Simulation of bi-material nano-second laser ablation

M. Dal1, L. Carvalho2, A. Semerok2, W. Pacquentin2, H. Maskrot2.

1. Laboratoire PIMM, Arts et Métiers, CNRS, Cnam, HESAM Université, 151 Bd de l’hôpital 75013 Paris, France.
2. CEA, DEN/DANS/DPC/SEARS/LISL, F-91191 Gif-sur-Yvette, France.

Introduction All these phenomena will be described in the next

subsections.
Among the new cleaning processes, laser ablation
technique generates a great interest since 1990 and has
led to industrial patents [1, 2, 3]. Number of studies
have been done on this topic in United States, France,
China, Japan and United Kindom. This process is
based on ejection of contaminated matier
consequently to energy absorption coming from laser
pulses. After ablating, the matier can be collected by
vacuum and insulated to avoid environment
contamination. As this is a non-contact method, the
treatment can be provided safely regarding
contamination and as only oxide is ablated, the
contamined material amount to be treated remains
quite weak. Moreover, this technique is particularly Figure 1: Scheme of laser absorption phenomena.
suitable for metallic treatment due to its ability to the
laser beam to interact only with the contamined oxide.
In these conditions, the laser ablation leads to high Governing Equations / Numerical Model /
cleaning rate [4, 6]. Simulation / Methods / Use of Simulation
Nevertheless, the process has to be optimized to Apps
reduce the residual contamination level. Indeed,
literature indicates several parameters limiting the Geometrical assumption:
treatment efficiency. One of the negative effects is the As the laser beam is circular and the workpiece
diffusion of contaminant in the bulk due to the laser quite large regarding the beam size R0, it is possible to
heating [6] and the entrapment of contaminant in consider the problem in cylindrical coordinates system
surface defects [5, 7, 8]. Both of them need an increase (Figure 2). Moreover, for this understanding stage, the
of the laser decontamination knowledge mainly study is focused on a single laser impact.
treated experimentally at this time.
In addition to an experimental study, authors propose
in this paper a numerical analysis of the process. The
supposed ablation phenomena are phase explosion,
mechanical fracturing and vaporization.
To reduce the scale of the numerical problem, a single
pulse is studied but results will be extrapolated to real
treatment conditions. The physical model and
numerical assumptions will be firstly explained. After
result validation, an analysis of ablation mechanisms
will be proposed.

Theory / Experimental Set-up

Figure 2: Simulated geometry (304L Stainless Steel: clear
In order to be sufficiently predictive for understanding gray, Fe3O4 Iron Oxide: dark gray).
ablation phenomenon, the present model has to
consider heat transfers in both phases, fluid flow and Heat transfers:
mass transfers. Moreover, as the material is
𝜕𝑇
evaporating, a recoil pressure is applied at the liquid- 𝜌" 𝑐$ %& /////⃗
+ 𝜌" 𝑐$ 𝑣⃗. ∇𝑇
gas boundary and the liquid boundary motion has to be " 𝜕𝑡 "

considered numerically (Figure 1). = /∇⃗. 1𝜆" /////⃗

∇𝑇4 + 𝑆" (𝑟, 𝑧, 𝑡)
Where rm, cp m and lm are the material dependent - [BF], [FC] and [CD] Thermal insulation
thermal properties. T is the computed temperature and /////⃗. 𝑛
−𝜆" ∇𝑇 /⃗ = 0
Sm(r,z,t) the laser bulk power absorption set only in the
oxide layer. The exponent eq indicates that the fusion Where, hcv is a convection coefficient, e is the material
latent heat is included in an equivalent [9] specific emissivity, ssb the Stefan-Boltzmann constant, 𝑚̇ is
heat: the evaporation rate coming from Hertz-Langmuir
𝐿" (DCDE )F
(C ) equation [13] and Lvap the latent heat of vaporization.
𝑐$%& = 𝑐$ " + exp GD F
√𝜋Δ𝑇 ?
𝑀
As the oxide is not perfectly opaque, a certain amount 𝑚̇ = 𝑝KJq (𝑇)r (1 − 𝛽L )
of power is distributed in the oxide bulk. This 2𝜋𝑅𝑇
absorption is assumed to follow a Beer-Lambert law:
With psat the saturation pressure (Clausius-Clapeyron
𝑆" (𝑟, 𝑧, 𝑡) = 𝜑IJK%L (𝑟, 𝑡) expCMNOPQR (SCST) 𝐴V%PWR equation [13]:
vwx y y
( C )
𝑝KJq = 𝑝_ exp ` Df D
With AFe3O4 the oxide volume absorption, z0 the z
coordinate of A and B points (Figure 2). jlaser is the Tv is the boiling point, M the molar mass and p0 the
spatial and temporal distribution of the laser beam. ambient pressure.
F
𝛼𝑃" L
C F Fluid flow:
𝜑IJK%L (𝑡) = 𝐷(𝑡) exp ?`T
The fluid flow problem is computed through the
2 𝑓 𝜏 𝜋 𝑅_?
resolution of two equations: Momentum conservation
and mass conservation.
D(t) is the time distribution of the laser impulsion
shown in Figure 3 and coming from experimental 𝜕𝑣⃗
measurements. Pm is the mean value of laser power, R0 𝜌" z /⃗)𝑣⃗ {
+ (𝑣⃗. ∇
the standard deviation of this Gaussian expression 𝜕𝑡
D
(with ½ factor), f the frequency and t the pulse time /⃗. |𝑝𝐼 + 𝜇" (∇
=∇ /⃗𝑣⃗4 • + ///⃗
/⃗𝑣⃗ + 1∇ 𝑓f
duration.
///⃗
∇. 𝑣⃗ = 0

Where the dynamic viscosity µm is assumed to be the

same in both materials. It should be noted that the mass
conservation is written here for a not compressible
material and Boussinesq approximations are not made
here due to the very fast time dynamics avoiding
convective eddies.

Boundary conditions:
- [AE] and [ED] Axial symmetry
𝑣⃗. 𝑛/⃗ = 0
- [AB] surface tension and recoil pressure
Figure 3: Measured time distribution of laser power D(t). 𝜎. 𝑛/⃗ = 𝛾(𝑇)𝜅 + 𝑝L%e
- [BF], [FC] and [CD] no slip
Boundary conditions: /⃗
𝑣⃗ = 0
- [AE] and [ED] Axial symmetry
−𝜆" /////⃗
∇𝑇. 𝑛/⃗ = 0 With g the liquid material surface tension, k the local
- [AB] convection, radiation and evaporation heat curvature of the interface and prec the recoil pressure
losses coming from the saturation pressure and considering
−𝜆" /////⃗ m
∇𝑇. 𝑛/⃗ = −ℎef 1𝑇 − 𝑇ghi 4 − 𝜀𝜎Kl 1𝑇 m − 𝑇$,ghi 4 the retro-diffusion coefficient br [12].
− 𝑚̇𝐿fJ$
 (1 + 𝛽L ) Boundary conditions:
𝑝L%e = 𝑝_ + 𝑝KJq
2 - [AE] , [ED] ,[BF] and [FC] no motion
in the r direction and free displacement
To complete the description, the initial conditions in in the z direction
velocity is zero and the pressure is set to the
- [DC] free along r and locked along z
atmospheric one. The laser evaporation thermal and
mechanical phenomena are described in more detail in - [AB] normal velocity
literature [10, 11, 13]. 𝑣l = 𝑣⃗. 𝑛/⃗ − 𝑣
////⃗.
J 𝑛 /⃗

Chemical species transport: With 𝑣⃗. 𝑛/⃗ the conservative part coming from the
As the aim is to compute the oxide residual content, fluid flow calculation and 𝑣////⃗. /⃗ the normal
J 𝑛
the proportion of each element (steel and oxide) has to "ablation velocity" ( ////⃗.
𝑣J 𝑛/⃗ = 𝑚̇/𝜌" )
be computed. The model used here is the component
conservation thanks to the Fick law: The resulting motion of the mesh in the domains
is computed in order to optimize the mesh quality
𝜕𝐶
+ 𝑣⃗. /////⃗
∇𝐶 = ∇ /⃗. 1𝐷gii (𝑇) ∇𝐶
/////⃗ 4 through an arbitrary hyper-elastic model.
𝜕𝑡
Model and process parameters:
For the whole boundaries, no material flux is
considered. The diffusion coefficient Diff(T) is The parameters used in previous models are
assumed to be temperature dependent through an summarized in tables 1 and 2.
Arrhenius law in the solid phase and related to the
Stokes-Einstein approximation in the liquid [14]: Table 1: Model Properties

„ Properties 304L Fe304

𝐷_ eC`D (𝑇 < 𝑇" V%PWR )
𝐷gii (𝑇) = ƒ 𝑘 𝑇 l [W/m/K] 24.5 15
l
(𝑇 > 𝑇" V%PWR )
6𝜋𝑎‰Š 𝜇

Where D0 is the pre-exponential factor, Q is the

r [kg/m3] 8020 5180
activation energy, kb is the Boltzmann constant and aEu
is the atomic size of Europium element. cp [J/kg/K] 609 621
It should be noted that the diffusive properties of oxide
in steel with the temperature are not very accurate.
Nevertheless, as the dynamic is very fast the fluid D0 [m2/s] 5 10-6 4.5 10-7
transport is supposed to be more sensitive.

Liquid deformation and material ablation: a [%] 29 70

In this model the liquid - gas interface has to
move due to the recoil pressure compensated by A [cm-1] 36 104 32 104
surface tension and due to the material
withdrawal by vaporization. The former
phenomena are computed form pressure Tm [K] 1700 1863
equilibrium coming from fluid flow calculation
([AB] boundary condition in Fluid flow Tv [K] 3173 2896
paragraph), thus naturally conservative in mass.
The later one is coming from the mass ablated
rate calculation 𝑚̇. As this mass leaves the
studied system, it has to be uncorrelated to the
mass conservation.
 pulse and 10 ns during the cooling. Secondly, the
Table 2: Common and process properties oxide layer is spatially meshed with triangular 10
nm elements.
Properties 304L and Fe304 After a mesh convergence analysis, this
configuration seems to allow correct computation
h [Pa.s] 0.004
of laser absorption and oxide concentration
calculation.
g [N/m] 1.6 In the rest of the element sizes growth with a
factor of 1.1 (Figure 4).
•Ž
[N/m/K] 1.2 10-4
•D In order to reduce the Degrees Of Freedom to be
solved, a linear interpolation is set in elements for
br 0.17 all physics. The numerical problem is thus
composed by more of 2500000 DOF, solved with
R0 [m] 32.5 10-5 2x8@2.4GHz CPU in approximately 4 hours.

Results and discussion

Pm [W] 15
Before presenting results, the aim of this
T¥ [K] 293 numerical analysis has to be recalled. This model
is made to predict the ablated material amount
and to understand ablation phenomena. The first
hcv [W/m2/K] 10 part of the results is the model validation
regarding similar experiment. In a second step,
the ablation phenomena will be analyzed in the
Mesh properties and time step: current process parameter range.
This problem has the particularity to be multi-
scales in space and in time. In other words, the Model validation
oxide layer is 130 nm thick and the whole Firstly, the thermal field can be validated trough
simulated piece is 30 µm thick (figure 2) and the oxide ablation amount and melted zone
pulse duration is 150 ns (figure 3) whereas the comparisons with experiments. The figure 5 is an
process period is 50 µs. Thus, the time and space extrapolate 3D view showing oxide concentration
discretizations have to be chosen carefully. decrease with the time during one spot laser
Firstly, in order to consider correctly the whole processing. It can be observed that the ablated
laser pulse, a time step of 5 ns is set during the zone remains constant after 500 ns, i.e. few

Figure 4: Mesh definition.

hundred nanoseconds after the laser pulse. Two compared to experimental picture in figure 6. A
mechanisms are responsible to the oxide amount particular caution has been taken to scales
decrease in the center part, the material removing equality between experiment and simulation
due to evaporation and the lateral fluid motion results. The experimental picture shows three
involved by recoil pressure. different areas, the ablated zone (lighter zone is
The isolevel lines indicate the fusion and the metal), oxide melted zone where the initial
vaporization temperatures. The larger one can be roughness seems to be modified and the solid
compared to the experimental melted zone (figure oxide zone. The comparison indicates that the
6). ablated zone corresponds to material part which
In order to validate the thermal field, numerical is higher than the boiling point and the melted
key temperatures (melting and vaporization) are

Figure 5: Dynamics of oxide removal.

oxide limit is also consistent with calculated Oxide layer removal analysis:
melted zone. In this part, the previous model will be used to
In addition to this upper view analysis, the melted explain phenomena occurring during oxide layer
thickness was measured on transmission ablation. The most probable reasons of oxide
electronic microscopy pictures. The experimental withdrawal are:
thickness is estimated to 0.7 µm whereas the — phase explosion,
numerical model predicts 1.2 µm. This — mechanical fracturing,
discrepancy can be explained by the uncertainties — vaporization.
on material properties at high temperature. The first one appears when the metal start to
Indeed, in the model, thermal properties are evaporated before oxide with both oxide and
different for oxide and metal, but not temperature metal in liquid phase. The second one is similar
dependent. Moreover, as the phase change is very to glass cutting, i.e. the solid oxide is broken due
fast, it is quite difficult to detect the transition to its thermal expansion. The last one, is the
zone accurately on TEM pictures. The classical evaporation process, which means that
experimental information is probably slightly the oxide leaves the sample as vapor.
underestimated. To validate or invalidate one of the previous
The computed minimal oxide residual ablation origins, the temperature evolution seems
concentration is close to 28 %. As the measured to be crucial information. The time evolution of
initial content of contaminant was 1 % in the point A and E temperatures are plotted in figure
oxide, we can estimate its residual concentration 7. The different material melting and boiling
as 0.28 % numerically. This content has also been points are added to the plot. The intersections of
measured experimentally to 0.2 % (by GDMS). lines and temperature curves indicate when each
phase start to melt and evaporate.
This figure indicates that the oxide melts and
evaporates before metal. Moreover, the metal
start to melt nearly when the oxide starts to
evaporate. When the metal start to evaporates, the
calculated evaporation rate of oxide is more than
20 times higher than metal. This fact indicates
that the phase explosion is the weakest likely
assumption.
In addition to this analysis, the high evaporation
rate indicates the vaporization is a sensitive
phenomenon in this context. For instance and for
this case, with an evaporation radius equal to the
laser beam radius (32 µm) during approximately
100 ns the ablated mass during 1.44 106 pluses
(experimental conditions) is computed to 5.1 mg.
The experimentally measured vaporized mass is
estimated to 3.3 mg. The gap between these two
values is mainly due to the context and
Figure 6: Experiment and numerical melted and assumptions. Indeed, numerically, the ablated
evaporated zones
mass of one pulse is multiplied by the number of
As the thermal and species fields are quite well experimental spot while the experimental data
estimated numerically, the model is supposed to concerns overlapped spots (80 %). Thus, the two
be sufficiently predictive to allow understanding evaporation cases are not perfectly similar, in the
of the oxide layer withdrawal. numerical case, only oxide is ablated, whereas in
the experiment, a part of previously ablated area to nuclear decontamination, Applied Surface Science,
is reheated. As the oxide absorbs more energy 208, 298–305 (2003).
than metal, the evaporation considered 6. R.L. Demmer, R.L. Ferguson, Testing and
numerically is higher. evaluation of light ablation decontamination,
Technical report, EG and G Idaho, Inc., Idaho Falls,
(1994).
7. Nuclear Energy Agency, R&D and Innovation
Needs for Decommissioning Nuclear Facilities
(2014).
8. A.J. Potiens, J.C. Dellamano, R. Vicente, M.P.
Raele, N.U. Wetter, E. Landulfo, Laser
decontamination of the radioactive lightning rods,
Radiation Physics and Chemistry, 95, 188-190 (2014).
9. C. Bonacina, G. Comini, A. Fasano, M. Primicerio,
Numerical solution of phase-change problems,
International Journal of Heat and Mass Transfer, 16
1825-1973 (1973).
10. C.J. Knight, Theoretical modeling of rapid surface
Figure 7: Temperature and evaporation rate evolution with vaporization with back pressure, AIAA Journal, 17, 5p
time on points A and E (Figure 1). (1979).
11. A. Matsunawa, V. Semak, The simulation of front
Conclusions keyhole wall dynamics during laser welding, Journal
of Physics D: Applied Physics, 30, 798-809 (1997).
As a conclusion, an ablation model has been done 12. K. Hirano, R. Fabbro, M. Muller, Experimental
considering thermal, fluid flows and species effects. determination of temperature threshold for melt
As the experimental context is quite extreme, the surface deformation during laser interaction on iron at
experimental validation is difficult. Nevertheless, the atmospheric pressure, J. Phys. D. Appl. Phys., 44,
few information compared between the model and 435402 (2011).
experiment results allow to be quite confident with the 13. A.A. Samokhin, Effect of Laser Radiation on
simulation. It provides a lot of information to analyze Absorbing Condensed Matter, Nova Science
and leads to propose interpretation of the most likely Publishers (1990).
mechanism occurring in our ablation case. 14. A.D. Pasternak, D.R. Olander, Diffusion in liquid
metals, AIChE Journal, 13(6) 1052–1057 (1967).
References

1. W.H. Kasner, T.A.Wojcik, H.E. Feree, Laser

decontamination method, Patent EP0091646B1
(1986).
2. S. Shuttleworth, M.R. Chandratillake, V.J.
Robinson, Material removal by laser ablation, Patent
WO1995027986A1 (1995)
3. J.P Cartry, G. Clar, A. Martin, Apparatus for
working by laser, especially for the decontamination
of a pipe of a nuclear reactor, US Patent 5256848
(1993).
4. A. Leontyev, Laser decontamination and cleaning
of metal surfaces: modelling and experimental studies,
PhD thesis, Université Paris Sud - Paris XI, (2011).
5. Ph. Delaporte, M. Gastaud, W. Marine, M. Sentis,
O. Uteza, P. Thouvenot, J.L. Alcaraz, J.M. Le
Samedy, D. Blin, Dry excimer laser cleaning applied