1. Introduction

Pump-probe experiments are based on the principle that the pump pulse triggers a series of events by modifying the equilibrium state of a system. The temperature-jump method consists in providing heat to a sample producing a temperature change in a timescale shorter than the process whose dynamics is under investigation. Nanosecond to femtosecond optical pulses have been used to obtain ultrafast T-jumps, allowing the observation of conformational changes in molecules of biological interest in water solutions [1], or of properties related to the H-bonding in liquid water [2]. Heating is achieved by exciting a vibrational transition of water. Vibrational relaxation and heat transfer through intermolecular interactions provide the T-jump in a timescale corresponding to the slowest process between the pulse duration and the thermalization of the sample, that is around 5 ps for liquid water [3] and around 10 ps for ice [4].

The pulse energy needed to increase the temperature of the sample by 1K is given by c_vρV/α where c_v is the specific heat (for water 4.2 Jg⁻¹K⁻¹), ρ is the density, V is the volume of the sample, and α is the fraction of absorbed light [5]. The studies performed so far using the T-jump technique involved a temperature increase of ∼10–20 K in liquid water or a partial melting of ice [4]. Here instead we want to obtain the complete melting of an ice sample with typical dimensions as those contained in high-pressure cells, to gain access to the investigation of melting and freezing dynamics in a wide P,T range. The feasibility of ultrafast time-resolved experiments at high static pressure, which provided unique information on the rotational dynamics of water as a function of density, has been recently demonstrated [6]. At present however the access to transient phenomena at high static pressures is extremely limited and provided by indirect, steady state measurements [7, 8]. Samples compressed up to 1 Mbar range between 1 and 6×10⁻⁶cm³. The amount of energy needed to entirely melt a sample of these dimensions, estimated by replacing c_v with the latent heat of fusion of water (335 Jg⁻¹), thus ranges between 2 and 10 mJ/pulse. The source described in this paper is an optical parametric generator and amplifier (OPA) pumped by a 20 ps, 10 Hz repetition rate, Nd:Yag laser providing up to 55 mJ/pulse at 1.064 μm. The wavelength of 1.9 μm matches vibrational combination bands of water (ν₁+ν₂; ν₂+ν₃) allowing the heat transfer into the ice samples. Absorbance of the sample at this wavelength is ∼0.2 for a typical thickness of ∼50μm. The pulse duration is in the same range as that of ice thermalization and the energy exceeds that needed to melt samples in a high-pressure cell, as will be shown in the following. Both features are basic requirements for future investigation of ice melting and nucleation in a sub-ns timescale, using for instance femtosecond broadband IR probe pulses from a synchronized source. Optical setups providing ps pulses with output wavelength around 2 m, energy up to 11 mJ/pulse and 1–10 KHz repetition rates have been recently developed [9–11]. However, we need to exceed this maximum energy value in order to be able to study bulk samples in high-pressure cells, and at the same time we are not interested in high repetition rates, because as we shall see the full relaxation of our investigated processes require at least 100 ms. The source presented in this paper provides energies up to 15 mJ/pulse with maximum 10 Hz repetition rate, with conversion efficiency exceeding 20%, while using a very simple and cost-effective design.

2. T-jump source

The pump beam for our OPA is the fundamental (1064 nm) of a 20 ps mode-locked Nd:Yag laser with maximum repetition rate of 10 Hz, energy up to 50 mJ/pulse, beam diameter 8 mm and vertical polarization (Ekspla PL2143A). A schematic view of the OPA is shown in Fig. 1. The non linear crystals are two identical BBO crystals (BBO I and BBO II in Fig. 1) for Type I generation (2418.2 nm (o)+ 1900.0 nm (o)= 1064.0 nm (e), θ = 22.0°, ϕ =90 °, 6×10×15 mm, with AR coating for 1064 and 1900 nm). Type I generation was chosen as it gives a larger bandwidth, required to excite a larger population of the water sample, and shorter pulses with respect to Type II [12]. BBO I is used for generation and preamplification at 1.9 μm using 20 % of the pump beam transmitted by a beamsplitter (CVI BS1-1064-80-1012-45S) and demagnified by a factor 0.25 on the beam diameter by a telescope (L1, f=100 mm, and L2, f=−25 mm). The elements indicated in Fig. 1 as M1 to M8 are highly-reflective mirrors at 1064 nm, M9 to M15 are highly reflective at 1900 nm (CVI TLM1-1900-0-1037 for i=0° and TLM1-1900-45P-1037 for i=45° and P polarization), D1 to D5 are highly reflective at 1064 nm on UVFS substrate, with anti-reflex coating at 1900 nm on both sides (CVI Y1-1025-45S-UV-AR1900). After the first pass through BBO I, the pump is reflected by D2 whereas the signal and idler are transmitted. The idler is also transmitted by M9 to avoid its amplification in the second pass (light at 2.4 μm is partly absorbed by the BBO and if not eliminated would cause its breakdown). Both the pump and the signal are reflected back into BBO I for the second pass by roof reflectors, one of which is mounted on a linear stage. The roof is needed to prevent back-reflection into the Nd:Yag laser. The preamplified 1.9 μm beam is sent to BBO II, after magnification by a factor 2 (L5, f=−50 mm, and L6, f=100 mm), temporally and spatially coincident with the 80% part of the pump beam reflected by the beamsplitter BS, and demagnified by a factor 0.6 (L3, f=250 mm, and L4, f=−150 mm) for a double amplification pass. M3, M4, and M13, are mounted on linear stages. The residual pump and idler beams are separated from the signal by reflection on D5 and transmission through M14 respectively. M11 and M15 are mounted on flipmounts in order to send independently the unamplified and amplified pulses to a 1/8 m monochromator (Spectral Products CM110, grating 300 grooves/mm, peaked at 2.5 μm and working range 1.5 – 6.0 μm AG0300-02500-303) coupled to a large area MCT photo-conducting detector (VIGO System PC-4-4x4-BNC-BaF₂) to characterize the spectral distribution of the output signal [Fig. 2(a)]. The maximum conversion efficiency is found for phase matching at 1.93 μm. The energy at 1.93 μm, measured with a piroelectric detector (Gentec-Eo QE25LP-S-MB) is reported in Fig. 2(b) as a function of the pump energy. The autocorrelation profile [Fig. 2(c)] has been measured by second order autocorrelation, exploiting the two-photon absorption at 1.9 μm on a Si photodiode (Hamamatsu S1722-02), which directly acted as the nonlinear element [13]. The FWHM of the autocorrelation curve, fitted as a gaussian profile, is 20.6 ps, corresponding to a pulse duration of ∼15 ps. The spatial profile of the output beam was sampled by the MCT photo-conducting detector mounted on a motorized xy translation stage, in steps of 125 μm in both directions, through a pinhole with 10 μm diameter [Fig. 2(d)]. The divergence angle is 1.5 mrad, giving a beam quality factor M²=3.0 in the x axis and M²=3.6 in the y direction.

 

Fig. 1 Scheme of the OPA. Wavelengths present in the path are indicated in μm.

Download Full Size | PDF

 

Fig. 2 Characterization of the T-jump beam. (a) Spectrum of the output beam (black), of the preamplified beam after the second pass in BBO I (blue), of the beam after the second pass in BBO II in the absence of seed beam (red). (b) Energy/pulse at 1.9 μm as a function of the energy/pulse of the pump at 1.06 μm. (c) Autocorrelation profile (see text). (d) Spatial profile (see text).

Download Full Size | PDF

3. Melting and freezing of ice Ih and VI

Water was loaded either in a room-pressure cell with amorphous silica windows and a teflon spacer, or in a high-pressure membrane cell (SAC) equipped with z-cut, low-fluorescence sapphire anvils and a copper-beryllium gasket. In both cases the sample dimensions were 50 μm in thickness and 400 μm in diameter. The fluorescence of a ruby chip ∼5 μm in diameter was used as pressure gauge [14] in the high-pressure experiment. The second harmonic (532 nm) of a cw diode-pumped intracavity-doubled Nd-YVO (Verdi-Coherent) laser was focused onto the sample with a focal length of 150 mm. Two calcite polarizers were placed before and after the sample [P₁ and P₂ in Fig. 3(a)] and the transmitted light was measured by a fast avalanche photodiode (Hamamatsu) and amplified with a DC-800MHz AVTECH amplifier and analysed by a 1GHz oscilloscope (R&S RT1014). P₁ was set to the angle maximizing the laser through-put, whereas P₂ was rotated to obtain the maximum extinction condition. With no sample or an isotropic sample the signal on the photodiode is not detectable. For liquid water in the SAC, a constant residual signal is due to depolarization by the sapphire anvils. With ice I_h or ice VI in the cell the depolarized signal has a constant non-zero value which is at least five times larger than the background in the case of the SAC. The I_h sample was prepared by cooling water to 250 K in a Peltier cryostat, and then increasing the temperature to T₀ = 270 K. The ice VI sample was prepared by compressing water to 1.5 GPa in the SAC at room temperature, and then lowering the pressure to 1.12 GPa, close to the melting. The FTIR spectrum, measured with a Bruker IFS120 HR, confirmed the crystallization of the sample. The pulse at 1.93 μm was focused with a f=50 cm BK7 uncoated lens. The observed focal length at this wavelength is 60 cm. The sample was placed at 55 cm from the lens to have it almost completely illuminated by the IR beam, avoiding irradiation of the metallic gasket. The low absorbance at 1.9 μm guarantees a longitudinally uniform irradiation. After each pulse, at 5 or 10 Hz repetition rate, a fast decrease (< 0.3 ms) of the depolarization signal is observed.

 

Fig. 3 (a) Scheme of the experiment. (b) Phase diagram of water in the range where the present measurements have been perfomed. In the inset: FTIR spectra in the overtone region showing a marked difference between liquid and ice VI phases.

Download Full Size | PDF

After reaching a minimum, the signal returns to the original value, consistently with the recovery of birefringence due to the complete recrystallization [Fig. 4(a)]. This effect was not observed on the residual depolarized light from the cell windows or from a cell containing liquid water, so we can exclude contributions to the signal due to other effects induced by the IR pulse. The observed waveform is independent of the pulse energy at 1.93 μm when it exceeds 7 mJ/pulse [Fig. 4(b)], with a minimum equal to the signal measured from the liquid phase. This indicates the complete melting of the sample. Indeed, 7 mJ is the expected value for melting a sample of these dimensions, taking into account reflection losses on optical surfaces. For lower energies the minimum value reached by the depolarization signal depends on pulse energy, indicating a partial melting. The waveforms shown in Fig. 4(a) are measured with 7.5 mJ/pulse and averaged over 1000 pulses. The loss of birefringence is completed in ∼ 0.3 ms. This is a much larger timescale than that reported for the first events in breaking the ice lattice at room pressure [4]. In the present case in fact we measure the evolution of a property related to the loss of the crystal lattice, in contrast to the local effects, related to hydrogen bonding, studied by transient infrared spectroscopy. The signal recovery, due to freezing, can be fitted in both samples as a double exponential. The time constants for the exponential decay are τ₁ = 2.75 ms, τ₂ = 28.7 ms for the room pressure sample, and τ₁ = 1.98 ms, τ₂ = 39.6 ms for the high pressure one. τ₁ could be related to freezing (nucleation and growth) processes [15–17], while τ₂ could be related to thermal diffusion in the water sample. The influence of thermal diffusion on the measured waveform will be investigated performing measurements with cells windows having different thermal conductivities and samples of different thickness, as well as performing measurements as a function of temperature.

 

Fig. 4 (a) Birefringence signal measured pumping at 5 Hz with 7.5 mJ/pulse of the T-jump beam. Black curve: room pressure, T₀ = −2°C. Red curve: P = 1.12 GPa, T₀ = 23°C. (b) Maximum birefringence signal variation as a function of the pulse energy.

Download Full Size | PDF

4. Conclusions

Successful generation of high-energy pulses at 1.9 μm, with 15 ps duration, allowed to melt bulk samples of water in different P, T conditions. Pressure and temperature are independently tuned and this allows us to investigate the involved processes in a wide range of conditions found in nature. Melting in a timescale similar to the intrinsic thermalization timescale of the H-bonded lattice of ice opens the possibility of applying time-resolved spectroscopies, such as infrared transient absorption, to investigate dynamics at phase transitions at a molecular level.