OptaDOS photoemission module #38

victorchanglee · 2021-05-19T10:42:59Z

Dear all,

This includes a new module for simulating photoemission in OptaDOS.

Details of the implementation are included in documents/photoemission

A .cell .param and .odi example files are also included in examples/Photoemission

Please, let me know about any issue or question about how to use the module.

Kind regards,
Victor

This pull request includes a new module for simulating photoemission in OptaDOS. Details of the implementation are included in documents/photoemission A .cell .param and .odi example files are also included in examples/Photoemission

ajm143 · 2021-05-20T13:33:45Z

@victorchanglee I'm confused by a ddome_bin file? What does it contain, and where does it come from? An ome_bin is the optical matrix elements. A dome_bin is the diagonal optical matrix elements, so what does ddome mean? And is that a consistent naming scheme?

ajm143

Looks really nice and useful.

These are some initial thoughts on the pull request. I will do more later, but I think it's good for @victorchanglee to have a look. And @Rebecca-Nicholls also needs to see the comments.

ajm143 · 2021-05-19T13:13:27Z

optados/src/photo.f90

+  integer :: max_layer
+  real(kind=dp) :: e_fermi
+  real(kind=dp) :: q_weight 
+  real(kind=dp), parameter :: epsilon_0 = 8.8541878176E-12_dp   !A^2 s^4 kg^-1 m^-3


Constants need to live in constants.f90 and be consistent throughout the code, please.

Thanks, will move that to the appropriate file. It was initially coded so that the original optados code is modified as little as possible

optados/src/photo.f90

ajm143 · 2021-05-19T13:20:42Z

optados/examples/Si2_PDIS/Si2.bib

@@ -0,0 +1,50 @@
+@article{DFT-HK,


This shouldn't be checked in.

ajm143 · 2021-05-19T13:21:16Z

optados/examples/Si2_PDIS/Si2-out.cell

@@ -0,0 +1,46 @@
+# CELL file written 10:16:14 (GMT+2.0) 18th July 2020 from run Si2


This shouldn't be checked in.

ajm143 · 2021-05-19T13:21:56Z

optados/examples/Photoemission/Cu.odi

@@ -0,0 +1,20 @@
+TASK              : photoemission


Might be worth commenting this file more, to explain what the new keywords do here.

ajm143 · 2021-05-19T13:23:40Z

optados/documents/photoemission/user_guide.tex

@@ -0,0 +1,970 @@
+\documentclass[a4paper,11pt,twoside]{book}


I'm not convinced that we should have a separate manual. I'm quite happy for an extra chapter in the manual.

Our longer term plan is to have an html / online manual. But for the moment, I think this should just be a chapter of the manual.

ajm143 · 2021-05-19T13:24:43Z

optados/examples/Photoemission/Cu.param

+SPIN_POLARIZED       : FALSE
+WRITE_FORMATTED_POTENTIAL : TRUE
+WRITE_FORMATTED_DENSITY : TRUE
+MIXING_SCHEME : Pulay


Are all of these parameters necessary? It would be helpful not to have red-herrings in here.

You are right, they are not a requirement. They has to do more with other results in my work.

ajm143 · 2021-05-19T13:25:39Z

optados/examples/Si2_PDIS/Si2.odo

+ +============================================================================+
+ +                 Projected Dispersion Curve Calculation                     +
+ +============================================================================+
+


By all means have a section here that says, "Photoemission by V.C. N.H. et al, please cite..."

ajm143 · 2021-05-20T13:25:57Z

optados/src/build.f90

@@ -0,0 +1,17 @@
+module od_build


Please don't check this in.

ajm143 · 2021-05-20T13:26:51Z

optados/src/electronic.f90

+    return
+
+101 call io_error('Error: Problem opening cst_vel file in read_band_curvature')
+102 call io_error('Error: Problem opening dome_bin file in read_band_curvature')


Surely not a dome_bin ?

Rebecca-Nicholls · 2021-05-20T13:50:16Z

@victorchanglee Hi Victor - thank you for all of this! I'm in the middle of teaching at the moment but should be able to look at it in ~three weeks time. Andrew has sent you some comments in the meantime. Best wishes, Rebecca

victorchanglee · 2021-05-20T18:06:30Z

@victorchanglee I'm confused by a ddome_bin file? What does it contain, and where does it come from? An ome_bin is the optical matrix elements. A dome_bin is the diagonal optical matrix elements, so what does ddome mean? And is that a consistent naming scheme?

@ajm143 It contains the second order kp perturbation, which is the band curvature/effective mass. This is currently under "testing"

It is related to how the parallel momentum in ARPES is interpreted. In ARPES, the angle is determined by the parallel momentum of the electrons inside of a crystal. It is usually assign to the crystal momentum or wavevector k. However, the nature of the Bloch theorem makes k only preserved for a defined unit cell or in other words, if we define a supercell, the bands will be folded and thus changing the interpretation of k.

Related to this issue, several band unfolding schemes has being developed (for example, CASTEP has a band unfolding tool), however, in my thesis I took a different look at this momentum interpretation of the band structure.

I used two different approaches, the first one is applying the momentum operator to get the momentum expectation value, which is in fact, the kp first order perturbation already computed in CASTEP in .dome

The second approach I used is a semi-classical momentum, for which I used the band gradient to obtain the electron velocity and the band curvature to get the effective mass. The band curvature is computed as the second order perturbation in the kp method, for which Phil Hasnip and Peter Byrnes provided me with a piece a code to modify Castep.

There is a key word "momentum" that switch between one momentum interpretation or the other. It is under testing since both three approaches gives different answers.

ajm143 · 2021-05-25T13:40:10Z

@pjh503 This is technically a CASTEP issue. But seems to fit better with this PR.

We seem now to have a ddome filename. I think that name could be confusing.

ome are the optical matrix elements j,i
dome is the diagonal optical matrix elements I=j. Hence d stands for diagonal
ddome second order k.p perturbation, I=j. Hence the first d stands for derivative, and the second d for diagonal.

I'm worried that this might get messy if we use d to mean two different things. What are your thoughts?

pjh503 · 2021-05-28T09:38:35Z

I agree that ddome_bin would be confusing, the question is what it should be called. I prefer "d" to mean "derivative", but unfortunately we've already used it for "diagonal".

I would also prefer the name to reflect the fact that it is a derivative with respect to "dk", as we can imagine cases whether we look at the derivative with respect to other things (e.g. strain, phonon mode).

ajm143 · 2021-05-28T09:42:24Z

.dome_dk ? I agree we're using d to mean two different things here... maybe we get away from omes and call it de_dk or similar?
.bands_dk ?

ajm143

The looks good, and is almost totally in the OptaDOS coding style (despite us never publishing such).

There seems to me to be a lot of redundant code, in photo.f90 that could just call other functions instead. You shouldn't feel bad about generalising functions in other modules to make them do what we both want. It will be much easier for us all to maintain.

How do we test your code? What version of CASTEP generates the necessary OptaDOS input files?

I'm very keen to get this code into the stable branch.

ajm143 · 2021-08-12T13:24:03Z

optados/src/photo.f90

+
+    epsilon_zero = 55.26349406_dp
+    ge = 10.24633_dp
+    kB=8.617333262E-5_dp


Put constants in the constant module please.

I moved all the constants to the constant module.

How do we test your code? What version of CASTEP generates the necessary OptaDOS input files?

I will prepare some input/output examples for the code. There are two input files required to run some of the options in the code that are generated by a modified version of CASTEP 18.1

One is the free electron matrix, which I generated by modifying the spectral.f90 in CASTEP, in which the final state is substituted by a free electron planewave.

the second file is the band curvature which I also modified the spectral.f90 in CASTEP for this purpose.

I think ideally both files should be written as a tool file in CASTEP but at the time, it was faster and easier for me to just modify the spectral.f90

I think the CASTEP code freeze is coming up. When will your modifications to CASTEP be in a release ?

ajm143 · 2021-08-12T13:24:55Z

optados/src/photo.f90

+      width = (1.0_dp/11604.45_dp)*temp
+      qe_factor=1.0_dp/(4*pi**2*s_area)
+      norm_vac = gaussian(0.0_dp,width,0.0_dp)
+      kB=8.617333262E-5_dp


put contestants in constants module please.

optados/src/photo.f90

ajm143 · 2021-08-12T13:30:12Z

optados/src/photo.f90

+
+    real(kind=dp) :: z,epsilon_zero
+
+    epsilon_zero = 55.26349406_dp


please put the constants in constants.f90

ajm143 · 2021-08-23T09:43:45Z

optados/documents/photoemission/user_guide.tex

@@ -0,0 +1,970 @@
+\documentclass[a4paper,11pt,twoside]{book}


Our longer term plan is to have an html / online manual. But for the moment, I think this should just be a chapter of the manual.

ajm143 · 2021-08-23T09:57:57Z

optados/src/photo.f90

+
+  !***************************************************************
+  subroutine calc_absorp_photo
+    !***************************************************************


ajm143 · 2021-08-23T09:58:10Z

optados/src/photo.f90

+  !***************************************************************
+    subroutine calc_reflect_photo
+  !***************************************************************
+  ! This subroutine calculates the reflection coefficient


ajm143 · 2021-08-23T09:58:39Z

optados/src/photo.f90

+
+ !===============================================================================
+  subroutine setup_energy_scale
+ !===============================================================================


This is a copy? Why not call the original ?

ajm143 · 2021-08-23T09:59:45Z

optados/src/photo.f90

+  !===============================================================================
+  subroutine jdos_utils_calculate_delta(delta_temp)
+  !===============================================================================
+  ! Copy from the jdos_utils.f90 


Why not just call jdos_utlis ?

This function is slightly trickier to call.

In the original jdos_utils, the delta function is computed on the fly (it is not allocated) in a variable called dos_temp to generate the weighted_jdos which is integrated over the BZ.

In photoemission, I want to generate the photoemission intensities as function of k, so I need the Dirac delta function to be allocated.

I think the solution is to allocate the dos_temp in calculate_jdos.

ajm143 · 2021-08-23T10:00:47Z

optados/src/photo.f90

+  end subroutine transverse_energy_spread
+
+  subroutine binding_energy_spread
+  ! This subroutine applies a Gaussian broadenning to the binding energy


This looks like a copy of the function above? Why not make a function that does both ?

I made changes so now it only generates a plot of the photoemission intensities (QE) as a function of the binding energy, which is what experimentalist usually obtain from UPS/XPS/ARPES.

Several changes: - Constants moved to constants.f90 -Merged the optical properties in the photoemission properties with the now public functions in optics.f90. There are still a few functions that are copies with small changes. I included all the input requires to run the photoemission module for a graphene sheet, including the free electron matrix. I will try to update the code with further small changes.

ajm143 · 2021-09-10T09:08:17Z

optados/src/electronic.f90

@@ -405,7 +405,7 @@ subroutine elec_read_band_curvature

       if(on_root) close (unit=curvature_unit)

-       ! Convert all band curvatures to eV Ang
+       ! Convert all band curvatures to eV Ang^2
       band_curvature=band_curvature*bohr2ang*bohr2ang*H2eV


Be very careful here. OptaDOS does a double conversion here bohr2ang twice because of an error in the units of the optical matrix elements coming out of CASTEP. Are you sure that the band curvature has the same error?

I think it has same error. I compared the calculated band curvature from a near-free electron band calculated as a chain of Li atoms in CASTEP with a polynomial fitting of the same band.

After unit conversion, the difference in curvature with both methods is around 0.02, which I think it is more likely to be an issue of the polynomial fitting.

OptaDOS photoemission module

1c7a715

This pull request includes a new module for simulating photoemission in OptaDOS. Details of the implementation are included in documents/photoemission A .cell .param and .odi example files are also included in examples/Photoemission

ajm143 assigned jryates, ajm143 and Rebecca-Nicholls May 20, 2021

ajm143 requested review from jryates, ajm143 and Rebecca-Nicholls May 20, 2021 13:09

ajm143 requested changes May 20, 2021

View reviewed changes

ajm143 reviewed Aug 23, 2021

View reviewed changes

ajm143 reviewed Sep 10, 2021

View reviewed changes

Merge branch 'develop' into develop

06e544b

		@@ -0,0 +1,46 @@
		# CELL file written 10:16:14 (GMT+2.0) 18th July 2020 from run Si2


		real(kind=dp) :: z,epsilon_zero

		epsilon_zero = 55.26349406_dp

OptaDOS photoemission module #38

Are you sure you want to change the base?

OptaDOS photoemission module #38

Conversation

victorchanglee commented May 19, 2021

ajm143 commented May 20, 2021 • edited Loading

ajm143 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Rebecca-Nicholls commented May 20, 2021

victorchanglee commented May 20, 2021

ajm143 commented May 25, 2021

pjh503 commented May 28, 2021

ajm143 commented May 28, 2021 • edited Loading

ajm143 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajm143 commented May 20, 2021 •

edited

Loading

ajm143 commented May 28, 2021 •

edited

Loading