CHAPTER 4

 

KrF NANOSECOND EXCIMER LASER MICROMACHINING OF AL₂O₃-TiC CERAMICS IN AIR

 

The micromachining of Al₂O₃-TiC composite ceramics in air using KrF (l=248 nm) nanosecond laser pulses was studied. The influence of the processing parameters was analysed aiming to optimise the machining process. Surface topography, surface roughness, ablation rate and surface finish machining quality were investigated as a function of the following parameters:

- Laser fluence;

- Number of pulses;

- Laser beam spot size;

- Laser beam angle of incidence;

- Machining with static or moving sample.

 

4.1 INTRODUCTION

 

Processing of Al₂O₃-TiC, Al₂O₃ and TiC

 

Most studies on excimer laser micromachining concern materials such as metals [1-3], polymers [4-7] and ceramics [8-10]. Excimer laser processing of Al₂O₃ has been thoroughly investigated due to Al₂O₃ potential applicability in industry processes [11], and the fact that although Al₂O₃ has a wide bandgap E_g (reported values vary between 7.3 [12] and 9.9 eV [13]), it nevertheless is easily machined at relatively low fluence using UV photons with energies below E_g, in the range 5-6.4 eV. Early studies on excimer laser ablation of alumina and sapphire showed that the presence of impurities and structural defects in polycrystalline materials strongly affects the absorption of laser energy. Lowndes et al. [14] showed that the ablation threshold for alumina is much lower than that for sapphire. Above this threshold, the number of pulses required for incubation of ablation is larger for sapphire than for alumina and decreases with increasing fluence for both materials, while above ~4 J/cm² the ablation rate after incubation is similar for both materials. These results were interpreted in terms of the need to generate a sufficient concentration of defects, which act as absorption centres, before efficient ablation of these wide bandgap materials can occur. Recently, Sciti et al. [15] studied the microstructural changes induced in 95 % a-Al₂O₃ by irradiation with a KrF excimer laser. SEM analysis of irradiated areas showed that alumina melts and flows under repeated irradiation. The thickness of the melted layer (<1 mm at 7.5 J/cm²) decreased with increasing laser fluence, which was attributed to a corresponding reduction in surface melting and increased material removal by vaporisation. On the processed surface a network of microcracks formed, due to the low thermal shock resistance of alumina. Rothenberg and Kelly [16] found significant thermal stress-induced exfoliation in sapphire after machining with 532 nm Nd:YAG pulses, but when 266 nm (Nd:YAG fourth harmonic), 248 or 193 nm excimer radiation was used, typical thermal or exfoliational features were absent, suggesting that the ablation mechanism is electronic. While these studies were focused on a low-fluence regime, just above the ablation threshold, thus avoiding the presence of laser-induced plasma, those by Lowdes et al. [14] and Sciti et al. [15] focused on a high-fluence regime, which can explain the observed differences. Ihlemann et al. [17] studied nanosecond excimer laser ablation of several oxide ceramics. For alumina processed with 248 nm radiation, measurements of the absorption of radiation by the bulk showed that absorption in the material that is not ablated increases due to defect formation. The observation that more than 50% of the pulse energy does not reach directly the sample surface, lead to the suggestion that the laser-induced plasma absorbs a significant portion of the laser pulse energy, which is subsequently transferred to the surface by the plasma, causing thermally induced ablation (plasma mediated ablation).

Studies on excimer laser ablation of TiC focussed on the deposition of thin films by pulsed laser deposition (PLD). In this technique, the laser beam incident on the target leads to removal of material, which is deposited on a suitably positioned substrate to form a film [18]. Growth of TiC films by PLD has been reported using frequency doubled Nd:YAG (l=532 nm) [19, 20] and KrF excimer (l=248 nm) [21, 22] lasers. D’Alessio et al. [19] studied the pulsed laser deposition of TiC films using a frequency doubled Nd:YAG laser (l=532 nm, t=10 ns, repetition rate=10 Hz). The experiments were performed in dynamic vacuum (1.5 ´ 10^-4 Pa). The ablation rate dependence on fluence was evaluated measuring the target weight loss. The authors observed a non-linear behaviour of ablation, and identified three distinct ablation regimes (Figure 4.1.1) from the ablation threshold (0.5 J/cm²) to 3 J/cm², from 3 to 8 J/cm², and above 8 J/cm². They suggested that the regimes resulted from different ablation mechanisms: in zone A) it was suggested that vaporisation of TiC is non-congruent [23]; in zone B) vaporisation is congruent, leading to compact high quality TiC films; and in zone C) the ablation rate is significantly larger than for lower fluence, a feature that was attributed to the expulsion of large target fragments due to an explosive ablation mechanism. In order to gain insight into the pulsed-laser deposition of TiC, the ablation plume has been analysed using several techniques. Time-resolved emission spectroscopy was used for characterisation of the ablation plume formed by irradiation with KrF laser pulses in vacuum and resulted in the identification of various species, depending on laser parameters. Radhakrishnan et al. [22] observed that for fluences ³ 8 J/cm², Ti II peaks dominate the spectra, while below 8 J/cm², Ti II peak intensity is significantly smaller and comparable to Ti I intensity. The C II peak intensity also drops off sharply as laser fluence decreases, and no emission from C II is measurable below 10 J/cm². The abrupt variation of the species that dominate the spectra supports the hypothesis of D’Alessio [19] who suggested that TiC has distinct ablation regimes. Zergioti et al. [24] investigated the plume produced during PLD of TiC using a KrF laser by means of ion mass spectrometry. The production of Ti ions occurs above the ablation threshold of TiC, which was determined at ~0.5 J/cm². C ions were observed above 0.8 J/cm², and no dimers or clusters were observed in the range of fluence investigated (up to 4 J/cm²). Conversely, Santagata et al. [25] observed a small concentration of clusters (n up to 5), (n between 2 and 5) and (n between 2 and 4) during laser ablation of TiC in vacuum using a frequency doubled Nd:YAG laser (l=532 nm). The different ion species observed by Zergioti and Santagata suggests that the radiation wavelength affects considerably ablation. A possible explanation is that the photon energy of KrF laser radiation (5 eV) is comparable to the dissociation energy of TiC (5.5 eV), making possible the dissociation of the carbide.

 

Fluence (J/cm2)

Figure 4.1.1: TiC mean weight loss per pulse laser as a function of fluence (l=532 nm, t=10 ns, 10 Hz, 1.5 ´ 10-4 Pa) [19].

 

Surface features

 

Despite the technical interest of ceramic matrix composites, only few studies on excimer laser ablation of this kind of material were published [26]. Processing of Al₂O₃-TiC ceramics with excimer laser radiation was previously limited to the study of Man et al. [27, 28], who showed that conical features develop when processing is carried out using fluences in the range 1-6 J/cm² and 248 nm nanosecond duration pulses. The formation of cone or columnar-like topography in ceramic materials (Figure 4.1.2) irradiated with excimer laser radiation has been observed for a wide range of processing conditions [10, 28-31]. Similar artefacts were observed in a variety of other materials processed with UV lasers, including semiconductors [32, 33], polymers [30, 34, 35] and metals [36]. The mechanisms that have been called upon to explain the formation of these topographies can be divided into two major groups:

·        Mechanisms that consider surface structures develop due to preferential removal of material around them;

·        Mechanisms that consider surface structures grow due to transport of material to their top.

 

Mechanisms of preferential removal form features that lie below the original surface, while mechanisms based on the transport of material may lead to features that grow above the original surface. Among the mechanisms of the first type one can distinguish:

·        Shadowing or differential mechanisms, where an ablation-resistant phase (previously existing or formed due to laser treatment) as compared to the surrounding or underlying material, leads to the growth of cones or columns as material is preferentially removed around it;

·        Mechanisms related to a variable distribution of laser fluence on target, leading to localised variations of the material removal rate.

 

The ablation-resistant regions required for shadowing mechanisms can be intrinsic impurities of the material [34, 37], redeposited ablated material [35, 38] or chemical heterogeneities, either intrinsic, or formed due to laser treatment [29, 39]. To test this hypothesis, Dyer et al. [34] deliberately seeded films with alumina and rare-earth particles, and demonstrated a correlation between the concentration of these impurities and the density of cones developed. Hopp et al. [37] investigated the development of conical features on polycarbonate processed with ArF excimer laser radiation. The authors suggested that the development of cones is due to shadowing and diffraction effects originating from particles of impurities and interference effects between the incoming and reflected radiation. Foltyn [29] showed that chemical segregation at the surface or at grain boundaries of a multicomponent material could be responsible for texture formation if these components have different ablation rates.

 

Develop PZT for example The process of interest here is the propagation into and withdrawal from the bulk target of a melt front. As resolidification begins, higher melting point components of the liquid freeze first, thus leading to a heterogeneous chemical distribution. Little segregation is likely to occur in a single pulse laser of a few to several hundreds of nanoseconds, but in usual processing conditions the cycle is repeated hundreds of times. Nevertheless, Foltyn questions about the mechanism that leads from surface segregation to the formation of islands in the surface that become cone tips. A possibility related to segregation is constitutional supercooling [Von allmen] which results from spatially periodic fluctuations in solidification velocity, and which leads to a cellular structure with impurity-rich boundaries (typical cell sizes, however, are much smaller than cones).

a) SEM micrograph of cones formed in mica after processing with 100 pulses at 1 J/cm2 (l=193 nm) [31].	b) SEM micrograph of Si microcolumns after irradiation with 1000 pulses in air at 3 J/cm2 (l=248 nm) [33].
Figure 4.1.2: Surface features on materials processed with excimer laser radiation.

 

Growth of surface structures due to a heterogeneous distribution of intensity on target requires a non-uniform energy distribution of the incident beam, or may be due to surface roughness [40, 41], to the polarisation of the incident laser beam [37], or to particles on the processed surface [34]. Dyer et al. [34] studied low-fluence excimer laser ablation of polyimide and observed that the cone apex angle depends on fluence. According to Dyer, due to near-field diffraction, the fluence in the vicinity of the edge of an impurity particle on the surface will show a finite gradient rather than a step falloff, which will lead to the development of a sloping wall as ablation proceeds (i.e., a cone or column structure). This wall will become steeper with repeated ablation until the increase of the surface area is such that the local fluence is reduced to the ablation threshold. A similar mechanism was proposed by O’Brien et al. [43] and Thomas et al. [30], only they consider that the heterogeneity of the initial surface is sufficient to start the process. If one considers ablation of a surface whose normal is at an angle a to the incident beam (Figure 4.1.3a), then the effective fluence F_eff on that area is less than the fluence of the beam F:

 

(4.1.1)

 

The authors considered that ablation from a rough surface of such material will lead to preferential ablation of those areas of the surface where the normal is at a small angle to the incident beam. Ablation will roughen the surface until the average surface angle is such that the effective fluence is reduced to the threshold fluence F_th. This process results in the formation of a surface consisting of protuberances growing from the original areas of steep slope, which act as nucleation centres. From equation (4.1.1) the final angle a_f at which the surface is mature is given by

 

(4.1.2)

 

Usoskin et al. [40] developed the model further (Figure 4.1.3b), showing that columns can grow from an initial surface relief, as a result of the combined effects of focusing of radiation due to reflection on the protuberances lateral walls, increase of optical reflectivity on the lateral walls with increasing angle of incidence, and decrease of energy absorbed at surfaces non-normal to the laser beam, due to increase of the irradiated area. These mechanisms related to a variable distribution of laser fluence on target explain the variation of the apex angle of cones or columns with laser fluence, but require an initial heterogeneity. Conversely, shadowing mechanisms do not require an initial heterogeneity; they explain the variation of chemical composition often observed along individual features, but not the variation of their shape with fluence. Both mechanisms explain the preferential orientation of surface features towards the laser beam [40] and the presence of unaltered material at the core of the features [44].

 

a   Beam     a)	b)
Figure 4.1.3: a) XeCl laser processed polyimide with 300 pulses at 70 mJ/cm2. The substrate was seeded with ~0.05 mm-diameter alumina particles [34]. b) Numerical calculations for ablation of Co at an incident angle of 22.5°: curve 0 corresponds to the initial surface wave; curves 1-3 correspond to the surface after 250, 500, and 1000 pulses, respectively [40].

 

The development of surface features was also explained by hydrodynamic or vapour phase transport of material. Hydrodynamic mass transport has been proposed to explain cone formation in Si [32, 45] and metals [46]. Melted material is supposed to stream up the walls of protuberances and solidify at their top. The liquid can move up due to capillarity or thermocapillarity effects [47], thermal expansion of the substrate [32, 48] or pressure gradients [46, 49]. More recently, Pedraza et al. [33, 50] suggested a vapour-phase transport mechanism to explain columns formation in Si. In this mechanism, columns grow due to deposition on the melted tip of columns of silicon from the intense flux of silicon-rich vapour, produced by ablation from regions between columns. Whether the mechanism for the formation of surface features in Si is hydrodynamic [32, 45, 51] or vapour-phase transport [33, 50] is still the subject of an intense controversy.

The conditions for formation of surface structures have been investigated as a function of several processing parameters, namely laser fluence [31, 34, 42, 46, 52], beam polarisation [37], angle of incidence [37, 53, 54], and sample scanning [53, 55-57]. Riet et al. [52] observed that the height of surface features in Si, Ti, Cu, Co, Fe and FeSiGaRu targets decreases with increasing fluence. A similar result was obtained for brass, bronze, tin, duralumin, Ni, Ge [46], and mica [31]. Hopp et al. [37] verified that the shape of the cones base depends on beam polarisation. When radiation is linearly polarised the bases are elliptical, while randomly polarised radiation leads to circular bases. Since the direction of the electric field in the polarised beam coincided with the minor axis of the ellipse, the authors suggested that the elliptical shape of the cone-like structures is due to the polarisation dependence of the reflection coefficients (Chapter 3). Various studies have shown that the height of surface structures decreases and their width increases when the angle between the laser beam and the normal to the surface increases. This was attributed to the variation of intensity as a result of the variation of the angle of incidence and effective irradiated area [54], shadowing by existing structures [37] and transfer of liquid due to a non-symmetric plume pressure [46]. Wagner et al. [55] studied the excimer laser ablation of PET when the sample is scanned under the beam. Different textures were observed on the machined ramps, and each occurred within a certain range of ramp angles. Varying the azimuthal angle of incidence while scanning the sample frequently leads to suppression of surface structures observed otherwise [53, 56, 57].

 

Thomas et al. [30] have shown that Al₂O₃ can be modified just above the ablation threshold to create cone-like features that are substantially rougher than the original surface. They do not specify the type or purity of the alumina, showing that cone formation takes place at 3.9 J/cm² using 248 nm. Rothenburg and Kelly [16] irradiating single crystal Al₂O₃ at 266 nm just above the ablation threshold, reported that after a few hundred pulses a pattern of conical structures had formed in the bottom of the etch pit. This pattern was attributed to diffraction effects in the beam. In support of this hypothesis, it was noted that scanning the target produced linear features in the scan direction, instead of cones, and that irradiation with a spatially incoherent KrF laser produced no cones.

4.2 EXPERIMENTAL

 

As part of this work, an excimer laser micromachining system was installed. The system is constituted by a nanosecond pulse duration excimer laser (Figure 4.2.1), an optical delivery and imaging system, positioning equipment, and auxiliary diagnostic equipment (Figure 4.2.2). The excimer laser can operate at several wavelengths and the projection system was designed to achieve on-target fluences above 60 J/cm² for 248 nm laser radiation.

 

Figure 4.2.1: Excimer laser Compex 110, fluorine version (Lambda Physik).

 

 

a)	b)
Figure 4.2.2: a) Projection optics (JPSA Associates, USA), positioning system and auxiliary equipment. b) Complete imaging system.

 

The excimer laser is a Compex 110 from Lambda Physik. Table 4.2.1 shows the optimum gas mixtures for the different wavelengths available with the Compex 110 (Fluorine version). Table 4.2.2 shows the laser specifications for these gas mixtures. The work developed in this chapter was carried out using a KrF gas mixture (l=248 nm), which allows to obtain laser pulses with a duration t=30 ns (FWHM) and a maximum energy of 350 mJ.

 

 

 

 

Table 4.2.1: Optimum gas mixtures for the Compex 110 (fluorine version) [58].

 

Table 4.2.2: Specifications for the excimer laser Compex 110 (Fluorine version) for optimised gas mixtures. 1) Measured at low repetition rate (5 Hz). 2) Measured at maximum repetition rate. 3) Typical value, FWHM [58].

 

The imaging system consists of beam delivery, illumination and projection optics (JPSA Associates, USA) (Figure 4.2.3 and Table 4.2.3). For rigidity, the optics are attached to a granite table top. The granite slab was fixed to a rubber layer and a low density agglomerated wooden block, which were bolted to a supporting aluminium frame mounted on top of damping feet to guarantee a vibration-free mounting. The optics are optimised for 248 nm radiation, guaranteeing minimal beam energy losses at this wavelength. An attenuator is located just at the beam output (part 2 in Figure 4.2.3) and allows to adjust the laser fluence by varying the pulse energy between 90-10%. In order to achieve fluences above 60 J/cm², a variable telescope (part 5 in Figure 4.2.3) working as a field lens is used to reduce the beam cross section (thus increasing the fluence), and imaging is performed with 30´ demagnification. This high demagnification limits the maximum image size and minimum fluence achievable, and consequently lower demagnifications may be required. The imaging system has two different mask holders, extension tubes that can easily be added to increase the optical path in fixed increments, and a periscope that may be used to continuously vary the mask-objective distance, thus allowing for variable demagnification in the range 8-30´. The projection optics are located in a lightweight gantry system that provides high stiffness to weight ratio and good static and dynamic rigidity (parts 13 to 16 in Figure 4.2.3). The objective used was a doublet with a focal length of 100 mm, consisting of a best fit convex/convex lens with a meniscus corrector for spherical aberration (Figure 4.2.4). Laser treatment was performed by geometrically projecting an aperture inserted in the laser beam path on the surface of the sample. To machine a square or rectangular shape, a rectangular variable aperture was used, consisting of two orthogonal micrometer adjustable slits mounted together in a single housing. Masks of different shapes were produced by laser drilling thin copper foils.

 

 

Microtech MBD Camera   Excimer laser Compex 110   a) Top view.	1 - Manual beam shutter 2 - Variable attenuator 3 - 50 mm beam tube 4 - Telescoping bezel tube assembly 5 - Variable telescope 6 - 1180 beam tube 7 - Beam switcher 8 - Telescoping bezel tube assembly 9 - 508 mm beam tube 10 - RVA/Mask holder (high demagnification position) 11 - 670 beam tube 12 - Mirror 13 - RVA/Mask holder (standard demagnification position) 14 - 500 mm beam tubes (for higher demagnification imaging) 15 - Periscope assembly 16 - MBD Camera Head 17 - Beam tube clamp and post assembly 18 - Beam switcher post assembly
a) Side view.
Figure 4.2.3: Schematic view of the excimer laser micromachining system (adapted from [59]).	Table 4.2.3: Parts list (refer to Figure 4.2.3).

 

 

Laser beam   a)	b)
Figure 4.2.4: a) Orientation of doublet in lens cell. b) Normal imaging using the doublet (adapted from [59]).

 

User-friendly software was developed for PC based control of the laser beam energy, laser triggering and adequate positioning of the workpiece using an X-Y motorised table (Aerotech) with 1 mm resolution. The control system is shown schematically in Figure 4.2.5. The laser is triggered by the PC, or using the table controller (Unidex 12). Vertical positioning of the sample is achieved by a manual Z-axis, while the projection system allows micrometer adjustment of the distance between the objective and the substrate. A video camera connected to a monitor allows to position the workpiece as required (along the X-Y plane), and observation of the processed area in real-time. An optical microscope is used for detailed observation of machined samples.

 

Figure 4.2.5: Scheme of the micromachining controlling system.

 

The experiments described in this chapter were performed in air at atmospheric pressure on substrates of an Al₂O₃-TiC composite ceramic consisting of 70 wt. % of alumina and 30 wt. % of titanium carbide (R_a»0.2 mm). For comparison purposes, titanium carbide (99.5 wt. %) and alumina (96 wt. %) targets were also used. The laser fluence was varied up to 90 J/cm². The pulse energy was measured using a calibrated pyroelectric detector (Molectron, J50HR-085), with an error of ± 7%.

A CCD video camera with a fast electronic shutter and spectral sensitivity in the range 400-1000 nm (FlashCam, PCO) was used to acquire images of the ablation plume (Figure 4.2.6). An external trigger starts the exposure after the preset delay (with a duration adjustable in increments of 1 mm up to 1 ms). The falling edge of the trigger signal activates the exposure and the CCD readout process, asynchronously. The frames were sent to a frame grabber (Matrox) controlled via PC, and stored for analysis (Figure 4.2.7). Since the CCD-camera is activated at the falling edge of the trigger signal, the laser sync-out signal was inverted and used as the trigger to the camera. This allowed to activate the camera before the output of the laser pulse.

After laser processing, the surface of the samples was observed by scanning electron microscopy (SEM). The surface roughness (R_a) was measured by optical profilometry. Optical profilometry results were also used to calculate the average ablation rate per pulse, by dividing the ablation depth after a predefined number of pulses by the number of pulses. The ablation depth was estimated by considering the mean line throughout the sampling length along the processed surface. Stereoscopic SEM images and stereoscopic pair analysis software were used to calculate the effective surface area of processed surfaces and the ratio of the effective surface area to the projected area. Chemical analysis was performed by energy dispersive spectroscopy (EDS), as described in Chapter 3.

 

Figure 4.2.6: CCD camera and objective.

Figure 4.2.7: Set up for imaging.

FOR RESULTS, DISCUSSION, CONCLUSIONS AND BIBLIOGRAPHY CHECK THE THESIS MAIN PAGE