chun wa wong quantum mech rvw 2

8/12/2019 Chun Wa Wong Quantum Mech Rvw 2

1/29

A Review of Quantum Mechanics

Chun Wa WongDepartment of Physics and Astronomy, University of California, Los Angeles, CA 90095-1547

(Dated: September 21, 2006)

Review notes prepared for students of an undergraduate course in quantum mechanics at UCLA,Fall, 2005 Spring, 2006. Copyright c 2006 Chun Wa Wong

I. The Rise of Quantum Mechanics

1.1 Physics & quantum physics:

Physics = objective description of recurrent physi-cal phenomena

Quantum physics = quantized systems with bothwave and particle properties

1.2 Scientic discoveries:

Facts Theory/Postulates Facts1.3 Plancks black body radiation:

Failure of classical physics:

Emissivity = 14

uc, max T = const ,

energy density = u = E ,E exp = kB T, (1)

( = density of states in frequency space).Planck hypothesis of energy quantization (E n =nh ) explains observation:

E = h n =0 nenxn =0 enx = h

eh/k B T 1, (2)

where x = h/ (kB T ).

1.4 Einsteins photoelectric effect:

Photoelectrons ejected from a metal by absorbedlight of frequency have maximum energy:

E = h W (work function) (3)that depends on the frequency of light and notits intensity.

1.5 Bohrs atomic model:

Electron of mass m in a classically stable circularorbit of radius r = a and velocity v around theatomic nucleus:

mv 2

r =

e2Gr 2

, or T = U 2

, (4)

Electronic address: [email protected]

(e2G = e2 / 40 in Gaussian units). Hence

E = T + U = U 2

= T. (5)

Bohrs two quantum postulates (L quantization andquantum jump on photon emission):

L = mva = n , n = 1 , 2, ...,h = E n i E n f . (6)

For ground state:

E 1 = U 1

2 =

e2G2a

= 13.6 eV= T 1 =

2

2ma 2,

a = 2

me 2G= 0 .53A (Bohr radius) , (7)

For excited states:

r n = n2a, E n = E 1n2

. (8)

1.6 Compton scattering:Energy-momentum conservation for a photon of labenergy E = hc/ scattered from an electron (of mass m) initially at rest in the lab (before=LHS,after=RHS; 1=photon, 2= recoiling e):

p = p1 + p 2 ,

E + mc2 = E 1 + m2c4 + p22c2 . (9)Photon scattered into lab angle has longer wave-length 1 :

1 = e (1 cos ), where e = hmc = hcmc2 = 2 .4 pm (10)

is the electron Compton wavelength.

1.7 de Broglies matter wave:Massive matter and massless light satisfy the sameenergy-momentum and momentum-wavelength re-lations

E 2 = p2c2 + m2c4 ,

p = k = h

. (11)


2/29

2

Thus moving matter is postulated to be a wave witha motional de Broglie wavelength = h/p , leadingto the diffraction maxima for both matter and lightof

2-sided formula: 2 d sin = n ,1-sided formula: D sin = n .

II. Waves and quantum wavesClassical and quantum waves have identical mathemat-

ical properties.

2.1 Waves in physics:

Waves rise and fall, travel, have coherence in space and in time , interfere, and diffract.Two classes of waves:

(a) Inertial waves in massive media: ocean waves,sound waves, vibrations of violin strings;

(b) Noninertial waves in vacuum: EM waves,matter waves.

2.2 Mathematical description of waves:

The 1-dimensional wave equation (1DWEq)

2

x 2 1v2

2

t 2(x, t ) = 0 (12)

is an equation of state that gives a wave function (x, t ) describing an unfolding event in spacetime.The wave equation is a partial differential equation

(PDE) because it depends on more than one vari-able, here the two variables x and t .In contrast, the solution x(t) of an equation of mo-tion such as a Newton equation describes how asingle point x(t) on an object evolves in time. Sim-ilarly, a snapshot of an object is not necessarily awave. A wave function ( x, t ) is needed to describea waves coherent structure in both space and time .

However, a wave (any wave) can have particle prop-erties if a point x can be dened on it such that onecan describe how its position changes in time.The wave equation (12) describes linear waves sat-

isfying the superposition principle that a sumof solutions is also a solution :

= a 1 1 + a 2 2 . (13)

Traveling waves:

(a) f (x vt)= a wave traveling to the right (+x),(b) g(x + vt)= a wave traveling to the left (-x),(c) the composite wave a1f (x vt) + a2g(x + vt)is also a wave.

LDEs (linear DEs) with constant coefficientshave exponential solutions:

ddx

eikx = ike ikx ,

ddt

eit = ieit . (14)

These DE are called eigenvalue equations be-cause the constant ik is called an eigenvalue of thedifferential operator d/dx , while the solution eikx iscalled its eigenfunction belonging to its eigenvalueik .

The 1DWEq (12) has four distinct (i.e., linearlyindependent) eigenfunctions:

(x, t ) = ei (kx t ) . (15)

Energy in both classical and quantum waves is car-ried by the intensity or energy ux

I ||2 . (16)Its quadratic dependence on is responsible for theinterference between two traveling waves i =Ai ei i with real A i :

|1 + 2|2 = |1|2 + |2|2 + 2Re( 12)= A21 + A

22 + 2 A1A2 cos( s),

where s = k x k

t . (17)

In the limit k

0,

ddk

= vg = group velocity

= v = k

= wave velocity . (18)

Examples: Certain waves satisfy the relation =Ak p . Then vg /v = p. Specic examples are p = 2for transverse vibrations of a thick bar, p = 3 / 2for short ripples moving under surface tension, and p = 1 / 2 for long waves on deep sea.The result (17) is similar in structure to the squaredlength of a sum of two vectors in space that inter-sect at an angle :

|A 1 + A 2|2 = A21 + A22 + 2 A1A2 cos , (19)Two waves of the same amplitude traveling in op-posite directions interfere form a standing wave :

= A[cos(kx t) + cos( kx + t)]= 2 A cos(kx)cos(t) (20)

with factorized space and time dependences.


3/29

3

2.3 Matter-wave quantization:Light wave of momentum p and energy E = pc isdescribed by the wave function:

ei(kx t ) = ei [( p/ )x(E/ ) t ]. (21)

de Broglie and Schrodinger suggested that the RHSexpression holds also for matter waves so that:

i

xei ( p/ )x = p ei ( p/ )x = p ei ( p/ )x ,

i t

ei(E/ ) t = H ei (E/ ) t = E ei (E/ ) t . (22)

The differential operators that appear are calledthe momentum operator p and the Hamiltonian (orenergy) operator H , respectively.Commutation relations: The rules of differen-tial calculus dictates that a differential operatordoes not commute with its own variable:

x x = 1 + x

x , or

x , x = 1;

t

t = 1 + t t

, or t

, t = 1 . (23)

These results give the fundamental commuta-tors of wave mechanics

[ p, x] =

i, [ H, t ] = i . (24)

2.4 Wave packet and the uncertainty principle:A wave packet is a superposition of waves of thetype (21)

f (x, t ) = 1 2 g(k)ei [kx (k ) t ]dk, (25)

with a linear superposition function g(k) sometimescalled a spectral amplitude .Expectation value (= mean value) of a propertyA(x, t ) of the wave:

A(x, t ) P (x, t )A(x, t )dx, whereP (x, t ) = |f (x, t )|

2

|

f (x, t )

|2dx

(26)

is the normalized probablity density of nding thewave at the spacetime point ( x, t ).The expectation value of a time-independent prop-erty B(k) of the wave can be calculated by usingthe spectral amplitude g(k):

B (k) P (k)B(k)dk, whereP (k) = |g(k)|

2

|g(k)|2dx (27)

is the normalized probablity density of nding thewave in k space. (We shall not consider morecomplicated properties that also depend on thetime t or the energy E .)Uncertainty relations: A wave of any kind sat-ises the uncertainty relation

x k 12 , where

( x)2 = (x x)2 , x = x ,( k)2 = (k k)2 , k = k . (28)

The expression for matter waves is usually writtenas the Heisenberg uncertainty principle :

x p

2. (29)

Spreading wave packets: A Gaussian wavepacket has the spectral amplitude

g(k) = ek 2

, with k = 12 . (30)

It gives rise to a wave function at time t = 0 of

f (x, t = 0) ex 2 / (4 ) , with x = . (31)

Hence the wave packet at time t = 0 has the mini-mal uncertainty product of

x k = 12

. (32)

For t = 0, however, the uncertainty product couldbe greater than the minimal value of 1/2. This hap-pens if the system is dispersive , meaning a nonzerosecond Taylor coefficient in the following expan-sion:

(k) (k = 0) + kddk 0

+ 12

k2d2dk2 0

= (0) + kvg + 12

k2. (33)

The resulting approximate wave function at anytime t can then be evaluated to the closed form

f (x, t ) ex2t / (4 t ) , with

x t = x vg t, t = + i 2 t. (34)

Thus the position x of the wave moves in time withthe group velocity

vg = ddk 0

. (35)

Its position uncertainty increases in time to

x(t) = | t | = 1 +t2

2 1/ 2

. (36)


4/29

4

For a massive particle in free space E 2 = p2c2 +m2c4 . Hence

= d2E dp2

= m2c6

E 3 = 0 . (37)Thus the spatial width of the Gaussian packet of a massive particle spreads out in both positive and

negative times.

III. Wave mechanicsWave mechanics describes physics by using the

Schr odinger wave eq. and its wave functions.

3.1 Schr odinger wave equation (SchWEq) :First quantization: The classical E-p relationE 2 = p2c2 + m2c4 is quantized into a wave equationby replacing E , p by the differential operators

E H = i

t

p p =

i

x : (38)

i t

2

(x, t ) =

i

x

2

c2 + m2c4 , (39)

where the wave function ( x, t ) has to be added forthe operators to operate on. In this way, a classicalstate of motion spreads out into a wave structure.Nonrelativistic limit: Likewise, the NR energy

E = T + V = p2/ 2m + V (not containing the restenergy mc2) is quantized into the SchWEq:

i t

(x, t ) = 12m

i

x

2

+ V (x) , (40)

Commutation relations: A differential operatordoes not commute with its variable. Hence

[ p, x] =

i, [ H, t ] = i . (41)

Group velocity: E 2 = p2c2 + m2c4

EdE =c2 pdp. Hence

vvg = E p

dE dp

= c2 . (42)

3.2 Borns interpretation of ||2 :The squared wave function gives the probabililtydensity of nding system at spacetime location(x, t ):

|(x, t )|2 = P (x, t ) (43)

in the sense that the probability for nding thestate somewhere in space at time t is

P (t) = P (x, t )dx= |(x, t )|2dx = 1 . (44)

By choosing P (t) = 1, the wave function is said tobe normalized .

3.3 Conservation of ||2:The probability density can be moved around, butcannot be created or destroyed in the absence of explicit creative or destructive physical processes:

0 = ddt

P (x, t ) = t

+ dxdt

x

P (x, t )

= t

P (x, t ) + x

jP (x, t ). (45)

This is called the continuity equation for the con-served probability current density

jP (x, t ) = v P (x, t ), (46)

where v = dx/dt .

IV. Wave mechanics in one spatial di-mension

Wave mechanics is applied to simple one-dimensionalproblems to illustrate how the SchWEq is solved and howits wave character affects the properties of a system.

4.1 Solving the Schr odinger wave equation(SchWEq) :Separation of varables: Since solutions areunique up to a superposition, we may look for asolution of the factorized form

(x, t ) = (x)(t). (47)

Then

(x) H(t)(x) (t)

=(t) p

2

2m + V (x)

(t) (x) = E. (48)

On simplication, the rst expression is clearly afunction of t only, while the middle expression is afunction of x only. Hence the separation constant E must be a constant independent of t or x.The PDE can thus be broken up into two ordinaryDEs:

H(t) = i t

(t) = E (t),

12m

i

x

2

+ V (x) (x) = E (x), (49)


5/29

5

x 0

x 0

E

V 0

V 0

0

k 2 q2

q2

BCs: ' '

FIG. 1: Wave numbers and wave functions at a potential step.

where E appears in both ODEs as the energyeigenvalue . The rst ODE gives the time-dependent factor

E (t) = ei (E/ ) t , (50)

but the second equation is a second-order ODEwhose solution requires the imposition of twoBoundary conditions at a suitable boundary x =x0:

< (x0) = > (x0 ),

< (x0 ) ddx

< (x)x 0

= > (x0 ). (51)

Note that if the rst BC is violated, (x0) becomesinnite and the DE (differential eq.) cannot be sat-ised. If the second BC is violated, (x0) becomesinnite. The DE also cannot be satised, except forthe special case where V (x0 ) has a compensatinginnity at x0 .

4.2 Particle in a box of length L:With the two BCs (0) = (L) = 0, the solutionsturn out to be the same as those describing classicalvibrations of a violin string of length L:

n (x) = 2L sin nL x (52)for a normalized wave function.

4.3 Potential step:Figure 1 shows a particle of mass m incident from

the left with energy E . The wave numbers k , q aredened as follows:

E = 2k2

2m , E V =

2q 2

2m ; (53)

or k2 = q 2 k20 , where k20 = 2mV 0/ 2 . The parti-cles matter wave is partially reected at the bound-ary x = 0, and partially transmitted across it (withno further reection downstream):

x < 0 : < (x) = eikx + Be ikx ,x > 0 : > (x) = AT eiqx . (54)

x 0 x L

E

V 0 2

0

k 2

q2

FIG. 2: Wave numbers for a rectangular potential barrier.

Its probability current density is conserved acrossthe boundary:

j< = k

m1 |B |2

= j > = q m |AT |

2 . (55)

Its reection and transmission coefficients at theboundary are:

R = |B |2 , T = 1 R. (56)4.4 Potential barrier:

A particle of incident energy E higher than theheight V 0 of a rectangular potential barrier is par-tially reected and partially transmitted by a potet-nial barrier. The wave functions for x < 0 beforethe barrier and for x > L after the barrier remainthe same as Eq.(54), but the presence of the sec-ond boundary at x = L causes reected waves to

be present in the barrier region0 < x < L : < (x) = Ceikx + De ikx . (57)

The two boundary conditions at the second bound-ary x = L are enough to determine the two extracoefficients C and D . The reection coefficient re-mains R = |B |2 , but its value is changed. Thetransmission coefficient is

T = 1 R = 1

1 + k2 q22kq

2sin2(qL)

. (58)

Quantum tunneling: The formula for T workseven when the energy E falls below the height V 0 of the potential barrier, as shown in Fig. 2. Then q 2 =

2 < 0, and q = i becomes purely imaginary.The function [sin( qL)/q ]2 in the formula is analyti-cally continued to the function [sinh( L )/ ]2 . Theresulting transmission coefficient T across the nitebarrier is shown in Fig. 3. The fact that T does notvanish for E < E 0 , as it does in classical mechanics,is called quantum tunneling . Note that tunnelingcan occur only when the barrier thickness is nite.


6/29

6

1 2 3 4 5 E V 0

0.2

0.4

0.6

0.8

1T

FIG. 3: Transmission coefficient across a rectangular potentialbarrier.

1 2 3 r R

0.5

0.5

1V

Alphaemission Fusion

FIG. 4: Tunneling across the Coulomb barrier in decay andin nuclear fusion.

4.5 Applications of quantum tunneling:Alpha decay: An particle inside a nucleus feelsan attractive nuclear potential. As the particleleaves the nucleus in alpha decay, it sees the repul-sive Coulomb potential outside the nucleus. Thesetwo potentials combine to form a potential barrierthat the particle must tunnel through on its wayout, as illustrated in Fig. 4. The decay probablilityis thus fT , the product of the frequency f of the particles hitting the inside wall of the potentialbarrier and the transmission coefficient T throughthe potential barrier (that gives the tunneling prob-ability per attempt).Nuclear fusion: is just the time reversed situationof emission. Here a charged particle, say a proton,is trying to enter a nucleus from the outside by tun-

neling through its very high and thick Coulomb po-tential barrier. The transmission coefficient is typi-cally very small, but increases very rapidly with theenergy E . If a number of atomic nuclei are exposedto a gas of protons of increasing temperature, thehigh-energy tail of the Boltzmann energy distribu-tion of the proton gas will allow more and moreprotons to get into atomic nuclei to initiate nuclearfusion reactions. For example, the pp reaction

p + p d + e+ + (59)(where e+ is the positron and is a particle called

a

W

Sample External field

b

W

SampleScanningtip

FIG. 5: Tunneling through a nite potential barrier (a) incold emission, and (b) in a scanning tunneling microscope.

the electron neutrino) powers the conversion of mass into solar energy in the interior of the Sun.The reaction rate depends on the temperature T as T 4 .The luminosity of a star usually increases as M 3 of

its mass M , or T 24

of its central temperature. Suchan unusually strong dependence led Bethe in 1938to look for another source of hydrogen burning. Hediscovered that this alternative route is a certainCNO cycle that involves certain carbon, nitrogenand oxgen isotopes in intermediate steps.Cold emission: Electrons are bound in metals bytheir work function W . However, they can be in-duced to tunnel out of the metal by applying anexternal electric eld to change the potential toa triangular penetrable potential, as illustrated inFig. 5(a).

In scanning tunneling microscopes, the tunnel-ing is induced by the proximity of the scanning tip,as shown in Fig. 5(b). Since the tunneling currentincreases greatly as the thickness of the potentialbarrier decreases, one has a very sensitive device formapping the geometrical shape of a surface downto atomic dimensions.

4.6 Bound states in a potential well:If V (x) constant as x , the lower of thesetwo contant values is usually taken to be the zeroof the energy scale. Two situations can then berealized:

(a) A system with E > 0 has continuous energyspectrum with wave functions that are not lo-calized or square-integrable, while

(b) a system with E < 0 has a discrete energyspectrum with localized wave functions thatare square-integrable . The resulting physicalstates are called bound states in a potentialwell.

The wave numbers associated with a bound state inan attractive one-dimensional square-well potential


7/29

7

x a x a

E

V 0

2V 0

q2

FIG. 6: Wave numbers associated with a bound state in anattractive square-well potential.

are summarized in Fig. 6, where

|E | = E = 22

2m , E V =

2q 2

2m . (60)

Such an attractive one-dimensional potential hasat least one bound state, the ground state of

even parity whose wave function is proportional tocos qx. This is because such a wave function al-ready bends over at the boundaries, and can al-ways be matched to a decreasing exponential func-tion outside the potential well.

V. The mathematical structure of quan-tum mechanics

Physical states in quantum mechanics are linear waves.Both quantum states and classical linear waves can besuperposed to form other waves, in the same way thatordinary vectors in space can be added to form other

vectors.5.1 Fourier transform (FT):

The superposition principle for linear wave func-tions (solutions of linear wave equations) is formal-ized in the methods of Fourier series and Fouriertransforms (FT). Both methods and their general-izations are used in quantum mechanics.The FT of a wave function f (x) in one spatial vari-able x is dened by the integral

F{f (x)} g(k) = 1 2 f (x)eikx dx. (61)

For example,

F ddx

f (x) = ikg(k) ip

g(k). (62)

Thus the momentum operator p = ( /i )/x yieldsthe momentum value p = k associated with thespectral amplitude g(k).Fourier inversion formula:

F 1{g(k)} f (x) = 1 2 g(k)eikx dk. (63)

This is how a wave packet is constructed, namelyby taking a linear superposition of waves. Also

F 1 ddk

g(k) = ixf (x). (64)

Thus a position operator x = i/k gives the valueof the position x associated with the wave function

f (x).The FT shows that waves can be described eitherby wave functions f (x) in space, or by their spectralamplitudes g(k) in k-space. Functions of one spacecan be constructed from those in the other space.Hence wave properties are completely specied ei-ther in ordinary space or in k-space, a simplica-tion that follows from a waves coherent structure.Table I gives some simple examples of FTs.Dirac -function: Since

F 1{F{f (x)}}= f (x), (65)we nd (by writing out the two integral transformson the LHS in two steps) that

(x x)f (x)dx = f (x), (66)where

(x x) = 12 eik (xx

) dk (67)

TABLE I: Some Fourier transforms.Property If f (x) is then F{f (x)} is

f(x) g(k)

e x2 1

2 ek 24

unit step function a 1 2 2k sin kaDerivatives ( ddx )

n f (x) (ik )n g(k)

Derivatives ( ix )n f (x) ( ddk )n g(k)

Translation f (x a) eika

g(k)

Attenuation f (x)e x g(k i )CC b Real g(k) = g(k)

Real and even g(k) = g(k) = g(k)Real and odd g(k) = g(k) = g(k)

a from a to ab Complex conjugation


8/29

8

is called a Dirac -function.Dirac has given a simple intuitive construction of his -function as a rectangular function of width and height 1 / with a unit area under it:

(x x) = lim01/, when |x x| /2,0, otherwise . (68)

5.2 Operators and their eigenfunctions:When a differential operator operates on its eigen-function , it gives a function that is proportionalto the eigenfunction itself:

p eikx =

i

xeikx = k eikx ,

x ei (x/ ) p = i p

eixp/ = x eixp/ . (69)

The proportionality constant is called its eigen-value.Two functions or eigenfunctions can be combinedinto an inner product :

(, ) (x)(x)dx= g (k)g (k)dk.

Dirac denotes this inner product by the bracketsymbol | .

5.3 Dirac notation:

Functions in function space and states of a physi-cal system in quantum mechanics are mathemati-cally similar to vectors in space. Dirac emphasizesthis similarity using a mathematical notation thatworks for vectors, functions and quantum states.A vector (function, state) can be expanded in termsof basis vectors (functions, states). Dirac distin-guishes between two kinds of vectors (functions,states) column vectors (functions, ket states)and row vectors (complex-conjugated functions,bra states), where the name bra and ket comes fromthe two parts of the word bracket:

| = ket = 1|1 + 2|2= 12

= column vector

| = bra = 1 1|+ 2 2|= ( 1 , 2 ) = row vector= | , (70)

For simple vectors, the basis states |1 and |2 are just the unit basis vectors usually denoted e1 ande2 , respectively. The symbol is called an adjointoperation. It combines the matrix operation of

transposition with the complex conjugation of allmatrix elements.The Dirac bracket | is a scalar product of vectors or an inner product of functions

| = ( 1 , 2 ) 12

= 11 + 22 , (71)

where the complex conjugation on the bra com-ponents is needed to keep the self scalar product| nonnegative. For two orthonormal vectorsor states:

i| j = ij . (72)Eigenbras and eigenkets: One goes beyond sim-ple vector spaces by using the eigenfunctions of op-erators such as position x and momentum p. Infact, we immediately associate their eigenfunctionswith abstract eigenstates in the following eigen-bra or eigenket form:

x|x = x|x , and x|x = xx| = x x|, p| p = p| p , and p| p = p p| = p p|, (73)

where the eigenvalues x and p are real, as is true forall physical observables, i.e., properties that can bemeasured experimentally. The eigenvalues x and pnormally range continuously from to .Projection and quantum destructive mea-surement: A simple 2-dimensional vector |A canbe decomposed into its components Ai = i|A withrespect to the basis |i :

|A = |1 A1 + |2 A2=

N

i=1|i i| |A . (74)

In a similar way, an N -dimensional vector space ismade up of the N one-dimensional subspaces pro- jected out by the projection operators

P i = Mi = |i i| (75)in the sense that

P i |A = A i |i (76)leaves only that part of the vector that is in thisith subspace. P i is also called a measurement sym-bol (hence Mi ). This measurement is not a classi-cal nondestructive measurement but a quantum de-structive measurement that destroys all other com-ponents of the original vector |A . The projectioncollapses the state |A into just the part Ai |i inthe ith subspace.Completeness relations: The original vector |Acan be restored by adding up all its projections


9/29

9

Ai |i . Hence an identity operator can be con-structed for this linear vector space

1 =N

i=1|i i| (77)

for discrete basis states that satisfy the orthonor-

mality relation i| j = ij .The idea of completeness or closure holds also foreigenstates of operators such as x and p that havecontinuous eigenvalues:

1 = |x x|dx= | p p|dp. (78)

The resulting basis states |x or | p are said to becontinuous basis states.Representations: A discrete representation is aspecication of a vector (or state) |A by its com-ponents Ai in a discrete basis. A continuous rep-resentation uses the continuous basis states |x togive the components A(x) = x|A in the expansion

|A = dx |x A(x). (79)Wave functions: The (position) wave function of a state or wave | is just its component (x) =x| in the position representation. Its compo-nent ( p) =

p| in the momentum representa-

tion is called its momentum wave function. A wavefunction can be expanded into an arbitrary discreterepresentation of basis states |i

x| (x) =N

i=1

x|i i|

=N

i=1i (x)i , (80)

where i (x) = x|i is the wave function of the basisstate|i . Incidentally, N must be innite to match

the number of points on the straight line.A wave function can also be expanded in anothercontinuous representation, say the momentum rep-resentation:

x| = (x) = dp x| p p|=

1 2

dp ei ( p/ )x ( p). (81)

This is just the inverse Fourier transform relation.

Dirac -function: The equivalence

(x) = (x x)(x)dx = x| = x|x x| dx (82)

shows that

x|x = (x x). (83)It immediately follows that the Dirac -functioncan be represented in innitely many ways. Forexample,

x|x =N

i=1

x|i i|x =N

i=1

i (x)i (x). (84)

Quantum measurement and wave-functioncollapse: Electrons emitted from a source passthrough a single slit and fall on a distant observingscreen on the other side of the slit: As a coherentmatter wave, each electron forms a diffraction pat-tern on the observing screen in the direction y of the diffraction pattern. (Diffraction = spreading of a wave into the geometrical shadow regions.)A detector of acceptance y placed at positiony = d on the observing screen detects electronsfrom signals of their passage through the detec-tor. If (y) = y| is the wave function of self-diffracting electrons normalized in some way, thepart that enters the detector is

(d) = d+ y/ 2

d y/ 2dy|y y| . (85)

The probability of an electron entering the detectoris

P (d) = d+ y/ 2

d y/ 2dy|(y)|2 . (86)

Note that the electron described by the wave func-tion (y) can be anywhere on the screen before thedetection. However, once it is known to have goneinto the detector, e.g., by the detector signal it gen-erates, its wave function has collapsed into (d).This is how a quantum measurement can modifythe state itself.

5.4 Operators and their matrix elements:Matrix mechanicsOperators as matrices: Operators can be visu-alized as matrices because they satisfy matrix op-erations. Hence quantum mechanics is also calledmatrix mechanics . For example, an operator Aacting on a state always gives another state:

A| = c| (87)


10/29

10

in much the same way that a square matrix multi-plying a column vector always gives another columnvector. It can happen occasionally that the resultis actually proportional to the original state |

A|a = a|a . (88)These special states are called the eigenstates of A. The proportionality constants a are called theireigenvalues . Note how the eigenvalue a is used tolabel the eigenstate |a itself. This operator equa-tion is called an eigenvalue equation .Noncommuting operators: Operators, like ma-trices, do not generally commute. For example,

[ p, x] =

i. (89)

In matrix mechanics, commutators like this specifythe fundamental wave nature of the matter wavesdescribed by these operators. It is the starting

point from which other results can be derived. Incontrast, the wave property of matter is describedin wave mechanics by postulating the quantizationrule

p =

i

x, (90)

and the wave function (x) on which p operates.Operator matrix elements: Operators like ma-trices have matrix elements. The matrix elementsof a matrix or operator M can be expressed conve-niently in the Dirac notation as

M ij = i| M | j , (91)a two-sided representation in terms of a suitablebasis |i . A 2 2 matrix or operator can thus bewritten in operator form as

M = M 11 |1 1|+ M 12 |1 2|+ M 21 |2 1|+ M 22 |2 2|. (92)

The ket-bra combinations on the RHS are them-self operators or square matrices. The off-diagonaloperators are the transition operators

| | =12

(1 2 ) =11 1221 22

, (93)

while the diagonal operators (with = ) are projection operators.An operator in the representation of its own eigen-states is a diagonal matrix. Thus

x|x|x = x|x x = x (x x)x| p|x = x|x

id

dx = (x x)

id

dx .(94)

Note how the operator p becomes a diagonal differ-ential operator in the x-representation, and how itcomes out to the right of the Dirac bracket to avoid operating on it. The Hamiltonian operator too hasinteresting diagonal representations:

E |H |E =E |E E = (E E )E , if continuousn

|n E n = nn E n , if discrete.

(95)

Eigenvalue eq. as wave eq.: Matrix mechanicscan be shown to be equivalent to wave mechanics bywriting the operator eigenvalue equation H |E =E |E as a Schr odinger wave equation:

E x|E = x|H |E = dx x|

p2

2m + V (x) x E (x)

=

dx (x x) 12m

id

dx

2

+ V (x)

E (x)=

12m

id

dx

2

+ V (x) E (x). (96)

5.5 Hermitian operators and physical observ-ables:

Hermitian operator: is one for which

H = H T = H (itself) . (97)

It has the important properties that

(a) Eigenvalues are real.

(b) Eigenstates can be orthonormalized.

Physical observables can have only real values.Hence they are described by Hermitian operatorsin quantum mechanics.

Dirac notation: For any operator A:

|A =

(x)[ A](x)dx

= {[ A](x)}(x)dx= A| . (98)

The Hermiticity of an operator H is conrmed bythe equality |H | = H | of the two distinctintegrals involved. Note that if (x) is a columnwave function (called a spinor), the transpositionpart of the adjoint operator will come into play tochange it into a row spinor.


11/29

11

5.6 One-dimensional harmonic oscillator:The 1DHO problem can be solved elegantly in ma-trix mechanics. First write its Hamiltonian

H = p2

2m +

12

m2 x2 (99)

in the compact form

h =H

= 12

q 2 + y2 ,

where y = xx0

,

q = x0

p =

1i

ddy

. (100)

Here the oscillator length x0 = /m is denedby the equation = m2x20 . (101)

The dimensionless position and momentum opera-tors do not commute:

[q, y] = 1/i. (102)

The dimensionless Hamiltonian h can be factorized:

h = aa + 12 ,where a = 1 2 (y + iq )

a = 1 2 (y iq ) (103)are called the step-down and step-up operators, re-spectively. They do not commute

a, a = 1 . (104)

Number operator and its eigenstates: Thenumber operator N = aa is Hermitian. It has realeigenvalues n that can be used to label its eigen-states

N |n = n|n . (105)It then follows that |n is also an energy eigenstate:

H |n = ( n + 12 ) |n . (106)The eigenvalue n of the number operator is a non-negative number or integer 0, 1, 2, ... This im-portant quantization property follows from non-negative nature of its Hamiltonian (meaning thatH 0) and from the nonzero commutators

[ N , a] = a, [ N , a] = a. (107)The commutators show that a and a change thestate |n as follows:

a|n = n + 1 |n + 1 , a|n = n|n 1 . (108)

Oscillator quanta: The quantum number n (qu.no. = any number used to label a quantum state)gives the number of oscillator energy quanta, eachof energy , that can be emitted from that state.Thus a increases the number of energy quanta inthe state by 1, and is called a creation operator,while the destruction operator a decreases n by1. Physical processes that causes the absorptionor emission of energy quanta can be expressed interms of these operators.

The ground state which by denition cannot emitany energy has the qu. no. n = 0. It has azero-point energy / 2 that comes from the wavespreading of its wave function from the classicalpoint x = 0 of stable equilibrium.

Wave functions: The impossibility of energyemission from the ground state is stated by the op-erator equation

a

|0 = 0 . (109)

Its y-representaion gives the rst-order differentialequation for the ground-state (GS) wave function:

y + ddy

0(y) = 0 . (110)

The resulting normalized GS wave function is

0 (y) = 0ey2 / 2 , where 0 =

11/ 4

. (111)

Excited states can be created by adding energyquanta to the GS:

|n =(a)n n! |0 . (112)

Consequently, its wave function can be constructedby successive differentiations of the GS wave func-tion:

n (y) = y|(a)n n! |0

= 0 2n n! y

ddy

n

ey2

/ 2 . (113)

Operators in the number representation: Inthe number representation |n , the only nonzeromatrix elements of a and a lie along the diagonalone line away from the main diagonal of the matrix:

a =

0 1 0 .0 0 2 .0 0 0 .. . . .

, a =0 0 0 . 1 0 0 .0 2 0 .. . . .

. (114)


12/29

12

Hence the dimensionless position and momentumoperators are the square matrices

y = 1 2(a + a) =

1 2

0 1 0 . 1 0 2 .0 2 0 .. . . .

,

q = 1i 2(a a) = 1i 2

0 1 0 . 1 0 2 .0 2 0 .. . . .

.

(115)

5.7 Uncertainty principle:The quintessential wave property described by theHeisenberg uncertainty principle takes on a sur-prisingly elegant form in matrix mechanics: If thesquare uncertainty of the expectation value of aphysical observable A in the quantum state | isdened as

( A)2 = ( A A)2 , (116)then the product of uncertainties of two physicalobservables A and B in the same quantum statesatises the inequality

( A)( B ) 12 [ A, B ] . (117)Simultaneous eigenstates: |a, b exist for com-muting operators A and B :

A|a, b = a|a, b , B |a, b = b|a, b . (118)Then A = 0 = B .

5.8 Classical correspondence:Bohrs correspondence principle: Quantummechanics reproduces classical mechanics in thelimit of large quantum numbers. Example: Theprobability density of quantum state |n of the1DHO agrees with the classical value:

limn

QMn (x)2

CM (x). (119)For a periodic system of period T , the classicalprobability density can be dened in terms of the

probablity dP (x) of nding the classical point masswithin a width dx of the position x during a mo-ment dt in time t :

dP (x) = CM (x)dx = 2T

dt, or

CM (x) = 2v(x)T

, (120)

where v(x) = dx/dt is the velocity at position x.See Fig. 4-18, 19 of Bransden/Joachain for a com-parison between quantum and classical s for the1DHO.

Ehrenfests theorem: The expectation value of a physical observable satises a classical equationof motion. The expectation value of an operatorA(t) in the time-dependent state

|(t) = ei Ht/ |(0) is (121)

A(t) t (t)|A(t)|(t)= (0) |ei Ht/ A(t)ei Ht/ |(0) . (122)

A time differentiation of the three time factors gen-erates three terms:

ddt

A(t) t = A(t)

t t + i

[ H, A(t)] t . (123)

Examples:

ddt

x t = i

[ H, x] t =

1m

p t ;

ddt

p t = i [ H, p] t =

dV dx t

. (124)

VI. Quantum mechanics in three spatialdimensions

Different directions in space are independent of oneanother in quantum mechanics in the sense that

[x i , x j ] = [ pi , pj ] = [x i , pj ] = 0 , if i = j. (125)

QM problems in three-dimensional space can be solvedby using methods of both wave and matrix mechanics.

6.1 Separation of variables in rectangular coor-dinates:

If a Hamiltonian H = H 1 + H 2 + H 3 is separableinto terms referring to different dimensions in rect-angular coordinates, its energy eigenstates can befactorized into the product states |n = n1n2n3 =|n1 |n2 |n3 , one in each spatial dimension:

H |n1n2n3 = ( E 1 + E 2 + E 3)|n1n2n3 . (126)The method works for other operators too:

p2| p1 p2 p3 = ( p21 + p22 + p23)| p1 p2 p3 . (127)Examples:

(a) 3-dim HO: E n = (n +3 / 2), n = n1 + n2 + n3 .

(b) Free particle: p2 = p21 + p22 + p23 in the state

| p1 p2 p3 .(c) Particle in a box of sides L i : The discrete mo-

mentum spectrum is dened by the quantumnumbers n i = ki L i / = 1 , 2,... .


13/29

13

6.2 Separation in spherical coordinates:Kinetic energy: In the spherical coordinates r =(r,, ), p 2 separates into a radial part and an an-gular part:

p 2 = p2r +L2

r 2 ,

where L = r p =e

x e

y e

zx y z px py pz

= Lx ex + Ly ey + Lz ez (128)

is the orbital angular momentum operator. Notethat the position and momentum operators in atypical term of L such as ex y pz commute becausethey refer to different directions in space. However,L is quite a different type of operator from r or p ,because its components L i do not commute withone another:

[L i ,

L j ] = i

Lk , (i , j ,k in RH order) . (129)

On the other hand,

[L2 , L i ] = 0, for any component i. (130)

The two commuting operators L2 , Lz can be usedto dene the simultaneous eigenstate |m :

L2|m = 2( + 1) |m ,Lz |m = m|m , (131)

where and m are called the orbital (angular mo-

mentum) and magnetic quantum numbers, respec-tively. This eigenstate describes completely an an-gular state in spherical coordinates. The resultingangular wave function

|m = Y m (, ) (132)is called a spherical harmonic .In the position represention, the Lz equation canbe written as the differential equation

i

Y m (, ) = mY m (, ). (133)

The solutions in the variable are

Y m (, ) = f ()eim . (134)

These solutions are periodic in with a period of 2:

eim (+2 ) = eim . (135)

Hence m must be an integer (0 , 1, 2,... ). It willturn out that is a nonnegative integer boundedfrom below by |m|.

The kinetic energy operator in spherical coordi-nates is

T = T r + T L = 12m

p2r +L2

r 2, (136)

Central potential: Central potentials are spheri-

cally symmetric and depend only on the radial dis-tance from the center of potential (or force):

V = V (r ) = f (, ). (137)

For central potentials, the Hamiltonian separatesin the radial and angular coordinates

H = ( T r + V ) + T L . (138)

As a result, the wave function of denite energy E factorizes as follows:

Em (r ) = r

|Em = RE (r )Y m (, ). (139)

The radial factor satises the Schr odinger equation

12m

p2r + 1r 2

2( + 1) + V (r ) RE (r )

= ER E (r ), (140)

where

p2r = 22r = 2 1r 2

ddr

r 2 ddr

=

2

1

r

d2

dr2 r. (141)

Note that the wave function RE (r ) depends on as well as the energy eigenvalue E . An en-ergy eigenstate in three-dimensional space has onlythree state labels.Laplace equation: of electrostatics is also im-portant in QM because its solutions describe thequantum states of a static or resting free particle(V = 0 , p = 0 and E = 0):

2(r ) = 0 . (142)

The solutions regular at the origin r = 0 are thesolid spherical harmonics (or harmonic polynomi-als)

Y m (r ) = r Y m (, ). (143)Solutions of the Laplace equations are sometimescalled harmonic functions.The solutions in rectangular coordinates are sim-pler. They are all rectangular polynomials withterms of the form xn 1 yn 2 zn 3 having the same de-gree n = n1 + n2 + n3 . All rectangular polynomials


14/29


15/29

15

Wave functions: are

nm (r ) = Rn (r )Y m (, ). (156)

The radial functions Rn are functions of

= r, where 22

2m = E. (157)

Then

Rn () = un ()

= ev() (158)

can be written in two other forms with each func-tion satisfying a different differential equation:

2r Rn (r ) =

( + 1)r 2

2m 2

e2Gr

+ 2 Rn ,

u n () =( + 1)

2 0

+ 1 un ,

v() + 2( + 1 )v + ( 0 22)v = 0 . (159)where 0 = 2 /a and v() = dv()/d . All solu-tions have

0 = 2 n, giving E n = E 1n2

, (160)

in agreement with the Bohr model.Polynomials v ( ): The factor v() in Eq.(158)is a polynomial of :

(a) v0() = 1 gives solutions with = n

1,

(b) v1() = c(a constant) gives solutions with = n 2,(c) polynomials of degree , namely

v () =

j =0

cj j , (161)

give solutions with = n 1 .Hydrogen states: The properties of the states of the hydrogen atom are summarized in Table III.The spectroscopy notation used is:

[] = s,p,d,f,... for = 0 , 1, 2, 3,... (162)

The spectroscopic symbol for a quantum state is2s +1 []j , giving the spin degeneracy 2 s + 1, theorbital angular momentum [ ] and the total spinquantum number j .

VII. Spin and statisticsThe study of orbital angular momentum states leads

us to intrinsic spins and quantum statistics.

TABLE III: States of the hydrogen atom. The radial wavefunctions are expressed in terms of the dimensionsless distancer = r/a .

n E n m Spect.Symbol Rn (r )

1 E 1 0 0 1s e r

2 E 1 / 4 1 0, 1 2p re r/ 2

0 0 2s (1 r/ 2)e r/ 2

n E 1 /n 2 max = n 1 n[max ] r max e r/n n[]

7.1 Angular momentum:The orbital angular momentum operator

L = r p =e x e y e z

x y z px py pz

= Lx ex + Ly ey + Lz ez (163)

is very different from x and p because its compo-nents do not commute with one another:

[Lx , Ly ] = i Lz , or L L = i L. (164)However, every component L i commutes with

L2 = L2x + L2y + L

2z . (165)

The angular momentum states |m are the simul-taneous eigenstate of the two commuting operatorsL2 , Lz :

L2 |m = 2( + 1) |m ,Lz |m = m |m , (166)

The nature of these states can be studied by usingtheLadder operators: for angular momentum states

L = Lx iLy . (167)

These operators have the structure of

x

iy = r sin ei (168)

that in wave functions have the magnetic quan-tum numbers m = 1. Hence L can be calledthe m = 1 components of L in spherical coor-dinates. Operators and states are both matrices,though states are not square matrices. They canbe multiplied together. In spherical coordinates,the m value of a product of functions or matricesis the sum of the m values of the individual func-tions or matrices. Hence we expect that L|mto be a state of magnetic quantum number m 1,respectively.


16/29

16

Before demonstrating this result, let us rst notehowever that the ladder operators, like their con-stituent rectangular components of Lx , Ly , com-mute with L2 itself. Hence they do not change theorbital quantum number when acting on the state

|m . However, they too do not commute with Lz :[Lz , L] =

L. (169)

These commutators can be used to show explicitlythat L+ is the step-up operator for m, increasingits value by 1, while the step-down operator Ldecreases m by 1:

| = L|m = Am |m 1 . (170)Am can be found by calculating the normalization| using

L2 L2z = L2x + L2y = LL Lz . (171)The result is

L|

m =

( + 1)

m(m

1)

|m

1 . (172)

Possible m values: m2 is bounded because

L2z < L2 , or m2 < ( + 1) . (173)

Hence mmin m mmax . From Eq.(172):L+ |mmax = 0 mmax = ,L|mmin = 0 mmin = . (174)

So the total number of steps N on the ladder is

mmax mmin = 2 = N 1. (175)Assuming that ladders with any number of stepsexist in nature, one nds

Odd N = 2 + 1 = N 1

2 = integer ,

Even N = 2s + 1 s = N 1

2 = half integer .

(176)

The orbital quantum numbers s are integers be-cause they are also the integer degrees of the har-monic polynomials Y m . The half-integer ss arethe spin quantum numbers of an intrinsic spin an-gular momentum S. The word spin is also usedfor any angular momentum.Angular-momentum representations: L2 , Lzand therefore also L,

Lx , Ly are N N matricesfor those states with N steps in their ladder.Example: For spin s = 1 / 2, the ladder has onlytwo steps, with spin states up and down.Then the spin operators are 2 2 matrices: S2 =(3/ 4) 2diag(1 , 1). The Pauli spin matrices arethe components of = 2 S/ :

z =1 00 1

, x =0 11 0 , y =

0 ii 0 . (177)

7.2 Addition of angular momenta:

Product states: Two angular momentum opera-tors L and S are independent if [ L i , S j ] = 0. Thesimultaneous eigenstates of the commuting opera-tors L2 , Lz , S2 , S z are the product states

|msm s = |m |sm s . (178)They have the degeneracy

d(, s) = (2 + 1)(2 s + 1) . (179)

Coupled states: Two independent angular mo-menta L and S can be added to the total angularmomentum:

J = L + S. (180)

The four operators L2 , S2 , J 2 , J z can be shown tocommute among themselves. Hence their simulta-

neous eigenstates |sjm exist:L2|sjm = 2( + 1) |sjm ,S2|sjm = 2s(s + 1) |sjm ,J 2|sjm = 2 j ( j + 1) |sjm ,J z |sjm = m|sjm . (181)

Clebsch-Gordan coefficients: The two distinctways (product or coupled) of specifying the statesof two angular momenta are related by a change of representation:

|sjm = m m s |msm s msm s | |sjm ,|msm s =

j|sjm sjm | |msm s , (182)

where each sum involves only one summation in-dex because m + m s = m. The transformation(Clebsch-Gordan) coefficients are chosen real, andtherefore

sjm |msm s = msm s |sjm . (183)Examples: Two spin 1/2 states can be coupled toa total spin of S = 1 or 0:

|12 12 SM = 11 = |, |12 12 1 1 = |,|12 12 10 = 1 2 |+ ;|12 12 00 = 1 2 |, (184)

where () denotes the spin up (down) magneticstate of the spin 1/2 system.


17/29

17

Vector model: The length J of the vector sumJ = L + S satises a triangle inequality :

|L S | J L + S, but

jmin = |s| j jmax = + s, (185)for quantum angular momenta involves the quan-tum numbers.Construction of total angular momentumstates: The j th ladder of states has 2 j + 1 steps.The topmost state m = + s of the longest ladder jmax = + s is simply

|s + s + s = |ss . (186)Then step down to nd the remaining states on theladder:

J |sjm =

j ( j + 1) m(m 1) |sjm 1

= ( L + S )m m s |msm s

msm s |sjm , (187)where the RHS is a sum over two terms involvingL|m and S |sm s , respectively.The topmost state of the next ladder with j = + s 1 is the remaining or unconstructed statewith m = + s 1. It is constructed by orthonor-malization from |s + s + s 1 , using the prod-uct representation. By convention, its overall phaseis taken to be the Wigner (aka Condon-Shortley)phases. The other states on the ladder are foundby stepping down. This construction method worksfor all the remaining j ladders.

7.3 Identical particles in quantum mechanics:Spin-statistics theorem: Identical particles of integer (half integer) spins satisfy the Bose-Einstein(Fermi-Dirac) statistics. They are called bosons(fermions).Two identical particles are indistinguishable intheir probability densities

(2, 1) = (1, 2), (188)

but their wave functions may differ by a negativesign

(2, 1) = (1, 2) forbosonsfermions . (189)

Boson (Fermion) wave functions are symmetric (an-tisymmetric) in the particle labels.Examples: If , are single-particle (s.p.) states:

(1, 2)[ (1) (2) (2) (1)] . (190)

TABLE IV: electronic structure of the elements. The quan-tum numbers of some atomic ground states are also given asthe spectroscopic symbol 2 S +1 [L]J .

Z Element Spatial Conguration Ground State

1 H 1s 2 S 1 / 22 He (1s)2 1 S 0

3 Li (He)2 s 2 S 1 / 24 Be (He)(2 s)2 1 S 0

5 B (He)(2 s)2 2 p 2 P 1 / 26 C (He)(2 s)2 (2 p)2

7 N (He)(2 s)2 (2 p)3

8 O (He)(2 s)2 (2 p)4

9 F (He)(2 s)2 (2 p)5 2 P 3 / 210 Ne (He)(2 s)2 (2 p)6 1 S 0

If = is the same s.p. state:Bosons : (1, 2) = (1) (2) = 0 , but

Fermions : (1, 2 ) = 0 . (191)

These important results are called:Bose-Einstein condensation: The ground stateof a system of identical bosons is one where everyboson is in the lowest energy single-particle state.Pauli exclusion principle: No two identicalfermions can populate the same single-particlestate.Many identical particles: The wave function of N identical bosons is symmetric under any per-mutation of the N particle labels. For identicalfermions, the wave function changes sign under anodd permutation of the N particle labels, but re-mains unchanged under an even permutation.

7.4 The Periodic Table:Electronic structure of the elements: is sum-marized in Table IV:Helium atom: Its two electrons with spins point-ing and can populate the same 1 s spatial state.They complete the principal qu. no. n = 1 shell .

The shell n can accomodate 2 n2

electrons.To get the lowest energy, both electrons in the Heground state (GS) occupy the lowest s.p. spatialstate nm = 100. The total wave function

0(1, 2) = 100 (r 1 )100 (r 2 ) SM =00 (1, 2)= 0(2, 1) (192)

must be antisymmetric in the two fermion labels 1and 2. Hence the spin function (not a state)for the total intrinsic spin must be the antisymmet-ric function with S = 0.


18/29

18

The atomic Hamiltonian

H (1, 2) = Ze2Gr1

Z e2Gr2

+ e2Gr 12

(193)

gives the GS energy (with Z = 2)

E 0(He) Z 3E R + E 12 109eV + 30eV . (194)where E R = 13 .6 eV is the Rydberg energy.Closed subshells: The ms value of the spinstate |sm s changes when the quantization axiz e zchanges direction. However, if all ms states arelled by 2s + 1 particles, they will remain lledwhen the arbitrarily chosen direction ez changes.Hence the total M S is 0, and this unique state of 2s + 1 particles has S = 0. Similarly, the completelling of the 2 + 1 states of different m gives aunique spherical symmetric spatial system of 2 + 1particles that has L = M L = 0. This is why theground states of He, Be and Ne are all 1S 0 states.

Spin-orbit interaction: The atomic Hamilto-nian contains a relativistic term ALS L S, ALS >0. It gives a negative (attractive) energy of (ALS / 2)(+1) for the s.p. state with the total an-gular momentum j = 1/ 2. For the j = + 1 / 2state on the other hand, the energy is repulsive,namely ( ALS / 2). This explains why the B groundstate in the spatial 2 p conguration has the quan-tum numbers 2 P 1/ 2 of the lowest-energy single elec-tron state outside the closed n = 1 shell and theclosed 2s subshell.The F ground state is one electron short of theclosed n = 2 shell. The lowest energy is obtained by

putting the missing electron or hole in the higher j = 3 / 2 state to give a 2P 3/ 2 ground state, i.e.,leaving a higher state unpopulated by populatingall the lower-energy states in the 2 p subshell.

7.5 Free Fermi gas:Box normalization: For a particle extending overall space, the plane-wave wave functions eik r arenot square-integrable. This feature causes awk-wardness including the appearance of Dirac -functions. The problem can be circumvented bysubdividing space into identical cubes of sides Lsatisfying a periodic boundary condition (BC). The

wave functions in each identical cube can then benormalized.One-dimensional space: The periodic BC

kx (x + L) = eik x (x + L ) = kx (x) = eik x x (195)

gives the discrete spectrum

kx L = 2n x , with nx = 0 , 1, 2,... (196)and the normalized wave functions

kx (x) = 1 L eik x x . (197)

The corresponding states |nx can be labeled by thequ. no. nx . They form a discrete basis for wavefunctions in the representative 1D box.Three-dimensional space: The normalized 3Dwave functions in the representative cube are

r |n = 1L 3 / 2 eik r , where ki = 2L n i , (198)and n = n1n2n3 .Fermi gas: For identical fermions, each spatialstate can be populated at most singly (the spindegeneracy being treated separately). Then n iscalled a (vector) occupation number. A free parti-cle of mass m has energy E k = ( k)2c2 + m2c4 .So the lowest energy state of a free Fermi gas of N fermions in the cube is one where the states of lowest k or occupation number n = |n | are popu-lated. With spatial isotropy (spherical symmetry),the occupation pattern is that of a sphere in n ork space of radius k = kF :

N = dN = d3n = L2 3 d3k=

kF

0

dN dk dk =

L2

3 43 k

3F . (199)

HeredN dk =

L2

34k 2 (200)

is called the density of states in k-space. The radiusof the Fermi sphere in k-space

kF = (6 2# )1/ 3 , where # = N

L3, (201)

is called the Fermi momentum. It does not dependon the size of the cube. For electrons with their spindegeneracy d(s) = 2, the total electron number isN e = 2N .The energy per particle in a free Fermi gas is

E N

= E k gas = dN dk E k dk

dN dk dk

= (3/ 5)F , if nonrelativistic;(3/ 4) ckF , if very relativistic . (202)

The nonrelativistic energy

F = 2k2F

2m (203)

at the surface k = kF of the Fermi sphere is calledthe (NR) Fermi energy.Chandrasekhar limit: A white dwarf star con-tains Z protons, Z electrons and A Z neutrons.The main contributions to its energy are the repul-sive kinetic energy of the highly relativistic elec-tron gas and the attractive gravitational potential


19/29

19

energy of its mass M . Both energies are inverselyproportional to R , the radius of the star, but theyhave different M dependences:

E Tot = bM 4/ 3

R cM 2

R

> 0, unbound ,= 0 , critical ,< 0, collapse.

(204)

The white dwarf will collapse gravitationally if itsmass exceeds the critical mass

M c =bc

3/ 2

. (205)

VIII. Approximation methods for time-independent problems

Few problems in quantum mechanics are exactly solv-able. We rst consider some approximation methods forHamiltonians that are time-independent .

8.1 Time-independent perturbation theory:2 2 matrices: can be diagonalized exactly. TheHermitian matrix

H = 0 bb a , (206)

where both a and b are real, has the eigenvaluesand orthonormal eigenvectors

E 1,2 = 12 (a E ),

E =

a2 + 4 b2 a + 2

b2

a ;

1 = cos sin , 2 = sin cos ,

tan = E 1

b ba

. (207)

Many problems of physical interest involving onlytwo dominant states can be solved approximatelyby this method.Perturbation theory: Let the Hamiltonian H =H 0 + H contains a main part H 0 whose energyeigenvalues E (0)n , eigenfunctions n or eigenstates

|n are already known. If a complicated but weakperturbation H is now added to the system, theeigenvalue equation

( H 0 + H )n = E n n (208)

can be solved by a systematic expansion in powersof that can be set back to its numerical value of 1 at the end of the expansion.This expansion is done in both E n and n :

( H 0 + H ) (0)n + (1)n + ...

= E (0)n + E (1)n + ...

(0)n +

(1)n + ... . (209)

The equation is solved by equating terms of thesame power of on the two sides of the equation:

0 : H 0(0)n = E (0)n

(0)n (0)n = n ,

1 : E (1)n = H nn ,

(1)n = k= n

H knE (0)k

E (0)n

k ,

where H kn = k|H |n . (210)

8.2 Calculation of E (1)n :Many perturbations are made up of sums of fac-torable terms of the type

H = Q(r ) A (211)

involving a radial factor Q(r ) and an angle-spin fac-tor A for a system with intrinsic spin S. Then

E (1)n = Qn

A ,

where Qn = 0 |Rn (r )|2 Q(r )r 2dr, = msm s or sjm. (212)

Examples:

L S = 12 J 2 L2 S2 sjm ,Lz + 2 S z m sm s = ( m + 2 m s ) ,

S z sjm = m 2 j ( j + 1)

J S sjm . (213)

8.3 Variational method:Variational principle: For any wave function :

E = H |H |

| E 0(GS) . (214)

The approximate wave function gives a higherenergy when it contains excited-state componentsof higher energies.Variational wave function: is any wave func-tion carrying a number of parameters b that can bevaried in order to minimize the energy

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

. (215)

The energy minimum E (b0) where dE (b)/db = 0 isthe best variational estimate because it is closest tothe true ground-state (GS) energy E 0 .He atom: The ee repulsion between the two elec-trons in the He atom can be eliminated approxi-mately by reducing the actual nuclear charge Z = 2to the effective charge z 27/ 16.


20/29

20

Hartree mean-eld screening: In an atom withZ electrons, the mutual ee interactions can be re-placed by a mean eld experienced by each electroncharacterized by an effective nuclear charge of

z(r ) = Z, for r 01, for r (216)

8.4 WKB approximation:Since any complex number can be written in theamplitude-phase form Aei , one can look for a so-lution of the 1D Schrodinger wave equation in thisamplitude-phase form. The WKB approximationis a semi-classical result used between the two clas-sical turning points x1 and x2 of the potential V (x)where the local wave number

k(x) = 2m 2 [E V (x)] (217)is real. It is obtained by ignoring the A(x) =d2A(x)/dx 2 term. Then

(x) = A(x)ei (x )

W KB (x) = C

k(x)ei

xx 1

k (x )dx . (218)

WKB quantization: At a turning point wherethe potential has a nite slope, the wave func-tion can penetrate into the classically forbidden re-gion where the local wave number becomes purelyimaginary k i. Under this analytic continua-tion , the amplitude factor becomes complex, A 1/

k e

i/ 4

/ , thereby contributing an addi-tional phase of / 4. The resulting WKB-modiedBohr-Sommerfeld quantization condition is

x2

x 1k(x)dx = n

m4

, n = 1 , 2, ..., (219)

where m is the number of turning points where thepotential is not an wall. The resulting WKBenergies are often quite good. They even agreewith the exact energies in the case of the one-dimensional harmonic oscillator potentials.

IX. Approximation methods for time-dependent problemsTime-dependent problems are more complicated partly

because one has to keep track of the time evolution of astate, but mostly because the Hamiltonian itself mightbe time-dependent.

9.1 The time-dependent two-state problem:is exactly solvable for the time-independent Hamil-tonian (206). Let a state be prepared initially as

| |(t = 0) = d1|1 + d2|2 , (220)

where |i are the time-independent eigenstates of H . Then it evolves at time t into

|(t) = ei Ht/ | = d1|1(t) + d2|2(t)

= d1eiE 1 t/ |1 + d2eiE 2 t/ |2 . (221)Unperturbed basis: It is often useful to express

|(t) in terms of the time-dependent unperturbed

basis states

|1(t) = eiE (0)1 t/ 1

0 ,

|2(t) = eiE (0)2 t/ 0

1 (222)

of energies E (0)1 = 0 and E (0)2 = a:

|(t) = c1(t)|1(t) + c2 (t)|2(t) . (223)The probabilities |ci (t)|2 can be found by using theenergy eigenstates

1 =cos sin , 2 =

sin cos (224)

in Eq.(221):

|c1(t)|2 = d1 cos d2ei t sin 2

,

|c2(t)|2 = d1 sin + d2ei t cos 2

, (225)

where = ( E 2 E 1)/ . These probabilities arein general functions of t. Indeed, their time depen-dences betray the fact that the i are not energyeigenstates, for otherwise they will be stationary

states with time-indepedent probabilites.The expansion (223) is particularly interestingwhen the basis states |i are themselves physicallyobservable, as happens inNeutrino oscillations: Neutrinos come in threedistinct varieties or avors: e , , , associatedwith the electron e, muon and tau , respec-tively. Although they are physically distinct fromone another, they are not mass eigenstates (energyeigenstates at zero momentum). The mass eigen-states are linear combinations of these neutrino ba-sis states of different avors. For this reason, theavor probabilities are not constants of motion, butinstead they oscillate in time.Solar neutrino oscillations: The neutrinos ini-tially produced in the solar interior where massesare converted to energy are e = 1 , and not = 2 , i.e., c2(t = 0) = d1 sin + d2 cos = 0,or d1 = cos , d2 = sin . Allowing only oscil-lation into for simplicity, the probabilities fornding e and at a later time t are

|c1(t)|2 = 1 sin2 2 sin2( t/ 2),|c2(t)|2 = sin 2 2 sin2( t/ 2). (226)


21/29

21

The neutrinos may have very small masses, so that

E i pi c + m2i c3

2 pi. (227)

The the probability for nding e of energy E at adistance L = ct from the Sun is

|c1 (t)|2 = 1 sin2 2 sin2[A( m2c4)L/E ]; m2 = |m21 m22 |,A =

14 c

= 1 .27 GeV(eV) 2km1 . (228)

Experimental data for solar neutrinos of energiesbroadly distributed around a mean value of E =3 MeV give

( m2)c4 = 7 .3 105 (eV) 2 , tan 2 = 0 .41. (229)9.2 Time-dependent perturbation theory:

Expansion in exact eigenstates: Suppose H =H (t), and its energy eigenvalues E j and eigenstates|j are known. Then any arbitrary quantum stateat time t can be expanded in terms of these energyeigenstates:

|(t) =N

j =1

dj |j (t) =j

dj eiE j t/ |j . (230)

Expansion in basis states: This same state canalso be expanded in terms of a known set of basisstates

|n (t) = eiE

(0)n

|n that are eigenstates of

a known Hamiltonian H 0 of energies E (0)n :

|(t) =n

cn (t)|n (t) =n

cn (t)eiE (0)n t/ |n .(231)

It happens occasionally that both | and | statesare physically observable. Then it is of interest toexpress one observational probabilities in terms of the other expansion coefficients. For example,

|cn (t)|2 =N

j =1U nj dj eiE j t/

2

, (232)

is a generalization of what is done for neutrino os-cillations to higher dimensions. Here U , with

U nj = n |j , (233)is an N N unitary matrix if both sets of states arenormalized. U is then made up of the eigenstates|j stored columnwise.TD perturbation: The expansion (231) can alsobe used if H = H 0 + H contains a time-dependent

(TD) perturbation H . Unfortunately, the TD ex-pansion coefficients cn (t) must be obtained morelaboriously by actually solving the TDSchEq

i t |(t) = ( H 0 + H )|(t) . (234)

The resulting differential equations (DEs) of mo-

tion for ck (t),i cb(t) =

k

H bk ei bk t ck (t), (235)

are a set of coupled rst-order DEs that can besolved easily in a PC. Here bk = ( E

(0)b E

(0)k )/ .

Systematic TD perturbation theory: is con-cerned with the Taylor expansion in (eventuallyset to 1):

cb(t) = c(0)b (t) + c

(1)b + ... (236)

Eq.(235) can then be separated term by term intoseparate equations for c( n )b :

0 terms : i c(0)b (t) = 0 c(0)b (t) = const ,

1 terms : i c(1)b (t) =k

H bk ei bk t c(0)k ,... (237)

These rst-order DEs in time can be integrated di-rectly, making it easy to calculate rst-order wavefunctions in TD perturbation theory.Sinusoidal perturbation: Let the sinusoidal per-turbation

H = V cost, 0 t t0 only, (238)be turned on for a nite period of time. After theperturbation has ceased, a state initially in |awill nd itself in the state |b , b = a, with theprobability

c(1)b (t t0)2

V ba2

2

j 20 (z)t20 2

, (239)

where z = 12 (ba )t0 , j0(z) = sin( z)/z,

ba = [E (0)

b E (0)a ]/ . (240)

Thus most of the excitations are close to the twostates |b with energies E

(0)b = E

(0)a .

Golden rule: The transition rate during the in-teraction lasting a time t0 from a state a to a groupof states b is thus

W ba 1t0 |c(1)b |2bd( ba )

2

V ba2

2

b, (241)


22/29

22

where b = dN b/dE b is the density of nal states.Example: The Bohr transition of an atomic elec-tron from a higher atomic state b to a lower statea with the emission of a photon of energy =E (0)b E

(0)a is caused by the electric dipole interac-

tion

H = V cos(t). (242)

Here V = D E 0 is the electric-dipole interactionbetween the electric dipole operator D = q r of theatomic electron and the electric eld E = E 0 of the emitted photon. The latter can be obtainedfrom the photon energy density in a cube of side L:

L3

= 0E 20

2 , (243)

where 0 is the permittivity of free space. WhenH is used in the Golden Rule, the resulting transi-tion rate is the Einstein coefficient for spontaneousphoton emission by the atom:

A = W ba = 43

3

c3 140 |D ba |

2 , (244)

9.3 Sudden, adiabatic and impulsive perturba-tions:In sudden perturbation, the Hamiltonianchanges suddenly by a nite amount to the nalvalue H a . If the change is so sudden that the state

| at t = 0 has not changed, the state at timest > 0 is just

|(t) = ei H a t/

|

=n

eiE ( a )n t/ |n n |=

n|n (t) n | . (245)

Adiabatic perturbation: When a time-dependent Hamiltonian H (t) changes very slowlyin time, we expect the instantaneous eigenvaluesE ( t )n (t) and eigenstates |

( t )n dened by the eigen-value equation

H (t)|( t )n = E ( t )n (t)|( t )n (246)to give a good approximation. That is

E n (t) E ( t )n (t), (247)a result known as the adiabatic theorem.If one watches the system for a sufficiently longtime, however, sooner or later additional featureswill appear. They can be described convenientlyby using the time-dependent basis |

( t )k :

|(t) =k

ck (t)e( i/ ) t0 E

( t )k (t

)dt |( t )k . (248)

The new features include the appearance of a ge-ometrical phase geon and of transitions to otherstates k = n.Impulsive perturbation: An impulsive pertur-bation contains a -function in time:

H (x, t ) = A(x) (t). (249)

The TDSchEQ can be solved exactly to show thatthe -function makes the state discontinuous at t =0:

|(0+ ) = 1 A

2i 1

1 +A

2i |(0) , (250)

X. ApplicationsIn this chapter, we see how quantum problems of great

physical interest have been solved exactly or approxi-mately.

10.1 Solving special matrix eigenvalue problems:

Special matrices of arbitrarily large dimensions of great physical interest can sometimes be solved an-alytically by symmetry considerations.

To begin, it is easy to verify by inspection the eigen-values and eigenvectors of the Pauli matrix x :

0 11 0

11 =

11 ,

0 11 0

1

1 =

1

1 . (251)

Indeed, three simple rules often go a long way innding the eigenvectors and eigenvalues of certainHermitian matrices:

(a) The vector v1 = (1 , 1,..., 1), where all compo-nents are 1, is an eigenvector of a matrix wherethe matrix elements of each row is a permu-tation of those of the rst row. Its eigenvalueE 1 is just the sum of the matrix elements of any row.

(b) Eigenvectors of a Hermitian matrix can be

made othogonal to one another.(c) The sum of eigenvalues is equal to the trace of

the matrix.

Example 1: The vector v1 = (1 , 1, 1) is an eigen-vector of the 3 3 matrix ((0,1,1),(1,0,1),(1,1,0)),with eigenvalue E 1 = 2. The three vectors v2 =(2, 1, 1), v3 = (1, 2, 1), v4 = (1, 1, 2) areall orthogonal to v1 , but not linearly independentof one another. Each has the eigenvalue -1. Theorthogonal eigenvectors can be taken to be v 1 , v2 ,and v 5 = ( v3 v4)/ 3 = (0 , 1, 1).


23/29

23

Example 2: The vectors v1 , v2 and v5 are alsothe orthogonal eigenvectors of the Hermitian ma-trix ((1,1,1),(1,1,1),(1,1,1)), with the eigenvalues 3,0, and 0, respectively. This is a special case of thegeneral result that two matrices M 1 and M 2 havecommon eigenvectors if

M 2 = M 1 + aI, (252)

where I is the identity matrix. Since the parame-ter a can be complex, we see that non-Hermitianmatrices can also have orthogonal eigenvectors.Example 3: The arbitrarily large N N matrixwhose matrix elements are all 1s has one collec-tive eigenvector v1 = (1 , 1,..., 1) with eigenvalueN . The remaining eigenvalues are all 0. The or-thogonal eigenvectors can be chosen to be v2 =(N 1, 1, 1,..., 1), v3 = (0 , N 2, 1,..., 1),..., v N = (0 , 0, ..., 0, 1, 1).BCS theory of superconductivity: can be un-

derstood conceptually by using a large dimensionalN N matrix Hamiltonian with the same smallnegative matrix element everywhere. All eigen-states have energy 0, except the collective eigen-state v1 = (1 , 1,..., 1) that has energy = N .The energy gap that separates the ground statefrom the excited states makes it hard for the sys-tem to be excited. If the ground state containscurrent-carrying electrons, then at sufficiently lowtemperatures (below a certain critical temperatureT c), these electrons cannot lose energy by inelasticcollisions with the crystal lattice in the conductor.When this happens, both dissipation and resistiv-

ity vanish, and the medium becomes superconduct-ing. (BCS = Bardeen, Cooper and Schrieffer whodiscovered this fundamental theory of superconduc-tivity.)

10.2 Stark effect in an external electric eld:A charge q moved a distance z against a constantexternal electric eld E = E ez has the (dipole) in-teraction energy

E = q E.dr = q E z. (253)The resulting perturbing Hamiltonian is

H = E D z , where D = q r (254)is called the electric dipole operator of the charge.This operator is odd in parity, changing sign whenz z. Hence all rst-order energies vanish:

E (1)n = n|H |n = 0 . (255)Since z = r cos is independent of the azimuthalangle , D z has magnetic quantum number m = 0.

Hence H has nonzero matrix elements only be-tween two opposite-parity states i, j with the samemagnetic quantum number: mi = m j .Example: In the n = 2 shell of the hydrogen atom,there are four states: nm = 200; 210, 21 1. H connects only the two opposite-parity m = 0 states.For these two states alone, it takes the form of a2

2 matrix b((01) , (1, 0)) whose eigenstates are

E 1,2 = b : 1,2 = 1 2 (200 210 ) . (256)

Their energies E 1,2 thus move away from the undis-turbed energy of the two states with m = 1 by anamount proportional to b, which is proportional toE . [The state |m has parity ( 1).]

10.3 Aharonov-Bohm effect:Gauge transformation: In classical electromag-netism (EM) the electric and magnetic elds areuniquely dened, but the EM potential (the scalarpotential and vector potential A ) are not unique.They can change by a gauge transformation

A A = A + , =

t

, (257)

where is any scalar eld, without changing theEM elds

B = A ,E =

At

. (258)

Relativistic electrodynamics: The basic dy-namical variable in classical mechanics is the 4-momentum ( p ,iE/c ). To add EM, one needs anEM 4-vector. There is a unique qualifying candi-date: the EM 4-potential ( A , i/c ). Hence electro-dynamics for a particle of mass m and charge q canbe built up from the sum ( p q A , i(E q)/c ).This combination transforms like a 4-vector un-der Lorentz transformations, thus guaranteeing thecorrect result in different inertial frames. As a re-sult, the EM 4-potential plays a more fundamentalrole in quantum mechanics than the EM elds.The energy-momentum relation for relativistic elec-

trodynamics is then(E q)2 = ( p q A )2c2 + m2c4 . (259)

One can nd from this the nonrelativistic kineticenergy T NR :

T E mc2

T NR = q + 12m

(p q A )2 . (260)Quantum electrodynamics: We are now in aposition to quantize the mechanical terms in NR


24/29

24

electrodynamics, leaving the EM potential unquan-tized for simplicity:

H = 12m

(p q A )2 + V + q. (261)

Aharonov-Bohm phase: The EM interaction re-sides primarily in the phase (or gauge) of quantumwave functions. For the special case where = 0,this important result can be demonstrated readilyby showing that the wave function in the pres-ence of the vector potential A differs from the wavefunction without A by an A -dependent phase g:

(r ) = eig (r ). (262)

If true, the phase factor eig serves the purpose of removing the A dependence from the Schrodingerequation when moved to the left of a gauged fac-tor:

i q A eig (r ) = eig i (r ). (263)

Since

i eig (r ) = eig ( g) +

i (r ), (264)

g has to satisfy the differential equation

g = q A , or dg = q A dr . (265)Hence

g = q

r

r 0A (r ) dr , (266)

where the line integral is in general path-dependent .The integral is then said to be non-integrable . Thisnon-integrability can be made explicit by integrat-ing around a closed circuit c once and simplifyingthe result with the help of Stokess theorem:

g = q

c A (r ) dr =

q

S

(

A )

dS =

q , (267)

where the total magnetic ux across the surfaceS enclosed by the circuit c depends on the magneticeld A on and inside c. Such a dependence onproperties of the system outside the path c showswhy quantum mechanics in particular, and wavesmore generally, describe nonlocal phenomena thatare spread out in space.

Magnetic ux quantization: Suppose the pathc is located in a region free of any EM eld, thenthe wave function must return to its original value

after one complete circuit. Hence its Aharonov-Bohm phase can only be an integral multiple of 2 :

g = 2n = q

, or = n0 . (268)

As a result, the magnetic ux enclosed by c canonly exist in integral multiples of a quantized unitux

0 = h/q . Experimental measurement deter-

mines that q = 2e in superconductors, thus show-ing that the super current is carried by pairs of electrons, now called Cooper pairs.Gauge interactions: Other fundamental inter-actions also work through the phase of quantumwave functions. Hence they are called gauge inter-actions. Indeed, the information contained in thephase of a wave function is usually more importantthan that contained in its amplitude. A picture of Snow White can be made to appear in our eye asa coherent wave. One can take the phase informa-tion from Snow Whites wave function and use it

with the amplitude part of the wave function froma picture of Happy, one of the seven dwarfs. Acomposite wave like this has been constructed the-oretically. Believe it or not, one sees Snow Whitein the resulting hybrid picture, and not the dwarf.

10.4 Magnetic resonance:Rabi measured the magnetic moments of atomicstates by nding the Zeeman splitting of the en-ergies of different magnetic substates by resonance matching . Atomic magnetic moments are tradition-ally expressed as gB in units of the Bohr magnetonB = e / 2m, the theoretically value g being called

the Lande g factor. The Hamiltonian involved isH = H 0 + H , (269)

where H 0 = H 1 + H LS + H 2 contains the usualNR Hamiltonian H 1 , the relativistic spin-orbit termH LS , and the Zeeman Hamiltonian

H 2 = M B z=

gB Bz

J z = 0 J z (270)

that gives the splitting between the magnetic sub-states of the same j . Finally a small oscillatorymagnetic eld is applied in the x-direction to givea perturbing Hamiltonian

H = x J x cos(t), where x = gB Bx . (271)

This perturbation causes transitions between themagnetic substates. Any quantum state

|(t) =m

cm (t)|m (t)=

mcm (t)ei m t |m , (272)


25/29

25

where m = E (0)m / , can be expressed in term of the

unperturbed states |m (t) . The time-dependentexpansion coefficient cm (t) then satises the TD-SchEq

i cm (t) =k

H mk ei mk t ck (t), (273)

where mk = m k .Example: The problem can be solved approxi-mately for cm (t) for the atomic n[]j = 2 p1/ 2 con-guration that has only two magnetic substates of energies E (0) = 0/ 2. For example, if the sys-tem is initially in the upper state ( c+ = 1 , c = 0),then the probability of nding the system in theupper and lower states at a later time t are respec-tively:

P ++ = |c+ |2 = cos 2(R t/ 2) + sin 2 sin2(R t/ 2),P + = |c|2 = cos 2 sin2(R t/ 2), where (274)

R = ( )2 + ( x / 2)2 , = 0 ,sin = R

, cos = x / 2

R. (275)

Magnetic resonance: As the applied frequency passes the Zeeman frequency 0 , the minimumpopulation of the + state falls sharply to 0 at 0and then rises sharply back up to 1 again, thusgiving a clear signal for the measurement of 0 .The third frequency x controls the period of oscillation of the population between the two mag-netic substates. At full resonance ( = 0), onends = 4/ x .

XI. Scattering theoryInformation about the dynamical properties of micro-

scopic systems like atoms, nuclei and particles can beobtained by scattering beams of projectiles from targetscontaining them, as rst demonstrated by Rutherford,Geiger and Marsden who elucidated atomic structure byscattering particles from a gold foil.

11.1 The scattering cross sections:The differential cross section

dd

= 1L

dRd

(276)

is the angular distribution dR/d of the reactionrate Rper unit luminosity L. L itself is the productIN T , where I is the ux or current density of theincident beam (number of incoming particles persecond per cross sectional area of the beam, or N i /afor a beam of uniform cross section a) and N T isthe number of target particles illuminated by thisbeam. L depends on both beam and target. d

is the differential solid angle in a suitable inertialframe.Example: Two beams of the same cross sectionalarea A collide head-on where their paths cross eachother. Each beam is made up of bunches of N i (i =1, 2) particles per bunch. The target can be takento be one bunch of N 2 particles in beam 2. Thenthe effective incident ux is I = f N

1/A , where f is

the frequency of collision between bunches.Rutherford cross section: For the scattering of -particles of charge Z 1 = 2 from an atomic nu-cleus of charge Z 2 , Rutherford found from classicalmechanics that in the CM frame

dd

=d4

2 1sin4(/ 2)

,

where d = Z 1Z 2e2G

T CM , (277)

is the distance of closest approach (or turning

point) for head-on collision at the NR kinetic en-ergy T CM in the CM frame. Note that for backscat-tering = , the Rutherford differential cross sec-tion d2 / 16 gives a direct measurement of the posi-tion d of the classical turning point.Total cross section: is proportional to the totalreaction rate:

= d = ddd = dRL = RL . (278)11.2 Quantum theory of scattering in a nutshell:

Partial-wave expansion: of a plane wave intospherical waves of good angular momentum around the origin of coordinates has the Rayleighform:

k (r ) = eik r =

=0

i(2 + 1) P (cos ) j(kr ),

r

=0

i(2 + 1) P (cos ) i2k

ei (kr / 2)

r ei (kr / 2)

r (279)

This is the time-independent wave function of a sta-tionary scattering state. The time-dependent wavefunction carries an additional time factor eit ,where = k2 / 2m. The additional time factorshows that the wave function eikr /r describes aningoing spherical wave collapsing towards the ori-gin, while the wave function eikr /r describes anoutgoing spherical wave expanding out from the ori-gin.Scattered wave: When a target particle is placedat the origin of coordinates, the radial part j of the


26/29

26

wave function (279) will be replaced by the func-tion R(r ) from the SchEq with a potential in it. Atlarge distances r , the ingoing waves are still collaps-ing towards the origin, and therefore do not know if a potential is present there. In elastic scatterings,an outgoing spherical wave has the same normal-ization as before, and can be changed at most by aphase, here taken to be 2 , that resides in a phasefactor S = e2i called an S -matrix element:

(r ) r

=0

i(2 + 1) P (cos ) i2k

ei(kr / 2)

r S ei (kr / 2)

r= k (r ) + sc (r ) where (280)

sc (r ) r

=0

i(2 + 1) P (cos )

S 1

2ikei (kr / 2)

r

= f ()eikr

r . (281)

is the scattered wave function, the additional wavefunction generated by the potential. Containingonly outgoing spherical waves, it can be writtenasymptotically ( r ) in terms of an angle-dependent quantity called the scattering amplitude

f () =

=0

(2 + 1) f P (cos ),

where f = S 1

2ik =

1k

ei sin (282)

if the scattering is elastic, meaning that the phaseshift is real.

Flux: or probability current density is

j =

2im ( ) jk + jsc , (283)

where jk = k/m is the ux of the plane wave, and

jsc = km

e r |f ()|2r 2 (284)is the ux of the scattered wave at large distances.

Differential cross section: A detector of areadA = r 2d then measures the differential cross sec-tion

dd

= 1Lk

dRd

= 1 k/m

( jsc e r ) dAd

= |f ()|2 . (285)

Total cross section: is the result integrated overthe solid angle d:

tot = dd d = f ()f ()d=

tot , , (286)

where the two sums over , one from each f , hasbeen simplied to only one sum by using the or-thogonality relation for Legendre polynomilas

1

1P (cos )P (cos )d =

42 + 1

, and (287)

tot , = 4k2

(2 + 1) sin 2 . (288)

Optical theorem: If is real, one nds that

Imf ( = 0) =

(2 + 1)Im f =

(2 + 1)sin2

k

= k4

tot . (289)

This theorem holds even when the phase shift = + i becomes complex.

11.3 Breit-Wigner resonance:The partial-wave total cross section =, max sin2 reaches a maximum value of , max =(4/k 2)(2 + 1) whenever (k) = / 2. This maxi-mum value is called the unitarity limit.

If (E ) rises through (E = E R ) = / 2 sifficientlyrapidly, a Taylor expansion about E = E R needsto be taken only to the terms

2 + ( E E R )

d dE E R

. (290)

Then tan / 2

E E R, where

2

= 1

(d /dE )E R> 0. (291)

Resonance: In the neighborhood of a resonance,

the partial-wave total cross section has the simpleE -dependence:

= , maxtan 2

1 + tan 2

= , max(/ 2)2

(E E R )2 + ( / 2)2. (292)

This function has a sharp maximum at E = E R andfalls rapidly to half its maximal value at E E R =/ 2. Hence is called the resonance width (orfull width at half maximum, FWHM).


27/29

27

Decaying state: The partial-wave scattering am-plitude near resonance has the simple form

f = 1k

ei sin = 1k

1cot i

= 1k

/ 2(E E R ) + i/ 2

. (293)

At the complex energy E = E R i/ 2, the time-dependent probability density decays exponentially|(r , t )|2 = |(r )ei(E R i / 2) t/ |2

= ||2e t/ . (294)

11.4 Yukawas theory of interactions:Yukawa showed in 1935 that both the Coulombinteraction mediated by the exchange of masslessphotons between charges, and the strong or nuclearinteraction mediated by the exchange of massivebosons called mesons, can be

chun wa wong quantum mech rvw 2

Documents