1. Introduction

Two-dimensional photon echo (2DPE) spectroscopy has emerged in the past decade as a powerful tool for uncovering a wide variety of phenomena in condensed phases [1–3]. Experiments carried out in the visible and infrared spectral ranges have transformed the understanding of processes ranging from electronic energy transfer to bond making and breaking in liquids [4–10]. Knowledge of photo-induced dynamics in small molecules and biological systems motivates applications of 2DPE in the 200-300 nm wavelength range [11, 12]. However, measurements conducted in the deep UV are challenged by dispersion management and the suppression of undesired photo-ionization processes. 2DPE experiments reported near 267 nm (i.e., the third harmonic of a Ti:Sapphire laser) are at the present technical frontier [13–18]. Further progress into the deep UV must contend with experimental difficulties that grow steeply as the wavelength becomes shorter.

In this contribution, we describe the development of transient grating (TG) and 2DPE experiments employing 200 nm laser pulses. This work leverages our recent construction of a four-wave mixing spectrometer operational near 267 nm [16]. Several technical issues relevant to the present study were addressed using this apparatus. Here, we begin by characterizing 60 fs, 200 nm laser pulses generated through filamentation of laser beams in both air and argon. The 200 nm light intensities, bandwidths, and pulse durations are compared in the two gases. It is shown that TG and 2DPE experiments are readily carried out at 200 nm with a (passively phase-stabilized) diffractive optic-based four-wave mixing interferometer [19]. These proof-of-principle experiments, which examine the DNA nucleoside adenosine, are performed in a two-color configuration, where 200 nm light is used to probe the dynamics initiated by a pair of 267 nm laser pulses.

2. Generation and compression of 200 nm laser pulses

One of the primary challenges facing nonlinear spectroscopies at 200 nm is the attainment of sufficiently short laser pulses. Motivations for using gases as nonlinear media in deep UV laser pulse generation have been discussed in earlier work [20–23]. Nonetheless, it is useful to briefly consider why nonlinear optical crystals such as BBO are not well-suited for our applications. For example, the phase matching bandwidth associated with 200 nm sum-frequency generation in a 30 μm thick BBO crystal is approximately 240 cm⁻¹ with incident 800 nm and 267 nm laser pulses [24]. A 30 μm thick BBO is chosen for illustration because this would result in the greatest conversion efficiency for the 90 fs, 800 nm pulses available in our laboratory. The key issue is that the bandwidth of the 200 nm pulse does not exceed the bandwidths of the incident pulses. By contrast, it has been demonstrated that bandwidths exceeding 1500 cm⁻¹ can be produced at 200 nm in gaseous media [22, 23]. Moreover, as will be shown below, the bandwidth achieved at 200 nm is actually greater than the bandwidths of the incident laser pulses because of self-phase modulation in the gas. The bandwidth we obtain at 200 nm is more than two times larger than the bandwidth of the 800 nm fundamental pulse produced by our laser system.

In earlier work, efficient generation of 200 nm light was achieved by confining the nonlinearity to a hollow-core waveguide filled with argon [20, 21]. Our initial attempts utilized this approach, but we were unable to easily obtain high-quality laser beams with the requisite pulse energies (>50 nJ) without damaging the waveguide. Therefore, we instead use filamentation in a pressurized cell to produce 200 nm laser pulses with energies exceeding 100 nJ. The experimental apparatus used for pulse generation and characterization is depicted in Fig. 1 . In this setup, 800 nm and 400 nm laser pulses with 90-100 fs durations are focused into a 30 cm long gas cell, which is filled with either air or argon. The focal lengths and incoming pulse energies, which are given in the figure, are chosen based on the conversion efficiency and beam quality at 200 nm. Within the filament, 200 nm light is generated through a cascade of four-wave mixing processes [22, 23]. To begin, the third-harmonic is produced by way of the nonlinearity, $k_{267}=2k_{400}-k_{800}$, where the subscripts denote the laser wavelengths. The 267 nm pulse then feeds a second process, $k_{200}=2k_{267}-k_{400}$, that yields 200 nm light.

 

Fig. 1 (a) Setup used for 200 nm pulse generation and compression. (b) Diffractive optic-based interferometer used for TG and 2DPE measurements. The 200 nm laser pulse probes the holographic grating induced in the sample by the pair of 20 fs, 267 nm pulses.

Download Full Size | PDF

The 200 nm laser beam is collimated, sent through a prism compressor, then undergoes only two reflections on aluminum-coated mirrors before it enters the diffractive optic-based interferometer shown in Fig. 1(b). This interferometer, which was designed for operation near 267 nm [15, 16], is quite lossy at 200 nm. The pulse energy at the sample position is only 2 nJ (98% loss in interferometer). For this reason, one-color, 200 nm four-wave mixing experiments are not presently possible. Still, TG signals are readily detected when a pair of 20 fs, 267 nm laser pulses is used for optical gating. These 267 nm laser pulses are derived from a hollow-core fiber setup described elsewhere [15].

The 200 nm output of the filamentation process is summarized in Fig. 2 . Similar laser spectra are generated in argon and air; however, the pulse energy is 30% larger in argon than it is in air near 760 Torr. The spectral widths increase slightly with pressure between 380 Torr and 1140 Torr. A spectral width of roughly 500 cm⁻¹ is measured at 760 Torr, which corresponds to a Fourier transform-limited electric field duration of 42 fs. Notably, this 500 cm⁻¹ spectral width is more than two times larger than the widths of the incident 800 nm and 400 nm laser pulses. As mentioned above, one of the primary advantages of using gaseous nonlinear media is that self-phase modulation enables the production of 200 nm pulses that are significantly shorter than those used to drive the nonlinearity.

 

Fig. 2 (a) Spectra of 200 nm light generated in air and (b) argon at the pressures indicated in the figure legends. (c) Pulse energies measured at 200 nm in air and argon. (d) Spatial profile of 200 nm laser beam derived by processing a photograph with the image utility in Matlab. (e) Photographs of filaments generated with individual 800 nm and 400 nm laser beams are shown below the filament obtained when both beams are overlapped. These filaments were produced in air at atmospheric pressure.

Download Full Size | PDF

The spatial quality of the laser beam is quite good, although it is slightly elliptical (see Fig. 2(d)). The width is roughly 7% greater in the vertical dimension than it is in the horizontal dimension. At the exit of the prism compressor, the beam is roughly 1 mm in diameter and it diverges negligibly on the remainder of its 1 meter path to the diffractive optic. We also note that the spatial quality degrades considerably at argon pressures above 1000 Torr. Figure 2(e) shows that the filament produced by the overlapped beams is displaced by roughly 5 mm with respect to those generated by the individual beams. The combined filament is also nearly twice as large as those corresponding to the individual beams. Filaments produced by individual laser beams are known to behave similarly with variation in the light intensity [25].

TG spectrograms obtained with three different prism compressor configurations are compared in Fig. 3 . The interval, $T$, is the delay between the time-coincident 267 nm pulse-pair (i.e., the gate pulses) and the 200 nm pulse. The prism compressor compensates for the ~1160 fs² group delay dispersion accumulated in the exit window of the pressurized cell (3 mm CaF₂) and the diffractive optic (1 mm fused silica). In each measurement, the prisms are inserted to minimize the amount of second-order dispersion at 200 nm, whereas the amount of third-order dispersion increases with the prism separation. Third-order dispersion dominates the phase with fused silica prisms separated by 10 cm. Some improvement is observed with a fused silica prism separation of 6.3 cm, but higher-order dispersion is still apparent in the spectrogram. The best result is obtained with CaF₂ prisms, which impart less third-order dispersion than fused silica prisms near 200 nm. While these measurements are useful for evaluating the sensitivity of the pulse width to the type of glass and prism separation, an accurate determination of the pulse duration requires use of a thinner medium. The spectrograms in the top row of Fig. 3 are temporally broadened by group velocity mismatch (GVM) between the 267 nm and 200 nm pulses, which walk off by approximately 108 fs in the estimated path length of 100 μm.

 

Fig. 3 (Top Row) TG spectrograms acquired with 20 fs, 267 nm gate pulses in a 250 μm thick fused silica window. Fused silica prisms are separated by (a) 10 cm and (b) 6.3 cm. (c) Calcium fluoride prisms are separated by 10 cm. These 200 nm pulses are generated in air at atmospheric pressure. (Bottom Row) TG spectrograms acquired with 20 fs, 267 nm gate pulses in a 50 μm thick BBO crystal. (d) Measurements conducted with 200 nm pulses generated in (d) air and (e) argon. (f) Pulse durations at 200 nm derived from wavelength-integrated TG signals obtained with a 50 μm thick BBO crystal.

Download Full Size | PDF

TG spectrograms measured with 200 nm pulses generated in air and argon are compared in the bottom row of Fig. 3. In these data, a 50 um thick BBO is used as the nonlinear medium to suppress effects of GVM. The 267 nm and 200 nm pulses walk off by approximately 43 fs when the electric field polarizations are aligned to the ordinary axis of the crystal. Our numerical simulations estimate that the wavelength-integrated temporal width of each spectrogram is broadened by a factor of 1.25 compared to a hypothetical medium with no GVM. The spectral bandwidth measured in argon is slightly larger (see Fig. 1); however, the pulses (apparently) cannot be compressed to shorter durations because of higher-order dispersion. As shown in Fig. 3(f), the pulse durations range from 56 to 63 fs depending on the pressure and the gas. The time-bandwidth products are near 0.65 if Gaussian pulse envelopes are assumed. Of course, the quality of the compression can be improved by using both gratings and prisms in the compressor [26]. This will be possible if the optics are upgraded for 200 nm light. The present amount of loss in the interferometer prevents the introduction of gratings in the compression scheme.

3. Transient grating and photon echo spectroscopies

TG and 2DPE spectroscopies are demonstrated in this section using the experimental setup shown in Fig. 1. For these measurements, signal detection by spectral interferometry is accomplished using the second 200 nm laser beam produced at the diffractive optic as a reference field [19]. The sample consists of a 100 μm thick jet of adenosine in aqueous solution (OD = 0.5) [27]. Adenosine is chosen as a model system because its response to photo-excitation at 267 nm has been characterized in previous work [28]. The present experiments are conducted in a two-color configuration in which light absorption at 267 nm precedes emission at 200 nm. Figure 4 shows that the two laser pulses are resonant with separate electronic states. These electronic resonances are primarily localized on the base (not deoxyribose ring), and thus are also found in the adenine nucleobase.

 

Fig. 4 Experiments conducted on adenosine in aqueous solution at pH = 7. (a) Spectra of laser pulses overlaid on absorbance spectrum of adenosine. (b) Real (absorptive) parts of TG signals acquired with 267 nm pump pulses and either 267 nm (red) or 200 nm (blue) probe pulses. (c) Signal detected at 200 nm plotted with respect to the coherence time, $\tau$. Absolute value of rephasing 2DPE signals acquired at (d) $T$ = 250 fs, (e) $T$ = 500 fs, and (f) $T$ = 1000 fs.

Download Full Size | PDF

In adenosine, internal conversion between the photo-excited ππ* state and the ground state takes place in less than 500 fs [28]. Several picoseconds are then required for the solute, which is in a “hot” ground state following internal conversion, to transfer more than 4 eV of excess vibrational energy into the surrounding solvent. It has been shown that vibrational cooling governs the decay of the ground state bleach nonlinearity near 267 nm [28]. Because the resonance at 200 nm shares a common ground state, its bleach recovery should report on essentially the same dynamics observed at 267 nm. We test this prediction by comparing measurements that utilize 267 nm and 200 nm probe pulses in Fig. 4(b). The similarity in the two transients suggests that the desired nonlinearity is detected (i.e., photo-ionization processes do not dominate the signal), which is consistent with the relatively low 0.3 GW/cm² peak power of the 200 nm pulse [15]. We also remark that the comparable signal qualities found with 267 nm and 200 nm probe pulses indicate similar amounts of noise in the signal phase; the real (absorptive) signal component is plotted, not the absolute value.

2DPE signals are acquired by scanning the delay between the two 267 nm pulses, $\tau$, with sub-cycle steps of 0.16 fs. The signal detected at 200 nm with $T$ = 250 fs, which is displayed in Fig. 4(c), reflects the waveform associated with absorption of 267 nm light. The signal oscillates at the period of the ππ* resonance (~0.9 fs), whereas macroscopic dephasing controls the shape of the decay envelope. The signal is Fourier transformed at each detection wavelength, ${\lambda }_t$ (i.e., pixel on CCD detector), to obtain the 2DPE spectra shown in the bottom row of Fig. 4. These spectra represent the absolute value of the 2DPE rephasing pulse sequence.

In essence, 2DPE spectra correlate absorption and emission frequencies as a function of the delay between these two events, $T$. The signals reported here reflect correlations between two different electronic resonances. That is, the solute evolves in coherences between different pairs of electronic states during the delays, $\tau$ and $t$. Correlations in $\tau$ and $t$ can generally be exposed when the laser bandwidth is greater than the homogeneous line widths of the electronic resonances in the solute [29]. It is not necessary for the bandwidth to exceed the total absorbance line width, which is generally dominated by macroscopic (inhomogeneous) dephasing in a room temperature liquid. Thus, the present 500 cm⁻¹ bandwidth at 200 nm should be sufficient for resolving some dephasing processes in aqueous solutions. Solutes in non-polar solvents may also be well-suited for measurements at 200 nm because their line widths are generally narrower.

A preliminary inspection of the 2DPE signals in Fig. 4 suggests that the line shape changes fairly little with $T$, which is consistent with earlier 267 nm, one-color experiments at ambient temperatures [15–17]. Notably, these spectra are measured at delay times, $T$, for which the three incoming laser pulses are not overlapped in the sample; 2DPE spectra cannot be reliably measured during pulse overlap because the quasi-instantaneous, off-resonant response of the solution dominates the signal [15, 16]. One prominent change observed in the 2DPE spectra with increasing $T$ is an increase in the signal amplitude at ${\lambda }_t$<199.5 nm. We tentatively assign these dynamics to solute-to-solvent vibrational energy transfer because of the agreement in time scales [28]. Previous studies of vibrational cooling in DNA nucleosides and other systems suggest that the “hot” ground state wavepacket gives rise to a signal component that shifts towards shorter wavelengths with increasing $T$ [28, 30]. This spectroscopic signature of vibrational cooling is consistent with our data.

4. Conclusions

Our development of TG and 2DPE experiments employing 200 nm laser pulses is motivated by studies of elementary relaxation processes in small molecules and biological systems. In this work, pulse durations near 60 fs have been achieved at 200 nm using filaments produced in both air and argon. To our knowledge, time-resolved spectroscopies in solution involving shorter 200 nm laser pulses have never been reported. As discussed in related work at 267 nm, diffractive optics are useful in these applications because they facilitate sensitive interferometeric signal detection and, in turn, the use of low laser fluence [14–16]. Our experimental setup is already capable of conducting high-quality TG experiments with sub-100 fs time resolution. Thus, studies of a wide range of photochemical processes in small molecules are now possible.

For 2DPE spectroscopies at 200 nm, it will be important to increase the laser bandwidth and implement more robust methods of dispersion management. A previous study shows that broader bandwidths can be obtained with shorter incoming 800 nm and 400 nm laser pulses [22]. We envision that compression to the desired 10-20 fs pulse durations will be possible with greater laser fluences because a compressor based on both gratings and prisms can then be employed [26]. In our setup, the available laser fluence can be increased by a factor of 15 if the present aluminum-coated mirrors are replaced with dielectric-coated mirrors. If necessary, it is also possible to increase the amount of laser power used to generate the 200 nm pulses.