8.

04 Quantum Physics

Lecture VII

Double slit: mathematical model of interference pattern and photon scattering

To develop some more insight into interference, and the correlations between quantum system and (classical) apparatus that lie at the heart of the quantum measurement problem, we will postulate rules on how the interference pattern is formed, and how photon scattering changes the electrons wavefunction.

Figure I: Double-slit interference. Rule 1. Wavefunction at slits 1, 2 is: 1 = Aei1 2 = Aei2 (7-1)

Complex numbers A real. For incident plane wave, and no observation: 1 = 2 Rule 2. The wavefunction at the detection point D is given by D = 1 ei2L1 / + 2 ei2L2 / D = 1 e
ikL1

(7-2)

(7-3) (7-4) VII-1

+ 2 e

ikL2

Massachusetts Institute of Technology

8.04 Quantum Physics where is the de Broglie wavelength and k = particle.

2

Lecture VII the associated wave vector of the

Rule 3. The intensity (and therefore particle arrival rate) at the detection point is proportional to: 2 |D |2 = 1 eikL1 + 2 eikL2 2 = |A|2 ei(kL1 +1 ) + ei(kL2 +2 ) = |A|2 ei(kL1 +1 ) + ei(kL2 +2 ) ei(kL1 +1 ) + ei(kL2 +2 ) = |A|2 1 + 1 + ei(kL1 +1 kL2 +2 ) + ei(kL1 +1 kL2 +2 ) = |A|2 {2 + 2 cos(kL1 + 1 kL2 2 )} = 2|A|2 { 1 + cos(k (L1 L2 ) + (1 2 ))} (7-5)
background term interference term

Changing the relative phase 1 2 of the wavefunction at the two slits acts to shift the interference pattern in the detector plane. Changing 1 2 by is equivalent to changing L1 L2 by , i.e. to an angle change. If we average over random relative 2 phases 1 2 of the wavefunctions at the two slits, the interference pattern and the interference term disappear.

Figure II: Average over many interference patterns with random shifts. Rule 4. If an electron at position x scatters a photon with incident wavevector kin into a new direction characterized by a photon wavevector kout (pin = h kin , pout = h kout incident, scattered photon momenta) its wavefunction at position x acquires a phase shift sc = (kin kout ) x = QM statement about momentum conservation Wavefunctions at slit 1,2 before scattering: 1 = Aei1 = A after scattering kin to kout : 1 = Aei(kin kout )x1 at slit 1 Massachusetts Institute of Technology 2 = Aei(kin kout )x2 at slit 2 (7-7) (7-8) VII-2 2 = Aei2 = A (7-6)

8.04 Quantum Physics

Lecture VII

Figure III: Electron passing through double slit scattering a photon. Interpretation. Momentum transfer onto electron. A photon scattering event into a given direction does not destroy or collapse the wavefunction. It merely shifts its phase. Repeating the previous calculation, the intensity in the detector plane is now:
2 |D | = 2|A|2 {1 + cos[(k (L1 L2 ) + (kin kout )(x1 x2 )]}

(7-9)

Note. Do not confuse k wavevector of the particle kin ,kout wavevector of incident, scattered photon. For a given single scattering event characterized buy outgoing photon wavevector kout there is an associated (and perfectly well dened) shift of the electron interference pattern in the detector plane. The momentum h kout (direction kout ) of outgoing photons is perfectly correlated with the shift of the electron interference pattern in the detector plane.

Reconstructing the interference pattern from noise via correlations (conditional interference)
If we repeat the measurement many times, each one will have a dierent scattered photon direction kout , a dierent corresponding phase shift (kin kout ) (x1 x2 ), for the electron interference pattern, and the electron pattern on the screen will not show fringes. Massachusetts Institute of Technology VII-3

8.04 Quantum Physics

Lecture VII

Figure IV: Averaging over interference patterns with random phases destroys interference. Figure V: A subset of scattering events into a single angle, as recorded by a detector, shows interference. However, if we record the photons and select only those events where a photon was recorded in a given, predetermined direction, this subset of electron arrival position will slow perfect interference with a shift determined by the chosen detector direction (direction kout ). The underlying interference is always there, it is the averaging over dierent phases of the interference pattern, or the correlated photon directions, that causes the intergerence pattern to wash out (disappear). The above argument still holds if the photons y o into vacuum, and the electrons are detected rst. If we average over all photon directions, there is no electron intergerence pattern. However, if we post-select only those electron arrivals that correspond to a photon observation in a certain direction (even if that photon observation is performed after the electron arrival on the screen) an interference pattern with the predicted shift is observed. Rule 5. A measurement is the interaction of our detector system (interaction here: electron scatters photon) and the averaging over many states of the detector (here: scattered photon directions). The averaging over the detector states is crucial. Before that averaging, we have a correlated (entangled) system, where the states of the quantum system (here: phase of the electron wavefunction, or position of the interference pattern) are correlated with the states of the detector: If the photon was scattered into direction 1, then the interference pattern is at position 1, and if photon was scattered into direction 2, then the interference pattern is at position 2, and if . . . would describe the entangled state before averaging. Massachusetts Institute of Technology VII-4

8.04 Quantum Physics

Lecture VII

Note that if the wavelength of the light is too long to optically resolve the double slit (photon x2 x1 = d), then even the maximum phase shift (obtained for 4 kout = kin ; i.e. (kin kout = 2kin ) of magnitude 2kin (x2 x1 ) = photon (x2 x1 ) 1, and the interference pattern continues to exist even when we average over all scattered-photon angles. If, on the other hand, the double slit is observed with a microscope with sucient resolution (i.e., using suciently short photon wavelength), then the microscopes objective collects photons scattered into dierent angles, and the averaging over kout , and the corresponding electron phases (kin kout ) (x2 x1 ), washes out the electron interference pattern.

Figure VI: A microscope resolving which slit the electron passed through uses interference of the light scattered at dierent angles to image the electron. These angles correspond to dierent shifts of the electron interference pattern in the detector plane, i.e. the electron interference is washed out.

Atom model: Spectra and quantization of energy

Thomson model
In order to have an elastically bound electron that radiates monochromatic radiation, J.J. Thomson constructed a plum-pudding model of the atom where pointlike electrons are conned within a uniform positive charge distribution of radius1 A. Force on electron at x:
x |qQ| R |qQ| 3 F = = x 2 40 x 40 R3 F |qQ| x = = x m 40 R3 m
3

(7-10) (7-11) 1 2 (7-12)

x = x For hydrogen:

|qQ| = 40 R3 m

Massachusetts Institute of Technology

VII-5

8.04 Quantum Physics Q = +qe = 1.6 1019 C, q = qe m = me = 9 1031 kg R = 1010 m = = 6 1015 =

c c = 2 = 150 nm 1 s

Lecture VII

The Thomson model yields a harmonically oscillating electron and the correct order of magnitude for the wavelength (shortest H wavelength Lyman , = 121 nm), but cannot explain the other spectral lines.

Rutherford scattering
Source. Marsden & Geiger 1906-, Rutherford 1911 Discovery of the nucleus through large-angle scattering. Marsden & Geiger mea-

Figure VII: Thomson plumpudding model of the atom with a negatively charged electron imbedded into a uniformly distributed positive nuclear charge. sured the angular distribution of scattered -particles. According to the Thomson model, the positive charge is distributed evenly throughout the atom, so it should cause little deection when the atoms are arranged in a solid. However, the observed distribution is dramatically dierent at large scattering angle To observe large scattering angles, we need to scatter o massive particles so that momentum conservation allows large angles. Observations can be quantitatively explained by assuming that the mass of the nucleus is concentrated in a small volume. For quantitative description of observed scattering angle dependence we need the concept of cross section. Massachusetts Institute of Technology VII-6

8.04 Quantum Physics

Lecture VII

Figure VIII: Marsden and Geiger experiment on particle scattering; explanation of results by Rutherford.

Figure IX: Observed angular dependence of -scattering by gold foil, and prediction from the Thomson model.

Inadequacy of Thomson model The electrons are 4 1850 7000 times lighter than an -particle, conservation of me momentum then results in a small deection angle e m 104 per -electron p collision. We expect a diusion-type process for many scattering events (random walk in angle), this should result in a Gaussian distribution of scattering angles of width = M e , where M is the average number of -particle scattering events. So we Massachusetts Institute of Technology VII-7

8.04 Quantum Physics expect for the fractional scattering into a particular angle :
2 2 N () 2 2 2 2 M e = Ae 2( ) = Ae = Ae N

Lecture VII

(7-13)

This would be the expected angular dependence of scattering events for Thomson model (only -electron scattering causes deection of -particle). However, the observed angular distribution has a much longer tail at large scattering angles (see gure above).

Figure X: Scattering of the heavy particle o an electron causes only a small deection angle e , and for many scattering events a Gaussian distribution of scattering angles for the particle.

Scattering problems and cross section

Figure XI: Scattering o a potential. Problem is solved if outgoing angles , , can be calculated as function of impact parameter b and incident angle i = (b, i ) = (b, i ) For spherically sysmmetric potentials, V = V (r), scattering is independent of : = i and problem solved if we know = (b) , i.e. if we can calculate the scattering angle as a function of impact parameter b. Massachusetts Institute of Technology VII-8

8.04 Quantum Physics

Lecture VII

For Coloumb scattering of a particle with charge Z q o a scattering center corresponding to a charge Zq , one can derive (see, e.g., Goldstein, Classical Mechanics) cot 80 = bE 2 ZZ q 2 (7-14)

where E is the particle kinetic energy.

Figure XII: Coulomb scattering. However, since the impact parameter b is not observable in a typical scattering experiment, we need a formalism to average over all impact parameters. We dene the total cross section, tot as the ratio of the total scattering rate R1 to the incident intensity (for one scattering center) tot = R1 [particles/s] = [m2 ] I [particles/(m2 s)] (7-15)

At density n, the number of atoms inside the volume A l is N = nAl. Then in

Figure XIII: Scatterers inside a volume. Massachusetts Institute of Technology VII-9

8.04 Quantum Physics

Lecture VII

the weak-scattering limit N tot A the total scattering rate is proportional to the incident rate Rin RN = N tot I = ntot lIA = ntot lRin (7-16) and the fraction of scattered particles is RN N tot = ntot l = A Rin (7-17)

The fraction of the particles removed from the beam is simply the total cross section N tot for N particles, divided by the beam area. To describe the angular dependence of scattering, we slightly generalize the cross d section concept. We dene the dierential cross section d (, ) as the ratio of the dR scattering rate per solid angle d to the incident intensity I (again for one scattering center). d dRi /d [particles/(ssterad)] m2 = = (7-18) d I [particles/(m2 s)] sterad For spherically symmetric potentials, the dierential cross sectio depends only on the angle . A detector of area A at distance R from the scatterer subtends an angle: d A = 4 4R2 or d = (7-19)

A (7-20) R2 Particles with impact parameters between b and b + db scatter with angle between

Figure XIV: Scattering into a solid angle d and detector area A. and + d corresponding to a solid angle d = 2 sin d. Thus the cross-sectional Massachusetts Institute of Technology VII-10

8.04 Quantum Physics

Lecture VII

Figure XV: Dierential scattering cross section area d = 2bdb corresponds to the solid angle d = 2 sin d, and the dierential cross section can be written as: d 2bdb = (7-21) d 2 sin d To eliminate the dependence on the impact parameter b (a quantity not directly 80 observable in the experiment), we use the relation cot 2 = ZZ q 2 bE from Eq. 7-14 between and b, and dierentiate: 80 E cos /2 db = d (7-22) ZZ q 2 sin /2 1 sin2 /2 1 cos2 /2 2 = 2 d (7-23) sin2 /2 d = (7-24) 2 sin2 /2
d Substitution of b and db into the expression for d yields 2 d ZZ q 2 1 1 = cot 2 d 80 E 2 2 sin /2 sin d 2 2 ZZ q 1 = 80 E 4 sin4 /2

(7-25) (7-26)

Dierential cross section for scattering Z q o Zq : d Z 2 Z 2 q 4 1 = 2 2 2 d 256 0 E sin4

2

(7-27)

Setting Z = 2 for the He-4 particle we arrive at the Rutherford scattering formula for -particles d Z 2q4 1 () = (7-28) 2 2 2 d 64 0 E sin4 2 To obtain scattering rate into detector at xed , calculate solid angle that detector occupies and use d RN () = N ()I (7-29) d Massachusetts Institute of Technology VII-11

8.04 Quantum Physics N : # of scatterers : detector solid angle I : incident intensity

Lecture VII

The Rutherford scattering formula displays strong suppression for large scattering angles, but still predicts much more large-angle scattering than the Thomson model for scattering o the light electrons. For larger incident energy E there is less deection and less scattering. d Note. The total cross section tot = d d diverges for Coulomb scattering because the potential has innite range, so there remains some deection even for very large impact parameter. The concept of cross sections remains very useful in the quantum theory of scattering. For very large scattering angles (and correspondingly small impact parameter) the experimental ndings deviate from the Rutherford formula. As the -particle enters the region of the nucleus (R fm), the Coulomb potential is modied by the internuclear forces. Measurements of this deviation at large scattering angle yield the result for the radius of the nucleus R = 1.2 1015 m A = nnucleus = 2 1017 kg m3 (7-30)

where A is the number of nucleons (protons or neutrons). The density of the nucleus is constant and enormously large. (Steel has a density of 8 103 kg/m3 .)

Massachusetts Institute of Technology

VII-12