9 On the Investigation of Excited State Dynamics with (Pump-)Degenerate
209
(Fig. 9.3). Here it is interesting to understand the interplay between the time and
spectral resolution [27]. Although the time resolution of such snapshots is a priori given by the delay T and the pulse durations (Fig. 9.2(a)), the measurement of
each vibrational spectrum also modifies the time resolution. In order to measure a
spectrum, a DFWM transient with a given length (delay τ 23 ) must be Fourier transformed. The longer the DFWM transient, the higher will be the spectral resolution,
but at the same time, the lower will be the time resolution of the snapshot, and viceversa. In other words, the interplay between spectral and time resolution in pump
DFWM may be fine adjusted by the length of the transient measured during the
delay τ 23 .
9.2.2 Setup Description
In our setup [21, 22, 28–31], the laser for the DFWM sequence is initially generated in a broadband noncollinear optical parametric amplifier (nc-OPA), which is
pumped by a commercial femtosecond laser system (1 KHz, 300 µJ at 800 nm).
After pulse compression in a prism compressor, typical pulse durations are about
11 and 15 fs depending on the central wavelength. After beam-splitting the nc-OPA
output in three DFWM beams (pump, Stokes and probe beams), two delay stages
are used to control the delay (Fig. 9.2(a)) between pump and Stokes (τ 12 ) and between Stokes and probe (τ 23 ). In general, the relative delay τ 23 is scanned with a step
length of about 2–3 fs in order to obtain enough data sampling for high-frequency
modes. All three beams are focused using the same concave mirror (f = 25 cm) in
a folded BOXCARS phase-matching geometry (Fig. 9.2(b)). Typical spot diameters
are around 60–100 µm and the energies are 20 nJ for pump and Stokes and 15 nJ for
the probe.
The combination of DFWM with an initial pump is straightforward. The IP pulse
is generated in a second nc-OPA with similar pulse parameters as the DFWM ncOPA. In such a way, the spectrum can be individually tuned from the DFWM’s
nc-OPA, which is an important requirement for the detection of excited electronic
states (see Sect. 9.2.3). The delay T between the IP and the DFWM sequence is
controlled by an additional delay stage. The IP beam is focused using an additional
concave mirror in an extended folded BOXCARS phase-matching geometry. This
allows controlling individually the spot size of the IP beam. Typical IP pulse energies used in pump-DFWM experiments depend on several parameters like spectral
overlap between excitation and absorption spectra, which will be discussed below.
In general, in order to achieve a high signal-to-noise-ratio (SNR), energies in the
range between 30–80 nJ are used.
The detection of the (pump-)DFWM signal is performed using two photomultipliers at two different wavelengths. An interferometric filter with a FWHM of 10 nm
is used in front of each detector. In order to avoid signal attenuation by using a 50 %
beam splitter, the signal beam is split with high-pass filters into two beams. In this
case, the low-frequency spectral region is reflected to one detector while the highfrequency spectral region is transmitted to the other detector. This always allows one
209
(Fig. 9.3). Here it is interesting to understand the interplay between the time and
spectral resolution [27]. Although the time resolution of such snapshots is a priori given by the delay T and the pulse durations (Fig. 9.2(a)), the measurement of
each vibrational spectrum also modifies the time resolution. In order to measure a
spectrum, a DFWM transient with a given length (delay τ 23 ) must be Fourier transformed. The longer the DFWM transient, the higher will be the spectral resolution,
but at the same time, the lower will be the time resolution of the snapshot, and viceversa. In other words, the interplay between spectral and time resolution in pump
DFWM may be fine adjusted by the length of the transient measured during the
delay τ 23 .
9.2.2 Setup Description
In our setup [21, 22, 28–31], the laser for the DFWM sequence is initially generated in a broadband noncollinear optical parametric amplifier (nc-OPA), which is
pumped by a commercial femtosecond laser system (1 KHz, 300 µJ at 800 nm).
After pulse compression in a prism compressor, typical pulse durations are about
11 and 15 fs depending on the central wavelength. After beam-splitting the nc-OPA
output in three DFWM beams (pump, Stokes and probe beams), two delay stages
are used to control the delay (Fig. 9.2(a)) between pump and Stokes (τ 12 ) and between Stokes and probe (τ 23 ). In general, the relative delay τ 23 is scanned with a step
length of about 2–3 fs in order to obtain enough data sampling for high-frequency
modes. All three beams are focused using the same concave mirror (f = 25 cm) in
a folded BOXCARS phase-matching geometry (Fig. 9.2(b)). Typical spot diameters
are around 60–100 µm and the energies are 20 nJ for pump and Stokes and 15 nJ for
the probe.
The combination of DFWM with an initial pump is straightforward. The IP pulse
is generated in a second nc-OPA with similar pulse parameters as the DFWM ncOPA. In such a way, the spectrum can be individually tuned from the DFWM’s
nc-OPA, which is an important requirement for the detection of excited electronic
states (see Sect. 9.2.3). The delay T between the IP and the DFWM sequence is
controlled by an additional delay stage. The IP beam is focused using an additional
concave mirror in an extended folded BOXCARS phase-matching geometry. This
allows controlling individually the spot size of the IP beam. Typical IP pulse energies used in pump-DFWM experiments depend on several parameters like spectral
overlap between excitation and absorption spectra, which will be discussed below.
In general, in order to achieve a high signal-to-noise-ratio (SNR), energies in the
range between 30–80 nJ are used.
The detection of the (pump-)DFWM signal is performed using two photomultipliers at two different wavelengths. An interferometric filter with a FWHM of 10 nm
is used in front of each detector. In order to avoid signal attenuation by using a 50 %
beam splitter, the signal beam is split with high-pass filters into two beams. In this
case, the low-frequency spectral region is reflected to one detector while the highfrequency spectral region is transmitted to the other detector. This always allows one