115103102

INDEX
S. No Topic Page No
Week 1
1 Propagators I 1
2 Propagators II 11
3 Second quantization I 23
4 Second quantization II 34
5 Second quantized Hamiltonian 43
6 Tight Binding Hamiltonian, Hubbard model 55
7 Magnetism 80
8 Singlet and Triplet State: Magnetic Hamiltonian 99
9 Antiferromagnetism in Hubbard model 111
Week 2
10 Green's function and representations in quantum mechanics 128
11 S matrix and free electron Green's function 142
12 Wick's theorem and normal ordering 158
13 Green's function and Feynman diagrams 173
14 Feynman diagram 187
Week 3
15 phonon Green' function and Hartree Fock approaximation 203
16 Finite temperature Green's function and Matsubara frequencies 221
17 Dyson's equation and disorder in electronic systems 241
18 Introduction to electrodynamics, Meissner effect 260
19 London penetration depth, Type I and II superconductors 293
Week 4
20 Cooper's problem, BCS gap equation 309
21 BCS theory, Transition temperature 328
22 Ginzburg Landau Theory, Coherence length and penetration depth 346
23 Quantum Hall Effect 359
24 Spin Hall effect, 2D topological insulator 381
25 Bose-Einstein condensation 398
Advanced Condensed Matter Physics
Prof. Saurabh Basu
Department of Physics
Indian Institute of Technology, Guwahati
Lecture – 01
Propagators I
Good morning everyone. In this course of Advanced Condensed Matter Physics, we are
mainly going to learn quantum many body theory as its applied to various solid state
systems or crystal lattices and other systems, continuum systems, but this is the
interacting version and it actually builds upon the first course of solid state physics and
preliminary knowledge of quantum mechanics would be needed. This is the course is has
a lot of mathematics embedded in it and the mathematics is to develop the formalism to
deal with interacting systems and you may not have encountered interacting systems too
much.
So, these course will give you a preliminary idea of how to deal with interacting systems
and because interactions are very important, inter particle interactions are very important
in condensed matter physics and so, we have to learn this technique and that is where all
the mathematics comes and once when we learn it we will apply it to various things that
we know such as magnetism, such as superconductivity and if there are time then we
would go ahead and discuss the kane Mele model and topological insulators the 2016
Nobel prize was awarded on the topology in condensed matter physics we briefly touch
upon that.
(Refer Slide Time: 02:18)
1
So, before we start talking about the contents of or rather the details of the course, let us
talk about the topics that would be covered in the next few weeks that will be doing the
course together. So, we will start with a brief recap of quantum mechanics and this is
essential because we have to introduce concepts such as Hilbert spaces single particle
Hilbert space and we you know many particle Hilbert space and so on and then we will
talk about second quantization, this is in contrast to the first quantization which we see in
quantum mechanics and this second quantization actually has the foundation of doing a
quantum many body theory that we are going to show.
Then there will be applications of second quantization and how to write down a
Hamiltonian particular, Hamiltonian for in the second quantized form and we apply to
magnetism and a celebrated model called as Hubbard model which has been used in the
context of interacting condensed matter system to a very large extent. We will then talk
about greens functions and these greens functions will be acting as if they are like wave
functions work in single particle quantum mechanics and these greens functions have to
be learned at 0K temperature.
Where we would talk about wicks theorem and Feynman diagram and these greens
function will be also learned in at finite temperature because as you know that all the
condensed matter physics is done at finite temperature. So, and in that connection will
learn Matsubara frequencies and finally, we will apply it to the theory of
superconductivity and learn what are called as Meissner effect and MCS theory.
So, BCS theory is a celebrated theory in superconductivity, which explains all the
phenomena related to the weak coupling superconductors or the so called conventional
superconductors and maybe will briefly discuss the high temperature superconductors,
which are the unconventional superconductors in some sense and that is where the course
will stop.
2
So, technically speaking condensed matter physics deals with either one body problem or
many body problems. The reason behind saying that is that a two body problem can
always be reduced to a one body problem and three body problems are unsolvable, but
physicists and chemists both they deal routinely with Avogadro number of particles,
which means that 10 to the power 23 number of particles and when you talk about such
large number of particles then you need to the, you must be talking about particles which
are close to each other, such that they are coming within the de Broglie wavelength of
each other.
In that sense in order to describe such a system with so many particles, one really needs a
quantum many body theory. Sometimes in this context it is advantageous to talk about
instead of talking about a large number of non interacting particles, one can also talk
about a few or relatively, few interacting particles and these particles are called as quasi
particles and this description holds; however, that description would break down when
these quasi particles, the time over which they are created if that time is shorter than the
or rather longer than the time over which they decay, in which case the quasi particle
descriptions will not hold.
Now we shall begin our discussion with quantum mechanics and in particular we will
talk about Hilbert spaces. So, the question is that what is a Hilbert space. So, a Hilbert
space is an infinite dimensional vector space and we will. So, a Hilbert space is infinite
dimensional vector space.
Let us just talk about a single particle Hilbert space to begin with. So, let us assume that
a single particle or a collection of particles which are non interacting. So, we are talking
about 1 such particle is represented by Hamiltonian H1. So, this subscript 1 stands for 1
particle and it has a Eigenfunction. So, this is the Hamiltonian of the system and the
Eigenfunctions are given by │ χ > and this χ is actually formed of. So, this χ is written as
or rather pronounced at χ, χ and this is formed of some complete set of states λ.
3
So, λ is the complete set of states here, set of states and obey λ obeys are a relation which
is. So, the │λ>< λ │ is equal to an identity matrix or a unit matrix which has only 1 as
the diagonal elements and 0 for all the off diagonal elements and so these λ form the
complete set of states for the Eigen functions of H1.
So, which means that H1, when it acts on these λ it gives an Eigenvalue equation which is
ε λ.
So, H1 acting on the │ λ>. I just touch upon what we mean by Ket and so, it returns an
energy Eigenvalue which is ε λ and returns the state ε λ. So, Dirac introduced this
notation of Bra and Ket and all throughout Quantum mechanics if you follow any book
on quantum mechanics modern book on quantum mechanics, they write a Ket with like
this and they write the Bra like this ok. So, these are the representations that are
followed. So, that is just as we have written in the last slide. This was an outer product
which is made out of the bras and now we are writing the Eigen value equation with the
Ket here.
So, this will be made more clear as we go along and let us give some examples of such
single particle Hamiltonian and let us talk about the most trivial example which is free
particle and the by free particle what we mean is that the free particle in a continuum.
The free particle Hamiltonian is represented, 1 particle Hamiltonian is represented by.
So, this -( )2/ 2m is h square, where is the planck’s constant which is h by 2 pi, m is the
mass of the particle and this is the Laplacian which in Cartesian coordinate system has a
form del square, del x square plus del square, del y square plus del square, del z square ;
however, often we go to another coordinate system in order to solve the problem and in
which case we have to write the Laplacian or the del square operator in other coordinate
systems such as spherical polar coordinate system or cylindrical coordinate system,
which is usually slightly more complicated, but they are available everywhere.
4
So, also let us write a spin half particle in a magnetic field B. So, H1 =-Bσz . So, this
magnetic field is in the z direction , it is assumed for simplicity. So, that is why we have
taken the z-component of the spinner matrix, which is a 2 by 2 matrix and is called as a
Pauli matrix. There are 3 Pauli matrices, σx σy σz .
So, where σz is the Pauli matrix in the in the z direction and let me make things a little
more involved and write down for 2 particles, these are non interacting particles will
write down H2 because this involves 2 particles. So, we are writing it as
H2 =1/√2 [ φ μ (1) φ ν (2) +- φ μ (2) φ ν (1 )]
And so, this is the Hamiltonian for 2 particles where this φ are the functions or they are
the Eigen functions, this μ represents an index which could be anything such as the
momentum or the spin or for that matter any other thing that is appropriate for the
description of the system.
So, in this case we have talked about the Hilbert space at the single particle level and
written down the Eigenvvalue equations.
Now let us generalize this to the many particle Hilbert space. So, before we proceed, we
have put a plus and a minus sign and we wish to assert that this plus sign is for Bosons
after an Indian scientist called S N Bose and the minus sign is for Fermions after named
after the scientist called Fermi, Enrico Fermi. So, to continue with the many particle
Hilbert space, we can write down the wave function as, So, this is μ1 , μ2 .
Now, I no longer want to write mu, nu because there are several of these indices that will
come and I am basically interested in writing down an N particle system and this has 1, 2
3 and so on and this is equal to 1/√N ∑ (-1)P φ μ 1(1) φ μ2 (2) ................ φ μN (N)
, and this is a many particle state there is a minus(-1)P, now (-1)P is where this P denotes
the number of permutations and if P is even, then we there is no sign associated with it.
So, there is a positive sign here because minus 1 to the power n even numbers a 2n is
5
even. So, we will simply write this without that (-1)P; however, we have when we have p
equal to odd, then we have a negative sign ok.
So, just to go back to the 2 particle problem, you see that for fermions there is a negative
sign. What I mean is that see here phi mu is for the particle at 1 and phi nu is for the
particle at 2 and this has been swapped ,the indices have been swapped. So, that the phi
mu is now for the particle at 2 and this φ μ is for the particle at 1 and there is a minus
sign and because we have made 1 swaps in going from the first term to the second term,
there is a negative sign coming which is apparent from this many body wave function as
well and this is equal to, so this whole thing (-1)P is replaced by 1 for bosons. Also if you
take a note of this for the fermions because of this(-1)P and there is a summation
involved, you can this summation is over all the particles this for the fermions, one can
write down this wave function in of the form of a matrix and this is called as the anti
symmetrised Slater determinant.
So, this wave function is the determinant of that matrix which you write and basically
every time you make a swap of the particle visa we the index that you are using here you
are getting a negative sign. So, there could be more examples.
So, this is and Let us give an example of many particle system. So, there is N-non
interacting particles, in which case we will write the Hamiltonian. Now this the
Hamiltonian is for many particle system even though they are non interacting this will be
a ∑ (pi 2/ 2m) . So, that i is actually the index for each particle. So, this is equal to a
summation over i and Hi, where Hi corresponds to a single particle Hamiltonian for
written for a particle i.
And similarly when we have N distinguishable, non interacting particle in a magnetic
field B, we can write that as H = -B ∑i=1 N σzi so, this is these are some of the examples
of the many particle Hamiltonian and the many particle Hilbert space will be the here
will be the product of the Hilbert space of the individual particles ok..
6
Let us now talk about propagators and how the propagators are interesting and useful in
quantum mechanics. So, we will start with the description of propagators in quantum
mechanics and we will see later on how this smoothly connects to the definition of
greens function which as I said earlier is an important topic for us to learn in this course.
So, we start with a time dependent Schrodinger equation, which if at all we use later we
call them as TDSE, Time Dependent Schrodinger Equation and which is written as
i =H
. In principle I should have written it with a partial derivative this one for the ;
however, I ignored that and wrote it as a full derivative, strictly speaking it should have
been a partial derivative which is written as del-psi del-t, it will not affect too much our
discussion that is that is going to flow. Now this can be solved by an a proposal of the
form un(x) . So, that is the answers or the proposal that is for
this problem take a note of this where Un-s are the basis functions.
So, for writing every state or an Eigenfunction we need a basis, just like what we have
introduced earlier Lambda was the basis for the Eigenfunction Kai, here Un is the basis
for this psi. So, the Un-s would have this obey this relation and if you put call this as
equation 1 and if you call this as equation 2, if you substitute 2 in 1, 1 is going to have a
relation which is i =EnCn
7
. Here you should also note 1 point in equation 2, that the time dependence of the wave
function or the Eigen function on the left is carried only by the amplitudes.
So, these are the amplitudes which carry the information about time while they are
independent of the of space and the space information is carried by the basis vectors
which span the entire Hilbert space for the problem. So, if we substitute 2 in 1, we get an
equation which is i =EnCn
, this is just the recast of the form that appears in equation 1. So, it is the same as the
Eigen for or rather Schrodinger equation written for the amplitudes Cn and the this will
have a solution which is equal to a Cn(t)= Cn(0)*
It is also good to mention here that the Hamiltonian does not explicitly depend on time in
this case, in which case we could not have written this because then En-s are not
independent of time and the system loses its time translational or time reversal
invariance, which will neglect at the moment and the solution comes in the form of this
the amplitude Cn(t), that is as a function of t is Cn(t)= Cn(0)*
and you can trivially verify that if you put this form in the equation that appears on the
left, then this satisfies this equation.
Let us call this as equation 3 and we will and then this equation 3, if its substituted in
equation 2 and then we shall have the * un(x)
8
Please note that this in this equation four ,Cn(0) is arbitrary and cannot be determined,
unless the wave function psi is known at time t equal to 0. By that what I mean is that
Cn(0)= un*(x) dx. In the modern bracket notation this can be written as <
│ > .So, if the wave function is specified for all x values that is all special values
at time t equal to 0, then one can determine Cn(0), which is the value of the amplitude or
the expression for the amplitude at t equal to 0. That tells that if we use this Cn(0) in 4.
So, if you put 5 in 4, then * un(x)
. Just to remind you that this Un star is actually the complex conjugate of Un and when
we write this definition it is implicitly assumed that definition in 2 for a wave function it
is implicitly assumed that the Un-s are complex quantities.
So, they will have they can be. So, there is a Un and there is a Un star and both would be
existing. So, if I have the . Another thing that you
should notice that in equation, let us call the top one of 6, we have used the which is a
dummy variable because this is a, is a function of x and now we are have introduced a
dummy variable which is summed over.
So, this equation, let us call it we can call it as equation 7, introduces a propagator for
. So, if we know , instead of solving the second order differential equation

given by the Schrodinger equation, which is written as equation 1 here, instead of
solving that we can actually solve this integral equation which is equation 7 and if we
know .
So, not only that, it takes contribution of at all the points and builds up this entire
integral from the contribution of psi from all the x prime points in order to have the value
of at x t and of course, at a given space time point of x 0 is or rather (x,t) is
proportional to psi of x 0. Now here my definition of the propagator un*
* un(x)
9
So, that is the definition of propagator and this propagator is formed of the basis
functions that we have used for writing down the Eigenfunctions . So, if you write it in
the modern notation it is un( ) │un(x)>
. So,that is the form of the propagator and this propagator as I said derives contribution
from all the points and sums them up to arrive at . So, once again I repeat that
instead of solving equation 1 in order to obtain psi of x t, we can actually solve an
integral equation which is equation 7 and which is for that one needs the knowledge of
and one also needs the knowledge of the propagator which can be obtained from
the basis functions as its given in equation 8.
10
Prof. Saurabh Basu
Lecture – 02
Propagators II
So, we have derived a formula for the propagator as we call it. So, it is G(x ,x′,t) which is
expressed in terms of these basis functions.
Which are un (x) and un (x′) and it takes the system from some x to x′ at a given time t.
11
So, as I said earlier that we can get ψ(x,t) by solving the Schrodinger equation, time
dependent Schrodinger equation which we have written as TDSE or we can also get it by
solving or rather computing the green of the propagator. So, what we want to show is
that or rather emphasize is that this ψ(x,t) at a given point is the sum of contribution of
coming from all the initial wave function, at all possible points x′ and that is taken into
account and the contribution from a particular point is proportional to G(x ,x′ ,t) and that
is why it is called as the propagator.
So, it propagates the wave function from x to x′ at a given t, now let us try to cast this
problem in terms of the propagator and. So, we have to find a equation of motion for the
propagator, just like the wave function the equation of motion is given by the time
dependent Schrodinger equation which we write it as TDSE and we should also be able
to find corresponding equation rather the equation of motion for the propagator.
So, the way to do it is that if you go back to equations 7 where we have written the
, I will simply multiply or rather operate this by this operator which is. So, we
will do it as (i – H) ’.
12
.So, this tells me that this will act, this operator here will act on this function which is
and this operator will actually act on the and will give
me an equation which is equal to
So, once again to iterate that in order to find out the equation of motion for the
propagator which is .I have taken this operator (i – H) which is
basically the time dependent Schrodinger equation and that is put here and since that is
equal to 0 we get this equation which is i There could be a, which is inadvertently I
have missed that it could be there is the
. So, let us call this as equation 9 and this as equation 10.
So, this is the Schrodinger equation for the propagator. So, far we have not done much
accepting that we have written down the time dependent Schrodinger equation in terms
of the propagator and one sort of similarity is very apparent is that in this problem we are
trying to cast the G the propagator with the same role that it has in ordinary quantum
mechanics.
And the reason we want to do would be apparent later or rather understood later in which
we will connect this propagator with the greens function and this greens function will
help us in formulating a number of things or rather compute the number of
experimentally observed quantities in terms of the greens function so here we may like to
actually talk about specifically t greater than 0.
13
So, what I mean is that this t that appears in your these all these slides is only greater
than 0. So, what happens to the wave function or the propagator as you move backward
in time is not discussed or rather it will be discussed in a more general sense, but right
now we want to concentrate on the fact that I only evolved a system positive in time. So,
that t > 0.
In which case I would define a propagator called as the retarded propagator and this
propagator is written with a GR with the same indices as we have for G. So, a GR is
defined as GR(x ,x′,t) which is equal to a G(x ,x′,t) and a θ function I will tell you the
meaning of the θ function.
So, this is equation number (11) where this θ function is defined as it is equal to 1 for t
greater than 0, it is equal to 0 for t less than 0 and this is also called as the step function
or it also has a name which is called as a Heaviside step function and, So, this
immediately makes it clear that a G(x ,x′,t) exists for all time positive and negative;
however, GR(x ,x′,t) exists or nonzero only for t greater than 0 and for t less than 0 it
vanishes because of this theta function the way it is defined here.
So, this is my definition of GR. So, it is very clear that GR makes a discontinuous jump,
jump from 0 from t equal to 0 to finite value. So, now the job of course, is trivial to find
out a Schrodinger equation for GR, we have already found one for G now we will have to
find the same thing for GR which is called as a retarded propagator. So, that is the job at
14
hand. So, I need to do a time derivative, a partial time derivative on the G(x ,x′,t) and
θ(t) more like a chain rule that we have for the derivatives for equation 1. So, let us see
how one does that.
So, I have i GR i i
, Now, what is the derivative time derivative of a function that can be easily
understood if I plot this figure to actually graphically represent theta function. So, this is.
So, this would be it is rather one minus function that I have drawn as a function of t
and a theta function would look like of t versus t and it looks like. So, this is t equal to
0.
So, this is my function and I have to take a derivative of the function, look this
function is 0 for t less than 0 anyway this function is equal to 1. So, this value is 1, this
value is 1 as well and this value is constant for all t greater than t equal to 0 and it only
has a discontinuity at t equal to 0. So, the derivative of the theta function is actually a
delta function. So, which is d/dt or a if you write it more correctly.
So, this is the definition of the time derivative of the theta function. So, I will simply get
a derivative so a delta function here which is a derivative of the theta functions. Now,
15
this if I recall the the Schrodinger equation that we have written down for G it becomes
automatically i GR H i
. Now, in order to have a delta t being multiplied with it we need to this would only be
non 0 a delta function is only non 0 at t equal to 0. In fact, it is infinite for t equal to 0
and at all other t it is 0. So, a delta function really looks like an infinite discontinuity at t
equal to 0.
So, thus it make sense that there is an infinite a slope for the theta function at this point
that is shown here. So, this, so to have t equal to 0 I will simply write this as. So,
. So, my equation for the greens function the retarded rather the retarded
propagated which is what we were calling it is this and if I rearrange this if I bring it to
the other side I will simply GR i
So, this is just written notationally because that will be this is by if you go back then we
have done this that the, actually this greens function from equation 8 or rather the
propagator in equation 8 could actually be written as delta of x minus x prime and that is
the delta function in time.
So, this is my Schrodinger equation for the retarded propagator. So, I will use a slightly
shorthand notation for the Schrodinger equation for retarded propagator. So, this
equation is what we need to solve for if we need to know the retarded propagator and this
propagator needs to be computed for a given problem in order to know the time
evolution of the wave function for that problem and this is going to be an important
equation for our analysis just in a while. So, now, let us, let us look at that how we can
connect this to a physically meaningful quantity such as a GR .
16
See omega is the energy variable rather it is the w is the energy. So, this is a energy
variable that is written as, this is an energy variable and with just the added to it it
becomes a energy dimension of energy and since the it is, it is conjugate to time. So,
omega is conjugate to time. In fact, the response of the system is often expressed in
terms of G(x,x′,w) which is at a given omega value or other energy value one can
compute the propagator.
So, we will we can write a GR here because this is what we are interested in and for
getting this we will have to do an integral GR
So, basically this tells you that there is a Fourier transform of GR t is being
taken, now remember that this has a form which is like un of x prime let me just make
sure that I use the same notation. So, it is un* .
So, there is a sum over n and if I use it there, then I will have an integral of the form
which is dt
We are not going to minus infinity because we are using the retarded propagator which is
only valid for all times t greater than 0 in which case we will have to solve this integral
and this integral diverges as time goes to infinity. So, this divergence is removed by a
trivial factor by introducing a small complex damping into this equation this will be done
over and over again. So, I will simply do it mathematically here. So, I take a small
imaginary part here and I will write it with dt, and
17
This integral is convergent. In physics we do not want to deal with divergent integrals
and because this integral is it diverges as t goes to infinity we need to put a damping
factor ideally a plus and my damping both would have been fine in this particular case
we would discuss it with a plus sign and I will tell you the reason that we are taking a
plus sign here and then my GR it simply becomes.
I am skipping one step which should fill it up and this is
GR
Where actually eta goes to 0 in the limiting case in the limit.
So, this would be my equation maybe this is my equation 12 and this I will write it as
equation 13. So, this is my propagator rather the retarded propagator, we should put R
there a retarded propagator written in the x omega space that is the this variables are
kept, the space variables are kept intact and the time derivative time variable has been a
Fourier transformed into the energy and we write it as a function of omega.
If you further want to get rid of this x and x prime variables in favour of the momentum
variables one should write it with GR and k omega where k would be a momentum vector
connecting the points x and x prime now what is the relationship between. So, the
relationship between propagator and greens function, I have used it a couple of times, but
18
in those all these situations we actually meant the propagator this G also represents a
greens function. Right now we are talking about greens function in quantum mechanics
or rather in a single particle description, but we will also talk about many particle greens
function in our subsequent discussion.
So, this greens function is defined as. So, the so this G+ is the greens function and this
greens function, I will tell you what the plus means this is
G+ = GR
. So, G+ is called as the Retarded Greens function and we will take a specific example
to compute Retarded Greens function as say for the free particle or for that matter any
case.
So, again the Schrodinger equation for the retarded Greens function would simply be
(H - ) G+ =i
And this is our equation for the Retarded Greens function. So, we this is the equation or
rather the equation of motion for the Retarded Greens function and it is also of
importance.
19
to calculate G+ in which case you need to do a Fourier transform of G+
now that would .
.Now, this integral can be a thought of or rather in order to compute this integral one can
think of to be a complex variable now this limit of omega actually goes from minus
infinity to plus infinity ok.
So, one really has omega going from minus infinity to plus infinity on a real line
however, because of this i eta one can think of omega to be a complex variable and there
is a pole if you are conversant with complex integrals, There is a pole at e n minus I eta
which means that it is in the lower half of the complex plane because your eta is a
positive number an infinitesimal small positive number. So, we will have a pole at
, it should be also divided by h cross let us call it then where
And, so, it is in the lower half plane.
So, it makes sense to actually do this complex integration by closing the contour from
below and closing it like this in a clockwise sense and call this thing as the gamma the
contour to be gamma. So, what I do is that in order to do this integral from minus infinity
to plus infinity I convert this into a closed integral over this, over this entire loop that is
shown and I will subtract out the curved portion from that.
So, that I get what I asked you want, this is the first term which is a closed loop integral
can be found out from the Cauchy’s formula and which is the 2 pi i into sum of residues
and the second term will vanish because of the Jordan’s lemma. So, I will write that
briefly. So, this from the Cauchy’s formula and this one vanishes because of Jordan’s
lemma.
So, in which case what one gets is that let us call this integral to be off. So, this involves
an integral let us call that integral as In.
20
So, this In = 2πi *
This is of course, for t greater than 0 and this integral is equal to 0 for t less than 0. Now,
to get back to our Retarded issue had we added a minus in this expression in order to
make it a convergent integral in which case we had to actually talk about only t less than
0 which is called as the Advanced Propagator which would have given rise to an
Advance Greens function.
So, right now we are talking about Retarded Propagator because most of our physical
problems or the visible observables are related to the Retarded Greens function. So, then
my G R which is the Retarded Propagator is related to the Retarded greens function
which is GR G+
and which is nothing, but GR un In
and this is my either you call it a. So, so this is how to get my retarded propagator or the
retarded greens function for this problem where my G R is the retarded propagator and
G plus is the Retarded Greens function.
So, this is in contrast with my advanced propagator which we will call it as G A as the
advanced propagator and. So, and G- is my Advanced Greens function and in the next
we will give an example how to calculate this propagator for a given problem. Just once
again to remind you that instead of solving a differential equation given by the
21
Schrodinger equation you can solve an integral equation for the propagator or the greens
function to have the same amount of information that is available by solving psi from the
time dependent Schrodinger equation.
22
Prof. Saurabh Basu
Lecture – 03
Second quantization I
To summarize what we have learned so far, we have defined what is a propagator which
propagates a wave function from an initial time which could be at t equal to 0 or may be
from a finite time t equal to t1 and it propagates to a different time which is at t2 which
where t2 is greater than t1 and that is called as a propagator and we have also calculated
the propagator or rather have related the propagator to the Green’s function for a
particular problem. Now, we have to learn, how to compute a propagator for a given
problem and we take the simplest possible example that is a free particle in continuum.
So, we are going to calculate the propagator or the Green’s function.
So, Green’s function or equivalently the propagator for a free particle is what we are
going to compute. So, just for your information free particle means that the energy
momentum relationship is given by E=p2/2m= 2/2m for a free non relativistic particle.
So, we are going to write the for this the basis function is a plane wave. So, it can be
expressed in plane wave states and hence if you remember the basis states that we have
written down earlier which now we can write it as un(x) = A eikx = A = up(x)
23
Where p is the momentum of the particle which is a well-defined quantity and the
quantum number corresponding to the momentum is a good quantum number.
So, if we have these as the basis states the basis states also obey this orthogonality
relation and. So, this is up(x) up(x’) =
and my if I if I write down the in terms of the energy eigen states then my
uE(x)= )
Where if you note that this is the normalization factor and this is well. So, this is the
normalization and this is the plane wave part written in the energy basis.
So, we will take this as the basis states and calculate the propagator or equivalently the
Green’s function.
So, this will call it as equation one and now let us write down also the orthogonality
relation for the basis states as uE(x) uE’(x) dx= ok.
So, these are the orthogonality relation that the basis states will follow and then the
propagator or the Green’s function can be written as
G = uE uE
24
and then we have if we put the values that we have or rather the expressions that we
have taken in equation one then in this expression which is 2, then we get this as
So, this is the Green’s function or the propagator for the free particle and we have solved
this integral and to solve the integral one can make a substitution K=
dk = and following that I can write down :
G *
And this is the form that I have and I have to evaluate this evaluate this integral and this
integral can be evaluated easily and one can write down G .
So , which is G?
So, this is my equation 4 and this is the final form of the propagator. So, as we have said
that for a particular problem such as this a free particle in a continuum, if we find this
quantity this will take a wave function say from x equal to at t equal to 0 to t equal to
some finite t by the operation of this propagator provided that my psi at x equal to 0 is
known for all possible values of x.
25
So, just to repeat that is :
. So, for a given problem that is a free particle in a continuum denotes the
propagator. So, G is the propagator and we also know how to link it to the Green’s
function and in particular we have learned for the retarded Green’s function, but we can
also write it for the advanced Green’s function in a similar manner where we just need to
write the advanced Green’s function GA =
So, to remind you the theta function when minus t is greater than 0 then that is this theta
function is equal to one and if it is not satisfied otherwise the theta function will give us
0. So, this is the definition of the advanced Green’s function.
So, to summarize we have so far, whatever we have done is that we have defined a
propagator and from the propagator we have learned how to calculate Green’s function
and how this propagator or the Green’s function can talk about time evolution of wave
function of psi of x t in quantum mechanics and then we have solved it for a for an easy
problem rather easy problem here that is on the free particle and calculate it the
propagator for the free particle.
26
So, the next thing that we have on the cards is to talk about second quantization and we
would first define what is second quantization and then what do we do with this. In fact,
we would write all the Hamiltonian in the second quantized formalism for us to attain to
a problem in solid state physics. So, the second quantization is formalism or a technique
which is different than the first quantization which is we all know as the quantum
mechanics.
So, what happens in first quantization or rather, the conventional quantum mechanics ah?
So, in quantum mechanics we have observables which are represented by observables are
represented by operators and the state functions or the eigenstates states are represented
as by functions which have continuous functions say that of the space variable x or x y
and z or r theta and phi.
So, this is what we have learned in quantum mechanics recall your discussion that you
have for the either the hydrogen atom or the harmonic oscillator in either of the cases we
have written down a Hamiltonian in terms of operators which are the Laplacian operator
and operators that go as it depends on space such as either x square in one dimension for
the oscillator or it says one over r for the hydrogen atom and then when we solved a
second order differential equation which goes by the name Schrodinger equation and
then we have got the energy eigenvalues for the harmonic oscillator problem the energy
27
eigenvalues came out as n plus half h cross omega where omega is equal to root over k
by m and in the hydrogen atom problem.
The energy eigenvalues came out as minus 13 point psi 6 by n square electron volt and in
both the cases n is the quantum number for the problem and n is quantized as either 0, 1,
2, etcetera for the harmonic oscillator and n equal to 1, 2, 3, etcetera for the hydrogen
atom problem and the Eigen states that were computed from solving by solving the
Schrodinger equation they are simple functions of r which are in the case of hydrogen
atom they are known as Laguerre polynomials in the name in the case of the harmonic
oscillators they are known as Hermite polynomials.
So, this is what we are talking about that the states are represented by functions now in
the second quantized formalism. So, second quantization we are going to represent the
observable anyway by operators which were done in quantum mechanics as well;
however, the states are now going to be represented by are represented now by operators
as well operators and these operators are known as a creation and annihilation operator,
they are also called as creation and destruction operators, but this destruction and
annihilation they are used interchangeably we will also probably do that.
So, the commutation and the anti-commutation of the creation and then annihilation
operators will define the property of the wave function and in the case of fermions this is
an anti-symmetric wave function and whereas, in the case of bosons that is a symmetric
wave function.
28
So, the commutation or the anti-commutation, rather, I will write and anti-commutation
relation of the operators and what we mean by operators of the creation and the
annihilation operators which you have seen in some form when you did the operator
methods in quantum mechanics and so, we extend that to many particle system you have
used it for the harmonic oscillator problem.
So, these are these operators they obey certain commutation relations or anti
commutation relations which we are going to learn and these commutation or the anti-
commutation relations they fix the nature of the wave function of the behaviour of the
wave function; whether the wave function would be anti symmetric or symmetric with
respect to interchange of particles that is known as anti-symmetric or symmetric property
and these relations will actually decide that decide the symmetric as I said that is for
bosons and anti-symmetric which is for fermions nature of the wave functions basically
the typical state is made up from a vacuum where a state like this is made of from the
product of the single particle states and where
│ ( │vac> =( │0>
. So, vacuum is a null state which has no particle. So, a lambda successively operating on
this m times will create that many that is capital N number of particles. So, these are
going to give me single particle states and I am going to get a full wave function starting
from a single particle states by successively multiplying or rather operating the creation
operator on the vacuum.
29
So, the vacuum is also written with a 0 and we can write it like this. So, that is my state
and as I said that in the second quantized notation the states are formed by the creation
operators or the annihilation operators the annihilation operators i have not written, but
one can define the property of the annihilation operator as one which is not written with
a dagger and which it acting on the vacuum gives me 0 which means that vacuum has no
particles.
So, there is no way that one can annihilate a particle from there now one important thing
is that that a m dagger and a m are Hermitian conjugate to each other Hermitian rather
Hermitian adjoint to each other and they are distinct operators and they are not self
adjoint. So, not a self adjoint and they are distinct operators. So, now, with this
background let us go back and compute or rather at least initially define the states for
both fermions and bosons and how they look like. So, we will the method of second
quantization is what we are going to discuss.
Now, keep in mind that there are a lot of books which they do this method of second
quantization or write the Hamiltonian or any operator in the second quantized formalism.
So, you need to actually focus on any one book because the notations that are used from
one book to another very like which will be sometimes difficult for a reader to follow.
So, follow these notes along with a book that is going to be prescribed and then you will
learn this thing better.
30
So, what we are trying to say is that lets take a single particle Hamiltonian H
H│ │ and returns me the same state. So, this lambda can in principle be
any quantum number for that which is suitable for that particular problem which could
be momentum or energy and so on. So, we are uncommitted about that at this point and
we write this as the eigenvalue equation now we can write for the wave function say for
2 particles and they could be. So, 2 particles, we want to generalize this equation and
these 2 particles can be fermions or bosons.
So, and then one can write down psi for a fermions at two. So, these are say space
variables x1 and x2 which is equal to one by root 2 and I can I will use both 2 notations
which is x1 lambda 1 and x2 lambda 2. So, whatever that state lambda is. So, we are
writing it in terms of the space variable. So, it is a. So, this would represent a state in the
position space and this is equal to and there is a minus sign and there is an x1 lambda 2
and there is a x2 lambda 1 and that is the state and as I said that I will use another
notation for that. F(x1 , x2) =1/ *[< x1│ 1>< x2│ 2> - < x1│ 2>< x2│ 1>]
So, we can also write this as this is equal to 1/ [ 1 2(x2) - 1(x2) 2(x1) ]
possibly this is a more simple notation for you because we are writing down the wave
function psi is the quantum mechanical wave function and it is expressed in the position
space and so, x1 and x2 refer to the positions of the 2 particles or the 2 fermions that we
are talking about and there is a necessity and necessarily, there is a negative sign in
between because we have swapped particle number one and particle number 2 while
writing the second term as compared to the first term; however, this will not be there for
a boson.
31
And one can write down the wave function for a boson equal to B(x1 , x2) =1/ *[< x1│ 1><
x2│ 2> + < x1│ 2>< x2│ 1>] and this as earlier can be written down as 1/ *[ 1 2(x2)
+ 1(x2) 2(x1) .\
So, we can write it with. So, either for fermions or bosons we can write it with a in a
more compact notation, we can write it with a F/B(x1 , x2) = 1/ *[ 1 2(x2) + 1(x2)
2(x1) and this p equal to minus 1 for fermions and it is equal to plus 1 for bosons and in
a similar fashion we can write down a many particle wave function.
Let us now write it with a big psi which is nothing, but a lambda 1 lambda 2 and all the
way for N particles and that is equal to.
= =
So, there is a normalization that is used here which is N factorial and there is a lambda
equal to 0 to infinity and n lambda factorial. So, this and there is a sum over k and there
is a P which is 1 minus I will write it in Sgn k divided by 2 and this is equal to all these
single particle states lambda k 1 and product with lambda k 2 and product and all the
way up to lambda kN as I said that this there is a particular index or the quantum number
32
that is used here which are k 1 k 2 k N which can be either momentum or spin or position
or energy depending upon the suitability of the problem.
So, the question is what is this p to the power one minus s g n k? So, just to say that that
when we actually swap particles that is if we make transposition of particles that is move
particles from one place to another which are these things are being swept this lambda
one lambda 2 indices being swapped and in that case we have S g n k is equal to it is
equal to one or minus 1. So, the 1 is plus 1 is for even number of transpositions number
of transpositions and minus 1 is for odd number of transpositions.
So, it is clear that for sign of if it becomes plus 1, then I have p to the power 0 which
means it is equal to 1. So, if you trans or rather swap particles even number of times then
one does not pick up a sign for the fermionic wave function; while if it is done odd
number of times, then one picks up a negative sign. However, this is only for fermions
for bosons it is always equal to 1.
33
Prof. Saurabh Basu
Lecture – 04
Second quantization II
So, we named this as equation 1.
And to remind you that we have this factor which is 1 divided by root over of N factorial.
And then there is a product of n lambda factorial and lambda really goes from 0 to
infinity. So, this is our normalization factor and these normalization factor normalizes
the many body wave function psi which is written in equation 1; the left hand side of
equation 1.
And the, this quantity which is P to the power 1 minus S g n K divided by 2; it

symmetrizes the wave function. So, what we mean by symmetrizing the wave function is
that as been told earlier that for the Fermions any change in the position of the particle or
transposition of two particles would bring in a negative sign. And if that is done twice
then it brings back another negative sign which means it becomes positive whereas, for
Bosons it is always positive no matter how many times the transposition is done.
So, this is our many body wave function that we are going to talk about now let us look
at a typical many particle state as this.
34
So, we are giving an example of a many particle states let us say that it is written with a
ket which is 1, 1, 1, 1 and then 2, 2 and then 3, 3, 3 and then its 4 and then it is 6, 6 and
etcetera. So, what is meant by this notation is that at the real space that is i equal to 1 as
the position space ah; for i equal to 1 there is one particle the for i equal to 2; there is
another one again one particle then at i equal to 3 then there is another one particle i
equal to 4 one particle, i equal to 5 two particles i equal to 6 two particles and so, on.
Now, this notation for a many particle system can be simplified; if we simply write this
as 4 and then 2 and then 3 and then 1 and 0 and 2 and so, on. So, in going from this many
particle state on the top from in the written in the position basis we have gone to a many
particle state where we have simply written down the number of particles that is at
position. So, so these are there are 4 number of particles at the first 4 positions rather. So,
there are 4 occupancy in the first 4 sites or locations and then it is 2, then there are 3 and
4 ,1 and there is no 5. So, if we put a 0 and then there is there are 2; 6 and so, on.
So, now this is written in a basis which is called as the Fock basis. So, what is the Fock
basis; so, we have written down this many particle states or the wave function in. So,
telling that how many particles occupy. So, occupy position i; so, for Fermions of course,
we know by the exclusion principle Pauli’s exclusion principle.
So, the numbers will take values the numbers that is written here the numbers will take
values 0 and 1 and for bosons, they can be any because Bosons do not have any
35
exclusion principle embedded there. So, as many as possible or as many as number of
particles are there, they can occupy the same energy level unless it is been or rather
constrained by any other principle or any other you know symmetry argument.
Now, this writing it in the occupation basis or the particle number basis is called as the
Fock basis. And we will call this as a occupation representation and this is same as the
Fock basis or Fock representation. And this is after the name of Russian scientist after
Russian scientist of V. A. Fock.
So, now, we will write this Fock basis as. So, this is written with a curly F a curly F like
this and so, the many body state would be written as in this occupation number is this I
am writing it with a ket we could have written this with the ket as well. So,
n1.,n2………………..>.
So, that is some many body state where are the coefficients and these are the
Fock basis. So, the occupation number basis. So, we have occupation n1 at a given side
and then occupation n 2 and occupation n 3 and so, on. So, n 1 and n 2 will all be equal
to either 0 or 1 for Fermions whereas, for Bosons there could be any number.
Now, in the; so, in general we have =N . So, the total number of particles is equal to
N; however, in a grand canonical sense we can relax this constraint and we can say that
we have a Hilbert space which is large enough to accommodate infinite number of
particles. So, then the total Fock space is written as Fock space is written as it is equal to
36
a notation that I will use I will just tell in a while its N equal to 0 to infinity and this is
equal to the power N where this is known as the direct sum.
Now this Hilbert space must necessarily include F 0 which is vacuum means that it does
not contain any particle and it necessarily should be to include in the family of basis
states. So, now, we have a state which is given by this and written in the occupation basis
or the Fock basis and in the Fock basis; the total Fock space is written in terms of the this
F to the power N and written as a direct sum for from n equal to 0 to infinity. And we do
not have this constraint of this one any longer that is sum over N i which is basically the
occupancy of each state has to be equal to some finite number N that is not there any
longer.
So, what is this occupation number representation useful for? So, what is the benefit
derived from starting from a representation which we have written in the position basis.
And then we have tried to go on to the occupation number basis or the Fock basis what is
the benefit of doing that. In fact, it of course, as you can see that this looks much more
simpler as compared to the representation that has been used earlier that is the position
representation.
However, in the position representation there is an enormous summation involved here in

equation 1 which is a sum over K and then there is a P to the power 1 minus S g n k by 2;
it is been told earlier and its written here as well that S g n k is either equal to k plus 1 or
minus 1 and when it is equal to plus 1 then there is no sign that is this factor is equal to 1.
However, when there is a S g n k is equal to minus 1 then we have sign. So, that is the
becomes P equal to 1 where P is equal to plus 1 for Bosons and P equal to minus 1 for
the Fermions.
And from there we wanted to go to this occupation number basis and even though it
looks simpler, but this enormous sum that appears here how is it going to be taken care
of. And remember that this sum is important for the symmetrization of the state and this
symmetrization gives the Slater determinant for the Fermions which is the anti
symmetric property of that matrices are encoded in transposition of particles; for
Fermions which would pick up a negative sign. As one does a transposition or changes
the position of one particle.
37
So, how that is going to be encoded into this occupation number basis? And in principle
we are need to be worried because we are talking about 10 to the power 23 number of
particles and transposition of them leading to the right symmetrization for a particular
wave function need to be taken into account. So, this is done very elegantly in the using a
second quantized formalism.
So, this is where the second quantization is important and will tell you how it is
important. So, let us define a creation operator which is acting on a many particle system
such as this. So, n1.,n2………………..> and that by definition we are now using a
definition that we are going to we are introducing ,introducing it for the first time
PSi n1.,n2…ni+1…..> .
And so, what it does is that a i which is a creation operator. So, this is the definition of
operation of a creation operator. So, creation operator acts on the ith particle and raises
the occupancy of the ith particle from n i to n i plus 1; brings in a coefficient which is
root over n i plus 1 additionally to take care of the symmetrisation requirement brings in
another factor which is P to the power S i and where = Si. So, we will have to
sum over all the occupancies up to that ith side just before the ith side and that ith side of
course, the occupancy is raised by one.
Now, it is important to note that; so, in a fermionic system. So, for Fermions this n i is is
modulo 2; so, what we mean by modulo two is that when you divide a number a by n it is
called as a modulo n and what it means is that; when you divide a by n the answer is the
remainder of this division. So, for Fermions the division will leave a number which is
38
either 0 or 1 because i am dividing by 2. So, n is equal to 2 here and so, n i is modulo 2
means that automatically it means that it can acquire values 0 or 1. And that is the reason
or rather that takes care of the fact the that the occupancy of Fermions at each side could
be either 0 or 1 and so, that emphasizes the Pauli’s exclusion principle.
Now, let us take a; so, basically; so, P to the power S i; it makes accommodates or rather
incorporates the symmetrisation, the proper symmetrization of the many particle state.
So, let us give an example take a many particle state which n 1; n 2, n 3 and so, on. And
let us operate by a creation operator which operates on this third occupation and then we
can write this n1.,n2……n3…………..> = P(n1+n2) n1.,n2…n3+1…..>.
So, this operation of the creation operator acting on n 3. So, as usual a P takes a value P
takes plus 1 for Bosons and minus 1 for Fermions and so, this P s i here n 1 plus n 2 this
takes care of care of symmetrization of the state; which means that if n 1 plus n 2 is equal
to 1. Then we have an anti symmetric state because for Fermions it is an anti symmetric
state because it is minus 1 whole to the power 1; however, if n 1 plus n 2 is equal to 0
then we have a symmetric state or n 1 plus n 2 equal to 2; we have a symmetric state.
And so, this is how the in this Fock basis; the symmetrization of the many particle state
is encoded and it is very elegantly done in the second quantized formalism you will get
more glimpse of it as we go along.
39
So, now let us then write down a wave function many particle wave function in the Fock
basis as n1.,n2,n3…..> = (ai+)ni 0> let us call this as equation 2. And if you
remember that we have insisted of having this vacuum or that F 0 that we had written
which is null state or a particle without any particles as an important element in the
family of the of the Fock basis.
And this is we actually create a many particle state by successively operating the creation
operator on the vacuum and so, this complicated permutation entanglement; which is
there in equation 1; here by this summation over K and P 1 minus S g n k by 2; now that
is encoded here and how would it be encoded? It will be encoded through the
commutation relation of these operators. And that is why this formalism of second
quantization is so, helpful for dealing with many particle systems. So, so, this basically
tells you that there is an enfold application of this operator a dagger on the vacuum state
to build a many particle state.
Let us try to understand that how the symmetrization is built in by the commutation
relation of this operators; let us take only two particles to begin with. So, we have two
particles and say they are located at a i or rather i and j and they are; so, we just write
them as a i dagger and a j dagger as the creation operator for these two particles. So, as if
they are acting on vacuum.
So, a i dagger and a j dagger acting on vacuum gives me that state which is what I want
to consider. Let us call it just a two particle state and let us look at the commutation
relation of these two particles. And needless to say here that i is not equal to j because for
Fermions it is anyway forbidden to have i equal to j you cannot have two particles being
formed at the same site or being created at the same site.
So, let us define the commutation relations by this ( - P )=0 . So, what I have
done is that; I have written them as i and j and then in the next term; I have written with a
minus i mean with a j and i in addition to that we have a minus P.
40
And so where P is equal to where P equal to 1 will stand for denotes commutation
relation and P equal to minus 1 denotes anti commutation relation.
Let us see clearly; what it means is that we have a i dagger a j dagger and I will write a P
sub script here which is equal to 0. So, the notation is that if I have two operators A and
B and written with a P as a sub with a square bracket and this is equal to A B minus
PBA. So, for commutators; this relation is P equal to 1 which is equal to A B minus B A
and for anti commutation.
So, this is the commutation relation of A and B and this is the anti commutation relation
for A and B. So, if this commutator is equal to 0 then we say that this two operators
commute. And when two operators commute it is usually correct that they have same set
of Eigen functions they have common Eigen functions. And if they do not commute then
that does not hold and these this is the anti commutation relation of since the anti
commutation relation ok. And if we go back to our a i dagger a i dagger then that is equal
to 0 because the identical operators would commute.
41
And so, if we summarize; here these relations for all i and j a i dagger; a j dagger equal to
P equal to 0. So, ah; so, similarly we have we can write this as. So, for both P equal to 1
and P equal to minus 1 and just like the similarly if we want to write it for the
annihilation operator; just like that a creation operator takes N component Fock state
from this to this F N to F N plus 1; we can also talk about annihilation operators on
which we will take my Fock state N to N minus 1.
And we have relationships similar relationships; commutation relationships for the

annihilation operators as well and they are written in a compact form by this relation
which is equal to this and a i; a j P equal to 0 and a i dagger a j dagger P equal to 0.
So, the enormous complex the complexity of the of the many particle space and the
complexity that is related to symmetrization of the many particle wave function; is taken
care of in the Fock basis by choosing or rather from the commutation relations or the anti
commutation relationships of the operator. And this is a phenomenal improvement of the
situation as we can see in the next discussion.
42
Prof. Saurabh Basu
Lecture – 05
Second quantized Hamiltonian
So, we started with discussing propagators and a propagators for a for; in the context of
quantum mechanics and then we introduced greens function and how the propagator is
related to the greens function. And then we have started talking about second
quantization. So, which is basically the beginning or start of the course and so, it is a
good time to talk about the references which we should be referring to.
So, the references are some of the references are I will probably include more references
as we go along, some of them are quantum, many particle system by Negele and Orland
it is a Levant books publication and it has an Indian edition; the quantum theory of many
particle system which is a classic by Fetter and Walecka, it is a McGraw hill publication,
many particle physics by Gerald D Mahan, it is a Springer publication and quite a
popular book, then it is a Condensed matter field theory by Altland and Simmons which
is a Cambridge University press publication and then for other things such as super
conductivity will mostly refer to Tinkham which is Dover publications and in addition to
43
that we will refer to some other scientific articles other than books or may be research
papers which we shall a site from time to time.
So, with this let us go back to the issue second quantization that we have been discussing
and let us now convince ourselves that the one that the symmetrisation of the quantum
many particle state which was encoded in a complicated factor and which have to be
summed over as we have seen in previous lecture, in equation one to be precise that can
be taken care of by using states which are formed by creation and annihilation operators
and this symmetrisation is actually taken care by the commutation relations of these will
operators the creation and the annihilations operators. So, let us take a simple example.
And where I can write down a quantum state as .
So, this if you change 1 and 2, it would go to - and we have seen earlier that
for bosons, it does not make a difference that that is there is no sign that comes in front
of this the second term on your right; however, for fermions, we get a negative sign.
So, this is negative sign comes for fermions and just that this negative sign can also be
obtained if we consider the anti-commutation relation of this 2 operators which is written
as this which is equal to 0 and this immediately means that the =0
and the consequence is that that if I change the position, I will get a minus sign as you
can see from here. T
44
So, this minus sign that you are getting is actually taking care of the swapping of particle
or transposition of the quadrates of the particle and hence we shall pick up a minus sign.
So, I think by now, it is clear that how the tremendous work of symmetrising ten to the
power 23 number of particles which is Avogadro number of particles in any system how
that huge symmetrisation requirement can be taken care of by writing the many particle
state in terms of the creation and the annihilation operators. So, now, we shall go ahead
and talk about.
So, the how to write down a single particle operators and may be 2 particle operators or
rather many particle operators in this second quantized form in terms of the creation and
the annihilation operators. So, we now know how to write down a state psi in terms of a
dagger, etcetera the next thing is talking about one body operators.
So, this is the next thing. So, representation of one body operators and there are many
examples of one body operators and they would in principle be written as this and then
there is a sum and then there is a n. So, we always put a cap when we particularly want
to denote operators and. So, this single particle operator this sub one or the subscript one
actually talk about that it is a 1 body operator. So, it is a summation over such operators
for n particle where n is any number in principle can go from 0 to infinity and so, I am
writing down a 1 body operator as a sum of the one body operators of a system where n
refers to the number of particles or any other quantum number that is relevant for the
45
problem and similarly we can have an example for this is. So, one example is the kinetic
energy operator.
So, this kinetic energy operator talks about or this quantifies the rather it is it gives you
the observable for the kinetic energy for a given system and we can write this as T=
where m is the mass of the particle ok. So, this is the total energy of a system
which consists of the individual energies of the particles which are represented like this,
then we can have one body potential which is written like V of.
So, this will be like V= Vn(xn) here p is also an operator which can have a form
which is my minus ih cross Del which is the differential operator and similarly we can
also have the spin operator. So S= Sn and so on. So, there are many examples of
that. So, the question is that how to write one body operators in terms of the creation and
the annihilation operators. So, we will write then as a dagger. So, that is the question that
we are going to handle now.
So, now let us look at this problem and let us define occupation number operator,
operator which is n lambda which is equal to a lambda dagger a lambda.
So, this operator operates on a particle state say lambda j which. So, this is the; so, this is
the operator and it is acting on a many particle state and gives me a n lambda j and then
returns me back the state n lambda 1 n lambda 2, etcetera. So, it basically counts the
number of particles in the state lambda. So, I will write this because this is important
counts the number of particles in the state particular state lambda.
46
So, this operator operates on a particle state say λj which. So, this is the; so, this is
the operator and it is acting on a many particle state and gives me a
n^λ | n λ , nλ ,..........> =n λ j | n λ , nλ ,..........> . So, it basically counts the number of
j 1 2 1 2
particles in the state lambda. So, I will write this because this is important counts the
number of particles in the state particular state lambda.
In this case we have taken lambda j as one of the representative states of this many
particle system. So, if we have such a definition for an occupation number operator then
we can write down any one particle operator say let us call it as O 1 as we have written
earlier which considered is assumed to be diagonal in lambda basis what it means is that
a 0^1=∑ O λ | λi > < λ i | . So, this is the meaning that is diagonal in the lambda basis and
i
^ | λi > .
where my O λ =< λ i | O i
The lambda is a particular state as we have shown that. So, now, with this nomenclature
let us calculate the matrix elements of O 1. So, the matrix elements of O one is written as
I <nλ ' , nλ ' .... | 0^1 | n λ , nλ ........>
1 2 1 2
am sure that you know that the matrix elements of an
operator is computed in this fashion. So, you squeeze it between 2 states which are from
the Hilbert space.
So, n lambda prime and n lambda form the Hilbert space of the problem and then you
have a, this is equal to a
<nλ ' , nλ ' .... | 0^1 | n λ , nλ ........> =∑ O λ n λ <nλ ' nλ ' ... | nλ ' , nλ ........ '>
1 2 1 2 i i 1 2 1 2
and so on. So, that is my matrix elements of the one body operator which I am trying to
write in terms of the creation and the annihilation operators.
So, this equality actually holds for any set of states. So, one can write down for O 1 the
second quantized representation as in the following.
47
So, your O 1 is this and so, this is a lambda equal to 0 to infinity and then there is a O
lambda n lambda O1= .
So, because we have said if you look at equation let us call this as equation 1 and let us
call this as equation 2 this as equation 3. So, if you look at equation one then my operator
O 1 is represented as the matrix element of that operator in the lambda basis as you can
see and then there is a lambda dagger a lambda which is nothing, but the occupation
number n lambda that is written here and let us call this as equation 4.
So, the interpretation is straight forward that single a 1 body operator enrages a single
particle state at a time and then the others other states are simply just the spectators and
in a diagonal representation one simply counts the number of particles in a state lambda
which is given by n lambda and multiplied it corresponding to the Eigen value of the one
particle operator between its Eigen states lambda and now one can generalize it to a non
diagonal basis and can write this as this operator now in terms of a general basis where
the operator is not diagonal in the lambda basis you can still write it as. So, this is the
representation of a general representation of a 1 body operator in terms of the creation
and the annihilation operators. O1=
So, this is called as the second quantization or the second quantized notation for a 1 body
operator let us take an example let us call this is equation five and now let us write down
for a spin operator. So, we are giving examples and this I will write this Si operator and
write the matrix elements of the Si operators between 2 states which is given by alpha
and alpha prime. So, this is basically nothing, but the components of the spin in ordinary
48
quantum mechanics you have studied spins even in a non-relativistic mechanics quantum
mechanics we add spin by hand where as in the Dirac theory the relativistic theory the
spin comes naturally.
So, the. So, in for a spin half object the spin operator is a 2 by 2 operator and for spin
one, it is 3 by 3 operator for spin half it has a special name called as a Pauli spin
matrices. So, the components of the spin, this is equal to half of a sigma i and alpha alpha
prime there is a h cross here in place of one, but we have taken it as this h cross has been
taken as one here.
So, this is the representation of rather of a spin operator and this spin operator is going to
be now written in terms of the creation and the annihilation operators. So, just before
that. So, just for completeness let me introduce the Pauli spin matrices as. So, I have a
sigma x which is 0 1 1 0 a sigma y corresponding to the special direction y and there is a
sigma z which is 1 0 0 minus 1 the important thing about them is that the trace of all
these is 0 identity sigma x and sigma y are of diagonal matrices where sigma z is
diagonal. In fact, because it’s diagonal it is chosen as a preferred variable or the z axis is
chosen as the proffered direction in most of the problems that we do in physics and. So,
what I mean by trace equal to 0 is that the sum of diagonal elements is equal to 0 you can
see that check that here.
So, now I can write down a spin operator as simply S=
where lambda actually denotes some other quantum number and here it is basically the
space variable. So, lambda denotes a space variable such as x and y. So, this is the
representation of the spin operator in the second quantized notation where we have
written as. So, this is the matrix element is the matrix element as we have shown in
above and then and etcetera they are creation and annihilation operators let us do
an another example which is of used to us let us do it for the kinetic energy and. So, we
can also do it for.
49
Kinetic energy and a one body potential and this will specifically write in the real space
basis.
So, a Hamiltonian consisting of the kinetic energy and a one body potential is written as
H= (r) [ ] (r) So, these rs are so, these are vectors and written as
this and. So, this is the Hamiltonian which consists of a kinetic energy and a one body
potential the kinetic energy is given by and. So, and this is just a potential at a
given site r and a p is written as minus i h cross del where del is a operator it is a
differential operator that acts on space.
And similarly, a local density operator can be written as (r) (r)
and in the above example, you see that we have written it because r is a space variable in
continuum in the continuum representation we have written it as an integral instead of a
sum and this integral is taken in d dimensional. So, it is a ddr. So, where as in 3
dimensions we would write it as d cube r or you can write it as a volume integral dV the
same thing.
So, now having done the one body operator, let us talk about the twp body operator
representation of two body.
50
So, these 2 body operators are used to denote pair wise interaction between particles. So,
whenever 2 particles are interacting with each other by a potential which could be of any
form which could have form such as one over r which is called as a Coulomb interaction
or for that matter any other form of interaction then this called as a 2, we need a 2 body
operator to represent that and in principle, we can do it for any n body operator.
However, interaction potential is beyond the 2 body term is not something that we
encounter in physical problems or we see in physical problems and that is why we shall
restrict ourselves to a 2 body operator and that could be good enough for writing a
Hamiltonian for a system for a many particle system. So, let us see how we can do that
and for doing that we know that in classical problems there is no sort of it can be written
straight forwardly because of the distinguishability of particles; however, in quantum
many body theory or quantum mechanics the indistinguishability makes it somewhat
harder to write.
So, let us take a symmetric potential symmetric two body potential by symmetric what I
mean is that V(rn,rm) where m n and m, they refer to 2 sites may be and this is equal to
V(rm,rn) . So, the potential or the interaction potential that one body exerts one body at n
exerts on another body at m is the same as the interaction potential that is exerted by the
body at m and on n. So, this is we have now. So, we will write this as. So, we will take a
real space representation and these are vectors and we have say we are going from
initially we have started with a coordinate representation or a position representation for
a many particle state and then we had gone to a occupation number or the number
51
operator formalism and now, in this particular case, we are coming back to the real space
representations, but ultimately you will have to write down a Hamiltonian in the same
basis or rather if you have to solve it, you have to do it on the same basis; however, here
we prefer to do it in the real space basis and we write it as this V r n r m, we can put
vectors here as well and this is equal r 1, r 2 and there is a r N.
So, this is where we have to restrict the sum to all N values which are less than m
otherwise we would be double counting because V r n r m is same as V r m r n if we do
not have this constraint then we would be double counting this interaction term;
however, it is difficult to keep track of these indices one being less than the other. So, it
is easier to introduce a factor of half and write this as n equal to or n not equal to m and it
goes like this.
V r1.,r2……rN.> = V(rn,rm) r1.,r2……rN.> = 1/2 V(rn,rm) r1.,r2……rN.>
So, this is my representation of the two body operator written. So, as it acts on a basis
state that is given by this. Now we can take a a two body interaction term in vacuum as
this is equal to a half and again written as a
V=1/2 (r) (r’) V(r,r’) (r) (r’)
So, that is the that is the form of the potential or a 2 body potential this see that it
contains two creation operators and two annihilation operators as opposed to the one
body term which contain one creation operator and one annihilation operator.
Now, this can be written as in the general sense and in any basis just like what we have
done it for the one body operator we will write the two body operator as. So, this is
written in terms of scattering events I will describe how and this is equal to lambda
52
lambda prime and a mu prime and then this is O mu, mu prime lambda lambda prime
and there is a mu dagger a mu prime dagger a lambda a lambda prime.
So, this is the matrix element of the 2 body operator between 2 states mu, mu prime and
lambda lambda prime. So, what happens is that a state form lambda and so, one particle
from a state lambda and another particle from the state lambda prime they scatter on to 2
states mu and mu prime. So, these are annihilating lambda and lambda prime are
annihilating and they are getting created in the state as in the mu and mu prime states.
So, this is the matrix element of the 2 body operator in the mu, mu prime and lambda
lambda prime states and these are the corresponding creation and annihilation operators.
So, it could happen that mu equal to mu prime or mu equal to lambda and mu prime
equal to lambda prime all that can happen, but in principle this can be denoted as
scattering event. So, I will draw that scattering event a pictorial representation of that
scattering event can be like this and there is some interaction there and it could be going
like this. So, there is a lambda coming. So, a particle from a state lambda is coming a
particle from a state lambda prime is coming and they are getting created in mu and mu
prime, it could also happen that the lambda goes to mu prime and lambda prime goes to
mu, but that is equally a possibility for this scattering problems this.
So, this is a simplest vertex that we can think of or there could be more complicated
vertices which take away you know some energy momentum which we will learn when
we are doing going to do a Feynman diagrams. So, in those cases we will see that we
have a vertex having a structure and that structure will depend upon the specifics of the
interaction term that could actually depend upon the interaction term can actually depend
upon a. So, here the interaction amplitude of the interaction term is constant, here, it
could depend upon r or it could depend upon spin or it could depend upon momentum
and then we have them getting recreated here. So, we have a lambda and the lambda
prime and they are getting recreated as mu and a mu prime.
So, these are the basic representations of the 2 particle operators. So, now, I will
summarize that; we now know how to write down a many particle state in terms of
creation and annihilation operators and we also know how to write down one particle
operator and 2 particle operator in the second quantized notation.
53
So, once again I reiterate that the name second quantization is justified as we have been
able to write down both the state and the operators that represent observables in the
quantized notation or what we call as second quantization where both of these quantities
are represented as operators the creation and the annihilation operators and to end the
discussion the proper symmetry requirements of fermions or bosons are encoded in the
anti commutation and the commutation relations respectively in these problems. So, we
shall wind up this discussion with an example which is for the 2 body spin-spin
interaction.
That is we are going to talk about a Hamiltonian where the only interaction that is there
is a via 2 spins and they would be interacting in a particular way and this is a in a spin
only model and that we would be writing it as a half factor to avoid double counting and
then there is a
V=1/2 (r) (r’) (r) (r’)
. So, this is a typical example of a 2 body interaction for a spinner system where 2 spins
are interacting via potential which is given like this, we will get into more examples
later.
54
Prof. Saurabh Basu
Lecture – 06
Tight Binding Hamiltonian, Hubbard model
So, we shall carry on with the applications of this second quantized formalism that we
have learnt so far. And this time we are going to apply it to real materials or rather the
models which actually mimicries materials. So, in that context we will look at tight
binding approximation today.
So, I am assuming that you have done tight binding approximation in your solid state
physics course when you wanted to calculate the electronic structure or the electronic
band structure of metals and in that context you have learned that we assume that in the
vicinity of each lattice point, the full periodic crystal Hamiltonian, which is given by H.
So, H is the full Hamiltonian is approximated by H1 which is an atomic Hamiltonian and
localized at a lattice site.
So, the approximation is about full periodic Hamiltonian, which we assume that it is now
a localized atomic Hamiltonian which is localized or rather only confined to a lattice site.
So, the overlap of the atomic wave functions is still enough for us to consider that there
is we are talking about a real material and not talking about isolated atoms or ions, but it
55
is not so much that the independent electron or rather the isolated atom information is not
completely dumped.
So, it is just the balance between that we are saying that the wave function is just
restricted to a site, and it is minimally overlapping with the wave function to a
neighbouring site, but; however, we are still acknowledging there is some very minimal
overlap so, that the isolated atomic description is not completely overthrown, but the
presence of other atoms are being acknowledged at the same time. And this
approximation is useful in describing energy bands that is that are relevant to two
partially filled d-shells of transition metal atoms and describing the electronic structure
of insulators.
So, this Hamiltonian within the tight binding approximation is what we are going to see
now. So, we have tightly bound electrons to the iron cores and we are pretty much this is
called as the atomic limit and this word is used to denote that the electrons are typically
bound to the iron cores. To write down the Hamiltonian in a formal manner, let us
introduce the wave function of the electrons in this atomic limit.
As a psi of R n will tell you what n is R is of course, a site or lattice site where an atom
or an iron is located and n is actually an orbital which we are going to come to and this
has a normalization factor like this, and then you have an exponential minus i k dot R
and in the momentum space it is written as this.
56
So, let us put a comma in between and there is a K that is only restricted to the first
Brillouin zone. I mean when I say B Z. So, B Z for me stands for Brillouin zone and we
say the mostly we say the first Brillouin zone. So, we use this notation that the BZ
actually talks about the first Brillouin zone.
So, this is the wave function of the electron which is localized at a given iron core or at a
given lattice site and so, there is a free part which is given by exponential minus i k dot R
and there is also this psi k n, where n is as I said is the orbital index, and R is a location
of the lattice. And you can write also the psi k n and just by taking the Fourier transform
and which will come with a simple.
So, psi k n is written as 1 by root over N and this sum over R exponential i k dot R and
psi R n. So, thus for this system which comprises of vanishingly weak inter atomic
overlap, this are called as the Wannier basis. And these are localized written the wave
functions are written in the localized Wannier basis and now this for an interacting
Hamiltonian the Wannier basis is certainly not the Eigen basis of the problem, but;
however, we can still they are still helpful and let us give you a picture physical picture
that what we are trying to talk about.
So, this is my energy axis and this is what I say that it is my interaction is in this along
this axis. And when we have this each of these orbital’s what happens is that, at 0
interaction these are discrete energy values. So, this one corresponds to say n equal to 1
and the other corresponds to say this n equal to 1 and these are n equal to 2 and so on,
but; however, when the interaction becomes larger than some value these start you know
branching out like this, and they spread out like this and these form the energy bands and
again I assume that you have done the first course of solid state physics in which the
formation of energy bands are taught.
So, beyond a certain interaction potential between the energy levels we would have these
individual orbitals from transform from being single particle discrete single particle
states, and to they will fan out or they will branch out like this and for being energy
bands.
Now, they will be called as insulators or conductors depending on the position of the
Fermi level and let us say. So, as I told this is the energy axis. So, let us say the Fermi
level is somewhere here and let us call that as EF and then if the Fermi level falls in the
57
spectrum of the available energy states within that spectrum, and then it is called as a
metal and otherwise it is called as an insulator. So,.
So, how do we actually exploit this Wannier basis to write down the ore to form the
Eigen states of the general Hamiltonian the full periodic Hamiltonian? So, two things
that need to be kept in mind one is that that the Wannier states represent orthonormal
basis of single particle states. The word orthonormal means it is both orthogonal and
normalized and it is also we can see that one can actually drop the band index because
only one band index or orbital index as we have said band index n is important is
important as we have shown yet only n equal to one is important. So, we can drop the
index n and drop n ok.
+
So, now allow us to introduce a transformation of the form 𝑎𝜎+ (𝒓) = ∑𝑘 𝜓𝑹∗ (r) 𝑎𝑘𝜎
+ ∗ +
or we can write the Fourier transform of this as 𝑎𝑘𝜎 (𝑟) = ∑𝑘 𝜓𝑹𝒊 (r) 𝑎𝑘𝜎 . So, this is how
the operators are defined in terms of the Wannier functions look these psi’s are the
Wannier functions that we have just spoken and we have dropped the band index and as
we introduce new operators the creation operators in terms of the Wannier basis.
+ 1
And a momentum space representations is equally possible which is given by 𝑎𝑘𝜎 = √𝑁
+ + 1 +
∑𝑖 𝑒 𝑖𝑘.𝑅𝑖 𝑎𝑖𝜎 and similarly the Fourier transform of that is 𝑎𝑖𝜎 = ∑𝑘 𝑒 −𝑖𝑘.𝑅𝑖 𝑎𝑘𝜎
√𝑁
58
So, this transformation can be used to write down the generic single particle Hamiltonian
+ 1 +
as H0=∑𝑘 ∈𝑘 𝑎𝑘𝜎 𝑎𝑘𝜎 = ∑𝑖,𝑖′ ∑𝑘 𝑒 𝑖𝑘.(𝑅𝑖−𝑅𝑗) ∈𝑘 𝑎𝑖𝜎 𝑎𝑖′𝜎
𝑁
+
=∑𝑖,𝑖′ 𝑎𝑖𝜎 𝑡𝑖𝑖′ 𝑎𝑖′𝜎
So, this 𝑡𝑖𝑖′ is equal to 1 over n sum over k again k belonging to the first Brillouin zone
1
and this is equal to 𝑡𝑖𝑖′ = 𝑁 ∑𝑘 𝑒 𝑖𝑘.(𝑅𝑖−𝑅𝑗) ∈𝑘 and so, let us take a particular. So, this is
the formula for the or rather the Hamiltonian and the t i i prime represents this, and let us
see the example of a square lattice.
And what is the square lattice you have these are the positions of the lattice points where
which is occupied by the atoms or the ion cores. So, this is my x direction and this is my
y direction and let us say that this is called which is called as a lattice spacing which is of
a and then this direction is called as the x cap direction and this is as the y cap direction.
So, we have for a particular site we have four neighbours. So, an electron actually can
hop in a lattice on a on a two dimensional square lattice from this point let us call this
point as o or let us call this point as x or will be indistinguishable in this matrix of the
lattice points that are shown here.
So, from this lattice point x, it can go to x 1 which is on the positive x axis towards the
positive x axis it can go to x 2 which is in the negative x axis, and it can go to y 1 which
is in the positive y direction and it can also go to y 2 which is in the negative y direction.
59
So, the neighbours for in a square lattice is given by plus a x cap and a minus a x cap and
a plus a y cap as I said a is called as a lattice constant and there is also a minus a y cap.
So, these are the four neighbours of a lattice site for a 2D square lattice two dimensional
square lattice ok. So, we so, to write down the tight binding model we need an
information about the neighbours, which are the most dominant contributor to the
Hamiltonian. So, then I can write down this epsilon k which is the single particle energy
spectrum as written in this slide here.
So, this epsilon k can be written as a 1 over n again taking a Fourier transform and there
is. So, it is x cap and y cap and this will be like exponential minus i k and then it will be
all the let us call these neighbours by a name let us call them as delta. So, it is dot delta
and there is a sum over in you can write it x cap y cap, but let us write it as delta sum
over delta and this delta are these all these neighbours and this is equal to.
So, this is simply I need to find the neighbours ∈𝑘 = 1/𝑁 ∑ 𝑒−𝒊𝒌.𝜹 , where k is just a
generic vector in two dimension which is k x x cap plus a k y y cap. So, this will give me
a 1/N{[𝑒−𝒊𝒌𝒙𝒂 + 𝑒𝒊𝒌𝒙𝒂 ]+[ 𝑒 −𝒊𝒌𝒚𝒂 + 𝑒 𝒊𝒌𝒚𝒂 ]} by. So, this is will be simply equal to.
So, I should have ok. So, these are simple to see you should verify them. So, k y a and
exponential i k y a. So, this is nothing, but
60
∈𝑘 = −2𝑡(cos(𝑘𝑥 𝑎) + cos(𝑘𝑦 𝑎))
and hence my epsilon k becomes equal to with an amplitude that is given by that as
minus p and then there is a sum. So, because this will take care of this 1 over n and I will
have.
So, this 1 over n will not be there if I do the sum and I have a minus 2 t cosine k x a and
a cosine k y a as the dispersion or the energy momentum relation for an electron a tight
binding electron in a square lattice. Now this formalism is very important because this as
I said that in a variety of real materials this tight binding approximation is very helpful in
finding the e versus k relation and this can be plotted on a two dimensional k x k y plane
and one can find that that how the dispersion looks like over the first Brillouin zone what
I mean by first Brillouin zone is that for k x to be.
So, the first Brillouin zone of a square lattice is also a square lattice and it looks like this
which is a. So, this is my k y and this is my k x. So, these are points which are pi and pi
here. So, that we have a this point to b like pi pi and this point is like pi minus pi and this
point is like minus pi minus pi and this point is minus pi pi and this is 0 0 and this point
is called as pi 0. They have specific names we are not going into that you can look at any
book on electronic structures and they have names such as m k gamma etcetera points
and. So, this is my first Brillouin zone which I need to do all these sums that we have
talked about. So, k x is between minus pi to pi and same for y.
61
So, now we can we know a lot of information about this square lattice or the energy
dispersion tight binding dispersion in square lattice, we can calculate the electron
velocity by taking a derivative with respect to k either k x or k y and then we can
calculate the velocity of the electrons at one of those given points pi pi pi 0 0 0 etcetera
this is called as a zone centres 0 0 is called as a zone centre and pi pi is called as a zone
corner.
So, we can calculate at thus these special points we can calculate the velocity and we can
also find out the effective mass of electrons by which is which comes out from the
inverse of the second derivative of this of this energy dispersion or with respect to k x or
k y. Now so, this is also widely applicable to other and let me write in bracket as
complicated lattice geometries and by complicated we mean other kinds of lattice
geometries such as the triangular lattice, the Honeycomb lattice, Kagome lattice and so
on and one of the famous examples is given by Graphene.
And this Graphene was discovered in 2004 by two scientists called Novoselov and Gaim
and this was their from Manchester United kingdom and they were awarded Nobel Prize
in 2010.
So, this is the citation for the Nobel Prize that they have got this was in 2010 from the
from Sweden and then this is the structure of grapheme, it is one atom thick honeycomb
structure. So, it is the best 2 D material that you can find in nature and it is just like a thin
62
sheet absolutely one atomic layer thin sheet of carbon atoms, which are which are there
arranged in the hexagonal form or the honeycomb form and to find out the Tight binding
dispersion of electrons in graphene we note this figures and this gives the unit vectors of
graphene. So, graphene is a it has two atoms per unit cell. So, one atom is by the filled
circle and the other is the filled and the open circle with the filled circle inside, these are
the two atoms per unit cell and these are the basis vectors that one can see.
So, they are basically the carbon atoms are arranged in the hexagonal lattice as we have
shown, and they are connected by very strong covalent bonds of the sigma orbitals and
they derive from the s p 2 hybridization of these atomic orbital’s and the remaining p z
orbital has a weak overlap and forms a narrow band of the pi orbital states. One should
look up the structure of graphene more carefully and understand that to the first
approximation these pi electrons they can be represented or denoted by a tight binding
Hamiltonian with a hopping energy minus t between the neighbouring atoms.
And to describe the the spectrum energy spectrum of the system and one actually can
think of a unit cell consisting of two atoms as I told earlier, enclosed in a in this forming
𝑎 𝑎
this honeycomb structure they are given by say 𝒂1 = (√3,1) 2 𝒂2 = (√3, −1) 2 a is
called as a lattice constant.
It has a value that is known which is equal to 2.46 angstrom and so, the tight binding
model for this though slightly more complicated than the square lattice which is what we
63
have written earlier because of the presence of two kinds of atoms per unit cell, we can
still write down in simple enough fashion as there is a r and r prime. So, and this is equal
to H= ∑𝑘,𝜎(𝑡𝑎1+ (𝒓)𝑎2 (𝐫 ′ ) + h. c) you will often find that in the tight binding
Hamiltonian written in literature it is written with a Hermitian conjugate to make the
Hamiltonian real or rather to give the Hamiltonian is still is.
So, the energy spectrum is real. So, that is why this Hermitian conjugate is added and
this sum is overall the nearest neighbour basis vectors of the unit cell that is shown there
this is a problem that is given to you to solve I am only outlining the solution of this
problem and this Hamiltonian can now be written in the momentum space as rather.
0 −𝑡𝑓(𝑘) 𝑎
+
So, this is equal H= ∑𝑘,𝜎 (𝑎1𝜎 +
𝑎1𝜎 )( ) ( 1𝜎 )
−𝑘𝑓 ∗ (𝑘) 0 𝑎2𝜎
This is the form of the Hamiltonian and it has simply has to be diagonalized in order to
find the energy Eigen values it is just a 2 by 2 matrix that one has to diagonalize 𝑓(𝑘) =
1+ e-ik1a + 𝑒 𝑖(−𝑘1 +𝑘2)𝑎 , where k vector is written as k 1 having two components k 1
and k 2 because we are talking about a two dimensional system and finally, you should
be able to write.
The dispersion as
∈𝑘 = ±𝑡|𝑓(𝑘)| =±√(3 + 2 cos(𝑘𝑥 𝑎) + 2 cos(𝑘𝑥 − 𝑘𝑦 ) 𝑎 + 2cos(𝑘𝑦 𝑎))
So, that is the dispersion and this of course, because of the plus minus sign it has two
bands and let me write down this together there. So, this is the dispersion and there is an
upper band and there is a lower band upper band is given by the positive sign and then
64
there is a lower band that is given by the negative sign and in fact, you should try to plot
it on a do a three dimensional plot of epsilon k being in the z axis, and a k x and a k y say
in the. So, let this be k x and the k y and if you plot epsilon k, you will see some nice
structures which I leave it as an example or rather a problem that needs to be solved. And
you will find that these two bands actually touch at 6 points in the Brillouin zone and
near those 6 points the energy dispersion is not parabolic it is rather it is linear.
So, these linear dispersions are called as the Dirac cones and because linear dispersion is
found in relativistic particles such as a photon, and the energy dispersion close to those
points where they touch they represent or rather resemble that of massless electrons and
that is why they are called Mass less Dirac fermions. So, most of it we leave it as an
exercise for you to solve let us go to the interaction term and write that interaction term
in some simple form for the two particle interaction.
So, interacting Hamiltonian; a bit of a summary if you like that we have done the second
quantized formalism, we have introduced the creation and the annihilation operators and
we have written down generic Hamiltonian which includes single particle Hamiltonian
and a Hamiltonian which consists of two particle interactions. As we have said earlier
that there could be higher body interactions but usually they are too weak and most of
the time they are unsolvable and the two particle interactions are good enough for
describing most of the systems that we come across, and from there on we have done the
single particle Hamiltonian and that Hamiltonian has only a single particle term which
65
can be solved it is a kinetic energy mostly it is a kinetic energy part of the Hamiltonian,
there could be also an onsite potential which it may include.
And we have also now seen today it is about the tight binding Hamiltonian, let us now
talk about the interaction Hamiltonian, again we will write it in a generic fashion. So, it
+ +
is an electron electron interaction and we have Ve.e = ∑𝑖,𝑖′ ,𝑗,𝑗′ 𝑈𝑖𝑖′ 𝑗𝑗′ 𝑎𝑖𝜎 𝑎𝑗𝜎 𝑎𝑖′𝜎′ 𝑎𝑗′𝜎′
So, once again just to make the notation clear that there are two particles interacting via
some interaction strength 𝑈𝑖𝑖 ′ 𝑗𝑗′ and these two particles are at i and i prime and they
have spin σi and σi ′ and after the scatter.
So, interaction is a type of scattering which we are going to see elaborately later and then
they go on to sites which are j and j prime, and maybe the sigma becomes sigma prime
and sigma prime becomes sigma or sigma retains as sigma and sigma prime become
retains is retained as sigma prime. This is a generic Hamiltonian and if you like the the
amplitude of the term is written as this
𝑈𝑖𝑖′ 𝑗𝑗′ = ½ ∫ 𝑑𝑑 𝑟 ∫ 𝑑d r’ 𝜓∗𝑅𝑖′ 𝜓∗𝑅𝑖 V(r-r’) 𝜓𝑅𝑗′ 𝜓𝑅𝑗
. So, maybe we have not kept that same notation let us just make it a little more careful.
So, we have and let us not keep this and so, we write this as R i and R i prime and this is
r j and r j prime ok. So, that and then the full Hamiltonian looks
+ +
H= ∑𝑖𝑖′(𝑎+
𝑖𝜎 𝑡𝑖𝑖′ 𝑎 ′ ) +∑𝑖,𝑖′ ,𝑗,𝑗′ 𝑈𝑖𝑖′ 𝑗𝑗′ 𝑎𝑖𝜎 𝑎𝑗𝜎 𝑎𝑖′𝜎′ 𝑎𝑗′𝜎′
𝑖𝜎
So, that is the interaction term we have put in and this is the full Hamiltonian that one
usually has to solve and most of the time this Hamiltonian has no solution that we have,
and there are there could be many approximations in which we can write this and we
have U we can
𝑈𝑖𝑗′ 𝑖′𝑗′ = V𝑖𝑖′ and or it could be simply you know all these.
66
So
𝑈𝑖𝑗′ 𝑖′𝑗′ = V𝑖𝑖′
𝑈𝑖𝑗 𝑖′𝑗′ = V
and these are some of these things that we want to discuss now.
So, let us now discuss specific types of interactions and call these as maybe 1 and maybe
this is 2, let us see specifically what one does. So, when we have U i j i prime and j
prime equal to depends only on two indices that is 𝑉𝑖𝑖′ as it is shown in the above
equation, then I have Hint= ∑𝑖≠𝑖′ 𝑉𝑖𝑖′ 𝑛𝑖𝜎 𝑛𝑖′𝜎′
+
is the interaction term and this is the interaction term that we have for. So, 𝑛𝑖𝜎 =𝑎𝑖𝜎 𝑎𝑖𝜎
and this is let us call it as H interaction and this is the form. So, this is going to give me
a charge density wave type of modulation.
So, because you have a density which is 𝑛𝑖𝜎 at a given site and in another site it is going
to be probably in the neighbouring site it is going to be n i prime sigma prime. So, one is
going to have a charge density wave, and this may lead to charge density wave
instabilities which those physics we will discuss later. Now in the atomic limit what can
happen. So, let us define what is the atomic limit and atomic limits means the hopping
which is denoted by t that is set equal to 0 or it tends to 0 becomes vanishingly small and
the atoms are well separated with the overlap between the neighbouring orbitals to be
minimal if in this limit.
67
So, the atomic limit just to repeat it again it talks about the overlap of the atomic orbitals
being minimum and the resultant interaction is called as the in that case the interaction is
known as Hubbard interaction and this was by john Hubbard, who first wrote down this
Hamiltonian way back in the 60s 1960s and in which case we have for this as we have
written it for the second one that is or the interaction is completely independent of the
indices i j and i prime j prime and it is equal to a constant and then we can write down
this interaction.
+ +
So, let us write it fresh to be ∑𝑖,𝜎,𝜎′ 𝑈 𝑎𝑖𝜎 𝑎𝑖𝜎 𝑎𝑖′𝜎′ 𝑎𝑖𝜎′ = ∑𝑖 𝑈 𝑛𝑖↑ 𝑛𝑖↓ - U/2 ni
. Now this last term is just a constant because it acts like a chemical potential which acts
on a single site or rather it is a single particle operator, which will just shift the energy up
or down and hence we can neglect that term.
So, the neglecting this term which we understand is going to shift the energy by a
constant amount and as if it is just acts like a chemical potential. So, we will drop that
+
and we will write the Hubbard Hamiltonian as H= ∑<𝑖𝑗>,𝜎 𝑎𝑖𝜎 𝑡𝑖𝑖 ′ 𝑎𝑖 ′ 𝜎 + 𝑈 ∑𝑖 𝑛𝑖↑ 𝑛𝑖↓
So, this is known as the Hubbard Hamiltonian and in fact, even though it is as I said that
it is discovered in the sixties of the last century, there is still physics which are not
understood and is often thought as the you know the research focus for many groups and
it is still under active you know research. And especially it was under focus immediately
after the discovery of the high temperature superconductors, in which it thought to away
68
from half filling it thought to give rise to the correct physics and it is been debated since
then in a number of variety of occasions.
So, well stop this discussion at this moment and let us go to another discussion in which
we will talk about maybe let us talk about a magnetic Hamiltonian.
And see that how this magnetic Hamiltonian gives rise to magnetic properties of
materials, and by a magnetic Hamiltonian what I mean is that it is written as
H=J∑<𝑖𝑗> 𝑺𝑖 . 𝑺𝑗 and where maybe it is just for the moment I assume that, J is greater
than 0 and the sum is over nearest neighbour sites and this is why it is called a magnetic
Hamiltonian because there is no Fermions or there is no kinetic energy of the Fermions it
is basically two frozen spins are interacting via a strength J and so, the ground state
properties of this model at low temperature has a long range anti ferromagnetic order and
this long range anti ferromagnetic order is called as an nil state.
So, what I mean by long range anti ferromagnetic order is that, the spins are alternately
pointing up and pointing down and there is a long range order in the system which means
that the spins at i and spins at some distant j are linked or related and so, this the long
range order is technically defined by a correlation function of S i and S j when i i and j
are large. And the signature of a long range order is given by this correlation to decay in
some exponential fashion as a function of the distance.
So, this long range order is achieved below a certain temperature, which is a
characteristic temperature for antiferromagnets and is called as an nil temperature and an
effective theory can be written down in terms of the partition function and hence
69
calculating the susceptibility and so on. It is difficult to solve this Hamiltonian in higher
dimensions. In one dimension we can hope to solve this Hamiltonian with certain
approximations and this is what I am going to show you now that how this Hamiltonian
can be solved in one dimension and that to a restricted version of the Hamiltonian and
how it can be solved using some transformations. So, let us write down the Hamiltonian
in one dimension and that is equal to H= J∑𝑖 𝑺𝑖 . 𝑺𝑖+1 .
So, every spin at site i is interacting with it is neighbour at i plus 1 and these are assumed
to be all spin half particles; that means, these values of s is are either up or down which
we can represent by plus 1 or minus 1. And this model was exactly solved by Bethe and
then onwards by many others and it is known that there is no anti ferromagnetic long
range order for this model, and the spin correlations have a power law that is the spins at
site a and spins at site j where i and j are distant sites in the lattice, it does not have
exponentially decaying correlations rather it has a power law of correlations and at the
wave vector given by k equal to pi and. So, there are these physics which has already
been known since the exact solution by Bethe, we are not attempting that solution rather
we would take a simpler model which is called as the quantum x y model.
So, we will do a S equal to half quantum x y model in 1-dimension. And let us write
down this Hamiltonian which is the starting point of our discussion now and this is equal
to H= J∑𝑖 𝑺𝑖𝑥 . 𝑺𝑖+1𝑥 + J∑𝑖 𝑺𝑖𝑦 . 𝑺𝑖+1𝑦 .
So, the interaction for the spin components are only taken to be in the x and the y
directions; however, the system is a one dimensional chain of spins there is no Sz
component or the interaction between the s z components are neglected and that is why
this name is x y model. So, this is the Hamiltonian that we want to solve and see that
what it gives us.
70
So, if you define operators which are 𝑆±

𝑖 = 𝑆𝑖𝑥 + i𝑆𝑖𝑦 . So, I these are called raising and
lowering operators.
So, I am making this or rather constructing this raising and lowering operators from the x
and y components of the spin at a given site, and there is a plus and a minus sign are
related to the plus and the minus which means S i plus equal to S i x plus i S y and
similarly for the minus sign. So, if I use these operators to write down the Hamiltonian,
this you should do it and convince yourself. So, the Hamiltonian H= J/2 ∑𝑖 𝑆+ −
𝑖 𝑆𝑖+1 +
+
𝑆−𝑖 𝑆𝑖+1 . As I said you should check that that how this Hamiltonian written in let us calls
it as equation 1, let us call this as equation 2 and let us call this transformation as
equation 3 and equation 4. So, you should see that how equation 2 becomes equation 4
via this transformation.
Now, if you write further that these a size are simply 𝑺𝒊 = ½ 𝝈𝒊 where sigma is are the
Pauli matrices sigma i vector is a matrix or rather it is a it is equal to sigma x i, sigma y i
and sigma z i. So, this is the definition of sigma i there is a h cross which has been taken
as 1 here, otherwise you could have written it as h cross over 2 it really does not matter
in this case even if we take h process 1 it is and it is a normal practice to take h cross
equal to 1 for this particular cases. And in which case your sigma i plus minus becomes
equal to your S i plus minus ok. So, you have a size now the spin vectors are defined in
terms of the Pauli spin matrices and the raising and the lowering of the Pauli spin
matrices are nothing, but the raising and the lowering of these spin operators s operators.
So, now what we can using this we want to write down the same Hamiltonian in 4 in
terms of the Pauli matrices and that is written as H= J/2 ∑𝑖 𝜎+ −
𝑖 𝜎𝑖+1 + ℎ. 𝑐
71
+
. So, the Hermitian conjugate will be 𝜎𝑖− and a 𝜎𝑖+1 . So, the Hilbert space is two
dimensional, Hilbert space is two dimensional and it is consists of spin up and spin down
fermions.
Now it is important to realize there are no fermions so far in the Hamiltonian 5 however,
these spin matrices they correspond to fermions and these are the spin half objects
correspond to fermionic particles and that is why these up and down spins they
correspond to fermionic operators or rather fermions and so, the only thing that you
should keep in mind is that the sigma is are not fermion operators at the sigma i actually
at different sides they commute and not anti commute as the fermion operators would
have done.
So, if there is any hope to transform this Hamiltonian into a Fermionic Hamiltonian we
have to do a careful transformation and which is what we are going to do. So, this
Hamiltonian is written in terms of Pauli speed matrices and we shall transform this
Hamiltonian to a Fermionic Hamiltonian and how do we do that let us define
𝑧 𝑧
𝐶+
𝑖 = (∏𝑗<1 𝜎 𝑗) 𝜎+
𝑖 𝐶𝑖 = (∏𝑗<1 𝜎 𝑗) 𝜎−
𝑖
So, these are the transformations that we do for from the Pauli spin matrices to the
fermionic operators. And now writing down a transformation does not guarantee that
these are fermionic operators unless they satisfy the anti commutation relations that we
have spoken about. So, we have to check that whether they correspond to the or rather
they satisfy the anti commutation relations the c operators. So, one can check.
72
So, please check this that

𝐶+
𝑖 𝐶𝑖 = σi
+σ
i
−
and this is true because you are the product of

𝑧 𝑧
∏𝑗′ ,𝑗<1 𝜎 𝑗 𝜎 𝑗′ =1
1
So, if that is the case then we can write this down as a 4
(𝜎𝑥 + 𝑖𝜎𝑦 ) ( 𝜎𝑥 − 𝑖𝜎𝑦 ). So,
sigma i will be half of sigma x plus I sigma y and sigma i minus will be half of sigma x
1
minus i sigma y, and this if you expand it becomes equal to 4 [2- i[𝜎𝑖𝑥 , 𝜎𝑖𝑦 ]] and this is
1 1
equal to a 4
{ 2- 2𝜎𝑖𝑧} which is nothing, but equal 2 (1 − 𝜎𝑖𝑧 ) so,
𝜎𝑖𝑧 =1- 2 𝐶𝑖+ 𝐶𝑖
So, if I do the reverse transformation. So, my

𝜎𝑖𝑧 =1- 2𝐶𝑖+ 𝐶𝑖 .
Now it is known that 𝐶𝑖+ 𝐶𝑖 is equal to either 0 or 1 for fermions at a given site that is a
restriction on the number of occupancy of the number of fermions at a given site going to
the Pauli’s exclusion principle.
Now, you see that if 𝐶𝑖+ 𝐶𝑖 is equal to 0, then my 𝜎𝑖𝑧 equal to 1. So, and if 𝐶𝑖+ 𝐶𝑖 equal
to 1 then this is equal to 𝜎𝑖𝑧 equal to minus 1. So, 𝜎𝑖𝑧 can take value 𝜎𝑖𝑧 can take values
plus 1 and minus 1 for n i to be 0 or 1. So, we can write 𝜎𝑖𝑧 equal to minus 1 whole to the
power C i daggers C i. Now this is a clever way of writing if C i dagger C i equal to 1
then sigma i z is equal to equal to minus 1 if C i dagger C i equal to 1, then it is equal ton
73
minus 1 and if it is equal to 0 then it is equal to plus 1. So, that is the that is the thing that
we have decided or rather we have gotten from this earlier step. So, my sigma i z can be
written in terms of the fermionic operators as this all right.
So, if we have this for the 𝜎𝑖𝑧 then we have

+
𝜎𝑖+ = ∏(−1)𝐶𝑖 𝐶𝑖
𝐶𝑖+
𝑗<1
. So, this is in the exponent. So, let me write it little neatly here ok.
So, that is my that is my sigma i raising or the plus 1 and for the
+
𝜎𝑖− = ∏(−1)𝐶𝑖 𝐶𝑖
𝐶𝑖
𝑗<1
So, you have to convince yourself that this is the right transformations between the
raising and the lowering spin matrices Pauli spin matrices with the fermionic operators
and it has to be also checked that {𝐶𝑖 , 𝐶𝑖+ } =1.
So, the fermionic anti commutation relations should give you 1 at a given site and this
means that 𝐶𝑖 𝐶𝑖+ plus a 𝐶𝑖+ 𝐶𝑖 which should give you 1 and that can be checked from
the fact that. So, check this I am almost doing it for you, but you should also do it
+ +
yourself to check that so, C i. So, {𝐶𝑖 , 𝐶𝑖+ } ={ 𝜎𝑖− , 𝜎𝑖+1
+
}/ ∏𝑗<1(−1)𝐶𝑖 𝐶𝑖
(−1)𝐶𝑗′ 𝐶𝑗′
+ +
So, it is which is ∏𝑗<1(−1)𝐶𝑖 𝐶𝑖
(−1)𝐶𝑗′ 𝐶𝑗′ and this is the definition of the anti
commutation relations in terms of this Pauli matrices and let us only take this the
74
numerator. So, the numerator is of the right hand side, which is sigma i plus minus with a
sigma i plus and that can be written as.
{(𝜎𝑖𝑥 + 𝑖𝜎𝑖𝑦 ),( 𝜎𝑖𝑥 − 𝑖𝜎𝑖𝑦 ) } = 𝜎𝑖− 𝜎𝑖+ − 𝜎𝑖+ 𝜎𝑖− = 2(1 − 𝜎 𝑧 ) + 2(1 + 𝜎 𝑧 ).
So, this is equal to 4 and interestingly this denominator which contains this term the
∏ (−1)𝑛𝑗 (−1)𝑛𝑗 ′
𝑗,𝑗′<1
. So, it gives rise to four possibilities for n j to be 0 and 1 and n j prime to be 0 and 1. So,
if you look at this then n j equal to 0 and n j prime equal to 0, then this term that the
above term gives 1 and similarly for n j equal to 1 n j prime equal to 1 again the above
term gives 1 and for n j equal to 1 and n j prime equal to 0 again the above term gives 1,
and n j equal to 0 and n j prime equal to 1 the above term again gives 1. So, these four
ones will actually cancel with the four that you are seeing here and I will finally, get the
commutation relations that.
75
We want that is {𝐶𝑖 , 𝐶𝑖+ } =1. So, the Hamiltonian takes a form which is equal to which
is as we have written it down earlier. So, we will simply write down one of the terms
which are 𝜎𝑖+ 𝜎𝑖+1
−
and now this thing would be written in terms of the fermionic
+
operators and the fermionic operators are a product for∏𝑗<1(−1)𝐶𝑖 𝐶𝑖 , I want to write
down the fermionic operators explicitly we could have written it n j as well.
+ +
So, there is a (∏𝑗<1(−1)𝐶𝑖 𝐶𝑖
𝐶𝑖+ )( ∏𝑗<𝑖+1(−1)𝐶𝑖 𝐶𝑖
𝐶𝑖+1). So, this is the term if you
recognize that this term is there in the Hamiltonian as written in equation let us call this
as equation. So, this is an equation 5. So, so let us call this equation as equation 6 and we
have checked. So, this is one term in the Hamiltonian that is written there.
+
And this if you simplify it becomes equal to 𝐶𝑖+ (−1)𝐶𝑖 𝐶𝑖
𝐶𝑖+1 . So, this is what comes
and this can be rewritten as 𝐶𝑖+ (1 − 2 𝐶𝑖+ 𝐶𝑖 )𝐶𝑖+1 as you know that this is equal to
either plus 1 or minus 1 which is written by these (1 − 2 𝐶𝑖+ 𝐶𝑖 )
because C i dagger C i can be either 0 or 1 and this can be hence written as C i dagger C
i plus 1.
So, now it becomes a fully Fermionic Hamiltonian and the Hermitian conjugate as we
have seen in equation 5 will come out as
𝐶𝑖+ 𝐶𝑖+1 .
So, we will write down the full Hamiltonian completely now in fermionic terms with a
H= J/2 ∑𝑖 𝐶𝑖+ 𝐶𝑖+1 + h.c
76
And so, this is the called as the tight binding Hamiltonian which we have already seen
earlier. So, tight binding Hamiltonian now we do not have any spin index so, these are
for spin less fermions in one dimension so, spin less fermions in 1 D. So, this is up the
problem that it boils down to starting from the Hamiltonian one as we have seen it here
or rather to which is a simplified version of one and written in one dimension.
So, this is the spineless fermions and there is a it can be exactly solvable. So, exact
solution exists.
And why I write is that, it is quite unusual for an interacting problem or of many body
Hamiltonian to have an exact solution, even if it is the case in one dimension. So, we will
solve it using those. So, you rerun the whole thing in momentum space. So, that we can
1
write down a 𝐶𝑖 = ∑𝑘 𝑒𝑖𝑘𝑥𝑖 𝐶𝑘
√N
So, k belongs to you know mind I mean in this range rather.
So, minus pi to pi which is the first Brillouin zone and if you do this momentum space
transformation one gets that H= J/2N ∑𝑖,𝑘,𝑘′ 𝑒 −𝑖𝑘𝑥𝑖+𝑖𝑘𝑥𝑖+1 𝐶𝑘+ 𝐶𝑘′ +h.c that is going
to be there and this is equal to J/2N ∑𝑖,𝑘,𝑘′ 𝑒 −𝑖𝑥𝑖(𝑘−𝑘′) 𝐶𝑘+ 𝐶𝑘′ +h.c as will be there
and this is is basically the lattice spacing and this is.
77
So, this is equal to you are the definition along with this 1 over n is the definition of delta
k k prime. So, this is becoming it becomes equal to J and there is a. So, this will give me
a J ∑𝑘 cos(𝑘𝑎) 𝐶𝑘+ 𝐶𝑘 .
So, this is simply J ∑𝑘 cos(𝑘𝑎) 𝑛𝑘 . If you like take a equal to 1 the lattice spacing equal
to 1 that does not cause any problem, but this is the exact solution of the Hamiltonian
that we have started with equation 2 and what does this Hamiltonian tell you? That this
tells me that the spectrum is like this and these are these points.
So, this is the Fermi energy so, 0. So, this is my E versus K let me write it neatly this is
equal to your epsilon and this is K and these are the points. So, this is that corner of the
Brillouin zone. So, minus pi to plus pi and this is it cuts the Fermi energy at this is at pi
over 2.
So, we have the Fermi points are at minus pi over 2 and plus pi over 2 so, the ground
state. So, in the ground state of the Hamiltonian one gets Fermi it should be in three
dimension, it should be Fermi surfaces; however, in one dimension these surfaces
actually boil down to point. So, we will write as Fermi points at k equal to plus minus pi
by 2. So, what it tells you is that there are gap less excitations so, which means that
another fermion or one extra fermion can be added without any cost of energy at the
Fermi level.
78
So, when you want to add a particle at in a system in a system of fermions because all
the states all the way up to 0 to k by 2 are filled which are inside this parabola they are
completely filled all the way up to k by 2 plus k by 2 and minus k by 2. So, if you want
to introduce one Hamiltonian then sorry if you want to introduce one fermion that
fermion has to be added at the Fermi level, and it costs no energy because the excitations
are gapless; however,. So, it is known that the quantum x y model gives gapless
excitations that is one can what I mean is that one can add one more fermion without any
extra cost of energy.
However this situation will change if we add the z term. So, if we have a term which is z
and z plus 1, this inclusion of this term of this term makes the spectrum gaped ok. So, we
are really talking about a Magnetic Hamiltonian in one dimension which has an exact
solution and this solution seemed to give us a magnetic metal or a kind of gapless
excitations and in the event that you want to add a z component of the spin interaction
along with the already the terms that appear in equation 2, this gapless situation will
vanish and you will have a gapped situation which would be corresponding to an
insulating scenario.
79
Prof. Saurabh Basu
Indian Institute of Technology Guwahati
Lecture – 07
Magnetism
So, we have learned magnetism, magnetic metal in the previous discussion in which we
have seen that how spin only modeled in one dimension actually can be solved exactly in
give rise to a magnetic metal. If you include the spin interaction in Z direction if the gap
will open up and then will have a magnetic insulator depending upon what is the
magnitude of the gap. So, let us now, look at more carefully magnetism.
And we do it as an example of the on going discussion, but it also helps us to understand

magnetism in particular and we will have going to talk about ferromagnetism to be more
precise.
So, let us talk about ferromagnetism where by ferromagnetism what we mean is that the
spins are all pointing in same direction. So, when the spins all of the all the spins are
pointing in the same direction we get a magnetization of the system. So, there is a bulk
magnetization that exists and this is put into use in various applications of magnets. So,
the first thing that comes to our mind is that the magnetization verses temperature curve.
80
So, magnetization verses temperature and as I told that we are going to talk about
ferromagnetism in particular.
So, that looks like, the magnetization is in the y axis and this is the temperature in the x
axis, so it is transition which is like this for external magnetic field equal to 0 which we
denote by H and this point is called as Tc. So, when the bulk magnetization vanishes and
the system loses its complete magnetization and it becomes a para-magnet and the same
plot in magnetic external magnetic field in presence of external magnetic field looks like
this. So, this is of the order of typically Tcs of the order of for some materials which will
see such as nickel its of the order of it could vary from something like few Kelvin may
be around 40 Kelvin to about may be even 1000 Kelvin and so 1000 Kelvin, if we do it
more carefully.
And will list out various values of Tc. So, what happens is that at Tc the magnetization
vanishes and the ferromagnet converts into or there is a transformation face
transformation into a paramagnet where which results in random alignments of fence and
we have shown to graphs, one is in the presence of external magnetic field which is a
outer curve and the inner curve is a with the magnetic field equal to 0. So, now, what
happens is at small values of field the magnetization is proportional to the field. So, this
is the external field and we can write proportionality relation with this where chi is called
as the susceptibility, and H as we said is the external magnetic field.
So, this is valued for low fields off course we have non-linear phenomenon which
involves higher order of the external field where the magnetization ceases to be linearly
depending on the field that very large fields, but we will only stay within this domain
where the magnetization is proportional to the field and we can talk about it relation
linear relation between magnetization and the magnetic field to be having this kind of
relation. So, now, let us look at how the susceptibility which is an important measure of
magnetization how that behaves with temperature.
81
And so, susceptibility verses temperature. And we can have 1 over chi plot verses T
which simply looks like a line like this a straight line like this which has the slope or
rather it can be represented by formula which is like χ=CC/T −θ where T is the
temperature and theta is some characteristic temperature which we are going to list out in
a while. And also a very close to the transition that is a ferromagnet to the paramagnet
transition we have the T actually or rather than chi actually depends on the temperature
in an interesting way and which is often called as a universal behavior and which can be
understood that if you look at the log of chi which is base 10 its chi verses temperature
and then it looks like straight line again.
−γ
( T −θ )
And this tells that the χ= ( TC ) So, this is the straight line if we simply you know
erase this that it looks like a straight line and, so it is a straight line like this and where
gamma is some coefficient or rather some number which is an indicator of the
universality class of the transition and so at large temperature rather, at large temperature
the susceptibility which is chi obeys a law called as Curie’s law.
And so this can be written as what we are shown in the, so this is written as some it is, so
chi is equals to C divided by T minus theta that tells us that 1 over chi goes as T minus
theta divided by C. And also in the vicinity of Tc in the vicinity of the transition which
occurs at a temperature T equal to Tc the chi actually goes as T minus Tc divided by T C
82
whole to the power minus gamma and the and this is will tell you as I told that it tells us
about the universality class of the transition. So, let us list out a few values for known
ferromagnet. So, we have let us, there are materials and there is a Tc and this
characteristic temperature theta.
So, if you list out few known materials of which shows ferromagnetism which is iron
which is about 1041, theta is 1093 nickel which is another known ferromagnet which is
has an Tc of 631 Kelvin, these are all in Kelvin. And this is the 650 Kelvin and a
gadolinium has a 293 Tc, 293 Kelvin which is almost room temperature and slightly
higher value of theta. And a chromium Cr Br 3 so this has a Tc of which is very low
which is 37 Kelvin and there is a no data may be available for C r B r 3 for the value of
theta.
So, these are some ferromagnets some known ferromagnets. So, before we go to the
many body theory of understanding ferromagnets let us see how we can understand it in
a simpler language of ferromagnetism and let us do models of ferromagnetism and write
down a simple model and try to solve it if possible.
83
So, in presence of a magnetic fields let us call it as B. So, this magnetic field is offer
interchangeably used by H and by B, B is usually called as magnetic induction and H is
called as magnetic field, but; however, sometimes a difference is not spelt out very
explicitly. And one can write down H=g μ B ⃗

B∑⃗
Si − ∑ ¿ first term is that a
i ¿ij>¿ J ij ⃗
Si . ⃗
Sj¿
coupling between an external magnetic field B. So, this is the B is the magnetic field, a
external magnetic field and S i are the spin vectors all right.
So, and off course we have to define other things also g is, it is it is called as the Lande g
factor and mu B is the Bohr or it is the magnetic moment, it is the magnetic moment and
Jij is a nearest neighbors spin spin coupling. So, this is Hamiltonian which this system of
non interact or the system of spins which are placed in a external magnetic field. The
first term denotes how an individual spin couples to magnetic field and the second term
is that there is an interaction between 2 spins which are at neighboring sides i and j. So,
when I write this form that is i and j in angular brackets they mean that i and j are nearest
neighbors. So, there is a nearest neighbor spin interaction the amplitude or the magnitude
of the spin spin interaction of the coupling is given by J i j which could be a constant for
all i and j which may not be a constant for pairs of any pairs of i and j.
This model is quite complicated to solve in 3 dimensions and often there are now,
solutions exact solutions unless you are ready to make some approximations. And let us
solve the scalar version of the model. So, this is what we are going to do and scalar
84
version of the model and which is we already know that is called as the ising model and
so that ising model looks like, so my spin vector has only can only assume 2 orientations
which is up and down that is called as the ising model. So, my Hamiltonian now,
⃗ ⃗
becomes H=g μ B B ∑ Siz − 2 J ∑ ¿ but now, it is scalar form of the spin that are
i ij>¿ ⃗
z
Si . ⃗
z
S j¿
considered.
And we have discussed this before that in order to avoid the double counting we have put
this factor and this is the Hamiltonian that we need to solve.
And we define the magnetization as let us call it with m i which is equal to S i Z S i z.
Now, with this notation my Hamiltonian an H=g μ B B ∑

i
m i − 2 J ∑ ¿ this is written
ij>¿ m . m ¿ i j
( g μB ) J
n
as H=
2
B∑ σi − ∑ σ . σ , where α i=2 mi. So, basically these are the alpha i S
2 i=1 i ( i+1)
i
will take values which are either plus 1 or minus 1 and if you notice that this sum or
rather the second term contains a Z N by, so Z and N, N by 2.
So, these are the total number of spins that is there and if we add a term which is alpha
N, N plus 1 we have to understand that alpha N plus 1 is same as alpha 1. So, the N plus
1 is spin is made as the same the as the first spin. So, that we have an open chain and so
on and this open chain now, has been put into a closed form.
85
So, the last one and the first one are connected and we have a closed system such that my
alpha N plus 1 equal to alpha 1 and this is called as the periodic boundary condition and
this periodic boundary condition is enforced on the problem in order to solve it. So, its
periodic boundary condition and this is the open chain. So, the open chain has been a
converted into a periodic or rather close chain by applying a periodic boundary
condition. Now, how to solve this Hamiltonian? One way of solving is the Hamiltonian
is to write down the partition function and see that if the partition function has any closed
form available.
( − β En ( α i ))
So, the partition function is written as Z=∑ e . Now, I will repeat for convenience
αj
that this is the canonical partition function which means that it is partition function for a
system which is in a contact with the heat bath at temperature T. And this is the form of
the partition function where the energies or which is we could have written H as well, so
( − β H ( αi ) )
we this is the same as what we mean by this is actually alpha i and e Z=∑ e and
ni
as you saw it here that H is a function of alpha i. So, this H is function of alpha i and so
we are to exponentiate it multiplied with beta and then take the negative sign here and
then sum over all alpha i S that are possible.
But look at this simplicity of this formula that alpha i S can only take values which are
plus 1 and minus 1 because of the fact that the S i S or the spins are only the Z
86
components of the spins and Z components take a value which is the Z component of the
spins take values which are plus 1 and minus 1.
So, even it looks like a complicated formula to solve or rather summation to compute it
is not so will see that here and we will use a technique which is called as the transfer
matrix technique to solve this. So, in order to do understand ferromagnetism from very
simple considerations we are written now, a general Hamiltonian where the spins are
interacting with the neighboring spins and they are in presence of a magnetic field and
there is a Zeeman coupling as it is called as a B dot S coupling at each site B is uniform
magnetic field external magnetic field is uniform.
And that problem has very complicated solution and in 3 dimensions it is not possible to
solve unless you do some approximations you can solve with numerically by doing exact
diagonalization of finite number of spins, but that is not the focus here focus is to
actually solve this model by using analytic techniques.
Then that Hamiltonian was converted to a one dimensional Hamiltonian which is like an
ising modeling in presence of a magnetic field, and we are trying to solve the ising model
in presence of a magnetic field and in one dimension. And this requires us to compute
the canonical partition function which is written as this. So, this Z can simply be written
as alpha i equal to plus minus 1 and these are there are N of those, so this is alpha. So,
this alpha 1 and there is alpha 2 equal to plus minus 1 and then
Z=∑ ∑ ........ ∑ K ( α 1 , α 2) K ( α 2 ,α 3 ) ........... K ( α N ,α 1 )

α α2 αN
So, the advantage of this expression is that each of those keys will see what these keys
are, each of these keys involve 2 spins at a time. So, this term involves the first and the
second spin where for alpha equal to 1 alpha 1 equal to plus 1 and minus 1 alpha 2 equal
to again can be plus 1 and minus 1, similarly for alpha 2 and alpha 3 and then alpha 3
and alpha 4 and continuing all the way for N spins and this term is equal to alpha N alpha
1. So, each of these k alpha 1 alpha 2 you should work it out carefully from this formula
that formula for the partition function that is written over here. So, this is equal to a
minus beta g mu B B. So, I should write here the beta equal to inverse of temperature or
its 1 over K T, where K B as its written here is called as the Boltzmann constant which
has the value which is often you will find it any book.
87
So, this divided by 2 and alpha 1 plus alpha 2 divided by 2 and plus a beta j by 2 alpha 1
alpha 2 and this is the form of each one of those K expressions that are written inside the
sum. So, and this can be a summed over each one of those case can be summed over for
values of alpha 1 and alpha 2 or alpha 2 and alpha 3 or alpha 3 and alpha 4 each having
value plus 1 and minus 1. So, this requires us to solve a matrix problem, where a this k
actually looks like matrix written like this, where this is for alpha 1 equal to 1 alpha 2
equal to 1 and this is alpha 1 equal to minus 1 and alpha 2 equal to minus 1. In this basis
e (− x+ a) e( −a )
one can write this as .
e− a e ( x+a )
(g μB)
x=
(2 kB T )
So, where my .
J
a=
(2 kB T )
So, what is shown here is the following that each of those K’s written in the alpha basis
which are, there is one term for which alpha equal to 1 and alpha equal to alpha 1 equal
to 1 and alpha 2 equal to 1. So, that is is given by; so this should be more carefully
written as, we have a 1 1 and that is equal to. So, this is the 1 1 element and alpha 2 equal
to minus 1 here. So, we have to write it as if we have these things. So, there are 4
combinations alpha 1 equal to 1, alpha 2 equal to 1 alpha 1 equal to 1 alpha 2 equal to
88
minus 1 alpha 1 equal to minus 1 alpha 2 equal to 1 and alpha the other term is alpha 1
equal to minus 1 and alpha 2 equal to minus 1.
So, if you write it that basis without writing it like this then it looks like each of these
K’s will look like. So, one term is this, this term is the first term. So, this is the first term
and there is another term which is alpha 1 equal to 1 alpha 2 equal to minus 1. So, which
tells that this term cancels out because one of the term is plus 1 and the other is minus 1;
however, this becomes exponential minus beta J by 2 which is written by a, I will show
you that.
So, this terms are first terms when one of them is plus 1 and the other is minus 1 and this
one of them is, this one for alpha 1 equal to 1 alpha 2 equal to 1. This is alpha 1 equal to
1 alpha 2 equal to minus 1 this term is alpha 1 equal to minus 1 alpha 2 equal to 1 and
this term is for alpha 1 equal to minus 1 alpha 2 equal to minus 1. So, this is these are the
4 terms. So, do not. So, this is written here you should do it and convince yourself to get
matrix of this kind for these 4 a unique combinations.
And this is the matrix that we have to solve and. So, x is equal to g mu B by 2 K T and a
equal to J over 2 K T and so we have the full Z the total partition function can be written
as the trace of K and N to the power N because if you see this is not only true for alpha 1
alpha 2 it is true for alpha 2 alpha 3, it is also true for alpha 3 alpha 4 is true for alpha 5
and alpha 6 and so on and alpha N and alpha 1. So, we have similar K matrices coming
and so the entire thing entire partition function will be the product of all these matrices
and since we are doing a sum it will convert into a trace. So, trace of those K’s raise to
the power N for N spins and this is the form of the partition function.
Now, should understand that all the internal spins or all the internals alphas that is what
are the internal alphas alpha 2, alpha 3, alpha 4 , alpha 5, and alpha N minus 1 there are
all summed over they are all summed over. And we have the only alpha 1 and alpha 2
sorry alpha 1 and alpha N are not summed over or rather they remain and that is what
give raise to this formula. So, it is a trace of that and because it is a 2 by 2 matrix the key
has 2 Eigen values, and those Eigen values let them be lambda 1 and lambda 2. So, my Z
since it is equal to trace of K to the power N.
So, this is same as . So, this is the general thing that you should keep in mind because I
am taking a tress of N 2 by 2 matrices. So, finally, the this is going to be 2 by 2 matrix
89
and those will have Z=Tr ( K N )=λ N1 + λ2N each one will have 2 Eigen values lambda 1 and
N
λ2
N N N N
lambda 2. So, the whole thing will come as Z=Tr ( K )=λ 1 + λ2 = λ1 1+
( ( ))
λ1
Now, you can see that if my lambda 1 is a bigger Eigen value this term goes to 0 for N
going to be very large, because if this is the leading Eigen value. Lambda 1 is a leading
Eigen value lambda 2 by lambda 2 goes to smaller with each power of N that is the spins
and when N goes to very large and left with only. So, if my lambda 1 is greater than
lambda 2 among the 2 Eigen values my Z becomes only 1 and I need to solve only one of
the Eigen values of this matrix which is shown here, and that the bigger one and then I
raise it to the power N. So, I can one can easily do that and one gets a
( − 4 a ) ( 1 /2 )
λ 1,2=e ∗ ( coshx ± ( sinh e
a 2x
) )and that is it.
So, these are the lambda 1 and lambda 2 where the lambda 1 corresponds to the plus sign
and the lambda 2 corresponds to the minus sign and we can a simply take lambda 1
which is equal to exponential a and this term plus this term that is what is going to be
used in calculating the partition function. So, what we do after calculating the partition
function?
90
We calculate the free energy and the free energy is calculated using
F=− K B T lnZ=− Nk B T ln λ1 . It is because we are decided to take the larger Eigen value
it is the log of lambda 1 and the magnetization can be derived from the free energy and
which is equal to minus del F del B off course in the limit B going to 0 that is what the
definition says.
( g μB B )
−(∂ F )
So, will have M = ( ∂ B ) =lim
μ
Ng B
( sinh
( 2 kB T ) )
2 ( 1 /2 )
❑
( g μ B B ) − 2 J /k
(( sinh 2
(2 k B T )
+e
) B t
)
So, this is has to be written with. So, this whole to the power half and let us put that is
the form for the magnetization. And interestingly it can be seen that M goes to 0 as B
goes to 0 ok.
So, if there is no external field one dimensional such spin model does not show any
magnetization this is called as the spontaneous magnetization. So, there is no
spontaneous magnetization of a 1D ising model if we only put the external magnetic
field then there will be magnetization there and for B going to 0. So, we have not still
done the B going to 0 limit. So, which if we do that my magnetization looks like
( N g2 μ2B ∗ e J/ k T )
[ ]
B
M= B
(4kBT)
So, that is the form for the magnetization as a function of a temperature and also. So, this
is that chi which one can rather this is not the chi because you have a temperature which
(∂ M )
is I mean. So, this is basically the chi because the chi is nothing, χ = and this is
(∂ B )
eJ / k T
b
equal to χ= and apart from this factor. So, this is equal to,
T
1
=T e − J/ T =T ( 1 − J /k B T +............... ). So, I have a expanded the exponential and kept
χ
only the first order term and this is equal to T − J /k B .
91
C
Now, you see that this is very similar to χ = . So, our intention was to derive the
T −θ
curies law from a Hamiltonian from a calculation and this gives the derivation of the
curies law where the 1 over the chi looks like T minus some characteristic temperature
which is J over K Boltzmann’s constant. And if you simplify this then chi looks like
formally looks like as if some constant divided by T minus theta theta is some
characteristic temperature and which is here in this case its J over K. So, this is the
simple discussion of ferromagnetism in model Hamiltonian which arises out of spin spin
interaction.
And once again point it out that this model or has no spontaneous magnetization which
means that as you switch off the external field the magnetization would vanish. At small
a field the magnetization looks like this edges shown here and the susceptibility looks
like it has a form which is like this we have dropped this the initial factor here just kept
only the T part, and associated the energy or rather the which depends on J over K is has
the characteristic temperature which is theta and then we have derived how chi depends
on temperature which is simply a 1 over T minus theta where theta is some characteristic
temperature. So, this is called as Curie’s law.
Now, let us see the same thing or rather ferromagnetism from Hubbard model.
92
And in particular we are going to derive what is called as Stoner criteria all right. So,
what was the interaction term? So, will write down the Hamiltonian the Hubbard
+¿
Hamiltonian H=− t ∑ C(i σ ) C ( j σ )+U ∑ n i↑ ni ↓ ¿ .
ij>, σ i
So, this is called as the onsite interaction between the electronics spins and since
exclusion principle prohibits up spins to be present the same site, but there could be a up
and down spins present and which would have interaction of this for. It is very clear that
if U is switched off that is which is called as a atomic limit will have a metal given by
the only a kinetic energy term. And if you use a much larger than T will have an
insulating term which means that the electrons will not be able to hop from one side to
another and the spectrum will be a large gap opening up due to U if U is much greater
than T ok.
So, this as we said that this has no exact solution in 3D and we are even in 2D we are
going to make some approximations and this approximation is called as the mean field
approximation or in this particular case it is called as the Hartree Fock approximation I
will tell you what it is. So, in order to solve it will have to use this approximation which
is also called as the Hartree Fock approximation. We will see this more elaborately when
we talk about greens function which we are going to start very soon; Hartree Fock
approximation.
And suppose you are sitting in class and you have friends who are sitting right beside
you and there are some friends who are sitting very far away from you you cannot
communicate in any way while the class is going on, but however, the people who are
sitting right in the vicinity that is in your just near neighborhood, they can actually speak
to you or they can point out they can communicate in some form with you. This is a
really a many body phenomenon because you have some interaction with the people who
are right at the vicinity and you have a lesser interaction with people who are far off and
even lesser interaction with the people who are farther off. However, if I make an
approximation that you are sitting in the class and the rest of your friends who all the
other friends other than you can be considered as as if you are in an external or an
effective field because of all the students present all the friends present in the class.
93
That is I am not making any distinction between the friends who is right in neighborhood
and the friend who is farther off, I am treating both of them to be equally or giving
equally importance to them and saying that you are freely and average field due to all
your friends in the class. So, if you want to go beyond that want to make a better
approximation then you would say that somebody who is in the neighborhood has has
more interaction with you than someone who is at the farther of positions and there could
be fluctuations or there could be you know corrections to these average field picture or
the mean field picture.
So, we will go with the mean field model where you are considered as if you are facing
or rather feeling an effective field from all of your friends. So, we would particularly
concentrate on one spin in the system and this that spin as if that spin is facing an
effective field from all other spins in the system. So, to do that what we need to do is that
we can split this term let us write it. So, I will write this
U ∑ ni ↑ ni ↓ → U ∑ ❑<ni ↑ >ni ↓+ U ∑ ❑ ni ↑< ni↓ >¿

i i i
So, you see here the the up the downs spins are facing an average field which is U into
the average up spins density. So, as soon as I take an average or an expectation of this
between the see the state of the system of the ground state of the system this becomes a
number.
So, this is the field that each of those spins down spins will be freely and similarly
without making any bias towards the up spins the up spins will be facing this field. So,
this is the average field or the mean field that each of those spins will be facing or
feeling. Now, understand that my n i up one kind of spins is much much greater than the
other kind of spins because of the ferromagnetism. So, this is saying that one species out
numbers the others and this is the essence of ferromagnetism. So, you know that in a
ferromagnet we have predominantly or majority spins are pointing in the in one
particular direction. So, in this model we do not have to have external magnetic field to
be present and now let us simply define magnetization.
We will not define bulk magnetization rather we will define this the what is called as the
sub lattice magnetization which is at a given site m i this is equal to n i up minus n i
down OK. And, so the magnetization is same at all sites because of the that all the up
94
spins are facing a potential which is U into the sum of all the down spin densities. So,
how do we calculate this magnetization? This magnetization m is same as m which is for
ϵF ϵF
all sites. So, m=∫ dE N ↑ ( E ) −∫ dE N ↓ ( E ) . Let us write this as you can more clearly write
ϵA ϵB
this is epsilon A, I will tell you what epsilon A is, some energy and to epsilon F and dE
and N down E.
So, we are simply computing the average up spin density which is the the density of
states of the up spin and integrated over energies from epsilon A to epsilon F and I will
just tell you what epsilon A is and minus the down spin density integrated over epsilon B
2 epsilon F.
So, this model here as given rise to a band picture which looks like this. So, there is an
up spin band. So, this is the down spin band and there is an up spin band like this and my
Fermi energy is here.
So, this is my Fermi energy and my; so this is the down spin this is the up spin and this is
my Fermi energy is and my epsilon A is right here or epsilon up if you want. So, that is
the bottom of the band for up spin and this is the bottom of the band for down spin. Now,
you understand that if I have to know what is my total up spins or average number of up
spins then I have to take the density of states of up spins and integrate those from here
95
this point which is known as epsilon A to epsilon F that is I will have to integrate over all
these place where the up spin is there and in order to do it for the down spin I will
integrate it from epsilon B to epsilon F which is all these down spin.
So, I will try to. So, the magnetization will be the shaded area on the right minus shaded
area on the left. And so what is my; so if I take a middle point here. So, this is my delta
let us call it as delta I will define what delta is and this is my minus delta. So, this is also
delta, but this is this is for delta equal to 0, there is a reference line this line is reference
line and you have delta. So, from that reference line all the way are up to the Fermi
energy the gap is known as delta and from there again the reference line to all the way to
the epsilon A which is a bottom of the up spin is called as delta.
ϵF ϵF
So, if you have this nomenclature then my m=∫ dE N ( E+ Δ ) −∫ dE N ( E − Δ ). So, the

−Δ Δ
density of the states is at energy E plus delta minus a delta to epsilon F. So, that is this
point marked by say let us call is as P. So, the from P to epsilon F, so that is my the
second one which is dE N E minus delta. So, this is my N up E as written earlier and this
is my N down E you know that N up is much greater than N down which is by the
magnetization is large.
And so now, since E is greater than delta E is much greater than delta because delta is the
gap. So, we can do a tailor expansion of the density of the states and each of those
N ( E+ Δ )= ( dEdn ) Δand N ( E − Δ )=( dEdn ) ( − Δ ). And then I can write down skipping one
step which you should fill up which is equal to 2 delta where I have not defined delta.
So, delta is equal to delta equal to by 2. So, M equal to 2 delta and I have delta to E f d N
dE and a dE and now, this delta is much smaller than epsilon F. So, we are talking about
epsilon F which are of the order of few electron volt delta is very small.
So, epsilon F B is much greater than delta we can approximate delta to be equal to 0. So,
we can or rather change this the lower limit of the integration to 0 and then this m
Ck
becomes equal to M =2 Δ ∫
Δ
( dEdn ) dE=2 Δ N ( ϵ )=m u N ( ϵ )
F F
96
So, this will tell you since magnetization is not equal to 0, we get a condition for
1
ferromagnetism to occur which is equal to N ( ϵ F )= . So, this is called as the Stoner
U
criteria. So, let us see what Stoner criterion is.
And so this tells the ferromagnetism will occur in a system in which if I have m verses U
which is the electron electron interaction, see the difference between this problem and
the problem that we have done earlier that is writing down Hamiltonian in a spin only
Hamiltonian and solving it via computing the partition function is that that model did not
have any electron, if it is a only a spin model. Now, we are having an electronic model.
So, there are electrons which can hop from one place to another which is a more realistic
model because the spins have to be carried by somebody, spins are not they do not exits
on their own.
So, the electrons carry their spin. So, at spin only or rather the electronic model is more
reliable and more realistic than a spin only model. So, here a Hubbard model is a
electronic model. So, this electronic model is we are trying to find how ferromagnetism
can be derived from the from this electronic model and in process we have found out a
criterion called as the Stoner criterion. And which tells you that the density of states for
the electrons at the Fermi level has to be inverse of this electron electron interaction
which is the very surprising condition that we have found and. So, so ferromagnetism,
97
for ferromagnetism to occur we need a critical a critical electronic coupling, a critical
electron electron coupling is required which is you see some critical coupling, so that the
1
N ( ϵ f )= .
(U C )
So, it is inbuilt in the problem what is the density of the states at the Fermi level. So, the
electrons contribution to the density of the states of the Fermi level is decided by the
system itself. However, that has to match with the critical electron electron coupling for
ferromagnetism to occur. So, that gives you a more realistic picture about
ferromagnetism and what happens is that, this is like this. So, you have a U =C U C here.
So, below that there is a paramagnet and above that this is a ferromagnet and we have
this as the; so this is the Stoner criterion. So, in order for ferromagnet to happen
U N ( ϵ F )> 1.
And this is a very stringent condition that is the electron electron interaction multiplied
by the density of the states of the Fermi level will have to exceed unity that is the value
has to be more than 1 in order to have ferromagnetism. And is quite a stranger criterion
and iron, nickel and cobalt satisfy well this criteria. So, this is called Stoner criterion.
So, to summarize the discussion that we had just now, is that we wanted to understand
magnetism in particular ferromagnetism we have written down a spin only model and
have solved it in one dimension and with ising kind of interactions. And we found that it
gives a ferromagnetism where it is same as the Curie’s law and it gives you characteristic
temperature for the ferromagnetic transition which is given by theta or Tc which we have
called as Tc at times we have called theta at times.
However, if we revisit the same problem from an electronic model then we get a Stoner
criterion and Stoner criterion tells that the electron electron interaction multiplied by the
density of the states at the Fermi level has to exceed 1 for ferromagnetism to occur.
98
Prof. Saurabh Basu
Lecture - 08
Singlet and Triplet States: Magnetic Hamiltonian
Let us now discuss how one gets a magnetic Hamiltonian involving only spins. We are
particularly talking about Ising kind of Hamiltonians or Heisenberg kind of
Hamiltonians. Mainly we would be talking about ising Hamiltonians where the spin can
only have either pointing up or down these are the two possible orientations and let us
see that how we can derive Hamiltonian which we have introduced or rather we have
talked about when we spoke on magnetism during our lectures.
So, to be want to study magnetic Hamiltonian and a derivation of a magnetic

Hamiltonian would require that we know the addition of spins or; so say addition of 2
spins, 2 and these are spin half particles. So, we have 2 spin half particles and we would
see that how one actually at the spin vectors.
So, let us consider 2 particles with spin vectors S 1 and S 2. The total angular momentum
I mean what I mean by angular momentum is that the total spin angular momentum is
99
S=S 1 +S 2, , where S 1 and S 2s are the spin vectors for the 2 particles that we are
considering. So, just to remind you that both are spin half..
Now, the direct product space is that consists of it is a 4 dimensions. So, the direct
product space is 4-dimensional and we can use the bases use S m s basis for each. So,
what I mean by that is that the Eigen value for the spin operator S is has Eigen value S
and S z has Eigen value m s. So, we can form the basis of each of the particles by this S
m s and the total space will be product of such 2 such S m s that is S 1. So, total product
total space is S 1 m s 1 and S 2 m s 2.
±ℏ
And let the since we have for each one of them, so ms = . So, let us represent the
2
states by up and down. So, each of these half h will correspond to say a up and this
minus up will correspond to minus half h cross. So, this h cross over 2 and this is minus
h cross over 2
And hence will have we can write it in 2 ways. So, the space the spin space or the direct
product space is either you call it alpha 1. So, may be this is called as the alpha and this
is called as the beta or so its α 1 α 2 which means both are in up spin states. α 1 β2 means
one of them in up spin other in down spin state and α 2 β1 the first one is in the down
and the second one is in up or both of them are in the down. And this is one option.
100
Whereas, the other option is that we can write it as up up as the states up down and down
up and down down ok. So, this is other option and we can simply choose one of them,
but let us choose this option in order to write the wave function for the or and. I mean to
discuss this problem of 2 spins. So, what is the total value of M s =m s1 +m s2 =1,0,0 ,−1 ,
1 when they both add up half plus half and this is when half minus half is minus half and
this minus half minus half. And the total spins quantum number S=S 1 +S 2 which can
take value 0 and 1 ok. So, for S equal to 0, for S equal to 0 we have they just one Eigen
function and that Eigen function let us write it with the form which is
1
χ 00= ↑↓−↓↑
√2
So, this is this called as the singlet wave function and this is anti symmetric, what I mean
by anti symmetric is the following. But, you have 2 particles, so the first one is in upstate
and the second one is in down state here the second one the first one is in down state
second one was in is in upstate and now, if you interchange up down one gets a negative
sign. So, this is that is why it is called as the anti symmetric and for S equal to 1 we
would need. So, for S equal to 1 will have 3 combinations because will have to take care
1
of χ 11 =↑↑ state χ 10= ↑↓−↓↑ and a χ 10=↓↓
√2
Now, all these are called as triplet and triplet states because there are 3 in number and
one can easily check that they are symmetric because if the first particle is written swap
to the second particle where the function remains the same. So, these are the states or the
wave functions for 2 particles both spin half and our for a system consisting of or
comprising of 2 spinner particles and all possible combinations have been taken. We get
4 states and those 4 states are 1 singlet and 3 triplet us states the singlet state is the anti
symmetric with respect to the change in the position of the particle and the triplet states
are symmetric with respect to the change in the position of the particle.
So, these are the states, but what about the Eigen values, because in order to solve a full
quantum mechanical problem we both we need both the information on the Eigen values
and the Eigen functions. So, let us see that all these 3 states. So, now, we will talk about
the Eigen value these states are |↑↑>,|↓↑>,|↑↓>,|↓↓> ¿ ¿ and. So, this forms the
basis of the problem of a 2 particle problem, 2 particle spin half system.
101
So, they are Eigen states of S 1 square S 2 square S 1 z and S 2 z. So, the total spin total
spin S can be 0 or 1 ok. So, now, we can see the how these total spin operators act on
those each of these states. So, the total spin operator which is S z which is equal to S 1 z
plus S 2 z that acting this state acting on or let us write it here as well. So,
S z |↑ ↑>=(S 1 z + S 2 z )|↑ ↑>= ℏ |↑↑>
So, I will get a h cross by 2 and up up. So, as we have said that these are Eigen states of
these operators. So, I get a Eigen value equation which is S z acting on a up up state
gives me h cross by 2 as the Eigen value returns me the up up state as well.
Similarly, for S z|↑↓>= 0 because S 1 z will give me a plus h cross by 2 and S z 2 will
give me a minus h cross by 2. And similarly will also have S z acting on the down up
state should also give 0 and S z|↓↓>=−ℏ|↓↓>¿ ¿ sorry this is the minus h cross by 2
for a each of they should be simply h cross. So, this is for each one of them is an h cross
by 2. So, there is a there are 2 h cross by 2 which make a h cross. So, S z on down down
will give me a minus h cross and so on.
So, these are the Eigen values of these S z operator and what about; so further more we
2
S 2 =( ⃗S1 + ⃗
have ⃗ S 2) = ⃗ S 1 . ⃗S2 ;
S 2 + ⃗S2 +2 ⃗ S 1 and S 2 will commute with each other
1
because they are they pertain to different particles. So, S 1 square will be h cross. So, it is
102
half into half plus 1, half into half plus 1 this acting on. So, S square acting on any of
these states so, say we will talk about up up say for example.
So, this is equal to half into half plus 1, sorry if S into S plus 1. So, that is this and then
1 1 1 1
S 2|↑↑>= ℏ2
⃗
2 2 ( )
+1 + ℏ2
2 2 ( )
+1 + 2 ⃗S1 . ⃗
S 2 . So, we will leave it for the moment and let
S1 . ⃗
us see that what we can do for the ⃗ S 2 . So, ⃗
S1 . ⃗
S 2 if you see it is equal to S
S 2 =S1x S x2 +S1y S 2y +S z1 S z2 . Now, if you introduce this ladder operators for the spins. So,
S1 . ⃗
⃗
S plus can be written as S + =S x +iS y ,S− =S x−iS y .
1 z z
S 2= ( S +1 S−
S1 . ⃗
Now, this will give me ⃗ − +
2 +S1 S2 ) +S1 S 2
2
And hence what we can do is that we can see that
3 3
⃗
4 [
S 2|↑↑>= ℏ 2 + ℏ 2 +2
4
ℏ
2 ( )]
|↑↑>=2 ℏ2|↑↑> . Now, we have a 2 S 1 dot S 2. Now,
for the up up state this will raise the spin and hence it will be 0 because the up is the
maximally align state and though S 2 minus can give you a nonzero contribution, but S 1
plus will give you 0 and similarly S 2 plus will give 0 and that is why these 2 terms do
not contribute and that simplifies the problem and then we are left with S 1 z S 2 z for
which we know the operation. So, that is why we have done this
103
And this is 2 into h cross by 2 and this whole thing and this whole thing multiplied or
rather acted upon by this. So, it is a Eigen value equation and this is if you simplify it
would becomes equal to 2 h cross square up up and so on. And similarly for the down
down as well one gets a same answer by doing the same technique one gets this as. So,
S 2|↓↓>=2 ℏ2|↓↓> and so they have, these states up up and down
on these acting on ⃗
down have total spin S equal to 1 and m S plus minus h cross ok. But of course, S equal
to 1 should have 3 states which are equal to m S plus minus h and 0.
So, the third states, m S equal to plus minus h is there. So, m S equal to 0 state is
obtained by a particular operation. So, by the application of S minus on up up state; Let
us see how one gets it or you can also consider or S plus on the down down states. So,
S− |↑↑>= ( S−1 +S−2 )|↑↓>=ℏ ¿ ¿
Now, this is something that you should have done in quantum mechanics this gives me
an Eigen value which is these are not Eigen states of up up, but I will operate on this and
give me this S 1 will give me a h cross and will give me up down sorry it will be a down
up the first one will load. So, it is a down up plus an up down S 2 will lower the other
1 1
one with an Eigen value which is given by h cross. So, ℏ S−|↑↑>= ¿ ¿ does not
√2
matter we have written down the second term ahead of the first term and this will
correspond to S z equal to 0.
So, these 3 will be called as the triplet states and.
104
So, the singlet states are of course which corresponds to, so these are the triplet states.
So, the 2 that is coming over here with spin S equal to 1 and m s equal to one which is
here and the other one comes from here. So, these are the 3 states and now, will just look
at the singlet state which corresponds to S equal to 0 m s equal to 0 let just call it as we
1
can call it as the | χ 00 >=|00 >= ¿¿ .
2
3 2 ℏ 2 2
So, why is it a singlet state? So, S z|00 >=
[ 2
ℏ −2()
2 ]
−ℏ |0,0 >=0|0,0> . So, which
means that m S value of this is equal to 0 and this has S equal to 0. So, we have found
out all the 4 Eigen states of this of this 2 particle problem.
So, let us now, look at the spin Hamiltonian consisting of the these if you want to
construct a Hamiltonian only consisting of the these 2 spins which is like a as I said like
a ising Hamiltonian or a Heisenberg Hamiltonian if S has a full rotational symmetry. So,
let us just discuss the construction of a magnetic Hamiltonian. So, we have, we have
2
S 2 =( ⃗S1 + ⃗
⃗ S 2) = ⃗ S 1 . ⃗S2 .
S 2 + ⃗S2 +2 ⃗ Now, the Eigen value of, Eigen value of
1
3
S⃗2 = ℏ 2 + 2 S⃗1 . S⃗2
2
105
S1 . ⃗
So, for the singlet state that is S equal to 0 will have to put S equal to 0 the S 1, ⃗ S2
−3 2
has an Eigen value which is equal to minus half, ℏ because this is equal to 0 if you
4
1 −3 2
S1 . ⃗
put the right hand is equal to 0 the ⃗ S 2 will have a Eigen value which is
2 2
ℏ ( )
−3 2
which is ℏ .
4
Whereas, for the triplet state which corresponds to S =1, so that will have
3
1 ( 1+1 ) ℏ 2= ℏ2 + 2 ⃗S1 . ⃗S2 . So, this is equal to two. So,
(2 ℏ − 32 ℏ ) =⃗S . ⃗S
2 2
for the
1 2
2 2
triplet state.
1 2 1
So, this is equal to 2 minus 3 half just half. So, this is equal to ℏ . So, ℏ2 is the
4 4
S1 . ⃗
Eigen value in short e value I am writing, for the operator ⃗ S 2 in the for a for a 2
S1 . ⃗
particle problem. So, they just summarize this quick result. So, for singlet states ⃗ S2 ,
106
−3 2 1
so this is singlet and triplet. So, this singlet one has ℏ and this is ℏ2 . So, this is
4 4
S1 . ⃗
the Eigen value of ⃗ S2 .
1
Now, if we write down Hamiltonian which is H= E +3 Et )−( Es −E t ) ⃗S1 . ⃗S2 . We have
4( s
written it in a particular way this term, where Es is the energy of the singlet state and Et
is the energy of the triplet state. And why have we written it in this fashion is that
1
H|00>= E +3 E t )−( Es −Et ) ⃗S1 . ⃗
S 2|00> we can skip the comma in between. So, that
4( s
is the singlet state.
−3 2 1 −3 2
So, with Es = ℏ ,E t = ℏ 2 , one can simply check that H|00>= ℏ |00> .
4 4 4
1 1
Similarly H|↑↑>= ℏ 2|↑↑> , H|↓↓>= ℏ 2|↓↓> and
4 4
1 1
H|↑↓+↓↑>= ℏ 2
4 √2 [
|↑↓+↓↑>
]
So, that says that we have arrived at a Hamiltonian which is which gives us for a 2
particle problem which gives us the correct energy Eigen values for a 2 spinner particles
for a system of 2 spinner particles and this is that Hamiltonian.
107
Now, you can see that if you, if you redefine the 0 of the energy we may omit the
( Es +3 Et )
constant which is common to all the states all the all the 4 states then we can
4
write down a spin Hamiltonian, spin Hamiltonian as a H=J ∑ ⃗

S 1 . ⃗S2 where J is
<ij>
nothing, J=E s −Et . Here of course we have the singlet energy to be lowered which is
equal to minus 3 by 4 h cross square and Et being one-fourth h cross square. So, J will be
negative. Now, if we if we say that such Hamiltonian's can be written for n particles with
the pair wise interaction between the particles then we can write the generic Hamiltonian
for a magnetic system or spin half system we can extend it to spin having any values it
should be then it is a H=−J ∑ ⃗

S 1 . ⃗S2 .
<ij>
And this is Heisenberg Hamiltonian if S has the full rotational symmetry and it is just the
ising Hamiltonian if S is taken as plus minus half ah, but however, it gives magnetic
properties of the magnetic systems such as antiferromagnet or ferromagnet. And of
course if J is positive now we are not restricting ourselves to only 2 particles, where we
know that J is negative ah, but we also consider go consider J to be positive as well. So,
if J is positive in this particular model in this Hamiltonian given by 1, 1 favors we can
write it with the minus sign putting a minus sign from outside then this favors parallel
108
arrangements of spins and which are essential for ferromagnetism. And if J is negative
then 1 favors anti parallel arrangement and it is anti ferromagnetism is.
We have seen these phenomenon from the purely electronic model which is Hubbard
model, but; however, we have also got an exposed to this kind of spin only models which
S1 . ⃗
are, which are there. So, if J is positive then the energy is lowered if the ⃗ S 2 that is
the S i n j is they point the spin vectors pointing the same direction which are in the sense
we talk about ferromagnetism where as if j is negative then; that means, that the whole
energy would be negative if S i and S j are anti parallel, aligned which are the features of
anti ferromagneticism and.
So, this can be actually compared with the magnetic dipolar interaction like this which is
1
U= [ m⃗ 1 . m⃗ 2−3 ( m⃗ 1. r^ )( m⃗ 2. r^ ) ] . So, these are the 2 magnetic moments and these are
r3
related this you are familiar in the context of classical electromagnetic theory and the
relative distance where the relative distance between m 1 and m 2 are involved ah, but
here we have a purely spin Hamiltonian which neglects all spatial symmetries.
Now, this is H written as H=J ∑ ⃗

S 1 . ⃗S2 has there are a large number of approximations
<ij>
that are going on namely i j’s are the nearest neighbors one does not have to be one can
include longer than nearest neighbor that is next to next nearest interactions as well. And
109
we can also write this as H= ∑ J ij ⃗S1 . ⃗
S 2 J to be inside and it does not have to be
<ij>
constant and it can depend from one bond to another and so these are, these are possible
Hamiltonian and they have all been explored in the context of spin systems.
110
Prof. Saurabh Basu
Lecture – 09
Antiferromagnetism in Hubbard model
So, we have discussed ferromagnetism, now we shall discuss Antiferromagnetism so,

what we mean by Antiferromagnetism is that there is a 2 sub lattice in a given crystal
lattices, in which one sub lattice call it as, A sub lattice we make this discussion more
clear.
A sub lattice contains primarily up spin density or predominantly up spin density and the
other sub lattice call it B sub lattice contains predominantly down spin density which
means down spins. So, you have ordering so, this ordering is up, down, up, down as you
go from one lattice site to another and this is called as the Antiferromagnetism
So, we will discuss Antiferromagnetism and again we shall talk about Hubbard model
which is what we have done earlier and the Hubbard model is once again for
convenience I am writing it as H=−t ∑ Ci+σ C j σ +U ∑ ni ↑ n i↓ the kinetic energy

<ij> , σ i
term which is the first term this term which is neighboring sites hopping between
neighboring sites i and j and sigma is a spin.
111
So, this angular bracket over i j that you are seeing here a, denotes that they are
neighboring sites. So, hopping is over the neighboring sites and then we have an
interaction which goes as n i up and n i down and runs over all sites i. So, this is the
interaction between the densities at a given site. So, this is an onsite interaction between
taking place between up spin density and down spin density we cannot have a up spin up
spin at a given site which is excluded by the Pauli’s exclusion principle
So, similarly now the problem with the solving this model is that you have this term as a
2 particle term which we know that kinetic energy is operators is actually a 2 body
operator or rather one particle operator in that sense. So, it causes hopping from one site
to another where as if you open this term it will be a 4 body operator and written together
you cannot find a suitable basis so, as to solve them exactly.
So, we will not talk about exact solution, but rather what we should talk about is mean
field approximate solution and this mean field solution that we are going to talk about
now is called as the Hartree - Fock Approximation. So, let us see what Hartree - Fock
approximation means now this will be again discussed at length when we do the green’s
function problem so, Hartree - Fock approximation.
So, by this what we mean is will write down the interaction term or call it H int . So, by
H inte =U ∑ ni ↑ n i↓=U ∑ C +i↑ Ci ↑ C +i↓ C i↓ this approximation what we are going to do is

i i
that we are going to split this 4 operator term into 2 operator terms and take all possible
combinations that come our way and so, this H int Hartree - Fock is equal to
HF
H inte =U ∑ < ai+↑ a i↑ >−U ∑ <C +i↑ Ci+↓> +U ∑ C So, the first the combinations are
i i
taken with these 2 and then the combinations will be taken between the first and the
third. So, I will have a term such as c i up dagger c i down and then there will be a
combination taken between. So, this will be this should come in with the negative sign
because you have changed to one electron operator you have swapped the c i down with
the c i up. So, this should come with the negative sign.
And so, this is the negative and then there will be term which is between the first and the
fourth which is again sorry this is dagger and then there will be one between the first and
the fourth so, that cause 2 swaps which would eventually get me a positive sign. So, that
112
is c i. So, I will write it once again I need bigger space so; I will just let me erase this and
so, that we can write it slightly elaborately in the next page because I would need larger
space here. So, I am cutting it out and let us just go to the next page and write down this.
So,
HF
H inte =U ∑ < Ci+↑ C i↑ > Ci+↓ C i↓ + < C +i↓ Ci ↓> C +i↑ Ci ↑−Ci+↑ C i↓ <Ci+↓ C i↑ >−C +i↓ Ci ↑ <C +i↑ Ci ↓>
i
Now, you see by doing this by doing this we have converted. So, what we have done is
that we have taken expectation values of 2 of the operators at a time. So, that the Hartree
Fock Hamiltonian can be reduced to a uniformly 2 body term or just like the kinetic
energy so, that we can found out a basis and can diagonalise them. So, the 4 operator
term has been decoupled into taking a expectation value of 2 of the operators at a time
and we have taken all possible combinations in doing.
So, now, it so, happens that the terms which are the first 2 terms that is this terms are
important for considering antiferromagnetism where as this other terms are not important
for antiferromagnetism, but they are important for something else which we will see
later. So, if you drop these 2 terms and off course there is another term that we are
dropping is that when we have the expectation of all the 4 that is 2 at a time. So, there
113
will be a term which is so, neglected term is which is like terms such as this. So,
U ∑ < Ci+σ C iσ ' > < Ci σ ' Ci σ >

+
So, the expectation of so, these are constant terms because we are taking the average
values of each of these 2 and it just turns out to be a constant and which is dropped from
the problem. But; however, so this total or rather this Hartree - Fock or the interaction
HF
term is written as H intt =U ∑ [< ni ↑> ni ↓ + <ni ↓> ni↑ ] and this is our Hamiltonian that we
i
are going to look at along with the kinetic energy term. So, now, consider a specific case
as an Antiferromagnet.
So, consider an Antiferromagnet and what we mean by Antiferromagnet is that as I said

earlier that there are 2 sub lattices A and B and so, when site is that of A sub lattice then
it contains predominantly up spin and a site B it contains predominantly down spin.
So, if that is the case then we have so, an up spin Hamiltonian feels a potential average
U <ni ↓ >down spin densities and similarly a down spin Hamiltonian feels a potential . So,
A A
i i
B B
U <ni ↑ >if we introduce notations such as the total density be <n↑ > + <n↓ > =1
So, . So, the total density of up spin and down spin should be equal to 1 as we said that
A contains predominantly up spin density.
114
So, if the site is i site belongs to that of A n up would be much larger than n down and
vice versa if i belongs to a site in B then n down will be much bigger than n up, but;
however, both of them put together should give me 1 and we can also introduce
magnetization which is given as the difference in these 2 densities so,
A A
i i
<n↑ > B −< n↓ > B =±m
Now, this could be the way is written it could be plus m or minus m depending on
whether we are on are on the A sub lattice or B sub lattice if we are on the A sub lattice
then the up spin density will be bigger than the down spin density and in which case m
will be positive and otherwise will have a negative sign here. So, you can see that this
whole scenario can be slightly simplified if we write the m as a minus 1 whole to the
power i into m; that means, that if we start scanning the lattice site as 1, 2, 3, 4 etc will
alternately get a b a b a b and so, on.
So, depending on whether we have m i equal to even then will have m equal to positive
and if we have i equal to odd then will have m equal to negative so, this is the situation.
So, let me go back once again and just let you know that what is my system so, we are
planning to study antiferromagnetism and we have written down Hubbard model and we
want to understand antiferromagnetism from the Hubbard model just like we have
understand understood ferromagnetism from the Hubbard model.
Now as I said that this Hamiltonian cannot be solved exactly because the 1 contains 2
electron operators the other contains the second term contains 4 electron operators. So, it
is there is no suitable basis you can solve it one can off course do an exact
diagonalisation of the problem which is often done, but it is usually for a smaller number
of sites and we want to do it analytically here.
So, we employ a Hartree - Fock approximation which is equivalent to splitting this 4

operator term into 2 operator terms and if we take unrestricted combinations we get 4
terms as it shown here let us call this as equation 1 or maybe we can call this one as
equation 1 and this one as equation 2.
So, in equation 2 we have taken all possible combinations and we have said that the only
the first 2 combinations are relevant for studying antiferromagnetism and the other 2
combinations are not relevant immediately because they are they correspond to how an
115
up spin is correlated with the an up spin creation operator is created with the down spin
annihilation operator and vice versa and these are not relevant for studying
antiferromagnetism because here we are interested in the to see that how a down spin,
what a down spin potential is and what an up spin potential is so, that we can write down
a Hamiltonian separately for an up spin or a down spin.
So, this is the form of the Hartree Fock Hamiltonian and let us make an assumption to
begin with that there are 2 lattice sides A and B and the whole system is formed of
interlacing these 2 sub lattices A and B sub lattices. A contains predominantly up spin
there could be a small down spin density, but very small and B contains predominantly
down spin density and it could also have a small up spin density. So, now, to make
notations clear we have written down the total density for i to be in A or B sub lattice
equal to 1 and the magnetization is defined as this.
Now, we have to solve this Hamiltonian because now we have been able to write it in
terms of 2 particles 2 operators 2 Fermion operators for each one of the terms that
appear there. So, for up spin V up i equal to U into down spin density as we have written
which is equal to now from this I solved for each one of the densities. So, these are really
the average values that appear there. So, these are averages and this is equal to by
U Um
V ↑ (i ) =U <ni ↓ >= − ( −1 )i solving these 2 for each one of the n up and n down
2 2
U Um
and for down spin my potential for a given site is V ↓ (i ) =U <ni ↑ >= + ( −1 )i
2 2
So, these are the potentials for the up spin and the down spin and we have also written it
down in terms of m etc. Now we can also define a quantity called as the delta so, define
delta which will say eventually that it is the magnitude of the gap for an Anti-
Um
ferromagnetic insulator it is ( −1 )i that is the second term that appears above and
2
let us keep numbering them. So, let us call this as number 3 and number 4 so, in equation
4 the second term would be replaced by delta.
116
U
So, that again VV ↑ (i ϵ A )= − Δ=V ↓ ( i ϵ B ) . So, there is A sub lattice symmetry which
2
U
you can see it from here. So, the V ↑ (i ϵ B )= + Δ=V ↓ ( i ϵ A )
2
And suppose we write this now we can write down the Hamiltonian for each species of
spin that is say for up spin. So, let us write down a up spin Hamiltonian to be simply H
H ↑ (i ) =−t ∑ ¿
+¿
¿ ij>¿C i↑ C j↑ + ∑ V ↑ (i) ni↑ ¿¿
and this is nothing, but C +¿C
i↑
¿
i↑
.
iϵ A ,B
So, this is the Hamiltonian for the up spin for i to be any lattice site i can belong to either
A or B and if i belongs to A then surely j will belong to B and the visa versa it is true that
is if i belongs to B then j will belong to A. Now this way we can also write down the
Hamiltonian for the down spin exactly in the similar manner when we will have a simply
a all these ups will be replaced by downs and we now know what V up and V down are
from the definition that we have given in equation 5 let us call this is equation 6.
Now in order to solve it because we have a 2 sub lattices we can write as a this
Hamiltonian as a 2 by 2 Hamiltonian in the sub lattice basis and for that we will have to
transform into the momentum space and the momentum space Hamiltonian H up
H ↑ ( k )=∑ ϵ k C +k ↑A C Bk↑ + ∑ ( − Δ ) C +k ↑A CkA− Q↑+ ( Δ ) C+k B↑ CkB− Q ↑

k
117
this is the up spin Hamiltonian.
Now, the sub lattice indexes are made explicit in the super script and there is a off course
also a constant term which you will be able to find and will write it the, but the constant
term as said is dropped and is not considered in our problem because that will just shift
the energy up by a constant term and down by a constant term on little simply a shift. So,
it is not important so, what is important is that, we should now be able to write down a 2
by 2 Hamiltonian from the Hamiltonian that appears in 7 and it is it so, write it the task is
that write H up K in sub lattice basis and so this H up K is nothing, but minus delta
epsilon K let me take out a bit more space here.
− Δ ϵk
And so, that I can write down the basis also which will help you. So, H ↑ ( k )= ( ϵk Δ )
+A +B
and now you will have a C K it is written in C C (k ↑) Ck − q ↑. So, it is in that so, it is
C +(k A↑) C+k −B q ↑ and sorry this is not plus this is minus we have written that is minus. So, this
A
− Δ ΕK Ck ↑ U 1 0
is minus and minus here will. So, H ↑ ( k )= ( ϵk )( ) ( )
B
+
Δ Ck −q↑ 2 0 1
and we have said
that we are neglecting it is term which is like this and this is say equation 8.
So, now, we can so, it is 2 by 2 Hamiltonian and it can easily be solved let me write
down the Hamiltonian here this can be solved with the energy Eigen value. So, I am
118
neglecting this part because it is a constant and so, E K if you solve this with this you
know how to solve 2 by 2 matrices and find the Eigen values. So, E K up becomes equal
to Ek ↑=± √ Δ 2+ ϵ 2K and similarly E over H down. So, this is my equation number 9.
Δ ϵk
So, H down K without the constant term can be written a H ↓ ( k )= ( ϵk − Δ )
and will again
have the Eigen values as Ek ↓=E k ↑=± √ Δ2 +ϵ 2K . So, this is your solution of the problem.
So, do say we have been able to do a Harte fock decomposition of the Hubbard
Hamiltonian and have been able to write it as a 2 by 2 in sub lattice basis and 2 by 2
matrices in sub lattice basis and then diagonalise it to find that there is a symmetry of the
up and down spins both give me Eigen values which are same as ± √ Δ2 +ϵ 2K .
Now, this tells me that this gives rise to density of states how do we calculate density of
η
the states density of states is calculated as N ( ϵ )=∑ δ ( ϵ − Ek ) =lim ∑
k η→0 k ( ( ϵ − Ek )2 +η 2)
This is usually done when you have to calculate the density of states which is an delta
function and the delta function will give me a peak as soon as mug the energy will hit a
particular value of E K and E K will have range of spectrum of values over the full
brillouin zone and if you are be a talking about square lattice we have a Brillouin zone
from minus pi over a to plus pi over A.
Where A is the lattice constant and in order this in order to calculate it computationally a
delta function can be delta function is ideally just the straight line having a peak and
infinite peak at particular value of this x axis here whenever omega will hit a key I will
get a peak and infinite peaks you have to say and I will have to sum over all these keys in
order to calculate the total densities of sates. So, it can be computationally calculated if a
give it small width to this by expressing it as a Lorentzian. So, this is the lorentzian and
this will help us to calculate the density of states.
119
And this density of states comes out as a so, this it will look like this and so, this is the
occupied band which is the below Fermi energy. So, this is the Fermi energy and this is
the basically we are plotting density of states verses energy, energy E and this is plus
delta and this is minus delta and we have empty whole band which is here. So, this is
empty and this is filled so, this is the, it is called as the particle band.
So, it is filled and this is called as the whole band and that is empty ok. So, this is the
density of states and this is where the Fermi energy lies. So, while write it here the Fermi
energy instead of there that is the density of states verses energy plot and sometimes you
will find that this is the axis turned that is in the x axis energy and the y axis will have
density of states, but it means the same thing.
Now you see that remember that we have 2 delta equal to U into m minus 1 to the power
i. So, if u goes to 0 that is if you switch off the interaction term in the Hubbard model
then you will have only kinetic energy. So, which means that this will this gap will go to
0 the gap that appears here so, this is the gap. So, this gap will go to 0 and will have a
system which is which looks like this which is a metal and so, this is my E f and this is
for U equal to 0. So, these are all occupied and it corresponds to metallic vision because
you have only the whole states are empty, but there are infinite similar to the particle
states so, there are transitions possible between the particle states and the whole states.
120
The interesting thing is that as soon as and infinitesimal U is switched on. So, I will write
this because this important as soon as an infinitesimal U is switched on a gap opens up
and we have an insulator, but remember we have started with an anti ferromagnetic
system.
So, the antiferromagnetism has to show up now going back to this formula we can now
drop the minus 1 whole to the power i and only consider the magnitude because we
know that it will vary from one site to another it will alternatively become up and downs
spin depending upon whether your on A sub lattice or B sub lattice. Now in this problem
you have been able to diagonalise it and found out the Eigen values, but what about the
Eigen vectors and what do they give us.
So, let us try to find the Eigen vectors.
2
√ 1− β
( )
k↑
αk ↑
So, let us calculate the Eigen vectors of ( )
βk ↑
=
√( E k↑
ϵk
2
− Δ ) +ϵ 2
k
and do a lit bit of a
algebra to find out the Eigen vector let us call them as alpha K and beta K for the up spin
similarly there will be there for the down spin is as well. So, this we only can calculate
one of them because the other one will obey the orthonormality condition which says
that alpha K up square plus beta K up square equal to 1. So, suppose we calculate only
beta K and say that alpha K is simply equal to this and beta K comes out to be this you
121
have to do it you have to put one of the Eigen value say plus Eigen value and then do it
and then also do it for the minus Eigen values. So, we are showing it for the Eigen
negative Eigen value which is equal t Ek =− √ ϵ 2k + Δ 2 you can also do it for the positive
Eigen value things will not be anything different.
So, I have a epsilon K up minus a delta whole square plus epsilon K square. So, this is
my amplitude and off course the state will be some amplitude into some space part and
in is. So, that could be function of r in some trivial or some non trivial manner so, this is
your and then you can calculate the magnetization. So, magnetization is m=n↑ − n↓you
know that we are talking about a particular site. Now here we are talking about a up spin
or a sub lattice site and this will be nothing, but equal to 2 over N, I will tell you why it is
2 over N and K.
Now I will have to sum over all K, but which K is the K is hat belong to the filled band
that is the lower let us call it lower Hubbard band or lower band would be sign and this is
2
∑ α 2k ↑ − α 2k ↓, why it is 2 over N, which means it is 1 by N by 2 the reason is that N
N k ϵ lower
is the number of sites and N by 2 is the number of unit cells and so, we are actually
summing over the unit cells is that clear so far.
So, we are talking about the magnetization which is the up spin density minus a down
spin density and how do we find those densities will have to sum over all occupied levels
and we are summing over only the levels which are occupied that is the whole levels are
empty and we have N sites off course in the lattice, but there are N by 2 unit cells
because each unit cell contains 2 sites each one of A and B.
So, a unit cell contains A sub lattice and the B sub lattice so, this is the A and this is the
B sub lattice. So, these are the unit cells for the problem and that is why we are summing
over the unit cells and that is why there is a 1 divided by N by 2. So, this N by 2 came
from a 1 divided by N by 2. So, this factor has gone in here. So, then we can solve for
this and what we can get or what we shall get is the following.
122
2 Δ 1 mU
So, will get m= ∑ = ∑ so, we are taking difference between the up spin
N k Ek N k E k
densities alpha K up square minus alpha K down square at a give site which is
predominantly containing an up spin ok. So, alpha K down square in principle very small
and alpha K up square is large.
So, that I should get a magnetization which is close to 1 and let us see that what happens
1 U
when we have so, this is nothing, but equal to 1= ∑ because my m U equal to 2
N k Ek
delta. This we have defined it earlier and a since m is not equal to 0, I can cancel m form
1 1 1
both sides and I will get =
U N ∑ Ek
. Now K belonging to the lower Hubbard band I
k ϵ LHB
will write it here LHB means a lower Hubbard band which is occupied.
So, this is equal to 1 over E k. So, just like the stoner criteria we got a self consistent
criterion for antiferromagnetism which is similar to the stoner criterion that we have
obtained in the last discussion. So, that is myself consistent equation and I should this is
the condition for antiferromagnetism to occur.
Now let us see a specific case which is of importance let us call c the strong coupling
limit and by strong coupling limit what I mean is that U is much much greater than t and
123
in principle one can just simply say that U tends to infinity. So, the Fermi on, the
electrons are extremely correlated at a given site.
So, it becomes energetically impossible for if there is an up spin at a given site down
spin to come and sit at the same site is nearly impossible. So, this means that implies that
it is energetically unfavorable for electrons of opposite spins to occupy the same site. So,
if this is the case this is called as the strong coupling limit and or rather strong correlation
limit in which case what we have is the following that.
mu
We have a Δ= is very large because U is large and which case let us look at the beta
2
K that appears there. So, we have equation 9 there and let us call this is equation 10 so,
the equation. So, beta in equation 10 let us call this is equation 11. So, beta in equation
10 or beta up or beta K up beta K up in equation 10 takes the form.
ϵk
So, β k ↑= . Now for E K to be which want to consider so, if we take E K to
√( E k↑ − Δ ) 2+ ϵ 2k
be minus delta, you see here is E K equal to minus delta. So, we take the top of the filled
band so, top of the filled band then I have if I put it here then I get a minus delta minus
delta whole square which gives me 2 delta whole square which is 4 delta square and that
can certainly be much greater than epsilon K square.
124
ϵk
So, epsilon K square can be neglected in the denominator. So, I will get a β k ↑= and
2Δ
this is since delta is large beta K up is very small and this is what we expect that A sub
lattice being sub lattice which contains predominantly up spin the beta K up will be very
small and then what we have is that α k ↑=1 because alpha K is since α k ↑=√ 1− β 2k ↑ and
this is nearly equal to 0. So, alpha K up is equal to 1 and if that is the case then my m
becomes equal to 2 over N and sum over K now you see that it had this had m U over E
K.
So, I can write that as simply as E K to be this can be written as this sum and 1. So, this
2 2 N
is m= ∑
N k
1= ∗
N 2 ( )
=1, K over all unit cells and you are summing over 1 which
means your summing over the number of unit cells which should give me 1.
So, this is my magnetization so, magnetization is equal to 1 which means that N up

minus N down equal to 1 which means N down equal to 0 and N up equal to 1 and the
1 1 1 1
self consistency condition becomes equal to = ∑ = now you see that E K
U N k (2 Δ ) ( 2 Δ )
becomes E K is same as E K minus delta and that is equal to. So, square root and so, on
this is equal to 2 delta. So, this is the self consistency equation and this is nothing, but
equal to 2 1 over 2 delta because of since you have sum over K summed over 1 gives me
equal to N.
125
And this is the self consistency condition is that the ( 2 Δ )=U and that immediately says
that m equal to 1. So, in the strong coupling limit the magnetization is 1 this is the
magnetization per sites the weights defined as a n i up minus n i down, but; however,
Hartree - Fock picture is somewhat away or some what sort of the realistic picture would
be some quantum fluctuations associated with it, because it is a mean field picture or a
average field picture and this because of the quantum fluctuations is value of the
magnetization goes down and these quantum fluctuations we are not including it, but we
will learn later how to include quantum fluctuations.
So, quantum fluctuations reduce this value to a 0.6 in 2 dimensions. So, in 2 dimensions
the fluctuations are large so, this quantum fluctuations actually reduce the value by 40
percent and the magnetization will be in the strong coupling limit the magnetization will
be 0.6 that is the sub lattice magnetization.
So, this gives rise to an anti ferromagnetic insulator as soon as you switch on a U it will
be an insulator and each site will order the A sub lattice will have a up spin density B
will have down spin. A will have up spin again and B will have down spin and this
corresponds to ordering vector which is given by.
So, ordering wave vector wave vector for the spin orientation is ⃗
Q = ( π , π , π ) let us just
try to see it in one dimension how it looks like. So, in one dimension will have sort of so,
⃗
SQ =S 0 ei Q . ⃗r. So, this is how the profile will go and so, in one
this enables us to write ⃗
126
dimension if I take a exponential i pi x as the modulation, this i pi will give me negative
sign.
So, as I go from one site to another I mean I have to write this as delta so, this will be a
delta. So, how the, if I have an ↑ →i, ↓ →i+δ. So, the ordering becomes so, as I go from
up i to i plus delta i should get from the ordering of the spin.
The spin vector will change it is orientation from up to down and that is why this pi, pi,
pi that kind of a ordering will come and this can be actually probed in from scattering
experiment. So, the neutrons scattering experiment will detect these Eigen vector and see
that there is a ordering of this kind in the 2 dimensional plane it is. So, in order to do
neutrons scattering for a 2 dimensional thing one has to actually take a small angle
neutrons scattering so, that you can see the surface well so, the spin orientation in the
surface.
127
Prof. Saurabh Basu
Lecture – 10
Green's function and representations in quantum mechanics
So, we shall talk about greens functions and in particular we will start with Green’s
functions at 0 temperature.
So, T equal to 0 and then we will learn how to extend it to finite temperature, but before
we go on to discuss Greens function at T equal to 0 we will have do a number of things
and the thing that we want to start with is a basic definition of greens function. Now if
you remember that we have introduced greens function in the context of propagators and
the retarded propagator was identified as Greens function and also the advanced
propagator was introduced in that context.
So, I will give brief introduction to what Greens functions are and how they are relevant
to condensed matter of physics or many body phenomenon in condensed matter physics.
So, general introduction is what we plan to start with and this is slightly mathematical in
128
nature in the sense that definition is mathematical in nature, but you will find that what
we get at the end of it is quite useful in the context condensed matter physics. So, let us
have a linear operator L L f ( x )=g(x ) . So, this is a linear operator and f (x) is any
arbitrary function which is a function of one variable and it gives a g(x).
And the Greens function is defined as so, we write greens function by G (x, x’) where
the L the same linear operator operating on LG( x , x ')= δ ( x−x ' ) . Now let us say this
is equation 1 and this is equation 2. So, one can write down the solution of a equation 1
in terms of the greens function as so, f (x)=∫ G(x , x ' )g ( x ' )dx ' . So, the solution of 1
is given by 3 which invokes the greens function. Now this is the general definition of
mathematical definition of greens function, we have to understand that how it is of
relevance to the study that we are presently involved in and for that.
Let us take this Schrodinger equation and this is equal to H acting on a wave function
δ | ψ( t)>
H | ψ (t)> =iℏ that is the definition of that is the Schrodinger equation that
δt
we are all familiar with and the solution of this is given by psi of t in terms of the greens
129
function as d x. So, this is a in principle of function of both x and t and so, we will write
it with | ψ (t )> =∫ dx ' G( x , x ' ,t ) ψ ( x ' ,0) .
So, this is the integral solution of the differential equation that we have written above
which is Schrodinger equation and the psi the wave function at any given space time
point can be generated from psi x 0 or rather this is x prime 0 by taking contribution
from all x prime points and introducing the greens function here. So, where the
−iHt
G(x , x ' ,t )=< x | e ℏ
|x '> and this is very easy to see, but this comes from the
completeness relation of the wave functions that is ∫ dx ' | x '> < x ' |=1
So, this is the, this is how the wave function at a given space time point can be
calculated starting from a psi of x 0 here we are written x prime as a dummy variable and
all x primes are being summed over we have introduced this concept earlier. So, I will
−iHt
skip here and go on to define Green operator which is defined as G(t)=−iΘ(t )e ℏ
.
So, just to remind you, this theta function takes a value 1 when t is greater than 0 and it
takes a value 0 when t is less than 0. So, Θ(t) is the usual Heaviside function
Heaviside step function. So, let us go to the Fourier transform of this green operator.
130
So, the Fourier transform of the green operator is G of epsilon which is a canonical
variable which is energy, which is canonical to the time and this is equal to
i ϵt it ( ϵ− H )
G( ϵ )=∫ dt G(t)e ℏ =−i ∫∫ dt Θ e ℏ
. So, this is the Fourier transformed green
operator and also we can write down theta function the integral representation of the
1 e−i ω t
Θ(t)=
2πi
∫ ω−i η . So, this is the it can be easily shown by doing a contour
d ω
integration that this is equal to this the right hand side is equal to 1 when t is greater than
0 and x equal to 0 when t is less than 0.
Now, eta is an infinitesimal parameter and so, one can check easily by direct integration
and now recognize that there is another form of the derived delta function where
1 1
δ ( ω)= ∫ dt e i ω t . And so, G( ϵ )=−∫ d ω δ ( ω + ϵ −H ) and this gives me
2π ω −i η
1
when integral of this is . This establishes a very important relation from the
ϵ +i η−H
point of view of condensed matter physics that for a given system if we know what
G( ϵ ) is or G( ω ) is, if you wish to write it that way then it gives you the energy
spectrum as a pole of the greens function.
So, energy spectrum is the pole of the greens function so, this is the basic introduction to
the greens function now we will get ahead with this form or rather the discussion of the
greens function, but before that we wish to say that greens function will be written in
representation which is called as the interaction representation. So, there are 3 kinds of
representation that are used in quantum mechanics and so, they are called as Schrodinger
representation, Heisenberg representation and interaction representation and their silent
features are discussed.
131
So, one is the Schrodinger representation and the Schrodinger representation is where the
wave function is time dependent and the operators are all time independent. So, to
distinguish it strong the Heisenberg representation and where the wave functions are
time independent and operators are time dependent. So, let us take this discussion a little
farther and let us write down in the Schrodinger representation so, will call it as S for the
Schrodinger representation. So, in S representation we have the Schrodinger equation is
δ ψ (t)
iℏ =H ψ(t) H is independent time and psi carries the time dependence and
δt
i Et
−
ψ (t)=e ℏ
ψ(0) .
So, now, this is in S representation and in H representation we have in the H

representation we have operator any operator which could be the Hamiltonian in this
i Ht −i Ht
case is given by O(t)=e ℏ
O (0)e ℏ
where H is the Hamiltonian and the equation of
^ )
d 0(t
the motion for the operator is given by iℏ =[O(t), H ] . So, this is the let us
dt
write it a little neatly and this is equal to so, this is the equation of the motion and these
are in the Heisenberg picture. Now whichever picture you take the expectation values of
the operators there should be independent of the representation. So, keeping that in mind
132
let us take an operator O and calculate it is expectation values between 2 operator
between 2 states. So, that is given by < ψ1 ( t)| O(0) | ψ2 (t)> in the Schrodinger
representation.
So, this can be written as so, this is in S this can be written as

iH t −i H t
< ψ1 ( 0)| e ℏ
O(0)e ℏ
| ψ2 (0)> and this is in the Schrodinger representation and
similarly in the Heisenberg representation my wave functions are independent of time
and the time dependence is carried by the operators and then this can still be written as
just the same by using the relation a for the operator the time dependence of the operator
in terms of the Hamiltonian, < ψ1 ( 0)| O(t)| ψ2 (0)> .
iH t −i H t
So, this can be written as < ψ1 ( 0)| e ℏ
O(0)e ℏ
| ψ2 (0)> . So, this is in the
Heisenberg representation and you can see that both the representations off course give
rise to the same expectation values which is what it should be because a expectation
values are physical observable sorry physical quantity we should not depend on the
representation in quantum mechanics that you are considering.
133
So, then the interaction representation is where the both. So, operators are time
dependent and wave functions are time dependent as well. So, the operators have a time
i H 0t −i H 0 t
dependence which is given by O(t)=e ℏ
O( 0)e ℏ
where it is assumed that the
total Hamiltonian it is written as a non interacting part which is H 0 and there is an
interaction part which is H’. So, the operator the time evolution of the operator only
involves the non interacting part of the Hamiltonian and farther it is assumed that H 0
does not commute with H.
iH0t −i H t
So, the wave function has a time evolution which is given by ψ (t)=e ℏ
e ℏ
ψ (0)
i(H 0− H )t
now remember that you cannot write it as exponential e ℏ
because this is possible
only when H naught and H commute since they do not commute you cannot write it as
combining the exponential and you have to keep both of these things separately. So,
basically what I mean is that e A e B=e A +B and when [A,B]=0 if they do not commute
then you cannot write it that is why you cannot combine the exponential.
Let us just check the expectation values of the operators in this representation. So,
expectations values and this is
134
iHt −iH 0 t iH 0 t −iH 0 t iH 0t −iH t
^
< ψ1 ( t)| (0)(t)| ψ2 (t)>=< ψ1 (0)| e ℏ
e ℏ
(e ℏ
O(0)e ℏ
)e ℏ
e ℏ
| ψ2 (0)>
So, if you look at this then these cancel and as well this and this cancel giving me exactly
iH t −iH t
the thing that, I we want that is < ψ1 ( 0)| e ℏ
O(0)e ℏ
| ψ2( 0)> . So, it is clear that the
expectation values are independent of any representation that you take and it is been
shown to be the same in Schrodinger, Heisenberg and interaction representation. So, will
just tell you in while that why these representation are being introduced and the utilities
of that, now let us show that how the time dependence of the wave function is governed
by the interaction term H prime.
So, once again just remind you that my H is written as H=H 0 + H ' where H naught is
the non interacting Hamiltonian part of the Hamiltonian and H prime is the interaction
part. So, let us see that how the wave function the time dependence of the wave function
iH0t −i H t
is related to the H prime. So, this is equal to ψ (t)=e ℏ
e ℏ
ψ (0) This we have
taken the definition of psi reference just while back to remind you that that while put the
definition once again here.
135
iH0t −i H t
So, ψ (t)=e ℏ
e ℏ
ψ (0) . So, once you have to keep the first term constant and take
a derivative of the second term and then again you have to keep the second term constant
and take the derivative of the first term and this is what comes after that and so,
iH t −iH t
δ | ψ (t) 0
=i e e ( H 0−H ) e ℏ ψ (0)
ℏ
δt
i H0t −i H t
Here become exponential minus i H naught t. So, that is −i e e ℏ
H' e ℏ
ψ (0) . So,
this will be this can be written as. So, this is so, delta t of psi t is nothing, but equal to
i H0t −i H t i H0t −i H 0 t
this is nothing, −i e e ℏ
H' e ℏ
(e ℏ
e ℏ
ψ (0)) . So, this exponential i so, there is
this term which is which gives me H prime and. So, these things will combine to give me
δ | ψ (t)>
these time derivative or time evolution of the wave function =−i H ' (t) ψ( t)
δt
.
So, this is the equation of the motion for the wave function which involves H prime you
can fill in line here or step here and let us introduce an operator which is
iH0t −i H t
U (t )=e ℏ
e ℏ
with U 0 U at t equal to 0 is 1. So, the equation of the motion for U,
δ U (t)
I am writing it short hand. So, equation of motion is e o m for U is again I am
δt
going to take derivative with respect to time meaning that first one will kept constant and
second one will be taken derivative and so, on.
iH t i Ht
δ U (t) 0
So, then it becomes equal =−i e (H 0−H ) e ℏ . So, that is the equation of
ℏ
δt
motion for U and if you look at the equation of motion for U and equation of motion for
psi they look identical.
136
δ U (t)
So, this can also be written as =−i H ' (t)U (t) . So, this is the equation of
δt
motion for U so, EOM for psi of t and U of t are identical. Now since that is correct then
t
we can solve either of psi of t or U (t)=U (0)−i ∫ dt 1 H ' (t 1) U (t 1) . So, this is the
0
solution of this equation that appears at the top of the page which is for the equation of
motion for U. So, U has the solution has the time evolution which is given by this, but
the only problem with this form is that both the left hand side and right hand side are
they involved the same variable unknown variable which is U of t. So, if you repeat this
procedure for the right hand side as well will get an iterative solution by doing it so, U of
t 1.
We can write down again an integral expression of the kind that we have written for u of
t and then again will get a U t 2 for which we can write down again repetitive solution,
integral solution and will get a series which is like this. So, U of t is equal to now since U
of 0 has taken to be equal to 1. So, we can write simply equal to
t t t1
2
U (t)=U (0)−i ∫ dt 1 H '(t 1)+(−i) ∫ dt 1∫ dt 2 H ' (t1 ) H ' (t2 )+ .............
0 t n−1 0
137
So, this is the second term and there will be farther terms which finally, can be combined
∞ t t1 t n−1
n
with U (t)=∑ (−i) ∫ dt1∫ dt 2...... ∫ dt n H '(t 1) H ' (t 2) H ' (t 3)...... H '( t n )
n=0 0 0 0
So, this is the solution series solution of the, for the operator that we have introduced U
of t and hence it should also be the solution for psi of t as we have seen that the psi of t
and U of t have the same equation of the motion. Now it is important to introduce time
ordering here and that is going to be quite helpful.
So, introduce time ordering operator let us call it as t and what it does is that it does. So,
T [H '( t 1 )H '(t 2) H ' (t 3)]=H ' (t 3) H '(t 1) H ' (t 2) t 3 is greater than t 1 is greater than t
2 so, that is what it does this is an important step which I think you should take a note of
this that we have a time we have introduced the time ordering operator which acts on a
series of interacting Hamiltonians written as a different times t 1 t 2 t 3 and it puts as you
apply the time ordering operator it puts the Hamiltonian or the interaction part of the
Hamiltonian with the earliest time that is the time which is greatest on the left and then
orders it accordingly and then next one goes after that and the next one after and so, on
this is the time ordering of operator.
138
So, a this also step function that we are using which is not exactly the Heaviside step
function, but it is something very similar to the that step function is interesting or rather
is important here interesting in this context and this is theta of x equal to 1 for x greater
than 0, x equal to 0 for x less than 0 x equal to half for x equal to 0. So, how the t is
written so, it is take 2 times H prime. So, the time ordering operator operates on
Hamiltonian at 2 distinct times
T [H '( t 1 )H '(t 2)]=Θ(t 1−t 2) H ' (t 1) H ' (t 2 )+Θ(t 2−t 1 ) H ' (t 2 ) H '( t 1 ) .
So, that you have taken into account both the possibilities which t 1 is greater than t 2
and t 2 is greater than t 1 if in a particular case if t 1 is greater than t 2 then only the first
terms survive and the second term will be dropped to 0 and remember one thing that if
[H ' (t 1 ), H '(t 2)]=0 would have commuted and they commute then these ordering is
unimportant, but since they do not the ordering becomes important. So, let us look at the
time ordering operator a little more carefully.
So, let us just write
139
t t1 t t2 t t2
1 1 1
∫ dt 1∫ dt 2 T [H ' (t 1 ) H ' (t 2 )]= ∫ dt 1∫ dt 2 H ' (t 1 )H ' (t 2 )+ ∫ dt2∫ dt 1 H '(t 2) H ' (t 1)
2! 0 0
2! 0 0
2! 0 0
Now, if you see the first and the second terms they are identical under the swapping of t
1 to t 2 and visa versa this t 2 back to t 1, then we can write down that
t t t t1
1 1
∫ dt 1∫ dt 2 T [H ' (t 1 )H ' (t 2 )]= ∫ dt 1∫ dt 2 T [ H '(t 1) H ' (t 2)] . So, you do not
2! 0 0
2! 0 0
know that something like this had appeared in our solution for U of t and this we are able
to write down as time ordering of 2 operators. So, many of n operators will have a 1 over
n factorial and simply we can write down that term.
So, we will just so, for 3 time ordering operator just give one more example, example for
3 time times t 1 t 2 and t 3 and this is equal to
t t t t t1 t2
1
dt 1∫ dt 2∫ dt 3 T [ H ' (t 1) H '(t 2) H ' (t 3)]=∫ dt 1∫ dt 2∫ dt 3 T [ H '(t 1) H ' (t 2) H ' (t 3)]
3!∫
0 0 0 0 0 0
. So, we really do not need to care about which time comes first or which time is the
earlier time as compared to others.
140
So, we can simply write it with the time ordering operator and now also note that all the
time integrals go from 0 to t and this time ordering takes care of everything. So, using
this time ordering operator the solution of the operator that we had introduced is
∞ t t1 t2
U (t)=1+∑ ∫ dt 1∫ dt 2∫ dt 3 T [ H '(t 1) H ' (t 2) ..... H ' ( t n )]

n=1 0 0 0
This is the complete solution of U of t and hence reminding you the hence it is also the
complete solution of psi of t this because it is an infinite series because this n there is a
sum goes from one to infinity we can write this as an exponential with the time ordering
t
operator outside. So, U (t)=T exp [−i ∫ dt 1 H '(t 1)] . So, that is the solution of U(t)
0
and and also. So, same solution persists or rather you should write is applicable to psi of
t.
So, thus if we know the U of t then we can also know how psi of t evolves and for that
we have to know the interaction term act different time intervals H t 1 t 2 etc. So, if for a
given, so the idea is this that a particle is actually introduced into an interacting path and
it undergoes a scattering with other particles sorry it is acted upon by an external
interaction potential that is also possible and which.
So, it undergoes H prime at t 1 t 2 t 3 etc and this whole solution can be written as the
time ordering operator multiplied by the exponential of this which involves the
interaction term act given t 1 and has been integrated from 0 to t. So, this is the solution
of psi of t and will see that how to one gets a Greens function from here.
141
Prof. Saurabh Basu
Lecture – 11
S – Matrix and free electron Green’s Function
So, in the context of Green’s function at 0 temperatures we have introduced the 3

representations that we would be working in, namely the Schrodinger representation
Heisenberg representation. And interaction representation and we say that we would be
working in the interaction representation mode and in that context we have computed the
time evaluation of the wave function psi. And have introduced another operator U which
was defined and it was found that the time evaluation of this operator U and the wave
functions psi are similar and then finely the solution was written in an integral form.
Let us carry this discussion forward and introduce another important quantity called as
the S-Matrix.
So, let us see what S-Matrix is and how it is simplifies our discussion? So, to remind
ourselves the wave function in the interaction representation was written as
iH 0 t −iH t
ψ ( t ) =e ℏ
e ℏ
ψ ( 0 ) one can refer to the discussion earlier and this was actually written
as U of t. Hence this was written as U ( t ) ψ ( 0 ). Now so, this is how the wave function
142
would evolve from the wave function at time t equal to 0. So, one is to operate by this
iH 0 t −iH t
U ( t )=e ℏ
e ℏ , where H is the total Hamiltonian equal to H naught plus H prime and H
naught is an non interacting part of the Hamiltonian and H prime is the interaction term.
So, now one can also write it as. So, we can define in S matrix like this and by this which
connects the wave function at t and the wave function at t prime. So,
ψ ( t ) =S ( t , t ' ) ψ ( t ' )=S ( t ,t ' ) U (t ' ) ψ ( 0 )
Now this psi of t as we know that can be written as U ( t ) ψ ( 0 ) hence this

S ( t , t ' ) U ( t ' )=U ( t ) → S ( t , t ' ) =U ( t ) U +(t ' ).
So, this is the definition of S matrix which is constructed from the U operators, at time t
and t prime it is a S ( t , t ' )=1=U ( t ) U +(t ). And now since we know the solution of U of t
we can find out S t t prime see the solution of U of t is in terms of the interaction term H
prime which we have shown earlier.
So, now let us some properties of the S matrix and what are the properties one of them is
S ( t , t ' )=1=U ( t ) U +(t ) which is equal to 1. So, this 1 means it is a unit matrix. So, that is
that can be easily shown let us show this, because U of t is like
iH 0 t −iH t iH t −iH 0 t
(e ℏ
e ℏ )( e ℏ
e ℏ )=1.
Remember that when you take a dagger you have to write the operators so, this is like
taking a transpose and that has to be taken in the reverse order. So, we have written this
as a so, it is a transpose and then it is taken as a conjugate and then it is written
backward. So, this is off course equal to 1, because these 2 will become equal to one then
farther these 2 will become equal to 1. So, that is 1 property.
143
Let us see the second property of this matrix. So, property number 2, which is equal to ,
S + ( t , t ' ) =U (t ' ) U + ( t )=S ( t ' ,t ) which means that S dagger t t prime is same as . So, t
S ( t , t ' ) S ( t ' , t )=S ( t , t ' ' ) his says that when we multiply these S matrices the resultant
matrix will be the first one and the last the first time index and the last time index. And
this can be you know extended to more than 2 products. And the reason that this is true is
that let us give a short proof of this one as well.
So, your ψ ( t ) =S ( t , t ' ) ψ ( t ' )=S ( t ,t ' ) S ( t ' , t ' ' ) ψ (t ' ' ) =S ( t ,t ' ' ) ψ ( t ' ' ) . And if you
compare the this one and this one it is easy to see that what is written in 3 follows from
that.
So, these are the properties and now the equation of motion every time will use this word
and write it as EOM. So, the equation of motion of ψ is . Now you see that the equation
δ S (t , t ' ) δ U (t ) U+ (t )
= =−i H ' ( t ) S ( t ,t ' ) of motion exactly takes the same form as U
δt δt
of t or psi of t. So, the S matrix also has a solution which looks like it is a
[ ]
S ( t , t ' )=T exp − i ∫ dt 1 H ' ( t 1 ) . The only difference is that the S matrix is function of 2
t'
time variables t and t prime, that is why this integral does not go from 0 to t rather it goes
144
from t prime to t and it is the same interaction term and it is exponential related and with
the time ordering operator.
This is exactly the same as we are written for the U of t or the psi of t. So, far we have
introduced the 3 representations and have computed closed form for the wave functions
psi the operator U and the S matrix and now the whole problem.
So, far is that that our ψ ( 0 ) is unknown and it is actually a genuine problem, because we
need to know the psi of 0 because everywhere it is we are getting that the psi of t is
generated from psi of 0 if psi of 0 is not known, then we cannot get psi of t even if we get
the U of t or S t t prime etc. However, there is one thing that is known is the Eigen
solutions of H 0 are known H 0 to remind you is the non-interacting Hamiltonian. So, let
us write that because it will help you. So, it is a non-interacting Hamiltonian and it is
assumed that we can solve it exactly and we know the wave functions and the ground
state energy.
So, by Eigen solutions what we mean is that ϵ ( 0 ) and ϕ ( 0 ) are known where ϵ ( 0 ) is the
energy of the non-interacting system and phi 0 is the wave function for the non-
interacting system. These are known these are the known quantities. And now we have to
generate psi of 0 from these ϵ ( 0 ) and ϕ ( 0 ) in particular ϕ ( 0 ) because psi 0 is the t equal
to 0 wave function; for the interacting system which is not known; however, the non-
interacting ground state or the energy the wave function is known.
145
So, we need to get that now this is facilitated by a theorem called as the Gellmann-low
theorem. So, the Gellmann-low theorem says that the interacting wave function at psi
equal to 0. Now when I say wave function at the back of my mind i have ground state
wave function, because that is the lowest energy state and in condensed matter physics as
we have said earlier also that, we are only interested in the low energy states and which
is the particularly the ground state wave function and the ground state energies. So, while
psi of 0 can actually mean any state at t equal to 0, but in we actually mean that it is a
ground state.
So, this is written in including the | ψ ( 0 ) >=S ( 0, −∞ ) | ϕ ( 0 )>. So, it is clear that psi of 0,
which is the wave function of the ground state wave function for the interacting system
can be generated from the wave function of the non-interacting system, which is known
by this S matrix, which are where the 2 time variables are 0 and minus infinity that is the
theorem statement of the theorem. Why does the theorem makes the sense? It make sense
because the interacting many body ground state must has evolved from the non-
interacting ground state in the distant past.
So, also it is known that let me write down because this is important makes sense
because the interacting many body ground state must have evolved from the non-
interacting ground state in the distant past, which means at t equal to minus infinity that
is how this S matrices is written as 0 and minus infinity.
Also it is true that ψ ( t ) =S ( t , 0 ) ψ ( 0 ) if we operate by S(0, t) and in that case we have a

S ( 0, t ) →ψ ( 0 ) =S ( 0 ,t ) ψ ( t ). So, my interacting many body ground state at t equal to 0 is
obtained from the from the many body ground state at a finite time t using this S Matrix.
146
(Refer Slide Time: `17:05)
And why is this done is because for t equal to minus infinity my

| ψ ( 0 ) >=S ( 0 , −∞ ) | ψ ( − ∞ )> =S ( 0 , −∞ ) | ϕ ( 0 )>you want to write it with the brackets. So,
| ψ ( − ∞ ) >=| ϕ 0 > which means that at t equal to minus infinity the interaction is not even
switched on. So, the interaction is switched on only after t equal to minus infinity.
And so, this is the non-interacting ground state now there is an additional property that
we would be needing at this in order to define the Green’s function. So, at t equal to plus
infinity we have | ψ ( ∞ )> =S ( ∞ ,0 ) ψ ( 0 ) . Now it can also be assumed that the interaction
is switched off at t equal to plus infinity. So, the interaction is switched on after t equal to
minus infinity and is switched off after t equal to plus infinity. And in which case the psi
infinity. So, psi infinity and phi 0 may be same and even if they are not same they should
only differ by a trivial phase factor or differ by a trivial phase factor say a chi.
So, we can write which is considering the case in which it differs is

ϕ 0 e i χ =ψ ( 0 )=S ( ∞ , 0 ) ψ ( 0 )=S ( ∞ ,− ∞ ) ϕ 0. I am not writing the always you can if you
wish you can write it.
So, this phase factor Is e i χ =< ϕ 0 | S ( ∞ ,− ∞ ) | ϕ 0 >. So, this is an interesting and important
property or rather and equation that you have would need while defining the Green’s
function. So, let us finally, the get to the Green’s function at T equal to 0.
147
And will write it in the Heisenberg representation. So, at T equal to 0 in Heisenberg

representation there is a reason to write it initially in Heisenberg representation and then
will switch over to the interaction representation.
So, the Green’s function is defined as G and some parameter
G ( λ , t −t ' )=− i <ψ 0 |T [ C λ ( t ) C +λ ( t ' ) ] |ψ 0 > So, that is the definition which you should
keep in mind lambda is any parameter it could be energy, it could be momentum, it could
be some other property, which is like spin or something which is equal to minus i times
the ground state expectation value of C lambda C lambda dagger at t and t prime with the
time ordering involved there.
That is the definition and so, lambda will relevant quantum mechanical parameter. Now
it could be for an electron gas for an electron gas lambda could be set of 2 quantum
numbers namely k and spin sigma or for other problems it could be something else. And
once again to remind ourselves that psi naught is the interacting many body ground state
and a c lambda and C lambda C lambda t and C lambda dagger t prime are written in the
Heisenberg representation as C λ ( t )=eiHt C λ ( 0 ) e −iHt
I am dropping ℏ and taking is equal to one and this is in the Heisenberg representation,
this we have discussed at length.
148
So, this is my operator. So, for t greater t prime greater than t the Green’s function is
defined as G ( λ , t −t ' )=i <ψ 0 | [ C +λ ( t ' ) C λ ( t ) ] | ψ 0 > .
Since we have resorted to a particular case where t prime is greater than t i the time
ordering says that we have to write down t prime on the left, in order to do that i gained
an additional negative sign, because of the anti-commutation relation of the Fermi on
operators, which absorbs this negative sign that written in equation one and also I now
do not need the time ordering operator because I have sorted the particular case this just
an example that we are wanted to site.
So, let us just look at things such as.
So, what is my | ψ 0 >=S ( 0 , −∞ ) | ϕ 0 > it is already written earlier. And now we have to go
to the interaction representation, because this is what we had planned now in order to go
to the interaction representation we write down the operators in the interaction
representation as this. From now on we would take h cross equal to 1 so, not put that in
exponent the denominator of the exponent.
So, this is how the time evolution of the operators are written with the
C λ ( t )=eiH t C λ ( 0 ) e −iH t. And so, if i take if i want to write the C λ ( 0 )=e −iH t C λ ( t ) eiH t. So,
0 0 0 0
what I did is that i left multiply a exponential i H 0 t and right multiply a exponential i H
naught t.
149
And that gives C lambda at t equal to 0 in terms of C lambda at t. Now these operators’
operators have to be converted to interaction representation. And in order to do that we
can write down a C λ ( t )=eiH t e− iH t C λ ( t ) e iH t e −iH t Remember the time evolution of the
0 0
operators in the Heisenberg representation is given by e i H t C λ ( 0 ) e− i H t. And now my C

lambda 0 is same as this, which is from the step earlier. So, I will have to write a in a
interaction representation my C lambda as this and this can be written as
C λ ( t )=U + ( t ) C λ ( t ) U (t ) and remember that write it neatly.
So, these U of t S are the operators that we have written earlier where U ( t )=e i H t e −i H t .
0
So, this is how the interaction that operators in interaction representations are written and
this can be written as C λ ( t )=S (0 ,∞) C λ ( t ) S( t , 0)
So, I introduce the S matrices which we have learned and written down the operators and
this was precisely the reason that we have introduced the S matrices and U in matrices,
and now we know the complete solution of the S matrices and the U matrices which
would be used in order to define the Green’s function.
So, let us write down. So, this will be slightly bigger expressions, but nothing to worry
these are well established and they can how they can be used to deal with physical
problems we shall see.
150
So, G lambda t minus t prime i t minus t prime this is what we have learned that if t is
greater than t prime the theta function. So, this is the theta function which will take a
value equal to 1 and if not the it will take a value equal to 0. Now I am introducing the
non-interacting ground states and writing it as minus infinity 0 S 0 t C lambda S t 0 S 0 t
prime C lambda dagger t prime.
Now the entire thing has to be written for the right part, which is
G ( λ , t −t ' )=− iΘ ( t −t ' ) <ϕ 0 | S ( − ∞ ,0 ) S ( 0 , t ) C λ ( t ) S ( t , 0 ) S ( 0 , t ' ) C +λ ( t ' ) S ( t ' , 0 ) S ( 0 , − ∞ ) | ϕ 0 >
+iΘ ( t ' − t ) <ϕ 0 | S ( − ∞ ,0 ) S ( 0 , t ' ) C +λ ( t ' ) S ( t ' , 0 ) S ( 0 , t ) C λ ( t ) S ( t ,0 ) S ( 0 , − ∞ ) | ϕ 0 >
So, the total Green’s function it is big no doubt, but you do not have to worry because
ultimately we are going to bring it a very simple form simple looking form and the utility
to physical systems will be shown. So, this a these are the 2 terms which are taken for t
greater than t prime t less than t prime. Now what we can do is that the extreme left and
back end for the first term left end of the first term or rather it is both the terms of the
terms can be replaced by ϕ 0 | S ( − ∞ ,0 )=e −i χ < ϕ 0 | S ( −∞ , ∞ ) S ( − ∞ , 0 ) by the property of
the S matrix that we have written is this. And now exponential i chi as we have shown
< ϕ0 | S ( ∞ , 0 )
that this is equal to <ϕ 0 | S ( − ∞ ,0 )= .
<ϕ 0 | S ( − ∞, ∞ ) | ϕ 0>
151
So, this can be written as a phi a S infinity 0 these 2 can be combined as S infinity 0 and
then we can write it
i
G ( λ , t −t ' )=− [Θ(t−t ') < ϕ 0 | S(∞ , t) C λ (t) S (t ' , t) Cλ+ (t ') S (t ' ,−∞)| ϕ0 > ]
ϕ 0 | S ( −∞ , ∞ ) | ϕ 0 >
− [Θ(t ' −t) < ϕ 0 | S(∞ , t ') C +λ (t ' ) S(t ' , t)C λ (t)S (t ,−∞)| ϕ0 > ]
So, these are the 2 terms and we have introduced the S matrices properly and so, the let
us see the first term. So, this term is
Θ ( t −t ' ) <ϕ 0 | S ( ∞ , t ) C λ ( t ) S ( t , t ' ) C +λ ( t ' ) S ( t ' , − ∞ ) | ϕ 0 >
= Θ ( t −t ' ) <ϕ 0 |T C λ ( t ) C +λ ( t ' ) S ( ∞ , −∞ ) | ϕ 0 >
this is an important step have a look at this and this says that C S ( ∞ ,− ∞ ) involves
operators in 3 time intervals. So, intervals and these are infinity to ( ∞ , t ) , ( t , t ' ) , ( t ' ,− ∞ )
So, the time ordering operator automatically will sort these time intervals, because if we
put the time that is largest will to the left will put it to the left.
So, proper sequencing of the time will be done by the time ordering operator and in
which case we can write it and the second term also can be written it in the similar
manner. So, this term also can be written it for the similar manner and again it will be 3
152
intervals, but; however, we have t prime is greater than t. So, there will be t t prime
instead of that will have t prime t and there will be a t minus infinity.
So, if we combine both my Green’s function will be written as
<ϕ 0 | T C λ ( t ) C +λ ( t ' ) S ( ∞ ,− ∞ ) | ϕ 0>

G ( λ , t −t ' )=− i
ϕ 0 | T S ( ∞ , −∞ ) | ϕ 0 >
In fact it is not required to put the S infinity minus infinity, because the time ordering
will take care of the time that are appearing in the time variables that are appearing in the
problem.
And in which case we can drop the S infinity and minus infinity and we can also sort of
write the wave function without the write the Green’s function without the denominator
as G ( λ , t −t ' )=− i< ϕ 0 | T C λ ( t ) C +λ ( t ' ) | ϕ 0 >. So, this is the final form of the T equal to 0
Green’s function.
So, it says that the operators C lambda t and C lambda dagger t prime are taken the
expectation is taken the time ordered expectation is taken with respect to the non-
interacting ground state, which are known. And one has to multiply it with a minus i
factor and the physical consequence or the physical situation that would it would
correspond to is that if we create a particle given by C lambda dagger at time t prime and
later take out the particle from the assembly of this many particle system at T equal to t.
153
The particle are rather the system will have the information that what kind of interactions
are gone into the inside the system and the Green’s function will provided information
about that. It is more important, now to relate physical quantities to this Green’s function
which is what we will do. We will now do one example of this Green’s function and the
most simple case that is the non-degenerate electron gas at which means it is a non-
interacting system.
So, example so, it is a degenerate non interacting electron gas and let us take a simple
ℏ2 k2
metal at T equal to 0. So, all electronic states for ϵ k = and being the electronic mass
2m
for epsilon k greater than mu are empty and for epsilon k less than mu are filled.
This you must have studied in the first course of solid state physics, which is the model
for a metal and we do not need to have t equal to 0 we will relax this condition. That
since we are talking about the chemical potential we are restricting ourselves to T equal
to 0, when we rather we can talk about the Fermi energy when we talk about T equal to
0, but a in principle this the illustration will carry over to T naught equal to 0.
Now since all the energies are measured from the Fermi level let us define a variable
Define rather and we have for a uniform spherical Fermi surface, we have
1
<ϕ 0 | Ck+ C k | ϕ 0> =<n k > =Θ ( k F −k ) = lim β ΕK +1
that is the Fermi distribution.
β→∞ e
154
So, this is the definition that we have the ground state expectation non interacting ground
state expectation value of the <ϕ 0 | Ck C +k | ϕ 0> =1 −< nk >=Θ ( k − k f ) but the Fermi
function.
And for the other term because we have 2 terms in the Green’s function, because of the
time ordering and this is equal to C k C k dagger phi 0, which is nothing, but equal to 1
minus n k and that is equal to theta of k minus k, f the states which are all empty at t
equal to 0 or t tends to 0.
So, the so the Green’s function, now this Green’s function, because we are talking about
a non-interacting problem will call it an unperturbed Green’s function this word will be
clarified more as we go along is G lambda t minus t prime, it is equal to
G0λ ( t −t ' )=− iΘ ( t −t ' ) < ϕ0 | C λ ( t ) C +λ ( t ' ) | ϕ 0 >+i Θ ( t ' −t )< ϕ 0 | C +λ ( t ' ) C λ ( t ) | ϕ 0 >, because the
operators swap the positions. And this is equal to
G 0λ ( t −t ' )=− i [ Θ ( t − t ' ) Θ ( Εk ) −Θ ( t ' −t ) Θ ( − Εk ) ]
So, the Fourier transform of this of this is G now let us write it because we are talking
about the unperturbed principle let us write it with 0. So
∞ ∞
0
[
G ( ⃗k , ω ) =−i Θ ( ξk )∫ dt e
0
it ( ω− ξ +i η)
k
−Θ ( − ξk ) ∫ dt e
it ( ω −ξ
0
k
−i η )
]
155
Θ (ξ k ) Θ (− ξk )
and this tells me that this becomes equal to t
[ +
ω − ξ k +iη ω − ξk −i η
.
]
Let me write it a little more clearly and it is a omega minus xi k minus i eta.
Θ (ξ k ) Θ (− ξk )
So, this can be written in a compact form as a
[ +
ω − ξ k +iη ω − ξk −i η ]
, where eta k equal
to some η k =η Sgn ( ξ k ) =η Sgn ( ϵ k − μ ) . Which says that whether epsilon k is greater than
mu or epsilon k is less than mu eta will take form, which is plus or minus where eta is
positive number.
So, will have a i eta or i minus i eta depending on whether we are talking about the
particle or the whole will clarify this more as we go along, but this term the term the first
term that is the term on the left is called as the electron propagator and whereas the term
on the right would be called as the whole propagator, because this corresponds to filled
states and this corresponds to empty states.
So, this is equal to plus eta when epsilon k is greater than mu. So, that is the first term
and x equal to minus eta, when epsilon k is less than mu that is the second term. So,
finally, the Green’s function if we introduced spin so, introducing additional properties
156
Θ ( k − k F) Θ (k F − k )
0
such as spin etc. One can write the G α β ( k ,ω )=δ α β
[ +
ω − ξ k +iη ω −ξ k −i η ]. So, in fact,
this correspond to the electron particle term the second term, sorry for messing that up
earlier and the first term corresponds to the whole term which are all empty where k is
greater than k f.
So, at t equal to 0 none of these states are occupied and all the states are occupied, which
are called as the particle states for k f greater than k.
So, this is the Green’s function that we compute for a degenerate free electron gas and
so, it gives the entire energy spectrum of the poles of the Green’s function gives the
energy spectrum for both the particles and the holes and they just differ by of course, the
numerator and in the poles there will fine, whether the complex term actually pushes the
pole up of the real line or it pushes down of the real line we will clarify all these things in
our subsequent discussion.
157
Prof. Saurabh Basu
Lecture – 12
Wick’s theorem and normal ordering
We have introduced Green’s function at 0 temperature and we have also provided an

expression what the Green’s function looks like. So, the Green’s function is a time
ordered product of a creation and an annihilation operator and it has to be. So, this is the
unperturbed Green’s function that I am talking about and expectation value of these time
ordered product of one creation and one annihilation operator has to be computed with
respect to the non interacting ground state which is assumably known.
Now, we will have to understand how to compute full Green’s function that is Green’s
function for interacting systems. In fact, that is one of the main aim of introducing
Green’s function. So, in order to calculate Green’s function for the interacting system
will have to understand one theorem which is very important in this context.
It is called as a Wick’s theorem. So, this was introduced by Wick’s in 1950 and before
we go that go to that of how to use Wick’s theorem in the context, let us write down the
full greens function. So, the full Green’s function and why we are using the word full
158
Green’s function is because we have computed the non-interacting or the unperturbed
greens function.
So, the full Green’s function is written by expanding the S matrix that has been
introduced earlier and the s matrix in this particular case is s infinity minus infinity in a
series and it acquires a form. Now we are writing it a little more specifically for a
translationally invariant system where the momentum k is a good quantum number it is
equal to sum over n equal to 0 to infinity minus one whole to the power n plus one
divided by n factorial.
(−1)n+1 +∞
∞ ∞
∑ ∫ dt 1............. ∫ dt n < ϕ0 | T C k (t)H ' (t 1 )........ H ' (t n) C +k (t ' )| ϕ 0 >

n! −∞ −∞
G( ⃗k ,t−t ' )= n=0
< ϕ 0 | S (∞ ,−∞)| ϕ 0>
Now, it does not matter even if we forget the denominator because it just acts as a
normalization we can ignore that and still we will have the numerator to compute and the
numerator can be written as for a particular case say we go up to t 1, t 2 and t 3 that is we
include 3 orders of the interaction term at a distinct times t 1, t 2 and t 3, then we will
have to compute a ground state expectation of the time ordered
< ϕ 0 |T Ck (t ) H ' (t 1) H ' (t 2) H '(t 3)C +k (t ' )| ϕ 0 >
So, this is an example that we are considering in which case in this series that we have
written above in the numerator there 3 distinct times that are taken t 1, t 2, t 3 and we
have written down the Green’s function the full Green’s function including these 3
terms. So, remember that there will be one annihilation operator at the left and there is
one creation operator at the right and this time ordering will take care of all the
combinations of times of t v, t 1, t 2, t 3, t prime and will ultimately put the time which is
largest at the left.
159
So, let us assume that the interaction term that we are going to consider has a coulomb
form. So, we will write down a coulomb interaction at t 1 is equal to half this is a
coulomb interaction same as e square over r where r is the distance between the 2
particles and; however, it is written now in k space. So, . So, these are the that Fourier tr
1 4 π e2 +
H ' (t 1)=
2
∑ q2 C k+q , σ Ck+' −q , σ C k ' ,σ ' Ck , σ e it ( ξ
1 k+ q
+ ξ k' −q− ξ k' − ξ k )
.
So, that we are writing it in. So, these are the terms for h prime given at a time t 1 and
similarly there will be 3 terms which are h prime t 1 t 2 and h prime t 3 all will have to be
written because it is an interaction term expectedly it will have to creation and 2
annihilation operators.
So, how many creations and annihilation operator we have operators in G each one of the
H primes will give. So, each H primes will give 2 creation and 2 annihilation and how
many of them are there. So, there are 3 of them are there. So, a 3 H prime will give 6
creation and 6 annihilation operators and the Green’s function itself has or the in non-
interacting Green’s function has one creation and one annihilation.
So, these two will have to be added and will have seven creation operators and seven
annihilation operators and each of them will have different times I mean in the sense that
the H primes will have one time t 1 the other h prime will have t 2 and t 3 etc. So, all
these fermionic operators C and C dagger will have will correspond to times t 1, t 2, t 3
160
and then there is a t and t prime which are coming from the non interacting greens
function. So, there are 4 or rather 5 time indices that are appearing here namely. So, they
are the time indices are. So, a time index in these cases a t t 1, t 2, t 3, t prime.
And when we consider a higher order problem that is a, suppose, we have 5 H primes.
So, we go up to the fifth order in order to compute the full H prime the full sorry the
Green’s function then will have more number of time indices we need to be ordered now
keeping track of all these time orderings become a problem and because of that taking all
sort of time combinations that is t greater than t 1 and. So, these are t t 1 t 2 t 3 and t
prime. So, we can have combinations such as t greater than t 1, but less than t 2 less than
t 3 greater than t prime and so on and so forth.
So, this keeping track will be difficult. So, computing a full Green’s function often can
be big problem a problem that is becomes formally quite complicated and messy;
however, things are not that difficult and a solution is provided by the Wick’s theorem
and this is what we are going to talk about, but before we go into Wick’s theorem let us
also see that.
We should have equal number of creation and annihilation operators in a Greens

function. So, this is basic requirement because suppose we write a time ordered of
< ϕ 0 |T C1 (t 1)C1+ (t 1) ......C n (t n)Cn+ (t n)| ϕ 0 >
161
This should only exist an unequal number of a creation and annihilation operators will
yield 0 expectation value now why is that. So, suppose we have unequal numbers that is
there is one C dagger which is extra which is not being compensated for and that will act
on the non-interacting ground state and will create a state which will have 0 expectation
or 0 overlap with the non-interacting ground states.
So, that is why this there has to be equal number of the creation and annihilation
operators further the job of the creation operator is to create a particle in a certain state.
So, creation operator creates a particle in state say m the destruction operator or the we
are talking about annihilation let us call it annihilation though they mean the same thing.
Operator must destroy the state m state n such that m has to be equal to n. So, that the
system is returned back to the ground state and it has finite expectation with respect to
the ground state ok. So, say for example, we have a < ϕ 0 |T Cα (t)C β+ (t ' )| ϕ 0> =0
unless alpha equal to beta.
And if alpha is not equal to beta then it is equal to 0 and if alpha equal to beta this will
correspond to a Green’s function g t minus t prime and similarly another example can be
given which is a larger term inside the expectation term. So,
< ϕ 0 |T Cα (t)C β+ (t 1 )C γ (t 2) C +δ (t ')| ϕ 0 >=0 . So, just to avoid confusion so, this is a
equal to again equal to 0 unless alpha equal to beta gamma equal to delta or alpha equal
to delta gamma equal to beta.
So, it has to returned to the ground state to have a nonzero expectation in the ground
state because if a state is created by acting an operator on the phi 0 if that state is not
destroyed that particular, the new state will not have an overlap with the original state of
the problem or the ground state of the problem.
So, that is why it has to vanish and so, there are in when you have larger number of such
terms inside the phi zeros then you will have a many such orderings that are going to
appear and each ordering will then have to be dealt separately and not only that the main
question still lies that how do we actually compute a Green’s function which is given
here.
So, a Green’s function such as phi 0 the time ordered of C k t H prime t 1 H prime t 2 h
prime t 3 C k dagger t prime and phi 0, how will they have be how will they be
162
calculated and to bring to a form that makes sense. So, then this is done by the Wick’s
theorem and Wick’s theorem uses a concept which is known as normal ordering. So, let
us learn what normal ordering and then will see what Wick’s theorem does.
So, normal ordering; so, normal ordering means that all the annihilation operators will be
placed to the right of the creation operators.
So, this is the process in which you will use the anti-commutation relations of the
Fermion operators and place all the creation all the annihilation operators to the right of
the creation operator. So, all the creation operators will be on the left and all the
annihilation operators will be in the right.
So, take an example and let us call it a p a +q ar a +s where p q r s are some states which
are the some quantum numbers which are valid quantum numbers for a given problem.
So, this can be written as now as I said that all the annihilation operators will be to the
right. So, first start with this a p a q dagger if I have to move a p to the right of a q
dagger.
Then what I have to do is that I have to use the anti-commutation relation which tells me
that it is a a p a +q ar a +s =( δ pq−a +q a p ) ar a +s . So, is this clear that a p a q dagger is nothing,
but delta p q which is if p equal to q if this is equal to one. So, if p equal to q, then it is 1
minus a q dagger a p. So, we have changed the ordering of the indices and I have just to
163
remind you that we have a p a q. So, this anti-commutation relation is equal to is equal to
delta p q.
So, a p a q dagger plus a q dagger a p equal to delta p q is the anti-commutation relation

and so, we have done the first step was the second step second step is simply multiplying
it out. So, it is δ pq ar a +s −aq+ a p ar a +s . Now let us do this procedure again on the first
term and I will have a d δ pq ( δ rs −as+ ar )−a +q a p ( δrs −a +s ar ) .
Now, I will change the ordering of this such that that is pushed to the right that is that the
annihilation operator is pushed to the right. So, it is delta r s minus a s dagger a r. So,
these are the two terms that we get. Now, I will simply again multiply it out
δ pq δrs −δ pq a +s ar −a +q a p δrs +aq+ ( δ ps −as+ a p)a r
So, now we have delta δ pq δrs −δ pq a +s ar −δrs aq+ a p + δ ps a+q ar−a+q a s a +p ar now that is the
complete simplification because in each of the terms wherever there is a creation and an
annihilation operator we have been able to push the annihilation operator to the right and
that tells me that if I take, now the ground state expectation of this term. So, it is phi 0
and this entire thing that we have written here you see the terms which are the first in the
second term the third term the fourth term and the fifth term they all will go to 0.
Because there is all in each of the cases the annihilation operator is standing on the right
and the non-interacting ground state may not have a state which is which contains a state
which is to annihilation. So, this will annihilation state at r, but the non-interacting
ground state may not have a state, but what definitely will not have a state is a vacuum
one thing to remember that phi 0 is not the vacuum.
164
However, we can redefine vacuum redefine vacuum and these phi 0 can be considered as
a redefined vacuum which really does not have a state which is occupied which can be
annihilated.
So, it seems that. So, how do we redefine a vacuum let us give me an example. So,
redefining vacuum I am giving an example of that you have studied the harmonic
oscillator in your quantum mechanics course and the harmonic oscillator has
p2 1 2 2
Hamiltonian H= + mw x .
2m 2
And you know the full story that goes with it that is if you solve this quantum
mechanical Hamiltonian that is the Schrodinger equation with this as the Hamiltonian
then you would get an energy spectrum which is n plus half h cross omega and there is a
0 point energy that exists for n equal to 0 and the way functions are going to be Hermite
polynomial multiplied by a Gaussian where Hermite polynomials are polynomials with
certain properties such as when the indices of these polynomials and even the polynomial
is even and when the indices are odd the polynomial is odd you can look up any quantum
mechanics book for this purpose, now if in this harmonic oscillator. So, we are talking
about simple harmonic oscillator quantum mechanical simple harmonic oscillator.
165
p mw
If you define operators which are like P=
√ mw ℏ √
, Q= ℏ x and in that case you
1
can write down the Hamiltonian H= (P 2+Q 2) these are operators that have been
2
made dimensionless. So, this Hamiltonian is dimensionless you are not seeing any
omega etc h cross omega which are should have come with the Hamiltonian we have
made it dimensionless and there are commutation relations that [P , Q]=−i satisfied
which come from the p and x position and momentum commutation relations in quantum
mechanics.
if you further define the raising and the lowering operators by this accuse and piece such
1 1
that we can write it as a += (Q−iP) and a= (Q+iP) with a and a dagger to
√2 √2
1 1
be commutation of that to be one then my H= (a + a+aa + )=a + a+ . Now if I
2 2
~ 1
redefine this half comes from the 0 point energy and if we redefine a H=H − .
2
Now, we do not have a 0 point energy and we have been able to define the ground state
without the 0 point energy. So, we have now n starts from one. So, we have so, such that
your H acting on 0 gives you a 0 and a 0. So, there is no the 0 point energy has been set
to be equal to 0 and this is what. So, the first a non 0 energy is of course, the h cross
omega and the 0 point energy is 0. So, this is the way one can redefine vacuum. So, if we
redefine vacuum in our case as well then all the annihilation operators acting on the
ground state which is now the vacuum should give me 0.
So, the only thing that is nonzero is this part. So, this is the only thing that is nonzero and
will only thing that is nonzero here and rest all of them are our 0 ok.
166
So, this is what it gives and hence we can write this as

< ϕ 0 | a p a+q a r as+ | ϕ 0 > =δ pq δrs =< ϕ 0 | a p aq+ | ϕ 0 > < ϕ0 | ar a +s | ϕ0 > . Now these are the greens
functions.
Accepting that we have not written the time indices explicitly, but these will give the
greens functions or they will give the or number operators if the time indices are same
will come to this. So, these are not only these are greens function, but these are non-
interacting greens functions and these one example of this non interacting Green’s
function we have done why we attended to that example of degenerate electron gas free
electron gas.
So, the nutshell is that the total Green’s function or the full Green’s function will have to
be expressed in terms of the non-interacting Green’s function and it will be done the
normal ordering. Now, the time ordering will be taken care of by the Wick’s theorem and
let us show how so, now, application of Wick’s theorem.
You might wonder that we have not defined Wick’s theorem, but we are talking about
applications of the Wick’s theorem we are just going to come to Wick’s theorem, but
seeing one application how it works will probably be able to for you to realize better
what those words mean when we talk about it. So, the time ordering in the full Green’s
function has to be done in a simple way.
167
So, keep in mind that we have equal number of creation and annihilation operators. So,
let take an example which is what we have said is
< ϕ 0 |T Cα (t)C β+ (t 1 )C γ (t 2)C +δ (t ')| ϕ 0 > this is really a hypothetical situation because
you will never have 2 creation and 2 annihilation operators ever because you will have a
H prime which itself will come with 2 creation and 2 annihilation operators; further the
Green’s function that is that requires one more creation and annihilation operator. So, at
least you will have 3 creation or 3 and annihilation operators.
For the first order Green’s function to be calculated we are just taking a hypothetical
example now what we will do is that we will have to write to terms with the normal
ordered product. So, will make them write into non interacting greens function. So,
assume the normal ordering is done and we write a C beta dagger t 1 and a phi zero. So,
the first 2 terms have been paired and the third and the fourth terms have been paired
well will have to write it properly. So,
< ϕ 0 |T Cα (t)C β+ | ϕ 0 > < ϕ 0 | (t 1 )C γ (t 2)C +δ (t ' )| ϕ 0 > remember each of these terms that
you see there are they are non-interacting greens functions.
And now there is a minus sign you will see why because you will have to take this C
gamma one move and a second move. So, 2 moves 2 swaps of this C delta dagger will
create 2 negative signs which will be positive, but; however, I will also take the
< ϕ 0 |T Cα (t)C β+ (t 1 )| ϕ0 > < ϕ0 | (t 1)C γ ( t 2 )C +δ (t ')| ϕ 0 >
- −< ϕ 0 |T Cα (t)C δ+ (t ') | ϕ 0> < ϕ 0 | (t 1)C γ (t 2 )C β+ (t 1 )| ϕ0 >
Remember there will be no terms which are consisting of both of them being annihilation
operator or both of them being creation operators because that will have 0 expectation
value.
168
Anyway so, let us come back to this in a while and let me complete this thing.
So, this will be equal to δα β δ γ δ < ϕ0 | T C α (t) Cβ+ (t 1) | ϕ 0> < ϕ 0 | (t 1)C γ (t 2) Cδ+ (t ') | ϕ 0>
- −δα δ δ γβ < ϕ0 | T C α (t) C +δ (t ' )| ϕ 0 > < ϕ 0 | (t 1 )C γ (t 2) Cβ+ (t 1) | ϕ 0>
Once you get to the habit of writing things will be very smooth and you know the rules
of changing signs and writing these things. So, I get 2 delta functions coming from 2 of
these brackets and so on. So, what happens is that I have to now multiply because there
169
are 2 greens functions if I multiply a the whole term as. So, I have a minus i square
because the Green’s function if you remember comes with the minus i.
So, (−i)2 < ϕ 0 | T Cα (t)C β+ (t 1 )C γ (t 2)C +δ (t ' )| ϕ 0 >= δα β δ γδ G0 ( γ , t−t 1)G 0 ( γ , t 2 −t ') -
−δα δ δ γβ G 0 ( γ ,t−t ')G 0 ( γ ,t 2−t 1) .
So, even if it is hypothetical this term is hypothetical.
We still have been able to simplify it and written it in terms of 2 greens functions along
with this delta alpha beta and delta gamma delta these delta functions and 2 greens
functions and similarly 2 delta functions and the greens functions. So, this is Wick’s
theorem, let us go back and see what Wick’s theorem are and we have included a
summary I will go through it slowly and try to explain it. So, that you understand these
concepts clear these will be required in computing the Green’s function the full Green’s
function for an interacting system and which is what we plan to do next. So, it is in
nineteen hundred and fifty it was put forward by Wick’s and some mathematical
technique to compute the interacting Green’s function which is the full greens function.
In terms of the non-interacting Green’s function by considering all possible contraction

between pairs of operators so, these are the all possible one possible is C alpha C beta
dagger and C gamma C delta dagger the other possible combination is C alpha C delta
dagger and C gamma C beta dagger; so, considering all possible contraction between
pairs of operators.
Now you see that more number of H primes that you have more will be the number of
combinations that you need to take the idea is to rearrange the order of factors in the
operator products which are between the ground states using the usual commutation or
anti-commutation rail rules in the context of Fermions they are anti-commutation and in
the context of Bosons they are the commutation rules.
So, that the annihilation operator always stands on the right the expectation values this
will be. So, when you write these operator products these are the normal ordering of
course, required that they have to be the annihilation will have to be on the right, but
when you write down greens functions then the creation operators will have to be. So,
this has to be replaced by the creation operators the expectation values of the rearrange
terms which is between the ground states which are phi zeros with respect to the
170
unperturbed ground states then vanish ok. So, this is telling you about ok. So, I change
this because this talking about the normal ordering once more.
So, let us talk in a sort of systematic manner. So, this talk about the normal ordering, but
of course, the process introduce additional terms which survive to yield a nonzero and a
meaningful result which is what we have seen here that we have a nonzero result which
is meaningful which gives rise to 2 Green’s function is what we have done these 2 greens
functions. So, this is what the process does.
So, this is that normal ordering now for from any arbitrary time order product of creation
and annihilation operators construct the product of the contractions. So, you contract
them which means you swap their positions in a physically meaningful and
mathematically meaningful way obtained by pairing of all the terms in a particular way
thereby forming a fully paired product you see we have formed a fully paired product.
So, there is a C alpha C beta dagger and there is C gamma C delta dagger and similarly
the other term repeat the process by pairing of the terms in a different possible way to
form another fully pair product and continue this process until the operators have all
been paired off in all possible ways to keep matter simple.
We have shown you that how these pairing takes place in for only 4 fermions operators,
but as I told for Fermion operators is never possible in a full Green’s function you should
have at least 5 Fermion or rather 3 Fermion operators and 3 3 annihilation operators and
3 creation operators which means 6 Fermion operators is minimum what we need for a
Green’s function and so, will have to repeat the process of pairing.
So, this is that statement that you repeat the process by pairing of the terms in different
possible way to form another fully paid product and continue this process until all the
operators have been paired off in all possible ways Wick’s theorem says that the ground
state expectation value of the original operator is the possible combination of all the
product pair products obtained in this way.
So, this Wick’s theorem does this time ordering in a simple manner and the normal order
product gives only a few possibilities that could survive and by putting all the
annihilation operators to the right of the creation operators there by making most of these
terms to be equal to 0 and only a few to survive .
171
So, to know the rules of applying Wick’s theorem is that you form a time ordered
product of the operators then go to the normal ordering. So, which means that put all the
annihilation operators to the right of the creation operators now you need to take into
account a sign change each time a Fermion operator changes it is position. So, one
should keep the count of the number of interchanges needed to achieve these desired
pairing.
So, we are now ready to write down the full Green’s function for a given problem and
we shall do it; for the simplest case which is coulomb interaction for a for an electron
system so, a system of fermions.
172
Prof. Saurabh Basu
Lecture – 13
Green's function and Feynman diagrams
So, let us have a quick review on the topics that have been discussed, so, far in the
context of Green’s functions. So, the Green’s function consists of time order product of a
creation and an annihilation operator followed by n interaction terms, which are all
written at distinct times that is the first point. So, this was just to say that it is of the form
of a G( ⃗k ,t−t ' )→< ϕ 0 | T C k (t)Ck+ (t ' ) H ' (t 1) ..... H '( t n )| ϕ0 > this is the expectation
value of the time order product is taken.
So, this is 1 annihilation term and the there is a creation term and this thing equal to. So,
this along with a minus i is same as is same as G k let us say that let us make this same.
So, it is G k a and p minus t prime.
So, this is what it is proportional to with a factor of minus i will see all that in a short
while. Now each order of H prime included here in this expression will give a sort of that
order of the Green’s function. So, if I include 1 H prime I will get a first order Green’s
function, if I include 2 H primes I will get a second order Green’s function and so, on.
173
So, these are in principle we have infinite number of these H primes. So, a full Green’s
function contains infinite number of H primes, what are the implications of a each of
these H primes is what we are going to come next?
So, a fully interacting Green’s function contains a 1 creation and 1 annihilation and n of
those H primes. So, each of the third point is each of the interaction terms H prime itself
is a collection of to creation and to annihilation operators. This is what we have learned
when we are dead second quantization, that each interaction term at least as 2 creation
and 2 annihilation operators .
So, there could be higher order of interaction terms, but we are mostly going to be
talking about interaction terms, where 2 particles come and interact and they exchange
their you know spin momentum etc and they go on as 2 different particles there could be
you know production of particles as a result of scattering, but let us not talk about that
immediately will sort of only be constrained with each of these H primes having 2
creation and to annihilation operators.
So, H prime has C dagger C and C dagger C with a momentum and spin indices as case
requires. So, the expectation value of the time ordered product is taken with respect to
the non-interacting ground state, this is what we have learned it is actually the ground
state of the many body system, but then we learned how to construct a non-interacting
ground state from an interacting many body ground state, and we have shown that there
is a phase factor that comes, which is which connects the many body ground state as well
to the to that of the non interacting ground state. And properly done we can simply take
the time ordered product to be taken between or the expectation to be taken between the
non-interacting ground states.
All the operators point number 5 all the operators are written in the interaction
representation. This is important, because these Green’s function is fully in the
interacting interaction representation, in which both the wave functions and operators
they carry the time dependence. Somehow it is only important for us that the operators
carry the time dependence, but nevertheless we should write those operators the creation
and the annihilation operators to be in the interaction representation.
174
In fact, the difference between the interaction representation and the Heisenberg
representation is that that the Heisenberg representation has similar time evolution as that
of the non-interacting system for the interaction representation.
So, now the non-interacting so, the interacting Green’s functions are written in terms of
the products of non-interacting Green’s function at the number operators. This is done by
doing a normal order product and each of the or rather the interacting of the full Green’s
function is split up into several of the non-interacting Green’s functions. And all the
number operators and then of course, the interaction vertex what we mean by that is,
what we are going to come next the interaction vertex, will appear and or the strength of
the interaction will appear and the fully Green’s function will contain information about
the nature of the interaction vertex or their strength of the interaction?
Such products are most conveniently computed using Feynman diagrams. So, we are
next going to discuss Feynman diagrams, but before that let us write down 1 example of
a Green’s function and see that what it means or how it can be split up decomposed into
the non-interacting Green’s function?
So, we start with an example and let us write down apart without those factors of minus i
etc will simply write down a time order product and this means this left angular bracket
means that there is it is taking with respect to phi naught 1 can write it also explicitly, but
it is understood.
175
So, it is a <T C k (t)Ck+1 (t 1) C +k 2( t 1 )C k 2+q (t 1 )C k 1−q (t 1 )C k+ (t 1 )>
So, let us just write it as so, there will be nothing here and there will be nothing here, but
as I said the angular brackets denote that we are taking the expectation with respect to
the non-interacting Green’s function. Now what we have done is that we have simply
written this as the simply this is the H ' (t 1) . So, we have written down a first order
Green’s function containing one H prime 1 annihilation operator here and a creation
operator here.
So, that is our, that is the term that we need to calculate. So, let us write down or rather
number these operators so, that we can take the proper combination in order to write the
non-interacting Green’s function. And so, this is 1 this is 2 let us let me write a different
color for that. So, that. So, this is 1, this is 2, this is 3, this is 4, this is 5, and this is 6.
So, what are the combinations allowed? So, possible combinations so, a will be a is 1 2 3
4 and 5 6 that is combination number 1. So, we can take a combination of 1 and first and
second term third and fourth term and 5th and 6th are also we can take first and third
term.
So, the combinations are being formed by 1 creation and 1 annihilation operator. So, 1 3
2 4 and 5 6 so, we have a third combination, which is 1 3 2 5 and 4 6. So, that is the third
combination and d to be 1 2 3 4 or rather 3 5.
1 2 3 5 and 4 and 6 fifth combination is 1 and 4 3 and or 2 and 4 1 and 6 sorry 1 and 1
and 6 2 and 2 and 5 and 3 and 3 and 4, so, 1 and 6 2 and 4 and 3 and 5.
So, this is the fifth combination and there will be another 1 there will be another one,
which is 1 6 and 2 5 and 3 4 ok. So, these are the 6 combinations and the 6 combinations
are coming, because of we have 3 operators 3 creation operators and 3 annihilation
operators. So, there are 3 factorial combinations possible. So, all these 6 combinations
for just for the first order Green’s function have to be evaluated.
So, let us first write down each one of them and specifically we shall take some time and
treat this as a tutorial such that you learn, these decoupling and writing it writing the full
greens function the first order Green’s function as in terms of the non-interacting Green’s
function.
176
So, the first term is according to the combinations prescribed 1 2 3 4 and 5 6. So, will
take a combination of C k t C k 1 t 1 then C k t 2 dagger and C k a 2 and then C k 1
minus k, and then C k dagger remember that that is not going be the Green’s function
unless you multiply it is going to be a non-interacting Green’s function if you multiply it
with a factor of i, because the Green’s function itself is defined with a factor of minus i.
So, a becomes equal to + + +

<T C k (t)Ck 1 (t 1)> <TC k2 +q (t 1)C k2 (t 1)> <T C k 1−q (t 1)Ck (t 1)>
and these are the 3 combinations that I get.
And so, I will have to multiply each one of them by a negative sign and also do not
forget to bring in the negative sign that comes, because of the swapping of the operators
that is very important. So, setting that aside for the moment we would write this 1 as
δk k G0 ( ⃗k , t−t 1) ok. So, just remind you that a factor of i is what we have dropped so,
1 1
far, but will get it back once when we do more thorough calculations. So, this is the first
term. So, your k has to be equal to k prime that is the conservation says and this is a G 0
with order Green’s function which is a function of k and t minus t 1.
Now, this one necessarily demands that because k 2 plus q has to become equal to k 2,
which means that this is necessarily a δq =0 η( ξ k ) . And now this is going to be a
number operator, because it is at same time at time t equal to t 1. So, it is not a
propagator or a Green’s function that propagates the system from sometime t 2 sometime
177
t prime. So, it just loops back to the same time. So, will write it with a n and this xi k,
where xi k could be epsilon k minus mu where epsilon k is this other single particle
energies.
Similarly so, this is the second term and similarly the third term is here. So, that is again
equal to δq =0 G0 ( k 1, t 1−t ') . This is how we have now written it the full Green’s
function at the first order in terms of the non-interacting Green’s function G 0 and a
number operator which is n of xi k and another G 0, which is there? And how to compute
this is what we will see and that is where the Feynman diagrams play a role, but before
that let us writes down all the terms. The second term so, again I want to remind you that
I am treating it as a tutorial you should do each and every term by hand and satisfy
yourself that this is what is coming?
Because this forms the basis of what we are going to do next with the Feynman
diagrams. Similarly for b we have
<T C k (t)Ck+2 (t 1)> <TC k2 +q (t 1) C +k1 (t 1)> <T C k 1−q (t 1) Ck+ (t ')> .
So, this tells that it is equal to δk k G0 ( k⃗1 , t−t 1 ) . And similarly this one will have
1 2
δk =k +q η ( ξk 1) and this will be simply a G 0 and k 1 and again this will give me a so,
1 2
this is from t 1 to t 1, which means it is time that if the time is repeated provided I have
written it correctly yeah that I have written it correctly.
So, this will be, but this is not at and given q equal to 0. So, this will be like G 0 and k 1
and this will be like a t 1 k 2 plus q will come back to this. And will have a
δk −q , k G (1) (k , t 1 −t ') . And so, these ones so, let us write down the number C,
1
<T C k (t)Ck+2 (t )> <TC k 1−q (t 1 )C k+1 (t 1)> <T C k 2+q (t )C +k ( t ')> and so, on.
So, this tells me that this is equal to delta δk k η( ξk ) . So, this is δq =0 η( ξ k1 )

1 2
So, this is δk 1
, k 2+q G(0) ( k , t−t ') . So, these are the 3 terms there and similarly for d, we
have <T C k (t)Ck+1 (t 1)> <TC k1−q (t 1 )C +k 2 (t 1 )> <T C k2 +q (t 1) Ck+ (t ')> and so, on.
So, this is equal to δk , k G(0) (k , t−t 1) once again to remind you that this G 0 will have
1
to add an i. So, that without that it does not become G 0 we are writing it loosely without
178
that factor of i. So, this is k t minus t 1 and now this is of course, will have a delta k 1
minus q, which is equal to k 2. So, this is k 2 and this will give me a so, just like the
δk −q , k η( ξk 2)
1 2
Similarly this will also be k 2. So, this will also be let me correct it here this was written
mistakenly. So, this will be like n and xi k 1. So, similarly it will be xi k 2 and then this
will be like δk , k +q G(0 ) (k , t 1−t ' ) . So, that is the fourth 1 and similarly the fifth 1 can
2
be written as something interesting about the fifth 1 which will come.
So, we have <T C k (t)Ck+ (t ' )> <TC k 2+q (t 1)C +k1 (t 1)> <T C k 1−q (t 1) Ck+2 (t 1)> . So, these
are the 3 terms this term will give me a delta k. So, this is anyway k. So, delta k k and
there is nothing, but there is a G 0 which is at k 2 t minus t prime.
Similarly this will b G 0 (k , t−t ' ) δk 1,

k 2+q η ( ξk 1 ) δ k 2,
k 1−q η ( ξk 2 ) anyway. So, these are 5
terms and to write down the sixth term we have
<T C k (t)Ck+ (t ' )> <TC k 2+q (t 1) C +k2 (t 1 )> <T C k 1−q (t 1)C +k2 (t 1)> and again this is equal
to a G0 (k , t−t ') there is nothing to write here. So, let me it is this part.
And right this one has to be written as a delta this is k 2. So, this is equal t
δq =0 η( ξ k2 ) δq=0 η( ξk 1) and so, on.
179
So, it is 3 5 3 and 5 k 2 and k 1 minus q. So, this is actually k 2. So, this is the last 1 is 3
5 3 is t 1 this t 1 this k 2 and k 1 minus k 2 3 is k 2 and this k 1 minus q this 2 ok. So, this
is k 1 minus k 2 equal to. So, so this is anyway.
So, these are the terms that are that need to be written down including the 0th order
Green’s functions. And now if you write down second order Green’s function you will
have many more terms and we can write down and then there will be a second order
Green’s function will have how many terms a second order. So, how many creation and
annihilation operator to H prime will give me 4 4 8 4 creation operators and for
annihilation operators and 1 will come from the Green’s function definition. So, there are
5 creation and 5 annihilation operators. So, this will contain 5 factorial terms.
Which means that equal to 5 into 4 into 3 into 2, which is equal to 20 into so, it is 120
terms that will be there a second ordered itself and the third order you can understand
that it will go as. So, third order will have. So, 67 so, 7 factorial number of terms so, the
question is that how actually to compute all these terms, even if we say want to write
down the first order Green’s function that is containing 1 H prime.
We got 6 terms and maybe some terms are 0 we do not know as yet, but even then the
nonzero terms how to compute them that is the question and the computation of these
terms are done by a Feynman diagrams.
180
So, this is will introduce Feynman diagrams here and tell you the basic features of the
Feynman diagram. So, so Feynman introduced this idea of representing the terms in
these above discussions or above expressions by diagrams and these diagrams are
actually very very important in a number of physical processes, which are themselves I
mean interesting scattering problems so, the rules of the diagram.
So, rule number 1 is that a solid line with an arrow pointing from smaller time t prime,
which is a smaller time to t, which is larger time for free electron Green’s function with a
momentum information ok.
So, basically that tells that this is like this with a t prime going towards a t and there is a
momentum information. So, this line carries a momentum p. So, the system or rather the
free electron Green’s function propagates the free system from a smaller time t prime to
a larger prime t and a corresponding to a given momentum p.
So, this is written as this is equal to G0 ( p , t−t ' ) , this is the notation for that this is for
the electron Green’s function, what about the phonon or the photon Green’s function?
Let us just write it for the photon or phonon. So, a dashed line with no arrow for the
phonon Green’s function and no arrow; because phonons can move in either direction on
the line, that is in time.
Sometimes this so, it is this line and there is a t prime to t and similarly there is a say a q
momentum will not put a arrow, sometimes this is also written it with a springy line like
a spring.
So, this is t prime and t it is also written like that and it is written as D0 (q ,t−t ' ) . So,
these are the electron and phonon non-interacting Green’s function third very
importantly.
181
The electron density η ( ξ p ) is shown via a loop to imply that it starts and ends at the
same time.
So, it just shows like this we show it by an arrow this is p and this is time it starts from
the same time, I mean it starts from time t and comes back to the same time and this is
written η ( ξ p ) this is what we were writing earlier.
Forth any interaction term v q is represented by a wiggly line so, such as this. So, this is
it may carry a momentum q. So, this is the v of q v as a function of q it could have a
energy information, which means that it will be a v q omega. So, for example, we can
write down 1 of the terms in this which had, so a. G0 ( p)v (q) G0 ( p−q )G 0 ( p) will be
shown as a line and then there is a wiggly line.
So, it comes from we have not written the time information, but the time is actually t2 t -
t1 and t 1 - t 2 etc. So, these are this and so on. So, this is say p this is p - q and this is
again p. So, an electron was coming with momentum p and it is scatters of by a potential
v (q), which takes away a momentum q here. So, that is why the wiggly line is
represented by wiggly line is the v (q) which carries a momentum.
Then so, this is the this the leftmost side is the propagated here the v (q) is by the wiggly
line and p minus q is the propagated here and then there is a propagator there.
182
So, the total Green’s function or the total expression for the interacting Green’s function
at the first order is written as a G 0 and then there is a wiggly line and then there are you
know sort of is a propagator from p sometime, t 2 t 1 or. So, this is like t 2 or t’, which
we have not put it here so, maybe t 1 and then maybe t 2 and then maybe t and so on. So,
these are the way to write down this each one of those terms, that we have got.
And. So, rule number 5 is that at the vertex at the vortex, there is a momentum and
energy conservation.
183
So, what I mean by vertex is these points that is a vertex and this is a vertex there are
energy and momentum conservation. So, what we have to do is that? So, integrate over
all the intermediate momenta and some over all integrate over all.
So, integrate over all intermediate let us not write it as momenta as energies and some
overall intermediate. So, basically it is a; you have to integrate over all let us not
introduce this word some and over all intermediate momentum.
So, that is the rule number 5, that is rule number 5 and the rule number 6 immediately it
will not be clear, but nevertheless will put it any way multiply by a factor of i to the
power n, because each Green’s function to convert a line into a Green’s function or each
of this time ordered product into a Green’s function we need to multiply it with i.
So, each of the terms will have to be multiplied with i and hence n of those terms will
have to be multiplied with i to the power n. And then we will have to also multiply it
with a minus 1 whole to the power F, where F are the number of closed Fermion loops.
So, these are the number operators. So, if we have 1 number operator then will have to
multiply it by minus 1 to the power 1, if there are 2 of them then it is there will be no
sign because minus 1 whole to the power 2 it will have to be multiplied by and so, on.
So, these all these rules put together from 1 to 6 as written there will help us in writing
down these Green’s functions the fully interacting Green’s function in terms of the non-
interacting Green’s function and then by summing over the all the intermediate
momentum and energy will get the finally, the interacting or fully interacting Green’s
function.
So, let us write down at least 1 term we have written down 1 term already let us write
down another term, which is which was written down by us.
184
So, example is time ordered
V q <T C k (t) Ck+2 (t)> <TC k 1−q (t 1 )C +k 1 (t 1)> < T C k2 +q (t 1) C +k2 (t ' )>
So, this term can be written as a line which will go like this from a t prime to t and then
there will be a term which is and there is an interaction term. Of course, that we are not
writing it explicitly, but the H int will come with a v q here and this. So, this will be here.
So, this is a because this will correspond to δq =0 . So, this interaction term corresponds
to V q=0 and this corresponds to a. So, this is 1 Green’s function, taking it from some
t prime to some intermediate t 1 and this another.
So, this is that Green’s function. So, this is that Green’s function here, this is that again
that Green’s function which is here and this is that term which is here ok? So, this is the
so, this is a fermionic bubble. So, there is 1 fermionic bubble. So, will have to multiply it
by minus 1 to the power 1 and since there are 3 such terms will have to multiply it by i to
the power n which is what we have prescribed i to the power n. So, it is i to the power 3
and rather there are 2 Green’s functions. So, will have to write it with i square and there
will be a minus sign coming because of this.
So, this is momentum is k here momentum is. So, this does not carry any momentum. So,
this is and this has momentum xi k 1 and this is this continues with the momentum k. So,
a free particle Green’s function propagates from minus t, I mean from t prime to t 1 with
185
a momentum k, then there is a interaction vertex which carries no momentum it is
connected to a fermionic bubble and then it carries on with.
So, the Green’s function or the propagator carries on with a momentum k from time t 1 2
t, will do more examples of Feynman diagrams and will in the next discussion will try to
compute them by computing the internal summing over the internal moment and the
energies will calculate and for a specific case.
186
Prof: Saurabh Basu
Lecture – 14
Feynman diagram
In the previous class, in the discussion that we had regarding the Feynman diagrams, we
had just learned how to write down fully interacting greens function at any order, in
terms of the non-interacting greens function. And then preliminary discussion was
conducted to write down the Feynman diagrams. So, draw the Feynman diagrams and
the rules were stated out now will do the same problem once again, will read on that
exercise and write down each one of the terms in the expansion of the greens function in
terms of the Feynman diagrams.
Just to let you know that we have this one <T C k (t) H ' (t 1)Ck+ (t ' )> . This gives rise to
a first order greens function. And similarly, this <T C k (t) H ' (t 1) H ' (t 2)C k+ (t ')> gives
rise to a second order and so on. This has to be dagger t prime and this is second order
greens function and so on.
So, let us write down at least a first order greens function, and let us write it down the as
we have seen earlier. Let us just write down a term; which is let us write down the first
187
order itself. And so, my H prime gives some so,
1
H ' (t 1)= V q C +k 1 (t 1)C k+2+ q (t 1 )C k 2 (t 1)Ck+1−q (t 1 ) .
2 k∑
, k ,q
1 2
So, all are the all the creation and the annihilation operators are at time t 1 as written here
the half factor is included to avoid double counting. And V q is the strength of the
interaction term, which can have some dependence on q as we shall see. So, then the
greens function can be written as now I will write it because it is a first order greens
function I will write it with a g, with a superscript inside the bracket as one. now I will
leave this bracket to be filled in later , where there will be terms such as is some powers
of i’s that will come in, i means the square root of minus 1. And maybe other factors
such as maybe minus 1 etc that will also come in.
But will try to take that into account. So, this is the minus infinity to plus infinity. And
there is a the internal time has to be integrated over. And now I have a k 1 and k 2 and q.
And I have a C k time ordered of C k. So, when I write this now, it means, there is a phi
0 which is the non-interacting ground state. So, this is implied so, it is a C k of t, then
there is a V q. So, we could have written V q here as well.
so, we can write down the
(1)
∞
Vq +
G =∫ dt1 ∑ <TC k (t ) C k 1 (t 1)C k+2+ q (t 2)C k+2 (t 1) Ck+1−q (t 2 )C +k ( t ' )>
−∞ k 1 ,k 2 ,q 2
188
So, this is the first order greens function. And this first order greens function as we
because there are 3 creation and 3 annihilation operators.
So, there are 3 factorial terms, which means 6 terms that will be there. And let us just
write down the terms. So, the first term will call it as so, G 1 and this corresponds to a, as
was written earlier according to certain type of combinations. And as I said that there is
some bracket, that I am leaving it out and will fill in later
∞
−V q
G(1), a= ∫ dt 1 ∑ <TC k (t)C +k 1(t 1 )> <TC +k 2+q (t 2 )C +k 2 (t 1 )> <TC +k 1−q (t) C +k (t ' )>
−∞ k1 ,k ,q
2
2
So, these are the 3 terms so, this tells that this is equal to so, δk , k 1
δq =0 δ k =k , q=0 ;
1
now there is an inconsistency that I am saying. So, will redefine and the greens function,
give me a moment, let us redefine the greens function as C k k 1.
So, this could be if we redefine the greens function as the interaction term as so, this is k
1, there is a k 2 + q and there is a k 2, and there is a k 1- q. So, that is coming here k 1 -
q. So, then I will have a k 1 so that, and then there is a k2 + q then there is a k1- q ok. So,
if we redefine the interaction term as this is C k 1 dagger and k 2 + q and k 2 and k 1 - q.
So, which means that 2 particles are coming, one with a k 1 momentum and the other is k
2 plus q.
189
And then they interact with a vertex which is V q, and then they go over as a k 2 and k 1
- q. So, the interaction vertex that retains you q will. So, we have just simply redefined
the momentum of the creation and the annihilation operators. And in that case the first
order greens function looks like this. And now I will have a term which is not so, this is
equal to k 1. And then there is a k 2 + q, and then there is a k 2 there. So, let me write it
down little go neatly. So, this will be t, and then there is a C k 2 + q t 1, and then there is
a C k 2 dagger t 1, now it is fine.
And then there is a C k 1 minus q and then there is a C k. So, this is equal to k 1 equal to
k and q equal to 0. This is delta q equal to 0, and this is k equal to k 1 instead of k 1 + q.
So, that is the first term, and this can be written as will again leave this bracket, and there
is a minus infinity to plus infinity. And there is a
∞
(1), a
G = ∫ dt 1 ∑ V q =0 G0 (k ,t−t 1) η ( ξ k2 )G0 (k ,t '−t 1 )
−∞ k2
, now this one if you see that it is at the same time, and because q equal to 0 it become C
k 2 C k 2 dagger.
Now, what can be done is that I can change the sign of this, and can write it with a
negative sign and can write it with a number operator which is equal to a xi k 2. And
then it is equal to a G 0 k. So, these zeroes were written earlier as in the so, this; and now
we have k k and t prime minus t 1. So, how would the diagram look like? So, this
diagram this particular diagram would look like there is a line which is from t prime to t,
there is a momentum which is k, now at t equal to t 1 I have an electron bubble which is
momentum k 2, and it at t equal to t 1, and then it continues as k. So, you see that this
term; let us write it with a different color. So, this term is here this term is this bubble,
and then this term is here ok. And the V q equal to 0 so, this is a q equal to 0, q equal to
0, and that is the so, this is V q equal to 0 is the wiggly line.
So, we have 2 greens functions as non-interacting greens functions. There is a number

operator; or rather there is a bubble electron bubble which corresponds to this n xi k 2.
And then there is a G 0 ; which is again the unperturbed or the non-interacting greens
function, which is taking it from t 1 2. So, this is equal to so in fact, so, there is a t prime
to t 1 is actually so, we should do it this labeling properly. So, this is these labeling has to
190
be. So, this corresponds to the green term, that is the last term then this corresponds to
the first term, and then there is a bubble.
So, this is the full description of this, and we will have to evaluate in order to evaluate
this diagram we will have to compute the integral that is appearing which we are going
to do in a short while from now, but let us all I mean write down all these terms which
we see here.
Then of course, there is a term let us call a G 1 a one
(1)b
∞
Vq
G = ∫ dt 1 ∑ <T Ck ( t)C +k 2 (t 1)> < T C k2 +q (t 1)C +k1 (t 1)> <T C k 1−q (t 1)Ck+ (t ' )>
−%Infinite 2
So, this corresponds to a greens function, this corresponds to another non-interacting

greens function. And so, there are 3 non-interacting greens function, and this term can be
written as so, this term is equal to so, there is a line which goes from t prime to t.
And so, there is a so, there is a time t 1 and so, this is k, and this is equal to k1, and this is
equal to k again, and there is this term that goes from so, this is really this the term that is
there. So, I mean this is really t1; which means that these 2 t is though they look different
they happen at the same time. So, there is an instantaneous interaction, and the vertex
carries a finite momentum q.
191
And so, again let us do this color; so, this color the green color that we see is a non-
interacting greens function that propagates from t prime to t 1 with a momentum that is
case because k equal to k1 - q. And then there is a greens function that propagates from
so, then there is a red one that is here, that is so, that talks about a greens function; which
goes from t 1 to t 1 , but with a momentum that is equal to k 1. So, k 1 equal to k2 + q,
and similarly the other one has this one is here; which goes from t 1 to t, and the vertex V
q, now carries a q a momentum q which is not equal to 0. So, this is b so, let us take the
third term.
So,
(1)c
∞
Vq + + +
G = ∫ dt 1 ∑ < T C k ( t)C k 2 (t)> <T C k 1−q (t 1) Ck 1 (t 1)> <T C k 2+q (t )C k ( t ')>
−∞ 2
So, that tells me that this gives me δk , k , this of course, gives me that this is δq =0 .
2
And this is δk 1 , k 2+q . So, that is that is the, those are the 3 terms. Now this is written as
again a line like this and then. So, there is a this term is like the first; term where we have
a t prime and a t here, and now this is propagating with k, and this is again propagating
with k, this is that V q equal to 0. And this is that k1 which is happening at t equal to t 1.
192
So, that is the so, again the last term corresponds to the greens function the free greens
function in the. So, showing it once more this is here, this is here, and this is this one ok.
So, that is the, those are the 3 terms that appear here.
And likewise, we can write down G 1 d we are doing it all of them (Refer Time: 22:00)
to all of them separately such that you get a feel of these things happening there.
∞
So, G(1)d =∫ dt 1 ∑ V q <T C k (t) C +k1 (t 1 )> <T C k 1−q (t 1) C +k2 (t 1)> <T C k 2+q (t)C k+ (t ')> .
−∞
Again, that this gives me δk , k , this gives me delta. So, k 1 so, this is equal to this
1
gives me delta k and k 2 plus q k 2 plus q because k 1 is equal to k. So, and this is equal
to delta of. So, this k 1 minus k 1 minus q equal to k 2 k 1 minus q is same as k 2. And
this is δk 2+ q ,k and this is again, written as t prime t there is a vertex which is with q
and this is k 2 and this is t 1 and this is t 1.
So, this the last term is a greens function on the on the left and then there is a term which
is then there is a term which is the vertex is carrying a momentum q. And there is a free
electron that propagates from t 1 to t 1 with momentum k 2. And then it again carries on
with a momentum k from t 1 to t that is those are the 3 terms.
193
Similarly, let me write down for completeness G 1, e now this is something that is
important you should see it. So, then this is equal to again that bracket, and then
∞
G(1)e = ∫ dt 1 ∑ V q <T C k (t 1) C +k (t ,)> <T C k 2+q (t 1 )C +k 1 (t 1 )> <T Ck 1−q (t 1 )C +k 2 (t 1 )> .
−∞
So, this tells me that so, there is a free greens function that propagates with a momentum
k from t prime to t. This one tells me that there is a δk 2+ q ,k 1 .
This one says that, δk 1−q , k 2 . This 2-momentum conservation means the same thing.
So, this can be drawn as this. Now this is a disjoint from so, there is one that propagates
from. So, it is like this, and then it is like this. So, there is one that propagates with k 1.
And there is a then it propagates with we can do it. So, there is this way this propagates
like this. And then there is a wiggly line Vq which is equal to here. So, this is k2, this is t
prime, this is t. And this thing is happening at t1, but these 2 are disjoint, you see that,
there is a the greens function that takes from k1 to k2 + q.
So, this is that k1 to k2 + q, and then there is a k 1 minus q going back to k 2. So, this is
that the 2 greens functions are shown on top and bottom of that shell-like structure and
the wiggly line in between will be the V q which carries the momentum the interaction
term that carries the momentum.
194
So, this is k, and the last term is
(1)e
∞
Vq + + +
G = ∫ dt 1 ∑ <T Ck ( t)C k (t ')> < T C k2 +q (t 1)C k2 (t 1)> <T C k 1−q (t 1)Ck (t 1)> .
−∞ 2
So, you see again, this greens function propagates without any interaction or rather
without any disturbance from t prime to t. So, will have to write that, there is a t prime to
t goes with a k. Now there is there are 2 so, this corresponds to delta I mean the q equal
to 0.
And this also corresponds to δq , 0 . So, I have 2 electron bubbles which are like this,
and they are connected with a q. So, one of them is k1, the other is this at some
intermediate time that happens and at k2. So, these are the, this is the Feynman diagram
for this particular case. So now, you see let us just go back once again. We have a greens
function with a bubble, this then there is a there is a greens function.
Which there is there is an interaction vertex which carries momentum q. Then again is
the greens function and bubble, then again, a greens function with a vertex carried
carrying a momentum q, there is a disjoint diagram there is a disjoint diagram.
195
So, if you summarize, so, there are 2 terms which look like this, then there are 2 terms
which look like this. And then there are 2 disjoint terms, terms one of them look like this,
and so, if you summarize. The other one looks like ok.
So, now let us see some higher order diagrams, some typical higher order diagrams. So,
we can have a term like this. So, at higher order diagram as you understand that will have
2 vertices. So, 2 wiggly lines at least I mean that, that is the second order diagram a third
order diagram will have 3 wiggly lines. So, it is one of them then there is one like this.
196
Then there is one like this, then there are so, then there is one term like this. So, anyway
we have 2 wiggly lines and we can have a term like this; which is a disjoint diagram. So,
of course, these are the last 3 are disjoint diagrams, and then there are join diagrams, and
this and so on there will be many of them have just shown some typical second order
diagrams.
So, we can in principle write down, any order for the diagrams, now we just have to
write down the corresponding expressions for that, like the first one let us if we call it as
one, will include integral, and then there will be two t
∫ dt 1∫ dt2 G(0) G(0) G(1) V q 1 V q 2 η( ξk 1) η( ξk 2 )

So, that is the first term so, one will have the form. So, notice that these diagrams are
identical, and similarly these diagrams are identical and so on. So, there are ways to
actually find out such symmetries in these diagrams, and I mean calculating one diagram
and then multiplying it with the number of such similar diagrams will actually one does
not have to compute those many diagrams as actually that come in.
Let me give you an example the symmetry.
So, consider a typical third order diagram. So, I have 3 so, there is a q 3, there is a t 1,
there is a t 2, there is a q 1, there is a q 2 and so on. and then there is a t prime, and then
197
there is a t t and ok. So, so, these are that that is a typical diagram. So, then there is a t 3
here.
t 3 add this let me write it down neatly. So, there is a t 3 here so, now, if we permute
between t 1 t 2 and t 3, they give rise to the same diagram. So, will have you know 3
factorial such diagrams, in any case I mean the; when we consider a third order diagram,
we have a third order diagram has 7 creation and 7 annihilation operators. So, how many
combinations we are going to get? 7 factorial combinations.
And how many of them are? So, this is equal to 5040 combinations. This tells that
actually we do not have to calculate all the 5040 diagrams. This permuting between t 1 t
2 and t 3 correspond to 3 factorial diagrams. So, which means that corresponding to
those 3 factorial, which is equal to 6 diagrams you have to only compute one. And
similar such symmetries can actually be understood or deciphered and so, all these things
will look the same. And in fact, if you think about it, that there are so, you calculate one
diagram, and according to the symmetry you multiply it by n factorial of each one of
those diagrams.
And, but there is also a n factorial in the denominator which cancels. So, all these will
only correspond to one diagram corresponding to the symmetries that you find.
198
Now, we will talk about what we have been leaving out is the multiplicative factor in the
in the perturbation expansion. So, how see how the multiplicative factors expansion. And
what I mean is that what we were writing as this and we are leaving it out.
Actually, there will be so, there are, 2 n plus 1 creation operators in the nth order. So,
there will be 2 n plus 1 creation operators, and 2 n plus 1 annihilation corresponding to
the nth order. So, so, there will be 2 n plus 1 non-interacting greens function.
So, will have so, each of the greens function will have a factor of i. So, there will be a i
into 2 n plus 1, because each one of those <C k C +k > =iGk and then will have to
multiply it by a minus 1 whole to the power n, this comes from the factor in the S matrix
expression. So, this you see that the S matrix expression also has a factor expansion. This
has a factor of minus i to the power n. And there is a minus i that comes from the overall
interacting greens function because G equal to minus i and then the time ordered product.
So, if you multiply all that it becomes i to the power n. So, typically in nth order greens
function will have a i to the power n. Since we were talking about a first order greens
function, then we have should have a factor of i in each of these cases you can put this
factor of i and so on. But that is not the end of the story, there is another thing that comes
in.
199
That is a plus minus 1 term or minus 1 whole to the power F where F is the number of F
number of Fermionic loops.
So, how do we understand that? So, we have a term which is like this, and so on so, this
corresponds to <TC k C +k 1 C +k 2 C k 2+q C k1−q C +k > I am not writing the time indices
explicitly. There is this, then there is this. So, if we change the order of this, this brings in
of minus 1, and then of course, this remains as it is. So now, this so, basically this
corresponds to delta q equal to 0, and one of the fermionic bubbles. So, we get a minus
sign.
Look at this other term. So, this was the term a that we have written the number b is this.
And so, this is equal to <TC k C +k 1 C +k 2 C k 2+q C k1−q C +k > and so on, now there is a so, this
one has to so, this will be one sign, this will be another sign so, a factor of plus 1 of plus
1 that is twice minus 1.
So, anyway one has to keep a track of how many swaps one makes and each time there is
a bubble you are sure to get a minus sign if there are 2 bubbles then one gets a plus sign
here.
So, let us talk about the disconnected diagrams and see what they contribute. And these
diagrams are called as the vacuum polarization graphs. So, at each order of the
200
perturbation theory, all the prefactors that we get while doing the expansion they are all
multiplicative.
So, there is a numerical factor that n factorial which comes from the symmetry of all the
diagrams, they cancel with the denominator, and we have a factor of you know i to the
power n and a plus minus 1 depending on how many fermionic bubbles we have. And
this means that the diagram actually the factorize as take an example for example of a
second order diagram.
So, this is a disconnected second order diagram, and this can be easily understood that,
this is does factorized as a term which is this, and multiplied by with the term which is
this. So, so, in general, this plus this, and then bubble like this, and plus this, and 2
bubbles like this and so on. Can be written as so, this is equal to a term like this, and then
multiplied by a term which is 1 plus this kind of a plus there are bubbles and so on.
But remember this is nothing but the denominator, which is equal to S infinity minus
infinity.
So, this cancels exactly the second bracket or the the second bracket cancels exactly with
the denominators.
So, from this factorization, it is clear, that only the connected diagrams survive. Now this
is a big simplification, because this simplification tells us only the connected diagrams
201
will have to be computed, and we really do not have to compute terms such as such as
these ones or these ones or these ones and so on. So, the top diagrams will be good
enough to calculate and we can do away with that.
So now, how to compute we have not still yet done, the integration over the time. So, one
important thing in a is that one has to has to integrate over internal energies and sum over
internal momentum. So, then each of the greens function is written as which is I’m
(0) 1
writing it now in the G ( k , ω )= where ηk =Sgn( ξ k )
ω− ξk +i ηk
So, a second order a typical second order diagram, which is equal to like this and this and
this. So, there is a q 1 and omega 1.
There is a q1 omega 1. So, this is omega2 – omega1, k - q1. And this is k omega and so,
this is q2 - q1 so, q2 - q1 and so on. So, these are k omega. So, that diagram has to be
written now it is a second order diagram. So, we should put a
d Ω1 d Ω 1 2
i 2(−1)∫ ∫ V q ∑ G(0), 2( k , ω)G (0 ) (k−q , ω−Ω)G(0) ( q2 , Ω2)G(0) ( q2−q1 ,Ω 1−Ω2)
2π 2π q ,q
1 2
So, this are the and there is a loop that is forming, and this loop will give rise to a minus
1. This one has to calculate with this G 0 etc. Put there and they do a method of complex
integrals which you must have seen in complex mathematical physics, course a under
this complex analysis and complex integration.
And that has to be done and then, one has to evaluate these integrals accordingly. And
that will give us the full greens function evaluated at a given order. For example, in this
case we are evaluating it at the second order.
202
Prof. Saurabh Basu
Lecture – 15
Phonon Green function and Hartree Fock approximation
So, having discussed electrons greens function elaborately. let us just touch upon the
phonon greens function and before that let us emphasize the importance of greens
function for a moment.
So, the greens function the pool of the greens function rather, gives the excitation energy
is of the fully interacting system.
We have started looking at different orders of the greens function that is weather the
greens function contains one interaction term or 2 or 3, and so on and it turns out that
they are quite complicated, because a third order greens function would necessitate
calculation of over 5000 diagrams without taking into account the symmetry
requirements. But never the less greens functions are very useful because the poles give
the excitation energy. And will also see that even a fully interacting greens function that
is taking into account all orders of interaction, it can be solved exactly, and the form is
not to complicated, and this is done by using Dyson equation which is what we will
learned.
203
But before that let us review a little bit on the phonon greens, function ah; however, most
of the most of our discussion will stick to the electrons greens function. So, the phonon
greens function is written as, let us write it with a D and q is the momentum, will write a
lambda and it is t minus t prime has the same form as we have seen for the fully
interacting greens function and D(q , λ , t−t ' )=−i <T Aq , λ (t) A −q , λ (t ')> this capital
A is are combination of the Bosonic operators of the operators for phonon will see in a
moment,
Ah now q is of course, the momentum that we have seen earlier, and lambda is a
polarization. Most of the cases the interactions that we considered in the for the phonon
system they do not mix polarization. So, it is safe to get rid of this polarization index
lambda. And the +
A q , λ =a q λ + a−q λ or equivalently one can write it as A q equal to a q
so, this is say involving the lambda, and then we can sort of remove the lambda as well
and can write it with.
So, will remove lambda in this step and write it without that, and these are the
combinations. So, +
A q =aq + a−q . So, the full phonon propagator in the interaction
representation is written as D q or we can write it inside has we are written about.
−i <T A q (t) A−q ( t ' )S (−∞ ,∞)>

So, D(q ,t−t ' )= . So, this is the full phonon greens
< S(∞ ,−∞)>
function, very similar to the electron greens function that we have seen including the S
matrices which were written earlier. So, if you open up the A q’s the capital A q’s in
terms of the small A qs one can write down the greens function ah, or let us write down
the non-interacting phonon propagator, without the S matrices.
204
So, this is
D0 (q ,t−t ' )=−i <T A q (t ) A−q (t ' )> =−i <T (a q e−iω t + a−q e i ω t )(a−q e−i ω t ' + a+q e i ω t ' )>
+ q q q q
And so, understanding that the angular brackets are for the non-interacting ground state
and. So, at 0 temperature is known about phonons or Bosons that that <a q a+q > =1
whereas, the <a +q aq > =0 .
So, combining these information the 0th order greens function for phonons is written as
D 0 (q ,t−t ' )=−i[Θ(t−t ') e−i ω (t−t ' )+ Θ(t−t ')ei ω (t −t ' ) ]
q q
And this is the form for the
0th order greens function for phonons. And now the Fourier transform is given by
∞
D0 (q , ω )= ∫ dt e iω t D0 ( q , t)
−∞
1 1
And if we do that then this would be − where eta is a eta is
ω−ω q +i η ω + ωq −i η
greater than 0 and infinitesimal is small. So, what I did is that, I have just put this D 0 q t
from the line above of the step over when then did that integration infinitesimal alright.
So, this is the form for the zeroth order greens function.
205
2 ωq
And if you simplify it looks like D(0 ) (q , ω )= So, this is the non-
ω − ω2q +i η
2
interacting phonon greens function. So, at finite temperatures what happens is that these
expectations have values that are that involve the number operator for a given q, which is
this is <a q a+q > =N q +1 , and this is equal to the number operator in Nq which is given
+ 1
as the Bose distribution function given by <a q aq > =N q= βN q
.
e −1
So, the total I mean the greens function above is a given by at finite temperature. So, it is
D(0 ) (q , t−t ' )=−i[( N q +1)e−i ω | t−t '| +(N q )e i ω | t−t ' | ] .
q q
And that is the 0th order propagated for phonons, and then one can do a Fourier
transform and write it in omega A q omega space, let us take a typical electron phonon
interaction the we shall not considered phonon phonon interaction, because phonons are
the quasi particles. So, it is assumed at the quasi particles are a non-interacting. So,
electrons can interact with phonons and electrons of course, can interact with electrons,
here will only consider electron phonon interaction.
So, interaction is of the form V = ∑ M q A q Ck++q σ C k σ . So, there are 2 Fermion

q ,k , σ
operators, one creation one annihilation an along with that the phonon operator which is
A q and M q is the strength of the interaction.
206
Now, something interesting can be noticed here, that <T Aq > =0 , because each
A q →a q , a+q kind of operators.
And the ground state expectation of each one of the terms <a q> =<a +q > =0 . Because
if you create a Bosons in the ground state, then you have one Bosons more which will
have 0 overlap with the original ground state and same with the other term which is a q.
Does the terms involving linear powers of A q capital A q would all be present? And not
only that all the terms with odd powers of A q will all be absent. So, it is only even
powers of A q which will be present, and the one can write down a linear in terms of I
mean quadratic. So, quadratic in V is what survives, and A quadratic term in V is then
the full greens function can be written as t minus t prime and it will be
(−i)3 ∞ ∞
(0)
G(k , t−t ')=G (k ,t−t ' )+ ∫ dt ∫ dt ∑ M M <T A q 1(k 1) Aq 2 ( k 2)>
2 −∞ 1 −∞ 2 q , q q 1 q 2 1 2
* ∑ <T C k+σ (t) Ck++q (t 1)C k1 σ (t 1 )C k 2+q σ ' ( t 2 )C k 2 σ ' (t 2 )C k σ (t ' )>
k 1 ,k 2 , σ , σ '
and. So, there are these so, one creation with sigma the other could be this is sigma
prime, and then there is a C k sigma coming from the definition of the greens function
for electrons is this.
207
So, these are the terms that are that will be present. And now in order to write the on the
electronic part, as the non-interacting means functions, we have we have studied how it
looks like.
You just worry about the phonon propagator at the moment.
So, that gives me easily <T Aq ( t 1 ) A q (t 2 )>=i δq +q D(0) (q ,t 1−t 2) . So, that is the;
1 2 1 2
write it neatly so that is so, delta q 1 plus q 2 will be equal to 0, which means q 1 equal to
minus q 2 and that will be with a i and then we have to in a compute the non-interacting
phonon greens function which is what we have learnt how to calculate it here. So, that
will be the energies square.
So, the electron greens function of course, as a many contributions, and there will be a
since there are 5 rather there are 3 greens a 3 electron 3 creation operators, and 3
annihilation operators for the electrons, there will be 3 factorial combinations which
would allow us to write down them down in terms of the non-interacting prince function
for the electrons so, which we have learned. Now let us discuss more on the so, will stop
with the phonon greens function here, and discuss with continue discussing with the
electron greens function. And so, there are some things about the electron greens
function which we have seen shortly or rather briefly. But we haven’t discussed them in
great detail.
208
So, let us discuss them the ones that we have we have started coming back to the electron
greens function yesterday or the in the previous discussion, you might have seen that that
we had diagrams of this kind. So, this is a q = 0 and there is a k, and then we are written
it with time, but if we Fourier transform will get an omega here, will get a omega here,
and this is some k⃗1 or k 2 depends on which of the diagrams you are calculating, and
this will correspond to energy k 1 and so on.
And this is q equal to 0 and omega equal to 0, it is an instantaneous interaction. And

similarly, there is another term that we had considered earlier, that is this one, in which
there is a k omega there is a k - q omega minus omega. And there is a q omega that is
being carried by the vertex and then finally, the free electron greens function is given by
this. So, what happens is that, at this point let us call this as A, at this point there is a
vertex or there is a weekly line which has a strength a V q, and has an energy
dependence which is goes has omega so that takes away some momentum and energy.
So, the electron is left with k k minus q and omega minus omega finally, they reunite
here at the point b and continues as k and omega. So, these 2 classes of diagrams is what
we had seen. So, this is let us call this as a and this as b. And c that what can this they tell
us, and how we actually use them to calculate something that is of important to us. So, a
is known as the Hartree term. And b is known as the fock term. So, we shall see that, the
Hartree term gives 0 contribution for the coulomb interaction.
209
And since were discussing the first order perturbation, there is no change in the wave
function, we only get renormalized energies. And the first order correction is computed
as the expectation value of the interaction Hamiltonian between the ground state. So,
correction to the unperturbed energies < ψ | H ' | ψ > → < H ' >=< ϕ 0 | H ' | ϕ0 > . And in
this case, we know that this H prime would have to be between the non-interacting states.
So, the underlying assumption is that in presence of the perturbation, the system hasn’t
gotten too far away from the ground state. So, the ground status are still good states for
the problem, and we simply want to calculate the change in energy and for for the
interaction term to be squeezed between the non-interacting downstairs.
And this can be written as, it is equal to
1
< ϕ 0 | H ' | ϕ 0 >= V q < ϕ 0 |C k+ σ C +k σ C k + q σ ' C k −q σ | ϕ 0 >
2 k , k q∑
1 2 ≠0, σ , σ '
1 2 2 1
If you look at the coulomb term that we had written earlier this is exactly the form that
4 π e2
we are taken. And what is V q? V q= and this blows up at equal to 0.
q2
So, q equal to 0 term from the summation will have to be dropped. And now look at the
pairings, will write down the pairing so, if we can take a pairing between this and this,
210
and this and this, both correspond to q equal to 0. And we can also take pairing between
the first and the third one, and the second and the 4th one, both will correspond to delta k
1 is k 2 plus q. And of course, also delta sigma prime. Now you see that the Hartree term
corresponds to q equal to 0; which is not included here at all, any of the 2 terms I mean
any of that the pairing comes from these pairing these lined terms.
The terms that are being connected by the horizontal lines they are not included. So,
Hartree terms Hartree diagrams, they do not contribute for the coulomb interaction term.
So, this is one nice information which see every information in this regard is chooseful,
because we had to calculate a large number of diagram. So, were trying to reduce by
using symmetries as much as possible. So, Hartree terms, now there are 2 such terms, if
you remember one with the k 1 and the other with the k 2 and both will not contribute.
So, Hartree terms do not contribute for coulomb interaction.
So, this is something that is interesting and important. So, will just put a line here now let
us look at the fock term. And the fock term is a is simply. So, this lower one is the fock.
So, this is the Hartree, and this is the fock. So, this term looks like so, this demands that
k 1 is equal to k 2 plus q, remember all these keys are vectors, they are a wave vectors
ah, but we are writing it without the vector notation you can put them, but it becomes
someone Messy to write every time with vector, but understand their vectors.
And this also demand that sigma equal to sigma prime. So, there is so, there is also a
Fermion loop that brings a negative sign. So, my H prime
1
< H ' >=−∑ < Ck+σ C k σ > ∑ V <C +k+ q ,σ C k+q ,σ >
kσ 2 q≠0 q
And that is the combination, now if we call this as a minus Hartree fock energy. So,
including the Hartree because Hartree does not give any contribution, then we can write
down this term as a k sigma and there is a Hartree fock, and energy of the Hartree fock,
and then C k sigma C k sigma. So, the expectation value of H prime between the non-
interacting ground states is equal < H ' >=∑ ϵ HF <C +k σ C k σ > which is the Hartree fock
kσ
energy given by this term which is the sum over there is a half factors well. Sum over q
naught equal to 0 of V q and then this expectation value.
211
So, that is the total Hamiltonian including this energy correction, at the Hartree fock
level is equal to.
So, this is this is including so, this is just the ok, let us just write the Hamiltonian which
will are only writing the interaction term. So, this is interaction we can put a interaction
Hamiltonian. So, that was not the total Hamiltonian. So,
H '=∑ ϵ HF < Ck+σ C k σ > =∑ ϵ k <C +k σ Ck σ > that is the Hamiltonian.

kσ k
Ah now if this Hamiltonian has some other terms; which are at the 0th level then that has
to be added to this term and so on. So, let us look at the fock term more carefully. So,
this is equal to i ; now I will do the Fourier transform and go over to the omega space.
dΩ 1 2
This is i∫ ∑ V G(0) (k−q , ω−Ω) G(0) ( k , ω )

2π 2 q q
That is like saying if you go to this there are 2 G 0s with k omega on the left side and
right side of the vertex, and then there is one G which is here and then the V there is a V
q omega that we are written V q. But it is also function of omega will see that how that
goes. Now let us call this thing entire thing here up to this to be something called
because we are summing over q, and we are so, let us call this a sigma omega we are
summing over q and we are also summing over capital omega.
212
So, this is equal to a sigma of omega. So, this is what we get from the Hartree fock term,
where we writing one of the terms is equal to the part of the Hartree fock term to be
equal to sigma of omega. Now look at n equal to 2 a typical n equal 2 diagram. So, there
is a k omega, there is a k1 – q1 omega - omega one there is again k omega, there is now k
minus.
So, this could be just k – q2 omega minus omega 2, and then there is a k omega and this
carries this one carries Aq1 omega one and there is A q 2 omega 2.
So, this one can be simply written down as

G(0) ( k , ω ) ∑ (k , ω )G (0) (k , ω ) ∑ (k , ω )G(0) ( k , ω ) and because, we are summing over
that internal energy is which are capital omega one capital omega 2 etc, and this and so
on. ∑ (k , ω ) is defined above.
So, sigma can actually be a function of it is actually a function of both k n omega,

because you are summing over q so, it is a k omega. This is called as ∑ (k , ω ) is
called as the free energy. And let me this is an important things so, let me box it in red
colour.
So, this is called as a free energy. So, diagrammatically a second order term is written as
this, and then there is a sigma, then it is this, and there is a sigma and then it is this and
213
so on. So, at the second order, we have 2 such sigmas, and in the third order there will be
3 such sigmas.
And if we considered an infinite series, and then series we can write down of full greens
function let us write it with a double arrow; which is equal to a single let us let us take it
to the new page because this is going to be used.
So, full greens function is written with a double arrow, double line with an arrow, and
the non-interacting or the unperturbed one is written with the so, this is the G(k , ω ) .
This is G(0) (k , ω ) . Now I will have a line and a ∑ (k , ω ) and a G so, this is a
G(0) (k , ω ) ∑ (k , ω ) G(0) ( k , ω ) . I am dropping all these dependencies.
So, this plus, and then there is a plus sigma sigma. So, there is a G(0) ( k , ω ) ,
G(0) (k , ω ) ∑ (k , ω ) G(0) ( k , ω ) and so on. And going up to infinite order, one

∞
n
can write down a GP series, with a G(0) ( k , ω ) ∑ [G(0) ( k , ω )∑ (k , ω )] and this is
n=0
G(0 ) (k , ω ) 1
equal to G(k , ω )= is equal to
1−G(0) (k , ω ) ∑ (k , ω ) (0) −1
G (k , ω ) −∑ (k , ω )
So, this is called as Dyson equation.
214
This is what we were saying earlier, that we could do a perturbation theory with up to
first order or up to second order or up to third order. Now in some given problems, a
perturbation theory can fall insufficient up to a particular order or given order, in which
case one has to actually sum over infinite number of terms containing infinite number of
H primes. Now formally that may look somewhat difficult because if we know that after
so, we go from first order, we have first order we had 6 terms, second order we had 20
terms, or something like that, if I remember it correctly there are.
So, so, second order so, so, the from the 0eth order it goes up to so, 0th order there is one
term. first order there are 6 terms, second order there are 120 terms and third order there
are 5040 terms. So, this is the magnitude of the problem that we otherwise need to find
out. But here up to infinite order which is taking into account repeated interactions,
infinite number of times one arrives at an expression for the greens function; which looks
very simple with a G 0 and a self-energy to be computed and we have given the form of
the self-energy. And this one is with a G is 0 as we already know, that this is equal to a
(0) 1
non-interacting greens function G ( k , ω )= eta k being the sign of xi k.
ω− ξk +i ηk
215
1
And in which case our full greens function looks like, G(k , ω )=
ω −ξk −Σ(k , ω )+i ηk
. And so, sigma is equal to a term like this, flush a term like all these and so on and so
on.
So, consequences of Dyson equation so, this is the first one of them is that less, far less
actually a number of diagrams to calculate. and second is of course, the interacting
greens function, a full means interacting greens function, function looks like a non-
interacting one, non-interacting one except for the energies are now replaced by well
they have to resemble same, minus ∑ (k , ω ) at this stage.
This is exactly the form of the G 0, now the energies are replaced by this
ξ k → ξk −Σ(k , ω)
Now, let us calculate, calculate H of which is a function of k explicitly.
So, a sigma is written as, we are taking into account that same fock term. So, this is a
d3 q d Ω (0)
Σ=i ∫ 3
V q∫ G (k −q , ω −Ω)
(2 π ) 2π
So, it is looks like that if we can calculate this integral, then we would be done in a doing
an infinite order perturbation theory, or rather we can get the Dyson equation on
216
calculate the full greens function. And know what the self-energy is. So, this is the
expression for the self-energy. And so, this requires us to calculate integral such as now
there is one thing that one should look at is that we are talking about really equal time
greens function, that is these 2 are have the same time; which means that the frequency
associated with it is infinite is small or going to 0, in which case we have the
d Ω (0)
i∫ G ( k−q , ω−Ω)=−n(k −q)
2π
This was once introduced while writing it, and then we had again written it as a green
function. So, so, this is that so, as omega goes to 0 will take the form of a number
operator n k minus q. the reason is that this is
dΩ 1 dΩ 1
limit ω →0 i ∫ =i ∫
2 π ω −Ωk − ξk−q +i ηk 2 π −Ω−ξ(k−q) −i ηk
So, limit omega going to 0 because these are equal time it happens instantaneously this
interaction. And now we have a minus omega minus xi k minus q plus i eta k, and the
pole is at is at ω =−ξ k−q−i ηk . Now it is always true that a corresponding to a minus
energy, there is also a plus energy.
So, this if it is true, then we have the pole −ξ k−q= ξk−q →Ω= ξk−q −i ηk .
217
Now, consider k minus q is greater than 0, and i eta k equal to i eta then of course, then it
is taking the sign of the xi k. So, the pole is Ω= ξk−q −i η . So now, the pole is in the
lower half. So, pole is not in the upper half. So, the integral is equal to 0. So, if you want
to a close it from top.
Ah now also so, pole is above the pole is in the below the real axis, real axis so, if the
loop is enclosed in the upper half, then the integral is 0. Now suppose a ξ k−q < 0 , in
which case my . i ηk =−i η So, pole is equal to Ω= ξk−q +i η . One simple pole, one
simple pole in the upper half plane and doing contour integral as you have
dΩ
∫ 2 π G(0) ( k−q , ω−Ω) it is equal to 2 π i into some of residues pi i into some of
residues. And the residue here is equal to 1. Because it is equal to
Ω−ξ k−q−i η i
=1=2 π i I now you multiply it with .
Ω−ξ k−q−i η 2π
i i
So, the required expression multiply with which was there here ∗2 π i=−1 ,
2π 2π
and one gets a minus 1 into k minus q ok.
d3q
So, then you recognize that a minus so, the Σ (k , ω )=−∫ V q n(k −q) . I can q is
( 2 π)3
being integrated over, I can use q as a dummy variable, and redefine k minus q A q. So, q
218
d3 q
becomes k minus q 2 that becomes equal to −∫ 3
V k−q n( q) So, I can simply
(2 π )
write this as, because n q is the number operator which will have a value equal to 1 at t
equal to 0, for q to be less than k infinity, where n q will have a value equal to 1.
d 3 q 4 π e2
So, I simply get at − ∫ . And this integral if you perform it becomes
| q | <k F (2 π )3 | k−q |2
k 2
2 1−( )
−e k F kF ϵF + k
equal to π [1+ 2( k /k ) ln| k −k |] k. And so, this is at this level it is independent
f f
of omega of omega.
So, omega means I should write it with the capital omega. So, it is it is simply so, the at
this level at this level. It is simply renormalizes the bandwidth. There is an interesting
cartoon on the perturbation theory that one can present this is a book by Matuk; in which
he gives an analogy between a propagator or a full bring function and the drunken man.
So, the drunken man if you look at it carefully the though the resolution is not good. He
started from one, which is his original location. And then while going back home, home
is at 2 here.
219
He is undergoing repeated sort of deviations into various bars. So, these are alices bar,
and then there are other bars which are written here. So, he goes and gets scattered by
each one of them, and then he multiply time goes and he comes out by they are shown in
the path, and then he goes here and then he before he goes home he goes to an Dixie bar
and all that.
So, basically the he is full propagation is given by the full brings function G, and he is
outside the bar the travel is given by the G 0, and then he could have gone. So, one
option is that he could have gone directly from here to hear, or he could have just visited
one bar; which is like one sigma. And so, it could have been like this bar that he visits
and then goes back home, but, so, one is G 0 and then his scatters at the bar and then
goes back home. I could do at 2 bars and then finally, goes back home.
So, which means that he goes hear, and then he goes here, and then he goes home and so
on. So, his entire trajectory will be actually formed by taking the combinations of all
these parts that he takes and this full propagation would be described by a perturbation
theory that we have presented here. G=G(0) +G(0) Σ G(1) +G(0 ) Σ G(0) ΣG(0)
220
Prof. Saurabh Basu
Lecture – 16
Finite temperature Green's function and Matsubara frequencies
We shall now talk about Finite temperatures because the experiments for condensed
matter systems are always done at finite temperature. And an important goal of the
physics that we are doing is to explain experiments or experimental data and for that we
need to go to finite temperatures. So, far we have been talking about 0 temperature
greens function.
And now we shall talk about Finite temperature greens functions.
So, what happens at finite temperature is that all quantities get weighted by the factor.
1
So, by e(−β E) where beta equal to β= which is called as the inverse
kB T
temperature and k is the K with a subscript b is called as a Boltzmann constant. So, now,
what happens to the operators at finite temperatures? So, which means that the operators
have to be weighted by this operator where H is the Hamiltonian of the system.
221
So, now so, at finite temperatures the operators that we need to write down the greens
function they have to be weighted by this quantity or the weighting factor which is
exponential minus beta H beta being the inverse temperature or 1 over k t H is the
Hamiltonian for the system.
So, we write down a greens function as. So, greens function will write down later, but it
Tr [e−β H C k σ (t)C k σ (t ' )]

is written down as the −β H
.
Tr [e ]
So, this is the a form of the greens function and the trace implies the T is actually the
trace, which implies that we have to take the trace over the complete set of states n.
And the operators this is pretty much we are still at 0 temperatures. So, the operators are
written as C k σ (t)=ei H t C k σ (0)e−i H t H being the H 0 plus H prime right.
So, this we have seen and so, H prime now appears at two places. So, remember that we
have done perturbation theory with in powers of H prime, now H prime is occurring at
these e±i H t , e−β H . So, this appears at both places we were writing a curly H does not
matter. So, so H prime appears at both these places.
So, the perturbative expansion in terms of H prime was earlier included in this term. And
should we now again do a perturbation expansion in terms of H prime with this. So, that
222
is the question? Now the answer to this question was given by Matsubara in the year
1
1955 in which he realized that β= can be considered as complex time.
kB T
So, if you consider this as complex time this beta then these two factors which are
written here and here they can be combined together. And so, basically if they can be
combined together, then we can do a single perturbation theory with respect to H prime
and can write down the complete green function or the full greens function in terms of
that.
But what finally, was done is exactly the opposite time is regarded or treated as complex
temperature; this is an important concept in the finite temperature calculations that will
see henceforth. Another motivation of the Matsubara formalism so, we henceforth we
call this as a Matsubara formalism is that the occupation numbers for Bosons the
occupation numbers which are the distribution.
So, these are Bosons and fermions and this you know that this is written as
1
n B= β ℏ ωq are the Bosons energies and this is written as exponential
e −1
1
n F= βξ k and where epsilon ks are the energies. So, these are the occupation
e +1
numbers.
223
Now these occupation numbers can be expanded in a series and let us call this 1 as n F
let us call 1 this as n F and let us call this one as n B.
So, n F is written as which is as we have written
1 1 1 ∞ 1
n F ( ξk )= = + ∑ and the n B for the energies omega q,
e +1 2 iπ
β ξk
β n=−∞
(2 n+1) − ξk
β
1 −1 1 ∞ 1
which as we have written n B= = + ∑ or we can keep
β ℏ ωq
e −1 2 β n=−∞ i
(2 π n) −ℏ ω k
β
this h cross here or we can drop it later.
Now, these are the expressions for the occupation densities of the occupation numbers
for the fermions or Bosons, we will not provide proof for that a rigorous proof for that,
but definitely we will tell you that it looks like that there are poles or singularities, if you
are familiar with the complex analysis or the complex integration these are called as the
poles.
So, there are poles at odd multiples of i pi by beta and these have poles at even multiply
even multiples of pi over beta a this we will prove and we simply say that this comes
from a theorem and the theorem states that, that any Meromorphic functions; please see
the definition of Meromorphic functions maybe expanded as a as a summation over it is
poles and residues at the poles.
224
Let me box this thing I am simply taking the statement and using it to know more about
this you can look at a complex analysis book by Churchill. So, what it says is that the
Bosons occupation functions?
So, n B has a poles at poles at even multiples of pi over beta in the along the imaginary
axis. So, let us n f it has poles at odd multiples of pi over beta, but which lie on the
imaginary axis ok.
So, these are even things are 2 2 n pi i by beta and this is 2 n plus 1 pi i by beta.
So, both the summations it is apparent that can be written as
1 1
∑ i ω −ω an d ∑ does not matter I mean the momentum index does not
n n q n i ωn− ξk
matter for fermions. So, this is for Bosons and these are for fermions.
So, for fermions they are odd integers of pi over beta and for Bosons there even integers
of pi over beta. Now it is interesting to note that at this form either of the forms they
1
correspond to a non-interacting greens function, function is of the form and
i ωn− ξk
this non interacting greens function is called as the Matsubara greens function.
Matsubara is a Japanese physicist.
225
So, now, in the Matsubara formalism as we have said the time is a complex quantity and
it is actually written as it is actually an imaginary quantity if you, it is a and which is
written as τ =it and the greens functions these greens functions the Matsubara greens
functions are functions off of tau and bounded between minus beta 2 beta 2 plus beta.
So, that is the range of tau for the Matsubara greens functions which are so, tau is
nothing, but the time, but it is just that it is a complex time and these greens functions are
functions of these tau bounded between minus beta 2 plus beta. Now let us take any
function which is a function of tau and 1 bounded between these values minus beta and
plus beta we can always do a Fourier series.
So, take an arbitrary function f tau, which can be expanded in Fourier series of the form.
1 ∞
nπ τ nπ τ
So, f ( τ )= a 0+ ∑ [a n cos ( )+ bn sin( )] .
2 n=1
β β
So, where the coefficients a n and b n are defined by
β
1 ∞
nπ τ nπ τ 1 nπτ
f ( τ )= a 0+ ∑ [a n cos ( )+ bn sin( )] , the
2 n=1
β β
bn = ∫ d τ f ( τ )sin (
β −β β ) .
Look at Arfken mathematical physics book or kreyszig for information on these Fourier
series.
226
1
So, in a compact notation we can write f (i ω n)= β (an +ib n ) So, this is the rather the
2
so, this is f as a function of these frequencies that we had talked about and which can be
obtained from the Fourier transform of f tau in the following fashion.
1 ∞ −i nβπ τ inπ τ
1 ∞
So, f ( τ )= ∑ e f (i ω n ) , such that our f (i ω n )= ∫ f ( τ )e β
.
β n=−∞ 2 n=−∞
So, these are the properties of these functions, we shall later show that these functions
are nothing, but the greens function the Matsubara greens function that we are going to
talk about and as said earlier that these functions are needed in order to have the
experimental quantities, which are always done at finite temperatures.
So, there are further simplifications possible and for Bosons the simplifications are like
this that we have an additional property, which is equal to f ( τ )=f ( τ +β ) for
f ( τ )=−f ( τ + β) minus beta to be in the range for tau to be in the range minus beta
and 0.
And so, this is a sort of periodicity of the function tau we are going to show that and for
fermions, we are we have an anti-symmetric situation in which we so, this is in again in
the limit tau to be between minus beta and 0 and let us show this for bosons.
227
β i πn τ 0 i πn τ
1
So, f (i ω n)= [∫ d τ f ( τ )e β
+ ∫ f ( τ )d τ e β
] .
2 0 −β
And now if we change the variables in this term in this term if we change variable from
β i πn τ
1
tau to tau plus beta, then I get f (i ω n)= (1+ e in π )[∫ d τ f ( τ )e β
]
2 0
So, both of them can be combined to write it in this thing 1 should check this n pi tau
over beta. So, because of this first factor here f of i omega n equal to 0 for n to be n to be
β iπ nτ
odd thus for bosons we can write it as f (i ω n)=[∫ d τ f ( τ )e β
] and for omega n to be
0
even number because for odd it is equal to 0.
1
So, 2 n pi by beta. And similarly your f ( τ )= e−i ω τ f (i ωn) but now n is denotes
β∑
n
only odd sorry only even integers and this is minus i omega n tau, n integer means the
omega n will correspond to even factors or rather even multiples of pi over beta that is
what. So, n is restricted in that sense.
f of i omega n so, this is for fermions; So, we have used this property which is because
the tau going to tau plus beta, then we have used that f tau of beta equal to f tau and have
written down this second term; here and this second term becomes equal to 0 if n is equal
228
to odd and that is why the Bosonic frequencies are even multiples of pi over beta. Let us
see these for fermions.
For fermions we have this odd property and then I will again split up the sum or rather
the integral into 2 terms, which is half 1 minus exponential i n pi. Again I will do the
same trick of writing down the minus beta over 0 to will change the variable from tau 2
tau plus beta and use the anti-symmetric property of f of tau with respect to f tau plus
β iπ nτ
1
beta and this is equal to f (i ω n)= (1−e i n π )[∫ d τ f ( τ )e β
] .
2 0
So, now it says that f of i omega n equal to 0 if n is even for n to be odd we have f i
omega n. So, we have omega n to have only odd values of pi over beta.
β iπ nτ
1
So, this is going to be f (i ω n)=[∫ d τ f ( τ )e β
] and f ( τ )= e−i ω τ f (i ωn) for
β∑
n
0 n
these things ωn =(2 n+1) π ok.

β
So, these are things that we have learned now we have written down that f is an arbitrary
function, but; however, that arbitrary function having some symmetry it tells us that there
are these the frequencies for fermions and bosons they get quantized or rather they are
229
discrete in terms of either even multiple for pi over beta for bosons and all multiple odd
multiples of pi over beta for fermions.
So, these Fourier these functions will be shown these arbitrary functions f will be shown
as the greens function. And the merit of the Matsubara greens function is that it directly
leads to physical results such as the electrical properties such as conductance or transport
properties such as conductance or you know magnetic properties such as magnetic
susceptibilities etc and they are complex functions of i omega n as you can see here. And
now if you need to go to 0 temperatures, then 1 can do an analogy continuation. So, will
write this limit to 0 temperatures can be taken by analytic continuation of
i ω n → ωn +i η
So, eta goes to 0. So, you need to a plus or minus depend on whether we want to go to.
So, this is for the retarded greens function and which are directly related to the
experimental data. So, we have extended the 0 temperature formalism for an arbitrary
function to finite temperature and found that, the there are restrictions or conditions on
the frequencies for bosons and fermions and these show up in the form of poles of the
Matsubara greens function and we shall now define the greens function themselves.
So, let us write down the Matsubara greens functions.
230
So, let us write start with the electron Matsubara greens function, but as we know that
the only difference comes in the way the frequencies are quantized. So, we would only
write down for the electron or the fermions, let us write down for the use the word
Fermion rather than electron.
So, the fermions we write it down for, but essentially you can write it down almost
similarly very similarly for the bosons as well. Here we will use a curly G a calligraphic
G to denote the greens function and to distinguish it between with this with the 0
temperature greens function.
So, g(k , τ , τ ' )=−<T τ C k σ ( τ )C +k σ ( τ ' )> . So, now, the 2 times are involved or rather
complex times are involved and we have taken 1 k 1 can taken a k and a k prime and
write the 2 operators fermions operators with k and k prime that would not change
anything.
Important thing is that we shall show that the greens function identically depends on tau
minus tau prime and not tau and tau prime individually, thus at the end 1 can drop 1 of
the time indices and can write the resultant greens function as a function of 1 time only
will have to come to that.
So, let us write down k tau and tau prime. So, as we said that this will be written down in
terms of the trace and this t tau is again the time ordering operator, which orders the time
that is earliest time tau earliest tau to be closest to minus beta.
Now, this is the difference between the 0 temperature greens function that we do not
have a minus infinity here, if the tau is bounded between minus beta to plus beta which is
why all these properties symmetry properties of the bosons and the fermions came.
So, will write this once more that let us before we write down the greens function let us
at least define the T tau a time ordering operator which arranges, which arranges the
fermionic operators such that the earliest tau, tau closest to beta closest to minus beta is
put at the extreme right.
So, this is a definition of the time ordering operator and so, it serves the same purpose as
the time ordering operator in 0 temperature formalism excepting that that was pushed to
231
minus infinity or plus infinity, here it is pushed towards the 1 extremity which is beta I
mean minus beta in this particular case.
So, we keep using the calligraphic greens g for denoting Matsubara greens functions. So,
we have the left hand side is written with 2 time indices the tau indices will see that how
these 2 time indices can eventually be dropped? For that let us write down k as tau tau
prime the so, the time ordering can be trivially taken into account by the theta operators,
which we have learnt that tau is greater than tau prime this is equal to 1 this function is
equal to 1 else it is equal to 0. And
g(k , τ − τ ' )=−Θ( τ − τ ') Tr [e−β (k −Ω) e τ k C k σ e−( τ −τ ' )k ]
+ +Θ( τ '− τ )Tr [e−β (k−Ω) e τ 'k C +k σ e( τ − τ ')k Ck σ e− τ k ]
you should write this neatly and see for yourself.
Now this k is an operator. So, K=H −μ N where H is the Hamiltonian of the system
and we decide to work in the grand canonical ensemble that is why this chemical
potential has been used mu is the chemical potential and is the number of particles.
And this omega is called as the grand potential and this grand potential is the basically
the just like the free energy in canonical ensembles. So, the k t log z would give us a free
energy and this is actually the Ω=−k B T ln Z G , where Z G equal to just write it where
Z G is the partition function in the grand canonical. So, this G corresponds to this G
corresponds to the grand canonical ensemble.
So, remember that this is an operator while this is a scalar quantity and this distinction
will be needed in the subsequent discussion. So, now, we shall use a theorem of the
cyclic variation. So, it says that the trace is unchanged by a cyclic variation of the
operators.
232
That is so, Tr ( ABC)=Tr (BCA)=Tr (CAB) and so on.
So, these are cyclic cyclically changing the traces the trace remains invariant. So, what
we why we said is that we want to use the exponential minus beta tau term or
exponential minus beta tau term tau prime term push it to the left. So, that G k there is
another thing that has been achieved is that the right hand side is completely a function
of tau minus tau prime.
So, we can start writing down the left hand side to be a function of tau minus tau prime
and this is equal to a minus theta tau minus tau prime, it is just mathematically a little
cumbersome, but there is nothing very difficult about it you can have to just do it 1 once
by yourself to get convinced.
g(k , τ − τ ' )=−Θ( τ − τ ') Tr [e− τ ' k e−β (k−Ω) e τ k C k σ e−( τ − τ ' )k Ck+σ ]
+ +Θ( τ '− τ )Tr [e− τ k e−β (k −Ω) e τ 'k C +k σ e( τ ' −τ )k Ck σ ]
So, these are the 2 terms where we have used the cyclic variation of the trace. Now
consider the commutation of these 2 operators. Now there is the operator in these 2 terms
the 2 exponential a k only because as I said omega is a scalar quantity. So, these 2
operators are both k in the 1 is with the exponential minus tau t the other is with minus
233
− τ' k −β (k−Ω) − β (k −Ω) − τ k
beta k. So, they should commute. So, commutation says e e =e e
does not matter in either of the terms the first term there is a tau prime
So, we have written down here an exponential minus beta k minus omega an exponential
minus tau prime k. So, these are the commutation and now we can use this commutation
relation and can push the exponential k minus omega to the other side.
And write down the g(k , τ − τ ' )=−Θ( τ − τ ') Tr [e− β (k −Ω) e( τ − τ ' )k Ck σ e−( τ − τ ')k Ck+σ ] +
+ +Θ( τ '− τ )Tr [e−β (k−Ω) e−( τ− τ ' ) k C +k σ e( τ − τ ')k Ck σ ]
So, that is the greens function and since as we have been saying that they depends upon
tau minus tau prime on both the sides drop 1 index 1 variable rather.
So, now that tells us that
g(k , τ )=−<T τ C k σ ( τ )C +k σ (0)>
g(k , τ )=−Tr [e−β(k−Ω) T τ ( e τ k C k σ ( τ )e− τ k C +k σ (0))]
So, this is the form of the greens function finally, which is what we have been trying to
get at it is a Matsubara greens function a function of I mean we have taken this to be k it
could be any suitable quantum label for the problem. And now consider these additional
234
symmetry properties that we have so, considered tau to be between minus beta and 0;
that means, consider tau to be negative and use the property f ( τ )=−f ( τ + β)
We have been telling this that those arbitrary functions can actually be like our greens
functions, which are relevant for us and in this limit of minus beta less than tau less than
0.
Then we can write down the g(k , τ )=−Tr [e−β(k−Ω) C +k σ e τ k Ck σ e− τ k ] and that is the
form of the greens function. And now 1 can use that cyclic property of trace several
times and can write it down as.
So, will write this give this clue here use cyclic in variance of trace G k tau equal 2 equal
to g(k , τ )=−Tr [eβ Ω e τ k C k σ e−( τ +β)k C +k σ ] and so on.
And so, exponential beta omega is not recycled, because it is not an operator. Recycle
means it is not just the way the trace of a b C and b C a that it has not been done it has
been kept, where it is it has not been recycled and then finally, I mean one can do a
regrouping by adding a adjusting exponential plus minus beta k, we have written down
as g(k , τ )=−Tr [ e−β(k−Ω) e( τ +β )k C k σ e−( τ +β)k C +k σ ] .
And so, this is the form of the greens function and that the term so, the term in the right
that is this term is equal to nothing, but equal to −g(k , τ +β ) for we have taken that
235
from 0 less than tau less than or rather minus beta less than minus beta less than tau less
than 0. And so, this is the property that we were talking about and so, this equality is
satisfied by as the same equality was satisfied by the function the arbitrary function that
we had talked about.
And the, I this above identity this allows the greens function to be expanded in a Fourier
series of the type.
That
g(k , i ωn)=∫ d τ e iω τ g(k , τ )

n
1
g(k , τ )= e−i ω τ g (k , i ω n)
β∑
n
. So, these properties are exactly save as that arbitrary function with omega n to the odd
multiples of pi over beta.
So, your i omega n are odd multiples of pi over beta. So, this is formally representing the
greens function the Matsubara greens function will do some examples with this and let
us do 1 example and it is right down example.
236
So, we write down the non-interacting greens function Matsubara greens function non
interacting Matsubara greens function ok.
So, H=H 0 =∑ ϵ k C +k σ C k σ ; k=k 0=H 0−μ N =∑ ξk Ck+σ C k σ where the

k ,σ k,σ
ξ k =ϵ k −μ . So, the tau evolution of the operators can be written as is that the
C k σ ( τ )=e τ k C k σ e− τ k =e−ξ τ C k σ
0 0 k
C +k σ ( τ )=e τ k C +k σ e− τ k =e ξ τ C +k σ .
0 0 k
So, this is the tau evolution tau is remember tau is complex time. So, the tau tau
Evolution of these are like this and similarly the tau evolution of the creation operator is
like this.
And this can be proved using what is called as the Baker Hausdorff Theorem, which says
1 1
e A C e− A =C+[ A ,C ]+ [ A ,[ A , C]]+ [ A ,[ A ,[ A ,C ]]] and so on.
2! 3!
237
So, this can be applied to our case where we have an

e τ k C k σ e− τ k =Ck σ +[ τ k 0, C k σ ]+ ................ so on. And
0 0
τ ξk [Ck+σ C k σ ,C k σ ]= τ ξk Ck σ
where k 0 =∑ ξ k C +k σ Ck σ .
k ,σ
( ξk τ )2 −τ ξ
And hence we can write down a C k σ=C k σ−( ξk τ )C k σ + C k σ .....=e Ck σ . So, k
2!
that is how we can write down the greens function for the non-interacting problem,
which for which I have introduced this 0 and
g(0 ) (k , τ )=−Θ( τ ) e−ξ τ <C k σ C +k σ > +Θ(− τ )e− ξ τ <C +k σ C k σ >

k k
− τξk
And this can be written as exponential e [Θ( τ )[1−nF ( ξ K )]−Θ(− τ ) n F ( ξk )] . So,
1
the the 2 combined so n F is nothing, n F ( ξk )=<C +k σ C k σ > = β ξk . It is easy to
e +1
obtain the frequency dependent greens function, which is equal to k and i omega n which
is nothing, but a 0 to beta d tau exponential i omega n tau G 0.
238
β β
i ωn τ
g ( k , τ )=−(1−n F )∫ d τ ei τ ( ω −ξ )
(0 ) (0)
And a g (k , i ωn )=∫ d τ e n k
0 0
−(1−n F )(eβ (i ω −ξ )−1)

n k
= .
i ωn − ξ k
Remember that the second term in the numerator here can be simplified if we use
e iβ ω =−1 ,i β n=(2 n+1)i π , and then I can write down the greens function the zeroth
n
− β ξk
0 (1−n F )(e + 1) 1
order greens function g (k , i ωn)= = because my
i ω n−ξ k i ω n−ξ k
1
1−n F = − βξ k
(e +1)
So, the temperature information is encoded in the frequencies omega n. So, now as we
said that the 0 temperature result would be obtained by analytic continuation of
i ω n → ωn +i η to the real axis and in the and so, in the same spirit as the 0 temperature
the minus the plus sign would relate it to retarded greens function and the minus sign
would be an advanced greens function.
So, this is how the 0 temperature limit is taken from the finite temperature and the reason
that the 0 temperature limit is relevant is that, because the bench marking temperature
239
scale in Fermionic system is given by the Fermi temperature. And the Fermi temperature
for metals it is of the order of 70000-80000 kelvin.
So, even a temperature which is room temperature 300 Kelvin can be taken as 0
temperatures. So, sometimes in fermionic systems, we may need or we can get by with 0
temperature properties or 0 temperature formula that is relevant even when the
temperature of the system is or the experiments are done at room temperature.
240
Prof. Saurabh Basu
Lecture – 17
Dyson's equation and disorder in electronic systems
So, let us do the Dyson’s equation for the Matsubara greens functions.
So, this is in line with what we have done for the 0 temperature greens function. So, from
the discussion so, far it is evident that the Matsubara greens function formerly looks
similar to that 0 temperature greens functions.
Excepting that, now we are talking about a complex time being bounded by minus beta
to plus beta, and the information about the temperature enters through the quantization of
the frequencies. And in case of fermions the frequencies are quantized in odd multiples
of pi over beta and for the bosons they are quantized in terms of even multiples of pi
over beta.
241
So, we need to now write down a Dyson’s equation in which we want to avoid doing a
finite order perturbation theory and want to invoke all orders of H prime into the sum
and we have seen that this is quite a doable exercise.
−βΩ −β k τk −τ k +
g (k , τ )= e Tr [ e T τ( e C k σ e )C k σ ]
So, consider the greens function and that is that
is the expression for the Matsubara greens function H to remind you that we are writing
it with a calligraphic g omega is the grand potential.
−βk
And K = H −μ N = H 0 −μ N + V and e =Tr (e ) . So, this is something that we have
−β Ω
seen and where V is the potential which typically represents a 2 body interaction term
say it can represent an electron-electron interaction of the coulomb, type which is what
we have been most familiar with. So, now, so, the K 0 problem that is this part has
already been solved. So, this is the K 0.
So, suppose we have been able to solve this exactly that is without the interaction term it
is been solved exactly. And also that our these

[H 0 , N ]=[ H , N ]=0 . And so, the
eigenstates of eigenstates of the number operator are simultaneous eigenstates of both H

0 and H. So, let us write down the operators in the interaction representation.
242
And write those unitary operators which we have introduced earlier, but with real time as
τ k0 −τ k −1 τ k − τ k0
U ( τ )= e e ;U ( τ )= e e if you remember that we have written down similar
operators with iT and with H 0 and so, this is not H, but I will write it as k. We had
written it earlier for the 0 temperature problem with tau as i T and k 0’s as H 0 and k as
H and so on.
Similarly we would have a needless to say that K and K 0 do not commute, I mean rather
there is a part of K 0 which does not commute there is a part of K which does not come
out with K 0.
So, as we have written it earlier the operators can be written as
C k σ ( τ )= e τ k ϵ k σ e − τ k = e − ξ
0 0 k τ
this has been proved earlier that is the tau evolution of the or
the imaginary time evolution of the operators.
And hence we can put them into the this
−βΩ −β k 0 β k 0 −β k τ k − τk0 τk0 −τ k0 τk0 − τk +

g (k , τ )= e Tr [ e (e e )( e e )(e C k σ e )( e e )C k σ ]
And this can be written as this can be written as we have a
−Tr [e−β k U ( β)U −1 ( τ )C k σ U ( τ )C +k σ ]

0
g(k , τ )= −β k 0
Tr[e U ( β)] .
Ah just show in a moment that how this thing comes.
243
So, we have U exponential minus beta omega that can be written as
e − β Ω =Tr [ e −β k ]=Tr [e − β k ( e β k e − β k )]=Tr [e− β k U ( β )]

0 0 0
So, a similar substitution can also be has also been made in the numerator for the
exponential minus beta k and that is how this exponential minus beta omega is replaced
by trace of exponential minus beta k 0 u of beta in the denominator. So, the equation of
motion, which in short will call as EOM can be for U of eta can be written as del del tau
u of tau will have 2 there are 2 factors in U consisting of exponential tau k 0 and
exponential minus tau k.
So, one has to be kept constant the other will be differentiated and vice versa will have to
δU ( τ) τ k
=e (k 0−k )e−τ k =−e−τ k H ' e τ k
0 0
be done and in which case we can write this as δ τ .
δU (τ)
=−(e τ k H e− τ k )(e τ k e τ k )=H ' ( τ )U ( τ )
0 0 0
So, δ τ
244
So, this part has been introduced and such that we can write this as a minus H prime T
sorry not T, but this will be tau this tau and u of tau. So, so we have seen this kind of an
equation, we can write down the equation of this differential or the solution of this
τ
U ( τ )=T τ exp [−∫ H ' ( τ )d τ ']
differential equation as 0 .
So, this is the solution of the unitary operator that we have introduced and all these
discussions are simply complementary to the 2 the ones that we have done for T equal to
0.
So, this is the solution for u of tau.
τ2
S ( τ , τ 0 )=T τ exp[−∫ d τ ' H ' ( τ ')]

τ1
Similarly we can write it for the . So, that is the definition
−1
S ( τ 2 , τ 1 )= U ( τ 2 ) U ( τ 1 )
of S matrix other properties of S matrix that includes that
245
And also the S ( τ 3 , τ 2)S ( τ 2 , τ 1)=S ( τ3 , τ 1) all these properties we have seen for the S
matrix. So, with all these inputs we can write down the greens function as
−Tr [e−β k T τ S (β , τ )C K σ ( τ )S( τ )C +K σ (0)]

0
g(k , τ )= −β k 0
Tr[e S (β )] .
So, at the end what is going to happen is that the disconnected diagrams as we have seen
are going to cancel with the denominator and will be left with only the numerator and in
which case the
∞ β β
n
g(k , τ )=− ∑ (−1) ∫ d τ1. ....... ∫ d τ n < T τ C k σ ( τ ) H ' ( τ 1) ....... H ' ( τn )C k+ σ ¿ (0)>
n=0 0 0
. So, that angular bracket is closing. So, this is the expression for the full greens function
including all the all orders of the interaction term H prime. As we told that the H prime
could be of the coulomb type or for that matter any other type and we are uncommitted
about the nature of the interaction, but if there is an interaction electron electron
interaction in the system the full greens function Matsubara greens function will be like
this where as u n equal to 0 will be the 0th order 0 and 1 will be till the first order and so
on. So, we can take do a perturbation theory in terms of H prime and can write down a
greens function to a particular order that is desirable to us.
Now, the question is that if for a given problem doing a perturbation theory up to a given
order is not a satisfactory treatment of the problem. It could happen that there are
suppose one takes care up to second order of H prime, and then it turns out that the H
third order is equally important as the second order and the fourth order could be as
important or at least halfway as important as the third order and so on.
246
In which case we have to go to higher orders and in principle we have to include all
orders of H prime and write down a like equation and this is what we are planning to do
here.
So, just to say that in this particular things only the connected diagrams survive all right.
β
So, now using the Fourier transform g(k , i ωn)=∫ d τ e iω τ g(K , τ ); ωn=(2 n+1) π
n
0
β
g(0) (k ,i ωn )
So, we write down Dyson’s equation as g(k , i ωn)= . So, it is
1−g(0) ( k ,i ωn ) Σ(k ,i ωn )
exactly a similar expression we have seen which we have derived for the 0 temperature
greens function.
Sigma being the self-energy and for a particular case we have also seen how to compute
the self-energy all these rules of writing down the diagrams or rather you know sketching
down the diagrams and getting out information about the scattering process follows
exactly in the same manner that we have seen. So, the rules just to give you once again
247
the rules is that with each electron line attached G 0 the interaction interaction is
represented by a wiggly line.
Conserve momentum and frequency at each vertex and also do not forget that the
Fermion frequencies are odd multiples of pi by beta whereas, the for bosons frequencies
are even multiples of pi by beta or we can write it as i pi by beta.
So, having done this formally let us try to do 1 example with the Matsubara function and
see that how it enriches our understanding of problems in condensed matter physics. In
principle disorder or defects is indispensable in condensed matter systems. So, these are
necessarily present and even if one tries to make an ultra-clean sample these are always
there and they constantly affect the experimental results depending on the degree at
which they are present. Sometimes they are necessary like in quantum hall effect they are
absolutely necessary to have the hall plateaus to be quantized either in integer or in some
chosen rational fraction of the quantity e square over H.
And sometimes they are non-desirable altogether in any case will do an example of this
kind, but before that let us do the equation of motion of the Matsubara greens function.
248
all right. So, we can we can take this k k prime and tau we can we can keep 1 index for
the momentum or we can keep two indices it does not matter, let us just for generality let
us take two indices and these are g(k , k ' k τ )=−<T τ C k ( τ )C +k ' ( 0)> Now
differentiating with respect to tau we have so, d d T equal to a minus d d T of.
Now, this time ordering as we know that can be easily replaced by the theta function. So,
d g( k , k ' k τ ) −d
will do that = [Θ( τ ) <C k ( τ ) Ck+' (0)>−Θ(− τ )< Ck+' (0)Ck ( τ )> ] . So,
dτ dτ
this is this is these are the 2 terms.
Now, to remind you that the derivative. So, this is not d d T this is d d tau and the
derivative of a theta function is a delta function. So, we have a minus delta tau delta k k
prime, now that comes because delta tau gives me tau equal to 0 in which they become
equal time greens function.
So, they will have to have delta k k prime otherwise this angular bracket will become
equal to 0 this is all what we have learnt, when we were dealing with Fermion operators
minus theta of tau and these are I am writing down for the time derivative to be in terms
of the equation of motion of each of the operators, which are
=− δ ( τ ) δk k ' −[Θ( τ )<[H , C k ( τ )]C +k' (0)> −Θ(− τ ) <C +k' (0)[ H ,C k ( τ )]> ]
d C k (t)
So, we have used =[H , C k ( τ )] where H is the total Hamiltonian for the
dt
system including the there is H 0 and v.
We have not taken for no particular reason though we have not taken the k here, but we
could you could just write down k instead of H without any difficulty. So, minus i so,
there are 2 minus signs. So, we will absorb that and write down the left the left hand side
d g( k , k ' , τ )
with a minus sign. So, this is = δ ( τ ) δk k ' + <T τ [H , C k ( τ )]C +k' (0)>
dτ
249
So, the important thing is that if H or k for that matter contains interaction term as
coulomb term ok, which is a 4 operator term or a 2 body term and then this side that is
this one will involve. So, H contains a 4 operator term and we are taking commutation
with C k. So, this will involve higher order greens functions such as 2 particle greens
function. So, when you try to write down the equation of motion for 2 particle greens
function here, it goes and demands that knowledge of a larger particle greens function
that is a 3 part of the greens function etc that should be available.
And then a 3 particle the equation of motion for a 3 part of the greens function would
require even higher particle greens function. However, there is a way to terminate that
series we are not going to discuss that here, for our case it is sufficient to talk about that
let us suppose.
H is only quadratic having a form. So, we are not talking about interaction terms
immediately let us just start with the free particle once. So, it is quadratic in electron
operators and it has a form which is H=∑ ( H 0 )k k ' C +k' C k

kk'
250
So, this gives that the [H , C k ( τ )]=−∑ (H 0 )k k ' C k ' ( τ ) that is the that is what is given
kk'
by the commutation of H with C k tau which you can check easily let us do that here. So,
H C k tau minus C k tau h so, this is the meaning of H and C k tau.
So, this is equal to [H , C k ( τ )]=HC k ( τ )−C k ( τ ) H=C k+ C k Ck −Ck C +k Ck =0 because

there are 2 annihilation operators while 1 can write this down change the order and write
it as (1−C +k C k )C k . So, C k will multiply with 1 and give me a C k and again seek a
dagger C k C k will give me 0 and this will give me a simple C k and that is what is
written there. So, the equation of motion yields.
So, will write it as u M as a as an abbreviation for the equation of motion. So, the
d g( k , k ' , τ )
equation of motion is = δ ( τ ) δk k ' + ∑ ( H 0)kk ' g (k ' ' , k ' , τ ) So, that is the
dτ k ''
form of the greens function of the equation of motion of the greens function, the first
term contains just the delta function which is operative at tau equal to 0 and it is a
diagonal terms that if k becomes equal to does not become equal to k prime, then that
term cancels out and the other 1 is the off diagonal term.
Now, let us imagine that ( H 0 )k k ' =ξk δk k ' +v k k' which means this is the diagonal term
and this is the off diagonal term. Why we are doing this we want to see the disorder
effects, which would be included through the second term which is of diagonal and the
kinetic energy is actually the diagonal term.
251
Now, the equation of motion is written as
−d
( −ϵ k )g (k , k ' , τ )= δ ( τ ) δk k ' +∑ v k k' ' g(k ' ' , k , τ ) So, that is the equation of
dτ k ''
motion for the Hamiltonian to have a diagonal term as well as an off diagonal term as I
told the diagonal term corresponds to say operators such as kinetic energy whereas, the
off diagonal terms can denote quantities such a disorder or defects in the problem.
So, this is my equation of motion now this differential equation can be converted into an
algebraic equation into an algebraic equation by using Fourier transforms and what do I
1
mean by that that is so, I have g(k , k ' , τ )= e−i ω τ g (k , k ' , i ω n) .
β∑
n
ω n
So, this is the the Fourier transform that we have and the inverse Fourier transform is
g(k , k ' , i ωn)=∫ d τ ei ω τ g( k , k ' , τ )

n
So, you see the g k k prime tau is has a Fourier transform in terms of the g k k prime i
omega n omega n 2 pi n i pi over beta. So, now, what we can do is that we multiply both
sides multiply both sides by exponential i omega m tau and integrate tau, from 0 to beta
252
which yields i omega m minus xi k or it could be here as we have not introduced the mu
explicitly w (i ω m−ϵ k ) g( k , k ' ,i ωm )=δ kk ' + ∑ v k ' 'k g(k ' ' ,k , i ωm ) .
k''
So, that is what comes when you multiply it by exponential i omega m tau and integrate
from 0 to beta and then we have used here.
β β
1
g 0, rather we have used d τ e i( p − p ) τ =δ mn and ∫ d τ e ip τ δ ( τ )=1
β∫
m n m
0 0
(0 ) 1
Now, we know that g g (k , i ωm )= So, then the full greens function is
i ωm − ξ k
written as g(k , k ' , i ωm)=g(0 ) (k , i ω m) δk k ' + g(0) (k ,i ωm ) ∑ v k k ' ' g(k ' ' , k ' ,i ωm ) .
k' '
So, this is the form of the greens function when we have the Hamiltonian contains a term
which is diagonal in k and there is also an off diagonal terms in k.
253
Though we are saying k we really mean any quantum numbers which is suitable for the
which forms suitable basis for the problem will have. So, there could be a lambda and
lambda prime in principle, which would the calculation would go through exactly. Now
you see that both the sides depend on i omega m since both sides depend on a single
variable at. So, at that at a particular value of the variable they will all i mean this
equation will be valid.
So, that tells that I can actually drop i omega m it is not required to keep it. So, in both
the sides and we can write down this as an infinite series. So,
g(k , k ' )=g(0) (k) δk k ' + g(0 ) (k ) ∑ v k k ' ' g( k ' ' , k ' ,i ωm )
k''
So, that is the greens function that we write.
we can propose a solution of the type in terms of an infinite series and which is of the
∞
form that g(k , k ' )=∑ g(n) ( k , k ' ) .
n=0
254
And so, if all the off diagonal elements vanish so, this is the first approximation if all the
off diagonal elements vanish. Then we have the g(k , k ')=g(0) (k) δk k ' . So, when we
have no of diagonal elements only diagonal elements the solution of the greens function
is purely the unperturbed greens function so the solution.
So, now if we insert this solution that we have written here into the equation that is here.
So, if we insert 1 into another, then we can iterate the solution and take out the zeroth
order because we have already written the zeroth order. So,
g(n ) (k , k ' )=g(0) ∑ v k ' ' g(n−1) ( k ' ' , k ') this is for n greater than equal to 1.
k''
So, n equal to 0 is taken out. So, if we iterate this then what we get is the following. So,
the first order g(1) (k , k ' )=g(0 ) v k ' ' g(0) (k ' ' , k ' ) .
And a g(2) (k , k ' )=∑ g(0) (k )v k , k g0 ( k 1) v (k , k ') g 0 (k ' ) v is the the term that we have
1 1
k1
taken as the off diagonal term here.
So, and so on. There will be g 2 and g 3 and so on.
255
So, for a general n, one can write
g(n ) (k 1 , k ' )= ∑ g(0) (k ) v k k g(0) ( k 1) .......... g 0 (k n−1)v k

1 n−1
k' g(0) (k ') .
k 1 ,k ¿
So, at the order we are simply able to write this as n non interacting greens function. So,
that which are solvable or which can be found out or we have already found them out
which is like i omega m minus i k so, all these things. So, depends on which order 1 once
one can find out this greens function till the order, if you want it till the first order or the
second order you simply need to plug in this v k k 1 or v k k 2, whichever the diagonal
terms are and these are the non-interacting greens function which we have already have
learned how to write them.
So, this is the pretty much the solution for this problem in which we are simply talking
about still not talking about the interaction term, but talking about the Hamiltonian being
quadratic in terms of the operators, what are these calculations good for let us apply them
to a disorder potential.
So, what it means is that at every lattice site there is a potential which could be randomly
varying from 1 lattice site to another lattice site. So, and which are which could be purely
random or which could be correlated, but this denotes an extra bit of potential at each
N
lattice site and this could be written as a V (⃗r )=∑ U (⃗r − R⃗ j ) . So, what we did is that
j=1
we have taken u to be the potential and there are j of j such disorder sites where the
potentials are present. So, which are given by all these are R u R minus R j and then
when we sum over all these n sites which have all these disorder potentials we get the
total disorder potential which is given by V of R.
So, V of R is the is the total impurity potential impurity or disorder whatever you want to
call it since we are talking about discrete lattice sites. So, it is a impurity is a more
appropriate word in this context from n static impurities distributed over G. So, clearly V
is not diagonal in the k basis ok.
256
And how do we know that we can write down V =∑ < k ' |V ( ⃗r ) | k > C k ' C k . So, this is
+
kk'
the kind of potential that we were talking about in our previous discussion when we
spoke about off diagonal term in the Hamiltonian it is still non interacting, but there is an
off diagonal term in the Hamiltonian which is something similar to that.
So, what is this matrix element this matrix element is given by k prime
N
1 ⃗ ⃗
<k ' | V (⃗r ) | k > = ∑ ∫ d 3 r ei ( k'−k ).⃗r U (⃗r − R⃗ j ) .
vol j−1
So, if ⃗r − R⃗ j=r⃗' r becomes equal to, ⃗r = R⃗ j+ r⃗' and then this becomes equal to
N
1 ⃗ ⃗ ⃗ ⃗ ⃗
= ∑ ∫ d 3 r ' e i( k ' −k). ⃗r U ( r⃗' )e−i( k ' −k). R = U (k '−k ) ρ ( k ' −k)
j
vol j−1
You just removing the vector signs knowing fully well that let us put this just to have
completeness and U of U of r prime and an exponential exponential minus i k prime
minus k dot R j where R j are the impurity sites. This is of course, you see that this is
independent of j, which means the impurity sites and we can simply write this as you k
1 ⃗
prime minus k rho k prime minus k where U ( ⃗k )= ∫ d3 r e−i k .⃗r U (⃗
r ) . So, this is the
vol
257
Fourier transform of these impurity potential and rho of k, which is simply equal to a
N
⃗ ⃗
ρ ( ⃗k )=∑ e−i k . R which gives the concentration or the density of the impurity.
j
j=1
So, it is clear that.
That in this problem in this problem we have a v k k ' =U (k '−k ) ρ (k ' −k ) which is the
∞
off diagonal contribution. And in which case we have g(k , k ' )=∑ g(n) ( k , k ') that is
k=0
a solution, and where g(n ) (k , k ' )=∑ g(0) U (k −k 1 ) ρ (k −k 1 )g(0) (k 1 ) +……… and all
k1
that other terms with k 2 for the second order and for the third order and so on.
So, one can iterate the solution , but in the first order this is the contribution coming from
the off diagonal term which comes specifically from disorder. So, we have been able to
write down a greens function or Matsubara Greens function function for the impurity
problem ok.
258
So, this is one of the examples that we wanted to do for with the Matsubara greens
function rest of the calculations as we have said exactly follows that of the T equal to 0,
now the only difference being that there we have dealt with temperature and frequencies.
Now here we would deal with complex time and complex rather frequencies, which are
having complex values and the temperature information actually enters through the
frequency terms.
259
Prof. Saurabh Basu
Lecture – 18
Introduction to electrodynamics Meissner effect
So, we are going to talk about superconductivity. So, having seen a various phenomena
in condensed matter physics such as magnetism then metallic behavior, insulating
behavior ferromagnetism, anti ferromagnetism and doing a thorough analysis with regard
to the greens function, let us now talk about superconductivity and initially we are going
to give you a glimpses of the historical achievements of superconductivity, that is how it
got developed in the earlier years.
And also we will try to give you some recent experiments on superconductivity which
are not very recent, but still the compared to the time that it was discovered there are
some recent developments and then we will talk about the most important thing which is
called as a BCS theory, the microscopic theory about superconductivity.
So, in these few slides we have tried to first give you the outline and then discuss with a
little more details and let us look at the periodic table.
260
And which are the superconductors or which form which has this phenomenon of super
conductivity, which is vanishing of electrical resistance below a certain temperature
certain characteristic temperature, which is a characteristics to that particular system. If
you look at it there are examples of superconductor the first superconductor was in
mercury and they are in lead and in tin and in indium antimonide and so on and also
some thallium based compound and in aluminum for example and so there are examples
of superconductivity here.
So, if we a nicely you know split the periodic table into two halves, the
superconductivity occurs at the right half of the periodic table; whereas the left half if
you see that there are nice magnets I mean including these are nice magnets and there are
superconductors which are denoted by the circles. So, superconductivity in the earlier
days have been seen in these materials and they from the colour code that is presented in
this periodic table, there are basically basic metals are they form superconductors and
there are these are somewhat like transition metals and semi metals and so on. However,
both of the magnets and the superconductors they have a lot of electrons at the Fermi
level.
However, in magnets the wave function of the electron or the you know the range over
which the electrons electronic wave function are distributed, have a certain range
whereas, for the superconductors the range is quite large and that is what pretty much
distinguishes metals or other the magnets or the elements which show magnetism with
the ones that show superconductivity.
261
So, let us go ahead with this and let us talk about a few books that are of importance, one
can look at books such as M. Tinkham is a very good book, G.D. Mahan we have already
been talking about there is a book called Tilley and Tilley their book called as by
David’s, D. Pines J.R. Schrieffer 1 of the discoverer of this BCS theory and there are
numerous review articles on superconductivity which you might want to look at it. So,
these are our starting point and let us have a historical overview superconductivity was
discovered in 1911 by gentleman called Kamerlingh Onnes in Leiden that is the name of
the place this in Europe and so after 3 years he had liquefied helium to reach
temperatures of the order of a few Kelvin.
262
Now why are they related or they have been placed one after another is that it is
important to achieve low temperatures in order to see superconductivity,
superconductivity does not occur at any temperature or at room temperature it happens at
very low temperatures. So, one has to cool system a metal, so to say to that temperature
below which it shows superconductivity and it is intimately related to the liquid helium
temperature, which is why it is mentioned which is about 4 Kelvin 4.2 Kelvin.
So, a liquid helium temperature is needed in order to cool a system and that is why the
superconductivity discovery of superconductivity is also intimately related to the ability,
to our ability of achieving low temperatures. Then the microscopic theory which is
known as the BCS theory, goes after the name we will talk about this a burden cooper
and Schrieffer that was given out in 1957, which is just about you know 60 years back
from now and then a new class of superconductors were actually discovered in 1986 they
are called high temperature superconductors. So, the super conductivity in these
conventional materials which are which are seen above in the sense that discovered by
Kamerlingh Onnes and then a few years the researchers have dealt with
superconductivity around with at most transition temperatures to be 20 Kelvin.
So, below 20 Kelvin they showed superconductivity and above that the
superconductivity was discovered, by the way Kamerlingh Onnes saw super conductivity
in ultra coal ultra clean sorry ultra clean mercury samples, which had shown a sudden
263
drop in resistivity at certain temperature below certain temperature other and we did a
very small narrow temperature range. So, this was about this was about 1986 and then
early 2000 there are superconductors which are based on iron. So, iron and
superconductivity they do not seem to go hand in hand because, irons are supposed to be
Ferromagnetic good Ferromagnetic and the Ferromagnetic imply that there are the spins
are all pointing in the same direction, that is what a Ferromagnetism is about.
However a superconductor requires a pairing of up and down electrons, I mean 2

electrons 1 having an up spin with a momentum k and the other with a momentum minus
k and a down spin. So, apparently there is an ambiguity, but there are instances and
experimental evidences of iron based superconductors.
So, what Kamerlingh Onnes did is that while he was doing an experiment with ultra
clean mercury and then subsequently with lead and tin, we have shown in the periodic
table that there are lead and tin etc showing superconductivity. So, the electrical
resistance electrical resistance that disappeared completely in a very small range of
temperature, we have some supporting figures for that you will see and that particular
temperature below which the resistivity vanishes is a characteristic of that particular
material and the complete disappearance of the resistance is demonstrated by the
existence of persistent currents, that flow in superconducting rings without any decay for
a year and in fact, the nuclear resonance experiments are predicted that they would only
264
be decaying in a time scale of 10 to the power 5 years, which is this is like 10 to the
power 5 years and that is why they are they are called as Perfect conductivity.
Now this word also has to be taken with caution because, very good metals have perfect
conductivity or rather infinite conductivity as T goes to 0 the temperature goes to 0;
however, they are different from superconductors in the sense that perfect conductors do
not show a property called as Meissner effect, which is said later and superconductors
show this property of Meissner effect and superconductors show perfect diamagnetism,
which is again what we was discovered by Meissner and Ocshefeld felt and this is
something that we are going to discuss; this is the electromagnetic properties of
superconductors which are very important.
So, super conductivity can be destroyed, why we are talking about destruction of
superconductivity is that then we understand how robust the superconducting state is.
So, there it can be destroyed by two means, that is one is thermal effects that is if you
raise the temperature of the system or if you apply an external magnetic field, both these
can actually destroy superconductivity and the system will go on to a so called metallic
state. It does not have to we will see that in high Tc superconductors they do not go on to
a state which is exactly metallic, but there are some deviations there as well. In any case
265
so there is a critical temperature and a critical magnetic field that exists below is the
superconductivity is robust and above which superconductivity vanishes, which means
that we have metallic state that emerges beyond that. So, one superconductivity is
destroyed there is a emergence of a metallic state.
So, we are going to be concerned about mainly 2 aspects of superconductivity and one of
them being the electromagnetic response as we have just said which takes care of the
Meissner auction field effects and so on and this was discovered by or rather they are
described by London equations, will give us a small overview now and then we will do it
later on in an elaborate fashion.
And then also we would be concerned about the microscopic theory of

superconductivity, which is known as BCS theory and you will see that how it was
actually coopers problem it is called cooper instability. It is shown that 2 electrons which
are known to be having the same charge and hence would repel they can be shown that
they actually have an attractive potential operative between them, when they are in
presence of a filled pharmacy and this attractive interaction is mediated by via phonons.
So, also we may talk on the phenomenological theory of superconductivity which are
known as Ginzburg Landau theory. So, this phenomenological theories in general are
very good starting points when microscopic theories are unavailable and they provided
you take into account the correct symmetries of the problem and you know that how a
266
system loses superconductivity, then you can write down a free energy functional in
terms of some wave functions or rather some order parameters which are going to give
you the; I mean particularly this the phase transition is captured by writing down the free
energy functional in terms of the order parameters. So, a very brief a description is first
presented about each 1 of them and then we will go on for a little elaborate discussion.
So, to start with it was we said that it is a electrodynamics. So, these were by 2 brothers
F. and H. London F. London and H. London in 1935. So, they actually were wrote down
equations where the electric field and the magnetic field variations, inside a
superconductor can be described in terms of the number density of the superconducting
electrons or the super current density.
So, it is important to say that normal current does not flow through a superconductor,
because of this super conductors being perfect diamagnets; the normal currents will be
killed and they cannot penetrate into the superconducting sample. Whereas, the super
currents what we actually mean by super currents will be explained later, they can be
they can propagate or rather they can they flow in the superconducting sample and it
nicely eels starting from these equations, which are nothing but some variations of the
Maxwell’s equations of electrodynamics. one can get diamagnetism from and hence
Meissner effect this is what we have talked about.
267
Then we have this microscopic theory that should come into the discussion, which was
again put forward in 1957 by 3 people called Bardeen cooper and Schrieffer and the
superconductor. So, it will show that how the energy gap in a superconductor varies with
temperature. So, what we mean by an energy gap is that there is a ground state and there
is a first I mean there is an excited state, where superconductivity is basically destroyed.
So, when the system is in the superconducting state, the ground state becomes occupied
by all these pairs of electrons. So, they form pairs which are actually pairs of up and
down electrons and then there are many of these pairs that are present and the ground
state is actually a many body ground state of all these pairs.
So, their swoops of pairs and in order to break a pair, we can populate this the first
available excited energy level by breaking the pair and that breaking can be done as we
have said either by applying temperature or by the magnetic field. So, how do these
electrons have attractive interactions, we have already said that after that that the
interaction is mediated via phonons they can still have attractive interaction and these
BCS theories they properly explain the thermodynamics and the transport properties such
as the vanishing of resistivity below a certain critical temperature and then the Ginzburg
landau theory as we told it is a phenomenological theory.
268
It does not have any microscopic basis and it was put forward ahead of BCS theory,
when BCS theory was unavailable and what it did is that it is it wrote down the free
energy functional in terms of powers of the order parameter or basically the complex
wave functions that describe a transition from 1 state to another.
So, it concentrates on the superconducting electrons and not on it is excitation it is based

on the landaus general theory of second order phase transition, that is why this name
Ginzburg landau; it is coming and in terms of free energy and so free energies is a
functional of the order parameter.
So, 1 is to identify the order parameter and know that how it vanishes that the phase
transition then 1 can write down a free energy functional and do calculations of say
electrodynamics and it also of very importantly introduces 2 length scales of the
problem, that distinguishes type 1 and type 2 superconductors we will see what a type 1
and type 2 superconductors. So, broadly all these conventional superconductors which
were discovered before 1986, they all can be plugged into 1 of the other type of
superconductors and so let us go to slightly more elaborate discussion of
superconductivity.
269
So, it is this gentleman which we said H. Kamerlingh Onnes and he was working with
very clean absolutely clean mercury and it found he found that the resistivity above 4.2
Kelvin it has a linear behavior here and suddenly in the vicinity of 4.2 Kelvin. In fact,
within a temperature range that is very narrow it is of the order of 10 to the power minus
5 Kelvin, it does not show it that way, but it is it is true that there are. So, this fall off is
extremely sharp and so below 4.2 Kelvin the resistivity becomes so small almost
negligible, so of the order of 10 to the power minus 5 ohm.
So, the left of this transition is called as a 0 resistance state and this is a mercury the plot
is that of a mercury showing transition Tc being, which is the temperature that
demarcates metallic regime and 0 such state regime is that temperature is 4.2 Kelvin. So,
what happens is that there is a sudden loss of electrical resistance electrical resistance is
denoted by the symbol rho, below a certain critical temperature Tc, there are very sharp
superconducting transition as we have seen; the magnetic fields are completely expelled
from superconductors.
So, if there are magnetic fields that are going to be completely expelled from the material
from the specimen and they cannot enter other than a very small depth of the specimen
which are called as penetration depth, we will see that and that is why because there is
absolutely no magnetic field, so that is why they are called as perfect diamagnets which
270
M
the diamagnets susceptibility becomes equal to minus 1. So, χ =limit H →0 =−1 . So,
H
So, M and H are actually oppositely directed and these this is the definition of perfect
diamagnetism, in even the best of the systems that show diamagnetism other than the
superconductors the or rather a chi is really small it is of sometimes of the order of 10 to
the power minus 4 or 5. However, in superconductors it is equal to 1, then as we have
said earlier that there are lost less energy conduction and generation of strong magnetic
fields.
So, in the vicinity of the superconductor there are generation of very strong magnetic
field we will show you at the end of videos showing that how magnetic field is expelled
and Japan is actually making train which would be realizable around 2027 or 2028 these
are called as a maglev trains and this maglev’s is the short form of magnetically
levitated.
So, let us see that what happens for a good metal and a superconductor, in a good metal
the resistivity decreases and as T goes to 0 it takes it gives you a value which is finite.
So, this is r equal to r naught equal to 0, whereas for a superconductor it decreases and
then it suddenly drops to 0 at a temperature which is equal to T equal to Tc. So, that is
the basic difference between good metal and a superconductor.
271
So, there are several things that 1 had to take care of in experiments and so Kamerlingh
Onnes felt that the cold wires resistance also would dissipate. So, this suggested that he
would there would be a steady decrease in electrical resistance allowing for better
conduction of electricity and then owns passed a current through a very pure mercury
wire and measured the resistance as it steadily lowered the temperature and that is how
the experiment went off; much to his surprise he saw that there was no resistance at 4.2
Kelvin implying extremely good conduction of electricity and this is the discovery of
superconductivity.
So, what the question is why is resistance 0 in a superconductor electrical resistance is

caused by electrons, that lose energy by scattering between or with each other. So, there
are new channels in which the energy can be dissipated that opens up and because of the
scattering between electrons the energy dissipates, the electron pairs are all in their
ground state and hence cannot lose energy.
For superconductors the electron pairs the they form pairs and they go into the ground
state. So, there is no way that they can lose energy, so the super conductivity is a
phenomenon of the ground state. However, the need to acquire at least a kinetic energy
equal to the energy gap to lose energy by breaking a pair. So, there are you need to give
it an energy a kinetic energy such that the pairs dissociate and they break and the there
are some electrons that are made available in the excited energy state.
The question is that is the resistance really 0 the experimental answer is that a lower limit
of, what we have what we have said already with 10 to the power 5 years is established
for a decay of decay time of current in a superconducting loop, which is 1 0 to the power
minus 5 precision over a year; which is an extremely good signature that the resistance is
really equal to 0 and the current is really persistent.
272
So, we have said this that what kill superconductivity, so we need to provide enough
energy for the pairs to dissociate and the pairs remain within the coherence length of
each other.
So, there is a thermal energy that can be given and there is also a critical current density
this is something that we have not said earlier, there is also a critical current density
which superconductor can hold and there is also a critical magnetic field which is needed
for it to the break super conductivity. So, Tc Jc and Hc are the critical values
demarcating a superconducting state from a non superconducting state. So, that question
is that what drives superconductivity what causes a superconducting transition.
273
So, the density of states at the Fermi level call it D(E f) or n Ef depending on which book
you are following. So, the density of states at the Fermi level plays a critical role, these
are the most unstable electrons in a solid unstable because they can actually cross the
barrier and go on to energy levels that are available to them.
Whereas the ones that are actually much farther away from the Fermi energy are very
robust they do not make any transition. So, these unstable electrons they can easily
rearrange themselves to lower their kinetic or lower their energies. So, electrons near the
Fermi level are able to lower their energy when the superconducting gap opens up. So,
these electrons in the vicinity of this thing can actually lower their energies and. So, this
is like acting like a refrigeration because so when there is a superconducting gap that
opens up.
So, what we are trying to say is that it is only the electrons in the vicinity of this can
move into the gap and the other electrons they do not make any transition and make any
attempt to go and occupy other energy levels, that is there is a selective cooling which
are the most energetic electrons that are near the Fermi surface and these electrons are
actually going into states which are which are occupied and thereby causing cooling to
the rest of the system.
So, this is one typical this thing called diagram called as then for the T greater than Tc is
for a normal metal, where we plotted the density of states on the x axis and E on the y
274
axis in a lot of other books, we will see that this is access an interchange but they carry
the same meaning and for a superconductor there is a Fermi energy at the Fermi energy
there is a gap that opens up and that gap has to be some you know accessed or rather it
has to be overcome, in order to have the electrons going into the normal state or rather
the superconductivity is destroyed and this corresponds to a gap which is given by the
metallic gap or rather the gap that we see in usual cases. So, this is the gap energy and
we need to get over the gap energy or rather have supply larger energies than the gap
energy in order to describe super conductive destroy superconductivity.
So, each electron gains delta and the pair gains to delta and so the number of electrons
that are involved in this in this process is half of Eg because, this is for the twice of the
gap and then multiplied by the density of states at the Fermi level. So, these are the
number of electrons that are available for causing superconductivity in a sample.
So, what are the experiments which see super conductivity or sort of make inferences
about superconductivity, they are infrared absorption which does the measurement of the
energy gap, they are tunneling which are which do the measurement of the
superconducting density of seeds and show these that a superconductor has a gap that
appears at the Fermi level photo emission. Photo emission tells you that if there is a
momentum dependence of the gap whether the gap is homogeneous at all k values or
275
there is a particular fashion or rather an orientation in which the gap is smaller along
some direction of k or it is bigger in some other direction, NMR it probes short pairs.
So, usually these pairs have are defined by a coherence length, which is basically the
resulting the size of the resulting wave function and there are also smaller pairs. In fact,
this high Tc superconductivity is caused by or it is believed to be caused by shorter pairs.
So, NMR probes this existence of shorter pairs and of course XRD which is x ray
diffraction it does the crystal structure determination of all materials including the
superconductor.
So, we will see this magnetic levitation in a video that we have said and this is one of the
properties, now how does the magnet levitate in presence of a superconducting sample
because, there is a very large magnetic field density in the vicinity of the
superconducting sample; that is because the magnetic flux is very large in the vicinity
because all the magnetic field lines are pushed out of the superconductor and they bunch
together.
So, much so the magnetic energy is able to lift with the weight of such a disk or a ring,
superconductors also have electronic and magnetic properties. So, that is we already
have said that they have a negative susceptibility and they acquire a polarization opposite
to that of the applied magnetic field and that is why the superconducting materials and
magnet they actually repel each other, so this is the reason for that. If the temperature
276
increases the sample would lose it is super conductivity and the magnet will not be able
to float on the superconductor because, superconductivity is destroyed so the Meissner
effect goes along with that. So, there will be no more exclusion of the flux and so the
magnetic energies go away which used to support the weight of the ring. this is in this
cartoon there is a sumo wrestler which is shown that he is standing on a platform and the
platform has a small gap as it is shown here between the grounds and that so the
superconducting platform.
So, because this plank or the layer of the material on which the sumo wrestler stands is
shown here this 1 the top 1 and the bottom 1 is a superconducting sample. So, this
levitation can actually support the weight of a sumo wrestler this is what is being said in
this diagram.
And this is the maglev train which is not yet operational, but they are planned to be
operational they would reach velocities more than even the bullet trains and the reason
being that they would be levitated and they would not be having any super conduct or
any friction from the rails which usually impede the velocities; these are nice looking
trains and they are probably artists impressions of how the trains are going to look like.
277
And this is of course some of them are only our a futuristic ideas and ambitions that we
will have that there are large scale applications of superconductors which are which can
begin to appear, like American superconductor 2G wires.
So, there are ybco that is yttrium barium copper oxide and Y2O3 yttrium oxide nano
particles for flux pinning, one can have the superconducting power cables such that there
are persistent current and you do not have to add sustain the current 1 starts it will keep
going on and various other things I mean there are you know weight reduction of the
wind turbine, 10 megawatt generator which will weight 120 metric tons instead of 300
using conventional copper wires and so on and these are the applications which we can
see in the future.
278
There are medical applications, so these are these ones are actually superconducting
magnet coils, which produce very large uniform magnetic fields that can penetrate
through the patient’s body and will give information if there is an anomaly or there is a
pathological defect in the human body and then the actual treatment can proceed a lot of
medical diagnostics are now dependent on MRI doing MRI and so they crucially have lot
of applications in the medical sector.
279
So, to talk about those important developments and some dates etc, 1935 a London
brothers proposed these equations on E and H and these are called as the London
equations, 1950 Ginzburg and Landau proposed a phenomenological called as the
Ginsburg landau theory, 1957 is a BCS theory microscopic it is a variational theory and
cannot be obtained by doing any finite order of perturbation theory and 1986 is Bednorz
and Muller discovered superconductivity at high temperatures and then there are lot of
other, like the iron based superconductors are called as a Pnictide superconductors, they
are being discovered including the magnesium boride and the arsenide and now 1 is
talking about the topological superconductors.
So, pictorially what happens in the meissner effect, we will do all these calculations as
we go along. So, the meissner effect it was discovered by Walter Meissner and Robert
Ochsenfeld in 1933, so this is the total expulsion of the magnetic flux. So, that is why the
superconductors are called as the perfect diamagnets, applying a magnetic field to the
superconductor by replacing the magnet above, it creates surface current that produces a
magnetic field to perfectly counter the applied field.
So, this is T greater than Tc though the magnetic fields penetrate as T is lowered below
Tc this it becomes this blue ball becomes superconducting and this it starts repelling the
magnetic fields and now you see that in this region there is a large flux of the magnetic
fields because it is bending around the obstacle and this as it says that what happens is
280
that it creates some surface currents and the surface currents produce a magnetic field in
a direction that exactly cancels the applied field.
So, let us have a distinction between the perfect conductor and the super conductor. So,
this is a superconductor so it is a superconductor or rather let us start with 1. So, it is a
conductor upon cooling it becomes a superconductor, if you switch on the magnetic field
the magnetic fields get say expelled and these are called Meissner effect and the
magnetic field if you on. So, if we relaxed that is if we raise the temperature so this is
raising temperature. So, superconductivity is destroyed and the flux lines will penetrate.
Now what happens for an ideal conductor if it is an ideal conductor the magnetic field is
on, so it is cold magnetic field is on the magnetic field is off that it is it completely the
flux lines completely vanish and if it is put in a magnetic field then the magnetic field
will penetrate if you cool, the magnetic fields will still penetrate and when we switch off
the magnetic field.
Then for an ideal conductor we will have that these things will make circles the magnetic
field lines will make circles whereas, this step is different than what we have for
superconductors. So, superconductors it completely sort of. So, if we switch off the
magnetic field then we do not have such trapping of the flux lines and that distinguishes
a superconductor from an ideal conductor.
281
Now, there are 2 kinds of superconductors we have already said type 1 and type 2
superconductors. So, type 1 superconductor is that they are pure metals which is what we
have seen in the periodic table, they are perfect diamagnets and they have lower critical
field. So, the M versus H diagram which is called as a phase diagram of superconductors
it, as you increase H the M keeps going linearly and then suddenly it drops as you further
increase eight.
So, as we know that if we give if you apply magnetic field superconductivity is

destroyed and then it gives rise to a normal stage. So, this is the superconducting state
inside and there is a normal state here. So, this is another sort of this thing for how the
magnetic field falls off inside the, so this is the distance inside the superconducting
material. So, it falls off magnetic field falls off within a length scale lambda, whereas the
wave function actually grows like this.
So, this is called as ξ and this length is called as λ as you can see this length is
called as lambda and this called as a penetration depth and this is called as a coherence
length. So, 1 can actually talk about a single quantity kappa, which is ratio of λ< 1
ξ √2
gives you a type 1 superconductor whereas, kappa greater than 1 by root 2 gives you type
2 superconductor.
282
Which means that if lambda is less than xi, so lambda by xi less than 1 by root 2 which
means lambda is lambda is less than xi over root 2, then we have a type 1
superconductors and if there is lambda is greater than xi over root 2 there is a root 2
there, then we have a type 2 superconductors and similarly the magnetic field versus
temperature shows that this the regime bounded by the red line is called as a
superconducting state and above that which is the normal state.
So, there is a Tc beyond which superconductivity destroys and there is also a Hc beyond
which superconductivity destroys and so this is a type 2 superconductors it occurs in
alloys. So, there is a combination of different elements it is alloys or inter metallic
compounds and they exhibit a much higher Tc. So, the phase diagram is not exactly like
what we have seen here it is dissimilar and there is a linear increase and then there is a
fall off like this, inside is a superconducting state a part of it; but a part of it what is
called as a vortex state and beyond that it is a normal state.
So, what happens is that the magnetic field can partially penetrate into the sample and
beyond that when the magnetic field is increased beyond the super conductivity is
destroyed; however, the magnetic flux lines that get trapped and this trapping also
happens in some nice regular ordering and in particular it happens in a form of a
hexagonal lattice, which is called as the abriskosov lattice. So, this is the lambda and
283
this is the psi and now you see that it penetrates much larger into the superconducting
sample.
So, this is d is the distance inside the superconducting sample, whereas the wave
function of the of the electrons are smaller and this is a typical diagram showing that
there are superconducting rings these are superconducting rings and there is there is a
magnetic field lines that are that are applied and so the current super current goes in
these superconducting rings or in the ring fashion the current actually circulates in about
the magnetic field lines and these are the experiments demonstrating persistent current.
So, finally just small glimpse of what came from Bardeen cooper and Schrieffer, so these
are the photographs of them and this explains superconductivity at low temperature for
all conventional metals and it was started with coopers work he realized that the atomic
lattice vibrations were directly responsible for causing attraction among the electrons,
which allow them to pair up into teams pairs that is that parcel ops all obstacles which
cause resistance.
So, when an electron pairs up with another electron forms a bound pair, this bound pair
seamlessly move without having any resistance or rather without having any collision
suffering any collision from another pair. So, this team or the pairs of the electrons which
are called as the cooper pairs, they move without any resistance through the sample. The
284
features of BCS theory the first microscopic theory it works well for low temperature
superconductors, it is variational and non perturbative we will say all that.
So, no small expansion parameter the superconducting gap is non perturbative, the
predicts Tc for materials under 20 Kelvin; the phonons are essential components of the
theory London’s electrodynamics is explained meissner effect is included at the linear
response level. So, there is a tremendous success including all these things.
just to have to make the discussion ongoing discussion a little more interesting, there is
an analogy of dancing room scenario that was given by Schrieffer; that there are these
males and females pairs that are dancing on a dancing floor and they are completely
unmindful of another and they keep dancing without suffering any collision from each
other. So, there are a large number of them and they can you know they pass on without
suffering any collision between each other, the super conducting ground state is exactly
like that.
285
We can talk about this in details later we will do that, but it is just that there are. So,
these electrons actually can form a pair in presence of phonons, that is what is said here
we will come to this.
Similarly this ones.
286
And these are some of the superconducting transition temperatures that are listed here,
for materials which are like tungsten mercury sulfur based organic superconductors
Nb3Sn and so on.
So, you see that suddenly from 23 to 40 there is a large jump and from of course 40 to 92
there are large jumps and so on. Now this is that these are these high Tc, so high Tc
superconductors.
287
this is how the transition temperature has increased since 1908 it was going fine till about
1986, this is where 1986 is then it shot up and went all the way up to 164 Kelvin. So,
there are you know Arsenide’s which are discovered around this time.
these are the people who discovered this Bednorz and Muller, this these high Tc
superconductors the they actually got a Nobel prize in 1987 and there are still no theory
because of a number of reasons one of them being that the normal state is not known,
pairing is not phonon mediated, experimental data contradict each other coherence length
is too short there is something called a pseudo gap.
It is like a gap but it violates other properties of an energy gap, there is no controlled
expansion parameters and many others and that is the reason that even the best of the
minds working in the field have not been able to come up with the general consensus that
what the theory that should be there, which would explain all experiments.
288
And if one looks at the phase diagram which is a temperature versus doping, so this
doping gives the holes which are rather the carriers of superconductivity in this case.
there is a superconducting dome that is here, but as you go up you go into a place which
is most of it is not known and that is the reason that the theory is not known; whereas, in
BCS superconductors all these form nice metals which are pretty much here. So, had all
these area here would have been like the Fermi liquid, we would have already had a
theory these are some layered structures of these high Tc.
289
So, basically the essential ingredient is a copper oxide plain in all these things. So, it has
a Tc of about 40 Kelvin 77 Kelvin and 135 Kelvin is a lanthanum strontium copper oxide
this is called ybco this is called ybco, this is a yttrium barium copper oxide and this is
called bisco it is abysmal strontium calcium copper oxide and so on.
So, let us show you some cartoon for this for the levitation and magnetic levitation in lab
experiments and.
290
So, this is a superconducting sample that black thing and now there is a coin like thing is
kept their, it is kept on top of it. Now this superconducting specimen is just a metal, now
they are as if the temperature is being lowered by liquid helium or some other coolant,
now it became superconducting now you see placing it on top.
It is just floating this is called as superconducting levitation and it is simply levitating

and if you just move it still does not fall off onto the material. So, this black thing that I
see is a superconductor now because, the temperature is lowered and you see there is no
trick there. So, it is been there is a film has been swiped in order to show that there is
nothing there so and now it is spinning also it is still floating as you try to bring it closer
it gets repelled, because the magnetic flux lines that has enough magnetic energy in order
to support the weight.
Now because it is kept in a kind of open container the temperature is being raised, it
lifted that thing itself the material itself and now you have kept taken it out slowly the
superconductivity is destroyed and meissner effect is gone the flux are no more expelled
and the material just falls crashes on top of the superconducting specimen.
291
So, this is the direct lab demonstration of meissner effect.
292
Prof. Saurabh Basu
Lecture – 19
London penetration depth, Type I and II superconductors
So, to continue with the discussion on superconductivity.
I will repeat a few things for your convenience and so, the sudden drop of resistivity
below a certain temperature is called as a superconducting phenomena or this gives rise
to superconductivity and the temperature below which the resistance vanishes is called as
a critical temperature, which is a property of a particular material and so that T C the
superconducting transition temperature is usually of the order of few Kelvin for the
conventional superconductors.
Now, this we had discussed yesterday that there are classes of unconventional
superconductors which are also known as high temperature superconductors and they are
not as extensive knowledge about them as they exist for the conventional ones.
However the T C is really vary from ah few Kelvin to about twenty 3 Kelvin for this is
for Nb 3 Ge and the features are one 0 electrical resistance or resistivity and no change in
crystal structure and this is verified by x ray diffraction both below T C and above T C
293
that is in the normal state and in the superconducting state and third is that it is the state
the superconducting state is characterized by a conductivity to be infinite. B the current
density to be still finite .c is that the electric field goes to 0 and (d.) is that the magnetic
field is constant and this cannot be explained by classical electrodynamics because ohms
law says that j equal to sigma E for j to be finite j is the current density.
For j to be finite sigma has to go to sigma tends to infinity then E has to be equal to 0 0.
−δ B⃗
So, this is the third there is a c condition and also that ∇× E
⃗= that gives you B
δt
to be constant this is number d. So, these are some of the features of the superconducting
state.
Let us now look at Meissner effect. So, the expulsion of the magnetic field from the bulk
of the superconducting material is called as Meissner effect. So, this complete and
sudden vanishing of the field or rather as the system goes into a superconducting state is
really something that is that distinguishes it from an ideal or a perfect metal, let us see
that.
So, Meissner and it is usually known as Meissner effect, but it is also said as Meissner
and Ocshenfeld in 1933 they discovered that there is a complete exclusion of the field
lines magnetic field lines. So, the way the experiment can be done is that one can take a
294
superconducting sample and then put a lot of iron filings around that sample and then
apply a magnetic field and along with that cool the temperature of the specimens.
So, that it enters into a superconducting state as it enters as it is in a superconducting

state then the magnetic fields will be pushed outside the sample and then the iron filings
would get lined up in a regular fashion outside the sample. So, let us think that this is
what the sample is like. So, the flux lines will go through this if it is t is greater than Tc
and as T becomes. So, the iron filings are all scattered in around this superconducting
sample and now as T goes below T C, then these flux lines the magnetic field lines are
pushed out like this and because of that the iron filings will nicely aligned around the
superconducting sample.
So, this is the tabletop experiment for seeing Meissner effect and it is seen for
superconductors such as lead and team and so on. So, now in certain materials it is only
above a certain critical magnetic field which is of the order of a few or states there is no
expulsion of the magnetic flux.
So, whence the superconducting super conductivity disappears the material reverse into
the normal resistive state and the magnetic field fully penetrates through it and so, there
are materials in which magnetic field can penetrate up to a certain extent beyond that if
you increase still the value of the strength of the magnetic field then it gets pushed out
and this is called as a type 2 superconductors which allow some magnetic for some range
of magnetic field for the magnetic field to penetrate inside the sample.
295
Now, let us do a thought experiment of to show the difference between perfect conductor
and superconductor this has been told yesterday or rather in the last discussion through
slides I just wanted to make it a little more clear by drawing the diagram. So, this is a
typical superconductor and this is a perfect conductor and so, say the state a is that this is
a superconducting materials and this is cooled below T C and then it becomes
superconducting and then magnetic field is applied and then there are you know the
expulsion of the magnetic field as we have said and then magnetic field is withdrawn
field withdrawn superconductor goes back to its original state no memory and what
happens to a perfect conductor.
So, this is a perfect conductor and. So, there is an external it is in an external field. So, it
is cooled below T C the flux still penetrates, but it is resistance less in presence of
magnetic field now when field is withdrawn it goes into a state which is. So, this is
magnetic field turned off of and it has a memory. So, because the perfect conductor does
not have Meissner effect this is what happens.
So, if you compare. So, this is a, this is b, this is c, and this is d, and now we have this let
us call it as e, this is f and this is g. So, if you compare d and g, it is easy to see that a
perfect conductor depends on its history and Meissner effect does not happen in perfect
conductor and that is what distinguishes perfect conductors from superconductors. So,
296
the total exclusion of the magnetic field from the inside of the superconductor is a
property which is called as diamagnetism and you can understand this as follows.
So, we have B=μ0 ( H ext + M ) external is the externally applied magnetic induction, we
have not written in vector form, but they are vector equations. So, since B equal to 0
inside a superconductor the induced M M exactly cancels H external. So, susceptibility
M
χ= =−1 and the as we have said earlier that nothing is more diamagnetic than a
H ext
superconductor.
In fact, in best of the diamagnetic metals have susceptibility which is extremely small of
the order of 10 to the power minus to 10 to the power of minus 5 to 10 to the power
minus 6. Now, let us a look at the electrodynamics on superconductors and particularly
we are going to talk about the penetration depth.
So, let us look at penetration depth and before we do any calculation let us say that ah.
So, we have claimed several times the magnetic field is totally expelled, but the fact is
that that it enters only up to a certain distance which is called as the penetration depth
now consider the following geometry to compute the distance through which it it
penetrates. So, this is the superconductor and one has applied an external magnetic field
here and lets call this direction B x and this direction B y ok.
297
So, this is the geometry of the sample now the Maxwell’s equations can be written as
δD
⃗ −δ B⃗
B =μ 0 ( J⃗s +
∇× ⃗ ) these are the last 2 Maxwell’s equation ∇× E
⃗= . So,
δt δt
this is the equation Maxwell’s equation that we all are aware of this is classical
electrodynamics we are simply doing classical electrodynamics and on super conductors
and trying to get some information on how much magnetic field can penetrate inside a
sample.
Before it gets completely expanded. So, there is a certain critical depth up to which the
magnetic field can penetrate and this by no means contradicts the statement that we have
made earlier that in superconductor the classical electrodynamics laws are not strictly
valid. We are simply trying to get some information out on a quantity called as
penetration depth and to tell you a priori the reason is that that this is one of the
important length scales of the problem and this length scale along with another one
called as the coherence length determines the material to be of type one or of type 2. So,
what happens is that since the displacement current.
So, let me go through it again B =μ 0 J⃗s , J s is the super current density and del D
∇× ⃗
del t is the variation of the displacement current density or displacement current and this
is that E and B are the electric and the magnetic fields respectively.
298
Now, the displacement current variation or displacement current rather this is a, which is
the time derivative of the displacement vector. So, the displacement current is far smaller
than the super current. So, in this term you can neglect this one. So, neglect this and
B =μ 0 J⃗s
write the Maxwell’s equations as a ∇× ⃗ and J s is nothing, but it is equal to
J⃗s=ns e v⃗s which is the super current density is the density of the super electrons
multiplied by the electronic charge.
And the velocity of these super electrons and the equation of motion of these super
electrons can be written as in an electric field to be m v⃗s=e ⃗
E that we are talking
about. So, if we call this as 1 this as a 2 and this as 3 and call these equations may be as
A and B. So, what happens is that if you put 3 or rather if we have if we take
differentiate 2 with respect to time. So, J⃗˙ s=nS e v⃗˙s and I can replace v s dot from this
equation 3 so, n s E square E over m from 3.
d J⃗s n S e2 ⃗
E
So, that is the equation for j s dot. So, this is equal to = this is how the
dt m
super current varies with time now if I put 4 let us call this as 4 if I put 4 in B putting 4 in
δ ⃗B −m δ J⃗s −m δB
⃗
B, I will have = 2
∇×( )= 2
∇×(∇×( ))
δ t ns e δt ns e μ0 δt
299
So, that is curl and curl off for this and hence what we have is that the dB dt. Now, I can
δ ⃗B δ ⃗B
write it as a complete differential is equal to =− α ∇ ×(∇ ×( )) . So, this is the
δt δt
m
equation for dB dt and α= that is what this is defined as alpha. So, now, this
μ 0 ns e2
δ ⃗B ⃗ .d⃗
⃗ (∇ B ⃗ 2( d ⃗
B
is written as the right hand side is written as ∇×( ∇×( ))= ∇ )− ∇ )
δt dt dt
Now this term is equal to 0 because one can swap the order of space and time derivative
d
and can write this as (∇ . ⃗
B)=0 which one knows by the second Maxwell’s
dt
equation that this is equal to zero. So, this term becomes equal to 0 and hence my
δ ⃗B d⃗
B
∇×( ∇×( ))=− ∇⃗ 2 ( ) . So, if you put this into the equation number let us call
δt dt
d⃗
B ⃗2 d ⃗
B
it as 5 putting 6 in 5 =α ∇ . So, doing a time integral this can be written as
dt dt
⃗ =α ∇⃗ 2 B
B ⃗
So, this is the equation that gives the space variation of B and now since the variation
according to this diagram the variation can be in the x direction. So, let us convert this 3
dimensional equation here by to one dimension because we can safely assume that d
δB δB
= =0 and hence we can write it down as a being equation 7.
δ y δz
300
2
We can write it down as ah B x= α δ 2 B x this is really ah partial differential if you
δx
like it is like del del and so on and this has a simple solution which is equal to B external
well this is like let us not write be external. Now we can write it as the solution of this
−x x
λ λ
can be written as is B ( x)= A e + B e L L
nothing, but.
So, alpha is lambda l square and one can easily see that that these because as x goes to 0.
So, this superconductor is say from 0 to some d that is the width of the superconductor in
that case at x equal to 0 B is equal to B external and so on. So, this can be written as
−x x
B ( x)=B ext e λ + Bext e λ
L L
and of course, this will blow up as x becomes large. So, we
can simply. So, this is again sorry B external. So, we can only talk about this term and
this term says that the B which is inside which exists inside the super conducting sample
falls off as exponential minus x over lambda l.
Bext
So, at x equal to lambda l B ( x)= so, which becomes equal to E minus 1. So, this
e
is called as lambda l is called as the penetration depth lambda l is of the order of about
500 angstrom in ordinary or conventional superconductors all right. So, we will come
back to this discussion when we talk about the other energy scale for the superconductors
301
namely the coherence length and distinguish between the type 1 and 2 superconductors;
meanwhile let us do the thermodynamics.
So far, we have been doing electrodynamics now let us do thermodynamics of. So,
thermodynamic route is usually the simplest route to study phase transitions and so, let
us consider Gibbs free energy of a superconductive.
So, if the magnetization is m and the magnetic induction is H then the work done in
bringing the superconductor into a region where the magnetic induction H external from
with the magnetic induction is H external; that exists from a region which is far away
where the H external is equal to 0 is given by the work done expression which I am
going to write let me write this because these are important points. So, the magnetization
is M M and magnetic induction is H the work done in bringing a superconductor from
infinity where H external equal to 0 to a region where H equal to H external exists is
H ext
H 2ext
given by W =−μ 0 ∫ M dH=μ 0 .
0
2
So, that is the work done in order to bring it from infinity to a place where H external
magnetic field or magnetic induction exists. So, we have m equal to minus H for a
superconductor. So, let us write down the Gibbs free energy ah. So, Gibbs free energy
per unit volume is given by g let us write it with a
302
H 2ext
gs (T , H ext )=g s (T , 0)+ μ0
2
2
Bext
gs (T , Bext )=g s (T , 0)+
2 μ0
because of the relationship between H so, your B external equal to mu H mu 0 H

external.
So, in the normal state the magnetization is very small thus the magnetic work done is
also vanishingly small.
Hence gn (T , H ext )=gn (T , 0) . So, in the superconducting state g s t and at a given

magnetic field or magnetic induction there is a extra term B external square by 2 mu 0 or
mu 0 H external square by two; however, in the normal state the magnetization is known
to be very small and we do not have this magnetic energy contribution as it is seen here.
So, now, this tells that equation one tells you that there is a B dependence for g s
whereas, there is no B dependence for or B or H dependence for g n. So, g n T H external
it increases quadratically with H external. So, it increases quadratically we take H
external. So, for some H external equivalently B external the normal and the
superconducting states have same energy have same Gibbs free energy.
303
So, what it says is that if you plot the Gibbs free energy. So, this is Gibbs free energy as
a function of either B external or H external does not matter they just get scaled by this
and. So, for the superconducting state it is like this and for the normal state it is like this.
So, this crossing point is called as H C or bc depending on which language you want to
use ah. So, below H C this is of superconductor SC and this is normal.
So, the sc state the superconducting state has lower energy for all values of external
fields or induction below H C or bc and as the external field crosses this H C, then the
superconducting state has higher energy and the normal state has lower energy as you
can see it is here.
So, then beyond that the normal state stabilizes. So, what one can look at is that we can
call this as equation one or rather and this as equation 2 and if we equate one and 2 at H
equal to H C.
H 2c
H external equal to HC, then gn (T ,0)−g s (T ,0)= ah. So, this is positive and
2μ0
because this is positive and because this is positive the superconducting state is more
stable than the normal state below H equal to H C. So, let us give an example for lead at
T equal to 0 H C equal to 0.08 Tesla, thus at T equal to 0 the superconducting state is
stabilized by 4.25 into 10 to the power minus 25 joule per mole and this is really a small
energy. So, it is quite amazing that such a small energy actually stabilizes a
304
superconducting state, but that is true of the order of 10 to the power minus 25 joule per
mole, let us now talk about the critical.
Fields which is what we are just introduced and the variation of the critical fields with
T 2
temperature is given by we give this without truth is H c (T )=H 0 [1−( )] where H
Tc
0 equal to H external at T equal to 0.
So, this is the externally applied magnetic induction at T equal to 0. So, that is H zero.
So, it is clear that at t equal to T C this term vanishes. So, H C of t goes to 0. So, there
are many things that are important in this particular context and we have talked about
electrodynamics and we have talked about thermodynamics and we found that some
important information is encoded in both of them, now let us talk about.
So, another interesting comment from Pippard which is encoded in Pippards local
electrodynamics or rather non-local electrodynamics and let us see what it says if you
look at this equation J⃗s=ns e v⃗s this is what we have written this looks like that
J⃗s (⃗r )→U s (⃗r )
So, it is a local equation. So, J s at r is simply determined by the v s of at that point r

what Pippard say that the current density at a point r depends on. So, J s at r it depends
on E the electric field at r prime. So, which is centered around r in a radius say l ok. So,
305
at a given point r the electric field the current density at this point will depend upon
electric field in the region which is spread all over r prime where r prime is centered
around r with a radius which is l.
So, every point inside this circle is r prime and all of that those r primes will contribute to
the current density at r. So, this is the non-locality which Pippard thought i s more
relevant and realistic and this can be you know thought of to be a spread over a region of
radius l and. So, this l is actually related to the characteristic dimension of the electron
wave function and so, this l is related or rather let us write it.
The wave function of electrons should also have similar characteristic dimension that is
extent which is called as let us call that at xi which is called as a coherence length. So,
electrons within an energy range K T C where T C is the transition temperature play a
major role in pairing. So, this is pairing of electrons.
So, the momenta of these electrons have an uncertainty of the order of d
Δ E kB T c
Δ p= = thus the position uncertainty. So, we are just using a Heisenberg’s
vF vF
ℏ ℏ vF (ℏ v F )
uncertainty relation the position uncertainty is a Δ x= = thus ξ =a
Δ p k BT c kb T c
306
Where a is a number which is of the order of 1 and a equal to 0.8 in BCS theory. So, this
introduces and others in length scale for the problem which is known as the coherence
length. So, a coherence length is the second length scale in addition to the penetration
depth that we have introduced earlier. So, Pippard suggested that. So, basically that j s at
R should be written as an integral rather than writing it as n s into E into v s.
So, this is equal to some constant which is not so, important. So, this is R;
−R
J⃗ ( ⃗
r )=const
∫ R⃗ [ R⃗ − ⃗A ( r⃗')]e ξ
d r⃗'
.
S
R4
So, this is how the non-locality in the super current density comes that the super current
density at a given position R depends on the electric field in this particular fashion that
there is a region which is centered around R prime that every point in that region
contributes to the electric field and ah characteristic length scale comes out of it and we
have introduced this earlier.
Now, we can talk about dimensionless introduce a dimensionless parameter kappa which
is equal to lambda by xi and these are. So, for typical superconductors lambda equal to
500 angstrom as we have already said psi equal to about 5000 angstrom.
307
So, kappa is typically less than 1, but; however, in 1957 Abrikosov found that for some
class of superconductors this kappa can be greater than 1 and this is used as a
distinguishing feature of type 1 and type 2 superconductors. So, to remind you in type 2
superconductors the flux lines penetrate the sample till some threshold value of the
magnetic field and if we increase it beyond a certain threshold then superconductivity
disappears.
So, a threshold value of value of kappa for which such flux penetration occurs or starts to
occur is kappa its say that value of kappa is called kappa c which is equal to 1 over root
2. So, I mean this is in terms of that so, ok. So, this says that at this value of at this value
of kappa the flux penetration starting at a lower critical field H C 1 H C 1 and reaching
an upper critical field.
So, the scanning tunneling microscope data STM data confirms penetration of magnetic
flux which are called as vortices and these vortices actually line up in the form of a
hexagonal lattice in a regular lattice in the form of an hexagonal lattice and they arrange
in hexagonal lattices and which are called as called as Abrikosov lattice.
So, we have studied mainly the electrodynamics and thermodynamics of

superconductors. So far, and have gotten quite a few information useful information
about superconductors and the next thing would be doing BCS theory which is a
microscopic theory of superconductors.
308
Prof. Saurabh Basu
Lecture – 20
BCS Theory
Let us look at BCS Theory the Bardeen cooper Schrieffer theory; the first microscopic
theory of superconductivity, which was extremely successful in describing the behavior
and the properties of the conventional superconductors, which are also known as weak
coupling superconductors.
So, I said that it was the first microscopic theory and it was formulated by Bardeen
cooper and Schrieffer and that is where the name BCS came in 1900 and 57. So, which
was approximately 50 years after the discovery of superconductivity experimentally,
which as we have seen before that it was done by Kimberling owns in 19 O 8.
This theory successfully describes the superconducting properties of weak coupling

superconductors will try to make it clear that what weak coupling the word weak
coupling means such as Aluminium and other materials mostly they are metals and not
all metals from or give rise to super conductivity, but there are certain metals which give
this also will discuss which are the potential candidates for superconductivity.
309
So, the basic idea is that the electrons lying within Debye energy of a filled Fermi sea ok.
So, that 2 electrons are needed and they are lying within the Debye frequency or Debye
energy of a filled Fermi sea in a metal they can form bound pairs.
So, this is something that has to be contrasted from the usual behavior of electrons which
would repel each other; however, if you can create a situation in which they are in a very
close proximity of a filled Fermi sea, the electrons do not interact with the Fermi sea
other than by exclusion principle. And then these 2 electrons actually can have an
attractive interaction between them and can form bound pairs ah. And it is important to
say that not all electrons participate in pairing and only the electrons which are in the
vicinity of the Fermi surfaces do not in the say the near the center of the Fermi surface
they do not take part in pairing, but it is only which lie in the vicinity of the Fermi
surface or close to the Fermi energy they do.
This paired state requires a many body description and this is what is the central focus of
our discussion at least for this one and a discussion that is going to be held after this, now
this hint that the Debye energy scale is important for this problem it actually signals that
there is an involvement of electron phonon interaction and this was proposed by Frohlich
as early as 1900 and 50. So, before this theory came up.
Now, this electron phonon interaction describes isotope effect this give me a minute to
say what that is. So, the dependence on the sorry this should be phonon, on the phonon
310
parameters were experimentally demonstrated by the fact that the transition temperature
we have seen this concept of transition temperature. A temperature that divides
superconducting state with that of a non-superconducting one or rather a metallic state
the temperature at which this transition occurs is called as a superconducting transition
temperature.
So, this transition temperature T C becomes a function of the ionic mass of our different
isotopes of the same metal. Now that tells that if that happens then the lattice, because
the ions are involved and hence the lattice is involved and then the phonons are involved
as well. And this is an experimental demonstration of an electron phonon interaction or
the rather the phonons are playing an important role in the whole process.
So, if you want to write this goes as a delta T C over T C this is equal to a minus half
delta M over m where M is an ionic mass and delta M is the difference in ionic masses of
2 isotopes, or very simply this is written as a T C to be a function of you know 1 by
square root of m where M being the ionic mass.
So, these are experimental demonstration that the phonons are involved and which is also
central to the discussion that we are going to have.
So, coming to the success of the BCS theory a priori of course, we have not talked about
BCS theory yet, but it is important to know that why should we be doing it? In a course
311
like this there are many successes. In fact, it is it gives the excitation energy gap the
energy gap. So, called in a superconductor there is certain amount of energy that is
required to be supplied to the superconductor in order to break the or rather the break the
cooper pairs and promote them to the excited state.
So, this is done experimentally done by the electron tunneling which was done by giver
in 1900 and 60 ah. So, when an electron is sent across a metal in metal superconductor
junction the superconducting energy gap to delta, it should not have been able to pass
through because there is a additional barrier that is there which the electron of which the
metallic side does not have.
But, however, there is a process called as a Andreev reflection, which we are not going
to discuss here and because of that Andreev reflection there is a pair if not the electron
there is a pair that can propagate into the superconducting region and this shows up in a
peak, which is called as a un reef peak and this happens just below the energy of the
excitation energy of the of the gap of a superconductor. And hence it is it gives these
electron tunneling measurements they yield the magnitude of the gap and there are other
experiments, which actually probe the symmetry of the gap the symmetry of the gap in
the momentum space.
So, to say it explains Meissner effect we have seen extensively what Meissner effect is it
explains the specific heat jump, it explains isotope effect as we have seen and the
important books in this context are parks R D parks he had 2 volumes of super
conductivity, which were published in 1969. And then there is a Rickayzen which there
is a book by Rickayzen which was in 1965 and then this book by Schrieffer in 1964 one
of the proposers of BCS theory.
312
So, what exactly is happening how are the electrons interacting with themselves why an
attractive interaction, why an attractive potential such that they form bound pairs? And
what do these bound pairs have in common with the superconducting state? So, it can be
understood as the instability of the Fermi gas.
So, what happens is that the ground state of the free Fermi gas as we said that this
corresponds to the filled Fermi sphere corresponds to all completely filled states up to
the Fermi level and all states beyond the Fermi energy they are all empty this is the
definition of the Fermi distribution function, but this Fermi gas is going to get unstable
against the formation of at least one bound pair irrespective of regardless of how weak
the interaction is so, long it is attractive and this is was known as coopers instability.
So, to rephrase this we have a field Fermi sea or a Fermi sphere if by some means, what
means we are going to discuss just in a while if by some means, that 2 electrons form
bound pair in the vicinity of that Fermi sphere lying within the energy scale which is
given by the phonon frequency or phonon energy.
Then this Fermi gas the free Fermi gas is going to be unstable and this is called as the
cooper’s instability. To see how such attractive interactions come about now considered
2 electrons added to a filled Fermi sea, the electrons interact with each other, but not
with the Fermi sea except via the exclusion principle ah, one sort of hand waving way of
seeing it is that suppose you have a horse moving through a dusty field.
313
And when it runs through the dusty field it gathers a lot of dust around it and anybody
say another horse would not see that horse, but would see a cloud of dust that is moving
forward. And pretty much something similar to that is happening that an electron passes
through the lattice and because of these the ions being present this picks up in a positive
charge that it looks or another electron it looks like a positive charge and they form
bound pairs.
Now, it could also be argued that why does not the this other electron also is perceived as
a cloud of positive charge, which when they will repel and we will see that this happens
at different time scales and that is why there is not enough time that both of them would
be like an ionic cloud moving together. And they would repel rather it happens at a
shorter life time scale that the electron one electron sees the other ionic cloud and gets
attracted to it and forms a bound pair, but this is just a hand waving way of seeing
looking at it we will see this more carefully.
So, as has been told that the first hunch or the inkling of the BCS theory what was put
forward by cooper, in 1956 just a year ahead of the discovery of BCS theory.
So, he pointed out that a normal metal will become unstable to the formation of pairs if
the I mean no matter how small the attractive interaction between the pair is and so, he
proposed that such pairing would be preceded or would be facilitated by the presence of
the lattice and hence phonons.
314
So, the electrons would scattered in a particular way and the electron interaction in
V ⃗q 2 ω D M 2q
presence of phonons is given by V s (⃗q , ω )= + I will
ϵ (⃗q , ω ) ϵ (⃗q , ω )[ ω2− ω2D (⃗q )]
explain what all these quantities are. So, the first term is the screened coulomb potential
and it is always positive and hence repulsive.
So, in no way that this term can give rise to an attractive interaction the first term. Now
the second term is generally weaker than the first term and caused by electron phonon
interaction the strength of which is given by M q, now this term can become negative if
omega square is less than omega D square. So, in general this, the frequencies lie around
the Debye frequency as has been told earlier.
So, in a small range that in the vicinity of omega D if omega becomes less than omega
D the energy denominator becomes small and negative. So, the whole interaction term or
this whole second term becomes negative and large and this can cause superconductivity.
So, it will happen in a thus I mean V s (⃗q , ω ) now when the second term actually takes
over the first term, the second term being negative then the whole interaction can become
negative in the vicinity of the of omega D. So, it is possible that the 2 electrons to bind if
they construct a relative wave function which selectively uses a frequency range that is
given by. So, if you plot this then it looks like that it is so.
315
So, this is a V S q omega and this is omega and it is in this range. So, it is in this range
that it is negative and one can have an attractive interaction. So, once again to repeat it is
that this is possible for the two electrons to bind, if they can construct a wave function
which selectively chooses the frequency the frequency range that is attractive. In fact,
Alkali metals like sodium potassium etc alkali metals have very weak electron phonon
interaction and hence do not give rise to superconductivity ok.
So, now so, this the interaction has to be in this regime that is shown here and it is
usually is as a competition of two terms; one term being necessarily positive and
repulsive which is the screen coulomb term and the other term which arises because of
the electron phonon interaction the strength of that being M q, but because of a
denominator which is the competition between the energy of the pair versus the energy
of the Debye energy.
So, and then since omega is close to omega D omega is measured from the field Fermi
sea omega is close to omega D. If it can happen that omega is less than omega d then this
term the denominator becomes small and negative and then we can have an attractive
interaction and that attractive interaction will actually give rise to super conductivity, but
we have not still derived or at least explain that how this interaction term comes from
and that can be understood in 2 different ways; one is that rigorously taking an electron
photon matrix element I draw all the diagrams and calculate this term the second term
particularly. There is another way of doing that which is what we are going to do. So, let
us do the origin of ok.
316
Ah. So, we have a scattering between K ( ⃗

k ,↑),(−⃗
k ,↓) electron ah. So, one could ask
the question that why is it that one has one electron has a momentum K the other electron
has the momentum minus K ah. It can be shown that with the momentum being chosen
in this particular fashion a 2 particle state yields the minimum energy which is the most
stable configuration.
And for a bound pair to happen because they are in the same state we cannot expect a
both up pair to occur or an up down pair to occur. So, the pairing has to be between up
and down. So, these are the momentum and spin of the participating electrons which are
participating in the bound pair. So, let us see that how these pairing takes place just to
say that it is a spin independent interaction and so, spin index can be dropped ok. So, let
us say the case one the case one is that that there is a direct electron interaction which is
facilitated by the coulomb forces.
So, let us consider an initial state. So, this case one is coulomb interaction and it is a
direct process what do I mean by direct process is that there is an initial state i before
scattering and there is a final state f after scattering. So, the matrix element of such a
electron interaction in such a direct process will just write e and dir let us just give a little
of bit of space here and this is written as
⃗ ⃗
( H dir dir
e−e )ij =<i| H e−e | f > =∫ e
i k .⃗r
U c (⃗r )e−i k ' .⃗r d 3 ⃗r
317
So, this is the matrix element and this matrix elements is the incoming state momentum.
So, the electron comes with a momentum K and leaves with a momentum minus I mean
leaves with a momentum K prime this minus sign is coming, because you are taking a
ket here whereas, this is a bra here. So, this is exponential I K dot r.
So, we have assumed that both i and f are plane wave states i and f are plane wave states.
So, this term cannot give any negative contribution and hence it cannot account for the 2
electrons to form a bound pair. So, this is positive and hence it is repulsive. So, we are
not yet there for getting the attractive interaction.
Now, let us look at a case 2 which is an indirect process it takes place via exchange of
phonons and this there is an intermediate process involved and it is a that is why it is a
second order process remember your second order perturbation theory, where the way it
is written is that there is an intermediate there is an initial state, and then there is a
perturbation, and then there is a final state I am sorry there's an intermediate state here,
and then there is another matrix element and the final state and then here one gets the
energy so, I minus the intermediate energy square.
<i | H ' | int> <int | H ' | f >

So, this is the correction let us call it as E(2 )= So, there are 2
( Ei −Eint )2
ways that this indirect thing can take place; one is number one in which the electron one
let us just label them even though they are indistinguishable this is for our convenience.
318
So, with momentum K emits a photon which is later reabsorbed by electron 2 with
momentum minus K. Let us just say this process once more. So, there is an electron
called electron one which is a momentum K it emits a photon, which later at a later time
the electron 2 absorbs this which had a momentum minus K and this process would be
looked upon as this.
So, this is the K which is electron number one and this is an electron minus K this is
electron 2 this one m it is K m it is a photon of momentum q. So, this becomes K plus q
and this becomes minus K minus q and because 2. So, this is time axis. So, this is how
time increases and we are particularly you know talking about that. So, this is the this is
how the time grows.
So, later on when after it emits it gets reabsorbed. Now the same thing can be thought of
for the reverse case that is to is that electron 2 with momentum K, minus K emits a
photon which is later reabsorbed by electron 1 with momentum K.
So, this is a similar thing it is just that the alignment of the figure will be a little different.
So, they will be K here and it goes like this and now because the process happens in the
reverse order. So, we will draw it this way. So, this is electron 1 with K again it is K plus
q here and it is minus K and this is minus K minus q.
319
And we are talking about. So, this is the again the time. So, these are the 2 possibilities
of the indirect process that one emits the photon and the other absorbs it and there is a
time delay between that and in principle both can occur and both will have to be
considered. Now you see that in both cases the initial and final states are same.
So, let us do this that process 1 process one had initial is ϵi =2 ξk which is e
ℏ2 k 2
ξ k= −μ and final is ϵ f =2 ξk ' and k⃗' =⃗
k+q .So, that is the that is for the
2m
process 1 and similarly for the process 2 it is initial is ϵi =2 ξk and ϵ f =2 ξk ' well I
will you know I mean this k’s are vector, but we are ignoring it at the moment. And so,
even if the initial and the final states are say I mean the energies are same ah, but the
inter I mean the initial and the final states are same, but the intermediate states are
different and let us say the intermediate step.
So, for process one the intermediate
ϵIint =ξk ' + ξ k + ℏ ω q

ϵIIint = ξk + ξ k' + ℏ ωq
. So, energies are still same of the intermediate state now calculate the matrix elements
including these intermediate states.
320
So, H e-e in direct using the same notation as earlier between i and f you have to sum
over the intermediate states and it is i H e indirect, I will just write it intermediate and
intermediate H e e indirect indirect
<i | H indirect | int > < int | H indirect |f >

(H indirect e−e e−e
e−e )i , j= ∑ 2
int ( E i ,f −E int )
1 1 1
=∑ <i | H indirect
e−e | int > [ + ]< int | H indirect
e−e |f >
int 2 E f − E int E f −E int
So, these are simply numbers because these are matrix elements. So, this is summing
over over both 1 and 2. So, this is written here. It is equal to
1 1 q ) |2
2 ωq | V c ( ⃗
−
ω− ω q ω + ωq = which is so, in in a essence this exactly looks like
ω 2 − ω2 q
the second term that we have gotten with an energy denominator, which is here given by
this.
And if you want how this comes then you can look at this thing that E
1 1
( E f −E int )−1= −1
and ( Ei−E int ) =
2 ξk ' − ξk − ξk ' −ℏ ωq 2 ξ k −ξ k' − ξk −ℏ ω q
So, this matrix elements are positive definite this one; however, because of this energy
denominator and in that narrow range that, we have talked about this could be attractive
321
and hints would give rise to the attractive interaction. Now let us look at the a 2 particle
wave function BCS gap equation ok.
So, block actually argued lowest energy state for a 2 particle system corresponds to 0
total momentum. So, the 2 electrons are equal and opposite momentum 2 electrons which
are taking part in binding. So, we can write down the orbital wave function it is only the
⃗ ⃗
i k .⃗r −i k . r⃗
orbital part that we are writing it down to have r ψ0 ( r⃗1 , r⃗2 )=∑ g k e e 2
where g K
is some weight or the amplitude of the wave function.
Now, when we write the full wave function the orbital part would be symmetric or anti-
symmetric depending on the spin part beings anti-symmetric or symmetric, because the
full wave function consisting of the orbital part and the spin part will have to be anti-
symmetric. So, if we choose symmetric for this then the spin part would be anti-
symmetric and if we choose anti-symmetric part for this then the spin part of the
symmetric, but remember that we have done this earlier for 2 spins we can have 4 states
one of them is called as a singlet state which is an anti-symmetric.
And the other 3 are called triplet states which are symmetric which means that if you
interchange the particles then the wave function does not change. So, because the spin
part for a singlet wave function is anti-symmetric. So, the orbital part will choose will be
chosen as symmetric.
322
Now, this is also shown that it is generally true for a 2 particle system that the singlet one
has a lower energy. So, we take a singlet choose a singlet wave function and BCS theory
corresponds to singlet pairing. So, we have the singlet wave function as up down or
down up and this is anti-symmetric and because this is anti-symmetric the orbital part
has to be symmetric.
So, we take this orbital part as and also we will have to take that K less than I mean K
greater than K because we are writing down the 2 particle wave function which is
residing just outside the Fermi surface. So
ψ0 ( r⃗1 , r⃗2 )=∑ g k cos [ ⃗k .( r⃗1−⃗
r 2)][| ↑↓>−|↓↑> ]
Now this can be plugged into the call this as equation 1 and this is plugged into the
Schrodinger equation.
So, this is ( E−2 ϵ k )g⃗k = ∑ V ⃗k k'⃗ g k⃗' . So, all these are K’s are of course, vectors. So, V
k ,> k f
K K prime are the matrix elements of the interaction potential. And the
1 ⃗ ⃗
V k k '= ∫ r )ei ( k−k ' ).⃗r d 3 ⃗r , but just to keep our discussion simple we will take this
V (⃗
vol
as coulomb term or rather you can consider this V r to be arising from that kind of a an
interaction which is mediated by phonons.
323
So, r is the r is the distance between two electrons with momentum K and distance
between two electrons and the V K K prime of course, it ah. So, V K K prime describes
scattering of K prime minus K prime to K and minus K.
So, we are to solve for this is equation number 2 solve for g K such that such that total
energy E is less than 2 E F is basically that is the condition for a bound pair to exist. It is
very hard to do this calculation for general potential. So, what you can do is that a
V k k ' =−V for ϵ F≤ ϵk k ' ≤ ϵf + ℏ ω D
So, this is as we have said earlier that it will be overall negative because mediated by
phonons of course, that has to be satisfied by the participating electrons, they have to
choose the energy range where it becomes negative, but suppose it becomes negative.
And the K and K prime for both the electrons would have to lie between ϵ f + ℏ ωD it is
equal to 0 otherwise.
gk '
Now, with this we have gk =V ∑ . Now sum over K so, it becomes one over
k' 2 ϵk −E
V this V is the strength of the interaction and we just reiterate that this V could be
infinitesimally small a super conductivity will still occur.
324
1 1
So, =∑ now we can replace the summation by integration and we had to
V k>k 2 ϵ k −E
F
just invoke the density of states, but as we have told several times that the super
conductivity is a phenomena that happens very near to the Fermi surface. So, while we
convert the sum into an integral we do not need to take the energy dependence of the
density of states rather we can take the density of states at the Fermi level.
ϵF +ℏ ω D
1 dϵ 1 2 ϵ F −ϵ + ℏ ω D
So, this is equal to =N ( ϵ F ) ∫ϵ = N ( ϵ F )ln( )
V F
2 ϵ k −E 2 2 ϵ F−E
So, that is the equation that one gets and now for conventional superconductors N
epsilon f into V is usually of the order of 0.3 or less than 0.3. So, this is what is meant by
the weak coupling approximation which demands that N epsilon f V has to be less than 1
or may be much lesser than 1. So, this is weak coupling superconductors.
2 2 ϵ − ϵ +2 ℏ ω D
So, then we have =ln ( F ) And now if I want to free this log
N ( ϵ F )V 2 ϵ F −E
2
2 ϵ F −ϵ +2 ℏ ω D N ( ϵ )V
we have a ( )=e . And if you solve for E it becomes equal to. So,
F
2 ϵ F−E
I can I can multiply this here and just doing one more step I would land up with this
−2
N ( ϵ F )V
E=2 ϵ F −2 ℏ ω D e
325
So, this is the energy of total energy of the 2 particle system. So, this is with respect to if
we take it with respect to the Fermi surface of the 2 electrons this is comes with a
negative sign. And this negative sign talks about that the, they have a bound their bound
they are in bound state. So, a bound state can form off infinitesimally this is not
committed to the magnitude of V any magnitude of V would do.
So, by an infinitesimal attractive interaction also you see that the binding energy is non-
analytic it comes as an exponential and hence no orders of perturbation theory, can bring
this result. Our perturbation theory in orders or in terms of V the strength of the attractive
interaction between the electrons just one more thing that we need to do here is that ah.
So, for the wave function we have.
cos (⃗k . ⃗r )
The amplitude of the wave function which is gk = ∑ where
k>k F 2 ξk + E '
ξ k =ϵ k −μ =ϵ k −ϵ F , T equal to 0 and E '=2 ϵ F −E> 0
Ah. So, E prime can now be called as a binding energy and this K has to be greater than
1
K F, just a few comments this weighting factor which is given by has a
2 ξK + E '
1
maximum at for xi K equal to 0, which are valid or which are applicable for
E'
electrons at the Fermi surface and it falls off at positive values of. So, if you note that the
326
E prime the binding energy is much smaller than the h cross omega D for N epsilon F to
be into V to be less than 1 ah.
So, these make sure that the detailed behavior of V K K prime is not important. Also the
small energy range allows the estimation of the range of the cooper pairs, which we
know is called as the coherence length and that coherence length is given by, this is what
has been told earlier and this is of the order of one in BCS theory it is of the order of 0.8
and that gives the good estimation of the coherence length of the cooper pair wave
function and that matches with the experimental values which are of the order of maybe
3 to 5000 angstrom.
So, this is a theory which gave the calculation had given rise to a gap the main features
of the gap is that it is valid for any strength of the electron electron interaction. So, long
is attractive it is a non perturbative the result is non perturbative and hence you cannot do
a perturbation theory of any order in order to get this result. And also it allows the
estimation of the extent of the cooper pair wave function which is called as the coherence
a ℏ vF
length, ξ o= .
kB T c
327
Prof. Saurabh Basu
Lecture – 21
BCS theory, Transition temperature
So, having laid the foundation of BCS theory that is how cooper established that there
could be an electron attractive interaction mediated by phonons.
Let us go over to start BCS theory and before we start, let me tell that it involves some
mathematics that is algebra which you should do because every step cannot be shown in
the in the class. So, go through the steps yourself before you convinced that the results
that we are coating are correct. So, postulate many body ground state as this is a sum
over. So, | ψ0 > = ∑ g k C k+↑ C−k

+
↓ | 0> .
k>k F
Here 0 is not vacuum, but it is the filled Fermi sea the g K is some amplitude of the wave
function and CK up dagger C minus K down daggers at 2 a pair is the pair created with
momentum K and minus K with an up and down spins and this is a many minus body
wave function superconducting many minus body wave function and then we have say
328
M electrons and we want to choose N of them to make N by 2 pairs and the number of
M!
ways, it can be done is
(M −N /2)! ( N /2)!
For m equal to 1023 this combination is equal to (1020)20 . So, in principle we have
to solve for this many g K's in order to be able to write down a proper many body state
and which is an impossible task. What can be done? Is that one can treat the problem
statistically and in order to do that one can also take a grand canonical ensemble such
that we do not keep the particle number of fixed and instead talk about an average
number of particles.
So, we shall talk in grand canonical ensemble and talk about average noise, we call it as
N bar instead of the number of particles.
So, BCS ground state BCS rolled down the ground state as a
| ψG > =∏k⃗ , k⃗ ,........ k⃗ ( u⃗k + v ⃗k C k ↑ C−k ↓)| ϕ0 > . So, this is the filled Fermi sea as we have
+ +
1 2 N
been talking about so, the probability of a pair to exist.
So, a pair formed off k ↓)→| v k |2

( k⃗↑,−⃗ and that this is unoccupied is given by
k ↓)→| u k |2
( k⃗↑,−⃗ and of course, the normalization says that the | uk | 2+| v k |2 =1
329
which means that the probability that a K up and minus K down would be either
occupied or unoccupied and the total probability is equal to 1.
take 2 states K one and K 2. So, the amplitude with uk 1 u k 2 represent no pairs in these
2 states uk 1 v k 2 one pair K 2 and - K 2 and v k 1 v k 2 clearly distinguish your use and
v’s there are 2 pairs for K 1 - K 1 and K 2 - K 2 both are occupied. So, these are the
notations. So, a pair would be unoccupied with a probability u K mod square and a pair
would be occupied would be given by v K square mod v K square.
So, average number of particles N̄=< ∑ n^k σ > which is an operator and this is equal
k ,σ
to we can write this N̄=< ψG | ∑ Ck+↑ C k ↑+C +k ↓ C k↓ | ψG >

k
Ah so, C K down dagger C K down and this expectation has to be taken between the
ground state that we have written here. So, that is the ground state postulate of the BCS
ground state. So, we have to take this thing here and this can be written as because there
is no preference over one spin on another.
So, we can simply write this as N̄=2 < ψG | ∑ C +k ↑ C k↑ | ψG > this I am writing it once,
k
but then later on I will skip. So, this will be a phi 0 u K star plus v K star C minus K
down C K up that is that is the psi g here the brass psi g and now I have a C K up dagger
330
C K up and now I have a u K plus a v K C K up dagger C minus K down dagger and
now I will have terms which are.
So, this is the same case as it is there in the operator n K sigma and now, I will also have
all the other terms in which
N̄=2 ∑ < ϕ 0 |(u∗k + v ∗k C−k ↓ C k↑ )C +k↑ Ck ↑ (uk + v k C k ↑ C−k↓

+
) ∏ k≠l (u∗l + v ∗l C−l↓ C l ↑)(ul + v l C +l↑ C−l↓
+
)| ϕ 0 >
k
. So, this corresponds to K naught equal to l and the top 1 is for K equal to l. So, in
principle u K and v K are complex quantities.
So, let us write that u K and v K are generally complex that is why the stars are written
separately and. So, what we have done here is that we have used a
< A ϕ | ψ >=< ϕ | A + | ψ>
Now, look at the term l not equal to K that is a the term that is written later now you can
understand that this will have a
| ul | 2+u∗l v l C l+↑ C−l↓ +ul v ∗l C−l ↓ C−l↑ +| v l |2 C −l ↓ C−l ↑ C +l↑ C −l now you can understand
+ +
↓
that.
So, this is coming from the product of these terms and so, they are four terms which are
here now you can see that this term gives you 0 because it will create a pair and so, we
331
will change the occupancy of pairs in the ground state. So, it will have 0 expectation
value. So, this term is not equal to 0, but then when you take it between this then that is
equal to 0 this is what I mean and similarly, this term will also yield equal to 0 when
taken between the filled Fermi sea now this simply adds a normalization that is it creates
a pair and then it annihilates a pair.
So, ultimately what happens is that. So, l not equal to K gives you | ul | 2+| v l |2=1 and
now let us look at the l equal to K this if you look at it carefully you will have a term
which is a | uk | 2 C +k↑ Ck ↑ | ϕ 0 > =0 and u K v K because this tells you that there are no
states to annihilate for K greater than K f and u k v k the cross minus terms both
u∗k v k and uk v ∗k r will give 0 for the same reason as its told above.
So, the only term that contributes is a term with | v k |2 . So, this N̄=2 ∑ | v k |2 . So,
k
that is a result. So, 2 come because of the pair and v K square is the probability of
occupied states.
Now, second thing is a fluctuation in N bar. So, that is given by

<( N − N̄ )2> =< N 2−2 N N̄ + N̄ 2> = N̄ 2− N̄ 2 . So, this if you again repeat the same
calculation and use the same logic to cancel out terms this is given by 4 ∑ u2k v 2k and
k
this is positive definite.
332
In fact, v K as a function of K goes from one to 0 whereas, u K as the function of K goes
from 0 to one. So, in a all happening in an energy range which is given by K t c. So, if
you write down the variation of this. So, this is my. So, this is my ϵK or ψK which
is equal to ϵ k −μ and the v K drops from 0 to 1 and this is my superconducting gap
delta this is minus delta 2 plus delta and this is my.
So, this thing is my −ℏ ω D to this as ℏ ω D . So, in this range v K becomes from one
Tc
to 0 and so, the sum above goes as ( ) N̄ . So, this is the practically if you want to
TF
estimate the fluctuation in N. So, this goes as that. So, let us write down a many a BCS
many minus body ground state.
Oh, rather, let us write down a Hamiltonian and the Hamiltonian
H=∑ ξk σ C +k σ C k σ + ∑ V k l C +k↑ C−k↓

+
C−l ↓ C l ↑ that is the Hamiltonian and the of
k,σ kl
course, we know that ξ k = ϵ k −μ we have already argued that BCS theory cannot be
obtained by doing a perturbation theory of any order.
So, we will do a variational calculation instead. So, and the variational calculation with
these small u K v k’s which are the occupancies will be used as a variational parameter.
333
So, what we have to do is that we have to take this variation and it is a
δ < ψ̄G | ∑ ξk σ ηk σ +∑ V k l C k+↑ C−k

+
↓ C −l↓ Cl ↑ | ψ̄G >=0
kσ l ,l
So, let us look at the kinetic energy term or the single particle the first term here. So, this
is the kinetic energy and this is the potential energy and this is given by let us call this as
2
kinetic energy operator which is a < K^ >=< ψG | ∑ ξ k σ C +k σ C k σ | ψ¯G >=2 ∑ | v k | ξ k and
k ,σ k
similarly for the potential energy.
We have < V^ >=< ψG | ∑ V k l C +k ↑ C−k↓

+
C−l↓ Cl ↑ | ψ̄G >=∑ V kl u∗l v∗l ul v k . So, these are
kl k ,l
the kinetic energy and the potential energy these are the expectation values of those with
of course, the constraint as | uk | 2+| v k |2 =1 now because of this constraint one can
actually take a pair.
And reduce so, a pair of variables like this and reduce the number of variables from 2 to
one see it is always very difficult to do a variational calculation with 2 variational
parameters then you have to look for the minimum in a space in a 2 minus dimensional
space it is much easier to look for a minimum in a 1, on a 1 minus dimensional line and
that is now given by this single variable theta K.
334
So, we can take (uk , v k )=(sin θk , cos θk ) you can take the other combination that is u
K equal to cos theta K and v K equal to sin theta k, but it seems that this combination
works better. So, now, what we do is that we take a variation with respect to theta K of
this. Now I will use a dummy variable K prime
δ [ ξ (1+cos (2 θ ))+ 1 [ V sin(2 θ ' )sin(2 θ )]]=0 in order to do a

δθ ∑k'
k k ∑
4 k ' ,l k ' ,l k l
variational calculation and see that what is the extremum value of theta K which;
minimizes the energy.
So, −2 ξk sin(2 θk )+ ∑ V k , l cos (2 θk )sin (2 θ l)=0 if we define

l
−1
Δl = ∑ Δ sin(2 θl ) , then using this and putting it into this equation one gets a
2 l kl
nice equation such as −2 ξk sin(2 θk )=∑ V k ,l cos(2 θ k ) sin(2 θl )=−2 Δ k cos (2 θk ) now
l
−Δk sin (2 θ k ) −Δk

this equation gives us tan (2 θ k )= → =
ξk cos (2 θk ) ξk
335
Δk
Now, if we use the definitions that 2u k v k =sin (2 θ K )= 2 2
. So, this is my sin 2
√ξ +Δ
k K
−xi k
theta K definition of sin 2 theta K which is also equal v 2k −u2k =cos(2 θk )= 2 2
,
√ξ +Δ
k K
there is a little bit of understanding that needs to be done here is that this particular
choice, we could have taken the other choice also that is a vice versa, but this choice fits
all the definition and we have taken the cos 2 theta K to be negative because if xi K is
large which means that epsilon K is much much greater than the chemical potential mu
then v K should go off to 0 which is a apparent from this diagram.
So, that is why the cosine 2 theta K is taken with a negative sign alternately we could
have taken the sin 2 theta k, but that would not have satisfied the conditions or the
−1
boundary conditions that we have. So, hence the quantity Δ k= ∑ V sin(2 θk )
2 l kl
−1 Δ
Δ k= ∑ Vkl 2 l 2
2 l √ ξ l + Δl
So, that is the definition of delta K so, a trivial solution. So, we have to solve for these in
order to solve for delta K a priori let us say that the delta K is really the energy gap or the
superconducting energy gap and we have to solve for it in order to find that what is the
or how does the gap vary with different parameters especially say v and xi l or how does
that enter into the expression of the gap.
In order to see that we can also look at this expression and see that the trivial solution is
delta equal to rather this is delta equal to 0 is the trivial solution. Now you see delta is
there on both the sides of the equation here it is just a standalone delta K and here it is
sum over l delta l and K is not equal to l. So, for a given momentum value l delta l is the
value and delta K has to be computed by summing over all those delta l, but since we are
solving for delta l we do not know what it is. So, we cannot. So, the unknown quantity
appears on both the sides.
336
So, a trivial solution as we said Δ=0 so, what does Δ=0 gives it gives that
2u k v k =0 because you see that 2 u K v K equal to delta k. So, if delta K equal to 0 2
K u K v K equal to 0 from this equation.
So, 2 u K v K equal to 0 and that tells that we also have v 2k −u2K =−1 but then we have
the normalization condition is that | uk | 2+| v k |2 =1 that it tells that . v 2k =0 So, at
Δ=0 implies that v 2k =0 . So, there are no pairs and hence this should correspond
to the normal state.
So, the trivial solution is important because it talks about the normal state, but at the
same time it gives a meaning to delta now delta can be used as an order parameter for the
superconducting transition because at normal state delta is equal to 0 and delta is not
equal to 0 for the superconducting state. So, now, go to the original Cooper’s
proposition.
1
That V kl =−V if | ξk −ξl |<ℏ ωD thus Δ k = V ∑ Δl sin(2 θl ) for
2 l
| ξk −ξl |<ℏ ωD equal to 0 otherwise. So, this tells that as if delta does not depend upon
K it simply depends upon it just a number. So, that is even more convenient because
delta is now can we a truly thought as a number which when is nonzero will give a
superconducting state; however, when it is 0 it will give rise to a normal state. So, then
337
we have delta l which has to be put here. So, this says that if it is a sum over all l delta
loses the K dependence.
So, it is just a just a number and has no K dependence. So, we can write this as
1 1
Δ= V Δ and we can cancel delta from both sides and this gives rise to an
2 √ Δ 2+ ξ2l
V 1
equation which is 1= .
2 √ Δ2 +ξ2l
So, there is no delta l. So, it is equal to xi l square now this is the equation for delta you
may not see delta on the left hand side to solve for, but there is delta in the right hand
side and in the denominator and in the square root of a denominator that tells that it is a
highly no linear equation and you have to solve it either by a root finding method or one
of the root finding methods such as Newton Raphson or bisection method if you want to
do it using a computer that is numerically.
We can also solve this problem analytically by converting this sum over l to an integral
which is of this form remember the V is the strength of the attractive potential. So, it is
the once when we convert a summation into an integral we need to bring the density of
states or we can write down the density of states like this and do this integral such as.
338
So, there is a 2 plus delta square plus psi square by psi is the variable. Now, the integral
will be from 0 to h cross omega d this is what we have said earlier and Cooper had
explained that how the pairs have to be formed within an energy shell of h cross omega d
from the Fermi surface. So, it is measured from the Fermi surface and the now
ℏ ωD ℏ ωD
1 dξ N ( ξF ) dξ
= ∫ N (ξ ) = ∫
V 0
2 2
2 √ Δ + ξl 2 2 2
0 2 √ Δ + ξl
ℏ ωD
2 dξ −1 ξ ℏ ω
And that tells us that this is equal to =∫ =2sinh ( Δ ) |0 D
if we put
N ( ϵ F )V 0
2
√Δ +ξ 2
l
ℏω
these values and rearrange then we will get Δ= .
1
sinh( )
N (ϵF ) V
Now, since we are talking about for week coupling super conductors N ( ϵ F )≪1 thus
−1
N (ϵ F )V
Δ=2 ℏ ωd e .
So, we are getting a similar expression for the energy gap as we have done by solving the
2 particle Schrodinger equation this tells that this much of energy has to be supplied in
order to break a pair and go from a superconductor to a normal state and in BCS theory
339
this energy gap is a scale which is given by the temperature scale. So, Δ ~ K BT c and.
−1
N ( ϵF )V
So, a k B T c =2 ℏ ω d e
It is actually 2 here as we have written there and very accurate calculation shows that this
is equal to 1.14 and so, the TC expression is obtained from here we will just do it in a
minute. So, this gives the how the energy gap depends upon the phonon energy spectrum
and how the density of states at the Fermi level come into the picture and the strength of
the attractive interaction is also there which is v and we also have an epsilon f multiplied
by v is much smaller than one which is relevant for a weak coupling superconductor.
So, let us give you all these occupation probabilities or what are also called as coherence
2 1 ξk 1 ξk
factors that we have a | v k | = (1− )= [1− 2 2 ] . So, that is
2 Ek 2 √Δ + ξK
1 ξ
| uk | 2=1−| v k |2= (1+ k ) . So, these are important because these decide the behavior
2 Ek
of the gap the K dependence of the gap now let us go to the finite temperature.
So, we have seen that lets. So, let us write down the BCS Hamiltonian once more which
is a mean field BCS Hamiltonian. So, this is equal to
340
H=∑ ξk Ck+σ C k σ−∑ Δ k C +k↑ C−k↓
+
+Δ∗k C− k↓ C k ↑ this would be obtained if you take
k
the mean field decoupling of the PE term of the potential energy term.
These kind of d couplings we have done earlier and this equation or rather, this
Hamiltonian can be diagonalized using a Bogoliubov Valatin transformation where the C
operators are transformed into quasi particle operators of this form C k↑=u∗k γ k +v k γ k
0 1
and the C−k↓

+ ∗
=−v k γ k +u k γ k
0 1
and. So, gammas are quasi particle operators.
So, they have. So, gamma dagger they have usual anti commutation relations as the C K
daggers now a generic form of the gap.
Now, we call delta as the gap because we have established that delta is nonzero for the
superconducting state and 0 for the normal state. So,
Δ k =−∑ V kl <C−l ↓ C l↑ >=−∑ V kl u∗l v l <1− γ +l γl − γl+ γ l >

0 0 1 1
l l
and this at finite temperature is given by each one of them will be given by a Fermi
distribution function which is or this is E l and the E l.
341
Thus <1− γ +l γl −γ +l γ l > =1−2 f ( El )
0 0 1 1
So, to say. S Δ k =−∑ V kl u∗l v l (1−2 f ( E l))
l
β El
and this is nothing, but tanh ( ) .
2
So, with v K prime equal to minus v which is coopers assumption I get this Equation as
Ek
tanh( β )
1 1 2
= ∑ again converting that sum into the integral and using the
V 2 k Ek
density of states to have a value that is at the Fermi level.
ωc
βc ℏ
2
1 tanh( x) β Ek
So, this is equal to a = ∫ dx where x= this integral as a
N ( ϵ F )V 0 x 2
2 eγ
standard value which is given by ln ( π βc ℏ ω C ) where beta C is equal to one over K
T C and gamma is the Euler constant which has a value 0.577 and hence this or 14 a
2 eγ
π =1.13 round 14 and that if you simplify this it comes out as
−1
N ( ϵF )V
k B T c =1.14 ℏ ωc e , this is exactly the formula for TC that we have got.
342
So, this is the how TC varies with the phonon frequency and the density of states at the
Fermi level and v so, thus delta 0.
Δ(0) 2
So, that is a value of the gap at = ~ 1.76 nearly and that tells you that this is
k B T c 1.14
2Δ
a feature of BCS super conductivity that =3.52 this is followed by all weak
kB T c
coupling phonon mediated superconductors ok. So, this is the a feature or rather a
property of this that the twice of the energy gap verse divided by TC or K TC should be a
number which is three point five 2 it deviates strongly for the high temperature
superconductors.
343
So, let us summarize what we have seen so far we have generically talked about the
properties of superconductor and those properties include the Meissner effect that is the
electromagnetic response then the thermodynamic response and the conducting
properties or rather the resistivity how it suddenly drops to 0 below certain critical
temperature or threshold temperature and so on.
And then we have gone on to talk about what is the origin of attractive interaction and
this tells explicitly that there is a phonon part involved the role of phonon is very
apparent because of the isotope effect that we have seen because the Debye frequency
actually scales as the ionic mass. So, the lattice is or rather the TC scales with the ionic
mass. So, the involvement of the lattice is very clear and there we can actually for us
narrow energy range we can get an attractive interaction between the electrons.
So, the wave function of the electrons will have to actually select this energy range to
form a bound pair and then we have written down a many minus body ground state and
have taken a Hamiltonian which is a generic Hamiltonian where the an interaction is
taking place between 2 particles for a particular form of this interaction term that is v K
prime equal to a minus v for the epsilon xi K minus i K prime to be falling in this energy
interval h cross omega d.
We find that the gap or rather we have been able to write down the equation of a gap and
also at finite temperatures have able to write down the temperature dependence or rather
344
the TC how the TC that threshold temperature the critical temperature depends on these
quantities such as v that is a attractive interaction the density of states and the phonon
frequency call it omega d or omega C.
So, this is in a nutshell what is super conductivity all about and. So, as you either destroy
superconductivity by using temperature or thermal effects or you destroy
superconductivity by applying magnetic field the state goes on to a normal state a
metallic state; which is apparently not the case for the high minus temperature
superconductors and they have many other complications there are very little consult
consensus. Whether phonons are involved into this pairing mechanism or there is
something else that is involved in any case.
And also these behavior or rather this 2 delta by K T see that we have found to be 3.52 is
a value there is much higher maybe 5 between 5 and 6 and which are not at all called as
the weak coupling superconductors and these high temperature superconductors they
belong to a class of a non-weak coupling rather strong coupling superconductors.
345
Prof. Saurabh Basu
Lecture – 22
Ginzburg Landau Theory, Coherence length and penetration depth
So, in this session we are going to learn the Ginzburg Landau Theory of
superconductivity.
Remember this was evolved before the microscopic theory that is BCS theory was put
forward and it is a phenomenological theory of second order phase transition which was
extremely successful in describing a few properties of these superconductors such as the
penetration depth and the coherence length and thereby a ratio of them could be formed
which helps in distinguishing between type 1 and type 2 superconductors.
So, in 1937 landau proposed a theory for second order phase transition now by second
order phase transition what we mean is that, that involves latent heat this is one of the
modern definitions of phase transition in this theory there is an order parameter that
continuously vanishes across the phase transition. So, this order parameter is nonzero for
T less than T c and it is equal to 0 for T greater than T c.
346
So, this is an indicator of the phase transition say for example, in a paramagnet to a
ferromagnet transition the magnetization can be the order parameter. So, the
magnetization let us call it by m so, magnetization is not 0 for T less than T c which talks
about a ferromagnetic state and magnetization being 0 for T greater than T c. So, this is a
ferromagnet and this is a paramagnet.
Similarly, in superconductors we can talk about the energy gap of the excitation
spectrum this can be taken as the order parameter call it as delta and Δ≠0 for T <
Tc, where T c is the transition temperature for a superconductor which is a characteristic
of a material. So, this corresponds to a superconducting state and Δ=0 for T > T c
this is called as the normal state or a non-superconducting state.
Now, since the order parameter continuously evolves across the boundary of the phase
transition it should be possible to write down the free energy as a power series in the
order parameter. So, this is the whole idea of GL theory a Ginzburg Landau theory I am
abbreviating as GL theory. So, it says that free energy should be expanded in terms of
order parameter, powers of in terms of a power series of order parameter.
Now, it can be noted that free energy is a scalar; however, the order parameter can be
higher dimensional quantity such as a vector or a tensor or a complex quantity. So, this is
what we are going to do that is write down the free energy functional the meaning of the
word functional is function of a function.
347
So, we will write this free energy functional which is a function of the order parameter
and order parameter could be a function of say the special parameter such as or any other
parameter that is relevant to the problem it is usually the space variable which is taken as
that that the order parameter being a function of that now we will study the GL theory.
So, the order parameter in GL theory is the wave function of the so, called super
electrons call that as ψ (r ) such that the density of the super electrons, what I mean by
super electrons is that, they are superconducting electrons which is equal to
ns ( ⃗r )=| ψ ( ⃗r ) |2
So, the plan is like this that we will do two cases one is absence of an external field
magnetic field and then also we will do it in presence of an external magnetic field and
finally, we would compute two different quantities one is called as the coherence length
and b called as the penetration depth.
348
So, the free energy in H external equal to 0; So, there is no external field and we write
2 β 4
down the superconducting free energy F s=F r + α | ψ | + | ψ | note that we are
2
keeping the expansion only up to the biquadratic that is aquatic terms of psi and the other
thing is that there are no odd powers of psi involved because that would be anti-
symmetric with respect to the change in the order parameter.
So, F s is free energy of a superconductor F n free energy of a normal metal and alpha
and beta are phenomenological coefficients expansion coefficients. So, minimum of F s
so, we want to minimize the free energy with respect to the order parameter and find out
where the optimum value which minimizes the F s lies minimum of F s is obtained by
d Fs
taking =0=2 α | ψ |+2 β | ψ |3=2 | ψ |( α + β)
d | ψ|
So, psi mod square that does minimize this let us write it with a
β 2 2
ψ0 2= −α → F n−F s= α ( α )− ( α ) = α and it is almost obvious that alpha and beta
β β 2 β 2β
both could be temperature dependent.
2
Now, in the first term so, let us write this once more that F n - F s is a α
2β
349
2
Fn −F s= α . So, in the first order that is not too far from T c that is at temperature
2β
which is not too far from T c α ~( T −T C ) . The reason is that alpha equal to 0 at T
equal to T c alpha is less than 0 at T greater than T c and alpha > 0 at sorry this is T less
than T c and this is T greater than T c ok. Now this happens because | ψ |=0 for T >
a (T c −T ) ( 21 )
Tc. So, | ψ |=[ ] here for T < T c.
β
So, the important thing for us to understand is that alpha changes sign across the phase
transition. However beta should not change sign because if both of them change sign
then of course, will not have this equation valid which is psi mod square equal to minus
alpha by beta and moreover beta also should not have any temperature dependence so,
that we will have some pathological.
So, if beta has to be help if beta does not have to change sign across the transition then it
has to be like T minus T c whole square because the whole square does not say in sign,
but in that case your alpha by beta will be like one over T which would be divergent as T
goes to 0 and that should not happen. So, beta should be a constant so, basically it is a
positive constant and independent of temperature of course, it means that being constant
means that it is independent of temperature.
350
So, at least for small deviations from T equal to T c this is what should happen. So, alpha
has a form which is like a α =a(T −T c ) a is a constant and hence will have the
| ψ |=0 for T > T c and psi mod equal going as T Tc - T whole to the power half with
as and b s as constant, a is a constant here.
Now, consider a superconductor in an external field magnetic field that is. So, the way it
e⃗A
is taken care of is that the ⃗p→⃗p − this you must have seen in your classical
c
mechanics course when we talked about a charged particle in a magnetic field. Now this
A is the vector potential which is related to the field as H ext (⃗r )=∇× ⃗
A (⃗r ) . So, it is
always derivable from a vector potential by this relation this tells that the
∇ ⃗ − ie A
⃗ →∇ should be replaced in presence of a magnetic field by this.
ℏc
Now, the free energies have to be written down the same free energy that we have
written down earlier. So,
2
2 1 4 ℏ 2 ⃗ ie A 2 H ext
F s=F n+ α | ψ (⃗r )| + β | ψ (⃗r ) | + |[ ∇ − ] ψ (⃗r )| + and plus a magnetic
2 2m ℏc 8π
energy because of the external field which is given by H external square by 8 pi. Now
free energy ⃗ ψ (r ), ∇
F=F[ ψ (⃗r ) , ψ(⃗r )∗ , ∇ ⃗ ψ (r )∗ , A (⃗r )] .
351
So, minimizing it
−ℏ 2 A (⃗r ) 2
ie ⃗ 2 H 2ext
δ F s=[ (⃗v − ) ψ (⃗r )+ α ψ (⃗r )+ β | ψ (r )| ψ (⃗r )]+
2m ℏc 8π
So, let us just write it as H external square over 8 pi, right now ignore this term it is not
that it is not important or it is small because we do not need it for our discussion now.
So, we ignore that and this when we put this quantity δ F s=0 that would tell that
these bracket has to vanish this curly bracket has to vanish for delta F s to be 0 this
bracket will be equal to 0 and hence we get the first Ginzburg Landau equation as
−ℏ2 ⃗ ( ⃗r ) 2
ie A
[ (⃗v − ) ψ (⃗r )+ α ψ (⃗r )+ β | ψ(r) |2 ψ (⃗r )]=0 .
2m ℏc
So, note the difference between the earlier case when we do not have magnetic field is
that psi was homogeneous we could mention that here as well that psi is and does not
depend upon r basically. So, this is called as a first GL equation and in which we have
neglected the next the second term there.
Now to get the second GL equation minimize F s with respect to A the vector potential
and that gives a very familiar equation which is called as the amperes law which is like
4π ⃗
∇×H ext (⃗r )= j(⃗r ) . So, this is the second GL equation where your
c
352
⃗r )= −ie ℏ [ ψ∗ (⃗r ) ∇ ψ (⃗ ∗ e2 2
j(⃗ r )− ψ (⃗r ) ∇ ψ (⃗r ) − | ψ | A (⃗r )] so, that is the second GL
2m mc
equations.
And let us now talk about the boundary conditions and in doing so, we would get or we
would sort of obtain the relations or the expressions for the penetration depth and the
coherence length. So, let us talk about an inhomogeneous order parameter which we
have been talking about since we have introduced the magnetic field.
So, let us consider a half region of so, this is x equal to 0 so, this is where the
superconductor exists. So, superconductor exists for x greater than 0 and x less than 0 we
do not have a superconductor it could be a normal metal or it could be a magnetic metal
for example, so there is a non-superconductor will just simply write it as non-
superconductor.
So, these are junction systems which are routinely studied in experimental physics
experimental condensed matter physics and one can easily make a junction of a
superconductor and a non-superconductor. Now for deriving the expression for
coherence length put A equal to 0. So, we do not need the external field there and in that
case I get from the first GL equation and the variation is taken to be in one dimension.
−ℏ2 d 2 ψ
So, this is 2
+ α | ψ |+ β | ψ |3 =0
2m dx
353
So, we have said that earlier that alpha is negative in the superconducting state. So, take
α =−| α | and define a quantity called as xi which is called as the coherence length
ℏ2
and this ξ = also this is simple algebra also write equation let us call this as the
2 m| α |
first one as equation 1 first GL equation, the second GL equation to be equation 2 and
this to be equation 3.
β 2 2
So, now write equation 3 in addition define that ( ) ψ =f . So, with these
|α|
2 3
definitions one can write down equation 3 as −ξ f ' '−f +f =0 . So, I have cast this
first GL equation in terms of a scaled variable which is called as a F.
df
Now multiply by a f '= and then one should be able to write this as
dx
d − ξ2 f ' 2 1 2 1 4
[ − f + f ]=0 . So, if d dx of this equal to 0 which means that
dx 2 2 4
− ξ2 f ' 2 1 2 1 4
− f + f =const . Now far from the boundary that is into the
2 2 4
superconducting state. So, this is the boundary and if you are too much into the
superconducting state f '=0 or ψ '=0 .
354
So, the variation of the order parameter with as a function of x should be equal to 0 and
|α|
hence this gives that f 2=1→| ψ | 2= which is what we have obtained. Then this
β
2 2 1 2 2 x
equation which is equation 4 becomes as ξ ( f ' ) = (1−f ) →f (x )=tanh( ) .
2 √2 ξ
|α| 1 x
So, that gives that ψ (x)=( β )2 tanh( ) . So, that tells that ψ (x) has a
√2 ξ
variation like this. So, that gives the extent of the wave function for the cooper pair or the
super electrons and there is a characteristic length that emerges which is equal to ξ .
So, ξ is the measure of the distance over which the order parameter responds to a
perturbation. In this case the perturbation is the presence of a boundary that lies between
the superconductor and maybe a normal metal again, since α =a(T −T c ) the
temperature dependence of the coherence length so, this is called as the coherence
length.
(ℏ 2) T −1 /2 T −1/ 2
So, ξ (T )= (1− ) . So, ξ (T) diverges as (1− ) . So, as T
(2 m a T c ) Tc Tc
goes to T c this diverges in this fashion you can see clearly that the divergence is like a
square root divergence. So, this is the behavior of the coherence length.
355
Let us look at the other quantity which is the Penetration depth, look at this expression
the second GL equation if you drop the first term and only look at the second term that is
this term let me just mark it in red. So, if you look at this term then this exactly looks like
the London equation this without the first term that is a usual current term.
So, we are purely looking at the current due to the external field and then it looks like
4π 2 mc 2
that j(⃗r )= 2 ⃗ A (⃗r ) and immediately the λ =
L and this is the vanishing
c λL 4 π e2 | ψ |2
or rather the dependence of the of psi.
So, the space dependence of psi we will yield the space dependence of λ remember
x
this ψ actually falls off as tanh ( ) . So, that will determine that the how
√2 ξ
lambda l falls off as a distance and to look at the temperature dependence the
mc 2 β T −12
λ L (T )= (1− ) . So, this is λL as a function of T this is the
4 π e2 a T c TC
penetration depth.
356
T −1/ 2
So, λ L (T ) diverges again as (1− ) exactly like xi of T and one can also
Tc
λ L mc β 1/ 2
define a dimensionless parameter called as κ = = ( ) which has only one
ξ eℏ 2π
unknown parameter which is beta which appears in the Ginzburg Landau theory. Now,
1 1
we know that κ> is termed as the type 2 superconductors and κ< the type
√2 √2
1 superconductors.
So, to summarize that without doing explicitly a microscopic theory which we have done
for the BCS case here simply writing down the free energy functional as in powers of the
order parameter and minimizing it with respect to the order parameter one can actually
get the 2 energy scales that are relevant for these superconductors namely the penetration
depth and the coherence length and not only that they are how they diverge for T close to
T c is also tailed which are of this order as 1 by T minus T c whole to the power minus
half.
357
So, to conclude the chapter on superconductivity we have given one a historical

introduction and then we have said about properties such as Zero resistance, Meissner
effect etc. And then we have talked about the origin of the attractive interaction between
electrons mediated via phonons this is known as Cooper’s instability.
And then of course, we have done BCS theory and in which the gap superconducting gap
is obtained and the superconducting gap is seen to be non-analytic in powers of the
strength of the attractive electron interaction. So, one cannot do a Perturbative theory in
order to get this result then we have done a variational theory and also gotten the
behavior of the superconducting gap as a function of temperature and then finally, we
have done a Ginzburg Landau theory to obtain one coherence length and to penetration
depth.
And these are all required for you to learn because this story of superconductivity is
fairly old now you understand it is more than it is about 110 years old since it is first
discovered and then all also it was the new class of superconductors were discovered in
1986 and a large amount of work had gone in. Since then however, we have not touched
that part because of poor knowledge or still you know evolving knowledge in that
particular area, but we have done these studies of the weak coupling superconductor so,
called the BCS superconductors in somewhat details.
358
Prof. Saurabh Basu
Lecture - 23
Quantum Hall Effect
So, in this set of special lectures we shall do quantum hall effect to begin with and then
we will see that how it gets connected to more exotic phenomena. So, we are familiar
with hall effect, hall effect is an effect which was discovered in 1879 by Edwin hall in
which the current is made to flow in a thin laminar sample in say the x direction and
there is a voltage that is generated in the y direction because of the migration of the
charge carriers abide d by the Lorentz force equations and at equilibrium one measures
transverse voltage which is known as the hall voltage.
So, this is the quantum version of that and in which will see that the resistivity versus
magnetic field plot which we have seen even in the classical hall effect, here there are
some strange features in this data and we will discuss that how it comes about.
So, there are in principle two kinds of hall effects which are integer quantum hall effect
which is what we are going to do mostly and there are also fractional quantum hall effect
I will just mentioned what they are, but will not get into the discussion because they are
more difficult and they need interactions to be taken into account. So, the first
359
experiments exploring quantum hall effect were performed by von Klitzing in 1918 and
then approximately which is 100 years later after Edwin hall discovered the classical hall
effect as we have just mentioned.
And Klitzing was awarded Nobel Prize in 1985 for this discovery and you will see that
why this experimental discovery is so important so, as to you know warrant a Nobel
Prize.
So, we shall talk about only IQHE as I just mentioned. So, what is the system? The
system is a thin 2 dimensional material or quasi 2 dimensional material in which the
electrons are restricted to move in a 2 dimension and the extent in the third dimension is
fairly limited. So, the motion is confined in the in the plane let us call it as x and this as y
direction. So, this is a 2 dimensional electron gas and it is written as 2 DEG in short and
there is a strong magnetic field that is been applied which is perpendicular to it.
So, it is coming out of the plane and this is the magnetic field is of the order of 5 to 30
tesla, the experiments are conducted at very low temperature so, that the quantum effects
become important. Disorder is an integral part of this of this effect or this experimental
discovery and it plays an important role in a seeing the phenomena better. In fact, in the
absence of disorder it can be shown that it reduces to classical Hall Effect.
360
So, this is the plot of row x x and row x y so, let me point out which are the so, what is
happening is that. So, this in this system a current is made to flow in this direction let us
call it as j x and there is a voltage that is developed here in the transverse direction and
because of the segregation of charges here and this voltage is called as hall voltage. So,
there is also a resistivity in the direction of the flowing current which we will call it as
rho x x which is generally called as the longitudinal resistivity or it is also called as the
magneto resistance and there is the other one which is the called as the transverse
resistivity and also called as the hall resistance or resistivity they are two different things
because we know that the resistivity is a property of the sample while resistance has also
the geometrical properties built into it.
So, in principle R which is the resistance which is what the experimentalist measure that
has a relationship with the R=ρLρLL( 2−d ) and this you can understand it well by noting
down the definition of a resistance that you have read in school for wire of diameter d
L L
and length l. So, that goes as R=ρLρL =ρLρL 2 =ρLρLL−1 and that is the story with d equal to
A L
3 and.
Now, this difference between the resistance and the resistivity that go away a for d equal
to 2. So, one can safely talk about either resistance or resistivity at d equal to 2, because
the geometrical factor appearing here will not contribute anything to the formula extra.
361
So, R becomes numerically equal to d and that is why we can talk about either of either
resistivity or resistance and so, these are the plots for. So, these step like plateaus are the
plot for the hall resistance which will denote by a row x y and these spikes that you see
are for the longitudinal resistance or the magnitude resistance. So, these are our plots.
in the classical hall effect, the whole resistance or their hall resistivity as a function of b
was a straight line, here it is far from being a straight line, there are plateaus at some
integer values will discuss what that is and at i equal to 2 and i equal to 3 and i equal to 4
and 5 and 6 and 8 and 10 and so, on and so, these are in units of e square over or rather h
over e square. So, the resistivity the whole resistivity for this plateau is a to h over e
square this is 3 h over e square and 4 h over e square and so, on.
Exactly that and these integers are independent of the material, independent of the
amount of disorder present and is very very universal and so, much so, that these integers
are correct up to ninth or 10 to the power minus 9 or 9 decimal places. So, the effects or
the experimental observations at that robust and absolutely independent of the material
used and hence this was really a discovery that has taken the scientific community by a
storm and this lot of work since 1980 that have started and still going on in various
fields.
So, this is another important thing about this resistivity that the longitudinal resistance is
0 most of the time, but it shows a peak as the hall resistivity makes a transition from one
plateau to another. So, it is only when it makes a transition from one plateau to another,
it shows a peak and you see the peaks are all there accompanied by the jump from one
plateau to another. So, you see the rho x x which is the longitudinal resistivity of the
magnitude resistance, it is 0 which means that resistivity 0 means it is a perfect metal or
a perfect conductor and suddenly the thing the resistivity shoots up giving rise that there
is a lot of resistance there. So, the system is undergoing a transition a series of transition
from being insulated to metal to insulator to metal and so, on.
362
So, these are some of the interesting features of the quantum hall effect which are which
are very important and for us to understand that what is going on. So, just to summarize
that we have said this the hall resistivity and the longitudinal resistivity, the hall
resistivity is given by ρL xy and longitudinal resistivity by ρL xx exhibit very interesting
h 1
behavior as we have just seen, a while ρL xy has plateaus ρL xy = ( ) where gamma is
e2 γ
an integer up to an accuracy 10 to the power minus 9. And because of this robustness of
these plateaus and remember that we have said that disorder is an integral part of the
system and since disorder does not do anything which means the fact that translational
invariance is broken has does not have any effect on this.
So, there must be some other mechanism that is protecting the flatness of the plateaus h
over e square is thus taken as a quantum of resistivity it is also called as a Klitzing
constant by the name of the discoverer it has a value approximately 25.8 kilo ohm and
you can see that for γ equal to one here this denotes exactly one quantum of resistivity.
So, it becomes 1 over 1 which is just simply h over e square.
363
There are more features which we unfold 1 by 1, the center of each plateau occurs at the
hn n
magnetic as a magnetic field assumes a value which is B=ρL = ()
ϕ , where
γe γ 0
ϕ 0=h /e is called as a flux quantum. So, as it takes a fractional multiple of this phi 0
or rather this sum coefficient multiplied by phi 0 then that is where the plateau occurs.
So, it could happen that the value of the magnetic field and so, n is the density of the
material hence n is the electronic density.
So, what it is saying is that the value of the magnetic field and the electron density have
to conspired to give rise to a plateau. So, this is the flux quantum phi 0, which is equal to
h over e. So, these are precise the values of magnetic field at which the gamma number
of landau levels are filled at this moment it is an unknown word, but we will see what
landau level spin and the hole resistivity takes a value as we have given in equation 1
which is this.
The quantization also persists over a range of the external magnetic field and another
striking feature which is we have already told is that there robust to disorder and it may
sound odd in the context of condensed matter physics because a large number of
phenomena gets adversely affected by presence of disorder while this is not.
364
So, the longitudinal resistivity which is ρL xx it shows a surprise as it is 0 most of the time
as we have told. So, it is 0 here, 0 here, 0 here and so, on except when there is a jump in
ρL xy from one plateau to another. So, the above features can be understood from a
classical picture. So, let us see what the classical picture unfolds in this particular case, it
has got nothing to do with quantum mechanics, but these will help us in understanding.
So, ohms law states that J=ρLσEE , this is another way of writing V equal to IR, everyone
knows sigma is a is a conductivity. So, sigma is conductivity, J is current density and E
is the electric field. So, sigma is a number if the current density J and the electric field E
are pointing in the same direction. So, J x equal to sigma E x and so, on and in which
case sigma is a number, but in presence of a magnetic field sigma is a tensor, it is a 2 by
σE xx σE xy
2 matrix having a form. So, σE=ρL
(
−σE xy σE xx )
. So, it tells you that diagonal elements are
same of diagonal elements are anti symmetric.
So, these are the features of this matrix and so, if we write the conductivity tensor we can
also write the resistivity tensor.
365
So, the resistivity tensor is written as ρ−σ−1=( ρ xx ρ xy ) this preserves the similar
−ρ xy ρ yy
properties with an of diagonal elements being anti symmetric. So, the it is easy to see
ρ xx −ρ
that the components have the relation that σ xx= 2 2
; σ xy= 2 xy2
ρ + ρ yx
xx ρxx + ρxy
1
Now, if ρxy =0 then we get σ xx= ρ . So, this is familiar and σ xy =0 is equal to
xx
0. So, this is 1, 2 is that if ρxy ≠0 then σ xx , σ xy both exist. So, this is clear to easy
to see that if rho x y is not equal to 0 then both of them will exist and provided of course,
your rho x x is also not equal to 0.
366
So, now consider ρxx =0 , σ xx =0 if ρxy ≠0 . So, that is clear from here if ρxx=0 then
σ xx=0 if of course, ρxy =0 is not equal to 0. So, what do we want to call the system
with rho x x equal to 0? So, we have ρxx=0 σ xx=0 , on one hand this tells that it is
a perfect conductor and this says that it is a perfect insulator. So, one is implying the
other, but one inference is that it is a perfect conductor the other inference is that it is a
perfect insulator.
So, the question is that what is going on? And in order to understand this let us just look
at the Drnde’s model.
367
So, in the Drude model this is in the preliminary solid state physics course were are in
the free electron theory of studied Drude a model, which gives you the assumption that
the electrons are non interacting and they only interact when they collide with each other
which happens after every time scale given by the relaxation time. So, a between these
relax 2 scatterings the electron propagates as free particle. So, that is the assumption and
σ0 l
the conductivity is given by a σ xx= 2 2
where τ= and omega c some
1+ ωc τ vF
cyclotron frequency where l is the mean free path and this is the velocity and sigma 0 is
n e2 τ
taken as σ 0= with n as the density m being the mass.
m
So, thus imply that the relaxation time becomes extremely large so, that tells that no
scattering. So, in this particular case the current is flowing perpendicular to the field and
it has a form which is we are talking about 2 dimensional transport. So, this is like
E 0 ρ xy j x ρ j
E=( x )=(
⃗ )( )=( xy y ) and so, on. So, it is easy to see that that the Ex
E y −ρxy 0 j y −ρxy j x
is the so, Ey is in the direction of Ex and jx in the direction of opposite to the direction of
E y and that tells that E and j are perpendicular vectors.
368
So, if E and j that is the electric field and the current density of perpendicular then
E . ⃗j=0 but remember that
⃗ E . ⃗j
⃗ has the meaning of work done in accelerating the
charges. And so, the fact that E . ⃗j=0 it means that there is a steady current.
⃗
Implies that there is a steady current flowing in the sample which does not require any
work to be done and hence no dissipation.
So, σ xx=0 tells us that no current is flowing in the longitudinal direction. So, this is
like an insulator, while ρxx=0 tells that there is no dissipation of energy. So, this is
the like a perfect conductor when we add the word perfect he also like a perfect
conductor. So, so, σ xx=0 equal to ρxx=0 is accompanied by by that
−1 ν e2
σ xy= ρ = and mu being and integer. So, this quantization of mu being an integer
xy h
to the accuracy of 10 to the power minus 9 is independent of the details of the system
and hence the effect is universal.
So, this goes back to this plot so, this plot tells exactly that the ρxy will become a
quantized number in units of h/e 2 and it will make a transition from one plateau to
another and during which the longitudinal or the magnet or resistance longitudinal
resistivity of the magnet or resistance will show a sharp peak and this at least
qualitatively explains that these are universal features of a 2 dimensional electron gas
369
and persists even in presence of disorder. An important thing is that since disorder is not
protecting the plateaus, that is it is getting or rather it is contributing to the robustness of
the plateaus as we have seen and a magnetic field already has broken time reversal
symmetry. So, there has to be a another quantity or another you know sort of preserving
factor that preserves this plateau and which is what has been attributed to the topological
properties of the system that is protecting these plateaus.
So, what happens is that in presence of the magnetic field ah. So, their electrons go round
in such circular orbits and the radius of this or rather radii of this orbits are proportional
to the magnetic field. So, this does not lead to any transport, but; however, the transport
occurs along the edges, where the electrons actually skip and these 8 states are robust no
matter whatever a kind of perturbations you put. So, these 8 states will be robust and will
carry current and will give rise to this resistivity curves a while the bulk of the material
that is interior of the 2 dimensional electron gas the electrons will not contribute to any
transport of charges. Let us try to understand these features better and let us talk about a
simple language and non-interacting.
So, we want to talk about landau levels so, electrons in a magnetic fields. So, these are
non interacting electrons in a magnetic field. So, remember what happens in for a
charged particle e to be put in a magnetic field the canonical momentum ⃗p→⃗p −q ⃗
A
over well c will drop those are old units we can work in new units and now q is equal to
370
minus e for electron. So, we can write this as ⃗p→⃗p + e ⃗
A , where A is the vector
⃗ ⃗
potential and it has a relationship with the magnetic field given by ∇× A=⃗
B , that is
the magnetic field we are talking about ok.
1
So, the Schrodinger equation that we have to solve is A )2 ψ=E ψ . So, that
(⃗p + e ⃗
2m
is the time independent Schrodinger equation where the p has been replaced by
⃗p→⃗p + e ⃗
A , now Schrodinger equation is gauge invariant. So, even if we change A to
A ⃗ χ (⃗r )
A− ∇
⃗→⃗ call it chi an arbitrary scalar function which is a function of r only,
then size simply picks up a sign which is given by and because if psi picks up only a sign
only a sorry face not sign it is a phase this quantity which is a physical quantity remains
in variant.
So, just to revise it once again that we write down the Schrodinger equation, the A is a
gauge invariant and because of the gauge invariance psi picks up a phase
−ie
( r)
) ϕ (⃗
ψ (r )→e ℏ
ψ( ⃗r ) , because I only picks up a phase and not anything else, the psi
mod square will not have this phase because it is psi mod square is psi star psi and you
will have a simple you know these will be same and hence we can do a gauge
transformation on this and still get the same result.
371
So, now choose a particular gauge for calculation and that gauge is let us call it
A x =−B y . , While A y = A z =0
This is called as the landau gauge that is the definition, this is fine because ⃗ ⃗
∇× A=⃗
B
and so, B is in the z direction as is a requirement of the hall effect. So, we have taken in
principle we have taken a 3 dimensional Schrodinger equation, but; however, very soon
you will see that that it boils down to actually a 2 dimensional motion and the B is
actually appointing the external magnetic field is pointing in the z direction and if that is
the case then in this gauge Schrodinger equation I am writing it in short. So, it becomes
1 2 p 2y p2z
equal to [ ( p −e B y ) + + ] ψ (⃗r )=E ψ (⃗r ) . So, as any quantum mechanics
2m x 2 m 2m
problem we have to find out the Eigen values and the Eigenvectors for this particular
equation.
So, you can see that the z direction the particle motion is like a free particle . So, we have
p2z
the energy in the z direction is like a free particle ϵ= and there is no other term
2m
which involves z. So, the motion is that of a free particle. So, now, we can now neglect
the z direction because it is like a free particle and see what happens in the x and y
direction.
372
1 2 p 2y
And to see that let us write down [ ( p −e B y ) + ] F( x , y)=ϵ F ( x , y ) is no
2m x 2m
longer a function of z because in the z direction the wave function is like let us call it as
say F of z is some with some normalization it is like exponential I k z.
So, we are only concerned about the 2 dimensional motion and this is what we were
talking about earlier that even if you have taken 3 dimensional case in presence of a
magnetic field for a non interacting electrons it boils down to a problem of 2 dimensional
motion. And this is equal to some epsilon and F x y so, this is like the Hamiltonian in 2
dimension H x y ah. It is easy to see that p x is a constant of motion, [ p x , H ( x , y)]=0
y p x is a constant of motion because nowhere in this Hamiltonian we have x the variable
position variable x and p x is equal to this is equal to 0. So, that tells that p x is a constant
of motion, remember p y is not a constant of motion because of the gauge chosen where
we have a y remaining so, p y does not commute with y and hence it is not a constant of
motion.
So, there is no term because there is no term linear in x this is what happens and that also
as I told because of the gauge, but it does not matter even if you can taken term A to B B
x then the x should have been p x would not have been a constant of motion while p y
would have been so, x and y simply gets interchanged. So, if p x is a constant of motion
2 π ℏ nx
then p x is actually quantized by px = where nx is of this quantized in this
Lx
form which is which takes values which are like 1, 2, 3 and so, on.
So, one can write this equation let us call this the original equation to be 1 and this
p 2y 1 eB 2 2
equation to be 2. So, one can write equation [ + m( ) ( y − y 0) ]f ( y)=ϵ f ( y ) .
2m 2 m
px
So, this is the equation, where y 0= . So, that is the equation that is the Schrodinger
eB
equation that we have to solve and it is very clear that this equation is the equation for
the same equation same Schrodinger equation, when written for harmonic oscillator in
px
the y direction and oscillating about this point is 0 given by y 0= and then of
eB
course, we know that we have solved the problem.
373
In the sense that we know that it is of the harmonic oscillator form. So, it is a simple
eB
harmonic oscillator with frequency ( )=ω . So, the Eigen values are simple epsilon
m
equal to n plus half e B by m and h cross.
1 eB
So, that is our so, it is ϵ =( n+ )( )ℏ . So, this is the eigenvalue for this problem and
2 m
these are called as the landau levels. Not only we know about the energy Eigen values
we also know about the Eigen functions so, the 2 dimensional Eigen functions. So, it is a
2
(y− y 0 )
1 ik x −e B eB( y− y 0 )
free particle we told that the F( x , y)= e Nn e
x ℏ
Hn( ) . So, it is
√ Lx ℏ
a free particle in that direction in the k x direction and some normalization, this is the
eB( y − y 0)
Gaussian which is Hn( ) and now the Hermite polynomial you must be
ℏ
knowing the properties of the Hermite polynomial when n equal to even the polynomial
is even when n equal to odd the polynomial is odd and it is given by y minus y 0 divided
by h cross and that is the complete solution of the problem.
There are more things to understand here is that so, these are landau levels and these are
the landau level energies and now you see that a p x is a constant of motion. So, that tells
374
nx
that all n x that satisfy all n x that satisfy this quantization condition k x =2 π ,
Lx
h
where y 0=( ) n . All n x values are valid solutions and since all n x values are
B Lx x
valid solutions the problem is hugely degenerate and gives rise to the enormous
degeneracy of the landau levels because each n x value that satisfy this relation L x is the
dimension of the sample in the x direction will be a valid solution and hence it is a
enormous this degeneracy will be there.
Now, what is the extent of degeneracy? So, let us try to understand. So, let us take this
energy value so, your y is y naught. So, this is the point about which the particle is
h 2 π nx
undergoing a simple harmonic motion. So, this is given by y 0=( ) k x ; k x=
B Lx Lx
Lx
Now, if we put that these k (ne )max=(k x )max which is the point in the y-direction
2π
about which the simple harmonic motion takes place to be the. So, take ( y 0 )max=L y
which is the size of the sample in the y dimension a y direction so, it cannot y 0 cannot
be greater than L y. So, the particle is undergoing a simple harmonic motion about y 0
and y 0 the maximum of y 0 is the sample dimension in the y direction.
375
e B Lx Ly e B A
So, the degeneracy is then given by g=(n0 )max = = , where A is the area
h h
of the sample. So, let us make 2 comments here one is that the degeneracy. So, if we put
all these if y 0 max to be equal to L y and put k x max in terms of n x max one, one gets
this degeneracy. So, degeneracy is proportional to B that is a magnetic field. So, as you
increase magnetic field the degeneracy will increase and the second is that the trajectory
is centered about y 0.
So, this is the story about that electrons non interacting electrons are put in a magnetic
field we have taken a simple gauge called as a landau gauge and have solved the
problem. The problem says that the motion in the z direction is that of a free particle so,
one can neglect that. So, now, the Hamiltonian becomes 2 dimensional and so, the
motion is constrained in a plane in the x direction because of the choice of the gauge the
particle moves as free particle with the momentum vector quantized in terms of 2 π ,
L
some integer multiple of 2π and in the y direction it executes a simple harmonic

L
motion.
So, we can find we can we get this effective Schrodinger equation is a 1 dimensional a
simple harmonic oscillator problem we know what the energy Eigen values are, we know
what the energy Eigen vectors are. Because, any a quantum number in the x direction is
provides a valid solution for the problem, the problem is enormously degenerate and this
degeneracy is found and this degeneracy is proportional to B and it is also proportional to
the area of the sample and the trajectory is a simple harmonic which is centered at y 0.
Now, with this so, having discussed this landau levels and degeneracy let us talk about
how this problem, now you should be finding a similarity between this problem and the
quantum hall effect. There are also the electrons were put in a strong magnetic field in
the z direction here also we are talking about a single electron and finding that how the
motion is a in a presence of a magnetic field.
376
B Ix
So, now the hall voltage is defined by VH= and hence the hall coefficient
ne
VH B
RH= = . So, this is a classical result which is well known.
I x ne
B
So, the degeneracy of the landau levels is given by g=e A so, the degeneracy so,
h
g eB
this is the degeneracy so, degeneracy per unit area which is = =N . So,
A h
eB Nh
RH= = . So, N is the current density, if n becomes equal to some ν into N,
ne2 ne2
where ν is an integer equal to 1, 2, 3 etc then R H, let us do it in a new page.
377
Nh h n
Then RH= = 2 . So, ν= . So, this was the quantization that was
2
ν Ne ν e eB
h
discussed at the beginning of the lecture. So, even starting with a classical picture if you
think that this the assumption is that if the degeneracy per unit area is a quantity which is
like you can take it as a number as n and then the hall resistivity it is quantized in terms
h h
of or rather it is a where N is the electron densities N and if n is taken as the
e2 ne2
gamma into this capital N which is taken as the degeneracy per unit area then it is
quantized actually as h by e square and L over mu.
So, thus a quantized hall resistance is always expected this is important, if a carrier
density n n and the magnetic field B B are adjusted in such a way that that the filling
factor. So, this called this mu is called as a filling factor, mu of the landau levels satisfies
mu equal to n by e B h becomes an integer. So, it is satisfying this becomes an integer.
So, just to rephrase this that even from a classical picture we got this quantization of the
plateaus or the hall resistivity or the resistance, here what we have taken is that this is
called as a filling factor where the total electron density is taken as a filling factor
multiplied by the degeneracy per unit area. So, your gamma is simply equal to n by N.
So, that is the electron density divided by this N, which is the degeneracy or the
degeneracy per unit area if these two things and the b that can be adjusted in a manner
378
such that this quantity becomes an integer then there will be quantization of the hall
resistance.
So, what happens is that under this condition the magneto conductivity let us call it as
σ xx so, it is a longitudinal becomes 0. Since, the electrons are moving like free
particles this is what we have got that the electrons in the x direction that it moves like
free particles exclusively perpendicular to the to the electric field. Thus there is no
scattering in the direction in the direction of the electric field is possible. So, whenever r
ρxy acquires a plateau ρxx or ρ yy in this particular direction will drop.
So, this discussion on a quantum hall effect at least in a as a you know new person or
introductory sort of discussion of this quantum hall effect to you should be sufficient to
make you understand the importance of this particular experiment and the universality of
this experimental observation which is independent of the nature of the material or
details of the material and is independent of the disorder is a very important thing. And
the hall effect where these all these labs were very sophisticated measurement of
resistivity is by is there, people do the hall effect measurements almost on any sample
that comes because it gives rise to not only the magneto resistance of the magneto
conductance also the hall resistance and helps us in understanding the quantization of the
plateaus.
379
Remember this thing that we have said that this has to be a number depending on the
value of B, because the degeneracy is very high the B has to be large and. So, in classical
Hall Effect this was missed and one really saw this as a straight line which is a row x y
versus B or r h versus B was seen as a straight line, which is having a constant slope
which is given as 1 over any. However, in large values of the magnetic field, this is not
h
the case where hall resistivity has intermitted plateaus at integer values of , the
e2
same thing in a, happens in fractional quantum hall effect where the quantization occurs
h
at rational fractions of in units of .
e2
Thank you.
380
Prof. Saurabh Basu
Lecture – 24
Spin Hall Effect, 2D topological insulator
So, in this part of the ongoing discussion we shall talk about spin hall effect and also
introduced just introduced this Hamiltonian called as a Kene Mele Hamiltonian which is
related to the spin hall effect. However, the level of rigor that will be used in this
discussion will be far lesser than what we have done so far, we will simply make a few
mentions and tell you the physics behind it, however, will not derive anything rigorously
as we have done earlier.
So, here we are going to talk about spin hall effect. We have talked about hall effect
which in principle means a charged hall effect where the charge migrate charges they
migrate to different edges of the opposite edges of the sample.
So, that there is a transverse field that develops. Here we are going to talk about the same
effect with spins that is spins will segregate up and down spins will segregate on two
edges of the sample and that will give raise to a spin hall effect. So, we have seen that the
plateaus in the quantum hall effect are robust, very robust to the accuracy of 10 to the
power minus 9 for their flatness because each of these plateaus take integer values of h
381
over e square and these do not go away even in presence of disorder. And in any case the
time reversal invariance is gone owing to the presence of the magnetic field a which is an
essential component of the hall effect that we have seen.
Now, this robustness of the plateaus or the plateaus being a exactly occurring at integer
multiples of h over e square or in the context of fractional quantum hall effect they occur
at rational fractions of multiplied by h over e square. So, those come from the bulk
boundary correspondence. What I mean is that, so the system will only conduct through
the boundaries or the edges and the bulk remains completely insulating. So, physical
picture is that this is a two dimensional electron gas and you will have the bulk where the
electrons execute a cyclotron orbit while it is skips along the edges and where the entire
you know the conductance takes place.
So, this distinctness in the behavior of the bulk and the edge that yield a name which is
called as a topological insulator. So, it is a very important and a recent trend in the study
of condensed matter systems and in which there are lot of classes of topological
insulators that are being predicted and some of them have been discovered and they
show unusual conductance properties which are very distinct from ordinary insulators or
which are called as the band insulators.
So, the schematics of a quantum hall state is just what we said that in the bulk we have
electrons executing cyclotron motion. There is a magnetic field which is perpendicular to
382
this which is coming out of the page and there are these skipping orbits. So, these are
known as the skipping orbits, and these are electrons executing a circular motion.
And the band diagram for all these classes of topological insulators so to say, they
actually cross at the boundaries of the sample. So, this is where the crossing of the Fermi
levels occur and it is only at the edges the bulk is separated away from the Fermi level
and hence they do not conduct. So, another schematic is that that the bulk remains
insulating and there are edges where the charges propagate in two different directions
and this gives rise to the conductance that we have seen in the quantum hall systems
which are two dimensional electron gas with disorder.
So, an ordinary insulator as I said that it is also known as band insulator has no such
distinction. So, in which case they are both the edge and the boundaries are insulating,
both the edge and the this should be bulk not boundaries edge means boundaries. So,
both the edge and the bulk please make this correction are insulating a while it is of
course, meaningless to talk about conductance properties of ordinary insulators, but it is
quite relevant to study the transport characteristics of a topological insulators.
So, since 1980 that is the discovery of the quantum hall state there is a search for another
class of topological materials where similar bulk boundary distinction exists. So, this
bulk boundary distinction is an important component of the topological insulator which
has to be present in order to be called as a topological insulator. So, it is realized that a
383
time reversal invariant term can achieve the goal. So, you have to get a time reversal
invariant Hamiltonian and this immediately eliminate the usage of the magnetic field
which evidently would have violated the time reversal invariance.
So, in summary what we need is another class of topological insulators. Once again the
word is coming; the word topology is coming because there is a distinction between the
bulk of the sample and the edges of the sample. So, the topology is important and this
topological insulators have only conducting states at the boundaries. So, we cannot use a
magnetic field because magnetic field would immediately destroy the time reversal
invariance. Now, what is understood is that a spin orbit coupling which is strong enough
to give a visibility to this effect could be a thought of an agent which can do the job, that
is get another class of topological insulators other than of course, which we have seen the
the quantum hall systems.
So, this is what was what was thought about it at that point of time which is since 1980
onwards or rather 84-85 onwards when people have fully understood the quantum hall
state and understood that the plateaus are actually protected by the edge states. So, as
long as the edge states would continue to exist there will be transmission or there will be
conductance facilitated by the boundaries even while the bulk remains completely
insulated.
384
So, in order to understand spin orbit coupling this you might have been exposed to in the
context of hydrogen atom or real hydrogen atom there could be L S coupling or which is
known as the spin orbit coupling L is the orbital angular momentum and S is the spin
angular momentum. However, in hydrogen atom you found it that it is alpha square
times smaller than the unperturbed energy where alpha is called as a fine structure
constant which has a value one over 137. So 137, alpha square is of the order of 10 to the
power minus 4.
So, it is the unperturbed energy in hydrogen atom is of the order of this is minus 13.6 by
n square electron volt. So, if we take a n equal to 1 then its minus 13.6 electron volt in its
of the order of an electron volt whereas, this is 10 to the power of minus 4 times smaller.
But when you also should realize that the spin orbit coupling term came with a Z square
and Z for hydrogen atom equal to 1 which is called as the atomic number the number of
a protons or a electrons that are present in the system protons present in the nucleus or
electrons present in the system which is equal to 1.
However the spin orbit coupling can become important for Z to be large such as you
know there are Z which is the uranium has Z equal to 92 etc. So, for them spin orbit
coupling can still play an important role. So, this is the SO coupling let us call it as SOC
this is what we have called this as a SOC is Z square times this number which is here.
However, in real materials which are not atoms the spin orbit coupling can still be very
significant and so much so that it cannot be neglected.
So, there are a number of reasons that spin orbit coupling arises in real materials, one of
them being when there is a bulk inversion symmetry being broken. So, it is a three
dimensional structure which is inhomogeneous and its somewhere at the middle of the
bulk or the somewhere inside the bulk the environment that the electrons face in going
from below that point or below that line or below that surface and above that surface.
They are quite different that is called as a bulk inversion symmetry being broken and in
which case add restless spin orbit coupling is an important quantity to consider.
Likewise, when the surface inversion symmetry is broken there is another spin orbit
coupling goes by the name Rashba spin coupling its by a two people called by (Refer
Time: 12:20) its popularly known as Rashba spin orbit coupling possibly a Japanese
scientist. And it is called our soc and surface inversion symmetry is mainly broken in 2D
385
materials. So, the environment at the surface is very different. So, above the environment
is very different from below I mean above the surface the environment above the surface
is very different from the environment below the surface and hence this is important in
2D materials.
So, as a spin orbit coupling if you open literature quantum mechanics book we cannot
actually derive it from the first principle in non-relativistic mechanics it should come
from Dirac equations. However, we are skipping those derivations and this is the spin
which is interacting with the momentum and this is crossed with a delta V 0, where V 0
is of the order of one over r which is the ionic potential.
So, there is a ionic potential has a gradient at the surface and because of this gradient this
is like an electric field. So, this is like a σ .(⃗p × ⃗E ) and that is pretty similar to what
we have in the Rashba spin orbit coupling. So, there are two of them as I told that is
Dresselhaus which looks diagonal which is like px σ x , where sigmas are the Pauli
matrices and px σ x − p y σ y this is the strength of that beta is the strength and h cross is
the Planck’s constant which we all know about and the Rashba spin orbit coupling is
given by ⃗ ). ^z , where p is the momentum and sigma are the two by two Pauli
(⃗p×σ
matrices.
386
So, this spin orbit coupling does the spin selection which otherwise would have been
done by the magnetic field in terms of. So, the magnetic field when they act on the
system of charge carriers there is a Lorentz force which segregates the charges from into
the two different you know ages of the sample, same kind of things would take place in
system in presence of a spin orbit coupling. So, this is like a two dimensional system
where these are up and down and in presence of the spin orbit coupling that down would
go here and the up would go here. And so there is an accumulation of up spins and there
is an accumulation of down spins here. So, if you have a voltage measuring system
which measure the spin current or the spin voltage then they could detect that there is a
significant accumulation of up spins on one edge and down spins on another engine this
is called as a spin hall effect.
So, it is a counter part of the charge hall effect and here we show an experimental figure
picture rather where you have a band red band on the left corner, and a blue band on the
right corner and these are also shown in black and white where there is a accumulation of
spills being shown in a across the edges of the sample. The only problem is about the
detection of this voltage the voltage is a usually very small even less than a micro volt
and usually of the order of nano volts. And this spin is this accumulation is detected in
non magnetic semiconductors by optical methods which are a known as the Kerr rotation
spectroscopy.
We will not go into details of that a just know that there are systems which in presence of
the spin orbit coupling SOC is being present here the spins will actually get lined up on
the opposite spins will get lined up in two different transverse edges of the sample and
there will be a spin current that is going to be generated. So, this is that spin current
detection, and a multi meter or an ammeter which measures a spin current or a voltage
that is being measured to spin hall voltage.
387
This spin hall effect has a great deal of practical relevance in the a field of spintronics
and it is even now it is growing at a rapid pace to this field of study which deals with
spin polarization, manipulation and detection of spins that are intelligently used for
applications. So, what we used to do with charge will do it with the spins, the spins will
carry the information will process the information and will do the same electronics that
the electron used to do and now, it will be done via the spin. And a large number of a
advantages such as spins have no joule heating, spins do not scatter by impurities, and so
there is a dissipation less communication. So, these are the advantage of spintronics over
electronics.
So, a meanwhile a quantized version of the spin hall effect or rather quantum version of
the spin hall effect which is called as a Q quantum spin hall effect has been proposed in
graphene. We have seen graphene when we calculated the band structure of graphene
and we saw that graphene has a band structure which is linear close to the Fermi level or
there are some points which are called as the Dirac points in the vicinity of which we
have a linear dispersion and there are only two distinct points in the Brillouin zone. So,
the dispersion the two bands of graphene the type binding graphene that touch at 6
points.
However, there are only two distinct points which are called as k and k prime and at
these points are important and these are called as the valleys, so there are these k and k
388
prime points which we have seen when we are talking about a applications of second
quantization and we were talking about the study of tight binding Hamiltonian in
different systems. So, there it was proposed that this quantum spin hall effect could be
observable in graphene.
So, in principle it is possible; however, the spin orbit coupling in graphene unfortunately
is way too small it is only of the order of a few Kelvin for spin hall effect or quantum
spin hall effect to be observed. Nevertheless it remained as a good theoretical proposal
and if that would have been possible suppose one can a manipulate which in a principle
it is possible to manipulate the spin orbit coupling the strength of the spin orbit coupling
in materials. So, in graphene it should be possible and there are a certain add atoms
heavy add atoms which are often used in order to induce a strong Rashba in orbit
coupling in graphene then it would have been I mean it should be in principle possible
and probably the future of such a discovery.
So, this has given rise to a theoretical studies and have given rise to a model Hamiltonian
which predicts in principle the quantum spin hall effect in graphene. We will have a
glimpse of that in just a while.
So, just to distinguish between a spin hall and a quantum hall, now quantum hall had
only one edge. So, this edge I mean one state per edge and here we are talking about n
equal to 1 or mu equal to 1 rather. Here there are two of them and one for each spin and
389
the spins are propagating in opposite directions. So, if the green denotes a down spin
then its opera propagating from right to left and if red denotes an up spin then it is
propagating from left to right and same happens in the other edge as well ok.
So, these are called as helical edge states as compared to the Chiral edge states what we
have seen in the quantum hall state. So, this defines or schematically it defines a
quantum spin hall state where the there will be quantized a conductance which is in; so if
e2
the resistance is H over E square. So, there are the conductance will be . So, this
H
2 e2
will give rise to conductance even at 0 biasing.
h
So, there are some more cartoons which will help you to understand better. See the
normal insulator we have a the blue one is the valence band and the red one is the
conduction band and the Fermi level is the dotted line that is shown here. So, the there is
a distinct difference between the red and the blue with regard to the dotted line which
means they never touch each other and hence there are no conduction and this is called
as an ordinary insulator or a band insulator you can call it a normal insulator.
Whereas in the quantum spin hall insulator you see they are crossing only at the edges.
So, there are crossing here and the crossing is being shown nicely here and so there are
of course, one line is shown, but there are two lines correspond to each spin which would
390
be shown here. So, again schematically the entire sample would be insulating while the
edges would be conducting as we have seen, and a similar picture is being shown here
and so these are the red one is. Now, representing a down and the blue one is the up and
so on, so there are these two edges which are each having this two a states one for up and
one for down and propagating in different direction. This has been taken from some
presentation of a Lorentz Molenkamp, who is one of the first person to talk about this
quantum spin hall effect and the same thing is shown here in a 6 terminal geometry say
for graphene.
So, as we were talking about the theoretical prediction spade and this goes by the name a
Kene and Mele. In fact, they have built upon this model on what Holden proposed at
least about 20 years back in 1988 and they have built upon the model and they have
added another degree of freedom which will just see.
And by the way 3 people got Nobel Prize in 2016 and all for these topological
implications in condensed matter physics one of them is called Kosterlitz the other is
called Thouless and the third is Haldane so, for this is in 2016 and for study of topology
in condensed matter physics.
So, let us see what the Kene mele model is, we will simply introduce the model and will
not derive it; however, it can be derived; however, that will be a long exercise now,
which we do not want to get in. So, in the quantum Hall Effect the time reversal
391
symmetry as we have said is broken by a strong perpendicular magnetic field. Now,
Kene and Mele they have shown that the spin orbit action in a single plane of graphene it
leads to a time reversal invariant quantum spin hall effect which has as we know now,
that a bulk energy gap and a pair of spin filtered edge states been filtered means spin
resolved edge states at the boundary. And in the model the perpendicular component of
the spin S Z is conserved that simplifies discussion because now, the model reduces to
two copies and one for each spin of which was introduced by Haldane as I was telling
that it was actually adding another degree of freedom which is this spin degree of
freedom and which in, so this model of Haldane which exhibits integer quantum hall
effect in the absence of a magnetic field.
So, all this work had been motivated by a seminal paper of Haldane in 88 in which it said
that you can also have quantum hall effect without the presence of a external magnetic
field, which otherwise we have been so familiar to. So, they consider a tight binding
model of graphene which generalizes Haldane model and further includes spins with
time reversal invariant spin orbit perturbation. The Hamiltonian is written as, so there is
a tight binding term there is an onsite term, staggered onsite term, there is an intrinsic
spin orbit term, and there is a Rashba term.
So, it consists of, so they wrote down this Hamiltonian and it can be shown rigorously
that this Hamiltonian does not break time reversal invariance. So, it is there they are time
reverse I mean all the terms put together they preserve the time reversal invariance and.
So, these states that you see here they cannot be destroyed by any perturbation which is a
time reversal invariance. You have to break the time reversal invariance in order to the
break this. So, what I mean by time reversal invariance is that as soon as H of t equal to
H of minus t time reversal invariance is preserved, time reversal invariance is let us call
it TRI is preserved and if this is not true then of course, the time reversal invariance is
gone. So, TRI means time reversal in variance.
So, the first two terms are spin independent. First term we have already been exposed to
when we did the tight binding model nearest neighbor hopping them on a honeycomb
lattice which is a structure of a triangular lattice with two carbon atoms per unit cell this
you already know. The term yields a Dirac semi metal as is the case for graphene and
then the second term is the sub lattice potential and which has the asymmetry between
the A and B sub lattices with xi i equal to plus minus 1. So, what it means is that for the i
392
to be on the A sublattice xi i equal to plus 1 and for i to be on the B sublattice it will be a
minus 1. So, this is your A sublattice here and your B sublattice here. So, these will take
this thing.
So, this term creates a gap in the spectrum. So, what was told is that this is the Dirac
spectrum. So, close to one of the points, one of the Dirac point you see that this is linear.
So, this is like this at the. So, this is your Fermi level and this is linear; however, this
term splits that. So, if there is a gap that is created, so there are you know gaps that is that
are created.
So, the third term is the mirror symmetric spin orbit interaction which involves spin
dependence second neighbor hopping with the complicated forms for nu i j not to
complicate I will show what it is which also could be plus or minus 1 depending upon
this d 1 and d 2 vectors which are unit vectors going from site i to site j, but site j is not
the nearest neighbor is the next nearest neighbor term sigma Z is of course, the Pauli
matrix. And this term is the most crucial term for getting a non trivial topological state
which we are always aiming at.
The fourth term is the nearest neighbor Rashba term which violates the Z going to minus
Z mirror symmetry and the familiar component for our discussion.
393
So, the third term is important. So, this is that intrinsic spin orbit coupling and. So, there
are A and B sub lattices being shown is shown with A filled circle and B is shown with
an empty circle. So, as long as you have a clockwise hopping from the neighbor to from
from one side to its next year a neighbor.
So, from this to this, it is a clockwise hopping. So, all these things correspond to in
νij =+1 and when you take the anti clockwise hopping. So, then νij =−1 . So, this
is an important term that is this condition is very important for having the time reversal
in variance and the gap is surely to open up in the bulk but they will be still edge states
and this is what is shown in this that now, you see that there is gap that is created in the
bulk because of all these other terms only tight binding would have given you this kind
of dispersion with no gap.
394
Now, there are these are spin result spectrum. So, there are these red lines and green lines
correspond to up and down and now, the two spectra has separated spectra have
separated because of the presence of spin orbit coupling. So, this gap is there so without
these greens lines joining the conduction band and the valence band we would have had
a normal ordinary insulator or a band insulator.
However; because of this we have a topological insulator and exactly for some lambda R
and lambda SO. So, lambda R is for the Rashba spin orbit coupling and lambda SO is for
the intrinsic spin orbit coupling. And only we see the quantum spin hall effect the QSH
only in this regime and if you go beyond this that is lambda R by lambda SO, if it
exceeds these value that is within this you know the kind of structure, that a circular
structure that you are seeing here you will have a band insulator or an ordinary insulator.
So, this is a topological insulator and that is a an ordinary insulator.
395
And this is the comparison between the quantum hall state and the quantum spin hall
states. So, in a quantum hall state this is shown particularly you know with an impurity.
So, there is no phase space for back scattering. So, an impurity will not do anything to a
quantum hall state and that is why its robust to presence of impurities. And the same
thing I just you know the dimension just goes up to from rather 1 to 2 or 2 to 4 that we
have these spin full conditions coming by where there will be helical edge states that are
going to propagate on two edges of the sample. So, there are only one state at the edge
there are while there are two states in quantum spin hall. So, quantum hall has 1 per H
and quantum spin hall at 2 per H.
So, we will stop this discussion this is very technical and it will require a lot of time and
effort to completely understand. Even if one had not seen experimentally a quantum spin
hall phase in graphene, there are other material strain semiconductors where a spin hall
the quantum spin hall phase has been observed which depends on the band inversion
properties and we do not sort of discuss it here. But that is in other thing that has been
done by 3 people every Huges and Junk where experimentally that is a theoretical
prediction of seeing quantum spin hall effect and Molenkamp as said earlier. Molenkamp
has actually seen that experimentally in strain semiconductors which are given by the Hg
Te Cd Te which are mercury telluride, and cadmium telluride, sandwich structures is a
quantum well structures where a quantum spin hall effect is observed experimentally.
396
And also there is a theoretical prediction which has finally, you know been realized in
actual experiments.
Thank you.
397
Prof. Saurabh Basu
Lecture – 25
Bose Einstein Condensation
We shall study Bose Einstein Condensation. This is particularly relevant in the context of
125th birth anniversary of Essen Bose the Indian scientist, whose been associated with
this Bose Einstein condensation. We will say a few things about him as we go along. So,
this is a special topic and it is distinguished from the earlier topics where we have talked
about interacting systems mostly. However, this happens in a this condensation happens
in a system of ideal bosons which means that without any interaction.
So, this is the picture of S. N. Bose and of course, Einstein here. So, this is Albert
Einstein and so we both of them are a responsible for these you know discovery of this
phenomena and will let us learn about them.
398
So, what happened or rather how S N Bose got interested in these things? So, what
happened is that their discussion on the emission of electromagnetic radiation as a
function of temperature. You must have seen that when a piece of iron or a metal is
heated it initially becomes red hot and then it becomes white hot and so on. So, the
spectrum of the incident radiation changes its frequency or wavelength from one to
another and experimentally it is found that the spectrum the incident radiation intensity
of the radiation versus this wavelength. So, this is the wavelength if you cannot see it
these are on small phones and this is the intensity. So, that looks like this. However,
there is a non monotonicity at a given value of the wavelength.
And this was discovered experimentally and the classical existing classical theories
predicted that either it is like this or it is like this which are according to the Wien’s
displacement law or and the Rayleigh’s law. However this non monotonicity nobody got
theoretically. So, Bose understood that there are a rather in fact, Planck proposed with
just conviction and knowing sort of theoretical backing; however, Bose said that there
has to be a new statistics for the photons or the incident or the emitted radiation and this
laid the Bose Einstein statistics to be proposed which he did in consultation with Einstein
and with help from Einstein.
So, it said that the emitted radiation has an energy dependence which goes as h nu or in
quantum of h nu and this was a birth of quantum mechanics in some sense. However, the
399
statistics governing this photons are the statistics is given by the Bose Einstein
distribution
So, let us look at how the condensation phenomena come into the picture. So, say that
there are indistinguishable particles. So, bosons are indistinguishable particles. So,
bosons photons are bosons phonons are Bosons. So, they are indistinguishable and let us
see that how a simple counting procedure can give rise to a condensation like
phenomena.
So, let us take 2 boxes and 2 balls marked by as A and B here and let us consider them as
classical particle or Maxwell Boltzmann particles, so the particles over Maxwell
Boltzmann statistics which means that they are distinguishable. So, if they are
Boltzmann particles then we can have A to be in 1. So, the first one refers to the particle
nomenclature or the name of the particle and one is the corresponding to the box index.
So, A could be in 1, B could be in 1 as well because there is no restriction on the number
of particles to be occupying any quantum state or it could be that A could be in 2, B
could be in 2 or it could be that A could be in 1 and B could be in 2 or it could be that A
could be in 2 and B could be in 1.
So, there are 4 possibilities and if you look at these 4 possibilities there are these 2
possibilities are that they are together. So, they are bunched up in the same box. So, there
are 2 out of 4 is the possibility of them being together for a classical particle or a set of
400
classical particles which are represented by the Maxwell Boltzmann statistics which
means they are distinguishable.
Now, coming to the Bose particles which are indistinguishable we could have there is no
now, no difference between A and B it is just for our reference that we have written as A
and B, but they are just both of them to be say A. So, both of them to be in box 1 is one
possibility because the bosons do not have that restriction of occupying the same
quantum state as the fermions. Fermions obey exclusion principle which we have said a
number of times, during the course of this particular advanced condensed matter physics
and then both of them could be in the second box remember these nomenclatures are just
hypothetical in the sense for our own convenience. They are both A’s. So, both of them
are in 1, both of them are in 2 or both 1 is in 1 the other is in 2 or the reverse happens.
So, now if you look at the bunching probability then you will see that 2 out of 3 are
bunched in the same quantum state. And so this is the crux of Bose Einstein
condensation that the statistics says that if they number of particles that is a large number
of particles can actually occupy one given quantum state then they will bunch up or
rather they will occupy crowd in that quantum state and that quantum state at very low
temperatures would be the ground state of the system. So, a macroscopic accumulation
of particles in the ground state is what is known as the Bose Einstein condensation.
So, this is what is written that A1 B1 and A2 B2 get two-third of the weight as opposed
to half. So, this is equal to half which is 50 percent and this is two-third is equal to 67
percent. So, there is a larger possibility and this could actually happen in a many particle
sector for us to understand Bose Einstein condensation of course, we are going to go to
details into that, but this is a very simple idea summarizing that how indistinguishability
of particles can lead to a condensation phenomena.
401
So, the more weightage to situations where particles occupy the same energy state is the
central idea or the central focus basic idea of a BEC. And so, Einstein upon receiving a
note from S N Bose he understood that there is a lot of merit in Bose’s derivation of the
Planck’s formula, and he says that Bose’s derivation of the Planck’s formula appears to
me an important step forward and the method used here gives also the quantum theory of
an ideal gas as I shall show somewhere.
402
So, pictorially let us see what happens. So, there are a, there is a ground state of a system
which corresponds to E = 0 and these are the spectrum of the quantum of the excited
states and we have just shown them almost like a continuous spectrum because in an
infinite system or a thermodynamic system they could be infinitesimally close to each
other. The reason that we have shown the ground state to be separate from the excited
state is something that I am going to discuss later because of the density of states going
as a particular fashion as a function of E, which is E to the power half or square root of E
this is getting 0 weight which it should not. So, a priori you cannot assign a 0 way to any
given quantum state, so we are going to consider this ground state separately as
compared to the other excited state.
Now, see say there are N particles N bosons in a given system where N 0 would occupy
the ground state just to let you know that the ground state has in principle infinite
occupancy. We will show that at lower temperature. And N ex is the number of particles
occupying all the excited states put together.
So, now the number of bosons N can actually be tuned using temperature. It is a function
of temperature which comes out of the Bose distribution function. Now, N 0 is as I said
at very low temperatures is practically infinite. Now, incidentally depending on certain
conditions and I will also speak about those conditions N ex may be finite in certain
circumstances and in fact, to get any x to be filed rather to get N naught to be infinity
you need to go to very low temperature and precisely lowering that temperature lowering
t to the desire value to 71 years I will tell you why 71 years because 1924 was the
theoretical prediction of BC by Boson Einstein.
And it was 1995 as written here experimental realization of BC came in sub system of
rubidium atoms and for the first time this actually show a BEC. So, a very large amount
of development has taken place since then and the field of what is called as the ultra cold
atoms has developed enormously. And, so if the one of the example is that, so a beam of
atoms in the same quantum state is called as an atom laser and this is possible one has
actually made atom laser and all these are based on the condensation phenomena that
was put forward by Bose and Einstein.
403
So, let us see how to make BEC. So, you have to heat the atoms say to vedyam atoms at
about 600 Kelvin to make a gas of them it is somewhat counter intuitive because we are
talking about cooling and then first you have to heat so that we have a very low density
of atoms. So, that it forms something like a gas and then confine them in a beam. So, this
is a laser beam, so when you, they say a system of atoms. So, if you confine them in such
you know and also from top and bottom.
So, then these atom loses its motion and when it loses its motion it cools down and the
kinetic energy becomes less, and the temperature corresponding temperature becomes
less and this is what is meant by confining them in a beam. Use a magnetic trap will
speak a little about them as well you can also talk about an optical trap this is something
that these are part of laser cooling we talked a little about laser cooling. And of course,
there evaporate cooling which is opening the you know or rather reducing the depth of
the trap such that more energetic atoms would escape.
404
And then finally there were achievements that 1998 which we are not too keen on ah;
however, this is helium 4 was achieved the cooling effects are not very important the
cooling effects that we are going to talk about because it happens at 4 Kelvin or say a
few Kelvin not 4 Kelvin, but a few Kelvin about 3 .3 Kelvin, 3.13 Kelvin. Then in 1995
as we said that Eric Cornell and Curl Wieman the people here they created the BEC in
rubidium 87 atoms and then Wolfgang Ketterle in he created BEC in sodium 23 in 1995
just 2 months later after this discovery of rubidium. So, all 3 of them got Nobel Prize in
2001, and these are Kapitsa Allen and Misener these are the people who have liquefied
helium in 2000, in 1938.
405
So, this is that old picture that which is connecting the 1925 story when the theory was
proposed and 1995 there are experimental evidences. Will try to say a little on this, but
just a priori this is still not a condensate here and it starts becoming a condensate and this
is a condensate here. So, this is not a condenser and this starts becoming a condensate
and so on.
So, what happens is that. So, they have cooled the atoms and then they release the trap in
which the atoms were held and when the trap is released the atoms fly away and then a
high speed camera device images them and they are the image is converted into Fourier
transformed into k space and there is a case phase picture. So, a peak here corresponds to
a macroscopic accumulation of particles corresponding to a given k value which is
momentum value of the particles and this case the momentum is equal to k equal to 0.
So, there is a macroscopic accumulation of particles in a given quantum state which is
what BEC is all about.
406
So, just to summarize and the developments that had you know come across in various
fields or rather in this trapped ultra cold atoms. So, 87 rubidium was cooled in 1995 and
it was in Jila in by Eric Cornell. Then it was lithium in July 1995 in Rice University by
Randy Hulet, then sodium in September 1995 by Ketterle, hydrogen in MIT and then of
course, helium again in 2001 by the group of Allen aspect.
So, now let us start discussing something more you know concrete. So, that you
understand the concepts of a BEC. So, let us just talk about the basic features of BEC.
So, that things are put in perspective.
407
We have said a few things, but will still repeat it. So, the condensation occurs in the
momentum space. So, it is a K equal to 0 is macroscopically occupied. So, this is an
important key concept rather. So, it is not a real space accumulation of particles. So, it is
K equal to 0 is occupied, and it is very important that the particles are non-interacting
and we are talking about an ideal Bose gas. In fact, whether the liquefaction of helium is
a BEC that is the question that one had to understand, because helium is still an
interacting fluid and possibly.
So, there are confusion whether the lambda transition in helium actually falls into the
class of a BEC and you will not get into that. Rather let us talk about a few things such as
the cooling techniques. And to begin with let us talk about two, one is called as a
magnetic evaporative cooling and the second is let us call it as a laser cooling ok.
So, we will briefly describe both of them. So, this is the magnetic evaporative cooling.
408
So, this you must be knowing that a system of two coils where the currents are actually
moving in the same direction produces. So, this is there are two coils where there is the
same current is moving in same direction and the distance between the radii of these two
coils is say a or rather this is say 2 a. So, this is 0 to, so this is minus a and this is at a.
And then what happens is that then the magnetic field variation because there are current
flowing in the coil there will be generation of magnetic field. Now, because there are 2
coils, the magnetic field or the magnetic induction would be a superposition of the effect
from both the coils. And this would be like symmetrically it will, so these fall off as one
over r whereas, it is fairly constant here and this is a shear minus a to plus a.
So, this is the situation for a Helmholtz coil or Helmholtz double coil as it is said. Now,
what person in MIT called David Pritchard did he produced a similar coil with again to
such coils; however now, the currents are flowing in different directions in both the coils.
So, one with respect to the other and now you will have a magnetic field variation to be
having a minima, as opposed to a maxima and which you can do a simple calculation to
see that this is of course, symmetric about 0 probably did not draw it as symmetric what
it is.
So, this if you can load the atoms here then this acts like a trap for it it is a magnetic trap
for it. So, any atom which has more energy will actually escape. So, atoms will escape
and leaving only say something like 10 to the power 9 number of atoms in the system
409
which is still you know it is far away from BEC, because BEC requires something
around 10 to the power 6 number of atoms. But at least this does one step and it makes a
lot of atoms which have which are energetically more I mean they have higher energy
they energetically more favorable atoms would escape and leaving behind the slower
ones. So, we have a gas of cool atoms and then so this is a one technique, the other one
has we say that the laser cooling.
So, this is a very nice technique to slow down atoms and hence a strip their kinetic
energies and thereby reducing the temperature. So, what happens is that when an
electromagnetic radiation of frequency omega falls on an atom, atom whose energy
levels are given by ω0 or ℏ ω 0 then this I will do without proof, then the probability
1
amplitude for absorption is proportional to ~ where gamma is the line
( ω −ω0 )+ i Γ
2
width line width of the radiation this is a natural line width.
So, this happens when the atom is at rest; but when the atom is moving then the Doppler
effect comes into play and if we write this. So, what happens when a Doppler effect
comes into place, that when the atom moves towards light source, towards light source
which means electromagnetic radiation, so the omega increases the effective omega
increases. And this is called as blue shift and when the reverse happens that is atoms
moves away from light source omega decreases and this is called as red shift.
410
So, this shift in frequency is say equal to Δ ω =−⃗v . ⃗k where v is the velocity of the
atom and k is the momentum. So, if you need to slow down atoms then it is beneficial for
the atoms to be moving towards the light source or the incident radiation. So, if you can
make the atoms move towards the light source then they will slow down and slowing
down means the kinetic energy becomes less and when the kinetic energy becomes less
the equivalent temperature because the energy is always expressible in terms of
temperature with the Boltzmann relation they equal to k T. So, the T decreases.
So, what happens is that the that the denominator here. So, this denominator it takes a
k )+i Γ ]
v .⃗
form of [( ω −ω0−⃗ and so the quantity so Δ=−ω + ω0 rather omega
2
naught minus omega 0 is called as the detuning parameter.
So, the Doppler effect yields v .⃗

Δ ' =Δ+ ⃗ k and so basically now, an atom moving,
moving along to counter propagating laser beams this is what I had shown in one of the
earlier slides that the atoms are actually held between counter propagating laser beams.
Counter propagating means a propagating in opposite directions with detuning. Then for
both of them for both Δ ' 1=Δ+| ⃗v || ⃗k | . So, this is a 0 angle between them. So, it is a v
k cos theta, but cos theta is equal to 0 because they are moving towards each other and
the atom is moving along the line. So, there is no angle that it is making and
v ||⃗
Δ ' 2=Δ−| ⃗ k| .
411
So, it is clear that the some more energy is actually absorbed from one of the beams as
corresponding to the other. So, we are talking about the 2 laser beams. So, one of them is
absorbing more energy from the atom as compared to the other, but this further means
2 | ⃗v | | ⃗
k|
that the atomic momentum is reduced by , ok. So, this forces exerted by the
c
beams on the atoms are not balanced. So, there is a resulting in a net force opposite to v
and thereby reducing the kinetic energy and so, reduces kinetic energy and causing
cooling.
Now, this ends the story on cooling from our side, but you understand that all these
engineering of having cold atoms possible in 70 years and they are the best minds that
were working in the subject it is still took a very long time, for getting them down to a
temperature which is extremely small and B E C really happens in sort of micro Kelvin
in the micro Kelvin range. However, well do a calculation and well show that actually
for the atoms such as helium the condensation or rather the condensation like
temperature is just about a few Kelvin maybe 2 to 3 Kelvin.
However, understand that actually the rubidium atoms required a temperature to be of

the order of a micro Kelvin and so let me write that the temperature is a of the order of
finally, and there is only a part of the cooling techniques that I have talked about there
are the sisyphus cooling, and sympathetic cooling and so on, and so this temperature
comes to about 10 to the power minus 6 Kelvin and the number of atoms or number of
particles that should be left is about 10 to the power 6.
So, this is the condition for BEC to be formed in lab and. However, as I said that will
not, will not discuss the exact magnitude of the temperature, but well do an order of
magnitude calculation.
Now, since we are done with the cooling part let us understand that why is that we are
talking about Bosonic atoms when the atoms are actually new neutral or rather the
charge neutral and if we are talking about the electronic the spin then that electronic spin
is always half and these alkali atoms rubidium being an alkali atom. So, this that that
should correspond to electrons having spin half and that cannot be put in the same class
as Bosonic atoms. So, let us understand why they are called as Bosonic atoms.
412
We are just talking about alkali atoms because more it started with alkali atoms as I have
shown you rubidium, sodium etc potassium lithium again. So, what happens is that even
if the electrons have the spin half it is the nuclear spin that comes into picture but
because the nuclear magnetic moment, nuclear magnetic moment μ N < μ e which is the
nuclear magnetic moment for electrons by at least of 4 orders of magnitude. And thus to
have nuclear spins to play an important role we really need very low temperature.
So, let us say that F is the atomic spin ,I is the nuclear spin, and J is the electronic spin.
So, F =⃗I + ⃗
⃗ J and so it has a range | ⃗I + ⃗J | to | ⃗I − ⃗
J| equal to which is a electronics
spin equal to half. So, we have F and for rubidium ok, let us for all the alkali atoms such
as rubidium, sodium, lithium etc we have I which is the nuclear spin is 3 / 2. So, the
atomic spin F = to 1 or 2, so they are Bosonic atoms with integral spins. And as I said for
the nuclear spin to play a role the temperature has to be low.
Now, let us start doing some calculations in order to understand the condensation
phenomena in an ideal Bose gas. So, let us start with the ideal Bose gas.
413
And the discussion is you can follow a statistical mechanics book very good statistical
mechanics book by Pathria which gives a very good description of this phenomenon
before we proceed let us write down the grand canonical partition function. So, the word
grand canonical grand canonical partition function is that where we also allow in a in
addition to exchange of energy among the particles we also allow for exchange of
particles between the system and the bath.
So, it is the canonical is about exchanging energies between the system and the bath and
this is also in addition to that allowing the number of particles. So, the number of
particles is not constant. So, a grand canonical partition function as you know that the
partition function is written with Z and this Z comes from a German word probably it is
called as; there could be a problem with the spelling, but it means zustandssumme means
that it is a sum over states; And this G, subscript G that corresponds to the grand
canonical.
∞
− βϵ n
So, this is equal to Z G = ∑ exp( βμ N ) ∑ e i i
let us you know you can write n i and
N =0 ni
so on.
∞
So, this is the canonical partition function and. So, Z G = ∑ exp( βμ N )Z C is pretty
N =0
much the formula for the grand canonical partition function. And you can follow pathria
414
to see that the grand canonical partition function the log of that which is related to the
free energy a minus K T of that is equal to F which is equal to a PV. So, this is equal to
V
ln Z g=P =−∑ ln (1−Z f e−βϵ ) i
KBT i
So, let us make this notations all clear. So, P is equal to pressure v 2 be volume t is of
course, temperature, Z f is called as the fugacities which is equal to exponential beta mu
beta is 1/KT and mu is the. So, let us write it here mu is chemical potential and beta
equal to 1/KT. And, so this is by and large the expression for this. So, this can actually
be the equation of state for an ideal boost gas because we are writing PV/KT which you
know for a classical ideal gas PV/KT is equal to some r or it is n r or something like that.
So, it is a constant whereas, in an ideal boost gas at quantum gas which follows an
indistinguishability constraint is given by this. And similarly the number of particle
1
which is equal to N=∑ < ni> =∑ − βϵ i y. So, for Boson is minus 1 and for
i i zf e −1
Fermions it is equal to plus 1 epsilon i’s are the single particle states. So, what I mean by
signal particle states is that if there are free particles. So, there they will go as the in k
space they will go as h cross square k square over 2 m.
Now, before we proceed what we have to do is that we have to convert these or rather
these are the working equations for any system whether you are talking about
condensation or you are talking about studying other thermodynamic or statistical
mechanics mechanical properties you have to start with these equations for both
fermions and boson.
Now, we need in order to calculate them we need to, calculate them analytically we need
to convert them into integrals. And you know when a sum is converted to an integral you
need the density of states you need the number of states in the energy range E and E plus
B E and that is an important quantity in condensed matter physics as I might have
already told a number of times because this is what creates a difference between a 2
dimensional the properties of a 2 dimensional system with that of a 3 dimension because
the density of states go down. In fact, the nano the whole branch of nanoscience and
nanotechnology when we talk about quantum wires which means that we are confining
415
electrons or the charge carrier in one dimension while in quantum dots we do it in 0
dimension and in a 2 dimensional electron gas this is in 2 dimension and so on.
So, we need the density of states in order to calculate or rather convert the sums into
integrals. Now, this is I will not go into that, but it is a very simple calculation first
course of solid state physics would do it that the density of states in short called as DOS.
1
( ) ℏ2 k 2
So, that goes as ϵ 2 for a ϵ= this is in 3D it goes as constant also let us write
2m
−1
( )
2
it as a ϵ(0) which is also which is a constant in 2D and it goes as ϵ in 1D ok. So,
for this dispersion a free particle dispersion.
So, if you see this that the density of say, so we are talking about a 3 dimensional system
with k square dispersion. So, we are talking about a d-dimensional system, where d = 3.
So, d = 3 for us and we have a dispersion which is s going as, dispersion going as k s ,
so s is equal to also, s is equal to 2 for us. In fact, whether a BEC will occur or not will
crucially depend upon this value d and s and a you can actually convert it into a single
parameter. So, either you call it as by d or d by s. So, depending on certain values of s by
d or d / s it is Bose Einstein condensation is possible.
416
So, we convert using this density of states, we convert the first equation let us call them
as first equation and second equation and let us write 1 gives. So, 1 means equation 1 and
(∞) 1
P −2 π dϵ 1
so this is equal to minus = 3 (2 m)(3/ 2) ∫ ϵ2 −1 βϵ−1 − ln (1−Z f )
kB T h 0 Zf e V
Now, I should have actually done it for all epsilons, but however, you see if we have
epsilon to the power half and for the ground state which is epsilon equal to 0, you would
attach 0 weight. This is what I was talking about earlier that you cannot have a 0 weight
assigned to a given state then that state is completely unimportant for our computation.
So, we will have to take that state out and write it for that state will write this thing as
rather it is a minus sign it is a minus 1 by V log of 1 minus Z f. So, this is that
contribution for the epsilon equal to 0 and this is for epsilon not equal to 0.
Now, ideally in this integral we should separate out the epsilon equal to 0, but since this
is an integral just having one point less does not make any difference in this integral. So,
we can still put it as 0 to infinity. So, so the integral becomes 0 to infinity and this
integrand and then this is the epsilon equal to 0 component of that equation number 1.
So, now let us call this as equation number 3, and 2 gives in the same manner
(∞) 1
N 2π dϵ Z
= 3 (2 m) ∫ ϵ 2 −1 βϵ −1 + f
(3 / 2)
let us call this as equation 4.
V h 0 Zf e 1−Z f
Now, if you look at these 2 equations we have clearly as separated them into two terms,
one corresponding to epsilon equal to 0 the contribution to the pressure corresponding to
epsilon equal to 0 and the contribution to the density of particles or the number of
particles to be f for epsilon equal to 0 and for epsilon not equal to 0. This is in one of my
slides that I have shown the ground state energy to be separate as compared to the
excited state energies.
So, now, let us see the second terms of both of both 3 and 4.
417
So, see what the second terms are. Let us write that second term as this its equal to
1 ZF N0
f= = . You understand N naught, N naught is the occupancy or the
V 1−Z f V
eβμ 1
number of particles that the ground state can hold and N 0= βμ = −βμ . So, this
1−e e −1
is the N 0 by V that is the second term in equation 4. So, this is 4.
Now, when your Zf << 1, now Z f is actually between 0 and 1 ok so, it can take a
maximum value 1 and can take a minimum value 0. So, if Z f is close to 0 that is a
smaller much smaller than 1 then we have, this corresponds to very large temperature.
And if it talks about very large temperature then you have the number of particles is very
low because this f is equal to f is a very small and so N 0 is very small and of course, that
corresponds to a no Bose Einstein condensation because you have a classical physics that
is taking over. So, thermal effects should you know drive all the particles away.
Now, you try to understand that as you reduce the temperature N 0 increases. So, at very
low temperature N 0 is very high and so as Z f goes to 0. So, that is, so μ is actually a
chemical potential which is negative. So, if Z f goes to 0 then N0/ V is significantly
large. In fact, it is so large that your N = Nex + N0 somehow if you can show that your N
ex is equal to a finite number then since N0 is infinitely large. So, if your N is infinite or
rather very large then all the particles will go to the ground state because it is has infinite
418
occupancy whereas, a very smaller number of particles would actually go to the excited
states because it has limited occupancy.
Now, if you look at 3 then you will see that this is equal to the 0 epsilon equal to 0
eβμ 1
contribution is N 0= βμ = −βμ and this tells that for a Z f << 1 this is equal to
1−e e −1
negligible. So, this minus 0 over V log of this thing is very small that of course, we know
that the pressure due to the all the particles that would go to the ground state is very
small because there is no almost no particles at large temperatures this is the limit for
large temperature.
However if you go to the other limit that is Z f going to 1, I am sorry the Z f should go to
one here not 0. So, Z f is 1. So, if Z f goes to 1, so this is for Z f going to 1. So, Z f going
1
to 1; however, this thing would take a form which looks like ln N 0 . Now, N naught
V
could be large, but log of that would be you know still small. So, the second term in
equation 3 can still be neglected even if at low temperature that is Z f going to 1, of the
fugacities going to 1; however, that cannot be neglected in equation 4. This is the main
central message of this discussion; that even the pressure contribution from all the
particles in the ground state of the system could be infinitesimally small; however, the
number of the number density is significant.
So, now well write this. So, basically our 3 becomes its
3
∞
P −2 π
= 3 (2 m k B T ) ∫ x(1/ 2) ln (1−Z f e− x ) dx where
2
x=βϵ , that is the thing.
kB T h 0
419
And so beta has come out and the 4 a gives let us call this as maybe 5 and 4 gives
2 m K B T ( 32 ) 1
N −N 0 N ex dx
= =2 π ( 3
) ∫ x 2 −1 x
V V h z f e −1
Now, these are called as the Bose Einstein integral and a g n of z. So, this these are Bose
Einstein integral lifts in short called as BE integral and this BE integral takes a form its
1 ∞ (n−1) dx
equal to gn (Z f )= ∫ x −1 x
. So, these integrals look very similar to that
Γ( N ) 0 Z f e −1
excepting for its N equal to say 3 by 2 and so on. So, and this is equal to it can be shown
this I leave it to you it is a matter of you know doing a partial integral integration, sorry
integration by parts of this integral and a express it in the form of this and finally, what
one gets is, so let us call this a 6.
P 1
So, 5 becomes equal to = g5 /2 (Z f ) . We are less interested in this formula
KBT λ3
though it is important nevertheless; however, we are interested in this formula the one
that we are going to write later. So, let us call this as 7 and this is equal to
N −N 0 N ex 1
= = 3 g3 / 2( Z f ) . That is the second equation that is equation 6. So, 6 yield
V V λ
this.
420
So, our the number of particles which is what we wanted in the excited states is given by
h
some quantity which is this where l λ = it is called as a thermal de Broglie
√2 m k B t
wavelength. So, this is equation a thermal de Broglie wavelength.
Now, we are almost done, we have obtained an equation for the number of particles in
the excited states. If this quantity is finite is not infinitely large which would depend
upon certain criteria then we are done we would again leave this thing and show that
l 2 3
∞
Zf Zf Zf
your, gN ( Z f )=∑ n =z f + n + n +...... . to the power n and so on.
l=1 l 2 3
Now, for small z that is that if much smaller than one the classical limit you can be
satisfied with Z f, ok. So, if you put that equal to a Z f then of course, that becomes will
depend on the excited state occupancy is you know is goes as exponential beta mu.
However, at low temperature that is when Z f goes to in the limit goes to 1, you have to
take all of these terms into consideration and cannot stop at a finite terms, but fortunately
this is equal to for g 3 (z f =1)= ξ (3/2)=2.612

( )
2
So, this is a finite value this is what we were hinting at time and again that the excited
state occupancies which are coming out from this equation number 8 has an the excited
421
state occupancy has a finite value. So, if there are more number of particles they will all
go to the ground state. If there are a macroscopically large number of particles they
would all go to the ground state and this is what the condensation. So,
3
2 (2 m π k B )(3 /2) 3
N >VT 3
ξ ( ) and the xi is this is not the way to write it is like this, it is
h 2
called as the Riemann zeta function.
So, the Bose Einstein integral is related to the Riemann zeta function when you take this
entire sum. The entire sum has a closed form which is called as Riemann zeta function
and for an argument equal to 3 by 2 it has a value equal to 2.612. So, if N is greater than,
this N is the total number of particles then BEC occurs and for the BEC to occur the
3 3/ 2
N (2 m π k B )
critical condition is that your =T 2 2.612
V h3
h2 (2/3 )
N
So, if you put everything there then what we get is that a T c= ( ) .
2 π m k B V ξ (3 /2)
So, this is the expression for T C which means the temperature if its lowered below this
then all the particles will go to the ground state and the excited state because the excited
states have a a finite occupancy they will avoid the excited states and go to the ground
state because the temperature is also very slow.
422
Now, this we will just do it for helium 4 which as we said that is not a a prototype case
for Bose Einstein condensation, but still if you take this values m=6.65×10−24 gm .
And so the density or the inverse density which is called as a specific volume
V
v= =27.6 cm3 /mole then if you put all these things T C comes out to be equal to
N
3.13 Kelvin and 13 Kelvin and this is close to the observed lambda point transition of
helium liquid helium, liquid helium which is equal to 2.17 Kelvin.
So, there was a initially misconception that the liquification of helium is actually or
rather they heal the lambda point transition in helium is actually a BEC transition. We
will not elaborate on that, but what we have got is a condition for the Bose Einstein
condensation also we have explained various steps that are associated with the cooling
process and finally, when the cooling happens the whole atom the 10 to the power 6
atoms number of atoms.
They form a structure which has very low temperature which is like 10 to the power
minus 6 to minus 7 Kelvin which is like a less than a nano, I mean about micro Kelvin
temperature which is the probably the coldest temperature in the universe and they are
imaged as I said that releasing the traps and letting them fly apart. So, when they fly they
are imaged and they are finally, Fourier transformed into k space to see there is a
macroscopic accumulation of particles in the k space.
So, this is a k space phenomena. It is a real momentum space accumulation of particles B

the BEC is a a a example of that and so this the credit goes to Bose and Einstein who
have proposed this in 1924, nearly a 100 years from now, 100 years earlier. However, as
I said the realization had take place a very large number of years.
423
THIS BOOK IS
NOT FOR SALE
NOR COMMERCIAL USE
(044) 2257 5905/08 nptel.ac.in swayam.gov.in

115103102

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

115103102

Uploaded by

Copyright:

Available Formats

INDEX

relation which is i =EnCn

equation which is i =EnCn

equation 2 and then we shall have the * un(x)

So, if you put 5 in 4, then * un(x)

. So, if we know , instead of solving the second order differential equation

proportional to psi of x 0. Now here my definition of the propagator un*

the modern notation it is un( ) │un(x)>

(Refer Slide Time: 00:41)

. So, let us call this as equation 9 and this as equation 10.

(Refer Slide Time: 10:59)

delta function. So, which is d/dt or a if you write it more correctly.

(Refer Slide Time: 17:14)

(Refer Slide Time: 21:50)

I am skipping one step which should fill it up and this is

Where actually eta goes to 0 in the limiting case in the limit.

(Refer Slide Time: 26:46)

now that would .

, it should be also divided by h cross let us call it then where

And, so, it is in the lower half plane.

So, this In = 2πi *

and which is nothing, but GR un In

(Refer Slide Time: 01:25)

(Refer Slide Time: 04:35)

dk = and following that I can write down :

(Refer Slide Time: 08:13)

(Refer Slide Time: 11:00)

(Refer Slide Time: 22:11)

(Refer Slide Time: 26:05)

(Refer Slide Time: 27:39)

So, we named this as equation 1.

(Refer Slide Time: 00:29)

And the, this quantity which is P to the power 1 minus S g n K divided by 2; it

(Refer Slide Time: 06:59)

However, in the position representation there is an enormous summation involved here in

(Refer Slide Time: 14:16)

(Refer Slide Time: 18:43)

And we have relationships similar relationships; commutation relationships for the

(Refer Slide Time: 01:00)

(Refer Slide Time: 02:58)

And where I can write down a quantum state as .

(Refer Slide Time: 06:00)

(Refer Slide Time: 09:43)

So, now I can write down a spin operator as simply S=

And similarly, a local density operator can be written as (r) (r)

V r1.,r2……rN.> = V(rn,rm) r1.,r2……rN.> = 1/2 V(rn,rm) r1.,r2……rN.>

V=1/2 (r) (r’) V(r,r’) (r) (r’)

(Refer Slide Time: 30:32)

(Refer Slide Time: 35:29)

V=1/2 (r) (r’) (r) (r’)

(Refer Slide Time: 00:49)

(Refer Slide Time: 05:45)

(Refer Slide Time: 12:19)

(Refer Slide Time: 16:44)

(Refer Slide Time: 20:28)

(Refer Slide Time: 23:19)

(Refer Slide Time: 26:39)

called as a lattice constant.

(Refer Slide Time: 29:41)

(Refer Slide Time: 32:48)

∈𝑘 = ±𝑡|𝑓(𝑘)| =±√(3 + 2 cos(𝑘𝑥 𝑎) + 2 cos(𝑘𝑥 − 𝑘𝑦 ) 𝑎 + 2cos(𝑘𝑦 𝑎))

(Refer Slide Time: 35:51)

have spin σi and σi ′ and after the scatter.

𝑈𝑖𝑖′ 𝑗𝑗′ = ½ ∫ 𝑑𝑑 𝑟 ∫ 𝑑d r’ 𝜓∗𝑅𝑖′ 𝜓∗𝑅𝑖 V(r-r’) 𝜓𝑅𝑗′ 𝜓𝑅𝑗