Franck-Hertz Experiment

Arthur Sowinski (author), Paul Hyde (lab partner)

December 17th, 2010

In this experiment the discrete nature of atomic energy levels was studied. A beam of electrons was
emitted into a vacuum tube in order to excite the electrons of a mercury (Hg) vapour. Energy vs
current graphs were obtained and analyzed. The data collected supported the simple quantum model
of the atom. In addition, three values for the excitation energy of the mercury atom electrons were
found, 4.93 +/- 0.14 eV, 4.98 +/- 0.08 eV, and 5.02 +/- 0.08 eV, of which the first two agreed with the
literature value of 4.9 eV.
Introduction
The beginning of the 20th century in physics saw the introduction of the quantum theory of matter.
Rutherford's existing model of the atom proposed a positive nucleus, around which negative electrons
orbited. In 1913 Niels Bohr proposed a model of the hydrogen atom in which the electron can move in
certain orbits around the nucleus without losing energy, known as stationary states, or energy levels.
The electrons radiates energy (in the form of photons) only when it moves from one stationary state to
another. The frequency of the emitted photons is related to the change in difference in energy between
the two stationary states, h being Planck's constant, and Ef and Ei the energies of the final and initial
states, respectively [1] :
(1)

Because only certain energy levels are allowed, in order for the electron to move to a higher level (be
excited), it must absorb some energy which is equal or greater to the difference Ef – Ei.

In 1914, J. Franck and G. Hertz performed their experiment, which provided the first direct evidence of
discrete electron energy levels.
In this experiment, mercury is housed in a tube, and turns into a vapour when heated to 180 - 200 °C.
A heated cathode, and anode are placed at either end of the tube, and a voltage grid is present in
between them. The heating of the cathode causes electrons to be released from it, and they get
accelerated by the voltage grid. The kinetic energy of an emitted electron is given by its charge
multiplied by the grid voltage Vo [1],
(2)

therefore increasing the grid voltage increases the kinetic energy of the electrons. The electrons travel
through the Hg vapour, and may collide with an electron belonging to an Hg atom. Although mercury
has 80 electrons in the neutral state, rather than just one, they are still expected to have quantized
energy levels, and obey Eq. 1. Since the Hg electrons need to obtain some minimum value of energy in
order to change energy levels, an incident electron needs to have at least that same value of kinetic
energy in order to transfer the energy [1]:
(3)

If it doesn't, an elastic collision will take place, and the incident electron will keep its kinetic energy.
For low grid voltages, it is expected that the incoming electrons will not have sufficient kinetic energy
to excite the Hg electrons. Therefore, they are expected to keep their kinetic energy, and arrive at the
anode, creating a measurable current. However, as grid voltage is increased, the kinetic energy of new
emitted electrons increases, and eventually they have enough kinetic energy to excite an Hg electron
from state i to state f.

If this is the case, an inelastic collision will take place, the incident electron will lose kinetic energy
(Ef – Ei), and the Hg electron will move up an energy level. Energy eVo - (Ef – Ei) will remain with
the accelerated electron. If this is sufficient to overcome the potential at the anode, the electron will
flow through the anode and contribute to the anode current.
Therefore, if the current at the anode is plotted as a function of grid voltage, one would expect a
steadily rising curve that would abruptly go down, leaving a peak. The rising portion represents the
electrons gaining kinetic energy with increasing grid voltage, but not yet having enough to excite Hg
electrons. The anode current therefore rises along with the grid voltage. The peak would represent a
critical voltage at which the incoming electron has just enough kinetic energy to excite the Hg electron.
After the peak, there should be a drop in current, since the electrons give away some or all of their
energy, and not all can contribute to the anode current.

Of course this one-peak plot just represents transitions to the first Hg electron energy level. For the
next energy level, the same situation is expected to occur, and the same for all the energy levels.
Therefore a grid voltage vs anode current plot is expected to be made up consecutive of rise – peak –
fall plots, one for each energy level. Each peak is expected to correspond to the grid voltage, thus
electron kinetic energy required to excite an Hg electron to a higher state [1].

Therefore the distance between consecutive peaks corresponds to the critical potential for excitation of
Hg electrons (in Volts) or difference in energy between the Hg ground state and the excited state ( in
electron Volts), and is the main quantity to be determined in this experiment. The literature value for
this energy is 4.9 V [1].

Another secondary result can be obtained from this experiment, that being the work function of the
cathode material. The cathode consists of a filament which heats up due to the cathode voltage, and it
causes the coating material to emit electrons which are then used for the main experiment. From the
photoelectric effect[1],
(4)

From Eq.1, it is known that hf corresponds to the energy difference between the excited and ground
states for an electron. Rearranging ;
(5)

Φ is the work function of the cathode material, and corresponds to the minimum energy required to
eject an electron from its surface. Eq.5 therefore indicates the actual kinetic energy carried by the
electron, since as it is first accelerated by Vo, it already has acquired energy Φ which was required to
emit it from the cathode. Since the excitation energy Ef – Ei will be calculated in the experiment, the
work function Φ can be determined as well.

Experiment
The mercury in this experiment was contained in a vacuum tube (Franck-Hertz tube), housed in a
furnace (Vulcan, model no. 27587). The entire setup of anode, cathode, voltage grid, vacuum tube and
furnace were all packaged together as "Franck-Hertz apparatus." This was connected to a variac which
acted as the power supply to the furnace. In addition, a thermocouple was fed through an opening in
the furnace and placed in direct contact with the vacuum tube, in order to measure its temperature.
Three voltages sources (Agilent E3611A power supply) were used to supply voltages to the F-H tube
components, V1 ( cathode voltage), V2 (anode voltage), and the Vo (grid Voltage). A current
preamplifier (Stanford Research Systems Model SRS70 Low-Noise current preamplifier) was
connected across the anode terminals, and served to amplify the current created by the moving
electrons. It was set to a low-pass filter, and with a scaling of 1 μA/V.
In order to obtain the final output current and voltage, two large multimeters (Agilent 34401A digital
multimeter) were used. One was connected in series with current preamplifier, the other was connected
across the output of the grid voltage source.
The two multimeters were both connected to a computer via DAC cables. A data collecting program
called Rebecca was run on the computer in order to facilitate data collection. It would record a point of
data on a graph, with voltage on the x-axis and current on the y-axis at some preset time interval. Had
the program not been used data collection would have involved individually reading hundreds of
voltage and current data sets off of the multimeters, which would have taken a large amount of time.

For each day of data collection, the variac was turned on about an hour ahead of time, and set from
around 30 V to 50 V. This was done in order to give the F-H tube time to heat up to the required
temperature of around 180 - 200 °C, without risking overheating. To collect data, Rebecca was set to
record a (grid voltage, current) point 7 times a second. The knob controlling grid voltage was turned
manually, very slowly from 0 to 30 Volts. This was done for various combinations of cathode and
anode voltages V1 and V2, each one ranging from 0.5 to 2 Volts. After sufficient good quality plots
were made, the plot files were imported into Origin.
Figure 1: The experimental setup.
Results
Out of of the plots obtained, three were selected for analysis, they are displayed below. These were
chosen their well defined peaks, and densely and evenly distributed data points.

Figure 2 : Anode current for varying grid voltage with temperature 190 °C, V1 = 1.5 V, V2 = 2.0 V.

Figure 3 : Anode current for varying grid voltage with temperature 196 °C, V1 = 2.0 V, V2 = 1.5 V.
Figure 4 : Anode current for varying grid voltage with temperature 200 °C, V1 = 1.5 V, V2 = 2.0 V.

As can be seen, the plots chosen all exhibit the expected behaviour of steadily ascending peaks
predicted in the introduction, thus supporting the quantum nature of the mercury atom.

Each one featured five distinct peaks. For each plot, the grid voltage at each peak was determined. To
get values for the voltages at the peaks, the graphs were printed on paper, and the approximate
locations of the peaks on the voltage axis were measured using a ruler and performing appropriate
conversions. The uncertainty in Vo was estimated to be +/- 0.25 V based on the uncertainty in the ruler
measurements. By equation 2, the grid voltage in V can be expressed as the electron energy in eV.

Table 1: Electron energies at anode current peaks (from Fig. 1)

Peak Number Electron Energy ( +/- 0.25 eV)
1 6.1
2 10.8
3 15.9
4 20.1
5 26.1

Table 2: Electron energies at anode current peaks (from Fig. 2)

Peak Number Electron Energy ( +/- 0.25 eV)
1 5.6
2 10.5
3 15.1
4 20.3
5 25.6
Table 3: Electron energies at anode current peaks (from Fig. 3)
Peak Number Electron Energy ( +/- 0.25 eV)
1 6.1
2 10.6
3 16.1
4 20.8
5 26.1

In order to get a value for the energy required to excite an Hg electron, the electron kinetic energies
were plotted as a function of peak numbers. The Resulting plots are expected to be straight lines, with
the slope corresponding to the energy difference between the excited and ground states.

The Fit Linear function was used in origin to attempt to fit each set of data to a straight line, and the
slopes and intercepts calculated. The reduced chi-square, a statistic which indicates quality of fit, was
calculated for each fit manually, using [2]:

(6)

yi and σi are the energy values, and the associated uncertainty, respectively, and a + bxi is the fitted line
equation. υ is the number of degrees of freedom in the data, given by the number of data points minus
the number of constraints[2], for this experiment there are 5 points, 2 constraints (2 axes), therefore
υ = 3.

Figure 5: Electron kinetic energy as a function of peak number (from table 1),
the resulting slope of the line is 4.93 +/- 0.14, the intercept is 1.01 +/- 0.48, the reduced Chi-square is 3.36.
Figure 6: Electron kinetic energy as a function of peak number (from table 2)
the resulting slope of the line is 4.98 +/- 0.08, the intercept is 0.48 +/- 0.26, the reduced Chi-square is 0.98.

Figure 7: Electron kinetic energy as a function of peak number (from table 3),
the resulting slope of the line is 5.02 +/- 0.08, the intercept is 1.01 +/- 0.48, the reduced Chi-square is 1.56.

The measured excitation energies are therefore 4.93 +/- 0.14 eV, 4.98 +/- 0.08 eV, and 5.02 +/- 0.08 eV.
The first two agree with the expected value of 4.9 eV within their uncertainties, the third however does
not, although it is still a reasonable value. The corresponding reduced Chi-square values are 3.36, 0.98,
and 1.56 respectively, implying a relatively poor, excellent, and fairly good fit, respectively. The larger
discrepancy in the first value is due to the outlying point for peak #4 of the plot in Fig. 5 (as well as
peak #5, to a lesser extent). The fact that the obtained values agree with the expected one well can be
attributed to the high precision used in this experiment. Since the program Rebecca was used to
automate data collection, large amounts of data points were obtained, as seen in Figs.2 – 4, and they
were therefore densely spaced. For this reason it was easy to read off accurate locations of the voltage
peaks.
The linear fits done for the plots in Figs.5-7 give the electron kinetic energy, as a function of the
excitation potential (Ei – Ef), which is the slope, peak number, and a constant intercept. By comparing
this with Eq.5, it becomes clear that the intercept corresponds to Φ, the work function of the cathode
coating material. A seen in the Figs.5-7 captions, the intercepts obtained were 1.01 +/- 0.48 eV, 0.48
+/- 0.26 eV, and 1.01 +/- 0.48 eV. These values have uncertainties much too large too be considered
reliable data, and therefore they are inconclusive. The values are also inconsistent, as the first and the
last one are identical, yet the second one is only half that magnitude. The work function for the
cathode coating material therefore remains unknown. The large uncertainties in the intercept can be
attributed to the fact that the fitted lines don't actually pass through the vertical axis since there was no
peak of value 0, therefore the uncertainty goes up.

The three anode current as function of grid voltage plots used in the analysis (Figs.2-4) were chosen as
the best plots from a large amount of data. The values of V1 and V2 were either 1.5 V or 2.0 V, and the
temperature was at least 190 °C. Prior to this, values of V1 and V2 ranging from 0 V to 1.5 V were
were used, and temperatures in the 180 °C range were attempted as well. The data obtained at these
values was not very good, which is why higher values were attempted. At temperatures that were too
low, no data was obtained, due to no mercury vapour being present in the tube.

For low temperatures, in the 180s, a problem arose in which the (Vo,I) curves got cutoff near the end,
and the current suddenly dropped drastically. The curve would however continue albeit from a lower
point. Two examples follow:

Figure 8: Anode current for varying grid voltage with temperature 180 °C, V1 = 1.5 V, V2 = 2.0 V.
Figure 9: Anode current for varying grid voltage with temperature 184 °C, V1 = 1.5 V, V2 = 2.0 V.

For the plot in Fig.8, the current is seen to drop to zero, after which if fizzles out somewhat. In Fig.9,
however, the current actually drops to below zero, and continues to rise. A large current drop implies
that for some reason most electrons suddenly stopped flowing, or in Fig.9, a current in the opposite
direction started flowing through the anode. The low temperatures would simply suggest that less
mercury vapour, with slower moving atoms is present in the tube, and it is unknown why that would
cause such a discontinuity. Since the data collection at lower temperature was done earlier than that at
higher temperatures, it is possible that an equipment problem was responsible for the discontinuity, as
the equipment was still being troubleshooted the first few days.

For low values of V1 and V2, another problem arose, with a major discontinuity eliminating all current
values from some region. Two examples are shown:

Figure 10: Anode current for varying grid voltage with temperature 193 °C, V1 = 0.5 V, V2 = 1.0 V.
Figure 11: Anode current for varying grid voltage with temperature 193 °C, V1 = 0.5 V, V2 = 1.5 V.

This problem was only seen for cases where the cathode voltage V1 was less than the anode voltage
V2. It is still completely unknown why this may have occurred. Interestingly, it is noted that in both
cases the span of the discontinuity in μA corresponds to the value of V1 in volts.
Overall the systematic nature of these discontinuities seems to suggest that they are due to some
problem with the equipment, rather than some physical phenomenon not accounted for. The fact they
disappeared as more data was being collected seems to support that idea as well.

Systematic errors aside, the main reason for increasing the values V1 and V2 was to get more well
defined peaks in the Current Voltage plots, and thus better data. An example is given where the values
are low:

Figure 12: Anode current for varying grid voltage with temperature 193 °C, V1 = 1.0 V, V2 = 0.5 V.
As can be seen in Fig.12, there are no systematic discontinuities, however, the peaks aren't as well
defined as for those plots where V1 and V2 were 1.5 V or 2.0 V, especially the first one. A lower
cathode voltage implies that less electrons are emitted from the cathode, which is fine. However a
lower anode voltage means that electrons which have suffered an inelastic collision and given away
their energy must overcome less potential in order to arrive at the anode, and therefore a larger fraction
of the electrons contribute to the current. The current therefore doesn't go down as drastically
following a peak, hence the peaks aren't as well defined. With a higher anode voltage, less electrons
have the energy remaining required to arrive at the anode, meaning less current following a peak, and
well defined peaks, as seen in Fig.2-4.

One improvement which could be made to this experiment would be to fully automate the data
collection. To do this, not only would data points be recorded automatically, but the grid voltage would
be increased slowly, automatically, and continuously as well, rather than turning a knob manually. As
can be seen in Figs.2-4, in some regions the data points are further apart than in others. If the grid
voltage was increased automatically this could be eliminated, and all points would be closely
concentrated. Nevertheless, the points were well concentrated around the peaks, which is really all
that was required in this experiment. Automating the grid voltage change would either involve doing it
physically, by for example attaching an electric motor rotating at constant velocity to the knob, or
digitally, by writing a computer program to automate it. Both of these ideas were considered but
eventually rejected due to lack of time and knowledge.
Conclusion
In this experiment the energy difference between the excited and ground states for electrons in a
mercury atom was measured using a mercury vapour in a vacuum tube and electron beam. The results
obtained graphically qualitatively demonstrated support for a quantum theory of the mercury atom. In
addition, the energy difference between the excited and ground states was calculated on three different
occasions to be 4.93 +/- 0.14 eV, 4.98 +/- 0.08 eV, and 5.02 +/- 0.08 eV, of which the first two values
agreed with the literature value of 4.9 eV. An attempt to measure the work function of the cathode
material was made but was inconclusive as the data required was inconsistent and contained very large
uncertainties.

Acknowledgements

For incorporating equations in to the report, an online LATEX equation editor was used,
http://rogercortesi.com/eqn/index.php.

OriginPro 8 was used for data plotting and analysis.

Special thanks to Gilles Roy for providing the computer program Rebecca, providing and helping to set
up much of the equipment, and helping with troubleshooting.

References

1. Tipler, Paul Allen, and Ralph A. Llewellyn. Modern Physics. New York, NY: W.H. Freeman, 2008.
Print
.
2. Bevington, Philip R. Data Reduction and Error Analysis for the Physical Sciences. 3rd ed. New
York, NY: Mc-Graw Hill, 2003. Print.