QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: Dipankar Saha on July 1, 2015, 22:31
-
How can I get some states, at negative energy values....in the Ph.-DOS diagram??
(Even though the "Acoustic sum rule" is checked.....!!!!)
Best_
Dipankar
-
In ATK..... whatever "transmission eigen values" that we get ......., how they are related to
the Trace[Gr Sigma1 Sigma2 Ga] .....??
Thanks & Regards__
Dipankar
-
Most likely, the atomic positions in your structure are not accurately optimized before calculating the PhononDOS.
If that still does not help, and if you are using classical potentials with a long range (e.g. Coulomb interactions) then you need to increase the maximum interaction range parameter to be larger than the largest cutoff distance of the potential.
-
Most likely, the atomic positions in your structure are not accurately optimized before calculating the PhononDOS.
If that still does not help, and if you are using classical potentials with a long range (e.g. Coulomb interactions) then you need to increase the maximum interaction range parameter to be larger than the largest cutoff distance of the potential.
Thank you Julian for your reply....../ Optimization May not be the reason in this case.... / However, I didn't get that.... while using classical potetials...... how can I "increase the maximum interaction range parameter" (Because, in calculator, all I have .... "potentialSet= " and "calculator= ").....??
-
The maximum interaction range is an optional parameter that can be set in the DynamicalMatrix analysis object. When calculating the dynamical matrix, this parameter specifies the distance around a displaced central atom (i) in which the derivative of the forces on atoms (j) are taken into account to calculate d^2 U / dr_i dr_j .
If this parameter is not specified, the default values are used (which are the covalent distances multiplied by a fuzz factor of 4.0). In most cases this works well, but when e.g. a classical potential is used, which has a large cutoff radius (e.g. CoulombDSF (http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.coulombdsf.html (http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.coulombdsf.html)) with a default of 9 Angstrom), then the maximum interaction range parameter has to be increased to be larger than this cutoff radius, i.e. larger than 9 Angstrom, to avoid discontinuities.
You can find out if large cutoff values are used by your potential, if you send the script to the Editor with the "Script details" parameter set to "Show defaults". Then all the details of the classical potential are reflected in the script.
Which exact potential are you using?
Apart from that, negative frequencies might also occur if you are sitting exactly at a local maximum (e.g. due to symmetries in your structure). In this case, small random displacements to the atoms followed by another accurate geometry optimization should lead you to a minimum.
-
Thanks a lot Julian for this detailed answer...... :) / It's actually the parameterization of SW.... (i.e., StillingerWeber_MoS_2013) ..../ I could not really follow the steps....as you said, like... "if you send the script to the Editor with the "Script details" parameter set to "Show defaults"...then all the details of the classical potential are reflected in the script...." ; but still looking at source paper, I find that they....fitted the phonon spectrum with SW parameters for the two-body and the three body(angle bending as well) potentials.....with a max. cut-off of 4.27 Ang. / In such a case... should I need to change the default "atomic_displacement=0.01*Angstrom" in dynamical matrix..??
Regards_
Dipankar
-
For the Stillinger-Weber and Tersoff-potentials the default values usually work well. You don't need to specify the maximum interaction range in the DynamicalMatrix parameters.
In your case, I'd rather suspect that there is something odd with the structure. Negative frequencies mean that there are soft modes present in the system, either by atoms sitting in a local saddle point.
You could try to calculate the phonon bandstructure (without checking Accoustic sum rule and Symmetrize) and then the location of the negative frequencies might give you some hints. If the negative frequencies occur at gamma, then you might need to increase the maximum interaction range, if the negative frequencies occur elsewhere, then it is likely that your structure is not in a local minimum.
-
Thanks a lot.... Julian..... :)
That's a nice way.....!!! / Let me....try that.... I will let you know...
Best _
Dipankar
-
Interestingly.... making the "acoustic_sum_rule= ", symmetrize= " True or False... produces no change in dispersion...!!! :o / On the other hand .... "max_interaction_range= " set to default.... or, a very specific range of values ..... could only provide the reqd. Ph-DOS... (otherwise it's arbitrary)!!!
Instead...I went for further opti. of the strcut. ...which somewhat soothes out the density states at negative energy values(see, the attachment)...../ So your guess, that it is due to the struct. ....seems to be correct.... :)
-
In ATK..... whatever "transmission eigen values" that we get ......., how they are related to
the Trace[Gr Sigma1 Sigma2 Ga] .....??
Besides, it will be of great help ....if you can please comment on this.....!!! :)
-
Dr. Julian Schneider,
For any particular energy 'E', one way to find the Transmission co-eff. is......
T(E) = Summation of [tk tkdagger delta(E-Ek)] for all 'k' states at energy 'E'
Which is again same as..... Trace[Gr Sigma1 Sigma2 Ga] .... :)
On the other hand, we have transmission eigen values of the transmission matrix...which can also be summed up...
for getting the transmission co-efficient...
1) How can one get these tk s (or, as you say transmission amplitudes).....??
Actually, why I'm looking for this...is simply to find an analogy...with the ideal transmisson....either 0 or, 1.
2) Is it incorporated through the delta function....??
-
See http://quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.transmissioneigenvalues.html
-
Dr. Anders Blom,
To be more specific,
can I extract the individual channel contributions form the total transmission....??
If yes.... then, which is the parameter...I should look into ??
Thanks in advance...... :)
Best_
Dipankar
-
That is precisely what the transmission eigenvalues are
-
Okay.... :) / Thanks a lot...!!!
-
Dr. Julian Schneider,
I wanted to know that_
1) For...obtaining kph(T) .....there should be some del_T between the electrodes {+(del_T)/2 and -(del_T)/2} .... / When we invoke a method called "thermalConductance" on the "phonon_transmission_spectrum" object .....how these TL and TR are set....(say for any specific T= 300 K) ??
2) Besides, maintaining the same analogy as electron transmission....here, EI is replaced by w2 M .....
How this M is formed....??
It will be of great help..if you can please answer those.... :)
Thanks & Regards_
Dipankar
-
1) An instead of using finite temperature difference between the phonon occupancies of the reservoirs, the differential ( nL (TL) - nR(TR) ) / (TL - T R) is evaluated as dn(T)/dT at the specified temperature.
2) As it says in the manual M is a diagonal matrix with matrix elements corresponding to the masses of the atoms.
-
Thanks Julian for your reply..... :)
-
Keeping the same analogy as electron transmission...... EI is replaced by w2 M ..... / How that is diff. from a case.... where in E2I ....the E gets replace with h_bar*w ??
Thanks & Regards_
Dipankar
-
Please do reply...!!! :)
_______
Besides,
Do you calculate the 'dynamical matrix'... with help of the force constant matrix??
-
You need to consult the ATK manual before asking these questions.
1) http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.phonontransmissionspectrum.html (http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.phonontransmissionspectrum.html) explains the main parts of the phonon transmission spectrum calculation, including the similarity to the electronic transmission spectrum. It also refers the reader to Ref. 69 for even more details.
2) Yes, as explained in the reference manual, http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.dynamicalmatrixparameters.html (http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.dynamicalmatrixparameters.html), the dynamical matrix is calculated from the change of all forces on all atoms when all other atoms are displaced a tiny bit.
-
Thanks a lot Jess for the replies ....!!!
___
However, in your reply 1) ....you mentioned a tutorial as well a paper (by Markussen et al.) ...which describe almost the same...as long as the answer to my question is concerned ....!!!
Rather, the answer would have been ....somewhat related to the Eigenvalues of the density matrix....!!! (which you have tried to point out ...as a reply... to my other query)
Anyways.... thanks again.... :)
-
By the way...... I think there is a mistake (although trivial...but, still....!!).....in the final expression of thermal conductance as written in the following tutorial__
http://quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.phonontransmissionspectrum.html
Best_
Dipankar
-
As shown in the Fig. attached ..., there occurs a negative freq. near the gamma point (although, the 'ASR' and the 'symmetrize' are checked). How to avoid this....? ( May be this not due the stability of the struct. .....)
Best_
Dipankar
-
If this is a phonon bandstructure, negative frequencies implies that the structure is less than perfectly relaxed.
-
I guess not..../ Perhaps, in this case, it might be related to the accuracy of the numerical calculation...!!!
-
ok
-
I tried various options_ e.g., increasing the repetition in Dij , mesh cutoff etc. ? Only thing left is the "max. interaction range"..... / As it's not out of any classical-potential based calculation....thereby, I'm not quite sure here.... about the value that needs to be set.....
-
Do you use a sufficient cutoff energy and k-point sampling in the relaxation and in the phonon calculation?
-
k-point grid is fine....!!! But, how to know about the " sufficient cutoff energy"?? Then obviously, you have to look into some other parameters for their convergence...../ I said... I increased the mesh cutoff....; meaning that even with the lower energy cutoff value...the results (of interest) were already well converged....
Regards_
Dipankar
-
So if you increase the mesh cutoff and restart your geometry relaxation from the relaxed structure you already have, you see no residual forces?
-
Dr. Jess Wellendorff,
No, I didn't restart the geometry relaxation again with the higher mesh cutoff.... / I was just using the previously relaxed struct. with a higher mesh cutoff... to avoid the small bulge exactly around the gamma point.....
Besides_
what's the role of ASR then??
Best_
Dipankar
-
ASR?
-
The accoustic sum rule correction corrects the non-zero accoustic frequencies at the gamma point. However, your gamma point frequencies are all zero. The negative bulge appears at finite wavevectors.
Assuming that it is not a numerical effect, and you have optimized your forces sufficiently, there is still the chance that the structure is not in a minimum but at a saddle point, e.g. stabilized by the symmetry.
In this case further optimization does not help, but you might try and rattle the atoms a bit to break the symmetry and then optimize again. If the resulting structure is different from the original structure, then this might be the origin for the negative frequencies.
Moreover, although it is generally possible to calculate phonon frequencies for strained systems, a compressed cell sometimes enhances such effects (see e.g. the tutorial http://www.quantumwise.com/documents/tutorials/latest/Phonon/index.html), especially for lower dimensional systems, such as graphene, which tends to buckle if you compress the cell. In this case, optimizing the cell can help to remove negative frequencies (I don't know if you already tried optimizing your cell).
Finally, be aware that you optimize the small cell, whereas the actual phonon calculations are carried out in the repeated supercell. If you have instabilities that occur at a larger scale, they may be stabilized by the periodicity of the small cell. A good example for such behaviour is again the graphene sheet that tends to buckle (e.g. because you compress the cell). You won't see the buckling in the small cell, because the wavelength of the buckling is larger than the cell length, but in the repeated supercell, the buckling may well be possible.
It don't know if any of these causes apply to your system, but it is something to think about when you encounter negative phonon frequencies.
-
Thanks a lot Julian for all the details..!!! :)
..................
Okay, let me try that...../ I will rattle the struct. a little bit...., and then again optimize...!! /
{ Yes obviously the cell was properly optimized (both force and stress)....; and the relaxed struct. could accurately reproduce the results (electronic) .... / Even for the phonon dispersion, any other occurrence of negative bands had been completely avoided....../ Now the only issue is that, the small bulge around the gamma point.....}
Anyways, I will let you know the outcome...... :)
Regards_
Dipankar
-
Besides, you didn't say anything about the max. interaction range value...../ I guess ( for this particular case), you are not suspecting it as the probable cause...!!!
-
If you have imaginary frequencies at the gamma point (if you switch off ASR correction), then this is often caused by the max_interaction_range. In your case, I would not suspect the max_interaction_range in the first place, although I wouldn't completely rule out this possibility.
-
Yes, you are correct..... If I rattle , and re-optimized....although it's getting back to the previously optimized struct.... (and, the lattice constants etc. are also remaining the same); but the co-ordinates are getting slightly altered...!!
For example, in case of any particular atom_
0.249999774742, 0.654389668865, 0.503484641034 is getting changed to ~
0.252655254706, 0.651356151454, 0.504421688472 ...... etc.
Thanks & Regards_
Dipankar
-
Hello,
While forming the dynamical matrix.... we usually set a repeat parameter. Now, consider..... I have a 2D super-cell (large enough), which I need to treat as the unit cell for the Frozen Ph. calculations .....what should be the min. repetition 1x3x3 (x,y,z) ?? ? /
(How the calculated result will be affected...if I take...the custom repetition 1x1x1 ?... Interestingly, in that case also...the total no. of atoms gets multiplied by a constant no. .... why?)
Regards_
Dipankar
-
1x1x1 never works, unless you are only interested in the Gamma point.
1x1x1 will not give more atoms, where do you see that it does?
-
Thank you Anders.... :) / That's the thing I wanted to know..... / I'm sorry...I wrongly said "no. of atoms".... Rather, I wanted to mean the no. of displacements.....
Best_
Dipankar
-
Sure, you always have to displace all atoms (6 times) to compute the dynamical matrix.