1. Introduction

In view of the demand for fabrication of high quality devices for diverse applications such as photorefractive elements, or precursors for the attachment and adhesion of biological materials, various microprinting techniques have been developed. One of them, LIFT, compared to other microprinting techniques is a direct write process, useful for the transfer of a broad range of materials such as dielectrics [1], semiconductors [2], metals [3],superconductors [4] and biological materials [5] for various applications in microelectronics,optoelectronics and biology. In the LIFT technique, a single laser pulse is focused through a transparent substrate onto a thin film of the material to be transferred [6,7]. A micrometersized part of the film material is ejected and is transported onto a receiver surface. The control of the morphology and the size of the structures obtained are of great importance for the success and the implementation of the targeted application. Nanosecond lasers were initially used for the LIFT process, but the technique has advanced by using femtosecond lasers which give the benefit of reduced heating effects and precise material processing [1,8].

Experimental parameters like beam intensity and carrier to receiver distance can, to a certain degree, affect the properties of the transferred material. However, with a single laser pulse, as used until recently, it is not possible to intervene and exploit the fundamental processes that occur in the material following ultrafast laser excitation. For example, energy transfer from the electron system to the lattice can occur in time windows that extend well beyond the laser pulse duration [9]. In laser ablation of metals it has been found that energy dissipation by electronic transport affects the intensity of the desorption [10]. In contrast, in wide band gap semiconductors, strong electron-phonon coupling due to carrier trapping modulates the optical damage on the material’s surface [11]. Generally a hydrodynamic model is assumed to describe the complex process of material removal during ablation of surfaces.According to this model, the energy transfer from the electronic system to the lattice and heat dissipation are considered to describe the ultra fast phase transitions from the solid phase to a super heated one, resulting in removal of material in a mixture of a molten and a fractured phase [12]. In LIFT, of metals in our case, where the fluences used are generally lower than those traditionally used for ablation, and the dispatch of material is the result of pressure build-up in a confined geometry [6,13] and where the morphology of the transferred film is important, it is not yet clear which mechanisms are responsible for the spatial characteristics of the transferred material. A promising but still unexploited way of accessing these fundamental processes during LIFT of metals and therefore possibly tailoring the properties of the transferred film is the use of temporally shaped laser pulses.

Temporally shaping of laser pulses by Fourier synthesis of the spectral components of an ultra short laser pulse has been proven to be an effective technique to control and direct a variety of physical and chemical processes [14]. It has been applied to control ionization processes [15], to improve the emission efficiency of high harmonic soft X-Rays [16] and more recently for materials processing [11,17].

In these first experiments we used two pulses of equal intensity obtained by this technique with the aim to demonstrate the control which may be achieved in the LIFT processing technique. Metallic thin films of Au, Zn and Cr have been chosen for these experiments because of their broad range of electron-phonon coupling constant values, g, in order to investigate the influence of the strength of electron relaxation towards the lattice to the morphology of the transferred films. The coupling constant ranges from g(Au)=2.1×10¹⁶ Wm^-3K^-1, to g(Zn)=6.4×10¹⁶ Wm^-3K^-1 to g(Cr)=47×10¹⁶ Wm^-3K^-1 [18–20]. It is demonstrated that by controlling the temporal separation of double laser pulses we are able to control the microprinting process, in regard to the size and the morphology of the transferred material. Furthermore, pump-probe transient absorption measurements have been employed in these samples to investigate their femtosecond dynamics at intensities lower than the damage thresholds. Our experimental results strongly suggest that for Cr and Zn the morphology of the transferred structures is affected by the strong coupling of the electrons with the lattice. In contrast, in Au where this interaction is weak, it is the fast ballistic electron motion and the resulting multiple scattering of electrons with the film’s surfaces as well as the internal electron thermalization that mainly influence the morphology and size of the transferred film.

Exploitation of the time scales in which ultrafast processes like electron-phonon scattering and electron ballistic motion occur, may allow the control of the spatial characteristics of the transferred material for optimizing the outcome of the LIFT process in a variety of research, analytical and diagnostic applications.

2. Experimental details

The LIFT experimental set up has been described elsewhere [1]. In this work thin,polycrystalline Au, Zn and Cr films with a thickness of 40 nm deposited on quartz were used as carrier substrates, while Si(100) was used as a receiver surface. The carrier-receiver pair was placed in a miniature vacuum chamber, at a distance of 100 μm, which was pumped down to a pressure of 10^-2 mbar. Amplified Ti:Sapphire laser pulses at 800 nm, 1 kHz repetition rate, with durations of 200 fs were used. The repetition rate was scaled down electronically to 1 Hz. Prior to amplification a programmable liquid crystal Spatial Light Modulator (SLM) was inserted into the Fourier plane of a 4f zero-dispersion configuration [21], allowing temporal pulse shaping of the incoming beam to two pulses with the temporal separation determined by phase modulation. The phase mask for the generation of the pulse shapes was determined numerically using an iterated Fourier transform method [22]. These generated shapes are then amplified thus compensating for spatio-temporal and energetic fluctuations that are inherent in this system [23,24]. Moreover, by utilizing a SLM to introduce two pulses of various pulse separation, mechanical instabilities that would be unavoidable when using an interferometer are minimized. The resulting double pulses of equal intensity have duration of 200 fs and their temporal separation was programmed to vary from 0 to 10 ps. Representative pulse profiles that are generated with the above method are shown in Fig. 1. The total energy (E_p) irradiating the target was varied from 0.01 to 0.9 μJ and was constant for single and time separated pulses. The amplified shaped laser beam illuminated a 1.6mm diameter circular aperture and was focused onto the carrier film interface using a 20x microscope objective resulting in a spot diameter of approximately 5 μm. One shot of single (Δt=0) or time separated double pulses was used for each transferred dot from the film. The dots were examined with Scanning Electron Microscopy (SEM).

The laser source used for the pump-probe experiments was a Ti:Sapphire oscillator delivering 100 fs long pulses at 800 nm and with a 80 MHz repetition rate. The pump and probe beams are focused onto the samples, close to normal incidence. Lock-in detection is used for measuring the transient changes in the differential transmission, ΔT/T, of the probe beam.

 

Fig. 1. Temporal pulse profiles generated with the method described in the text. Red and blue profiles in (b) are a guide to the eye to represent the underlying double pulses.

Download Full Size | PDF

3. Results and Discussion

In this first study of the effect of pulse separation on the transferred materials, we employ two pulses of equal intensity at various time separations Δt, and total pulse energy E_p. Several dots from each of the samples were transferred and were subsequently inspected and analyzed with respect to their morphology and size. Representative SEM images of the transferred dots are shown in Fig. 2, for E_p=0.3 μJ.

Initially, we determined the energy regimes for optimum transfer. The LIFT threshold was determined for the three materials examined using single laser pulses and found to be about 0.07 μJ for Au, 0.01 μJ for Zn and 0.035 μJ for Cr. These values are in good agreement with earlier reports for Au and Cr [25,26]. The absorption measured for the samples used for the LIFT experiments were 6, 49 and 41 % respectively. Therefore the thresholds of the absorbed energies needed for observing transfer were 0.0042 μJ for Au, 0.0049 μJ for Zn and 0.014 μJ for Cr which seem to scale with the melting temperatures of these materials (1338 K for Au, 693 K for Zn and 2130 K for Cr) [27]. In all the cases the increase in pulse energy resulted in an increase of dot size. For energy values close to the transfer threshold we observed sparse agglomerates for all the materials. Therefore, all the comparisons presented in the following discussion have been made for energy regimes where we can safely define the formation of a transferred continuous film and not just sparse agglomerates. At these intermediate energy values the morphology of the transferred dots are characteristic for each of the materials. Specifically:

For Au and for intermediate incident energies (0.15 to 0.6μJ) the morphology is similar to Fig. 2 (a), Δt=0, i.e. the transferred dots exhibit both a molten and a fractured phase of the material. For Zn when increasing the energy beyond the transfer threshold the agglomerates become denser and a crater appears in the central region of the transferred films (thus making the central region rougher). For incident energy values of 0.07 to 0.3 μJ the morphology of the transferred Zn films look similar (see Fig. 2 (b), Δt=0). Additionally, the presence of agglomerates exists for all the above incident energies used for Zn, thus making the morphology of Zn always rougher in comparison with Au and Cr. For Cr, with increasing the energy from 0.15 to 0.9 μJ the morphology of the obtained dots is similar to Fig. 2 (c), Δt=0,i.e. a smooth transfer with the presence of a splashing on the outside part, with the splashing effect being more intense at higher energies. Note here, that the splashing effect was characteristic of the Cr only and was not observed for the other two materials at any of our incident energy values. For incident energies higher than the intermediate energies for each of the materials mentioned above (higher than 0.6 μJ for Au, 0.45 μJ for Zn and 0.9 μJ for Cr) the film quality deteriorates to the point that we cannot draw any conclusions about their shape and morphology or make any comparisons between the materials. Therefore we have restricted ourselves to these intermediate energy regimes.

Now we turn to the main observations which are drawn when looking at the results depicted in Fig. 2 with regard to the double pulse temporal separations. By splitting the pulse in two separate pulses of half the energy and a temporal separation of 0.1 ps the morphology of the transferred Au dot is significantly altered. For Au, there exists a transition from the rough phase to a much smoother phase that implies that the material has been transferred in a molten state. As can be seen from the sequence of the pictures in Fig. 2(a) for Au, the molten state appears suddenly at 0.1 ps and persists for all pulse separations with no significant variation thereafter, up to 10 ps.

For the case of Zn, the situation is somewhat different (Fig. 2(b)). At Δt=0 only the central part of the transferred material exhibits a rough morphology while the larger, outer part is smoother. With progressing pulse separation, the outer parts of the transferred Zn dots remain unchanged while the central part evolves to a smoother deposition state, which is similar to the case of Au. However, this transition from a rough towards a smooth morphology occurs in Zn with a much slower rate than for the case of Au. Specifically, the crater that appears at Δt=0 and persists for Δt=100 fs, is still distinguishable from the outer parts of the transferred dot at 500 fs and has almost disappeared for Δt=5 ps, while for Δt=10 ps the central region where initially the crater was formed appears to be smoother than the outer regions.

For the case of Cr, the morphology of the deposited dots using total pulse energy E_p=0.3 μJ at various separation times is depicted in Fig. 2(c). In contrast to the results obtained for Au and Zn, no change has been observed in either the single pulse or the double pulse transfer at any separation time. The morphology trends with Δt for all three materials that have been investigated have been observed for all the intermediate energy values used.

Apart from the observations of differences in the morphology obtained for the transferred materials, there exists a systematic change in the diameter of the transferred dots observed for the case of Au. In Fig. 3 we plot the measured average diameter of the deposited dots as a function of the two pulse separation time for various total pulse energies. The error bars in Fig. 3 show the standard deviation of the sizes of several transferred dots that were measured on the SEM machine prior to photograph development. It is evident that the size of the transferred dots increases with pulse energy, a fact which is consistent with previous observations [3]. At the threshold energy necessary for observing transfer i.e. for E_p=0.07 μJ, the diameter remains unchanged with the pulse separation time, within the experimental error.When double pulses with separation times from 0 to 0.5 ps are used, a gradual increase in the size of the dot appears for pulse energies of 0.15 μJ and higher. During the first 500 fs the average diameter increases by 15% for E_p=0.15 μJ, 16% for E_p=0.3 μJ, 51% for E_p=0.45 μJ and 46% for E_p=0.6 μJ. For pulse separations greater than 0.5 ps, the average size of the transferred dots tends to reach a saturation value which is more evident for the higher pulse energies. For the cases of Zn and Cr, no observable variation of the average transferred dot sizes with respect to the pulse separation times has been noticed.

To interpret these results we must first distinguish whether the observed changes are due to processes occurring during, before or after the transfer of the material from the carrier to the substrate. We consider that LIFT is effectively an ablation process under confined geometry [7]. Ablation takes place as a result of a heat wave leading to the formation of a longitudinal pressure wave at the carrier-metal interface [28]. For the material to be removed,the longitudinal pressure wave needs to have a minimum at the metal/carrier interface meaning that the pressure wave needs to travel two times the thickness of the thin films,which is 80 nm in total. The required time for this process is 24.6 ps for Au, 19.2 ps for Zn and 12.9 ps for Cr. These values have been calculated taking into account the sound velocities in these materials (3240 ms^-1 for Au, 4170 ms^-1 for Zn and 6200 ms^-1 for Cr). Therefore we can safely assume that all the changes that we observe during the first picoseconds are made while the material is still stationary. Thus, we may consider that these changes are due to the fast electronic and lattice processes occurring inside the material following ultrafast laser excitation. In order to determine the time scales for these processes we employed a pumpprobe,transient transmission spectroscopy technique [29].

Figure 4 shows the normalized differential transmission, ΔT/T, versus the temporal delay between the pump and the probe pulses for the Au, Zn and Cr films. The time constants for the decays are 3.8 ps and 1.5 ps for Au and Zn respectively, while the response for the case of Cr appears to be instantaneous, i.e. symmetric and following closely the laser pulse temporal profile. The differential transmission is following the induced changes in the refraction index of the materials, Δε, and to a first approximation its decay reflects the rate of electron-phonon scattering and the subsequent lattice heating [30,9]. The rate of the lattice heating in the case of Au is the slowest of the three materials used, and the one for the case of Cr can be considered instantaneous for our experimental resolution. The trends observed for the decay of ΔT/T and therefore for the electron-lattice energy transfer are in agreement with the electron-phonon coupling strengths that are reported in the literature for Cr, Zn and Au [18–20]. Previous measurements in the transient behavior of Au [31], Cr [20] and Zn are comparable with the time scales of our transient results. Small deviations may be attributed tothe variations in the technique, the wavelengths and the properties of the material that areused.

 

Fig. 2. SEM images of dots deposited by the LIFT technique using single (Δt=0) and double pulses of various separation times with a total energy of 0.3 μJ in all the cases. (a) Au (b) Zn (c) Cr. The white arrow on the left side denotes the direction of increasing pulse separation, Δt, for all of the materials.

Download Full Size | PDF

Based on the above considerations, we may interpret the observed differences in the morphology and size in the LIFT experiments on the basis of the ultrafast dynamics recorded for the different materials. Starting from the case of Cr, we did not observe any systematic changes of either the morphology or the size with respect to the pulse separation time. At the same time, the transient transmission data show that the response of the system is extremely fast. Therefore we conclude that all changes of the system if any (morphology, size) occur during the excitation by the first laser pulse, and therefore we can not observe any change in the LIFT process, regardless of the pulse separation. For the case of Zn on the other hand, the pump-probe data show a decay with a time constant of 1.5 ps which corresponds to the rate of electron-phonon scattering. The concurrent observation that the morphology of the transferred Zinc films also changes within a picosecond time scale is a strong indication that electronphonon scattering and the subsequent heating of the lattice is the major mechanism responsible for the observed morphological changes. As long as the second pulse arrives before efficient dissipation of the heat injection due to the first pulse, the illuminated area is subject to an enhanced heating effect which results in a distorted morphology which decays with a picosecond rate.

On the other hand, the situation for the transferred Au dots is significantly different. The morphology changes abruptly within 100 fs and for longer pulse separations it remains mostly unchanged. At the same time, Fig. 3 shows that the dot size increases for pulse separations of a few hundred fs and for longer than 0.5 ps the sizes remain unchanged. Electron-phonon scattering occurs with a time constant of 3.8 ps in Au and therefore it cannot be the process that affects significantly the size or the morphology of the film (which occur within a few hundred fs). Faster processes within the electron system must be more influencial for these spatial parameters of the deposited films. We assume that in the case of Au the as yet nonthermalized electrons that exhibit mainly a ballistic motion [32] with velocities in the range of 106 m/s have enough time to scatter by average multiple times with the film surface and therefore distort it by induced strain or even by Coulombic explosion [33]. This assumption is also supported by the internal electron thermalization time for Au thin films that is in the order of a few hundreds of fs [31]. As long as the pulse separation remains short (Δt<100 fs) the enhanced contribution from the ballistic electrons generated by each of the pulses results in an enhanced distortion in the morphology (Fig. 2(a)). Moreover, we propose that the internal electron thermalization is the mechanism responsible for the observed enlargement in the transferred film diameter for pulse separations Δt<500 fs. The coupling of the electrons to the lattice increases during the thermalization process [30,32]. The rate of electron-lattice coupling therefore increases from the moment of excitation and has been practically maximized by 500 fs for Au [30] and remains constant thereafter. The resulting increased rate of phonon excitation and the subsequent loosening of the lattice may give rise to an increase in the diameter of the transferred films for pulse separations Δt < 500 fs. On the contrary, in the case of Zn the more efficient electron-phonon coupling in the bulk of the film, which is in competition with the ballistic electron motion [32], does not allow for such very rapid changes in either the morphology or the diameter of the films.

 

Fig. 3. Dependence of dots diameter in Au, deposited by LIFT, on double pulse separation time and total pulse energy.

Download Full Size | PDF

 

Fig. 4. Time dependence of normalized differential transmission ΔT/T for Au, Zn and Cr thin films.

Download Full Size | PDF

As mention earlier, the materials in this work were chosen according to their electron-phonon coupling strengths. These values have been measured in different experiments that used fluences much lower than the fluences used typically in LIFT experiments. Usually in experiments where the electron-phonon coupling strength is measured (for example thermoreflectance measurements) the excitation is weak therefore making the approximation that the lattice temperature of the irradiated system remains unchanged and that the electron temperature is elevated by a few hundred degrees. Under such conditions, the electron-phonon strength G is typically considered constant [34,35]. However, in an experiment where the material is under very strong non-equilibrium conditions, as is the case in a LIFT experiment,this may not be the case [35]. In such cases, the broadening of the Fermi distribution can be so strong that density-of-state effects may come into play resulting in a changing electronphonon coupling strength that is dependant on the electron temperature. Therefore, the coupling “constants” of the materials used in this work may vary with time as the electron temperature varies. On the other hand, the interesting effects that are observed in our experiments occur in times where it is not even justified to define an electronic temperature. As a suggestion for future modeling of such phenomena we propose that the possible variations of the electron-phonon coupling strength be related to the energy that is stored to the electron and lattice sub-systems and not to the temperature. Nevertheless, we justify the choice of the materials used in our experiments according to the starting values of their electron-phonon coupling strengths as these values determine to a large extent their initial ultrafast dynamics.

The importance of electron phonon coupling during the ablation process has been shown for a variety of materials. For example, during ablation of wide band gap semiconductors with shaped laser pulses the scattering of electrons from defects and the electron-phonon scattering determine the intensity of the effect as well as the morphology of the material left behind. Stoian et. al. [36] have studied the morphology of Al₂O₃ and α-SiO₂ surfaces following ablation by shaped femtosecond laser pulses. They found that for Al₂O₃ where the electronic decay time lies in the 100 ps regime, the ablated spots remain unchanged for double pulse separation times of the order of a few ps. On the contrary, when using α-SiO₂ in which the electronic decay via efficient electron-defect scattering is much faster (few ps) the picosecond pulse separation modulated the material’s surface. The strong influence of electron-phonon interaction in the ablation process was also shown for metals [10]. Thereby, it was observed that the desorption of metals following ablation with ultrashort laser pulses becomes more efficient when energy is transferred to the lattice via electron-phonon interaction, while fast electronic transport is in competition with the above because it is responsible for fast and efficient energy dissipation away from the surface [32]. Furthermore, we find that in the LIFT of metallic thin films where the imposed boundary conditions play a significant role, electronphonon interaction and efficient scattering of fast ballistic electrons from the film’s surfaces as well as the internal electron thermalization are processes that determine the spatial characteristics (size, morphology) of the transferred films.

Finally, Colombier et al. [37] have employed pulse shaping routines to optimize ablation rates from an Al surface. They found that it may be possible to access different temporal regimes by employing different temporal envelopes of the pulsed excitation in order to select a desired ablation pathway. They conclude that while long, multipicosecond profiles are more suitable to tune the plume characteristics, ultrashort, subpicosecond time profiles, on the contrary, are preferred where precision in laser structuring is needed. These results further support our arguments for the need for accessing primary mechanisms on ultrashort time scales that may have a strong influence on the structural characteristics of the transferred films.

4. Conclusions

In conclusion, we have reported the effect of double pulses on the LIFT process for Au, Zn and Cr. We find that the morphology and the size of the deposited dots can be affected by the temporal shape of the excitation pulse, and depends on the time scales of the ultrafast early stage processes occurring in the material of the film. For Cr and Zn where the electronphonon coupling is relatively strong, we find that the morphology of the transferred films is determined by the electron-phonon scattering rate, i.e. very fast and within the pulse duration for Cr, and in the few picoseconds time scale for Zn. For Au where the electron-phonon coupling is weak but the fast ballistic transport of electrons is very efficient we find that the frequent collisions of electrons with the film’s surfaces determine the film morphology and that the internal electron thermalization rate which controls the electron-lattice coupling strength may determine the films’ sizes. In order to obtain the desired control of the spatial characteristics of the transferred films, pulse shapes accessing the time scales of the above fundamental processes must be used.