PHYS 625, Spring 2015

Homework 1 Solution
(Dated: February 13th, 2015)

PROBLEM 1

Figure 1

(a) The two lattice vectors as shown in the picture above are a1 = a(1, 0) and a2 = b(0, 1). Let the total number of
sites be N. The hopping matrix elements being ta and tb in the x and y directions, one can write the Hamiltonian as
X X
H = ta ci,j, ci+1,j, tb ci,j, ci,j+1, + H.C. (1)
i,j, i,j,

Expanding the operators in momentum basis

1 X ikRij
ci,j, = e ck, (2)
N k
1 X ikRij
ci,j, = e ck, (3)
N k

We get the Hamiltonian

ta X X  ikRij   ik0 Ri+1j  t X X
b
 0 
H= e ck, e ck0 , eikRij ck, eik Rij+1 ck0 , + H.C. (4)
N i,j, 0 N i,j, 0
k,k k,k
ta X  0
 t b
X 0

ck, ck0 ,
X X
= ck, ck0 , eikRij eik (Rij +a1 ) eikRij eik (Rij +a2 ) + H.C. (5)
N 0 i,j
N 0 i,j
k,k , k,k ,

Using the identity

1 X i(kk0 )Rij
e = k,k0 (6)
N i,j

we simplify (5) further

0 0
ck, ck0 , k,k0 eik a1 tb ck, ck0 , k,k0 eik a2 + H.C.
X X
H = ta (7)
k,k0 , k,k0 ,

ck, ck, eika1 + eika1 tb

X X
ck, ck, eika2 + eika2



= ta (8)
k, k,

(2ta cos(kx a) 2tb cos(ky b)) ck, ck, (kx , ky )ck, ck,
X X
= (9)
k, k,
 2

Hence, we arrive at the energy dispersion

kx ,ky = 2ta cos(kx a) 2tb cos(ky b) (10)

Setting a = b, for ta = tb , the dispersion relation becomes kx ,ky = 2ta (cos(kx a) + cos(ky a)). For ta  tb , kx ,ky
2ta cos(kx a) The following figure depicts the first Brillouin zones for different ratios of the hopping parameters ta
and tb .

4 200
3 3

3 150

2 2

2 100

1 1
1 50
ky a

ky a
0 0 0 0

-50
-1
-1 -1

-100
-2
-2 -2
-150
-3

-3 -3
-200
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
kx a kx a

(a) (b)

Figure 2: Isoenergetic lines for different ratio of x and y hopping parameters (a) ta = tb = 1 and (b) ta = 100tb

(b) Including the next nearest neighbor interaction as shown by the grey dotted lines in the schematic figure Fig.
1 for cuprates (ta = tb = t0 ), the Hamiltonian in (5) becomes

t0 X X  ikRij   ik0 Ri+1j+1    0 

H= e ck, e ck0 , + eikRij ck, eik Ri+1j1 ck0 , + H.C. (11)
N i,j, 0
k,k
t0 X X 0 0

= ck, ck0 , eikRij eik Ri+1j+1 + eikRij eik Ri+1j1 + H.C. (12)
N
k,k0 , i,j
X  0 0 0 0

0
= t ck, ck0 , k,k0 eik a1 + eik a2 + H.C. (13)
k,k0 ,

2t0 (cos(k a01 ) + cos(k a02 )) ck, ck,

X
= (14)
k,

where as before we have used the relation (6) to sum over the lattice points. The lattice vectors are defined as
a01 a1 + a2 and a02 a1 a2 , such that we obtain the dispersion relation

0 (kx , ky ) = 2t0 (cos(kx a + ky b) + cos(kx a ky b)) = 4t0 cos(kx a) cos(ky b) (15)

PROBLEM 2


(a) Consider the graphene lattice with lattice vectors a1 = a 3/2, 1/2 and a2 = a (0, 1) as shown in Fig. 3(a).
We know that the reciprocal lattice vectors b1 (b11 , b12 ) and b2 (b21 , b22 ) follow the condition

ai bj = 2ij (16)

which gives us

3/2b11 b12 /2 = 2/a; b12 = 0 (17)

3/2b21 b22 /2 = 0; b22 = 2/a (18)
 3

(a) (b)

Figure 3: Graphene lattice in real space and k-space

or
4
b1 = (1, 0) (19)
3a
4
b2 = (1/2, 3/2) (20)
3a

(b) Starting with the Hamiltonian in Problem 3.4

3
ai, bi+, + H.C.
XX
H = t (21)
i, =1

as suggested, we use the momentum basis representation of the operators

1 X ikRi
ai, = e ak, (22)
N k
1 X ikRi+
bi+, = e bk, (23)
N k
(24)

to rewrite the Hamiltonian as

3
t X X X i(kRi k0 Ri+ )
H= e ak, bk0 , + H.C. (25)
N i, 0
k,k =1
3
!
t X
X 0 X 0
= ak, bk0 , ei(kk )Ri eik + H.C. (26)
N 0 i
k,k , =1

P3
Again, using the relation (6) and defining gk =1 eik as defined in problem 2.4, we get

gk ak, bk, + gk bk, ak,

X
H = t (27)
k,

which in matrix representation can be expressed as

  
X 0 gk ak
ak bk


H = t (28)
gk 0 bk
k,
 4

This gives
the
energy eigenvalues

k = t|gk |. Noting
as E that
the vectors i connecting nearest neighbors are
1 = a/ 3, 0 , 2 = a/(2 3), a/2 and 3 = a/(2 3), a/2 we simplify gk as follows
 
|gk | eikx a/ 3 + ei(kx a/(2 3)+ky a/2) + ei(kx a/(2 3)ky a/2)  (29)
 
  
eikx a/ 3 1 + ei( 3kx a/2+ky a/2) + ei( 3kx a/2ky a/2)  (30)
 

 
Hence Ek = t 1 + ei( 3kx a/2+ky a/2) + ei( 3kx a/2ky a/2) , as we set out to prove.
 

(c) Defining 3kx a/2 and = ky a/2,
 2
2
|gk | = 1 + ei(+) + ei()  (31)
 

= 3 + 2 cos( + ) + 2 cos( ) + 2 cos(2) = 3 + 4 cos() cos() + 2 cos(2) (32)

The energy eigenvalue therefore becomes

 1/2
Ek = t 3 + 4 cos( 3kx a/2) cos(ky a/2) + 2 cos(ky a) (33)


(d) Expanding about the point K 2/( 3a) + kx , 2/(3a) + ky
 1/2
Ek = t 3 + 4 cos( + 3akx /2) cos(/3 + aky /2) + 2 cos(2/3 + aky )) (34)
    1/2
= t 3 4 cos( 3akx /2) 1/2 cos(aky /2) 3/2 sin(aky /2) + 2 1/2 cos(aky ) 3/2 sin(aky )
(35)

   1/2
= t 3 4 1 3a2 kx2 /8 1/2(1 a2 ky2 /8) 3ky /4a + 2 (1 a2 ky2 /2)/2 3(aky )/2 (36)
 1/2
= t 3 2 1 + 3ky a 3ky a + 3a2 kx2 /4 + 3a2 ky2 /4 (37)

1/2
= 3ta/2 kx2 + ky2 = 3ta/2k ~kvF (38)

Thus we arrived at the linear dispersion relation with vF 3ta/(2~). For K0 (0, 4/(3a)), we rewrite
 1/2
Ek = t 3 + 4 cos( 3kx a/2) cos(2/3 + ky a/2) + 2 cos(4/3 + ky a) (39)
    1/2
= t 3 + 4 1 3kx2 a2 /8 1/2(1 ky2 a2 /8) 3/2(ky a/2) + 2 1/2(1 ky2 a2 /2) + 3/2(ky a)

(40)
 1/2
= t 3 2 1 + 3ky a 3ky a + 3a2 kx2 /4 + 3a2 ky2 /4 (41)

1/2
= 3ta/2 kx2 + ky2 = 3ta/2k ~kvF (42)

Note: There is a typo in the problem statement, K0 should be (0, 4/(3a)).

PROBLEM 3

Consider the current density in the presence of an electromagnetic field

e X
j(r) = ((pi + e/cA(ri )) (r ri ) + (r ri ) (pi + e/cA(ri ))) (43)
2m i
e X e2
= ((pi (r ri ) + (r ri )pi )) (A(ri )(r ri ) + (r ri )A(ri )) (44)
2m i 2mc
 5

Since A(ri ) commutes with (r ri ), the second term becomes

e2
jD (r) = A(ri )(r ri ) (45)
mc
In the second quantized form

e2 X e2 e2
Z X
jD (r) = d3 r0 (r0 )A(r0 )(r r0 ) (r0 ) = A(r) (r) (r) = A(r)n(r) (46)
mc mc
mc

For the paramagnetic current density, we have

Z
e X
jP (r) = d3 r0 (r0 ) ((p(r0 )(r r0 ) + (r r0 )p(r0 ))) (r0 ) (47)
2m
Z
i~e X
= d3 r0 (r0 ) ((r0 (r r0 ) + (r r0 )r0 )) (r0 ) (48)
2m

where we have used the coordinate space representation of the momentum operator (p(r) = i~r ). Now using
product rule

(r0 ) [r0 (r r0 ) (r0 )] = r0 (r0 )(r r0 ) (r0 ) r0 (r0 ) (r r0 ) (r0 )

   

(49)

for the first term, we get

Z
i~e X
jP (r) = d3 r0 r0 (r0 ) (r0 )(r r0 ) r0 (r0 ) (r r0 ) (r0 ) + (r0 )(r r0 ) (r0 (r0 ))




2m
(50)

The first term can be identified as probability density at the boundaries, which vanishes on integration. We can then
simply write
i~e X 
jP (r) = (r) (r (r)) r (r) (r)


(51)
2m

(b) Let us now consider the Fourier transform of the paramagnetic current density as
Z
jP (q) = d3 r0 jP (r0 )eiqr0 (52)

Again using the plane wave expansion for the field operators
1 X ikr
(r) = e ck (53)
V k
1 X ikr
(r) = e ck (54)
V k

and (51), we express (52) as

Z
jP (q) = i~e
XXh 0 0
i
d3 r (ik)ei(kk )r ck0 ck + (ik0 )ei(kk )r ck0 ck eiqr (55)
2mV k,k0

0
1
d3 rei(kk q)r = kk0 ,q , we find
R
Integrating over r, noting that V

jP (q) = ~e
XXh i
(k + k0 )ck0 ck kk0 ,q (56)
2m 0 k,k

jP (q) = ~e (2k + q)ck+q ck

XX
(57)
2m
k
 6

PROBLEM 4

(a) (i) Let us use the given definitions for the spin operators in terms of the bosonic operators to simplify each side
of the given identity

LHS = SZ2 + SZ (58)

2
= (s a a) + (s a a) (59)
2 2
= s + (a a) 2sa a + s a a (60)
2 2
= s + s + (a a) 2sa a a a (61)

RHS = s(s + 1) S S+ (62)

 p  p 
= s(s + 1) a 2s a a 2s a aa (63)
= s(s + 1) a 2s a a a


(64)
2
= s + s 2sa a + a (aa 1)a (65)
2 2
= s + s + (a a) 2sa a a a (66)

(ii) Similarly,

LHS = S+ S S S+ (67)
p  p   p  p 
= 2s a aa a 2s a a a 2s a a 2s a aa (68)
p p
= 2s a a(a a + 1) 2s a a a 2s a a a


(69)
= (a a + 1) 2s a a a 2s a a a



(70)

= 2s a aa a a a + a a aa = 2s 2a a = 2SZ = RHS (71)

(iii)
p 
LHS = SZ S+ = (s a a) 2s a aa (72)
p
= 2s a a(s a a)a (73)
p
= 2s a aa(s a a + 1) (74)
= S+ (SZ + 1) (75)

Also,
 p 
LHS = SZ S = (s a a) a 2s a a (76)
p
= a (s a a 1) 2s a a (77)
p
= a 2s a a(s a a 1) (78)
= S (SZ 1) (79)

Hence proved.
(b) For nearest neighbor interactions,

Si Si+ = SZi SZi+ + SXi i+

SX + SYi SYi+ (80)
1 i
1 i
= SZi SZi+ + i

i+ i+ i

i+ i+


S+ + S S+ + S S+ S S+ S (81)
4 4
i i+ 1 i i+ i i+


= SZ SZ + S S + S S+ (82)
2 +
 7

Restricting
up to quadratic terms, we approximate the spin raising and lowering operators as S+ 2sa and
S 2sa . This gives the Hamiltonian as
 
JX i i+ 1 i i+ i i+


H= SZ SZ + S S + S S+ (83)
2 2 +
i,
J X  
= (s ai ai )(s ai+ ai+ ) + s ai ai+ + ai ai+ (84)
2
i,

Js2 X Js X  
= + Js ai ai ai ai+ + ai ai+ (85)
2 2
i, i,

(86)

Now using the plane wave expansion for the bosonic operators
1 X ikri
ak = e ai (87)
N i
1 X ikri
ak = e ai (88)
N i

such that the operators {ak , ak0 } satisfy the canonical commutation relations. The Hamiltonian can then be
rewritten as

Jsm X X  ik0 ri  ikri
Js X X h ik0 ri+  ikri
i

H = E0 + e ak0 e ak e ak0 e ak + H.C. (89)
N i 0
2N 0
k,k i, k,k

(90)

where we have assumed that each lattice site interacts with m nearest neighbors. We define E0 Js2 /2 as the
group state energy. Simplifying further, using (6)

Js X ik
ak ak + C.C. ak ak
X

H = E0 + Jsm e (91)
2
k k
X
cos (k ) ak ak
X
= E0 + Jsm ak ak Js (92)
k k,

(93)

One can then find the dispersion relation as

!
1 X
~k = Jsm 1 cos(k ) (94)
m

1
P
It can be seen from the above expression that as k 0, m cos(k ) 1 and k 0.