Experiment T12

Detection principles
Advanced lab course for bachelor students in physics

Leon Mans
Vincenzo Sanfratello

RWTH Aachen University

October 25, 2022
Contents

1 Theory 2
1.1 β ± -decay and its spectrum . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Fermi correction & Kurie diagram . . . . . . . . . . . . . . . 3
1.2 Charged particles in magnetic fields . . . . . . . . . . . . . . . . . . 3
1.3 Energy loss of electrons in matter . . . . . . . . . . . . . . . . . . . 3
1.3.1 Ionization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Bremsstrahlung . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Momentum resolution 5
2.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.1 Magnetic field properties . . . . . . . . . . . . . . . . . . . . 6
2.3 Momentum spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.1 Kurie Diagram . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3 Energy loss in matter 12

3.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.1 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Kurie diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 Multiple Scattering 18
4.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2 Experimental procedure . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3 Analysis for different materials . . . . . . . . . . . . . . . . . . . . . 19
4.4 Analysis for different positions . . . . . . . . . . . . . . . . . . . . . 24
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1
Chapter 1

Theory

In this chapter a brief overview of the main ideas and equation is given.

1.1 β ±-decay and its spectrum

The β decay works via the weak interaction. It is observed in nuclei with great
imbalances in the number of neutrons in relation the number of neutrons. The
two possibilities are

n −→ p + e− + v̄
p −→ n + e+ + v
The β spectrum is continuous and has a maximal energy boundary. By Fermi’s
Golden Rule the transition probability is described as

2π   dN
2
w= ⟨Ψf |Ĥs |Ψi ⟩

h̄ dE0
By approximating the wave function as a plane wave and by further approximating
the wave function as constant over the nucleus’ region we can write
 2
⟨Ψ |Ĥ |Ψ ⟩
  g 2 |Mf i |2
f s i  =
V2


where g parametrizes the strength of the weak interaction and |Mf i | is the matrix
element for a given transition. Taking into account that the total energy is dis-
tributed between the positron (electron) and neutrino (anti-neutrino) the following
formula can be derived

dw q
= K|Mf i |2 (Emax − Ee )2 (Te + me ) Te2 + 2Te me
dTe
where Te = E − me

2
1.1.1 Fermi correction & Kurie diagram
It is necessary to take the Coulomb interaction between the nucleons in the atom
and the decayed β-particle into account. The following equation can be derived

2πη ZαEe
F (Z, η) = , η = ±
1 − e−2πη pe
where + (-) is for electrons (positrons). Finally, we can write

dw
= K|Mf i |2 F (Z, η)p2e (Emax − Ee )2 (1.1)
dpe
for the momentum distribution.
r
dw/dpe
By plotting F (Z, pe )p2e
versus the electron’s energy we obtain the so called Kurie
plot. By assuming that |Mf i | is constant, which is assumed to be so for allowed
transitions, the Kurie diagram should show a linear decrease with increasing ener-
gies. By fitting the data and extrapolating the root it is possible to compute the
maximum energy for that specific decay. See section 2.3.1 for an example of such
a plot.

1.2 Charged particles in magnetic fields

The Lorentz equation tells us that a charged particle with charge ”q” will be
accelerated by an electrical and magnetic field as

F⃗ = q(E
⃗ + ⃗v × B)
⃗

Therefore a charged particle will be forced into a circular motion by the magnetic
field and be accelerated along the electrical field. Thus in total we expect a helix-
like motion. By assuming the velocity vector to be perpendicular to the magnetic
⃗ = 0 and by equating the Lorentz force with the centripetal force we obtain
field, E

v2
⇒m = qvB
R
⇔ mv = p = qBR
And by using natural units
B R
p = 0.3
mT mm

1.3 Energy loss of electrons in matter

When electrons travel through matter, different interactions lead to them loosing
part or all of their energy. Traversing electrons can interact with shell electrons of
atoms or ionize them fully. Another phenomenon is the emission of Bremsstrahlung
when the electrons interact with the nucleus of an atom via the Coulomb force.

3
1.3.1 Ionization
By evaluating Z ∞
∆p = F⃗C dt
−∞
where F⃗C denotes the Coulomb force, the energy loss per unit length can be derived
via
(∆p)2
∆E = ne dV, ne : electron density
2me
as
2πNA r02 me c2 Zρ τ 2 (τ + 2 τ 2 /8 − (2τ + 1)ln2
! ! !
dE
= ln + + (1 − β 2 ) − δ ,
dx e
β2 A 2(I/me c2 )2 (τ + 1) 2

cm2
" #
dE
= M eV
dx g
2
where r0 = 4πε0eme c2 , τ = γ − 1 (γ from special relativity), ρ is the density of the
traversed matter and δ is a parameter correcting for the polarisation inside the
traversed matter caused by the electron. I is the mean ionization energy for the
traversed atoms.

1.3.2 Bremsstrahlung
Electrons are accelerated by the atoms nuclei via the Couloumb interaction. The
energy loss per unit length is given as

dE Z2 2 183 E
! !
= 4αNA re ln E :=
dx A Z 1/3 X0
g
X0 is the radiation length; [X0 ] = cm2

4
Chapter 2

Momentum resolution

The moment spectrum of the β − decay of Sr90 is measured.

2.1 Setup
The source is placed into a holder in such direction that the decayed electrons
will travel through a magnetic field created by a two coil setup as shown in Fig.
2.1. The strength of the magnetic field can be varied by adjusting the current
feeding it. A hall magnetic sensor is placed inside the magnetic field as far away
as possible of the electron’s path in order to measure its value. A GM-counter
with a horizontal opening parallel to the ground is put in place to detect incoming
electrons, whose trajectory is been bent by the magnetic force.

Figure 2.1: Experimental setup for the momentum spectrum measurement.

5
2.2 Experimental procedure
The idea is simple; the magnetic field is varied in small steps by adjusting the
current and the detected β − particles are counted by the GM-detector and saved
to a text file. The hall sensor provides the value of the magnetic field.

2.2.1 Magnetic field properties

A source of error could be the magnetic field not being homogeneous in space or
time. To ensure that this is not the case, a measurement of the magnetic field is
performed on a total of 47 points following a grid on the metal surface of one the
coils (In our case we used the right side in Fig. 2.1). A plot with a colormap is
presented in Fig. 2.2. We conclude that the coil setup provides the experiment
with a sufficiently homogeneous field.

Figure 2.2: Colormap of magnetic field. Measured at given grid points.

6
 Two measurements, one at 0.5 A and one at 1 A, are performed over time to
visualize the possible time dependence of the magnetic field. As shown in Fig. 2.3
we also conclude that the time dependence will not be an issue with the setup
given since the magnetic field stayed perfectly constant in both measurements (up
to sensor resolution).

Figure 2.3: Time dependence of magnetic field for given current values.

7
2.3 Momentum spectrum
The measurement is presented in Fig. 2.4. The uncertainty on each data point
follows a Poisson distribution. There exists an uncertainty on the x-Axis be-
cause of the finite resolution of the hall sensor. The sensor used has two ranges.
−2 0
(0, 200) mT with a resolution
√ up to 10 and (0.2, 2) T with a resolution of 10 .
By using σB = Interval/ 11 the uncertainty on the x-Axis turns out too small
to be shown in a plot.

Figure 2.4: Momentum spectrum in dependence of the magnetic field. Gaussian

fit shows agreement.

By converting the magnetic field values to momentum values it also possible

to plot the spectrum in dependence of the particle’s momentum. In Fig. 2.5 the
momentum distribution with and without the Fermi correction is displayed.

Fig. 2.6 shows the momentum as a function of the magnetic field. As expected,
the momentum is proportional to the magnetic field’s strength.

8
Figure 2.5: Momentum spectrum in dependence of the energy. Left: Without
Fermi correction. Right: with Fermi correction.

Figure 2.6: Momentum in dependence of the magnetic field. Linear relation is

visible.

9
2.3.1 Kurie Diagram
A Kurie plot is created as shown in Fig. 2.7 (no Fermi correction). A linear fit is
performed on a specifically chosen interval. By calculating the intersection of the
linear function with the x-Axis the maximum energy of the β − decay of Sr90 can
be estimated (See Fig. 2.8

Figure 2.7: Kurie diagram of Sr90.

Figure 2.8: Kurie diagram of Sr90 and linear fit for estimation of Emax .

The uncertainty on the Kurie diagram is calculated by Gaussian error prop-

agation. This uncertainty is then fed into a python script that calculates the
parameters ”a” and ”b” of the linear fit y = aE + b. The script also provides an
uncertainty on the parameters. The intersection is computed by
−b
0 = aEmax + b ⇔ Emax =
a
10
 v
u  !2
u σb 2 −bσa
σEmax = t
+
a a2
The final value obtained by this method is Emax = 2285.6 ± 51.1 keV when con-
sidering the Fermi correction and Emax = 2280.9 ± 48.4 keV without it. It is also
possible to calculate the mass of the β-from the maximum energy and maximum
momentum that were measured.
q
m = E 2 − p2 (2.1)

The maximum momentum can be obtained from Figure 2.5 where the count
rate becomes constant, which means that no particles with this momentum can be
found, only the measurement background. The maximum momentum is pmax =
2220 ± 40 keV , which means that m = 543.5 keV . The error on this value is
calculated to be

1q
σm = (EσE )2 + (pσp )2 = 270.0 keV (2.2)
m
This error is quite large due to the fact that the mass is calculated from the
difference of two large values that have errors themselves.

2.4 Conclusion
Momentum spectrum

The momentum spectrum measurement is in agreement with the expectations.

The Fermi correction shifts the lower momentum values up and the higher mo-
mentum values down. This is as expected for β − particles (electrons) since we
expect the nucleus to decelerate electrons on their way out, thus more states of
lower momentum should exist after the Fermi correction than before.

Kurie diagram

We observe a rapid decrease for the very first values, followed by a maximum
and a linear decrease thereafter in both Kurie diagrams. Towards the end the
data flattens out. A linear fit is performed within a manually chosen interval. We
notice the rather poor value of our χ2 -test. Besides an incorrect estimation of the
uncertainty of each term in Eq. 2.3.1, a source of error is the chosen interval used
to perform the fit. Since the measurement does not follow the ideal linear function,
we are forced to use our best guess on the starting and ending point of the interval.
This has resulted in the uncertainties on both the Kurie diagram and the fit to be
too small. Still, the (Fermi corrected) value obtained Emax = 2285.55 ± 51.08 keV
has a relative error of ∼ 1.3% in relation to the reference value of 2200 keV , even
though it is not within the uncertainty interval.

We conclude that the observed β − -decay of Sr90 is a allowed transition. This

comes from the fact that the observed Kurie diagram displays a linear decreasing
behaviour as is expected for a constant matrix element |Mf i | in Eq. 2.3.1.

11
Chapter 3

Energy loss in matter

The energy spectrum of different radioactive sources are measured behind different
materials.

3.1 Setup
The source is placed in the source holder and the shielding material is placed in
the material holder. On the other side is a scintillator in front of a photo cathode
and a photo multiplier which is connected to a PC that saves the data. Each
measurement takes place over a timespan of 60s.

Figure 3.1: Experimental setup for the energy spectrum measurement.

3.2 Experimental procedure

3.2.1 Calibration
First of all, different radioactive materials with different energy peaks are placed
in the source holder and their spectrum is measured. The signal isn’t analysed in
terms of energy, but in terms of channels that correspond to an energy and count
to events that are equal to that energy. With the different spectra, the energies
that are equivalent to the channels are measures.

12
 Figure 3.2: Experimental setup for the energy spectrum measurement.

After these peaks have been identified, their energy value is divided by the
channel number at which the peak was found. This gives 9 values for the en-
ergy distance between two channels. The mean of these 9 values is taken and
Estep = (2.570 ± 0.022)keV . This scale can be used to plot the energy spectrum
of the Strontium-90 source for different thicknesses of aluminum in front of the
scintillator.

13
Figure 3.3: The energy spectrum of the Sr90 for different aluminum thicknesses.

The aluminum layer leads to a reduction in peak height for all thicknesses, with
a thicker layer leading to smaller peaks. It can be seen that for higher energies,
the thinner layers produces more events and the thicker layers less events. Now
the maximum energies that are reached are examined and plotted against the
corresponding thickness in Figure 3.4. The error is assumed to be 40 keV for all
maximum energies. Now it is assumed that the differences between the maximum
energies represent the energy loss due to the aluminum plates.

Figure 3.4: The maximum energy in relation to Al-thickness.

14
 As can be seen from the χ2 /ndeg value which is close to 1, the maximum energy
decreases linearly with more plates of aluminium. The slope of the curve is then
the energy loss of the Sr90 electrons in aluminum per mm. The slope of the curve
is − dE
dx
= −394.5 ± 23.5 keV
mm
.
The literature value for energy loss in aluminium is 1.518 Mcm eV g
2 and with the
g M eV g keV
density of aluminum of ρ = 2.7 cm3 it is 1.518 cm2 ρ = 410 mm . That means that
the measured value and the literature value are compatible, as the literature value
lies in the error range of the measured value.
The maximum energy of the case without any aluminum shieldung is also shifted
from the literature value of 2285.6 keV due to the influence of the aluminum oxide
cover over the scintillator. It is 1.6 mm thick and has a density of ρao = 3.94 cmg 3 .
Assuming the same energy loss value as aluminum, the energy loss per millimeter
value turns out to be 1.518 Mcm eV g keV
2 ρao = 577 mm . Which means that the electrons

lose an additional 923.2 keV in the material in front of the scintillator, which
would put the truth maximum energy for the case without shiedling to 2703.2
keV , which is greater than the expected value. This could either mean a wrong
nominal value or a wrong measurement.

3.3 Kurie diagrams

We plot the Kurie diagram using the kinetic energy as its measured in the scintil-
lator setup for Sr90 in Fig. 3.5

Figure 3.5: Kurie diagrams for Sr90 for different aluminum absorber thicknesses

15
 Similarly to section 2.3.1 a linear fit is performed on the Kurie diagrams and
the maximum energy is extrapolated. Fig. 3.7 show the fit for each thickness used.
Table 3.1 lists the results of the maximum energy for each fit.

Figure 3.6: Kurie plot for no absorber energy loss setup for Sr90 .

Kurie diagram results

Absorber thickness [mm] Emax [keV] σ [keV]
0 1947.7 28.4
1 1897.7 74.9
2 2042.8 137.8
3 2114.9 152.3
4 2107.5 150.8

Table 3.1: Kurie diagram results. Sr90 for given absorber thickness.

16
Figure 3.7: Kurie plots for 1, 2, 3 and 4mm (from top to bottom) aluminum
absorber thickness for Sr90 and fit is shown.

17
Chapter 4

Multiple Scattering

The angular scattering distribution of the β-particles of a Strontium-90 source is

measured after passing through different materials.

4.1 Setup
A Sr90-source is placed in the source holder and the scattering material is placed in
the position between the source and the GM-counters. The particles are detected
by four Geiger-Müller counters which are placed on a moveable angular scale so
that their relative angle to the position straight from the source, which is defined
as 0◦ , can be changed. A Rasberry Pi records the data from the GM-counters.

Figure 4.1: Experimental setup for the multiple scattering measurement.

4.2 Experimental procedure

The black flag which signifies the middle of the GM-counters is placed at 60◦ and
the scattering material is placed in the material holder. Then a measurement with
the duration τ is taken where the GM-counters record the number of times they’ve
been hit by the β-particles of the source. The black flag is then moved to an angle
which is 5◦ smaller than the previous angle. This is the angular thickness of one

18
GM-counter and a smaller step size wouldn’t provide more accurate results. Then
another measurement is started until the angle −60◦ is reached and the angular
spectrum is fully covered.

4.3 Analysis for different materials

Initially, the rates of the different GM-counters are plotted against the black flag
angle at which they were measured and normal distributions are fitted to each
data set of each GM-counter with the following formula:
2 !
A 1 x−µ

Gauss(A, µ, σ, x) = √ exp − (4.1)
2πσ 2 σ

Figure 4.2: The rates of the GM-counters with regard to the black flag angle.

Now each normal distribution is shifted to the center of the plot, for each angle
value α for each data point, the following transformation is applied:

αnew = α − µGMi (4.2)

with µGMi being the mean of the gaussian fit. To this readjusted set of data
points, a new normal distribution is fitted. The measurement time τ was 20s for
each angle both measurements.

Figure 4.3: The rates of the GM-counters with readjusted angles.

19
 The same can be done for the other absorbers.

Figure 4.4: The rates of the GM-counters with readjusted angles.

With both Figure 4.3 and Figure 4.4, it can be seen from the χ2 /ndeg value
close to 1 that the gauss functions fit the data. In the case of 3mm and 4mm
it has to be noted that the distributions of the different GM-counters don’t fit
to a normal distribution anymore, which is deductable because the different GM-
counter distributions aren’t centered on top of each other as it is the case with
Figure 4.3. As an alternative, the displacement of the GM-counters themselves
with regard to the black flag are used as an alternative to the new unreliable
µGMi . With αGM1 = 25◦ , αGM2 = 8◦ , αGM3 = −10◦ and αGM4 = −27◦ , the
following plots are created by fitting a normal distribution to the shifted values.
The measurement time τ was 30s for each angle both measurements.

Figure 4.5: The rates of the GM-counters with readjusted angles according to the
position of the GM-counters.

20
Figure 4.6: The rates of the GM-counters with readjusted angles according to the
position of the GM-counters.

For the fits in Figure 4.5 and Figure 4.6, even though the χ2 /ndeg values still
indicate a fit that fits the data, especially the fit in Figure 4.6 resembles a constant
line more than a Gaussian. At the ends of the distributions, the measured rate
becomes higher than in the middle, indicating that the particles are scattered away
from the center for thicker materials.

Now the different norm factors A and the standard deviations σ can be plotted
with regard to the thickness of the aluminium.

Figure 4.7: The norm factor A for different thicknesses.

As can be observed in Figure 4.7, the normalisation factor, which is larger

if more particles are measured on the other side of the aluminium, tends to get
smaller with thickness, as would be expected. Only the last value is larger than
the previous one, which could be because Figure 4.6 doesn’t really show a gauss
distribution but a straight line, so the value has a high uncertainty,

21
 Figure 4.8: The standard deviation σ for different thicknesses.

Figure 4.8 shows that the σ gets bigger with thickness, as would be expected
because the particles have to pass through more material which increases their
scattering angle. The last value has a high uncertainty again, for the same reason
as the last value in the A-plot.
Now the same gauss fit can be done for the shifted GM-counter distributions
for different materials. For each of them, τ was 10s except for the steel, for which
it was 60s to get enough hits. Also, the data points were shifted with the mean of
the gauss distributions of each fit for every GM-counter except for the steel shift,
which was again done by shifting by the positions of the GM-counters themselves,
just like Figure 4.5 and Figure 4.6, as the singular distributions didn’t fit to a
gauss distribution.

Figure 4.9: Angular distribution for plastic.

22
 Figure 4.10: Angular distribution for paper.

Figure 4.11: Angular distribution for plexiglas.

Figure 4.12: Angular distribution for steel.

As expected, the measured rate is highest for the least dense material paper,
which can be seen in Figure 4.10. Every distribution can again be described by a
normal distribution, as the χ2 /ndeg values are close to 1. But again for the densest
material steel, the gauss distribution approaches a straight line.

23
4.4 Analysis for different positions
All of the previous measurements were done in position A. Now this position can
also be compared to position B, a different distribution can be expected.

Figure 4.13: The different positions for the Sr90 source.

Figure 4.14: Angular distribution for different positions.

It can be observed that the data in Figure 4.15 behaves in the following way:
The distribution for the source at position B closer to the GM-counters looks
sharper, with a higher peak and a smaller count rate for larger angles in comparison
with the data for the source in position A which is farther away. One would expect
a smaller standard deviation, but just about the same mean for position B.

24
 Figure 4.15: Angular distribution for different positions with gauss fit.

This hypothesis is tested my fitting normal distributions to the data. Both

have a a slightly larger χ2 /ndeg than 2, but overall, the gaussian distribution fits
to the data if it is plottet between −15◦ and 15◦ , as the other values represent
the background of the measurement. For these fits, σA = 7.55 ± 0.78 and σB =
6.64 ± 0.62. As expected, the σ for position A is larger than the σ for position B.
The means are µA = 0.80 ± 0.81 and µB = 0.26 ± 0.80, so they are compatible
with each other as expected.

4.5 Conclusion
It can be concluded that a thicker and denser material leads to a gauss distribution
with a higher standard deviation and a smaller normalisation factor, as fewer
particles reach the GM-counters. At a certain thickness or density, the normal
distribution approaches a constant line. If the source is farther away from the
GM-counters, that would be equivalent to a thicker layer of air being in between
the source and the counters, which means that the standard deviation also grows
larger in this case.

25