Contents

1 Theory 1
1.1 Emission and absorption of light . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Spectral lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Rubidium atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Preparation 3
2.1 Fineness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 frequency calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Absorption spectroscopy 6
3.1 experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 absorption spectroscopy results . . . . . . . . . . . . . . . . . . . . . . . . . 6

4 Saturation spectroscopy 10
4.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.2 saturation spectroscopy results . . . . . . . . . . . . . . . . . . . . . . . . . 10
List of Figures

2.1 Resonance spectrum of the FPI . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2 Calibration of the oscilloscope from time to frequency . . . . . . . . . . . . 5

3.1 setup for the absorption spectroscopy taken from ?? . . . . . . . . . . . . . 6

3.2 photo of the setup ?? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.3 Isotopic shift of the 52 s1/2 hyperfine splitting of Rubidium . . . . . . . . . . 7
3.4 Absorption spectrum of Rubidium-gas . . . . . . . . . . . . . . . . . . . . . 7
3.5 First two dips of the absorption spectrum of Rubidium-gas fitted . . . . . . 9
3.6 Last two dips of the absorption spectrum of Rubidium-gas fitted . . . . . . 9

4.1 photo of the saturation setup . . . . . . . . . . . . . . . . . . . . . . . . . . 10

4.2 setup for the saturation spectroscopy taken from ?? . . . . . . . . . . . . . 10
4.3 Saturation spectroscopy spectrum . . . . . . . . . . . . . . . . . . . . . . . . 11
4.4 Saturation spectroscopy spectrum fitted . . . . . . . . . . . . . . . . . . . . 12
4.5 Lorentz hyperfine peaks of Rubidium gained from saturation spectroscopy
by subtracting the fitted doppler peaks. . . . . . . . . . . . . . . . . . . . . 12
4.6 Lorentz shaped peak 1 and peak 2 of Rubidium gained from saturation
spectroscopy by subtracting the fitted doppler peaks. . . . . . . . . . . . . . 12
4.7 Lorentz shaped peak 3 of Rubidium gained from saturation spectroscopy by
subtracting the fitted doppler peaks. . . . . . . . . . . . . . . . . . . . . . . 12
4.8 Estimated position of peak 4 of Rubidium gained from saturation spec-
troscopy by subtracting the fitted doppler peaks. . . . . . . . . . . . . . . . 13
1. Theory

I this experiment we want to have a closer look on optical spectroscopy. Every atom and
molecule has it’s own unique absorption and emission spectrum, which shall be determined
for Rubidium.

1.1 Emission and absorption of light

Every molecule or atom with two different energy levels E1 and E2 can interact with
photons of frequency ν if
hν = E1 − E2 (1.1)
is fulfilled. There are three different phenomena which can accrue. The first is the absorption
of the photon, which will change the energy level of the atom from E1 to E2 . If the atom
already is excited, a photon can stimulate an emission of a photon with the same frequency
and same direction. The energy level of the atom will decrease from E2 to E1 during
the emission. The last option is spontaneous emission, in which case no further photon
is needed. The emitted photon also has the frequency ν = (E2 − E1 )/h but arbitrary
direction.
For this experiment absorption will fundamental.

1.2 Spectral lines

Unfortunately the energy levels don’t result in infinitely narrow absorption and emission
lines. Due to Heisenberg’s uncertainty principle
ℏ
∆E · ∆t ≳ (1.2)
2
the energy levels posses a finite natural linewidth. Those spectral lines can be represented
by Lorenzian profiles.
Another reason for an increase in the linewidth is collisional broadening. When using a gas,
collisions between atoms are inevitable. Inelastic collisions are an additional relaxation path
for excited atoms, shortening the average lifetime of excited states. Due to Heisenberg’s
uncertainty principle this leads to an increase of the spectral linewidth.
In the case of elastic collisions, energy of an atom’s electronic system is transferred to
the electromagnetic radiation field. This causes phase disturbances between the emitted
photon before and after the collision. Hence the elastic collision leads to a broadening and
a shift of the spectral linewidth.
Doppler broadening also leads to a broadening of the linewidth. It’s caused by the motion
and therefore also temperature of the atoms. A spectral line dominated by Doppler
broadening can be represented by a Gaussian profile. Usually Doppler broadening is about
two orders of magnitude larger than the natural line width. In part two of the experiment
one will see how Doppler broadening can be avoided.

1
2 P4 Laserspectroscopy

1.3 Rubidium atom

Rubidium posses only one valence electron and consequently behaves hydrogen-like. 72.17 %
of natural Rubidium is 85 RB and 27.83 % 87 Rb. Both isotopes are effectively stable but
possess a different nuclear spin and therefor a different hyperfine splitting. The splitting of
the hyperfine structure is cause by the coupling of the magnetic moment of the core with
the magnetic field induced by the electron.
Comparing the spectrum of both isotopes further, an additional shift, called isotopic shift,
stands out. This can be explained by different physical properties like mass, volume and
charge distribution of the nucleus.
In this experiment the isotopic shift of the 52 s1/2 hyperfine splitting should be examined.
2. Preparation

2.1 Fineness
To determine the wavelengths of our absorption and saturation spectrum, we have to
convert the time axis into a frequency axis. We can take advantage over the fact that the
Fabry-Pérot interferometer (FPI) lets only pass eigenemodes. When using those eigenmodes,
the quality of the FPI will effect the accuracy of spectroscopy frequencies in the end. The
finesse describes the quality of the FPI and is defined as
F SR
F = , (2.1)
∆v
where ∆v is the spectral linewidth and F SR is the free spectral range between two peaks:

F SR = vm − vm−1 (2.2)

With this in hand the finesse can be experimentally determined by fitting Lorentz peaks to
the resonance spectrum of the FPI. In figure 2.1 the fitted data can be seen. By taking the
mean of the peak distance the FSR is determined and by averaging the width of the peaks
we get the spectral line width. Note that the first peak is neglected in this calculation,
since this peak is the last of the previous trigger sequence, therefor the time distance is
meaningless. By paying attention to the standard deviation when building the mean for
the FSR and ∆v, one will get an error by applying the Gaussian error propagation. This
yields an finesse of:
F ≈ 14.4 ± 0.064. (2.3)

The finesse can also be calculated using the reflectively R = 99.85 ± 0.05 % of the confocal
resonator in the following way:
πR
F = = 1046.4 ± 0.00157. (2.4)
1 − R2

To get a similar result as the experimental value the reflectivity should be approximately
90 %. The large discrepancy between the literature value and the experimental value could
arise from a smudged resonator.

2.2 frequency calibration

For the given confocal FPI this resonance condition is
c0
ν= · m = m · 2.5 GHz (2.5)
4l · n
with c0 the velocity of light in vacuum, n = 1.0 the refractive index of the FPI and
l = 30 mm the length of the resonator. Using the equidistant frequencies of neighboring

3
4 P4 Laserspectroscopy

Figure 2.1: Resonance spectrum of the FPI with fitted lorentzian peaks.

Table 2.1: Data of the relevant resonant peaks of the FPI. The first peak is omitted
since it is a artifact of the previous cycle.
Peak number mean [ms] mean error in [s] width γ [s] width error [s]
1 0.34 9.4·10−7 1.93 · 10−4 2.8 · 10−6
2 2.95 9.1·10−7 1.62 · 10−4 2.6 · 10−6
3 5.31 9.0·10−7 1.48 · 10−4 2.6 · 10−6
4 7.56 10.0·10−7 1.39 · 10−4 2.9 · 10−6
5 9.73 9.9·10−7 1.22 · 10−4 2.9 · 10−6
6 11.82 10.0·10−7 1.09 · 10−4 2.9 · 10−6

peaks, each of the peaks in figure 2.1 can be assigned a frequency value relative to its
neighboring peak. By plotting these frequencies over the time the peak appeared on the
oscilloscope, a conversion between time and frequency can be found. This can be seen in
figure 2.2. Note, that the first peak is not considered, since it is an artifact of the previous
cycle. The error on the time values is given by the error on the means of the lorentz peaks
as calculated by the fit function, while the error on the frequencies is estimated to be
0.1 GHz. The result is the following conversion formula:
Hz
f = (−9.30 ± 1.15) · 108 Hz + t (1.13 ± 0.01) · 1012 (2.6)
s
Chapter 2. Preparation 5

Figure 2.2: Calibration of the oscilloscope from time to frequency

3. Absorption spectroscopy

3.1 experimental setup

The first set up of the experiment is for the absorption spectroscopy, which means that the
photo diode will be right behind the Rubidium gas cell, just separated by a neutral density
filter and a lens. The setup can be seen in figure 3.1 and 3.2.

Figure 3.1: setup for the ab-

sorption spectroscopy taken Figure 3.2: photo of the
from ?? setup ??

3.2 absorption spectroscopy results

Taking a closer look at absorption spectroscopy, one will notice that the Doppler shift is the
dominant effect compared to the natural line width. Hence for the fit for the absorption
data Gaussian-distributions are used. Figure 3.4 shows the result of measured values
together with the fitted curve. Figures 3.5 and 3.6 show the same spectrum in more detail
and in a higher resolution. Which is why only the date from those is taken in account to
measure the absorption peaks.
For all four peaks the position and the the width can be determined, which is shown in
table 3.2. The distances between each peak to the others allows to assign the values to
Rb85 or 87 Rb. Comparing it to the given configuration in figure 3.3 gives that the first
dip is 87 Rb F=2 second dip is 85 Rb F=3 third dip is 85 Rb F = 2 forth dip is 87 Rb F =
1. The difference to the literature values, shown in table 3.4 and table 3.5 is within the
margin of error. The error on the frequency difference is rather large, since the error for
the frequency calibration carries over to the fitted means of the peaks.

With the given equation:

2
σ·c m

T = (3.1)
ω0 8kB ln(2)

one can determine the temperature using the laser frequency ω0 = 394.349 THz, the speed
of light in vacuum c, the boltzmann constant kb , m85 = 84.91 u for the mass of 85 Rb and

6
Chapter 3. Absorption spectroscopy 7

Figure 3.3: Isotopic shift of the 52 s1/2 hyperfine splitting of Rubidium

m87 = 85.47 u for the mass of 87 Rb.

For the dip width σ the FWHM of the respective dip is used.
The temperature is calculated for all four dips, the results of which can be seen in table 3.1.
All calculated temperatures are significantly higher than expected result especially for the
first and last dip. These two dips stem from the 87 Rb and the resulting temperature is far
from the expected 300 K. Even though the temperatures derived from the 85 Rb dips are
closer to the expected value they are still too high.

Table 3.1: Calculated temperature from the doppler broadening of the absorption
spectroscopy.
Peak number mass [u] temperature [K]
1 85.47 412 K
2 84.91 397 K
3 84.91 358 K
4 85.47 439 K

Figure 3.4: Absorption spectrum of Rubidium-gas with dips fitted with a linear slope
background. Mean position of the dips, as well as width of the fitted doppler broadened
dips can be found in table 3.2.
8 P4 Laserspectroscopy

Table 3.2: Gauss fitted absorption dips of doppler broadened hyperfine dips from full
spectrum
Peak number Mean [GHz] width σ [GHz]
1 1.48 ± 0.20 0.62 ± 0.05
2 2.76 ± 0.09 0.61 ± 0.02
3 5.82 ± 0.20 0.58 ± 0.05
4 8.31 ± 0.62 0.64 ± 0.15

Table 3.3: Gauss fitted absorption dips of doppler broadened hyperfine dips from partial
spectrum
Peak number Mean [GHz] width σ [GHz]
1 1.46 ± 0.12 0.66 ± 0.03
2 2.74 ± 0.05 0.62 ± 0.01
3 5.81 ± 0.10 0.58 ± 0.03
4 8.29 ± 0.33 0.58 ± 0.09

Table 3.4: distance between the gauss peaks in full spectrum

gap distance mean position [GHz] literature value [GHz]
∆f1,2 1.28 ± 0.28 1.30
∆f2,3 3.06 ± 0.28 3.04
∆f3,4 2.49 ± 0.82 2.50

Table 3.5: distance between the gauss peaks in partial spectrum

gap distance mean position [GHz] literature value [GHz]
∆f1,2 1.28 ± 0.17 1.30
∆f2,3 3.07 ± 0.15 3.04
∆f3,4 2.48 ± 0.43 2.50
Chapter 3. Absorption spectroscopy 9

Figure 3.5: First two dips of the absorption spectrum of Rubidium-gas fitted. Values of
mean and width of the dips as well as the errors on them can be found in table 3.3.

Figure 3.6: Last two dips of the absorption spectrum of Rubidium-gas fitted. Values of
mean and width of the dips as well as the errors on them can be found in table 3.3.
4. Saturation spectroscopy

4.1 Setup
For the saturation spectroscopy a polarizing cubic beam splitter (PBS) is placed in front
of the Rubidium gas cell. The photo diode is replaced with a mirror and the photo diode
itself is put in the second beam of the beam splitter, as one can see in figure 4.1 and 4.2.

Only if the laser is in resonate with a transition, both beams will interact with particles
of zero velocity. When travelling through the Rubidium gas cell the first time, the laser
depopulates the particle’s ground state and saturates the transition. Hence the absorption
will be decreased on the way back through the medium. Hence for the resonating wavelength
a Lamb-dip will occur in the Doppler-broadened spectrum, which can be measured.

Figure 4.2: setup for the

Figure 4.1: photo of the sat- saturation spectroscopy taken
uration setup from ??

4.2 saturation spectroscopy results

At first the frequency calibration was repeated in the same way as for the absorption
spectrum. The Lamb-dips within the Doppler-broadened absorption spectrum can be seen
in figure 4.3. By fitting the Doppler spectrum as seen in figure 4.4 and subtracting the
fit from the measured date, the Lorentz shaped saturation peaks can be isolated. The
resulting spectrum is shown in figure 4.5.

As before the location and the width of the peaks determined by fitting them. The subtracted
spectrum is very noisy as the gaussian fit does not perfectly negate the doppler-broadened
spectrum. Therefore for a more accurate calculation the fits were made individually for
the zoomed in spectra, as seen in figure 4.6, 4.7 and 4.8. The first and last peak do not
show the distinct lorentz-shape as they are probably not captured accurately most likely
due to errors during the experiment. The determined values of peak position and width
are shown in table 4.1. Noticeable are the smaller widths of the peaks compared to the

10
Chapter 4. Saturation spectroscopy 11

Table 4.1: Values of position and width of lorentz peaks in the saturation spectroscopy
spectra.
Peak number Mean position [GHz] width γ [GHz]
1 2.22 ± 0.005 0.018 ± 0.004
2 3.54 ± 0.003 0.051 ± 0.005
3 6.63 ± 0.002 0.023 ± 0.002
4 9.12 Not determined

Table 4.2: distance between the Lorentz peaks of saturation spectrum

gap distance mean position [GHz] literature value [GHz]
∆f1,2 1.32 ± 0.008 1.30
∆f2,3 3.09 ± 0.005 3.04
∆f3,4 2.49 2.50

absorption spectrum, which should result in more accurate values. As the last lorentz peak
could not be fitted due to the large noise in the data only the position is determined with
a peak finding algorithm while the width is not calculated.

Interestingly the experimentally determined value for the hyperfine splitting through
absorption spectroscopy is closer to the literature value than the values determined by
saturation spectroscopy. This is against expectations, since the lorentz shaped peaks are
engulfed in the gauss curve of the doppler shift in absorption spectroscopy, while they
are visible in the saturation spectroscopy. Therefore the results should have been more
accurate in the saturation spectroscopy.

Figure 4.3: Full saturation spectroscopy spectrum of Rubidium gas

12 P4 Laserspectroscopy

Figure 4.4: Saturation spectroscopy spectrum of Rubidium gas with the Doppler peaks
fitted.

Figure 4.5: Lorentz hyperfine peaks of Rubidium gained from saturation spectroscopy
by subtracting the fitted doppler peaks.

Figure 4.6: Lorentz shaped peak 1 and peak 2 of Rubidium gained from saturation
spectroscopy by subtracting the fitted doppler peaks.

Figure 4.7: Lorentz shaped peak 3 of Rubidium gained from saturation spectroscopy by
subtracting the fitted doppler peaks.
Chapter 4. Saturation spectroscopy 13

Figure 4.8: Estimated position of peak 4 of Rubidium gained from saturation spec-
troscopy by subtracting the fitted doppler peaks.