A Broadband Ultrafast Transient Absorption Spectrometer Covering the Range from Near-Infrared (NIR) down to Green

Abstract

We present a new development for pump–probe absorption spectroscopy that allows the simultaneous measurement from the green part of the visible spectrum (510 nm) over the whole near-infrared range to >1600 nm, corresponding to 0.77-2.40 eV. The system is based on a sub-picosecond supercontinuum generated in bulk material used as a broadband probe that is dispersed with a custom-made prism spectrometer and detected by an InGaAs array with extended sensitivity to the visible. Two versions, with and without probe referencing, are implemented for operation at laser repetition rates of a few hertz and kilohertz, respectively. After presentation of the optical configuration of the spectrometer, its performance is characterized and further illustrated on two time scales, with the ultrafast radiolysis of isopropanol induced by a picosecond electron pulse and with the instantaneous response of a BK7 plate to a femtosecond light pulse. The photophysics of the dye IR-140 is resolved from the femto- to picosecond regime. Stable and easy day-to-day routine use of the spectrometer also can be achieved in non-optical laboratory surroundings. For operation in a hazardous environment, the optical probe beams can be transported to the detector unit by optical fibers.

Keywords

Near-infrared spectroscopy NIR Pump–probe spectroscopy Transient absorption spectroscopy Time-resolved spectroscopy Ultrafast kinetics Femtochemistry Radiolysis Ultrafast optics Supercontinuum Multichannel detection

INTRODUCTION

Pump–probe spectroscopy is a versatile tool for the study of ultrafast processes in physics, chemistry, and biology.^1–4 These processes are usually triggered by an ultrashort light or particle pulse and can be studied from the very earliest steps of physical or chemical relaxation on the femto- and picosecond scale up to a few nanoseconds. Longer time scales are made accessible by flash photolysis techniques using in the traditional approach a (quasi-) continuous probe beam that is recorded with a fast photodetector.^5,6 The acquisition of absorbance changes in the ultraviolet (UV), visible (Vis) to near-infrared (NIR) induced in the sample under study reveals dynamics or kinetics of electronically excited as well as ground states of transient species. For ultrafast transient absorption spectroscopy, a short optical probe pulse is delayed relative to the pump pulse by means of a retroreflector mounted on a mechanical translation stage. The highest temporal resolution, on the scale of few tens of femtoseconds, can be achieved with temporally compressed probe pulses, e.g., delivered from noncollinear optical parametric amplification (OPA) pumped by the laser that generates also the pump pulse or that is at last synchronized with it.^7,8 However, the corresponding two-color pump–probe experiments do not directly reveal the spectral evolution that is often helpful to identify and to characterize transient species or relaxation processes as vibrational cooling and solvation.

A broadband femtosecond transient absorption spectrometer can be made with a supercontinuum probe generated by an ultrashort laser.⁹ In combination with a polychromator and multichannel detection, the temporal evolution of ultrafast processes can be recorded in a reasonable acquisition time with a resolution well below 100 fs and sensitivity better than milli-absorbance units (milliunits of optical density (mOD)) over a large spectral band.^10–13 The latter is in general limited by the supercontinuum power distribution, the detector's sensitivity, and dynamic range. Line detectors based on Si as the photoactive material typically exhibit significant response over the range from the UV to 1100 nm; in practice, the spectral operation range of transient spectrometers is further restricted at the fundamental wavelength (~800 nm) of the Ti:Sa-amplified laser source frequently used for pump and probe generation. After continuum generation, this wavelength region still contains the major part of the overall laser energy. Shortpass filters are often used to avoid saturation of the detector, giving rise to an effective spectral bandwidth from about 300 to 750 nm limited by the probe light level. Recently, a novel probe setup was implemented using pulses at 1180 nm for continuum generation to overcome this limitation.¹⁴ These NIR pulses were produced by OPA of a supercontinuum generated by a part of the Ti:Sa amplifier energy; with this approach, a gap-free probe range from 415 to 1150 nm could be obtained. On the other side, towards longer NIR wavelengths, InGaAs diode arrays enable probing in the range from about 900 to 1700 or 2000 nm, respectively. Femtosecond transient spectrometers based on these detectors are in use, covering the range from 1000 to 2000 nm or from 850 to 1800 nm.^15,16 Whereas the first approach runs a multifilament continuum directly with the laser source, the latter uses OPA pulses at 1900 nm to generate a continuum whose anti-Stokes side, with its known large spectral broadening, is used as probe.

We describe here a new and easy-to-run broadband transient absorption spectrometer that allows the simultaneous measurement over the spectral range from 510 to >1600 nm. Based on a state-of-the-art femtosecond laser source, the ultrabroad probe continuum can be generated directly without additional nonlinear light conversion processes. Selection of an adapted band-block filter and adjustment of the continuum generation allow use of the probe spectrum without a gap at the fundamental laser wavelength. A prism-based polychromator guarantees uniform high transmission over the large spectral range without the need for higher order filtering. In combination with a new InGaAs linear image sensor, the spectra can be read over the large range down to the visible at a kilohertz rate, with a high signal-to-noise ratio. Besides the system adapted for kilohertz operation using no reference beam (Fig. 1a), a second configuration is implemented for operation with low repetition rate sources in hazardous environments (Fig. 1b). The latter version is extended for single-shot referencing and transport of the probe beams to the detector unit by optical fibers.

Fig. 1.

Scheme of the ultrafast broadband spectrometer using a supercontinuum probe generated in a YAG crystal, a prism-based polychromator, and an InGaAs array as photodetector. Configuration for (a) kHz operation and (b) for low repetition sources in hazardous environments with single-shot referencing and probe beam transport to the detector unit by optical fibers. VDF, variable neutral density filter; CM, concave mirror; L, achromatic lens; F1, filter RG850; F2, filter KG2.

SETUP OF THE TRANSIENT ABSORPTION SPECTROMETER

The ultrafast absorption spectrometer is configured on a regenerative Ti:Sa amplifier system that delivers 2 mJ pulses around 780 nm with a pulse duration of about 100 fs at a repetition rate up to 1 kHz. Without restriction to the basic performance, the described principles can be applied to other ultrafast, state-of-the-art laser sources emitting in the NIR. Thus, spectral broadening of several hundreds of nanometers can be achieved in bulk material using different ultrafast laser sources with varying pulse duration, wavelength, and repetition rate (e.g., Bradler et al.¹⁷). These supercontinua are known to exhibit a smooth energy distribution over a large spectral range with good temporal coherence and high stability at reasonable energy conversion efficiency. Common materials are sapphire and calcium fluoride crystals for generation of continua applied both as seed for (non)collinear OPA^7,8,18,19 and as broadband probe light.^{10–1320–23} The first material shows excellent stability against photodegradation, whereas the latter material shows better spectral coverage down to UVB. With a reasonably flat intensity distribution and good energy density on the level of 10 pJ/nm, it is particularly the anti-Stokes side that is used for the above-mentioned applications. In contrast, both types of continua exhibit an exponential decrease of intensity by almost a factor of 10 over 100 nm on the Stokes side. This strongly limits the use of the infrared part of continua generated in these bulk materials with common NIR laser sources. Here, we have chosen a single crystal of yttrium aluminum garnet (YAG) for continuum generation optimized for NIR output. About 1 μJ of the laser beam is split off, and its central part (selected by an aperture) is focused weakly in a 4 mm thick disk of YAG. The focusing is realized at a low numeric aperture of ≤0.02, as proposed previously,^24,25 to avoid multiphoton ionization and optical breakdown before the effective buildup of a filament needed for efficient wavelength conversion.²⁶ Increasing the focal length enhances the spectral coverage of the anti-Stokes side of the sapphire continuum towards the UV by about 0.2 eV.²⁷ These primary findings were recently confirmed by an experimental study on continuum generation including the Stokes side that also revealed enhanced results on the NIR broadening for decreasing numerical aperture.¹⁷ The cited work exhibits optimized NIR coverage for YAG of minimum 4 mm thickness in a series of five selected optical crystals. After continuum generation, the region of the fundamental laser wavelength with a full width half-maximum of 10 nm still contains >95% of the overall energy. This part must be strongly reduced to avoid a possible, significant photoexcitation of the sample and saturation of the detector. As the Vis and NIR parts of the continuum cover about 1000 nm, the ratio of spectral energy density relative to the laser pump wavelength is typically lower than 0.0005. We therefore use a notch filter (Edmund Optics 67-112) with an OD of 4 at 785 nm to adapt the pump energy to the level of the continuum. Its full width at half-maximum of 40 nm is also well suited to reduce the highly intense region next to the pump wavelength that is mainly due to self-phase modulation.

The continuum is operated as single filament to guarantee homogenous, symmetric beam quality and spatial stability and to avoid time-consuming day-to-day beam alignment. Its intensity per wavelength is not arbitrarily scalable with the input energy and inherently limited to the level of several picojoules per nanometer (approximately several 10⁷ photons/nm) in the NIR, decreasing rapidly above 1200 nm by more than an order of magnitude. The corresponding detected signal for the spectrometer dispersion of typically 5 nm per pixel toward the infrared is below the saturation charge of the InGaAs array (112 × 10⁶ electrons) with its quantum efficiency of around 0.8 in the NIR up to 1600 nm. Therefore, to obtain the best spectral coverage toward the infrared, the losses on the optical path to the detector unit must be limited. For the beam steering, mirrors with protected silver coatings are used. They exhibit high reflectivity of at least 0.96 over the whole spectral range of interest. Beam focusing and collimation is also performed with reflecting optics up to the polychromator to minimize chromatic aberrations and chirp over the large spectral range. Spherical mirrors are used under low angle of incidence in a folded beam-steering geometry to minimize the astigmatism. As shown recently,^12,22,27 probe light referencing can be omitted for laser sources with high correlation between successive pulses. However, for systems with low pump repetition rates as described in the section, Radiolysis of Isopropanol, a reference beam that accounts for the shot-to-shot fluctuations of the probe light is necessary to obtain reasonable measurement sensitivity for the necessarily limited data averaging. To split off the reference beam, a broadband beam splitter (metallic type neutral density filter, LOT-Oriel 040FN46-25) is used to reflect the signal beam toward the sample and transmit the reference beam. Thus, no additional chirp is added to the signal continuum. The optical layout around the sample, i.e., the pump and probe geometry, can be configured as in state-of-the-art femtosecond transient absorption spectrometers.^10,12 Of course, the established literature concepts of optical flow cells and jets that consider the pump and probe group velocity dispersion and their velocity mismatch to obtain the highest time resolution can be applied without restriction.

After passage through the sample, the continuum beams are coupled to the spectrometer unit. We have chosen a prism-based polychromator primarily due to the high and broadband optical transmission that can be realized with this approach. The losses due to reflections on the prism can be kept typically around 25%, with negligible dependence on the wavelength. Adapting the Brewster configuration for polarized light can further significantly reduce the losses to the percent level. In contrast, with a grating used as the dispersive element, peak efficiency up to 90% can be achieved over a small spectral interval around the blaze wavelength, decreasing rapidly to below 30% toward the wings of the operating range. As a consequence, the effective spectral range would be significantly reduced by a grating spectrometer. Moreover, the second diffracted order of the Vis part would overlay the first order in the NIR above 1100 nm. The corresponding order filtering can also be avoided with a prism-based polychromator. The spectral resolution of such a system follows the dispersion relation of the prism material and can be quite constant in energy over the large detection range, whereas it is in first approximation constant in wavelength for a grating.

The prism spectrometer is set up as a portable unit in a classical way, collimating the rays coming out of the slit with lens L1 and focusing the rays dispersed by the prism with the lens L2 on the detector array (Fig. 1); the slit is imaged to the focal plane according to the ratio of the focal lengths of L2 and L1. The equilateral prism is operated in the Brewster angle configuration that corresponds also to the minimum beam deviation for the chosen material. In this alignment, the beam path is parallel to the base of the prism; as a consequence, the maximum effective aperture of the prism is obtained. N-SF11 was selected as highly dispersive material in the Vis and NIR that allows a compact design of the spectrometer, with resolution adapted for ultrafast spectroscopy. For this material, the angle of minimum deviation is obtained for an angle of incidence of 61° at 1100 nm (63° at 600 nm, 60.6° at 1600 nm), corresponding almost to the Brewster angle of 60.3° (60.7° at 600 nm, 60.1° at 1600 nm). The losses of p-polarized light by reflection on the prism surfaces are therefore nearly zero.

The length of the detector and the desired bandwidth of the spectrometer, in combination with the dispersion of the prism, fix the focal length f of the focusing lens L2 after the prism. For the 12.8 mm size of the photodiode array, the choice of f = 100 mm results in a nominal bandwidth of the detector unit of over 1100 nm for the spectral region of interest. Thus, the full nominal sensitivity range of the detector array from about 500 to 1700 nm is accessible (see below). In contrast, a similar spectral coverage based on a prism with low dispersive material in the NIR, such as fused silica or calcium fluoride, can be obtained with f around 500 mm.

This compact design is of particular interest because the spectrometer input can be fiber coupled. This approach is necessary for radiolysis applications (see section, Radiolysis of Isopropanol) where the detection unit must be protected against the hazardous radiation and electric noise generated by the electron excitation source. These perturbations can add important parasitic noise and even destroy the detection electronics. Therefore, the broadband signal and reference beams are coupled each into a fiber and transported over 14 m to a shielded room (SEDI-ATI Fibres Optiques, HCG 400 with numerical aperture 0.22). Under the described experimental conditions, no distortion of the probe spectrum due to the fiber transport was observed (see also Supplemental Material); no nonlinear effects are expected to occur in the fiber for the given probe power of

<0.1 MW and a corresponding intensity of <0.3 GW/cm² for the fiber coupling. The end of each fiber is directly positioned in front of the entrance slit of the spectrometer. The imaging of the slit and the fiber core, respectively, on the detector array defines the further constraints of the spectrometer setup. The 0.4 mm diameter of the fiber core was chosen to be distinctly higher than the focal size of the Vis-to-NIR continuum obtained by focusing in the fiber; the high ratio between fiber core and focal size allows stable coupling of the probe beams to the fibers, compensating for their spatial fluctuations and drifts. Thus, the laser pointing becomes important due to the beam transport over 15 m from the laser source to the pump–probe setup in the accelerator room, including three optical tables. On the other side, the InGaAs detector has a height of 0.5 mm. So, the magnification of the spectrometer must be <1.25 to image the full beam height defined by the fiber core to the detector. The collimating optics L1 at the entrance of the polychromator is therefore chosen to f = 100 mm. Inside the polychromator, collimation and focusing are realized with achromatic doublet lenses with free aperture of 48 mm (Qioptiq G322-302-300). The prism has side length and height of 60 mm and an effective horizontal aperture of 52 mm at 61° angle of incidence. In the configuration with beam referencing, the signal and reference beams are positioned parallel in the vertical plane at a distance of 24 mm and coupled in the polychromator. This results in an effective aperture of 24 mm for each beam. After the focusing lens L2, the beams are steered by mirrors in opposite directions to have sufficient space to place each InGaAs sensor board in the corresponding focal plane.

We have chosen the mentioned achromatic lenses inside the prism spectrometer and also for the free space coupling to it as these components allow quite easy and compact implementation without significantly deteriorating the performance, i.e., light transmittance and spectral resolution. An alternative setup with concave mirrors that have to be operated under a small angle of incidence to keep the astigmatism small is hard to realize without reducing the aperture of the beam path. This is particularly true for the operation with reference beam. The wavelength dependence of the focal length is considered by tilting the detector unit by 15°, reducing the effective detector width only slightly by 0.4 mm (in comparison, the use of a singlet lens requires a tilt of 30°, yielding a 1.7 mm reduction of the effective detector width). In the described configuration the size of the beam focused on the detector array is about 150 μm defining the spectral resolution (see Performance of the Transient Absorption Spectrometer below). For applications where no housing of the prism spectrometer is needed, lens L as well as the slit can be omitted. In this case, the focal spot of the supercontinuum inside the sample is directly imaged to the detector array with L1 and L2 of adapted focal length.

For the multichannel detector, we used a Hamamatsu G11608-512DA photodiode array with 512 pixels of size 25 × 500 μm². Its spectral response range is specified between 0.5 and 1.7 μm. Because of a new development, the response of this InGaAs photosensor is significantly enhanced in the Vis. The quantum efficiency is given as 35% at 500 nm, increasing to 75% at 900 nm, and to >80% above 1000 nm, the typical value of standard InGaAs sensors in the range from 1000 to 1600 nm. The conversion efficiency of the photodetector is set to a saturation charge of 112 × 10⁶ electrons per pixel. The data readout can be performed at the kilohertz rate and is triggered at the laser repetition frequency. The external triggering allows easy implementation of single-shot spectral referencing. Thus, signal S and reference R can be read simultaneously with two 16 bit A/D converters working in parallel. The absorbance change is calculated relative to the signal S₀ passing the unexcited sample. For this purpose, the pump pulse is blocked every second laser pulse. So, for kilohertz operation without reference

A (λ, t) = - log (\frac{\sum S (λ, t)}{\sum S_{0} (λ)})

Each single shot spectrum is detected separately, which permits the rejection of outliers relative to the average.^† In the case of single-shot referencing according to the configuration in Fig. 1b.

A (λ, t) = - log [\sum ((\frac{S (λ, t)}{R (λ)}) (\frac{R_{0} (λ)}{S_{0} (λ)})) / n]

with n single measurement pairs per delay step. Both the spectral distribution of the probe light and the absorbance change can be monitored online at video rate. Therefore, the alignment of the setup can be optimized, in particular the continuum generation, on parameters such as spectral coverage and shot-to-shot stability.

The calibration of the prism spectrometer and its multichannel detector is performed with filters exhibiting specific and well-known spectral features. The filters are positioned before the coupling to the spectrometer, in the collimated continuum beam. A pellicle beamsplitter with a dielectric coating (Thorlabs BP150) guarantees a series of reference points evenly distributed over the whole spectral detection range without introducing any displacement of the supercontinuum probe on the detector line. Its reference spectrum is recorded under an identical angle of incidence with a continuous wave spectrometer (Cary 5000 UV-Vis-NIR, Agilent Technologies). In the case of the radiolysis configuration with fiber coupling, a set of nine bandpass laser line filters (Thorlabs FB550-10, FL05632.8-3, FB810-10, FL051064-3, FB1200-10, FB1300-10, FB1400-10, FB1510-12, Edmund Optics 62 187) and one color glass absorption filter (BG36) can be used. The transmission curves as a function of pixel are measured using the supercontinuum as light source, i.e., under the conditions of ultrafast transient absorption measurement. The set of spectral reference points are fitted with the dispersion function of the prism spectrometer in the plane of the multichannel detector to obtain the relation between number of pixel N and wavelength X:

N (λ) = (f / p) sin (δ (λ) - δ (λ_{0})) + C

Here, f is the focal length of the lens L2, p the pitch of the detector, δ(λ) the output angle from the prism, and C the pixel offset. According to Snellius's law

δ (λ) = {sin}^{- 1} [n (λ) sin (A - {sin}^{- 1} (\frac{α}{n (λ)}))]

with α the angle of incidence on the prism, A its apex angle, and n(λ) the index of refraction. As in Megerle et al.,¹² the fit is realized with two free parameters: the effective focal length f and the pixel offset C. The effective focal length f considers a possible focal length shift after the focusing lens L2 that can be due to imperfect collimation and astigmatism. A dataset of the wavelength calibration is shown in the Supplemental Material.

PERFORMANCE OF THE TRANSIENT ABSORPTION SPECTROMETER

The spectral intensity distribution of a supercontinuum used as the probe is shown in Fig. 2. It is recorded with the setup according to Fig. 1a and its InGaAs detection unit under kilohertz measurement conditions. In addition to the majority of the laser fundamental that is blocked by the notch filter placed before the sample cell, the neighboring spectral regions must also be reduced to keep the signal level distinctly below saturation of the detector and thus to guarantee linear response to the light intensity. For this purpose, we use a KG2 and a RG850 filter (Schott AG) directly in front of the focal plane of the prism spectrometer in the Vis and NIR part of the spectrum, respectively. The two filters are positioned to cover the corresponding spectral regions up to the laser wavelength. The resulting spectral distribution exhibits an intensity variation of less than a factor of 10 within the range of 650-1350 nm. The lower light level toward the green and infrared still has sufficient photons to be exploited for transient absorption spectroscopy. Due to the distinctly increased saturation charge of the InGaAs diode array, this level corresponds to similar photon numbers as detected by ultrafast systems operating in the UV-Vis, commonly using a charge-coupled device sensor. It can be read in with high sensitivity because of the dynamic range of 16 000 resulting from 16-bit analog-to-digital conversion and the low readout noise at the kilohertz acquisition rate. The resulting effective bandwidth of the setup is >1000 nm, with measurement sensitivity better than 0.25 mOD (see below). To show the impact of the notch filter on the spectral distribution, we replaced it by an additional RG850 filter (dashed line in Fig. 2). Due to the notch filter, there is a dip of signal of about 60% beyond 1200 nm that recovers around 1600 nm to the level without the filter. Consequently, the shot noise level is slightly increased in this spectral region. However, for many applications, the remaining energy of the fundamental is in the same order as the one of the pump beam and could therefore induce a significant photoresponse of the sample. We therefore use and characterize the setup with the notch filter.

Fig. 2.

Logarithmic display of the supercontinuum probe light as detected by the broadband spectrometer. The fundamental wavelength at 780 nm and the high-intensity regions next to it were adjusted by a notch filter placed before the sample, and by KG2 and RG850 filters placed directly in front of the detector array in the anti-Stokes and Stokes side, respectively. Dashed line indicates spectrum with an additional RG850 replacing the notch filter.

The spectral resolution of the ultrafast transient absorption spectrometer is determined under measurement conditions directly with the continuum probe beam, in a way similar to the calibration procedure. The transmission curves of nine bandpass filters are measured, as shown in the top of Fig. 3. Δλ is derived from the corresponding full width at half-maximum with the known bandwidth of the filters. The experimentally obtained values exhibit in first approximation a reciprocal dependence on the derivative of the prism dispersion, as expected for a prism-based spectrometer (Δλ∝ λ/(dn/dλ)). The energy resolution is <10 meV in the Vis, increasing to 30 meV toward 1500 nm (Fig. 3, bottom). This resolution is more than sufficient to precisely resolve electronic transitions of molecules in solution with absorption bandwidths on the order of 250 meV.

Fig. 3.

(Top) Transmission spectra of a series of bandpass filters placed in the supercontinuum before the prism spectrometer. (Bottom) Spectral resolution deduced from the measured transmission curves of the filters with known full width at half-maximum.

In Fig. 4, we show the absorbance sensitivity of the setup that can be obtained in a typical measurement series: 10 temporal scans with average of 1000 single measurements per delay step. The overall acquisition time for scanning over 400 delay steps is below 3 h. As a measure of the sensitivity, the standard deviation of absorbance change ΔOD over 40 points of the baseline before time zero is given, i.e., without pump excitation. As Fig. 4 reveals, a sensitivity of at least 0.25 mOD is achieved over the whole spectral range, the average is 8.5 × 10-⁵ OD. These values were obtained with an energy stability of the laser source of the approximately 2% root mean square (RMS) recorded at the exit of the laser that is situated in a room next to the transient spectrometer. From ~650 nm toward the blue detection edge and from 800 to 1050 nm, the sensitivity seems to be predominantly governed by the shot noise of the probe light. In comparison, the spectral regions above 1050 nm and directly below the fundamental that are on the same or even higher light levels (see also Fig. 2) are less sensitive. This can be explained with increased shot-to-shot fluctuations in these spectral components of the continuum pulses. Such observations on wavelength-dependent fluctuations were made in the literature, among others, on the anti-Stokes side of continua generated in YAG.¹⁷

Fig. 4.

Probing sensitivity for a typical measurement of 10 temporal scans with averaging of 1000 per delay step at 500 Hz optical pump rate. The standard deviation of 40 points of the baseline is given. Over the whole spectral range, a sensitivity of at least 2.5 × 10-⁴ OD is achieved; the average is 8.5 × 10-⁵ OD.

In general, the time resolution of a femtosecond pump probe setup depends on the specific experimental configuration. Important parameters are the duration and wavelength of the excitation pulse; the geometry of pump and probe in the sample and relative to each other; as well as the thickness of the sample, including cell windows. The temporal resolution as well as the wavelength dependence of time zero can be revealed using the coherent artifact measured under experimental conditions.^28,29 This phenomenon is due to nonlinear interactions between the pump and probe pulses and therefore to their temporal overlap in a medium. Here, we have generated the coherent artifact by slightly focusing (with a concave mirror, f = 500 mm) few 0.1 μJ of the fundamental laser wave with pulse duration of about 100 fs in a 0.54 mm thick Bk7 plate. The angle between the pump and probe beams was less than 4°. Figure 5 shows the two-dimensional representation of the induced transient signal over the spectral detection range. The amplitude is on the level of few milliunits absorbance (mOD). The time zero was arbitrarily set to the green part of the broadband probe. Its wavelength dependence is due to the group delay dispersion of the supercontinuum from its generation in the YAG crystal to the sample. As the figure reveals, the temporal dispersion between the spectral detection limits is <0.7 ps. The typical chirp curve in the normal dispersion regime is superposed by the contribution of the dielectric coating of the notch filter. The latter adds group delay particularly at the limits of its blocking region (~760 and 800 nm) due to the intrinsic phase variations. At these specific wavelengths, the overall duration of the coherent artifact is increased to ~400 fs compared with ~250 fs found typically over the spectral detection window. The time resolution, often defined as the full width half-maximum of the Gaussian convolution of the pump and probe pulses, is distinctly below this value. Higher time resolution can be obtained in particular with shorter pump pulses, e.g., compressed noncollinear OPA pulses, and shorter interaction length of the pump and probe pulses as realized in a jet sample. As shown in previous work,^12,28 the curve of the coherent artifact can be used to correct the chirp of the supercontinuum probe to obtain transient spectra with temporal precision on the very low femtosecond level over the spectral detection range.

Fig. 5.

Transient signal of the coherent artifact at the time zero induced by the femtosecond laser pulse at 780 nm in a 0.54 mm thick BK7 plate.

APPLICATION ON THE ULTRAFAST PHOTOPHYSICS OF THE DYE IR-140

With the new broadband transient absorption spectrometer, we measured the relaxation dynamics of the organic dye IR-140 dissolved in ethanol after excitation with the laser pulse at 780 nm. The spectral evolution in the Vis-NIR spectral range from the femtosecond scale up to 1 ns is presented in Fig. 6b and Fig. 6c; the steady-state absorption and emission are shown in Fig. 6a. The transient spectra were recorded with an average of 500 measurements of absorbance change per delay step; 340 delay steps per scan were distributed over the chosen time range. A quasi-exponential time scale was used after 1 ps to adapt the measurement statistics to the kinetics of population transfer. The data were averaged over four scans acquired in about 30 min (70% of time for measurement, 30% for translation of the delay stage). The sample was provided in a static cell with 0.2 mm thick windows and 0.5 mm optical path;^‡ the sample concentration was set to 0.05 mM corresponding to a pump transmission of 40% to guarantee homogenous excitation over the full sample length. A few 0.1 μJ of the laser source were slightly focused in the sample to a diameter of approximately 250 μm. The polarization between pump and probe beam was set to the magic angle of 54.7°.

Fig. 6.

(a) Steady-state absorption (black) and fluorescence (red) of IR-140 in ethanol. (b) Transient spectra at selected time delays upon photoexcitation with the 100 fs laser pulse at 780 nm. (c) Two-dimensional representation of the full data set that was acquired in 30 min at an optical pump rate of 500 Hz.

The transient data of IR-140 show a complex behavior both in the spectral as well as in the time domain pointing out the need of broadband time-resolved measurements. A detailed discussion of the photophysics in several solvents and for different excitation wavelengths can be found in the literature.^30–32 Here, and very similar to the literature, several spectral bands that are partially overlaying can be observed; they exhibit a temporal evolution on different time scales (see Fig. 7). Although for all bands the overall decay is almost determined within 1 ns with very similar kinetics, the molecular dynamics on the femto- and picosecond scale is strongly wavelength dependent. The transient spectra exhibit an excited state absorption (ESA) band centered at 565 nm and strong negative absorbance change with several features around the S₀ - S₁ electronic transition around 800 nm. The corresponding ground state absorption band is asymmetric with a shoulder at the short wavelength side (Fig. 6a). Steady-state absorption and emission studies have revealed that two species are responsible for this absorption that are attributed in the literature to the cis and trans isomer of IR-140, peaking in ethanol at 760 and 803 nm, respectively. With the pump pulse at 780 nm, both isomers are excited to their S₁ electronic state. As a consequence, the observed negative absorbance change is composed by the ground state bleaching (GSB) of these two species and, from the excitation wavelength toward lower energies, by their stimulated emission bands. Relative to the known steady-state fluorescence peaks at 820 and 840 nm, the transient spectra locate the stimulated emission at 815 and 838 nm, with the higher energetic emission transition peaking in the wing of the main emission. The slight blue shift can be explained by the superposition with the GSB.

Fig. 7.

Kinetic traces probed in the different electronic transitions of IR-140: excited state absorption at 560 nm, pure stimulated emission at 950 nm, ground state absorption at 760 nm below the emission contributions and superposed by emission of cis and trans isomer at 819 and 840 nm. The display after 20 ps is logarithmic.

With the revealed transient spectral signatures, the different species and their temporal behavior can be separated: over 900 nm, the stimulated emission can be isolated; below 780 nm, the GSB is free from contributions of stimulated emission; and for the ESA below 550 nm, the GSB is negligible. The observed slow recovery of the ground state is correlated both in relative amplitude and time behavior with the decay of ESA and stimulated emission indicating direct radiative and nonradiative conversion pathways. The wavelength dependent temporal evolution below 20 ps is typical for relaxation within the excited state surface. Detailed information can be gained by methods such as singular value decomposition and Bayesian data analysis.⁴ This example shows how precise transient spectra recorded over a broad probing range provide insight into complex molecular systems and their nonequilibrium evolution.

OPERATION IN HAZARDOUS ENVIRONMENT ON LOW REPETITIVE EXCITATION SOURCES

The configuration designed for operation with low repetition sources in hazardous environments (Fig. 1b) must address several drawbacks. Applied to ultrafast radiolysis as described in the next section, single-shot referencing is necessary due to the pump rate of 10 Hz; the signal and reference beam must be transported by optical fibers to the shielded room housing the detector unit. Both factors imply a significant reduction of detected probe photons. In addition, the use of fibers depolarizes the probe light and increases the reflection losses on the entrance and exit sides of the prism to about 15% in each case. The typical intensity distribution of the supercontinuum detected with the described setup is shown at the top of Fig. 8. Due to the photon losses, the notch filter is sufficient to adapt the probe light level around the laser fundamental below the saturation level of the detector; no additional filters are needed. The reduction of the light level reduces the effective spectral coverage toward the limits of the detection window. Of course, both low repetition rate and reduced photon content per pulse have impacts on the measurement sensitivity. Moreover, the long-distance beam transport between laser source and sample cell, including three separate optical tables, adds laser beam instability in terms of shot-to-shot fluctuations and drifts. Another important source of noise is the pump pulse. In the case of a picosecond electron accelerator, short- and long-term instabilities can be easily higher than 10% RMS. To minimize the effect of pump drifts on the temporal evolution of transient absorption, one strategy is to drive many temporal scans with low averaging per delay step. A typical measurement series contains 40 scans over 260 delay steps each, averaged over three single absorbance change measurements repeated at least on two days. The overall acquisition time on one day is slightly above 1 h, with the duration of one scan < 2 min and therefore also below typical drift times of the electron pump source. The bottom of Fig. 8 shows the measurement sensitivity obtained by such a measurement taking about 1 h. It is the standard deviation of the 25 baseline points of the dataset A(λ,t) on ultrafast radiolysis of isopropanol discussed in the next section. In spite of this low averaging of 120 shots per delay step, measurement sensitivity on the milliunits OD level is obtained over the spectral detection window, below milliunits OD from about 680 to 1400 nm, reaching the level of 0.5 milliunits OD over a width of 500 nm in the central part of the detection window. The sensitivity curve follows the probe light distribution, indicating shot noise limitation of the configuration with single-shot referencing.

Fig. 8.

(Top) Logarithmic display of the intensity distribution of the probe supercontinuum detected by the fiber-coupled spectrometer with single-shot referencing for pulse radiolysis. The fundamental wavelength was adjusted by a notch filter. (Bottom) Probing sensitivity for a typical radiolysis measurement of 40 temporal scans with averaging of three per delay step at 10 Hz electron pump rate. The standard deviation of 25 points of the baseline before time zero is given.

RADIOLYSIS OF ISOPROPANOL INDUCED BY A PICOSECOND ELECTRON PULSE

As second example of an application that illustrates the performance of the new transient absorption spectrometer, we present ultrafast radiolysis measurements on liquid isopropanol. This kind of experiment can be seen as a test of the setup under severe conditions compared with standard optical pump excitation. It involves the previously mentioned low repetition rate; the long-distance beam transport between the laser source and sample cell, including three separate optical tables and the lower probe light level; as well as the very limited access to the setup. A picosecond high-energy electron pulse acts as the pump that ionizes a part of the solvent molecules during its passage through the sample. The experiments are performed at experimental area 1 of the ELYSE radiolysis facility (Université Paris Sud at Orsay, France), an electron accelerator based on the radiofrequency photocathode gun technology.^33,34 The electron pump is generated with the same laser source as the probe continuum. The pump and probe pulses are therefore intrinsically synchronized as is typical for ultrafast setups. For the present measurement, the accelerator output was set to a typical configuration: an electron bunch with a charge of 3.6 nC and a mean energy of 7.6 MeV is focused to a diameter of 3–4 mm into the sample cell that is placed 3 cm beyond the accelerator exit. Under these conditions, it exhibits a typical pulse duration around 10 ps and a pulse to pulse RMS jitter <1 ps. These values were confirmed at the position of the sample cell by single-shot electro-optic sampling of the electric field co-propagating with the relativistic electron bunch.^35,36

As the accelerator is a low repetition source operated here at 10 Hz, a reference pulse of the probe beam is recorded for each laser shot delivered at 20 Hz (see the above section, Setup of the Transient Absorption Spectrometer). For this purpose, the reference beam is split off before the sample cell. As mentioned above, both the signal and reference beams are coupled to optical fibers of low OH content and transported to the polychromator situated in the detection room. The optical probe beam is coupled into the sample by means of a thin aluminum mirror with a substrate of 0.2 mm thickness. Whereas the electron pump passes through this optical component with low losses and distortions of its beam quality, the continuum light is reflected and directed collinear to the pump beam through the sample. The irradiated volume of the sample is exchanged during measurement to avoid accumulation of possible products. The thickness of the optical flow cell is 5 mm with 1 mm fused silica windows. This sample thickness is a good trade-off to obtain a distinct change of absorbance without losing significantly the time resolution. In the described configuration, it is mainly the electron pulse duration and the mismatch of velocity of the electron pulse and the optical pulse that determine the rise time of the absorption signal; the increase from the 10% to 90% level takes < 20 ps. For pump electrons of several million electron volts, the energy transfer is in the low linear regime; therefore, the irradiation can be considered to be almost homogenous over the sample thickness. During the passage through the sample, only a few percent of the electron's energy are transferred to the solvent molecules that are in turn dissociated to ionic and radical products (or charged and uncharged radicals). As products of this initial interaction with the electron pulse, nonthermalized electrons can further ionize solvent molecules in the neighborhood, and so-called spurs are formed.

In the case of the solvent isopropanol, ionization forms the main primarily products of radiolysis. The solvated electron in isopropanol is known to absorb in the Vis and NIR. Its decay proceeds on timescales of nano- to microseconds, mainly by reaction with neutral and protonated solvent molecules.³⁷ Here, the new transient Vis-to-NIR spectrometer for the first time enables observation of the buildup of this species in isopropanol at ambient temperature, i.e., the dynamic solvation of the electrons generated by the ionization of solvent molecules. The full spectral and temporal information on this process is shown in Fig. 9 in the range of 600 to 1600 nm within the first 200 ps after electron pulse radiolysis. Initially revealing a broad absorption band centered around 1300 nm (0.95 eV) and covering the whole detection range, the temporal evolution of the transient absorption is characterized by a spectral shift of about 500 nm to shorter wavelengths and a narrowing of the absorption band. After 200 ps, an absorption band peaking around 800 nm (1.55 eV) is formed that does not exhibit any change on the further picosecond scale. Its broad, structureless shape is typical for solvated electrons in alcohols, and its peak wavelength coincides with the value of the solvated electron in isopropanol known from longer time scales.³⁸ The solvation process is therefore considered to be complete within 200 ps. The observed blue shift directly following the excitation pulse is typical for solvation processes observed via transient absorption and can be explained with a higher and/or differently oriented dipole moment of the populated lower lying electronic state relative to the one of the probed states.^22,23,39,40 In the case of electron solvation in protic solvents, it is very large and accompanied by a distinct narrowing of the absorption band as observed here.⁴¹ The corresponding solvation dynamics of isopropanol exhibit a fast absorption buildup at the center of the final absorption band that is completed within 100 ps. Toward shorter wavelengths, the increase takes slightly longer. In contrast, the overall absorption decrease in the NIR wing is longer lasting; it exhibits a further deceleration toward the final absorption center. Due to this wavelength dependence, the shift of the absorption center is a more meaningful value to characterize the solvation time than the dynamics at specific wavelengths.⁴¹ This example illustrates the importance of accessing the complete evolution over a large spectral window for ultrafast spectroscopy.

Fig. 9.

Transient absorption of isopropanol in the NIR to Vis after radiolysis with an ultrashort electron pulse. The measurement time was about 1 h at an electron pump repetition rate of 10 Hz.

CONCLUSIONS

We have described in detail a new broadband ultrafast transient absorption spectrometer that allows the simultaneous measurement from about 510 to 1650 nm. The corresponding probe continuum is generated with the laser fundamental at 780 nm in a YAG crystal. This large spectral range becomes accessible as a result of the available photon number of the continuum that is adapted to the NIR, the high optical transmission of the setup, and particularly the prism-based polychromator unit as well as the extended detector sensitivity. The combination of the used supercontinuum, a notch filter placed before the sample, and two color glass filters in front of the diode array enables measurement over the fundamental laser wavelength without a spectral gap. With a kilohertz laser system using no reference beam, a sensitivity of absorbance change ΔOD defined as the standard deviation over the baseline can be achieved on the level of 10-⁴ for a typical scan time of few hours (for a laser energy stability on the range of 2% RMS). The continuum is generated directly with the fundamental laser wavelength and in single-filament operation; consequently, the setup can be run easily in day-to-day routine without time-consuming alignment. Single-shot referencing can further increase the sensitivity, accompanied by a slight increase of the complexity of the setup and reduction of the spectral range. A version with fiber coupling to the polychromator unit was realized for application in the hazardous environment of an electron accelerator used for picosecond radiolysis. In this configuration and thanks to single-shot referencing, a sensitivity of 0.5 mOD over a wide spectral range was achieved during about 1 h overall scan time in spite of the low pump frequency of 10 Hz and the severe conditions of the infrastructure.

The presented system, with its resolution of <30 meV, enables time-resolved studies on condensed phases such as liquids or films in a spectral region of interest for a variety of applications. Thus, electronic transitions or vibrational overtones and combination bands of molecules can be probed in solution in the accessible spectral range. For ultrafast radiation physics and chemistry, the NIR is of particular interest because the solvated electron, one of the major products of radiolysis, exhibits its absorption band there in low-polarity solvents. Moreover, and as shown here, the first steps of electron solvation are accompanied by transient absorption extended over a large spectral range that is distinctly red shifted relative to the final equilibrium state. In semiconductor systems, the dynamics of free charge carriers and their subsequent lower-lying states can be probed exclusively in the NIR and simultaneously the ground state recovery via interband transition in the Vis. The large spectral detection range can thus simplify the disentanglement of possible parallel and consecutive relaxation pathways toward the development of electronic devices and photovoltaics based on organic materials or nanoparticles. In general, full information on the ultrafast spectral evolution in the Vis and NIR, combined with known adapted analysis procedures, can give detailed insight into the early steps of radiation-induced processes that are often decisive for the branching to the final species and their conversion efficiencies.

Footnotes

ACKNOWLEDGMENTS

We are very grateful to James Wishart, Brookhaven National Laboratory, for carefully reviewing the grammar of the manuscript. We thank Edmund Optics SARL/GmbH, particularly Elodie Hamel, for providing information on the notch filter.

SUPPLEMENTAL MATERIAL

All supplemental material mentioned in the text, including two figures, is available in the online version of the journal, at .

†

No rejection was used for the presented dataseis.

‡

For the presented application on IR-140 in ethanol, no photochemistry and, consequently, no accumulation of a photoproduct is expected. However, fluctuations of the absorbance change on the scale of seconds are observed occasionally that could be attributed to thermal effects; these fluctuations could be avoided using a flow cell or a jet that changes the irradiated sample volume between two excitation pulses.

References

Zewail

A.H.

“Femtochemistry: Atomic-Scale Dynamics of the Chemical Bond”. J. Phys. Chem. A. 2000. 104(24): 5660–5694.

Berera

van Grondelle

Kennis

J.T.M.

“Ultrafast Transient Absorption Spectroscopy: Principles and Application to Photosynthetic Systems”. Photosynth. Res. 2009. 101(2–3), 105–118.

Greetham

G.M.

Burgos

Cao

Clark

I.P.

Codd

P.S.

Farrow

R.C.

George

M.W.

Kogimtzis

Matousek

Parker

A.W.

Pollard

M.R.

Robinson

D.A.

Xin

Z.J.

Towrie

“ULTRA: A Unique Instrument for Time-Resolved Spectroscopy”. Appl. Spectr. 2010. 64(12): 1311–1319.

Ruckebusch

Sliwa

Pernot

de Juan

Tauler

“Comprehensive Data Analysis of Femtosecond Transient Absorption Spectra: A Review”. J. Photochem. Photobiol., C. 2012. 13(1): 1–27.

Bonneau

Wirz

Zuberbuhler

A.D.

“Methods for the Analysis of Transient Absorbance Data”. Pure Appl. Chem. 1997. 69(5): 979–992.

Schmidhammer

Roth

Riedle

Tishkov

A.A.

Mayr

“Compact Laser Flash Photolysis Techniques Compatible with Ultrafast Pump–Probe Setups”. Rev. Sci. Instrum. 2005. 093111(9): 1–7.

Cerullo

Nisoli

De Silvestri

“Generation of 11 fs Pulses Tunable Across the Visible by Optical Parametric Amplification”. Appl. Phys. Lett. 1997. 71(25): 3616–3618.

Wilhelm

Piel

Riedle

“Sub-20-fs Pulses Tunable Across the Visible from a Blue-Pumped Single-Pass Noncollinear Parametric Converter”. Opt. Lett. 1997. 22(19): 1494–1496.

Alfano

R.R.

The Supercontinuum Laser Source: Fundamentals with Updated References. New York: Springer, 2006. 2nd ed.

10.

Dobryakov

A.L.

Kovalenko

S.A.

Weigel

Pérez-Lustres

J.L.

Lange

Müller

Ernsting

N.P.

“Femtosecond Pump/Supercontinuum-Probe Spectroscopy: Optimized Setup and Signal Analysis for Single-Shot Spectral Referencing”. Rev. Sci. Instrum. 2010. 81: 113106.

11.

Buntinx

Naskrecki

Poizat

“Subpicosecond Transient Absorption Analysis of the Photophysics of 2,2′-Bipyridine and 4,4′-Bipyridine in Solution”. J. Phys. Chem. 1996. 100(50): 19380–19388.

12.

Megerle

Pugliesi

Schriever

Sailer

C.F.

Riedle

“Sub-50 fs Broadband Absorption Spectroscopy with Tunable Excitation: Putting the Analysis of Ultrafast Molecular Dynamics on Solid Ground”. Appl. Phys. B: Lasers Opt. 2009. 96(2–3), 215–231.

13.

Polli

Lüer

Cerullo

“High-Time-Resolution Pump–Probe System with Broadband Detection for the Study of Time-Domain Vibrational Dynamics”. Rev. Sci. Instrum. 2007. 78(10): 103108.

14.

Hermann

Niesar

Scharsich

Köhler

Stutzmann

Riedle

“Role of Structural Order and Excess Energy on Ultrafast Free Charge Generation in Hybrid Polythiophene/Si Photovoltaics Probed in Real Time by Near-Infrared Broadband Transient Absorption”. J. Am. Chem. Soc. 2011. 133(45): 18220–18233.

15.

Loevsky

Wand

Bismuth

Friedman

Sheves

Ruhman

“A New Spectral Window on Retinal Protein Photochemistry”. J. Am. Chem. Soc. 2011. 133(6): 1626–1629.

16.

Duncan

T.V.

Rubtsov

I.V.

Uyeda

H.T.

Therien

M.J.

“Highly Conjugated (Polypyridyl)Metal-(Porphinato)Zinc(II) Compounds: Long-Lived, High Oscillator Strength, Excited State Absorbers Having Exceptional Coverage of the Near-Infrared”. J. Am. Chem. Soc. 2004. 126(31): 9474–9475.

17.

Bradler

Baum

Riedle

“Femtosecond Continuum Generation in Bulk Laser Host Materials with Sub-μJ Pump Pulses”. Appl. Phys. B: Lasers Opt. 2009. 97(3): 561–574.

18.

Huber

Satzger

Zinth

Wachtveitl

“Noncollinear Optical Amplifiers with Output Parameters Improved by the Application of a White Light Generated in CaF₂”. Opt. Commun. 2001. 194(4–6), 443–448.

19.

Tzankov

Buchvarov

Fiebig

“Broadband Optical Parametric Amplification in the Near UV–VIS”. Opt. Commun. 2002. 203(1–2), 107–113.

20.

El Omar

A.K.

Schmidhammer

Rousseau

LaVerne

Mostafavi

“Competition Reactions of H₂O^•+ Radical in Concentrated Cl⁻ Aqueous Solutions: Picosecond Pulse Radiolysis Study”. J. Phys. Chem. A. 2012. 116(47): 11509–11518.

21.

Balcerzyk

Schmidhammer

El Omar

A.K.

Jeunesse

Larbre

J.-P.

Mostafavi

“Picosecond Pulse Radiolysis of Direct and Indirect Radiolytic Effects in Highly Concentrated Halide Aqueous Solutions”. J. Phys. Chem. A. 2011. 115(33): 9151–9159.

22.

Schmidhammer

Megerle

Lochbrunner

Riedle

Karpiuk

“The Key Role of Solvation Dynamics in Intramolecular Electron Transfer: Time-Resolved Photophysics of Crystal Violet Lactone”. J. Phys. Chem. A. 2008. 112(37): 8487–8496.

23.

Schmidhammer

Pernot

De Waele

Jeunesse

Demarque

Murata

Mostafavi

“Distance Dependence of the Reaction Rate for the Reduction of Metal Cations by Solvated Electrons: A Picosecond Pulse Radiolysis Study”. J. Phys. Chem. A. 2010. 114(45): 12042–12051.

24.

Nguyen

N.T.

Saliminia

Liu

Chin

S.L.

Vallée

“Optical Breakdown Versus Filamentation in Fused Silica by Use of Femtosecond Infrared Laser Pulses”. Opt. Lett. 2003. 28(70): 1591–1593.

25.

Ashcom

J.B.

Gattass

R.R.

Schaffer

C.B.

Mazur

“Numerical Aperture Dependence of Damage and Supercontinuum Generation from Femtosecond Laser Pulses in Bulk Fused Silica”. J. Opt. Soc. Am. B. 2006. 23(11): 2317–2322.

26.

Brodeur

Chin

S.L.

“Ultrafast White-Light Continuum Generation and Self-Focusing in Transparent Condensed Media”. J. Opt. Soc. Am. B. 1999. 16(4): 637–650.

27.

Schmidhammer

Ladungstransfer im Wechselspiel mit der Nanoumgebung: Organische Reaktivität auf der Femto- bis Millisekunden Skala. [Ph.D. Dissertation]. Munich, Germany: Ludwig-Maximilians-Universität, 2008.

28.

Kovalenko

S.A.

Dobryakov

A.L.

Ruthmann

Ernsting

N.P.

“Femtosecond Spectroscopy of Condensed Phases with Chirped Supercontinuum Probing”. Phys. Rev. A. 1999. 59(3): 2369–2384.

29.

Lorenc

Ziolek

Naskrecki

Karolczak

Kubicki

Maciejewski

“Artifacts in Femtosecond Transient Absorption Spectroscopy”. Appl. Phys. B: Lasers Opt. 2002. 74(1): 19–27.

30.

Mohanty

Palit

D.K.

Mittal

J.P.

“Photophysical Properties of Two Infrared Laser Dyes – IR-144 and IR-140: A Picosecond Laser Flash Photolysis Study”. Proc. Indian Nat. Sci. Acad. A. 2000. 66: 303–315.

31.

Wang

Jiang

Yang

Liu

Gong

Xiang

“Investigation on Photoexcited Dynamics of IR-140 Dye in Ethanol by Femtosecond Supercontinuum-Probing Technique”. J. Opt. A: Pure Appl. Opt. 2002. 4(2): 155–159.

32.

Wang

Jiang

Yang

Gong

Xiang

“An Investigation of Solvent Effects on the Optical Properties of Dye IR-140 Using the Pump Supercontinuum-Probing Technique”. J. Opt. A: Pure Appl. Opt. 2003. 5(5): 515–519.

33.

Belloni

Monard

Gobert

Larbre

J.-P.

Demarque

De Waele

Lampre

Marignier

J.-L.

Mostafavi

Bourdon

J.C.

Bernard

Borie

Garvey

Jacquemard

Leblond

Lepercq

Omeich

Roch

Rodier

Roux

“ELYSE - A Picosecond Electron Accelerator for Pulse Radiolysis Research”. Nucl. Instrum. Methods Phys. Res., Sect. A. 2005. 539(3): 527–539.

34.

Marignier

J.L.

De Waele

Monard

Gobert

Larbre

J.-P.

Demarque

Mostafavi

Belloni

“Time-Resolved Spectroscopy at the Picosecond Laser-Triggered Electron Accelerator ELYSE”. Radiat. Phys. Chem. 2006. 75(9): 1024–1033.

35.

Schmidhammer

De Waele

Marques

J.-R.

Bourgeois

Mostafavi

“Single Shot Linear Detection of 0.01-10 THz Electromagnetic Fields: Electro-Optic Sampling with a Supercontinuum in Balanced Detection”. Appl. Phys. B: Lasers Opt. 2009. 94(1): 95–101.

36.

De Waele

Schmidhammer

Marques

J.-R.

Monard

Larbre

J.-P.

Bourgeois

Mostafavi

“Non-Invasive Single Bunch Monitoring for ps Pulse Radiolysis”. Rad. Phys. Chem. 2009. 78(12): 1099–1101.

37.

Dey

G.R.

Remita

Mostafavi

“Radiolytic Reduction of Fe(II) in 2-Propanol”. Chem. Phys. Lett. 2006. 431(1–3), 83–87.

38.

Arai

Sauer

M.C.

“Absorption Spectra of the Solvated Electron in Polar Liquids: Dependence on Temperature and Composition of Mixtures”. J. Chem. Phys. 1966. 44: 2297–2304.

39.

Lustres

J.L.P.

Kovalenko

S.A.

Mosquera

Senyushkina

Flasche

Ernsting

N.P.

“Ultrafast Solvation of N-Methyl-6-Quino-lone Probes Local IR Spectrum”. Angew. Chem. Int. Ed. 2005. 44(35): 5635–5639.

40.

Karunakaran

. “Solvation Oscillations and Excited-State Dynamics of 2-Amino- and 2-Hydroxy-7-nitrofluorene and Its 2′-deoxyriboside”. J. Phys. Chem. A. 2008. 112(18): 4294–4307.

41.

Lampre

Bonin

Soroushian

Pernot

Mostafavi

“Formation and Solvation Dynamics of Electrons in Polyols”. J. Mol. Liq. 2008. 141(3): 124–129.