38

Chapter 3. Experimental

Overview :

The experimental set-up was designed to provide short (sub-100 fs) pulses of microjoule energies in the UV and in the red spectral region for photoexcitation and stimulated emission pumping of DCM and photoexcitation of LDS 750. It also produced a femtosecond broadband continuum ( 350-900 nm) to monitor the transient absorption and emission changes of the photoexcited dyes. Control of the temporal delay between excitation and probing pulses was achieved by a computer-controlled optical delay stage in the path of the pump (and, for the case of DCM, also of the dump) pulse. A/D conversion and storage of the measured data were effected by means of a PC A/D converter card and a home-programmed user interface. The pump-probe experiments on DCM were performed with a set-up based on that described in [Bing 95], producing tunable excitation pulses between 450 and 530 nm and a probe continuum of approx. 45 fs pulselength. The diameter of the pump beam on the sample was 150 µm, and the focus diameter of the probe continuum 80 µm.

The next three subsections will give a description of the short pulse generation, amplification and pump-probe set-up, whereas the following three subsections will be concerned with synchronization, signal detection and noise reduction. A description of the basic data correction procedures concludes the treatment of the femtosecond set-up. The last three subsections deal with the characteristics of the employed chemicals and the experimental conditions for the photostationary absorption, fluorescence, molecular beam and Raman measurements. All measurements were conducted at 20 °C, if not indicated otherwise.

39

3.1. Short Pulses

3.1.1. Generation

The colliding pulse mode-locked (CPM) laser is a ring dye laser pumped by a continously (cw) emitting argon ion laser, which here was a Lexel 3500-7 (Polytec) operating between 0.9 and 1.2 W at 514 nm only. The CPM generated optical pulses of 60 fs pulsewidth at a fundamental of 630 nm with a repetition rate of 76 MHz ( Figure 3.1-1 ). The home-built laser resonator consisted of the six-mirror, four-prism configuration of Valdmanis and Fork [Vald 86]. Fused silica prisms with an apex angle of 68.9° were used. The gain mirrors had a curvature of R=100 mm, the absorber mirrors had R=50 mm. A jet with a thickness of 100-150 µm of a solution of 3.4 mM rhodamine 6G in ethylene glycol was the gain medium, and a approx. 17 µm thick jet of a 0.89 mM solution of DODCI in the same solvent served as saturable absorber. The output power in one of the beams behind a 3% output coupler was around 8 mW, which corresponds to a pulse energy of about 0.1 nJ.

The principle of the CPM pulse generation is passive mode locking caused by the presence of a saturable absorber. As in all femtosecond laser oscillators, stable performance can be achieved because of the interplay of self phase modulation and group velocity dispersion [Mart 84]. Self phase modulation is due to the optical Kerr effect in the ethylene glycol jets and to gain and absorber saturation (see also 2.4). It contributes a positive frequency chirp to the pulses, generating new frequency components and therefore broadening the laser bandwidth [Sieg]. The four prism set introduces in total a negative group velocity dispersion (a negative chirp) [Bor 84, Fork 84], depending on the distance between the second and third / third and fourth prism. It should compensate the linear positive chirp from the glycol jets and the prism material itself. The amount of the latter can be adjusted by translating a prism in a direction normal to its base (see also 2.4).

The geometric stability range of a resonator can be calculated with the help of ABCD matrices [Sieg]. A similar calculation to that in [Bult] was used to obtain the stability range of the set-up. Starting from empirical distances between the gain and absorber mirrors, so that the fluorescence was imaged at a distance of 2 m from the left gain mirror and 4-4.5 m from the right one, the gain and absorber mirror distances were symmetrically varied in steps of 10 and 5 µm, respectively, to find a stable mode-locking regime. Care should be

40

Mode-locking was achieved in the most stable fashion near the laser threshold of 0.9 -1.2 W. Removing the absorber jet lowered the threshold to 240 mW. High absorber concentrations, raising the laser threshold above 1.3 W, led to instabilities due to competitive modes and were therefore avoided. For better solubility, 50 ml of methanol was added to the ethylene glycol when preparing the DODCI solution. The measurement quality was greatly improved by replacing the dyes already after 50 h of operation, although at that time no sign of degradation, e.g. amplitude instability, had shown yet. When replacing the DODCI solution, the corresponding pump was cleaned once with acetone.

Figure 3.1-1 : CPM-Laser set-up.

3.1.2. Amplification

Nonlinear processes, such as stable continuum generation and frequency doubling require a pulse energy in the mikrojoule regime. Three dye amplification stages provided pulse energies of 0.2 µJ after passing the first stage and a saturable absorber, 26 µJ after the second stage and up to 240 µJ after the third stage. The amplified pulses were broadened in time due to positive group velocity dispersion of the materials in the optical paths and had to be recompressed to 65 fs pulselengths using a prism compressor.

The pump laser for the amplification stages was a Q-switched, seeded Nd:YAG laser (Continuum), delivering approx. 70 mJ of the extra-cavity frequency-doubled fundamental at 532 nm at 30 Hz repetition rate. The pulselength in seeded operation was 4-5 ns, with a timing phase jitter of 1 ns rms between the electronic timing pulse for the Pockels cell

41

For further details regarding the optical geometry of the amplification stages, please see Figure 3.1-2 . The prism compressor consisted of two SF10 prisms placed at an apex distance of 72 cm (see 2.4). After compression and spatially limiting the beam to the central Airy disk, the pulse energy ranged between 140 and 160 µJ containing approximately 1% of amplified spontaneous emission (ASE).

42

Figure 3.1-2 : Laser pulse dye amplification set-up, after [Loch].

43

3.1.3. Diagnostics

To ensure an optimal performance, several parameters of the generated and amplified pulses were under constant control. The spectrum of the pulses behind the output coupler of the CPM oscillator was monitored simply by the second order reflection from a grating (d=2400 g/mm). It should be smooth, extending from approx. 615 nm to 645 nm, and stable especially with regard to its red edge. The pulsewidth was controlled by means of second harmonic generation from two parts of the beam superimposed at a finite angle in a nonlinear KDP crystal (non-interferometric autocorrelation). One of these parts was shifted in time with respect to the other with the help of a mirror mounted on a piezoelectric crystal for the pulses from the CPM oscillator or by a stepper motor driven translation stage for the amplified pulses, while the generated UV light was detected with a photomultiplier or with a photodiode, respectively. An autocorrelation trace of the amplified pulses is given in Figure 3.1-3 .

Figure 3.1-3: Autocorrelation trace of the amplified CPM pulses.

With the assumption of a sech² pulse shape, the pulselength is calculated as 0.65 of the intensity autocorrelation full width at half maximum (fwhm) [Sala 80] of 85 fs, which corresponds to a pulselength of 55 fs. To measure the pulsewidth of pulses shorter than 40 fs, a phase-sensitive (interferometric) autocorrelation technique would be chosen [Szab 88].

44

The spatial profile of the beam was checked to ensure a smooth TEM (0,0) mode output and prevent hot spots.

Another diagnostic means which has become a standard in the last few years of optical short pulse spectroscopy is the frequency-resolved optical gating (FROG). Here the intensity and phase of the pulse are measured simultaneously. The beam is split and the parts are mixed in a nonlinear medium, for example a Kerr medium, creating a Kerr polarization rotation of one part of the beam, which is then frequency resolved and serves as the input to an algorithm retrieving the electric field of the pulse [DeLong 94]. The amount and type of relative phase-shift (linear or quadratic chirp, see 2.4) can be directly perceived from the frequency-resolved output, which facilitates its compensation.

3.1.4. Handling

In general, the optical path length for a beam passing through a medium of length l and refraction index n is given as n l. The refraction index may be intensity-dependent (electrooptical Kerr effect), subjecting short pulses propagating through materials to intensity-dependent nonlinear processes such as self-focusing and self-phase modulation. The pulses are also broadened in time due to different propagation times for their spectral components, imposing a delay of the higher energetic parts relative to the lower energetic ones (group velocity dispersion).

Self-focusing is an induced lens effect. Assuming a single-mode beam with a Gaussian transverse profile propagating in a medium with refractive index n given by

n = n ₀ + n ( ), (3.1)

where n ( ) or n (I) is an optical-field induced refractive index change. If n is positive, the central part of the beam having a higher intensity experiences a larger refractive index than the edge and therefore travel at a slower velocity than the edge. As the

45

Self-phase modulation stems from the same intensity-dependent refractive index change n( ). Consider a time-dependent envelope E(t) of an optical pulse and an instantaneous response n( ) given by n₂ = n₂ I(t) of the refractive index. In a length l of the nonlinear medium, the pulse is subject to an intensity-dependent phase shift

(t) = ( / c) n₂I(t) l and a corresponding frequency modulation (t)= - , which appears as a broadened spectrum [Sieg].

Self-phase modulation is one of the contributing mechanisms for continuum generation, which may be desired for tunability reasons, but is an annoying effect if it takes place in the probe cell or in a nonlinear crystal intended for second harmonic generation (see 2.5). Therefore, care was taken to avoid self-focusing or self-phase modulation by expanding the amplified beam to one cm diameter with a telescope (f = -50 and f = 200 mm) just behind the third amplification stage. When any focus was required, usually the largest acceptable focused spot size was used. For a Gaussian beam of wavelength , the focused spot size d₀ behind a lens with focal length f is approximated as d₀ 2f / D [Sieg]. D is the diameter of the beam when passing the lens. The depth (length) of focus is /2 (d₀ / )² .

Group velocity dispersion is a consequence of the frequency dependence of the refraction index. For most transparent dielectric materials in the visible region of the electromagnetic spectrum the following approximation holds :

n₀ () = A + B ². (3.2)

The high frequency components of the pulse are more delayed than the low frequency ones, thus the pulse broadens in time. To compensate this so-called "positive dispersion" (wavelength-dependence) of the group velocities, a prism or grating set-up introducing a net group velocity dispersion with the opposite sign can be used [Bor 84, Fork 84, Treacy 69]. These frequency-dependent delay lines are called pulse compressors, as each can compensate a quadratic phase shift reducing the pulselength of the broadened pulse.

46

Figure 3.1-4: Prism pair used for pulse compression.

According to a Fourier theorem, the time-bandwidth product of any pulsed signal is constrained by the uncertainty principle f_rms t_rmsr 1/2, where f_rms and t_rms are the root-mean-square widths of the signal in frequency and in time. As a consequence, the achievable pulsewidth (fwhm) for a Gaussian pulse with a frequency width of f is limited to a minimum of 0.44 / f [Sieg].

In a prism compressor, the beam passes twice through two prisms separated by a length L. The group-velocity dispersion is defined as the third term of a Taylor expansion of the pulse phase in frequency:

(3.3)

The corresponding instantaneous temporal phase to the third term in equation 3.3 is proportional to the square of time, t², and the corresponding instanteneous frequency has a linear dependence on t. That is why a quadratic phase shift means a linear frequency shift (linear chirp), a cubic phase shift means a quadratic chirp etc.

For the prism set-up of Figure 3.1-4 , with e defined as AB + CD , is given as

, (3.4)

where n₀ is the refraction index of the prism material and is its first derivative in wavelength ( = (dn / d)_ë0 ) at the wavelength ₀; is the second derivative. The negative contribution to the group velocity dispersion scales with the distance L between the prisms, whereas the path e within the prisms adds positive group velocity dispersion. Values

47

Table 3.1.4 : Values for the refractive index and its wavelength derivatives for different media at 620 nm [Salin 87].

Medium	n0	n0' / µm-1	n0'' / µm-2
SF10	1.7244	-0.1079	0.5725
FeD 05-25	1.8011	-0.13326	0.67787
SiO2	1.4572	-0.02984	0.12144
BK7	1.5159	-0.03756	0.16641
H2O	1.33	-0.02729	0.1303

SF10 prisms were chosen for compression of the amplified pulses due to their high value for , enabling a more or less compact design of the compressor. When the prisms were polished with ceroxide once a year to avoid reflection losses, the overall energy transmittance of the compressor was 90 %, a large value compared to the maximum of approx. 50% transmittance of an even more compact grating compressor.

Propagation through materials does unfortunately not only involve a quadratic phase shift, but also cubic and higher order terms, which means that equation 3.3 is only valid as a first approximation. The cubic phase shift can be compensated by using a combination of prism and grating compressor, as these devices introduce a cubic phase shift with an opposite respective sign [Cruz 88]. Also, mirrors with chirped multilayer coatings have been shown to compensate quadratic and cubic phase shifts [Szip 94]. A monotonic variation of the multilayer period throughout the deposition process leads to a wavelength-dependent penetration depth of the optical field in the coating and to the desired group velocity dispersion.

As a consequence of the Fourier theorem mentioned above, a restriction in spectral width will impose an increase in pulselength. Therefore, only broadband dielectric high reflection mirrors centered at 616 nm or aluminium mirrors were used to propagate the beam.

48

3.1.5. Tunability

Nonlinear optical processes offer the possibility of changing the pulse's central wavelength, providing flexibility to experimental demands. Sum- and difference frequency generation, parametric oscillation and - amplification have been described. To obtain UV photons, here second harmonic generation (SHG) was used, which can be viewed as the degenerate case of sum frequency generation. A broadband probe pulse was created by continuum generation in water.

KDP (Kalium dihydrogen phosphate) was chosen as a crystal with a high susceptibility for second harmonic generation of 630 nm. To overcome the obstacle of different velocities of the fundamental (630 nm) and the second harmonic (315 nm), phase matching is required. As KDP is a uniaxial dichroic crystal, the refraction indices for 630 nm and 315 nm can be the same for a certain direction of propagation [Berg].

A 0.15 mm thick KDP slab was used, cut so that a beam entering the slab perpendicular to its surface, thus suffering minimal reflection losses, would propagate at 56.7° to the optical axis and induce a nonlinear polarization oriented at 45° in the x-y plane perpendicular to the optical axis. The second harmonic emerged polarized perpendicular to the fundamental (type I phase matching). From approx. 40 µJ at 630 nm focused with a lens of f=1330 mm into the crystal, 8 µJ at 315 nm were obtained and separated from the fundamental by reflection off four UV-high reflective dielectric mirrors transparent in the visible ( Figure 3.1-5 ).

Continuum short pulse generation in liquids and solids was first observed in 1970 [Alf 70]. Self-phase modulation and self-focusing occur when a femtosecond pulse is focused e.g. into water, fused silica or sapphire. The self-focusing leads to filament formation with beam diameter narrowing [Shen 84]. In the filaments efficient nonlinear optical processes such as stimulated Raman scattering, stimulated parametric four-photon interaction and cascading frequency conversion take place [Wang 89, Penz 93, Witt 96]. The parametric four-photon interaction is degenerate in the pump laser frequency (_P + _P-> _S + _I, _S is the signal, _I the idler frequency). The filaments impose a longitudinal phase-matching condition on this process. Transversally a phase-mismatch is allowed, which gives rise to an enlargened divergence of the white light emission compared to the fundamental [Witt 96].

If the spectral extent of the laser pulses covers a Raman frequency, stimulated Raman scattering changes over to the more efficient impulsive stimulated Raman scattering

49

In our case, after careful attenuation of the fundamental and dye filtering the continuum had a bandwidth of approx. 15000 cm^-1, extending from 375 nm to 850 nm. It was generated by focusing 36 µJ of the 630 nm fundamental with a lens of f = 200 mm into a water cell of 1.5 mm length, with 0.2 mm fused silica windows. The continuum was energetically most stable when the colour spots from the different filaments nearly faded into each other and the fundamental emerged as a double-horseshoe-like spot. From the higher noise level of the baseline (optical density in the absence of the pump pulse) on the low-frequency side of the fundamental, a strong Raman contribution to the lower energy side of the continuum is concluded.

A non-divergent, low energy white light continuum can be generated by keeping the energy of the fundamental pulses just above the threshold for continuum generation. If a part of this continuum is selected by interference filters of appropriate bandwidth and optically compressed (see 2.4), it can exhibit a pulselength drastically shorter than that of the fundamental pulses [Sim 91].

50

Figure 3.1-5: Pulse compression, second harmonic generation and measurement set-up.

3.2. Broadband Pump-Probe Measurement Set-up

The measurement set-up, as well as the frequency doubling and continuum generation arrangements are presented in Figure 3.1-5 .

To modify the spectral shape of the supercontinuum, which is otherwise dominated by the generating fundamental frequency, it traversed a thin (0.45 mm total thickness, 0.17 mm fused silica windows) cuvette containing a filter solution made from 14 different dyes in ethanol.

The continuum beam was then focused into the probe cell with a combination of curved mirrors (R= 224 and R=-1220), by which combination astigmatism can be avoided. To prevent any ASE present from interfering with the experiment, the central spot of the continuum was blocked on the concave mirror. The ASE should show far less divergence

51

The astigmatism of a given optical arrangement can be calculated with ABCD matrices. Here the angle of incidence upon the concave mirror was empirically found to be critical for astigmatism compensation and minimized to = 5.5°. The continuum beam was split by a 50 % broadband beamsplitter (from Laser Optics) into a probe and a reference part. The degree of reflection from the beam splitter was observed to be dependent on the angle of incidence ( Figure 3.2-1 ), increasing strongly for larger angles in the range above 650 nm. Due to pulse-to-pulse spatial changes of the continuum, this caused additional transmission changes leading to an increase of the noise level, which was dealt with by reducing the incidence angle of the continuum onto the beam splitter to = 12.5°.

Figure 3.2-1: Transmission of the beamsplitter for different angles of incidence. Solid and dotted curves alternate for successive angles.

Only the reflected part of the continuum was focused into a flow cell, which had fused silica windows and a 0.3 mm thickness of the probe solution. The continuum beam crossed the fundamental and, for the stimulated emission pumping experiments, also the UV beam in the probe cell at an angle of = 12°. The polarization of the 630 nm beam used for continuum generation was rotated to an angle of approx. 54° with respect to the UV beam polarization. As shown in [Bult], continuum generation does not change the polarization, so that there was an angle of 54° between the pump and probe polarizations, the so-called "magic angle" of exact 54.7 ° preventing polarization reorientation effects on the signal

52

As the probe part of the continuum had been reflected by the beam splitter and thereby transversally inverted, the reference part was treated likewise and reflected by an additional mirror. Both parts were collimated with identical achromatic lenses (f=50 mm) and focused (with f=30 mm) on the slit of two home-built Rowland type polychromators [Bing, Laur]. The gratings were from Zeiss, with a line spacing of 248 g/mm, blazed for 225 nm and a radius of 116.31 mm. In the focal plane of each a photodiode array was placed (S3904-512Q from Hamamatsu, 512 pixel, sensitive in the range 200-1000 nm). The intensities of both part of the continuum, I_Probe and I_Ref were recorded as a function of wavelength in a range of 420 nm between 350 and 930 nm. The transmission I_Probe / I_Ref was calculated and averaged over 30-200 laser shots. As every second pump pulse was blocked by the optical chopper, every second transmission value was summed up to give the baseline, that is the transmission without the pump pulse, which was then subtracted from the averaged transmission. Temporal delay between the pump and probe pulses was provided by a delay stage M-013 from Physical Instruments driven by a 5-phase stepper motor C-545 with 0.25 µm position accuracy and step width. It was controlled with a home-interfaced module SMA 05-Z from Ovis. The delay stage was positioned in the beam path of the fundamental behind the beamsplitter that reflected the part necessary to generate the probe continuum, so that the pump and dump pulses were synchronously delayed or rather, made to arrive earlier than the probe pulse. Delay of the fundamental used for continuum generation could cause systematic drifts in the output energy of the continuum, a rise of which might have led to saturation of the diode arrays.

53

3.3. Test, calibration and resolution

To check and find the temporal overlap of the pump- and probe pulses (time-zero), a dye, for example oxazine 750 in ethanol, was placed in a cuvette at the pump and probe beam intersection. Its long-lived stimulated emission can be induced by either 315 nm or 630 nm excitation and is easily found.

Since the stepper motor accuracy is specified as 0.25 µm corresponding to 1.7 fs temporal delay, the temporal resolution of the system was limited by the far larger pulselengths of the probe and the 630 nm pump or dump pulse. It was characterized by the rise of induced transmission of malachite green in ethanol ( Figure 3.3-1 ), which produces a signal over the whole visible spectrum. Its rise was assumed to be a convolution of a step function and the cross-correlation of pump and probe pulses; the width of the latter is then given by the time between 12% and 85% of the signal rise. For an angle of 25° between pump and probe beams, as present in the DCM experiments, the time resolution was approximately 120 fs. After decreasing the angle to about 14°, the time resolution improved to approx. 100 fs for the LDS 750 experiments.

Figure 3.3-1: Malachite green kinetics demonstrating the system temporal resolution. Rise and decay times as obtained from fits to (sums of) exponentials are given in ps.

54

Figure 3.3-2: Spectrum of a mercury lamp recorded with the home-built polychromators.

To ensure that no transient from the solvent complicated the interpretation, the pump energy during measurements of DCM or LDS 750 was reduced to a level where no signal was obtained from the pure solvent (0.5-3 µJ). Stationary absorption spectra were recorded before and after the measurement to check for signs of photodegradation of the probe solution.

As the continuum experiences group velocity dispersion when traversing the cell windows and the solution, the time-zero overlap of pump and probe pulses takes place at different times for different wavelengths. From the electronic response of the pure solvent [Kov 99a], the amount of this 'time-zero dispersion' can be derived. Therefore, measurements in the pure solvents with high excitation pulse energy (5-8 µJ) were also performed.

The main sources of noise were found to be pulse-to-pulse spatial fluctuations of the probe continuum, giving rise to differences in the transmission ratios of the broadband beam splitter, and misalignment of the probe continuum optical path. The probe or the reference continuum must not be cut off at its edge or be guided through the lenses in a way different

55

3.4. Synchronization and electronics

The Nd:YAG laser used to pump the amplification stages as well as the detection set-up had to be synchronized to the laser pulses delivered with 76 MHz by the CPM dye laser. The maximum repetition rate for the amplified pulses was limited by the Nd:YAG laser to 30 Hz. The corresponding trigger scheme is illustrated in Figure 3.4-1 . A photodiode (250 ps risetime, BP 28) monitored the CPM signal, which was then divided to 150 Hz and 30 Hz by a home-built frequency-divider (see annex, Figure A-1). The light chopper was triggered with 150 Hz, and the 30 Hz signal was fed into a multi-channel delay generator (Stanford Research, DG 535). It delivered TTL pulses of various lengths and delays used as trigger for the Nd:YAG flashlamp, Nd:YAG Q-switch pockels cell, the polychromators and the A/D converter card. The delay between the Nd:YAG flashlamp and the Q-switch trigger was optimized for maximum pulse energy of the Nd:YAG pulses to 355 µs. The 30 Hz input signal from the pulse divider was variably delayed so that a femtosecond CPM pulse arrived at the same time at the amplifier stage as the maximum of the 5 ns long Nd:YAG pulse. Control of this was effected by monitoring the amplified spontaneous emission and the CPM pulse reflected from the saturable absorber as mentioned in 3.1.3.

The polychromators were controlled by the unit C4070 from Hamamatsu and home-built control circuits described in detail in [Bing]. These produced a 600 kHz clock impulse and two start impulses, separated by 5 µs, of 100 kHz synchronized to the external trigger for read-out of the photodiode arrays. The photodiode arrays were read out at 200 kHz, the maximum frequency for A/D conversion of the converter card (Meilhaus PC30-PGH). Alternating between the probe and the reference array, intensity values were read into two channels and converted into 12 bit digital values. A trigger signal for the 2 x 512 A/D conversions necessary every laser shot was also delivered by the polychromators' control circuits. Synchronisation of the measurement cycles was provided by gating the timer

56

The commands for the start of the measurement cycle, A/D conversion, data averaging and storage and delay stage control were incorporated in a measurement user interface programmed in TurboPascal 7.0. It was an extension of that used in [Bing] and [Laur] and the converter card was adressed with the help of driver software provided by Meilhaus. This card had also 4 D/A channels and 24 digital in/outputs available for measurement and control purposes; two of the latter were used for delay stage control. For a typical measurement of 5 ps length with 10 fs steps and 150 averages around 45 minutes were needed.

A/D conversion took place in burst mode and was DMA controlled. The option of double-ended A/D conversion, where the shielding cable's voltage level is subtracted from the signal as a reference for each channel, showed to be noisier (by a factor of three) than the simpler single-ended conversion. Shielded BNC cables were employed and ground loops were avoided.

The dynamic range of 4096 counts of the converter card was set to correspond to 10 V. It could not be fully exploited since the photodiodes were already saturated at 8.9 and 7.9 V for the probe and the reference array. The dark current amounted to 170 and 110 counts, respectively, and was subtracted from the measurement. It fluctuated by about 6 counts, which was averaged out.

Two of the 24 digital outputs from the converter card were used to control the stepper motor of the pump pulse delay stage. The level of the first digital output set the direction of the move, and the number of low/high transitions of the second output corresponded to the number of steps to be taken. Owing to 'cross-talk' of the digital output channels, 'low' of any channel was at 0.5 V when one of them was constantly high. The output of the second channel was therefore fed to an additional monoflop, which provided an adequate output signal with the low level at 0 V. To prevent a mechanical hysteresis of the delay stage, moves backwards were realized by a slightly longer move backward than necessary and

57

Figure 3.4-1: Synchronization scheme.

3.5. Data correction

All data presented are corrected for background signal and for dispersion of the temporal zero-position. Stray light from the pump pulse contributes, if present, also for negative time delays and can be easily subtracted from the set of spectra. To obtain the wavelength-dependent delay times of maximum overlap between the excitation or dump pulse and the probe continuum, the changes in optical density were recorded for a pure solvent. The coherent electronic contribution and the signal from impulsive stimulated Raman scattering exhibited for high pump energy by the solvent are centered around the peak of the cross-correlation of the optical pulses [Kov 99a]. Thus the temporal zero-position can be measured as a function of wavelength, with an accuracy of about 20 % of the temporal resolution. The experimental kinetic traces were shifted each by the negative of their zero-

58

When this improved the signal quality, the spectra were smoothed with binominal averaging.

3.6. Chemicals

LDS-750 and DCM were obtained from Exciton and Lambda Physics and used as received. Apart from malachite green (Kodak), all dyes used for short pulse generation or amplification or filtering were from Lambda Physics. All solvents were of spectroscopic grade (Merck Uvasol), apart from propylene carbonate, which was of HPLC grade. DCM and LDS 750 concentrations were about 0.2 mM in all solvents except cyclohexane, where the DCM concentration was below 0.1 mM.

3.7. Photostationary spectra

Stationary absorption spectra were recorded with a Shimadzu UV-3101PC spectrometer. Fluorescence spectra were measured using a SPEX Fluorolog1680 0.22 m double monochromator spectrometer. Photon counts were corrected by comparison with a secondary emission standard to determine the intrinsic fluorescence quantum distribution.

3.8. Molecular beam spectroscopy

The experimental set-up was that described in [Mühl 99a]. DCM was sublimated at 240 °C, and subsequently cooled by adiabatic jet expansion with Neon as buffer gas (1.2 bar). The fluorescence was detected with a monochromator and photomultiplier, blocking the excitation light with a glass filter (GG475 from Schott). The excitation wavelength was then scanned from 423 to 484 nm.

59

3.9. Raman spectroscopy

All Raman measurements were performed at the Bundesanstalt für Materialforschung (BAM), Berlin-Adlershof.

The resonance Raman (RR) spectrum of DCM in methanol was detected in a flow cell and back-scattering geometry by means of a nitrogen-cooled CCD camera, which was part of a double-stage DILOR-XY-Raman spectrometer with a spectral resolution of 4 cm^-1. Excitation was at 457.9 nm of an Ar⁺-laser with a focus size of 70 µm and excitation power between 1 and 1.5 mW. The concentration of DCM in methanol was 0.5 mM. The RR spectrum of DCM in crystallic form was recorded with the same spectrometer using an Olympus-BH2 microscope and a 50x object lens. The excitation power was reduced to 0.01 W, with a focus size of 2 µm and a maximum irradiation time of 10 s.

Non-resonant Raman spectra were recorded with a nitrogen-cooled DTGS detector of type 418-S of a BRUKER-IFS 66 v Fouriertransform Raman spectrometer with a spectral resolution of 1 cm^-1. Excitation was at 1064 nm of a diode-pumped Nd:YAG laser in 180° backscattering geometry, with a laser power of 10 mW for the crystal spectrum and 1000 mW for the spectra of DCM solutions. The beam diameter was 100 µm in solution and 1 mm on the probe surfaces. The solutions were saturated with DCM at room temperature.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.