During the last two decades coherent two-dimensional (2D) Fourier transform (FT) spectroscopy has developed into a valuable tool for studying complex dynamics in molecular and material systems [1,2]. To date, most 2D FT experiments have employed three femtosecond fields with similar center frequencies ranging from 2D THz to 2D UV [3–6]. The most widely used 2D FT techniques include 2D infrared (IR) and 2D electronic spectroscopies (ES) to probe structural dynamics and energy transfer in complex systems such as liquids, proteins, catalysts, and photosynthetic complexes. An increasing need to directly correlate vibrational and electronic dynamics during ultrafast photochemical processes has motivated the recent development of 2D electronic-vibrational (EV) [7,8] and 2D vibrational-electronic (VE) spectroscopies [9,10]. These 2D FT experiments measure cross-peaks between the vibrational and electronic transition frequencies of a sample. For example, 2D EV (VE) spectra report on the modulation of a(n) vibrational (electronic) transition due to an initial resonant electronic (vibrational) excitation. Two-dimensional EV spectroscopy has been demonstrated using excitation pulses in the visible region (500–800 nm) and a mid-IR probe with a $300{\mathrm{cm}}^{-1}$ bandwidth generated from an optical parametric amplifier (OPA) [7,8]. These experiments have followed structural dynamics on the excited states of laser dyes [7,8], a model carotenoid complex [11], and chlorophyll [12]. Extending the excitation to UV frequencies and the detection to a broader range of mid-IR frequencies enables 2D EV experiments of biological media [13] and proton-coupled electron transfer systems [14,15] to directly explore correlations between UV electronic excitations and a complete set of broad and varied vibrational transitions in a single experiment. In this Letter, we report on the development of a coherent FT 2D EV experimental setup with near-UV (400 nm) excitation and broadband mid-IR (BBIR) detection that spans an octave ($1600–3200{\mathrm{cm}}^{-1}$) in a single 2D experiment. We use an acousto-optic programmable dispersive filter (AOPDF) to overcome inherent challenges with producing phase-stable pump pulse pairs in the UV, and we use a tunable BBIR source [16] as the probe.

Generating phase-stable pump pulse pairs at UV wavelengths for use in a 2D FT experiment becomes difficult, as a given time delay (${\tau }_1$) is a larger percentage of an optical cycle at shorter wavelengths. Experimentalists performing 2D UV spectroscopy have circumvented this problem in the fully noncollinear geometry using diffractive optics [17] and pairwise beam manipulation [18], and in the partially collinear (“pump-probe”) geometry using pulse shapers [19]. The introduction of an AOPDF for the UV using a KDP crystal [20,21] has aided the development of facile 2D UV setups employing AOPDF-based pulse shaping [6] and utilizing phase cycling schemes to improve signal-to-noise and isolate specific third-order signals of interest [22]. We use this AOPDF in our 2D EV setup to produce UV pump pulse pairs with the accurate time delays required for 2D FT experiments.

We use a BBIR probe in our 2D EV setup to measure the couplings of electronic transition frequencies in the UV with a large range of mid-IR frequencies. Typical generation of femtosecond mid-IR pulses relies on difference frequency generation (DFG) of an OPA output to yield tunable mid-IR pulses with $300{\mathrm{cm}}^{-1}$ bandwidths [23]. Recent 2D-IR and IR pump-probe experiments have utilized the large spectral bandwidth of BBIR pulses as probe fields to study hydrogen bonding dynamics in exquisite detail [24–26]. This Letter expands the scope of the experimental implementation of BBIR sources to include 2D EV spectroscopy.

2D EV is a third-order nonlinear spectroscopy where three field-matter interactions result in a signal field. The partially collinear 2D EV experiment, outlined in Figs. 1(a) and 1(b), consists of two collinear, near-UV excitation pulses, with wave vectors ${\mathbf{k}}_{\mathbf{1}}$ and ${\mathbf{k}}_{\mathbf{2}}$, that are generated with a variable coherence time, ${\tau }_1$, using an AOPDF-based pulse shaper. These two pulses are resonant with an electronic transition and can create a population in the sample’s ground or excited electronic state. After a time delay ${\tau }_2$, the noncollinear BBIR probe pulse (${\mathbf{k}}_{\mathbf{3}}$) impinges on the sample and generates the 2D EV signal (${\mathbf{k}}_{\mathbf{sig}}=\pm {\mathbf{k}}_{\mathbf{1}}\mp {\mathbf{k}}_{\mathbf{2}}+{\mathbf{k}}_{\mathbf{3}}$). In the partially collinear geometry, the collinear pumps result in the phase-matched, fully absorptive, third-order signal, $S({\tau }_1,{\omega }_3;{\tau }_2)$, copropagating with the probe pulse (${\tau }_3=0$) which acts as an intrinsic local oscillator for heterodyne detection. The 2D spectrum is parameterized by a fixed ${\tau }_2$; the FT over ${\tau }_1$ yields the excitation frequency, ${\omega }_1$, and the detection frequency, ${\omega }_3$, is obtained by spectral detection.

 

Fig. 1. (a) Optical table schematic; the Ti:sapphire output is split into the broadband mid-IR (BBIR) probe pulse (solid green), an 800 nm characterization pulse (dashed red), and the pump pulse (solid blue). The BBIR is generated via filamentation of the 800 nm fundamental and its second harmonic in gaseous media and compressed with material compensation (2 mm Ge) and a deformable mirror compressor (DMC). The pump pulse is chopped (500 Hz) and shaped in an acousto-optic programmable dispersive filter (AOPDF). The orientation of all beams before focusing to the sample (S) is shown, and filled circles indicate beams that generate 2D EV signal. The signal (Sig, purple) and reference (Ref, dashed green) are both dispersed in a spectrometer (Spec) and detected in a $2\times 64$ HgCdTe-pixel array (MCT) detector. $\lambda /2$, half-wave plate; Pol, polarizer; $2:1$, down-collimating telescope. (b) 2D EV experimental pulse sequence. (c) Spectral interferogram of two pump pulses, ${\mathbf{k}}_{\mathbf{1}}$ and ${\mathbf{k}}_{\mathbf{2}}$, at ${\tau }_1=200\mathrm{fs}$ (solid black), with the pump pulse spectrum overlaid (dashed black). (d) Extracted phase difference between pump pulses, $\mathrm{\Delta }{\phi }_{12}$ (black dots), is plotted for the center frequency (${\omega }_0=\mathrm{25,000}{\mathrm{cm}}^{-1}$) over experimental time, t.

Download Full Size | PDF

The three pulses derive from the output of a Spectra-Physics Spitfire XP Pro Ti:sapphire regenerative amplifier (800 nm, 35 fs, $\sim 3.6\mathrm{W}$, 1 kHz), and the schematic of their generation is shown in Fig. 1(a). To form the pumps, a 215 μJ portion of the 800 nm fundamental undergoes second-harmonic generation (SHG) in a 0.1 mm thick $\beta$-BBO Type I crystal. The resulting 400 nm pulse is isolated using dichroic mirrors, down-collimated and polarization rotated with a half-wave plate and linear polarizer before being amplitude- and phase-shaped in the AOPDF (UV-250-400 Dazzler, Fastlite) to generate ${\mathbf{k}}_{\mathbf{1}}$ and ${\mathbf{k}}_{\mathbf{2}}$ (125 nJ/pulse, 40 fs), with a controlled delay, ${\tau }_1$. The generation of the BBIR probe has been detailed previously [16], and its compression using a deformable mirror (DM) compressor is discussed in detail elsewhere [27]. Briefly, a portion of 2.5 mJ of fundamental undergoes SHG, and the resulting 800 nm and 400 nm pulses are matched in polarization, temporally overlapped, and focused into a sealed, 1.2 m stainless steel cell filled with gaseous media. The selection of the gas and its pressure produce the BBIR pulse (bandwidth $>1000{\mathrm{cm}}^{-1}$, 2–8 μm tunable) which is compressed in a 19-actuator DM compressor with the DM optimized to yield the shortest BBIR pulse [27]. The 2D EV spectra shown in this Letter use air (760 Torr) to generate the BBIR pulse (180 nJ, 21 fs); the pulse is characterized by the XFROG technique using the upconversion of BBIR and a small portion of 800 nm in a 0.1 mm thick Type I lithium niobate crystal. The compressed BBIR pulse is split with a beam splitter (ISP Optics BSP50-CF-50-2) to generate the probe and reference pulses; the probe is routed along a computer-controlled translation stage (Newport, ILS200LM) to adjust ${\tau }_2$.

The three input pulses overlap spatially and temporally at the sample. The pump pulses pass through a half-wave plate and polarizer before being focused to the sample with a UV-fused silica plano–convex focal lens (UV AR coated, $f=300\mathrm{mm}$). The pumps focus through a machined hole in the first of a pair of silver-coated off-axis parabolic mirrors ($f=101.5\mathrm{mm}$) which focuses all other beams at the sample. The pump and probe beams focus at the sample to $1/e^2$ spot sizes of 185 and 110 μm, respectively. The signal (reference) is spectrally dispersed in a 0.190 m Czerny−Turner spectrometer (Triax 190, Horiba Jobin Yvon, 75 grooves/mm grating blazed for 4.65 μm) and focused onto the lower (upper) of two 64-pixel arrays in a liquid ${\mathrm{N}}_2$ cooled HgCdTe array detector (IR-0144, Infrared Systems Development) for single-shot detection. Shot-to-shot normalization of the signal by the reference is performed before subtracting alternating shots (pump on–pump off via chopping the pump at 500 Hz) to remove the innate probe background in a pump-probe geometry. The signal is measured as a change in IR transmission ($\mathrm{\Delta }T$) of the sample following near-UV excitation, and the 2D EV signal is isolated from other third-order pump-probe signals through non-zero baseline subtraction and FT processing [9]. The full BBIR spectrum is stitched together from spectra collected using nine different grating positions and bandpass filters (2.5–5.1 μm or 4.5–8.5 μm) at the spectrometer entrance to eliminate higher diffraction orders from the grating.

It is necessary to measure the phase stability of the pump pulse pairs generated in the AOPDF to assess the experimental precision of ${\tau }_1$. FT spectral interferometry is used to obtain the interpulse phase difference, $\mathrm{\Delta }{\phi }_{12}$, for 12 min, which is the collection time of one 2D EV surface. With ${\tau }_1=200\mathrm{fs}$, a spectral interferogram is obtained every five seconds by collecting 1000 laser shots in an OceanOptics HR2000+ spectrometer (integration $\text{time}=1\mathrm{s}$). Figures 1(c) and 1(d) show the interferogram and the extracted $\mathrm{\Delta }{\phi }_{12}$ for the center frequency, respectively. We measure the standard deviation, $\sigma$, of $\mathrm{\Delta }{\phi }_{12}$ to quantify the phase stability of the pump pulse pair; ${\sigma }_{\mathrm{\Delta }\varphi 12}=0.0082\mathrm{rad}$ for ${\omega }_0$, which corresponds to 0.13% of a 400 nm optical cycle and a timing jitter of $\pm 0.0017\mathrm{fs}$. The average ${\sigma }_{\mathrm{\Delta }\phi 12}$ for the frequencies within the FWHM of the pump pulses is calculated to be 0.0091 rad. Even for small ${\tau }_1$ intervals (e.g., $\mathrm{\Delta }{\tau }_1=0.5\mathrm{fs}$), this jitter is negligible.

To demonstrate the capability of our 2D EV experiment with the BBIR probe, a 2D EV spectrum of a 0.25 mm thick Si wafer is collected at ${\tau }_2=200\mathrm{fs}$; two 2D EV surfaces (1000 shots per ${\tau }_1$) are averaged (Fig. 2, contour). When Si is excited with light of equal or greater energy than its bandgap (1.12 eV, 1107.2 nm), a change in carrier density results from the excitation of valence band electrons to the conduction band. The long-lived (nanoseconds) electron-hole pairs result in a pump-probe delay-dependent reduction of IR transmission. This effect is often used to characterize the duration of ultrafast mid- to far-IR pulses by fitting the decay width as it corresponds to the convolution of the pump and probe pulse durations [28].

 

Fig. 2. 2D EV spectrum of Si at ${\tau }_2=200\mathrm{fs}$. The real part of the FT is plotted as $-\mathrm{\Delta }T$ with $\mathrm{red}=\mathrm{positive}$ ($\text{green}=0$, $\text{blue}=\text{negative}$) contours plotted from $-1$ to 1 every 10%. The highest contour corresponds to a change in absorbance of $+0.05$ OD. Savitzky–Golay filtering is used to remove the effect of atmospheric water absorption in the BBIR spectra. The top and side panels show pump and probe spectra, respectively.

Download Full Size | PDF

 

Fig. 3. 2D EV spectrum of ${\mathrm{K}}_3[\mathrm{Fe}{(\mathrm{CN})}_6]$ in FA plotted as $\mathrm{\Delta }T$ at ${\tau }_2=500fs$. The spectrum is normalized to the absolute maximum value in the 2D plot and plotted from $-1$ to 1; the contour $\text{separation}=10\%$. The maximum contour corresponds to a change in absorbance of $-0.004$ OD. Top panel: 400 nm pump (dashed gray) and UV/Vis absorption (solid black) spectra. Side panel: BBIR probe (dashed gray) and transient-IR (solid black) spectra; 2D EV projection (dotted black).

Download Full Size | PDF

Similarly, we characterize our 2D EV experiment by detecting the change in BBIR transmission due to pumping Si with AOPDF-generated near-UV pulse pairs (400 nm, $\mathrm{FWHM}=10\mathrm{nm}$). A FT over ${\tau }_1$ (0–100 fs, $\mathrm{\Delta }{\tau }_1=0.5\mathrm{fs}$) resolves the near-UV pump frequencies coupled to the BBIR probe frequencies. The contour plot in Fig. 2 shows the real part of the FT of the 2D EV signal. The resultant 2D EV peak extends to all ${\omega }_1$ and ${\omega }_3$ positions with appreciable pump and probe spectral amplitude (Fig. 2, top and side panels). This is expected since Si has a bandgap well below the pump frequencies and a broadly uniform transmission curve in the mid-IR. In effect, Fig. 2 is simply a map of our 2D EV instrument response that shows the full range of excitation and detection frequencies usable in our 2D EV setup.

Notably, significant signal-to-noise in the 2D EV spectrum is obtained, even on the wings, as demonstrated by magnifying by factors of 4 ($1650–1750{\mathrm{cm}}^{-1}$ and $2700–3000{\mathrm{cm}}^{-1}$) and 22 ($1500–1650{\mathrm{cm}}^{-1}$ and $3000–3300{\mathrm{cm}}^{-1}$) on either side of ${\omega }_3$ in the 2D spectrum. More than an octave of mid-IR frequencies is clearly able to be associated with the pump in this 2D EV setup. Thus, the 2D EV experiment in Si represents the wide range of mid-IR frequencies capable of being directly correlated to photo-induced molecular dynamics arising from the pumps in our setup. The incorporation of the BBIR probe in our 2D EV experiment will allow molecular dynamics that display very broad spectral features or involve several well separated spectral signatures to be simultaneously correlated with ground and excited electronic state dynamics.

To demonstrate the viability of our 2D EV setup for probing condensed-phase molecular systems, we study potassium hexacyanoferrate (III) (${\mathrm{K}}_3[\mathrm{Fe}{(\mathrm{CN})}_6]$) dissolved in formamide (FA). The 2D EV spectrum of ${\mathrm{K}}_3[\mathrm{Fe}{(\mathrm{CN})}_6]$ in FA shown in Fig. 3 is obtained by averaging three 2D EV surfaces; 3000 laser shots are collected for each ${\tau }_1$ (scanned over 0-100 fs, $\mathrm{\Delta }{\tau }_1=0.5\mathrm{fs}$). The chemicals were used as received from Sigma Aldrich. The sample was held in a 0.14 mm path length enclosed by 1 mm ${\mathrm{CaF}}_2$ windows (ISP Optics) in a brass flow cell to refresh the irradiated sample volume at a 1 kHz rate. Solvent-subtracted linear spectra gave absorbances of 0.42 OD at $25000{\mathrm{cm}}^{-1}$ and 0.30 OD in the ${\nu }_{\mathrm{CN}}$ stretching region ($2109{\mathrm{cm}}^{-1}$), as measured with a JASCO V-630 spectrometer and a JASCO FT/IR 4100 spectrometer.

The goal is to correlate the ligand-to-metal charge transfer (LMCT), ${\nu }_{\mathrm{LMCT},\max }=23585{\mathrm{cm}}^{-1}$, with the cyanide (CN) stretching vibrations (${\nu }_{\mathrm{CN}}$). Previous work by Zhang et al. studied ${\mathrm{K}}_3[\mathrm{Fe}{(\mathrm{CN})}_6]$ in two different solvents using polarization selective transient-IR spectroscopy to monitor CN vibrations that reflect electron hole delocalization of the photoexcited LMCT state [29]. They identified the population of a secondary LMCT excited state via the appearance of two distinct IR-active ${\nu }_{\mathrm{CN}}$ modes and attributed the vibrational dynamics to localization of the LMCT state on a single CN ligand. However, isolating the relevant solute and solvent degrees of freedom for this hole solvation process requires further investigation.

The 2D EV spectrum in Fig. 3 contains two peaks. The ground state bleach (GSB) is the narrow, positive (red contours, Fig. 3) signal at ${\omega }_3=2109{\mathrm{cm}}^{-1}$ that results from a population of the ground electronic state following two interactions with the near-UV pump pulses and an absorption of the BBIR probe pulse by the ${\nu }_{\mathrm{CN}}$ modes in the sample. The broad excited state absorption (ESA) is the negative (blue contours, Fig. 3) signal shifted lower in energy at ${\omega }_3=2067{\mathrm{cm}}^{-1}$ that results from the weakening of the CN bond in the electronic excited-state. The dashed gray lines in the top and side panels show the pump and probe spectra, respectively; the solid black lines in the top and side panels show the electronic absorbance spectrum and the transient-IR spectrum, respectively; the dotted black line is the integrated projection onto the probe axis obtained by summing horizontally across the 2D EV spectrum.

As expected, the integrated projection of the 2D EV spectrum onto ${\omega }_3$ matches the transient-IR spectrum at ${\tau }_2=500fs$. The transient-IR spectrum is qualitatively similar to results in Ref. [29]. In the spectra reported here, the GSB and ESA peaks are blueshifted by $\sim 10{\mathrm{cm}}^{-1}$ compared to Ref. [29] due to the FA solvent. The third-order signal is found to be linearly dependent on pump power, which insures that no higher order signals are contributing to the observed signal. A full discussion of the amplitudes, positions, and the line shapes of the peaks in the 2D EV spectra at various ${\tau }_2$ delays will be presented in later publications.

In conclusion, we report the development of 2D EV spectroscopy with AOPDF-shaped near-UV pump pulses and an octave-spanning BBIR probe pulse. This Letter extends the reach of the 2D EV technique along both the pump and probe dimensions. We expect that BBIR probing in 2D EV experiments will prove useful for studying coupled electronic and vibrational dynamics in the high frequency and fingerprint region of the vibrational spectrum in the electronic excited state. Future developments include implementing broader and tunable UV sources into the pump dimension to explore the relevant electronic and vibrational coordinates driving electron and proton transfer in solution.