Laser in Physics, Chemistry, Biology and Medicine

generally weaker mode, for example in spatial space (volume of the active medium). The practical
application of this property is easy to see: By creating a situation of high mode competition, a
strong tendency towards single mode operation of the laser is achieved.

A spatial competition of individual modes occurs because the standing wave of a mode, as shown in the
figure, fully utilizes a certain area of the amplifier medium and therefore leaves no residual amplification
for another mode that would require the same area. However, if another mode has its antinodes in those
areas of the medium where the first has its valleys, it can in principle start to oscillate. The phenomenon
that a mode claims a certain area of the medium for itself is called Spatial Hole Burning.

Figure 3.22. Spatial Hole Burning.

One possibility of increasing the mode competition so that in most cases only a single mode oscillates is to
use the spatial hole burning effect. Choosing a ring-shaped resonator (ring laser) causes the antinodes to
change their position continuously. Accordingly, a mode requires practically the entire area of the medium
and there is no longer any possibility of amplification for other modes. A ring laser therefore has a very high
tendency to go into single mode operation. It should be pointed out, however, that this mode will not be
stable without further frequency-stabilizing elements in the resonator.

The simplest form of a ring resonator is shown in the figure. A right-hand and a left-hand traveling wave
can develop in this. The traveling wave is very similar to the standing wave, but with the wave bellies (and
valleys) moving clockwise or counter-clockwise at a constant speed. The existence of two opposing
traveling waves is undesirable and one of the two has to be suppressed in a ring laser.

42
Laser in Physics, Chemistry, Biology and Medicine

Figure 3.23. Ring resonator

3.4.2 Frequency spectrum of the laser

In summary, the frequency and frequency width of the laser is determined on the basis of the above
considerations by the following factors:

• The resonance frequency and frequency range of the amplifying medium fundamentally determine
the possible frequency range of the laser.
• Due to the resonator structure, not all frequencies can develop in the amplifier range, but only
those that correspond to a possible mode of the resonator (see Figure 3.24).

Figure 3.24. a) Frequency spectrum of a laser. b) Two short-term recordings of the mode spectrum of a free-running
argon laser. Both images were copied on top of each other to demonstrate the statistically fluctuating spectrum.

The longitudinal modes that will develop in the active resonator correspond to those of the passive one.
However, the following changes must be taken into account:

• The refractive index of the medium must be taken into account for the effective resonator length.
• The light waves are amplified in the medium after each reflection on the mirror. This shows that in
the corresponding formula for the quality for the passive resonator only 𝑅 (reflection coefficient of
the mirror) has to be replaced by the gain 𝐺(𝜈).
• By introducing frequency-selective (dispersive) elements (e.g. Fabry-Perot interferometer, etalons),
a few modes - in extreme cases a single mode - can be selected.

3.5 Frequency selection, single mode operation

As an example, the situation for a single-mode dye laser is shown in Figure 3.26. This laser is also a good
example of a tunable laser.

43
Laser in Physics, Chemistry, Biology and Medicine

Each laser is in principle "tunable" within its gain profile, that is, its frequency can be changed. However, if
the gain profile is only the Doppler-broadened resonance line, then the term "tunable laser" is generally
not used. To get a tunable laser, the gain profile must be significantly above the Doppler width. For
example, a dye enables a gain profile of typically a few hundred Å.

By narrowing down the wavelength with frequency-selective elements, one can then achieve that only a
certain frequency range from the gain profile really contributes to the laser light. If the laser works in
multimode mode, it is sufficient to adjust a prism or a birefringent filter to tune the laser. In single-mode
operation, however, after setting the coarse adjustment element (in Figure 3-3, step 1), a more complex
procedure must be selected.

After the rough selection, tuning takes place by “shifting” the active, oscillating laser mode on the
frequency axis. In most cases this is done by changing the length of the resonator. This can be achieved in a
number of ways, e.g. by mounting one of the resonator mirrors on a piezo element or by inserting a small
rotatable glass (quartz) plate approximately at the Brewster angle in the resonator [with this method, the
effective resonator length is changed by changing the path of the light in the material with 𝑛 ≠ 1; please
also note that the installation at Brewster's angle serves to minimize reflection losses]. As shown in Figure
3.25, however, shifting the laser mode alone would not lead to the desired goal, since it would move out of
the maximum of the transmission curve of the frequency-determining etalon and thereby promote a mode
moving into this. This is why it is necessary to readjust these elements with single-mode lasers.

Figure 3.25. Narrowing of the gain profile of a dye laser by two Brewster prisms made of heavy flint and mode
selection by two etalons. (a) Experimental setup; (b) Explanation of the mode selection by superimposing the gain
profile and losses of the selecting elements.

44
Laser in Physics, Chemistry, Biology and Medicine

Figure 3.26. Tuning a single mode dye laser.

3.6 Distinguishes ultrashort laser pulses

3.6.1 Mode coupling

The first step in generating ultra-short pulses is to use the mode structure of the laser. Due to the
molecular nature of the active medium, the gain profile of most lasers will be inhomogeneously broadened.
Accordingly, a large number of modes will generally oscillate. With a He-Ne laser, there will be about 10
modes [∆𝜈6UoomTH = 1500𝑀𝐻𝑧 (about 0.2 Å), resonator length → 𝜈„UeT = 150𝑀𝐻𝑧].

We can then write the total electrical field 𝐸IS# in the following form:

(…`%)/'

𝐸IS# = Ÿ 𝐸. 𝑒 /':[;+ 4`∆;,(-. .4`r%]

$(…$%)/'

In the equation, 𝐸. is the amplitude of the 𝑛-th mode and 𝑁 is the total number of modes. 𝜑. is any phase
constant. If these phase constants are not constant over time (which is the case in many cases, coherence!)
The individual modes will oscillate independently of one another and contribute to the frequency spectrum
of the laser radiation. But if they are constant in any way, the individual modes will interfere. With the
simplified assumption that all 𝐸. are equal and 𝜑. = 0 then results for 𝐸IS# :

45
Laser in Physics, Chemistry, Biology and Medicine

sin[(∆𝜈„UeT 𝑁𝑡)/2]
𝐸IS# (𝑡) = 𝐸& 𝑒 /':;+ 4
sin[(∆𝜈„UeT 𝑡)/2]

This is practically the same formula as for multi-beam interference on a grating, but here for a Fourier
transformation from the frequency space to the time period. This means that temporal laser pulses with a
time interval 𝑇 and a half-value width arise:

2𝑑 1
𝑇= =
𝑐 ∆𝜈†I</48

2𝜋
∆𝑇 =
𝑁 ∙ ∆𝜈

In the case of mode locking, the laser delivers short light pulses of length ∆𝑇 and with the repetition
frequency (repetition rate) 𝑇.
The length of the light pulses is therefore shorter, the greater the number of interfering modes, or the
wider the gain profile of the laser.

Example. 𝐻𝑒 − 𝑁𝑒 laser: ∆𝑡 ≈ 0.6𝑛𝑠, 𝑁𝑑-glass: 0.3𝑝𝑠, dye: 30 𝑓𝑠.

How can you now implement a mode-locked laser? How can we make the oscillating modes have a
constant phase relationship and interfere? One important method is called active mode locking. For this
purpose, a shutter is built into the resonator (for example an acousto-optic modulator or a Pockels cell), the
frequency of the shutter roughly corresponding to the mode spacing of the resonator. The carrier
frequency and the higher harmonic thus correspond to possible resonator modes. Since the laser modes
pass through the modulator, they are modulated with its frequency and corresponding sidebands are
generated. At a certain time all modes have their maximum at the position of the modulator, a situation
that repeats itself after every revolution 𝑇 = 2𝑑/𝑐. The side ligaments and the longitudinal modes
experience a fashion competition and vie for the maximum gain. The most efficient state is that where the
sidebands are attracted to the longitudinal modes. Since the higher harmonics of the modulation frequency
are phase-locked, so are the induced laser resonator modes.
The scheme and implementation is shown in the figure.

Another possibility is passive mode locking. A non-linear absorber (for example dye) is inserted into the
resonator, which serves as an absorber and thus as a loss at low laser powers. If a power peak from the
noise is so intense that the threshold value is exceeded despite the loss caused by the absorber, then this
spike will quickly increase significantly. All the more so since the absorber soon changes into the non-linear
range (it becomes saturated) and the losses for this spike are therefore very low compared to all other
spikes. This results in very short pulses, similar to active mode locking.

46
Laser in Physics, Chemistry, Biology and Medicine

Figure 3.27. Realization of the mode coupling.

For both pulsed and continuous wave lasers, passive mode coupling is an experimentally particularly simple
method with which pulse widths of less than 1ps can be achieved.

It can be clearly understood as follows: Instead of the active modulator, a “saturable absorber”, whose
“absorption level” has the shortest possible relaxation time, is placed in the laser resonator close to an end
mirror. In order to reach the laser threshold despite the absorber, the gain in the laser medium must be
correspondingly high. Shortly before the pump power has reached the threshold value, the emission of the
active medium consists of spontaneously emitted, induced amplified photon avalanches, the peak power of
which fluctuates more or less statistically. As a result of the non-linear saturation of the absorbing medium,
the most intense photon avalanche experiences the lowest absorption and thus the greatest net gain. It
therefore grows faster than its weaker competitors during the next resonator cycle, so it saturates the
absorber even more and thus further increases its net gain.

After a few rounds of the resonator, this pulse has become so strong that it accounts for the majority of the
total laser emission. After a settling stage, this therefore consists of a regular sequence of intense pulses,
the interval between which is 𝑇 = 2𝑑/𝑐 and which lasts as long as the pump power remains above the
threshold, which is now lower due to the saturation. The Fourier decomposition of this pulse sequence
gives the mode spectrum of the laser. The mode locking is achieved at the fixed points in time 𝑡 = 𝑡& +
𝑞 2𝑑/𝑐 when the saturation pulse passes the absorber. This explains the term “passive mode coupling”. A
good picture for understanding the term passive mode coupling is the following consideration: the intense
pulse opens a door when passing through the absorber, which the other modes use to run through them in
phase. This is particularly beneficial for them as they hardly suffer any losses. So they couple each other in
terms of their phase.

3.6.2 Pulse propagation in medium and self-phase modulation

With model locking, pulses with half widths of less than 1𝑝𝑠 can be achieved. In order to fall below this
value, additional methods must be used. One of the most used lately is the so-called pulse compression by
self-phase modulation.

With the help of this procedure it is possible to obtain pulses with a half width of a few femtoseconds. The
spectral broadening through self-phase modulation in a medium (mostly light guide) with subsequent
compression in a dispersing element, such as a grating, is used.

We now assume that a light pulse with the spectral amplitude distribution 𝐸(𝜔) passes through a medium
e= e e</0 e</0
that has the refractive index 𝑛(𝜔). The group speed 𝑣b = e>
= e> p𝑣oA 𝑘q = 𝑣oA + 𝑘 e>
with e>
=
e h h e.(>)
e> .
j m = − ." j e>
m, which determines the speed of the pulse maximum, has a dispersion:

47
Laser in Physics, Chemistry, Biology and Medicine

𝑑𝑣b 𝑑𝑣b 𝑑𝑘 1 𝑑' 𝜔

= =
𝑑𝜔 𝑑𝑘 𝑑𝜔 𝑣b 𝑑𝑘 '

If the last term in the equation is ≠ 0, then the shape of the pulse in the medium will change. This must
always be taken into account when sending a short laser pulse through a medium! In a normal dispersive
medium and low laser intensities, a situation then arises that is shown in the following figure.

Figure 3.28. Spatial (temporal) broadening of a pulse in a medium with a normal linear refractive index.

The pulse is delayed, broadened in time and a frequency chirp (the frequency is different for different times
of the pulse) is impressed. However, the frequency spectrum is not broadened, in contrast to self-phase
modulation, which we will discuss below.

It is now essential for pulse compression that, given sufficiently high light intensities, the medium with the
refractive index 𝑛(𝜔, 𝐼) has a non-linear component:

𝑛(𝜔, 𝐼) = 𝑛& (𝜔) + 𝑛' 𝐼(𝑡)

The phase of the light wave thus depends on the intensity of the light:
𝜔𝑛𝑧 𝑛& 𝑧 𝑛' 𝜔𝑧
𝜑 = 𝜔𝑡 − 𝑘𝑧 = 𝜔𝑡 − = 𝜔 j𝑡 − m−~ € 𝐼(𝑡)
𝑐 𝑐 𝑐

Furthermore, since the frequency 𝜔 is equal to the time derivative of the phase, it follows from the above
equation:

𝑛' 𝜔𝑧 𝑑𝐼
𝜔 = 𝜔& − ~ €
𝑐 𝑑𝑡

and further that the light frequency becomes smaller at the beginning of the pulse (𝑑𝐼/𝑑𝑡 > 0), but
increases at its end. This is called self-phase modulation. The spectral profile of the pulse is therefore wider
(next figure).

Figure 3.29. Spectral broadening of a pulse with a non-linear refractive index.

As shown above, however, the linear component of the refractive index 𝑛(𝜔, 𝐼) now causes the red
components in the medium to experience a greater speed and the blue components a lower speed (with
normal dispersion).

48
Laser in Physics, Chemistry, Biology and Medicine

The first step of pulse compression is actually the opposite, namely a temporal expansion of the pulse,
whereby the components have different wavelengths at different times (GVD Group Velocity Dispersion).
However, it is essential that the frequency spectrum has been broadened, which means that shorter pulses
are only possible.

3.6.3 Pulse compression

The second step is to compress the temporally and spectrally broadened pulse, which can be done in
different ways.

For example, if one sends the spectrally and temporally broadened light through two parallel gratings, as
shown in the figure, then the optical path 𝑆 is different for different wavelengths, and it increases with
increasing wavelength. So you can compensate for the higher speed of the red light through a larger optical
path and thus compress the pulse.

Figure 3.30. Pulse compression with a pair of grids.

1
𝑆(𝜆) = 𝑆% + 𝑆' = 𝐷 Žcos 𝛼 + − sin 𝛼 tan 𝛽•
cos 𝛽
𝑑𝑆 𝐷𝜆
→ =
𝑑𝜆 𝜆 +/'
𝑐𝑑' r1 − jsin 𝛼 − m s
𝑑

→ Optical path increases with increasing wavelength! (𝑑 ... grid spacing)

Another practical implementation with single-mode light guides is shown in the following figures.

Figure 3.31. Generation of 𝑓𝑠 pulses through self-phase modulation in light guides and pulse compression.

Instead of the grid arrangement to compensate for the dispersion, so-called “chirped mirrors” have recently
been used. These are mirrors that are coated with multilayers that are calculated in such a way that they
reflect different spectral components at different depths in the layer. Therefore, they cause a frequency-
dependent group speed delay, which ideally compensates for the self-phase modulation generated (Figure

49
Laser in Physics, Chemistry, Biology and Medicine

3.33). A particular advantage of these “chirped mirrors” is their uniform behaviour over a large frequency
range.

The following figure shows the principle of pulse compression using a prism compressor (allows relatively
large dispersion corrections, for example after a laser amplifier) and a chirped mirror.

Figure 3.32. Principle of a prism compressor (e.g. after a 𝑓𝑠 laser amplifier).

Figure 3.33. Chirped Mirror.

3.7 Pulse shaping

Any light pulse (Figure 3.34) can be represented either in the time period or in the associated frequency
Fourier space completely by a time-dependent (frequency-dependent) amplitude and a time-dependent
(frequency-dependent) phase (equations 3.22, 3.23).

Figure 3.34. A laser pulse is described by its amplitude and phase (both time-dependent).

50
Laser in Physics, Chemistry, Biology and Medicine

1
𝐸(𝑡) = N𝐼(𝑡)𝑒 /(=+ 4$r(4)) + 𝑐. 𝑐 (3.22)
2
1 1
𝐸´ (𝜔) = N𝑆(𝜔 − 𝜔& )𝑒 $/r(=$=+ ) + N𝑆(𝜔 − 𝜔& )𝑒 $/r(=`=+) (3.23)
2 2

The representation in the frequency domain is important because it can be easily implemented
experimentally and the frequency-dependent phase can then be influenced. This means that any pulse can
be shaped as desired. After the transformation back into the time period, a pulse of any shape (pulse
shaping, pulse shaping) is obtained.

The freely selectable phase function in 3.23 is particularly easy to interpret if it is expanded into a Taylor
series around the frequency 𝜔& (Eq. 3.24). The coefficients 𝜑% , 𝜑' and 𝜑+ correspond to easily
interpretable physical constants: 𝜑% is the group delay, which means that it shifts the pulse on the time
axis. 𝜑' gives the group delay dispersion, or, in other words, the spread of the pulse over time. 𝜑+ can be
related to the chirp of the pulse.

Figure 3.35. Various possibilities to influence a pulse transformed into the frequency space (shapes, pulse shaping).

𝜔 − 𝜔& (𝜔 − 𝜔& )'

𝜑(𝜔) = 𝜑& + 𝜑% + 𝜑' +⋯ (3.24)
1! 2!
𝑑𝜑
𝜑% = » (3.25)
𝑑𝜔 =$=+

𝑑' 𝜑
𝜑' = Ú (3.25)
𝑑𝜔 ' =D=
+

Since the parameter space for a phase function can be very large, it is very difficult (practically impossible)
to determine the optimal phase function and thus the optimal pulse shape with conventional methods.
However, what is known as a genetic optimization method (evolutionary algorithm) is available, which is
explained in principle in Figure 3.36.

The goal of this coherent control, the selective preparation of a target state, is one of the main
prerequisites for mode-selective chemistry. The laser pulse tailored to a specific stimulus, which drives the
system from the initial to the target state as a disturbance, can usually no longer be determined in advance
by quantum mechanical calculations in complex systems, since often the Hamilton operator of the system

51
Laser in Physics, Chemistry, Biology and Medicine

is no longer known is. One approach is to determine the required laser field in a control loop which uses a
signal obtained from the experiment as feedback. This optimization loop is run through until a pulse that
meets the requirements has been found. Previous experiments in this area were largely limited to the
wavelength range of 𝑇𝑖: 𝑆𝑎 lasers and their harmonics.

Figure 3.36. Principle of an evolutionary algorithm.

3.8 Frequency stabilization of a laser

In single-mode operation of a laser, frequency stabilization of the laser is practically inevitable. There are
several reasons why frequency stabilization is necessary. Even with extreme mode competition and the
resulting single mode operation without many aids, there is no control over the oscillating mode without
stabilization and only one mode will then be active, but different modes will oscillate at different times.
Control and stabilization is also necessary for sensible tuning of the laser, as otherwise the coordinated
tuning of the laser mode and the transmission curves of the etalons described above cannot be guaranteed.

It should also be noted that although the bandwidth of the active laser mode is only a few 𝑘𝐻𝑧 or less,
these values are ideal values, since in practice the resonator length 𝐿 cannot be kept completely constant.
Every change in length ∆𝐿 (due to temperature influences, pressure changes, etc.) manifests itself in a
frequency change. However, these changes are much slower than the formation of the laser mode itself, so
that we actually have a mode oscillating with a width 𝑑𝜈, but this constantly changes its mean frequency,
which results in a larger effective bandwidth 𝑑𝜈T55 ≫ 𝑑𝜈.

Figure 3.37. Effective bandwidth and frequency drift.

This effective bandwidth can also be kept as small as possible by frequency stabilization. With a
corresponding effort, it is possible to come very close to the ideal value 𝑑𝜈.

52
Laser in Physics, Chemistry, Biology and Medicine

The methods of frequency stabilization are based on the principle that the deviations of the current
frequency of the laser are compared with a very stable frequency standard (for example stabilized FPI,
atomic resonance line) and the resonator length is corrected proportionally to the change detected. In
addition to short-term (typically in the 𝑚𝑠 range) fluctuations in the resonator length, there are long-term
(typically minutes to hours) changes in the resonator length, the so-called long-term drifts.

To do this, a small part of the laser light is sent through an extremely well stabilized FPI or through a cell
with a type of atom that has a resonance line at the frequency used. For example, the parameters of the
FPI can be selected so that the laser radiation is exactly in the transmission maximum or precisely in its
flank at the turning point. An undesired change in the wavelength of the laser, for example due to drifts,
then results in a relative shift of the active laser mode in the transmission curve of the FPI from the target
position at the maximum or at the turning point. Associated with this is a (albeit small) change in the
transmitted light intensity. This change can be registered with a photodiode. Usually the quotient of
𝐼eSHhA sfa /𝐼& or 𝐼eSHhA sfa − 𝐼& is measured. If a change is found, this must result in a correction of the
frequency-determining elements of the laser. For example, a resonator mirror can be adjusted to
compensate for the change in frequency.

Stabilization on an external FPI enables stabilization at any frequency and also allows control and
stabilization when tuning. Of course, the FPI speaker must also be voted on.

Figure 3.38. Frequency stabilization.

Another, mostly additionally used stabilization when tuning is done by “attracting” the oscillating mode to
the frequency-determining (narrowest band) FPI. For this purpose, either the resonator length of the laser
is modulated a little periodically using one of its mirrors, which is mounted on a piezo, or the transmission
curve of the FPI. The resulting periodic change in intensity of the laser is detected using a lock-in technique.
As long as the mode is exactly at the maximum of the transmission curve of the FPI (desired position!), the
error signal obtained in this way at the Lock In (DC voltage signal) is equal to 0 due to the positive and
negative identical intensity deviation. However, if the mode drifts from the maximum (equivalent to a
change in frequency), the error signal becomes positive or negative ≠ 0 and can be used to correct the
resonator length.

53
Laser in Physics, Chemistry, Biology and Medicine

The main wavelength of the 𝑁𝑑-YAG (glass) laser is 1.0461𝜇𝑚, that of the 𝐸𝑟-YAG about 3𝜇𝑚. The ruby
laser delivers radiation at 6943Å. Due to the high power of 𝑁𝑑 solid-state lasers, they are very suitable for
frequency multiplication or frequency mixing. Most systems have at least one frequency doubling as
standard, which provides a frequency in the green wavelength range. However, frequency tripling to
around 350𝑛𝑚 is also typical. This wavelength can also be used as a starting point for further frequency
multiplication up to 120𝑛𝑚.

The YAG laser is also often used as a pump laser for dye lasers. Another important area of application is in
medicine. High-performance laser systems (e.g. laser fusion) are also based on these lasers.

4.2.2 Diode-pumped continuous wave YAG lasers

In recent years, continuous, diode-pumped solid-state lasers have increasingly been used to replace the
argon ion laser. The advantages are obvious: good efficiency, handy and very easy to stabilize.

Figure 4.14. Diode-pumped Yag laser Verdi.

A model (Verdi, Coherent) is shown in Figure 4.14. Nd:Vanadate (Nd:YVO4) is used as the laser material.
With the ring resonator design, up to 10𝑊 output power can be achieved at the frequency-doubled,
(single-frequency) wavelength 532𝑛𝑚. Diodes are used for pumping (809𝑛𝑚), which are coupled into the
laser head via light guides. Vanadate absorbs particularly well at the 809𝑛𝑚 wavelength.

Figure 4.15. Energy scheme of dyes resulting in wide gain profile.

60
Laser in Physics, Chemistry, Biology and Medicine

4.3 Tunable lasers

4.3.1 Dye laser

We now want to deal with one of the most important tunable lasers. The most important requirement for
this is a very wide reinforcement profile. In the case of dyes, this is due to the broad "energy bands" of the
individual electronic energy levels. Actually, these are only very closely spaced Ro-vibration levels of the
individual electronic states, but they practically overlap so much that we can speak of "energy bands". The
excitation takes place from the ground state into the higher Ro vibration levels of the excited electronic
state. The electrons collect in the lowest Ro-vibration levels of this level through collisions. From here, laser
radiation takes place in the various higher (originally unoccupied) Ro-vibration levels of the basic state. The
typical width of the various dyes is a few 100Å. How far this can really be used depends on the pump
source and is generally wider for pulse operation. Dyes can be used to cover the range from approximately
350𝑛𝑚 to 750𝑛𝑚. For the infrared range from 700 to almost 900𝑛𝑚, however, 𝑇𝑖: 𝑆𝑎 (𝐴𝑙' 𝑂+ ) has
recently become more suitable. With some models with the same resonator structure, you can cover the
entire wavelength range by exchanging the dye nozzle and the 𝑇𝑖: 𝑆𝑎 crystal.

Figure 4.16. Various dyes.

The realization of a pulsed dye laser is shown in the following figure. The wavelength selection takes place
with a grating which is arranged as a retroreflector and thus simultaneously serves as one of the resonator
mirrors. The beam expansion is necessary in order not to make the power density at the grid too high
(Hänsch type oscillator).

Figure 4.17. Tuning ranges of dyes.

61
Laser in Physics, Chemistry, Biology and Medicine

A complete system is shown in the next figure. In addition to the resonator (actual laser), it contains one or
two post-amplifiers. These consist of an active medium (dye), which is also population-inverted with the
pump laser (here excimer laser), but is not located in a resonator. The laser beam is now further amplified
in this population-inverted dye.

Another interesting aspect of the configuration shown above is the type of coupling of light from the
resonator selected here. This is not done by a semi-transparent resonator mirror, but by partial reflection
on a prism, which is otherwise mainly used as a beam deflector. Overall, after exiting the resonator, the
laser beam is brought to the correspondingly high output power in two downstream amplifiers. The cell for
the first post-amplifier is identical to that of the amplifier of the resonator, whereby, of course, a different
part of the dye cell is used geometrically. An (excimer) pump beam is split and pumps all three amplifiers.
The bandwidth that can be achieved with such a pulsed system is around 2 − 3𝐺𝐻𝑧 without and 900𝑀𝐻𝑧
with an etalon.

Next we want to consider a tunable continuous wave system which enables single mode operation.

Figure 4.18. Construction of a simple pulsed dye laser.

Figure 4.19. Pulsed dye laser with post-amplifier.

Figure 4.20. Ring dye laser.

62
Laser in Physics, Chemistry, Biology and Medicine

The present system is a ring dye laser that is pumped by an 𝐴𝑟-ion laser. The dye is a thin stream and is
created with a nozzle. The elements in the resonator discussed in earlier sections ensure single-mode
operation. An effective bandwidth of 1𝑀𝐻𝑧 is achieved by means of frequency stabilization. An
electronically controlled tuning range of 30𝐺𝐻𝑧 is possible by turning the Brewsterplate and at the same
time changing the thick etalon. A wavelength range from around 420 to 850𝑛𝑚 can be covered with dyes.
With a 𝑇𝑖: 𝑆𝑎 amplifier medium, the above model can be tuned from 690 to 1025𝑛𝑚.

An autoscan unit is also available for this system, which enables the entire range of a dye (or the 𝑇𝑖: 𝑆𝑎
range) to be automatically tuned.

Tunable lasers are certainly one of the most important tools for spectroscopy. Many new spectroscopic
data could and will be obtained with them. In addition, they are an indispensable instrument for many
applications in spectroscopy, such as laser analysis, etc. Dye lasers are also used in medicine.

4.3.2 𝑻𝒊: 𝑺𝒂 Laser

The 𝑇𝑖: 𝑆𝑎 (𝐴𝑙' 𝑂+ ) laser has proven to be a particularly suitable laser for ultra-short pulses. The active
medium of this solid-state laser is a sapphire crystal doped with 𝑇𝑖 +` ions, the laser level diagram of which
is shown in the following figure (Figure 4.21). The essential criterion for the good suitability as a short-term
laser is the broad emission spectrum, which enables mode coupling over about 100𝑛𝑚 and therefore
makes laser pulses in the femtosecond range possible in principle. Since the absorption cross-section of the
𝑇𝑖 ions is small, the size of the crystal must be large (a few 𝑚𝑚 to 𝑐𝑚). On the other hand, this results in
the fact that non-linear effects, such as those required for short-term lasers (we have already mentioned
Kerr lens mode locking above), are relatively strong and can be used well. Furthermore, this method offers
the advantage that the corresponding "saturation effect" can take place as quickly as required.

Figure 4.21. 𝑇𝑖: 𝑆𝑎 Laser.

63
Laser in Physics, Chemistry, Biology and Medicine

6. BASIC TECHNIQUES AND APPLICATIONS OF

LASER SPECTROSCOPY

6.1 Fundamental methods

In the following we want to discuss some of the basic techniques of spectroscopy, with the distinction
between each method given by the measurands determined in the experiment.

6.1.1. Absorption spectroscopy

When a laser beam passes through a medium, the intensity decreases according to the following
relationship:

𝐼(𝜔, 𝑧) = 𝐼4HI.# = 𝐼& 𝑒 $G(=)9 (6.1)

Equation 6.1 is of general validity and is often referred to as Beer's law. According to the above
considerations, the absorption coefficient 𝛼(𝜔) is:
𝜔JI 𝑔I
𝛼(𝜔) = ¼𝑛I − 𝑛> ½ ∙ 𝐵IJ (6.2)
𝑐 𝑔J

The quantity to be determined in absorption spectroscopy is the absorption coefficient 𝛼(𝜔). With the
arrangement shown in Figure 6.1, 𝛼(𝜔) can be determined, since from equation 6.2. It follows [applies to
the mostly fulfilled case that 𝛼 is small]:

𝐼& − 𝐼4HI.#
𝛼(𝜔) ≈ (6.3)
𝐼& ∙ 𝑥

Figure 6.1 Typical arrangement of a measurement based on absorption spectroscopy. A tunable laser is tuned across
possible absorption lines of a gas located in the absorption cell. To increase absorption, the laser beam can be
reflected back and forth in the cell more often in order to increase the absorption distance 𝑥. A Fabry-Perot
interferometer can be used for frequency calibration.

Since the absorbed intensities are generally very low due to the small absorption coefficients, the accuracy
and sensitivity of absorption spectroscopy is very limited. In general, one can assume that the primary
intensity 𝐼& is significantly greater than the absorbed intensity. As a result, the signal-to-noise ratio is
relatively high with this method.

71
Laser in Physics, Chemistry, Biology and Medicine

An increase in sensitivity can be achieved by using the lock-in technique. For this, either the laser frequency
or the laser intensity must be modulated periodically. With this method it is possible to measure small
signals that are hidden in the noise. Another possibility to increase the sensitivity is to have the absorption
cell inside the laser resonator (intracavity spectroscopy), where the available power is much higher.
However, for experimental reasons, this is only possible in special cases.

6.1.2 Fluorescence spectroscopy

The above-mentioned difficulty of absorption spectroscopy can be avoided by measuring the fluorescence
from the excited level caused by the laser excitation instead of the intensity absorption. One speaks then of
fluorescence spectroscopy or emission spectroscopy. If the atoms or molecules are transported from the
state 𝐸I to the state 𝐸J by the laser radiation, then the system will generally respond with an emission of
spontaneous photons (resonance fluorescence). In Figure 2.1 and Figure 5.2 this is described by the decay
(relaxation) constant 𝛾J . It should be noted that 𝛾J can also be a sum of different 𝛾/ , which then describe
different decay channels to lower energy levels. Then the mean inverse lifetime of the upper state is:
.
.
1
𝛾̅J = Ÿ 𝛾/ = Ÿ 1 𝜏
/
/D% /

𝜏/ are the lifetimes of the upper state with respect to possible decay channels into lower states 𝐸/ .

Figure 6.2. Estimation of the 𝑛 photoelectrons registered on the counter in a typical fluorescence spectroscopy
experiment. Let us assume that 𝑁* atoms were promoted to the energy state 𝐸* by means of laser radiation. Then
𝑁* 𝛾* are emitted in the entire solid angle. The solid angle component 𝛿𝛺 is detected with our detection optics. We
denote the losses caused by the optics with 𝛼. Of the photons arriving at the PMT, a fraction 𝜂 [quantum efficiency,
modern PMT reach values around 0.2] is converted into photoelectrons, which we register with the counter.

The crucial factor for fluorescence spectroscopy is the fact that the spontaneously emitted photons are
emitted isotropically into the solid angle 4𝜋. As a result, the observation can take place in any direction in
which there are no laser photons. This results in a low background noise, in contrast to absorption
spectroscopy. On the other hand, in most cases it is not possible to detect the photons in the entire solid
angle. One of the main problems in fluorescence spectroscopy is therefore to optimally map the emission

72
Laser in Physics, Chemistry, Biology and Medicine

volume onto the photon detector. In general, photodiodes or photon amplifiers (photomultiplier tubes,
PMT, channel plates) are used as detectors.

The photons registered at the PMT [we want to register the photons when 𝐸> decays into a level 𝐸/ ] can
then be described by steps 1–3 in Figure 6.2.

Figure 6.3. Optical system for the effective collection of the laser-induced fluorescence.

Let us ask ourselves which absorbed intensities ∆𝐼 we can still detect by means of the fluorescence if the
stimulating intensity 𝐼 is given. We make the following assumptions: photomultiplier noise ≤ 10 𝑐𝑜𝑢𝑛𝑡𝑠/
𝑠𝑒𝑐; signal-to-noise ratio ≈ 10, i.e. about 100𝑐𝑜𝑢𝑛𝑡𝑠/𝑠𝑒𝑐 signal can still be measured well. 𝛿𝜔 ∙ 𝛼 ≈
0.1, 𝜂 ≈ 0.2. The number 𝑛I of absorbed photons is proportional to the number of atoms in the lower
energy state, the absorption cross section 𝛽/> , the length of the interaction volume and the number of laser
photons 𝑛g . According to our assumptions, we need about 5 ∙ 10+ absorbed photons 𝑛I (this also means 5 ∙
10+ emitted photons under equilibrium conditions) to register 100 photoelectrons on the counter. Let us
imagine that we have 1𝑊 laser power to get the 5 ∙ 10+ absorbed photons [we can either imagine a small
cross section or a small number of atoms; after all, 𝑛I = 𝑁/ 𝛽/> 𝑛g 𝑥]. At 500𝑛𝑚 this gives ≈ 3 ∙ 10%! laser
photons. For these realistic data there is a ∆𝐼/𝐼 ratio of ≤ 10%, .

Perhaps we should also note here that so far we have assumed an idealized state in which no noise from
laser photons will get into the photomultiplier. This assumption is certainly unrealistic when we observe
photons with the same frequency as the excitation photons. In many cases, however, it is possible to
observe a transition with a different frequency and to use a filter or spectrometer in the observation beam
path. Then the ideal conditions assumed above are given quite well.

An improvement in the photon yield in fluorescence spectroscopy can be achieved under certain conditions
by using light guides or special mirrors around the fluorescence volume (see Figure 6.3).

6.1.3 Optogalvanic Spectroscopy

Optogalvanic spectroscopy is the exploitation of the fact that in a plasma, when a resonance transition of
one of the plasma constituents is selectively excited with a laser, the equilibrium between voltage and
discharge current is disturbed. This is the case because the excitation changes the occupation of the
individual atomic states, which in turn corresponds to a change in the ionization probability. This can be
measured via the resulting discharge current change. For spectroscopic investigations, a hollow cathode
lamp (cold cathode discharge) in which the anode is made of the material to be examined is usually used as

73
Laser in Physics, Chemistry, Biology and Medicine

the plasma source. First the discharge is ignited with a filling gas (usually helium or neon). Material is
atomized by the plasma from the anode and enters it, where it is excited and partially ionized.

Figure 6.4. Set-up for determining the absorption line of a resonance transition by means of the optogalvanic effect.

A typical arrangement for exploiting the optogalvanic effect is shown in Figure 6.4. Since the current
changes resulting from the excitation are small compared to the constant portion of the current, the laser
beam is periodically interrupted and a lock-in technique is used. With pulsed laser excitation, the current
change simultaneous with the laser pulse is detected.

Optogalvanic spectroscopy is primarily used for the spectroscopy of discharges. Due to the relatively simple
detection (no photon detector is needed) of an electrical voltage and the possibility to use hollow cathode
lamps for a variety of elements, optogalvanic spectroscopy is often used as a reference method for complex
laser spectrometers and in a wide range of applications.

Figure 6.5. Optogalvanic spectrum of a neon discharge recorded with a broadband dye laser.

An example is shown in Figure 6.5. At 7𝑚𝐴 discharge current (left part of the spectrum) you can see
predominantly the argon lines of the carrier gas, while at 10𝑚𝐴 the uranium lines also appear due to the
increased atomization (sputtering) yield (right part of the spectrum).

6.1.4 Ionization Spectroscopy

In the methods that we have seen so far, the common feature is the registration of photons. In the
following methods, not photons but a different physical quantity is measured.

In ionization spectroscopy, the atoms or molecules are ionized by means of laser radiation. Of course, this is
generally not possible by absorbing a single photon, because the ionization energies of the atoms are
almost always larger than the photon energies that can be achieved with lasers. However, absorption of
several photons (usually 2 photons are sufficient) is possible. We will deal with the cross-sections as well as

74
Laser in Physics, Chemistry, Biology and Medicine

the further details and applications of multiphoton ionization spectroscopy in a later chapter. A simple
scheme of a corresponding experiment is shown in Figure 6.7.

Provided that the atoms (or particles in general) can be ionized efficiently, the detection of the resulting
photo-ions is generally more efficient than the detection of photons that would be generated during
excitation to bound states. In the previous chapter we showed that in the best case we can achieve a
detection sensitivity of 0.02 for photons. A detection sensitivity of 1 can be achieved for ions.

With this high detection sensitivity of ions in combination with powerful pulsed lasers, ionization
spectroscopy achieves the highest detection sensitivity for particles among all known methods.

Figure 6.7. Scheme of a laser ionization spectrometer. The laser beam symbolizes one or more laser beams with the
frequencies 𝜔$ and 𝜔! running in parallel.

This method is so interesting and opens up a variety of new applications for laser spectroscopy because the
ions can be detected particularly effectively. We want to look at the ionization spectroscopy method
especially with regard to its use as a detection method for small and very small trace elements, i.e.
analytical purposes.

In practically all known cases, one photon is not enough to ionize atoms or molecules from the ground
state. For this reason, one often speaks of multiphoton ionization spectroscopy. In the following we will
always assume that the absorption of several photons is involved in the ionization process.

This applies to lasers with pulse lengths of up to a few hundred 𝑓𝑠 and the corresponding powers up to
around 10%% 𝑊/𝑐𝑚' . For shorter pulses and correspondingly higher powers, the ionization process is
fundamentally different: in this case the field strength is so high that the Coulomb field of the atom is
changed and tunnel or even overbarrier ionization can occur (see below).

In the case of classical multiphoton ionization, one can now distinguish between two possibilities for how
the particle can be ionized by absorbing several photons:

• Resonance Ionization Spectroscopy (RIS)

• Non Resonance Ionization Spectroscopy (NRIS)

75
Laser in Physics, Chemistry, Biology and Medicine

Figure 6.8. Possible scenarios for resonant (RIS) and non-resonant (NRIS) ionization of an atom by photon absorption.

Let us consider the sketch in Figure 6.8. The essential characteristic of RIS is that the particle is resonantly
excited in one or more steps by means of narrow-band laser radiation and then ionized in a final step with
an intense, relatively broad-band laser. There is a case with NRIS that is similar to two-photon absorption.
Two or more photons are absorbed in order to overcome the energy interval without the need for
intermediate resonant levels. It is easy to see that the effective cross-sections for the RIS process are one to
several orders of magnitude higher than those for the NRIS process. On the other hand, two or more lasers
are required in the first case, while one laser is sufficient for NRIS.

A brief note on the history of photoionization spectroscopy. The first such experiments were carried out at
the Institute of Spectroscopy of the Academy of Sciences of the USSR by Letokhov and co-workers in the
early 1970s. A number of experiments on multiphoton ionization were subsequently carried out and
proposed by this group. Another landmark is the first detection of individual atoms with this method (1977,
S. Hurst and co-workers in Oak Ridge, USA). Since 1980 this method has been used in a variety of ways by
different groups.

We now want to consider which atoms of the periodic table we can detect using resonance ionization
spectroscopy. This consideration is essential when considering using this method as an analytical method.
In this case it is important to be able to detect a spectrum of elements as wide as possible.

Figure 6.9. Further possible RIS schemes. The ionizing laser radiation 𝜔! can be a separate laser, but can also be
identical to 𝜔$ . 2ħ𝜔 means a frequency doubled laser.

The theoretical treatment of the excitation process is similar to that of normal laser excitation. However,
depending on the selected process ① − ⑤, its treatment will be correspondingly extensive and complex.

76
Laser in Physics, Chemistry, Biology and Medicine

Another difficulty arises in the treatment of pulsed laser radiation, which consists of a large number of laser
modes.

It can be shown that the ionization probability for non-resonant multiphoton ionization with 𝑛 photons is
given by:

𝑊/U. = 𝜎/U. 𝐼 .

The probability is dependent on the laser intensity. The cross-section 𝜎. consists in the most general case
of a sum over an infinite number of dipole matrix elements between atomic states (bound and free), but
can in many cases be reduced to the sum over a few quasi-resonant ones. In principle, it can be determined
experimentally using the above formula.

A promising way of ionization is the use of ultra-short laser radiation. The achievable power of >
10%' 𝑊/𝑐𝑚' allows to reach a field strength range in which the superposition of the laser field strength
with the Coulomb field during a period of light oscillation leads to a reduction in the Coulomb wall (see
Figure 6.10).

The main advantages of this ionization method are: a) The ionization cross-section is practically
independent of the material. b) Atoms (particles) with high ionization energy can also be ionized.

The intensity range in which the multiphoton ionization transitions to tunnel ionization is described by a so-
called Keldysh parameter 𝛾:

𝐼o
𝛾==
2𝑈o

Where 𝐼o is the ionization energy and 𝑈o :

(𝑒𝐸)'
𝑈o = = 9.33 ∙ 10$%, ∙ 𝐼 ∙ 𝜆'
4𝑚T 𝜔&'

The theory for atoms is quite well established for this. There is currently only a rudimentary theory for
molecules.

Figure 6.10. Principle of tunnel or overbarrier ionization with ultra-short pulses and the high field strengths that occur.

77
Laser in Physics, Chemistry, Biology and Medicine

6.2 High resolution spectroscopy

Methods for reducing the inherent Doppler broadening

The laser offers a number of advantages in terms of its application for spectroscopy. Above all, we can
achieve high power densities with the help of lasers. Furthermore, the radiation is directional and coherent.
Finally, the laser radiation is characterized by a narrow bandwidth, which is crucial in spectroscopy for the
achievable resolution of various closely spaced resonance transitions. Let us have a look at the following
table, in which the widths (broadening) of spectral lines under different conditions are compared with laser
bandwidths. The table can also be used as a reference for converting between energy quantities that are
frequently used in spectroscopy.

The table clearly shows that with modern lasers we can achieve bandwidths that are roughly comparable
to, or even below than, natural line widths. If we were to make spectroscopy in a thermally Doppler-
broadened gas, we would have to expect a residual broadening that is significantly higher than the
resolution of the laser. In order to fully utilize the advantages of the laser, it will be necessary to find
methods and procedures that make it possible to bypass the residual broadening and still do high-
resolution spectroscopy. Only in this way will it be possible for us to resolve lines and transitions that
exhibit sub-doppler splits.

Figure 6.11. Comparison of individual quantities used in spectroscopy to characterize the line widths.

In the following chapters we will discuss individual methods that make it possible to eliminate the Doppler
Effect and demonstrate them with examples.

6.2.1 Atomic beam spectroscopy

The principle of atomic beam spectroscopy is simple and is a direct consequence of the dependence of the
Doppler Effect on the excitation geometry. Let us consider the relationship between frequency shift 𝛿𝜈,
particle speed 𝑣 and the angle 𝜃 between the direction of the particle speed and the direction of the
momentum of the photons 𝑘 d⃗ :

𝑣 𝑣
𝛿𝑣 = 𝑣& cosp𝑘d⃗𝑣q = 𝑣& cos(𝜃) (6.4)
𝑐 𝑐

From equation 6.4 it can be seen immediately that the Doppler Effect disappears if the angle 𝜃 is chosen to
be 90°. A practical set-up of an atomic beam spectrometer is shown in Figure 6.12.

78
Laser in Physics, Chemistry, Biology and Medicine

Figure 6.12. Principle of atomic beam spectroscopy (shown here for LIF).

However, it is not possible to achieve condition 𝜃 = 0 exactly for every particle, since the particle beam
always has a divergence 𝜀. This also determines the residual Doppler broadening 𝛿𝜈 ∗ = 𝛿𝜈 sin 𝜀, where 𝛿𝜈
is the original Doppler broadening. In practice, atomic beam spectroscopy can achieve residual Doppler
widths that are in the range of the natural line width. In principle, it is possible to a certain extent to
improve the residual broadening in favour of the intensity by means of diaphragms.

Due to its simplicity, atomic beam spectroscopy has been broadened widely and is used for a large number
of spectroscopic measurements on various types of atoms and molecules. In the following we want to
discuss the measurement of the hyperfine splitting as an example, which demonstrates well the
possibilities of atomic beam spectroscopy.

Analogous to the fine structure of an atomic energy level, which results from the addition of the orbital
angular momentum and the spin to a total angular momentum (usually denoted by 𝐽⃗) (the physical cause is
the interaction of the magnetic moments of the two angular momentum, usually referred to as the spin-
orbital coupling, which has a small additional Energy contribution to the total energy of the system), the
interaction of the magnetic field of the total electron angular momentum 𝐽⃗ with the magnetic field of the
nuclear spin 𝐼⃗ results in additional energy. The energy states corresponding to this new total energy are
characterized by a new total angular momentum 𝐹⃗ , which results from the addition of the angular
momentum of 𝐽⃗ and 𝐼⃗. The resulting splitting is called hyperfine splitting, since it is much smaller than the
fine splitting due to spin coupling.

𝐹⃗ = 𝐼⃗ + 𝐽⃗

𝐹 = 𝐽 + 𝐼, 𝐽 + 𝐼 − 1, … , |𝐽 − 𝐼|

Figure 6.13. Hyperfine splitting and fluorescence spectrum of the 𝑁𝑎 − 𝐷! line when tuning a continuous single-mode
dye laser over the 588.9𝑛𝑚 resonance line. The fluorescence spectrum is shown for low and high laser power.

79
Laser in Physics, Chemistry, Biology and Medicine

The hyperfine splits are typically on the order of 1 − 100𝑀𝐻𝑧, so they can only be resolved with narrow-
band (single mode) lasers. As an example, consider the hyperfine splitting of the 𝑁𝑎 − 𝐷' line at 588.9𝑛𝑚.
The spectra shown in Figure 6.13 are measured with an apparatus analogous to Figure 6.12. A dye laser
with rhodamine 6G was used, whose radiation has a bandwidth of << 10𝑀𝐻𝑧. The hyperfine lines can be
resolved well for small laser powers (< 2𝑚𝑊/𝑐𝑚' ). At higher laser powers (> 50𝑚𝑊/𝑐𝑚' ) the saturation
broadening can be clearly observed, which means that the individual lines can no longer be resolved.

With increasing laser intensity, the excitation probability ℜIJ becomes larger, which results in a broadening
of the half-width of the line problem. The reason is that for the high transition probability in the centre of
the resonance curve, saturation is already achieved, but on the wings of the resonance curve (line profile)
the signal still increases. This effect is very important in laser spectroscopy and must be taken into account
due to the high power of the exciting radiation achieved with lasers.

From the intensities of the measured fluorescence signals, the effect of the optical pumping in a two- or
three-level system can be observed well. Only the transition 𝐹 ‰ = 3 → 𝐹 = 2 represents a (in this case not
ideal) two-level system. Accordingly, the signal for this transition is much larger than for the others, since
the atom can be pumped more often and is not relatively quickly (after a few pump cycles as with the other
transitions) "lost" into another level.

Another example is shown in Figure 6.13 and Figure 6.14. The spectroscopy of molecules is characterized by
the complexity of the spectra, since there are generally a large number of vibration and rotation levels,
which overlap each another. In order to resolve individual rotation and vibration levels, on the one hand
the Doppler Effect must often be suppressed and on the other hand the temperature must be reduced so
that only the lowest levels are occupied.

Figure 6.14. a) Hyperfine structure of the rotation line R(31) of the 𝑁𝑎! molecule. b) Hyperfine structure and isotope
shift of radioactive 𝑁𝑎 isotopes.

Figure 6.15. Atomic model of adiabatic cooling during expansion into vacuum and velocity distribution before and
after expansion.

80
Laser in Physics, Chemistry, Biology and Medicine

This can be realized by molecular jets generated by free adiabatic expansion of a gas into vacuum. In the
expansion nozzle, the molecular concentration is high, increasing the probability that the translation,
vibration, and rotation energy will be partially converted into expansion energy. This results in a molecular
jet with small translational, vibrational, and rotational energies, where the occupancy of each vibrational
state (𝜈) and rotational state (𝐽) is a function of the vibrational and rotational energies and the rotational
(𝜏H ) and vibrational (𝑇; ) temperatures, respectively.

Figure 6.16. Determination of the velocity perpendicular and parallel to the molecular beam axis by measuring the
corresponding Doppler profiles of an absorption line of the 𝑁𝑂! molecule.

81