CLÁUSULAS Primera. Objeto

The growth mechanisms of the thin films depend on the material system in question and on the thermodynamic conditions during the process. In particular, the energy of the incoming particles and the substrate temperature play an important role. [148] In PLD the particles are typically very energetic with kinetic energies up to several tens of electron volts. This allows deposition of dense, high-quality thin-films at lower substrate temperatures than is required in most other methods.

In addition to deposition of smooth films, PLD can be used to produce nanomateri- als. One approach is PLD in high background pressures, which produces nanoparticles and nanotubes. Furthermore, it can be used to deposit high surface area films. Fem- tosecond laser ablation offers another possibility as it inherently leads to formation of nanoparticles.

Effect of background gas pressure

The high kinetic energy can lead to recoiling of the incoming particles (atoms, ions) from the substrate [97, 99], implantation of the particles into the substrate and sputtering of the substrate or the deposited thin-film [149]. The energy of the particles can be controlled by increasing the background pressure or the distance between the target and the substrate.

4.3 Film growth in PLD

Figure 4.2: The increasing oxygen pressure slows down the particles in the plume, initially increasing the thickness of the grown Y ZO-films. This is caused by fewer particles recoiling away from the surface, reduced film-sputtering, increased incorporation of oxygen into the film and lower density of the formed film. The reduced density can be observed easily from the top- down perspective. At sufficiently high pressures the thickness of the film drops as fewer particles reach the substrate. The deposition time for all samples was 30 minutes. The pressure values in the picture are expressed in mbar.

At background pressures below 10−2 mbar the plume expansion is relatively unhin- dered by the background gas and is adequately described by the models of adiabatic expansion of a gas cloud into the vacuum. [130, 150] Higher pressures lead to increased collisions between the particles in the plume and the background gas, reducing the speed of expansion. The slightly decreased energy is typically observed as an increase in the growth rate of the film (Fig. 4.2). The increase can be attributed to fewer particles re-

(a) (b)

Figure 4.3: PLD in high background pressures leads to formation and deposition of nanoparti- cles (a). With even higher pressures the slow nanoparticles form loose, foam-like deposits with very high specific areas suitable for catalytic applications (b). Titanium dioxide (a) and nickel (b) samples were both deposited with fibre laser PLD.

coiling from the substrate and reduced sputtering. With even higher pressures, the film thickness increases as the particle energy is no longer high enough to deposit as dense films as in vacuum. If the background gas is reactive, like oxygen in the case of zirco- nium dioxide (Fig. 4.2), the incorporation of gas can also lead to thicker films.

With pressures above 10−1 mbar the increased collisions of the particles with the background gas lead to diffusion of the material over a larger area, possibly away from the substrate. Therefore, the deposited films are thinner.2

Even higher pressures lead to formation of a ”shock-front” at the plume-background gas interface and causes nucleation of nanoparticles (Fig 4.3 (a)). [97, 151–155] Increas- ing the pressure further slows down the nanoparticles and leads to deposition of foam- like structures with extremely high specific surface areas that are useful in e.g. catalytic applications (Fig. 4.3 (b)). [156, 157]

2_{Note that the films shown in Fig. 4.2 have been deposited with a high-repetition rate laser in the}

thermal evaporation mode (chapter 3, P1 and P2). However, the observations are similar to those obtained with nanosecond pulse excimer lasers [150] though the thermal evaporation mode itself in some respects is different.

4.3 Film growth in PLD

Effect of the pulse length

PLD with nanosecond pulse lasers produces high quality films, but it requires precise control of the process parameters. The laser needs to be very stable and have sufficiently high beam quality. The surface of the target needs to be ablated evenly, yet the quality of the surface can still degrade during long depositions. Deviation from the optimal parameters can lead to production of droplets that deteriorate the quality of the film.

The ablation process with femtosecond pulses is almost free of thermal effects and thus is less prone to droplet formation. In addition, the surface quality of the target is higher after femtosecond pulse ablation, as shown by micro-machining studies. [58] It has indeed been shown that use of ultra-short pulses in PLD avoids the droplet problems, but unfortunately introduces a new problem: the evaporated material forms nanoparti- cles. [84, 85, 158–160]

The nanoparticles can be either evaporated directly from the material or formed through collisions in the dense plasma. In ablation with nanosecond pulses, the laser interacts strongly with the plume and dissociates the evaporated molecules and clusters. In the case of femtosecond (and in most cases picosecond) pulses the laser pulse is over before any significant ablation occurs. Thus, the formed nanoparticles remain intact and can serve as nucleation centres during gas expansion and increase in size. Nanoparticle creation is thus an inherent feature of ultra-short pulse PLD.

Improvement of the homogeneity of the film thickness

Due to the directed nature of the plume, the films are deposited over relatively small areas and have considerable thickness variation. In an effort to improve the film homogeneity over the deposition area, a simulation model for the deposition process and resulting

film appearance was developed using Matlab®. The effective plume profile and the

deposition rate were estimated from the deposited film thickness profiles. The simulated thickness profile was transformed into the visual appearance of the transparent film by calculating the reflection spectra at each point and then converting the spectrum into the RGB-coordinates3(Fig. 4.4 (a) and (b)). A sample deposited on a two-inch silicon

3_{The conversion to RGB(Red-Green-Blue) was made by the Colorlab Toolbox collected by the Univer-}

(a) (b) (c)

(d) (e) (f)

Figure 4.4:Results obtained with the model developed for the calculation of the deposited thick- ness profile (a) and the visual appearance of the film (b). A photograph of the real sample shows good agreement with the simulation results (c). The thickness (d), the simulated appearance (e) and the photo (f) for the optimized process.

wafer, with the simulated parameters was found to match the simulation well (Fig. 4.4 (c)).

The simulations were then used to optimize the rotation of the substrate in order to maximize the homogeneity of the film thickness and again, the simulation was found to agree well with the actual deposition. It correctly predicted both the optical thickness and the thickness profile of the film (Fig. 4.4 (d)-(f)). Thickness variation over the two-inch wafer was reduced from the initial 1600% down to 20%. The radius of the homogeneous area with less than 5% thickness variation was increased from 2 mm to 15 mm.