Int J Adv Manuf Technol (2013) 68:69–83

DOI 10.1007/s00170-012-4709-8

ORIGINAL ARTICLE

One-dimensional multipulse laser machining of structural

alumina: evolution of surface topography
Hitesh D. Vora & Soundarapandian Santhanakrishnan &
Sandip P. Harimkar & Sandra K. S. Boetcher &
Narendra B. Dahotre

Received: 9 August 2012 / Accepted: 22 December 2012 / Published online: 24 January 2013
# Springer-Verlag London 2013

Abstract The development and understanding of laser–ma- Abbreviations

terial interactions have steered to the machining of advanced Nd:YAG Neodymium-doped yttrium aluminum garnet
structural ceramics. At one point, it was nearly impossible to CO2 Carbon dioxide
machine effectively using various conventional machining ρ Density (in kilograms per cubic meter)
techniques. Nevertheless, achieving a higher material re- Cp Specific heat at constant pressure (in joules
moval rate along with a good surface finish is a critical per kilogram per kelvin)
issue. In this study, a multistep computational model based k Thermal conductivity (in watts per meter per
on COMSOL™ Multiphysics was designed and developed kelvin)
to study the influence of multiple laser pulses on the evolu- T Temperature (in kelvin)
tion of surface roughness of alumina. The computational t Time (in seconds)
model employed the various heat transfer and hydrodynam- x x-coordinate
ic boundary conditions and thermomechanical properties for y y-coordinate
better prediction of surface roughness under various laser δm Normalization function around the melting
processing conditions. The results indicate that, as the pulse temperature (per kelvin)
rate increases, the surface roughness also increases. The δv Normalization function around the
results of the computational model are also validated by vaporization temperature (per kelvin)
experimental observations with reasonably close agreement. H′ Smooth Heaviside function
B Fraction of a particular phase (solid, liquid, or
Keywords Nd:YAG laser . Laser machining . Multiple vapor)
pulses . Alumina . Computational model . Surface roughness Tx Transition temperature between two distinct
phases, solid–liquid or liquid–vapor (in kelvin)
mod Modular function
H. D. Vora : S. Santhanakrishnan : N. B. Dahotre (*) u Velocity in the x-direction (in meters per
Laboratory for Laser Materials Processing and Synthesis, second)
Department of Materials Science and Engineering, v Velocity in the y-direction (in meters per
University of North Texas, 1155 Union Circle #305310,
second)
Denton, TX 76203-5017, USA
e-mail: [email protected] p Pressure (in newtons per square meter)
URL: http://mtse.unt.edu/Dahotre/ Ts Instantaneous surface temperature (in kelvin)
SiO2 Silicon dioxide
S. P. Harimkar
Fe2O3 Iron oxide
School of Mechanical and Aerospace Engineering,
Oklahoma State University, 218 Engineering North, R2O Base or fluxing oxides
Stillwater, OK 74078-5016, USA μm Micrometer or micron (10−6 m)
Rt Maximum height of surface profile (in
S. K. S. Boetcher
micrometers)
Embry–Riddle Aeronautical University,
600 South Clyde Morris Boulevard, Rp Maximum peak height of surface profile (in
Daytona Beach, FL 32114, USA micrometers)
70 Int J Adv Manuf Technol (2013) 68:69–83

Rv Maximum valley depth of surface profile (in the past, Samant and colleagues [1, 2, 14, 27, 28] used the
micrometers) experimental and computational approaches to understand the
mechanisms of material removal during one-dimensional laser
machining to various ceramics (alumina, zirconia, magnesia,
1 Introduction silicon nitride, and silicon carbide). Their studies show that the
higher material removal rate can be achieved by using multiple
Alumina, zirconia, magnesia, silicon carbide, and silicon ni- laser pulses and considered as a preferable machining tech-
tride are grouped as structural ceramics to be considered as one nique for structural ceramics [1, 2, 8, 26, 31]. However, the
of the most versatile groups of materials because of its high main focus of their works was to increase the material removal
temperature stability, ability to resist deformation at elevated rates by selecting the optimal laser processing parameters and
temperature, excellent wear and thermal shock resistance, disregarded the issues related to surface finish. Hence, the
chemical inertness, superior electrical properties, and lower undesirable surface finish produced during one-dimensional
density [1–4]. Therefore, structural ceramics have found appli- laser machining is still one of the critical issues to be resolved.
cations in a variety of industries, including automobile, aero- In light of this, the present authors have already developed
space, medical, printing, textile, and electronic [1–4]. the computational model based on COMSOL™ Multiphysics
However, due to their low fracture toughness (~3–5 MPam½) validated with experimental trials to understand the influence
and high hardness (~1,200–2,200 Knoop hardness), they are of a single laser pulse on laser machining of alumina and its
difficult to machine using conventional machining techniques, subsequent effect on surface finish [32]. It was observed that
such as grinding, cutting, and polishing [5–10]. Moreover, the surface roughness increases with increase in laser energy
numerous other investigators [11–16] reported that these ma- density during the application of a single laser pulse. In
chining techniques are also associated with many drawbacks, extension of this work and as part of the ongoing incremental
such as unacceptable tool wear, insufficient dimensional accu- efforts to develop a laser-based multidimensional (3D) ma-
racy, mechanical or thermal damage to workpieces, lower chining through understanding of the basic science behind the
material removal rates, and surface and subsurface cracks. machining phenomena, the present paper investigates the
Many researchers [17–21] have demonstrated that recent effect of multiple laser pulses on the evolution of surface
development of lasers have driven the machining of advanced topography during one-dimensional laser machining (station-
structural ceramics, which previously were nearly impossible ary workpiece and laser beam) of alumina using both compu-
to machine effectively using various conventional machining tational and experimental approaches. The computational
techniques. Particularly, a one-dimensional laser machining model is developed to simulate the multiple pulses and then
technique (both the workpiece and the laser beam are station- validated with the experimental observations to study the
ary) has been successfully implemented in many industrial effects of various process parameters on the evolution of
applications [21–24], such as aerospace (cooling holes in surface topography of the laser-machined alumina.
nozzle guide veins and gas turbines), automotive (fuel injec-
tion nozzles), and electronic (circuit boards). One-dimensional
laser machining is inherently associated with several defects 2 Laser–material interactions
and many dimension-related issues. Some of the prominent
issues reported are the formation of the recast layer on the walls The temperature rise and fall due to heating (absorption of
of machined surface, spatter formation on the top surface, the laser beam) and cooling (self-quenching by bulk mate-
nonparallel machined walls, lack in dimensional accuracy, rial and heat losses due to external natural convective cool-
larger heat-affected zones, and microcrack formation [21, 25, ing and radiation) control different physical phenomena in
26]. Considerable amount of experimental and computational the material during thermal processing. When a high-
works [1, 2, 14, 15, 27–30] have been published in the past to intensity laser beam (~106 J/m2) strikes the surface of a
improve the quality of laser machining by employing various material, the surface temperature quickly rises due to heat
types of lasers: CO2, excimer, and Nd:YAG. These lasers can conduction. This rapid rise in surface temperature leads to
be operated either by continuous wave (CW) or pulse mode surface vaporization and the subsequent release of neutral
(PM) (millisecond, nanosecond, picosecond, and femtosecond atoms or molecules in a gas cover adjacent to the laser–
lasers). In CW, the output energy of the laser beam is constant- material interaction zone [29, 33–35]. Furthermore, the
ly ON for a specific amount of time, whereas in PM, the lasers evolving vapor from the melt pool is significantly cooler
concentrate their output energy into shorter-time high-power and denser than the atmosphere, which results in a conden-
bursts. The PM lasers can either fire a single pulse or a series of sation of vapor particle back on the surface which generates
pulses (multiple) at regular intervals. In both cases, the instan- a recoil pressure on the melt pool underneath [29, 33–35].
taneous power densities can be extremely high (~107 J/m2), The commonly used pulse width for laser machining is 0.1
and as a result, a larger volume of material is removed [1, 2]. In to 10 ms [1, 2]. These time scales allow the surface to heat to
Int J Adv Manuf Technol (2013) 68:69–83 71

the vaporization temperature and remain there for some time [1, laser beam stationary. The average laser energy density
2]. For higher laser energy densities (~106 J/m2) and shorter (energy per pulse/surface area of laser beam) in each
pulse durations (0.1 to 10 ms), the resulting recoil pressure may case is very high (~106 J/m2), which results in a high
be extremely high, which induces a shock wave. This generates loss of material. Similarly, by delivering an increased
a hydrodynamic melt motion on the liquid melt pool, as pro- number of pulses, higher material removal rates can be
posed by various investigators [29, 33, 34]. As a consequence, achieved. In order to obtain higher material removal
the liquid metal is ejected out from the laser–material interac- rates, PM lasers deliver multiple laser pulses (which is
tion zone. The temperature of the expelled molten material preferred for laser machining of structural ceramics).
drops drastically as soon as it leaves the laser–material interac- Therefore, in the present study, the effect of multiple
tion zone. Furthermore, the effect of recoil pressure on the laser pulses on the evolution of surface topography is
molten material also nullifies as the surface temperature goes modeled to understand the laser machining mechanism
below vaporization temperature. At the end of a laser pulse, the for various conditions. Subsequently, this understanding
surface tension and gravitational forces drive the liquid material can be extended to predict the surface topography dur-
to return to its original place. Due to self-quenching effect and a ing multidimensional laser machining processes.
higher cooling rate (~105 K/s), the liquid material solidifies In PM laser machining, the multiple laser pulses of
instantaneously and forms a liquid crown or liquid pileup. short pulse width are delivered in a repeated manner
Further, Dahotre and colleagues [1, 2, 14, 15] reported that (Fig. 2a, b). The material removal during laser machining
the tangential stress exerted due to surface tension gives shape is carried out either by evaporation or by melt ejection,
to the solidified material. Thus, a pronounced surface topogra- which solely depends on the laser machining parameters,
phy is generated on the laser-machined surface (Fig. 1). such as average energy density, pulse width, and pulse
It was previously observed that, during the application of rate (repetition rate). The laser (or heating) is ON
a single pulse (~106 J/m2), the laser–material interaction (Fig. 2b) during the application of first pulse (till time
zone (~10−10 m3) reaches an extremely high temperature t1) with preset value of pulse width followed by cooling
(>8,000 K) within a very short period of time (<0.5 ms) until the end of the first pulse (time t2 =1/f, in second).
[32]. Therefore, it makes the situation enormously difficult The surface temperature increases from initial temperature
for in situ measurements of thermophysical properties under (Ti) to T1 during heating, while during cooling, the
these severe conditions (high temperature, short time of temperature decreases from T1 to T2 (where T1 >T2 >Ti).
interaction, and small interaction volume). In consequence, Subsequently, the second pulse is supplied from time t2
it is extremely challenging to investigate the evolution of to t3 and again cooling until time t4. This pulse train
surface topography through experimental analyses. Hence, successively delivers a specific number of pulses over the
the computational modeling approach (via finite element entire time of machining operation (until time tf). How-
method) is successfully used in the present study. ever, during the application of multiple laser pulses, the
temperature at the beginning of every pulse was always
higher than the initial temperature (Ti), causing a gradual
3 Modeling approach rise in average surface temperature (Fig. 2b). Further-
more, the variations in surface temperature, in turn, in-
For PM lasers, one-dimensional laser machining is car- fluence the mechanism of evolution of surface roughness
ried out by either supplying a single laser pulse or (topography). Hence, the distinct surface topography is
multiple laser pulses by keeping both workpiece and generated (Fig. 2c) due to the impact of multiple laser
pulses based on the laser pulsing parameter (Fig. 2d).
A two-dimensional axisymmetric computational model,
which incorporates fluid flow and heat transfer, is devel-
oped using COMSOL™ Multiphysics (Fig. 2c). The
model is used to predict the temperature, cooling rate,
and fluid velocity (all of which influence the surface
topology of alumina) after the surface is subjected to
various laser machining conditions. A two-step modeling
approach (Fig. 3) is used in the present computational
model to predict the surface deformation under the ap-
plication of multiple laser pulses. In step 1, the multi-
phase (solid–liquid–vapor) heat transfer model is
Fig. 1 Schematic of laser–material interaction and associated surface incorporated to evaluate the material loss due to evapo-
effects ration which is used to predict the geometry of a crater.
72 Int J Adv Manuf Technol (2013) 68:69–83

Fig. 2 Schematic of laser

machining processes: a Multiple laser f a b
multiple-pulse laser machining pulses 4
process, b temperature as a
function of time, c evolution of
3
surface topography, and d laser Alumina 2
pulsing parameters after
solidification
1 X

Cutting plane
Multiple c d
along X-X laser pulses Surface
topography

Alumina
after
solidification

The initial geometry used in the step 1 computational identify liquid–solid and vapor–liquid interfaces. In step
model is presented in Fig. 4. In order to get the 1, the elements identified as vapor phase are excluded
geometry-independent solution, a mesh sensitivity analy- from the geometry. The geometry of the crater (depth
sis was carried out for six different mesh sizes and the and width) evaluated from step 1 plays an important role
optimal mesh size was used for the computation (Fig. 3). in predicting the final surface profile. Therefore, the
The level set method is also employed in step 1 to crater geometry predicted from step 1 is considered as

Fig. 3 Flow diagram for multistep computational modeling approach

Int J Adv Manuf Technol (2013) 68:69–83 73

the material properties of alumina [1, 2, 36–39] and the

laser machining parameters used in the computational
model. The specific heat related to phase transition is
used (Cp term) in the computational model, which is
expressed by Eqs. 1–3 in Table 2 [40]. The governing
equation and the various boundary conditions employed
in the heat transfer model are also summarized in
Table 2.

3.1.1 Simulation of multiple laser pulses

In the present work, multiple pulses (10, 20, 30, 40,

and 50) are delivered in durations of 1 s. Thus, the
analytic function (β) was designed to simulate the rep-
etition of multiple laser pulses that are expressed by Eq.
7 enumerated in Table 2 [41]. As discussed earlier
(Fig. 2b), during the application of the first pulse, the
laser is ON with a specific value of pulse width (Pw =
0.5 ms) followed by cooling until the end of first pulse
(time t2 =1/f, in seconds). Hence, the laser ON time is
simulated by the rectangle function (lower and upper
limits of 0 and 0.5 ms, respectively). The β function
only works when t≥0 s and t≤1 s, so that it simulates
the pulse train of 1 s.

Fig. 4 Two-dimensional geometry for the computational model 3.1.2 Prediction of crater geometry

The level set method developed by Sussman et al. [42] is

an initial surface profile for step 2. In step 2, the mul- employed in step 1 in order to track the solid–liquid and
tiphase fluid flow model coupled with heat transfer and vapor–liquid interfaces as expressed by Eqs. 11–13 in
phase change kinetics is employed. The prime objective Table 2. For alumina, Tx can be Tv to distinguish the vapor
of step 2 is to estimate the contribution of the deforma- phase from the liquid phase and Tm for the liquid phase from
tion of the liquid material pool inside of the crater under the solid phase. The B ranges from 0 to 1. For Tx =Tm, B=1
various forces (recoil pressure, surface tensions, and indicates 100 % liquid phase and B=0 designates 0 % liquid
gravitational forces) and to predict the final surface to- phase. The interphases is tracked between B=0 and B=1.
pography. This integrated computational approach involves Similarly, for Tx =Tv, B=1 indicates 100 % vapor phase and
various physical phenomena for predicting the surface profile B=0 designates 0 % vapor phase. The elements coming
during multiple-pulse laser machining of alumina (Fig. 5). under vapor phase (B=1 or T>Tv) are identified and re-
moved from the geometry and the dimensions of the crater
3.1 Heat transfer model geometry are estimated which is further considered as an
initial geometry for step 2 (Fig. 5).
During laser machining, the material undergoes phase
change (solid to liquid to vapor or vice versa). In 3.2 Fluid flow coupled with heat transfer model
addition, the heat (latent heat of fusion and evaporation)
consumed or released during the phase transition signif- Multiphase fluid flow coupled with heat transfer and
icantly influences the magnitude of surface temperature, phase change kinetics is modeled in step 2. The new
recoil pressure, melt pool size, and fluid velocity. There- geometry is drawn by considering the material loss due
fore, the temperature-dependent properties of alumina, to vaporization from the step 1 results. The governing
such as density, thermal conductivity, and specific heat, equations for conservation of mass, momentum, and
are incorporated in the computational model. The energy for two-dimensional transient, incompressible
change in density and thermal conductivity with respect laminar flow coupled with heat transfer are expressed
to temperature (ranging from 293.15 to 3,500 K) is in Eqs. 14–17 listed in Table 3. The natural convection
illustrated in Fig. 6. In addition, Table 1 summarizes induced by the buoyancy-driven flow and volume force
74 Int J Adv Manuf Technol (2013) 68:69–83

Fig. 5 Schematic of a physical

model in two-step computa- Step-2: Fluid flow coupled
Step-1: Heat transfer model
tional modeling approach to with heat transfer model
predict the surface topography
a Heat transfer boundary e Fluid flow and heat transfer
conditions boundary conditions

3 6 8 3 8
1 9 1 6 9
Alumina Alumina
2 2
Updated geometry for two-phase fluid flow model
Geometry for two-phase heat transfer model
coupled with heat transfer model

b Heat flux boundary f Boundary conditions on 6:

condition on 6 T Tv heat flux, recoil pressure, surface
(B = 1) tension
Solid-liquid 3 8
Vapor-liquid 1 6 9
interface interface Alumina
2
T< Tv
(B = 0) Level-set method Boundary Conditions

c Crater g Recoil pressure & surface

Crater depth width tension forces

Solid-liquid Solid-liquid
interface Alumina Alumina interface

Remove the elements where its temperature goes

Flow of molten material
above the vaporization temperature or B=1

d Melt h Rp Rt = Rp + Rv
width

Melt depth Rv
Alumina Alumina

Quantification of molten material Surface topography after solidification

is estimated by the Boussinesq approximation presented flow interface shown by Eq. 19 in Table 3. This term
by Eq. 16 in Table 3 [43, 44]. Step 2 also utilized the imposed the velocity in the x-direction and overrides the
same boundary conditions of heat transfer and phase slip boundary condition. In the laser–material interaction
change kinetics as used in step 1 (expressed by Eqs. zone, where the temperature is at its highest, the surface
2–10 listed in Table 2).
The evaporation-induced recoil pressure (Pr) at the evap-
orating surface depends on the incident laser energy density
as expressed by Eq. 18 in Table 3 [1, 2]. In the present
model, the experimentally verified physical model of melt
hydrodynamics and laser-induced evaporation proposed by
Anisimov [33] and Semak [34] are used to simulate the
effect of recoil pressure. The temperature fields determined
from the computational model facilitate the calculation of
evaporation-induced recoil pressure at the evaporating sur-
face during laser machining of alumina. In the present
model, the recoil pressure is applied only when the temper-
ature is above the vaporization temperature (T≥Tv).
The forces at the liquid–air interface due to Marangoni
convection are incorporated in such a way that the shear
stress is proportional to the temperature gradient on the
surface [43]. In order to apply this condition, the weak term Fig. 6 Temperature dependent material properties of alumina density
is added from “weak contribution” feature of the laminar and thermal conductivity
Int J Adv Manuf Technol (2013) 68:69–83 75

Table 1 Material properties of alumina (99.6 %) and laser machining parameters

Property Nomenclature Value (units) References

Material properties
Melting temperature Tm 2,324 (K) [35–37]
Vaporization temperature Tv 3,273.15 (K) [35–37]
Latent heat of melting Lm 1,067.43 (J/g) [35–37]
Latent heat of evaporation Lv 1,066.5 (J/g) [35–37]
Mass of vapor molecule Mv 1.693e−25 (kg/atm) [35–37]
Thermal expansion coefficient α 8.4e−6 (1/K) [35, 36]
Temperature derivative of the surface tension γ −8e−5 (N/(m K)) [38]
Dynamic viscosity μ 1 (Pa s for T<2,190 K) and 0.001 (Pa s for [38]
T>2,500 K)
Laser machining parameters
Number of laser pulses f 10, 20, 30, 40, 50 pulses –
Pulse energy (Ep) and corresponding average laser energy Ep 1 (J) –
Ep
density (Ed) Ed ¼ f 3.5, 7.1, 10.6, 14.1, 17.7 (J/m2 ×107), respectively
ðp4D2 Þ
Pulse width Pw 0.5 (ms) –
Beam diameter D 0.6 (mm) –
Emissivity ε 0.7 [1, 2]
Heat transfer coefficient h 10 (W/m2 K) [16]
Absorptivity A 0.25 [38]
Half-width of the curve ΔT 30 (K) –
Reference point to represents the center of the laser beam xr 1,000 (μm) –
Standard deviation of the Gaussian laser beam ϕ 100 (μm) –
Ambient temperature Ta 293.15 (K) –
Initial temperature Ti 293.15 (K) –
Stefan–Boltzmann constant σ 5.67×10−8 (W/m2 K4) [36]
Universal gas constant R 8.314 (J/mol K) [36]
Gravitational acceleration g 9.81 (m/s2) [36]

Table 2 Governing equations and boundary conditions of heat transfer model

Boundary no. Boundary condition Equation No.

 0
Whole geometry Temperature-dependent specific Cp ¼ 880 þ d m ðLm Þ þ Lm
 H ððT  Tm Þ  ΔT Þ 1

Tm
heat at constant pressure (Cp) 0
þ d v ðLv Þ þ LTvv  H ððT  Tv Þ  ΔT Þ ðin J=kg KÞ
 ðT T Þ2   ðT T Þ2 
exp  m
exp  v
dm ¼ p ffiffi
ΔT 2
; dv ¼ pΔTffiffi2 2 and 3
ΔT p ΔT p
  h
2 
2 i
Whole geometry Governing equation, step 1 ρCp @T
@t ¼ k
@ T @ T
@x2 þ @y2 4
@T
 
6 Heat flux, natural convection cooling  k @y ¼ bPg  h½T  Ta   "σ T 4  Ta4 5
and radiation h
2
i
Pg ¼ A P ppD2 exp  ðxx rÞ
E
Average laser power density in Gaussian (in W/m2) 6
w ð4 Þ 2f2

distribution   
Analytic function b ¼ rect1 mod t; 1f 7
 4 
3, 8 Natural convection cooling and radiation  k @T
@y ¼ h½T  Ta   "σT  Ta 
4
8
1, 9  k @T
@x ¼ h½T  Ta   "σ T  Ta
4 4
9
@T
2 Insulation @y ¼ 0 10
4, 5, 7 Neutral –
Whole geometry Level set method B ¼ 1 for ðT > Tx þ ΔTÞ 11
B ¼ ðT T2ΔT
x þΔT Þ
for ðTx  ΔTÞ  T  ðTx þ ΔTÞ 12
B ¼ 0 for ðT < Tx  ΔTÞ 13
76 Int J Adv Manuf Technol (2013) 68:69–83

Table 3 Governing equations and boundary conditions of fluid flow coupled with heat transfer model

Boundary no. Boundary condition Equation No.

@u
Whole geometry Governing equations, step 2 @x þ @v
@y ¼ 0 14

 h i
@u @p
þ u @u @u @ u @ u
2 2
ρ @t @x þ v @y ¼  @x þ μ @x2 þ @y2 15

 h2 i
@p
ρ @v @v @v @ v @2 v
@t þ u @x þ v @y ¼  @y þ μ @x2 þ @y2 þ ρga ðT  Tm Þ 16

 h2 i
ρCp @T @T @T
@t þ u @x þ v @y ¼ k @x2 þ @y2
@ T @2 T
17
pffiffiffiffiffiffiffiffiffiffiffi
  1:69 ðMkT
v Lv Þ
s

6 Recoil pressure Pr ¼ Pg p ffiffiffiffi only when T  Tv 18

Lv 1þ2:2ð kTs Þ2
h i Mv Lv
@g @T
3, 6, 8 Surface tension test ðuÞ  @T @x 19
3, 6, 8 Open boundary –
Whole geometry Temperature-dependent surface tension (γ) gðT Þ ¼ 0:64  8:2  105 ðT  Tm Þ 20
and dynamic viscosity (μ) for 2,190–2,500 K h
3
i
μðT Þ ¼ 3:2  exp 43:210
RT (21)

tension is at its weakest. Therefore, the fluid moves from the contact stylus profilometer, contactless optical profilometer,
center to the edges of the molten pool. scanning electron microscopy (SEM), and atomic force
microscopy. However, the selection of appropriate measure-
ment techniques is vital since it predominantly depends on
4 Experimental approach the magnitude and periodicity of surface profiles. In the
contact stylus tracing method, the main requirement for
4.1 Laser machining proper measurement is the size of the stylus tip and stem,
which should be smaller than the roughness of the surface.
In the present study, the JK 701 Lumonics pulsed Nd:YAG On the contrary, the no-contact optical profilometer mea-
(1.064 μm wavelength) fiber optic laser system is used to surement is independent of the size of the stem and the
machine the alumina coupon (50×40×6 mm3; AdValue stylus tip. Conversely, measuring the surface topography
Technology Inc.). The alumina slab has a purity of using SEM requires the cross-sectioning of the sample.
99.6 wt% (<0.1 % SiO2, <0.05 % Fe2O3, and <0.1 % However, it is extremely difficult to cross-section the ce-
R2O, all in weight percent). Table 1 summarized the laser ramic sample (alumina) of finer dimensions (diameter and
machining parameters used in this study. In addition, Fig. 7 depth <1,000 μm) exactly across the center of the crater for
schematically represents the layout of laser-machined accurate measurement. In addition, piled up and resolidified
regions in alumina for various processing conditions. alumina around the crater lip tends to get removed during
cross-sectioning and subsequent polishing, thereby leading
4.2 Measurements of surface topography to inaccurate surface profile (roughness) measurements. In
light of this, the optical profilometer proved to be the most
Generally, there are various standard surface profile appropriate tool to conduct the topographic measurements
(topography) measurement techniques available, such as on the surface and within the crater without missing the
details of irregular pileup and resolidified layer around the
crater lip. Hence, the optical profilometer is employed for
surface profile measurements in the present work.
The NANOVEA® optical profilometer (model PS50)
with an optical pen of 3.5 mm at the scan speed of 1 μm/s
is used for the surface profile measurements. The ISO
25178-2:2012 (Geometric Product Specifications—Surface
Texture: Areal) specifications was followed during the sur-
face profile measurements. In addition, the ISO 4287:1996
standards are utilized for the line profile measurement on the
previously scanned surfaces to measure the amplitude
parameters (Rt, Rp, and Rv). The raw data obtained during
the surface profile measurements using the optical profil-
Fig. 7 Schematic of laser-machined alumina ometer is further post-processed by using the NANOVEA®
Int J Adv Manuf Technol (2013) 68:69–83 77

3D software. In addition, the standard post-processing pro- increases the surface temperature. Likewise, the pulses are
cedure suggested by the inbuilt NANOVEA® template is continuously delivered every 1/f, in seconds, until the end of
followed. The spatial filtering operator with a median the laser cycle at 1 s causes the sequential rise and drop of
denoising filter 3×3 is used to eliminate the outliers (bad temperature. In the present study, each individual laser pulse
points). The surface has been leveled by subtracting the least has constant laser pulse energy (1 J) and laser ON time
square plane from the whole surface area. Later, the median (Pw =0.5 ms). Due to the heat accumulation by the repetition
smoothing filter 7×7 was applied to filter out the noise of multiple laser pulses (Fig. 2b), the maximum surface
(microroughness) and improve the quality of the scanned temperature gradually increases with increase in number of
surface profile. Moreover, the surface was split into its laser pulses (Fig. 9a). But above a certain average laser
waviness and roughness component by using an areal energy density (>7.1×107 J/m2), the material experiences
Gaussian filter with a cutoff of 0.25 mm. Finally, the line the saturation state where the heat losses (natural convective
profile measurement was taken across the crater of the cooling and radiation) are balanced by the given heat input
previously processed sample, as schematically shown in (absorbed laser energy) and do not show substantial rise or
Fig. 8a. Thus, this procedure ensured the minimization or drop of the surface temperature (Fig. 9a). However, the
elimination of error in surface topography measurements. average laser energy density increases with increase in pulse
Further, a total of eight readings are taken for each process- rate which in turn increases the surface temperature
ing condition (8 line profiles×3 locations=24 readings) in (Fig. 9a). The average maximum temperature for all laser
order to achieve higher accuracy in surface profile measure- machining conditions are plotted to analyze the increase in
ments. From each line profile measurement, the roughness surface temperature more accurately (Fig. 9b). It is also
parameters Rp, Rv, and Rt were evaluated (Fig. 8b). observed that the average surface temperature increases with
increase in pulse rate (or average laser energy density). Even
for lowest average laser energy density (3.5×107 J/m2) used
5 Results and discussion in the current study (Fig. 9b), the average maximum tem-
perature (4,065 K) is considerably higher than vaporization
5.1 Prediction of material loss temperature (3,273 K), indicating the higher material loss
due to evaporation.
From step 1, the temperature history for various laser ma- Due to the rapid increase and decrease of surface temper-
chining conditions is extracted from boundary 6 (Fig. 9a). ature (Fig. 9a), the material experiences the higher heating
Due to higher laser energy density (>3.5×107 J/m2) and the and cooling rates (dT/dt) during laser machining. Further-
localized nature of the laser beam, the surface temperature in more, due to repetition of multiple laser pulses on the
a confined volume quickly rises. Self-quenching and heat alumina surface, the material initially experiences a sudden
losses due to external natural convective cooling and radia- increase in temperature which is evidenced by the initial
tion cause the drop in temperature. Both the increase and higher heating/cooling rates (Fig. 10a). However, once the
decrease of the temperature controls the different physical bulk material attains the saturation state, the further
phenomena taking place in the material (heat conduction, increases in temperature in subsequent laser pulses are not
melting, and vaporization). During one-dimensional as high as in the initial stage and caused the gradual drop in
multiple-pulse laser machining, the temperature rises when heating/cooling rates until the laser is switched off. As a
the laser is ON for a time duration of 0.5 ms (Pw) for each result, the heating/cooling rate decreases with increase in
laser pulse, followed by the temperature drop during the number of pulses for all laser machining conditions
laser OFF time ([1/f]−Pw). Based on the pulse rate (f=10, (Fig. 10a). The average value of heating/cooling rate is
20, 30, 40, 50), the laser is turned ON again for a second calculated to represent the corresponding value of heating/
pulse (for duration of pulse width, Pw = 0.5 ms) which cooling rate for each machining condition (Fig. 10b). It can
be observed that the average heating/cooling rate decreases
with increase in average laser energy density (or pulse rate)
for all laser machining conditions (Fig. 10b). Furthermore, it
is noticed that, as the pulse rate increases, the material
remains in the liquid phase for longer time and the surface
deforms continuously during this time.
The level set method is used to predict the crater depth
and width for various laser machining conditions. The crater
Fig. 8 Surface profile measurement: a tracks of line profile measure-
depth and width increases with the increase in average laser
ments and b two-dimensional representation of line profile measure- energy density or pulse rate (Fig. 11). In the present work,
ments along line 5–5 and corresponding roughness parameters the energy distribution of laser beam is in Gaussian, i.e., the
78 Int J Adv Manuf Technol (2013) 68:69–83

Fig. 9 Computational plots: a

maximum temperature as a
function of time and b average
maximum temperature as a
function of average laser energy
density

laser energy is highest at the center of the beam and it can be noticed that the shape of boundary 6 (Fig. 4) in step 1
gradually reduces as it goes towards the edges of the laser is flat, but due to the loss of material, it had changed to
beam. Therefore, the surface temperature is higher at the concave shape. Thus, the new surface geometry of boundary
center of the laser beam compared to the edge. As a result, 6 precisely defines the normal and tangential boundary
the material experiences preferential heat conduction. conditions (heat flux, recoil pressure, and surface tension).
Hence, the penetration of heat inside the material normal Even though the heat transfer and phase change boundary
to the top surface is higher than the lateral direction, causing conditions adopted in step 2 are the same as step 1, the
the higher crater depth compared to crater width (Fig. 11). change in surface geometry of boundary 6 (Fig. 5) strongly
During the repetition of multiple laser pulse, the material influences the corresponding boundary conditions of fluid
experiences a sequential loss of material due to evaporation. flow, heat transfer, and phase change kinetics and, hence,
This material lost also increases with increase in number of the generation of final surface profile.
laser pulses, which directly contributes in increasing the The recoil pressure, surface tension, and gravitational
crater depth. From this data, the updated geometry was then forces are the important factors to generate the hydrodynam-
created and utilized in step 2. ic melt pool motion. During the heating cycle, when the
surface temperature is above the vaporization temperature,
5.2 Prediction of surface topography the recoil pressure is the dominating factor to drive the
molten material out from the crater that further increases
The fluid flow coupled with heat transfer and phase change the crater depth as well as the liquid pileup. On the other
kinetics is incorporated in step 2. Based on the dimension of hand, when the temperature drops below the vaporization
crater depth and width provided by step 1 heat transfer temperature, surface tension and gravitational forces lead to
model, the new geometry (creation of crater) is designed the movement of the molten material and, therefore, liquid
in step 2 to include the material loss due to evaporation. It metal tends to return to its original place. At the end of the

Fig. 10 Computational plots: a

rate of heating/cooling (dT/dt)
as a function of time and b
average dT/dt as a function of
average laser energy density
Int J Adv Manuf Technol (2013) 68:69–83 79

However, in the computational model, the actual recoil

pressure is only applied on the melt pool when the surface
temperature reaches above the vaporization temperature.
Once the material reaches the saturation state, the further
increase in the number of pulses or average laser energy
density does not alter the magnitude of recoil pressure
(Fig. 12). In addition, the depth of the crater increases
drastically due to the increase in average laser energy
density (or pulse rate) and recoil pressure. However, above
a certain critical crater depth (~260 μm), the magnitude of
recoil pressure is not enough to drive the molten material
out of the crater even though the surface temperature is
above the vaporization temperature. Here, the magnitude
of surface tension and gravitational forces dominate over
the higher recoil pressure, causing the drop in velocity of
molten material. Hence, net velocity of the molten mate-
Fig. 11 Measured crater depth and width for various laser machining rial initially increases with increase in average laser ener-
conditions
gy density (3.5 × 10 7 and 5.3 × 10 7 J/m 2 ) (Fig. 13).
However, after reaching a certain critical crater depth
heating cycle, the higher cooling rates and self-quenching (~260 μm), net velocity of molten material drops down
that simultaneously affect the solidifying liquid material and for further increase in average laser energy density (>5.3×
the tangential stresses generated by the surface tension 107 J/m2). The velocity is high at the end of the laser on
forces provide the profile of the solidified material. time due to the maximum temperature and corresponding
The recoil pressure strongly depends on average laser maximum recoil pressure for all the average laser energy
energy density and the instantaneous surface temperature density machining conditions that existed at the end of the
(Eq. 18) and, hence, its magnitude gradually increases heating cycle (Fig. 13). However, as the surface tempera-
with increase in average laser energy density and the ture continuously drops down immediately after the laser
instantaneous surface temperature (Fig. 12). Figure 12 off time until the end of the cycle (1 s), it in turn causes
only shows the numerically calculated recoil pressure dur- the gradual drop in the velocity of the molten material.
ing various laser energy density machining conditions. The computationally predicted surface deformation plots
in cross-section for various laser machining conditions are
presented in Fig. 14a. The present computational model
assumes that the piled up and/or resolidified material is
symmetric around the plane x–x (Figs. 2c and 14b). In this

Fig. 12 Effect of recoil pressure as a function of time for various laser Fig. 13 The velocity profile for various laser energy density machin-
energy density machining conditions ing conditions during the last laser pulse
80 Int J Adv Manuf Technol (2013) 68:69–83

case, the predicted results are revolved around the axis (z- contributes to increasing the Rv value (Figs. 14 and 15).
axis, Fig. 14c) of the crater (along the circumference) and, Although the crater depth increases, the recoil pressure is
hence, a uniform ring-like structure formed along the crater not sufficient to eject the molten material out from the crater,
lip, as illustrated in Fig. 14d. Under this assumption, the resulting in a decrease in the Rp value (Figs. 14 and 15).
cross-sectional view (Fig. 14a) facilitates the measurement Hence, after reaching a certain critical crater depth
of the surface roughness parameters Rp, Rv, and Rt. During (~260 μm), the net velocity of the molten material decreases
the application of each laser pulse, the surface temperature with increase in average laser energy density. On the con-
of the material goes above the vaporization temperature, trary, a longer time necessary for solidification (~2–5 s)
leading to material loss due to evaporation. As the number allows the expelled molten material to solidify on the inside
of laser pulses increases, the material experiences material wall of the crater (known as recast layer) without the possi-
removal which increases the crater depth. In addition to this, bility of returning to its original place which will reduce the
the molten material expelled out by recoil pressure directly pileup height (Rp) (Figs. 8b, 14, and 15). Thus, distinct

a e
10
10 Pulse 3.5x107 J/m2
Pulse
b
a

b
3.5x107 J/m2
Rt =465.8 µm
Rt = 490 µm

20 7.1x107 J/m2
Pulse
20 Pulse
a

b
(b) Cross-se onal view
7.1x107 J/m 2
Rt =559.6 µm Rt = 604 µm
Z
c
30
Pulse
30 Pulse 10.6x10 7 J/m2
a

b
10.6x107 J/m2
Rt =967.5 µm Rt = 898 µm

Z 14.1x107 J/m2
40
40 Pulse
Pulse
a
(c) Revolve around Z-axis
b
Revolve 360° around Z-axis
14.1x107 J/m2 d
Rt = 1030.96 µm Rt = 1010 µm

50 17.7x107 J/m2
Pulse
50 Pulse
a

17.7x107 J/m2
Rt = 1870.06 µm Rt = 1920 µm
(d) Three-dimensional view

Fig. 14 Computationally predicted and experimentally measured sur- the z-axis, d schematic three-dimensional view of predicted surface
face topography (profiles): a computationally predicted surface topog- topography, and e experimentally measured surface topography (ISO
raphy, b schematic cross-sectional view of predicted surface 25178) and one out of eight corresponding line profile measurements
topography, c schematic of predicted surface profile revolved around along line a–b (ISO 4287)
Int J Adv Manuf Technol (2013) 68:69–83 81

dependent material properties [36–38] and boundary

conditions [1, 2, 34, 42, 43].
The present computational model incorporates temperature-
dependent material properties, such as thermal conduc-
tivity, density, heat capacity, absorptivity, viscosity, and
surface tension coefficient. The deviation of these ther-
mophysical properties may play a critical role in gener-
ating the hydrodynamic melt motion of the liquid metal
during the simulation. Furthermore, during laser process-
ing, the crater shape is also constantly changing, which
can vary the angle of incidence of the laser beam. There-
fore, the change in absorbance of the laser energy can
cause the variation in surface temperature and, conse-
quently, the temperature-dependent material properties.
Hence, the more accurate thermophysical properties can
further reduce the difference between computational and
Fig. 15 The change of surface roughness parameters for various experimental values. Moreover, as mentioned earlier, the
average laser energy density machining conditions in situ measurement of variable thermophysical proper-
ties is a very challenging task due to high temperature,
teardrop shape topography is generated on the wall of the very short duration, and very small volume of laser–
crater (Fig. 14a). Among all laser machining conditions, the material interaction. Therefore, very limited data are
highest laser energy density machining condition (10.6×107 J/ presently available in the open literature. In addition,
m2) provides the largest crater depth (Rv =1826.26 μm) with efforts are underway to minimize the difference between
the least material pileup (Rp =43.8 μm) on the surface. This the results of computational model and experimental
combination is responsible for the higher surface roughness observations.
(Rt =1870.06 μm) for higher energy density.

5.3 Validation of computational model 6 Conclusion

In order to validate the computational model, the com- COMSOL™ Multiphysics was used to develop a computa-
putationally predicted and experimentally measured val- tional model to understand the influence of multiple laser
ues of surface roughness for various laser machining pulses on the surface finish of alumina for various laser energy
conditions are summarized and presented in Table 4. It density machining conditions. Results indicate that the mate-
is observed that the surface roughness increases with the rial lost due to evaporation causes an increase in crater depth,
increase in average laser energy density. The difference whereas liquid expulsion created by the recoil pressure
between the computational model and the experimental increases the pileup height. However, after a critical crater
observations are within the range of 2–8 %, thereby depth (~260 μm), the magnitude of the recoil pressure was
strongly validating the computational model. The differ- insufficient to eject the significant amount of liquid material
ence of computational results can be attributed to the out of the crater and hence the liquid material solidified inside
adaptation of various literature-suggested temperature- the crater wall, leading to the formation of a typical teardrop

Table 4 Computationally predicted and experimentally measured surface roughness

Experiment Parameters

Average energy Laser pulses (1/s) Numerically Experimentally measured Difference (%)
density (J/m2) calculated Rt (μm) Rt ±standard deviation (μm)

1 3.5×107 10 465.8 490±5.92 5.19

2 7.1×107 20 559.6 604±3.80 7.93
3 10.6×107 30 967.5 898±5.16 7.18
4 14.1×107 40 1,030.96 1,010±4.84 2.03
5 17.7×107 50 1,870.06 1,920±3.89 2.67
82 Int J Adv Manuf Technol (2013) 68:69–83

shape topography. In this study, it was found that the surface 16. Kannatey-Asibu E (2009) Principles of laser materials processing.
roughness increased with increasing pulse rate (10, 20, 30, 40, Wiley, Hoboken
17. Chryssolouris G (1991) Laser machining: theory and practice.
and 50) or the increase in average laser energy density. A close Springer, New York
agreement and a similar trend were observed between the 18. Salonitis K, Stavropoulos P, Stournaras A, Chryssolouris G (2007)
predicted roughness values of the computational model with CO2 laser cutting of aluminium. Proceedings of the 5th Laser
experimental observations. Assisted Net Shape Engineering pp 825–835
19. Stavropoulos P, Salonitis K, Stournaras A, Euthimiou K, Chryssolouris
G (2007) Molecular dynamics simulation of laser ablation of iron.
Acknowledgments The authors, NBD and HDV, would like to ac- Proceeding of the 10th CIRP international workshop on modeling of
knowledge the financial support from the National Science Foundation machining operations, pp 549–553
(NSF-CMMI 1010494). The authors would also like to thank Dr. 20. Stournaras A, Salonitis K, Stavropoulos P, Chryssolouris G (2007)
Radovan Kovacevic, Southern Methodist University, for providing Finite element thermal analysis of pulsed laser drilling process.
the research facilities for surface roughness measurement. Proceeding of the 10th CIRP international workshop on modeling
of machining operations, pp 563–570
21. Dubey AK, Yadava V (2008) Laser beam machining—a review.
Int J Mach Tools Manuf. doi:10.1016/j.ijmachtools.2007.10.017
22. Yeo CY, Tam SC, Jana S, Lau MWS (1994) A technical review of
References the laser drilling of aerospace materials. J Mater Process Technol.
doi:10.1016/0924-0136(94)90073-6
23. Corcoran A, Sexton L, Seaman B, Ryan P, Byrne G (2002) The
1. Samant AN, Dahotre NB (2009) Laser machining of structural laser drilling of multi-layer aerospace material systems. J Mater
ceramics—a review. J Eur Ceram Soc. doi:10.1016/j.jeurceramsoc. Process Technol. doi:10.1016/S0924-0136(01)01123-2
2008.11.010 24. Li L, Diver C, Atkinson J, Giedl-Wagner R, Helml HJ (2006)
2. Samant AN (2009) Laser machining of structural ceramics: com- Sequential laser and EDM micro-drilling for next generation fuel
putational and experimental analysis. Dissertation or Thesis, Uni- injection nozzle manufacture. CIRP Ann Manuf Technol.
versity of Tennessee doi:10.1016/S0007-8506(07)60393-X
3. Tuersley IP, Jawaid A, Pashby IR (1994) Review: various methods 25. Lawrence J, Low DKY, Pou J, Toyserkani E (2010) Advances in
of machining advanced ceramic materials. J Mater Process Tech- laser materials processing technology. Woodhead Publishing Ltd.,
nol. doi:10.1016/0924-0136(94)90144-9 Cambridge
4. Vikulin VV, Kelina IY, Shatalin AS, Rusanova LN (2004) Advanced 26. Hanon MM, Akman E, Genc Oztoprak B, Gunes M, Taha ZA, Hajim
ceramic structural materials. Refract Ind Ceram. doi:10.1007/s11148- KI, Kacar E, Gundogdu O, Demir A (2012) Experimental and theo-
005-0017-2 retical investigation of the drilling of alumina ceramic using Nd:YAG
5. Shen JY, Luo CB, Zeng WM, Xu XP, Gao YS (2002) Ceramics pulsed laser. Opt Laser Technol. doi:10.1016/j.optlastec.2011.11.010
grinding under the condition of constant pressure. J Mater Process 27. Samant AN, Dahotre NB (2008) Computational predictions in
Technol. doi:10.1016/S0924-0136(02)00636-2 single-dimensional laser machining of alumina. Int J Mach Tools
6. König W, Wagemann A (1993) Machining of ceramic components— Manuf. doi:10.1016/j.ijmachtools.2008.05.004
process-technological potentials. National Institute of Standards and 28. Samant AN, Dahotre NB (2009) Differences in physical phenom-
Technology (NIST): Machining of Advanced Materials Proceedings. ena governing laser machining of structural ceramics. Ceram Int.
45:3–16 doi:10.1016/j.ceramint.2008.11.013
7. Chryssolouris G, Anifantis N, Karagiannis S (1997) Laser assisted 29. Harimkar SP, Samant AN, Dahotre NB (2007) Temporally evolved
machining: an overview. J Manuf Sci Eng Trans ASME 119:766– recoil pressure driven melt infiltration during laser surface modifica-
769 tions of porous alumina ceramic. Journal of Applied Physics.
8. Kacar E, Mutlu M, Akman E, Demir A, Candan L, Canel T, Gunay doi:10.1063/1.2710288. http://jap.aip.org/resource/1/japiau/v101/i5/
V, Sınmazcelik T (2009) Characterization of the drilling alumina p054911_s1
ceramic using Nd:YAG pulsed laser. J Mater Process Technol. 30. Sun S, Durandet Y, Brandt M (2005) Parametric investigation of
doi:10.1016/j.jmatprotec.2008.04.049 pulsed Nd:YAG laser cladding of stellite 6 on stainless steel. Surf
9. Harimkar SP, Dahotre NB (2006) Effect of laser fluence on surface Coat Technol. doi:10.1016/j.surfcoat.2004.03.058
microstructure of alumina ceramic. Advances in Applied Ceramics 31. Islama MU, Campbella G (1993) Laser machining of ceramics: a
105:304–308 review. Materials and Manufacturing Processes. doi:10.1080/
10. Triantafyllidis D, Li L, Stott FH (2002) Surface treatment of 10426919308934870
alumina-based ceramics using combined laser sources. Appl Surf 32. Vora HD, Santhanakrishnan S, Harimkar SP, Boetcher SKS,
Sci. doi:10.1016/S0169-4332(01)00639-0 Dahotre NB (2012) Evolution of surface topography in one-
11. Koepke BG, Stokes RJ (1970) A study of grinding damage in dimensional laser machining of structural alumina. J Eur Ceram
magnesium oxide single crystals. J Mater Sci. doi:10.1007/ Soc. doi:10.1016/j.jeurceramsoc.2012.06.015
BF00551000 33. Anisimov SI (1968) Vaporization of metal absorbing laser radia-
12. Kirchner HP (1984) Damage penetration at elongated machining tion. Soviet Physics J Exp Theor Phys 27:182
grooves in hot-pressed Si3N4. J Am Ceram Soc. doi:10.1111/ 34. Semak V, Matsunawa A (1997) The role of recoil pressure in
j.1151-2916.1984.tb09629.x energy balance during laser materials processing. J Phys D Appl
13. Conway JC, Kirchner HP (1986) Crack branching as a mechanism Phys 30:2541–2552
of crushing during grinding. J Am Ceram Soc. doi:10.1111/j.1151- 35. Quintero F, Varas F, Pou J, Lusquiños F, Boutinguiza M, Soto R,
2916.1986.tb04815.x Pérez-Amor M (2005) Theoretical analysis of material removal
14. Dahotre NB, Samant AN (2011) Laser machining of advanced mechanisms in pulsed laser fusion cutting of ceramics. J Phys D
materials. CRC, Boca Raton Appl Phys 38:655–666
15. Dahotre NB, Harimkar SP (2008) Laser fabrication and machining 36. Kolev NI (2011) Multiphase flow dynamics 5: nuclear thermal
of materials. Springer, New York hydraulics. Springer, Berlin
Int J Adv Manuf Technol (2013) 68:69–83 83

37. Gitzen WH (1970) Alumina as a ceramic material. American 41. COMSOL (2011) Users guide. COMSOL Multiphysics model
Ceramic Society, Columbus library V 4.2a
38. Paradis P, Ishikawa T (2005) Surface tension and viscosity mea- 42. Sussman M, Smereka P, Osher S (1994) A level set approach for
surement of liquid and undercooled alumina by contactless techni- computing solutions to incompressible two-phase flow. J Comput
ques. Jpn J Appl Phys 44:5082–5085 Phys. doi:10.1006/jcph.1994.1155
39. Samant AN, Du B, Dahotre NB (2009) In-situ surface absorptivity 43. COMSOL (2011) Marangoni convection model. COMSOL Multi-
prediction during 1.06 um wavelength laser low aspect ratio machin- physics model library V 4.2a
ing of structural ceramics. Physica Status Solidi A. doi:10.1002/ 44. Pierron N, Sallamand P, Matteï S (2005) Numerical modeling of
pssa.200925108 molten pool formation during an interaction of a pulse laser (Nd:
40. COMSOL (2011) Continuous casting model. COMSOL Multi- YAG) with an aluminum sheet. Proceedings of the COMSOL
physics model library V 4.2a Multiphysics User's Conference, pp 1–5