1. Introduction

Axicon-generated Bessel-like optical beams have found numerous applications in the field of imaging [1], optical tweezing [2], micromanipulation [3], two-photon microscopy [4], electric discharge guiding [5], laser micromachining [6] and so on. One of the applications is stealth dicing of glass with the thickness up to several millimetres thanks to the extended non-diffractive length of such beams [7–9]. However, the dicing efficiency is limited by the intra-distance between laser-induced modifications, which may be increased by inducing asymmetrical modifications via spatial and spatiotemporal Bessel beam control [10–14].

Bessel-like laser beams can be efficiently generated by the use of macroscopic axicons, which are usually fabricated via conventional milling, grinding and polishing processes. However, such technology cannot fulfil the ever-growing requirement for the miniaturisation of optical components with arbitrary design and integration into compact systems. For instance, micro-axicons have been applied for optical coherence tomography [15], compact imaging systems of several elements [16] and as beam shapers on the tip of optical fibres [17]. Micro-optics, resembling a cone profile, can be manufactured via electron-beam lithography [18], focused ion beam milling [2], micro glass blowing [19] and case-dependent techniques, such as selective chemical etching of doped optical fibres [17]. However, these technologies are complex, slow and cost-inefficient. The 3D laser lithography is a flexible technique, enabling fast prototyping of conical elements with the desired shape [20] and allowing the integration of different optical elements into the single monolith [21]. However, the relatively low damage threshold of polymers limits the usage of such axicons in high power applications. Although the measured laser-induced damage threshold of record-breaking polymer thin films was over 10 J/cm² for nanosecond pulses [22], this is still lower by an order of magnitude, comparing to the fused quartz [23]. Moreover, more comprehensive research has to be done on the validation of volumetric polymeric structures [24]. An alternative to aforementioned techniques is the precise ablation of optical glasses using ultra-short laser pulses with a subsequent CO₂ laser polishing step [25,26]. Compared to longer laser pulses, ultra-short pulsed laser processing is well-known for its excellent machining quality due to an efficient energy deposition and minimum heat-affected zone [27–30]. For the ablation of dielectrics e.g. optical glasses, electrons are excited from the valence band to the conduction band of the dielectric material via multi-photon absorption and avalanche ionization [31]. A further excitation and heating of these electrons take place by the incident laser radiation before the energy is transferred to the lattice leading to ablation of the material [32].

Several optics-manufacturing methods, based on ultrashort pulsed laser processes, have been demonstrated. Therefore, the fabrication of microlenses with a two-photon polymerisation process [33], the generation of micro-ball lenses inside PMMA [34], applying photosensitive glass consisting of femtosecond machining, etching and two annealing steps [35] or different femtosecond and wet etching processes [36,37] have been shown. Furthermore, different authors demonstrated combinations of both, an ultra-short pulsed and a CO₂ laser. Thus, the ultra-short pulsed lasers are used for the ablation of geometrical preforms, which are subsequently transformed into the desired optical component geometry by a CO₂ laser reshaping process, accompanied by a surface polishing [38,39]. Contrary to this fabrication method, for the laser-fabricated axicons in this paper, reshaping of the ultra-short pulsed laser ablated preform geometry should be minimised while a good surface polishing is ensured. In previous publications, Schwarz et al. evaluated the fabrication of a cylindrical lens [26], an axicon [25], an axicon array [40] and demonstrated the applicability of this two-step optics manufacturing process. In contrast to the aforementioned methods, this two-step laser processing enables the production of optical elements made of glass with a contact-less fabrication enabling a high degree of freedom in optical design.

In this paper, we investigate the all-laser-fabricated axicons and comprehensively compare them to conventional commercial axicons in terms of the shape geometry and quality of the generated Bessel-like laser beam. The experimentally captured intensity distributions were compared to modelling results. Furthermore, axicon-generated beams were applied for glass modification and dicing process. Modified samples were separated using the 4 point bending setup at different processing conditions. To the best of our knowledge, this is the first time non-conventional all-laser-fabricated micro-axicons from fused silica were applied for volumetric processing of glasses, utilising the mJ-level laser system of sub-nanosecond pulse duration.

2. Methods

2.1 Laser-based fabrication technology of axicons

Laser-fabricated axicons were compared to commercial precision and standard axicons (from Eksma Optics) with the same nominal apex angle of 170 deg. Commercial axicons were fabricated from fused silica and differed in the quality of the tip. Two types of laser-fabricated axicons, namely LF-1 and LF-2, were fabricated via femtosecond laser ablation and CO₂ laser polishing in natural fused silica (GVB solutions in glass). Figure 1(a) shows this two-step manufacturing process schematically. Firstly, an 1030 nm ultra-short pulsed laser (Pharos, from Light Conversion) with a pulse duration of 230 fs (at FWHM) was used to ablate the pre-defined axicon geometry in a layer-by-layer process. Therefore, hatches, consisting of parallel lines, were ablated while changing the scanning direction by 100 deg after each layer for a more homogeneous ablation. The distance between two adjacent lines as well as the pulse-to-pulse distance within each line was set to 12 µm, and the fluence was adjusted to remove 1 µm of the fused silica with each ablated layer to have a controllable and precise 3D ablation process. Subsequently, the fabricated preforms were polished with a CO₂ laser (Infinity, from Iradion) having a wavelength of 10.6 µm and maximum output power of 77 W. The polishing process was also conducted by scanning hatches with parallel lines having a distance of 25.4 µm while the maximum power of the laser was used without a focal lens. This polishing step was performed 4 times while after each pass, the sample is rotated by 90 deg to get a homogeneous polishing result. Two different scanning speeds, 7.8 mm/s and 8.2 mm/s, were applied for polishing the LF-1 and LF-2 axicons, respectively, resulting in the different quality, discussed in Section 3.1. The aperture size of laser-fabricated axicons was 2 mm. All laser-fabricated and commercial axicons were uncoated. For more details of this optics fabrication process, see Refs. [25,30,40].

 

Fig. 1. (a) Schematic of the two-step laser-based axicon fabrication with a femtosecond laser ablation step and a subsequent CO₂ laser polishing process. (b) Experimental setup for glass processing. (c) The magnified view of the generation of modifications and directions of observations with respect to a sample.

Download Full Size | PDF

2.2 Characterisation of axicons

The fabricated axicons were characterised using a stylus profiler Dektak 150 (from Veeco) to define the geometry of an axicon and a laser scanning microscope VK-X210 (from Keyence) to measure the surface roughness. The measured profile was compared to an ideal cone of the 170 deg-apex angle. The distance from the actual profile position to the cone surface was evaluated to obtain the shape error dependence on the transverse distance to the optical axis of an element.

The fundamental harmonic (1064 nm) of the DPSS laser Atlantic HE (from Ekspla) was used to conduct the experiments. The pulse duration at FWHM was 300 ps, and the maximum pulse energy equalled to 2.5 mJ at a pulse repetition rate of 1 kHz. The diameter of the incident Gaussian beam was adjusted using a demagnifying telescope and equalled to ∼1 mm at the 1/e² intensity level to avoid diffraction effects from the axicon aperture edges. The axicon-generated beam was imaged on the CMOS camera Beamage-4M (from Gentec) using a magnifying system to obtain the intensity distribution. The spectrum of spatial frequencies was measured by placing a positive lens behind an axicon, which carries out the Fourier transform of the Bessel beam, and by capturing the resulting intensity distribution in the focal plane by the CMOS camera.

In part of experiments, the axicon was tilted around the Y-axis of the coordinate system to add some astigmatic aberrations into the system to obtain the elliptical intensity pattern in the XY plane [Fig. 1(b)]. More details about the axicon tilt operation could be found in Refs. [13,41].

2.3 Modelling

The axicon-generated intensity patterns behind axicons were simulated by the use of principles of Fourier optics [42]. The field distribution in the observation plane was calculated by solving the Rayleigh–Sommerfeld diffraction integral, using the two-dimensional fast Fourier transform. The field in the source plane was calculated according to the parameters of the incident Gaussian beam, which phase was modulated by an axicon. The actual profile shape, obtained via stylus profilometry, was taken into account by fitting one side of an axicon by a polynomial function. Axicons were considered to be rotationally symmetrical, except the standard axicon, which has an elliptical cross-section. In this case, the additional term was added while defining the transverse distance to the optical axis of an axicon. More details about simulations and comprehensive formulae could be found in Refs. [10,13,43].

2.4 Glass processing

Single-shot modification and glass dicing experiments were carried out using the experimental setup, shown in Fig. 1(b). To obtain sufficient intensity for glass processing, the central core extent and non-diffractive length of the axicon-generated beam was reduced using the demagnifying system, which consisted of two positive lenses with the focal lengths of + 200 mm and + 30 mm. The vertical position of samples was controlled using a stage with a stepper motor 8MT167-100 (from Standa). The linear stages ALS25020 (from Aerotech) were used to move the samples in the XY plane. Single-shot laser-induced modifications were inscribed into the all-sides polished fused silica samples and then side-viewed using the optical microscope Nikon Eclipse LV100 in transmission and dark-field regimes. Modifications were viewed perpendicularly to XY and XZ/YZ planes, as shown in Fig. 1(c), as Obs 1 and Obs 2 directions. The peak intensity values, given in Section 3.4, were calculated according to relation (3) in [44], considering the increased half-angle of the Bessel cone and reduction of the Bessel zone length after passing the demagnifying system.

Glass dicing experiments were performed on 76 × 26 mm² soda-lime glass plates with a thickness of 1 mm. Samples were diced through the middle and then separated using the 4 point bending setup. The length of the inner and outer support spans was 16 mm and 60 mm, respectively. The breaking force was measured using the digital dynamometer FMI-S30A5 (from Alluris) with or without an additional lever, which extends the measuring range, and then recalculated to the flexural strength of the material. More details about flexural strength evaluation could be found in [41].

3. Results and discussion

3.1 Axicons’ geometry

The measured profile shapes of axicons are presented in Fig. 2(a). As can be seen, the most critical part of axicons is their tip. The laser-fabricated LF-2 axicon has a similar radius of the rounded tip (R = 0.6 mm) as the commercial precision axicon (R = 0.55 mm). Although the LF-1 axicon is slightly rounder (R = 1.3 mm), it is remarkably sharper than the commercial axicon of standard quality (R = 6.7 mm).

 

Fig. 2. (a) The measured profile shapes of axicons using a stylus profiler. A 10 µm-offset was added to the profile of the non-polished laser fabricated axicon to avoid overlapping with other curves. The estimated shape error for a laser-fabricated axicon before (b) and after (c) the polishing step and for the commercial precision (c) and standard (d) axicons. The inset graph in (d) shows the cross-section ellipticity versus transverse distance to the optical axis of the standard axicon.

Download Full Size | PDF

The estimated shape error versus the transverse distance to the optical axis is shown in Figs. 2(b)–2(d). The absolute shape error of the laser-fabricated axicon, measured in the range from -0.75 to 0.75 mm, was equal to 12.1 µm before the polishing step. However, the subsequent laser-polishing step reduced this value to 7.0 and 3.5 µm for LF-1 and LF-2 axicons, respectively. Additionally, the average surface roughness R_a of laser-fabricated axicons was reduced from 0.65 µm to 31 nm and 55 nm for LF-1 and LF-2 axicons, respectively. In comparison, the absolute shape error and surface roughness of the precision axicon is equal to 3.3 µm and 35 nm, respectively. All these axicons have a V-shaped error profile close to the tip, ranging from -0.1 to 0.1 mm transverse distance, and an almost constant shape error in the outer region close to edges. The shape error profile is slightly wavy of LF-1 and precision axicons, while the rough texture can be clearly recognised for the LF-2 axicon. Therefore, the LF-1 axicon has a more rounded tip and better surface quality, especially on the outer region, comparing to the LF-2 axicon, which has a sharper tip, but lower surface quality closer to axicon edges.

The maximum shape error of the standard axicon, measured in the range from -10 to 10 mm and the surface roughness R_a are equal to 60 µm and 40 nm, respectively. The shape error is non-flat even at high distances from the tip because the base angle of this axicon is slightly lower than nominal and equals to 4.9 deg even in the outer region. Also, it was detected that the standard axicon has an elliptical cross-section, which was evaluated as the ratio of major to minor axis length. The ellipticity dependence on the transverse distance is shown in the inset graph in Fig. 2(d). The ellipticity value, equalling to 1, represents an ideal symmetrical axicon cross-section. The cross-section ellipticity shows unstable and oscillating behaviour at 0-1.5 mm transverse distance. However, at longer distances, the ellipticity value tends to decrease and equals to 1.01.

3.2 Spectra of spatial frequencies

The experimentally captured and numerically simulated spectra of spatial frequencies of axicon-generated Bessel beams are presented in Fig. 3. The commercial precision axicon generates the Bessel beam, which has a reasonably good spatial spectrum, comparable to an ideal axicon, which generates a ring, which thickness reduces with the increase of a radius of the incident Gaussian beam [44]. The maximum spatial spectrum intensity of the precision axicon-generated beam is 21% lower, compared with an ideal axicon-generated spectrum, shown as the dark cyan curve in the modelling section in Fig. 3(b). Note that values herein are normalised to the LF-2 intensity level.

 

Fig. 3. (a) The experimentally measured and modelled spectra of spatial frequencies, obtained in the XY plane; (b) the cross-sections of spectra along the X-direction. The dark cyan curve in the modelling section (b) represents the spatial spectrum of the Bessel beam, generated by an ideal sharp-tip axicon.

Download Full Size | PDF

The LF-2 axicon-generated beam has a similar intensity to the precision axicon-generated beam. However, the spatial spectrum is more disturbed, and there are some low and high-frequency components, which could be potentially removed using spatial filtering [44,45]. The LF-1 axicon generates the smoother and cleaner spatial spectrum, as a result of the better surface finishing quality. However, due to the more rounded tip, the ring is broader, and the maximum intensity is 37% lower. Perturbations in the spatial spectra, generated by laser-fabricated axicons, may occur due to the shape deviations. In other words, the cross-sections of these axicons are not ideal, but slightly wavy rings in contrast to the precision axicons, which are fabricated with well-defined turning operation during milling.

The standard axicon generates the qualitatively different spectrum of spatial frequencies. The shape of a spectrum is broader and more disc-like, rather than a ring. The maximum intensity is 80% lower, comparing to the precision axicon.

3.3 Axicon-generated intensity distribution

The experimentally measured and modelled on-axis intensity distributions of Bessel beams, generated by laser-fabricated and commercial axicons, are presented in Figs. 4(a) and 4(c), respectively. Modelling curves are solid and dashed lines while dots represent experimental data. The LF-1 axicon generates the Bessel beam at ∼1 mm longer distance from the tip, comparing to the LF-2 axicon due to the rounder tip. Also, the maximum intensity is higher for the LF-1 axicon. LF-2 axicon generates intensity distribution with lower on-axis oscillations, comparing with simulation results, since the wavy axicon shape [Fig. 2(c)] is not fully rotationally symmetrical, as considered in simulations. The precision axicon generates a smoother intensity distribution, comparable to the one, generated by an ideal sharp-tip axicon, as shown in Fig. 4(c). On the contrary, the standard axicon generates intensity distribution, which is shifted about 10 mm along the beam propagation direction and has well-defined intensity oscillations and the highest maximum intensity among all tested axicons. However, we could expect the better performance of the standard axicon with the increased diameter of the incident Gaussian beam, since the relatively smaller part of the beam would be refracted by the oblate tip. Therefore, the on-axis intensity distribution would be more uniform and the spatial spectrum would turn to more ring-like.

 

Fig. 4. The experimentally measured and modelled on-axis intensity distribution (a, c) and central core diameter at FWHM (b, d) along the beam propagation direction for laser-fabricated (a, b) and commercial axicons (c, d). The modelling curves are solid lines. The dashed line represents the Bessel beam, generated by an ideal sharp-tip axicon (modelling data).

Download Full Size | PDF

The experimentally measured and modelled central core diameters at FWHM of Bessel beams, generated by the laser-fabricated and commercial axicons, are presented in Figs. 4(b) and 4(d), respectively. The central core diameter of the ideal axicon-generated beam is slightly below 10 µm, according to the modelling data. It is shown in the presented graphs as a dashed line for the reference. In all cases of real axicons, the central core diameter is the largest close to the tip, because of the incident wave refraction by lower angles due to the tip roundness.

LF-2 and precision axicons show similar central core size dependence on the distance to the axicon tip. Starting from ∼3 mm, it becomes comparable to the one obtained by an ideal sharp-tip axicon, except that for laser-fabricated axicons slight size oscillations occur, especially at longer propagation distances. In the case of the standard axicon, oscillations are periodical and more pronounced.

The experimentally captured intensity distributions in the XY plane at various distances from the tip are presented in Fig. 5. Laser-fabricated axicons generate beams, which maintain their Bessel-like intensity distribution with an intensive central core and surrounding concentric rings up to ∼15 mm propagation distance. However, with the increase of the propagation distance, the perturbations become more and more pronounced, and the generated beam even reshape into a tripod-like structure. This could be caused by an uneven contour of an axicon cross-section, as discussed in the previous section. This is evident for the LF-2 axicon, which has a wavy profile shape [Fig. 2(c)], and disruptions of the side lobes start to appear at shorter propagation distances comparing to the LF-1 axicon. On the other hand, the LF-2 axicon has a sharper tip, and the Bessel-like intensity pattern is generated closer to the axicon; therefore, the Bessel zone length is similar for both cases. In comparison, the beam disintegration is avoided over the longer propagation distance in the case of the precision axicon. However, the standard axicon generates exclusive intensity patterns with lower contrast between side lobes.

 

Fig. 5. The experimentally captured intensity distributions in the XY plane at various distances from a tip of axicon-generated Bessel beams.

Download Full Size | PDF

Some applications, e.g. glass cleaving [11,41], require the asymmetrical intensity distribution. Therefore, we have investigated one of the approaches – oblique axicon illumination. Axicons were rotated around the Y-axis of a coordinate system to add some astigmatic aberrations into the optical system for the introduction of the asymmetry in the transverse plane (XY). The experimentally captured and modelled intensity distributions at the 15 deg-tilt configuration for various distances are presented in Fig. 6. Experimental results are in a good agreement with simulations for commercial axicons; however, there is a slight discrepancy for the laser-fabricated axicons and the standard axicon.

 

Fig. 6. The experimentally captured and modelled intensity distributions in the XY plane at various distances of Bessel beams, generated by tilted axicons. The tilt angle equalled to 15 deg.

Download Full Size | PDF

Disruptions of the beam due to oblique illumination become more pronounced at longer propagation distances when Bessel rings gradually reshape into distinct maxima, the intensity of which becomes even higher than in the central core. In the ideal case, these maxima are orientated along the X and Y axes. However, laser-fabricated axicons generate slightly rotated intensity patterns in comparison to the precision axicon. This effect is more pronounced for the LF-2 axicon of the worse surface quality.

The asymmetry of the beam was evaluated as the ratio of measured distances between sidelobe maxima of the first ring along the Y and X directions. The ellipticity at 3.5 mm distance from the tip is equal to 1.15, 1.08 and 1.04 for LF-1, LF-2 and precision axicons, respectively. Therefore, the asymmetry dependence on the tilt operation is more pronounced for rounder axicons. According to numerical modelling, the sharp-tip axicon would generate a beam with ellipticity, equal to 1.03 at the particular distance from the tip.

As it was stated before, the standard axicon has an elliptical cross-section. As a result, the intensity pattern of the generated beam is asymmetrical even at 0 deg tilt. At 15.5 mm-distance from the tip, the ellipticity equals to 1.07 and is enlarged to 1.56 when axicon is tilted to 15 deg, and the major axis of the axicon cross-section is orientated along the Y-axis.

3.4 Single-shot glass modification

Single-shot intra-volume modifications were induced in the bulk of fused silica using axicon-generated Bessel beam and then side-viewed. Experiments were carried out using a demagnifying optical system with the nominal reduction factor of f₁/f₂= 200 mm/30 mm = 6.7. The measured central core diameter at FWHM after passing through the system was equal to 1.4 µm for the Bessel beam, generated with the LF-1 axicon. This corresponds to the 16 deg half-angle of the Bessel cone in the air. The relatively high half-angle and maximal peak intensity level below 2×10¹³ W/cm² ensured the stable Bessel beam propagation regime with the uniform energy deposition into the material [46–48].

Optical microscope images of modifications are presented in Fig. 7. Modifications were induced using 2 mJ pulse energy (on the sample), except for Fig. 7(b), when 0.5 mJ pulses were applied. The non-tilted LF-1 axicon induced modifications of 1.25 mm overall length, taking into account the pale tail, visible in Fig. 7(a). The length of the main part of modifications omitting tail was equal to 0.94 mm. The non-tilted axicon induced disruptive modifications, which consisted of the laser-induced narrow track, having 6.2 µm-diameter in the middle part, and transverse cracks, which propagated at chaotic directions. The estimated threshold intensity to generate modifications was 1.7×10¹² W/cm², while the maximum peak intensity was 16×10¹² W/cm². Therefore, more even although shorter modification tracks could be induced with lower pulse energy, as shown in Fig. 7(b). In this case, the track width was reduced to 2.8 µm. Another possible approach would be to use shorter laser pulses to exploit the more gentle modification mechanism of the material. The oblique illumination at 15 deg resulted in more determined modifications, as seen in Fig. 7(e). The pale tail disappeared, and modification length was equal to 1.14 mm; also, the modification tracks became slightly wider at longer distances from the tip due to the intensity pattern reshaping into several maxima, as seen in the presented intensity distributions in Fig. 6. Therefore, it supposes that side lobes have enough intensity to contribute for induction of the modifications. However, the most important result is tidier transverse cracks, which could be clearly seen in the dark field regime, shown in Fig. 7(g). At oblique illumination, the transverse cracks mainly spread along the YZ plane, which is parallel to the rotation axis of the axicon or, in other words, along the major axis of the elliptical intensity pattern, while spreading into the XZ plane is highly suppressed. This result is consistent with our previous findings [13]. The LF-2 axicon induced comparable modifications with 0.94 mm longitudinal length and 200 µm-width of random cracks at non-tilted configuration. At 15 deg-tilt, the longitudinal length of modifications, shown in Fig. 7(f), was 1.17 mm, and transverse cracks were directional.

 

Fig. 7. Laser-induced single-shot modifications in the bulk of fused silica observed perpendicularly to YZ and XZ planes. Modifications were viewed using an optical microscope in the transmission (a-f) and dark-field (g) regimes. Bessel beams were generated using laser-fabricated (a, b, e-g) and commercial (c, d) axicons, which were at non-tilted configuration (a-f) and tilted to 15 deg (e-g). The laser pulse energy was 2 mJ (a, c-g) and 0.5 mJ (b). The beam propagation direction was from left to right.

Download Full Size | PDF

The modifications, induced by Bessel beams, generated using non-tilted commercial axicons, are shown in Figs. 7(c) and 7(d). The longitudinal length of modifications was higher than obtained with laser-fabricated axicons and equalled to 1.68 mm and 1.75 mm for precision and standard axicons, respectively. The precision axicon generated a uniform modification track in contrast to modifications, induced by standard axicon generated beam, which were unsteady and discontinuous. The probable reason for such behaviour is the on-axis intensity oscillations and the broad spectrum of spatial frequencies, which results in the lower resistivity to spherical aberrations, occurring in the air/glass interface.

The images of transverse cracks in the XY plane and width dependence on the tilt angle are shown in Fig. 8. In the case of the laser-fabricated axicon, the width of transverse cracks in the XZ plane decreases with the increase of the tilt angle, while in the YZ plane, the maximum crack width of 250 µm is reached at 10 deg-tilt [Fig. 8(b)]. As seen in Fig. 8(a), the oblique illumination gives the possibility to control cracks orientation when cracks turn from a star-like to a plaice-like shape. It might also be noted that the LF-2 axicon generates similar modifications as LF-1 with the transverse crack width of 230 µm at 10 deg-tilt.

 

Fig. 8. (a) The optical microscope images in the transmission regime of laser-induced single-shot modifications in the bulk of fused silica, observed perpendicularly to the XY plane. Red dashed lines indicate the dominant direction of transverse cracks. (b) The transverse cracks width dependence on the tilt angle. Solid and dashed lines are for eye-guiding.

Download Full Size | PDF

The transverse cracks width is considerably shorter using the precision axicon and equals to 135 µm. The longer modifications, shown in Fig. 7(c), suppose that the laser energy is distributed over the larger volume; therefore, the intensity is lower in the material, leading to the smaller cracks width. Furthermore, the controllability of the direction of transverse cracks is less evident in this case, probably due to the higher symmetry of the generated pattern, because of the sharper tip of the axicon. Even shorter cracks of about 100 µm-width are obtained with the standard axicon. However, in this case, the directional transverse cracks are generated even with the non-tilted axicon, as seen in Fig. 8(a), where cracks align around 45 deg with respect to the X/Y axes, as indicated by red dashed lines. Such alignment corresponds to the position of the major axis of the elliptical cross-section of the standard axicon. Therefore, the direction of cracks could be easily controlled by simply rotating the axicon around its axis. When tilting is applied, cracks start to align along the YZ plane; therefore, they are controllable.

3.5 Glass dicing experiments

Dicing experiments were carried out with the laser-fabricated, precision and standard axicons in a 1 mm-thick soda-lime glass. Laser pulse energy hitting the sample and repetition rate were adjusted to 2 mJ and 1 kHz, respectively. The laser-fabricated and precision axicons were additionally tilted to 15 deg to obtain directional transverse cracks. The glass dicing performance was similar using LF-1 and LF-2 axicons, therefore, the results obtained with the former axicon are presented only. The tilt operation did not give a substantial change in the case of the standard axicon; therefore, herein only results with the non-tilted configuration are presented. As it was mentioned before, due to the elliptical cross-section of the standard axicon, the directional cracks were generated even at 0 deg-tilt. Therefore, the axicon axial position was adjusted so that cracks were aligned parallel to the cutting path to achieve the easiest separation of glass plates.

The flexural strength of modified glass plates versus dicing speed is presented in Fig. 9. The pitch or intra-distance between laser-induced modifications along the scanning direction is shown in the top label. The optical microscope images of diced and separated glass plates are shown in Fig. 10. In the case of the LF-1 axicon, the flexural strength at low dicing speeds was similar for both non-tilted and tilted configurations and equalled to ∼9 MPa. However, the significantly different behaviour of the flexural strength was observed when the dicing speed was increased. In the case of the non-tilted configuration, the flexural strength resulted in the steep increase with the increase of dicing speed, reaching 37 MPa at 150 mm/s which resulted in a pitch distance of 150 µm. Moreover, transverse cracks tended to spread into the bulk of the material, that resulted in the uneven cleavage plane and poor quality. In comparison, the deviation of transverse cracks was highly reduced by the tilted axicon configuration. This enabled us to achieve higher dicing speeds with satisfactory dicing quality, as seen in optical microscope images in Fig. 10. The flexural strength was below 11 MPa up to 125 mm/s and below 18 MPa even at very high dicing speed of 225 mm/s.

 

Fig. 9. The flexural strength dependence on the dicing speed of glass plates, modified using axicon-generated beams. The laser-fabricated and precision axicons were additionally tilted to 15 deg. Solid and dashed lines are for eye-guiding.

Download Full Size | PDF

 

Fig. 10. The optical microscope images of diced and separated glass plates at different scanning speed using different axicons and non-tilted and tilted configuration. Glass plates are on the right side.

Download Full Size | PDF

At 25 mm/s dicing speed and 0 deg-tilt angle, the flexural strength of glass plates, modified using the non-tilted precision axicon, was similar to the results, obtained with the LF-1 axicon. The flexural strength increased continuously with the increase of the dicing speed; however, it was lower than using the non-tilted LF-1 axicon in the high-speed range. The precision axicon generates more gentle modifications, consisting of randomly distributed cracks of 135 µm-width, in comparison to disruptive tripod-like cracks, induced using the LF-1 axicon [Fig. 8(a)], which tend to penetrate into the bulk material [Fig. 10]. Therefore, the more regular cleavage plane is created with lower flexural strength, using the precision non-tilted axicon.

In the case of the tilted precision axicon configuration, the minimum flexural strength of the modified material was equal to 5.4 MPa at 50 mm/s dicing speed. However, the flexural strength increased sharply at 100 mm/s speed and was higher compared to the non-tilted configuration at the high-speed range. Such result was predictable since the intra-distance between laser-induced modifications became larger than the transverse cracks width of ∼70 µm [Fig. 8(b)].

In the case of the standard axicon, the minimum flexural strength of the modified material was equal to 10 MPa at 75 mm/s dicing speed and 0 deg-tilt. Also, the cutting quality was relatively good due to the guiding effect of the cleavage plane even at high dicing speed. Overall, the standard axicon showed better performance compared to the precision axicon at higher dicing speed, but worse than the LF-1 axicon. The main reason is shorter cracks, probably due to the lower intensity. In the case of the precision axicon, the low controllability of transverse cracks is the main reason for low quality and high flexural strength, which may be caused by the fact that sharper axicon generates more symmetrical intensity distribution even at oblique illumination in contrast to axicons with rounder tips.

4. Summary and conclusions

In this report, we have comprehensively characterised the laser-fabricated axicons and compared with the commercial axicons. We have investigated the shape of axicons, spatial spectra, intensity distributions of generated beams at non-tilted and oblique illumination configurations and compared to numerical modelling results. Furthermore, the investigated axicons were applied for intra-volume glass modification and dicing with mJ-level laser pulses.

The overall quality of laser-fabricated axicons was more comparable to the precision axicon rather than to the standard axicon. The laser-fabricated axicons, having the tip shape error well below 10 µm, generated the decent spatial spectrum and Bessel-like intensity pattern with the long non-diffractive length. However, there were slight on-axis intensity oscillations due to oblate-tip and perturbations, which were more noticeable at longer propagation distance when the beam reshaped into a tripod-like structure. The surface quality was critical for the performance of laser-fabricated axicons. Despite the rounder tip and larger beam shift, the smoother axicon generated the intensity pattern of improved quality, which was prominent at oblique illumination.

Laser-fabricated axicons were applied for intra-volume glass modification. Although the longitudinal length of modifications was 27% lower compared to ones, generated with commercial axicons, the transverse spreading of laser-induced cracks was more pronounced with laser-fabricated axicons. Tilted axicon configuration forced the transverse cracks to orientate along the aimed direction. At 15 deg-tilt and 2 mJ pulses, the transverse cracks width and the longitudinal length of modifications were 200 µm and 1.1 mm, respectively. Such configuration significantly improved the 1 mm-thick glass dicing quality and allowed us to outperform the commercial axicons by extending the working window of dicing speed up to 125 mm/s with a flexural strength of the modified material below 11 MPa. To the best of our knowledge, this is the first time non-conventional micro-axicons were applied for volumetric processing of glasses.

We have demonstrated that all-laser-based technology allows the fabrication of high-quality axicons, which were applied for glass processing applications. This technology is advantageous in the flexible fabrication of millimetre-sized optical elements with the desired 3D surface, which is challenging with conventional fabrication techniques. Laser-fabricated axicons are not restricted to dicing applications, but could also be potentially applied for material modification, drilling, ablation or imaging applications.