Sie sind auf Seite 1von 131

Lecture notes for Biophysics II

Hana Dobrovolny
Department of Physics, York University, Toronto, ON
September 12, 2011
Contents
1 Images 5
1.1 Imaging theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.1 One-dimensional images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.2 Two-dimensional images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Image reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.1 Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.2 Fourier transform technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.3 Back projection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 Image ltering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.4 Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2 Optical tweezers 36
2.1 Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.1.1 Radiation force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.1.2 Gradient force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2 Sphere in a focused beam of light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3 Determining strength of the trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.1 Constant velocity calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.2 Brownian motion calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3 Photon interactions with matter 43
3.1 Types of interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1.1 Photoelectric eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1.2 Compton scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.3 Coherent scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.1.4 Pair production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2 Photon attenuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.3 Energy transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.3.1 De-excitation of the atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.3.2 Mass energy transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.4 Charged particle interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.4.1 Stopping power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.4.2 Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4.3 Measures of energy transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
1
4 Crystallography 69
4.1 Lattice structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.1.1 Bravais lattices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.1.2 Primitive cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.1.3 Volume of a cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.1.4 Index system for crystal planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2 Bragg diraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.1 Fourier analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.2 Reciprocal lattice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.2.3 Diraction in Fourier space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.2.4 Structure Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.2.5 The phase problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.3 Protein Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.3.1 Making protein crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5 Spectroscopy 92
5.1 Determining energy levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.1.1 Particle in a box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.1.2 Two-dimensional particle in a box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.1.3 Circular well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.1.4 Rotational spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.1.5 Vibrational spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.1.6 Franck-Condon principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.2 Absorption and emission rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6 Nuclear Magnetic Resonance 109
6.1 Magnetic dipoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2 Magnetization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.2.1 Average magnetic moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.3 Precession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
6.3.1 Dephasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.4 Relaxation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.4.1 Rotating coordinate system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.5 Oscillating magnetic eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.5.1 Pulses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
6.5.2 Fluctuations in the magnetic eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.5.3 Autocorrelation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.5.4 Measuring the magnetic resonance signal . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.5.5 Pulse sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.6 Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.6.1 Slice selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.6.2 Image reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
2
List of Figures
1.1 Impulse response of a temporal signal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Point spread function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Spatial frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 Projections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Sinogram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 Back projection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Examples of image lters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.8 The microscope. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.9 Microscopy techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.10 Staining in bright eld microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.11 Oblique illumination microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.12 Dark eld microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.13 Dispersion staining microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.14 Phase contrast microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.15 Dierential interference contrast microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.16 Interference reection contrast microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.17 Fluorescence microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.18 Confocal microscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.1 Gradient force in a laser beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2 Forces on a sphere in a focused beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3 Eect of numerical aperture on a laser trap. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.4 Power spectrum of a trapped particle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.1 Photon interactions with matter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.2 Photoelectric eect. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Applications of the photoelectric eect. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.4 Compton scattering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5 Angular dependence of energy in Compton scattering. . . . . . . . . . . . . . . . . . . . . . . 48
3.6 Compton scattering cross section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.7 Coherent scattering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.8 Attenuation coecient. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.9 De-excitation of the atom. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.10 Tracks of charged particles in matter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.11 Mass stopping power. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.12 Energy transfer at high energies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.13 Range. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.14 Energy transferred and energy imparted. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.15 Dose and kerma. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.1 Crystals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3
4.2 A two-dimensional lattice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3 Two-dimensional Bravais lattices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.4 Three-dimensional Bravais lattices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.5 Primitive vectors of BCC and FCC lattices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.6 The Wigner-Seitz primitive cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.7 Crystal planes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.8 Bragg diraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.9 DNA diraction pattern. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.10 Laue cones. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.11 Brillouin zones. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.12 The phase problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.13 Isomorphous replacement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.14 Extended X-ray absorption ne structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.15 Protein crystallization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.16 Vapor diusion methods of protein crystallization. . . . . . . . . . . . . . . . . . . . . . . . . 87
5.1 Wavefunctions and probabilities for a particle in a box. . . . . . . . . . . . . . . . . . . . . . . 94
5.2 Degeneracy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.3 The rigid rotor model of a diatomic molecule. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.4 Simple harmonic oscillator spectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.5 Simple harmonic oscillators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.6 Franck-Condon principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.1 Larmor precession. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.2 Dephasing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.3 Relaxation of magnetization in a constant magnetic eld. . . . . . . . . . . . . . . . . . . . . 115
6.4 Magnetic eld of a quadrupole magnet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.5 Conguration of magnetic elds in an MRI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.6 Phase encoding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.7 Image reconstruction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4
Chapter 1
Images
Images, whether formed by the eye, a camera or other detection device are important in medical and biological
physics. We have already seen many examples of imaging in medical physics such as x-rays, optical coherence
tomography (OCT) or positron emission tomography (PET). Imaging plays a crucial role in observing and
understanding biological processes, so we need to understand how images are formed and how to correctly
interpret them.
1.1 Imaging theory
Images are map some quantity as a function of position, typically in two dimensions. Before we attempt
to describe two-dimensional images, we will introduce the basic concepts of images with a one-dimensional
system. Since spatial one-dimensional images are dicult to conceptualize, we will use a temporal system,
namely sound, to develop the mathematics needed to understand images.
1.1.1 One-dimensional images
Imagine that you are listening to music coming from your radio. Suppose that the original soundtrack
is described by some function, f(t). Ideally, the music perceived by your ear will also be f(t). In reality,
however, the electronics of the radio and travel through the air to your ear distorts the original input (see Fig.
1.1). The majority of these distortions essentially add some delay to parts of the original signal. For example,
while some of the sound wave will travel directly to your ear, other parts will bounce o nearby structures
before being detected by your ear. The signal perceived by your ear, g(t), then will be a combination of the
original input and the input at earlier times. If we assume the system is linear, we can simply add all the
contributions to determine g(t):
g(t) =
_

f(t

)h(t, t

) dt

(1.1)
where h(t, t

) is a weighting function that describes how the signal is delayed.


Suppose f is a function at time t

0
you can imagine it to be some loud, sudden, short-lived noise like
a thunderclap. Then the noise you will hear is given by
g(t) =
_

(t

0
)h(t, t

) dt

= h(t, t

0
) (1.2)
This means that h(t, t

) describes how our system responds to an impulse at time t

, so its called the impulse


response of the system. If we know the impulse response of the system, then its possible to calculate the
output for any arbitrary input which means that if you want to calculate the output signal for any input
signal, you will rst need to determine the impulse response for your system by applying some sudden loud
noise and measuring the response.
5
Figure 1.1: Impulse response of a temporal signal. An input signal is transformed by the impulse
response function h(t) to the signal that is actually detected.
If the system response to an impulse is the same regardless of when it occurs, the system is called a
stationary system. For example, if you are listening to music in a room where nothing (including you)
is moving, this is a stationary system the impulse response will be the same no matter when you test
it. If, however, you move to another point in the room, the system is no longer stationary there will
be one particular impulse response before you move and a dierent one after you move. For a stationary
system, h(t, t

) depends only on the time dierence between the input and the detection of the signal:
h(t, t

) = h(t t

). Then the superposition integral takes the form


g(t) =
_

f(t

)h(t t

) dt

. (1.3)
This is called the convolution integral and is denoted g(t) = f(t) h(t).
Example: Convolution
Find the convolution of f(t) = e
t
and h(t) = e
2t
for x 0.
Solution: Note that the integral goes from 0 to rather than from to because f(t) = 0
when x < 0.
g(t) = f(t) h(t)
=
_

0
e
t

e
2(tt

)
dt

= e
2t
_

0
e
t

dt

= e
2t
Convolution theorem
Recall that any function can be approximated by a Fourier integral, which converts a function to an innite
series of sine and cosine functions:
f(t) =
1
2
_

[C
f
() cos t +S
f
() sin t] d, (1.4)
6
where
C
f
() =
_

f(t) cos t dt (1.5)


S
f
() =
_

f(t) sin t dt
are the Fourier transform pairs. We will use the Fourier transform to help characterize the convolution.
We can write the input signal, f(t), and the impulse response, h(t t

), in their Fourier integral form


f(t

) =
1
2
_

[C
f
() cos t

+S
f
() sin t

] d (1.6)
h(t t

) =
1
2
_

[C
h
() cos (t t

) +S
h
() sin (t t

)] d .
This means that the output function is
g(t) =
_

f(t

)h(t t

) dt

(1.7)
=
_
1
2
_
2
_

dt

[C
f
() cos t

+S
f
() sin t

] d
_

[C
h
() cos (t t

) +S
h
() sin (t t

)] d.
We can use the trigonometric relations sin(tt

) = sin(t) cos(t

)sin(t

) cos(t) and cos(tt

) = cos(t) cos(t

)+
sin(t) sin(t

) to expand the h(t t

) integral
h(t t

) =
_

[C
h
()(cos(t) cos(t

) + sin(t) sin(t

)) (1.8)
+ S
h
()(sin(t) cos(t

) sin(t

) cos(t))] d.
We can then integrate g(t) over t

to get
g(t) =
1
2
_

[C
f
()C
h
() S
f
()S
h
()] cos(t) d (1.9)
+
1
2
_

[C
f
()S
h
() +S
f
()C
h
()] sin(t) d.
Comparing this to the Fourier integral for g(t), we see that we must have
C
g
() = C
f
()C
h
() S
f
()S
h
() (1.10)
S
g
() = C
f
()S
h
() +S
f
()C
h
().
This is known as the convolution theorem and it tells us that the convolution can be determined from the
Fourier transforms of the original functions. We now have two possible methods for determining the output
function for an arbitrary input function. We can directly perform the convolution integral or we can calculate
the Fourier transforms for f(t) and h(t t

), use the convolution theorem to nd the Fourier coecients


for g(t) and calculate the inverse transform to nd g(t). Note that this can be done more compactly using
complex exponentials, in which case the Fourier transform is
F() =
_

f(t)e
it
dt, (1.11)
and the convolution theorem becomes
G() = F()H(). (1.12)
7
Example: Convolution theorem
Find the convolution of f(t) = e
t
and h(t) = e
2t
for x 0 using the convolution theorem.
Solution: We need to nd the Fourier transform coecients for both f(t) and h(t).
C
f
() =
_

0
e
t
cos t dt
C
f
() =
1
1 +
2
S
f
() =
_

0
e
t
sin t dt
S
f
() =

1 +
2
.
and
C
h
() =
_

0
e
2t
cos t dt
C
h
() =
2
2 +
2
S
h
() =
_

0
e
2t
sin t dt
S
h
() =

2 +
2
.
This allows us to nd the Fourier coecients of g
C
g
() = C
f
()C
h
() S
f
()S
h
()
C
g
() =
2
(1 +
2
)(2 +
2
)


2
(1 +
2
)(2 +
2
)
C
g
() =
2
2
(1 +
2
)(2 +
2
)
S
g
() = C
f
()S
h
() +S
f
()C
h
()
S
g
() =

(1 +
2
)(2 +
2
)
+
2
(1 +
2
)(2 +
2
)
S
g
() =
3
(1 +
2
)(2 +
2
)
.
Now we can take the inverse transform to nd g(t)
g(t) =
1
2
_

[C
g
cos t +S
g
sin t] d
g(t) =
1
2
_

_
(2
2
) cos t
(1 +
2
)(2 +
2
)
+
3 sin t
(1 +
2
)(2 +
2
)
_
d
g(t) = e
2t
8
1.1.2 Two-dimensional images
In medical and biological physics, we most often use two-dimensional images, so we want to extend the math-
ematics of images from one dimension to two dimensions. In two dimensions, an object can be represented
by a function f(x

, y

) in the object plane. We have developed many ways to look at the object. If we
are using direct visual observation (looking with our eyes), we detect light rays that have bounced o the
object and are processed by our brains. There are also other ways of observing the object x-rays measure
the attenuation coecient as a function of position or MRIs measure the relaxation time as a function of
position. The image we see, g(x, y) in the image plane, is given by
g(x, y) =
_

f(x

, y

)h(x, x

; y, y

) dx

dy

(1.13)
where h(x, x

; y, y

) describes the measurement process.


We can dene a space invariant image, similar to a stationary one-dimensional signal, as one in which
the object is not moving in space and where the measurement process does not change in time. In this case,
the image produced at (x, y) depends only on relative distances x x

and y y

and the image can be


determined from the two-dimensional convolution
g(x, y) =
_

f(x

, y

)h(x x

, y y

) dx

dy

(1.14)
g(x, y) = f(x, y) h(x, y).
Suppose we have an object given by f(x

, y

) = L(x

0
)(y

0
), which describes a single bright
spot, of amplitude L, in a plane, then the image is given by:
g(x, y) =
_

L(x

0
)(y

0
)h(x x

, y y

) dx

dy

(1.15)
= Lh(x, y; x

0
, y

0
)
So h(x, y; x

0
, y

0
) describes the distortion of the imaging system, analogous to the impulse response in a
one-dimensional system. h(x, y; x

0
, y

0
) is known as the point spread function (PSF) of the system (see Fig.
1.2). Note that, in reality, the PSF may vary for each point in the object plane, however, for simplicity we
assume that there is a single PSF for the entire object.
Example: Magnication
There are some specic distortions of the object that may be benecial when trying to observe
the object. One such distortion is the magnication of an object. Verify that magnication is
described by the following point spread function:
h(x, y; x

, y

) = (x mx

)(y my

) (1.16)
Solution: Substituting into the equation for the image:
g(x, y) =
_

f(x

, y

)(x mx

)(y my

) dx

dy

= f
_
x
m
,
y
m
_
,
which tells us that the point (
x
m
,
y
m
) maps to the point (x, y) or that the object is magnied by a
factor of m in both the x and y directions.
9
Figure 1.2: Point spread function. An object is transformed by the point spread function h(x, y), which
describes the measurement or observation process, to the image that is actually detected.
10
Convolution theorem in two dimensions
We can write a Fourier transform in two dimensions
f(x, y) =
_
1
2
_
2
_

dk
x
_

dk
y
[C
f
(k
x
, k
y
) cos(k
x
x +k
y
y) (1.17)
+ S
f
(k
x
, k
y
) sin(k
x
x +k
y
y)]
where
C
f
(k
x
, k
y
) =
_

dx
_

f(x, y) cos(k
x
x +k
y
y) dy (1.18)
S
f
(k
x
, k
y
) =
_

dx
_

f(x, y) sin(k
x
x +k
y
y) dy
are the Fourier transform pairs. We can write a similar expression for the Fourier transform of h(x x

, y
y

) and substitute them into the expression for g(x, y). Using the trigonometric relationships to expand
h(x x

, y y

) and integrating over x

and y

, we can derive the convolution theorem in two dimensions


C
g
(k
x
, k
y
) = C
f
(k
x
, k
y
)C
h
(k
x
, k
y
) S
f
(k
x
, k
y
)S
h
(k
x
, k
y
) (1.19)
S
g
(k
x
, k
y
) = C
f
(k
x
, k
y
)S
h
(k
x
, k
y
) +S
f
(k
x
, k
y
)C
h
(k
x
, k
y
).
Transfer functions
The optical transfer function (OTF) is dened as the Fourier transform of the point spread function:
C
h
(k
x
, k
y
) =
_

dx
_

h(x x

, y y

) cos(k
x
(x x

) +k
y
(y y

)) dy
S
h
(k
x
, k
y
) =
_

dx
_

h(x x

, y y

) sin(k
x
(x x

) +k
y
(y y

)) dy,
or if we use complex numbers we can write this more compactly
H(k
x
, k
y
) =
_
1
2
_
2
_

dx
_

h(x x

, y y

)e
i(kx(xx

)+ky(yy

))
dy (1.20)
The OTF is useful because it breaks up the point spread function into two components: the modulation
transfer function (MTF) which describes how the modulation is altered, and the phase transfer function
(PTF) which describes how the phase is altered. The three transfer functions are related through
OTF = MTFe
iPTF
(1.21)
Suppose we have an object of the form L(x, y) = a + b cos(k
x
x + k
y
y). The modulation of the object is
dened as:
M =
L
max
L
min
L
max
+L
min
(1.22)
=
(a +b) (a b)
(a +b) + (a b)
=
b
a
.
which is the ratio of the amplitude of the signal to the oset of the signal. The modulation transfer function
is the amplitude of the optical transfer function:
MTF(k
x
, k
y
) = |H(k
x
, k
y
)| =
_
C
2
h
(k
x
, k
y
) +S
2
h
(k
x
, k
y
) (1.23)
11
Figure 1.3: Spatial frequencies. The left image is the original. The center image has the high frequencies
removed and the right image has the low frequencies removed.
and it gives the ratio of image modulation to object modulation MTF = M
image
/M
object
.
The phase transfer function is the phase of the optical transfer function:
PTF(k
x
, k
y
) = tan
1
_
S
h
(k
x
, k
y
)
C
h
(k
x
, k
y
)
_
(1.24)
and it describes phase shifts from the object to the image. Note that for each angle, there may be a dierent
phase shift.
Spatial frequencies
Just as sounds are a superposition of waves in time, images can be thought of as a superposition of waves in
space. The lowest spatial frequency determines the eld of view. Low frequencies determine shape, contrast
and brightness. The highest frequency determines the resolution of the image. High frequencies determine
edges and sharp detail. The role of various frequencies can be seen in Fig. 1.3. The center image shows the
original with the high frequencies removed the image appears blurry with all sharp edges and small details
removed. The right image shows the original with the low frequencies removed the image has retained all
the edges, but other than those edges the image is a uniform gray.
1.2 Image reconstruction
Most often the measurements we make somehow convert a three-dimensional object to a two-dimensional
image. Many times we are interested in re-creating the full three-dimensional object by making a series of
two-dimensional images and mathematically reconstructing the three-dimensional object. In this section,
we will study the mathematical techniques used in image reconstruction. For simplicity, however, instead of
considering reconstruction of a three-dimensional object using two-dimensional images, we will consider the
simpler case of reconstructing a two-dimensional surface from one-dimensional projections. The techniques
developed here for two dimensions can easily be extended to three dimensions.
1.2.1 Projections
First, well state the problem more formally. Suppose we have a two-dimensional function f(x, y) that
describes the surface of an object. We measure the function along projections essentially integrating
f(x, y) along various lines F(, x

) =
_

f(x, y) dy

, getting several one-dimensional measurements, as


shown in Fig. 1.4. If we do this along several angles, can we reconstruct the original surface?
12
Figure 1.4: Projections. A projection is essentially an integration of the surface along a particular line.
Each arrow indicates a particular projection and results in a single numerical measurement.
Example: Projection
Suppose an image is given by
I =
_
_
5 2 9
3 0 3
6 2 7
_
_
.
What are the projections along the horizontal (x) and vertical (y) directions?
Solution: For the projection along the x-axis, we sum all the columns
F(x) =
_
5 + 3 + 6 2 + 0 + 2 9 + 3 + 7

F(x) =
_
14 4 19

,
and for the projection along the y-axis, we sum all the rows
F(y) =
_
5 + 2 + 9 3 + 0 + 3 6 + 2 + 7

F(y) =
_
16 6 15

.
The set of all projections can be plotted in a sinogram. The sinogram plots the projections with x

on the
x-axis, on the y-axis and F(, x

) as an intensity (Fig. 1.5), essentially stacking the projections we measure


as the detector rotates around the object.
13
Figure 1.5: Sinogram. An object and its sinogram consisting of a stack of projections at dierent values
of .
1.2.2 Fourier transform technique
It turns out that there are several possible methods for reconstructing images, one of which uses Fourier
transforms. We can nd the Fourier transform of the surface f(x, y) (Eq. (1.17)):
f(x, y) =
_
1
2
_
2
_

dk
x
_

dk
y
[C
f
(k
x
, k
y
) cos(k
x
x +k
y
y)
+ S
f
(k
x
, k
y
) sin(k
x
x +k
y
y)]
where the Fourier transform coecients are given by Eqs. (1.18).
First consider the slice k
y
= 0; along this slice, the Fourier cosine coecient is
C
f
(k
x
, 0) =
_

cos(k
x
x) dx
_

f(x, y) dy (1.25)
=
_

cos(k
x
x)F(0, x) dx
where F(x) is simply the projection along the y-axis. Similarly, the other Fourier coecient can be written
as a function of the projection along the y-axis,
S
f
(k
x
, 0) =
_

sin(k
x
x)F(0, x) dx. (1.26)
This tells us the projections along the y-axis determine the Fourier transform pairs along the k
y
= 0 axis.
This is a specic example of the Fourier slice theorem. You can write equations of this form along any line in
Fourier space, correlating slices in Fourier space to slices in real space. In general, the Fourier slice theorem
states that the following two calculations are equal:
1. Take a two-dimensional function f(x, y), project it onto a line, and do a Fourier transform of that
projection.
14
Figure 1.6: Back projection. (Left) Every point on the object surface contributes in some way to each
projection and every point in every projection contributes in some way to the reconstructed image. (Right)
The location of a point on the object can be described in either the coordinate system of the object (x, y)
or in the coordinate system of the projection (x

, y

).
2. Take that same function, do a two-dimensional Fourier transform rst, and then slice it through its
origin, parallel to the projection line.
This means that the projections can be used to reconstruct the original image through Fourier transforms in
the following way: Take the Fourier transforms of the projections each projection gives one slice in Fourier
space and put the slices together in Fourier space then take the inverse Fourier transform. Unfortunately,
this is not always an easy technique to implement; the object must be fairly densely sampled and the inverse
transform calculation can be time-consuming.
1.2.3 Back projection
Back projection is based on the idea that every point on the surface contributes in some way to each
projection and every point in every projection contributes in some way to the reconstructed image (Fig.
1.6 (left)). The general strategy for back-projection is to take the contributions from each projection to
reconstruct the original images.
Let the coordinate system of the object be (x, y) and the coordinate system of the projection at angle
to the object be (x

, y

), see Fig. 1.6 (right). We can nd equations to transform from the object coordinate
system to the projection coordinate system;
x

= xcos +y sin (1.27)


y

= xsin +y cos
and the inverse transforms
x = x

cos y

sin (1.28)
y = x

sin +y

cos
Suppose we have an object f(x, y). We dene the projection, F, at angle as an integral along the line
y

F(, x

) =
_
f(x, y) dy

(1.29)
=
_
f(x

cos y

sin , x

sin +y

cos ) dy

.
15
This is known as the Radon transform.
We now need to nd some way to convert the projections back into the image. We dene the back
projection as:
f
b
(x, y) =
_

0
F(, x

) d. (1.30)
This sums the projections over a half circle we only need to measure projections over half the circle, since
the other half will result in identical projections. We want to show that the back-projection is in some way
related to the original object. Consider the projections at the origin:
F(, 0) =
_
f(y

sin , y

cos ) dy

(1.31)
and at the angle

= +/2, which is the angle to the y

-axis (see Fig. 1.6 (right)),


F(

, 0) =
_
f(y

cos

, y

sin

) dy

. (1.32)
This looks like integration in polar coordinates, so re-name y

as r

,
F(

, 0) =
_

f(r

) dr

. (1.33)
We substitute this into the back projection, Eq. (1.30), which becomes
f
b
(0, 0) =
_

0
_

f(r

) dr

. (1.34)
We can change the limits of integration
f
b
(0, 0) =
_
2
0
_

0
f(r

) dr

; (1.35)
multiply and divide by r to make it look like polar integration
f
b
(0, 0) =
_
2
0
_

0
f(r

)
r

dr

; (1.36)
and convert back to Cartesian coordinates
f
b
(0, 0) =
_
2
0
_

0
f(x

, y

)
(x
2
+y
2
)
1/2
dx

dy

. (1.37)
Recall that this result was only for projections at the origin; the general result (away from the origin) is
f
b
(x, y) =
_
2
0
_

0
f(x

, y

)
((x x

)
2
+ (y y

)
2
)
1/2
dx

dy

(1.38)
which looks like the convolution of the object with
h(x x

, y y

) =
1
((x x

)
2
+ (y y

)
2
)
1/2
(1.39)
So the back projection is related to the object, but it is not an exact replica of the original object.
16
Example: Back Projection
Suppose an image is given by
I =
_
1 2
3 4
_
.
Find all the projections then reconstruct the image using the back projection method.
Solution: In addition to the horizontal and vertical projections, we can take projections along
the diagonal.
F(0

) =
_
1 + 3 2 + 4

F(0

) =
_
4 6

F(45

) =
_
3 1 + 4 2

F(45

) =
_
3 5 2

F(90

) =
_
1 + 2 3 + 4

F(90

) =
_
3 7

F(45

) =
_
1 2 + 3 4

F(45

) =
_
1 5 4

.
The back projection method essentially puts all the data from the projections back into the original
image
I =
_
4 + 5 + 3 + 1 6 + 2 + 3 + 5
4 + 3 + 7 + 5 6 + 5 + 7 + 4
_
=
_
13 16
19 22
_
.
You can see that although all the information has been put back, the back projection method
does not quite reconstruct the original image.
It would be more convenient if we could take the projections in some clever way so that when we calculate
the back back projection, we get the original object. That is, can we create a function G(, x

) from F(, x

)
such that back projection of G will give us exactly f(x, y) instead of the convolution of Eq. (1.38). To begin,
note that there is a function g(x, y) which when projected and back-projected leads to f(x, y):
f(x, y) = g
b
(x, y) =
_
2
0
_

0
g(x

, y

)
((x x

)
2
+ (y y

)
2
)
1/2
dx

dy

(1.40)
17
We can now use the convolution theorem. Since f is a convolution of g and h =
1
((xx

)
2
+(yy

)
2
)
1/2
, the
Fourier transforms are related by the convolution theorem,
C
f
(k
x
, k
y
) = C
g
(k
x
, k
y
)C
h
(k
x
, k
y
) S
g
(k
x
, k
y
)S
h
(k
x
, k
y
) (1.41)
S
f
(k
x
, k
y
) = C
g
(k
x
, k
y
)S
h
(k
x
, k
y
) +S
g
(k
x
, k
y
)C
h
(k
x
, k
y
).
Solving for S
g
and C
g
gives
C
g
=
C
h
S
f
S
h
C
f
C
2
h
+S
2
h
(1.42)
S
g
=
C
h
C
f
+S
h
S
f
C
2
h
+S
2
h
.
In this case, we have a specic functional form for h, so we can nd the Fourier transform,
C
h
(k
x
, k
y
) = 2(k
2
x
+k
2
y
)
1/2
(1.43)
S
h
(k
x
, k
y
) = 0,
which makes S
g
and C
g
C
g
=
1
2
(k
2
x
+k
2
y
)
1/2
C
f
(1.44)
S
g
=
1
2
(k
2
x
+k
2
y
)
1/2
S
f
.
So we can nd a function g(x, y) from f(x, y) that when projected and back projected will give f(x, y)
exactly. This means that we have to somehow manipulate the original object, f(x, y), to get the function
g(x, y) which will give the projections G(, x

) which when back-projected gives exactly f(x, y). Instead of


trying to manipulate the object, however, it is easier to manipulate the projections. So its more useful to
nd G(, x

) from F(, x

), entirely avoiding the function g(x, y) which we do not know. Consider projections
along the x axis. We can write G as a Fourier transform of g
G(0, x) =
1
2
_

[C
g
(k
x
, 0) cos(k
x
x) +S
g
(k
x
, 0) sin(k
x
x)] dk
x
, (1.45)
where we have set k
y
= 0 because of the Fourier slice theorem. Eq. (1.44) gives a relationship between the
Fourier coecients of g and f, so G can be written as a function of the Fourier coecients of f
G(0, x) =
_
1
2
_
2
_

[C
f
(k
x
, 0) cos(k
x
x) +S
f
(k
x
, 0) sin(k
x
x)] |k
x
| dk
x
. (1.46)
Since this is independent of the choice of axis, it must be true for all projections. This means that there is
a function that will transform the projections of f into the projections of g. We can nd that function from
the Fourier coecients. Since we have
C
g
(k
x
, 0) =
|k
x
|
2
C
f
(k
x
, 0) (1.47)
S
g
(k
x
, 0) =
|k
x
|
2
S
f
(k
x
, 0),
we must have
C
h
=
|k
x
|
2
(1.48)
S
h
= 0,
18
which means
h(x) =
_
1
2
_
2
_

|k
x
| cos(k
x
x) dk
x
(1.49)
If we take the each projection of the object, F(, x), and convolve it with h(x) then take the back projection,
we will recover exactly f(x, y). This process is called ltered back projection.
1.3 Image ltering
Filtering of images is the process of modifying an image to highlight certain aspects or minimize others.
Filters are deliberate distortions of the images and, so can be written as convolutions. When discussing
lters, it is sometimes easier to consider the image as consisting of discrete pixels. In this case, lters
are represented as matrices which are passed over the entire image by the convolution process. The lter
essentially replaces the value of a pixel with a weighted sum of its nearest neighbours. Some examples of
lters are discussed below.
Blurring
A blurring lter smooths out dierences between neighbouring pixels by replacing a pixel with the mean or
median of the pixel and some of its neighbours. A simple 33 blurring lter (larger ones can also be made)
has the form
F
blur
=
_
_
0 0.2 0
0.2 0.2 0.2
0 0.2 0
_
_
. (1.50)
Note that we have to normalize the lter i.e. the sum of all elements must be one or the resulting image
would be brighter than the original. In this example, all pixels used in the averaging are given equal weight,
but it is also possible to create lters with unequal weights and to vary the number of neighbouring pixels
used in the averaging. The eect of a blurring lter is shown in Fig. 1.7 (center left).
Motion Blurring
Motion blurring gives the impression that an image is moving in one direction. This is achieved by blurring
in only one direction. A motion blurring lter that gives the appearance of diagonal movement has the form
F
motion
=
_
_
1
3
0 0
0
1
3
0
0 0
1
3
_
_
, (1.51)
where we have again normalized the lter. Note that this example is a 33 lter, but again larger ones are
possible. An image ltered with a motion lter is shown in Fig. 1.7 (center right).
Edge Finder
An edge-nding lter can be seen as a sort of discrete version of the derivative; you take the current pixel
and subtract the value of the previous one from it, so you get a value that represents the dierence between
those two or the slope of the function. An edge-nding lter that will detect horizontal dierences is
F
edge
=
_
_
0 0 0
1 1 0
0 0 0
_
_
. (1.52)
This lter is not normalized; the sum of all the elements is 0. Since this is the case, most of the image will
be darker than the original, as can be seen in Fig. 1.7 (bottom left).
19
Sharpening
Sharpening an image is very similar to nding edges. The dierence is that we want to add the enhanced
edges to the original image. So we add an edge-nding lter with the identity lter (which just returns the
original image) and the result will be a new image where the edges are enhanced, making it look sharper.
An example of this type of lter is
F
sharpen
=
_
_
1 1 1
1 9 1
1 1 1
_
_
, (1.53)
where we have again ensured that the lter is normalized since we want to retain the original image. Note
that this particular lter nds edges in all directions. The result of a sharpening lter is shown in Fig. 1.7
(bottom right).
Example: Filters
Suppose an image is given by
I =
_

_
4 5 2 4 5
3 0 6 1 3
8 4 3 0 2
6 7 1 2 7
9 4 9 3 4
_

_
,
where you can assume all pixels outside the image are 0. Use the motion blurring lter given by
Eq. (1.51) to lter the image.
Solution: For each pixel in the image, we replace it with the weighted sum indicated by the lter
I
new
=
_

_
0 +
4
3
+ 0 0 +
5
3
+
6
3
0 +
2
3
+
1
3
0 +
4
3
+
3
3
0 +
5
3
+ 0
0 +
3
3
+
4
3
4
3
+ 0 +
3
3
5
3
+
6
3
+ 0
2
3
+
1
3
+
2
3
4
3
+
3
3
+ 0
0 +
8
3
+
7
3
3
3
+
4
3
+
1
3
0 +
3
3
+
2
3
6
3
+ 0 +
7
3
1
3
+
2
3
+ 0
0 +
6
3
+
4
3
8
3
+
7
3
+
9
3
4
3
+
1
3
+
3
3
3
3
+
2
3
+
4
3
0 +
7
3
+ 0
0 +
9
3
+ 0
6
3
+
4
3
+ 0
7
3
+
9
3
+ 0
1
3
+
3
3
+ 0
2
3
+
4
3
+ 0
_

_
I
new
=
_

_
1
1
3
3
2
3
1 2
1
3
1
2
3
2
1
3
2
1
3
3
2
3
1
2
3
2
1
3
5 2
2
3
1
2
3
4
1
3
1
3
1
3
8 2
2
3
3 2
1
3
3 3
1
3
5
1
3
1
1
3
2
_

_
.
1.4 Microscopy
One of the original imaging techniques used to study biology is microscopy. The microscope contains a series
of lenses that produce a magnied image of the object under study (Fig. 1.8). While the original use of
the microscope was to simply observe specimen under normal light (bright eld microscopy), the eld of
microscopy has advanced and now uses many dierent techniques for viewing objects. These include:
Bright eld
Oblique illumination
Dark eld
20
Figure 1.7: Examples of image lters. Image lters can be used to distort or enhance images. Some
exapmles include a blurring lter (center left), a motion lter (center right), an edge lter (bottom left) and
a sharpening lter (bottom right).
21
Figure 1.8: The microscope. The objective produces a slightly magnied real image that is seen as a
highly magnied virtual image when viewed through the eyepiece.
Dispersion staining
Phase contrast
Dierential interference contrast
Interference reection microscopy
Fluorescence
Confocal
Many of these techniques can be used in conjunction to highlight dierent aspects of the specimen. Fig. 1.9
shows the same specimen (algae lament) viewed with three dierent microscopy techniques. The image is
quite faint in the bright eld and many features are indistinct. The phase contrast and dierential interference
contrast images are clearer and sharper, but each highlights dierent features of the lament. Images taken
using a variety of microscope techniques can be seen at http://www.microscopyu.com/smallworld/gallery/
and http://www.olympusmicro.com/galleries/index.html.
Bright eld microscopy
Bright eld microscopy is the simplest microscopy technique. The sample is illuminated from below and
transmitted light is viewed through the eyepiece. The sample is seen because light is absorbed by some
parts of the sample, but passes through others. This technique is cheap to implement and very easy to
understand and interpret. Unfortunately, the contrast is typically low since biological samples tend not to
absorb much light, although this can be corrected to some extent with stains. Fig. 1.10 shows stained and
unstained mites under bright eld microscopy. Without stain, the mite is faint and dicult to see; with
stain, even small details such as the cilia on the legs are clearly visible.
22
Figure 1.9: Microscopy techniques. The images show an algae lament imaged using (A) bright eld
microscopy (B) phase contrast microscopy and (C) dierential interference contrast microscopy. Taken from
http://www.microscopyu.com/articles/formulas/specimencontrast.html.
Oblique illumination microscopy
For oblique illumination, the light source is moved o-axis, so that some light is reected from the sides
of the sample, producing a 3-D eect. This is another simple, easy-to-interpret technique that is cheap to
implement. It has the advantage of showing parts of the sample that are not seen in normal bright eld
microscopy. Unfortunately, the contrast is again typically low since biological samples tend not to absorb
much light, although again this can be ameliorated with stains. Fig. 1.11 shows a polyp under varying degrees
of oblique illumination. As the angle increases, the sample develops more contrast and a 3-D appearance
since light is being reected o the edges of the specimen.
Dark eld microscopy
Dark eld microscopy removes most of the transmitted light by physically blocking the central portion of the
illumination source. Light that enters the eyepiece has been scattered or refracted by the sample (Fig. 1.12
(left)). This is an extreme version of oblique illumination where none of the direct illumination is allowed
to form the image. This is another simple, easy-to-interpret technique that is cheap to implement. It has
the advantage of producing sharp images without the use of stains. Unfortunately, light levels tend to be
low, requiring a strong light source to see the image. The images produced by dark eld microscopy can be
quite striking as the edges of the specimen appear bright on a dark background as seen in the image of a
mosquito wing in Fig. 1.12 (right).
Dispersion staining
Dispersion staining takes advantage of dispersion of materials to produce colour images without staining.
Recall that dispersion is the relationship between the index of refraction of a material and wavelength.
Suppose two substances in contact have the same dispersion at some wavelengths, but not at others. Then
some wavelengths will pass through the contact region undisturbed (since there is no change in the index of
refraction) while others will be refracted or reected. We can collect the refracted and reected light (which
will be of a certain colour) to create an image. This technique can be used to identify substances, and is
commonly used to identify asbestos (Fig. 1.13). Unfortunately, it is of limited usefulness since the sample
under study must be immersed in a uid whose index of refraction is the same at all but a few wavelengths
23
Figure 1.10: Staining in bright eld microscopy. The images show two species of plant mites, Orthoty-
deus kochi (top row) and Raphignathus gracilis (bottom row), under bright eld microscopy either unstained
(left column) or stained with a potassium-based dye (right column). Taken from Faraji and Bakker (2008),
Eur. J. Entomol. 105: 793-795.
24
Figure 1.11: Oblique illumination microscopy. The images show a Cnidaria (coe-
lenterate) polyp Obelia in the hydroid generation stage, under (a) bright eld illumi-
nation and under oblique illumination at 5

(b), 10

(c), and 15

(d). Taken from


http://micro.magnet.fsu.edu/optics/olympusmicd/anatomy/micdoblique.html.
25
Figure 1.12: Dark eld microscopy. (left) Light rays used in dark eld microscopy are reected or
refracted from the sample. (right) A dark eld image of a mosquito wing.
and these few wavelengths should be in the visible spectrum. Pictures of objects imaged with dispersion
staining are available at http://www.microlabnw.com/index/gallery-ds.html.
Phase contrast
Phase contrast microscopy uses the phase shift introduced by the change in index of refraction of the sample
to produce an interference pattern with a reference beam. In phase shift microscopy, the light from the
illumination source is split into a reference beam and a beam that goes through the sample. The beams
are later combined to produce an interference pattern, which converts dierences in phase to dierences
in amplitude, which are detectable by our eyes. This method highlights dierences in densities which is a
particular advantage for biological samples, allowing otherwise transparent structures within cells to be seen.
Phase contrast is, however, susceptible to a particular aberration which causes halos around dark spots that
may obscure small details. Fig. 1.14, shows glial cells in both the bright eld and phase contrast modes. In
the bright eld mode, the cells appear semi-transparent with only the membrane and nucleus being somewhat
visible. In phase contrast mode, the same eld of view reveals more structural detail; cellular attachments
are visible and the internal structure is visible in much more detail. Images taken using phase contrast
microscopy can be viewed at http://www.olympusmicro.com/primer/techniques/phasegallery.html.
Dierential interference contrast
Dierential interference contrast (DIC) microscopy splits a single light beam into two beams polarized at
90

(Fig. 1.15 (top)). The beams sample two nearby parts of the sample and may develop dierences in
phase due to dierent optical properties at the sites. When the light is recombined, an interference pattern is
produced showing regions of contrast. Similar to the phase contrast method, DIC has much better contrast
and resolution than the bright eld mode. Fig. 1.15 (bottom) shows several specimens in both DIC and
phase contrast. The DIC images do not show the halo eect that is seen in phase contrast images, so that
the edges are sharper in the DIC images. However, the phase contrast method sometimes shows features
that DIC does not, such as the cell nuclei in murine kidney tissue (center column).
26
Figure 1.13: Dispersion staining microscopy. An image of asbestos using dispersion staining. The
index of refraction of asbestos matches that of the surrounding uid at all wavelengths except blue. Blue
light is reected/refracted from the asbestos strands making them visible. Copyright: 2006 Environmental
Microbiology Laboratory, Inc.
Figure 1.14: Phase contrast microscopy. Human glial cells shown in
(a) bright eld and (b) phase contrast microscopy. Image taken from
http://www.microscopyu.com/articles/phasecontrast/phasemicroscopy.html.
27
Figure 1.15: Dierential interference contrast microscopy. (top) Paths of the light
rays used for DIC. (bottom) Images showing human buccal mucosa epithelial (cheek) cell (left
column), murine kidney tissue (center column), and Obelia polypoid annulated stem peris-
arc (right column) shown in DIC (upper row) and phase contrast (lower row). Taken from
http://www.olympusmicro.com/primer/techniques/dic/dicphasecomparison.html.
28
Figure 1.16: Interference reection contrast microscopy. (left) The paths of light rays used to produce
IR images. (right) A mouse embryo imaged using IR microscopy. Image taken from Verschueren (1985), J.
Cell Sci. 75:279-301.
Interference reection contrast
The specimen is placed under a piece of glass. A beam of light is directed through the objective towards the
glass covering the specimen. When the cell membrane is close to the glass surface, there will be destructive
interference with the original beam, resulting in a dark spot. When it is further from the glass, there will
be less interference, resulting in lighter spots (Fig. 1.16 (left)). This technique is particularly useful for
measuring cell migration as it can detect parts of the cell moving towards or away from the glass surface.
The mouse embryo cell shown in Fig. 1.16 (right) shows the cell moving to the left there are dark regions
on the edge of the extension indicating that this part of the cell is close to the glass.
Fluorescence microscopy
When some substances are excited with one wavelength of light, they emit light at another wavelength.
Fluorescence microscopy uses light with the excitation wavelength to illuminate the specimen and collects
the emitted light to create the image (Fig. 1.17). Since excitation and imaging are done at dierent wave-
lengths, there is no interference between the illumination source and collected light. Unfortunately, the
emitted light levels can be low and tend to degrade over time through photobleaching. While most bi-
ological samples do not naturally uoresce, uorescent labels have been developed that can be used to
image biological samples and biological processes. There is a gallery of uorescence microscopy images at
http://www.microscopyu.com/galleries/fluorescence/.
Confocal microscopy
Instead of using a wide eld of illumination, confocal microscopes use a single point of light to illuminate a
small section of the sample. A full image is made by scanning the point source over the entire sample. This
greatly improves resolution by eliminating out-of-focus light (Fig. 1.18) and allows for 3-D reconstruction of
a specimen.
29
Figure 1.17: Fluorescence microscopy. (left) The paths of light rays used to produce uo-
rescence images. (right) An endothelial cell stained with uorescent labels; red labels the mi-
tochondria, green labels the F-actine cytoskeleton and blue labels the nucleus. Taken from
http://www.meade.com/dsi/gallery/molecular probes.html.
Figure 1.18: Confocal microscopy. Images of human medulla (left column), rabbit mus-
cle bers (center column) and sunower pollen grain (left column) imaged using standard wide
eld uorescence microscopy (top row) and confocal microscopy (bottom row). Image taken from
http://www.olympusfluoview.com/theory/confocalintro.html.
30
1.5 Problems
Problem 1: (a) The equation
C(x, t)dx =
1

4Dt
__

C(, 0)e
(x)
2
/4Dt
d
_
dx
describes the concentration of particles that are subject to diusion and drift, where D is the
diusion coecient. What is the impulse response of this system? (b) The equation
v(R) =
_
a
2

i
4
o

2
v
i
x
2
1
r
dx
describes the exterior potential due to current ow along an axon, where
i
and
o
are the
interior and exterior conductivities. What is the impulse response of this system?
Problem 2: The identity lter is dened as the lter that takes an input to itself Id[x(t)] = x(t). Show that
the impulse response of the identity lter is (t).
Problem 3: Show that the convolution function is commutative interchanging the order of variables gives
the same results.
Problem 4: Calculate the convolution, f(x) g(x), of f(x) = g(x) = e
|x|
.
Problem 5: Find the output of a system if the input is x(t) = 2u(t 10) and the impulse response is
h(t) = sin(2t)u(t).
Problem 6: A linear time invariant system has the following impulse response,
h(t) = 2e
at
u(t)
Use convolution to nd the response y(t) to the following input
x(t) = u(t) u(t 4)
Sketch y(t) for the case when a = 1.
Problem 7: Calculate the cosine and sine transforms of e
bx
, where b is a positive integer.
Problem 8: (a) Derive the expression for the Fourier transform of
df(x,y)
dx
in terms of the Fourier transform
of f(x, y). (b) Based on the result in (a), what is the Fourier transform of
d
2
f(x,y)
dx
2
?
Problem 9: Show that the Fourier transform of
(x) =
_

n=
c
n
e
inx
, 0 < x < 2
0, otherwise,
does not vanish outside of any nite interval.
Problem 10: Dene two rectangular functions as
f
1
(x, y) =
_
1, if 0 x X; 0 y Y
0, otherwise,
and
f
2
(x, y) =
_
1, if |x| X/2; |y| Y/2
0, otherwise.
Find the Fourier transforms of the two functions and explain any dierences between the two
functions in the spatial and frequency domains.
31
Problem 11: (a) Use the convolution integral to calculate the convolution g(t) of the function
h(t t

) =
_
0, t < t

e
(tt

)/
, t > t

,
where is a constant, with
f(t) =
_
1, 0 < t < T,
0, otherwise.
(b) Calculate the Fourier transform of g(t), h(t t

), and f(t) from part (a) and show that


they obey the convolution theorem.
Problem 12: Use the convolution integral to calculate g(x) from
h(x x

) =
a
a
2
+ (x x

)
2
and f(x) = cos(kx). This can be interpreted as a spatial frequency lter. Hint:
_

cos(kx)dx
x
2
+b
2
=

b
e
kb
,
_

sin(kx)dx
x
2
+b
2
= 0.
Problem 13: What are the two dimensional images whose Fourier transforms are shown below?
32
Problem 14: Calculate the two-dimensional Fourier transform of the function
f(x, y) =
_
1, a/2 < x < a/2, b/2 < y < b/2,
0, otherwise.
Problem 15: Calculate the two-dimensional Fourier transform of the function
f(x, y) = sech
_
x
a
_
sech
_
y
b
_
.
You may need the following relationship
_

0
sech (uz) cos(wz)dz =

2u
sech
_
w
2u
_
.
Problem 16: Calculate the two-dimensional Fourier transform of the function
f(x, y) =
_
1,
_
x
2
+y
2
< a,
0,
_
x
2
+y
2
> a.
Hint: convert to polar coordinates in both x-y and k
x
-k
y
and use the fact that
J
0
(u) =
1
2
_
2
0
cos(ucos(x))dx
, J
1
(u) =
_
uJ
0
(u)du,
where J
0
and J
1
are Bessel functions of order zero and one.
Problem 17: Find the Fourier transform of the point-spread function for magnication (Eq. (1.16)).
Problem 18: Show that if the Fourier transform of a function f(x) is g(k), then the Fourier transform of
f(x a) is e
ika
(k).
Problem 19: Use the convolution theorem to show that the sum of the squares of the Fourier coecients
of the image is equal to the sum of squares of the Fourier coecients of the object times the
square of the modulation transfer function.
Problem 20: How does magnication change the spatial frequencies in going from object to image?
Problem 21: This problem shows how increasing the detail in an image introduces high-frequency compo-
nents. Find the Fourier transform of the two functions
f
1
(x) =
_
_
_
0, x < 0,
1, 0 < x < 1,
0, x > 1
f
2
(x) =
_

_
0, x < 0,
_
3/2, 0 < x < 1/3,
0, 1/3 < x > 2/3,
_
3/2, 2/3 < x < 1,
0, x > 1.
Plot a(k
x
) = [C
2
(k
x
) +S
2
(k
x
)
1/2
for each function and compare the features of each plot.
Problem 22: Suppose that the object is a point at the origin, so that f(x, y) = (x)(y). Find the projection
F(x) and the transform functions C(k
x
, 0) and S(k
x
, 0). Use the results to reconstruct the
image using the Fourier technique.
33
Problem 23: Derive the coordinate transformations Eqs. (1.27) & (1.28).
Problem 24: An object is described by the function
f(x, y) = e
(x
2
+y
2
)/b
2
.
(a) Find the Fourier transform C(k
x
, k
y
) and S(k
x
, k
y
). (b) Find the projection F(, x

). Then
take the one-dimensional Fourier transform of F(, x

) using
C(, k) =
_

F(, x

) cos(kx

)dx

S(, k) =
_

F(, x

) sin(kx

)dx

,
where k = (k
2
x
+k
2
y
)
1/2
. Your answer should be the same as in (a). You will need the following
integrals:
_

e
ax
2
dx =
_

a
_

e
ax
2
cos(bx)dx =
_

a
e
b
2
/4a
_

e
ax
2
sin(bx)dx = 0
Problem 25: A system performs the operation g(x, y) = f(x, y)f(x1, y). Derive the MTF of the system.
Problem 26: Assume you have just measured the following projection function
F(, x

) =

be
(x

a cos )
2
/b
2
.
Find f(x, y).
Problem 27: Assume you have just measured the following projection function
F(, x

) =
a

2
e
x
2
/a
2
_
1 + cos
2

_
2
x
2
a
2
1
__
.
Find f(x, y).
Problem 28: Suppose an object is a point at the origin, f(x, y) = (x)(y). The projection is also a point:
F(, x

) = (x

). Calculate the back projection f


b
(x, y) (without ltering). You will need the
following property of functions
(g(u)) =

i
(u u
i
)
|dg/du|
u=ui
where the u
i
are the points such that g(u
i
) = 0. Note that the back projection does not recover
the original image.
Problem 29: Let f(x, y) = (x x
0
)(y y
0
). (a) Calculate F(, x

). You will need the following properties


of the function:
_
(b z)(z a)dz = (b a)
(az) =
(z)
|a|
.
(b) Use the function F(, x

) you found in part (a) to calculate the back projection f


b
(x, y). (c)
Show that f
b
(x, y) is equivalent to the convolution of f(x, y) with the function 1/
_
(x x

)
2
+ (y y

)
2
.
34
Problem 30: A 22 image has the pixel values
_
7 2
4 0
_
.
Compute the projections of the image at 0

and 90

. Compute a reconstruction of the image


using the simple back projection method.
Problem 31: If f(x, y) is dimensionless, determine the units of F(, x

) and f
b
(x, y). Do the image and the
back projection have the same units?
Problem 32: Calculate the projections, F(, x

) for the following objects: (a) f(x, y) = 1/[(xa)


2
+y
2
+b
2
]
and (b) f(x, y) = x/(x
2
+y
2
)
2
.
Problem 33: Consider the object f(x, y) = a/
_
a
2
x
2
y
2
for |x| < a, and 0 otherwise. (a) Calculate
the projection F(, x

). (b) Use the projection from part (b) to calculate the back projection
f
b
(x, y). (c) Compare the object and back projection.
Problem 34: An image is given by the 33 matrix
_
_
1 2 1
2 3 2
1 2 1
_
_
,
and is 0 outside this matrix. What is the output if the image is processed with the following
lters: (a)
h =
1
9
_
_
1 1 1
1 1 1
1 1 1
_
_
,
(b)
h =
1
5
_
_
0 1 0
1 1 1
0 1 0
_
_
,
Problem 35: The 55 image
f(m, n) =
_

_
0 0 0 0 0
0 10 10 10 0
0 10 10 10 0
0 10 10 10 0
0 0 0 0 0
_

_
,
is processed by two systems in cascade. The rst system produces the output g
1
(m, n) =
f(m, n)f(m1, n). The second system produces the output g
2
(m, n) = g
1
(m, n)g
1
(m, n1).
Compute the images g
1
and g
2
. Does the sequence of application of the two operators aect
the result?
Problem 36: What are the impulse response and transfer function of the 33 mean lter?
Problem 37: The output of a lter at (m, n) is dened as the average of the four immediate neighbours of
(m, n); the pixel at (m, n) is itself not used. Derive the MTF of the lter.
35
Chapter 2
Optical tweezers
Optical tweezers are an experimental technique that is used to study and manipulate small objects. Since this
technique is non-invasive, it has found wide application in manipulating and studying dielectric spheres, living
cells, DNA, bacteria, and metallic particles. Optical tweezers are routinely applied to measure elasticity,
force, torsion and position of a trapped object. The technique relies on forces caused by the electromagnetic
waves in a laser beam. We begin with an examination of the forces on a particle trapeed in a laser beam.
2.1 Forces
2.1.1 Radiation force
When light hits an object, it causes a force. If we consider light as a particle, then the cause of this force
is quite straightforward the light transfers some of its momentum to the object. This radiation pressure
always acts in the direction of propagation of the light. We can write the radiation pressure as a function of
the Poynting vector, S, which is dened as
S =
1

0
EB, (2.1)
where E is the electric eld, B is the magnetic eld of the electromagnetic wave, and
0
is the permeability
of free space (a constant). The Poynting vector gives the energy ux (energy per unit area per unit time) of
an electromagnetic wave and it can be related to radiation pressure
P =
n
c
< S >, (2.2)
where n is the refractive index of the object, c is the speed oight, and < S > is the average value of the
Poynting vector. The force due to radiation pressure on the surface of an object is
F =
n
c
_ _
(S
in
S
out
) dA (2.3)
where we integrate the dierence in pressure due to incoming and outgoing radiation over the surface area.
36
Example: Force on a mirror
Calculate the force due to radiation pressure from a 60 W light bulb on a mirror.
Solution: For light reecting o a mirror, S
in
= S
out
. The force then is
F = 2
n
c
_ _
S
in
dA
=
n
c
P,
where the integral represents the total power, P. Putting in the index of refraction for glass (1.5),
the speed of light (3 10
8
m/s), and the total power incident on the mirror (60 W), we nd
F = 2
1.5
3 10
8
60 W
= 4 10
7
N.
We see that for macroscopic objects, the force due to radiation pressure is very small.
2.1.2 Gradient force
Light incident on a particle creates a dielectric response, due to the polarizability of the constituent atoms or
ions. We can think of this force as a consequence of the refraction of light as it enters the particle. Refraction
at the surface changes the direction of propagation and thus the momentum of the light. This momentum
is transferred to the particle causing a force. For an atom or ion in a monochromatic, linearly polarized,
continuous light eld, E, the time-averaged induced dipole moment is
p = E, (2.4)
where is the relative complex polarizability of the particle to the surrounding medium. The interaction of
the induced dipole with the electric eld of the light creates an electrostatic potential
U = p E. (2.5)
Thus in a light eld with a spatially varying intensity, there is a gradient force
F = U
= p E (2.6)
= E E
=

2
E
2
.
We see that the gradient force is linearly dependent on the spatial gradient of the intensity of the electro-
magnetic wave. Note that the force acts towards the point of highest intensity, which for a focused laser
beam will be the focal point.
Since this is a restoring force, we can model this force as a Hookean spring with stiness, k,
F = kx. (2.7)
This linear regime is seen in Fig. 2.1 which shows the gradient force as a function of the distance from the
center of the trap. Although not immediately obvious from these simple expressions, the trap stiness is
greatest when the particle to be trapped is the same size as the beam waist; as particle size decreases, the
restoring force decreases rapidly, but decreases only modestly when the particle size increases.
37
Figure 2.1: Gradient force in a laser beam. Near the focus of the laser beam, a small particle experiences
a linear restoring force.
Example: Gradient force
In a typical optical tweezers setup, the trap stiness can be varied from 10
6
0.1 pN/nm. What
is the range of restoring force if a particle is displaced 0.1 m?
Solution: The gradient force is given by
F = kx
and substituting the trap stiness and displacement, we nd F = 10
4
10 pN.
2.2 Sphere in a focused beam of light
Suppose we focus a beam of light to a very narrow point. If we place a sphere within the beam, light will exert
several forces on it as shown in Fig. 2.2. The forces F
1
and F
2
represent the gradient (or refractive) forces
due to the edges of the laser beam. In reality, there will be forces pointing in all directions between F
1
and
F
2
due to all the light rays in the beam. It is clear, however, that the horizontal components (perpendicular
to the laser) of these forces will cancel, leaving only the vertical components (parallel to the laser). There
will also be a force due to reection from the spheres surface. Recall that this force acts in the direction
of propagation of the light. If the sphere is displaced towards the light source, the radiation force helps to
push the sphere back towards the focal point. If the sphere is displaced in the other direction, however, the
radiation force helps to push the sphere out of the trap. The net result is to shift the center of the trap
slightly away from the focal point of the laser.
For ideal optical tweezers, we want to minimize the reective force and maximize the refractive force.
The greatest contributions to the refractive force come from the edges of the beam. Increasing the numerical
aperture of the lens will increase the force by increasing the angle of incidence
NA = nsin , (2.8)
38
Figure 2.2: Forces on a sphere in a focused beam. Refraction forces tend to pull the beam toward the
focus, while reection forces tend to push the bead away from the light source.
while this does not increase the reective force (it is independent of angle of incidence), but it comes at the
cost of creating a shallower trap (Fig. 2.3).
2.3 Determining strength of the trap
2.3.1 Constant velocity calibration
In a typical experiment, a bead, cell or part of a cell sits in a viscuous uid. If we trap the bead and then
move it through the medium at a constant velocity, we can determine the strength of the trap. At a constant
velocity, there is no net force on the object. The force exerted by the trap must equal the viscuous drag
force. The viscuous is force is given by
F = 6rv (2.9)
where is the viscosity, r is the radius of the sphere, and v is the velocity of the object, all of which can be
measured. This means we can determine the force exerted by the trap.
2.3.2 Brownian motion calibration
The trap can also be calibrated by observing Brownian motion. The trapped object will experience small
random forces from Brownian motion. Since there is no net acceleration, all forces must balance and we can
write
F
vis
+F
trap
= F
Brown
(2.10)

dx
dt
+kx = F(t),
39
Figure 2.3: Eect of numerical aperture on a laser trap. Increasing the NA increases the trap strength,
but decreases the trap depth and increases the power lost.
where is the coecient of viscous drag and we have assumed an ideal trap with no reective force. Since
Brownian motion is due to thermal uctuations, it has an average value of zero. This also means that the
power spectrum (Fourier transform) is constant since there is no dominant frequency in the motion.
|F(f)|
2
= 4k
B
T, (2.11)
where k
B
is the Boltzmann constant, and T is the temperature.
Recall that, if the Fourier transform of g(x) is G(f), then the transform of its derivative is 2ifG(f).
We can transform the entire dierential equation:
2ifX(f) +kX(f) = F(f) (2.12)
Squaring and solving for X(f):
|X(f)|
2
=
k
B
T
(f
c
+f)
2
(2.13)
where f
c
= k/2. The power spectrum of the motion of a trapped particle is constant at low frequencies
and it bends at f
c
(Fig. 2.4). By measuring the power spectrum of a trapped particle, we can determine f
c
which can be used to determine k and therefore F
trap
.
40
Figure 2.4: Power spectrum of a trapped particle. The power spectrum of a particle trapped in a laser
beam can be used to determine the strength of the trap.
41
2.4 Problems
Problem 1: Estimate the radiation pressure due to a 75 W bulb at a distance of 8.0 cm from the center of
the bulb. Estimate the force exerted on your ngertip if you place it at this point.
Problem 2: Suppose that a laser pointer is rated at 3 mW. If the pointer produces a 2 mm diameter spot
on a screen, what is the radiation pressure exerted on the screen?
Problem 3: Laser light can be focused (at best) to a spot with a radius r equal to its wavelength . Suppose
that a 1.0 W beam of green laser light ( = 5 10
7
m) is used to form such a spot and that
a cylindrical particle of about that size (let the radius and height equal r) is illuminated by
the laser. Estimate the acceleration of the particle, if its density equals that of water and it
absorbs the radiation.
Problem 4: Consider the total internal reection of a plane wave with wavelength = 800 nm incident at
an angle = 70

from the normal of a glass-air interface. The plane wave is incident from
the glass-side. The normal of the interface is parallel to the gravitational axis and the air-side
is pointing to the bottom. A tiny glass particle is trapped on the air-side in the evanescent
eld generated by the totally internally reected plane wave. Calculate the minimum required
intensity I of the plane wave to prevent the glass particle from falling down ( = 1.25). The
specic density of glass is = 2.2 10
3
kg/m
3
and the particle diameter is d
0
= 100 nm. What
happens if the particle size is increased?
Problem 5: A spherical glass particle in water is trapped at the focus of a laser beam with = 800 nm.
The polarizability of the particle is
= 3
0
V
0

w
+ 2
w
where V
0
is the volume of the particle, and the dielectric constants of glass and water are
= 2.25 and
w
= 1.76, respectively. (a) Show that for small transverse displacements (x)
from the focus the force is proportional to x. Determine the spring constant as a function
of d
0
, , and P
0
, where d
0
is the particle diameter and P
0
is the laser power. (b) Assume
d
0
= 100 nm. What intensity is necessary in order to create a trapping potential V > 10kT,
where k is Boltzmanns constant and T = 300 K is the ambient temperature? What is the
restoring force for a transverse displacement of x = 100 nm?
Problem 6: (a) If the maximum trapping force from laser tweezers on a 1 m radius spherical particle in
water at 20

C is 10 10
2
N, nd the minimum ow velocity of the water that will just free
the particle the trap. (b) If the calculated ow velocity of the water is doubled, by what factor
must the intensity of the laser beam be increased to just maintain the trap?
42
Chapter 3
Photon interactions with matter
A large part of biomedical physics involves using light or other electromagnetic waves to observe natural
phenomena. So we need to know how light interacts with matter to understand what we are seeing. In
classical physics, this involves geometric optics reection, refraction, and transmission, but in the quantum
world, there are other possibilities. How the photon interacts with matter depends largely on the energy of
the photon, as shown in Fig. 3.1. At low energies the photoelectric eect dominates; at medium energies
scattering dominates; and at high energies pair production dominates.
3.1 Types of interactions
Interactions between subatomic particles are denoted by (, bc) where the rst symbol represents the incident
particle(s) and the second symbol represents the result of the interaction. In this chapter, we will study in
detail the following photon interactions with matter:
Photoelectric eect (, e): A photon ejects an electron from a material.
Compton and incoherent scattering (,

e): A photon scatters o an atom, losing some energy


and ejecting an electron.
Coherent scattering (, ): A photon scatters elastically o an atom.
Inelastic scattering (,

): A photon scatters inelastically o an atom.


Pair production (, e
+
e

): A photon produces a electron-positron pair.


3.1.1 Photoelectric eect
Suppose a photon of energy E = h
0
is incident on a material. If the photon is absorbed by an atom, an
electron may be ejected. The photon must have enough energy to overcome the binding energy B of the
electron (work function). If there is any left-over energy, it goes into kinetic energy of the electron,
T
e
= h
0
B, (3.1)
where T
e
is the kinetic energy of the electron, h is Plancks constant and
0
is the frequency of the incident
photon. Fig. 3.2 depicts this process showing photons with three dierent energies incident on a strip of
potassium, which has a work function of 2.0 eV. The rst photon has an energy of 1.7 eV which is too low
to free an electron from the potassium atom. The other two photons have energies larger than 2.0 eV and
are capable of freeing electrons from the potassium atoms. As the energy of the incident photons increases,
the velocity (kinetic energy) of the ejected photons increases.
43
Figure 3.1: Photon interactions with matter. Photon interactions depend on the energy of the photon;
at low energies the photoelectric eect dominates, at medium energies scattering dominates, at high energies
pair production dominates.
Example: Photoelectric eect
The work function of copper is 4.7 eV. What is the minimum frequency of light needed to free
an electron from copper through the photoelectric eect? Would this frequency of light free
electrons from aluminum which has a work function of 4.19 eV?
Solution: The minimum energy required to free an electron from copper would leave the electron
with no kinetic energy, so Eq. (3.1) becomes
h
0
= B

0
= B/h

0
= 4.7/(4.135 10
15
)

0
= 1.13 10
15
Hz
This is light in the ultraviolet range. This light would also free electrons from aluminum since
the work function of aluminum is smaller. In the case of aluminum, the ejected electrons would
have a kinetic energy of T
e
= 4.7 4.19 = 0.51 eV.
While it is clear that there is some energy threshold for the photoelectric eect to occur, once we have
reached that threshold, it is not necessarily true that every photon incident on the material will release
an electron. Recall that the cross-section expresses the probability that an interaction will occur. The
cross-section for the photoelectric eect is

Z
n
E
3
, (3.2)
where Z is the atomic number of the material, E is the energy of the electron and n varies between 4 and
5. As E increases, the cross-section rapidly decreases. The total cross-section is a sum of the cross-sections
44
Figure 3.2: Photoelectric eect. Photons incident on a material with energies greater than the work
function of the material can transfer their energy to an electron which is subsequently ejected from the
material.
for all the electron shells of the atom
=
K
+
L
+
M
+. . . (3.3)
Applications of the photoelectric eect
Since the discovery of the photoelectric eect, it has been put to good use. The most common application
is in solar cells which collect sunlight (or other light) to generate electricity. Solar panels convert photons
from the sun to a stream of electrons (current) which can then power other devices. Fig. 3.3 (left) shows a
schemetic of a photovoltaic cell. The cell is covered with an anti-reective coating to maximize absorption of
light. The semiconductor beneath the surface is used to create an electric eld that will capture any electrons
released by the photoelectric eect. The electrons can then be directed into a circuit to power devices.
Another application is in imaging. CCD cameras use a variant of the photoelectric eect; light reected
or emitted from the sample hits pixels in the camera. The pixel sensors can be thought of as buckets (Fig.
3.3) where each incoming photon is converted to an electron that is stored in the bucket. When the camera
reads a pixel, it empties the bucket, creating a current that is proportional to the number of photons that
hit that pixel.
3.1.2 Compton scattering
Assume a photon with energy E = h
0
is incident on an electron that is unbound and at rest (Fig. 3.4). The
photon will collide with the electron and impart some of its energy and momentum to the electron. After
the collision, the photon and electron will leave the interaction site with directions and velocities that can
be determined by conservation of momentum.
The initial momentum of the photon is determined by its energy
E
2
= (p

c)
2
+ (m
0
c
2
)
2
(3.4)
E = p

c = h
0
,
where p

is the momentum of the photon and c is the speed of light. The electron has no initial momentum.
After the collision, the electron emerges at an angle , with momentum p
e
, and total energy
E
e
= T
e
+m
e
c
2
, (3.5)
45
Figure 3.3: Applications of the photoelectric eect. (left) Photovoltaic cells convert light
(usually sunlight) to an electric current that can be used to power devices. Image taken from
http://science.nasa.gov/science-news/science-at-nasa/2002/solarcells/. (right) A CCD camera
converts incident photons to electrons which are stored in each pixel until they are read to produce an image.
Image taken from http://www.mssl.ucl.ac.uk/surf/ydac/nuggets/2000/001201/001201.html.
Figure 3.4: Compton scattering. An incident photon scatters o an electron, giving some energy and
momentum to the electron.
46
where T
e
is the kinetic energy of the electron. The photon emerges at angle and reduced energy corre-
sponding to frequency

. We can now apply conservation of momentum in the direction of the incident


photon
h
0
c
=
h

c
cos +p
e
cos . (3.6)
Conservation of momentum in the direction perpendicular to the direction of the incident photon leads to
h

c
sin = p
e
sin . (3.7)
Finally, conservation of energy gives
h
0
= h

+T
e
. (3.8)
We can nd an expression for the electron kinetic energy from Eqs. (3.5) and (3.4) as applied to an electron;
E
e
= T
e
+m
e
c
2
E
2
e
= T
2
e
+ 2T
e
m
e
c
2
+ (m
e
c
2
)
2
(3.9)
(p
e
c)
2
+ (m
e
c
2
)
2
= T
2
e
+ 2Tm
e
c
2
+ (m
e
c
2
)
2
(p
e
c)
2
= T
2
e
+ 2T
e
m
e
c
2
We now have four equations that can be solved for various unknowns. We can nd the wavelength shift
of the photon by squaring and adding the momentum equations (Eqs. (3.6) and (3.7)) and substituting Eq.
(3.8) into Eq. (3.9) to get

0
=
c

0
(3.10)

0
=
h
m
e
c
(1 cos )
which tells us that the wavelength shift is independent of the incident wavelength of light. The quantity

C
= h/m
e
c = 2.427 pm is called the Compton wavelength.
We can also determine the angular dependence of the energy of the scattered photon
h

=
m
e
c
2
1 cos +
mec
2
h0
, (3.11)
and the energy of the electron
T
e
= h
0

m
e
c
2
1 cos +
mec
2
h0
(3.12)
T
e
=
m
e
c
2
(1 cos )
1 +
h0
mec
2
(1 cos )
.
While the equations may be somewhat dicult to interpret, Fig. 3.5 shows the angular dependence of both
the photon and electron as a function of the photon scattering angle. The maximum electron energy occurs
when = 180

that is when the photon essentially hits head-on and is scattered directly back.
47
Figure 3.5: Angular dependence of energy in Compton scattering. Energy of a photon and electron
after a 500 keV photon undergoes Compton scattering.
Example: Compton scattering
When the kinetic energy of the electron is a maximum, what is the wavelength shift of the photon?
Solution: From Fig. 3.5, we see that the electron kinetic energy is a maximum when = 180

.
Substituting this into Eq. (3.10), we nd

0
=
h
m
e
c
(1 cos )

0
=
h
m
e
c
(1 cos(180

0
=
2h
m
e
c
The wavelength shift is twice the Compton wavelength.
Now that we have analyzed the actual interaction, we are again interested in determining the probability
that the interaction will occur. This is given by the cross section, or in the case of Compton scattering, the
dierential cross section which gives the probability of scattering the photon (or electron) within a given
solid angle. Derivation of the expression for the Compton scattering dierential cross section is beyond the
scope of this course, we will simply examine the expression which is
d
C
d
=
r
2
e
2
_
_
1 + cos
2
+
x
2
(1cos )
2
1+x(1cos )
[1 +x(1 cos )
2
_
_
, (3.13)
where x = h
0
/m
e
c
2
and
r
e
=
e
2
4
0
m
e
c
2
(3.14)
is the classical radius of the electron. At low energies (x 0), this simplies to
d
C
d
=
r
2
e
(1 + cos
2
)
2
. (3.15)
48
0 1 2 3
(radians)
0
5e-16
1e-15
2e-15
2e-15
3e-15
D
i
f
f
e
r
e
n
t
i
a
l

s
c
a
t
t
e
r
i
n
g

c
r
o
s
s
-
s
e
c
t
i
o
n
Low energy
Medium energy
High energy
Figure 3.6: Compton scattering cross section. Angular dependence of the dierential scattering cross
section for Compton scattering.
The dierential scattering cross section is plotted in Fig. 3.6. At low energies, the cross section is symmetric
around 90

with a low probability of the photon scattering at right angles and higher probabilities of the
photon either passing straight through or scattering directly back. At higher energies, the probability of the
photon scattering backwards decreases.
Incoherent scattering
Incoherent scattering is an extension of Compton scattering. For Compton scattering, we assumed that the
electron was unbound and at rest. In reality, the photon is more likely to interact with electrons bound
within an atom. However, within the atom there are many electrons in close proximity so the photon is
thus interacting with several electrons at once. Suppose the photon interacts with an atom that contains
Z electrons. The maximum possible value of the cross section occurs if all the electrons take part in the
interaction, since this gives the maximum possible target size. This gives us an upper bound for the incoherent
scattering cross section

incoh
Z
C
(3.16)
3.1.3 Coherent scattering
Coherent scattering or Rayleigh scattering occurs when the photon scatters o the entire atom and doesnt
interact with individual electrons within the atom. The atom experiences a slight recoil, but its kinetic
energy is negligible. If we substitute the mass of the atom in the kinetic energy equation for the recoil
electron in Compton scattering (Eq. (3.12)) then
x =
h
0
m
atom
c
2
(3.17)
is small and
T
atom
=
h
0
x(1 cos )
1 +x(1 cos )
(3.18)
49
is also small or negligible. In coherent scattering, there is no wavelength or phase shift as can be seen from
Eq. (3.10)

0
=
h
m
atom
c
(1 cos ). (3.19)
Coherent scattering can be treated classically, so we can determine the intensity of scattered light (rather
than the cross section)
I = I
0
8
2

4
R
2
(1 + cos
2
) (3.20)
where R is the distance between photon and atom, is the polarizability of the atom. The wavelength
dependence of the intensity is shown in Fig. 3.7. Note the strong dependence on wavelength which is the
reason why sky is blue blue scatters much more than any other colour. Also note the angular dependence
of the intensity. It has the same angular dependence as the dierential scattering cross section for Compton
scattering at low energy, Eq. (3.15).
Example: Coherent scattering
For what wavelength of light is the scattering only 2% that of light with a visible wavelength of
520 nm?
Solution: Since we want to compare two light beams, we assume that the measurements are made
under identical conditions. This means we assume they have the same I
0
, R, and we measure at
the same angle . We know that
I

= 0.02I
520
.
Substituting Eq. (3.20) on both sides and cancelling all the constants we get
1

4
=
0.02
(520 nm)
4

4
=
(520 nm)
4
0.02
= 1382.8 nm
3.1.4 Pair production
Photons with energies above 1.02 MeV (2 the rest mass of the electron) can produce electron/positron
pairs. Other particle/antiparticle pairs are possible if the energy of the photon is high enough. However,
this process will not simply occur in free space.
Suppose a photon is travelling in free space and it at some point splits into an electron/positron pair.
Conservation of energy requires
h
0
= 2E = T

+m
e
c
2
+T
+
+m
e
c
2
= T
+
+T

+ 2m
e
c
2
, (3.21)
where T

is the kinetic energy of the electron and T


+
is the kinetic energy of the positron. Conservation of
momentum requires that
h
0
= 2p cos (3.22)
so we have
E = p cos
E
2
= p
2
cos
2
(3.23)
(pc)
2
+ (m
e
c
2
)
2
= p
2
cos
2

p
2
=
(m
e
c
2
)
2
(cos
2
c
2
)
.
50
Figure 3.7: Coherent scattering. Wavelength dependence of the intensity of coherent scattering.
Note that the denominator is always negative (cos
2
c
2
) which means that p is imaginary. Therefore we
cannot assume that momentum is conserved which means that the process cannot happen in free space; it
requires the presence of a large atom/nucleus to absorb some of the photons momentum. Since the atom
has a mass that is large compared to the masses of the electrons and positrons, its kinetic energy is small
and is often neglected in calculations.
51
Example: Pair production
A photon of energy E

strikes an electron at rest and undergoes pair production, producing a


positron and another electron (e

,e

e
+
). The two electrons and the positron move o with
identical linear momentum in the direction of the initial photon. All the electrons are relativistic.
Find the kinetic energy of the nal three particles in terms of the mass of the electrons and the
speed of light.
Solution: Let p, T, and E be the momentum, kinetic energy, and total energy of each of the
particles after the collision. Since the particles all have the same momentum and mass, they will
also have the same kinetic energy and total energy. Conservation of momentum for this system
gives
p

= 3p
E

c
= 3p.
The conservation of energy gives
E

+m
e
c
2
= 3m
e
c
2
+ 3T
E

= 2m
e
c
2
+ 3T.
Between the two equations, we can eliminate the photon energy and solve for momentum,
p =
2m
e
c
2
+ 3T
3c
.
We also have the following two equations for the total energy of the particles,
E = T +m
e
c
2
E
2
= (pc)
2
+ (m
e
c
2
)
2
We can square the rst equation, substitute it into the second equation,
T
2
+ (m
e
c
2
)
2
+ 2Tm
e
c
2
= (pc)
2
+ (m
e
c
2
)
2
.
Finally, we substitute for p and after a little algebra, we nd
T =
2
3
m
e
c
2
.
3.2 Photon attenuation
We have discussed several ways that photons interact with matter. We rarely, however, observe any of these
interactions individually. More often we shoot a narrow beam of light through some material and we are
interested in how this beam attenuates through the material. The attenuation will, of course, be due to the
interactions discussed above, but we do not need to know the paths of individual photons to describe the
attenuation of the beam. The beam attenuation is described by
dN
dz
= N, (3.24)
52
Figure 3.8: Attenuation coecient. The attenuation coecient of a beam is the result of photons leaving
the beam because of interactions with the material.
where N is the number of photons in the beam, is the attenuation coecient, and z is the distance the
beam travels. Recall that the attenuation coecient is related to the cross section
=
N
A

A
, (3.25)
where N
A
is Avogadros number, is the mass density of the material, and A is the atomic weight. Since
the total cross section is the sum of cross sections for individual interactions, the attenuation coecient will
also be a sum of inidividual attenuation coecients
=
N
A

i
f
i

i
, (3.26)
where f
i
is the fraction of energy lost in process i. Fig. 3.8 shows how each type of interaction conttributes
to the overall attenuation of the beam. The relative contributions of each process to the overall attenua-
tion coecient change with the energy of the beam; at low energies where attenuation is highest it is the
photoelectric eect that most strongly attenuates the beam, at high energies pair production dominates.
It is not just the number of photons that attenuates, but also the energy of the beam. In many applica-
tions, we are, in fact, more concerned with the energy of the beam as it passes through the material. If a
beam of monoenergetic photons of energy E = h and particle uence passes through a thin layer dx of
material. The number of particles per unit area that interact is given by
d =
atten
dx. (3.27)
The energy uence is
d = hd = h
atten
dx. (3.28)
Integrating, we get
=
0
e
attenx
, =
0
e
attenx
. (3.29)
53
Compounds and mixtures
In biology, we rarely have materials made of a single type of atom. For mixtures and compounds, we assume
that each atom interacts independently of others
n
N
=

i
(N
T
)
i
=

i
(N
TV
)
i
dz (3.30)
where n is the mean number of photons that interact, N is the number of incident photons, (N
T
)
i
is the
number of target atoms per area of species i, and (N
TV
)
i
is the number of target atoms of species i per unit
volume.
Example: Attenuation
Let
1
= 0.02 cm
1
and
2
= 0.04 cm
1
be the linear absorption coecients for two processes in
a slab of tissue. Let the thickness of the slab be L = 5 cm and the initial number of particles in
the beam be N
0
= 10
6
particles. How many particles, N
L
, are transmitted, and how many are
absorbed by each process in the slab?
Solution: The number of particles transmitted determined by the total absorption coecient,
=
1
+
2
and is given by
N
L
= N
0
e
(1+2)L
N
L
= 10
6
e
(0.02+0.04)5
N
L
= 7.4082 10
5
7.408210
5
particles make it through the slab. This means that 10
6
7.408210
5
= 2.591810
5
particles are absorbed in the slab. The number absorbed by each process is determined by the
individual absorption coecients.
N
1
= 2.5918 10
5

= 8.6393 10
4
N
2
= 2.5918 10
5

= 1.7279 10
5
3.3 Energy transfer
To this point, we have only considered what happens to the photons as they move through the material, but
the interactions change the material as well. In particular, any energy lost by the photons is transferred to
the material. This section discusses energy transfer from photons to a material.
3.3.1 De-excitation of the atom
After some of photon interactions, we are left with an atom in an excited state. The excitation arises because
there is a hole in one of the inner shells of the atom because a photon has interacted with and knocked out
an electron. The hole can be lled through one of two processes:
Radiative transition in which a photon is released
Radiationless transition in which an Auger electron is released.
Both of these processes are depicted in Fig. 3.9. De-excitation of the atom involves having an electron drop
into the hole. To do this, the electron in the outer shell must somehow lose energy. In the radiative process,
54
Figure 3.9: De-excitation of the atom. An excited atom can de-excite through either a radiationless
process (left) in which an electron is knocked out of the electron or a radiative transition (right) in which a
photon is released.
the extra energy is emitted as a photon. In the non-radiative process, the extra energy goes to a nearby
electron which is then ejected from the atom. The emitted electron is known as an Auger electron.
In the radiative process, the energy of the emitted photon is equal to the energy dierence between the
levels
h
0
= B
K
B
L
, (3.31)
where B
K
is the energy of the outer energy level and B
L
is energy of the inner empty energy level. In the
radiationless process, the kinetic energy of the emitted electron is
T
e
= B
K
B
L
B
L
= B
K
2B
L
. (3.32)
3.3.2 Mass energy transfer
If we send a beam of photons through some material, some will interact with the material leading to secondary
photons, electrons and positrons. The secondary photons typically travel long distances from the site of the
interaction, so their energy is typically carried out of the material. Electrons and positrons lose their energy
fairly close to the site of the interaction. Damage to cells is caused by local ionization or excitation of atoms
and molecules.
If N monoenergetic photons of energy E strike a thin absorber of thickness dx. We dene the mass
energy transfer coecient,
tr
/, as

tr

=
1
NE
dE
tr
dx
, (3.33)
where is the density of the material and E
tr
is the energy transferred to the material. This coecient
characterizes the average fraction of energy absorbed from the beam by the material per unit length. We
can relate
tr
to
atten
by comparing Eq. (3.26) and Eq. (3.33);

tr

=
N
A
A

i
f
i

i
. (3.34)
We have already described the interactions which cause energy transfer; we simply have to sum all their
contributions to determine the mass eneergy transfer coecient. Coherent scattering produces no charged
55
particles, so we are left with

tr

=
N
A
A
(
photo
f
photo
+
C
f
C
+
pair
f
pair
). (3.35)
Recall that the photoelectric eect leaves an excited atom in the material. This atom will decay either
through uorescence or an Auger electron. The fraction of energy transferred to charged particles can be
written in terms of , the average energy emitted as uorescence per photon absorbed.
f
photo
=
h
h
= 1

h
. (3.36)
If we assume that the dominant term is due to holes in the K shell (holes in higher shells lead to lower
energy photons), then the probability that the hole is lled by uorescence (rather than the Auger process)
is W
K
. The energy of the photon emitted during this process is B
K
B
L
(Eq. (3.31)). So the average energy
emitted through uorescence is
W
K
(B
K
B
L
), (3.37)
so we have
f
photo
= 1
W
K
(B
K
B
L
)
h
. (3.38)
For compton scattering, the energy transferred to the electron is given by Eq. (3.12)
T =
m
e
c
2
(1 cos )
1 +
h0
mec
2
(1 cos )
The average energy transferred can be calculated by taking the average over all scattering angles
f
C

C
=
_

0
d
C
d
T()
h
0
2 sin d. (3.39)
This can integrated
f
C

C
= 2r
2
e
_
2(1
x
)
2
x
2
(1 + 2x)

1 + 3x
(1
2
x)
2

(1 +x)(2x
2
2x 1)
x
2
(1 + 2x)
2

4x
2
3(1 + 2x)
3

_
1 +x
x
3

1
2x
+
1
2x
3
_
ln(1 + 2x)
_
.
(3.40)
For pair production, any energy over 2m
e
c
2
becomes kinetic energy of the electron and positron. The
fraction of energy transferred to charged particles is
f
pair
=
h 2m
e
c
2
h
= 1
2m
e
c
2
h
. (3.41)
3.4 Charged particle interactions
Charged particle interactions with matter are dierent than photon interactions with matter. The cross
section of charged particle interactions is 10
4
10
5
larger than the cross section for photon interactions. A
beam of charged particles hitting a substance will typically only travel a short distance before being stopped.
This can be seen in Fig. 3.10 which shows particles, electrons, and positrons travelling through emulsions.
The average distance travelled by these particles is very short, ranging from nanometers to micrometers.
3.4.1 Stopping power
We can characterize what happens to a charged particle by measuring the stopping power and range. The
stopping power is the energy lost per unit length; or the expectation value of the kinetic energy lost per unit
56
Figure 3.10: Tracks of charged particles in matter. (left) Tracks of particles in a photographic
emulsion; the particles travel about 100 m. (center) Tracks of an e
+
/e

pair in a photographic emulsion;


the particles travel about 100 m. (right) Tracks of 1 keV electrons in a cloud chamber; the Auger electrons
travel about 1020 nm.
length S =
dT
dx
. The mass stopping power is the stopping power divided by the density of the material
S

.
The range is the total distance the particle travels before losing its energy.
As we saw with photons, charged particles can also interact with matter through several dierent pro-
cesses. The dominant interaction of projectile particles is with the electrons, S
e
, within atoms of the material.
The projectile may also interact with the nucleus S
n
, although this is a small fraction of the interactions
(1%). Finally, the projectile may emit a photon to lose energy (radiation), S
r
. The total stopping power
is the sum of the three processes
S = S
e
+S
n
+S
r
. (3.42)
Fig. 3.11 shows the mass stopping power for electrons, positrons, protons and particles in carbon as a
function of energy. The mass stopping power is higher for protons and particles, but once we normalize
(Fig. 3.11 (right)), all curves have roughly the same shape; initially rising to a peak then decreasing. At
higher energies, the curves will rise again. When the projectile has low energy, it moves past a target atom so
slowly that the electrons within the atom have time to rearrange themselves in response, essentially staying
in equilibrium during the entire process. The projectile will leave with nearly the same energy it had when
it entered. As the speed of the projectile increases, the process is no longer adiabatic and the projectile will
lose some of its energy during the interaction. At still higher speeds, the projectile moves through so quickly
that the atom has little time to respond and absorb any energy. The energy transfer is most eective when
the speed of the projectile is about equal to the speed of the atomic electrons.
The probability of a particle losing an energy between W and W+dW when travelling through a thickness
dx is
n
N
=
N
A

A
dx
d
dW
dW. (3.43)
The average energy lost is found by multiplying the probability by the energy and integrating over all energies
dT =
N
A

A
dx
_
Wmax
0
W
d
dW
dW, (3.44)
57
Figure 3.11: Mass stopping power. (left) The mass stopping power of various particles in carbon as a
function of the projectile energy. (right) The same curves scaled by z
2
.
where W
max
is the maximum possible energy that can be transferred. This is known as the stopping cross
section.
The maximum possible energy transfer can be calculated from conservation of energy and momentum.
Assume a projectile of mass M
1
and kinetic energy T collides with a particle of mass M
2
, initially at rest.
After the collision, M
1
has energy T

and M
2
has energy T
2
. Conservation of momentum (non-relativistic)
gives
M
1
_
2T
M
1
= M
1
_
2T

M
1
+M
2
_
2T
2
M
2
, (3.45)
and conservation of energy (non-relativistic) gives
T = T

+T
2
. (3.46)
Recognizing that W = T T

= T
2
, with a little algebra, we nd
W
max
=
4TM
1
M
2
(M
1
+M
2
)
2
. (3.47)
For a relativistic calculation, we nd
W
max
=
2(2 +T/M
1
c
2
)TM
1
M
2
M
2
1
+ 2(1 +T/M
1
c
2
)M
1
M
2
+M
2
2
. (3.48)
Note that when the masses are the same (electron-electron scattering, for example) W
max
= T; the projectile
can transfer all of its kinetic energy to the particle within the matter.
58
Example: Energy transfer
Suppose the particle with the matter, M
2
, initially has some kinetic energy T
2
. Does this change
the amount of energy that can be transferred? Do the calculation non-relativistically.
Solution: In this case, conservation of momentum gives
M
1
_
2T
M
1
+M
2
_
2T
2
M
2
= M
1
_
2T

M
1
+M
2

2T

2
M
2
,
and conservation of energy gives
T +T
2
= T

+T

2
.
We again have W = T T

, so we can make the following substitutions: T

= T W and
T

2
= W +T
2
(from energy conservation). After some messy algebra, we nd
W
max
=
4(M
1
M
2
(M
1
+M
2
)
2
_
T T
2
+ 2(M
2
M
1
)
_
TT
2
M
1
M
2
_
.
To check the result, we set T
2
= 0 and verify that we get Eq. (3.47) and we do.
High energy classical approximation
Recall that when the projectile has a high energy, the interaction is much like an impulse since the projectile
moves by so fast. We can do a classical approximation at high energies using an impulse approximation.
Suppose a particle of mass M, charge ze, and velocity V = c is moving past a stationary electron. The
impact parameter, b, is the perpendicular distance from the electron to the path of the projectile (see Fig.
3.12 (left)). The distance from the projectile to the electron is r (see Fig. 3.12 (right)). The distance along
the path to the point of closest approach is (see Fig. 3.12 (right)). The momentum transferred to the
electron is
p =
_
Fdt. (3.49)
The force acting on the electron is due to the electric eld of the charged projectile,
p = e
_
Edt. (3.50)
By symmetry, there is no net component of E parallel to the direction of travel (any force experienced as the
projectile approaches will be cancelled by an equal force as the projectile recedes), so it is the perpendicular
component of E that causes the momentum transfer.
The perpendicular component of the electric eld is
E

= E sin . (3.51)
The electric eld is created by the charge on the projectile
E

=
ze sin
4
0
r
2
. (3.52)
We can determine sin from geometry
E

=
ze
4
0
r
2
b
r
=
ze
4
0
r
2
b
(
2
+b
2
)
3/2
. (3.53)
59
Figure 3.12: Energy transfer at high energies. (left) The impact parameter, b, is the perpendicular
distance from the target particle to a line from the projectile. (right) A particle of mass, M and charge ze
moves at velocity v past an electron.
The momentum transfer is
p =
_
F

dt = e
_
E

_
dt
d
_
d (3.54)
If the fraction of energy lost by the projectile is small, its velocity does not change much
d
dt
= c = constant, (3.55)
which gives a momentum transfer of
p =
e
c
_
E

d
=
ze
2
b
4
0
c
_

d
(
2
+b
2
)
3/2
(3.56)
=
2ze
2
4
0
cb
We see that smaller impact parameters lead to greater momentum transfer and that smaller velocities lead
to greater momentum transfer.
The kinetic energy of the electron is
W =
p
2
2m
e
=
2z
2
e
4
(4
0
)
2
m
e
c
2

2
b
2
(3.57)
W does not depend on the mass of the projectile, but lower speeds lead to higher kinetic energies and smaller
impact parametes lead to higher kinetic energies.
The classical cross section (see Fig. 3.12 (left)) is
d =
d
dW
dW = 2bdb. (3.58)
We can use the equation for W to eliminate b
d
dW
dW =
z
2
e
4
8
2
0
m
e
c
2

2
dW
W
2
. (3.59)
60
There is some minimum W
min
where the impact parameter is large and the impulse approximation breaks
down. There is also a maximum possible energy transfer, W
max
, as we have seen earlier.
To get the stopping power, we multiply the cross section by W and integrate
S
e

=
N
A
e
4
4
2
0
m
e
c
2

2
Z
A
z
2
ln
_
W
max
W
min
_
. (3.60)
Stopping power is independent of the mass of the projectile, but does depend on its charge (hence the
normalization of Fig. 3.11). We can see that at high energies, the stopping power is inversely propotional
to (c)
2
, a result that is also seen in Fig. 3.11 (right). The full quantum mechanical calculation leads to a
similar result.
So far, we have only considered the stopping power of a projectiles interaction with an electron. When
the projectile enters the material, it may also interact with the nucleus. When a projectile interacts with
a nucleus, it is actually scattered (changes its direction of travel) because of the mass of the nucleus. The
energy transferred to the nucleus depends on the scattering angle. The calculation is complex because we
must consider
the electric force between nucleus and projectile
the electric force between the projectile and the electron cloud
Pauli exclusion principle,
and so it is beyond the scope of this course. Remember, however, that projectile interactions with electrons
account for 99% of the interactions within the material.
Compounds
As with cross sections, when dealing with a compound we assume that the constituent atoms of the target
act independently of each other. This means that stopping power can be summed
S

i
w
i
_
S

_
i
. (3.61)
This is known as the Bragg rule.
3.4.2 Range
The range can be calculated from the stopping power. Recall that the stopping power is the average energy
loss per unit path length. If we assume that the projectile loses energy continuously (continuous slowing-down
approximation (CSDA)), then
R
CSDA
=
_
T0
T
f
dT
S
e
+S
n
+S
r
(3.62)
However, this is only one possible denition for the range. As can be seen in Fig. 3.13, there is not an
obvious single denition for the range. In practice, several possible ranges are dened as the projectile moves
through the material:
median range, R
50
distance at which 50% of energy is lost
maximum range, R
m
distance at which all energy is lost
extrapolated range, R
ex
distance extrapolated from linear part of the graph to the abcissa
61
Figure 3.13: Range. A plot of the number of projectile particles transmitted as a function of distance shows
that there are several possible denitions for the range.
mean range, R dened as
R =
_
x
_
dF
dx
_
dx
_ _
dF
dx
_
dx
(3.63)
where F(R) is the fraction of transmitted particles as a function of distance.
If the transmitted curve is symmetrical about the mean, then R = R
50
. For heavy particles, R is the best
estimate of R
CSDA
. For light particles, R
ex
is closest to R
CSDA
.
Example: Mean range
What is the mean range if the transmission is given by F(R) = e
x/
?
Solution: The mean range is given by Eq. (3.63)
R =
_
x
_
dF
dx
_
dx
_ _
dF
dx
_
dx
.
First the numerator:
_
x
_
dF
dx
_
dx =
_

0

e
x/
dx
= e
x/
_
1 +
x

0
= ,
and the denominator
_ _
dF
dx
_
dx =
_

0

e
x/
dx
= e
x/

0
= 1.
So the mean range is R = .
62
Fluctuations in the range are called straggling. Straggling is more apparent for lighter particles like
electrons and positrons than for heavy particles like protons. Electrons and positrons get scattered often so
their range is badly approximated by R
CSDA
.
3.4.3 Measures of energy transfer
In biological systems, we are interested in how much energy is absorbed by matter locally after an interaction,
since this is what causes tissue damage. One estimate is the energy transferred, E
tr
, to particles within the
substance. The energy transferred refers specically to energy loss of a beam of uncharged particles (typically
photons) in a given mass of some substance. Energy transferred can be calculated by taking the dierence
between the incoming and outgoing energy of the beam.
Energy transferred is a stochastic quantity for a given mass, the energy transferred will be random,
depending on which particular interactions the beam experiences. We dene the kerma as the expectation
value of the energy transfer per unit mass
K =
dE
tr
dm
. (3.64)
Consider a beam of monoenergetic photons of energy h, we have
NE = S. (3.65)
We can write dm = Sdx, then the Kerma is
K =
S
tr
dx
Sdx
=

tr

(3.66)
Energy imparted, E, is the net energy into a volume from all sources (not just uncharged particles). Note
that this is not the same as energy transferred a photon interaction in one volume can send an electron
to another volume where it leaves its energy. In this case, volume 1 had an energy transfer, but volume
2 had energy imparted. An example of the dierence between energy transferred and energy imparted is
shown in Fig. 3.14. Two 10 MeV photons enter some water and undergo pair production (top) and Compton
scattering (bottom). The energy transferred and energy imparted are shown for each volume.
The absorbed dose is the expectation value of the energy imparted per unit mass
D =
dE
dm
. (3.67)
Since energy transferred and energy imparted are not the same, neither are kerma and absorbed dose. Fig.
3.15 shows the kerma and dose as a photon beam goes through tissue. The absorbed dose and kerma are
quite similar inside the tissue, but dier near the surface. While absorbed dose actually measures the amount
of energy left within the tissue, it is not necessarily a good indicator of biological damage since particle size
and mass will determine damage.
63
Figure 3.14: Energy transferred and energy imparted. Two 10 MeV photons enter some water. The
water is divided into small volumes and the energy transferred and energy imparted are shown for each
volume.
Figure 3.15: Dose and kerma. Absorbed dose and kerma tend to dier near the surface where photon
energy is higher.
64
Example: Energy transferred and imparted
A photon undergoes pair production with the subsequent interactions shown in the gure below.
What are (a) the energy transferred in the volume, and (b) the energy imparted in the volume?
h
1
h
2
h
2
h
3
h
4
Solution: (a) Since the photon transfers all its energy to the positron and electron, the energy
transferred is E
tr
= h
1
. (b) The energy imparted is
E = E
in
E
out
= h
1
2h
2
h
3
Buildup
When a beam of photons enters a substance, some photons will undergo an interaction that produces new
photons. We have so far neglected these secondary photons, but they can also undergo interactions. In fact
there can be a cascade of several generations of interacting photons. The secondary interactions produce
photons and electrons that contribute to the energy transferred and the energy imparted. In general, the
energy transferred and energy imparted will be larger when the secondary photons are included. The buildup
factor is dened as the ratio for the quantity including secondary radiation to the quantity for primary
radiation only. For example, for primary radiation the energy uence is
(x) =
0
e
x
(3.68)
To include the secondary radiation, we include the buildup factor
(x) = B(x)
0
e
x
(3.69)
65
3.5 Problems
Problem 1: The work function for cesium is 2.9 eV. (a) What is the maximum wavelength photon that will
produce a photocurrent? (b) If 400 nm photons are used what is the maximum kinetic energy
of the emitted electrons? (c) What maximum work function is needed to allow photoelectron
emission using green photons of 500 nm wavelength?
Problem 2: Photons of 400 nm wavelength are incident on a photocathode. As the anode potential is
made more negative, the photocurrent decreases until it reaches zero when the anode voltage
is 0.82 V. Find the work function of the photocathode.
Problem 3: For Compton scattering, derive an equation for the direction of the recoil electron, , in terms
of and
0
.
Problem 4: For Compton scattering, show that the sum of the energies of the scattered photon and the
recoil electron equals the energy of the incident photon.
Problem 5: For Compton scattering, show that a photon cannot transfer all of its energy to a free electron.
Problem 6: Gamma rays of energy 1.02 MeV are scattered from electrons which are initially at rest. Find
the angle for symmetric scattering at this energy (that is, the angles and are equal). What
is the energy of the scattered photons for this case?
Problem 7: A 0.012 nm wavelength beam of x-rays is incident on a foil target. (a) What is the wavelength
and energy of backscattered Compton x-rays? (b) How much energy is given to the foil target
for each backscattered x-ray?
Problem 8: A 1 MeV photon undergoes Compton scattering from a carbon target. The scattered photon
emerges at an angle of 30

. (a) What is the energy of the scattered photon? What is the energy
of the recoil electron? (b) What is the dierential scattering cross section for scattering at an
angle of 30

from one electron? From the entire carbon atom (Z = 6, A = 12)?


Problem 9: What is the energy of a Compton scattered photon at 180

when h
0
m
e
c
2
? At 90

?
Problem 10: Integrate equation (3.15) over all possible scattering angles and show that

C

8r
2
e
3
.
Problem 11: Find the limit of Eq. (3.13) as x .
Problem 12: In Compton scattering, the wavelength shift is independent of the wavelength of the incident
photon. Calculate the fractional wavelength shift (
0
)/
0
for an infrared photon (
0
=
10 nm), an ultraviolet photon (
0
= 100 nm), a low-energy (soft) x-ray (
0
= 1 nm), and a
high-energy x-ray (
0
= 0.01 nm).
Problem 13: A beam of 59.5 keV photons from
241
Am scatters at 90

from some calcium atoms (A = 40).


(a) What is the energy of a Compton-scattered photon? (b) What is the energy of a coherently
scattered photon? (c) What is the recoil energy of the atom in coherent scattering?
Problem 14: A 2.5 MeV photon passes near a stationary atom and produces an electron-positron pair. If all
the energy of the photon goes into creating the pair, what is the speed of each when produced?
Problem 15: Most diagnostic x-rays use photon energies in the range 20100 keV. For carbon, which mech-
anism is most important in this range: photoelectric eect, Compton scattering, coherent
scattering, or pair production?
66
Problem 16: Consider photons of three energies: 0.01, 0.02, and 0.03 MeV. What fraction of the photons at
each energy will be unattenuated after they pass through 0.1 mm of lead ( = 11.35 gcm
3
)?
Problem 17: What will be the attenuation of 40 keV photons in muscle 10 cm thick? Repeat for 200 keV
photons.
Problem 18: Assume that a patient can be modelled by a slab of muscle 20 cm thick of density 1 gcm
3
.
What fraction of an incident photon beam will emerge without any interaction if the photons
have an energy of 10 keV? 100 keV? 1 MeV? 10 MeV?
Problem 19: Muscle and bone are arranged as shown below. Assume the density of muscle is 1.0 gcm
3
and
the density of bone is 1.8 gcm
3
. The attenuation coecients are
E (/)
muscle
(cm
2
g
1
) (/)
bone
(cm
2
g
1
)
60 keV 0.200 0.274
1 MeV 0.070 0.068
Compare the intensity of the emerging beam that has passed through the bone and muscle and
just muscle at the two energies.
Problem 20: Describe how you could use dierent materials to determine the energy of monoenergetic x-rays
of energy about 50 keV by using changes in the attenuation coecient. What materials would
you use?
Problem 21: A beam of monoenergetic photons travels throuch a sample made up of two dierent materials
of unknown thickness x
1
and x
2
, as shown below. The attenuation coecients at two dierent
energies, E
a
and E
b
, are accurately known. They are
1
(a),
2
(a),
1
(b), and
2
(b). One
measures accurately the log of the ratio of the number of photons emerging from the sample
to the number entering R = ln(N
0
/N), at each energy so that R
a
and R
b
are known. Find an
expression for x
2
in terms of R
a
, R
b
, and the attenuation coecients.
67
Problem 22: You wish to use x-ray uorescence to detect lead that has been deposited in a patients bone.
You shine 100 keV photons on the patients bone and want to detect the 73 keV uorescence
photons which are produced. The incident photon uence is
0
= 10
12
photons m
2
. There are
10
14
lead atoms (1 nanomole) in the region illuminated by the incident beam. The photoelectric
cross section is 1.76 10
25
m
2
atom
1
. The uorescence yield is W = 0.94. Assume for
simplicity that the uorescence photons are emitted uniformly in all directions. The detector
has a sensitive area 12 cm and is located 10 cm from the lead atoms. How many uorescence
photons are detected?
Problem 23: A 5 keV photon strikes a calcium atom. The following events take place:
(a) A K-shell photoelectron is ejected.
(b) A K

photon is emitted. This corresponds to the movement of a hole from the K shell to
the L shell.
(c) An electron in the M shell goes to the L shell and an M shell electron is emitted.
Give the excitation energy of the atom, the total energy in the form of photons, and the total
energy in the form of electron kinetic energy at each stage. Use the following data for calcium:
Z = 20, A = 40, B
K
= 4000 eV, B
L
= 300 eV, B
M
= 40 eV. Ignore dierences in subshells.
Problem 24: Derive Eq. (3.48). Show that it simplies to the non-relativistic equation (Eq. (3.47)) when
v/c 1.
Problem 25: Prove that if a particle of mass M
1
and kinetic energy T collides head on with a particle of
mass M
2
which is at rest, the energy transferred to the second particle is 4TM
2
/M
1
or 2M
2
V
2
in the limit M
2
M
1
. The maximum energy is transferred when the particles move apart
along the line of motion of the incident particle.
Problem 26: (a) Suppose the transmission function of a projectile is known to be F(x) = 1x
2
. What are the
mean range and median range? (b) Find the mean and median range if F(x) = 11/(1x
2
).
Problem 27: Suppose that a photon of energy h enters a volume of material and produces an electron-
positron pair. Both particles come to rest in the volume, and the positron annihilates with an
electron that was already in the volume. Both annihilation photons leave the volume. Show
that the formal denition of energy transfer agrees with the common-sense answer that it is the
kinetic energy of the electron and positron, which is h 2m
e
c
2
. What is the energy imparted?
Problem 28: What are the energy transferred, and the energy imparted in the volume shown?
68
Chapter 4
Crystallography
An ideal crystal is constructed by the innite repetition of identical structural units in space. In the simplest
crystal, the structural unit is a single atom such as copper, gold or an alkali metal. The structural unit can
also be a more complex molecule, like a protein. Examples of crystals are shown in Fig. 4.1. Gold (left) is
a simple crystal with a single atom in a cubic structure. Diamond (center) has a slightly more complicated
crystal structure. Salt (right) has two dierent atoms forming the crystalline structure.
4.1 Lattice structure
Crystals can be described in terms of a lattice. An atom or group of atoms is attached to each lattice point.
These groups of atoms are called the basis. The lattice itself is determined by three translation vectors a
1
,
a
2
, a
3
such that the atomic arrangement looks the same when viewed from the point r as when viewed from
the point
r

= r +u
1
a
1
+u
2
a
2
+u
3
a
3
, (4.1)
where u
1
, u
2
, u
3
are integers The set of all points r

is a lattice. A simple lattice is shown in Fig. 4.2.


The translation vectors are called primitive if any two points r and r

from which the atomic arrangement


looks the same always satisfy
r

= r +u
1
a
1
+u
2
a
2
+u
3
a
3
. (4.2)
The primitive lattice vectors are the shortest possible vectors that satisfy this relation. The cell created by
these vectors is the primitive lattice cell. This denes the smallest cell that serves as a building block for the
crystal. There is always one lattice point per primitive cell each vertex of the cell is shared by neighbouring
cells, so it averages to one.
Example: Counting lattice points
Show that a simple cubic lattice, whose primitive cell is shown below, has only one lattice point
in the primitive cell.
Solution: There are eight lattice points on the cube. Each lattice point is shared by eight dierent
cubes, so only 1/8 of the point contributes. There are eight lattice points each contributing 1/8
of an ocial lattice point, for a net of one lattice point per primitive cell.
69
Figure 4.1: Crystals. Some examples of crystalline structures include gold (left), diamond (center), and
salt (right).
a
1
a
2
Figure 4.2: A two-dimensional lattice. A lattice is a set of points related by translation vectors a
1
and
a
2
.
70
Symmetry
Crystals are highly symmetric objects. A symmetric operation is one which changes the position of the
lattice points in such a way that there are lattice points at exactly the same places after the operation as
before. Crystals exhibit the following symmetries:
translational symmetry as seen before, crystals can be moved by some integer multiple of the primitive
lattice vectors
T = r +u
1
a
1
+u
2
a
2
+u
3
a
3
. (4.3)
mirror symmetry crystals can be mirror reected through some mirror plane.
inversion symmetry crystals can be inverted, r r.
rotational symmetry crystals can be rotated about an axis through a lattice point. Dierent lattices
have dierent rotational symmetries. Lattices exist that have one-, two-, three-, four-, and six-fold
rotational symmetry. These correspond to rotations through 2, , 2/3, /2, and /3. There are no
lattices with other rotational symmetries, such as 2/5 or 2/7 since these symmetries will not ll all
space.
4.1.1 Bravais lattices
There are ve basic types of lattices in two dimensions these are known as Bravais lattices. They are shown
in Fig. 4.3.
1. Oblique the primitive vectors have dierent lengths and the angle between them is not 90

2. Rectangular the primitive vectors have dierent lengths and the angle between them is 90

3. Centered rectangular the primitive vectors have dierent lengths and the angle between them is not
90

; it is composed of alternating oset rows and columns


4. Hexagonal the primitive vectors are equal in length and the angle between them is 120

5. Square the primitive vectors are equalt in length and the angle between them in 90

There are fourteen types of Bravais lattices in three dimensions, as shown in Fig. 4.4. They are grouped
into seven classes of cells based on their cubic geometry: triclinic, monoclinic, orthorhombic, tetragonal,
cubic, trigonal, and hexagonal. Within each type, there are variations determined by the placement of the
basis
Primitive basis is placed on the vertices of the conventional cell only
Body-centered basis is placed on the vertices and in the center of the conventional cell
Side-centered basis is placed on the vertices and in the center of two opposite sides of the conventional
cell
Face-centered basis is placed on the vertices and in the center of all sides of the conventional cell
71
Figure 4.3: Two-dimensional Bravais lattices. The ve basic Bravais lattices in two dimensions.
4.1.2 Primitive cells
Note that the basic geometric unit (conventional cell) is not necessarily the same as the primitive cell. For
example, the primitive cubic lattice has a unit cube as both the conventional cell and the primitive cell. For
body-centered and face-centered cubic lattices (bcc, fcc), the primitive cell is not the same as the conventional
cell (cube). One way to tell is to count the number of lattice points in the cell. The primitive cubic lattice
has one lattice point per cube, so this is the primitive cell for that lattice, but the bcc has two and the fcc
has 4, so the cube is not the primitive cell for these lattices.
72
Figure 4.4: Three-dimensional Bravais lattices. The fourteen basic Bravais lattices in three dimensions.
73
Figure 4.5: Primitive vectors of BCC and FCC lattices. The primitive vectors for the BCC lattice
(left) and the FCC latice (right) do not run along the edge of the cube.
Primitive vectors
For a bcc lattice, the primitive vectors connect a point at the vertex to the points in the center
of the cube. The vectors are given by
a
1
=
a
2
( x + y z)
a
2
=
a
2
( x + y + z) (4.4)
a
3
=
a
2
( x y + z), (4.5)
where x, y, and z run along the edges of the conventional cube. They vectors are shown in Fig.
4.5 (left).
For an fcc lattice, the vectors connect a point at the vertex to the points on the face of the cube.
The vectors are given by
a
1
=
a
2
( x + y)
a
2
=
a
2
( y + z) (4.6)
a
3
=
a
2
( x + z), (4.7)
where x, y, and z run along the edges of the conventional cube. Primitive vectors for the fcc are
shown in Fig. 4.5 (right).
While it is often dicult to visualize the primitive vectors or the primitive cell for a crystal, there is a
method for nding the primitive cell. It is known as the Wigner-Seitz method.
1. Choose a lattice point
2. Draw a line from this point to each of its nearest neighbours
74
Figure 4.6: The Wigner-Seitz primitive cell. An example of nding the Wigner-Seitz primitive cell in
two dimensions. The dashed lines connect the lattice point to its nearest neighbours. The solid lines are
perpendicular to the dashed line and enclose the primitive cell.
3. Draw planes (lines) perpendicular to the dotted lines (i.e., planes with a normal that is parallel to the
line)
4. Collect all of the space within the interior of the volume (area) formed by the planes (lines).
The method essentially cuts o the lattice point from all its nearest neighbours, leaving only a single lattice
point in the cell. This, of course, is the denition of a primitive cell. The method is depicted (in two
dimensions) in Fig. 4.6.
4.1.3 Volume of a cell
The volume contained by three vectors a,

b, and c is
V =a (

b c). (4.8)
Example: Volume calculation
Find the volume of the body-centered cubic primitive cell.
Solution: For bcc, the primitive vectors are given by Eqs. (4.4). Putting these into the volume
formula gives
a
2
a
3
=
a
2
2
( x + y)
a
1
( a
2
a
3
) =
a
3
2
.
The volume of a bcc primitive cell is V =
a
3
2
.
75
Figure 4.7: Crystal planes. Examples of the crystal planes used to describe crystals.
4.1.4 Index system for crystal planes
A plane is determined by three non-colinear points. For crystals, the planes are determined by the points
of intersection with lattice vectors (either primitive or nonprimitive). The planes are typically described by
the Miller index system. The indices are determined as follows: Find the points of intersection; take the
reciprocals of these intercepts, and then reduce to three integers having the same ratio. The result, denoted
as (hkl) is the Miller index of the plane. Negative numbers are written with a bar above i.e 3 for 3. Some
examples of crystal planes are shown in Fig. 4.7.
Example: Miller index
Find the Miller indices of the plane that intercepts the axes formed by the primitive lattice
vectors at (4,2,3).
Solution: We are given the intercepts, so we now take the reciprocals
_
1
4
,
1
2
,
1
3
_
. The lowest
common denominator is 12. We multiply the reciprocals by this number to get the Miller indices
for the plane: (364).
4.2 Bragg diraction
Crystal structure can be studied through the diraction of photons (x-rays), electrons or neutrons. The
waves are partially reected (10
3
to 10
5
at each surface) o the planes of the crystal as shown in Fig. 4.8
(left). The reected waves combine to make an interference pattern (Fig. 4.8 (right)). Simple algebra shows
that constructive interference occurs when
2d sin = n (4.9)
76
Figure 4.8: Bragg diraction. Bragg diraction occurs when x-rays are reected o the planes of a crystal
(left) resulting in interference patterns (right) that can be interpreted to determine crystal structure.
where d is the distance between planes. Bragg diraction only occurs when 2d
4.2.1 Fourier analysis
While Bragg diraction can be summarized by Eq. (4.9), properly interpreting the three-dimensional struc-
ture of crystals requires a more detailed analysis. Suppose n(x) describes some property of crystals (electron
density, charge concentration, etc.). Because of the periodic nature of crystals, we have
n(r +T) = n(r), (4.10)
which says that n(x) is also periodic. We can expand n(x) in a Fourier series
n(x) = n
0
+

p
[C
p
cos(2px/a) +S
p
sin(2px/a)], (4.11)
where the ps are positive integers and the factor 2/a ensures that n(x) has a period of a. We can write
the transform more compactly with an exponential
n(x) =

p
n
p
exp(i2px/a). (4.12)
To ensure that n(x) is real, we must have n

p
= n
p
(complex conjugate of n
p
must equal n
p
). The Fourier
coecients are
n
p
=
1
a
_
a
0
n(x) exp(i2px/a) dx. (4.13)
This can easily be extended to three dimensions
n(r) =

G
n
G
exp(iG r) (4.14)
where G is the set of vectors that leave n(r) invariant under crystal translations T = u
1
a
1
+ u
2
a
2
+ u
3
a
3
.
The Fourier coecients are given by
n
p
=
1
V
c
_
cell
n(r) exp(iG r) dV (4.15)
where V
c
is the volume of a cell.
77
4.2.2 Reciprocal lattice
To procede, we need to nd the vectors G. We construct the vectors of the reciprocal lattice, which are
dened as
b
1
= 2
a
2
a
3
a
1
(a
2
a
3
)
b
2
= 2
a
3
a
1
a
1
(a
2
a
3
)
(4.16)
b
3
= 2
a
1
a
2
a
1
(a
2
a
3
)
.
If a
1
, a
2
, and a
3
are primitive vectors of the crystal lattice, then b
1
, b
2
, and b
3
are primitive vectors of the
reciprocal lattice. b
1
, b
2
, and b
3
are orthogonal to two axis vectors of the crystal lattice, so we have the
property
b
i
a
j
= 2
ij
(4.17)
Example: Reciprocal lattice
Find the reciprocal lattice of a bcc lattice.
Solution: Recall the primitive translation vectors of the bcc lattice are
a
1
=
a
2
( x + y z)
a
2
=
a
2
( x + y + z)
a
3
=
a
2
( x y + z).
We can work out the primitive translation mathbftors of the reciprocal lattice
b
1
= 2
a
2
a
3
V
c
= 2
2
a
3
a
2
( x + y + z) ( x y + z)
=
2
a
( y + z).
The remaining vectors are
b
2
=
2
a
( x + z)
b
3
=
2
a
( x + y).
Careful inspection of these vectors shows that they are the primitive vectors of an fcc lattice (Eqs.
(4.6)).
The reciprocal lattice is a rather abstract construct, but in essence it is a lattice in the Fourier space
of the crystal. If we could resolve the crystal in a microscope, we would see the actual lattice in space.
In a similar manner, an x-ray diraction pattern is a picture of the reciprocal lattice. Since we know the
relationship between the real lattice and the reciprocal lattice, we can use the x-ray diraction pattern to
nd the real lattice.
Points on the reciprocal lattice are given by
G = v
1
b
1
+v
2
b
2
+v
3
b
3
, (4.18)
78
which means that G is a reciprocal lattice vector. We can show that these G vectors are the ones we need
in the Fourier series. Recall that G is the set of vectors for which n(r) remains invariant under crystal
translations T = u
1
a
1
+u
2
a
2
+u
3
a
3
;
n(r +T) =

G
n
G
exp(iG r) exp(iG T). (4.19)
The second exponential can be expanded
exp(iG T) = exp[i(u
1
a
1
+u
2
a
2
+u
3
a
3
) (v
1
b
1
+v
2
b
2
+v
3
b
3
)], (4.20)
and we can use the relationship between real lattice vectors and primitive lattice vectors, Eq. (4.17),
exp(iG T) = exp[i2(u
i
v
i
+u
2
v
2
+u
3
v
3
)] = 1 (4.21)
since u
i
v
i
+ u
2
v
2
+ u
3
v
3
is an integer. Thus we have n(r + T) = n(r). This means that the Fourier
representation of a function that is periodic in the crystal lattice contains components only at the reciprocal
lattice vectors.
4.2.3 Diraction in Fourier space
Suppose a wave, e
ik r
, is incident on a crystal. It undergoes reection within the sample and the wave that
emerges is e
ik

r
. Dene the scattering amplitude as
F =
_
n(r) exp[i(k k

) r] dV, (4.22)
where n(r) is specically the local electron concentration. We can write n(r) as a Fourier series
F =
_

G
n
G
exp[i(Gk) r] dV (4.23)
When k = G, we have F = V

n
G
. F is very small otherwise and can be neglected at other points. This
condition tells us the scattering amplitude is large when the change in wavevector equals a reciprocal lattice
vector.
Bragg diraction is caused by elastic scattering, so the photon does not change frequency or wavenumber
k
2
= k
2
. We can re-write the diraction condition
k +G = k

(k +G)
2
= k
2
(4.24)
2k G+G
2
= 0
2k G = G
2
,
which gives us the condition for constructive interference. You will show in your homework assignment that
d(hkl) = 2/|G|. Using this result we can re-write the diraction condition
2k G = G
2
2
_
2

_
|G| cos(k, G) = |G|
2
(4.25)
2
_
2

_
cos(k, G) =
2
d(hkl)
2d sin = n
which is consistent with the original simple Bragg condition for constructive interference (Eq. (4.9)). An
example of an x-ray diraction pattern for DNA is shown in Fig. 4.9.
79
Figure 4.9: DNA diraction pattern. The x-ray diraction pattern taken by Rosalind Franklin that
helped determine the structure of DNA.
80
Figure 4.10: Laue cones. The Laue cones show where reected waves must be in the reciprocal lattice.
Constructive interference occurs along the intersections of the cones.
Laue equations
If you break up G into its component vectors and take the dot product with k, the diraction condition
becomes
a
1
k = 2v
1
, a
2
k = 2v
2
, a
3
k = 2v
3
(4.26)
This denes three cones in space along the directions of a
1
, a
2
, and a
3
(Fig. 4.10). The reected wave must
satisfy all three conditions and lies along the intersection of the cones.
Brillouin zones
The rst Brillouin zone is dened as a Wigner-Seitz primitive cell in the reciprocal lattice. Construction
of the zone follows construction of a Wigner-Seitz cell but in the reciprocal lattice. More generally, the nth
Brillouin zone consists of the set of points that can be reached by crossing n 1 Bragg planes as shown in
Fig. 4.11.
4.2.4 Structure Factor
Recall the scattering amplitude is Eq. 4.22,
F =
_
n(r) exp[i(k k

) r] dV.
When the diraction condition is met k = G, the scattering amplitude becomes
F =
_
n(r) exp[iG r] dV. (4.27)
For a crystal of N cells the scattering amplitude is
F = N
_
n(r) exp[iG r] dV = NS
G
81
Figure 4.11: Brillouin zones. The rst three Brillouin zones of cubic lattices.
where S
G
is the structure factor.
We write the electron concentration as a superposition of electron concentration functions n
j
associated
with each atom j of the cell. Suppose r
j
is the vector to the center of atom j, then n
j
(r r
j
) is the
contribution of that atom to the electron concentration at r. The electron concentration at any point due
to all atoms is
n(r) =
s

j=1
n
j
(r r
j
). (4.28)
This allows us to re-write the structure factor
S
G
=
_
s

j=1
n
j
(r r
j
) exp[iG r] dV (4.29)
S
G
=
s

j=1
exp(iG r
j
)
_
n
j
() exp[iG ] dV.
The structure factor has two parts we dene the atomic form factor
f
j
=
_
n
j
() exp[iG ] dV. (4.30)
The atomic form factor describes scattering from an isolated atom; the structure factor describes scattering
from the unit cell of a crystal.
We re-write the structure factor:
S
G
=
s

j=1
f
j
exp(iG r
j
). (4.31)
Recall G = v
1
b
1
+v
2
b
2
+v
3
b
3
and r
j
= x
j
a
1
+y
j
a
2
+z
j
a
3
so
G r
j
= (v
1
b
1
+v
2
b
2
+v
3
b
3
) (x
j
a
1
+y
j
a
2
+z
j
a
3
) (4.32)
= 2(v
i
x
j
+v
2
y
j
+v
3
z
j
).
82
The structure factor becomes
S
G
=

j
f
j
exp[i2(v
i
x
j
+v
2
y
j
+v
3
z
j
)]. (4.33)
Example: Structure factor
Calculate the structure factor for the bcc atom.
Solution: The rst atom in the bcc lattice is at the origin: x
j
= y
j
= z
j
= 0
S(v
1
, v
2
, v
3
) = f exp(2i(0)) = 1
This atom actually represents the atoms at all eight vertices since they all contribute 1/8 of an
atom to the cell. The central atom is at x
j
= y
j
= z
j
= 1/2.
S(v
1
, v
2
, v
3
) = f exp(2i1/2(v
1
+v
2
+v
3
))
= f exp(i(v
1
+v
2
+v
3
)).
The structure factor then is
S(v
1
, v
2
, v
3
) = f(1 + exp[i(v
1
+v
2
+v
3
)]).
The structure factor for more complex molecules is calculated in much the same way: one atom at a time.
Find the vector from the origin to the center of each atom to calculate its contribution to the structure
factor. Then sum all the contributions for all atoms in the cell.
4.2.5 The phase problem
Recall the scattering amplitude is
F =
_
n(r) exp[iG r] dV (4.34)
We can interpret this as the Fourier transform of the electron concentration (density).So to nd the electron
concentration, simply take the inverse transform
n(r) =
1
V

G
F exp[iG r] (4.35)
This equation requires that we sum over all structure factors in reciprocal space. In practice, the intensity
of scattered waves decreases with increasing angle, giving a limit to how well we can calculate electron
concentration. The largest angle for which diraction can be measured determines the resolution of our
measurements.
Additionally, in a diraction experiment we do not really measure F, we actually measure |F|
2
, so we
have information about the amplitude, but not the phase of the structure factor. To correctly determine the
electron concentration, we really need both. The top row of Fig. 4.12 shows two images and their Fourier
transforms. If we combine the Fourier amplitudes of the duck with the phases of the cat and take the inverse
transform, we reconstitute an image that appears more like the cat than the duck. This means that the
phase information is crucial for assembling the correct picture of crystals. Several experimental techniques
have been developed to determine this information.
83
Figure 4.12: The phase problem. Combining Fourier amplitudes from one image with the phase informa-
tion from another image and taking the inverse transform shows that the phase information contains most
of the details of the image.
Molecular replacement
In molecular replacement, the phases for an unknown molecule are inferred from the phases for a similar
known molecule. We can calculate the Patterson function for both molecules
P(r) =
1
V

G
|F|
2
exp[iG r] (4.36)
Identical regions of the molecules will produce identical Patterson functions, so the phase of these atoms can
be inferred.
Isomorphous replacement
In isomorphous replacement, the crystal is grown alone and with a heavy metal that grows in the same shape
(isomorphous). The structure factor is measured for both compounds and we must have:
F
P
+F
H
= F
PH
(4.37)
Measurement of |F
P
|
2
and |F
PH
|
2
leads to two possible phases, the process needs to be repeated with a
second heavy metal to uniquely determine the phase, as shown in Fig. 4.13.
Anomolous dispersion
In anomolous dispersion, we use the fact that some molecules will not just diract x-rays. Molecules with
the correct energy level spacing will absorb some of the x-rays. We can measure the diraction pattern at
dierent wavelengths to identify the anomolous dispersion and use the anomaly to determine the phase.
84
Figure 4.13: Isomorphous replacement. Isomorphous replacement uses a heavy metal that grows in the
same shape as the unknown crystal, but measurement of both leads to two possible phases (left). Using a
second isomorphous crystal leads to a unique solution (right).
Measurement of x-ray absorption as a function of energy gives a sharp jump at the frequency where
absorption happens. At slightly higher energies, there is some oscillation as emitted waves are scattered by
nearby atoms. This is the extended x-ray absorption ne structure. The phase of the wave is given by
Phase =
4d

= 4d
p
h
(4.38)
=
4d
h
_
2m
e
(h E
ion
)
Once the phases are determined, we can construct a 3D picture of the electron density. This does not
identify specic atoms without a prior knowledge of the composition of the molecule. The molecular/atomic
components of proteins are known, crystallography determines the shape. Putting the parts into the electron
cloud is typically done manually.
4.3 Protein Crystals
Proteins are large, irregularly shaped molecules that do not pack into crystals very willingly or eciently.
A solvent, usually containing a salt, which keeps the protein folded is needed to prompt crystallization of
proteins. The crystallization solution ends up being 20-80% of the crystal (by volume). This makes protein
crystals very soft.
4.3.1 Making protein crystals
Crystal growth in solution is characterized by two steps: nucleation of a microscopic crystallite followed by
growth of the crystal. The solution conditions that favour nucleation are not always the same conditions
that favour growth of the crystal. The goal is to identify solution conditions that favour the development
of a single, large crystal, since larger crystals oer improved resolution of the molecule. The ideal solution
conditions should disfavour nucleation but favour growth, so that only one large crystal forms per droplet.
If nucleation is favoured too much, a shower of small crystals will form in the droplet, rather than one
large crystal. If favoured too little, no crystal will form whatsoever (Fig. 4.15). A method involving vapour
diusion is a good balance since solutions in which there is a small nucleus are allowed to slowly grow as the
concentration of protein increases.
The most common methods of protein crystallization involve the use of vapour diusion: the hanging
drop and the sitting drop. In either method, a small drop containing a 1:1 mix of protein and crystallization
85
Figure 4.14: Extended X-ray absorption ne structure. The x-ray absorption spectrum of a molecule
can be used to determine the phase of the structure factor.
Figure 4.15: Protein crystallization. Protein crystallization is a delicate balance between nucleation and
crystal growth.
86
Figure 4.16: Vapor diusion methods of protein crystallization. Protein crystals are typically made
using the hanging drop method (left) or the sitting drop method (right).
solutions is placed inside a sealed container with a reservoir of crystallization solution. In the hanging drop
method, the drop hangs over the reservoir (Fig. 4.16 (left)); in the sitting drop method, the drop sits on a
plateau next to the reservoir (Fig. 4.16 (right)). The concentration of crystallizing agent is lower in the drop
than in the reservoir, so water will vapourize from the drop to increase the concentration in the drop.
87
4.4 Problems
Problem 1: In a cesium chloride crystal, planes of cesium atoms alternate with planes of chlorine atoms.
On every other plane, the atoms lie at the corners of squares, positioned like the squares of
a checkerboard. Atoms on one plane are directly above the centers of the squares formed by
atoms on the plane below. The distance between neighbouring atoms on a plane is a = 4.11

A
and the separation of adjacent planes is 1/2a. Find a set of primitive lattice vectors and
associated basis vectors for this crystal.
Problem 2: What is the inherent symmetry of the plane square lattice?
Problem 3: Given a point with coordinates x, y in the xy plane, nd its coordinates after each of the
following operations: (a) a rotation by /2 about the z-axis; (b) a rotation by 2/3 about the
z-axis; (c) a rotation by /2 about the y-axis; and (d) a reection in the yz plane.
Problem 4: The angles between the tetrahedral bonds of diamond are the same as the angles between the
body diagonals of a cube. Use elementary vector analysis to nd the value of the angle.
Problem 5: The diagram below shows a plan view of cubic ZnS (zinc blende), looking down the z axis. The
numbers attached to some of the atoms represent the heights of the atoms above the z = 0
plane expressed as a fraction of the cube edge a. Unlabelled atoms are at z = 0. (a) What is
the Bravais lattice type? (b) Describe the basis. (c) Calculate the nearest neighbour distances,
Zn-Zn, Zn-S, and S-S.
Problem 6: Position vectors for the lattice points in two dierent lattices are given by (a) r = (10n
1
+
9n
z
+19n
3
)(a/10) x+6(n
2
+n
3
)(a/5) y +2n
3
a z and (b) r =
1
2
(2n
1
+n
2
)a x+
1
2

3n
2
a y +2n
3
a z.
Here n
1
, n
2
, and n
3
are integers and a is a length. In each case, nd a set of primitive vectors
and identify the Bravais lattice type.
Problem 7: Take the primitive lattice vectors for a body-centered cubic lattice to be a
1
=
1
2
a( x + y + z),
a
2
=
1
2
a( x + y + z), and a
3
=
1
2
a( x y + z). Express as linear combinations of these vectors:
(a) position vectors for the eight cube corners; (b) position vectors for the eight points within
the cubic unit cell that are one-fourth a body diagonal from cube corners; and (c) the lattice
vector from a y to a x, diagonally across the cell base.
Problem 8: Primitive lattice vectors for a body-centered tetragonal lattice are a
1
=
1
2
a( x + y)
1
2
c z,
a
2
=
1
2
a( x + y) +
1
2
c z, and a
3
=
1
2
a( x y) +
1
2
c z, where a is a side of the square base and c
is the height of the conventional unit cell. Initially c > a; the crystal is now compressed along
the z axis. (a) For what value of c does the lattice become body-centered cubic? (b) For what
value of c does the lattice become face-centered cubic? Give your answers in terms of a.
Problem 9: Find the position vectors of the eight cube corners and six cube face centers in terms of the
primitive lattice vectors a
1
=
1
2
a( x + y), b =
1
2
a( y + z), and c =
1
2
a( x + z) for a face-centered
cubic lattice.
88
Problem 10: There are only 14 Bravais lattices; why are the following not included in the 14? (a) base-
centered tetragonal, (b) face-centered tetragonal, (c) orthorhombic plus a body-centered point.
Problem 11: For each of the following sets of primitive lattice vectors, identify the Bravais lattice type and
give the dimensions of the conventional unit cell in terms of a, b, and c: (a) (a/2) x + (a/2) y,
a y, (a/

2) z; (b) (a/2) x + (a/2) y, a y, z; (c) a x + 2b y, b y, c z; (d)


1
2
a x +
1
2
b y, b y, c z.
Problem 12: Sodium chloride has a cubic unit cell (NaCl structure) with an edge of length a = 5.63 10
8
cm.
The atomic weight of sodium is 23 and the molecular weight of chlorine is 71. Calculate the
density of sodium chloride.
Problem 13: A unit cell for zinc has a base that is a rhombus with edge a = 2.66

A and internal angle
= 60

. The sides are rectangles perpendicular to the base and have length c = 4.95

A. There
are two zinc atoms per unit cell. Find the cell volume and the density of zinc.
Problem 14: SrTiO
3
has what is known as the ideal perovskite structure. Strontium atoms are the corners
of cubes, titanium atoms are at cube body centers, and oxygen atoms are at cube face centers.
Take the cube edge length to be a. (a) What is the Bravais lattice type? (b) Verify that there
are 3 oxygen atoms, 1 titanium atom, and 1 strontium atom for each primitive unit cell in the
crystal. (c) Write down a set of primitive lattice vectors and associated basis vectors for this
structure.
Problem 15: The face-centered cubic is the most dense and the simple cubic is the least dense of the three
cube Bravais lattices. The diamond structure is less dense than any of these. Suppose identical
solid spheres are distributed through space in such a way that their centers lie on the points of
each of these four structures, and spheres on neighbouring points just touch, without overlap-
ping. (Such an arrangement of spheres is called a close-packing arrangement.) Assuming that
the spheres have unit density, show that the density of a set of close-packed spheres on each of
the four structures (the packing fraction) is:
fcc:

2/6 = 0.74
bcc:

3/8 = 0.68
sc: /6 = 0.52
diamond:

3/16 = 0.34.
Problem 16: Find Miller indices for the following lattice planes: (a) a plane parallel to both a
1
and a
3
; (b)
a plane parallel to both 3a
1
+ a
3
and a
2
; (c) the plane containing the points 3a
1
, 2a
2
, and
1
2
(a
1
+a
2
+a
3
).
Problem 17: For a simple cubic lattice with cube edge a, place the origin at a lattice point and take the
primitive lattice vectors to be along cube edges. Find the Miller indices for a plane that
intersects the x-axis at 4a, the y-axis at 3a, and the z-axis at 2a. Then nd the separation of
adjacent planes parallel to the given plane. Finally, nd the intercepts on the crystal axes of
the parallel plane closest to the origin, exclusive of the one through the origin.
Problem 18: Let N
n
be the number of nth nearest neighbours of a given Bravais lattice point (e.g. in a
simple cubic Bravais lattice N
1
= 6, N
2
= 12, etc.). Let r
n
be the distance to the nth nearest
neighbour expressed as a multiple of the nearest neighbour distance (e.g., in a simple cubic
Bravais lattice r
1
= 1,r
2
=

2). Make a table of N


n
and r
n
for n = 1, . . . , 6 for the fcc, bcc,
and sc Bravais lattices.
Problem 19: Prove that the Wigner-Seitz cell for any two-dimensional Bravais lattice is either a hexagon or
a rectangle.
89
Problem 20: Prove that the volume of a Wigner-Seitz cell for the body-centered cubic lattice is one half the
volume of the conventional cell.
Problem 21: Consider a plane hkl in a crystal lattice. (a) Prove that the reciprocal lattice vector G =
hb
1
+ kb
2
+ lb
3
is perpendicular to this plane. (b) Prove that the distance between two
adjacent parallel planes of the lattice is d(hkl) = 2/|G|. (c) Show for a simple cubic lattice
that d
2
= a
2
/(h
2
+k
2
+l
2
).
Problem 22: Consider a general oblique lattice with lattice vectors a, b, and c. Let one (hkl) plane pass
through the origin, another intercept the axes at a/h, b/k, c/l. (a) What are the direction
cosines for a vector d drawn from the origin to the second (hkl) plane? (b) What is the
perpendicular distance between (hkl) planes of the simple cubic lattice?
Problem 23: X-rays with a 0.12 nm wavelength produce a rst-order diraction peak at a Bragg angle of
24

. What crystal spacing gave rise to this diraction?


Problem 24: 1.54

A x-rays are incident along the z axis on two identical atoms separated by 3.2

A. Their
relative displacement is in the yz plane and makes the angle
r
with the z axis. A detector is
moved in the yz plane around the atoms. For (a)
r
= 0

and (b)
r
= 45

, nd the positions
of the detector where it records maxima of scattered intensity.
Problem 25: Consider a simple cubic crystal structure with cube edge a = 3.50

A and suppose it is used
to scatter 3.10

A x-rays. Find all sets of planes that satisfy the Bragg condition and, for each
peak, nd the Bragg angle .
Problem 26: The primitive translation vectors of the hexagonal space lattice may be written as:
a
1
= 3
1/2
a
2
x +
a
2
y, a
2
= 3
1/2
a
2
x +
a
2
y, a
3
= cz.
(a) Show that the volume of the primitive cell is (3
1/2
/2)a
2
c. (b) Show that the primitive
translations of the reciprocal lattice are:
b
1
=
2
3
1/2
a
x +
2
a
y, b
2
=
2
3
1/2
a
x +
2
a
y, b
3
=
2
c
z.
Problem 27: Let (ijk) be a plane that intersects the lattice at points a
1
/i, a
2
/j, and a
3
/k. Show that the
reciprocal lattice vector given by ib
1
+jb
2
+kb
3
is perpendicular to this plane.
Problem 28: (a) Prove that the reciprocal lattice primitive vectors satisfy
b
1
(b
2
b
3
) =
(2)
3
a
1
(a
2
a
3
)
.
(b) Suppose primitive vectors are constructed from the b
i
in the same manner as the b
i
are
constructed from the a
i
. Prove that these vectors are just the a
i
themselves; i.e. show that
2
b
2
b
3
b
1
(b
2
b
3
)
= a
1
etc.
Problem 29: Prove that the distance between adjacent parallel planes in the real lattice is proportional to
the magnitude of the corresponding recprocal lattice vector, and determine the constant of
proportionality.
Problem 30: Show that the volume of the rst Brillouin zone is (2)
3
/V
c
where V
c
is the volume of a crystal
primitive cell. You will need the vector identity (c a) (a b) = (c a b)a.
90
Problem 31: Suppose that in a linear crystal there are identical point scattering centers at every lattice point

m
= ma where m is an integer. The total scattered radiation amplitude will be proportional
to F =

exp[ima k]. The sum over M lattice points is


F =
1 exp[iM(a k)]
1 exp[i(a k)]
by the use of the series
M1

m=0
x
m
=
1 x
M
1 x
.
(a) The scattered intensity is proportional to |F|
2
. Show that
|F|
2
=
sin
2
(
1
2
M(a k))
sin
2
(
1
2
(a k))
(b) We know that the diraction maximum appears when a k = 2h, where h is an integer.
We change k and dene in a k = 2h + such that gives the position of the rst zero
in sin(1/2M(a k)). Show that = 2/M, so that the width of the diraction maximum is
proportional to 1/M and can be extremely narrow for macroscopic values of M.
Problem 32: (a) Find the structure factor,S, for diamond. (b) Find the zeros of S and show that the allowed
reections of the diamond structure satisfy v
1
+v
2
+v
3
= 4n, where all indices are even and n
is any integer, or else all indices are odd.
Problem 33: What is the structure factor for the face-centered cubic unit cell? From which planes will x-rays
not be reected? Answer the same questions for the body-centered cubic unit cell.
Problem 34: BaTiO
3
has a simple cubic lattice and a basis with atoms having fractional coordinates: Ba
(0,0,0), Ti (
1
2
,
1
2
,
1
2
), and O (
1
2
,
1
2
,0), (
1
2
,0,
1
2
), (0,
1
2
,
1
2
). (a) Sketch the unit cell. (b) Show that
the structure factor for the (00l) Bragg reections is
S
hkl
= f
Ba
+ (1)
l
f
Ti
+ [1 + 2(1)
l
]f
0
where the fs are the atomic form factors.
Problem 35: For the hydrogen atom in its ground state, the number density is n(r) = (a
3
0
)
1
exp(2r/a
0
),
where a
0
is the Bohr radius. Show that the form factor is f
G
= 16/(4 +G
2
a
2
0
)
2
.
Problem 36: Consider a lattice with an n-ion basis. Suppose that the i th ion in the basis, when translated to
r = 0 can be regarded as composed of m
i
point particles of charge z
ij
e, located at positions
b
ij
, j = 1, . . . , m
i
. Show that the atomic form factor f
i
is given by
f
i
=
mi

j=1
z
ij
e
ik bij
.
91
Chapter 5
Spectroscopy
Spectroscopy is the study of any quantity as a function of wavelength or frequency. It is usually used to
refer specically to studies involving light as a function of wavelength or frequency. Typically, absorption
and emission spectra are measured to identify molecules or atoms since the absorption and emission spectra
are determined by energy levels within the molecule or atom. The molecule will emit light when it loses
energy, and will increase its energy when it absorbs light. The frequency of the light emitted or absorbed
is proportional to the dierence between the energy levels: h = E
2
E
1
. Thus, we need to nd ways to
determine the energy levels of molecules and atoms in order to understand the spectra that we observe.
5.1 Determining energy levels
In quantum mechanics, the state of a particle is described by its wavefunction, . The evolution of the
wavefunction is determined by the Schrodinger equation
ih

t
=

H, (5.1)
where

H is the Hamiltonian of the system. Or if the system is stationary (does not vary in time)
E =

H, (5.2)
where E is the energy eigenstate associated with

H. To determine energy levels, we apply Schrodingers
equation to a variety of systems and determine the expected absorption and emission spectra.
5.1.1 Particle in a box
A particle of mass m is conned between two walls at x = 0 and x = L. The potential energy is 0 inside the
box and immediately rises to innity at the walls. Since the potential is innite outside the box, the particle
cannot be found there and the wavefunction is 0. The Hamiltonian is

H =
h
2
2m
d
2
dx
2
+V (x) =
h
2
2m
d
2
dx
2
. (5.3)
Schrodingers equation for this system is
h
2
2m
d
2
dx
2
(x) = E(x). (5.4)
We assume a general solution of the form
(x) = Asin kx +Bcos kx (5.5)
92
where k = 2/ is the wave number. The derivatives of the wavefunction are
d
dx
(x) = k(Acos kx Bsin kx) (5.6)
d
2
dx
2
(x) = k
2
(Asin kx Bcos kx) = k
2
(x).
Substitution into Schrodingers equation gives
h
2
2m
k
2
(x) = E(x) (5.7)
or
E =
h
2
k
2
2m
. (5.8)
The wavefunction must be continuous everywhere, so we must have (x = 0) = (x = L) = 0
(0) = 0 = Asin 0 +Bcos 0 B = 0 (5.9)
(L) = 0 = Asin kL kL = n,
where n=1, 2, 3, etc.
The probability of nding a particle within a volume dV is

(r)(r)dV . The probability of nding the


particle somewhere in space must be 1 so
1 =
_

(x)(x)dx
1 =
_
L
0
A
2
sin
2
_
nx
L
_
dx
1 = A
2
_
L
0
1
2
_
1 cos
_
2nx
L
__
dx (5.10)
1 =
a
2
2
_
x
L
2n
sin
_
2nx
L
__
L
0
1 =
A
2
L
2
A =
_
2
L
We have now completely solved the problem. The wavefunction is
(x) =
_
2
L
sin
_
nx
L
_
(5.11)
with energies given by
E =
h
2
2m
_
n
L
_
2
=
n
2
h
2
8mL
2
(5.12)
where n=1, 2, 3, etc.
The wavefunctions and probabilities for this system are shown in Fig. 5.1. Note that we cannot have
n = 0 since that leads to (x) = 0 or no particle. This means that the particle always has some energy
the zero-point energy is at n = 1. The spacing between energy levels gets larger as n increases. The higher
energy solutions are harmonics of the lowest energy solutions. The particle can only be found at certain
positions in the box and it can move between this regions without entering any forbidden regions.
For spectroscopy, we are really interested in the transitions between energy levels,
E = E
n2
E
n1
=
h
2

2
2mL
2
(n
2
2
n
2
1
). (5.13)
93
Figure 5.1: Wavefunctions and probabilities for a particle in a box. Wavefunctions (left) and
probabilities (right) of the one-dimensional particle in a box. Higher energy solutions are harmonics of the
lowest energy solutions.
A photon with energy h = E can be absorbed or emitted by this system.
Example: Energy levels of -carotene
Carotenoids are conjugated molecules widely found in biological systems that are responsible for
red/orange pigmentation of animals and plants. They are long molecules of alternating single and
double carbon-carbon bonds. Electrons within the bonds can move along the entire length of the
molecule. To a rst approximation this is a particle in a 1-dimensional well.
Solution: -carotene has 11 double bonds and 10 single bonds. Single bonds are typically 1.46

A;
Double bonds are typically 1.35

A. This means that the length of the box is 29

A. There are
22 electrons within the bonds, so the rst 11 energy levels (2 electrons per level) are lled. This
means that the lowest possible transition is from the 11th to 12th energy levels
E =
h
2

2
2mL
2
(12
2
11
2
) =
23h
2
8mL
2
(5.14)
The wavelength of the light absorbed to cause this transition is
=
8m
e
cL
2
23h
= 1207 nm. (5.15)
94
Momentum
In quantum mechanics, we can only calculate the expectation value of any parameter. The expectation value
of the momentum is
p =
_
L
0

(x)
_
h
i

x
_
(x)dx
p =
2
L
_
L
0
sin
_
nx
L
_
_
h
i

x
_
sin
_
nx
L
_
dx
p =
2 hn
iL
2
_
L
0
sin
_
nx
L
_
cos
_
nx
L
_
dx (5.16)
p =
ih
L
_
sin
2
_
nx
L
__
L
0
= 0.
The particle travels with equal probability to the left and right, so the average is zero. The mean square of
the momentum is
p
2
=
_
L
0

(x)
_
h
i

x
_
2
(x)dx
p
2
=
2
L
_
L
0
sin
_
nx
L
_
_
h
2
i
2

2
x
2
_
sin
_
nx
L
_
dx
p
2
=
2h
2
L
_
n
L
_
2
_
L
0
sin
_
nx
L
_
sin
_
nx
L
_
dx (5.17)
p
2
=
2
L
_
nh
L
_
2
[x (Lnx)]
L
0
p
2
=
h
2
n
2
4L
2
=
h
2

2
.
The expectation value returns the de Broglie relationship.
Position
The average position is
x =
_
L
0

(x)x(x)dx
x =
2
L
_
L
0
xsin
2
_
nx
L
_
dx (5.18)
x =
L
2
.
The particle is in the center of the box on average.
5.1.2 Two-dimensional particle in a box
Most molecules do not have a simple chain-like structure. Molecules consist of two or three dimensions so
we need to extend the particle in a box to more dimensions.
Suppose and electron is trapped in a rectangular area with sides L
x
and L
y
with potential V = 0 inside the
box and V = outside the box. The Hamiltonian now consists of partial derivatives in the two dimensions

H =
h
2
2m
_

2
x
2
+

2
y
2
_
. (5.19)
95
Schrodingers equation then is

h
2
2m
_

2
x
2
+

2
y
2
_
(x, y) = E(x, y). (5.20)
Assume a solution of the form
(x, y) = X(x)Y (y). (5.21)
Substituting into Schrodingers equation:

h
2
2m
_
Y (y)
d
2
X(x)
dx
2
+X(x)
d
2
Y (y)
dy
2
_
= EX(x)Y (y) (5.22)
_
1
X(x)
d
2
X(x)
dx
2
_
+
_
1
Y (y)
d
2
Y (y)
dy
2
_
=
2mE
h
2
.
Since the sum of both terms is a constant, each term must be equal to a constant. We can now separate this
into two equations
_
1
X(x)
d
2
X(x)
dx
2
_
=
2mE
x
h
2
(5.23)
_

h
2
2m
d
2
X(x)
dx
2
_
= E
x
X(x)
and
_
1
Y (y)
d
2
Y (y)
dy
2
_
=
2mE
y
h
2
(5.24)
_

h
2
2m
d
2
Y (y)
dy
2
_
= E
y
Y (y),
where E
x
+ E
y
= E. These are the same equations as a one-dimensional particle in a box, so we know the
solution,
X(x) =
_
2
L
x
sin
_
n
x
x
L
x
_
(5.25)
Y (y) =

2
L
y
sin
_
n
y
y
L
y
_
,
for a complete solution of
(x, y) =
_
2
L
x
sin
_
n
x
x
L
x
_

2
L
y
sin
_
n
y
y
L
y
_
, (5.26)
with energy of
E =
h
2
8m
_
_
n
x
L
x
_
2
+
_
n
y
L
y
_
2
_
. (5.27)
This system shows degeneracy. Consider the states (n
x
= 1, n
y
= 2) and (n
x
= 2, n
y
= 1) on a
square L
x
= L
y
. These states have dierent wavefunctions as shown in Fig. 5.2, but have the same energy.
Rectangles may also have degenerate states if L
x
/L
y
is a rational number.
96
Figure 5.2: Degeneracy. Two wavefunctions are degenerate if they result in the same energy.
5.1.3 Circular well
Molecules are not typically square a circle is a better model for rings within the molecules. Suppose that a
particle is in a circular well of radius a (V = 0 inside, V = outside). The Hamiltonian in polar coordinates
is

H =
h
2
2m
_
1
r

r
_
1
r

r
_
+
1
r
2

2
_
(5.28)
and Schrodingers equation becomes

h
2
2m
_
1
r

r
_
1
r

r
_
+
1
r
2

2
_
(r, ) = E(r, ). (5.29)
We try separation of variables (r, ) = R(r)() which gives
h
2
2m
_
r
R(r)

r
_
1
r
R(r)
r
_
+
2mEr
2
h
2
_
+
h
2
2m
_
1
()

2
()

2
_
= 0. (5.30)
Since the sum of both terms is a constant, each term must be equal to a constant and we can separate this
into two equations.

2
()

2
= m
2
() (5.31)
1
r

r
_
1
r
R(r)
r
_
+
2mER(r)
h
2
=
m
2
r
2
R(r).
The equation for R(r) is the Bessel equation whose solutions are known. The equation for has a solutions
of e
im
which is single-valued only if m is an integer. This describes the number of oscillations around the
ring.
5.1.4 Rotational spectra
In addition to the changes in motion of the electrons, changes in motion of the entire molecule can create
emission and absorption spectra. We rst consider the rotational motion of the molecule. To begin, we look
97
Figure 5.3: The rigid rotor model of a diatomic molecule. A diatomic molecule can be modelled as
two masses bound together at xed length. The molecule can rotate around its center of mass.
at the rigid rotor model of the diatomic molecule (Fig. 5.3). In this model, the two atoms of the molecule
are represented as two masses connected by a rigid rod of xed length. Since we are considering rotation, it
is convenient to write the Hamiltonian in three dimensions:
H =
h
2
2
_
1
r
2

r
_
r
2

r
_
+
1
r
2
sin
_

sin

_
+
1
r
2
sin
2

2
_
, (5.32)
where is the reduced mass, and and are the polar and azimuthal angles in spherical coordinates. Which
means that Schrodingers equation in spherical coordinates is
h
2
2
_
1
r
2

r
_
r
2

r
_
+
1
r
2
sin
_

sin

_
+
1
r
2
sin
2

2
_
(, ) = E(, ). (5.33)
The rigid rotor assumes that the radius is xed, so the derivatives with respect to r are zero, so the equation
simplies to
h
2
2R
2
_
1
r
2
sin
_

sin

_
+
1
r
2
sin
2

2
_
(, ) = E(, ), (5.34)
where R is the distance between the two masses. We assume a solution of the form (, ) = ()() and
substitute = 2R
2
E/h. Substituting this into Eq. (5.34) and re-arranging this equation to separate the
two angles,
_
sin
()
_

sin

_
+ sin
2

_
() =
1
()

2
(). (5.35)
Since these two functions of dierent variables are equal, they must be equal to some constant,
_
sin
()
_

sin

_
+ sin
2

_
() = m
2
(5.36)

1
()

2
()

2
= m
2
.
The azimuthal equation simply has an exponential solution
() =
1

2
e
im
, (5.37)
where m = 0, 1, 2, . . .. The polar equation is not quite so simple. The solution to this equation is the
spherical harmonic functions with the requirement that = J(J +1) with J = 0, 1, 2, . . .. This requirement
leads to the quantization of rotational energy
E =
h
2
2R
2
J(J + 1). (5.38)
98
This result can be generalized by noting that I = R
2
is the moment of inertia of the rigid rotor, so for more
complex molecules, we can use
E =
h
2
2I
J(J + 1) (5.39)
The energy dierence between two successive energy levels is
E =
h
2
2I
[(J + 1)(J + 2) J(J + 1)] = 2
h
2
2I
(J + 1) (5.40)
Example: Rotational spectroscopy
The rst line in the rotation spectrum of carbon monoxide has a frequency of 1.15 10
7
Hz,
calculate the
1
2C-
1
6O bond length in carbon monoxide.
Solution: The rst line is the transition from J = 1 to J = 0,
h = 2
h
2
2I
I =
h
2
I = 1.45 10
42
kgm
2
.
We need to nd the reduced mass of carbon monoxide
=
m
C
m
O
m
C
+m
O
= 1.14 10
26
kg.
Now we can nd the bond length
r =

r = 1.13 10
8
m.
5.1.5 Vibrational spectra
The situations we have studied so far describe the potentials seen by electrons in dierent molecules and
the energy levels associated with them. Complex molecules can have other energy levels that are associated
with vibrations within the molecule. We will model vibrations as particles on a spring. Classically, a mass
m attached to a spring with spring constant k will oscillate (if there is no damping);
x = Asin(t). (5.41)
The frequency of oscillation is =
_
k/m and the potential energy of the oscillator is
V =
1
2
kx
2
=
1
2
m
2
. (5.42)
We can use the classical potential to set up Schrodingers equation for a harmonic oscillator

h
2
2m
d
2
dx
2
(x) +
1
2
m
2
x
2
(x) = E(x). (5.43)
99
We will make the equation unitless by changing variables,
=
E
h
y =
_
m
h
x,
Which gives

(y) + (2 y
2
)(y) = 0. (5.44)
We rst solve the equation in the limit of y . In this case, the equation reduces to

a
(y) y
2

a
(y) = 0. (5.45)
The asymptotic solution has the form

a
(y) = exp(y
2
/2). (5.46)
Assume that the complete solution is the asymptotic solution modied by a polynomial
(y) = u(y)
a
(y). (5.47)
Substituting into Schrodingers equation, we get
u

(y)e
y
2
/2
2u

(y)ye
y
2
/2
+ (2 1)u(y)e
y
2
/2
= 0 (5.48)
u

(y) 2u

(y)y + (2 1)u(y) = 0.
We assume that we can write a solution for u(y) as a power series
u(y) =

n=0
C
n
y
n
. (5.49)
Substituting into our equation, we get

n=0
C
n
[n(n 1)y
n2
2ny
n
+ (2 1)y
n
] = 0, (5.50)
Which leads to the recursion relation
C
n+2
= C
n
2n + 1 2
(n + 1)(n + 2)
. (5.51)
For large n, the power series diverges faster than the Gaussian converges. We require the wavefunctions to
remain nite, so the power series must truncate at some nite n. We must have
2n + 1 2 = 0
= n +
1
2
(5.52)
E = h
_
n +
1
2
_
.
The energy levels of the QM harmonic oscillator are quantized. The energy levels are evenly spaced, unlike
the particle in a box, and the dierence is h (Fig. 5.4).
100
Figure 5.4: Simple harmonic oscillator spectrum. The spectrum of carbon monoxide shows the equally
spaced energy levels of the vibrational states.
Example: Oscillating proton
The hydrogen atom within a water molecule can be modelled as a ball on a spring with the
chemical bond acting as the spring. The bond is the equivalent of a 500 N/m spring. What is
the wavelength of a vibrational energy transition?
Solution: The mass of the hydrogen atom can be approximated as the mass of a proton. The
frequency of oscillation is
=
_
k
m
= 5.5 10
14
/s. (5.53)
The wavelength for an energy transition is
=
h
2
c
E
=
h
2
c
h
= 3.4 m. (5.54)
Once we normalize, we nd the complete solution:

n
(x) =
_
m
h
_
1/4
1

2
n
n!
H
n
_
_
m
h
_
1/2
x
_
e

m
2 h
x
2
(5.55)
H
n
(x) are the Hermite polynomials and are generated by
H
n
(x) = (1)
n
e
x
2 d
n
dx
n
e
x
2
(5.56)
The wavefunctions and associated probabilities for the SHO are shown in Fig. 5.5. Classically, the particle
cannot move further than its initial amplitude. Quantum mechanically, there is a non-zero probability that
the particle will be in this region. The probability of the particle being in the classically forbidden region is
P =
_

A

(x)(x)dx. (5.57)
The amplitude is determined by the energy
E =
kA
2
2
A =
_
2E
k
. (5.58)
101
Figure 5.5: Simple harmonic oscillators. The wavefunctions (left) and probabilities of the quantum
mechanical simple harmonic oscillator.
For the ground state

0
(y) =

(mk)
1/4
h
1/2

1/2
e
y
2
/2
. (5.59)
The probability of being in the forbidden region is
P =
1

1/2
_

1
e
y
2
dy =
1
2
(1 erf(z)), (5.60)
which works out to 7.9% on each side or 15.8% total probability of being in the forbidden region.
Example: Vibrational modes of atoms
Consider
12
C
16
O, which has a bond strength of 1900 N/m. What is the frequency of vibration of
this atom?
Solution: The bond between two atoms can be modelled as a spring, except that there is a mass
at both ends. We correct for this by calculating the reduced mass,
=
m
1
m
2
m
1
+m
2
. (5.61)
The reduced mass for
12
C
16
O is
=
m
C
m
O
m
C
+m
O
= 1.14 10
26
kg,
and the vibrational frequency is

= 4.08 10
14
Hz
102
3D harmonic oscillators
In three dimensions, the Hamiltonian is

H =
p
2
2m
+
1
2
m
2
r
2
. (5.62)
Writing this in cartesian co-ordinates, we see that it is easily separated into x, y, and z components

H =
p
2
x
2m
+
1
2
m
2
x
2
+
p
2
y
2m
+
1
2
m
2
y
2
+
p
2
z
2m
+
1
2
m
2
z
2

H =

H
x
+

H
y
+

H
z
. (5.63)
Separation of variables will clearly work to nd the complete three-dimensional wavefunction

nx,ny,nz
(x, y, z) =
nx
(x)
ny
(y)
nz
(z). (5.64)
With energy given by
E = h(n
x
+n
y
+n
z
+
3
2
). (5.65)
5.1.6 Franck-Condon principle
A complex molecule will have many dierent types of transitions: the electrons can transition between energy
levels, the bonds in the molecule can change their vibrational state, or the molecule can change its rotational
state. The energy levels and possible transitions can be quite complex, as shown in Fig. 5.6. A particle in
the ground electronic state is more likely to transition to the second excited state, rather than remain in the
ground state, if the molecule moves to the rst excited vibrational state, because the wavefunctions of these
states overlap more. The Franck-Condon principle states that during an electronic transition a change in
vibrational state is more likely to happen if the wavefunctions overlap signicantly.
Although the possible transitions within a molecule are constant, the absorption and emission spectra
of molecules are typically not the same. The emission spectra tend to be red-shifted from the absorption
spectra. Electrons tend to be knocked from the highest occupied vibrational/orbital state to the next orbital
state. Vibrational states within this orbital state relax quickly to the lowest vibrational state before returning
to the highest occupied state through uoresence.
5.2 Absorption and emission rates
If we have a system with two energy levels, there are three possible types of transitions:
1. Absorption an electron in E
1
absorbs a photon and moves to E
2
. The rate of absorption is given
by
W
12
= N
1
B, (5.66)
where N
1
is the number of electrons in E
1
B is the Einstein coecient for absorption and is the
mean intensity of the photons.
2. Spontaneous emission an electron initially in E
2
falls to E
1
emitting a photon. The transition
rate for spontaneous emission is
W
spon
= N
2
A, (5.67)
where A is the Einstein coecient for spontaneous emission.
3. Stimulated emission a photon interacts with an electron in E
2
causing it to fall to E
1
, emitting a
photon. The transition rate for stimulated emission is
W
stim
= N
2
B

, (5.68)
where B

is the Einstein coecient for stimulated emission.


103
Figure 5.6: Franck-Condon principle. The Franck-Condon principle states that during an electronic
transition, a change in vibrational state is more likely to happen if the wavefunctions overlap.
104
In thermal equilibrium, the emission and absorption processes are balanced
N
2
(A+B

) = N
1
B. (5.69)
Re-arranging we get
A
B

+ =
N
1
B
N
2
B

(5.70)
A
B

= (
N
1
B
N
2
B

1).
But in thermal equilibrium, the Boltzmann equation also applies
N
1
N
2
=
g
1
g
2
e
h0/kT
, (5.71)
where g
1
and g
2
are the degeneracies of the two levels. Substituting into our equation:
A
B

= (
g
1
B
g
2
B

e
h0/kT
1) (5.72)
or
=
A
B

1
g1B
g2B

e
h0/kT
1
. (5.73)
At equilibrium, the intensity is given by Plancks function
=
2h
3
0
c
2
1
e
h0/kT
1
. (5.74)
The two equations for must be equal. They will only be equal for all T if
g
1
B
g
2
B

= 1
B
B

= g
2
g
1
, (5.75)
and
A
B

=
2h
3
0
c
2
. (5.76)
Since A, B, and B

are properties of molecules (constants) these relationships hold even when the molecules
are not in thermal equilibrium.
We have two equations and three unknowns; if we nd an expression for one of the coecients, through
some other means, then weve found them all. The absorption rate (and therefore the Einstein coecient)
can be determined from the wavefunctions,
B =
||
2
6
0
h
2
, (5.77)
where is the transition dipole moment
=
_

f

i
dV (5.78)
and the dipole moment operator is given by
= er. (5.79)
So the Einstein coecients are
B =
||
2
6
0
h
2
B

=
||
2
6
0
h
2
g
1
g
2
A =
2||
2

2
0
3c
2

0
hg
1
g
2
. (5.80)
105
5.3 Problems
Problem 1: If the classical term for kinetic energy was 4m
3
v instead of mv
2
/2 what would Schrodingers
equation be? What if the classical term for kinetic energy was mv
4
?
Problem 2: Consider a one-dimensional problem that yields a wavefunction that is zero everywhere except
between x = 0 and x = a, where it has a constant value of A. Determine the value of A.
Suppose now that the function has a value Ax between x = 0 and x = a, what is the value of
A?
Problem 3: For the one-dimensional particle in a box: (a) What is the wavelength of the rst excited state
in terms of L? (b) What is the probability of nding the particle between x = 0 and x = L/4
in the ground state? (c) What is the expectation value of x
2
in the ground state?
Problem 4: For a one-dimensional particle in a box, if L = 1

A, (a) calculate the longest wavelength tran-
sition for an electron that is initially in the lowest energy state. (b) Calculate the wavelength
of a photon that is absorbed after an electron makes a transition from the ground state to the
fourth excited state.
Problem 5: A particle of mass m moving in one dimension is subject to the following potential
V (x) =
_
0, x > 0
, x < 0
Find the wavefunctions and energies of the stationary states.
Problem 6: Set up the problem of a particle in a box with its walls located at L/2 and L/2. Show that
the energies are equal to those of a box with walls located at 0 and L. Show, however, that
the wave functions are not the same and in this case are given by
(x) =
_
2
a
sin
_
nx
L
_
neven
(x) =
_
2
a
cos
_
nx
L
_
nodd
Does it bother you that the wave functions seem to depend on where the walls are located?
Surely the particle knows only that it has a region of length L in which to move and cannot
be aected by where you place the origin.
Problem 7: Assume there are six electrons in a box with a length of 10 nm: (a) What is the highest
occupied state? (b) What is the wavelength of the highest occupied state? (c) What is the
lowest unoccupied state? (d) What is the wavelength of the lowest unoccupied state? (e) What
is the wavelength of the longest wavelength transition?
Problem 8: Predict the wavelength of the lowest energy transition in the following polymethine ion:
(CH
3
)
2
N
+
= CH CH = CH CH = CH N(CH
3
)
2
Assume that all the C C C N bond lengths equal 1.40

A. Note that N
+
and N contribute
1 and 3 free electrons, respectively.
Problem 9: Assume a potential of the form
V =
_
_
_
, x a
0, a x 3a
, x 3a
106
(a) Write the boundary conditions for the problem. (b) Write Schrodingers equation for this
problem. (c) Write the general solution to this problem. (d) In terms of a, what is the
wavelength of the ground state?
Problem 10: Assume that a two-dimensional box is rectangular with a length of 10

A along x and 20

A
along y: (a) write Schrodingers equation for this problem; (b) write the ground state wave-
function; (c) calculate the ground state energy for a single electron; and (d) calculate the
longest wavelength transition when there is one electron in the box.
Problem 11: Assume a square well with the potential
V =
_
_
_
, x < 0
0, 0 < x < L
V
0
, x > L
For this well, what is the equation that determines the energy eigenvalues? Show that if the
parameter 2mL
2
V
0
/h
2
is smaller than a critical magnitude, there are no bound states. What
is this critical magnitude?
Problem 12: For the following two-dimensional problem assume that the potential is given by
V =
_
_
_
, x 2a, y b
0, 2a x 3a, b y 2b
, x 3a, y 2b
(a) Write the boundary conditions for the problem. (b) Write Schrodingers equation for this
problem. (c) Write the general solution to the problem.
Problem 13: Many proteins contain metal porphyrin molecules. The general structure ofthe porphyrin
molecule is
This molecule is planar and so we can approximate the electrons as being conned inside a
square. The porphyrin molecule has 18 electrons. If we approximate the sides of the molecule
by 1000 pm, then what is the predicted lowest energy absorption of the porphyrin molecule?
Problem 14: What are the degeneracies of the rst 10 energy levels for a particle in a three-dimensional box
with L
x
= L
y
= 1.5L
z
?
Problem 15: For a simple harmonic oscillator: (a) what is the zero-point energy for an electron? (b) What
is the energy of the second excited state for an electron? (c) Write the wavelength of light
absorbed for a transition of an electron between adjacent levels?
Problem 16: For a simple harmonic oscillator, write (but do not solve) the probability of the ground state
being in the classically forbidden region.
107
Problem 17: Water (H
2
O) absorbs infrared radiation at 3 m due to the OH stretch. What wavelength
would you predict is absorbed by D
2
O?
Problem 18: For a simple
16
O
2
molecule, calculate the (a) the reduced mass and (b) the frequency change
when the isotope
15
O is present.
Problem 19: What are the reduced masses for molecules of Li
2
, N
2
, and AlH?
Problem 20: Derive the expression for the moment of inertia of a classical rigid two-particle rotor.
Problem 21: How many revolutions/sdoes a C
12
O
16
molecule make when J = 1, and J = 10? Assume
r = 1.128

A.
Problem 22: The J = 1 J = 0 transition of HCL
35
occurs at 6.264 10
5
Hz. Calculate the moment of
inertia and internuclear separation of HCL
35
.
Problem 23: The OH radical has a moment of inertia of 1.480 10
40
gcm
2
. (a) Calculate its internuclear
distance. For J = 5, calculate (b) its angular momentum; (c) its angular velocity. (d) Determine
the energy absorbed in the J = 5 J = 6 transition.
Problem 24: Write Schrodingers equation for the potential V (x) = Ax
2
+ Be
x
. If B A what will the
wavefunctions be approximately?
Problem 25: For the following questions, assume that a particle of mass m is in a valley between two hills.
You model the potential as
V (x) =
_
a(x
2
+bx
3
), |x| L/2
cL
2
, |x| L/2
where cL
2
is larger than the total energy of the particle. (a) Write Schrodingers equation for
these potentials. (b) How are a and b related to c? (c) Estimate the probability of nding the
particle in the |x| L/2 and |x| L/2 regions. (d) If b is very small and c = 0, what can be
said about the wafunctions in the region |x| L/2?
108
Chapter 6
Nuclear Magnetic Resonance
FIXME
6.1 Magnetic dipoles
Recall that all magnets are dipoles whose strength is described by the magnetic dipole moment, . When a
magnet is placed in a magnetic eld,

B, it experiences a torque
=

B. (6.1)
There is a potential energy associated with the torque
U =

B. (6.2)
Example: Charged orbiting particle
Suppose a particle of charge q and mass m travels in a circular orbit with speed v. The magnetic
moment is
= i

A, (6.3)
where i is the current and

A is the area vector. The angular momentum is
L = mvr. (6.4)
and the current is
i =
qv
2r
(6.5)
Putting these equations together, we get
= iA
= ir
2
(6.6)
=
qvr
2
=
q
2m

L
It is generally true that =

L where is the gyromagnetic ratio and depends on the geometry


of the system.
109
We can derive an equation of motion for a magnetic dipole. The torque is the rate of change of angular
momentum
=
d

L
dt
(6.7)
where the torque is given by Eq. (6.1)

L

B =
d

L
dt
(6.8)
(

B) =
d
dt
6.2 Magnetization
An electron has a magnetic moment because it orbits the mucleus. The electron (and other elementary
particles) will also have intrinsic angular momentum (spin). The spin can be thought of as the angular
momentum caused by the particle spinning on its own axis. The spin also creates a magnetic moment

s
=
s

S, (6.9)
where
s
is the gyromagnetic ratio of the spin.
The nucleus does not orbit, but it will have an intrinsic angular momentum created by coupling of the
spin states of the constituent particles. The nuclear angular momentum is quantized with vector

I and
quantum number I. The possible values of the z component are mh where m = I, (I + 1) . . . I. The
energy levels are given by

B =

I

B = mhB
Magnetic resonance imaging depends on the magnetization of the sample. The magnetization,

M of a
sample is the average magnetic moment per unit volume. Suppose we have a collection of spins of a single
nuclear species which do not interact with each other and are in thermal equilibrium with the surroundings
at temperature T. In the absence of an external magnetic eld, there will be no net magnetization since the
spins are randomly oriented. We apply an external magnetic eld B which causes a net magnetization

M = N < > (6.10)


6.2.1 Average magnetic moment
We will calculate the average value of the z component of the magnetic moment. We multiply each possible
state m by the probability that the particle is in that state. Since we are in thermal equilibrium, the
probability of being in a certain energy state is given by the Boltzmann factor
e
U/k
B
T
= e
m hB/k
B
T
(6.11)
The average magnetic moment is
<
z
>=
h
I

m=I
me
m hB/k
B
T
I

m=I
e
m hB/k
B
T
(6.12)
We can expand for small magnetic potential energies mhB/k
B
T 1 with
e
x
1 +x (6.13)
110
The numerator becomes
I

m=I
m+
hB
k
B
T
I

m=I
m
2
(6.14)
The rst term sums to zero and the second is
hB
k
B
T
I

m=I
m
2
=
hB
3k
B
T
I(I + 1)(2I + 1) (6.15)
The denominator becomes
I

m=I
1 +
hB
k
B
T
I

m=I
m (6.16)
In this case, the second term vanishes and the rst term is
I

m=I
1 = 2I + 1 (6.17)
Thus, the average magnetic moment is
<
z
>=

2
h
2
I(I + 1)
3k
B
T
B (6.18)
The z component of the magnetization is
M
z
= N <
z
>=
N
2
h
2
I(I + 1)
3k
B
T
B (6.19)
The magnetization is proportional to the applied magnetic eld. If the eld is strong, this approximation
breaks down since eventually all the spins are aligned and the magnetization reaches a saturation state.
The expectation value of a quantum mechanical operator behaves classically
d < >
dt
= (< >

B) (6.20)
We can re-write this in terms of magnetization
d

M
dt
= (

M

B) (6.21)
If

M is parallel to

B, there is no torque and the magnetization remains constant.
6.3 Precession
Suppose we apply a constant B eld in the z-direction
dM
x
dt
= M
y
B
z
,
dM
y
dt
= M
x
B
z
, (6.22)
dM
z
dt
= 0
111
Precession
Figure 6.1: Larmor precession. The net magnetization of an object precesses in the presence of a constant
uni-directional magnetic eld.
If the initial conditions are M
x
(0) = M

, M
y
(0) = 0, and M
z
(0) = M

, then the solution is


M
x
= M

cos(B
z
t),
M
y
= M

sin(B
z
t), (6.23)
M
z
= M

where
0
= B
z
is the Larmor precession frequency. This solution is pictured in Fig. 6.1. The magnetization
in the z direction does not change; in the x and y directions, the magnetization moves in a circle.
6.3.1 Dephasing
The magnetization is an average value of the magnetic dipole moment. While the magnetization may have
a specic precession frequency, the individual magnetic dipoles may have a slightly dierent frequency. This
leads to dephasing the precession of individual dipoles becomes decoherent (Fig. 6.2). This shows up in
the x and y components of magnetization as damped oscillations.
112
Figure 6.2: Dephasing. Individual dipole moments do not necessarily precess at the same frequency as the
net magnetization leading to damped oscillations.
6.4 Relaxation
Suppose that the system is perturbed out of thermal equilibrium. The populations within each energy level
will re-distribute to return to equilibrium. We assume that this process is exponential
dM
x
dt
=
M
x
T
2
dM
y
dt
=
M
y
T
2
(6.24)
dM
z
dt
=
1
T
1
(M
0
M
z
)
where M
0
is the equilibrium value of M
z
, T
1
is the longitudinal or spin-lattice relaxation time, and T
2
is the
transverse or spin-spin relaxation time. Typically, T
1
is longer than T
2
since it requires energy transfer with
the reservoirs while T
2
can change without input or output of energy (dephasing).
The relaxation needs to be included in the equation of motion
dM
x
dt
=
M
x
T
2
+(

M

B)
x
dM
y
dt
=
M
y
T
2
+(

M

B)
y
(6.25)
dM
z
dt
=
1
T
1
(M
0
M
z
) +(

M

B)
z
These are the Bloch equations. Note that the exponential relaxation is an assumption and is not strictly
true for all materials.
For a constant B eld in the z-direction, we have
dM
x
dt
=
M
x
T
2
+M
y
B
z
dM
y
dt
=
M
y
T
2
M
x
B
z
(6.26)
dM
z
dt
=
1
T
1
(M
0
M
z
)
113
Assume initial conditions of M
x
(0) = M
0
, M
y
(0) = 0, M
z
(0) = 0. For M
z
, we have a simple exponential
which can be integrated
dM
z
dt
=
1
T
1
(M
0
M
z
)
_
Mz
0
dM
z
(M
0
M
z
)
=
_
t
0
dt
T
1
(6.27)
ln(M
0
M
z
)|
Mz
0
=
t
T
1

t
0
M
z
= M
0
(1 e
t/T1
)
M
z
undergoes exponential decay to its thermal equilibrium value.
The solutions for M
y
and M
x
require a little more computational eort. Solve M
y
dierential equation
for M
x
M
x
=
1
B
z
_
dM
y
dt
+
M
y
T
2
_
. (6.28)
Substitute into dierential equation for M
x

d
2
M
y
dt
2

1
T
2
dM
y
dt
=
1
T
2
dM
y
dt
+
M
y
T
2
2
+
2
B
2
z
M
y
(6.29)
We get an equation for M
y
independent of M
x
d
2
M
y
dt
2
+
2
T
2
dM
y
dt
+
_
1
T
2
2
+
2
B
2
z
_
M
y
= 0 (6.30)
This is a dierential equation with constant coecients, we need to nd the roots of the characteristic
equation

=
1
2
_

2
T
2

4
T
2
2
4
_
1
T
2
2

2
B
2
z
_
_
(6.31)

=
1
T
2
B
z
i
So the solution for M
y
is
M
y
= e
t/T2
(A
y
cos(B
z
it) +C
y
sin(B
z
it)) (6.32)
Since the original equations are symmetric in M
x
and M
y
, M
x
has the same general solution
M
x
= e
t/T2
(A
x
cos(B
z
it) +C
x
sin(B
z
it)) (6.33)
Applying boundary conditions, we nd
M
x
= M
0
e
t/T2
cos(B
z
t) (6.34)
M
y
= M
0
e
t/T2
sin(B
z
t)
Initially the magnitude in the xy plane is M
0
, but decays exponentially, as the magnetic dipole moment
precesses. The magnetization moves to the z direction as shown in Fig. 6.3.
114
Figure 6.3: Relaxation of magnetization in a constant magnetic eld. Magnetization that has been
moved perpendicular to the magnetic eld relaxes exponentially to be parallel to the magnetic eld.
6.4.1 Rotating coordinate system
Since the magnetization rotates in the xy plane at a constant frequency, it is helpful to describe the motion
of

M in a rotating frame. We can convert from one coordinate system to another through
M
x
= M
x
cos M
y
sin (6.35)
M
y
= M
x
sin +M
y
cos
In this case is time-dependent: =
0
t. Note that M
z
is unaected by the change in the coordinate
system.
We can nd the time derivatives in the new coordinate system, recalling that the transformation is
time-dependent
dM
x
dt
=
dM
x

dt
cos(
0
t)
dM
y

dt
sin(
0
t)
+
0
M
x
sin(
0
t) +
0
M
y
cos(
0
t) (6.36)
dM
y
dt
=
dM
x

dt
sin(
0
t) +
dM
y

dt
cos(
0
t)

0
M
x
cos(
0
t) +
0
M
y
sin(
0
t)
We can use these to re-write the equations of motion in the rotating frame. In the absence of relaxation, the
equation for M
x
becomes
dM
x
dt
= M
y
B
z
dM
x

dt
cos(
0
t)
dM
y

dt
sin(
0
t) + (6.37)

0
M
x
sin(
0
t) +
0
M
y
cos(
0
t) =
0
(M
x
sin +M
y
cos )
dM
x

dt
cos(
0
t)
dM
y

dt
sin(
0
t) = 0
We nd a similar equation for M
y
dM
x

dt
sin(
0
t) +
dM
y

dt
cos(
0
t) = 0 (6.38)
Solving the rst equation for M
x

dM
x

dt
=
dM
y

dt
sin(
0
t)
cos(
0
t)
(6.39)
115
and substituting into the second equation
dM
y

dt
sin(
0
t)
cos(
0
t)
sin(
0
t) +
dM
y

dt
cos(
0
t) = 0 (6.40)
dM
y

dt
= 0
Similarly,
dM
x

dt
= 0 (6.41)
All components of the magnetization in the frame rotating at the Larmor frequency are constant.
General transformation
For any vector

M, the relationship between time derivatives in the laboratory frame and a frame
rotating with angular velocity

is
_
d

M
dt
_
lab
=
_
d

M
dt
_
rot
+



M (6.42)
For the magnetization, we have
_
d

M
dt
_
rot
= (

M

B)



M (6.43)
_
d

M
dt
_
rot
=

M
_

B +

_
At the Larmor frequency,

B, this rhs vanishes.


6.5 Oscillating magnetic eld
In addition to the constant B eld in the z direction, we add an oscillating eld in the x direction in the lab
frame: B = B
1
cos(t). In the lab frame, we have
dM
x
dt
= M
y
B
z
dM
y
dt
= M
z
B
x
M
x
B
z
(6.44)
dM
z
dt
= M
y
B
x
We will transform these equations to a frame rotating at Larmor frequency. In the x direction, the
oscillating B-eld has no eect; the equation is the same as for constant B
z
.
dM
x

dt
cos(
0
t)
dM
y

dt
sin(
0
t) = 0 (6.45)
In the y direction, we get
dM
x

dt
sin(
0
t) +
dM
y

dt
cos(
0
t) = B
1
M
z
cos(t) (6.46)
In the z direction, we get
dM
z

dt
= B
1
M
x
cos(t) sin(
0
t) B
1
M
y
cos(t) cos(
0
t) (6.47)
116
We can eliminate M
x
or M
y
to get equations for M
x
and M
y

dM
x

dt
= B
1
M
z
cos(t) sin(
0
t) (6.48)
dM
y

dt
= B
1
M
z
cos(t) cos(
0
t)
These are the equations of motion in the rotating system.
If =
0
then the motion is complicated. If the applied eld oscillates at the Larmor frequency, we can
look at the time-averaged equations
dM
x

dt
= 0
dM
y

dt
=

2
B
1
M
z
(6.49)
dM
z

dt
=

2
B
1
M
y

In the rotating lab frame, M


x
does not change. Dene a new angular frequency

1
=
B
1
2
(6.50)
This is the frequency of nutation rotation caused by the oscillating magnetic eld. The dierential equations
become
dM
y

dt
=
1
M
z
(6.51)
dM
z

dt
=
1
M
y

We can decouple the equations by dierentiating and substituting


d
2
M
y

dt
2
=
2
1
M
y
(6.52)
d
2
M
z

dt
2
=
2
1
M
z

This has a solution of


M
y
= a cos(
1
t) +b sin(
1
t) (6.53)
M
z
= a sin(
1
t) +b cos(
1
t)
If

M is initially along the z-axis then
M
x
= 0
M
y
= M
0
sin(
1
t) (6.54)
M
z
= M
0
cos(
1
t)
An B-eld oscillating in the x direction at Larmor frequency will cause the precession axis to switch from
z

to x

. The frequency of precession is not Larmor frequency, but at the nutation frequency,
1
.
6.5.1 Pulses
Since the oscillating eld causes the magnetization to rotate in the z

-plane, a pulse of a xed length of


time will cause a xed shift of the magnetization. A /2 pulse causes the magnetization to go from the
117
z

-axis to the y

-axis (Fig. 18.7). A pulse causes the magnetization to go from the z

-axis to the -z

-axis
(Fig. 18.8). This is easy to visualize in the rotating frame. In the lab frame, the magnetization spirals to its
end point. A /2 pulse can shift the magnetization by 90

. We saw a similar eect with relaxation. They


are not the same process; their paths in phase space are dierent even though the outcome is the same (Fig.
18.7c). The oscillating eld causes a rotation about the x

-axis, but does not change the magnitude of the


magnetization in the x

-direction. If the magnetization in the x

direction is not initially zero, this motion


is dicult to describe in the lab frame
We have now determined how to apply a perturbation of arbitrary size to the magnetization of the
sample. If we apply a pulse, we change the magnetization of the material. When the pulse is nished, the
magnetization returns to its original thermal equilibrium through relaxation. Relaxation in the xy-plane
does not require absorption or emission of energy, but changes in the z component do since changes in M
z
are caused by changes to the populations within the energy levels of the nuclei. A nucleus can only change
its energy level through emission or absorption of a photon. Recall that the spacing between energy levels is
h
0
, so the photon will have the Larmor frequency. Processes that can change energy of nuclei are absorption,
spontaneous emission or stimulated emission. These processes are caused by small local magnetic elds due
to neighbouring nuclei.
6.5.2 Fluctuations in the magnetic eld
Example: Magnetic eld in water
The eld due to a magnetic dipole is
B
r
=

0
4
2
r
3
cos
B

=

0
4
2
r
3
sin (6.55)
B

= 0
We can calculate the magnetic eld between the two hydrogen nuclei in water. The two protons
are in line with each other, so cos = 0 and there is only a component to the magnetic eld.
The magnetic dipole moment of a proton comes from its intrinsic spin: = S
z
. The magnetic
eld is 3.97 10
4
T. If the water molecule remains xed in time (like ice), this local magnetic
eld will remain constant. If the water molecule changes its orientation in the magnetic eld, the
local magnetic eld changes (Fig. 18.11). So there can be large uctuations in the magnetic eld
simply due to random motion of the molecules.
There will be a range of possible magnetic eld values within the sample caused by either internal or
external uctuations
B
tot
= B
internal
+ B
external
(6.56)
Since the magnetic eld is related to the frequency

tot
=
internal
+
external
(6.57)
We assume that the transverse relaxation time is determined by the spread in : T
2
= K
1
T

2
=
1
T
2
+
B
external
K
(6.58)
where T

2
is the relaxation time measured experimentally and T
2
is the non-recoverable relaxation time
caused by internal uctuations of the magnetic eld.
118
6.5.3 Autocorrelation
We can also describe the local uctuations in the magnetic eld through the autocorrelation function. Cor-
relation describes the amount of shared information in two signals. Autocorrelation describes the amount
of shared information between two parts of a signal a time apart. Autocorrelation measures how much
knowledge of past behaviour tells us about present behaviour. When particles are moving randomly, the
autocorrelation drops very quickly as increases. We can approximate this as exponential decay

11
() =
_

y
1
(t)y
1
(t +)dt e
||/
C
, (6.59)
where |
C
| is the autocorrelation time and is assumed to be the same in all directions.
The Fourier transform of autocorrelation is the power spectral density. Power spectral density gives the
power of a signal at dierent frequencies. For our assumed autocorrelation function:
P =
_

0
cos(t)e

|t|

C
dt
P =

C
1 +
2

2
C
The power spectral density is proportional to the power at frequency .
The transition rate 1/T
1
is proportional to the power at frequency
0
. We can combine the two to get
an expression for the longitudinal relaxation time
1
T
1
=
C
C
1 +
2
0

2
C
(6.60)
Re-writing,
T
1
=
1
C
C
+

2
0

C
C
(6.61)
At short autocorrelation times, the relaxation time does not depend on the Larmor frequency. At long
autocorrelation times, the relaxation time is proportional to
2
0
. The relaxation time is a minimum when

C
= 1/
0
.
We can do a similar analysis for T
2
to nd
T
2
=
2
C
C
_
1 +
2
0

2
C
2 +
2
0

2
C
_
(6.62)
At long autocorrelation times, the transverse relaxation time goes to 0. There is no minimum value, so the
transverse relaxation time simply increases as autocorrelation time decreases
6.5.4 Measuring the magnetic resonance signal
A sample of nuclei in a constant B eld will align with that eld and precess about the axis. If we apply an
oscillating B eld of a xed duration, we can change the precession axis to any arbitrary value (determined
by the duration of the pulse). After the axis changes, the nuclei will relax back to their original state. The
relaxation occurs with two time constants, T
1
along the constant B eld, and T
2
perpendicular to the original
B eld.
Suppose we have a sample at the origin in a constant B eld along the z axis. The magnetization will
precess in the xy plane. To change the angle of precession, we need to apply a second oscillating magnetic
eld along another axis. Place a coil along the x axis and apply alternating current at the Larmor frequency
to change the axis. The current is turned o in the coil after the axis has shifted and that same coil can now
be used to detect changes in magnetization as the nuclei relax. The detected signal is a damped oscillation
known as free induction decay.
119
The induced voltage is
v =

t
=

t
_ _
B dS (6.63)
The ux through the plane of the loop, radius a is the negative of the ux going through the going through
a hemispherical cap since
_ _
closed
B dS (6.64)
So we have
=
_
B
r
2a
2
sin d (6.65)
On the hemispherical surface, only the radial component of the magnetic eld contributes to the ux. For
a magnetic dipole, the eld is
B
r
=

0
4
2
a
3
cos (6.66)
So the ux is
=

0
4
4
x
a
_
/2
0
cos sin d (6.67)
Note that the net ux due to
y
will be zero. Integrating we get
=

0
4
2
x
a
(6.68)
We know that
x
= M
x
V with M
x
given by solutions to the equation of motion during relaxation
=

0
4
2M
0
V
a
e
t/T2
cos(
0
T) (6.69)
The induced voltage is /t
v =

0
4
4M
0
V
a
e
t/T2
_

1
T
2
cos(
0
t)
0
sin(
0
t)
_
(6.70)
Since T
2
1/
0
, this can be simplied to
v =

0
4
4M
0
V
a
e
t/T2

0
sin(
0
t) (6.71)
If we started with M
z
at thermal equilibrium that was then nutated to the xy plane, we can use the expression
for M
z
at thermal equilibrium
M
z
= N <
z
>=
N
2
h
2
I(I + 1)
3k
B
T
B (6.72)
For a spin-
1
2
particle, this becomes
M
z
=
N
2
h
2
4k
B
T
B (6.73)
and the voltage signal is
v =

0
4
NV
3
h
2
B
2
)
2k
B
Ta
e
t/T2

0
sin(
0
t) (6.74)
The voltage signal is a damped sine wave with decay determined by T
2
. The amplitude of the signal is
inversely proportional to the radius of the coil smaller coils are better. The amplitude of the signal is
proportional to B
2
0
larger constant B elds produce larger signals. The amplitude of the signal is inversely
proportional to the temperature.
120
6.5.5 Pulse sequences
There are many dierent pulse sequences that can be used for measuring relaxation times. Five classic
sequences are used in magnetic resonance imaging:
Free Induction Decay Sequence
Inversion Recovery Sequence
Spin Echo Recovery
Carr-Purcell Sequence
Carr-Purcell-Meiboom-Gill
Free Induction Decay
Start in equilibrium with constant B-eld in the z direction. A /2 pulse nutates M into the xy plane. The
signal is e
t/T

2
sin(
0
t) where T

2
includes both external and internal inhomogeneities. This sequence will
give us T

2
.
Inversion-Recovery
Start in equilibrium with constant B-eld in the z direction. A pulse switches M to point along z. As the
system is returning to equilibrium according to M
z
= M
0
[1 2e
t/T1
], a /2 is applied at an interrogation
time T
I
. This nutates the instantaneous value of M
z
= M
0
[1 2e
T
I
/T1
] to the xy plane. So we can
determine T
1
as well as T

2
. When T
I
/T
1
= ln 2 there will be no signal; if T
I
is less than this the signal will
be negative.
Spin-Echo
Start in equilibrium with constant B-eld in the z direction. Apply /2 pulse to nutate M into the xy plane.
Due to local uctuations, individual magnetic moments will precess at slightly dierent frequencies. After a
time T
E
/2, apply a pi pulse that rotates all the magnetic dipoles about the x-axis. Now the slower dipoles
are ahead of the faster ones, so the faster ones will catch up re-focusing the magnetization. This causes an
echo signal whose decay depends on T
2
rather than T

2
.
Carr-Purcell
Start in equilibrium with constant B-eld in the z direction. Apply /2 pulse to nutate M into the xy plane.
Apply a sequence of pulses at T
E
/2, 3T
E
/2, 5T
E
/2, to get a sequence of echos whose amplitude decays with
time constant T
2
. Corrects for possible diusion of nuclei between regions of dierent magnetic elds.
6.6 Imaging
If we put a body into the magnetic eld, all nuclei in the sample will undergo magnetic resonance at the same
time in the same way. We cannot dierentiate between dierent locations in the body. To select dierent
slices, we use a magnetic eld with a constant gradient. Since only pulses with the correct frequency,

0
= B
z
will cause nutation, this will select dierent slices of the body for imaging.
121
6.6.1 Slice selection
We apply a magnetic eld B
z
(z) = B
0
+G
z
z where
G
z
=
B
z
z
(6.75)
Since the eld varies continuously, if we want a slice of some nite thickness we need to include frequencies
between
0
and
0
+ where
=
G
z
z
2
(6.76)
So B
x
must be
B
x
(t) =
A
2
_
0+
0
cos(t)d
=
A

sin(t
t
cos(
0
t)
The gradient elds are created by specially wound solenoids. Recall that the magnetic eld inside a
solenoid is
B =
0
N
l
I (6.77)
where N/l is the number of turns per unit length. If we vary the number of turns per unit length linearly
along the length of the solenoid, we will get a linear gradient in the magnetic eld.
Another way to create a gradient eld is using two coils with current owing in opposite directions. For
two loops of radius R a distance d apart, the magnetic eld along the axis is
B
z
=

0
R
2
I
2
_
1
((z d/2)
2
+R
2
)
3/2

1
((z +d/2)
2
+R
2
)
3/2
_
(6.78)
A taylor expansion gives
B
z
=

0
I
2R
__
1 +
3
2
_
z d/2
R
_
2
_

_
1 +
3
2
_
z +d/2
R
_
2
__
B
z
=

0
I
2R
2
zd
so the eld is approximately linear.
Gradient magnets can also be created from a quadrupole magnet. If four magnets are placed in a specic
orientation, the dipole term will cancel this is a quadrupole magnet. The magnetic eld between the
magnets has a constant gradient i.e. varies linearly (Fig. 6.4).
The typical conguration of magnets within an MRI is shown in Fig. 6.5. Conguration of magnetic
elds in an MRI. The constant magnetic eld is created by a solenoid running along the patients body. The
slices in the x and y direction are created by quadrupole magnets.
6.6.2 Image reconstruction
B
x
still has the same form, but the amplitude is now
B
1
(t) =
A

sin(t
t
(6.79)
122
Figure 6.4: Magnetic eld of a quadrupole magnet. The magnet eld at the centre of a quadrupole is
approximately linear and can be used as a gradient magnetic for slice selection.
The angle through which the spins are nutated is
=
_

1
(t)dt
=

2
_

B
1
(t)dt
=
A
2
_

sin(t)
t
dt
=
A
2
(6.80)
so once we set the angle (ex. /2), we determine A.
This procedure selects a slice in the z-direction, but we still need to determine xy information. The same
gradient technique can be used to change the frequencies along the x-axis frequency encoding. We cannot
use the gradient again along the y-axis since this will not uniquely dene each (x, y) pair. Along the y-axis,
we use phase encoding (Fig. 6.6)
When the gradient eld is turned on, the magnetic dipoles spin at a frequency that is dependent on their
position. When the gradient is turned o, the magnetic dipoles have spun a dierent amount. They are left
with a permanent phase shift that is dependent on position
=
_
(t)dt = G
y
T
p
y (6.81)
The voltage readout actually gives Fourier transforms of magnetization. With no y gradient, we get
points along the k
y
= 0 axis. A phase shift along the y-axis will shift the readout line along the k
y
axis.
Once we have created the image in Fourier space, we can take the Fourier transform and recover the original
image (Fig. 6.7).
123
Figure 6.5: Conguration of magnetic elds in an MRI. The constant magnetic eld is created by
a solenoid running along the patients body. The slices in the x and y direction are created by quadrupole
magnets.
124
Figure 6.6: Phase encoding. There is a position-dependent phase shift in the signal when gradient magnets
are used for slice selection.
125
Figure 6.7: Image reconstruction. FIXME
126
6.7 Problems
Problem 1: Show that for a particle of mass m located at position r with respect to the origin, the torque
about the origin is the rate of change of the angular momentum about the origin.
Problem 2: Obtain an expression for the magnetization analogous to Eq. (6.19) in the case I = 1/2 when
one cannot make the assumption hB/k
B
T 1.
Problem 3: Calculate the value of M
2
x
+M
2
y
+M
2
z
for relaxation Eqs. (6.27) & (6.34) when T
1
= T
2
.
Problem 4: What is the solution to the Bloch equations if the initial conditions are M
x
(0) = M
y
(0) = 0
and M
z
= M
0
?
Problem 5: In one resonance experiment the sample is placed in a magnetic eld B
0
, along the z axis. Its
magnetization is M
0
, in the same direction. A transverse rotating eld with amplitude B
1
is
turned on and remains on for a time t = /(2B
1
). The eld rotates with angular velocity
B
0
in the same direction as the magnetization. Ignore relaxation eects. (a) Show that M
z
then obeys (d
2
M
z
/dt
2
) =
2
B
2
1
M
z
. (b) Show that M
z
= 0 when the rotating eld is turned
o. (c) What is the value of M
2
x
+M
2
y
when the eld is turned o?
Problem 6: Derive the following transformation matrices for counter-clockwise rotations. (a) Angle about
the x axis:
_
_
1 0 0
0 cos sin
0 sin cos
_
_
(b) Angle about the y axis:
_
_
cos 0 sin
0 1 0
sin 0 cos
_
_
(c) Angle about the z axis:
_
_
cos sin 0
sin cos 0
0 0 1
_
_
Problem 7: Calculate M
2
= M
2
x
+M
2
y
+M
2
z
for Eqs. 6.54.
Problem 8: Find the magnetic eld at one proton due to the other proton in a water molecule when both
proton spins are parallel to each other and perpendicular to the line between the protons. The
two protons form and angle of 104.5

and are each 96.5 10


12
m from the oxygen.
Problem 9: The magnetic eld at a distance of 0.15 nm from a proton is 4 10
4
T. What change in
Larmor frequency does this B cause? How long will it take for a phase dierence of radians
to occur between a precessing spin feeling this extra eld and one that is not?
Problem 10: Assume two dipole moments, one in a eld B
0
and the other in in a eld B
0
+ B
0
, start
aligned. How much later are they out of phase in their precessional motion?
Problem 11: Consider a collection of spins that are aligned along the x axis at t = 0. They precess in the xy
plane with dierent angular frequencies spread uniformly between /2 and +/2. If
the total magnetic moment per unit volume is M
0
at t = 0, show that at time T = 4/ it is
M
0
sin(2)/2 = 0.455M
0
.
Problem 12: What is the contribution to the transverse relaxation time for a magnetic eld of 1.5 T with a
uniformity of 1 part per million? The nonrecoverable relaxation time of brain is about 2.5 ms.
What determines the measured transverse relaxation in the brain?
127
Problem 13: Suppose the two dipoles of the water molecule shown below point in the z direction while the
line between them makes an angle with the x axis. Determine the angle for which the
magnetic eld of one dipole is perpendicular to the dipole moment of the other. For this angle,
the interaction energy is zero. This is called the magic angle and is used when studying
anisotropic tissue such as cartilage.
Problem 14: In solving this problem, you will develop a simple model for estimating the radio-frequency
energy absorption in a patient undergoing an MRI procedure. (a) Consider a uniform conductor
with electrical conductivity, . If it is subject to a changing magnetic eld B
1
(t) = B
1
cos(
0
t),
show that the electric eld has a magnitude of E
0
= R
0
B
1
/2 where R describes a circular
path at right angles to the eld. You will need the following relationship between electric and
magnetic elds
_
E ds =
d
dt
_ _
B dS
(b) Use Ohms law in the form j = E to show that the time average power dissipated per
unit volume of material is p = E
2
0
/2 = R
2

2
0
B
2
1
/8 and that if the mass density of the
material is , the specic absorption rate (SAR) or dose rate is SAR= R
2

2
0
B
2
1
/8. (c) If
the radio-frequency signal is not continuous but is pulsed, show that this must be modied by
the duty cycle factor t/T
R
, where t is the pulse duration and T
R
is the repetition period.
(d) Combine these results with the fact that rotation through an angle (usually or /2) in
time t requires B
1
= 2/t and that
0
= B
0
, to obtain SAR= (1/T
R
t)(/2)R
2
B
2
0

2
.
(e) Use typical values for the human body R = 0.17 m, = 0.3 S/m to evaluate this
expression for a /2 pulse. (f) For B
0
= 0.5 T and SAR< 0.4 W/kg determine the minimum
value of t for T
R
= 1 s. Also nd B
1
. (g) For 180

pulses, what is the dose in grey ?


Problem 15: Calculate the initial amplitude of a signal induced in a one-turn coil of radius 0.5 m for protons
in a 1 mm cube of water at 310 K in a magnetic eld of 1 T.
Problem 16: Suppose a proton NMR resonance peak occurs at a frequency of 453 MHz when measured at
a temperature of 300 K.Find the ratio of the average number of protons with spin-down to
spin-up in the sample.
Problem 17: Calculate the resonance frequency of a proton in a magnetic eld of (a) 7 T and (b) 20 T. (c)
Calculate the resonance frequency of a
14
N nucleus in the same magnetic elds.
Problem 18: Obtain an analytic expression for the maximum value of the rst and second echo amplitudes
in a Carr-Purcell pulse sequence in terms of T
2
and T
E
.
Problem 19: The general equation for M
z
in the Carr-Purcell pulse sequence is M
z
= M
0
+Ae
t/T1
. After
several pulses, the value of M
z
is ipping from b to b. Find the value of b.
128
Problem 20: Consider a spin-echo pulse sequence. Find (a) M
z
just before the pulse at T
E
/2, (b) M
z
just
after the pulse at T
E
/2, (c) M
z
just before the /2 pulse at T
R
, and (d) the rst and second
echo amplitudes as a function of T
E
, T
R
, T
1
, and T
2
. (The second amplitude is the same as all
subsequent amplitudes).
Problem 21: Use the rotation matrices given in Problem FIXME. Start with M = (0, 0, M
0
). Rotate M
about x

by /2, then about z

by , then about x

by , and nally about z

by . What are
the nal components of M? Identify what pulse sequence or physical process corresponds to
each rotation. Why would be nonzero in the rotating reference frame? What would be the
signicance if the nal M is independent of ?
Problem 22: This problem uses the matrices from Problem FIXME to examine the dierence between
the Carr-Purcell and the Carr-Purcell-Meiboom-Gill pulse sequences. (a) Start with M =
(0, 0, M
0
). Rotate about x

by /2, about z

by , about x

by , about z

by again, about
x

by , and about z

by . What is the nal result? This process corresponds to the rst two
echoes produced by a Carr-Purcell pulse sequence. (b) Repeat part (a), but change the two
rotations about x

to two + rotations about x

. Assume and use the approximations


cos( + ) = cos 1 and sin( + ) = sin to simplify your result. Keep only
terms in order . This process corresponds to the rst two echoes produced by a Carr-Purcell
pulse sequence in which the pulses have slightly wrong amplitudes. (c) Repeat the analysis
of part (b) but change the rotations about x

to be rotations about y

. What are the dierences


between CP and CPMG pulse sequences?
Problem 23: Show that an alternative expression for the eld amplitude required for a /2 pulse is B
1
=
B
0
/
0
t = B
0
/2t.
Problem 24: A certain MRI machine has a static magnetic eld of 1.0 T. Spins are excited by applying a
eld gradient of 3 mT/m. If the slice is to be 5 mm thick, what is the Larmor frequency and
the spread in frequencies that is required?
Problem 25: Consider a pair of gradient coils of radius a perpendicular to the z axis and located at z =

3a/2. The current ows in the opposite direction in each single-turn coil. (a) Obtain an
expression for B
z
along the z axis. (b) For a gradient of 5 mT/m at the origin and a = 10 cm,
nd the current required in a single-turn coil.
Problem 26: The slice selection gradient G
z
must be applied for a time which is at least as long as the
duration of the B
1
pulse. Suppose that = 6 2/(G
z
z). How much has the phase at the
top of the slice (z = z) changed with respect to the middle of the slice (z = 0)?
Problem 27: This problem shows how to extend the Bloch equations to include the eect of diusion of the
molecules containing the nuclear spins in an inhomogeneous external magnetic eld. Since M is
the magnetization per unit volume, it depends on the total number of particles per unit volume
with average spin components <
x
>, <
y
>, and <
z
>. In the rotating coordinate system
there is no precession. In the absence of relaxation eects does not change. In that case
changes in M depend on changes in the concentration of particles with particular components
of , so the rate of change of each component of is given by a diusion equation. For example,
for M
x
,
M
x

t
= D
2
M
x

If the processes are linear this diusion term can be added to the other terms in the Bloch
equations. Suppose that there is a uniform gradient in B
z
, G
z
, and that the coordinate system
rotates with the Larmor frequency for z = 0. When z is not zero, the rotation term does not
129
quite cancel the (MB)
z
term. (a) Show that the x and y Bloch equations become the same
one that would be seen if Eq. 18.17 were written
M
x

t
= +G
z
zM
y

M
x

T
2
+D
2
M
x

M
y

t
= G
z
zM
x

M
y

T
2
+D
2
M
y

(b) Show that in the absence of diusion


M
x
= M(0)e
t/T2
cos(G
z
zt)
M
y
= M(0)e
t/T2
sin(G
z
zt)
(c) Suppose that M is uniform in all directions. At t = 0 all spins are aligned. Spins that
have been rotating faster in the plane at z + z will diuse into plane z. Equal numbers of
slower spins will diuse in from plane zz. Show that this means that the phase of M will
not change but the amplitude will. (d) It is reasonable to assume that the amplitude of the
diusion-induced decay will not depend on z as long as we are far from boundaries. Therefore
try a solution of the form
M
x
= M(0)e
t/T2
cos(G
z
zt)A(t)
M
y
= M(0)e
t/T2
sin(G
z
zt)A(t)
and show that A must obey the dierential equation
1
A
dA
dt
= D
2
G
2
z
t
2
which has a solution A(t) = exp(D
2
G
2
z
t 2/3). (e) Show that if there is a rotation about y

at time T
E
/2, then at time T
E
M
x
is given by
M
x
(T
E
) = M
0
exp(T
E
/T
2
) exp(D
2
G
2
z
T
3
E
/12).
Hint: This can be done formally from the dierential equations. However, it is much easier to
think physically about the meaning of each factor in the expressions shown in (d) for Mx

and
M
y
. This result means that a CPMG sequence with short T
E
intervals can reduce the eect
of diusion when there is an external gradient.
130

Das könnte Ihnen auch gefallen