Scholarly article on topic 'Ablation Study of WC and PCD Composites Using 10 Picosecond and 1 Nanosecond Pulse Durations at Green and Infrared Wavelengths'

Ablation Study of WC and PCD Composites Using 10 Picosecond and 1 Nanosecond Pulse Durations at Green and Infrared Wavelengths Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Physics Procedia
OECD Field of science
Keywords
{"Laser ablation" / "picosecond pulses" / "nanosecond pulses" / diamond / "green wavelength"}

Abstract of research paper on Materials engineering, author of scientific article — Gregory Eberle, Konrad Wegener

Abstract An ablation study is carried out to compare 10 picosecond and 1 nanosecond pulse durations as well as 532 nanometre and 1064 nanometre wavelengths at each corresponding pulse duration. All laser parameters are kept constant in order to understand the influence of pulse duration and wavelength independently. The materials processed according to the electronic band structure are a metal and an insulator/metal composite, i.e. tungsten carbide and polycrystalline diamond composite respectively. After laser processing said materials, the ablation rate and surface roughness are determined. Analysis into the ablation behaviour between the various laser parameters and the materials processed is given, with a particular emphasis on the graphitisation of diamond.

Academic research paper on topic "Ablation Study of WC and PCD Composites Using 10 Picosecond and 1 Nanosecond Pulse Durations at Green and Infrared Wavelengths"

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Physics Procedia 56 (2014) 951 - 962

8th International Conference on Photonic Technologies LANE 2014

Ablation study of WC and PCD composites using 10 picosecond and 1 nanosecond pulse durations at green and infrared wavelengths

Gregory Eberle*a, Konrad Wegenera,b

aInstitute of Machine Tools and Manufacturing, ETH Zurich, Tannenstrasse 3, 8092 Zurich, Switzerland binspire AG, ETH Zurich, Technoparkstrasse 1, 8005 Zurich, Switzerland

Abstract

An ablation study is carried out to compare 10 picosecond and 1 nanosecond pulse durations as well as 532 nanometre and 1064 nanometre wavelengths at each corresponding pulse duration. All laser parameters are kept constant in order to understand the influence of pulse duration and wavelength independently. The materials processed according to the electronic band structure are a metal and an insulator/metal composite, i.e. tungsten carbide and poly crystalline diamond composite respectively. After laser processing said materials, the ablation rate and surface roughness are determined. Analysis into the ablation behaviour between the various laser parameters and the materials processed is given, with a particular emphasis on the graphitisation of diamond.

© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibilityoftheBayerischesLaserzentrumGmbH

Keywords: Laser ablation; picosecond pulses; nanosecond pulses; diamond; green wavelength

1. Introduction

It is apparent over the recent years that the laser processing of ultrahard materials, specifically diamond in the cutting tool industry, is becoming more widespread and offering competition to conventional methods, e.g. grinding and electro-discharge machining. Reasons include namely a force- and wear-free process, geometric flexibility, material independency and reduced processing times. The equivalency and advantages using laser technology to process ultrahard cutting tools have been previously demonstrated by Rabiey et al. (2011). An open ended question which often arises when it comes to laser material processing is: What is the appropriate laser source for the specified

* Corresponding author. Tel.: +41-44-633-79-47; fax: +41-44-633-14-92. E-mail address: eberle@iwf.mavt.ethz.ch

1875-3892 © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Bayerisches Laserzentrum GmbH

doi:10.1016/j.phpro.2014.08.115

application? In general, there consists a wide range of influencing variables, the most important being pulse duration, wavelength and fluence. The most common form of diamond used in the cutting tool industry is polycrystalline diamond (PCD) composites integrally bonded to a tungsten carbide (WC) substrate. Attributes include its isotropic properties, high toughness and minimal wear characteristics as well as its ability to be brazed onto tool bodies.

1.1. Previous works

The use of short nanosecond laser technology to process PCD and WC is extensively demonstrated by Harrison et al. (2005, 2006) for laser cutting and laser ablation using a Tp = 20 to 200 ns, X = 1064 nm and Ep = 18 mJ laser system. Additionally, Wu and Wang (2011) studied the laser ablation of PCD composites using a X = 1080 nm and Ep = 1 mJ nanosecond laser system. Extensive empirical studies are performed to correlate laser and processing parameters to measured qualities, namely roughness and material removal rate, with the goal of process optimisation. Within the pulse duration, Harrison et al. models material removal in which the PCD composite is heated and transforms into graphitic carbon at ~1000 °C, wherein the WC melts and both materials are eventually removed through vaporisation. The model holds well but diverges at high irradiances due to over-simplification, ignoring shock wave and shock changes as well as the presence of cobalt in the PCD composite. Zhang et al. (2007) carried out a comparative study into the laser cutting of PCD composites and WC using a Tp = 100 ^s, X = 1064 nm and Ep = 0.3 J as well as a Tp = 120 ns, X = 0.532 nm and Ep = 1 mJ laser system. The use of longer pulse durations is attributed to greater melting effects and striations. It is stated that shorter pulse durations and the green wavelength resulted in the best clearance face qualities due to lower thermal effects. Everson and Molian (2008) studied laser cutting of PCD composites using a Tp = 100 ns, X = 1064 nm and Ep = 1.5 mJ laser system. Geometric quantities, ablation rate, kerf width and edge radius are quantified. Following laser processing, it is determined through Raman spectroscopy that polycrystalline graphite remains on the clearance face. The rapid heating and cooling at the laser interaction zone generates amorphous carbon such that polycrystalline graphite remains according to the temperature distribution at the clearance face. Material removal is explained through the conversion of PCD composite to graphitic carbon leading to a volume expansion and development of thermal stresses. The stresses causes fragmentation leading to material removal. The use of nanosecond laser technology, though heavily investigated, is clearly attributed to the generation of graphitic carbon. As a result, a cutting tool with higher rates of wear, lower lifetime and overall poorer chipping performance is expected.

As a result, the recent use of ultrashort picosecond laser technology to process PCD composites and WC is demonstrated by Dold et al. (2012 and 2013) using a Tp = 10 ps, X = 1064 nm, Ep = 50 ^J laser system. Laser processing is faster than conventional grinding methods, no grain breakouts occurs and no thermal effects are observed. It is determined that the laser processing of PCD composites are independent of grain size. Additionally, lasered and ground cutting tools are tested in the turning of carbon fibre reinforced polymers (CFRP) where it is determined that process forces for lasered cutting tools are equal if not slightly lower. The laser processing of other forms of diamond such as chemical vapour deposition diamond (CVD-D) and nanopolycrystalline diamond (NPD) is also quite extensive in literature. However, its use in the tooling industry is still small compared to PCD composites due to brazing issues and grinding difficulties. The advent of ultrashort pulsed laser technology and its rapid industrialisation will allow the presence of these other forms of diamond to grow significantly.

This paper will investigate the influence of short (tp = 1 ns) and ultrashort (tp = 10 ps) laser technology when laser processing PCD composites and WC. Furthermore, at each pulse duration, investigations at X = 532 nm and X = 1064 nm are carried out while the effects of reducing the focussed beam diameter from D4a = 30 ^m to D4a = 15 ^m is also considered. From this direct comparison study, ablation behaviour and mechanisms are explained.

1.2. Absorption of laser radiation (electron system)

Absorption curves for diamond, highly oriented pyrolytic graphite (HOPG), non-hydrogenated evaporated carbon and tungsten are plotted in Fig. 1a. The following curves are assumed to hold valid under the condition of normal

incidence, room temperature and non-polarised radiation. Reflection is calculated according to the Fresnel equation in Eqn. (1) given measured values for index of refraction, n, and extinction factor, k:

(nm - nmat)2 + k2 Kir + nmat)2 + k 2

Transmission is calculated according to Eqn. (2) for an arbitrary material thickness of x = 300 ^m, where a is the absorption coefficient and X is the wavelength:

T = e-œx = e

As a result, diamond exhibits full transmission when X > 220 nm, otherwise, all materials exhibited no transmission. In doing so, absorption is approximated while taking into account reflection and transmission losses. Fig. 1a demonstrates the trend that for shorter wavelengths, absorption increases.

Diamond: Phillip and Tafl (1SS4) —HOPG: Djurisi and Li (1999) —Evaporated C: Hagemanri et al. (1974) —Tungsten: Weber (2002) —X = 532 nm —>,= 1064 nm

Wavelength [nm]

X Eg=9.0

Graphitic C: Robertson (2002)

X Eg=5.5

il=355nm

. E„=2.4

À=532nm

a.= 1064nm

Eg=0.55X

Eg=0.01

Eg=0.0

Fig. 1. (a) Absorption curves for various materials as a function of wavelength; (b) Band gap energies for various materials

For laser absorption to ensue, free electrons must be present. However, as shown in Fig. 1b, not all materials exhibit free electrons and in fact need to be generated by overcoming an electronic band gap, Eg. Silicon dioxide (i.e. glass) and diamond and are examples of insulators demonstrating that short wavelengths or a high density of photons are necessary for multiphoton absorption to occur and hence initiate electron mobility. Tetrahedral amorphous carbon (i.e. diamond-like carbon) and glassy carbon are examples of a semi-conductor having a moderate band gap while WC is considered a metal which has no band gap. HOPG is an example of a semi-metal and is characterised as having a negative band gap. According to the absorption curves and the electronic band gap, the ablation rate at X = 532 nm is expected to be higher compared to X = 1064 nm. Since absorption and electronic properties of PCD composites are unavailable in literature, values for diamond are used as a first glance. Nevertheless, the absorption properties of diamond is generally very poor compared to the other mentioned materials.

Laser pulses at Tp = 1 ns are dominated by linear absorption mechanisms dependent on optical properties (e.g. transparency, reflectivity, opaqueness) and level of defects (e.g. microcracks, impurities, porosity) in the material. To process reflective and transparent materials, high pulse energies are necessary resulting in material transformations and thermal stress induced defects. This in turn leads to improved absorption to subsequent laser pulses causing ablation, but often of poor quality. Laser pulses at Tp = 10 ps are dominated by non-linear absorption mechanisms and is independent of the optical properties of the material. In doing so, optical breakdown of transmissive materials such as diamond is possible leading to improved absorption to subsequent laser pulses and

causing ablation, often of excellent quality. The efficiency of optical breakdown is dependent on the multiphoton absorption process and peak power of the pulse, i.e. dependent on wavelength and pulse duration respectively. For materials that exhibit a band gap such as in the case of diamond, electrons from the valence band need to be excited, in accordance with the electronic wave of laser radiation, to the empty conduction band whilst leaving electron holes in the valence band. These free electrons, due to their mobility, further excite other bound electrons in the valence band leading to avalanche ionisation and optical breakdown. In contrast, WC and HOPG already exhibit free electrons and hence optical breakdown occurs immediately. In general, femtosecond and picosecond pulsed material ablation exhibits a deterministic behaviour while for nanosecond pulses and longer exhibits a statistical behaviour. Table 1 gives a brief summary, in line with the experimental approach of this paper, of the differences between picosecond and nanosecond pulses at a pulse energy of Ep = 18 J Hence, it is expected that the ablation efficiency of Tp = 10 ps to be far better compared to Tp = 1 ns.

Table 1. Comparison between picosecond and nanosecond pulses

Pulse duration, Tp 10 ps 1 ns

Wavelength, X [nm] 532 1064 532 1064

Energy per photon [eV] 2.33 1.17 2.33 1.17

Nr. of emitted photons within Tp [-] 4.8x1024 9.6x1024 4.8x1021 9.6x1021

Nr. of emitted photons needed to reach Ep = 18 ^J [-] 4.8x1013 9.6x1013 4.8x1013 9.6x1013

Peak power [W] 1.6x106 1.6x106 15.7x103 15.7x103

Primary absorption mechanism Multiphoton absorption Material and defect dependent

Nr. of photons to reach diamond band gap [-] 3 5 3 5

Nr. of photons to reach WC/sp2 graphitic carbon band gap [-] 1 1 1 1

1.3. Absorption of laser radiation (lattice system)

The aforementioned discussion evaluates the differences among pulse duration and wavelength absorption from the point of view of the electron system. However, the lattice must also be considered, especially during material processing. The lattice in the case of this paper is either the PCD composite or WC. Initially, regardless of the pulse duration, laser radiation excites electrons through an intense increase in the electron temperature. During this time, the temperature of the lattice remains unchanged. Upon the onset of avalanche ionization, electrons relax and begin to conduct energy in the form of heat into the lattice. The instance when electrons start to relax occurs at Tp = 0.5 ps, such that after Tp = 100 ps, an equilibrium between the electron and lattice system is reached. These values are valid for metals according to Walter (2010) and would be delayed by a few picoseconds given, for instance, an insulator. Hence, the longer the pulse, the more thermal energy is inputted into the lattice system while the significance of multiphoton ionisation diminishes. Given enough conducted thermal energy to increase the lattice temperature beyond ~1000°C, then the diamond in the PCD composite transforms into graphitic carbon. For Tp = 1 ns, this results in a sudden increase in absorption for subsequent pulses since prior, diamond is transparent at X = 532 nm and X = 1064 nm if the presence of cobalt in PCD is ignored. In doing so, nanosecond pulses can only process diamond through this graphitisation mechanism. For Tp = 10 ps, there exists two possibilities. The first and simplest possibility is that optical breakdown of the PCD takes place, i.e. a change in the refractive index of the otherwise transparent PCD to an opaque PCD. This leads to absorption and ablation of the PCD composite. The second possibility follows a similar behaviour to Tp = 1 ns since electron relaxation at Tp = 0.5 ps already occurs within the Tp = 10 ps pulse duration. Since the temperature of the electronic system is in the order of 103 °C, it is expected that electron-lattice conduction will lead to a temperature rise beyond the graphitisation onset temperature as investigated by Kononenko et al. (2005). However, since equilibrium between the electron and lattice system does not take place, growth of graphitic carbon is minimal. Since in the scope of this paper, presence of graphitic carbon is not measured at Tp = 10 ps, it is assumed that through the ablation effects, e.g. phase explosions, and through the processing strategy, i.e. tangential processing, the graphitic carbon is expelled away from the laser interaction zone. For instance, the evaporation velocity which is associated with the recoil pressure of material removal lies in the order of ~102 m/s

and ~100 m/s for picosecond and nanosecond pulses according to Besner (2010) and Chen (2005) respectively. Hence, for every pulse, the graphitisation process is repeated while for nanosecond pulses, upon graphitisation, subsequent pulses will only interact with graphitic carbon since the thermal penetration depth is deep enough to sustain the graphitisation process. This argument fully depends on the period when electron relaxation occurs and the electro-phonon coupling constant, which for polycrystalline diamond, cannot be found in literature. Further research into this topic will demonstrate which possibility dominates, however, this lies outside the scope of this paper.

2. Methodology

The material processed is fine grain PCD composite blanks whereby material properties are given in Table 2. The PCD composite blanks are composed of a 0.4 to 0.6 mm thick layer of PCD composite bonded to a ~1 mm thick WC layer resulting in an overall thickness of 1.6 ± 0.05 mm. The overall dimensions of the blanks are 10x3x1.6 mm. PCD composite refers to a mixture of 90 Vol% diamond grains and 10 Vol% cobalt such that through catalytic high pressure high temperature (HPHT) synthesis, the diamond grains coalesce. According to EDX measurements, WC is composed of (in wt%) 2.7C-3.2O-4.6Co-89.5W.

Table 2. Material specifications of PCD composite blanks acc ording to Diamond Innovations (2004) and WC

Property Value for PCD composite blanks Value for WC Unit

Average grain size 4 2-3 ^m

Diamond/cobalt content 90/10 - Vol%

Compressive strength 7.5 5 GPa

Elastic modulus 850 550 GPa

Thermal conductivity 500 85 W/m-K

Electrical resistivity 1.5x10-2 2x10-7 ^■m

Density 4.1 15.6 g/cm3

Mohs hardness > 9 9 -

Enthalpy of fusion 8.8 (value for C) 0.19 (value for W) kJ/g

Enthalpy of vaporisation 59.3 (value for C) 4.48 (value for W) kJ/g

Specifications of the utilised laser systems are given in Table 3. Either a MOPA Nd:YVÜ4 solid state laser from the company Time-Bandwidth Products (TBWP) or a Yb:YAG fibre laser from the company IPG Photonics are used. Additionally, each laser system outputs linear polarised radiation with similar beam qualities, M2 < 1.3. Frequency doubling to attain X = 532 nm is carried out using non-linear optics. The LASER LINE used in this study is a modified laser machine centre with a preinstalled high powered infrared laser source and an optical breadboard to allow ease of integration for the two fibre laser sources. The green picosecond laser source could not be integrated into the LASER LINE and hence experimentation is carried out in a laboratory environment.

Table 3. Specifications of the utilised laser systems

Pulse duration, Tp 10 ps 1 ns

Wavelength, X [nm] 532 1064 532 1060

Processing environment Laboratory EWAG EWAG EWAG

LASER LINE LASER LINE LASER LINE

Laser system TBWP Duetto TBWP Fuego IPG GLPM-20 IPG YLPR-0.3

Max. average power, Pavg [W] 7 40 to 50 20 18

Frequency, frep [kHz] 50 to 8200 200 to 8200 10 to 600 10 to 500

Max. pulse energy, Ep [J 100 @ 70 kHz 200 @ 200 kHz 33 @ 600 kHz 60 @ 300 kHz

A schematic of the experimental setup is illustrated in Fig. 2a. The experimental setup both in the laboratory and the LASER LINE processing environments exhibit the exact same optical and kinematic arrangement. The experimental setup begins with one of the aforementioned laser sources. To ensure that the properties of the focal spot are equal among the utilised laser systems, polarisation plates are used so that left-hand circular polarisation is attained and a variable beam expander allows for the focal spot diameter to be defined independent of the raw beam diameter. The focal spot diameter is measured using a beam analysis camera with a resolution of 1600* 1200 pixels and a pixel size of 4.4*4.4 ^m. Bending mirrors and iris diaphragms guide the laser beam and ensure parallel beam alignment into a scanhead respectively. A 163 mm f-theta objective lens focuses the laser beam onto the workpiece and a 4-axis (laboratory) or a 5-axis (LASER LINE) system positions and moves the workpiece relative to the focussed laser beam. A scanhead is used in order to deflect continuously and in two dimensions, the focussed laser beam in a pre-defined hatched pattern, e.g. a square meander. This deflection corresponds to a hatch size of 500 ^m. Simultaneously, the workpiece, i.e. the PCD composite blank is moved back-and-forth relative to the hatch, and subsequently moved about the z-axis in predefined steps in order to generate pockets as shown in Fig. 2b. After the pockets are generated, the ablated volume and the surface roughness of the pocket's wall are measured using an Alicona 3D optical microscope.

Fig. 2. (a) Schematic of the experimental setup; (b) Pocket generation on PCD composite and WC layers with an inset of the measured regions

The pockets are generated according to a defined set of laser and axes parameters shown in Fig. 3. The experimental set is designed to benchmark the quality of the generated pocket in terms of surface roughness, ablation rate and presence of graphitic carbon according to the primary influencing variables, i.e. pulse duration and wavelength. In order to do so, a secondary set of influencing variables are necessary, i.e. pulse energy, feed rate and focal spot diameter. Each experiment, i.e. each data point are repeated three times to establish a standard deviation amongst the results. To attain a comprehensive understanding of graphitisation, Raman spectroscopy measurements are carried out using a WiTec CRM 200 confocal Raman microscope. The laser source is a continuous wave, X = 532 nm, Pavg < 2 mW, dFWHM = 300 nm system whereby a 100x, 0.80 NA microscope objective lens is used. Grating is set to 600 g/mm and the spectral centre set to 2150 rel./cm. Additionally, focussed ion beam (FIB) cross sections are generated using a CrossBeam NVision 40 from the company Carl Zeiss SMT in which a current of 1.5 nA and an acceleration voltage of 30 keV are used to produce the final cross section with a surface quality suitable for observation and measurement purposes.

Primary influencing Pulse duration [-] 10 ps 1 ns

variables Wavelength [nm] 532 1064 532 1060

Secondary influencing variables Pulse energy [pj] 1.6 up to 16 16 up to 18,3* Feed rate [mm/min] 15 to 75

Constants Frequency [kHz] 300 Scan speed [mm/s] 2000 Focal spot diameter [um] 30 Pulse to pulse overlap [%] 78 Number of passes [-] 20 z-step size [|im] 20 Angle of incidence [°] 0 Polarisation state [-] Circular

*up lo 36.6 for WC since ablation rate was too low to carry out measurements

Fig. 3. Input and output experimental parameters

3. Results and Discussion

Previous research by Dumitru (2002) indicates that the single pulse ablation threshold of industrial diamond and of WC is 1.6 J/cm2 and 0.3 to 0.4 J/cm2 respectively when Tp = 150 fs, X = 800 nm, frep = 1 kHz and d1/e2 = 53 ^m. This corresponds to a pulse energy of 35 ^J and 6.6 to 8.9 ^J respectively. It is expected that the presence of cobalt in PCD composites further lowers the ablation threshold value. These values only serve as a reference since the approach of this paper focusses on multipulse ablation of cutting edges. All data depicted in Fig. 4 to Fig. 6 are fitted with a quadratic polynomial with the exception of Fig. 5 which is fitted with a linear function.

3.1. Ablation rate

The ablation rate, Q, is calculated according to Eqn. (3):

Vabl Vf

It is the product of the ablated volume, Vabl, and the feed rate of the workpiece, Vf, divided by the number of passes of the workpiece relative to the laser beam, PNr. By extracting the cross sectional profile of the pocket, orthogonal to the feed rate as shown in Fig. 2b, the ablated volume is then calculated for an arbitrary width of 1 mm.

Ablation rate vs. Fluence for PCD

—?-532nm, t ~1ns

—X=1060nm, =1 ns

—/,=532nm, x =l0ps

— 5.=1064nm, p r

Ablation rate vs. Fluence for WC

;.=532nm, t =1ns p

—/.=1060nm, i =1 ns p

—À=532nm, t =10ps —X=1064nm

Fluence [J/cm2]

Fig. 4. Ablation rate vs. fluence for (a) PCD composite and (b) WC according to pulse duration and wavelength

As shown in the Fig. 4, both Tp = 10 ps and X = 532 nm exhibit higher ablation rates for both materials than Tp = 1 ns and X = 1064 nm as well as X = 1060 nm. Additionally as expected, the ablation rate of diamond for Tp = 10 ps is lower compared to WC since more photons are required to achieve multiphoton absorption and hence ablation as outlined in Table 1. However, this phenomenon is reversed for Tp = 1 ns attributed to the graphitisation process, where the absorption properties of graphitic carbon is higher than WC leading to higher ablation rates. This ablation phenomenon performs exceptionally when using Tp = 1 ns and X = 532 nm since its ablation rate is comparable to Tp = 10 ps and X = 1064 nm. Therefore, although the multiphoton absorption process plays a minor role at Tp = 1 ns, the graphitisation mechanism allows for high ablation rates of theoretically transparent PCD.

3.2. Surface roughness

The surface roughness mentioned is the arithmetic mean value, Ra, measured over an area (depthxwidth) of 175x500 ^m and 66x550 ^m (175x550 ^m at Tp = 10 ps) for PCD composite and WC respectively. Surface roughness is measured on the pocket's wall parallel to the top edge emulating surface roughness at the clearance face of a cutting tool. Fig. 5 shows that the surface roughness using Tp = 10 ps pulses is significantly lower compared to Tp = 1 ns and relatively material independent. Due to its higher ablation efficiency, pulses at Tp = 10 ps are able to cleanly remove material out from the pocket. Conversely, pulses at Tp = 1 ns generally have a lower ablation efficiency, which in the case of WC is further hindered by the fact that at the bottom of the pocket's wall, solidified melt contributes to a poor surface roughness and poor absorption properties for subsequent pulses, e.g. due to multiple reflections.

_ 0.8 E

-= 0.7 of

u>' 0.6 ft

Ü 0.5

£ 0.3

Surface roughness vs. Feed rate for PCD

= 1060nm, T =1ns

=532nm, i =10ps

0.9 „ 06

3. 0.7

¡f 0.6 4)

£ 0.5

1 04 <u

I 0.3 CO

Surface roughness vs. Feed rate for WC

>.=532rim, r =1ns p

■■Â=1060nm, i =1 ns p

-Â=532nrn, tp=10ps -Â=1064nm, T =1 Ops

30 40 50

Feed rate [mm/min]

30 40 50 Feed rate [mm/min]

Fig. 5. Surface roughness vs. feed rate for (a) PCD composite and (b) WC according to pulse duration and wavelength

For Tp = 1 ns and X = 532 nm, the surface roughness for PCD composite is almost half compared to that of Tp = 1 ns and X = 1060 nm, however, comparable surface roughness for both wavelengths occurs for WC. Furthermore, the surface roughness for Tp = 1 ns, especially for WC displays a clear trend though there is significant scatter, demonstrating a statistical behaviour which is a characteristic of using a longer pulse duration.

3.3. Focal spot diameter

The main advantage of using X = 532 nm is not only its improved absorption properties, but also its ability to halve the focal spot diameter while maintaining the same raw beam diameter compared to X = 1064 nm. Results in Section 3.1 and Section 3.2 are based on a D4a = 30 ^m focal spot diameter and parameters according to Fig. 3. From this, the pulse train of the hatch is shown as scenario A in Fig. 6a (For clarity, the frequency of calculations and reality was reduced by a factor 6, i.e. 50 kHz as opposed to the experimental value of 300 kHz). However, if the same parameters are used for D4a = 15 ^m, with a modified scan speed of 1000 mm/s so as to attain the same pulse to pulse overlap, i.e. 78%, scenario B will occur. This is undesirable since the equivalent scan length versus scan time is not achieved.

Ablation rate vs. Pulse energy

■D4n=15nm, PCD ■D4it=3D wnn, PCD ■D4fT=15 Jim, WC ■D4(y=30 jim, WC

6 8 10 Pulse energy |jij]

Fig. 6. (a) Pulse train comparison between calculated and reality for scenario A when F = 4Ep/7e"6 J/cm2, frep = 50 kHz, D4a = 30 ^m, vscaii = 2000 mm/s, scenario B when F = Ep/2e-6 J/cm2, frep = 50 kHz, D4a = 15 ^m, vscan = 1000 mm/s, and scenario C when F = Ep/2e-6 J/cm2, frep = 100 kHz, D4a = 15 ^m, vscan = 2000 mm/s; (b) Ablation rate vs. pulse energy for PCD composite and WC at D4a = 15 ^m and D4a = 30 ^m

To overcome this shortfall, as opposed to modifying the scan speed, the repetition rate and average power are doubled. The pulse energy, scan speed and pulse to pulse overlap conform to Fig. 3 while the scan length versus scan time are the same when comparing scenarios A and C. Comparison of results for D4a = 30 ^m and D4a = 15 ^m using X = 532 nm and Tp = 1 ns for PCD composite and WC is depicted in Fig. 6b. Consequently, the ablation rate with a smaller focal spot diameter achieves higher ablation rates in both cases for PCD composites and WC. This is due to a four times higher fluence while the number of pulses per unit length is two times higher. A summary of the achieved ablation efficiencies at maximum pulse energy and for one pass is given in Table 4.

Table 4. Comparison of ablation efficiencies for PCD composites and WC at Tp = 10 ps and Tp = 1 ns as well as for green and infrared wavelengths

Pulse duration, Tp 10 ps 1 ns

Wavelength, X [nm] 532 1064 532 532 1060

Focal spot diameter [^m] 30 30 30 15 30

Pulse energy, Ep [J 16 16 15 15 18.3(PCD) / 36.6(WC)

PCD composite [mm3/J] 805 655 630 735 370

Tungsten carbide [mm3/J] 1240 1000 290 585 20

For Tp = 10 ps, it can be observed that there is almost a 20% increase in ablation efficiency for WC and PCD composites when using green compared to the infrared wavelength. For Tp = 1 ns, a 70% and a 1400% increase occurs for PCD composites and HM respectively when using green compared to the infrared wavelength. If one compares D4a = 15 ^m and D4a = 30 ^m for Tp = 1 ns, a 17% and 100% increase for PCD composites and WC is respectively achieved. Upon closer inspection, it can be noticed that the ablation efficiency of PCD composites for Tp = 1 ns, X = 532 nm and D4a = 15 ^m is better compared to Tp = 10 ps, X = 1064 nm and D4a = 30 ^m. This can be attributed to graphitisation effects when using nanosecond pulses improving laser absorption and hence material removal rates. This trend, however, no longer holds true when processing WC using the same aforementioned laser parameters.

3.4. Graphitisation

The most comprehensive method to characterise graphitic carbon is through Raman spectroscopy. In particular, it probes the vibrational and rotational modes of the material to give information about the crystallographic structure. Since carbon comes in various allotropes, Raman spectroscopy can differentiate between crystallographic diamond

and non-crystallographic diamond-like carbon, however, it cannot be used to quantify the thickness of said layer generated after laser processing. The generation of pockets simulates the partial generation of a cutting edge geometry where the pocket's wall represents the clearance face. As a result, Raman measurements are carried out on the pocket's wall since the pocket's bottom would eventually be fully removed. To quantify the graphitic carbon layer thickness, a FIB is used to generate a cross section on the pocket's wall in order to visualise and measure the layer. Raman spectra for the tested pulse durations and wavelengths are given in Fig. 7a and an example of a FIB cross section with close-ups is given in Fig. 7b. Strong fluorescence effects of the Raman spectra are observed attributed to the influence of the metallic cobalt and hydrogenation is not considered since there is no major source of hydrogen. The Raman spectra in Fig. 6a shows that for Tp = 10 ps, only one peak in the D-band associated with crystallographic diamond is present. On the other hand, for Tp = 1 ns, there exists two major peaks, one in the G-band associated with graphitic carbon and one in the D-band associated with graphitic carbon exhibiting defects. Furthermore, there is no peak in the 2D-band and the two peaks in the D- and G-band are still relatively distinct signifying glassy carbon (GC) composed of sp2 hybridised bonding. The broadening of the peaks, however, and absence of the 2D peak indications the onset of amorphous carbon with a low content of sp3 hybridised bonding, i.e. complete crystallographic disorder. Namely for longer pulse durations, PCD converts to polycrystalline graphite since the graphitisation process initiates through the nucleation of graphitic carbon crystallites. However, for nanosecond pulse durations, there exists rapid heating and cooling of the PCD composite. Instead of polycrystalline graphite, GC and eventually amorphous carbon is generated in line with the amorphisation trajectory as supported by Ferrari (2007). GC is characterised as exhibiting a network of randomly entangled graphite-like ribbon structures resulting in almost isotropic properties.

Raman Shift [cm" ]

—t —1ns, A=1060nm, E =1 8.3liJ —t =1ns, i.=532nm, E =18.3liJ

p p h P P f

—z =10ps, X=1064-nm, E =16jjJ =10ps, ^=532nm, E =16uJ

P P P K Pf

PCD 200 nm

Fig. 7. (a) Raman spectra of PCD after laser processing; (b) Typical FIB generated cross section for Tp = 1 ns, X = 532 nm, Ep = 33 ^J

The FIB cross section is generated at the wall of the pocket, perpendicular to the edge. The cross section in Fig. 6b consists of a platinum (Pt) protective layer, a layer of GC followed by the PCD composite. The PCD composite itself is composed of cobalt inclusions in white, and PCD in black. What can be concluded is that the GC layer is non-uniform, such that the top surface resembles the surface roughness Ra < 100 nm, and the interface with the PCD composite is abrupt. The GC has an average thickness of 0.5 ^m whereby Kononenko et al. (2005) explains that the thickness of graphitic carbon is a function of the distribution dynamics of the absorbed laser energy. What makes PCD composites intriguing compared to other forms of diamond, is that cobalt acts as additional defect sites reducing the graphitisation onset temperature and accelerating the graphitisation process as clarified by Kuznetsov and Butenko (2012). Curtaining is also present on the FIB cross section, attributed to processing a material composite exhibiting inherently two different milling rates, and continuous processing at the same angle of FIB incidence.

4. Conclusion

In summary, an extensive comparative study of Tp = 10 ps and Tp = 1 ns as well as the green and infrared wavelength is carried out by comparing the absorption properties as well as surface roughness and ablated volume from generated pockets. Furthermore, two characteristically different materials are studied, PCD composites and WC. It is determined that nanosecond pulses carrying out ablation through the graphitisation mechanism resulting in residual thermal influences on PCD composites. The use of picosecond pulses have no significant thermal influences and achieved higher ablation rates and lower surface roughness, however, its ablation mechanism has yet to be fully concluded. The advantage of green versus infrared wavelength is made apparent through higher ablation rates, especially aided by tighter focussing capabilities. Raman spectroscopy helped determine the specific carbon allotrope that remains after laser processing in which for Tp = 10 ps, no presence of graphitic carbon is measured while for Tp = 1 ns, glassy carbon is measured regardless of wavelength. Finally, using FIB techniques allows for viewing the graphitic carbon layer having an average thickness of 0.5 (im when Tp = 1 ns and Ep = 33 J

Acknowledgements

The authors would like to thank for support and use of the LASER LINE by EWAG AG. Acknowledgement is also given for the use of equipment by IPG Laser GmbH, Laser Zentrum Nord and the Electron Microscopy Center of the ETH Zurich as well as financial support from the Swiss Commission for Innovation and Technologies. Discussions with Mr. Paul Boerner are also gratefully accredited.

References

Besner, S., Meunier, M., 2010. Laser Synthesis of Nanomaterials, in "Laser PrecisionMicrofabrication". In: Sugioka, K., Meunier, M., Pique,

A. (Eds.). Springer, Berlin, pp. 163-187. Chen, Z., Bogaerts, A., 2005. Laser ablation of Cu and plume expansion into 1 atm ambient gas. Journal of Applied Physics 97, 063305-1-12. Diamond Innovations, 2004. Compax diamond tool blanks for machining.

Djurisi, A., Li, E., 1999. Optical properties of graphite. Journal of Applied Physics 85, 7404-7410.

Dold, C., Henerichs, M., Bochmann, L., Wegener, K., 2012. Comparison of ground and laser machined polycrystalline diamond (PCD) tools in

cutting carbon fiber reinforced plastics (CFRP) for aircraft structures. Procedia CIRP 1, 178-183. Dold, C., Henerichs, M., Gilgen, P., Wegener, K., 2013. Laser processing of coarse grain polycrystalline diamond (PCD) cutting tool inserts

using picosecond laser pulses. Physics Procedia 41, 610-616. Dumitru, G., Romano, V., Weber, H., Sentis, M., Marine, W., 2002. Femtosecond ablation of ultrahard materials. Applied Physics A 74, 729739.

Everson, C., Molian, P., 2009. Fabrication of polycrystalline diamond microtool using a Q-switched Nd:YAG laser. International Journal of

Advanced Manufacturing Technology 45, 521-530. Ferrari, A., 2007. Raman spectroscopy of graphene and graphite: Disorder, electron-phonon coupling, doping and nonadiabatic effects. Solid State Communications 143, 47-57.

Hagemann, H., Gudat, W., Kunz, C., 1974. Optical Constants from the Far Infrared to the X-Ray Region, in "DESY report SR-74/7". Deutsches

Elektronen Synchrotron DESY, Hamburg. Harrison, P., Henry, M., Brownell, M., 2006. Laser processing of polycrystalline diamond, tungsten carbide and a related composite material.

Journal of Laser Applications 18, 117-127. Harrison, P., Henry, M., Wendland, J., Brownell, M., 2005. Enhanced cutting of polycrystalline diamond with a q-switched diode pumped solid

state laser, ICALEO 2005. Miami, USA, paper #202. Kononenko, V., Kononenko, T., Pimenov, S., Sinyavskii, M., Konov, V., Dausinger, F., 2005. Effect of the pulse duration on graphitisation of

diamond during laser ablation. Quantum Electronics 35, 252-256. Kuznetsov, V., Butenko, Y., 2012. Diamond Phase Transitions at Nanoscale, in "Ultrananocrystalline Diamond". In: Shenderova, O., Gruen,

D. (Eds.) Elsevier, pp. 181-244. Phillip, H., Taft, E., 1964. Kramers-Kronig Analysis of Reflectance Data for Diamond. Physical Review 136, A1445-A1448. Rabiey, M., Walter, C., Kuster, F., Stirnimann, J., Pude, F., Wegener, K., 2011. A comparative study on the dressing of hybrid bond CBN wheels using a conventional SiC dressing wheel and a short-pulse fiber laser, Proceedings of the 11th international conference on management of innovative technologies and 2nd international conference on sustainable life in manufacturing, Fiesa, Slovenia. Robertson, J., 2002. Diamond-like amorphous carbon. Materials Science and Engineering R 37, 129-281.

Walter, D., 2010. Online-Qualitaetssicherung beim Bohren mittels ultrakurz gepulster Laserstrahlung. Doctorate thesis, University of Stuttgart. Weber, M., 2002. Metals, in "Handbook of Optical Materials". CRC Press.

Wu, Q., Wang, J., 2011. An experimental investigation of the laser milling process for polycrystalline diamonds. Advanced Materials Research 291-294, 810-815.

Zhang, G., Zhang, B., Deng, Z., Chen, J., 2007. An experimental study of laser cutting mechanisms of polycrystalline diamond compacts. Annals of the CIRP 56, 201-204.