1. Introduction: Fundaments of Femtosecond-Laser Material Processing
Advances in laser technology have produced lasers capable of emitting ultrashort pulses, with duration at the scale of femtoseconds. This characteristic have revolutionized laser material processing, a subject of research since the laser’s first demonstration, which opened new possibilities for optical and photonic device development [
1,
2,
3]. In femtosecond-laser micromachining, enough energy from the laser light is deposited onto the material to cause superficial or volumetric permanent change on the order of micro/nanometers. The specific features of the obtained microstructures depend on experimental parameters, such as wavelength, pulse energy, repetition rate, pulse duration, focusing objective numerical aperture and scan velocity. From a material science point of view, its optical and thermal properties are crucial to determining how matter will respond to intense light irradiation.
Once material modification at ultrashort pulse regime is related to aspects fundamentally different from those that rule the micromachining process with longer pulses, an understanding of light-matter interactions is of foremost relevance to understanding material processing. One of the main reasons for these differences lies in the fact that at an ultrashort pulse regime the laser energy deposition occurs within the pulse duration, i.e. before any relaxation or thermalization processes have been initiated. Material electronic configuration is responsible for the energy absorption and primary heating, while the atomic lattice remains nearly unaltered. Heat transfer from highly excited electrons to ions and further thermalization followed by diffusion begins long after the pulse has left the material [
4]. In this scenario of heavy non-equilibrium conditions, the decoupling between electronic and lattice systems allows their temperatures to be treated independently according to the well-known Two-Temperature Model [
5,
6,
7].
Another main reason why femtosecond-laser micromachining provided new avenues for material processing lies in absorption, which is governed by nonlinear processes due to the extremely high peak intensities delivered by ultrashort pulses. Although the promotion of nonlinear effects deeply influences the microfabrication of absorptive materials, the most striking phenomenon is the nonlinear absorption in transparent materials. Linear absorption effect occurs when an electron is promoted from the ground to the excited state by one-photon absorption. Clearly, the energy
hν of this single photon has to be enough to overcome the energy gap
Eg between those states, otherwise absorption does not take place. In nonlinear absorption, a group of photons with insufficient individual energy to perform the linear absorption acts together to produce material excitation and, eventually, ionization. There are two regimes involved in the nonlinear ionization: the photo-induced and avalanche mechanisms. The first one suggests that promotion of electrons to the conduction band is due to the laser field, specifically by either multiphoton ionization or tunneling. Multiphoton ionization is described by the simultaneous absorption of
n photons by a single electron, which is consequently excited from the valence to the conduction band. Here, the energy combined of all involved photons has to exceed the bandgap energy in such a way that ionization can occur,
nhν ≥ Eg. Photoionization can also be achieved by a tunneling process facilitated by the distortion of the atomic potential caused by the laser electric field. Both previously described ionization processes are, in fact, part of the same phenomenon modeled by Keldysh [
8] in 1965, differentiating themselves by laser intensity level and wavelength.
Avalanche ionization is the result of a repetitive sequence of events, which includes free carrier absorption and impact ionization. At first, an electron excited to the conduction band absorbs subsequent photons raising its energy up to a level that exceeds the minimum of conduction band energy. Later, this highly energetic electron collides with an electron occupying the top of the valence band and, through this impact ionization, both electrons end up in the lowest level of the conduction band [
9,
10]. The process repeats for the excited electrons, promoting new electrons from the valence band to the conduction band. This mechanism continues to occur during the presence of the intense laser light and leads to an exponential growth in the population of the excited electrons. A requirement for the avalanche ionization is the presence of seed electrons in the conduction band so that the process can start. In the case of femtosecond-laser pulses, these seeds mostly originate from the photoionization mechanisms mentioned before, but defect states close to the conduction band may facilitate the seed electron creation. The light absorption is strongly enhanced by the plasma formed as the electron density in the conduction band increases drastically and its frequency matches the laser frequency. At this critical point, plasma becomes nearly opaque to the incident wavelength and absorbs a great amount of energy that later evolves to energy relaxation and material modification.
In general, femtosecond-laser micromachining is classified into two major types of material modification: ablation and damage. Ablation is the process that usually occurs at the surface of the target material, in a timescale on the order of hundreds of nanoseconds, resulting in material removal. The rule of nonlinear absorption in femtosecond (fs)-laser ablation has been recently investigated [
11]. The non-equilibrium condition brings extreme complexity to the physical mechanisms involved during ablation. Among the most accepted ones are the Coulomb explosion [
12], material ejection, and evaporation [
13]. One of the great achievements of the ultrashort pulse ablation is the capability of producing a minimal heat-affected zone surrounding the laser spot region. It is because significant deposited energy is taken away with the early stages of the material removal and less heat is diffused into the lattice.
The process of bulk damage takes place in transparent material. Once nonlinear absorption is highly dependent on laser intensity, only at the focus there is enough intensity to produce ionization leading to optical breakdown. Such a characteristic enables the microfabrication of three-dimensional structures into the material bulk, without modifying regions outside the focal volume of the laser beam, inscribing, for example, waveguides [
14] and photonic crystals [
15]. Novel applications of fs-laser micromachining in micro-supercapitors [
16], microlenses [
17] and biosensors [
18] have also been proposed and demonstrated.
There are two main structural changes observed in bulk damage experiments. By using laser energy level close to the material threshold, changes in the refractive index [
3,
19] can be produced, which is attributed to material melting and subsequent fast resolidification [
20,
21,
22]. This variation in the refractive index can be positive or negative, according to an increase or decrease in the material density that is accompanied by the rapid cooling [
23]. In the opposite energy level range, well above the material threshold, the production of empty voids [
24,
25,
26] due to microexplosions that lead to a hollow or less dense regions in the focal volume (where material has been expanded) is observed [
27,
28]. When using an intermediate energy level, the induced modification can result in an alternated composition variation, denominated nanograting, which leads to a birefringence in the microstructured region [
29]. Other review papers in this field, presenting not only applications but also fundamental aspects of fs-laser micromachining, have been recently published [
30,
31].
As previously mentioned, produced damage depends on laser and material features, as well as on the experimental conditions, thus, different results can be observed in fs-laser micromachining. Optical and thermal properties of materials are crucial to determine how the material responds to intense light irradiation. Experimental parameters, such as pulse duration, pulse energy and numerical aperture of the focusing lens define the light intensity, which along with laser wavelength define the induced optical nonlinearity. Among the experimental parameters, the laser repetition rate plays an important role in thermal effects caused by fs-pulses. Because laser energy is first deposited into the electronic system and because it is only after the highly excited electrons are thermalized with the lattice that heat diffusion occurs, heat is carried away from the focal volume in a time scale of the order of 1 μs. Therefore, the interval between adjacent pulses directly influences the microfabrication process, which can be carried out in two distinct regimes. When using fs-oscillators that emit pulses with energy on the order of nJ at a repetition rate of MHz, the micromachining exhibits a cumulative behavior, since pulses are separated by a time much shorter than the characteristic time for heat diffusion. In this scenario, the energy deposited from a pulse cannot be dissipated before the next pulse arrives at the same place. Therefore, after sequential pulses, the accumulated heat inside and outside of the focal volume reaches critical level, leading to structural change. The focal volume acts as a heat source of micrometer dimensions emanating heat for as long as the laser exposure time. In contrast, amplified laser systems deliver pulses with energy on the order of μJ and with repetition rate in the range of kHz. In this way, the micromachining takes place in a repetitive regime, since pulses are separated by millisecond, which is significantly larger than the heat diffusion time. As a consequence, the material returns to equilibrium at room temperature before the next pulse strikes it again. This means that produced microstructures are restricted to the focal volume, presenting very little collateral damage in its vicinity [
32].
In the next sections, we present some details about how the fs-laser can be used to modify distinct materials at surface/bulk, aiming particularly at applications in optical waveguides, photonic structures and superhydrophobic surfaces.
2. Experimental Aspects of fs-Laser Material Processing
Although there are several fs-laser-based techniques employed for obtaining materials at micro- and submicro scales, one of the most common approaches relies on focusing the fs-laser into the sample, either in its volume or on its surface, depending on the desired goal. From the experimental viewpoint, the laser light is focused into the sample using microscope objectives with different numerical apertures (NAs), as shown in
Figure 1. Alternatively, depending on the desired size of the features, lenses with larger focal distances can be used. Hence, as illustrated in
Figure 1, a specific pattern is produced by either
x-y sweeping the laser beam on the sample, which is kept fixed, with the aid of galvanometric mirrors (
Figure 1a), or by using a three-dimensional (
x-
y-
z) stage to translate the sample in respect to the fixed focusing lens (
Figure 1b). In the first case (
Figure 1a), the movement in the
z-axis is achieved by an additional translation stage. The choice of which approach should be used depends on experimental details, such as processing time and area, laser repetition rate, etc. The movement of the sample or laser beam deflection is computer controlled, which defines not only the processing pattern on the sample but also the material processing speed, which is directly related to the number of pulses per laser spot for a given laser repetition rate. In order to visualize the material processing in real time, usually a CCD camera is coupled to the experimental setup.
When fs-pulses are focused onto a material, the irradiance reached at the focal volume can be high enough to induce optical nonlinearities, i.e., for intense and tightly focused laser pulses, the corresponding electric field can achieve magnitudes comparable to the ones that bind the electrons in atoms or molecules, leading to a nonlinear optical phenomenon (e.g. multi-photon absorption). Because the multi-photon absorption is localized into the focal volume, the resulting structural changes [
33,
34], photoreduction [
35,
36], degradation [
37] and ablation [
38], will also be confined to the vicinity of the focus, which can be exploited to advance materials processing.
The light irradiance required to produce changes in materials is determined by three experimental parameters: pulse energy (E), pulse duration (τ) and numerical aperture (NA) of the focusing objective. Usually, the highest value of E and the shortest pulse duration are specified by the laser system being used. When τ and NA are set, the nonlinear absorption will be exclusively dependent of E. In this circumstance, the threshold energy is defined as the smallest value of E for which changes in the material occur. When E is increased above the threshold, the sample-affected zone is typically larger than the focal volume, whereas for E, close to the threshold, changes are localized to the focal region. The feature size achieved by laser processing is essentially determined by NA, which defines the focal width and depth. The geometry of the structured region is also influenced by NA; while spherical features are obtained for high NA (typically higher than 0.6), asymmetric structures are observed when small NAs are used.
Fs-laser processing can be achieved in several types of materials, including polymers and glasses, using energy per pulse on the order of nanojoules, which can be accomplished through fs-laser oscillators that operate with repetition rate in the scale of MHz. When material processing is carried out using such oscillators, the heat diffusion time is longer than the interval between consecutive pulses, which confers a cumulative character onto the process [
39,
40]. On the other hand, when amplified femtosecond sources are used, which generally operate with a kHz repetition rate, the nature of the micromachining process is repetitive [
39]. Ti:sapphire laser oscillators, operating at a wavelength of approximately 800 nm and repetition rate of MHz, are still the most used laser sources for fs-laser processing of materials, although other sources based on fiber femtosecond oscillators have gained significant attention. The quality, resolution and properties of the fs-laser-processed samples are usually evaluated by microscopy techniques, such as optical, atomic force and scanning electron microscopy, as well as by standard spectroscopic tools, such as UV-Vis, Fluorescence and Raman.
Other experimental setups that include interference and diffraction processes have also been developed, allowing the fabrication of complex microstructures without the need of scanning the laser beam or translating the sample. For example, arbitrary shapes were microfabricated by using Spatial Light Modulators, which are used to apply amplitude/phase masks to the beam prior to focusing [
41,
42]. Diffractive optical elements (DOE), such as diffractive beam splitters play an important role in the multifocal micromachining [
43] and in the fabrication of periodic microstructures [
44,
45,
46]. The basic idea behind the use of DOE is to produce multiple laser beams that are later focused at different angles, producing an interference pattern. Such modulation in the laser intensity is then transferred to the structure micromachined in the sample.
Besides the micromachining setups mentioned previously, multiphoton polymerization (MPP) can also be used to fabricate micro- and nanometer-sized three-dimensional structures. In this approach, a photoinitiator molecule, when excited via multiphoton absorption, triggers a polymerization reaction, allowing the fabrication of microstructures [
47]. The spatial resolution achieved in such microstructures can be drastically improved by incorporating the concept of stimulated emission depletion (STED) [
48], which can reduce the resolution to tens of nanometers [
49]. Many applications could benefit from the combination STED-MPP, particularly the fabrication of photonic crystals operating in the visible range of the spectrum that require periodic structures on the order of few hundreds of nanometers [
50].