60

Chapter 4. Results

Overview :

Time-resolved pump-probe experiments and stimulated emission pumping (dump-probe) experiments have been carried out on DCM and LDS 750 in solution. To obtain additional information on DCM properties in the liquid phase, stationary UV/VIS absorption and fluorescence spectra and Raman scattering spectra were measured. The (0,0) transition energy in the gas phase for excitation into the first electronically excited singlet state could be derived from the fluorescence excitation detection of jet-cooled DCM.

In the next two subsections, results of the time-resolved and stationary measurements on DCM will be presented. The following subsections are concerned with semiempirical calculations of isolated DCM and simulations of vibrational and solvent relaxation of DCM solutions.

The results on LDS 750 will be presented at the end of this section.

4.1. Time-resolved spectroscopy

Pump-probe experiments were carried through on DCM in cyclohexane, toluene, tetrachloromethane, chloroform, methanol and acetonitrile for different pump pulse central wavelengths and pump energies. Dump-probe experiments were realized on DCM in methanol, acetonitrile and propylene carbonate with a pump pulse central wavelength of 315 nm and a dump pulse centered at 630 nm. The delay between pump and dump pulse was 100 ps, which should ensure solvent equilibration.

4.1.1. Spectral decomposition

Typical pump-probe spectra of DCM in acetonitrile are shown in Figure 4.1-2 a) for an excitation wavelength of 530 nm and a excitation pulse energy of 0.4 µJ. The pump-induced changes in the optical density (differential optical density) are plotted against the

61

Positive contributions to the differential optical density are observed in the spectral region from 400 to 480 nm, whereas in the region from 480 to 750 nm the signal is negative. The total spectrum consists of an overlap of different spectral contributions from the excitation-induced loss of population (hole) in the electronic ground state and the created population (particle) in the first or higher electronically excited states and their transitions to higher or lower electronic states. By its similarity to the stationary fluorescence spectrum, the large negative band centered at 624 nm for delay times larger than 1 ps is assigned to stimulated emission of the particle in the S₁ state. After spectral relaxation, another negative contribution is visible centered around 500 nm. This is assigned to the bleach of the ground state absorption. The remaining positive part of the signal is assigned to absorption of the particle (excited state absorption). It is evident that for a excitation pulse energy of 0.4 µJ the signal intensity in the latter spectral region remains approximately constant over a delay time range of 100 fs - 2 ps. The same observations and assignments can be made for pump-probe spectra of DCM in acetonitrile and in chloroform.

For excitation conditions with the pump pulse centered near the maximum of the ground state absorption band, as e.g. it is the case with 470 nm excitation of DCM in strongly dipolar solvents (see 2.1.2), the excitation-induced ground state absorption bleach should remain constant in its maximum frequency and experience only spectral broadening in time. Here, it was assumed that the corresponding hole distribution relaxes faster than the instrumental response function, so that the bleach band can be viewed as constant on the timescale of measurement. The constancy of the excited state absorption and ground state absorption bleach bands enable a spectral decomposition as follows. As the fluorescence lifetime of DCM is of the order of nanoseconds in most solvents, after complete spectral relaxation the form of the stimulated emission band should be given by the stationary emission cross section. From the stationary fluorescence spectrum the cross section for emission is calculated [Ernst] and scaled upon the red edge of a pump-probe spectrum for a delay time between several picoseconds for acetonitrile to several tens of picoseconds for methanol. The stimulated emission spectrum estimated this way is subtracted from the measured transient, so that the remaining difference spectrum consists of excited state absorption and bleach of ground state absorption ( Figure 4.1-2 b). These contributions are convoluted in time with the instrumental response function and subtracted from the time-

62

Figure 4.1-2: a) Pump-probe spectra of spectra of DCM in acetonitrile after excitation at 530 nm with 0.4 µJ excitation pulse energy. b) Spectral decomposition of the pump-probe spectra in a). The fluorescence spectrum has been converted to stimulated emission (gain). c) Isolated stimulated emission spectra obtained from the spectral decomposition.

63

Typical dump-probe spectra of DCM in methanol are shown in Figure 4.1-3 a). Their form is the negative of the pump-probe spectra: stimulated emission pumping creates a hole in the excited state population and ‘bleaches’ excited state absorption and stimulated emission, the contributions of which now have respective negative and positive signs. The particle in the ground state contributes additional absorption (positive sign). To ensure that no excitation into higher electronic states by absorption of the dump pulse occurred, pump-probe and dump-probe spectra were compared for delay times after spectral relaxation has been completed ( Figure 4.1-4 ), and coincidence was found.

The spectral decomposition for the dump-probe spectra was performed in a way similar to the spectral decomposition for the pump-probe spectra ( Figure 4.1-3 b). The dump pulse was centered near the peak frequency of the emission spectrum of DCM in strongly dipolar solvents (see 2.1.2), so that the maximum frequency of the stimulated emission bleach should remain constant in time. It was assumed that the bleach contributions relaxed fast and were constant on the timescale of the experiment. The stationary emission cross section was scaled onto the red edge of the dump-probe spectrum after relaxation and subtracted. Further separation of the remaining sum of excited state absorption and ground state absorption was only possible under the assumption of similar transition dipole moments for ground state absorption and emission from S₁. The integrals over the corresponding bands should then be the same [Stri 62]. If the form of the relaxed ground state absorption band is given by the stationary absorption spectrum, its height can be obtained by scaling the part of its area which shows approximate mirror symmetry to the fluorescence spectrum onto the band area of the scaled emission cross section. By subtraction of the thus determined ground state absorption and stimulated emission contributions from the relaxed dump-probe spectrum, the excited state absorption spectrum was obtained ( Figure 4.1-3 b). Together with the stimulated emission band it constituted the hole’s contribution to the experiment. This bleach spectrum was convoluted in time with the instrumental response function and subtracted from the set of measurements to achieve isolation of the time-resolved ground state absorption spectra. It should be noted that a temporary isosbestic point at 537 nm is already observed in the dump-probe spectra.

64

Figure 4.1-3 : a) Dump-probe spectra of spectra of DCM in methanol after stimulated emission pumping at 630 nm. b) Spectral decomposition of the dump-probe spectra in a). The fluorescence spectrum has been converted to stimulated emission (gain).

Figure 4.1-4 : Comparison of pump-probe (dashed) and dump-probe spectra (solid) of DCM in methanol at a delay time of 20 ps.

65

4.1.2. Spectral dynamics

4.1.2.1. Dump-probe experiments

Isolated ground state absorption ("particle") spectra of DCM in methanol are presented in Figure 4.1-5 . Negative contributions to the differential optical density and the spike at 649 nm stem from an imperfect subtraction of the "hole" spectrum and from stray light by the dump pulse, respectively. For delay times within the response function width of the experimental set-up, the absorption spectrum starts to rise at time zero centered above 600 nm and shifts during its rise to approximately 565 nm ( Figure 4.1-5 a). The subsequent spectral relaxation proceeds via the growth of a shoulder on the blue side of the absorption maximum and a decay on its low-energy side up to 400 fs, exhibiting a temporary isosbestic point at 537 nm ( Figure 4.1-5 b). Until about 10 ps, the absorption band moves to the blue ( Figure 4.1-5 c).

Isolated absorption spectra of DCM in acetonitrile are shown in Figure 4.1-6 . The initial spectra around time zero are also centered at about 633 nm and shift during their rise-time to approx. 577 nm ( Figure 4.1-6 a). After 100 fs they still show some structure, which is lost during the next 150 fs. On this timescale, the rise of a blue shifted shoulder and a decay of the low-energy side of the absorption band take place, similar to the spectral evolution in methanol. For acetonitrile the high-energy shoulder is centered around 500 nm, and an isosbestic point is observed at 519 nm. For later delay times up to around 1.5 ps, the isolated absorption spectrum shifts further to the blue and grows slightly ( Figure 4.1-6 c).

For DCM in propylene carbonate the relaxation for 120-300 fs ( Figure 4.1-7 b) parallels that in methanol and acetonitrile, with a decay on the low-energy side of the absorption spectrum, and a rise on the higher energy side. The isosbestic point is found at approx. 558 nm. In the course of the following 15 ps ( Figure 4.1-7 c), the absorption band maximum shifts from approx. 545 nm to the position of the stationary absorption spectrum and decreases in amplitude.

It is interesting that a second, weaker absorption band with maxima at 375 nm and 393 nm in acetonitrile and 377 nm and 398 nm in propylene carbonate remains constant after the initial rise.

66

Figure 4.1-5 : Isolated absorption spectra of DCM in methanol for different delays after stimulated emission pumping at 630 nm (dump wavelength indicated by large arrow in a).

67

Figure 4.1-6: Isolated absorption spectra of DCM in acetonitrile for different delays after stimulated emission pumping at 630 nm (indicated by large arrow in a).

68

Figure 4.1-7 : Isolated absorption spectra of DCM in propylene carbonate for different delays after stimulated emission pumping at 630 nm (indicated by large arrow in a).

69

4.1.2.2. Pump-probe experiments

Isolated stimulated emission spectra of DCM in methanol are shown in Figure 4.1-8 and Figure 4.1-9 a) for an excitation wavelength of 530 nm and an excitation energy of 0.4 µJ. The initial spectra are highly structured ( Figure 4.1-8 ). Over the first 600 fs ( Figure 4.1-9 a), a blue shifted shoulder of the main fluorescence band centered approximately at 530 nm disappears, while the main band around 600 nm grows. For longer delay times up to 8 ps, the emission band exhibits a further slight growth and a spectral red shift (not shown). For comparison, Figure 4.1-9 b) shows the spectral evolution of the emission over the same timescale as in a) but for an excitation pulse energy of 0.8 µJ. Instead of a shoulder, another emission band centered at approx. 505 nm declines in amplitude while the band centered at 630 nm grows. This growth continues with a smaller amplitude accompanied by a red shift until about 10 ps ( Figure 4.1-10 ). Spikes as well as positive spectral contributions remaining constant in the course of time are the consequence of the inexact subtraction procedure to isolate the emission bands and should be ignored.

Figure 4.1-8: Early isolated stimulated emission spectra of DCM in methanol after excitation at 530 nm with 0.4 µJ excitation pulse energy. The excitation wavelength is indicated by an arrow.

70

Figure 4.1-9: Isolated stimulated emission spectra of DCM in methanol for different delays after excitation at 530 nm a) with 0.4 µJ excitation pulse energy, b) with 0.8 µJ exc. pulse energy.

Figure 4.1-10: Isolated stimulated emission spectra of DCM in methanol for different delays on a picosecond timescale after excitation at 530 nm with 0.8 µJ excitation pulse energy.

71

Figure 4.1-11: Isolated stimulated emission spectra of DCM in acetonitrile for different delays after excitation at 470 nm with a) 0.2 µJ excitation pulse energy, b) 0.4 µJ exc. pulse energy, c) 0.9 µJ exc. pulse energy.

72

Figure 4.1-12: Isolated stimulated emission spectra of DCM in acetonitrile for different delays on a picosecond timescale after excitation at 470 nm with excitation energies of 0.4 and 0.9 µJ.

For the less polar solvents investigated (chloroform, toluene, tetrachloromethane and cyclohexane) the separation of the emission bands was not straightforward. Due to a large relative amplitude of the excited state absorption, the scaling of the stationary emission spectrum onto the red edge of the relaxed pump-probe spectrum often gave unsatisfactory

73

Figure 4.1-13: Pump-probe spectra of DCM in chloroform for different delays after excitation at 470 nm : a) with 0.4 µJ excitation pulse energy; the dotted curve indicates the spectrum after 20 ps, b) with 0.8 µJ exc. pulse energy, c) with 0.8 µJ exc. pulse energy on a picosecond timescale.

74

Figure 4.1-14 : Pump-probe spectra of DCM in chloroform for different delays after excitation at 530 nm with 0.7 µJ excitation pulse energy.

Figure 4.1-15 a) and b) give two examples for the time-dependendent changes of the differential optical density of DCM in toluene and for their dependence on excitation

75

Pump-probe spectra of DCM in tetrachloromethane for three different excitation energies and for the first 250 fs are shown in Figure 4.1-17 . The ratio between the two maxima of the excited state absorption bands at 482 and 516 nm changes with increasing excitation pulse energy in favour of the latter. Also the excited state absorption in the region above 545 nm is dominant when compared to the stimulated emission for higher pump energies, yielding positive differential optical densities. These tendencies can also be found for DCM in toluene, and spectral evolution on the short timescale is the same as described for this solvent.

The time-dependent changes of differential optical density in the pump-probe spectra of DCM in cyclohexane parallel those in tetrachloromethane and toluene ( Figure 4.1-18 ), with the difference that a very fast decay component of the high-energy side of the spectrum is not accompanied by a rise in the region above 535 nm.

76

Figure 4.1-15: Pump-probe spectra of DCM in toluene for different delays after excitation at 450 nm with a) 0.7 µJ excitation pulse energy, b) 1.4 µJ exc. pulse energy.

Figure 4.1-16: Pump-probe spectra of DCM in toluene for different delays on a picosecond timescale after excitation at 450 nm with 1.4 µJ excitation pulse energy.

77

Figure 4.1-17: Pump-probe spectra of DCM in tetrachloromethane for different delays after excitation at 450 nm with a) 0.3 µJ excitation pulse energy, b) 0.6 µJ exc. pulse energy, c) 0.9 µJ exc. pulse energy.

78

Figure 4.1-18 : Pump-probe spectra of DCM in cyclohexane for different delays after excitation at 450 nm with 0.8 µJ excitation pulse energy.

4.1.3. Dynamic Stokes shift

The broad emission and absorption bands of chromophores in solution can empirically be described by lognormal line shape functions g ( ) [Sian 69]:

(4.1)

where g₀ is the amplitude, ₀ is the maximum frequency, b is the asymmetry and is the width of the spectral distribution. The lognormal lineshape is essentially that of an asymmetric Gaussian. This function can be made use of to model absorption and emission bands, as long as their shape is not complicated by structure [Mar 87]. It was applied to the isolated time-resolved emission and absorption spectra with nonlinear least-squares fitting, using a Simplex search method. Any (sub)structure of the absorption and emission bands was ignored. The time evolution of the maxima and width of the spectral bands can then be extracted from the function parameters, either directly from ₀ and or by calculating the first and second moment of the distribution from b, ₀ and [Bult].

79

Figure 4.1-19 shows the spectral response function of DCM in acetonitrile (circles), obtained from spectral evolution of the solute in the excited state and in the ground state. Also displayed (straight line) is a fit to the spectral response function from fluorescence upconversion investigations of the relaxation of coumarin 153 in its first singlet excited state by Horng et al. [Horng 95]. There is good agreement between the data from different methods and molecules for delay times from 0.4 ps onwards. For earlier delay times the curves differ, the values for DCM exhibiting a kind of plateau up to 100 fs and declining steeply afterwards. The limited time resolution of approximately 120 fs in the dump-probe and of appr. 60 fs in the pump-probe experiments forbids the interpretation of the very early dynamics.

For the response function after stimulated emission pumping of DCM in propylene carbonate as presented in Figure 4.1-20 , good agreement with the spectral response from coumarin 153 is noted from approx. 0.6 ps onwards.

80

Figure 4.1-19 : Spectral (or solvent) response function S_ÄE(t) of DCM in acetonitrile calculated from the peak frequency ₀ of a) isolated transient ground state absorption spectra (solid and hollow symbols refer to different dump-probe measurements) b) isolated stimulated emission spectra (solid and hollow symbols refer to different pump-probe measurements). Also shown (solid lines) is a fit to the spectral response function of coumarin 153 from fluorescence spectra in acetonitrile [Horng 95].

81

Figure 4.1-20 : Spectral response function S_IE(t) of DCM in propylene carbonate from the peak frequency ₀ of isolated transient ground state absorption spectra. Also shown (solid line) is a fit to the spectral response function of coumarin 153 from fluorescence spectra in propylene carbonate [Horng 95].

The rigid coumarin 153 has been considered an ideal probe molecule for solvent dynamics, although very recently evidence was also given for intramolecular relaxation processes [Mühl 99a]. The deviation manifested in all strongly dipolar solvents for early times could be traced to the fact that [Horng 95] chose an average to calculate S_ÄE(t) from the first moment and the peak frequency of the fluorescence quantum distribution (not converted to cross section), whereas here the peak frequency of the stimulated emission or absorption band was used. It could also be due to intramolecular relaxation of either molecule. The latter hypothesis is supported for DCM by the fact that the early spectral dynamics in the excited state are characterized by the presence of several emission bands, and in the ground state by the rise and decay of shoulders of the structured absorption band. As a consequence, to model the early spectra correctly a sum of several lineshape functions would have to be employed. The good agreement of the spectral response functions of DCM and coumarin 153 for times above 0.4-0.6 ps is interpreted in the sense that solvent dynamics dominate the spectral evolution of both molecules on this timescale.

82

Figure 4.1-21: a) As in Figure 4.1-20 , but for a picosecond timescale. b) Intensity correlation function S_I(t) of dump-probe spectra of DCM in propylene carbonate.

83

4.1.4. Total intensity

The intensity of a spectral band in a pump-probe or dump-probe experiment is given by its integrated amplitude (divided by frequency to correct for the explicit frequency dependence of the absorption and stimulated emission cross section):

(4.2)

It is proportional to the population of the electronic state from which the transition occurs, and to the electronic transition moment [Lip 68, Tom 91], see also 5.1, and is therefore an indicator for state-to-state dynamics. The integral over the complete measured time-dependent spectra was calculated for delay times up to 2 ps, to see whether and how the observed spectral dynamics are reflected in the total intensity evolution.

Some time-dependent areas of the pump-probe spectra for different solvents are diplayed in Figure 4.1-22 b). The integration range was 350-750 nm for the nonpolar solvents (cyclohexane, tetrachloromethane and toluene), and 400-800 nm for all the others. The area decreases by 10-50% in the course of the first 500 fs due to the delayed rise of part of the stimulated emission band for all solvents. The intensity changes are largest in methanol and least pronounced in chloroform.

The curves were fitted to a sum of an instanteneously and a more slowly exponentially rising component with a negative total amplitude, convoluted with the instrumental response function (a Gaussian with a width of 60 fs). The time coefficients for the slow rise are listed in Table 4.1.1 and Table 4.1.2 under "bandintegral / pump-probe", with the amplitude ratio between the two rising components indicated in brackets. Also shown are the excitation energies for the measurements. While the values from different experiments vary over a range of approx. 0.18 ps, the average time coefficient is nearly solvent-independent around 0.2 ps.

For all solvents, the relative amplitude of the slowly rising component increased with excitation pulse energy in consecutive experiments, exemplarily demonstrated in Figure 4.1-23 for methanol.

84

Figure 4.1-22: Total intensity of a) dump-probe b) pump-probe spectra of DCM in different solvents and for several measurements.

Figure 4.1-23 : Total intensity of pump-probe spectra of DCM in methanol (symbols) and fits (solid lines) for different pulse energies (excitation at 470 nm).

85

Figure 4.1-24: Amplitude fraction of exponential decay of the total intensity of pump-probe spectra of DCM in acetonitrile (solid squares) and its time coefficient in ps (hollow circles) on the same axis as a function of excitation pulse energy.

The time-dependent area of the dump-probe spectra integrated over the range of 390-780 nm is displayed in Figure 4.1-22 a) for the solvents acetonitrile, methanol and propylene carbonate. After an initial rise, which for the solvents methanol and propylene carbonate is slower than the response function of the dump-probe set-up (approx. 0.12 ps), the curves exhibit a plateau and a slow decay up to around 0.6 ps for acetonitrile and methanol and up to 0.8 ps for propylene carbonate. The combination of a rise and decay with similar characteristic times rendered it difficult to adapt any functional description, as in this case the time coefficients e.g. for a combination of an exponential rise and exponential decay are highly correlated. Results from nonlinear least-squares fitting of a sum of an instanteneous initial rise and one or several exponential decays are listed in Table 4.1.1 and Table 4.1.2 under "bandintegral / dump-probe". A fast (0.2 - 0.3 ps) and a slower decay component were observed, the time coefficient of which strongly depends on the solvent, in acetonitrile being fastest (0.8 ps) and in propylene carbonate slowest (6.6 ps). The slow decay has been illustrated for propylene carbonate in Figure 4.1-21 b).

86

Table 4.1.1: Time coefficients (in ps) from precursor-successor fits to the transient spectra ("spectra") and fits to their integrated area ("bandintegral") in highly dipolar solvents. Relative amplitude fractions are shown in brackets.

	Methanol	h/µJ	Prop. Carbonate	h/µJ	Acetonitrile	h/µJ
Spectra	0.26	0.4			0.20	0.2
Pump-probe	0.28	0.7			0.22	0.4
	0.28	0.8			0.20	0.4
	0.20	1.1			0.23	0.9
	0.32	1.2
Average	0.27±0.03				0.21±0.02
Spectra	0.33		0.33		0.29
Dump-probe					0.17
Bandintegral	0.21 (0.56)	0.15			0.11 (0.52)	0.2
Pump-probe	0.25 (0.47)	0.4			0.12 (0.40)	0.25
	0.20 (0.45)	0.4			0.14 (0.44)	0.4
	0.21 (0.38)	0.4			0.14 (0.59)	0.4
	0.25 (0.29)	0.7			0.17 (0.53)	0.4
	0.20 (0.73)	0.8			0.13 (0.53)	0.4
	0.17 (0.69)	0.8			0.22 (0.77)	0.8
	0.24 (0.70)	1.1			0.21 (0.91)	0.8
	0.35 (0.68)	1.1			0.18 (1.00)	0.8
	0.24 (0.44)	1.2			0.20 (0.81)	0.9
					0.42 (0.78)	1.1
Average	0.23±0.03				0.19±0.03
Bandintegral	0.21, 3.9 (2:1)		0.58, 2.2 and 6.6		0.3 and 0.8 (9:1)
Dump-probe			(4:1:1)		0.35 and 0.8 (2:1)

87

Table 4.1.2: Time coefficients (in ps) from precursor-successor fits to the transient spectra ("spectra") and fits to their integrated area ("bandintegral") in moderately polar solvents. Relative amplitude fractions are shown in brackets.

	Chloroform	h/µJ	Toluene	h/µJ	CCl4	h/µJ
Spectra	0.34	0.4	0.14	0.4	0.16	0.3
Pump-probe	0.25	0.8	0.22	0.4	0.15	0.9
			0.21	0.9
			0.25	1.4
Average			0.21±0.02
Bandintegral	0.13 (0.42)	0.4	0.11 (0.4)	0.4	0.16 (0.2)	0.3
Pump-probe	0.17 (0.43)	0.7			0.35 (0.7)	0.9
	0.21 (0.32)	0.7
	0.25 (0.56)	0.8
Average	0.19±0.04

4.1.5. Precursor-successor modelling

The considerable changes in the form of the pump-probe and dump-probe spectra, the existence of isosbestic regions, and total area changes indicate population relaxation to be involved in the spectral relaxation of DCM after photoexcitation. The spectral dynamics in strongly dipolar solvents after 0.4 - 0.6 ps are attributed mainly to solvent reorientation (4.1.3). In consequence, the investigation of a possible photoreaction concentrates on the early time window. The time-resolved spectra were modelled by a global fitting procedure assuming a precursor-sucessor relationship between two spectrally distinct species. This kinetic model is rather simple; it was chosen because of its descriptive character rather than for any physical reason, as its parameter reveals information about the relevant timescale of the underlying population changes for each experiment.

Presuming spectral independence of the apparatus function, the bilinear structure of the time-dependent spectra can be used to model them as the product of a matrix containing

88

. (4.3)

Here is the convolution of the instrument response function R(t) with the time evolution E(t) of the species s :

, (4.4)

and is the spectrum of that species :

. (4.5)

The matrix of the residuals depends linearly on the spectral amplitudes :

. (4.6)

For any given matrix they can be determined by multiple linear regression. The nonlinear parameters defining the time evolution may be obtained independently of the amplitudes by a Simplex optimization [Press 92], whereas a gradient-based nonlinear fitting algorithm such as the Levenberg-Marquardt algorithm implicitly takes the gradient dependence on the linear parameters into account.

Corresponding to precursor-sucessor kinetics, the functions were set to be an exponential decay and exponential growth with the same time coefficient . This can be reduced to a constant and an exponential term, yielding two wavelength-dependent amplitudes which have to be combined to the spectrum of the precursor component:

Exemplary fits to the dynamic isolated absorption spectra of DCM in acetonitrile are displayed in Figure 4.1-25 along with the data. Figure 4.1-26 contains the spectra for the precursor and successor components. It should be recognized that for the dump-probe as well as for the pump-probe measurements, the precursor spectra are structured and broader than the successor spectra. A comparison with precursor-successor modelling of the directly measured dump-probe spectra yielded a difference of 0.01 ps in the time coefficient and a spectral difference of the two amplitudes very close to those from fits to the isolated absorption spectra.

89

Figure 4.1-25 : Fit (solid lines) with precursor-successor modelling to isolated transient ground state absorption spectra (squares) of DCM in acetonitrile.

Figure 4.1-26 : Precursor and successor spectra from the analysis of isolated absorption spectra of DCM in acetonitrile of the previous figure.

90

Figure 4.1-27 : a) Fit (solid lines) to isolated stimulated emission spectra of DCM in methanol (squares). b) Pecursor and successor spectra as obtained from analysis in a).

All optimizations were performed limiting the data to delays up to one picosecond, after which the continous spectral shift in strongly dipolar solvents demonstrated in 4.1.2 and 4.1.3 prevents the application of a kinetic model with species characterized by constant spectra. Precursor-successor modelling could also describe the time-dependent isolated emission spectra ( Figure 4.1-27 ).

The time coefficients obtained from the precursor-successor modelling are compared in Table 4.1.1 and Table 4.1.2 for dump-probe and pump-probe measurements to the values

91

In conclusion of section 4.1., the spectral response of DCM in strongly dipolar solvents after photoexcitation to the electronic excited state or to the ground state can be explained by solvent reorientation from 0.4 - 0.6 ps onwards.

For excited state relaxation, population changes of DCM were observed with time coefficients of approx. 0.2 ps from changes in the integrated intensity. The relative amplitude of the intensity changes grows with excitation energy. For acetonitrile, the characteristic time coefficient also increases with excitation energy. Modelling of the time-dependent spectra in the first picosecond with a precursor-successor kinetic scheme yielded a nearly solvent- and excitation energy-independent time coefficient of around 0.23 ( 0.04) ps (averaged over all solvents). In the unpolar solvents, this relaxation of DCM seems to continue on a longer (picosecond) timescale.

The time-dependent intensity of the dump-probe spectra in strongly dipolar solvents exhibits solvent-dependent, multiple relaxation times, but the largest changes are within the first 0.2 - 0.6 ps. The latter are assigned to ground state population relaxation. The transient ground state absorption spectra within the first picosecond were modelled with a precursor-successor kinetic scheme as for the excited state, yielding also a nearly solvent-independent time coefficient of 0.28 ( 0.07) ps (again, averaged over all solvents). The smaller intensity changes in the dump-probe spectra on a picosecond timescale are solvent-specific and follow the spectral response function, indicating transition moment or population changes with solvent reorientation.

92

4.2. Stationary spectroscopy

4.2.1. Absorption and fluorescence characteristics of DCM

To investigate the solvent dependence of the transition strength, the maximum extinction coefficient for the lowest energy optical transition of trans-DCM was determined in solvents of varying polarity. was found to be 30500 ± 3000 l mol^-1 cm^-1 in cyclohexane, 39700 ± 400 l mol^-1 cm^-1 in chloroform and 43300 ± 1300 l mol^-1 cm^-1 in strongly dipolar solvents, like acetonitrile, methanol and dimethyl sulfoxide. The value for given by [Lambda] for DCM in ethanol, 42500 l mol^-1 cm^-1, falls in between the latter two values, as should be expected.

The quantum yield of trans-DCM fluorescence in cyclohexane was determined from a comparison of the relative intensities of the fluorescence bands in methanol and cyclohexane solutions. The solutions were of identical optical density at the excitation wavelength (455 nm) and subjected to identical fluorescence excitation and detection conditions. With a quantum yield of 0.43 in methanol, the fluorescence efficiency of trans-DCM in cyclohexane was found to be only 7.3×10^-3 ± 5×10^-4.

Although the cis-isomers of substituted stilbenes are generally non-fluorescent, cis-DCM was investigated for fluorescence which might lead to unwelcome contributions to the dump-probe experiments. Transient absorption has been reported for cis-DCM in the wavelength range of 400 - 550 nm [Mey 90], so that this transition cannot interfere with trans-DCM stimulated emission pumping at 630 nm.

Solutions of trans-DCM in cyclohexane and chloroform were exposed to 308 nm XeCl excimer laser radiation (80 mJ, 10 Hz). The absorption spectrum was recorded in a 10 mm fused silica cell after every 500 shots (cyclohexane) and 135 shots (chloroform) and is presented for chloroform in Figure 4.2-1 . The spectra show isosbestic points, the spectrum of the photo-generated component agreeing with that published for cis-DCM in solution [Mey 90]. In accordance with the high quantum yield for trans-cis isomerization of DCM in chloroform, only 500 shots were needed to obtain an equilibrium between the trans- and cis-isomer populations. In cyclohexane, no absorption changes were found, obviously due to

93

The irradiated solution of DCM in chloroform was then excited at 375 nm, where the cis-isomer absorbs preferentially, and the fluorescence spectrum was compared to that of non-irradiated solutions ( Figure 4.2-2 ). A weak fluorescence band appears only after irradiation in the region of 400 to 500 nm. The quantum yield for photoproduct fluorescence was estimated from the comparison of the relative intensity of the trans-DCM fluorescence band to the photoproduct fluorescence band to be 5×10^-4. The emission ( Figure 4.2-2 ) is ascribed to the cis-isomer of DCM.

Nonwithstanding the low fluorescence quantum yield, care was taken to check for cis-DCM accumulation by recording stationary absorption spectra of the solution before and after time-resolved measurements, and no evidence for the presence of cis-DCM was found.

Figure 4.2-1: Absorption spectra of irradiated DCM in chloroform. The number of laser shots between subsequent spectra are indicated by arrows.

94

Figure 4.2-2: Fluorescence spectrum of photostationary DCM solution compared to that of a solution maintained in the dark. Excitation was at 375 nm. The new fluorescence band is assigned to cis-DCM.

4.2.2. Jet spectrum of DCM

The spectroscopy of isolated molecules permits information to be gained about their vibronic and electronic transitions, unimpeded by the line broadening and electronic coupling induced by the solvent environment.

Apart from the single peak visible in Figure 4.2-3 , no further vibronic lines could be resolved in the fluorescence excitation spectrum of isolated DCM. The low signal-to-noise ratio for jet-cooled fluorescence excitation is ascribed to a very low fluorescence quantum yield of isolated DCM, coincident with the decrease of trans-DCM fluorescence efficiency with diminishing solvent polarity. The (0,0) transition energy for the lowest electronic transition was estimated from the peak in Figure 4.2-3 to 22416 cm^-1.

Figure 4.2-3 : Fluorescence excitation spectrum of isolated DCM, from [Mühl 99b].

95

4.2.3. Raman spectra

From the lines in resonance Raman (RR) spectra, the vibrational frequencies coupled to the optical transition can be obtained. Fast electronic dephasing presumed, their intensity in the RR spectrum scales with the dimensional equilibrium displacement of the vibrational mode upon excitation and its frequency: I ²² [Mye 87].

Due to the strong fluorescence of DCM, RR spectra could only be realized with excitation around 460 nm. For longer excitation wavelengths within the DCM absorption band, the large fluorescence background obscures any Raman lines. The RR spectrum of DCM in methanol is presented in Figure 4.2-4 a), while the RR spectrum of DCM crystals (after background subtraction) is shown on the same frequency scale in Figure 4.2-4 b).

Figure 4.2-4: Resonance Raman spectra of DCM a) in methanol b) of DCM crystals. For the latter, the fluorescence background has been subtracted.

96

Table 4.2.1 : Resonance Raman frequencies (in cm^-1) for DCM from Stokes resonance Raman spectra in methanol and of DCM crystals. Also shown are relative displacements estimated from the intensities and frequencies of the peaks of the crystal spectrum.

Frequency / cm-1	Frequency / cm-1	rel. Displ.
DCM / Methanol	DCM / Crystal	DCM / Crystal
	1165	0.65
	1180	0.78
1189	1189	0.83
1213	1209	0.97
1308	1304	0.60
1339	1334	0.43
	1357	0.51
1387	1383	0.45
1420	1421	0.56
	1439	0.48
1503	1489	0.91
	1525	0.68
1554	1550	1.0
1603	1598	0.99
1624	1617	0.62
1655	1644	0.61

The RR bands are similar in relative intensity and close in frequency for DCM in crystallic form and dissolved in methanol, proving that indeed the lines in Figure 4.2-4 b) stem from DCM molecules. The frequencies of the RR bands are listed in Table 4.2.1 . As the fluorescence background is very strong, an analysis of the intensities of the RR peaks was not performed for the methanol solution. Values for the relative displacements D of the modes in the crystal spectrum were calculated after I ²² and are also listed in Table 4.2.1 . There is a notable shift towards higher frequencies between crystal and methanol surroundings for the modes around 1495, 1620 and 1650 cm^-1. The latter two can be assigned to either the olefinic C=C stretch or the pyrane ring bond stretch, respectively

97

For trans-stilbene, the RR mode in the first singlet excited state at 1565 cm^-1 has been assigned to the olefinic (ethylenic) C=C stretch [Qian 93]. It has been intensively studied to investigate vibrational relaxation in the excited state [see also Ham 92, Mat 95, Schultz 97]. Following the treatment of Qian et al. [Qian 93], a lower estimate for the vibrational energy relaxation time T₁ of DCM in methanol in the ground state can be deduced from the bandwidth \|[Ggr ]\| of the RR line at 1553 cm^-1. \|[Ggr ]\| was obtained to 17 cm^-1 after background subtraction and fitting to a Gaussian ( Figure 4.2-5 a), well above the frequency resolution of 4 cm^-1. The Gaussian lineshape of the Raman line implies inhomogeneous broadening from slow fluctuations of the solvent polarization not present for the crystal spectrum, where the same line has a Lorentzian shape ( Figure 4.2-5 b). Ignoring the inhomogeneous broadening and pure dephasing (i.e. setting T₂ 2T₁ in equation 2.37), T₂ = (c\|[Ggr ]\|)^-1 yields a lower limit of 312 fs for ground state vibrational energy relaxation. If only the (unknown) linewidth in methanol due to homogeneous broadening was considered, the estimated limit for vibrational energy redistribution would be even larger.

98

Figure 4.2-5: Fits to the profile of the C=C stretch around 1550 cm^-1 in the resonance Raman spectrum of DCM. a) In methanol, shown are the adaptation of a Gaussian (solid) and a Lorentzian (dotted) lineshape. b) In crystallic form, shown is the adaptation of a Lorentzian lineshape.

Table 4.2.2 lists the RR frequencies obtained in [Kov 96] and frequencies extracted from the dump-probe spectra of DCM in methanol and acetonitrile. This was done by assignment of the structure observed in the early spectra to spectral hole burning in analogy to [Kov 96]. The frequencies of the particle (S₀) and hole (S₁) contributions were extracted via calculation of the second derivative spectrum [Pel 94] and calculating the frequency difference of the peaks to the dump frequency. It is evident that the transient spectra do not yield the Raman transitions dominanting the RR measurement. The physical origin of the structure observed in the early spectra should therefore be reconsidered.

99

Table 4.2.2 : Vibrational frequencies obtained from the structure of the early spectra in pump-probe and stimulated emission experiments. Values from the analysis of quantum beats (see 5.1.4) are denoted by asterisks.

	Frequencies	/ cm-1
	Dump-probe	Dump-probe	Pump-probe
Solvent	Methanol	Acetonitrile	Acetonitrile
			3450
			3130
	2337	2586	2560
	1861	1727
	1496
S0			1430
		1209
			1130
	768	772
	460	452
	3464	3869	3460
	2987	2958	2940
	2175	2247	2300
S1	1502	1385
	970	858	1090
	465	454	440
		322	410*
	130		150*
		77	70*

100

Figure 4.2-6: Raman spectrum of DCM crystals.

101

4.2.4. Solvatochromic analysis

Its large Stokes shift, increasing with solvent polarity, renders DCM a particularly suitable molecule for a solvatochromic analysis. Assuming the validity of the continuum model for solvation (see 2.3), the maxima of the absorption and fluorescence spectra of DCM in solvents of various polarities ( Table 4.2.3 ) were plotted against the Debye reaction field factor F ( = (₀ - 1) /( ₀ + 2) - (n² - 1)/(n² + 2) ). These data were modelled with the transition frequency as a function of F, as determined by the solvent nuclear polarization (middle terms in the right-hand side of equations 2.29 and 2.30). The dipole moment µ_e in the excited state, the peak frequency ₀ of the isolated molecule and its Onsager radius R_o were treated as fit parameters. The dipole moment in the electronic ground state µ_g was set to 9 D obtained from the semiempirical calculations (4.3). Since in the middle terms of equ. 2.29, 2.30 depends linearly on F, the three parameters are correlated. From the various sets of parameters yielding similar (F)'s, only µ_e = 21 - 26 Debye went along with physically meaningful Onsager radii for both sets of data. The excited state dipole moment was therefore fixed at 23 D. Good convergence of the fits to the absorption / emission maxima was reached for values of 11.3 Å / 10.6 Å for R_o and 21875 cm^-1 / 17673 cm^-1 for ₀ ( Figure 4.2-7 ).

Figure 4.2-7 : Absorption (circles) and emission peak frequencies (squares) of DCM against the Debye reaction field factor F in different solvents. Also shown is modelling according to a reaction field treatment with µ_g = 9 D and µ_e = 23 D.

102

Table 4.2.3 : Solvent dielectric properties [Horng 95], peak maxima and widths of DCM stationary absorption and fluorescence spectra.

Solvent	0	n	F(0,n)	MaxAbs / nm	MaxFls / nm	FwhmAbs/ cm-1	FwhmFls/ cm-1
Cyclohexane	2.02	1.424	0.0	455	582	4302	5061
2-Meth. Butane	1.83	1.351	0.0	450	574	4395	5483
Toluene	2.38	1.494	0.02	460	571	4136	3471
Tetrachl.meth.	2.25	1.460	0.02	461	549	4170	4182
Chloroform	4.81	1.443	0.29	470	583	4092	2739
THF	7.58	1.405	0.44	462	603	4333	2633
Methanol	32.66	1.327	0.71	467	640	4545	2391
Acetonitrile	35.94	1.342	0.71	463	638	4587	2391
Prop. Carb.	64.92	1.420	0.7	471	643	4373	2361
DMSO	46.45	1.478	0.66	481	658	4335	2295

4.2.5. Simulation of stationary UV/VIS absorption and fluorescence spectra

Stationary absorption and fluorescence spectra of DCM in several solvents of low polarity are presented in Figure 4.2-8 . The fluorescence spectra of trans-DCM in solvents of low polarity show a drastic reduction in width with increasing solvent polarity ( Table 4.2.3 ). A prominent vibrational progression can be observed in the absorption and in the fluorescence spectra. The relative amplitudes of the bands due to this progression also change when increasing the solvent polarity.

103

Figure 4.2-8 : Stationary emission (left) and absorption (right) spectra of DCM in various solvents of low polarity (dots) and modelling (lines) with a progression of an effective harmonic vibrational mode. Fluorescence quantum distributions have been converted to cross sections; all spectra are normalized.

104

Franck-Condon factors giving the probability for any transition (n’, m’’) were obtained by numerically calculating the overlap of the corresponding Hermite polynomials up to values of nine for n and m. The resulting line spectrum was convoluted with a Gaussian lineshape to achieve a simulation of the absorption / emission spectrum. The broadening , that means the width of the Gaussian, as well as the (0,0) transition energy E and the vibrational level spacing were adapted by a nonlinear fitting procedure to the absorption or fluorescence spectra. It should be noted, however, that the effective vibrational frequencies together with the displacement of the vibrational equilibrium position had been presumed for the calculation of the Franck-Condon factors. Therefore, if the vibrational spacing resulting from the fits deviated by more than 70 cm^-1 from the assumed value for the vibrational frequency, the vibrational frequency was changed and the Franck-Condon factors were recalculated. The procedure was thus iterated until coincidence was achieved. Table 4.2.4 lists the values for the displacement (in units of and hence dimensionless), the broadening and the effective vibrational frequencies in the ground and excited states, _ground and _exc, for DCM in different solvents of low polarity. Missing structure of the stationary spectra in strongly dipolar solvents prevented them from being subjected to the same analysis. The fits to the stationary absorption and emission spectra are also shown in Figure 4.2-8 . They reproduce the spectral features well, if somewhat underestimating the contribution from the (0,0) transition, probably due to the contribution of low-frequency modes not accounted for. The dimensionless displacement decreases with solvent polarity from approx. 2.0 for methyl butane to 1.3 (emission) and 1.55 (absorption) for chloroform. The effective vibrational frequencies also differ between solvents, the ground state effective vibrational frequency declining from 1360 to 1150 cm^-1, whereas the excited state effective vibrational frequency shifts with increasing polarity from 1200 to 1270 cm^-1. In the resonance Raman spectrum of DCM, the modes around 1550 cm^-1 and 1210 cm^-1 are dominating. Although this is partly reflected in the above analysis, the

105

Table 4.2.4 : Dimensionless displacement , (0,0)-transition energy E_(0,0), width and frequencies _exc respective _ground of an effective harmonic vibrational mode as obtained from fits to the stationary absorption and fluorescence spectra of DCM.

		Absorption
Solvent		E (0,0) / cm-1	/ cm-1	exc / cm-1
2-Methyl Pentane	1.99	20942	552	1200
2-Methyl Butane	2.02	21104	530	1200
Cyclohexane	1.95	20775	545	1200
Tetrachl.meth.	1.85	20460	552	1270
Toluene	1.7	20410	658	1270
Chloroform	1.55	20112	784	1270
		Fluorescence
		E (0,0) / cm-1	/ cm-1	ground / cm-1
2-Methyl Butane	2.25	20231	572	1360
Cyclohexane	2.05	19805	582	1360
Tetrachl.meth.	1.85	19511	560	1270
Toluene	1.5	18553	672	1270
Chloroform	1.3	17696	719	1150

The difference between the values of the displacement between the absorption and emission can be explained by different coupling strengths of the vibrational mode to the electronic transition due to different geometries in the ground and excited states. This might be a change in bond angles or lengths as well as differences in the planarity of molecular

106

4.3. Semiempirical calculations

To investigate the influence of dimethylamino bond angle rotation on the properties of isolated DCM in the electronic ground state (S₀), semiempirical calculations were carried out. A complete geometry optimization was performed employing the AM1 Hamiltonian and a modification of the Davidson-Fletcher-Powell algorithm for energy minimization from the program package AMPAC 5.0 (Semichem). The resulting geometry of trans-DCM in the ground state was nearly planar, with both methyl groups symmetrically tilted by 17° out of the plane containing the nitrogen atom and the phenyl ring, and the dimethylamino group rotated by approx. 4° around the phenyl-nitrogen bond. The dimethylamino group was then rotated by varying the dihedral angle of the methyl group closer to the oxygen in steps of 10°; all other bond angles and bond lengths were optimized using the Eigenvector Following algorithm. The dimethylamino group rotation angle was calculated afterwards from and the angle between the methyl-nitrogen-carbon planes of the two methyl groups by = + / 2 - 90°. The heat of formation as a function of dimethylamino group torsion ( Figure 4.3-1 ) shows a second, very shallow minimum slightly above = 30° and increases to a maximum for = ± 90°. A rotation of the dimethylamino group by = ± 34° corresponds to a position where one of the methyl groups is in plane with the phenyl ring, leading to an energetic stabilization. The assymmetry in the E() curve is due to the fact that only one methyl group was rotated, while the other was allowed to relax. When the free methyl group is supposed to pass through the plane with the phenyl ring, it tries to stay in plane with it instead. The main and the shallow minimum are energetically spaced by 0.41 kcal / mol, corresponding to 1.43 cm^-1, which is within the accuracy of calculation for the heat of formation. It should be remarked that a slight torsion of the neighbouring phenyl ring was induced by the dimethylamino group rotation. At the energetic minimum, with = 4°, the dipole moment amounts to 9.1 D. For = 90°, it exhibits a minimum of 6.8 D.

107

Figure 4.3-1: Heat of formation of DCM as a function of dimethylamino group rotation.

Transition energies to the next four excited electronic singlet states (S₁ - S₃) were obtained by configuration interaction with 20 electronic configurations to 23700, 29370, 30840 and 33620 cm^-1. The geometry in the S₃ was similar to that in S₀, while in S₁ both methyl groups of the dimethylamino group were tilted by 11.5° symmetrically out of the plane with the nitrogen and the phenyl ring, and trans-DCM in S₂ was found to be planar. Compared to the value of 22416 cm^-1 for the (0,0)-transition energy from the jet spectrum, the transition energy to S₁ is overestimated by nearly 1300 cm^-1. There are further transitions around 40000 cm^-1, and several more between 45000 and 50000 cm^-1 ( Figure 4.3-2 ).

Figure 4.3-2 : Oscillator strength of transitions from trans-DCM electronic ground state. For comparison, the stationary absorption spectrum in 2-methyl pentane is also shown.

108

Figure 4.3-3: Dipole moments of DCM (in Debye) for Franck-Condon and relaxed electronic ground and excited states.

To check for the influence of rotation of other moieties, the dipole moment in S₁ and in S₂ was calculated after geometry optimization with configuration interaction for a fixed 90° rotation of the bonds connecting the pyrane ring or the anilino group to the central ethylenic part or for 90° rotation of the dimethylamino group. For all of these torsions, the dipole moment in S₁ and in S₂ decreased or did not increase by more than one Debye relative to the values obtained in the full optimization.

109

4.4. Simulation of solvation and vibrational relaxation dynamics

Calculations of time-dependent fluorescence spectra were performed based on the theoretical treatment of [Lor 87], where only effects of solvent and intramolecular vibrational relaxation were considered, with the aim of assessing whether these processes suffice to explain the observed spectral dynamics. The solute was described by a two-state model with a single harmonic vibrational mode coupled to the optical transition.

[Lor 87] solved the Liouville equation for the density matrix (t) of the solute-solvent system in the presence of an external electric field :

(4.9)

Here H is the sum of the Hamiltonians governing the nuclear degrees of freedom of the solute and of the solvent in absence of each other, and the interaction of the solute with the nuclear and electronical degrees of freedom of the solvent for the electronic ground and excited state. is the Hamiltonian of the radiation field, and is given by the transition dipole operator multiplied by the electric field E(t). \|[Ggr ]\| is a matrix operator representing vibrational relaxation.

The time-dependent fluorescence spectrum at the excitation frequency ₁ and the emission frequency ₂ is proportional to :

, (4.10)

where the operator denotes the time derivative of the number of photons in the emitted mode. The result for is expressed in dependency of the solvent coordinate U. U is the difference in interaction energy of the solvent nuclear degrees of freedom with the solute in its ground and in the electronically excited state: .

The calculations have been performed using the following expression for :

110

[Lor 87]

(4.11)

E_i denotes the amplitude of the pump- respective emission field. The excitation pulse envelope is assumed to be Gaussian in time with E₁(t) = E₁ exp(-²t²). V_nm is the transition matrix element between vibrational level n in the electronic ground state and m in the excited state, and _eg is the (0,0) absorption frequency of the fully soluted chromophore. P(a) is the probability that the solute occupies vibrational level a in the electronic ground state at thermal equilibrium; it is determined by the Boltzmann factor P(a) = exp (-W_a / k_BT).

_g and _e characterize the inhomogeneous broadening in the ground and excited state of the solute. In the following, _g = _e has been assumed. The values for , the ground and excitated state vibrational frequencies and the transition matrix elements V_ab and V_cd were taken from the simulation of the stationary emission and absorption spectra (see 4.2.5), neglecting homogeneous broadening. The total solvent reorganization energy _e - _g was extracted from the (0,0)-transition energy difference for emission and absorption (4.2.5). For acetonitrile, where the stationary spectra have not been simulated, the difference between the Stokes shift calculated for chloroform from the (0,0) transition energies and the peak difference of the absorption and emission spectra was subtracted from the peak frequency difference for DCM absorption and emission in acetonitrile. The reorganization energy was thus calculated to be 970 cm^-1 for cyclohexane, 2440 cm^-1 for chloroform and 4320 cm^-1 for acetonitrile.

The variables related to the solvation coordinate have been expressed in [Lor 87] as

111

Z(t) = exp (-t / _L). (4.12)

(see 2.3). The longitudinal relaxation time _L of the solvent was set here to the weighted averaged time coefficient from exponential fits to the Stokes shift of coumarin 153 for the respective solvent as presented in [Horng 95], that is 0.25 ps for acetonitrile and 2.08 ps for chloroform.

B and A_e(t) in the notation of [Lor 87] are defined by :

and (4.13) and (4.14)

and were calculated accordingly.

_dd,bb denotes the conditional probability that the solute occupies state d at time t, given that it occupied b at time zero. The conditional probabilities were obtained as in [Lor 87] using the model of Montroll and Shuler for vibrational relaxation of a harmonic oscillator of frequency coupled to a bath of harmonic oscillators of the same frequency (see 2.4). The formula for _dd,bb can easily be obtained by exchanging d for n and b for m in equation 2.41.

Calculations were performed for the solvents cyclohexane, chloroform and acetonitrile with _vr equal to (200 fs)^-1 and for acetonitrile with _vr = (20 fs)^-1. _L for cyclohexane was set to 0.25 ps, deduced from the spectral relaxation of photogenerated thiyl radicals [Loch 99]. The excess energy ₁(_eg was 3620 cm^-1 for all solvents. The pulselength was 50 fs, and the temperature 298 K.

As expected, the vibronic lines undergo a redshift equal to _e - _g and broaden from the excitation pulse-length determined spectral width to . When vibrational relaxation is much faster than solvent reorientation (such as for _vr = (20 fs)^-1, cf. Figure 4.4-1 ), the vibrational structure disappears within the first 100 fs. The subsequent relaxation consists only of the redshift, and the spectral width and form remain unaltered.

112

Figure 4.4-1: Time-dependent fluorescence lineshapes of DCM in acetonitrile after eq. 4.11 for a fast vibrational relaxation rate.

For a smaller vibrational relaxation rate coefficient of (200 fs)^-1, simulations for acetonitrile are presented in Figure 1-1 along with the isolated emission spectra of DCM in that solvent. The decay of a structured emission in the 18000-24000 cm^-1 region and the rise of a lower-frequency, nonstructured emission can be observed in both simulation and experiment. The lower-frequency emission in the simulations experiences a frequency shift about double of that in the empirical spectra, which in the simulations prevents the isosbestic region observed for delay times >300 fs around 17000 cm^-1 in the empirical spectra.

Simulations and isolated emission spectra for DCM in chloroform are shown in Figure 4.4-3 . The early simulated spectra are structured and dominated by vibrational relaxation. The smaller Stokes shift and longer reorientation time of chloroform compared to acetonitrile prevent large broadening of the spectra at intermediate times such as in the acetonitrile simulations. Therefore, the emission spectra in chloroform remain structured up to more than 500 fs. The main dynamic features only reflect those of the experiment if calculations for t > 300 fs are compared to earlier empirical spectra. This could partly be explained by faster solvent reorientation than represented by the average reorientation time.

113

Figure 4.4-2: a) Time-dependent fluorescence lineshapes of DCM in acetonitrile after eq. 4.11 for a vibrational relaxation rate of (200 fs)^-1. Also shown are isolated stimulated emission spectra in the same solvent after 470 nm excitation for an excitation pulse energy of a) 0.4 µJ and b) 0.9 µJ.

114

Figure 4.4-3: a) and b) Time-dependent fluorescence lineshapes of DCM in chloroform after eq. 4.11 for a vibrational relaxation rate of (200 fs)^-1 on different timescales. c) Isolated stimulated emission spectra in the same solvent after 450 nm excitation for an excitation pulse energy of 0.8 µJ.

115

The spectral dynamics in cyclohexane are better reproduced by the simulations ( Figure 4.4-4 ), although the isosbestic region is shifted for about 1000 cm^-1 to higher energies in the calculations. The spectral structure of the early empirical spectra is hidden by noise, so that no comparison of the scale of spectral broadening can be made. The Franck-Condon factors for vibronically excited DCM do not yield the large transition strength initially observed in the region 19000 - 22000 cm^-1, but the spectral amplitudes of the empirical fluorescence spectra are subject to a systematic uncertainty due to the spectral decomposition procedure.

Figure 4.4-4: a) Time-dependent fluorescence lineshapes of DCM in cyclohexane after eq. 4.11. b) Isolated stimulated emission spectra in the same solvent after 450 nm excitation.

116

4.5. LDS 750

Stationary absorption and emission spectra of LDS-750 are displayed in Figure 4.5-1 . The width of the absorption band is larger in acetonitrile than in chloroform (4220 versus 3580 cm^-1), whereas the emission bandwidth is larger for chloroform (1580 versus 1390 cm^-1). With increasing solvent polarity, the absorption spectrum shifts to higher frequencies. The emission spectrum remains nearly unaltered, exhibiting only a small red shift of 150 cm^-1 when going from chloroform to acetonitrile. The large width of the stationary absorption spectra of LDS-750 compared to the smaller width of the fluorescence spectra led to the assumption of several ground state conformers or isomers [Cast 87, Blanch 91, Smith 99]. Excitation of LDS-750 at 280, 332, 562 and 633 nm in acetonitrile did not change the stationary fluorescence spectrum, so that either the conformational relaxation in the excited state is fast or the conformers apart from one are non-fluorescent. The fact that the stationary absorption spectrum changes mainly its frequency position and not its shape argues against a distribution of different ground state conformers. Since they would be likely to show distinct dipole moments in the ground and excited state, their absorption spectra should exhibit different frequency shifts and the width of the total absorption spectrum should change.

117

Figure 4.5-1: Stationary emission (left) and absorption (right) spectra of LDS-750 in chloroform and in acetonitrile.

Pump-probe measurements were carried out on LDS-750 with the pump pulse centered at 630 nm in chloroform, methanol, acetonitrile and propylene carbonate. The spectral dynamics were found independent of the excitation energy in the range between 0.4 and 1.2 µJ. Transient spectra for LDS-750 in acetonitrile are shown in Figure 4.5-2 . Their overall composition from overlap of excited state absorption, ground state bleach and stimulated emission bands is essentially the same as for DCM (see 4.1.1). Due to pronounced changes of the excited state absorption band with delay time, spectral decomposition would be possible only by a global fitting procedure. The overall differential optical density rises for wavelengths higher than 693 nm and decreases below this isosbestic point on a longer timescale of about 600 fs. This is accompanied by the shift of the center of a broad stimulated emission band from about 700 to 723 nm.

In propylene carbonate, the initial emission band centered at 638 nm broadens and in part decays during the first 250 fs ( Figure 4.5-3 a). Its decline continues up to 5 ps, going along with a rise of a lower-frequency emission band. An isosbestic point appears correspondingly around 693 nm. The rising emission is centered at 723 nm; its amplitude changes parallelly to that of the higher-frequency emission from 250 fs to about one picosecond. Afterwards, its increase proceeds up to 11 ps in spite of only small simultaneous changes in the higher-frequency emission amplitude ( Figure 4.5-3 b).

118

Figure 4.5-2: Pump-probe spectra of LDS-750 in acetonitrile after excitation at 630 nm.

Figure 4.5-3: Pump-probe spectra of LDS-750 in propylene carbonate for different delays after excitation at 630 nm.

119

Figure 4.5-4: Pump-probe spectra of LDS-750 in chloroform after excitation at 630 nm.

Figure 4.5-5: Pump-probe spectra of LDS-750 in methanol for different delays after excitation at 630 nm.

120

In methanol, the spectral dynamics ( Figure 4.5-5 ) are similar to the features in chloroform. The initial fast (within instrumental resolution) decay and broadening of a small emission band centered at 635 nm is accompanied by a rise of the excited state absorption. From about 120 fs up to several picoseconds, the broadened emission band decays further in the region 590 - 660 nm, while another emission band rises in the red spectral region. The latter shifts during its rise from approx. 685 nm to a final peak wavelength of 718 nm, leading to the appearance of several consecutive "temporary" isosbestic points.

Instead of a global analysis, which is often deranged by high parameter correlation, fits to selected kinetic traces were performed using sums of exponentials and effectively instanteneous contributions convoluted with the instrumental response function. Results for wavelengths representative of the species associated with the two emission bands are listed in table 4.5 and compared to characteristic time coefficients for solvent relaxation [Horng 95]. For all solvents, the fast disappearance of high-frequency emission is reflected in a large instanteneous or fast decay, whereas a large fraction (> 0.5) of low-frequency emission rises with time coefficients >0.53 ps. This indicates the involvement of more than two states in the dynamics. The time coefficients for fast decay of the high-frequency emission are close to those for inertial solvent motion, whereas the time coefficients of the slower decay / rise component change with solvent as the timescale for diffusive solvent motion. Apart from the values for methanol, the latter agreement is also quantitative.

121

Table 4.5.1: Decay and rise times (in ps) from fits of sums of exponential functions to wavelength traces of LDS 750 in different solvents. Solvent relaxation times from [Horng 95] are also given.

	? / ps
Solvent	Decay 645 nm	Solvent Rel.	Rise 722 nm
Acetonitrile	0.08 (0.86)	0.089 (0.686)	inst. (0.35)
	0.56 (0.14)	0.63 (0.314)	0.53 (0.65)
Methanol	inst. (0.6)	inst. (0.101)	inst. (0.47)
	0.99 (0.25)	0.28 (0.340)	0.66 (0.53)
	7.7 (0.15)	3.2 (0.298)
		15.3 (0.261)
Propylen	0.13 (0.74)	inst. (0.116)	0.19 (0.45)
Carbonate	1.82 (0.26)	0.180 (0.429)	0.99 (0.38)
		2.03 (0.237)	2.6 (0.17)
		6.57 (0.218)
Chloroform	0.48 (0.53)	0.289 (0.356)	inst. (0.3)
	4.4 (0.47)	4.15 (0.644)	4.6 (0.7)

© 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.