Int J Adv Manuf Technol (2010) 49:553566

DOI 10.1007/s00170-009-2431-y

ORIGINAL ARTICLE

Nano-second pulsed DPSS Nd:YAG laser cutting of CFRP

composites with mixed reactive and inert gases
Reza Negarestani & Lin Li & H. K. Sezer &
David Whitehead & James Methven

Received: 27 August 2009 / Accepted: 6 November 2009 / Published online: 25 November 2009
# Springer-Verlag London Limited 2009

Abstract Superior structural capabilities and lightweight of

carbon-fibre-reinforced polymer composites have made
their applications increasingly noticeable particularly in
the aerospace and automotive industries for reduced fuel
consumption. Anisotropic and heterogeneous features of
these materials, however, have been prohibiting the
application of laser cutting on these materials in industrial
scale. In the present study the thermal degradation
characteristics in laser cutting of these materials are
investigated with a nano-second pulsed diode pumped solid
state Nd:YAG. A statistical analysis was performed for the
optimisation of the process parameters. Furthermore,
quality improvement was achieved by the use of low
oxygen content assistant gas simultaneously with the inert
gas shield. The controlled presence of oxygen as a burning
mechanism reduced the fibre pull out up to 55% at the same
time with a high processing rate.
Keywords Laser cutting . Carbon fibre composites .
Gas mixture . DPSS Nd:YAG laser

1 Introduction
Carbon-fibre-reinforced polymers (CFRPs), as any other
fibre-reinforced polymeric (FRP) composite, are consisted
of higher strength fibres bonded within a weak and adhesive
polymer. The combination offers a high performance and
R. Negarestani (*) : L. Li : H. K. Sezer : D. Whitehead :
J. Methven
Manufacturing and Laser Processing Group,
School of Mechanical, Aerospace and Civil Engineering,
The University of Manchester,
Manchester, UK
e-mail: rnegarestani@gmail.com

extremely light material. At the same time, however, this

mixture remains heterogeneous since each of the constituents
retain their individual physical properties. Machining of these
materials (especially cutting and drilling) is of interest to
produce intricate shapes with desirable tolerances mainly for
fastening and profiling applications. Nevertheless, in conjunction with anisotropic properties of the material which
depend upon the fibre orientation and laminar lay out, the high
mechanical strength and thermal resistance of carbon fibres
lead to significant challenges for conventional [1] and
unconventional [2] machining of these materials.
Lasers as a non-contact, fast, precise and flexible tool
have been successfully used for cutting both metallic and
non-metallic materials. However, laser cutting of CFRPs
composites show particular difficulties due to the anisotropic and heterogeneous properties of these materials [3].
The challenges are produced mainly by the variations in
thermal expansion coefficients of carbon fibres in the radial
and longitudinal directions [3] and considerable difference
in thermal properties of the fibres and the polymer matrix
[4] (see Table 1 [3, 5, 6]). Consequently, laser processing of
these materials generally result in severe thermal damages
such as heat-affected zone, pull out of fibres, laminar
delamination and fibre end swelling (Fig. 1).
Ablation and photochemical reactions in the laser
processing of composite materials using a UV beam have
been reported to considerably reduce thermal damages [7
10]. However, low material removal rate and lack of
flexibility are the common drawbacks for such systems
limiting their application in industrial scale. CO2 lasers (IR
beam) have also been used to investigate laser cutting of
CFRPs both in CW [4, 11] and pulsed mode [12]. However,
pulsed Nd:YAG laser (as another IR beam laser) has been
reported to give less thermal damage due to pulse-off
cooling [3]. Lau et al. [13] studied the quality factors in

554

Int J Adv Manuf Technol (2010) 49:553566

Table 1 Thermal and physical properties of the CFRP composite

used in the study [3, 5, 6]

Volume fraction
Density (kg/m3)
Thermal conductivity (W m1 K1)
Specific heat (J kg1 K1)
Vaporisation temperature (K)
Thermal expansion coefficient (m m1 K1)

Fibre

Epoxy

60%
1,800
50
710
4,000
0.5 L
5 T

40%
1,200
0.1
1,884
700
65

response to the effect of different process parameters on the

feature quality in laser cutting of composite materials using
a pulsed Nd:YAG system. They demonstrated the effectiveness of pulse width and the cooling gas on the quality.
Mathew et al. [14] provided optimised process parameters
in the laser composite cutting process. Despite the noted
improvements in the previous studies, achieved machining
quality does not satisfy the required specifications given in
[11], which suggests the acceptable extent of fibre pull out
as less than 150 m and the kerf width close to the beam

spot diameter with no fibre swelling. Thus further improvement is required for any possible practical applications.
Current work introduces a novel laser cutting approach to
achieve this. The technique is based on controlling the heat
input simultaneously with in-process cooling mechanisms
incorporated in the laser-cutting process using a mixture of
oxygen and nitrogen assisting gases as reported here for the
first time. Active O2 gas enhances the decomposition
process of the composite material by exothermal reaction
resulting in higher material removal rates (MRR) while the
inert N2 gas reduces the thermal damages by enhancing
thermal cooling during the process.
Assistant gas is an important process factor and, it can
affect the processing results in Nd:YAG laser cutting of
CFRPs [13]. Controlled mixture of oxygen and nitrogen
gases carried out in this study aims to utilise the combined
positive effects for improving the quality of the cut whilst
keeping the MRR at reasonable levels. Laser-cutting tests
with the developed technology is carried out using a high
power diode pumped solid state (DPSS) Nd:YAG laser
system. This system offers more reliability, higher efficiency,
narrower frequency linewidths and higher peak powers as
compared to arc lamp pumped laser systems used in previous
studies [13].

Fig. 1 SEM images of typical quality defects in laser cutting of CFRP composites a large heat affected zone, b fibre pull out, c delamination
between two lamina, d fibre end swelling

Int J Adv Manuf Technol (2010) 49:553566

2 Relationship between CFRP decomposition

and assist gas
Typically, epoxy structures are produced by combining a
resin and a hardener. During polymerization of the liquid
resin and the hardener, the cross-links form amorphous
structures. Therefore thermosetting epoxies (being heavily
cross-linked and amorphous) do not show true melting or
viscous flow upon heating and once exposed to excess
heating (e.g. during laser machining) they will decompose
[15]. Generally, thermolysis (i.e. thermal decomposition) of
the epoxy resin consists of preheating and decomposition.
Decomposition usually starts with dehydration and thereafter
chain scissions occur due to reduction in thermal stability of
other bonds e.g. CO and CN, with the heat absorption [16].
The resulting thermolysis products contain light gases,
various hydrocarbons and char.
Carbon fibres on the other hand are highly crystalline
structures consisting of high content of carbon e.g. above
95%. Because of the high bonding energies (of various
carbon atom to atom bonds) at low atmospheric pressures
(such as the current study) carbon elements undergo direct
vaporisation (degrade directly from solid to the gas phase)
at temperatures round about 4,000 K [17].
Control of thermal degradation and hence damage of
CFRPs in thermal processing is best observed in inert
atmosphere [18]. The decomposition of the matrix occurs at
relatively much lower temperatures (around 700 K) as
compared to the fibre vaporisation temperatures (4,000 K).
Considering higher thermal diffusivity of the carbon fibre
and rapid heatingcooling rates involved in pulsed laser
cutting, quicker but controlled decomposition of fibres in
particular, can lead in less thermal damage to the cut

555

surface through conduction. In oxidative medium, the

decomposition of fibres [18] as well as the epoxy matrix
[19] would be enhanced with the heat released from the
exothermic reactions. Oxidative decomposition of CFRPs is
mainly influenced by the carbon fibre oxidation in the form
of following two exothermal reactions [18].
C O2 ! CO2

1
C O2 ! CO
2

Figure 2 illustrates thermal gravimetric analysis (TGA)

and derivative thermal analysis (DTA) for the material used
in this study (60% carbon fibre40% epoxy polymer) in
oxidative and inert medium at two different heating rates.
As it can be observed from Fig. 2a, the material decomposes quicker in air (through oxidation of fibres) as
compared to nitrogen medium. From Fig. 2b, it is clear
that in nitrogen the decomposition shows only one weight
loss peak representing devolatilisation (at around 673 K). In
air on the other hand, a different decomposition mechanism
is evident through more stages of weight loss mechanism
(i.e. devolatilisation, char oxidation and then fibre oxidation
at around 1,073 K). It can also be seen from the figure that,
although increasing the heating rate decreases the weight
loss, the difference between oxidative and non-oxidative
environment is still valid. Therefore, the observed difference
can still be expected to be valid at even higher heating rates as
in laser processing.
Therefore, although presence of reactive gas can deteriorate
the quality (through excessive degradation) in laser processing, once controlled, can be useful for compromising effective

Fig. 2 a TGA and b DTA of the material used in experiments in inert and oxidative mediums at two different heating rates

556

Int J Adv Manuf Technol (2010) 49:553566

material removal and reduced thermal damages. Thereby,

mixing oxygen into the inert nitrogen assist gas is investigated
in this work. Presence of oxygen as a reactive medium can
enhance: (1) diffusion at elevated temperatures and (2)
chemical decomposition i.e. oxidation, of fibres. The nitrogen
is on the other hand more effective to dissipate the heat and
hence reducing the thermal damages. Properties of the oxygen
and nitrogen gases are provided in Table 2 [20, 21].

3 Experimental procedure
A 400-W Powerlase DPSS Nd:YAG laser was used in this
study. The beam is non-polarised with 1,064 nm wavelength and 350 m focused spot diameter. High repetition
rate of 3 to 15 kHz and short pulse duration (i.e. 2847 ns)
distinguishes the system from the millisecond pulsed Nd:
YAG lasers used in most of the previous CFRP laser cutting
studies [13, 14].
The material used in the experiments was 1.2 mm thick
fully cured 0/90/0/90/0CFRP lamina. The volume
fraction of the carbon fibres (7 m in diameter) is 60%
and the resin is E-765 Epoxy by nelcote. The samples
were clamped on an Aerotech 3-axis CNC stage with
maximum transmitting speed of 200 mm s1. Multiple pass
strategy was used for laser cutting of 18 mm slots and
5 mm outside allowance was considered for the stage
acceleration purpose. A schematic view of the experimental
set up is given in Fig. 3. The figure shows the double gas
jet inlets on the laser head used to feed the oxygen and
nitrogen into the nozzle. The assist gas flow is hence
coaxial to the laser beam.
3.1 Process parameters
As the first step, design of experiments (DoE) approach
was used to adopt optimum process parameters for
conducting the tests on mixture of oxygen in the gas
assisted. Response surface methodology (RSM) based on
central composite design (CCD) was applied. The RSM is a
collection of mathematical and statistical techniques used to
establish the relationships between a response of interest
and the independent variables of the process. The CCD is
Table 2 Properties of oxygen and nitrogen [20, 21]

Density (kg m3)

Thermal conductivity (W m1 K1)
Viscosity (106 N s m2)
Specific heat capacity (J Kg1 K1)
Heat of vaporisation (kJ kg)

O2

N2

1.30
0.0268
2.06
920
255.39

1.14
0.0265
1.78
1,042
208.21

Fig. 3 Schematic view of the experimental set up

the most commonly used RSM to study the quadratic

effects besides main and the factor interactions. Design
Expert software was used to generate the CCD for three
numerical factors (i.e. pulse frequency, pulse energy,
cutting speed) with three levels and three replicates for
each experiment. The design factors and the levels are
given Table 3. The range of the process parameters was
confirmed following a number of screening tests which
showed multiple pass cutting with low energy pulses
provides better quality as compared to high-power single
pass cutting. Number of passes for different parameter
combinations was optimised so as to get through cut in all
cases. Finally, the assist gas was nitrogen delivered at
8 bars through a 1-mm exit diameter converging nozzle.
The obtained optimum process parameters were used to
study the effect of pressure in mixing oxygen in the assist
gas.
3.2 Assist gas
The main objective of the current work is to optimise the
burning rate by varying the oxygen partial pressure in the
assist gas so that thermally induced damage in laser cutting
of CFRPs can be eliminated/reduced. Hence, following the
confirmation of the laser parameters, firstly the effect of
assist gas pressure was studied involving pure oxygen, pure
nitrogen and 50% O250% N2 assist gas at different
Table 3 Process parameters ranges in DoE
Name

Unit

Frequency
Pulse energy
Cutting speed

kHz
mJ
mm s1

Minimum

Maximum

3
7
50

7
25
200

Int J Adv Manuf Technol (2010) 49:553566

557

pressures. In laser metal processing assisted with an inert

gas, increasing the gas pressure increases the drag force by
gas flow on the cut front. When cutting with oxygen gas,
the exothermic-reaction-induced burning rate would also be
influenced by the gas pressure besides the enhanced drag
force [20]. As the final stage of experimental work, the
oxygen volume fraction in the assist gas was then analysed.
The results were evaluated by using optical microscopy
to assess the kerf width and fibre pull out both at the beam
entrance side (referred as top) and beam exit side (referred
as bottom). This was carried out using Polyvar optical
microscope with PC interface via a 12 Mega pixel camera
into I-solution software. The MRR and taper angle which
are interrelated with these factors were also studied. The
analysis of MRR was for the productivity interest while
taper angle is a suitable quality response to study the
variation of the two other factors i.e. kerf width on top and
bottom. The fibre pull out on the top surface is a benchmark
in recognising thermal damages to the material during laser
processing. Ideally, this extent should reach zero to
minimise the mechanical failure in service life. Hence, the
quantitative quality criteria aimed in this study (as in [11])
include reduction of top fibre pull out below 150 m and
top kerf width close to the beam spot diameter. By
analysing the bottom thermal damages comparative ratios
between the fibre pull out and kerf widths, both on top and
bottom, were defined and applied to analysing the variation
of the thermal damage inside the kerf.

significance of the coefficients. The order of the model was

adjusted to neglect the insignificant terms. Once the model
is suitably reduced the normal plot of residuals was
analysed to ensure that the model assumptions are not
violated. In all cases, the residuals were found to follow a
normal distribution, indicating the model was appropriate
for the data set.
Generally, the DoE analysis showed that the fibre pull
out on the beam entrance (i.e. top surface) is the major
quality defect and mostly influenced by the pulse energy.
Figure 4a shows the variation in the extent of top surface
fibre pull out with change in the pulse energy. On the other
hand, pulse frequency was identified as the most effective
factor for the fibre pull out at the beam exit (i.e. bottom of
the kerf; Fig. 4b). The kerf widths at the beam entrance and
the exit were the other quantitatively analysed quality
factors, and also found mostly sensitive to change in the
pulse frequency. Additionally, the cutting speed and pulse
energy were also found significant for the kerf widths at
the top and bottom surfaces, respectively. The 3D view of
the combined factors effects are shown in Fig. 4c and d
for the kerf widths at the beam entrance and beam exit,
respectively.
The MRR and the taper angle of the kerf walls were also
considered in the model. MRR was calculated according to
the number of passes and the kerf geometry (Fig. 5) at the
known scanning speed by Eq. 4 and the taper angle was
calculated using Eq. 5 based on the kerf cross section
geometry (Fig. 5).

4 Results
MRR

4.1 DoE analysis

Statistical models were built using linear regression
analysis to relate the quality responses (i.e. fibre pull out
on the top and bottom surfaces, top and bottom kerf widths,
taper angle and the material removal rate) to the design
factors given in Table 3. Quadratic model were chosen in
all cases. The model for the given response (r) is
represented as [22]:

hr b 0

k
X
j1

b j xj

k
X
j1

bjj x2j

XX

bij xi xj

ij

Where 0 is the response at the centre of experiment, j

is the coefficient of main effects, jj is the coefficient of
quadratic effects and ij is linear by linear interaction
effects. The regression coefficients given in the equation
were calculated using the least squares method and then
finalised by stepwise regression technique. A complete
analysis of variance technique was used to identify the

Removed volumecm3

process timemin

Wa Wb
2

 d  lc  109
nlc
60V

1

q tan

Wa  Wb
2d


5

Where, Wa (m) and Wb (m) refer to the kerf widths at

the top (beam entrance) and the bottom (beam exit) side
respectively, d (m) is the sample thickness, lc (mm) is the
cut length, n is the number of passes cutting through,
V (mm s1) is the scanning speed and (Rad) is the taper
angle. The significance of the frequency effect was once
more observed for both MRR and taper angle responses.
The change in the material removal rate with the pulse
frequency is shown in Fig. 6a. Effect of the pulse frequency
and the scanning speed on the taper angle is presented in
Fig. 6b.
In addition to the above analysis, the overall thermal
degradation was characterised using the ratio of the extent

558

Int J Adv Manuf Technol (2010) 49:553566

Fig. 4 Modelled influence of significant factors (in assistance of 8 bars N2) affecting: fibre pull out on top (a) and bottom (b), Kerf width on top
(c) and bottom (d)

of fibre pull out and the cut width at the top and the bottom
of the workpiece by defining two ratios as:
Ri

fi
Wi

fi : a; bg

and,
R'

Rb
Ra

Fig. 5 Kerf geometry used to

calculate the taper angle and
MRR

Where Ri is the ratio, fi (m) is the fibre pull out,

Wi (m) is the kerf width and a and b indices refer to top
and bottom surfaces, respectively. R is the arbitrary ratio
between the bottom and top ratios. Analysis then revealed
that the bottom ratio exceeded the top ratio (i.e. R1) in
most cases. This could be explained by excessive heat
accumulation towards the beam exit. Hence, for optimisation purpose, R was also incorporated in the statistical
model. Frequency yet again showed the most significant

Int J Adv Manuf Technol (2010) 49:553566

559

Fig. 6 Modelled influence of significant factors on a material removal rate and b taper angle

influence on R. The interaction of the pulse frequency and

the energy that according to the model was of secondary
importance is given in Fig. 7.
Based on these responses, the optimization was carried
out in the software for the thermal defects of the cut (i.e.
minimising the fibre pull out on the top and bottom
surfaces) and the geometry and processing time (i.e.
minimising taper angle and maximising MRR). The
optimum solutions predicted by the software are given in
Table 4. Therefore, by conducting further confirmative trail
tests, pulses of 7 mJ energy duration delivered at 5 kHz
frequency and with 125 mm s1 scanning speed was
identified as optimum process parameters. The optimised
parameters were used for the assist gas effect analysis
presented in the following sections.

4.2 Gas effect

4.2.1 Gas pressure effect
The first series of experiments were conducted to confirm
the optimum pressure for the assist gas. Here, cutting
with the mixture of O2 in the N2 assist gas at a constant
ratio (i.e. half by half proportions) was compared to pure
nitrogen and pure oxygen with three repetitions for each
experiment. The effects on various quality factors are
plotted in Fig. 8. It can be observed that the process
performance is improved at higher pressures in general. The
fibre pull out on both the top and the bottom surfaces was
considerably reduced at 8 bars (Fig. 8a and b) and the MRR
was maximum at this pressure for the assist gas (Fig. 8c). It
was then decided to use high level gas pressure (i.e. 8 bars)
in the following experiments that were conducted with a
range of oxygen volume fractions in the nitrogen assist gas.
4.2.2 Effect of oxygen volume fraction in the assist gas

Fig. 7 Modelled influence of significant factors on the bottom to top

ratio, R

To investigate the influence of oxygen content in the inert gas

shield, the oxygen and nitrogen gas mixture was studied over
the 0100% range at 12.5% intervals. Total pressure of the
assist gas was kept constant at 8 bars in all cases and
the experiments were repeated for three time at each level. The
various measured quality factors at varying oxygen levels are
plotted in Fig. 9. As it can be observed, a low content of
oxygen (i.e. in the range of 12.5%) leads to considerable
reduction of fibre pull out both at the beam entrance and the exit
showed a minimum in this range. It can also be observed that
variation of oxygen volume fraction in the assist gas influences
the kerf width both on the top and the bottom surfaces.

560

Int J Adv Manuf Technol (2010) 49:553566

Table 4 Optimum solutions predicted by the statistical DoE analysis

Solution

Process parameter
Pulse energy
(mJ)

1
2
3

Responses

Scanning speed
(mm s1)

7
10.25
7

125
200
50

Frequency
(kHz)

5
5
5

Thermal defects
Fibre pull out
at the beam
entrance (m)

Fibre pull out

at the beam
exit (m)

Taper
angle ()

Processing
time
MRR
(cm3 min1)

197.92
250.83
273.95

146.54
173.19
88.54

2.45
2.13
2.61

0.134
0.129
0.183

Figure 10 illustrates calculated quality factors variation in

response to the oxygen level. As shown in Fig. 10a, the MRR
generally increased with the increase in oxygen volume
fraction. From Fig. 10c where the extent of fibre pull out is
compared to the kerf width, a minimum is observed over that
oxygen content range. Although the taper angle showed an
elevated value for this range according to Fig. 10b, this could
be neglected since the overall kerf geometry was narrower as
a result of a decrease in the bottom and top kerf widths
(Fig. 9). Therefore (from Figs. 9 and 10), the optimum
results in terms of both quality and productivity of the laser
cutting CFRP materials were obtained for the mixture of
Fig. 8 Variation of different
responses corresponding 50%
O250% N2 assist gas and their
individual usage at
different pressures a top fibre
pull out, b bottom fibre pull out
and c material removal rate

Desirability
Geometry

0.88
0.86
0.81

12.5% oxygen and 87.5% of nitrogen assist gas with the total
pressure is 8 bars. The microscopic images of the result in
this set are compared to the pure oxygen and also pure
nitrogen gas streams in Fig. 11.

5 Discussions
5.1 Statistical analysis
As mentioned, the investigation was preceded by a
statistical analysis aimed for finding the best possible

Int J Adv Manuf Technol (2010) 49:553566

561

Kerf width at beam entrance side

Kerf width at beam exit side
Fibre pull out at beam entrance side
Fibre pull out at beam exit side

320
280
240

(m
m)

200
160
120
80
40
0
0

12.5

25

37.5

50

62.5

75

87.5

100

Oxygen volume fraction (%)

Fig. 9 Influence of oxygen content in 8 bars balanced assist gas with

nitrogen on fibre pull out and kerf widths on the beam entrance and
beam exit

system parameter combination for improved quality. The

significant process factor for beam entrance fibre pull out
was the pulse energy. From Fig. 4a, the fibre pull out
increases with the increase of pulse energy for a given
frequency and scanning speed. The effect is directly related
to the increased heat input which is conducted through the
fibres (due to high thermal conductivity of fibres) leading to
Fig. 10 Influence of oxygen
content in 8 bars balanced assist
gas with nitrogen on a material
removal rate, b taper angle and
c fibre pull out to kerf width
ratio on top Rt and bottom Rb

matrix recession. On the bottom surface the frequency was

the dominant factor. The bottom fibre pull out showed a
minimum for the middle value i.e. 5 kHz of the studied
frequency range. This can be explained by effective MRR
with sufficient pulse-off time i.e. thermal cooling interval.
For lower frequencies since the MRR generally decreases
(Fig. 6a) more passes are required to cut through the
material. This increases the heat input as well as heat
accumulation towards the beam exit side. The excessive
heat from the extra passes necessary for a through cut
increases the kerf width on the top surface rather than
extending the fibre pull out. Therefore, the taper angle and
the bottom fibre pull out show similar trends in response to
pulse frequency as in, Figs. 6b and 4b, respectively. For the
higher frequencies on the other hand, the bottom fibre pull
out also increases due to increased power irradiance and the
decreased interval thermal cooling.
For the top kerf width the statistical analysis predicted
the frequency as the significant factor. The factor interaction between frequency and the scanning speed was also
found significant in this case. As shown in Fig. 4c the top
kerf width slowly decreases with an increase in pulse
frequency from 3 to 5 kHz. From the same figure it can be
observed that the top kerf width shows a proportional and
slightly bending relation to the scanning speed. Here,
interaction time increases due to the decreased MRR at

562

Int J Adv Manuf Technol (2010) 49:553566

Fig. 11 a Top view and b bottom view of the laser cut kerf at 5 kHz frequency, 7 mJ pulse energy and 125 mm s1 in assistance of: (i) 8 bars pure
oxygen (66 passes), (ii) 8 bars 12.5% oxygen mixed with 87.5% nitrogen (72 passes) and (iii) 8 bars pure nitrogen (78 passes)

higher scanning speeds and subsequent increase in number

of passes for a through cut [23]. This together with the nonpolarised beam of the system can be attributed to increase
in the top kerf width at higher scanning speeds. The effect
is in fact opposite for the single pass cutting where the top
kerf width generally reduced with the increasing scanning
speed [14]. Overall, as observed from the contour lines, the
top kerf width shows a minimum value for the parameter
combinations of 5 kHz frequency and 125 mm s1 scanning
speed. Pulse frequency and energy as well as their
interactions were the significant factors for the bottom kerf
width. The model for the bottom kerf width showed a
minimum for lower bound of the interaction of these factors
(i.e. 3 kHz and 7 mJ) with a slightly bending and
continuous proportionally increase up to the higher bond
(i.e. 7 kHz and 25 mJ) as in Fig. 4d.
Material removal rate showed a proportional relationship
with the pulse frequency which was found to be a
significant process parameter (Fig. 6a). More effective
pulses that interact with the material as well as less pulseoff time elevate the MRR. Taper angle on the other hand,
was also most sensitive to the pulse frequency. The second
influencing factor was the interaction between the pulse
frequency and the scanning speed. As depicted in the
modelled relationship in Fig. 6b, more frequent laser
interactions at higher scanning speeds would result in less
taper angle. This can be justified first by less number of
passes that would be required due to increased power

irradiance with the increase in pulse frequency and

secondly by decreased interaction time due to high scanning
speeds. These effects together provide a mechanism in which
the kerf widths on top and bottom get a closer extent to each
other that reduces the taper angle. However, this could not
achieve optimisation as the kerf widths at the top and the
bottom increase at higher frequencies (see Fig. 4c and d).
From the contour lines in Fig. 6b it can be seen that the taper
angle shows a reduction for the ranges adjacent to 5 kHz
frequency and 125 mm s1 scanning speed. The effect is
mostly dominated with the top kerf width which showed
similar trend to these two factors (Fig. 4c).
As mentioned earlier, the other quality response that was
statistically analysed in the optimisation series of experiments was ratio R, which gives the relationship between
the fibre pull out to the kerf width ratio on the bottom
surface to that of the top surface (see Eq. 7). Pulse energy
and the frequency were modelled to have the most
significant influence on R. As it can be observed from
Fig. 7, R generally increased with reducing both the pulse
frequency and the energy. More gradient was pronounced
on the lower half of the studied ranges of these process
parameters while from higher bound of these factors (i.e.
7 kHz frequency and 25 mJ) the pulse energy towards the
middle of the graph the trend descended slightly. The upper
half levels of these factors showed less R, due to a more
pronounced thermal input (i.e. high pulse energy) with
more frequent pulses. This in turn increase the fibre pull out

Int J Adv Manuf Technol (2010) 49:553566

563

however, it would take the whole remaining 30% of the

process to open the bottom kerf width. The heat accumulation imposed by long processing period of the lower end
of the kerf, would affect the bottom side thermal damage.
The fibre pull out on the top surface as well as the kerf
width are also affected during processing of the lower
section of the kerf. It can be reasoned from Fig. 12a that
nearly 55% of the top fibre pull out occurs during the last
30% of the process. Furthermore, the difference in thermal
expansion behaviour of different lamina (depending on the
direction of fibres in each laminate [24]) causes delamination
effect at the edge along the lamina in which the fibres lye
transversely to the beam path.

on the top and bottom surfaces (see Fig. 4a and b) which is

not desirable. In general low pulse energies at intermediate
frequency i.e. 5 kHz could achieve moderate values of R.
5.2 Thermal degradation development
The development and analysis of R ratio, Eq. 7, revealed
that for nearly 60% of the experiments, the fibre pull out to
kerf width ratio at the bottom exceeded the top ratio i.e.
R1. Detailed analysis of the degradation mechanism
throughout the process is illustrated in Fig. 12. From
Fig. 12a it can be observed that an unsteady thermal
degradation development occurs throughout the process.
This could be attributed mainly to the considerable beam
divergence inside the kerf since the focal plane was set on
the top surface. Beam divergence as well as the beam
scattering inside the kerf reduce the effective beam intensity
and hence increase the number of passes required to process
the lower section of the kerf. This is shown in Fig. 12b
where about 80% of the kerf depth is processed in nearly
55% of the whole process. After the first 70% of the
process time, the beam approaches the bottom side;

The effect of gas pressure analysis showed that the top fibre
pull out reduces considerably at high levels of the gas
pressure (i.e. 8 bars) of the studied range (Fig. 8a). Oxygen,
because of the added energy of the exothermic reactions,
showed the highest fibre pull out on the top surface for the
lower pressures. For 8-bar pressure, however, the top fibre

44 Passes

28 Passes

56 Passes

68 Passes

78 Passes

Beam En tran ce

200 m

200 m

200 m

360 m

200 m

360 m

200 m

360 m

Depth of Cut

360 m

360 m

Delamination at the
edge of cut kerf

1600
1200
44

800

56

68

Beam Exit

Cut kerf depth (m)

Fig. 12 a Thermal degradation

development on beam entrance,
kerf depth and beam exit
sections; b kerf depth
development, at 7 mJ pulse
energy delivered at 5 kHz
frequency and 125 mm s1
scanning speed and using 8 bars
nitrogen gas

5.3 Gas pressure effect

78

28

400
200 m

0
0 10 20 30 40 50 60 70 80 90
Number of passes

200 m

200 m

564

pull out in case of oxygen was reduced considerably and

was only slightly more than that of the nitrogen. The
difference between the two can be attributed to higher heat
capacity of nitrogen as well as its inert behaviour while
their closeness is mainly due to their close thermal
conductivities which would make them showing similar
heat dissipation i.e. thermal cooling at elevated velocities. It
also implies that for given process parameters the increase
in oxygen gas pressure beyond a certain level, does not
enhance oxidation and hence the thermal damage. From the
same figure it is evident that the mixture of oxygen and
nitrogen gases showed an acceptable effect to top fibre pull
out trend of each individual gases. The top fibre pull out
was on a close trend to that with pure nitrogen, while the
material removal rate had improved remarkably compared
to pure nitrogen (Fig. 8c). On the bottom side the fibre pull
out showed (Fig. 8b) similar trends to that for the top
surface.
5.4 Oxygen content effect
A mixture of gases was found to improve the machining
quality due to controlled combustion, shear and cooling at
the cutting front. In oxidative decomposition, presence of
free radicals such as O and OH and the exothermal heat
reduce thermal stability of material which in turn reduces
the reaction temperature as well as the activation energy
[19]. The oxygen content analysis (see Fig. 9) showed that
the composition gas behaviour was more like pure active
gas for the ranges between 50% to 100% (referred as upper
range) while the high content of nitrogen in 0% to 50% O2
range (referred as lower range) resulted in responses with
closer behaviour to inert gas. As in Fig. 10a, the
proportional increase trend of material removal rate showed
slightly descending trend at 50% and 100% oxygen content
levels. Since nitrogen has lower heat of vaporisation
compared to oxygen (see Table 2), their mixture is prone
Fig. 13 Influence of presence
of a pure nitrogen and b pure
oxygen on decomposition of
matrix on the beam exit side;
scanning speed, 125 mm s1;
pulse energy, 7 mJ; pulse
frequency, 5 kHz and gas
pressure, 5 bars

Int J Adv Manuf Technol (2010) 49:553566

to more effective thermal degradation control as compared

to pure oxygen case. This together with nitrification of the
epoxy matrix in presence of nitrogen [19] could justify the
slight descending trend in MRR at 100% O2 while at 50%
O2 content the higher heat capacity of the equally weighted
nitrogen, partially absorb the exothermic heat and hence
slightly reduce the oxygen effect.
As illustrated in Fig. 9, fibre pull out on the top and the
bottom sides at 12.5% oxygen content was reduced
remarkably compared to pure nitrogen content. This is
because the reaction temperature decreases in presence of
oxygen [19] and hence more effective decomposition/
vaporisation is observed. This particular trend at 12.5%
oxygen content is in agreement with [19] where 10% O2
content was reported to show an inferior devolatilisation
temperature (503 K) and a peak decomposition temperature
(959 K) as compared to 5% and 20% oxygen content for
the epoxy.
Similarly from Fig. 9, the minimum bottom kerf width is
observed at 12.5% oxygen content level. The bottom kerf
width generally increased proportional to the oxygen content
on a slight ascending trend. The taper angle was generally
showing smaller values for the upper range of oxygen content
(Fig. 10b) due to higher material removal in presence of
more oxygen. Figure 10c, on the other hand, showed that the
fibre pull out to kerf width ratio on top i.e. Ra, and bottom
i.e. Rb, exhibit their minimum value at the 12.5% oxygen
content similar to the overall trends of Fig. 9. Ra and Rb
generally showed descending trend for the upper oxygen
content range which could be contributed to the less number
of passes that are used in this region. It can also be observed
from Fig. 10c that Rb values are always greater than the Ra
values that agrees with the modelled R trend (Fig. 7) where
for 7 mJ pulse energy and 5 kHz frequency at 125 mm s1
scanning speed the Rb was predicted to exceed Ra.
Oxidative decomposition/vaporisation of CFRPs occurs
at 9731,073 K [25] which is much less than the vapor-

Int J Adv Manuf Technol (2010) 49:553566

isation temperature of carbon fibres in inert atmosphere i.e.

4,000 K [17]. Since the oxygen content in the assist gas is
controlled in the current approach, excessive thermal
damage to the material (by thermal conduction along the
fibres) was prevented and hence the cutting quality
improved. The low oxygen content in the assist gas (i.e.
12.5%) balanced with nitrogen, benefits the accelerated
oxidation at a light and comparatively higher viscosity gas
flow which exhibits good heat capacity (cooling effect).
These embody an effective material removal mechanism
with sufficient heat transfer properties that would lead to
considerable reduction on the thermal damages on the beam
entrance and exit surfaces. As it can be seen from Fig. 11,
the optimum mixture (i.e. 12.5% oxygen with 87.5%
nitrogen), resulted in a very narrow heat-affected zone on
the top, i.e. 70 m, which represent a 54% improvement
compared to pure nitrogen case and a 55% improvement
compared to pure oxygen case. On the bottom side the
quality improved is 47% (i.e. reduced fibre pull out) as
compared to pure nitrogen and 59% as compared to pure
oxygen. Although top kerf width did not show remarkable
variation with pure oxygen or nitrogen, the bottom kerf
width was improved by 19% and 41% as compared to pure
nitrogen and oxygen, respectively.
The influence of the presence of oxygen on acceleration
of the decomposition of CFRPs was particularly visible on
the bottom surface of the cuts. Figure 13 presents a
comparison between the assistance of pure oxygen and
pure nitrogen at the same process parameters. As depicted
in Fig. 13a, in presence of pure nitrogen due to the inertness
of the gas a large heat-affected zone is pronounced (darker
colour surrounding the fibre pull out) which represents the
matrix that has been thermally affected but not enough to
be totally decomposed. In case of pure oxygen (Fig. 13b),
however, this region is eliminated as the oxygen would
accelerate the decomposition of thermally affected matrix.
This is in agreement with previous finding [19] that showed
the epoxy char residue can be reduced significantly (around
2.1%) in presence of oxygen as compared to pure nitrogen
(12.417.9%).

6 Conclusions
Large differences between the thermal properties of carbon
fibres and polymer matrix has brought up major challenges
in laser machining of CFRPs. High thermal conductivity of
fibres particularly leads to large extent of matrix recession
around the cut path. Statistical analysis predicted low pulse
energy at the intermediate level of pulse frequency and
medium to high scanning speeds to provide the optimum
possible results. Furthermore, a monitored mixture of
oxygen into the inert gas was investigated in order to

565

accelerate the vaporisation/decomposition process to reduce

the thermal damage since the oxidative thermolysis of the
material occurs at much lower temperatures (9701,070 K)
as compared to the vaporisation temperature of fibres
(4,000 K). The analysis revealed that low volume fraction
of oxygen (i.e. typically 12.5%) mixed with nitrogen gas
and total assist gas pressure of 8 bars are the optimum
parameter configurations to improve the machining quality
in laser cutting of CFRPs. The fibre pull out was reduced
by nearly 55% resulting fibre pull out of 70 m on the top
surface. The involved complexities in the material properties,
material thermolysis and other embodied features such as
paramagnetic characteristic of the oxygen gas welcome
further investigation into such a process.

References
1. Abrate S, Walton DA (1992) Machining of composites. Part I:
traditional methods. Compos Manuf 3(2):7583
2. Abrate S, Walton D (1992) Machining of composite materials.
Part II: non-traditional methods. Compos Manuf 3(2):8594
3. Fenoughty KA, Jawaid A, Pashby IR (1994) Machining of
advanced engineering materials using traditional and laser techniques. J Mater Process Technol 42:391400
4. Tagliaferri V, Di Ilio A, Visconti IC (1985) Laser cutting of fibrereinforced polyesters. Composites 16(4):317325
5. Mallick PK (2008) Fibre-reinforced composites: materials, manufacturing and design, 3rd edn. Taylor & Francis Group, New York
6. Voisey KT, Fouquet S, Roy D, Clyne TW (2006) Fibre swelling
during laser drilling of carbon fibre composites. Opt Lasers Eng
44:11851197
7. Yamaguchi S, Sekine K, Mori M, Kenmochi Y (1993) Processing
of composite materials by using laser ablation. J Photopolym Sci
Tech 6(3):409414
8. Lau WS, Yue TM, Lee TC, Lee WB (1995) Un-conventional
machining of composite materials. J Mater Process Technol
48:199205
9. Denkena B, Vlkermeyer F, Kling R, Hermsdorf J (2007) Novel
UV-laser applications for carbon fibre reinforced plastics. in
Applied Production Technology APT07. Bremen
10. Li ZL, Chu HY, Lim GC, Li L, Marimuthu S, Negarestani R,
Sheikh M, Mativenga P (2008) Process development of laser
machining of carbon fibre reinforced plastic composites. in
International Congress on Applications of Lasers and ElctroOptics, ICALEO. Temecula, CA, USA
11. Caprino G, Tagliaferri V (1988) Maximum cutting speed in laser
cutting of fibre reinforced plastics. Int J Mach Tools Manuf 28
(4):389398
12. De Iorio I, Tagliaferri V, De Ilio AM (1987) Cut edge quality of
GFRP by pulsed laser: laser-material interaction analysis. in
LAMP87. Osaka
13. Lau WS, Lee WB, Pang SQ (1990) Pulsed Nd:YAG laser cutting
of carbon fibre composite materials. CIRP Ann 39:179182
14. Mathew J, Goswami GL, Ramakrishnan N, Naik NK (1999)
Parametric studies on pulsed Nd:YAG laser cutting of carbon fibre
reinforced plastic composites. J Mater Process Technol 8990:198
203
15. Ashby MF, Jones DRH (1998) Engineering materials 2: an
introduction to microstructures, processing and design, 2nd edn.
Butterworth-Heinemann, UK

566
16. Levchik SV, Weil ED (2004) Thermal decomposition, combustion
and flame-retardancy of epoxy resinsa review of the recent
literature. Polym Int 53:19011929
17. Bundy FP (1989) Pressure-temperature phase diagram of elemental
carbon. Physica A: Statistical Mechanics and its Applications 156
(1):169178
18. Yin Y, Binner JGP, Cross TE, Marshall SJ (1994) The oxidation
behaviour of carbon fibres. J Mater Sci 29(8):22502254
19. Chen KS, Yeh RZ, Wu CH (1997) Kinetics of thermal
decomposition of epoxy resin in nitrogen-oxygen atmosphere. J
Environ Eng 123(10):10411046
20. Chen S-L (1998) The effects of gas composition on the CO2 laser
cutting of mild steel. J Mater Process Technol 73(13):147159
21. Shearer JS (1903) The heat of vaporization of oxygen, nitrogen
and air. Phys Rev (Series I) 17(6):469

Int J Adv Manuf Technol (2010) 49:553566

22. Myers RH, Montgomery DC (1995) Response surface methodology:
process and product optimization using designed experiments
probability and mathematical statistics, ed. Wiley, New York
23. Yung KC, Zhu HH, Yue TM (2005) Theoretical and experimental
study on the kerf profile of the laser micro-cutting NiTi shape
memory alloy using 355 nm Nd:YAG. Smart Mater Struc 14
(2):337342
24. Rodden WSO, Kudesia SS, Hand DP, Jones JDC (2002) A
comprehensive study of the long pulse Nd:YAG laser drilling of
multi layer carbon fibre composites. Opt Commun 210:319
328
25. Jiang G, Pickering SJ, Walker GS, Bowering N, Wong KH, Rudd CD
(2007) Soft ionisation analysis of evolved gas for oxidative
decomposition of an epoxy resin/carbon fibre composite. Thermochim Acta 454(2):109115