You are on page 1of 7

Rutherford Scattering

Evan Rule
May 8, 2013
Abstract
In order to investigate the nature of the atomic nucleus, we carry out a series of scattering
experiments to measure the deection of -particles passing through a thin metal foil. By
observing phenomena such as back-scattering, we can draw qualitative conclusions about the
structure of the atom, and by measuring the angular dependence of the scattering rate we can
make a determination of the atomic number of aluminum, which we nd to be Z
Al
= 13.
1 Introduction
Determining the structure of the atom was one of the primary goals of early 20th century physics.
Prior to 1911, the accepted model of the atom was the so-called plum pudding model, which
hypothesized that the atom consists of negative charges distributed throughout a positively charged
sphere. In 1909, Hans Geiger and Ernest Marsden, under the direction of Ernest Rutherford,
carried out a scattering experiment to probe the nature of the atom. The results of this experiment
contradicted the predictions of the plum pudding model and eventually led to the development of
the Bohr model of the atom.
Rutherford scattering occurs when a beam of -particles is directed normally onto a sheet of
thin gold foil in a vacuum chamber. The -particles interact with the atoms in the foil via the
Coulomb force and scatter elastically. While many -particles pass directly through the foil, the
interesting result occurs when particles are deected at large angles (in some cases > 90

). These
large deections suggest that the atom consists of a small, dense, positively charge nucleus, not a
large positively charged sphere as described by the plum pudding model.
2 Experiment and Data
In order to probe the nature of the atom, we employ a modernized version of the apparatus used
by Geiger and Marsden. By measuring the scattering rate N at dierent scattering angles , we
can make a determination of the angular dependency of the counting rate. Based on the scattering
theory developed from the results of Rutherfords experiments, we anticipate the relation
N() = N
0
c
f
d
f
d
d
, (1)
where N
0
is the incident -particle rate, c
f
is the atomic concentration of the foil, d
f
is the thickness
of the foil, is the solid angle into which the particles are scattered, and d/d is the dierential
cross section dened as
d
d
=

Z
1
Z
2
e
2
8
0
E

2
1
sin
4
(/2)
. (2)
1
Figure 1: An overview of the Rutherford Scattering experiment. (1) is the vacuum scattering chamber, (2)
is the discriminator preamplier, (3) is the digital counter. (Image taken from Leybold Didactic)
Here, Z
1
and Z
2
are the atomic numbers of the foil particles and -particles, respectively, and
E

= 5.5 MeV is the kinetic energy of the -particles.


2.1 Apparatus
The apparatus consists of a scattering chamber which can be evacuated via the vacuum pump to
< 100 Pa. Inside the chamber, the foil sample is secured in the swivel arm. A collimator slit
ensures that the -particles are normally incident on the gold foil. Knobs on top of the chamber
allow us to adjust the angle of the swivel platform, and thus the scattering angle. The source of the
-particles in our experiment is Americium-241. The chamber is also equipped with an -particle
detector, which is connected rst to the discriminator preamplier and then to the digital counter.
The digital counter and attached computer allow us to record the number of -particle detections
recorded in a given time interval. The thickness of both the gold and aluminum foil are measured
using a micrometer. To avoid damaging the foil samples, it is necessary to be extremely careful
when handling them and when evacuating the chamber.
2.2 Procedure
In preparation for the experiment, we insert the americium-241 sample and the gold foil into
the scattering chamber. For angles less than 30

, we make use of the 1mm collimator slit. After


moving the swivel holder to the edge of the chamber and aligning the slit diaphragm of the detector
vertically, we then close and evacuate the chamber using the vacuum pump. Beginning with the
swivel angle set to = 0

, we record the number of counts detected in a designated time interval.


For angles in the range 10

, we measure the number of counts within a 10 second interval for


40 consecutive intervals. For angles 15

and 20

, we measure the number of counts within a 60


second interval for 6 consecutive intervals. This ensures that we are taking data for roughly the
same amount of time as at the smaller angles. For angles 25

and 30

, we use a stopwatch to record


the amount of time needed to accumulate 5 counts. We do so because the counting rate is very low
for these large angles. We then set the swivel platform to = 5

and repeat the above procedure


until we have reached 30

.
After data collection for gold is complete, we carefully depressurize the chamber, remove the
gold foil and replace it with the aluminum foil sample. We then evacuate the chamber and set the
angle to 10

. We then record the number of -particle detections in a 60 second interval for 30


consecutive intervals. The rst 15 of these trials are conducted with the chamber uncovered and
exposed to the ambient lighting of the room. The last 15 trials are conducted with the chamber
2
Figure 2: A plot of count rate vs. scattering angle for gold foil. The red points show data taken in the
range 30

and the corresponding errors. Here, we assume that the distribution of measurements for each
angle is roughly Gaussian. The black line shows the best-t function (see Eq. (6)).
covered by an opaque black cloth. These results will allow us to check whether the excess lighting
can aect our results.
In order to observe back-scattering (i.e. scattering at large angles), we replace the 1-mm col-
limator with the 5-mm collimator. We then set the angle to = 120

and congure the digital


counter to record the total number of counts registered. Because the background is essentially zero
at large angles, we can be sure that any counts registered at such large angles are indeed genuine
detections.
Data and Error Estimation When our data collection is complete, we have measurements of
the number of counts, n() per given time interval, t, for various scattering angles. For each angle,
we can then dene the counting rate
N() =
n()
t
counts/s, (3)
where n() is the number of counts recorded in a time interval of length t seconds at scattering
angle . Fig. 2 shows our results, represented as a plot of log(N()). Note: due to the nature of
the angle grid-lines on the chamber, the uncertainty in our measurement of the scattering angle is

=2.5

. We can then use this distribution to compare our results to the theoretically determined
relation
N()
1
sin
4
(/2)
. (4)
3
Figure 3: This plot shows the number of counts recorded in a 60 second interval for 2,000 consecutive time
intervals. For these trials, the angle was set to 0

and there was no foil present in the chamber. By tting


this data to a linear function, we see that the vacuum in the scattering chamber shows no appreciable time
dependence on time scales less than 16 hours. The black solid line shows the best-t linear function.
Figure 4: This plot shows background data taken in the range 7.5

(red line) and the resulting best-t


Gaussian function (black line). Analysis reveals that the background is not centered at zero, but is biased
slightly towards positive angles. By nding the functional form of the background, we are able to account
for detector noise when we t our actual data.
4
3 Analysis
In order to verify that the vacuum inside of the scattering chamber remains constant with time, we
record the number of counts per 60 second interval for 2000 consecutive time intervals. The results
of this measurement are shown in Fig. 3. We then t this data to a linear function of the form
f(x) = ax + b. We nd the best t value for the parameter a 10
4
, which indicates that the
vacuum in our scattering chamber is constant to approximately 1 count per 60,000 seconds. Since
each session of data collection is completed in less than 16 hours ( 60,000 seconds), we conclude
that our data has no systematic time dependence.
Having established the time independence of our data, we make the assumption that our mea-
surements from repeated trials at a given angle follow a distribution that is roughly Gaussian, and
as a result, we can justify averaging over all trials for a given angle. Then, we t our background
data to a Gaussian function of the form
B() = Ae
(C)
2
/D
2
, (5)
from which we nd that the background is centered at C 0.34

. Note that this oset agrees with


the data in Fig. 2, which shows a small but denitive bias towards positive angles. Analysis of our
background t yields
2
= 291. Further details about the background are shown in Fig. 4. Once
we have determined the functional form of the background, we can then t our data to a function
of the form
N() = c
1
B() +
c
2
sin
4
(/2 C)
, (6)
where B() is the background function and C is the center of the background Gaussian. We employ
a weighted tting algorithm, where for each angle, the weight is given by the standard deviation of
all trials conducted at that angle. We obtain the best t value of
c
2
= (3.3 0.7) 10
8
s
1
. (7)
Analysis of this t yields
2
=372.
Using Eq. (1) and Eq. (2), we can check that our best t parameter is indeed reasonable.
Combining these two equations, we see that the theoretically predicted value for c
2
is given by
c
2
= N
0
c
f
d
f

Z
1
Z
2
e
2
8
0
E

2
. (8)
From the literature, we adopt c
f
= 5.9 10
22
cm
3
, Z
1
= 79, Z
2
= 2, E

= 5.5 MeV. We estimate


the value of N
0
as the amplitude of the Gaussian function which we used to model our background.
This assumption yields N
0
= 43.9 counts/s. Using the micrometer, we measure the thickness of
the gold foil to be d
f
= 0.006 0.003 mm. Thus, the only free parameter in Eq. (8) is the solid
angle . We can check our result for c
2
by using our best-t value to estimate the area of the
detector. If we assume that the distance from the gold foil to the detector is r 10 cm, then we
can use the relation
=
a
2
r
2
, (9)
to estimate the value of a, the radius of the detectors surface. Given the parameters as described
above, we determine the radius of the detector to be a 0.1 mm. While this value appears to be
smaller than expected, it is not an unreasonable size for the detector, though we can likely conclude
that our determined value of c
2
= 3.3 10
8
s
1
is less than the actual value of this constant.
5
4 Discussion
In general, a large
2
value is indicative of a poor t, which is apparent from Fig. 2, though the
logarithmic scale does exaggerate the residuals. From this plot, we see that our model underesti-
mates the rate of counts at every angle. This is likely an indication that in our eorts to account for
background noise, we have removed actual data points at angles 10

. By choosing to model the


background as a Gaussian function, we introduce a systematic error. Specically, the background
should be truncated beyond a certain angle. However, the Gaussian function extends, in theory, to
, the end result being that we account for background noise where there theoretically should
be none, yielding a suboptimal t for our data. It might be possible to correct for this eect by
modifying the Gaussian function with an appropriate window function, though this ltering would
likely lead to signicant complications when tting the data.
We can make a further check on our results by using Eq. (1) and (2) to obtain the relation
N
Au
N
Al
=
Z
2
Au
d
Au
Z
2
Al
d
Al
. (10)
Using the micrometer, we measure d
Al
= 0.010 0.003 mm. We consider = 10

as our ducial
trial, which yields
Z
Al
= 13.0 0.6, (11)
from which we determine that the most likely value for the atomic number of Aluminum is Z
Al
=
13. Here, the error was obtained by adding the relevant errors in quadrature. Indeed, the atomic
number of Aluminum is 13. Given our accuracy in determining this value, we can take this as
further evidence that our poor t in Fig. 2 is the result of some systematic error introduced by
our tting algorithm. At the very least, we can conclude that any systematic errors present in our
apparatus have an equal eect on both gold and aluminum scattering.
A nal systematic error which we investigate is whether or not shielding the scattering chamber
from sources of light has any eect on our measurements. To do so, we examine the results of
our measurements for aluminum, half of which were taken with the chamber covered and half with
the chamber exposed to the well-lit room. We nd that the average number of counts recorded
within a 60 second interval agree within error between the two sets of trials. However, the standard
deviation of the uncovered trials is nearly twice as large as the standard deviation of the covered
trials. This could be evidence that covering the apparatus improves the reliability of the detector,
but more data is needed to draw conclusions with any certainty.
Lastly, we recount that we observed back-scattering of -particles at angle 120

. Scattering
at such a large angle is evidence that the atom consists of a small, very dense nucleus which is
positively charged. This picture of the atom runs contradictory to Thompsons plum-pudding
model. Therefore, based on the results of the Rutherford scattering experiment, we can reject the
plum-pudding model of the atom in favor of a model much closer to that developed by Niels Bohr.
5 Conclusion
Based on our quantitative results, we can conclude that the counting rate varies with the scattering
angle according to
N()
1
sin
4
(/2)
. (12)
We determine the constant of proportionality for this relationship to be c
2
= (3.3 0.7) 10
8
s
1
, which agrees well with the theory of Rutherford scattering. We also make an experimental
6
determination of the atomic number of aluminum and nd that Z
Al
= 13. From our quantitative
results, we note that our best-t function for the angular dependence was less than ideal and that
the source of this error is likely the model which we employed to account for background noise.
With regards to our qualitative result, the observation of scattering at angles larger than 90

is an
indication that plum pudding model of the atom is incorrect. Alternatively, we propose a model in
which the atom consists of a small, dense, positively charged nucleus.
References
[1] Leybold Didactic, Rutherford Scattering Instruction Sheet, 1990.
7