Discrepency in Effective mass calculation

[PR]

Hello,
I am calculating the effective mass (m_l and m_t) for InAs bulk in Gamma and L valleys, using both of the two methods given in ATK 2014.2 version. The values in Gamma valley is quite close to the literature [1], but for L valley there is a discrepency in values.
Direction used :  [111] for m_l and [-110] for m_t for L valley.
Methods for effective mass calculation:
(1) Effective mass analyzer under tools tab is giving  m_l = 1.827  and m_t=0.121.
(2) Effective mass calculating directly from plugin given at bottom right corner in the band structure plot window is giving  m_l=0.609 and m_t=0.305.
While the values are m_l = 0.64 and m_t = 0.05  in literature [1]

Can you please tell me why the tool is giving different results with different effetive mass analyzers and which one is more reliable.
Also, why the values are not consistent with literatre. Where I am doing mistake or what  should I change to get the correct results.
Attaching the python script of the calculations.
Reference:
[1] I. Vurgaftman, J. R. Meyer and L. R. Ram-Mohan "Band parameters for III–V compound semiconductors and their alloys" J. Appl. Phys. 89 5815 (2001)

[Jess Wellendorff]

Hi,

1) I ran your script and find that both of the methods that you list (effective mass analysis object vs. effective mass from the band structure analysis tool) gives identical effective masses (1.827 and 0.121). Perhaps you did not specify the correct directions in method 2?

2) The values you mention for comparison (0.64 and 0.05) appear to be experimental values. It is very hard for theory to get the X and L valleys correct wrt. to experiments. Even though you fit the TB09LDA c-parameter to experimental Gamma-point band gaps and/or effective mass, the same quantities in the X and L points are not guaranteed to be correct. According to  https://engineering.purdue.edu/gekcogrp/publications/pubs_src/DOC119_e125207.pdf, the target theoretical value of the longitudinal effective mass at L is 1.90, which is not much different than the 1.83 you find with ATK.

[Forum Administrator]

One should be a bit careful with Vurgatfman's paper. If you read carefully the notes (always do that!) about the L masses in InAs, you will find that they are neither experimental nor computed, but "guesstimated" from the DOS effective mass.

That said, 1.8 is very high, so indeed the curvature is not very accurate with the basis set you use (did you try other ones?).

[Jess Wellendorff]

FHI-DZP and OMX tier0 and tier1 basis sets seem to agree pretty well wrt. the curvatures for InAS.

[PR]

Thank you all for your replies.
@ Admin, I have not tried for other basis sets, but as suggested by Jess Wellendorff, I will try FHI DZP and OMX tier basis sets.


Regards
PR.

[PR]

================================
For the X valley of bulk InAs:
values obtained using ATK: ml=mt=0.37, the value doesn't change on changing the direction ([010] for ml and directions perpendicular to it for mt), suggesting a isotropic mass, which is wrong!
This is also in contrast with the data (target values: mt=0.23, ml=1.3) in the reference cited in your reply (Boykin et. al, PRB 2002).
Could you please calculate at your end let me know the values you are getting?

Thanks.

[Anders Blom]

The X point in fcc is at 101 in fractional coordinates, so that's the direction you need for mL. So it's actually mT which is 010.

[PR]

I have again calculated the effective mass values for X valley in [101]  direction for m_l and getting value of m_l = 52.0858 and in [010] for m_t= 0.375864.
Here m_l is again questionable. It would be helpful if anyone can tell me now where I am doing wrong for m_l calculation.


Thanks
PR

[Jess Wellendorff]

Did you try using the HGH pseudopotentials (as in this tutorial: http://quantumwise.com/documents/tutorials/latest/InAs-2D/index.html/chap.bulk.html) and optimize the metaGGA c-parameter wrt. description of the X valley instead of the Gamma valley?