Tuesday, February 19, 2013

Homework 7: electrons in metals

This homework gives you a chance to explore electrons in metals, current flow and things like that. It will stay at a pretty high-level: we start with an expression for energy vs k, which we have discussed and tried to justify to some degree. We also take as "given", a straight-forward, easy to use relationship for velocity. Our metal is one-dimensional (1D) with a particular electron density, $10^8$ electrons/cm and a particular filling of the band (1/2 filled).

$E_k = E_o -(B/2) cos (bk)$, where B is in eV and typically around 1 to 8 eV.

$v_k = (1/\hbar) \partial{E_k}/\partial{k}$

1. Consider a metal with one well per 0.1 nm ($b = 0.1 nm$) and one (itinerant=non-local) electron per well. That leads to an electron density of 1 electron per 0.1 nm, or, equivalently, $10^8$ electrons/cm, right? With one electron per well we will get, i think, a half-filled band, that is, all the states between $k = -\pi/2b$ and $k = \pi/2b$ are filled. Assume that B = 4 eV.
a) Sketch $E_k$ vs k and indicate on your sketch where the boundary between filled and empty states is.
b) What is the speed of an electron in a boundary state, i.e., the state $\psi_k$ with $k = \pi/2b$ (also known as $k_f$). What is it in cm/sec?  (feel free to post here for a quick check and comparison with others)
c) When there are an equal number of electrons in k and -k states, the net current is zero. However, an electric field can skew the occupation boundaries so that that balance is disrupted. Suppose that the balance is skewed so that 1 in 1000 right-moving electrons are uncompensated. What is the net current carried by these electrons. [Hint: You can assume all of them have essentially the same speed and use the equation I=nev to calculate the current in electrons per second or in coulombs per second. Here n is the density of current carrying, uncompensated electrons (in electrons per cm), e is the charge of an electron, and v is the speed (from b). Do you get a current more or less in the micro-ampere range. What is it?

2. Compare the speed of the electron in problem 7 of the midterm, to the speed of an electron in the "band state" $\psi_k$, with $k=\pi/2b$ with B = 1 eV. Are they similar at all? Feel free to comment/discuss here. (Your first line could be: "how do we define the speed of the electron in problem 7?")

3. What is the nature of the time dependence of the crystal state:
$\psi_{o,k} (x,t) = \sum^\infty_{n= -\infty} e^{inkb} \psi_o (x-nb)$.
 Is this an eigenstate or a mixed state?
Please ask questions about this, comment or share your thoughts below...

4. Graph the function $f(E) = 1/(e^{(E-2)/.1} + 1)$ from about zero to 5. Describe it.

5. Compare the functions $f(E) = 1/(e^{(E-2)/.1} + 1)$ and $f(E) = e^{-(E-2)/.1}$. Where are most different? Are they similar over any region? Why?

6. A conduction band of a semi-conductor can be just like that of a metal in terms of the nature of the states $\psi_k$ and the E vs k relation, but it is completely different in terms of the occupation of the states. Typically in a semi-conductor only some states near the bottom of the conduction band are occupied at all.
a) Suppose 1 in 1000 of the states in the lowest quarter of the band (between $k = -\pi/4b$ and $\pi/4b$) are occupied and the rest of the conduction band is empty. How many electrons/cm is that in the conduction band? (units are electrons/cm)


Appendix:
This energy vs k relation (called a "dispersion relation"),
$E_k = E_o -(B/2) cos (bk)$,
refers to the energies of states of the form:

$\psi_{o,k} (x) = \sum^\infty_{n= -\infty} e^{inkb} \psi_o (x-nb))$.
These are the states made of "combinations of atomic orbitals", as we discussed and sketched last week.

Although it is not at all obvious, once you put in their time dependence (see problem 4), these states have a velocity given by the equation:
$v_k = (1/\hbar) \partial{E_k}/\partial{k}$
(which you might notice is proportional to k at low kb just like for free electron states.)

Oh, by the way, a note on terminology:
1) the "height" of the band, which is B since the band goes from a lowest energy of $E_o - B/2$ to a maximum energy of $E_o + B/2$ (at $k=\pi/b$, is generally called "bandwidth".

What looks like it should be called width, the interval from $-\pi/b$ to $\pi/b$, is called the Brillouin zone (or sometimes 1st Brillouin zone). The points $-\pi/b$ to $\pi/b$ are called the "Brillouin zone edge."

Solution notes:



54 comments:

  1. Thursday night section sounds good to me.

    ReplyDelete
  2. yeah thursday !!! aw yeah

    ReplyDelete
  3. I think for number 1 $k=\pm \pi/2b$ if we want it half filled.

    ReplyDelete
  4. Can someone post the speed of the electron from problem 7 on the midterm. I don't think this velocity was on my test (blue).

    Thanks

    ReplyDelete
    Replies
    1. It is not on anybody's test. How would you define speed for that electron?

      Delete
    2. Im confused on how to calulate speed for the electron, i know it has to do with the two energies but I don't know what to do with them?

      Delete
  5. Could we perhaps define the speed in problem 2 as being the width of the central barrier divided by the time required to tunnel through the barrier between the wells (from midterm problem 7)?

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

    ReplyDelete
    Replies
    1. I added a note on terminology in reply to an excellent comment that was sadly deleted :(

      Delete
    2. Original question: "Seems like band width should be -pi/b < k < pi/b ~ 2pi/b but what does that mean?"

      Sorry for the deletion! Deleted it when I found that K refers to "crystal momentum" from my notes.

      You mention that this interval can be called the "1st Brillouin zone". Does this mean that there are others and if so where might we encounter them?

      Delete
    3. Not in this class. complicated, technical and unnecessary imo. Most people I know just say Brillouin zone and never think about that.

      Delete
  7. If I do use (barrier width) / (tunneling time) for prob 7 electron, I get a speed which is on the same order of magnitude as that from 1b.

    Not sure if this is the right way to go about this though.

    ReplyDelete
    Replies
    1. It might be all right? How far did the electron go? How long did it take?

      Delete
    2. The time was 3.45E-15 seconds with a distance of 0.2nm ... but now that I think about it, just using the barrier width might not be sufficient since as I understand this is the time for the electron to move from being mostly in the first well to being mostly in the second ... so maybe I need to include some portion of the width of the wells in that calculation as well? Perhaps a total distance of 0.4nm (half of each well) instead of 0.2nm.

      Delete
    3. Slight correction to the above ... 0.4nm is the barrier width plus half of each well.

      Does blogspot have a plugin or perhaps an option to allow editing of comments? Maybe with a time limit ~5 minutes?

      My brief internet queries have turned up nothing.

      Delete
    4. I don't think that's possible. Your posts are great though. no worries. comments aren't expected to be perfect..

      Delete
    5. My method for determining the electron's speed isn't possible or editing comments? Or perhaps both?

      Delete
    6. editing comments.

      Your method for speed has the right units and it is based on distance over time, so it must be right...

      Delete
  8. I am wondering about 3. I don't think that this state is time dependent, but I am second guessing myself. It seems like an eigenstate as it is a summation of many $\Psi_{o}(x-nb,t)$ that all have the same energy. Wouldn't it only be time dependent if we attempted to combine states $\Psi_{o,k}(x-nb,t)$, and $\Psi_{o,j}(x-nb,t)$ with $k \neq j$?

    ReplyDelete
    Replies
    1. Technically all states are time dependent. It is the prob densities and expectation values that are sometimes not time dependent. Are you maybe suggesting that it is an energy eigenstate; that it would have the time dependence of an energy eigenstate?

      If so, what is that?

      Delete
    2. Ok, I see. I think I was mixing expectation values up with the actual wave function. Of course it has time dependence! And yes I do think that it is an energy eigenstate with the time dependence being $$\Phi(t) = e^{-i\omega_{k}t}$$ for all $\psi_{o}(x-nb)$. Is this correct?

      Delete
    3. well yes, but there is no need to distribute it in there. That is really the time dependence for one particular sum of those. one particular $\psi_{o,k} (x)$. Does that make any sense?

      Delete
    4. Yes, it does, thank you. And I just wanted to solidify that it was the same time dependence for all psi being summed, and of course can be factored out.

      Delete
  9. By the way, this is a great website for practicing Latex stuff: http://www.codecogs.com/latex/eqneditor.php

    ReplyDelete
  10. Ok, K=b/(2pi)for problem #1 kinda doesn't make sense. People keep asking whether it's correct or not, but nobody's answering. So is it right and I'm just looking at the problem wrong?

    ReplyDelete
    Replies
    1. Yes I think it ought to be $k = \frac{\pi}{2b}$ since $b$ is proportional to our $\lambda$ in this case. Harris provides a pretty good read on this in chapter 10, specifically pages 431 on.

      Delete
    2. I see. there was a mistake in the problem. i fixed that. thanks.

      Delete
  11. For number 6—if only one state in 1000 is filled in the lowest quarter of the band and there are no states filled outside of this quarter, wouldn't there only be one electron in the conduction band? I think I might be missing some aspect of the problem.

    ReplyDelete
    Replies
    1. Well the conduction band contains $2 10^8$ states per cm, so I don't see how you would get just one occupied state - only one electron in the conduction band? I think it would be much more and it would be in units of electrons/cm...

      Delete
    2. I was wondering about number 6 as well. We are dealing with $10^8$ electrons/cm and this is the same as the available number of states, or is that half the amount available as only half the band is filled? This would lead me to believe that a quarter band would be half the $10^8$ electrons/cm, and then only 1 in 1000 of those are filled. Is this correct, or am I missing a factor of 2?

      Delete
  12. Seems like the portion of the band we are looking at would be irrelevant for question 6 because we don't know what the total length of the band is?

    That being the case wouldn't we just be looking at a factor of 1000 less electrons/cm?

    ReplyDelete
  13. Concerning #3
    In my eyes, the expression for the wave function given is confusing. On the left hand side we have some wave function that has a subscript of k psi_0,k , does that mean that the sum includes only one k value?

    ReplyDelete
    Replies
    1. No, that means that every one of this k states (crystal states) comes from the ground state of the atom. It is a band label. In other words, it says that each of these states, which all have different k values, are part of the same band - the "o" band.

      Delete
  14. What is everyone getting for problem 1b? I'm using k=pi/(2b) and substituting that into (dE_k)/dk=(B/2)*sin(b*k), but I'm not seeing how to convert the units into cm/sec. Since k is in units of 1/(length), isn't (dE_k)/dk in units of eV*length?

    ReplyDelete
    Replies
    1. i multplied numerator and denominator by c so i got (hbar)c in the denominator, which has units of eV-nm or eV-angstrom. does that help?

      Delete
    2. I think maybe you didn't remember the b factor that comes from the derivative. That is:
      v= Bb/2hbar
      B: eV
      b: cm
      hbar: eV-sec

      so you get cm/sec, right?

      Delete
    3. I did make that typo involving b, but my actual issue was forgetting the 1/hbar factor from the formula for v_k.

      Thanks!

      Delete
  15. In 1c, when we are finding electron density for the uncompensated right moving electrons, do we take 1 in 1000 of 10^8 electrons, or just 5x10^8 electrons because only half of them are moving right to begin with?

    ReplyDelete
    Replies
    1. where do you get the 10^8 electrons? and i think its half

      Delete
    2. you think we use half? Now that i think more about it, an electric field that moves the boundaries wouldnt change the electron density/cm would it? it just unblanaces the momentum...

      Delete
    3. shoot i dont know now thats a good point..

      Delete
    4. Good point. I think that what you are saying is that the question, the way it is stated is a little ambiguous. I was thinking option 1 10^8/10^3 electrons/cm, but now that you mention it, it seems like it could be saying 5x10^7/10^3 electrons. Really either one would be fine.

      Since you pointed out the ambiguity, do you have a preference (for how to resolve that)?

      Delete
    5. im not sure... instead of "Suppose that the balance is skewed so that 1 in 1000 right-moving electrons are uncompensated." state it as
      "Suppose that the balance is skewed so that 1 in 1000 electrons are uncompensated to the right."

      Does that clear up the ambiguity?

      Delete
  16. wait for number one part b is it supposed to be pi/2b or b/2pi? :(

    ReplyDelete
    Replies
    1. pi/2b

      thats what he said in class today, said he would fix it on here...

      Delete
    2. I think it is fixed. Is it still incorrect in some part of the problem?

      Delete