1. Introduction

Laser-induced breakdown through optical absorption is a result of the interaction between high intensity ultra-short laser pulses and matter. The initial energy deposition consists of the generation of localized highly ionized plasma with a typical density of one electron per hundreds-of-molecules in the focal region of a laser beam [1–3]. In liquids, laser breakdown is accompanied by filamentation; an important aspect of femtosecond laser propagation that is characterized by a number of nonlinear effects including self focusing, intensity clamping, propagation at high intensity beyond Rayleigh range and supercontinuum white light generation [4]. It is one considerable interest over the past decades due to its potential applications ranging from micro- and nano-processing of dielectric materials for biomedical and therapeutic applications such as optical-to-acoustic conversion in photoacoustics and bubble dynamics in ultrasound generation [5–7].

Acoustic signals with longer propagation length give rise to localization of thermal expansion in 3D space and time that opens feasibility to probe materials which are optically turbid and not transparent. Moreover, it opens new possibilities to probe IR properties of materials with sub-wavelength nanoscale resolution which cannot be accessed by optical IR spectroscopy techniques due to larger focal spots [8]. For example, a spectral scan with focal spot placed at the atomic force microscope (AFM) tip with simultaneous height detection by AFM delivers absorption spectra via local absorption-heating-expansion [9]. With the current state of the art in 3D triangulation and imaging, there is a strong possibility of photoacoustic techniques to enter wider fields of applications in biomedical and material science engineering.

Femtosecond laser-induced acoustic wave generation contributes to a wider range of frequency emission spectra as compared to the case of nanosecond laser. In the case of nanosecond lasers, longer pulse duration leads to slow heating which is accompanied by a complex mechanism such as thermal expansion, cavitation, and bubble formation. An amplified photoacoustic performance of reduced graphene oxide-coated gold nanorods with photoacoustic amplitudes in the 4–11 MHz range was observed with 5–7 ns laser pulses [10]. The quantitave imaging of microvasculature in deep tissues with spectrum-based microscopy up to 10 MHz was reported using 8–10 ns laser pulses [11]. On the other hand, femtosecond laser pulses with shorter pulse duration induce rapid heating which yields to the generation of acoustic wave with wider frequency range. It can easily generate optical breakdown in water through multiphoton and impact ionization [12]. Two-photon absorption-induced photoacoustic imaging using 50 fs laser pulses showed a wide frequency emission spectra up to 30 MHz [13]. A broadband emission up to 20 MHz from the propagation of intense 45 fs laser pulses in water was reported [14]. Enhanced photoacoustics from gold nano-colloidal suspensions under 40 fs laser excitation was observed with a wide frequency range up to 25 MHz [15]. These aspects of photoacoustic generation over a wide range of frequencies by femtosecond laser pulses are highly beneficial for imaging technologies with high spatial resolution.

Pulse pre-chirping is known to have a strong impact in nonlinear dynamics and has been widely used in recent years. An optimal negative input chirp makes an equal spatial focusing and temporal compression lengths, yielding enhancements in non-linear effects resulting in high intensities, long plasma channels, and broader spectra and possibility to remote spectroscopy [16,17]. Hatanaka, et al. studied the chirp effect on X-ray emission intensity from CsCl aqueous solution jet irradiated by femtosecond laser pulses. The negatively-chirped pulses of approximately 240 fs duration produced up to ten times higher X-ray intensity as compared with transform-limited 160 fs pulses [18]. Similar studies were conducted by Yamaoka, et al. photoacoustic microscopy using ultra-short pulses with two different pulse durations in the range of femtoseconds to picoseconds. It was demonstrated that the required number of photons in photoacoustic microscopy using ultrashort optical pulses (t_p = 250 fs) was about 1/1000 lower than sub-nanosecond optical pulses (t_p = 600 ps) [19].

Nanoparticle-facilitated absorption of pulsed laser leads to the production of MHz ultrasound waves used for photoacoustic techniques. In particular, plasmonic Au nanoparticles have been investigated due to their biocompatibility and superior optical properties such as large absorption cross-section and spectra selectivity in the visible and near infrared ranges based on surface plasmon resonance. The mechanism of interaction between pulsed laser and Au nanoparticles is attributed to non-radiative relaxation dynamics of surface plasmon oscillations [19–27]. Tunability of surface plasmon resonance wavelength to the near-IR region where human tissue exhibits high sensitivity and transmission is considered to be highly beneficial. When Au nanoparticles are irradiated with ultra-short laser pulses, a rapid increase in temperature resulting in photothermal and photoacoustic effects was used for highly localized cell damage in photothermal therapy and contrast agents in photoacoustic imaging were used to visualize temperature maps in a desired location in tissues [28–33].

Here, we show the generation of MHz frequency coherent ultrasound by fast energy deposition with ultrashort laser pulses. Chirp effects on amplitude of photoacoustic signal and spectral properties of Au colloidal nanoparticles of spherical and rod-like shapes were systematically studied for generation of augmented photoacoustic signals.

2. Samples and procedures

Colloidal suspensions of Au nanospheres and rods for the ultrasound generation were prepared via syntheses described elsewhere [15]. Au nanospheres with an absorption peak at 520 nm (diameter of 20 nm) and nanorods with transverse and longitudinal absorption peaks at 520 nm and 800 nm (dimensions of 12 nm×35 nm) respectively were used for photoacoustic detection and measurements. The longitudinal band of Au nanorods was controlled by changing the aspect ratio of nanorod depending on the desired wavelength of interest. The 800 nm longitudinal absorption band of Au nanorod with aspect ratio of ~2.92 which coincides with the wavelength of femtosecond laser at 800 nm was chosen for the ultrasound generation. Au nano-colloidal suspensions with atomic concentration of ~1.4×10⁻⁴ mol/L, particle concentration of ∼ 4.21×10¹⁸ NPs/mL and volume of ~4×10³ nm³ were used in the experiments.

Figure 1(a) shows the experimental set-up. Femtosecond laser pulses (Mantis, Legend, Elite HE USP, Coherent, Inc.) were tightly focused by an objective lens (10×, NA = 0.28, M Plan Apo, Mitutoyo) into a 5-mm glass tube inside the water tank. The sample suspension in the glass tube was circulated continuously by a conventional pump. The photoacoustic measurements were carried out with a hydrophone (HNA-0400, ONDA) in the detection frequency range at 1–20 MHz and an ultrasound preamplifier (5678, Olympus). The distance between the hydrophone and the glass tube was kept constant at 1.5 cm. The signals were recorded and analyzed with a digital oscilloscope (DSO-X 3034A, Agilent Tech.). The geometry of glass tube in water was used to simulate the experiments with biomedical relevance where the photoacoustic signal generation is separated from detection. The white light continuum spectra was measured by a spectrometer (SD2000, Ocean Optics, Inc.).

 

Fig. 1 (a) Generation of ultrasound in aqueous solutions of Au nanoparticles with fs-laser pulses. Laser pulses are focused inside the glass tube using 10× numerical aperture NA = 0.28 objective lens. (b) Side-view of the optical image of white light continuum (WLC) in water at the focal region at different pulse energies E_p = 30,100 μJ/pulse at repetition rate of 1 kHz and pulse duration t_p ≃ 40 fs; the corresponding power per pulse P_p = 0.75,2.5 GW/pulse. Distance between focal spot and hydrophone was set 1.5 cm in all experiments. Arrows mark pulse propagation direction.

Download Full Size | PDF

Near-IR femtosecond laser pulses used for ultrasound generation: the pulse duration of t₀ = 40 ± 5 fs at the central wavelength of λ = 800 nm, the pulse energy was typically from E_p = 0.1 mJ to 1 mJ, at 1 kHz repetition rate. A pulse can be described by a Gaussian temporal envelope of the form $E\left(t\right)=E_0{e}^{-2\ln 2{\left(t/t_p\right)}^2}\cos \left(\omega t+\beta t^2\right)$, where ω is the cyclic frequency of light, t_p is the pulse duration at the full-width at half maximum (FWHM), β [1/fs²] is the linear chirp, and $E_0=\sqrt{2I_0/\left(c{\epsilon }_{0}n\right)}$ is the field amplitude, n is the refractive index, c is speed of light, t is time, I₀ = 2I_av is the peak intensity which is twice larger than the average, I_av for the Gaussian, and ε₀ is permittivity of vacuum. The instantaneous cyclic frequency ω_ins(t) = ω₀ + 2βt, where β > 0 corresponds to the positive chirp with trailing high frequency components.

Pulse pre-chirping is implemented to increase pulse duration up to ten times using 1D array of liquid crystal cells (FemtoJock, Biophotonics Solutions, Inc.). For the slow frequency with ω varying spectral phase φ(ω) it can be expanded into the Taylor series around the central frequency ω₀ with the few first terms as $\phi \left(\omega \right)=\phi \left({\omega }_0\right)+{\mathrm{\Phi }}_1\omega +\frac{{\mathrm{\Phi }}_2}{2!}{\omega }^2+\dots ,$ where φ(ω₀) is the absolute phase of the pulse in time domain, the first derivative φ′(ω₀) = Φ₁ is the group delay (GD) which defines a shift of envelope in time domain, φ″(ω₀) = Φ₂[fs²] is the group delay dispersion (GDD) or the second order dispersion which defines the chirp in time domain. Duration of the time broadened pulse at FWHM is defined as $t_p=t_0\sqrt{1+{\left[4\ln 2{\mathrm{\Phi }}_2/t_0^2\right]}^2}\equiv t_0\sqrt{1+{\beta }^2}$ where t₀ = 40 fs is the shortest spectral bandwidth limited pulse duration. For this study higher orders as well as Φ₁ were set to zero.

3. Coherent ultrasound generation: theory

Generation of coherent ultrasound by laser pulses is not an efficient process and depends on the absorbed energy, E_abs, and acoustic energy of coherent ultrasound, E_ac [34]:

Despite low efficiency, a substantial pressure amplitude of the coherent sound can be generated [34]:

4. Results and discussion

4.1. Water

Generation of white light continuum (WLC) by ultra-short laser pulses in wide bandgap materials is an efficient process when photon energy is less than the half of the bandgap of the host matrix; for shorter wavelengths two-photon absorption becomes an efficient absorption mechanism. Water absorption at ∼6.2 eV energy was for a long time assumed as the band gap energy. Recently, it was demonstrated that this energy corresponds to production of solvated electron complex out of the valence band in pure water [35] while the bandgap of the water is ∼9.5 eV. The use of fs-laser pulses with central wavelength of 800 nm is appropriate for efficient WLC generation. Side-view WLC images from the focal region reveals localization of emission around the focal region which would have an axial extent of $2z_R=\pi w_0^2/\lambda \simeq 24\mu \mathrm{m}$ with a waist of the beam (radius) w₀ = 0.61λ/NA ≃ 1.7 μm for the NA = 0.28 objective lens; the estimation assumes a Gaussian beam profile and aberration free linear focusing. The average power used in experiments for the E_p = 100 μJ pulse was I_av = 2.5 GW. Nonlinear WLC and filamentation are both related to self focusing. In water and glass, a strong WLC generation is observed upon excitation by near-IR fs-laser pulses. For a pulse duration t_p ≃ 40 fs, pulse power becomes larger than the critical self-focusing threshold P_cr = α₁λ²/(4πn₀n₂), where n₂ is the nonlinear refractive index due to Kerr effect n(I) = n₀ + n₂I with constant α₁ = 1.8962 for a Gaussian pulse [36]. For water at λ = 800 nm, n₂ = 4.1×10⁻¹⁶ cm²/W [37], refractive index of n = 1.33, the critical power of self focusing is P_cr ≃ 1.77 MW. However, filamentation along the propagation was not very strong L ∝ 100 μm (Fig. 1(b)). Filaments for photoacoustic generation of sound by WLC in glass (for silica n₂ = 3.5 × 10⁻¹⁶ cm²/W [37]) using longer t_p ≃ 150 fs pulses at the same λ = 800 nm wavelength [38] showed considerably longer traces. The better axial localization of energy delivery in water can be due to a strong light scattering and bubble formation along light propagation in the focal region. Such localization and energy deposition is helpful for photoacoustic applications due to a better localized pressure wave time-arrival onto the hydrophone.

Figure 2 shows WLC spectra for different pulse durations controlled by GDD, Φ₂. Negative Φ₂ values correspond to a spectrally blue-shifted wavelengths preceding the red-shifted ones, while for the positive chirp (Φ₂ > 0) the blue-part is trailing the red. As expected, the most spectrally broad spectra were generated when the pulse duration was the shortest, or Φ₂ = 0 (shaded region in Fig. 2). The highest peak intensities are reached at such setting; i.e., for the shortest pulse. There was a small difference between spectra of WLC generated with positive and negative chirp with the same absolute values indicating that the phenomenon is mostly dependent on intensity rather ionization which has a strong wavelength dependence.

 

Fig. 2 Generation of white light continuum (WLC) with positively Φ₂ > 0 (a) and negatively Φ₂ < 0 (b) chirped ultra-short laser pulses in water. Spectra are plotted as lg(Intensity). Arrows mark trend with increase of pulse duration. Pulse energy was E_p = 100 μJ/pulse.

Download Full Size | PDF

The representative photoacoustic signals in time domain generated by Au nanosphere colloidal suspensions were shown in Fig. 3; negatively-chirped pulse (800 fs) (a), transform-limited pulse (40 fs) (b), and positively-chirped pulse (800 fs) (c). The first peak represents the fundamental ultrasound signal (a presssure wave) and the second peak corresponds to the internal reflection of ultrasound waves from the inner surface of the glass tube. The time interval between the two peaks is ascribed to the difference in travel distances of the two signals in water solutions where the speed of sound is 1500 m/s [15]. The first peak signal amplitude was used as a measure of photoacoustic response from femtosecond laser-irradiated sample solutions. It was observed that higher photoacoustic signal amplitude was exhibited by negatively-chirped pulses rather than positively-chirped pulses. On the other hand, irradiation of transform-limited pulses result in lowest photoacoustic signal amplitude.

 

Fig. 3 Representative photoacoustic signals in time domain generated by Au nanosphere colloidal suspensions: negatively-chirped pulse (800 fs) (a), transform-limited pulse (40 fs) (b), and positively-chirped pulse (800 fs) (c) at pulse energy of E_p = 100 μJ/pulse.

Download Full Size | PDF

Figure 4 shows amplitude of photoacoustic signal of hydrophone at different frequencies over 1–20 MHz range. It is clearly observed that the shortest chirp-free pulses induce the least efficient ultrasound generation in pure water as well as in Au nano-colloidal solutions. This is consistent with the acoustic pressure scaling with parameter A = αc₀t_p (Theory sec. 3). Longer pulses generated stronger acoustic (pressure) signals. The central wavelength was at 800 nm which was close to the longitudinal (L-mode) resonance of nanorods [Fig. 4(c)]. The insets in Figs. 4(b) and 4(c) shows the calculation results of the light intensity distribution around single spherical and rod-shaped nanoparticles. Up to ten times more intense light field localization is expected for the resonance case Fig. 4(c). Au nanoparticles are homogeneously distributed and randomly oriented in the solution phase. In the case of Au nanorods, the relative orientation of nanorod particles and the incident laser polarization is not always the same as expected in the calculation, while in the case of Au nanospheres the calculation reflects the real condition of the solution. Therefore, as shown in Figs. 4(b) and 4(c), it is reasonable that the ultrasound intensity of nanorod is lower than expected from the calculation.

 

Fig. 4 Generation of photoacoustic (PA) signal at different values of positive and negative chirp values in water (a), Au nanosphere (b), and nanorod (c) solutions. Pulse energy was E_p = 100 μJ/pulse. Insets in (b,c) shows a light intensity |E|² distribution around nanosphere and nanorod for linearly polarized incident field E = 1 at a close to resonant condition (L-mode) for the nanorod (finite difference time domain code FDTD-Lumerical).

Download Full Size | PDF

In our previous work, it was demonstrated that higher photoacoustic intensity was exhibited by Au nanorods which is attributed to non-radiative relaxation dynamics of surface plasmon oscillations [15]. Au nanorods maximize the absorption of pulsed light which leads to an enhanced photoacoustic intensity. Similarly, Hatanaka, et al. [39] performed a study on the influence of temporal chirp on X-ray emission. The highest X-ray intensity was observed at the longer pulse width t_p = 372 fs as compared to the shortest pulse width t_p = 45 fs. The negative chirp was, also, slightly more efficient in X-ray generation. Spectral shape of the photoacoustic signal was different for different colloidal solutions as discussed next.

4.2. Au nano-colloidal suspensions

Spectrally broad WLC could be considered advantageous in delivering energy by absorption to Au nanoparticles in water, since Au has interband transitions at shorter wavelengths and spectrally broad extinction peak which is size and shape dependent. Apparently, in pure water and colloidal solutions sound generation followed similar scaling at different acoustic frequencies and showed stronger pressure waves formed for longer pulses (Fig. 4).

Interestingly, the photoacoustic signal was the smallest then WLC was the most spectrally broad at Φ₂ = 0 (Fig. 5). Figure 5 shows efficiency of photoacoustic response for different chirp settings (correspondingly different pulse durations). Negative chirp was slightly more efficient in sound generation as can be evidenced by steeper slope of the dependence (Fig. 5). However, the key control parameter to obtain strong photoacoustic signal was the pulse duration. Longer pulses were more efficient in sound production as follows from analytical prediction Eq. (2).

 

Fig. 5 (a) White light continuum (WLC) spectral width at different pulse durations controlled by chirp. (b) The photoacoustic signal integrated over the range of hydrophone 1 – 20 MHz as a function of pulse duration defined by the GDD, Φ₂. Duration of the time broadened pulse at FWHM is defined as $t_p=t_0\sqrt{1+{\left[4\ln 2{\mathrm{\Phi }}_2/t_0^2\right]}^2}$, where t₀ = 40 fs is the shortest spectral bandwidth limited pulse duration. Note, the vertical axis of WLC spectra in Figure 2 is logarithmic. Diameter of Au spheres was 20 nm and rods were 35-nm-long and 12-nm-wide. The slope difference between the dashed lines is 1.79 times. Pulse energy was E_p = 100 μJ/pulse.

Download Full Size | PDF

Speed of sound in gold c₀ = 3240 m/s is 2.19 times higher than in water, 1482 m/s. The slope difference of photo-acoustic signal in the case of water and Au nanorods is approximately 1.89 for the long pulse side [see lines at the negative chirps; Fig. 5(b)]. This might indicate a changing contribution to thermal expansion detected by transducer from water to increasingly from gold nanorods as pulse duration gets longer. Speed of sounds enters parameter A ≡ αc₀t_p which governs photo-acoustic amplitude (see, Eq. 3).

In pure water, there was strong nonlinearity in the photoacoustic signal generation around Φ₂ = 0. It was almost absent for Au nanoparticle solution. This can be understood by highly nonlinear absorption mechanism in water and linear absorption and avalanche evolution via free electron absorption in the case of Au nanoparticles. By changing chirp, an instantaneous laser frequency is changing accordingly, ω_ins(t) = ω₀ + 2βt. As governed by Eqn. 3, the strongest photo-acoustic signal is expected for the longer pulse and stronger absorbance, α. The steepest changes in PA amplitude where, indeed, observed for the Au nanorods at negative chirp. Since the nanorod had extinction (absorption and scattering losses) around 800 nm wavelengths, pre-chirping was also affecting spectral overlap of plasmonic resonance and the WLC spectrum centered around 800 nm.

5. Conclusions and outlook

It is demonstrated that negative chirp of ultra-short laser pulses at the wavelength overlapping with an extinction resonance of Au nanoparticles achieves the most efficient ultrasound generation at MHz-frequency range. Accordingly, it has been shown that size homogenization of Au colloidal dispersions can be achieved by WLC reabsorption after an Au foil ablation in pure water and a surfactant-free stable colloidal solution can be obtained [40]. Pulse pre-chirping effect on photoacoustic intensity and WLC emission spectra exhibited an opposite tendency; longer pulse duration generates higher photoacoustic intensity and narrower WLC spectra, while shorter pulse duration generates lower photoacoustic intensity and broader WLC spectra. This tendency is beneficial for the determination of the optimum pulse duration that can be used in photoacoustic imaging techniques.

Funding

SJ is grateful for partial support via the Australian Research Council DP130101205 Discovery project and by the nanotechnology ambassador fellowship program at the Melbourne Centre for Nanofabrication (MCN) in the Victorian Node of the Australian National Fabrication Facility (ANFF).