September 2010

September, 2010

Hi Everyone,

The third Thursday crept up on me this month, arriving only 15 days

into September, so for those of you on the African side of the world,

it's Friday by now. I hope you are all doing well, and that you've

had a chance to use some of the things you learned at the workshop.

Next month I'll be traveling when the third Thursday rolls around,

so Sinead will take over the newsletter.



Alison Hatt


1) Long-distance help with calculations

2) Free journal subscriptions and the arXiv

3) Article of the month: Calculating Band Offsets in Heterojunctions

4) Group-project update


Since the workshop, we mentors have been delighted to receive emails

from people asking for help with their Quantum Espresso

calculations. We want our interactions to be as beneficial as

possible, and you would probably like to get help quickly without

waiting for several days of back-and-forth email with someone in a

different time zone. Toward that end, here are a few things to keep

in mind when emailing us about problems with your calculations.

1) Send all relevant information. In order to figure out what's

going on and help trouble-shoot the calculation, we typically need

to see all the input and output files, the job script (if you are

using one), and any errors written to standard out (on the screen).

Often it will be easiest to make a tarball so you just have one

attachment. (gtar -zcvf <foo.tgz> <file1> <file2> ....)

2) Carefully describe your problem. What exactly are you trying to

achieve with the calculation? What have you tried so far? Very

often I find that the process of writing a careful email actually

helps me figure out a problem for myself, as it forces me to slow

down and really consider what's going on.

3) Provide feedback after receiving help. Did our suggestions fix

the problem? Did you figure out a different solution? This gives

us a chance to learn from you, too, and ensures that we don't keep

working on something you've already solved.

4) Consider cc'ing more than one mentor on your email. That way,

you can get help from the first person available instead of waiting

for a response from someone who may be in the middle of writing a

proposal. You also tap into a bigger pool of knowledge, in case one

of us has seen a similar problem before.

Good luck with your calculations, and keep the questions coming!


Last month I promised that the September newsletter would include

information about how to get research articles if you can't afford a

journal subscription. Here I'll share with you the little I know or

have been told, but I invite people to send me their tips as well.

Eventually I will compile all our combined knowledge on the


The first suggestion, which Prof. Martin told us about at the

workshop, is to request special access directly from the journal.

Apparently, the American Physical Society journals, like Phys. Rev.

Lett. and Phys. Rev. B., will provide free subscriptions to

universities in Africa. In order to get a subscription, a librarian

from the University must contact the journal and request it, so you

may need to get this process started yourself. I've yet to learn if

other journals have a similar policy, but will investigate this.

Another option is to look on the physics arXiv, which is an on-line

repository of articles in physics, math, and computer science.

There isn't a peer-review process for the arXiv, but many authors in

condensed matter physics put their journal articles on there even

before they are published. ( By the way, the name

is an awesome joke for physicists: Because we are familiar with the

Greek letter Chi, which looks like an 'X', we know that the word

"arXiv" is actually pronounced "archive". If you just laughed, you

are probably a nerd.)

If you have other suggestions for gaining access to journal articles

from Africa, please send me an email, write a comment on the

website, or start a discussion on the forum.


From Kris Delaney:

"Band Offsets in Lattice-Matched Heterojunctions: A Model and

First-Principles Calculations for GaAs/AlAs", A. Baldereschi, S.

Baroni and R. Resta, Phys. Rev. Lett. 61, 734 (1988).

This classic article introduced what has since become the standard

method for computing band offsets at planar interfaces using

first-principles electronic structure methods. When two dissimilar

semiconducting/insulating materials are joined at a planar

interface, the electrical behavior is largely determined by the

alignment of valence and conductions bands at the interface. This

is a well-known situation that occurs in all semiconductor devices,

ranging from CPUs to semiconductor lasers to thin-film photovoltaic

cells, and its accurate description is crucial to performing useful


Howeer, electronic-structure calculations of bulk (infinite)

crystals have no absolute reference energy, meaning the band

energies can be arbitrarily shifted without changing and physical

properties. This means that the only way to compute band offsets is

to explicitly perform a calculation with a supercell containing the

interface. (Surprisingly, despite the breaking of periodicity and

the long-range nature of the Coulomb interaction, one can reliably

compute the band alignment of arbitrarily thick slabs using a

supercell with very few layers of each material.)

Obtaining the band offsets from an electronic-structure calculation

of a supercell is simply a matter of tracing some suitable reference

energy across the interface, but these references usually contain

spatial oscillations from the underlying crystal structure. This

paper introduces the macroscopic averaging technique, which is a

reliable method to "smooth out" spatial oscillations in reference

levels, leading to accurate and unambiguous calculations of

interface band alignments.


The group project from the workshop is still underway! We are

working to put together a summary of the progress, to be sent by

email or posted on-line, so keep an eye on the website announcements

for information.