24.03.2013 Views

link to lecture transcript - UT-H GSBS Medical Physics Class Site

link to lecture transcript - UT-H GSBS Medical Physics Class Site

link to lecture transcript - UT-H GSBS Medical Physics Class Site

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Today, we continue our talk about Comp<strong>to</strong>n scatter.<br />

1


Let us first review what we’ve covered so far about Comp<strong>to</strong>n scatter cross sections.<br />

We’ve said that the Comp<strong>to</strong>n scatter cross section is equal <strong>to</strong> the classical scatter<br />

cross section in the limit of zero energy. It’s also equal <strong>to</strong> the classical scatter cross<br />

section for forward scatter, that is, θ = 0. The Comp<strong>to</strong>n scatter cross section<br />

becomes more peaked in the forward direction as the pho<strong>to</strong>n energy increases. The<br />

Comp<strong>to</strong>n scatter cross section for backscattered pho<strong>to</strong>ns, that is, θ = π, is lower than<br />

that for θ = 0 by about one order of magnitude for pho<strong>to</strong>ns of energy 1 MeV, and<br />

about two orders of magnitude for pho<strong>to</strong>ns of energy 10 MeV. This is what we<br />

know so far about Comp<strong>to</strong>n scatter.<br />

What we are going <strong>to</strong> do now is try <strong>to</strong> get an equation, first of all for the <strong>to</strong>tal scatter<br />

cross section, and then we will look at the energy gy dependence p of the cross section.<br />

The <strong>to</strong>tal scatter cross section is the attenuation coefficient per unit incident pho<strong>to</strong>n<br />

fluence. Thus, from the scatter cross section we will be able <strong>to</strong> obtain the<br />

attenuation coefficient.<br />

2


We previously derived an equation for the differential scatter cross section per unit<br />

solid angle. The first step <strong>to</strong>ward obtaining the <strong>to</strong>tal scatter cross section is <strong>to</strong><br />

convert the differential scatter cross section per unit solid angle <strong>to</strong> the differential<br />

scatter cross section per unit plane angle. We are going <strong>to</strong> do the exact same thing<br />

that we did with coherent scatter and use the relationship dΩ = 2π sin θ dθ.<br />

So dσ/dθ is equal <strong>to</strong> dσ/dΩ times dΩ/dθ, which is dσ/dΩ times 2π sin θ. This is<br />

exactly what we did before with the cross section for coherent scatter.<br />

3


Note that the sin θ term suppresses the cross section at θ = 0 and θ = π as it had done<br />

for the expression for the cross section for coherent scatter.<br />

So the final thing that we need <strong>to</strong> do is integrate over θ <strong>to</strong> obtain a <strong>to</strong>tal cross section<br />

which is equal <strong>to</strong> this quantity. We see here σ 0, which is the cross section for<br />

coherent scatter, and then we see a lot of fac<strong>to</strong>rs involving α, where α is related <strong>to</strong><br />

the energy of the pho<strong>to</strong>n beam.<br />

4


So σ 0 is the classical cross section, which is 8/3π times the classical radius of the<br />

electron squared, and the remaining term is the integral of the Klein-Nishina<br />

coefficient.<br />

The <strong>to</strong>tal Comp<strong>to</strong>n scatter cross section is directly related <strong>to</strong> the electronic<br />

attenuation coefficient. Remember, we’re looking at interactions with the electrons.<br />

Therefore, the attenuation coefficient we’re getting is an electronic attenuation<br />

coefficient, which is the <strong>to</strong>tal Comp<strong>to</strong>n coefficient. From this attenuation<br />

coefficient, we can derive the energy transfer coefficient, the energy absorption<br />

coefficient, etc.<br />

If we want these numbers, we can look them up in J&C Table A2A second column.<br />

5


Take a look at that table, and observe the energy dependence of the Comp<strong>to</strong>n scatter<br />

coefficient for free electrons. Remember we said that the classical cross section was<br />

independent of energy. Now let’s look at the correction due <strong>to</strong> multiplying the<br />

classical attenuation coefficient by the integral of the Klein-Nishina coefficient.<br />

At 1 keV the Comp<strong>to</strong>n cross section is roughly equal <strong>to</strong> the classical value – it’s<br />

roughly 10 -28 m 2 /electron.<br />

At 7 MeV, we are down 1 order of magnitude; at 100 MeV we are down two orders<br />

of magnitude.<br />

Now notice this. From 1 keV <strong>to</strong> 100 MeV, the scatter coefficient drops down two<br />

orders d of f magnitude. it d As A a function f ti of fenergy we go down d five fi orders d of f magnitude it d<br />

of energy from 1 keV <strong>to</strong> 100 MeV, but we only drop down two orders of magnitude<br />

in the Comp<strong>to</strong>n scatter coefficient. Because of this we say the Comp<strong>to</strong>n scatter<br />

coefficient is roughly independent of energy.<br />

6


Let’s look at this graph. Notice that this is a log-log plot. The <strong>to</strong>tal Comp<strong>to</strong>n scatter<br />

coefficient is on the vertical axis, and the pho<strong>to</strong>n energy is on the horizontal axis.<br />

7


At least for a while it’s very flat and then it starts trailing off. So we drop two<br />

orders of magnitude for five orders of magnitude of energy and we say the Comp<strong>to</strong>n<br />

scatter coefficient is essentially independent of energy.<br />

8


What about the dependence of the Comp<strong>to</strong>n coefficient on a<strong>to</strong>mic number Z? The model we are using is a free<br />

electron gas. We are looking at the probability of an interaction per electron. We are not talking about the<br />

nucleus. Nor are we talking about the size of the nucleus or the number of electrons in the a<strong>to</strong>m. Consequently,<br />

the electronic coefficient for Comp<strong>to</strong>n scatter is essentially independent of Z, and, as we have seen previously, it<br />

is also roughly independent of pho<strong>to</strong>n energy.<br />

So what is it dependent on? About the only quantity the linear attenuation coefficient is going <strong>to</strong> be dependent<br />

on is density. The mass attenuation coefficient is going <strong>to</strong> be essentially independent of everything.<br />

Consequently, when Comp<strong>to</strong>n scatter is the predominant interaction, the linear attenuation is going <strong>to</strong> be linearly<br />

dependent on the density.<br />

AAnother th way of f saying i this thi is i “Equal “E l masses attenuate tt t equal l amounts.” t ” That’s Th t’ actually t ll a very good d thing; thi and d<br />

the reason why that’s good results from the fact that radiation dose is defined <strong>to</strong> be energy absorbed per unit<br />

mass. If we look at the absorption of energy in one gram of bone versus one gram of soft tissue with the same<br />

incident pho<strong>to</strong>n fluence, we find that one gram of bone absorbs approximately the same amount of energy as one<br />

gram of soft tissue. Equal masses absorb equal amounts.<br />

How is that different from a situation in which the pho<strong>to</strong>electric effect is the predominant interaction?<br />

Remember that for the pho<strong>to</strong>electric effect, absorption is going <strong>to</strong> be very strongly dependent upon a<strong>to</strong>mic<br />

number. So high-Z materials absorb a greater fraction of pho<strong>to</strong>ns than do low-Z materials. This difference in<br />

absorption gives rise <strong>to</strong> contrast in diagnostic imaging and an enhancement of dose <strong>to</strong> bone in low-energy<br />

radiation therapy therapy. This was a problem in the pre-megavoltage pre megavoltage era when we were treating patients with 100 kVp<br />

or 250 kVp x-rays. Because there was so much pho<strong>to</strong>electric interaction, for a given amount of radiation, bone<br />

would receive a much higher dose than soft tissue.<br />

When we’re in the Comp<strong>to</strong>n regime, bone receives the same dose as soft tissue, and that’s good.<br />

Now, from the point of view of imaging, Comp<strong>to</strong>n interaction is not good because we don’t have as much<br />

contrast. Recall that equal masses absorb equal amounts, so that because bone is a little denser than soft tissue,<br />

we get a little bit of contrast between bone and soft tissue.<br />

From the point of view of shielding, the Comp<strong>to</strong>n interaction is very good because equal masses absorb equal<br />

amounts. So an equal mass of lead is as good a shield as an equal mass of concrete. Now if you were a builder<br />

what would you rather work with, lead or concrete? Concrete is a standard building material; builders know<br />

how <strong>to</strong> pour concrete. They need <strong>to</strong> make sure there aren’t any air bubbles or anything like that. It is easy <strong>to</strong><br />

pour four feet of concrete. Consequently when we shield a radiation therapy facility we are going <strong>to</strong> be looking<br />

at conventional building materials <strong>to</strong> do our shielding; we just use thick walls. Several inches of lead shielding<br />

may be just as effective a shield as a few feet of concrete, but the lead is much harder <strong>to</strong> work with and builders<br />

don’t like <strong>to</strong> use it.<br />

9


Next we want <strong>to</strong> calculate the energy transfer coefficient. We have calculated the<br />

<strong>to</strong>tal attenuation coefficient, which tells us the fraction of pho<strong>to</strong>ns attenuated per<br />

unit path length. We want <strong>to</strong> calculate the energy transfer coefficient <strong>to</strong> determine<br />

how much energy is transferred <strong>to</strong> the electrons as a result of the interactions. In<br />

order <strong>to</strong> do so, we take the linear attenuation coefficient and weight it by the<br />

fraction of the incident pho<strong>to</strong>n energy that is transferred <strong>to</strong> the electrons. This<br />

fraction is equation <strong>to</strong> the amount of energy transferred <strong>to</strong> the electrons divided by<br />

the incident pho<strong>to</strong>n energy. energy<br />

10


The energy transferred <strong>to</strong> the electron is given by this equation; we derived that in<br />

the previous <strong>lecture</strong>. The fraction of energy transferred <strong>to</strong> the electrons is this<br />

quantity divided by hν. So dσ tr/dΩ is ½ r 0 2 times 1+ cos 2 θ , that is, the classical<br />

quantity, times the Klein-Nishina coefficient times the ratio α(1 – cos θ) divided by<br />

1 + α(1 – cos θ).<br />

11


If we want <strong>to</strong> calculate the <strong>to</strong>tal cross section, hence the attenuation coefficient, we<br />

have <strong>to</strong> integrate over angles. The integration is rather messy.<br />

I am not going <strong>to</strong> integrate this for you as we can spend a couple of days doing the<br />

integration; I’d rather not do it.<br />

Th The bot<strong>to</strong>m b tt line li is i that th t we still till have h this thi ¾ σ0, which hihi is what htwe had hdf from the thlinear li<br />

attenuation coefficient times a horrendous quantity that has a very complicated<br />

energy dependence.<br />

12


If we plot the energy transfer coefficient as a function of energy, this is what we get.<br />

Notice that, unlike the linear attenuation coefficient, the energy transfer coefficient<br />

is suppressed at low energies.<br />

13


For high energy pho<strong>to</strong>ns, that is, large values of α, the energy transfer coefficient<br />

approaches the classical coefficient.<br />

14


The energy transfer coefficient has a maximum in the vicinity of ½ <strong>to</strong> 1 MeV. This<br />

means we have an enhancement of energy transferred <strong>to</strong> kinetic energy of charged<br />

particles at energies fairly close <strong>to</strong> what we like <strong>to</strong> use for radiation therapy. That’s<br />

really neat, because we want <strong>to</strong> transfer as much of this pho<strong>to</strong>n energy as possible <strong>to</strong><br />

charged particles, that is, the electrons, <strong>to</strong> produce secondary ionizations. So that’s<br />

good. We have a maximum energy transfer where we would like such a maximum<br />

<strong>to</strong> be.<br />

15


At low energies, however, the energy transfer coefficient is strongly suppressed.<br />

Comp<strong>to</strong>n scatter is not a good energy transfer mechanism at low energies. To recap,<br />

energy transfer via Comp<strong>to</strong>n scatter is very efficient at high energies, but at low<br />

energies Comp<strong>to</strong>n scatter is not an efficient mechanism for energy transfer.<br />

16


What about the scatter coefficient? Remember when we have attenuation, two<br />

things can happen. Some of the energy can be transferred <strong>to</strong> charged particles,<br />

while some of the energy can go <strong>to</strong> a scattered pho<strong>to</strong>n. So the scatter coefficient is<br />

going <strong>to</strong> be equal <strong>to</strong> the <strong>to</strong>tal coefficient minus the energy transfer coefficient.<br />

At low energies, the energy transfer is small, so the scatter coefficient is<br />

approximately equal <strong>to</strong> the <strong>to</strong>tal attenuation coefficient. Comp<strong>to</strong>n scatter is not a<br />

good process for transferring energy <strong>to</strong> charged particles at low energies, but this is<br />

also consistent with classical scatter, so we do wind-up getting the coherent scatter<br />

in the limit of low energies.<br />

At high energies, almost all of the energy gets transferred <strong>to</strong> charged particles.<br />

Consequently, q y the scatter coefficient approaches pp zero. So the scatter coefficient<br />

goes between the <strong>to</strong>tal attenuation coefficient at low energies and zero at high<br />

energies.<br />

Now, let’s use this information about coefficients <strong>to</strong> deduce some properties of<br />

energy transfer and energy absorption.<br />

17


For example, how do we determine the mean energy that’s transferred <strong>to</strong> charged<br />

particles?<br />

We now know how <strong>to</strong> determine mean energy transfer based on the definition of<br />

coefficients. The mean energy transfer is the initial pho<strong>to</strong>n energy multiplied by the<br />

ratio of energy transfer coefficient <strong>to</strong> attenuation coefficient.<br />

18


Using this equation we can now plot mean energy transferred via Comp<strong>to</strong>n<br />

interactions vs pho<strong>to</strong>n energy.<br />

19


More important is the energy distribution. We have incident pho<strong>to</strong>ns interacting<br />

with absorbing material via Comp<strong>to</strong>n interactions. Let’s look at a monoenergetic<br />

beam of pho<strong>to</strong>ns and determine the distribution of energies of the Comp<strong>to</strong>n<br />

electrons that are going <strong>to</strong> be produced.<br />

Let’s determine the probability distribution of electrons generated from a<br />

monoenergetic pho<strong>to</strong>n.<br />

20


This probability is represented by dσ/dE. We can write this as the quantity dσ/dθ<br />

times dθ/dE. We have already calculated dσ/dθ, so all we need <strong>to</strong> calculate is<br />

dθ/dE. We have an equation of energy as a function of scatter angle, and we can use<br />

this equation <strong>to</strong> get dE/dθ.<br />

21


We find that dE/dθ is the incident pho<strong>to</strong>n energy, hν, times α sin θ, divided by the<br />

quantity [1 + α(1 – cos θ)] 2 . We combine this quantity with dσ/dθ <strong>to</strong> obtain an<br />

expression for dσ/dE, the distribution of energies of Comp<strong>to</strong>n electrons.<br />

22


This figure is a plot taken from Johns & Cunningham. For each value of incident<br />

pho<strong>to</strong>n energy, we have a different curve for dσ/dE.<br />

Note first, that for each pho<strong>to</strong>n energy, there is a maximum energy of the Comp<strong>to</strong>n<br />

electron. What is this maximum electron energy? Recall that the maximum energy<br />

imparted <strong>to</strong> the Comp<strong>to</strong>n electron occurs when the pho<strong>to</strong>n is scattered 180°, or<br />

backscattered. We determined this quantity in the previous <strong>lecture</strong> <strong>to</strong> be equal <strong>to</strong> the<br />

energy of the incident pho<strong>to</strong>n multiplied by the ratio 2α divided by 1 + 2α. For<br />

example, for a 1 MeV pho<strong>to</strong>n, the energy of the scattered pho<strong>to</strong>n is 204 keV, so the<br />

maximum energy of the Comp<strong>to</strong>n electron is 796 keV. For a 500 keV pho<strong>to</strong>n, we<br />

can calculate the maximum electron energy <strong>to</strong> be a bit less than 350 keV.<br />

23


We also observe there are peaks in dσ/dE at both zero energy and the maximum<br />

energy. Low energy and high energy electrons are more likely than medium energy<br />

electrons.<br />

Notice, for example, for a high energy incident pho<strong>to</strong>n the differential energy cross<br />

section is relatively flat but as you get <strong>to</strong> the maximum energy you have a peak. For<br />

very low energy pho<strong>to</strong>ns we find there is a significant decrease in differential cross<br />

section as we increase the energy. We observe a distinct minimum in the energy<br />

distribution of the scattered electrons and then again a peak.<br />

24


We can always calculate what the maximum electron energy is from the Comp<strong>to</strong>n<br />

equation; this is the quantity that we call a Comp<strong>to</strong>n edge, an observed increase in<br />

the energy distribution of the scattered electrons.<br />

So if we are able <strong>to</strong> look at the scattered electron energy distribution using some<br />

kind of multi-channel analyzers or something like that we will find there’s going <strong>to</strong><br />

be a peak right at the maximum energy. This maximum is called the Comp<strong>to</strong>n edge.<br />

I think you are going <strong>to</strong> be seeing Comp<strong>to</strong>n edges in subsequent courses when we<br />

talk about scintillation detec<strong>to</strong>rs.<br />

25


How do we determine the <strong>to</strong>tal cross section?<br />

We can integrate the differential cross section over energies. It turns out we can do<br />

this integration numerically rather simply.<br />

26


Remember dσ/dE is relatively constant with electron energy. If you want <strong>to</strong><br />

estimate the electronic attenuation coefficient from a differential coefficient, it is<br />

rather easy <strong>to</strong> do so. This coefficient the electronic attenuation coefficient then is<br />

going <strong>to</strong> be directly proportional <strong>to</strong> the mass attenuation coefficient.<br />

27


Here’s an example of how we can estimate the electronic attenuation coefficient.<br />

We want <strong>to</strong> determine the electronic attenuation coefficient for a 0.8 MeV pho<strong>to</strong>n in<br />

a free electron gas. How do we determine this? First of all, the maximum energy of<br />

the electrons can be estimated from the graph or calculated from a formula <strong>to</strong> be<br />

0.606 MeV.<br />

28


Now we also look at the graph and see that dσ/dE is approximately constant with a<br />

value of 40 10 -30 m 2 /e MeV.<br />

To a first approximation, let’s forget about the Comp<strong>to</strong>n edge; we will just<br />

approximate the integration by assuming that dσ/dE is constant. To do the<br />

integration, we multiply dσ/dE by the energy interval, 0.606 MeV, and find that σ is<br />

the value of 24 10 -30 m 2 /e. Now if we look at the number in Johns & Cunningham<br />

we find it’s 23.5 10 -30 m 2 /e. So simply by assuming dσ/dE <strong>to</strong> be constant, we get<br />

a very good estimate of the <strong>to</strong>tal cross section.<br />

29


Here’s another example.<br />

Let’s determine the number of Comp<strong>to</strong>n interactions and the number of electrons<br />

that are set in<strong>to</strong> motion with energies between 0.15 MeV and 0.25 MeV when we<br />

have a slab of bone 6 mm thick, and bombard the bone with 10 4 pho<strong>to</strong>ns of energy<br />

0.5 MeV.<br />

We are going <strong>to</strong> do some accounting now. We are going <strong>to</strong> pretend we are<br />

accountants and we want <strong>to</strong> follow the energy and figure out where these Comp<strong>to</strong>n<br />

electrons go and what kind of energy these Comp<strong>to</strong>n electrons have.<br />

30


First, let’s recall that the attenuation coefficient is the fraction attenuated per unit<br />

absorber b b thickness. hi k How many pho<strong>to</strong>ns h are attenuated? d? We could ld use the h<br />

exponential approximation but when we’re estimating let’s use the thin absorber<br />

approximation. The number of pho<strong>to</strong>ns attenuated is the number of incident<br />

pho<strong>to</strong>ns times the attenuation coefficient times the absorber thickness.<br />

How many incident pho<strong>to</strong>ns do we have? We have been <strong>to</strong>ld that the number of<br />

incident pho<strong>to</strong>ns is 104 cdetp o<strong>to</strong>ss 0. . WWhat at is s the t e Comp<strong>to</strong>n Co p<strong>to</strong> attenuation atte uat o coefficient coe c e t for o ½ MeV eV<br />

pho<strong>to</strong>ns. We look that up in Table A2a in Johns & Cunningham and find that <strong>to</strong> be<br />

0.2892 10-28 m2 per electron.<br />

What’s the thickness of the absorber? The thickness of the absorber is the thickness<br />

of the bone times the density of bone. We multiply the thickness of bone, 0.6 cm,<br />

by the density of bone, 1.650 kg/m3 , which we get from the Table 5.3. Next we<br />

multiply by 10-2 multiply by 10 <strong>to</strong> convert the bone thickness from meters <strong>to</strong> cm<br />

2 <strong>to</strong> convert the bone thickness from meters <strong>to</strong> cm.<br />

Let’s convert that in<strong>to</strong> centimeters times the electron density. Remember we have<br />

<strong>to</strong> be able <strong>to</strong> calculate the electron density <strong>to</strong> be Z over A times Avogadro’s number.<br />

Z over A is about a half. So electron density is always going <strong>to</strong> be roughly half of<br />

Avogadro’s number. That’s a very good approximation for low <strong>to</strong> moderate Z<br />

material, such as bone. Electron density comes out <strong>to</strong> be 3.19 10 23 electrons per<br />

kilogram. So the electron areal density is 3.16 10 27 electrons per meter squared.<br />

31


So 3.16 10 27 electrons per meter squared times the Comp<strong>to</strong>n coefficient times the<br />

number of incident pho<strong>to</strong>ns gives us the number of pho<strong>to</strong>ns that are scattered.<br />

The result is 9.14 10 2 pho<strong>to</strong>ns.<br />

We find the assumption of a thin absorber is reasonable in this case. So now we<br />

kknow hhow many electrons l t are scattered. tt d<br />

32


What we now want <strong>to</strong> find out is how many electrons have energies between 0.15<br />

and 0.25 MeV.<br />

What is the differential cross section for pho<strong>to</strong>ns of incident energy ½ MeV<br />

producing electrons of energy around 0.2 MeV? We find the differential scatter<br />

cross section is roughly 70 10 -30 meters squared per electron per MeV. That’s a<br />

ballpark figure. Again, we’re estimating here. We will not do an exact calculation.<br />

To do the exact calculation you would have <strong>to</strong> plug and chug in<strong>to</strong> the equation for<br />

differential cross section, which I am not going <strong>to</strong> have you <strong>to</strong> do.<br />

33


We now have the differential scatter cross section. The energy interval we’re<br />

looking at from 0.15 <strong>to</strong> 0.25 MeV is 0.1 MeV. To obtain the cross section we<br />

integrate the differential cross section over the energy interval. Assuming the<br />

differential cross section is constant over that energy interval, a reasonable<br />

assumption, we find the cross section is 7 10-30 meters squared per electron. To<br />

calculate the number of electrons generated in this energy range we multiply the<br />

number of incident pho<strong>to</strong>ns, 104 , times the cross section, 7 10-30 meters squared<br />

per electron times the absorber thickness 3 16 1027 per electron, times the absorber thickness, 3.16 10 electrons per meter squared squared,<br />

giving us an answer of 220 electrons.<br />

In general when we want <strong>to</strong> calculate the number of electrons that are liberated, we<br />

take the differential energy cross section, multiply it by the energy interval, multiply<br />

it by the number of incident pho<strong>to</strong>ns, multiply that by the electron density. The cross<br />

section section, then then, is the probability of an interaction per unit incident pho<strong>to</strong>n fluence. fluence<br />

Is everyone comfortable with this kind of a calculation? The problem sets will give<br />

you more practice in doing calculations of this nature. Our basic goal in doing these<br />

problems is <strong>to</strong> be able <strong>to</strong> know where the energy is going. Remember, we need <strong>to</strong><br />

follow the energy.<br />

34


So far, our approximations have been based on the model of a free electron gas. For<br />

a free electron gas we have no binding energy. There is no force attracting the<br />

electrons <strong>to</strong> the a<strong>to</strong>m, but that’s not completely correct.<br />

So now we are going <strong>to</strong> make a minor refinement and account for binding energy.<br />

To do this, we have <strong>to</strong> use quantum mechanics. We need the wave functions for<br />

a<strong>to</strong>ms. You can only do quantum mechanics relatively accurately on fairly simple<br />

a<strong>to</strong>ms.<br />

The effects of binding energy are important only for low energy incident pho<strong>to</strong>ns<br />

for which the binding energy is comparable or at least not very much less than the<br />

energy of the incident pho<strong>to</strong>n.<br />

35


To account for binding energy, we multiply the differential cross section, dσ/dθ,<br />

based on the Klein-Nishina theory, times a quantity that amounts <strong>to</strong> an a<strong>to</strong>mic form<br />

fac<strong>to</strong>r. It represents the probability that an electron actually leaves the a<strong>to</strong>m. It’s a<br />

function of x, which is related <strong>to</strong> the scatter angle and the pho<strong>to</strong>n energy, and it’s a<br />

function of a<strong>to</strong>mic number Z.<br />

36


These values were tabulated by Hubbell in the Journal of Physical Chemistry<br />

Reference Data.<br />

So if we take this quantity, take everything <strong>to</strong>gether, and integrate over angles, we<br />

can obtain the Comp<strong>to</strong>n coefficient in the real world. We are going <strong>to</strong> have some<br />

slight deviations from the Comp<strong>to</strong>n coefficient for a free electron gas that account<br />

for binding energy.<br />

37


We can’t evaluate this integral in closed form, but if we look at Johns &<br />

Cunningham Table A4, we’ll see different values for Comp<strong>to</strong>n coefficients for each<br />

material. These values differ slightly from those for a free electron gas because of<br />

the fact there is binding energy.<br />

38


So what is the effect of binding energy? Here is a plot of Comp<strong>to</strong>n coefficients<br />

versus pho<strong>to</strong>n energy for a free electron gas, oxygen, and lead. We see the curve<br />

for the free electron gas on <strong>to</strong>p; below that is the curve for oxygen; the lowest<br />

values are for lead.<br />

First of all, notice the binding energy only has an effect on the Comp<strong>to</strong>n coefficient<br />

at low energies; once you get above 100 keV, the binding energy essentially has no<br />

effect.<br />

Binding energy is more important for higher Z materials. This also makes sense.<br />

So the impact of binding energy on Comp<strong>to</strong>n coefficient is most noticeable for high-<br />

Z materials.<br />

In general, if we look at energies where the Comp<strong>to</strong>n effect is the predominant<br />

interaction, the effect of binding energy is not significant.<br />

39


Let us now summarize what we know about Comp<strong>to</strong>n scatter.<br />

First of all, if we base our determination of the attenuation coefficient on the free<br />

electron model, the attenuation coefficient is nearly constant <strong>to</strong> 100 keV. It falls less<br />

than 2 orders of magnitude out <strong>to</strong> 100 MeV. So the coefficient is proportional <strong>to</strong> hν<br />

raised <strong>to</strong> a power something smaller than -1, maybe -2/3, from 100 keV <strong>to</strong> 100 MeV.<br />

To a ballpark approximation we can say it’s roughly independent of energy.<br />

At low energies, the binding energy of the electrons affects the Comp<strong>to</strong>n scatter<br />

coefficient suppressing attenuation in high-Z materials such as lead. But this<br />

suppression only occurs at lower energies.<br />

40


At higher energies there is no difference between any of the Comp<strong>to</strong>n attenuation<br />

coefficients so then σ/ρ is essentially independent of Z.<br />

The dependence of the Comp<strong>to</strong>n coefficient on a<strong>to</strong>mic number is really an effect of<br />

the binding energies. At higher pho<strong>to</strong>n energies you can forget about the binding<br />

energies; we are basically dealing with free electrons. So in summary, σ/ρ, the <strong>to</strong>tal<br />

attenuation coefficient, is roughly independent of Z, and depends on the hν <strong>to</strong><br />

something a little bit less than the magnitude of -1, say -2/3 or something like that.<br />

41


In this graph we see the differential angular cross section for Comp<strong>to</strong>n scatter as a<br />

function of scatter angle. Thompson scatter is the curve that’s symmetric in angle,<br />

Rayleigh scatter peaks at low angle scatter.<br />

42


Note that this graph is for very low energy pho<strong>to</strong>ns incident on a low-Z target, 10<br />

keV pho<strong>to</strong>ns incident on carbon.<br />

Coherent scatter at these low energies is significant and stronger than Comp<strong>to</strong>n<br />

scatter. Also the binding energy has a significant effect on Comp<strong>to</strong>n scatter at these<br />

low energies and that Comp<strong>to</strong>n scatter approaches the classical limit.<br />

43


Comp<strong>to</strong>n scatter involves the interaction between a pho<strong>to</strong>n and a single electron.<br />

The probability of Comp<strong>to</strong>n scatter is almost independent of Z; it decreases slightly<br />

with increasing energy.<br />

When Comp<strong>to</strong>n scatter occurs, some of the energy is scattered and some of the<br />

energy is transferred in<strong>to</strong> the Comp<strong>to</strong>n electron. The amount, the distribution of<br />

energy, depends on the angle of emission of the scattered pho<strong>to</strong>n and the energy of<br />

the incident pho<strong>to</strong>n.<br />

44


The fraction of energy transferred <strong>to</strong> kinetic energy of charged particles per<br />

collision increases with increasing pho<strong>to</strong>n energy. For low energy pho<strong>to</strong>ns, energy<br />

transfer is low, σ tr is much smaller than σ; Comp<strong>to</strong>n scatter is not an efficient means<br />

of energy transfer. For high energy pho<strong>to</strong>ns, σ tr is approximately equal <strong>to</strong> σ, and<br />

Comp<strong>to</strong>n scatter is an efficient method of energy transfer. And finally, in soft tissue<br />

Comp<strong>to</strong>n scatter is the most important process in the energy range of 30 KeV <strong>to</strong> 30<br />

MeV.<br />

We’ve spent a lot of time on Comp<strong>to</strong>n scatter. The reason we’re spending a lot of<br />

time on it is because Comp<strong>to</strong>n scatter is a very important process. It’s the most<br />

important process in the therapy energy range and matches the pho<strong>to</strong>electric effect<br />

for importance in the diagnostic energy range.<br />

45

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!