Download Materials processing with a tightly focused femtosecond laser vortex

October 15, 2010 / Vol. 35, No. 20 / OPTICS LETTERS
3417
Materials processing with a tightly focused
femtosecond laser vortex pulse
Cyril Hnatovsky,1,* Vladlen G. Shvedov,1,2 Wieslaw Krolikowski,1 and Andrei V. Rode1
1
Laser Physics Centre, Research School of Physics and Engineering, The Australian National University,
Canberra Australian Capital Territory 0200, Australia
2
Department of Physics, Taurida National University, Simferopol 95007 Crimea, Ukraine
*Corresponding author: [email protected]
Received June 30, 2010; revised September 2, 2010; accepted September 5, 2010;
posted September 20, 2010 (Doc. ID 130950); published October 8, 2010
In this Letter we present the first (to our knowledge) demonstration of material modification using tightly focused
single femtosecond laser vortex pulses. Double-charge femtosecond vortices were synthesized with a polarizationsingularity beam converter based on light propagation in a uniaxial anisotropic medium and then focused using
moderate- and high-NA optics (viz., NA ¼ 0:45 and 0.9) to ablate fused silica and soda-lime glass. By controlling the
pulse energy, we consistently machine micrometer-size ring-shaped structures with <100 nm uniform groove
thickness. © 2010 Optical Society of America
OCIS codes: 320.2250, 140.3390, 320.7160, 140.3440, 140.3300.
Femtosecond laser vortex pulses provide an opportunity
to investigate the effects of the optical angular momentum on photoionization processes [1] and also to obtain
information on how the ring-shaped intensity distribution
of vortex pulses affects the light-matter interaction.
Unfortunately, such studies pose a significant challenge
owing to technical difficulties hindering the synthesis of
high-power broadband laser vortex beams. The traditional methods utilizing spiral phase plates and holograms are inherently chromatic and therefore require
the introduction of correcting elements in order to compensate for the topological charge dispersion occurring
in polychromatic pulses. Uncompensated charge dispersion leads to a poor quality of the beam and the impossibility of using it for precision laser materials processing
[2]. Compensation schemes, however, come at the expense of making the beam-shaping setups significantly
more complicated and sensitive to alignment and remain
either bandwidth limited, or low throughput, or unsuitable for operation with high-intensity laser pulses ([3] and
references therein). So far, none of the proposed compensation techniques have been tested or used in laser
micromachining.
Recently we have synthesized high-quality doublecharge femtosecond vortex beams by using polarization
singularities associated with the beam propagation in
uniaxial birefringent crystals [4]. In this Letter we employ
this powerful technique to generate double-charge femtosecond laser vortex beams and use them for reproducible submicrometer structuring of fused silica (SiO2 )
and soda-lime glass samples. To the best of our knowledge, this is the first report on materials processing using
tightly focused femtosecond vortex pulses.
It was shown in [4] that a circularly polarized femtosecond Gaussian beam propagating along the optical axis of
a uniaxial crystal is converted into a superposition of two
polarization states with opposite handednesses. The state
with the handedness opposite to that of the input beam
carries a double-charge optical vortex (l ¼ 2), whereas
the state with the handedness coinciding with that of the
input beam represents a nonvortex beam. After the crystal, the wave propagating in free space becomes a super0146-9592/10/203417-03$15.00/0
position of independent solutions of the scalar wave
equation ð∇2⊥ þ 4πiλ−1 ∂z ÞΨ ¼ 0, and the vortex solution
can be written as Ψl¼2 ¼ Ψzþδ − Ψz−δ , where Ψzδ ¼
ðr 2 þ w20 ξzδ ÞGzδ expð2iφÞr −2 , Gzδ ¼ ðE=ξzδ Þ expð−r 2 =
ðw20 ξzδ ÞÞ, and ξzδ ¼ 1 þ iðz δÞλ=ðπw20 Þ. E is a constant
electric field amplitude, λ is the wavelength in vacuum, w0
is the 1=e2 beam waist radius, and δ denotes the axial shift
of the waists of the ordinary and extraordinary beams
with respect to each other due to double refraction in
the crystal. For paraxial beams, the shift is approximated
by δ ¼ dðn2o − n2e Þ=ð2n2e no Þ, where no and ne stand for the
ordinary and extraordinary refractive index, respectively,
and d is the thickness of the crystal along the beam
propagation direction z. The focusing of such beams
was studied theoretically in [5].
In our experiments we used the output beam of a
Clark-MXR femtosecond Ti:sapphire amplifier with a
central wavelength at λ ¼ 775 nm. Figure 1(a) shows
the experimental setup for the generation of doublecharge femtosecond vortex beams (see [3]). For consistency, the topological charge in our experiments was
l ¼ þ2. The cross section of the generated vortex beam
consists of a series of concentric rings, as shown in the
inset of Fig. 1(a). The number of rings is determined by
the input beam diameter at L1, the optical power of L1,
and the crystal parameters [3]. In our experiments the
beam expander, comprising L1 and L2, was adjusted
to allow only the first bright ring of the vortex beam
to enter the objective O1 (i.e., the first dark ring coincides
with the entrance aperture of O1) to be used for material
modification.
To compare the simulated and actual behavior of a
femtosecond vortex beam, we mapped the intensity distribution in the focal region of O1 with an imaging objective O2 having a NA significantly higher than that of O1,
viz., NA ¼ 0:9 (Nikon M Plan 100×) versus NA ¼ 0:45
(Olympus LUCPlan FLN 20×) [see Fig. 1(b)]. The images
were captured with a CCD camera and further analyzed
using ImageJ software. The analysis shows that the peakto-valley intensity variation along the maximum of the
first bright ring, which, according to our measurements,
has a diameter of ∼2 μm at the focus, does not exceed
© 2010 Optical Society of America
3418
OPTICS LETTERS / Vol. 35, No. 20 / October 15, 2010
15%. The average intensity along the first ring at the focus
(i.e., z ¼ 0) is ∼1:2 and ∼2 times higher than those measured at z ¼ 5 μm and z ¼ 10 μm, respectively. Based
on these numbers, the “confocal parameter” of the generated beam is ∼20 μm, which is ∼3 times larger than
that of a Gaussian beam with a 2 μm waist diameter.
The observed elongated tubular focus agrees well with
the simulations shown in Fig. 1(b), which were performed for O1 with NA ¼ 0:45 and λ ¼ 775 nm. Our simulations also show that the radial intensity distribution of a
vortex beam is well approximated by an expression describing the Laguerre–Gaussian pl beam (i.e., p zeros in
the radial direction and 2l angular nodes; l also determines the topological charge of a beam):
I¼
2jljþ1 r 2jlj expð−2r 2 =ðw0 jξjÞ2 Þ
P;
jlj!π
ðw0 jξjÞ2ðjljþ1Þ
where P is the total power and ξ ¼ 1 þ izλ=ðπw20 Þ.
According to this expression, the maximum intensity
Fig. 1. (a) Setup for the generation of double-charge femtosecond vortex pulses and materials processing with objective O1.
BF, beam filter viz., an aperture ∼1 mm in diameter located
∼5 m from the beam converter BC; λ=4, achromatic quarterwave plate; L1, negative lens (−50 mm); CR, 10-mm-long
c-cut CaCO3 crystal; L2, positive lens (þ125 mm); PBS,
polarization beam splitter; IB and VB, CCD images
(6:4 mm × 4:2 mm) of the incident and vortex beams, respectively; NB, nonvortex beam that is removed from the system;
S, glass sample. (b) Simulated (top) and experimental (bottom)
intensity distributions of a focused polarization-singularity vortex beam in the x–z and x–y planes. O2, imaging objective; CCD
images show the intensity distribution in the x–y plane at the
focus of O1 with NA ¼ 0:45.
I¼
2jljjlj expð−jljÞ
P
jlj!πðw0 jξjÞ2
is achieved at r 2l ¼ ðw0 jξjÞ2 jlj=2. For a vortex with l ¼ 2
and r 2 ¼ w0 jξj the peak intensity is by a factor of 2e−2
lower than for a Gaussian beam (i.e., l ¼ 0) with the same
1=e2 radius.
In experiments we explored two focusing regimes
using 200 fs (FWHM after BC) linearly polarized doublecharge vortex pulses. In the first case, O1 had a moderate
NA of 0.45 (Olympus LUCPlan FLN 20×). The beam was
focused onto the surface of a fused silica (i.e., SiO2 ) sample. Figure 2 shows topographic images of the ablation
rings produced with a single pulse at ∼150 nJ (after
O1), which were recorded with a Wyko NT9100 surface
profiler providing a 0:45 μm lateral and a subnanometer
vertical resolution. The depth of the ablation ring craters
is ∼40 nm, and the profile is uniform along the grooves.
The lateral dimensions of the craters agree with the intensity distribution in the focal region of O1 shown in
Fig. 1(b). The ∼0:2 μm3 volume of ablated silica agrees
with the amount of material removal per unit energy in
air of 1:5 μm3 =μJ reported in [6].
In the second set of experiments, samples of soda-lime
glass were irradiated using O1 with a high NA of 0.9
(Nikon M Plan 100×). After irradiation the samples were
examined under a Hitachi 4300SE/N scanning electron
microscope (SEM). The SEM images in Fig. 3(a) show the
ablation rings produced at different pulse energies. From
the leftmost image corresponding to the lowest energy,
one can see that (i) the diameter of the ablation ring
is ∼1 μm, which is half of the value obtained with
NA ¼ 0:45, and (ii) the width of the ablated annular
groove is less than 100 nm. According to measurements
performed with a Veeco MultiMode scanning probe microscope, the grooves are ∼90 nm deep, thus exhibiting
a ∼1:1 depth-to-height aspect ratio. The production of
such narrow and deep grooves is facilitated by the sharp
intensity distribution across the ring. At higher pulse
energies, the ablation signature is much more pronounced, but it still preserves an annular shape with
sharp and clean edges. It is also noteworthy that ablation
of material performed using a single vortex pulse is
highly reproducible [Fig. 3(c)].
Fig. 2. (Color online) Surface profile of a SiO2 sample ablated
with double-charge single ∼150 nJ femtosecond vortex pulses
using NA ¼ 0:45 focusing optic. Ablation rings are separated by
5 μm.
October 15, 2010 / Vol. 35, No. 20 / OPTICS LETTERS
Fig. 3. Ablation of soda-lime glass with double-charge single
femtosecond vortex pulses using an NA ¼ 0:9 focusing optic.
(a) Ablation rings produced at a pulse energy of ∼100, ∼150,
∼250, and ∼700 nJ from left to right. (b) Cavities produced
at a pulse energy of ∼1:5 μJ. (c) Region of a 200 μm ×
200 μm array produced at a pulse energy of ∼250 nJ. Ablation
rings are separated by 5 μm.
An interesting effect is observed when the pulse energy is increased to ∼1:5 μJ and the laser focus is positioned ∼2 μm subsurface [Fig. 3(b)]. In this case we
clearly observe the formation of deep cavities on the axis
of the focused beam where the residual central light
intensity is negligible [7]. The ∼1:5 μm deep cavity in
the end of the row has opened, whereas the cavities
to the left of it are still covered with material to a different
extent. The top shell of the cavity seen in the third image,
which has almost detached from the substrate, is strongly
charging and therefore looks pronouncedly bright on the
SEM image.
We suggest that the formation of nanocavities with a
single femtosecond vortex pulse is caused by the implosion phenomenon. By focusing a femtosecond vortex
pulse very tightly, we concentrate the electromagnetic
energy inside a micrometer-size toroidal volume. In
our experiments the peak intensity at the focus is nominally (i.e., aberration-free focusing) ∼4:5 × 1014 W=cm2 ,
which exceeds by far the ionization threshold (i.e.,
∼1013 W=cm2 ) of glass. In the regions where the light intensity exceeds the ionization threshold, the material is
rapidly atomized and converted into a toroidally shaped
3419
electron plasma. The volume of the ∼1 μm diameter, 1.5μm-long plasma toroid (see Fig. 3) is estimated at
∼1 μm3 , which translates into a ∼1 MJ=cm3 electron
thermal energy density (assuming a 50% pulse energy deposition efficiency) [8]. Within a few picoseconds after
the pulse, the electrons give their energy to the lattice
and two shock waves (SWs) emerge from the energy
deposition zone: one wave propagates toward the center
of the toroid, whereas the other wave propagates into the
surrounding material. The estimated ∼1 MJ=cm3 electron thermal energy density corresponds to a ∼0:5 TPa
pressure created inside the material that drives the SWs
[8]. This pressure is well above the Young modulus for
soda-lime glass, viz., ∼20 GPa. At a certain point the
SWs convert into sound waves. The stopping distance
can be estimated from the condition that the internal energy inside the volume confined within the SW fronts is
comparable to the absorbed pulse energy [8]. The stopping distance for our quasi-cylindrical geometry is estimated at ∼1 μm, which exceeds the radius of the
toroid. Therefore, the converging SW can reach the axis
of the toroid and raise pressure inside that region dramatically. The implosion phase is followed by the explosion
of the ultracompressed material near the surface, which
leads to the formation of cavities. Further research and
detailed modeling are necessary to ascertain the origin of
this novel phenomenon.
We acknowledge financial support from the National
Health and Medical Research Council of Australia and
the Australian Research Council. We also acknowledge
Dr. F. Brink and Dr. H. Chen for their assistance in obtaining the SEM images and K. Vu for his assistance in
measuring depth profiles of the ablated samples.
References
1. A. Picón, J. Mompart, J. R. Vázquez de Aldana, L. Plaja, G. F.
Calvo, and L. Roso, Opt. Express 18, 3660 (2010).
2. R. Ling-Ling, Q. Shi-Liang, and G. Zhong-Yi, Chin. Phys. B
19, 034204 (2010).
3. Y. Tokizane, K. Oka, and R. Morita, Opt. Express 17, 14518
(2009).
4. V. G. Shvedov, C. Hnatovsky, W. Krolikowski, and A. V.
Rode, Opt. Lett. 35, 2660 (2010).
5. A. Volyar and T. Fadeyeva, Opt. Spectrosc. 96, 96 (2004).
6. A. Ben-Yakar and R. L. Byer, J. Appl. Phys. 96, 5316 (2004).
7. Y. Iketaki, T. Watanabe, N. Bokor, and M. Fujii, Opt. Lett.
32, 2357 (2007).
8. E. G. Gamaly, S. Juodkazis, K. Nishimura, H. Misawa, B.
Luther-Davies, L. Hallo, P. Nicolai, and V. T. Tikhonchuk,
Phys. Rev. B 73, 214101 (2006).