Volumetric modi cations in fused silica using Gaussian and ... · Volumetric modi cations in fused...

Volumetric modifications in fused silica using Gaussian andBessel femtosecond laser beams

Domas Paipulasa, Mindaugas Mikutisa,b, Valdas Sirutkaitisa, and Saulius Juodkazisc

a Vilniaus University, Laser Research Center, Saultekio av. 10, LT-10223 VilniusbAltechna RD, Mokslininku st. 6a, LT-08412, Vilnius

c Centre for Micro-Photonics, Faculty of Engineering and Industrial Sciences SwinburneUniversity of Technology, Hawthorn, Vic 3122, Australia

ABSTRACT

In this work we present the results on volumetric fused silica modifications using femtosecond Gaussian andGaussian-Bessel laser beams. We show that for specific photonic device, like volume Bragg grating, fabrication,Bessel beams are more superior to Gaussian, as the recording process is much faster and fabricated devices havebetter efficiencies. Also Gaussian beam tend to be more efficient in formation of nanogratings that causes theappearance of birefringence in modified zones. This reduces optical quality of fabricated device and limits overallrecording velocity. We have successfully fabricated volume Bragg gratings in the bulk of fused silica that hadabsolute diffraction efficiencies reaching ∼90% using femtosecond Gaussian-Bessel beam, while gratings madewith Gaussian beam reached only 60%.

Keywords: Volume gratings, Femtosecond processing, Integrated optics

1. INTRODUCTION

Transparent material modification with ultrafast laser pulses has attracted great scientific and industrial interestin the recent decade. If femtosecond laser pulse intensities are kept slightly above materials damage threshold,it is possible to induce well-defined regions with altered optical properties. Most common optical changes arevariations in refractive index,1 induction of birefringence,2 or appearance of light scattering microcracks orvoids.3,4 This phenomenon opens doors for three-dimensional integration of various photonic devices in the bulkof transparent material using direct laser writing technique.

However, for some particular integration tasks, common direct laser writing method using Gaussian laserbeam is not efficient. In order to achieve material modification in large band gap materials, such as fused silica,strong focusing resulting in small focal spot size is required. This complicates and prolongs recording of large-size photonic devices, such as thick volume Bragg gratings and etc. In order to overcame this problem somelevel of beam shaping is required. Femtosecond filamentation can be utilized for similar device inscription,5

however, this process is rather nondeterministic and limits optimization of essential recording parameters. Herwe study the use of Gaussian-Bessel (GB) laser beam for photonic device inscription. We show that highlyefficient (absolute diffraction efficiency ∼ 90%) volume Bragg gratings can be formed in pure fused silica withoutany post-processing or thermal treatment using femtosecond GB laser pulses.

2. EXPERIMENTAL SETUP

Direct laser writing in the bulk of fused silica was carried out with Yb:KGW laser (”Pharos”, Light ConversionLtd.) having central wavelength of 1030 nm, pulse duration of 170 fs (FWHM) and pulse repetition rate of200 kHz. Laser beam was guided through automated attenuator and polarization controller and focused with0.42 numerical aperture (NA), long working distance objective (”Mitutoyo Plan Apo NIR 50X”) in the bulk ofthe sample. Fused silica glass sample (”Lithosil”, Schott) with refraction index of 1.46, was mounted on highprecision 3D positioning stage (Aerotech) and translated with respect to laser beam. Stage traveling speeds were

Further author information: (Send correspondence to D.P.): E-mail: [email protected]

Pacific Rim Laser Damage 2013: Optical Materials for High Power Lasers, edited by Jianda Shao, Takahisa Jitsuno, Wolfgang Rudolph, Proc. of SPIE Vol. 8786,

87860D · © 2013 SPIE · CCC code: 0277-786X/13/$18 · doi: 10.1117/12.2020258

Proc. of SPIE Vol. 8786 87860D-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/11/2013 Terms of Use: http://spiedl.org/terms

FOCUSING OPTICS

3D POSITIONING STAGEPOLARIZATION CONTROL

FEMTOSECOND LASER "PHAROS"

axicon Lens, f1

t1M

Si02

260

240

220

200

180

160

; 140m

120

100

c 80

6o

Lens, f2

3D stage

40

20

0

-8

Axial (z) coordinate, utn20 40 60 80 100

-6 -4 -2 0 2 4 6 8

Laterial (x) coordinate um

a) b)

in range from 1 to 160 mm/s. However, the recording was aimed at the maximum speed which was optimizedfor stage acceleration, deceleration and recorded grating dimensions. Recording sequence was fully automated(software ”SCA”, Workshop of Photonics) and fabrication process was monitored in real time with imagingoptics and CCD camera. Schematic representation of the experiment can be seen in Fig. 1.

In order to create Bessel type beam, fused silica axicon (apex angle – 179◦) was inserted before the 0.42 NAobjective. Additional plano-convex lens was used in order to image ”non-diffracted zone” through objective, thusforming 4-f type afocal optical system with 75x demagnification (Fig. 2a). This resulted in GB type beam thatis shown in Fig. 2b. Also, in GB experiments we used second harmonics of Yb:KGW laser, thus wavelength was515 nm.

Figure 1. A sketch of femtosecond direct laser writing system.

Figure 2. (a) Schematics of experimental setup for Gaussian beam transformation to GB. Axicon is used together withdemagnifying telescope (4-f system) to reduce original GB beam size by 75 times. (b) Measured axial and lateral intensitydistribution of GB after the telescope, inset shows GB beam profile, scale bar is 5 µm.

For Gaussian beam the spot diameter and confocal length at the focus can be expressed by: 2w0 = 4λfπDM

2

and 2zR =πw2

0

2λ , where w0 and zR are beam waist radius and Rayleigh length, D is beam diameter beforeobjective, f is focal length of objective and M2 = 1.2 is the quality factor of the used laser beam. For our



Translationspeed

10 mm/s

5 mm/s

1 mm/s

0.1 mm/s

0.05 mm/s

0.02 mm/s

0.01 mm/s

Pulse energy 200 nJ 180 nJ 160 nJ 140 nJ 120 nJ

E

411111r ears 1111

1W.

TOP VIEW AXIAL VIEW

MED

10 µm

objective and laser focal spot was 2w0 = 1.7 µm and 2zR = 4.5 µm.

Similar evaluation can be carried out with GB laser beam. For GB beam central core radii (ρ0), and theaxial extent of the ”non-diffracting” zone (zmax) can be expressed by following formulas:6

ρ0 =1.2024λ

π sin(α0), zmax =

w0 cos(α0)

sin(α0), (1)

where α0 = α(nax − n0)/n0 with n0 and nax being refractive indexes of the ambient and axicon, respectively;and α is axicon’s half-angle measured in respect of its base (in our case 0.5◦). One would find that for ideal 1/75demagnification ρ0 = 0.66 µm and zmax = 116 µm (see, Fig. 2b). As can be seen, the axial extend of GB beamis by an order larger than Gaussian beam while waist diameter remains almost the same. It is noteworthy thatthe Eq. 1 are valid only for an ideal axicon.

3. MODIFICATIONS IN FUSED SILICA

Typical modifications created at various writing conditions with Gaussian laser beam (by inscribing single line100 µm bellow sample surface) are shown in Fig. 3. As can be seen from this picture, all modifications have ellip-tical cross sections with maximum axial length not exceeding 12 µm at highest pulse energies, while lateral sizeis ∼ 2 µm. Such shape is common for Gaussian beam recorded modifications at perpendicular-to-beam sampletranslation direction. If needed, this elliptical shape can be corrected using various beam shaping techniques, assuccessfully demonstrated in waveguide integration applications.7

Figure 3. Modifications induced in the volume of fused silica sample with Gaussian laser beam. Top and axial views areshown for separate lines, made at various pulse energies and sample translation speeds. Beam was focused with 0.42numerical aperture lens 100 µm below the surface. Laser repetition rate - 100 kHz.

Other interesting feature of Gaussian-beam inscribed modifications is the fractalization of lines when sampletranslation speed is increased or pulse energy is decreased. It is natural to assume, that if sample translationspeed is increased to the levels when pulses cease to overlap on the sample, resulted line would be renderedinhomogeneous. However, in our recording conditions (rep. rate 200 kHz) pulses stop to overlap at speedsexceeding 200 mm/s, but fractalization becomes apparent at speeds as low as 1 mm/s. Such behavior is causedby the non-regular formation of nanogratings.

Nanogratings self-form in the laser affected zones after multipulse exposition of the same sample area or whenlaser pulse duration is sufficiently long (> 300 fs).8,9 If multipulse accumulation is not sufficient, nanogratingformation becomes non regular and causes apparent fractalization in laser-recorded lines. This fractalization alsois a source of strong light scattering. Thus for homogeneous line recording sample translation speed should bekept sufficiently low (∼ 100 shots per area in focus, which result in < 2 mm/s, at our recording conditions). Lines



; 'r

11

r1 3

'

0,33 pi. 0,45 p,J 0,61 pj 0,80 pJ 1,02 pJ 1,28 1.1J L56 p,J 1,84 1.1.J 2.14 1.1..T 2.48 t..T

recorded in these conditions have much lower scattering, but all posses strong birefringence, observed throughcrossed polarizers.

In contrast, modifications recorded with GB beam have quite different look. In Fig. 4 it is shown cross sectionsand top views of similar lines recorded at various recording conditions. Lateral diameter of such modification

Figure 4. Modification tracks created with GB beam in fused silica at various pulse energies. At each energy value sixseparate tracks are shown that were created at different pulse density values: (from left to right) 625, 1250; 1875, 2500,3125, 3750 pulses/mm. a) Top view and b) axial view are shown. Scale bar is 10 µm.

was not larger than in Gaussian beam case; however, axial length increased tenfold and reached around 80-100 µm after single shot. No significant dependence of modification’s axial length on laser pulse energy wasobserved, implying that axial length is determined only by the choice of axicon (and/or additional focusingoptics) according to Eq.1. However, as in Gaussian case, multipulse accumulation plays significant factor inmodified line (or in this case – plane) quality. As can be seen from Fig. 4, if pulse density is higher than 2000pulse/mm (meaning that each modified area is affected with at least 2 pulses), modification track consist ofrandomly spaced microcracks producing strong scattering. When there is no pulse accumulation, modificationbear smooth and regular appearance.

In the picture it is also visible that typical modified track (for example 1.02 µJ at 625 pulses/mm) is alsonot homogeneous: there is a prominent dark central region surrounded by bright zones at the both sides ofthe modified track. Bright regions could be linked with the areas where material refractive index is increased(so called Type I modification), where dark regions are linked to the formation of scattering centers in fusedsilica (Type II modifications). Indeed, these two types of laser-induced modifications are well known to exist infused silica, which development depend on the laser pulse intensity.9 By looking at the measured axial energydistribution of GB beam (Fig. 2b) it is clear that central part has twice as large intensity than the slopes, thusexplaining the appearance of two different types of modification.

Also filamentation of GB beam should have to be avoided during the recording as it is known to producescattering bead-like damage tracks along pulse propagation path.10 Typical example of such damage track isshown Fig. 5.

4. RECORDING OF VOLUME BRAGG GRATINGS

Volume Bragg gratings (VBG) are particular type of phase-gratings embedded into the volume of transparentmaterial. In comparison with other diffraction gratings, VBG have additional dimension of depth (thickness),that can rise diffraction efficiencies even up to theoretical 100%. Having this property, VBG are widely usedin many photonic applications.11 As grating depth/period ratio have to be considerably large, surface Bragggratings are uncommon, thus traditional grating recording technique relay on photosensivity of the transparentmaterial where interference pattern from UW cw-laser emission12 or ultrashort laser pulses13,14 are recordeddirectly in the volume of the material. So far hardened gelatin polymer, photo-thermo-refractive and varioustypes of UV-sensitive phosphate glasses were used as a medium for VBG recording.11,12,15 Femtosecond directlaser writing expanded the range of suitable materials where such devices can be recorded.



too

80

60

40

20

n,-On = 0.02- -An = 0.01-An = 0.005- - -An = 0.001

8020 40 60

Grating thickness, µm1010

Figure 5. An example of bead-like damage appearance along the beam path when GB pulse power is kept above filamen-tation regime.

Diffraction properties of VBG’s are analyzed by coupled wave model.16 According to this model, VBG canreach the highest efficiency operating only at the Bragg diffraction conditions, that can be expressed as:

mλ/n = 2d sin(θB); (2)

where θB is a so called Bragg angle, d is a period of the grating, λ the wavelength of diffracted beam, and n isrefractive index of the material where diffraction takes place (see Fig. 6). It is easy to see, that Bragg conditionis validated when incident and diffracted angles are equal in absolute values, so angle |αinc| = |αdif | = θB .

Figure 6. The dependence of VBG diffraction efficiency on grating thickness for various refractive index modulation levels(left). A sketch of VBG depicting incident (Iinc), diffracted (Idif ) and nondiffracted beams (Itr) as well as main gratingparameters: incident (αinc) and diffracted angles (αinc), grating period (d), and grating thickness (t) (right).

Diffraction efficiency is defined as ratio between diffracted and incident beam powers and can be expressedby the following formula:

η = sin2

[π∆nt

λ cos(θB)

], (3)

here ∆n is a magnitude of homogeneously modulated refractive index, t is the grating axial thickness. It easyto see that the product ∆nt determines VBG efficiency for single wavelength at Bragg condition (Fig. 6). Aslaser-induced variations of ∆n in non-photosensitive material is somehow limited, grating depth can be easilychanged by using multiplexing procedure: stitching several layers of gratings on top of each other using directlaser writing technique. Typical laser induced refractive index variations are in range of 0.001 – 0.005, thus inorder to fabricate efficient grating one needs to create grating with thickness of order 80 µm - 400 µm.

If we look at Fig. 7, the advantage of GB beams for VBG fabrication becomes immediately apparent. Usingjust Gaussian beam will need to fabricate many layers of gratings on top of each other in order to achievesufficient grating thickness (Fig. 7a), while with GB beam single layer already results in 90 µm thick grating.



NA= 0.42

1030 nm,300 fi

#.r

O 13D N

SNING 2STAGE

Gaussian- Besscl

Mill

ANGPS

G STAGE

b)

1.5 pm

90 pm

Figure 7. DLW approach for formation of thick volumetric Bragg gratings for a) Gaussian and b) Gaussian-Bessel laserbeams.

Table 1. VBG recording comparison between Gaussian and GB beams

Gaussian beam Gaussian-Bessel beam

Period d (µm) 2 1.5Thickness t (/micro m) 90 350Total layer number 20 4Sample translation speed (@100 kHz) (mm/s) < 1 mm/s 20 mm/s (not limited)Grating size (mm×mm) 1 × 1 6 × 6Recording time (h) >3 <1Diffraction efficiency at Bragg angle (@633 nm) 60%(p)/29%(s) 89%

Diffraction efficiencies of all VBG’s demonstrated in this work were measured using HeNe laser (633 nmwavelength) at Bragg condition. We estimated efficiency by measuring the total power of diffracted beam andcompared it to the beam that freely passes through sample when grating is moved out of the beams path.In such configuration absolute efficiency was evaluated and glass absorption and Fresnel reflection loses wereautomatically eliminated.

The highest achieved diffraction efficiency recorded with GB beam was reached at 633 nm wavelength andwas 89%. VBG thickness was 352 µm (formed by 4 layers of 88 µm each) and period – 1.5 µm, hence gratingdepth/period ratio was 234. Its lateral size was 6×6 mm2, which took less than an hour to record with ultra-shortGB laser pulses. The smallest period gratings recorded by axial multiplexing of N = 5 layers with GB beamswas 1 µm.

Gratings made with Gaussian laser beam also show relatively high diffraction efficiencies (∼ 60%), however,stronger focusing leads to the formation of Type II modifications with birefringence. Thus diffraction efficiencybecomes polarization-dependent. VBGs made from Type II modification can have smaller depths as inducedrefractive index variation is three times larger if compared to GB case; however this does not help to reducefabrication time as small modification’s axial length requires higher level of multiplexing (N > 20).

Comparison between main recording parameters is given in a Table 1. As can be seen, throughput of VBGrecording with GB beams is increased by 40 times, and diffraction efficiency of such gratings are better.

In order to evaluate spectral bandwidth, White Light Continuum (WLC) was shone on the VBGs andspectrum of transmitted (undiffracted) beam was measured in order to evaluate which spectra components were



10 ,. , ra) 0,9;

0,8

S#I-7.2*-7.8°

8.4°-9°-9.6°

CO 0,7,0,6

0,4450

- 10.2°

i i i i i

500 550 600 650 700 750 800 850

0,5

0,4

v 0,3

a°

0,2

0,1

0,0460

;

i

Iv i

ilá

III lU;,l¡

I I1III II ll

IIIiil; ulIIÌÌilil,

ilI IIIli iIIII

111 1I' I II;II

;II1I I I Ii I

:'!

11 14:I

II

I IIrI ':'Ii

I I

I il l! ; S IT; j 1 I 1 i

41 ¡

t j

:::

"I

":

, I ,;

;1'1 ;i , I

I

I

1

)

I

I1

,,,

,

!

- -7,65°- -7,95°- -8,25°- -8,55°- -8,85°- -9,15°- -9,50°- -960°i. , . , ,

480 500 520 540 560 580 600 620 640

diffracted, thus absent, from WLC spectrum.These results are shown in Fig. 8. Spectral bandwidth is inversely

Figure 8. Spectra of normalized white light continuum as observed after the VBG, tilted at various Bragg angles (a) forgratings made with Gaussian beam, b) for GB beam).

proportion to grating thickness. As gratings made with Gaussian beam are thiner, its spectral bandwidth (∆λ) islarge and could reach up to 150 nm. In comparison, thick gratings made with GB beam has spectral bandwidth<10 nm.

5. CONCLUSION

In conclusion we demonstrate that femtosecond Gaussian –Bessel laser pulses can be used for recording extendedtraces of optical modified material inside transparent dielectrics that could increase the range of possible appli-cations for laser fabrication. Using this technique we fabricated 6 × 6 mm2 footprint volume Bragg grating with1.5 µm period having diffraction efficiencies of 89% in an hour.

ACKNOWLEDGMENTS

This work was supported by research grant No. VP1-3.1-SMM-10-V-02-007 (Development and utilization of anew generation industrial laser material processing using ultrashort pulse lasers for industrial applications) fromthe European Social Fund Agency.

REFERENCES

[1] Davis, K. M., Miura, K., Sugimoto, N., and Hirao, K., “Writing waveguides in glass with a femtosecondlaser,” Opt. Lett. 21, 1729–1731 (NOV 1 1996).

[2] Shimotsuma, Y., Kazansky, P. G., Qiu, J., and Hirao, K., “Self-organized nanogratings in glass irradiatedby ultrashort light pulses,” Phys. Rev. Lett. 91, 247405(1–4) (Dec 2003).

[3] Glezer, E. N., Milosavljevic, M., Huang, L., Finlay, R. J., Her, T.-H., Callan, J. P., and Mazur, E., “Three-dimensional optical storage inside transparent materials,” Opt. Lett. 21, 2023–2025 (Dec 1996).

[4] Hashimoto, T., Juodkazis, S., and Misawa, H., “Void recording in silica,” Appl. Phys. A 83, 337–340 (2006).

[5] Yamada, K., Watanabe, W., Kintaka, K., Nishii, J., and Itoh, K., “Volume grating induced by a self-trappedlong filament of femtosecond laser pulses in silica glass,” Jpn. J. Appl. Phys. 42(11), 6916–6919 (2003).

[6] Brzobohaty, O., Cizmar, T., and Zemanek, P., “High quality quasi-Bessel beam generated by round-tipaxicon,” Opt. Express 16(17), 12688–12700 (2008).

[7] Ams, M., Marshall, G., Spence, D., and Withford, M., “Slit beam shaping method for femtosecond laserdirect-write fabrication of symmetric waveguides in bulk glasses,” Opt. Express 13, 5676–5681 (Jul 2005).



[8] Richter, S., Heinrich, M., Doring, S., Tunnermann, A., and Nolte, S., “Formation of femtosecond laser-induced nanogratings at high repetition rates,” Appl. Phys. A 104(2), 503–507 (2011).

[9] Hnatovsky, C., Taylor, R. S., Rajeev, P. P., Simova, E., Bhard-waj, V. R., Rayner, D. M., and Corkum,P. B., “Pulse duration dependence of femtosecond-laser-fabricated nanogratings in fused silica,” Appl.Phys.Lett. 87(17), 014104 (2005).

[10] Gaizauskas, E., Vanagas, E., Jarutis, V., Juodkazis, S., Mizeikis, V., and Misawa, H., “Discrete damagetraces from filamentation of Bessel-Gauss pulses,” Opt. Lett. 31(1), 80–82 (2006).

[11] Arns, J., Colburn, W., and Barden, S., “Volume phase gratings for spectroscopy, ultrafast laser compressors,and wavelength division multiplexing,” Proc. of SPIE 3779, 313–323 (1999).

[12] Efimov, O. M., Glebov, L. B., and Smirnov, V. I., “High-frequency Bragg gratings in a photothermorefractiveglass,” Opt. Lett. 25(23), 1693–1695 (2000).

[13] Toma, T., Furuya, Y., Watanabe, W., Itoh, K., Nishii, J., and Hayashi, K., “Estimation of the refractiveindex change in glass induced by femtosecond laser pulses,” Opt. Rev. 7, 14–17 (JAN-FEB 2000).

[14] Misawa, H., Juodkazis, S. Matsuo, S., and Kondo, T., “3D holographic recording method and 3D holographicrecording system,” US7542186 B2 patent (2009).

[15] Lumeau, J. and Glebov, L. B., “Modeling of the induced refractive index kinetics in photo-thermo-refractiveglass,” Opt. Mater. Express 3, 95–104 (Jan 2013).

[16] Kogelnik, H., “Coupled wave theory for thick hologram grating,” Bell Syst. Tech. J. 48, 2909–2947 (Mar1969).



Volumetric modi cations in fused silica using Gaussian and ... · Volumetric modi cations in fused...

Documents

Transcript of Volumetric modi cations in fused silica using Gaussian and ... · Volumetric modi cations in fused...