Berrys Geometric Phase

Introductory Lecture

Raffaele Resta
Dipartimento di Fisica, Universit di Trieste

2014

The landmarkDownloaded
paper,
1983-1984
from rspa.royalsocietypublishing.org on January 2, 2012

Very simple concept, nonetheless missed by the founding

fathers of QM in the 1920s and 1930s
Nowadays in any modern elementary QM textbook
.

Sakurai QM textbook, p. 464

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

Is this a paradox?

particles, even in the regions where all the fields (and therefore the forces on the
particles) vanish.
Feynman
Lectures (1962-63), Vol. 2, Sec 15-5
The free
paper
was at
shocking,
and its conclusions were challenged by several auNowadays
online
http://www.feynmanlectures.caltech.edu
thors; nonetheless experimental validations appeared as early as 1960 [25, 26].
The main message of Ref. [23] is at the basis of many subsequent developments

Figure 2.1: The Aharonov-Bohm interference experiment (From Ref. [24])

...E and B are slowly disappearing from the modern expression

of physical laws; they are being replaced by A and .
10

Feynman vs. the bad guys....

IL NUOV0 CIMENT0

VoL. 47 A, N. 4

Nonexistence of the Aharonov-Bohm

21 0~obre 1975

Effect.

P. B o c c m ~ i
Istituto di Fisica Teorica dell' Universitb . Pavia, Italia
Istituto iVazionale di ~isica Nucleate - Sezione di Pavia, Italia

A. L O Z ~ G ~
Istituto di Scienze Fisiehe dell'U~iversith - Milano, Italia

(ricevuto il 2 Giugno 1978)

Summary. - - In ~his paper the Akaronov-Bohm effect is investigated

and it is shown that iC has a purely mathematical origin. All the physical
consequences of quantum mechanics turn out to be dependent on the
field strengths and not on the potentials.

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

Michael Berry
Professor Sir Michael Berry, FRS

Berry knighted by the queen

Ig Nobel award, 2000: M. Berry & A. Geim

Berrys reaction at his website:

We are pleased to accept the Ig prize because we have always
considered it a duty to make physics more understandable and
bring it closer to nonscientists. We think the prize
acknowledges our contribution in this direction.....
.

From tinkering on the fringes to Nobel

glory award, 2010
The real Nobel

Our other blog

Short Sharp Scien

13:42 20 January 2012

One Per Ce

Research

New Scientist T

Sean O'Neill, contributor

CultureL

Big Wide Wo

Bookmark&shar

Categorie
Biotech
Business
Environment
Getting a job
Graduate
I'm a scientist
Interview
Interviews
Masters

Andre Geim receives the Nobel Prize from Swedish King Carl XVI Gustaf (Image: Sipa
Press/Rex Features)

Andre Geim

Andre Geim shared the Nobel prize in physics in 2010 for his co-discovery of graphene. He is
. the University
.
director of the Manchester Centre for Mesoscience & Nanotechnology at
of.

PhD
Research
.

Basics
Berry phase: Fundamentals
Parametric Hamiltonian, non degenerate ground state

n depends on a parameter (and has nondegenerate

ate):

H()|() = E()|()

parameter : slow variable

H()|()! = E()|()!.

|(3 )!
|(2 )!
|(1 )!
ei12 =

( 1 )|( 2 )
|( 1 )|( 2 )|
= Im log ( 1 )|(2 )

ei12 =
12

|(4 )!

#(
2 )! +
=1 )|(
12
23
|#(1 )|(2 )!|

+ 34 + 41

= Im log (1 )|( 2 )(2 )|(3 )( 3 )|(4 )( 4 )|(1 )

12Gauge-invariant!
= Im log #(1 )|(2 )!
Berrys phase 2010 p. 7

Basics

Parametric Hamiltonian, non degenerate ground state

H()|() = E()|()

|(3 )!
|(2 )!
|(1 )!

parameter : slow variable

( 1 )|( 2 )
|( 1 )|( 2 )|
= Im log ( 1 )|(2 )

ei12 =
12

|(4 )!

= 12 + 23 + 34 + 41

= 12 + 23 + 34 + 41

= Im log (1 )|( 2 )(2 )|(3 )( 3 )|(4 )( 4 )|(1 )

#(1 )|(2 )!#(2 )|(3 )!#( 3 )|(4 )!#(4 )|(1 )!

Gauge-invariant!

invariant!
.

From discrete geometry to differential geometry

A smooth closed curve C in space
ei =

()|(+)
|()|(+)|

If we choose a differentiable gauge:

i ()| ()

" ()|( + ) #

ei =
| " ()|( + ) # |

s,s+1

s=1

i $ " ()| ()
#
d
= A() d

d linear differential form,

d
C

= i ()| () d
i ()| () vector field
.

Berry connection & Berry curvature

Domain S:

S Rd

Berry connection
A() = i ()| ()
real, nonconservative vector field
gauge-dependent
geometrical vector potential
a.k.a. gauge potential

Berry curvature ( R3 )
() = A() = i ()| | ()
gauge-invariant (hence observable)
geometric analog of a magnetic field
a.k.a. gauge field
.

The Berry connection is real

()|() = 1

()|() = 0
= ()|() + ()| ()
= 2 Re ()| ()

()| ()
A() = i ()| ()

purely imaginary
(1)

real

Last but not least:

What about time-reversal invariant systems?
.

Berry connection vs. perturbation theory

|0 ( + ) |0 ()

n ()| [ H( + ) H() ] |0 ()

|n ()
E0 () En ()
n=0

| 0 () =

|n ()

n=0

n ()| H()|0 ()
E0 () En ()

A () = i0 ()| 0 () = 0
parallel transport gauge
.

Berry connection vs. perturbation theory, better

|0 () =

|n ()

n=0

n ()| [ H( + ) H() ] |0 ()
E0 () En ()

|0 ( + ) |0 () + |0 ()

Better:
|0 ( + ) [ |0 () + |0 () ] ei( )

[1 i () ] |0 () + |0 ()

A() d = i0 ()| 0 () d
= 0

d
.

Berry curvature: perturbation theory is OK

The Berry curvature is gauge invariant
() = A()
( R3 )
0 ()|H()|n () n ()|H()|0 ()
= i
[E0 () En ()]2
n=0

() singular at degeneracy points

Berry curvature: perturbation theory is OK

The Berry curvature is gauge invariant
() = A()
( R3 )
0 ()|H()|n () n ()|H()|0 ()
= i
[E0 () En ()]2
n=0

() singular at degeneracy points

Berry curvature: perturbation theory is OK

The Berry curvature is gauge invariant
() = A()
( R3 )
0 ()|H()|n () n ()|H()|0 ()
= i
[E0 () En ()]2
n=0

() singular at degeneracy points

Stokes theorem: C =

=
A() d =
() n d

.......only if is simply connected!

.

Berry phase

Loop integral of the Berry connection on a closed path:

=
A() d
C

Berry phase, gauge invariant modulo 2

corresponds to measurable effects

Main message of Berrys 1984 paper:

In quantum mechanics, any gauge-invariant quantity is
potentially a physical observable

Coupling to the rest of the Universe

cannot be cast as the expectation value of any Hermitian
operator: instead, it is a gauge-invariant phase of the
wavefunction
The quantum system is not isolated:
the parameter summarizes the effect of the rest of the
Universe
Slow variables: (e.g., a nuclear coordinate).
Fast variables: here, the electronic coordinates
For a genuinely isolated system, no Berry phase occurs
and all observable effects are indeed expectation values of
some operators
What about classical mechanics?
.

Semantics: why Geometric?

So far, everything time-independent.
Suppose instead that:
The energy of |() is E()
The parameter moves adiabatically on the closed path in
time t: (t), with (T ) = (0)
Then the state acquires a total phase factor ei ei(T )
The phase is independent of the details of motion: hence
geometric
The additional phase is the dynamical phase, and does
T
depend on the motion:
(T ) = ~1 0 dt E((t))
.

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

A quantum system in zero field

quantum
system

The parameter
No magnetic field, box centered at the origin:
[
]
1 2
p + V (r) (r) = (r),
(r) real function
2
Parameter the box position:

H(R) =

1 2
p +V (r R)
2

r|(R) = (r R)
If there is a magnetic field (somewhere):
H(R) =

1
e
[ p + A(r) ]2 + V (r R)
2
c

r|(R) = ei(r) (r R)
r
e
A(r ) dr
(r) =
~c R
.

Berry connection & Berry phase

Formal solution!
However: In the region where B(r) vanishes, (r) is a single
valued function of r, and r|(R) is an honest electronic
wavefunction.
What about the dependence on the slow parameter R?
Berry connection:
i(R)|R (R) = i(R)|R (R)
Berry phase:
=

e
~c

e
A(R)
~c

A(R) dR
C
.

thors; nonetheless experimental validations appeared as early as 1960 [25, 26].

The main message of Ref. [23] is at the basis of many subsequent developments

A closer look at the Berry phase

e
=
~c

A(R) dR =
C

~c

In this problem (and only in this problem):

The2.1:
geometric
vector potential
coincides
theRef. [24])
Figure
The Aharonov-Bohm
interference
experimentwith
(From
magnetic vector potential (times a constant)
10
e
hc

is the flux quantum: = 2 0

Only the fractional part of /0 is relevant

The Berry phase is observable (mod 2)
.

Bottom line (no paradox!)

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

Reexamining the Born-Oppenheimer approximation

H([R], [x]) =

~2
2 + Hel ([R], [x])
2Mj Rj
j

[x]: electronic degrees of freedom (orbital & spin)

[R]: nuclear coordinates Rj
i~ Rj : canonical nuclear momenta
Hel ([R], [x]) = electronic kinetic energy
+ electron-electron interaction
+ electron-nuclear interaction
+ nuclear-nuclear interaction
.

Recipe
Product ansatz: ([R], [x]) = [x] |el ([R]) ([R])
Solve the electronic Schrdinger equation at fixed Rj :
Hel ([R], [x]) [x] |el ([R]) = Eel ([R]) [x] |el ([R])
Use Eel ([R]) as the potential energy for nuclear motion:

~2

2 + Eel ([R]) ([R]) = E ([R])

2Mj Rj
j

Textbook example: Vibrational levels of a biatomic

molecule.
On many occasions, the nuclear motion can be considered
as purely classical (Schrdinger Newton).
.

A closer look at the Born-Oppenheimer recipe

Product ansatz: ([R], [x]) = [x] |el ([R]) ([R])
The operator Rj acting on ([R], [x]):
Rj ([R], [x]) = [x] |el ([R])Rj ([R])
+ [x] |Rj el ([R])([R])
Multiplying by el ([R]) | [x] and integrating in d[x]:

d[x] el ([R]) | [x] Rj ([R], [x])

(
)
=
Rj + el ([R]) |Rj el ([R]) ([R])
Nuclear kinetic energy, after [x] is integrated out:
TN =

)2
~2 (
i~Rj i~ el ([R]) |Rj el ([R])
2Mj
j

A term was missing!

Naive Born-Oppenheimer approximation:

(
)
TN + Eel ([R]) ([R]) = E ([R]),

TN =

~2
2
2Mj Rj
j

More accurate Born-Oppenheimer approximation:

TN =

)2
1 (
i~Rj i~ el ([R]) |Rj el ([R])
2Mj
j

The electronic Berry connection acts as a geometric

vector potential in the nuclear Hamiltonian
In most cases the correction is neglected: Why?
.

The hydrogen (or sodium) trimer, LCAO

1
Conical intersection
in|A
the
|2 = ( |B+|C 2
) trim
6

1 its equilateral configura

Homonuclear|1
trimer
= in( |C|B
)

2
1
|0 = ( )|A+|B+|C
) 1 (
|B|C
;
|2 =
3
6

Degenerate HOMO
B

|1 =

1 (
2

2|A

Equilateral geometry, 3 valence electrons: degenerate

HOMO (1 = 2 )
Broken-symmetry equilibrium geometry: isosceles
Jahn-Teller splitting (1 = 2 )
|1 is the HOMO,

|2 is the LUMO
.

Born-Oppenheimer surfaces

pseudorotation

Born-Oppenheimer surfaces

The Greek word "diabolos" means "the liar" or "the one that commits perjury", from the verb "diaballo",
which means "to throw in", "to generate confusion", "to divide", or "to make someone fall". Later the word
"diabolos" was used by Christian writers as "the liar that speaks against God". From this meaning come
many modern languages' words for "devil" (French: diable, Italian: diavolo, Spanish: diablo, Portuguese:
diabo, German: Teufel, Polish: diabe).

pseudorotation

Confusion about the provenance of the name may have arisen from the earlier name "the devil on two
sticks", although nowadays this often also refers to another circus-based skill toy, the devil stick.

Design
The design of diabolos has varied through history and across the
world. Chinese diabolos have been made of bamboo. Wooden
diabolos were common in Victorian times in Britain. Rubber diabolos
were first patented by Gustave Phillippart in 1905. [2] In the late
twentieth century a rubberised plastic material was first used. Metal
has also been used especially for fire diabolos.

conical intersection

The size and weight of diabolos varies. Diabolos with more weight
tend to retain their momentum for longer, whereas small, light
diabolos can be thrown higher and are easier to accelerate to high
speeds. Rubber diabolos are less prone to breakage yet are more
prone to deformations. More commonly used are plastic-rubber
hybrids that allow flex but hold their shape. One-sided diabolos are
also available but are more difficult to use. For beginners diabolos of
a diameter of min 9 cm are recommended.

a.k.a. diabolical point

Nuclear dynamics
Eel () = Eel ()
-independent
1
Eel () =
k ( 2 2 min )
2
Lowest BO surface:
minimum in min
1 2
= EJT
Eel (min ) = k min
2
Classical: Free motion at valleys bottom, M = 3m
& transverse oscillations
Quantized pseudorotations:
mn (, ) Hn () e 2 (min ) eim
m Z,
n = 0, 1, 2, . . .
Ground state: m=0, n=0

H. C. Longuet-Higgins
et al.intersection
(1958)
Conical

in the trim

Homonuclear trimer in its equilateral configura

Degenerate HOMO
B

|1 =

1 (
2

|B|C ) ;

|2 =

1 (
6

2|A

The electronic wfn r|el () changes sign (a phase)

The total wfn (, r) = r|el () () must be single-valued
Even the nuclear wfn must change sign
Different quantization rules!
mn (, ) Hn () e 2 (min ) eim
m half-integer,
n = 0, 1, 2, . . .
1
Ground state: m = 2 , n = 0
Observable effect in QM, no effect in CM
(the system does not visit the conical intersection)

particles) vanish.
The paper was shocking, and its conclusions were challenged by several authors; nonetheless experimental validations appeared as early as 1960 [25, 26].
The main message of Ref. [23] is at the basis of many subsequent developments

Molecular Aharonov-Bohm effect

Aharonov-Bohm effect (real B field):

=
A() d = 2
mod 2
0
C

Figure 2.1: The Aharonov-Bohm interference experiment (From Ref. [24])

Molecular Aharonov-Bohm effect (B = 0):

10

=
A() d = mod 2
C

Same as having a -like flux tube at the conical intersection

=

0
2

(half-quantum, a.k.a. flux)

.

The Berry phase

Berry phase: discrete algorithm

|( j+1 )
|(j )

j,j+1

j=1

= Im log ( 1 )|( 2 )( 2 )|( 3N) . . . ( N )|( 1 )

!

j,j+1
= points
Three
are enough
j=1

= Im log (1 )|(2 )(2 )|(3 ) . . . . . .

|( 3 ) |C|A

N=3
|(1 ) |B|C

|( 2 ) |B|A

( 1 )|( 2 )( 2 )|( 3 )( 3 )|( 1 ) =

.

1
8
.

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

What topology is about

Authors Pictures
Leonardo Da Vinci
Shop
Composers Posters
Composers T shirts
Famous Physicists

Math Mug (set theory):

If you consider the set of all sets that have
never been considered ....

The above shapes are topologically equivalent

and are of Genus 0

Math Mug - Topology

To a Topologist This is a Doughnut

http://cosmology.uwinnipeg.ca/Cosmology/Properties-of-Space.htm (7 of 11) [03/01/2002 7:23:17 PM]

Topological invariant: genus (=1 here)

Math Mug:
Real Life is a Special Case
(black background)

http://mathematicianspictures.com/Math_Mugs_p01.htm

Page 4 of 6

What topology is about

Authors Pictures
Leonardo Da Vinci
Shop
Composers Posters
Composers T shirts
Famous Physicists

Math Mug (set theory):

If you consider the set of all sets that have
never been considered ....

The above shapes are topologically equivalent

and are of Genus 0

Math Mug - Topology

To a Topologist This is a Doughnut

http://cosmology.uwinnipeg.ca/Cosmology/Properties-of-Space.htm (7 of 11) [03/01/2002 7:23:17 PM]

Topological invariant: genus (=1 here)

Math Mug:
Real Life is a Special Case
(black background)

http://mathematicianspictures.com/Math_Mugs_p01.htm

Page 4 of 6

From Wikipedia:

genus 0

genus 1

genus 2

genus 3
.

1+cos 8
sin 8
Hence, if t,b is to be real, like ~1 and q 2 , we must have
c1 = sin $9,
c2 = -cos 30,
c2

Topology & conical intersections

(2.12)

or
c1 = -sin $8, c2 = cos 38.
(2.13)
In either case, as we move round the origin keeping R constant and allowing 6 to
increase from 0 to 27r, both c1 and cz change sign, and so does t,b. This result is
a generalization
of one which has been1963:
proved 5 in connection with the Jahn-Teller
Herzberg
& Longuet-Higgins,
effect, 6 where one also encounters a conically self-intersecting potential surface.
It shows that a conically self-intersecting potential surface has a different topological character from a pair of distinct surfaces which happen to meet at a point.
Indeed, if an electronic wave function changes sign when we move round a closed
loop in configuration space, we can conclude that somewhere inside the loop there
must be a singular point at which the wave function is degenerate; in other words,
there must be a genuine conical intersection, leading to an upper or lower sheet of
the surface, as the case may be.
3.

THREE HYDROGEN-LIKE ATOMS

Berry
phase

A useful
illustration
of the above generalizations is a system of three hydrogen
atomsTopologically
near the verticestrivial:
of an equilateral
triangle.
the(0
internuclear
= 0 mod
2 = If
mod 2) distances
Topologically nontrivial: = mod 2 = (1 mod 2)
Topological invariant Z2
(Z2 = additive group of the integers mod 2)
.

Robustness of the topological invariant

Two-valued topological invariant:

The Z2 index is either 0 or 1 (mod 2)
The index is robust against deformations of the path C,
provided it does not cross the obstruction
The index is very robust against continuous deformations
of Hamiltonian & wave function,
provided the HOMO-LUMO gap does not close
We can even continuously deformate the wfn into the
exact correlated one (if ground state non degenerate)
Key role of time-reversal invariance
In modern jargon:
Z2 invariant is protected by time-reversal symmetry
.

Outline

Aharonov-Bohm effect, 1959

Elements of Berryology

Aharonov-Bohm revisited

Born-Oppenheimer approx. in molecules (B = 0)

The Z2 topological invariant

Born-Oppenheimer approx. in molecules (B = 0)

BO approx. for the H atom, B = 0

~2 2
+ Hel (R, r)
2M R
~2 2
e2
Hel (R, r) =
r
2m
|r R|
H(R, r) =

Lowest BO surface:
Eel (R) = const =
BO Recipe:
EBO (k) =

e2
,
2a0

r|el (R) e|rR|/a0

~2 2
e2
R (R)
(R) = E(R)
2M
2a0

~2 k 2
e2
,

2M
2a0

BO (R, r) e|rR|/a0 eikR

.

Compare exact with Born-Oppenheimer approx.

H(R, r) =
Separable using:

~2 2
~2 2
e2
R
r
2M
2m
|r R|
= M R + m r,
R
M +m

~2 k 2
e2

,
2(M + m) 2a0
~2 k 2
e2
EBO (k) =

2M
2a0
lim E(k) = EBO (k)
E(k) =

r = r R

mM
m+M

m/M0

Compare exact with Born-Oppenheimer approx.

H(R, r) =
Separable using:

~2 2
~2 2
e2
R
r
2M
2m
|r R|
= M R + m r,
R
M +m

r = r R

~2 k 2
e2
mM

,
=
2(M + m) 2a0
m+M
2
2
2
~ k
1
e
EBO (k) =
,
k

2M
2a0
a0
lim E(k) = EBO (k)
E(k) =

m/M0

BO approx. for the H atom, B = 0

(Neglecting irrelevant spin-dependent terms)

H(R, r) =
Hel (R, r) =

]2
1 [
e
i~R A(R) + Hel (R, r)
2M
c
[
]2
1
e
e2
i~r + A(r)
2m
c
|r R|

In a constant B field
Eel (R) = Eel = const
]2
1 [
e
Naive recipe:
i~R A(R) (r) Eel (r) = E(r)
2M
c
Same kinetic energy as if the proton were naked
Classical limit: the H atom is deflected by a Lorentz force
A neutral system is not deflected by a Lorentz force
.

Solution of the paradox

Screened Born-Oppenheimer approximation:

Schmelcher, Cederbaum, & Meyer, 1988
Better:
Berry Connection & Berry curvature (same as for B = 0)
]2
]2
1 [
e
1 [
e
i~R A(R)
i~R A(R) ~A(R)
2M
c
2M
c
A(R) genuine vector potential of magnetic origin
A(R) = iel (R)|R el (R) Berry connection

Detailed reckoning in the central gauge

]2
1 [
e
e2
i~r
Br
2m
2c
|r R|
[
]
2
1
e
e2
i~r +
Br
Hel (0, r) =
2m
2c
r

r|el (0) = 0 (r)

complex wfn, cylindrical symmetry
Hel (R, r) =

ie
r|el (R) = e 2~c rBR 0 (|r R|)
e
e
A(R) = iel (R)|R el (R) =
B R = A(R)
2~c
~c
]2
1 [
e
~2 2
TN =
i~R A(R) ~A(R) =

2M
c
2M R
.

Magnetic & geometric together

H atom
Paradox solved (both quantum nucleus & classical nucleus)
In the classical limit no Lorentz force
Hamiltonian (quantum & classical)
The Berry connection cancels the vector potential
Newton Eq. (gauge invariant):
The Berry curvature cancels the magnetic field

Molecule

(rotations & vibrations in a B field)

The two terms do not cancel

They are of the same order of magnitude
The geometric term is important even for classical nuclei:
geometric Lorentz force in Newton Eq.

B = 0 vs. B = 0 in Born-Oppenheimer

B=0

(time-reversal symmetric)

Conical intersections nontrivial geometric effects

The electronic wfn can be chosen as real
The Berry curvature vanishes (or is singular)
Classical nuclei not affected by geometric effects
The Berry phase only shows up when quantising the nuclei

B = 0

(time-reversal symmetry absent)

No singularity needed in the Born-Oppenheimer surface

The electronic wfn must be complex
The Berry curvature is generally nonzero
Classical nuclei are affected by geometric effects
The Berry curvature enters the Newton Eq. for the nuclei