Sei sulla pagina 1di 8

Laser micromachining of 'transparent' fused silica with 1-ps pulses and pulse trains

Peter R. Herman a , Anton Oettl b , Kevin P. Chen a , and Robin S. Marjoribanks b

a Department of Electrical and Computer Engineering, b Department of Physics,

University of Toronto

&

Photonics Research Ontario

Toronto, ON, M5S-3G4, Canada

ABSTRACT

Ablation rates and etched-surface morphology of fused silica has been studied with 1-ps Nd:glass laser pulses in a regime of near-diffraction-limited spot size. Shallow holes of 1.7-mm diameter were too small for the formation of laser-induced periodic-surface structures. Atomic-force and scanning-electron microscopy showed that reproducible etch depth and moderately smooth surfaces are attainable for low fluences of 5.5 - 45 J/cm 2 — the “gentle” ablation regime. Etch depth progressed linearly with the number of laser pulses until the onset of surface swelling and shock-induced microcracks after a critical number N c of laser pulses, scaling as N c = 1.7 + 80/F (fluence F in J/cm 2 ). Below this limit — for accumulated etch depths less than ~2 mm — three-dimensional surface structuring with sub-micron precision is possible with picosecond-laser pulses. In the strong ablation regime (F > 45 J/cm 2 ), surface morphology was poor and microcracking developed within 2-4 pulses. These shock-induced microcracking effects were eliminated when a mode-locked train of ~400 identical 1-ps pulses, each separated by 7.5 ns, was applied. Very smooth and deep (~30-mm) holes of 7–10-mm diameter were excised at a total fluence of ~100 kJ/cm 2 , establishing a new means for rapid and precise micromachining of fused silica and other brittle materials.

Keywords: laser micromachining, ablation, ultrafast lasers, mode-locked pulses, fused silica, transparent materials

1. INTRODUCTION

Ultrafast lasers of different sorts are proving to be an excellent source of new tools for precise micromachining of a wide range of materials, especially including wide-bandgap or transparent materials. Single-picosecond and femtosecond lasers give qualitatively new results, by extending our control over how a given pulse energy is delivered. Their intense optical fields drive distinctly different, nonlinear, absorption mechanisms at relatively modest fluence values of 1-10 J/cm 2 . Their shorter time scale of laser interactions reduces collateral damage and avoids plume-shielding effects. Further, laser energy can be preferentially deposited into exceptionally small volumes (<1-mm 3 ), affording an unprecedented degree of control and precision in sculpting surfaces 117 or changing the refractive index 18-20 of optically transparent materials. These advantages offer prospects for ultrafast-laser shaping of surface or internal microstructures to define miniature photonic components, now in high demand in optical communication networks. 21 Photonic devices require optically smooth surface morphology, controllable etch-rate, and sub-micron resolution. While several groups 2,9-17 have demonstrated smooth surface etching on various optical materials using picosecond to 20-fs laser sources, suitable processing windows are too poorly defined for application purposes. Etch rates are frequently haphazard, falling into ‘gentle’ and ‘strong’ ablation phases, 2,14 and affected by varying degrees of material incubation, laser beam self-focussing, and laser-prepulse energy. Control over ablation debris or melt, surface rippling, swelling, shock-induced microcracking, and bulk ‘clouding’ effects (radiation damage) must also be addressed. Lastly, much of the prevailing literature is centered on features greater than 10-mm diameter — too large for most photonic device applications. Correspondence: Email: hermanp@ecf.utoronto.ca; WWW: http://www.ecf.utoronto.ca/~hermanp; Telephone: 416 978 7722; Fax: 416 971 3020

148

Part of SPIE Conference on Commercial and Biomedical Applications of Ultrafast Lasers San Jose, California – January 1999 SPIE Vol. 3616 – 0277-786X/99/$10.00

This paper presents a systematic study of laser ablation of uv-grade fused silica (Corning 7940; 160-mm thick) with 1.2-ps laser pulses from a 1.054-mm Nd:glass laser. The purpose is to define a laser-processing window for 3-D microstructuring of transparent glass surfaces. Etch rates were examined over a wide single-pulse fluence range (up to 180 J/cm 2 ) using near- diffraction-limited spot sizes of 1-3 mm diameter. These tight-focussing conditions should hinder the formation of laser- induced periodic surface structures (LIPSS), improving the prospects for smooth surface-morphology. The development of shock-induced microcracks and surface swelling is observed to be the main limitation to smooth and controllable

micromachining in both the ‘gentle’ and ‘strong’ ablation regimes. Nevertheless, a laser processing window is described for practical 3-D microstructuring when features less than ~2-mm deep are to be formed. We also describe a new and very promising approach to smooth micromachining of brittle glasses at high fluences, without induced microcracking. A cursory investigation of the use of a train of ~400 pulses from a special mode-locked Nd:glass oscillator showed that smooth symmetric holes of 10-mm diameter and up to 30-mm deep could be excised without inducing fractures in the hole perimeter. High-repetition-rate (133 MHz) multi-pulse ablation therefore emerges as a new option for controllable and smooth etching

of brittle materials.

2. EXPERIMENTAL ARRANGEMENT

A feedback-controlled Nd:glass oscillator 22 (l = 1054 nm) operating at 1 Hz repetition rate provided a flat-topped train of

430 mode-locked pulses with pulse-to-pulse separation of 7.5 ns. A single high-contrast pulse of 1.2-ps duration was selected by an external Pockels cell and amplified 13-fold in a four-pass geometry. The ~3-mJ pulses were focussed by interchangeable GRIN lenses (f = 15.4, 11.0, 4.5 and 3.1 mm) to near diffraction-limited spot sizes of 3.2, 2.0, 1.0, and 0.8 mm (1/e 2 ) diameter, respectively. The polished glass samples were mounted on a precision x-y-z stage. Focussing was monitored by image-relaying the retro-reflected beam from the focal spot, with magnification, onto a CCD camera. On- target fluence was varied over the range 2 to 170 J/cm 2 by adjusting the amplifier gain, using neutral density filters, and employing different focal-length lenses. Excisions were made using between 1 and 60 pulses, of various fluence values. All samples were irradiated in air, and transverse nitrogen gas flow was at times used to reduce the accumulation of ablation debris. Self-focusing effects in air were not seen, at this pulse duration, peak power, and with the short focal-length lenses used. Laser focussing conditions also did not produce bulk discoloration or damage effects in regions beneath the excised holes (as evidenced by optical microscopy).

A waveplate was at times added, to pass the full oscillator train of 430 pulses for a parallel study of high-repetition-rate

machining. The 3-ms long pulse train was amplified and focussed onto glass surfaces as described above, accumulating a total fluence of ~40 kJ/cm 2 in a diameter of ~2 mm.

Three characterization methods were used to examine surface morphology and determine hole depths: Nomarski optical microscopy, atomic force microscopy (AFM), and scanning emission microscopy (SEM).

3. SINGLE-PULSE MICROMACHINING

In this section, results of 1.2-ps laser ablation of fused silica at repetition rates of 1 Hz or less are described. Surface

morphology of microholes formed by single laser-pulses is shown in the AFM photographs on the left side of Figure 1. Fluences of 9.1 and 38 J/cm 2 each produced moderately smooth holes of ~2.0-mm diameter (FWHM). Surface-profile traces, shown adjacent, reveal hole depths of 100 and 360 nm, respectively. A small ring structure is observable in the higher- fluence hole, a feature also reported by Ashkenasi et al. 13 for 3.2-ps ablation of fused silica. The excised surface contour was found to crudely follow the laser beam profile, with small-scale surface roughness of ±10% (rms) of the hole depth. This ±10% surface roughness was a general observation for the ‘gentle’ ablation phase, noted here for fluences, F, less than ~44 J/cm 2 . Even when several pulses were applied to the same area, surface roughness typically increased in absolute terms, but remained limited to ±10% of the final hole depth. For 1-mm structures, one can therefore expect surface roughness of ~100 nm, a value that will lead to only modest optical scattering losses for photonic device applications.

Figure 2 shows the progress of hole depth with the number of laser pulses, N, for fluence values of 9.6 and 38 J/cm 2 . (Accumulated fluence was used for the abscissa to better account for the ±10% variations of the laser energy). For 9.6 J/cm 2 (38 J/cm 2 ), the depth increases linearly with N (accumulated fluence) to an apparent peak value of 2.7 mm (2.2 mm) after 14

149

(6) pulses. The nominal plateau apparent for higher pulse-number is simply an artifact of the AFM tip, which cannot probe larger aspect-ratio holes: those deeper than their ~2-mm diameter. It is anticipated that the hole depth will in fact rise linearly with pulse number through and beyond this plateau region until incubation processes raise the density of defects or color centers to a critical value. Beyond this critical value, the ‘gentle’ ablation process is expected to give way to ‘strong’ ablation, a distinct regime wherein etch rates (depth per pulse) can be increased more than 10 fold. 2 For fused silica, Varel et al. 12 reported etch rates of 550 nm per pulse when 100’s of pulses at 1.3-ps duration were applied at 12 J/cm 2 fluence. Their rate is triple our 180 nm/pulse rate for the same fluence, but with N < 10. For reasons given below, the transition to strong ablation with increasing N was not studied here.

to strong ablation with increasing N was not studied here. nm nm Figure 1: AFM image

nm

strong ablation with increasing N was not studied here. nm nm Figure 1: AFM image (left)

nm

ablation with increasing N was not studied here. nm nm Figure 1: AFM image (left) and

Figure 1: AFM image (left) and depth profile (right) of a micro-hole in fused silica, drilled by a single 1.2-ps laser pulse with a peak fluence of 9.1 J/cm 2 (top ) and 38 J/cm 2 (bottom). The ~2-mm diameter holes (FWHM) have a 10% rms surface roughness.

The etch-depth data in Figure 2 show that material removal was initiated with the first laser pulse (N = 1), for both 9.6 or 38 J/cm 2 fluences. Incubation effects developing at fluences before the onset of ablation were not studied in the present work although such effects are already anticipated below our single-pulse ablation threshold of ~5.5 J/cm 2 . Kautek et al. 11 reported the need for 7 incubation pulses before low-fluence (1 J/cm 2 ) ablation of barium-aluminum borosilicate glass could proceed with 50-fs laser pulses. Such incubation processes are undesirable for practical applications, and were deliberately avoided here by applying fluences that exceeded the single-pulse ablation threshold of fused silica.

150

3000 2500 F = 9.6 J/cm 2 2000 1500 1000 F = 31 J/cm 2
3000
2500
F = 9.6 J/cm 2
2000
1500
1000
F = 31 J/cm 2
500
0
0
50
100
150
200
250
300
Etching Depth (nm)

Accumulated Fluence (J/cm 2 )

2000 F th = 11 J/cm 2 -1 1500 a eff = 1.28 x 10
2000
F th = 11 J/cm 2
-1
1500
a eff = 1.28 x 10 4 cm
“Strong”
Ablation
1000
“Weak”
500
Ablation
F th = 5.5 J/cm 2
a eff = 4.24x10 4 cm -1
0
10
100
Etching Rate (nm/shot)

Fluence (J/cm 2 )

Figure 2: Excised hole depth as a function of accumulated laser fluence (i.e., number of laser pulses) for single- pulse fluences of 9.6 and 31 J/cm 2 . In this ‘gentle’ ablation phase, the hole depth advanced linearly with the number of laser pulses. Incubation effects that superlinearly raise etch rates, associated with the onset of the ‘strong’ ablation phase, are anticipated at higher accumulated fluence. The apparent saturation of ablation depth at ~2700 nm and 2200 nm, respectively, is an artifact due to the limited aspect- ratio of the AFM scanning tip.

Figure 3: Etching depth per pulse in fused silica as a logarithmic function of laser fluence. Two regimes, gentle and strong ablation, are identified. Representation of the data (solid lines) by (1/a eff ) log (F/F th ) provide values for threshold fluence and the effective absorption coefficient in each regime.

Single-pulse etch rates were collected from the slopes of data in graphs like Figure 2 and plotted in Figure 3 as a function of single-pulse fluence. The etch-rate data follow a logarithmic fluence-dependence from an extrapolated ablation threshold of 5.5 J/cm 2 to ~44 J/cm 2 , the onset of strong ablation. This fluence window (5.5 to 45 J/cm 2 ) defines the gentle-ablation processing window for controllable etching of smooth features in fused silica. Thin layers, ~100-nm deep or less, could be precisely excised with appropriate choice of fluence. The logarithmic fluence dependence — normally associated with single-photon absorption mechanisms — is surprising here, considering the nonlinear mechanisms that are understood to drive absorption in this transparent material. Kautek et al. 11 have also reported a logarithmic fluence dependence for 20-fs to 3-ps ablation of barium aluminum borosilicate glass. For 1.2-ps ablation of fused silica, the slope of the solid curve in Figure 3 (for F < 44 J/cm 2 ) provides an effective penetration depth of 1/a eff = 235 nm, a value commensurate with the ~100- nm layer-by-layer resolution cited above. The 5.5 J/cm 2 threshold fluence is in accord with the damage threshold of 5±1 J/cm 2 reported by Varel et al. 9 for 1.0-ps ablation of fused silica. This group also reports, 13 in a later paper, etch rates of ~200 nm/pulse with 3.2-ps pulses at 10 J/cm 2 fluence. This etch rate is only slightly larger than our 180 nm/pulse value from Figure 3 for 1.2-ps pulses. Note again that the rates in Figure 3 are only valid where the number of laser pulses is small. The onset of a strong ablation phase after 10’s or 100’s of laser pulses explains the 2 or 3-fold faster etch rates reported by Varel et al. 12 for ablation of deep channels in fused silica.

A

higher fluence, F > 44 J/cm 2 in Figure 3, etch rates abruptly rise to values 2- or 3-fold faster than by simple extrapolation

of

the gentle ablation data. This enhanced rate is related to the incubation phenomenon described previously where now a

single pulse provides sufficient fluence to fully incubate the underlying glass material. The effective penetration depth rises

151

to 1/a eff = 780 nm, supporting rapid etch rates of up to 2-mm per pulse at a fluence of ~150 J/cm 2 . While rapid etch rates are attractive for many applications, this strong-ablation regime provides less control over etch depth as evidenced by the wider scatter of data points in Figure 3. Another disadvantage was the development of microcracks following 2 or 3 ablation pulses at high fluence.

Figure 4 shows the rapid development of shock-induced microcracks forming around the perimeter of laser-ablated holes. At the 140 J/cm 2 fluence, microcracks and surface swelling (noted by AFM) developed very quickly — by the third laser pulse — posing a significant limitation to precise shaping of smooth optical surfaces. At lower fluence, microcracking developed more slowly. Over the 5.5 to 170 J/cm 2 fluence-window studied here, these undesirable surface features appeared consistently after an onset number of laser pulses, N c , that approximately followed

N c = 1.7 + 80/F

(F in J/cm 2 ).

(1)

Because these N c values are small, peaking at ~25 near the threshold fluence for the gentle ablation region, there was no point in extending studies to integrate large numbers of pulses (N > 60). Therefore, the transition from gentle ablation to strong ablation with increasing N was not observed here, preceded by the early development of microcracks — our main limitation to smooth surface-structuring of fused silica. Varel et al. 12 also reported the formation of microcracks around deep (~1 mm) channels etched in fused silica by hundreds of laser pulses of 100-fs to 30-ps duration. Their study showed a favorable trend of reduced microcracking with decreasing pulse duration.

12.8 mm
12.8 mm
12.8 mm
12.8 mm
12.8 mm
12.8 mm

Figure 4: Optical microscope photographs of fused silica ablated by 1.2-ps Nd:glass pulses at 140 J/cm 2 . From left to right, holes were drilled by one, two, three, four, and five pulses. Shock-induced microcracks developed quickly, by the third pulse, for this large fluence.

Combination of N c in Equation 1 with the per-pulse-etch rates in Figure 2 provides a coarse guide to the maximum ablation depth one can attain without deleterious microcracking or surface swelling phenomena. The calculated maximum ablation depth is plotted in Figure 5 as a function of the single-pulse laser fluence. Excellent prospects for controllable shaping of smooth 3-dimensional surface structures is found by restricting etch depth to values below the curve in this figure. Therefore, structures up to 2-mm deep with 10% rms surface roughness are shown here to be possible, establishing a practical processing window for picosecond ablation of fused silica.

4. PULSE-TRAIN MICROMACHINING

The full flat-topped oscillator pulse train — ~400 pulses of 1.2-ps duration at ~100 J/cm 2 fluence each — was applied to fused silica surfaces. The optical microscope photo in Figure 6 shows a high aspect-ratio hole formed as a result of this single pulse train. A smooth symmetric hole of 8-mm diameter was excised to 30-mm depth (determined with optical microscopy) without inducing fractures, cracks, or swelling in the hole periphery. In comparison, the 1-Hz repetition rate applied in Section 3 caused microcracks after only 3 pulses for the same single-pulse fluence. Cumulative heating effects associated with the 133-MHz pulse-repetition rate in the pulse train are likely to improve the ductility of the surrounding glass, thereby mitigating the shock-induced microcracking in regions immediately surrounding the hole perimeter. Heat incompletely dissipated on a nanosecond time -scale may explain the expansion of hole diameters to 8-10 mm which is ~5¥ larger than the diameter of the focussed laser beam. In the operation of the feedback-controlled modelocked oscillator, the first dozen pulses of the train are also somewhat longer-duration (~10 ps); possibly this may also have an effect.

152

SEM examination showed the hole walls to be smooth from top to bottom except for a very small region ~1-mm 2 centered at the very bottom of the hole. An example of this damaged region is shown in Figure 7. The ~130-nm surface ripples are not related to LIPSS which typically generates structures the size of the laser wavelength (1.054 mm). Kruger et al. have seen similar 200-nm features in fused silica with 300-fs pulses 10 and in barium-aluminum borosilicate glass with 20-fs pulses. 16 They proposed surface plastification and melting followed by relaxation of internal mechanical stress as the ripple-forming mechanism. 16 The presence or absence of these surface structures depend on the number of pulses and the laser fluence and duration, 16 and may therefore be controlled or eliminated for multi-pulse ablation with slightly different laser conditions.

5 4 "Strong" Ablation 3 2 1 "Gentle" Ablation Fused Silica 0 0 20 40
5
4
"Strong" Ablation
3
2
1
"Gentle" Ablation
Fused Silica
0
0
20
40
60
80
100
120
140
Maximum Depth (mm)
mm)

Single-Pulse Fluence (J/cm 2 )

10 mm
10 mm

Figure 5: Maximum ablation depth as a function of single-pulse fluence, defined by the onset of microcracking and surface swelling. Picosecond laser pulses can therefore precisely shape smooth, 3- dimensional surface structures to vertical dimensions of ~2 mm (gentle ablation regime)

Figure 6: Nomarski micrograph of hole excised in fused silica by a mode-locked pulse train consisting of ~400 single 1.2-ps laser pulses with a pulse-to-pulse separation of 7.5-ns. The ~30-mm deep hole has smooth walls and shows no evidence of fractures, cracks, or surface swelling in the periphery. The total (integrated) fluence was ~100 kJ/cm 2 .

Multiple pulse trains were also applied to fused silica at 1-Hz repetition rate, resulting in a slight increase (~20%) in hole diameter and only a moderate advance in etch depth. Hole depths saturated quickly for 100-J/cm 2 single-pulse fluence. Varel et al. 12 also noted a saturation of hole depth when drilling deep channels in fused silica with femtosecond and picosecond laser pulses at 10-1000 Hz repetition rate. Surprisingly, the application of many pulse trains never led to the formation of microcracks, fractures, or swelling for any samples in our work. High repetition-rate multi-pulse ablation is clearly a promising new option for controlling the micromachining quality of brittle materials. The 7.5-ns pulse-to-pulse separation is sufficiently short to reduce the material cooling between laser pulses, thereby presenting a heated and ductile glass to succeeding laser pulses. This pulse separation is also sufficiently long to permit hydrodynamic expansion of the laser-produced plasma plume, reducing or eliminating obscuration of subsequent laser pulses. Further study of multi-pulse ablation is currently underway in our laboratory.

153

5. CONCLUSIONS Figure 7: High-magnification SEM of the hole in Figure 6 showing a ~130-nm

5. CONCLUSIONS

Figure 7: High-magnification SEM of the hole in Figure 6 showing a ~130-nm period ripple structure. Such structures were confined only to a ~1-mm 2 area at the bottom of hole. SEM showed that all other surfaces were smooth and free of fractures and ablation debris.

Ultrafast laser ablation of fused silica has been investigated for its prospects in shaping smooth photonic components. For 1.2-ps laser ablation, smooth holes with diameters as small as 1-mm were excised in a controlled and reproducible manner. A processing window free of microfractures, rippling phenomena, and swelling was defined for shaping 3-dimensional surface structures to vertical dimensions of 2-mm. The gentle ablation regime — fluences between 5.5 and 44 J/cm 2 — yielded a 10% (rms) surface roughness which is small enough for producing optical devices.

Further, our group has discovered an alternative approach to controlling surface microcracking with ultrafast-laser micromachining of fused silica. By applying the full oscillator pulse train — 430 pulses of 1.2-ps duration at ~100 J/cm 2 fluence each — smooth symmetric holes of 8-mm diameter and up to 30-mm deep were generated by one pulse train. No evidence of surface microcracking or swelling was found. Heating effects associated with the 133 MHz repetition rate may improve the ductility of the surrounding glass. Multi-pulse etching is therefore attractive as a new option to eliminate undesirable microcracking phenomena while micromachining glass and other brittle materials.

These results together emphasize that useful control of laser-micromachining in optically transparent materials depends greatly on the details of delivery of a given fluence — an observation which broadly motivates ultrafast laser-processing.

ACKNOWLEDGEMENTS

Support from Photonics Research Ontario and the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged.

REFERENCES

1 S. Kuper & Stuke, Appl. Phys. B 44, 199 (1987); S. Press & M. Stuke, Appl. Phys. Lett. 67, 338 (1995).

2 A.C. Tam, J.L. Brand, D.C. Cheng, and W. Zapka, Appl. Phys. Lett. 55, 2045 (1989).

3 J. Ihlemann, Appl. Surf. Sci. 54, 193 (1992).

4 J. Ihlemann, B. Wolff, and P. Simon, Appl. Phys. A 54, 363 (1992).

5 D. Du, X. Liu, G. Korn, J. Squier, & G. Mourou, Appl. Phys. Lett. 64, 3071 (1994).

6 P.P. Pronko, S.K. Dutta, J. Squier, J.V. Rudd, D. Du, & G. Mourou, Optics Comm. 114, 106 (1995).

154

7 B.C. Stuart, M.D. Feit, A.M. Rubenchik, B.W. Shore, and M.D. Perry, Phys. Rev. Lett. 74, 2248 (1995).

8 B.C. Stuart, M.D. Feit, S. Herman, A.M. Rubenchik, B.W. Shore, and M.D. Perry, J. Opt. Soc. Am. B 13, 459 (1996).

9 H. Varel, D. Ashkenasi, A. Rosenfeld, R. Herrmann, F. Noack, E.E.B. Campbell, Appl. Phys. A 62, 293 (1996).

10 J. Kruger and W. Kautek, Appl. Surf. Sci. 96-98, 430 (1996).

11 W. Kautek, J. Kruger, M. Lenzner, S. Sartania, C. Spielmann, and F. Krausz, Appl. Phys. Lett. 69, 3146 (1996).

12 H. Varel, D. Ashkenasi, A. Rosenfeld, M. Whamer, E.E.B. Campbell, Appl. Phys. A 65, 367 (1997).

13 D. Ashkenasi, H. Vare l, A. Rosenfeld, F. Noack, and E.E.B. Campbell, Nucl. Instr. & Meth. in Phys. Res. B 122, 359

(1997).

14 D. Ashkenasi, A. Rosenfeld, H. Varel, M. Wahmer, E.E.B. Campbell, Appl. Surf. Sci. 120, 65 (1997).

15 S. Ameer-Beg, W. Perrie, S. Rathbone, J. Wright, W . Weaver, and H. Champoux, Appl. Surf. Sci. 127-129, 875 (1998).

16 J. Kruger, W. Kautek, M. Lenzner, S. Sartania, C. Spielmann, and F. Krausz, Appl. Surf. Sci. 127-129, 892 (1998).

17 J. –X Zhao, B. Huttner, and A. Menschig, Micromachining with ultrashort laser pulses, appearing in SPIE Proc. 3616

(1999).

18 K.M. Davis, K. Miura, N. Sugimoto, and K. Hirao, Opt. Lett. 21, 1729 (1996).

19 E. N. Glezer, M. Milosavljevic, L. Huang, R. J. Finlay, T.-H. Her, J. P. Callan, and E. Mazur, Opt. Lett. 21, 2023 (1996).

20 D. Ahkenasi, H. Varel, A. Rosenfeld, S. Henz, J. Hermann, and E.E.B. Campbell, Appl. Phys. Lett. 72, 1442 (1998).

21 P.R. Herman, et al., Advanced-laser processing of photonic and microelectronic components at Photonics Research Ontario, appearing in SPIE Proc. 3618 (1999).

22 R.S. Marjoribanks, F.W. Budnik, L. Zhao, G. Kulcsár, M. Stanier, & J. Mihaychuk, Optics Lett. 18, 361 (1993).

155