How to set the K point sampling for DOS calculating(version 10.8)

[eastnobil]

For ATK10.8, the K point samplings for DOS calculating and caculator setting are segregative. How to set the K point sampling for DOS calculating? Is it the same with the calculator setting? I has calcualted a semiconductor GNR(graphene nanoribbon). From band structure the energy gap is observed. But the energy gap miss when the same K point sampling is set for DOS calculating.

[Anders Blom]

It's set in the same way as for the calculator, i.e. you specify a Monkhorst-Pack grid with a certain number of points. The proper setting for a 1D ribbon would thus be (1,1,N). Beyond that it's hard to venture a guess without seeing a script or a plot.

[eastnobil]

Thanks for your reply. The calculating detail will be depicted as following：
This is my calculating example which is an armchair graphene nanoribbon (see AGNR.jpg)
The calculator is set to "Extended Huckel" and the K Point Sampling is set to 1X1X15. After running this job, the bandstructure is obtained(see Bandstructure.bmp). The energy gap is found from that plot.
Using the same K points sampling, the DOS plot is shown in DOS(1X1X15).jpg. From this plot, The energy gap can not be found. When the K points sampling increases to 1X1X30 (see DOS(1X1X30).jpg), the energy gap appears.

Why can not observe the right result from DOS plot by setting the same K points sampling as calculator?

[Anders Blom]

I see your dilemma. The "number of k-points" you need to compare is not those used for the calculation, but those used for the band structure. If you have 100 k-points on the band structure plot, then you need the same for DOS to make it reflect the same results.

Ultimately, as will all quantities, it's a matter of increasing the k-points until the results don't change. Clearly 1x1x15 is too little, there is almost no structure in the DOS at all.