Lecture 16: Step potential reflection and transmission coefficients. Phase shift, wavepackets and ti

L16.1 Step potential probability current (14:59)

L16.2 Reflection and transmission coefficients (08:12)

L16.3 Energy below the barrier and phase shift (18:40)

L16.4 Wavepackets (20:51)

L16.5 Wavepackets with energy below the barrier (05:54)

L16.6 Particle on the forbidden region (06:48)

L16.1 Step potential probability current (14:59)

MITOCW | watch?v=z79v39lMR3k
PROFESSOR: I’ve put on the blackboard here the things we were doing last time. We began our study of
stationary states that are not normalizable. These are scattering states. Momentum
eigenstates were not normalizable, but now we have more interesting states that represent the
solutions of the Schrodinger equation, that are stationary states with some energy e.
Because they are not normalizable, but we cannot directly interpret any of these solutions as
the behavior of a particle. I kind of tell you a story, OK. So this is-- a particle is coming,
colliding, doing something. These are not normalizable states. So part of what we’re going to
be trying to do today is connect to the picture of wave packets and see how this is used to
really calculate what would happen if you send in a particle of potential.
Nevertheless, we wrote a solution that has roughly that interpretation, at least morally
speaking. We think of a wave that is coming from the left, that’s ae to the ikx. Now, in order to
have a wave, you would have to have time dependents, and this is a stationary solution. And
there is some time dependents. There is exponential of e to the minus iet over h bar, where e
is the energy.
So this could be added here to produce the full stationary state psi of x and t. But we’ll leave it
understood-- it’s a common phase factor for both the solutions of x less than 0, and x greater
than 0, because the whole solution represents a single solution. It’s not like a solution on the
left and a solution on the right. It’s a single solution over all of x 4 a psi that has some definite
energy.
So we looked at the conditions of continuity of the wave function and continuity of the
derivative of the wave function, at x equals 0. Those were two conditions. And they gave you
this expression for the ratios of c over a and b over a. We could even imagine, since we can’t
normalize this, setting a equal to 1, And. Then calculating b and c from those numbers.
We have two case a k in a k bar. The k is relevant to the wave function for x less than 0. The k
bar is relevant for x greater than 0. And they have to be different because this represents a
DeBroglie wavelength, and the DeBroglie wavelength encodes the momentum of the particle,
and the momentum of the particle that we imagine here classically is different here where this
has this much kinetic energy, and in the region on the right, where the particle only has a
much smaller kinetic energy. So that’s represented by k bar.
And k bar, being the energy proportional to the energy minus v 0, while k squared has just the
energy, it’s smaller than k. So this is what we did. We essentially solved the problem, and this
qualifies as a solution, but we still haven’t learned anything very interesting from it. We have to
understand more what’s going on.
And one thing we can do is think of a particular limit. The limit, or the case, when e is equal to
v 0, exactly equal to b 0, what happens? Well, k bar would be equal to 0. And if k bar is equal
to 0, you’re going to have just the constant c in here.
But if k bar is equal to 0, first b is equal to a, because then it’s k over k, so b is equal to a. And
c is equal to 2a. And the solution would become psi of x equals-- well, a is equal to b. So this is
twice a cosine of kx, when a equal to b is a common factor, call it a. And this thing is the sum
of two exponentials with opposite signs. That gives you the cosine.
And for c, you have 2a. And since k bar is equal to 0, well, it’s just 2a. It’s a number. This is
unnormalizable, but even the original solution is unnormalizable, so we wouldn’t worry too
much about it. So how does that look for x over here? You have a cosine of kx, so it’s going to
be doing this to the left. That’s for x less than 0.
And at this point, it goes like that. Just flat side effects, and here’s 2a. so there’s nothing wrong
with the solution in this case. It’s kind of a little strange that it becomes a constant, but
perfectly OK.
What really helps you here is to find some conditions that express the conservation of
probability. You see, you have a stationary state solution. Now, stationary states are funny
states. They’re not static states, completely static states. For example, if you have a loop, and
you have a current that never changes in time. This is a stationary condition, even in
electromagnetism.
So what we imagine here is that we’re going to have some current, probability current, that is
coming from the left, and some of it maybe bounces back, and some of it goes forward. But
essentially, if you think of the barrier, whatever-- you look a little to the left of the barrier and a
little to the right of the barrier.
Whatever is coming in, say, must be going out there, because probability cannot increase in
this region. It would be like saying that the particle suddenly requires larger and larger
probability to be in this portion of the graph. And that can’t happen.
So probabilistic current gives you a way to quantify some of the things that are happening. So
probability current plus j effects, which was h bar over m, imaginary part of psi star d psi dx.
So let’s compute the probability current. Let’s compute for x less than 0. What is the probability
current j of x? We then call it the probability current on the left side. I would have to substitute
the value of the wave function for x less than 0, which is the top line there, into this formula.
And see what is the current.
In fact, I believe you’ve done that in an exercise some time ago. And as you can imagine, the
current is proportional to the modulus of a squared, the length of a squared, the length of b
squared also enters. And the funny thing is, between those two waves, one that is going to the
right and one is going to the left, that that’s very visible in the current. This is h bar k over m.
A squared minus b squared. It’s a short computation, and it might be an OK and a good idea
to do it again. It was done in the homework. And x greater than 0. J right of x would be equal
to h bar k bar. Now that, you can almost do it in your head. This c into the ik bar x. Look what’s
happening.
From psi star you get a c star. From psi you get the c, so that’s going to be a c squared. The
face is going in a cancel between the one here and the one on psi, but the derivative will bring
down an ik bar, and the imaginary part of that is just k bar. So the answer is this.
And these are the two currents. Now, if we are doing things correctly, the two currents should
be the same. Whatever current exists to the left, say a positive current that is coming in, to the
right must be the same. Another pleasant thing is that this current doesn’t depend on the value
of x, as x is less than 0. Nor of the value of x when x is greater than 0. And that’s good for
conservation. It would be pretty bad as well, if you look at two places for x less than 0, and you
find that the current is not the same. So where is it accumulating? What’s going on?
So the independence of these things-- from x, this is constant, and a constant is very
important, because this constant should be the same. Now, whether or not about current
conservation, it’s encoded in Schrodinger’s equation. And we solved Schrodinger’s equation.
That’s how we got these relations between b, a and c So it better be that these two things are
the same.
So Jay elsewhere-- jl, for example. I won’t put off x, because it’s just clear. It doesn’t depend
on x. 1 minus b over a squared a squared. I’m starting to manipulate the left one, and see if
indeed the currents are the same. And now, I get h bar k over m. 1 minus b over a squared. I
put the modulus squared, but so far everything is real in this-- k, k bar, b, a, c-- all real. Is that
right? No complex numbers there. So I don’t have to be that careful.
B over a squared is k minus k bar over k plus k bar squared a squared. And I gfit that here,
maybe. H bar k over m. Now what? This is squared, so the si, squared passes here to the
numerator minus the difference squared, that’s going to be a 4k k bar over k plus k bar
squared a squared.
And yes, this seems to be working quite well. Flip these k’s-- these two k’s, flip them around.
So the answer for jl so far is h bar, now k bar, because I flipped that, m, and now I would have
4 k squared over this thing, which is 2k over k plus k bar squared of a squared. And this
quantity, if I remember right, is just c squared, as you can see from there.
So indeed, this is j right, and it worked out. This is j right. So j left is equal to j right by current
conservation. So that’s nice. That’s another way of getting insight into these coefficients. And
that kind of thing that makes you feel that there’s no chance you got this wrong. It all works
well.

L16.2 Reflection and transmission coefficients (08:12)

MITOCW | watch?v=bX-k26w-tsU
PROFESSOR: We can write, however, J left as J of A, minus J of B. Where J of A would be h bar k over m A
squared. And J of B would be h bar k over m B squared.
You see the current that exists to the left of the barrier has two components and-- it’s very
intuitive. It’s the current that would have been brought alone by the incoming A-wave. Minus
the current that would have existed alone from the reflected B-wave. B is the reflected wave.
So that’s very nice. There’s no interference between these two terms. We can really think of a
current that is associated to the incoming wave, and a current that is associated with the
reflected wave.
So this suggests how you should define a reflection coefficient. Reflection coefficient R would
give me the amount of current I get reflected, compared to the amount of current that there is
incident. You see, the incident current is going to be partially reflected and partially
transmitted.
So an idea of a reflection is the value of the reflected current divided by the incident current.
It’s a definition, but it’s a reasonable definition.
And then, if it is this ratio-- because of these expressions-- it happens to be B over A squared.
And that’s an interesting number. Now, there’s some physics in it. It tells me how much of the
probability gets reflected as a function of the probability that is incident.
So that’s a good measure. If you get a reflection coefficient of 1/10, then you would expect
1/10 of the particles to be reflected. Now we don’t have particles yet. This is non-normalizable
solution. But, still, this will be the intuition very soon.
Now we could have a transmission coefficient, as well. And here is something where we
sometimes make a mistake. T is going to be the transmission coefficient. Transmission
coefficient.
And how should we define it? There is a temptation to define it-- well, coefficient B over A
gives me this. Then maybe it should be C over A. But actually-- while c over a gives you some
idea of how big the wave to the right is compared to the wave of the left-- that’s not what we
should call a reflection coefficient.
And the reason is that-- I will call this current J C. And that’s the amount of probability–
because it’s a current associated to the wave C. And that’s the amount of probability that is
being carried by the transmitted wave. That is the probability. Not necessarily C over A.
So the transmission coefficient will be defined to be J C divided by J A. And then J C divided by
J A-- J C has an h-bar, k-bar. And J A has a k. So this is not equal to this ratio, but is actually
k-bar over k, C over A.
So it’s not just this number. The reflection coefficient-- and transmission coefficients-- really
originate from probabilities. And the probabilities for this current. And, therefore, there is no-- it
would have been very hand-wavy, and actually wrong to think it’s C over A.
These definitions-- because after all, this is a definition-- makes some nice sense. Because
you have J L-- we said is equal to J right, but J L is J A minus J B, is equal to J C. And
therefore R plus T. The reflection coefficient plus the transmission coefficient-- which is J B
over J A plus J C over J A is equal to J B plus J C over J A. And you see here that J A plus J C
is indeed equal to J B.
I’m sorry. I got this wrong. Yes. Where is the eraser. I should have passed the B to the other
side.
This force implies J A equals J B plus J C. And this ratio is equal to 1, which is something you
usually want when you define reflection and transmission coefficients. They should add up to
1.
So now we’ve got an idea. Yes, with this solution I can understand the reflection and
transmission coefficients. But do these apply to particles? Well, the good news is that it roughly
applies to particles-- as we will see with the wave packets soon.
If you send the wave packet, it’s going to have some uncertainty and momentum. It’s going to
have some uncertainty and energy. But for some energy-- suppose the wave packet doesn’t
have that much uncertainty-- basically, the probability that the wave packet bounces is the
reflection at that energy, that is the main energy that the wave packet sends.
If your wave packet is very broad over energies, then it’s a more complicated thing. But as
long as the wave packet is such that it basically has one narrow band of energy, the reflection
coefficient associated to this calculation is the reflection coefficient or reflection probability for
the wave packet that you’re sending in.

L16.3 Energy below the barrier and phase shift (18:40)

MITOCW | watch?v=EkpbxgEslE4
PROFESSOR: Let’s do E less than V not. So we’re back here. And now of the energy e here is v not is x
equal 0. X-axis. And that’s the situation. Now you could solve this again. And do your
calculations once more. But we can do this in an easier way by trusting the principle of analytic
continuation.
In this case, it’s very clear and very unambiguous. So the big words, analytic continuation,
don’t carry all the mathematical depth. But it’s a nice, simple thing. We first say that the
solution is the same for x less than 0. So for x less than 0, we write the same solution.
Because the energy is greater than 0, or all what we said here, the value of k squared, a into
the ikhd e to the minus i k x. It’s all good. And k squared is still 2 m e over h squared. The
problem is the region where x is greater than 0. Because there you have an exponential. But
now you must have a decaying exponential.
But we know how that should work. It should really be an e to the minus some kappa x. So
how could I achieve that? If I let k bar replace-- everywhere you see k bar, replace it by i
kappa. Park then one thing that happens is that-- you learn from here, from this k bar squared,
would be minus kappa squared equal to that.
So kappa squared would be 2 m, v not minus e over h squared. The sign is just the opposite
from this equation. That’s what that equation becomes upon that substitution. Now that
substitution would not make any difference, how they put a plus i or minus i, I wouldn’t have
gotten my sine change, and this.
But if I look at the solution there, the solution psi becomes-- on the region x greater than 0,
turns into c, e to the i. kappa bar is a i kappa x. Therefore, it’s equal to c, e to the minus kappa
x, which is the right thing. And that sine that I chose, of letting k equal i kappa proves
necessary to get the right thing.
So it’s clear that to get the right thing, you have that. And now you know that, of course, if you
would have written the equation from the beginning, you would have said, yes, in this region,
there is a decaying thing. And looking at the Schrodinger equation, you have concluded that
kappa is given by [? that. ?]
But the place where you now save the time is that, since I just must do this change in the
equations, I can do that change in the solutions as well. And I don’t have to write the continuity
equations again, nor solve them. I can take the solutions and let everywhere that was a kappa
bar replaced by i, that was a k bar, replace it by i kappa bar.
So what do we get? It should go here and believe those circuits. OK, so b over a, that used to
be. Top blackboard there. Middle, k minus k bar becomes k plus i-- no, minus I k bar, minus i
kappa. And k plus i kappa. so it has changed.
Suddenly this ratio has become complex. It’s kind of interesting. Well, let’s make it clearer by
factoring a minus i here. So this becomes kappa. And you need plus k, so this must be plus i k
over i. This would be kappa minus i k. So this is just minus kappa plus i k over kappa minus i k.
But when you see that ratio, you’re seeing the ratio of two complex numbers of equal length.
And therefore, that ratio is just a phase. It’s not any magnitude. So this is just a phase, and it
deserves a new name. There is a phase shift between the b coefficient and the a coefficient.
And we’ll write it as e minus-- the minus I’ll keep. e to the 2 i delta.
That depends on the energy. I’ll put delta of the energy, because after all, kappa, k, everybody
depends on the energy. So let’s call it 2 i delta of e. And what is delta of e? Well, think of the
number kappa plus i k. This is the k, i k here.
The angle-- this complex number has an angle that, in fact, is delta. e to i delta is that phase.
And delta is the arc tangent, k over kappa. So I’ll write it like that. Delta. Now you get a delta
from the numerator, a minus delta, as you can imagine, from the denominator. And that’s why
you get a total 2 i delta here.
So delta of e is 10 minus 1, k over kappa. And if you look at what k and kappa were, k over
kappa is like the ratio of the square root of the energy over v not minus the energy. So delta of
e is equal to 10 minus 1 square root of energy over e v not minus and energy.
Now I got the question about current conservation. What happens to current conservation this
time? Well, you have all these waves here. But on the region x greater than 0, the solution is
real. If the solution is real, there is no probability current on the right. There’s really no
probability that you get this thing, and you get current flowing there. And you get this pulse, or
whatever you send to keep moving and moving and moving to the right.
Indeed, the solution decays. And it looks like the one of the bound state in this region. So
eventually there’s no current here, because there’s no current here. No current there. No
current there. Because the solution drops down. But it’s a real solution anyway. So there’s no
current there.
So Ja-- Jc is equal to 0. 0. Solution is real for x greater than 0, and any way goes to 0 at
infinity. So the fact that it’s real is a mathematical nicety that help us realize that it must be 0.
But the fact that there’s no current far away essentially telling you better be 0.
So if the current Jc is 0, Ja must be equal to Jb. And therefore that means a squared is equal
to b squared. And happily that’s what happened because b over a is a complex number of
magnitude one. So the fact that b and a differ by just the phase was required by current
conservation.
A over b is a number that has norm equal to 1. So that’s a consistent picture. This phase is
very important. So what happens for resolution for x less than 0? Well, psi of x would be a into
the i k x, plus b all the way to the left there.
Your solution is a plus b equal minus i k x. Of course, we now know what the b is, so this is
minus a from the ratio over there. e to the 2 i delta of e. E to the minus i k x. For x less than 0.
And for x greater than 0, psi of x is going to c e to the minus kappa x. And I’m not bothering to
write the coefficient c in Terms of a.
Now this expression for x less than 0 can be simplified a little. You can factor an a. But it’s very
nice, and you should have an eye for those kind of simplifications. It’s very nice to factor more
than an a and to factor one phase like an i delta. I delta of e, because in that way, you get e to
the i k x minus delta of e from the first term, where the two delta appearances cancel each
other.
Because the first term didn’t have a delta. But then the second term will have the same
argument here, of the exponential but with a minus sign. Minus i k x. And I claim also a minus
delta of e. And this time minus and minus gives you plus. And the other e to the i delta gives
you back the 2.
But now you’ve created the trigonometric function, which is simpler to work with. So psi effects
is equal to 2 i a e to i delta of e, sine of k x minus delta of e. And if you wish psi squared, the
probability density is proportional to 4 a squared, sine squared of k x minus delta of e.
So this can be plotted. Sine squared is like that. And where is x equal 0. OK, I’ll say x equal 0,
say is here. So this is not really true anymore. But this point here is x 0. It vanishes. Would be
the point at which k x 0 is equal to delta of e. And the sine squared vanishes. So this is not a
solution either.
Solution is like that. And then this is for psi squared. And then it must couple to the k
exponential on this side. So that the true solution must somehow be like this and well,
whatever. I don’t know how it looks. That is the e to the minus 2 kappa x decay and
exponential.
It must decay. There’s continuity of the derivative and continuity of the wave function. So that’s
how this should look. A couple more things we can say about this solution that will play a role
later. I want to get just a little intuition about this phase, delta of e, this phase shift.
So we have it there. Delta of e, I’ll write it here, so you won’t have to-- 10 minus 1, square root
of e over v not minus e. So this is interesting. This phase shift just applies for energies up to v
not. And that corresponds to the fact that we’ve been solving for energies under the barrier.
And if we solve for energies under the barrier, well, the solutions as we’re writing with these
complex numbers, apply up to energies equal to the barrier, but no more.
So we shouldn’t plot beyond this place. And here is delta of e. The phase shift. And when the
energy is 0, when your particle you’re sending in, or the packet eventually is very low energy
here. Then the phase shift is the arctangent of 0, which is 0.
As the energy goes to the value v not, then the denominator goes to 0. The ratio goes to
infinity. And the arctangent is pi over 2. So it’s a curve that goes from here to here. And it’s not
quite like a straight line. But because of the square roots, it sort of begins kind of vertical, then
goes like this. It’s not flat either, in the middle. So maybe my curve doesn’t look too good.
Those more vertical here. Wow, I’m having a hard time with this. Something like this.
In fact, it’s kind of interesting to plug the derivative, d delta, d energy. A little calculation will
give you this expression. You can do this with mathematica or v not minus e. And shows, in
fact, that here is v not, and here is v delta, v e.
We could call it delta prime of e, because we wrote the phase shift as a function of the energy.
So that the delta, d e is really delta prime of e. And it sort of infinite-- goes to a minimum and
infinite again, in that direction. That’s how it behaves.

L16.4 Wavepackets (20:51)

MITOCW | watch?v=NXPvXI603RA
PROFESSOR: In order to learn more about this subject, we must do the wave packets. So this is the place
where you really connect this need solution of Schrodinger’s equation, the energy eigenstates,
to a physical problem. So we’ll do our wave packets.
So we’ve been dealing with packets for a while, so I think it’s not going to be that difficult.
We’ve also been talking about stationary phase and you’ve practiced that, so you have the
math ready. We should not have a great difficulty.
So let’s new wave packets with-- I’m going to use A equals 1 in the solution. Now, I’ve erased
every solution, and we’ll work with E greater than v naught to begin with. The reason I want to
work with E greater than v naught is because there is a transmitted wave. So that’s kind of
nice.
So what am I going to do? I’m going to write it this following way. So here is a solution with A
equals to 1. e to the i kx plus k minus k bar over k plus k bar e to the minus i kx. And this was
for x less than 0.
And indeed, when there was an A, there was a B, if A is equal to 1, remember we solve for the
ratio of B to A and it was this number so I put it there. I’m just writing it a little differently, but
this is a solution. And for x greater than 0, the solution is C, which was 2k over k plus k bar e to
the i k bar x.
Now we have to superimpose things. But I will do it very slowly. First, is this a solution of the
full Schrodinger equation? No. Is this a solution a time-independent Schrodinger equation?
What do I need to make it the solution of the full Schrodinger equation? I need e to the minus i
Et over h. And I need it here as well.
And this is a psi now of x and t. There’s two options for x, less than 0 and x less than 0, and
those are solutions. So far so good.
Now I’m going to multiply each solution. So this is a solution of the full Schrodinger equation,
not just the time-independent one, all of it. It has two expressions because there’s a little
discontinuity in the middle, but as a whole, it is a solution. That is the solution.
Some mathematicians would put a theta function here, theta of minus x and add this with a
theta of x so that this one exists for less than x, x less than 0, and this would exist for less
greater than 0. I would not do that. I will write cases, but the philosophy is the same.
So let’s multiply by a number, f of k. Still a solution. k is fixed, so this is just the number. Now
superposition. That will still be a solution if I do the same superposition in the two formulas,
integral dk and integral dk. That’s still a solution of a Schrodinger equation.
Now I want to ask you what limits I should use for that integral. And if anybody has an opinion
on that, it might naively be minus infinity to infinity, and that might be good, but maybe it’s not
so good. It’s not so good. Why is not so good, minus infinity to infinity?
Would I have to do from minus infinity on my force. Does anybody force me? No. You’re
superimposing solutions. For different values of k, you’re superimposing. You had a solution,
and another solution for another, another solution. What goes wrong if I go from minus infinity
to infinity? Yes.
AUDIENCE: [INAUDIBLE] the wave packet’s going to be in one direction, so it [? shouldn’t ?] be [? applied.
?]
PROFESSOR: That’s right. You see here, this wave packet is going to be going in the positive x direction,
positive direction, as long as k is positive. It’s just that the direction is determined by the
relative sign within those quantities. E is positive in this case. k is positive. This moves to the
right.
If I start putting things where k is negative, I’m going to start producing things and move to the
left and to the right in a terrible confusion. So yes, it should go 0 to infinity, 0 to infinity.
And f of k, what is it? Well, in our usual picture, k f of k is some function that is peaked around
some k naught. And this whole thing is psi of x and t, the full solution of a wave packet.
So now you see how the A, B, and C coefficients enter into the construction of a wave packet. I
look back at the textbook in which I learned quantum mechanics, and it’s a book by Schiff. It’s
a very good book. It’s an old book. I think was probably written in the '60s. And it goes through
some discussion of wave packets and then presents a jewel, says with a supercomputer,
we’ve been able to evaluate numerically these things, something you can do now with three
seconds in your laptop, and it was the only way to do this.
So you produce an f of k. You fix their energy and send in a wave packet and see what
happens. You can do numerical experiments with wave packets and see how the packet gets
distorted at the obstacle and how it eventually bounces back or reflects, so it’s very nice. So
there is our solution.
Now we’re going to say a few things about it. I want to split it a little bit. So lets go here. So
how do we split it? I say the solution is this whole thing, so let’s call the incident wave that is
going to be defined for x less than 0 and t, this is x less than 0. And the incident wave packet is
dk 0 to infinity f of k e to i kx e to the minus i et over h bar.
And this is just defined for x less than 0, and that’s so important that write it here. For x less
than 0, you have an incident wave packet. And then you also have a reflected wave packet, x
less than 0 t is the second part dk f of k k minus k bar over k plus k bar e to the minus i kx e to
the minus i et over h bar.
It’s also for x less than 0, and we have a psi transmitted for x greater than 0 and t, and that
would be 0 to infinity dk f of k 2k over k plus k bar e to the i k bar x e to the minus i et over h
bar. Lots of writing, but that’s important.
And notice given our definitions, the total psi of x and t is equal to psi incident plus psi reflected
for x less than 0, and the total wave function of x and t is equal to psi transmitted for x greater
than 0. Lots of equations. I’ll give you a second to copy them if you are copying them.
So now comes if we really want to understand this, we have to push it a little further. And
perhaps in exercises we will do some numerics to play with this thing as well. So I want to do
stationary phase approximation here. Otherwise, we don’t see what these packets, how
they’re moving.
So you have some practice already with this. You’re supposed to have a phase whose
derivative is 0, and it’s very, very slowly at that place where there could be a contribution. Now
every integral has the f of k. So that still dominates everything, of course.
You see, if f of k is very narrow, you pretty much could evaluate these functions at the value of
k naught and get a rather accurate interpretation of the answer. The main difficulty would be to
do the leftover part of the integral. But again, here we can identify phases. We’re going to take
f of k to be localized and to be real. So there is no phase associated with it, and there is no
phases associated with these quantities either, so the phases are up there.
So let’s take, for example, psi incident. What is this stationary phase condition? Would be that
the derivative with respect to k that we are integrating of the phase, kx minus Et over h bar
must be evaluated at k naught and must be equal to 0. So that’s our stationary phase
approximation for the top interval.
Now remember that E is equal to h squared k squared over 2m. So what does this give you?
That the peak of the pulse of the wave packet is localized at the place where the following
condition holds. x minus de dk, with an h bar will give an h k t over m evaluated at k naught
equals 0. So this will be x equals h bar k naught over m t. That’s where the incident wave is
propagating.
Now, look at that incident wave. What does it do for negative time? As time is infinite and
negative, x is negative, and it’s far away. Yes, the packet is very much to the left of the barrier
at time equals minus infinity. And that’s consistent because psi incident is only defined for x
less than 0. It’s only defined there.
So as long as t is negative, yes, the center of the packet is moving in. I’ll maybe draw it here.
The center of the packet is moving in from minus infinity into the wall, and that is the picture.
The packet is here, and it’s moving like that, and that’s t negative. The psi incident is coming
from the left into the barrier, and that’s OK.
But then what happens with psi incident as t is positive? As t is positive, psi incident, well, it’s
just another integral. You might do it and see what you get, but we can see what we will get,
roughly. When t is positive, the answer would be you get something if you have positive x. But
psi incident is only for negative x.
So for negative x, you cannot satisfy the stationarity condition, and therefore, for negative x
and positive time, t positive, psi incident is nothing. It’s a little wiggle. There’s probably
something, a little bit-- look at it with Mathematica-- there will be something. But for positive t,
since you only look at negative x, you don’t satisfy stationarity, so you’re not going to get
much.
So that’s interesting. Somehow automatically psi incident just exists for negative time. For time
near 0 is very interesting because somehow stationary [INAUDIBLE], when you assess, you
still get something, but you’re going to see what the packet does as it hits the thing.
Let’s do the second one of psi reflected. d dk this time would be kx with a different sign, minus
kx minus E t over h bar at k naught equals 0. For the reflected wave, the phase is really the
same. Yeah, this factor is a little more complicated, but it doesn’t have any phase in it. It’s real,
so [INAUDIBLE].
So I just change a sign, so this time I’m going to get the change of sign. x is equal to minus h
bar k naught over m t. And this says that for t positive, you get things. And in fact, as t is
positive your are at x negative. And remember psi reflected is only defined for x negative, so
you can satisfy stationary, and you’re going to get something.
So for t positive, you’re going to get as t increases, a thing that goes more and more to the left
as you would expect. So you will get psi reflected going to the left.
I will leave for you to do the psi transmitted. It’s a little different because you have now k bar,
and you have to take the derivative of k bar with respect to k. It’s going to be a little more
interesting example.
But the answer is that this one moves as x equals h bar k bar over m t. k bar is really the
momentum on the right, and since psi transmitted exists only for positive x, this relation can be
satisfied for positive t. For positive t, there will be a psi transmitted.
The psi transmitted certainly exists for negative t, but for negative t, stationarity would want x
to be negative, but that’s not defined. So for negative t on the right, yes, psi transmitted maybe
it’s a little bit of something especially for times that are not too negative.
But the picture is that stationary phase tells you that these packets, psi incident, pretty much
exist just for negative t and psi reflected and psi transmitted exist for positive t. And these are
consequences of the fact that psi incident and psi reflected exist for negative x. The other
exists for positive x, and that coupled with stationarity produces the physical picture that you
expect intuitively, that the incident wave is just something, part of the solution that exists just at
the beginning. And somehow it whistles away. Some of it becomes transmitted, some of it
becomes reflected.

L16.5 Wavepackets with energy below the barrier (05:54)

MITOCW | watch?v=yqrMAZkQOwI
PROFESSOR: E less than v0. So you have an incident wave, e to the ikx, incident, and a reflected wave that
you have, e to the minus ikx-- remember minus the other face-- and e to the 2i delta of E. So
this is the incident wave, and this is the reflected wave.
They correspond to your Ae to the ikx plus Be to the minus ikx. Remember, when the energy
was less than v0, the range of B over A was minus e to the 2i delta. And since I take A equals
1, you get this thing.
So suppose I construct a psi incident of x less than 0 and t as the sum 0 to infinity dk f of k e to
the ikx e to the minus iEt over h bar. Whew. So I superimpose the incident thing here.
Then the reflected one should be superimposed too and would be 0 to infinity, a minus in front
because there’s a minus, dk f of k e to the minus ikx e to the 2i delta of E e to the minus iEt
over h bar. Whew. That’s the reflected wave superimposed.
So now you’ve constructed everything. Here the reflected wave is more interesting than the
transmitted wave, because there’s no real big transmitted wave. It just whistles out.
But the reflective thing is interesting. If you’re doing the experiment, you send in a particle, you
want to see what you get back. That’s going to tell you what kind of potential you can expect it
encountered.
So let’s do the stationary phase for this one, for the reflected. Let’s see how it moves. We
know how the incident moves. The incident moves with x equals-- we’ve done it there-- hk0
over m t.
But how about this one? Well, this one you would have to do d dk of minus kx plus 2 delta of E
minus Et over h bar, all that at k0 equals 0. And you’ll probably remember that this thing was in
your midterm and your first test. You had this wave, and you had to analyze, what did
stationary phase do? And it does that.
So what do you get? Well, when you take the derivative, you have to take the derivative of
delta with respect to energy, that’s delta prime, and then derivative of energy with respect to t.
Let me save you a little time. The answer is minus h bar k0 m t minus 2h bar delta prime of E.
OK. That’s what you get. That’s how this packet moves.
And what does it do really? Remember, this is defined for x less than 0. So this is valid-- forget
about this little term here-- this is valid for t positive. For t positive, you’re going to get this to
satisfy. So this is a big wave packet for t positive. It’s the reflected wave. That’s what you
would expect. This is the reflected.
Now, if this factor was not here, it is as if, well, the incoming packet hit the origin at t equals 0.
And this will be perfect bouncing in which the packet gets reflected. And at t equals 0, it starts
to move to the left. And as t increases, it’s moved more and more to the left. You see it there,
because x must be negative.
But if there is this term, it really doesn’t start to move to the left until t is bigger than that so that
x is negative. So only at t equal to this amount of time the packet reflects. So there’s a delay,
and the delay is 2h bar delta prime of E.
So this is a technology people use in scattering theory to figure out what kind of potential you
have, figure out how much things get delayed from the bouncing. Now, this derivative-- we’ve
plotted it there-- delta prime of E. You get a big delay for low energy, for energies near v0, and
in the middle it’s not so big.

L16.6 Particle on the forbidden region (06:48)

MITOCW | watch?v=lA8-N_ARHTw
PROFESSOR: We have found in this solution with some energy like this, that there’s a decaying exponential
over this side. And the question is often asked, well what happens if you tried to measure the
particle in the forbidden region? Must be a problem. If you find the particle in the forbidden
region, it has energy E that is less than v0, so you you have found the particle with negative
kinetic energy. How does it look? How can it happen? What’s going on? Can you really find the
particle in the forbidden region? And then how does this negative kinetic energy look like?
The answer is that it’s kind of funny what happens here. You can make two statements. It
would be contradictory, contradictory if you could make-- could say the following things. One,
that the particle is in the forbidden region, forbidden region. And two, that the particle has
energy less than v0. Because then it would mean negative kinetic energy. So if you can say
these two things, it seems contradictory.
So quantum mechanics evades this problem. Now, this is not discussed as far as I can see,
except in some lecture notes of Gordon [? Boehme. ?] And because the argument is not 100%
precise, but they think the spirit of the argument is clear. So I want to share it with you.
So here is the catch. This particle, remember it’s governed by e to the minus kappa x is the
forbidden region. So the length scale here where you can find it, the particle. The length scale
is, this forbidden region stretches to about x of the order 1 over kappa. If you are going to find
it, it is in the region of a distance 1 over kappa. At 10 1 over kappa you’re not going to find it.
The exponential is too small.
But remember, what was kappa? Kappa squared was 2m v0 minus E over h squared. That’s
what it was. Now if you want to see and declare that you have this particle, you would have to
be able to measure position with some precision, with a precision a little smaller than this.
Otherwise if you measure with precision 10 times that, well maybe it’s to the left, maybe it’s
somewhere else. So you need to measure position with delta x a little smaller than 1 over
kappa, otherwise you cannot really tell it’s inside the forbidden region.
But now the problem is that if you do a position measurement, and you localize the wave
function, there is some momentum uncertainty. The particle that you’re looking at, as opposed
to the particle to the left, has no momentum. It’s a different kind of wave function. There’s no
momentum really associated or well-defined momentum to it.
So because you make a position, you’re localizing x, whatever wave function you have. You’re
going to have some uncertainty, and some momentum that is going to be kind of bigger than h
bar over delta x. So a momentum that is bigger than, or a little bigger, than h bar kappa. If
delta x is less than that inequality, it goes in the same direction. So there’s going to be an
uncertainty P.
And therefore, this particle has now some kinetic energy due to this uncertain momentum. So
uncertainty in the kinetic energy is how much? It’s P squared over 2m, where P is this
uncertain momentum. So this is equal to h bar kappa squared over 2m, which is equal to v0
minus E.
So actually, if you think about it, here is v0. This difference is v0 minus E. And you were going
to say, oh, I found the particle, it has negative kinetic energy. But no. The uncertainty principle
says, you found it localized? OK. Your kinetic energy, I’m sorry, no. There’s an uncertainty.
How much? v0 minus E.
So whatever you wanted to prove, it has been disproved. You can’t do it. The total energy,
total energy is now E plus the uncertainty in the energy, which is E plus v0 minus E. And it’s
therefore greater than or equal to v0. And no real contradiction.
So the uncertainty principle sort of conspires to prevent you from finding a particle with
negative kinetic energy. And if you do detect a particle in the forbidden region, it will have total
energy 0, or total kinetic energy 0. It will be a normal particle. Nothing strange about it.