobility.

If you refer to how to get quantitative growth kinetics: This cannot be done based on the .driv output. Instead you will need a somewhat more complicated procedure ( See here: http://board.micress.de/viewtopic.php?f=17&t=17 )

Bernd

---

original message from Bernd