Opt Quant Electron (2007) 39:939–946

DOI 10.1007/s11082-007-9158-5

Model of the femtosecond laser inscription by a single

pulse

Jovana S. Petrovic · Vladimir Mezentsev ·

Holger Schmitz · Ian Bennion

Received: 15 July 2007 / Accepted: 20 November 2007 / Published online: 11 December 2007
© Springer Science+Business Media, LLC. 2007

Abstract A comprehensive model of processes involved in femtosecond laser inscription

and the subsequent structural material modification is developed. Different time scales of
the pulse-plasma dynamics and thermo-mechanical relaxation allow for separate numerical
treatments of these processes, while linking them by an energy transfer equation. The model
is illustrated and analysed on examples of inscription in fused silica and the results are used
to explain previous experimental observations.

Keywords Femtosecond laser inscription · Numerical model

1 Introduction

Direct femtosecond micromachining of optoelectronic components is based on modification

of the refractive index of material by intense femtosecond laser pulses. Material modification
is a consequence of multi-photon ionization initiated by the focused femtosecond pulse and
subsequent avalanche ionization and heating due to the absorption of the rest of the pulse.
The transfer of the energy stored in excited electrons to ions causes structural material mod-
ification, hence a refractive index change. Precise control of the resultant refractive index
profile can be achieved by controlling the laser pulse power, duration, repetition rate and
focusing conditions. However, fabrication is mainly based on empirical results as a complete
understanding of the physical mechanism involved does not exist yet. Complex pulse-plasma
dynamics, as the first stage in the inscription, has been studied by a number of groups which
has resulted in a number of models for onset of the pulse–material interaction (Carr et al.
2005; Gaeta 2000; Tzortzakis et al. 2001). On the other hand, material changes induced by
ultra-short pulses have been modelled using the laser field as a direct source of thermal
energy (Bennett and Li 2001; Zhang et al. 2004). This paper reports on plasma density cal-
culated by extensive simulations of the nonlinear pulse propagation and suggests how to link

J. S. Petrovic (B) · V. Mezentsev · H. Schmitz · I. Bennion

Photonics Research Group, Aston University, B4 7ET Birmingham, UK
e-mail: [email protected]

123
940 J. S. Petrovic et al.

it with a thermo-mechanical model of the material. The examples are based on inscription in
fused silica by different laser pulse powers in a realistic experimental setup. The presented
model is a simplified phenomenological alternative to initio molecular dynamics simulations
(e.g. Sen et al. 2003).

2 Physical model

As the time scales of the processes involved in femtosecond inscription differ greatly, mod-
elling can be simplified by treating subsequent processes independently and linking them
together through the initial conditions. Electrons excited by multi-photon ionisation reach
thermal equilibrium in several femtoseconds and serve as seed for the avalanche ionisation
through which the trailing part of the pulse is absorbed. Absorbed energy that is not spent on
the ionisation is stored in electrons as their kinetic energy. Whereas the electron recombina-
tion time has been measured to be 170 fs in fused silica (Sun et al. 2005), which is similar
to the pulse duration, the mechanical and thermal changes on this time scale are negligi-
ble. Electron-ion energy transfer lasts several ps, whilst elastic deformation spreads with the
speed of sound through the focal region of diameter 1 µm in 0.1 ns, and the thermal diffusion
is on the µs timescale. Hence, they can be decoupled from the plasma generation due to the
different time scales.
Due to the high pulse intensity the pulse propagation is strongly nonlinear, with the Kerr
effect and nonlinear plasma absorption as the dominant effects for the pulses with duration
of about 100 fs. When the pulse power exceeds critical power, diffraction cannot counteract
self-focusing and the collapse of the pulse occurs which leads to a damage in the material
(Marburger 1975). However, due to the fast free electron generation via MPA, plasma defo-
cusing and absorption may arrest the collapse. Interplay of these effects has been suggested as
useful for ‘one-step’ fabrication of subwavelength or geometrically complex refractive index
changes (Mezentsev et al. 2006a). Here, both the subcritical, P < Pcr , and critical regimes,
P > Pcr , of self-focusing defined by Pcr = λ20 /(2πnn2 ), where λ is the laser wavelength, n
is the linear and n2 the nonlinear refractive index of silica, are studied.
With regard to the profile of the induced index change and the background physical pro-
cesses there are two general scenarios of femtosecond inscription. (i) Plasma created by a
strong or a very tightly focused pulse exerts a high (∼GPa) pressure on the surrounding
material thereby generating a shockwave that moves outwards from the focal region. This
results in a rarified region with decreased refractive index surrounded by a densified region
with the increased refractive index. In the extreme case a void is formed (Martinez et al.
2006; Juodkazis et al. 2006). This regime can be used for fabrication of point defects that
form gratings and optical memories. (ii) Pulses with the power below the threshold of the
structural damage of material cause smooth index change suitable for fabrication of wave-
guides (Nguyen et al. 2006). As it does not treat the shock waves, the scope of the model
proposed here is limited to inscription with moderate pulse powers and focusing conditions.
Time scale of the pulse–material interaction has been studied by using different pulse
repetition rates. In the case of a high repetition rate (82 MHz) profile of the induced index
change depends on the number of pulses which is explained by the arrival of the subsequent
pulse before the thermal diffusion has finished (Schaffer et al. 2003). This effect has not
been observed when a low repetition rate (1 kHz) was used. Our model simulates material
modification over a time long enough to be used as one cycle in both high and low repetition
rate regimes.

123
Model of the femtosecond laser inscription by a single pulse 941

3 Numerical model

Propagation of the pulse with the duration tp , wavelength λ and frequency ω and its interac-
tion with the material are described by the nonlinear Schrödinger equation (NLSE) coupled
to the Drude’s equation for plasma generation (Feit and Fleck 1974; Feng et al. 1997):
∂ 2u
iuz + κ⊥ u − δ + σ |u|2 u = −iγ (1 − iωτ )ρu − iµ|u|2(K−1) u (1)
∂t 2

∂ρ
= ν|u|2 ρ + |u|2K (2)
∂t
Here u is the electric field

normalised to the intensity at which the K-photon absorption
starts, given by IMPA = K ρBD K h̄ω/β (K) τp , and ρ is the plasma density normalised to the
breakdown density ρBD . Terms on the LHS of (1) describe diffraction, dispersion and Kerr
effect, respectively. The first term on the RHS accounts for the absorption and defocusing
by plasma and the second for the multiphoton absorption. Plasma is generated via avalanche
and K-photon absorption, whereas the recombination was neglected in this model. Since the
radial dimension of the pulse changes from several mm prior to focusing to nearly 1 µm at
the focus, adaptive mesh refinement was implemented. Details of the numerical procedures
can be found in Mezentsev et al. (2006b).
Normalised parameters assumed the following values: κ = 43.8, δ = 0.005, σ = 63.95,
γ = 922.87 (Tzortzakis et al. 2001), and the pulse was the one generated by the Ti:Sapphire
laser with the wavelength λ = 800 nm, duration tp = 60 fs and energies up to 10 µJ.
Changes in the material following the relaxation of plasma were obtained by solving the
set of thermo-mechanical equations coupled through the dependence of glass viscosity on
temperature. Thermal diffusion was modeled by the heat equation
∂T
ρg Cp − ∇(k∇T ) = Q (3)
∂t
in which ρg , Cp and k are the density, specific heat and thermal conductivity of glass, respec-
tively. The temperature field was derived under the assumption that the total kinetic energy
of electrons is transferred to the thermal energy of ions:
ρ(r)Ek
Tpl (r) = . (4)
ρg Cp
The kinetic energy of electrons is obtained as the total energy
 absorbed from the pulse abated
by the energy spent on ionisation: Ek = Eabs − Eg ρ(r) dr. High temperature gradient
which is due to the good confinement of plasma causes fast heating and expansion of the
glass that is followed by rapid cooling, which results in incomplete relaxation of material.
When exposed to a long term stress or high temperature, glass structure develops a strain that
aims to relieve the stress and to return the glass to the state of equilibrium. The result is the
contraction or expansion, therefore the change in density. This phenomenon, called creep,
σ N −1
can be described by the Norton law ε̇c = 1.5τ N s, with the creep strain εc , relaxation time τ
σY
and yield stress σY (Benham et al. 1996; Bennett and Li 2001). The relaxation time is propor-
tional to the viscosity η = C exp (Ea /kb T ) (Krol et al. 1986; Doremus 2002). Measurements
of viscosity and relaxation time reported in literature do not give unique values for activation
energy Ea and constant C of the vitreous silica, largely due to the difference in composi-
tion of the samples. In the examples given here, the parameters of Homosil Ea = 5 eV and

123
942 J. S. Petrovic et al.

C = 5 × 10−15 from Krol et al. (1986) were  used. σ is von Mises stress and s is the deviator-
ic stress defined as sij = σij − (1/3)δij σkk , where σij are elements of the stress tensor.
Dynamic change in fictive temperature Tf was neglected. Nevertheless as it influences glass
transition, it will be included in the future versions of the model.
Total strain is composed of elastic strain εe , creep εc and thermal strain εth = α(T − Tref ),
with α = 9 × 10−6 and Tref = 300 K. Displacements and stress equilibrium equations are
defined in the cylindrical coordinate system and are radially symmetric in accordance with
the model of plasma generation:
∂σr τrz σr − σφ
+ + =0 (5)
∂r z r

∂τrz ∂σz ∂σz τrz

+ + + =0 (6)
∂r ∂r ∂z r
The components of tensile stress σr , σφ and σz and shear stress τrz are related to strain ε
and therefore displacement u via the constitutive strain–stress relation σ = Dε, (see e.g.
Benham et al. 1996). Equations 3, 5 and 6 have been solved by finite element solver released
by Comsol (2006).
Refractive index change can be calculated from the change in density by using the
Clausius–Mossotti relation scaled by empirically obtained constants as in Zhang et al. (2004).
The change in density is obtained directly from the volumetric strain: dρ/ρ = −εvol . As the
experimental values for the index change are of the order of 10−4 or 10−3 , the density of
glass in the heat equation was considered constant.

4 Results and discussion

Dynamics of the pulse propagation and plasma generation was studied for three different
pulse powers: 0.5Pcr , 2Pcr and 5Pcr in two focusing configurations: tight focusing by micro-
scope objective with NA = 0.45 and weak focusing by lens with f = 0.5 mm and effective
NA = 0.1. Whereas the complex pulse dynamics is a large topic and has been published else-
where (Mezentsev et al. 2006b), the results more relevant for the current model are plasma
density distribution and energy absorbed from the pulse.
The first column in Fig. 1 shows the spatial distribution of the normalised plasma den-
sity obtained by integration over the whole time domain for the case of strong focusing.
As a consequence of the intense nonlinear absorption in the focal region, pulse focusing
is counteracted by defocusing on plasma, preventing the collapse of the pulse. The electron
concentration exceeds critical plasma density only for the pulse of power 5Pcr and only in the
radius of a few hundreds of nanometers, which corresponds to the size of the voids observed
in experiments (Martinez et al. 2006). In general, the maximal electron concentration in the
case of strong focusing was of the order of 1021 cm−3 based on which a permanent index
change may be expected. With the increase in pulse power, the maximum of plasma density
moves several micrometers deeper into the sample, as shown in Table 1. Plots (1b), (2b), (3b)
in Fig. 1 show the drop in pulse energy along the propagation axes. In order to estimate the
elevated temperature of the material, it is assumed that all absorbed energy which has not
been spent on ionisation is spent on the thermalisation of plasma, hence the kinetic energy
of one electron is Eke = (Eabs − EI )/N . The electrons were accelerated to 7.31, 6.79 and
6.61 eV, for the pulse powers of 0.5Pcr , 2Pcr and 5Pcr . This corresponds to the electron
temperatures of 56,470, 52,467 and 51,108 K, respectively.

123
Model of the femtosecond laser inscription by a single pulse 943

Table 1 Maximal plasma density and its location, number of ionised electrons, energy spent on ionisation
and pulse energy absorbed by silica during the inscription by a pulse with NA = 0.45

P /Pcr ρmax (1021 cm−3 ) z(ρ = ρmax ) (µm) Ne EI (µJ) Eabs (µJ)

0.5 1.31 175.4 1.497 × 1010 0.0182 0.0357

2.0 1.62 174.8 8.839 × 1010 0.1075 0.2035
5.0 1.95 178.1 2.629 × 1011 0.3196 0.5977

Fig. 1 (a) Density distribution of plasma generated in silica by 120 fs pulses with powers: (1) 0.5Pcr ,
(2) P = 2Pcr , (3) P = 5Pcr . (b) Correspondent drop in pulse energy along the propagation axis

123
944 J. S. Petrovic et al.

Fig. 2 Evolution of the von Mises stress in silica glass caused by the pulse with P = 2Pcr and NA = 0.45

123
Model of the femtosecond laser inscription by a single pulse 945

In the case of weak focusing the maximal electron concentration is of the order of
1019 cm−3 which is close to the inscription threshold estimated in Stuart et al. (1996). Electron
energies are comparable to the previous case: 5.31 eV (41,043 K), 5.51 eV (43,749 K) and
5.66 eV (42,589 K) for the pulse powers 0.5Pcr , 2Pcr and 5Pcr , respectively, but the number
of electrons is small and the increase in glass temperature is only several tens of K. In the
rest of the paper we consider only the inscription by the tightly focused pulse.
Results from Table 1 were plugged into Eq. 4 to obtain the initial condition for the thermo-
mechanical model of the material response. As the significant mechanical change occurs after
the pulse has left the focal region and electrons have recombined, the source in Eq. 3 is zero.
The glass was heated to 1,010, 1,159 and 1,361◦ C by pulses with powers 0.5Pcr , 2Pcr and 5Pcr ,
respectively. While these temperatures are below the melting point of glass Tm = 1,700◦ C,
the stronger pulses heat the glass above the transition temperature Tg = 1,100◦ C causing the
structural change. This is followed by the nonlinear relaxation due to the rapid decrease in
glass viscosity with temperature.
Initial stress field is calculated as thermal stress and emulates the shape of the electron
plasma distribution (compare Fig. 1b with Fig. 2). It extents beyond the focal region which
may explain the effects observed in fibre Bragg gratings inscribed by femtosecond laser.
Studies of these gratings revealed that the change in the refractive index detectable by the
optical microscope with resolution n = 10−5 cannot account for the grating birefringence
(Dubov et al. 2006), as well as that the gratings become stronger when annealed (Smelser
et al. 2005). Both effects could be explained by the presence of the stressed material around
the major index change at the focal point.
Creep strain increased as the glass tried to restore the equilibrium and it stabilised in 6 ns
at a very small value of 2.5 × 10−13 which was not enough to explain the expected index
change. The very small value of creep strain is due to the long relaxation time τ . Since the
parameters of silica glass found in literature vary greatly and are mainly based on measure-
ments over much larger time scales, the model should be calibrated by fitting the results to
the experimental data. The defects induced by plastic strain were neglected but may also
contribute to the index change.
In 1 µs maximal temperature dropped to 360 K while the von Mises stress decreased about
20 times to approximately 1 MPa. Therefore, the thermal diffusion and the stress relaxation
are not finished in 1 µs, which is in agreement with the results of experiments with MHz
pulse trains.

5 Conclusion

A model of femtosecond inscription by a single pulse based on the separation of the pulse-
plasma dynamics and thermo-mechanical processes due to the difference in their time scales
has been proposed. These models have been linked by the thermal energy transfer equation.
Simulations of inscription in realistic experimental conditions were able to explain and con-
firm experimental results. Subject to the proper calibration to experimental data, model should
be able to render an estimate of the refractive index change, the work on which is in progress.

References

Benham, R.P., Crawford, R.J., Armstrong, C.G.: Mechanics of Engineering Materials, 2nd edn. Prentice Hall
(1996)

123
946 J. S. Petrovic et al.

Bennett, T.D., Li, L.: Modeling laser texturing of silicate glass. J. Appl. Phys. 89, 141 (2001)
Carr, C.W., Feit, M.D., Rubenchik, A.M., De Mange, P., Kucheyev, S.O., Shirk, M.D., Radousky, H.B., Demos,
S.G.: Radiation produced by femtosecond laser-plasma interaction during dielectric breakdown. Opt.
Lett. 30, 661 (2005)
COMSOL AB Comsol Multiphysics 3.3. http://www.comsol.com (2006)
Doremus, R.H.: Viscosity of silica. J. Appl. Phys. 92, 7619 (2002)
Dubov, M., Allsop, T.D.P., Martinez, A., Mezentsev, V., Bennion, I.: Highly birefringent long period gratings
fabricated with femtosecond laser. Optical Fibre Communication Conference (OFC) (2006)
Feit, M.D., Fleck, J.A.: Effect of refraction on spot-size dependence of laser-induced breakdown. Appl. Phys.
Lett. 24, 169 (1974)
Feng, Q., Moloney, J.V., Newell, A.C., Wright, E.M., Cook, K., Kennedy, P.K., Hammer, D.X., Rockwell,
B.A., Thompson, C.R.: Theory and simulation on the threshold of water breakdown induced by focused
ultrashort laser pulses. IEEE J. Quantum Electron. 33, 127 (1997)
Gaeta, A.L.: Catastrophic collapse of ultrashort pulses. Phys. Rev. Lett. 84, 3582 (2000)
Juodkazis, S., Nishimura, K., Tanaka, S., Misawa, H., Gamaly, E.G., Luther-Davies, B., Hallo, L., Nicolai, P.,
Tikhonchuk, V.T.: Laser-induced microexplosion confined in the bulk of a sapphire cystal: evidence of
multimegabar pressures. Phys. Rev. Lett. 96, 166101 (2006)
Krol, D.M., Lyons, K.B., Brawer, S.A., Kurkjian, C.R.: High-temperature light scattering and the glass tran-
sition in vitreous silica. Phys. Rev. B 33, 4196 (1986)
Marburger, J.H.: Self-focusing: theory. Prog. Quantum Electron. 4, 35 (1975)
Martinez, A., Dubov, M., Khrushchev, I., Bennion, I.: Structure of Fibre Gratings Directly Written by Infrared
Femtosecond Laser 2006 Conference on Lasers and Electro-Optics (CLEO) JTuD13 (2006)
Mezentsev, V., Dubov, M., Martinez, A., Lai, Y., Allsop, T.P., Khrushchev, I., Webb, D.J., Floreani, F.,
Bennion, I.: Micro-fabrication of advanced photonic devices by means of direct point-by-point femto-
second inscription in silica. Proc. SPIE 6107, 61070C (2006a)
Mezentsev, V., Petrovic, J., Dreher, J., Grauer, R.: Adaptive modeling of the femtosecond inscription in
silica. Proc. SPIE 6107, 61070R (2006b)
Nguyen, N.T., Saliminia, A., Chin, S.L., Vallee, R.: Control of femtosecond laser written waveguides in silica
glass. Appl. Phys. B 85, 145 (2006)
Schaffer, C.B., Garcia, J.F.: Bulk heating of transparent materials using a high-repetition-rate femtosecond
laser. Mazur E. Appl. Phys. A 76, 351 (2003)
Sen, S., Dickinson, J.E.: Ab initio molecular dynamics simmulation of femtosecond laser-induced structural
modification in vitreous silica. Phys. Rev. B 68, 214204 (2003)
Smelser, C.W., Mihailov, S.J., Grobnic, D.: Formation of type I-IR and type II-IR gratings with an ultrafast
IR laser and a phase mask. Opt. Express 13, 5377 (2005)
Stuart, B.C., Felt, M.D., Herman, S., Rubenchik, A.M.: Nanosecond-to-femtosecond laser-induced breakdown
in dielectrics. Phys. Rev. B 53, 1749 (1996)
Sun, Q., Jiang, H., Liu, Y., Wu, Z., Yang, H., Gong, Q.: Measurement of the collision time of dense electronic
plasma induced by a femtosecond laser in fused silica. Opt. Lett. 30, 320 (2005)
Tzortzakis, S., Sudrie, L., Franco, M., Prade, B., Mysyrowicz, A., Couairon, A., Bergé, L.: Self-guided prop-
agation of ultrashort IR laser pulses in fused silica. Phys. Rev. Lett. 87, 213902 (2001)
Zhang, X.R., Xu, X., Rubenchik, A.M.: Simulation of microscale densification during femtosecond laser
processing of dielectric materials. Appl. Phys. A 79, 945 (2004)

123