Compton Scattering

Raphael Nydegger ([email protected])

Niklaus Häfliger ([email protected])
January 19, 2010

TU
AS
UN I VE RSI T

RI
CEN S I S

MDCCC XXXIII
Contents
1 Theory 2
1.1 Compton Effect . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Compton Cross Section . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Differential Cross Section . . . . . . . . . . . . . . . . . . . . . 2
1.4 Used Constants . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Experiment 3

3 Data and Analysis 6

3.1 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 Zero Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.3 Solid Angle of the Detector . . . . . . . . . . . . . . . . . . . 8
3.4 Effective Thickness of the Target . . . . . . . . . . . . . . . . 8
3.5 Rγ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.6 Verification of Compton Relation . . . . . . . . . . . . . . . . 10
3.7 Total Cross Section . . . . . . . . . . . . . . . . . . . . . . . . 11
3.8 Differential Cross Section . . . . . . . . . . . . . . . . . . . . . 13

4 Discussion 15
4.1 Compton Relation . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2 Total Cross Section . . . . . . . . . . . . . . . . . . . . . . . . 15
4.3 Differential Cross Section . . . . . . . . . . . . . . . . . . . . . 16

A Raw Data and Statistical Analysis 17

1
1 Theory
1.1 Compton Effect
A photon, emitted by a 137 Cs source, gets scattered at an electron. The pho-
ton (gamma quant) and the electron change their momenta, while the total
energy is conserved. We derive the following relations about the respective
energies of the photons before (Eγ ) and after (Eγ0 ) the scattering under the
angle θ:
Eγ0 1
= (1.1)
Eγ 1 + γ(1 − cos(θ))
Eγ
γ := (1.2)
me c2

1.2 Compton Cross Section

The cross section σ is a hypothetical area of the electron perpendicular to
the trajectory of the photon. The photon scatters if it hits this area. The
cross section is described by the following equation:
Rscattered · M
σ= (1.3)
Rγ · Z · d · NA · ρ
We determine the cross section by measuring the decrease of the intensity
the photon ray experiences when crossing a plate of thickness x. The intensity
follows the exponential decrease described in the following formulas:

Rγ (x) = Rγ (0) · e−µx σ (1.4)

Z · NA · σ · ρ
µ := (1.5)
M

1.3 Differential Cross Section

The differential cross section is the hypothetical area which the photons must
hit to scatter under a specific angle θ. We can derive the differential cross
section from formula (1.3), which results in the following relation:

dσ(θ) Rscattered (θ) · M

= (1.6)
dΩ Rγ · Z · d · NA · ρ · ΩD
The theoretical prediction for the differential cross section by Oskar Klein
and Yoshio Nishina (derived in 1929) is:

2
 dσ(θ) r2
= e (κ(θ) − κ(θ)2 sin2 θ + κ(θ)3 ) (1.7)
dΩ 2
Eγ0 (θ)
κ(θ) := (1.8)
Eγ
e2
re = = 2.818 · 10−15 m (1.9)
4π0 c2
By integrating formula (1.7) over the total solid angle we get a formula
for the theoretical prediction of the total cross section.

πre2 2γ 2 (1 + γ) γ 2 − 2γ − 2
σ= (4 + + ln(1 + 2γ)) (1.10)
γ2 (1 + 2γ)2 γ

1.4 Used Constants

For our calculations we used the following constants:

M : molar mass of aluminium 0.02698 kg/Mol

Z: number of electrons in an Al atom 13
NA : Avogadro’s constant 6.022 · 1023 mol−1
ρ : density of Al 2698.9 kg/m3
0 : electric field constant 8.85 · 10−12 As V−1 m−1
c : speed of light 2.998 · 108 m/s
re : electron radius 2.818 · 10−15 m

2 Experiment
For our experiment we use a 22 Na source for the calibration and a 137 Cs source
for the actual measurements. The photons scatter at an aluminium target
and are afterwards detected with an Nal(TI)-Detector composed of a scintil-
lator, a photomultiplier and an integrator. The outputs of the integrator are
amplified and then digitized by a PCI card. Since the amplitude of the final
signal is proportional to the energy of the detected photon, this allows us to
measure those energies. The Maestro software program automatically cre-
ates a histogram of the photon count for each channel (corresponding to an
energy interval). Additionally we can rotate the source, so we can measure
a variety of angles between the target and the detector (we were constrained
between zero and about 130◦ ). Below follow some pictures to illustrate the
configuration of the experiment.

3
Figure 2.1: Cs source in a lead block

Figure 2.2: Opening of the detector

4
Figure 2.3: Detector with photomultiplier (grey)

Figure 2.4: Amplifier

5
 Figure 2.5: Maestro software plotting data in a histogram

3 Data and Analysis

3.1 Calibration
For the calibration we measured the background and the spectrum of a 22 Na-
Source for 600 seconds. We then fitted two gauss-curves into the two specific
peaks for sodium and derived a function for the channel-energy-relation. The
fits with Matlab are shown in figure (3.1).

6
 Calibration
700

600 Energy Spectrum of an 22Na−Source

500 ← 511 keV Peak

Photon Count per Channel

400

300

200

1274.6 keV Peak

100 ↓

−100
0 500 1000 1500 2000 2500
Channel

Figure 3.1: Calibration spectrum

So we get the calibration function

E = 0.8114 · C − 21.9326 (3.1)

where E is the energy and C is the channel number.

3.2 Zero Angle

To determine the zero angle we took 10 measurements for 120 seconds each
from 340 to 360 degrees with the 137 Cs-Source. The degrees are relative to
the rotating base in which center we later placed the target.
We first fitted a Gauss-curve to the data for every angle to get the number of
photons which had been detected for every measurement. These results are
shown in figure (3.2) below. In this figure we also plotted the gauss-curve
which we fitted through the data to get the zero angle. The error on the zero
angle results from the fitting of the curve.
The errors on the number of photons detected are so small you can’t
really see them in figure (3.2). We therefore also show an enlarged area of
figure (3.2) in figure (3.3).
We obtained the preliminary result of 349◦ during the experiment and
used that as our zero angle for further angle measurements. So when we

7
refer to angles in whole integer numbers such as 60◦ , in reality this always
means 60.4◦ .
5 Determination of Zero Angle
x 10
8

6
Number of Photons detected

0 Zero Angle = 348.6021

Error = 0.48861
−2

−4

−6
330 335 340 345 350 355 360 365 370
Angle

Figure 3.2: Determination of the zero angle

3.3 Solid Angle of the Detector

For calculating the solid angle of the detector, we measured the distance
from the target to the detector as R = 18.8 ± 0.5 cm, and the radius of the
collimator as r = 1.205 ± 0.01 cm. Since the solid angle is calculated as

r2 π
ΩD = (3.2)
R2
we get a solid angle of ΩD = 0.0129 with an error of
s 2  2
2rπ −2r3 π
mΩD = mr + mR = 7.2 · 10−4
R2 R3

3.4 Effective Thickness of the Target

The thickness of the target was 3 mm, however it was placed at an angle
with respect to the ray of photons. Since we made sure never to place it

8
 5 Determination of Zero Angle
x 10
7.5

7
Number of Photons detected

6.5

5.5

4.5

344 345 346 347 348 349 350 351 352 353
Angle

Figure 3.3: Zoomed area of figure (3.2)

◦
at an angle of more than 45√ , we can be certain that the effective thickness
was√between 3 mm and 3 · 2 mm. For our calculations we used the value
3 · 2 − 0.5mm = 3.74 mm ± 0.5 mm.

3.5 Rγ
We measured the total output times the probability of the detector going off
of the 137 Cs-source with an 70 mm aluminium shielding for 120 seconds and
used formula (1.4) to calculate the output of the source per second without
shielding. The data and the results of the analysis are shown in figure (3.4)

9
 Determination of emmitted Photons by Cs−Source
1400

1200 With 70mm Shielding:

Area(70) = 398291.6291
Number of Photons per Channel

1000 Error(70) = 2360.4276

800 Without Shielding:

Area(0) = 1559599.7136
600 Error(0) = 9242.7809
Counts per second = 12996.6643
400 Error on cps = 77.0232

200

0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Channels

137
Figure 3.4: Output of the Cs source (Rγ )

3.6 Verification of Compton Relation

One of the goals of the experiment was to verify formula (1.1), the Compton
relation.
For each angle, we fitted the characteristic peak with a Gauss function,
which allowed us to determine the channel and therefore the energy of the
peak using the calibration function (3.1). Detailed plots of that data and the
fitted curves can be found in the appendix.
The plot of the measured energies is plotted versus the theoretical pre-
diction in figure (3.5) below.

10
 Verification of Compton relation
700
Theoretical Curve
650 Experimental Values
Corrected Values (see text)
600
Photon energy after scattering [keV]

550

500

450

400

350

300

250

200
0 20 40 60 80 100 120 140
Angle [Degrees]

Figure 3.5: The Compton relation

While there appears to be agreement between theory and experiment,

the data points for Θ = 50◦ and Θ = 60◦ don’t follow the general trend.
Taking a closer look at the data for Θ = 50◦ we found that the background
spectrum looked very much like the spectrum with scattering, leading us to
believe that we made a mistake when recording this spectrum. When we
instead took the average of the background from Θ = 40◦ and Θ = 60◦ , we
arrived at a value much closer to the one predicted by theory. For Θ = 60◦
we remembered that the source appeared to have been shifted by about 3◦
while we left the experiment unattended. So when we plot that energy at
57◦ we get very close to the theoretical value as well.
The errors on the energy we get from the fit are very small, between 0.5
and 1.5 keV. A bigger source of errors was determining the angle. Since oc-
casionally, the group doing the experiment next to us would bump into the
source and move it by a little bit, this would give an error of one degree. In
addition there was the error on the zero angle, which adds up to 1.5 degrees.

3.7 Total Cross Section

We measured the decrease of detected photons by placing aluminium plates of
different thicknesses between the 137 Cs-source and the detector. The results

11
are plotted in figure (3.6). The slope of the fitted line gives the µ in the
formula (1.4) and with formula (1.5) we get the total cross section σ. The
error on σ is calculated as follows:
M
mσ = · mµ (3.3)
Z · NA · ρ

Total Cross Section

13.8

13.6
mu = 19.5139
13.4 error on mu = 1.0775e−05
Log ( Number of Photons )

total cross section = 2.4918e−29

13.2 error on tcs = 1.3759e−35

13

12.8

12.6

12.4
−10 0 10 20 30 40 50 60 70
Aluminium Shielding in mm

Figure 3.6: Total cross section

12
 12000

10000
Counted Photons per Channel

8000 Fitted Peak at 669.0344 keV

Error = 0.069056 keV

6000

4000

2000

0
0 500 1000 1500 2000 2500
Channel

Figure 3.7: Eγ

From figure (3.7) we know the emitted energy Eγ from our 137 Cs-source
and we can compute γ := mEeγc2 = 1.30919. We get value of σ = 2.4918133
·10−29 ± 1.4 · 10−35 m2 . With formula (1.10) we compute the theoretical value
for the total cross section and get σtheoretical = 2.5495 · 10−29 m2 .

3.8 Differential Cross Section

The differential cross section was determined from the same data and the
same fits as the verification of the Compton relation. For the differential
cross section we don’t need to know the energy of the photons, but only the
photon count.
The areas and their respective errors were the result of the Gaussian fits
of the spectra, which can be seen in the appendix.
The error on the differential cross section mσ is calculated as follows:

v !2 !2 !2 !2
u dσ dσ dσ dσ
u ∂ dΩ ∂ dΩ ∂ dΩ ∂ dΩ
m dσ =t mRsc + mRγ + md + mΩD
dΩ ∂Rsc ∂Rγ ∂d ∂ΩD

which result in the following:

13
  2  2
M M Rsc
m dσ = mRsc + 2
mRγ
dΩ Rγ ZdNA ρΩD Rγ ZdNA ρΩD
 2  2 ! 12 (3.4)
M Rsc M Rscattered
+ md + mΩD
Rγ Zd2 NA ΩD Rγ ZdNA Ω2D

md = 0.0005 m
mRγ = 77
mΩD = 7.2 ·10−4
With the errors above we arrive at the following numbers:

Table 1: Differential cross section

Θ [◦ ] Rsc [s−1 ] mRsc [s−1 ] dσ

dΩ
(experimental) [m2 ] m dσ [m2 ] dσ
dΩ
(theory) [m2 ]
dΩ

30.4 7.18 0.14 1.460 · 10−29 0.214 · 10−29 0.5028 · 10−29

40.4 6.14 0.18 1.249 · 10−29 0.185 · 10−29 0.3797 · 10−29
50.4 2.58 0.15 0.525 · 10−29 0.082 · 10−29 0.2833 · 10−29
50.4 1 5.71 0.16 1.161 · 10−29 0.167 · 10−29 0.2833 · 10−29
60.4 8.70 0.15 1.770 · 10−29 0.258 · 10−29 0.2156 · 10−29
70.4 3.98 0.11 0.810 · 10−29 0.119 · 10−29 0.1715 · 10−29
80.4 4.53 0.13 0.921 · 10−29 0.136 · 10−29 0.1444 · 10−29
90.4 4.84 0.15 0.984 · 10−29 0.146 · 10−29 0.1287 · 10−29
100.4 3.9 0.13 0.793 · 10−29 0.118 · 10−29 0.1202 · 10−29
110.4 3.79 0.12 0.771 · 10−29 0.114 · 10−29 0.1162 · 10−29
120.4 3.11 0.18 0.633 · 10−29 0.099 · 10−29 0.1148 · 10−29

See figure (3.8) for a plot of this data.

1
This datapoint was calculated with a hypothetical reconstructed background, see the
appendix for details

14
 −29 Differential Cross Section
x 10
2.5
Experimental Values
Corrected Value (see text)
Theoretical Curve
2
Differential Cross Section [m2]

1.5

0.5

0
0 20 40 60 80 100 120 140
Angle (Degrees)

Figure 3.8: Differential cross section

4 Discussion
4.1 Compton Relation
Our data fit the Compton relation quite well, however for large angles there
was a systematic deviation for which we have no explanation.

4.2 Total Cross Section

The result of the total cross section is quite close to the theoretical value.
However, our error is much smaller than the difference between theory and
experiment. We think that this is due to the fact that our detector is too
large do differentiate between photons that were scattered under small angles
(smaller than about 3.6 degrees) and photons that were not scattered at all.
This is consistent with the fact that our measured cross section is smaller
than the theoretical prediction.

15
4.3 Differential Cross Section
Our result for the differential cross section differed from the prediction by
a large margin. It is our hypothesis that the problem lies first of all with
some of the data points as well as our measurement of Rγ , the total rate of
emitted photons. Looking at the spectra, in general the measurements for
large angles seem to be much better than the ones for small angles, because
the background spectra look more uniform there.
We also believe that our measurement for Rγ was flawed. We placed 70
mm of aluminium between the source and the detector in order to reduce
the measured rate in a predictable way in order to reduce dead time. The
instructions stated that 20 % or less were acceptable. While our shielding
got the dead time down to exactly 20 %, it seems that this was not enough.
It is our hypothesis that the software program corrects for dead time of the
PC components, however there is also dead time of the detector. It seems
plausible to us that this way the measured rate could be off by a factor
of about seven, which is what would be needed to match up the theory
with at least the data taken at large angles. See figure (4.1) for a plot of
the differential cross section assuming Rγ to be seven times larger than the
actual value.

−30 Differential Cross Section

x 10
8
Experimental Values
Corrected Value
7 Theoretical Curve

6
Differential Cross Section [m2]

0
0 20 40 60 80 100 120 140
Angle [Degrees]

Figure 4.1: Hypothetical differential cross section

16
A Raw Data and Statistical Analysis
The data for the verification of the Compton relation as well as the differential
cross section crucially depended on fitting a Gauss curve through the recorded
spectra, so this is what we spent most of our time on. First of all, we found
it hard to fit a proper curve through the recorded spectrum at 8192 channels.
There was too much randomness involved in the distribution of the energies
into all the channels. So we transformed the histograms to contain only 2048
channels, which gave a much clearer picture. This is illustrated in figure
(A.1).

Spectrum at 90 degrees, 8192 Channels Spectrum at 90 degrees, 2048 Channels

100

30 90

80
25
70

20 60
Photon Count

Photon Count

50
15
40

10 30

20
5
10

0 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 50 100 150 200 250 300 350 400 450 500
Channels Channels

Figure A.1: Comparison between 8192 and 2048 channels on the same section
of a spectrum

For each spectrum, we subtracted the background measurement from the

spectrum with the target, which resulted in a very clear peak in most cases.
On this data we then fitted a Gauss curve using the least square method.
The raw data, the spectra without the background as well as the fits are
show in figures (A.3) to (A.12).
Special treatment was given to the measurement at 50◦ . Looking at figure
(A.5) it is obvious that the background spectrum looks almost exactly like
the spectrum with the target. We are not quite sure how this happened,
but it seems evident, that this can not be the true background measurement.
Using this background results in a peak that is much lower than predicted
by theory, and the energy at the peak shifted considerably as well. Instead
of discarding that datapoint outright, we tried to see what it looks like when
we take the average of the backgrounds at 40◦ and 60◦ . This plot is shown
below in figure (A.2).

17
 50 Degrees − Reconstructed Background 50 Degrees − Corrected for Reconstructed Background
100 100
With target Data
Reconstructed Background Gaussian fit
90 90

80 80 Peak channel: 596.2 +/− 1.4

Peak energy: 461.8 +/− 1.1 keV

70 70 Area: 5136 +/− 150

Count/s: 5.71 +/− 0.17

60 60 χ2/ndf: 1.5577
Photon Count

Photon Count
50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.2: Spectrum at 50 degrees with a reconstructed background

30 Degrees − Raw Data 30 Degrees − Corrected for Background

100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 740.6 +/− 0.9

Peak energy: 579 +/− 0.7 keV

70 70 Area: 6461 +/− 129

Count/s: 7.18 +/− 0.14

60 60 2
χ /ndf: 2.0378
Photon Count

Photon Count

50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.3: Spectrum at 30 degrees

18
 40 Degrees − Raw Data 40 Degrees − Corrected for Background
100 100
With target Data
Background Gaussian fit
90 90

Peak channel: 665.3 +/− 1.4

80 80
Peak energy: 517.9 +/− 1.2 keV

70 70 Area: 5523 +/− 162

Count/s: 6.14 +/− 0.18

60 60 χ2/ndf: 2.8383
Photon Count

Photon Count
50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.4: Spectrum at 40 degrees

50 Degrees − Raw Data 50 Degrees − Corrected for Background

100 100
With target Data
Background Gaussian fit
90 90

Peak channel: 584.4 +/− 1.8

80 80
Peak energy: 452.2 +/− 1.5 keV

70 70 Area: 2322 +/− 136

Count/s: 2.58 +/− 0.15

60 60 χ2/ndf: −3.6407
Photon Count

Photon Count

50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.5: Spectrum at 50 degrees

19
 60 Degrees − Raw Data 60 Degrees − Corrected for Background
100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 558.5 +/− 1.1

Peak energy: 431.2 +/− 0.9 keV

70 70 Area: 7829 +/− 132

Count/s: 8.7 +/− 0.15

60 60 χ2/ndf: 2.0934
Photon Count

Photon Count
50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.6: Spectrum at 60 degrees

70 Degrees − Raw Data 70 Degrees − Corrected for Background

100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 484 +/− 1

Peak energy: 370.8 +/− 0.8 keV

70 70 Area: 3586 +/− 99

Count/s: 3.98 +/− 0.11

60 60 χ2/ndf: 1.7364
Photon Count

Photon Count

50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.7: Spectrum at 70 degrees

20
 80 Degrees − Raw Data 80 Degrees − Corrected for Background
100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 426.1 +/− 0.8

Peak energy: 323.8 +/− 0.6 keV

70 70 Area: 4078 +/− 119

Count/s: 4.53 +/− 0.13

60 60 χ2/ndf: 2.9404
Photon Count

Photon Count
50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.8: Spectrum at 80 degrees

90 Degrees − Raw Data 90 Degrees − Corrected for Background

100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 384.7 +/− 0.9

Peak energy: 290.2 +/− 0.7 keV

70 70 Area: 4355 +/− 137

Count/s: 4.84 +/− 0.15

60 60 χ2/ndf: 4.1376
Photon Count

Photon Count

50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.9: Spectrum at 90 degrees

21
 100 Degrees − Raw Data 100 Degrees − Corrected for Background
100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 353.1 +/− 0.7

Peak energy: 264.6 +/− 0.6 keV

70 70 Area: 3507 +/− 113

Count/s: 3.9 +/− 0.13

60 60 χ2/ndf: 2.8765
Photon Count

Photon Count
50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.10: Spectrum at 100 degrees

110 Degrees − Raw Data 110 Degrees − Corrected for Background

100 100
With target Data
Background Gaussian fit
90 90

80 80 Peak channel: 324.9 +/− 0.6

Peak energy: 241.7 +/− 0.5 keV

70 70 Area: 3415 +/− 104

Count/s: 3.79 +/− 0.12

60 60 χ2/ndf: 2.4457
Photon Count

Photon Count

50 50

40 40

30 30

20 20

10 10

0 0

−10 −10
0 500 1000 1500 0 500 1000 1500
Channels Channels

Figure A.11: Spectrum at 110 degrees

22
 120 Degrees − Raw Data 120 Degrees − Corrected for Background
160 160
With target Data
Background Gaussian fit
140 140

Peak channel: 303.5 +/− 0.9

120 120 Peak energy: 224.3 +/− 0.7 keV

Area: 2795 +/− 165

Count/s: 3.11 +/− 0.18
100 100
2
χ /ndf: 5.2781
Photon Count

Photon Count

80 80

60 60

40 40

20 20

0 0

0 500 1000 1500 0 500 1000 1500

Channels Channels

Figure A.12: Spectrum at 120 degrees

23