Sunday, May 17, 2015

HW 7.

Some people asked for extra time and I think it would be okay to just turn this in Wednesday in class.

For any of these problems, assume a scattering rate of \(\tau = 10^{-13} s\) unless otherwise specified.

1. a) Show that you can combine two 1D Bloch states to obtain a standing wave.
b) How are their k values related?

2. Suppose the E vs k relation for the partially filled band of a metal is:
\(E(k) = E_o - (B/2)  cos(ak)\) where k ranges from \(-\pi/a\) to \(+\pi/a\) and a = 0.157 nm.
a) What is the relationship between the density of atoms, N, and the integral of g(k) dk from \(-\pi/a\) to \(+\pi/a\) ?
b) If the density of electrons is N, what is \(k_f\)?
c) If the density of electrons is N/2, what is \(k_f\)?

3. (Same system) When an electric field is applied, the Fermi boundaries (in k space) shift.
a) For an electric field of 10 Volts/cm applied, by what amount is the Fermi boundary shifted if the density of electron is N? Illustrate.
b) For an electric field of 10 Volts/cm applied, by what amount is the Fermi boundary shifted if the density of electron is N/2?

4. When the Fermi boundary is shifted, electrons can be viewed as belonging to two classes: those that have an occupied partner state with which they can form a standing wave state and those that do not. What percentage electrons fall into each class for:
a) n=N
b) n=N/2

5. Some people say that the current density of metal is proportional to: the density of electrons in states that have no partner, multiplied by the fermi velocity.  (Let's use a bandwidth of 6 eV for this problem.)
a) Argue for or against that as a good approximation. What other factor do you need?
b) What is the fermi velocity for a half-filled band (n=N)?
c) What is the fermi velocity for a quarter-filled band (n=N/2)?
d) How does that compare with: the thermal velocity of electron in a semiconductor with an effective mass of 1? the average speed of an electron in a semiconductor conduction band with an electric field of 10 V/cm applied?

Fermi Boundaries:
A Fermi boundary in k-space is the set of points that separate a regions of occupied states from a region of unoccupied states. In 1D it is just two points -- one at kf, the other at -kf. In order to explore the concept of a Fermi boundary we therefore need to go up to at least 2D.

6. Consider a two-dimensional (2D) metal for which the E vs k relationship for the conduction band  is: \(E(k) = E_o -  b cos(ak_x) - b cos(ak_y)\) where kx and ky each range from \(-\pi/a\) to \(+\pi/a\) and a = 0.157 nm. This range of k includes all the states of the band (2N states), however, they are not all filled.
a) What is the bandwidth of this band?
b-e) Figure out and illustrate the Fermi boundary for the cases where the band is roughly: 1/8, 1/4, 1/2 and 3/4 filled, respectively.*
f) Which one of these is simplest? Which one of this is not like the others? Describe and discuss your results and illustrations.
g) To what 2D lattice structure do you think this E vs k relationship corresponds? How come?
* If doing this for a specified filling fraction (1/8, 1/4...) is too difficult, then maybe we could use a sequence of fermi energy values instead. Would that be more manageable? What do you think? What might be a reasonable selection of values of Ef that meet our goal of exploring?

7. One can make this anisotropic in the following way:
 \(E(k) = E_o -  b cos(a_xk_x) - \frac{b}{4} cos(a_yk_y)\)
a) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o -b/8\).
b) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o -b/4\).
c) Figure out and illustrate the Fermi boundary for the case where the fermi energy is ... your choice.
d) What would you guess is a lattice structure to which that this might correspond??

6.2 (problem added on Wednesday)  I think it might be interesting to also examine the fermi boundaries for an "upside down" E vs k relationship. That is, suppose:
\(E(k) = E_o + b cos(ak_x) + b cos(ak_y)\) where kx and ky each range from \(-\pi/a\) to \(+\pi/a\) and a = 0.157 nm.
a-d) Figure out and illustrate the Fermi boundary for the cases where the band is roughly: 1/8, 1/4, 1/2 and 3/4 filled, respectively.
e) Discuss

8. The previous problems are for a simple structure, but it is not a very realistic or stable structure  (because the atoms are not very efficiently (closely) packed). A more realistic 2D metal could have an E vs k relationship for the conduction band of:
\(E(k) = E_o -  b cos(ak_x) - b cos(\frac{1}{2}ak_x + \frac{\sqrt{3}}{2}a k_y) - b cos(\frac{-1}{2}ak_x + \frac{\sqrt{3}}{2}a k_y) \).
a) Explore the nature of the Fermi boundaries for this band for a some particular examples of filling (corresponding to different values of Ef). What structure would you guess that this could be the and structure for? (educated guess --based on symmetry considerations perhaps)

Here are some possible suggestions:
b) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o -b/2\).
b) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o)
b) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o + b/2\)
b) Figure out and illustrate the Fermi boundary for the case where the fermi energy is \(E_o + 0.9 b\)
f) Life changing extra credit opportunity: To what 2D lattice structure do you think this E vs k relationship corresponds? Explain your thinking.

10. Special project opportunity: Suppose you have a solar cell like that in the previous homework, problem 7. Suppose you want to design a circuit to be able to efficiently generate heat in the resistor (which is a stand-in for energy use).  The system should be able to work in a range of brightness of the incident light.  Let's say that the number of photons absorbed per second in the junction might range from 10^16 to 2*10^17 photons/second. Can you find a single value of R that would provide pretty good energy conversion over that entire range?


18 comments:

  1. yes, D(k) = g(k). I just used a different symbol by mistake. It is in the numerator, right?

    ReplyDelete
  2. It seems that I have a few things in mind for 1(a) that may lead to a good discussion [Though that may be me pulling an all nighter for another class talking].

    In order to obtain a standing wave by combining two 1D bloch wave, should we use superposition using different k and a values and combine them? Also, would the new wave be time independent?

    ReplyDelete
    Replies
    1. I think if you were to try and combine two waves with different wave numbers it would not give you a standing wave there would still be some right or left moving term left over. Also different values of 'a' would lead to the same issue, making the functional form of the standing wave more complex, that and 'a' is more of a structural constant of the crystal so in 1D it would be nice if it stayed the same.

      Delete
  3. For question 5 when we are calculating values of \(V_f\) can we use a band width of around \(8eV\)? Or is there a different energy value we should use?

    ReplyDelete
  4. For #2 (b): Not really sure how to find k_f? I know that we can assume the delta k is equivalent to that, as we did in class on Monday, but, what would we use for the bandwidth for the electric field? Or is that something that I shouldn't really worry about?


    ReplyDelete
    Replies
    1. This Integra might help:

      \(\int_{-k_f}^{k_f}\frac {Na}{\pi}dk\)

      Delete
    2. This Integra might help:

      \(\int_{-k_f}^{k_f}\frac {Na}{\pi}dk\)

      Delete
    3. This Integra might help:

      \(\int_{-k_f}^{k_f}\frac {Na}{\pi}dk\)

      Delete
    4. Hey Freddy Hernandez Sanches de la Cruz Santos Cisneros,
      Looking at the work you did it seems like you related the occupation number to the size of the energy band. So for \(1/8\) occupation you got the "Fermi Plane" (maybe bad wording) to be \(E_f = B/8\). Very elegant approach. It should be correct too because we are using a constant density of states.

      Delete
    5. Ultimately I think it is the fraction k-space associated with occupied states compared to the amount of k space associated with all states that tells you the filling. I wouldn't necessarily count of the constant density of states thing.

      Delete
    6. Illustrate which regions are occupied. (and which are not)

      Delete
  5. What I described works for part b and c, not a, I apologize.

    ReplyDelete
  6. Are you saying that you think delta k does not depend on kf or the density of electrons?

    ReplyDelete
  7. This comment has been removed by the author.

    ReplyDelete
  8. Anyone got any ideas on 6.g? I feel like there's some sort of key relationship between the lattice structure and fermi surface, but I just can't make any sort of connection between your domain in k-space that yields occupied states and lattice structure. I can't really find anything online either, because (for those of you haven't tried a google search for fermi surface yet) all of the material is for 3 dimensions and extremely formalized.

    ReplyDelete
    Replies
    1. One approach we found helpful in office hours was to look at the arguments of the Cosine function for problem 6, 7, and 8. Although for 7 you can look at the how the energy changes in x Vs y directions. I think it was Nick and Freddy who said it would be helpful to see the argument as a vector pointing to a nearest neighbor.

      Delete