Journal of Materials Processing Technology, 25 (1991) 139-151 139

Elsevier

A study into the laser spot-welding of sheet metals

using oxygen and argon as assisting gases

B.S. Yilbas
Erciyes Universitesi, Muhendislik Fakultesi, Kayseri, Turkey
R. Davies
School of Mechanical Engineering, The University o[ Birmingham, Birmingham B15 2TT, UK
and
Z. Yilbas
Erciyes Universitesi, Muhendislik Fakultesi, Kayseri, Turkey
(Received September 21, 1989; accepted in revised form Marc h 27, 1990 )

Industrial Summary

The present study examines the mathematical analysis of a one-dimensional heat-transfer model
for the pulsed Nd YAG laser-welding process. A quasi-steady solution for the temperature rise in
the weld region is attempted, the solution being obtained for the surface temperature to reach 90%
of its steady-state value. In addition, the laser spot-welding of sheet metals has been carried out
with oxygen and argon assisting gases at atmospheric pressures. It is concluded that the use of
oxygen reduces the tensile strength whilst the use of argon increases the tensile strength of the
resulting welds, when compared with air ambients.

1. Introduction

Laser welding is especially useful when it is essential to limit the size of the
heat-affected zone and to reduce the roughness of the welded surface. One of
the important applications of laser welding is spot welding, where the diameter
of the weld section is of the order of 1 mm and the heat-affected zone is of the
order of a fraction of a millimetre. Laser techniques for the welding of single
and multi-core wires, and the spot welding of sheet metals, have been studied
previously [ 1 ].
The following main conditions must be satisfied in laser spot-welding:
(i) Highly accurate radial positioning of the component being welded.
(ii) Accurate positioning ofthe component in the focus of the laser to prevent
large variations in the power density.

0924-0136/91/$03.50 © 1991--Elsevier Science Publishers B.V.

140

(iii) Creation of a protective or enhancing atmosphere. If the materials to be

welded are easily oxidized in air, then welding must be carried out in a
protective atmosphere using, for example, argon.
(iv) Provision of symmetrical laser heating. For instance, for the laser welding
of two wires, the laser beam must be focussed on the contact of the wires
in order to optimise the properties of the weld.
In the present study, a mathematical analysis of a one-dimensional heat-
transfer model has been carried out for the pulsed Nd YAG laser-heating pro-
cess. A quasi-steady solution for the temperature rise inside the weld region is
attempted. On the experimental side, the laser spot welding of sheet metals
has been undertaken, the workpieces including stainless steel, titanium and
nickel, with oxygen and argon at different pressure levels being used as assist-
ing gases. It should be noted that the use of oxygen at atmospheric pressures
results in increasing heat transfer and causes a high mass-removal rate in the
weld region. On the other hand, argon, being an inert gas, results in a reduced
temperature rise on the surface of the workpiece, which causes a low material-
removal rate. Metallurgical changes occurring in the weld zone have been ex-
amined using photographs of weld cross-sections.

2. Mathematical analysis of the heat-transfer mechanism for laser welding

The heat transfer taking place during laser welding has been studied previ-
ously by Tiziani et al. [2] and Mazumder and Steen [3]. The analyses con-
ducted mainly covered the CO2 continuous wave (cw) welding process, and
consequently is not applicable directly to the present case, in which pulsed-
laser spot-welding is studied. An analytical model governing laser pulsed-heat-
ing is therefore required.
Criteria for the use of analytical methods for laser welding are: (i) the clas-
sical models of welding do not fit the dynamics of laser welding accurately;
(ii)for practical purposes it may be sufficient to take into account only the
quasi-steady solution for heat development and transfer.
In justification of the second of the above criteria,there are some parameters
which affect the heat-transfer mechanism during laser welding, but may be
considered to be small and therefore negligible when compared with the main
parameters of laser heating and associated heat-transfer, these neglegible pa-
rameters being plasma absorption of the incident laser beam, plasma heating
of the workpiece surface and temperature-dependent thermophysical proper-
ties of the workpiece surface. Whilst these are omitted from the present anal-
ysis, the model proposed is, however, sufficient to explain the physical nature
of the laser/workpiece interaction and the temperature rise inside the work-
piece material.
 141

The justification for using classical thermodynamic heat-equations is the

order of difference between the time taken for an electron to transfer its energy
to the lattice and the intervals considered during which heat fluxes are pre-
dicted. The heat-transfer equation in its most precise form may be solved only
if constant surface-regression is postulated [4]. Although the importance of
such a solution may be questioned in the face of evidence which points to
changing thermodynamic processes, the following theory establishes condi-
tions which explain the evolution of these processes. The attainment of equi-
librium is governed by the pulse duration at a given power-intensity.

2.1 Surface evaporation

The liquid surface-layer formed during a laser pulse moves into the material
at a rate determined by the quantity of vapour expelled. As the temperature of
the liquid molecules is increased so the additional energy needed to free them
from the binding forces decreases. The latent heat of vaporization therefore
decreases with temperature, until, at the critical temperature and above, it
becomes zero. The latent heat L at a given temperature Ts has been taken as
an elliptic function of temperature, which appears adequate for the accuracy
sought [5]. Hence:
L = L o ( 1 - T 2 / T 2 ) 1/2 (1)
where Lo is the latent heat of vaporization at absolute zero and Tc is the critical
temperature.
• The rate of change of latent heat with temperature can be expressed as [6 ]:

d T - T ~- (Cp2-Op,) (1/2- V , ) L \ ~ ] - \ - ~ ] p _ J (2)

where Cpl and Cp2 are the specific heats at constant pressure, 1/1 and V2 are
the specific volumes and 1, 2 are the subscripts for liquid and vapour states,
respectively.
Although integration of the latent heat over the temperature range from zero
to Tc is impossible, it can be used to show that little inaccuracy is involved in
taking the normal latent-heat as the latent heat at absolute zero. If now:
(OV2~ {aVl'~
or / (3)
(i.e. the specific volume of gas V1 is much greater than the condensed liquid
V2 and its rate of change with temperature at constant pressure is correspond-
ingly greater), then:
dL L . L [ a v2"~
dT-T~-(Cp2-Cp,)-~-~2k--~) p (4)
142

and applying the perfect gas law:

RT2 01/2 R
V2 - and
p OT - p
so that:

dL L+(cp_cp,)_L=Aep (5)
dT-T
ACp is extremely small for temperatures up to room temperatures [ 7 ] and so
little error will result in taking Lo as the latent heat at room temperature.
According to Maxwell's law, the velocity distribution of molecules is [8]:
/ m .~,i2 f ~,~
.exp[ mVz ).dVz (6)

where V~ is the velocity in the direction normal to the surface, T is the tem-
perature of solid, liquid or gas, k is the value of Boltzmann's constant, and m
is the mass of the atom. On the other hand:
No. of atoms with velocity Vz to Vz + dV~ per unit volume
[( Vz)dVz - No. of atoms per unit volume (7)

Only those molecules whose velocity is greater than that given by:
1 2
2mVmin=L(T) (8)
where Vminlies in the z-direction, will escape from the retaining potential. If n
is the number of atoms per unit volume, then the number of atoms with veloc-
ities Vz to V~+ d Vz per unit volume is nf(Vz) d Vz. The number of atoms with
these velocities passing a unit area per unit time is therefore nf(Vz) V~dVz. All
these atoms for which V~>> Vmin do not return to their equilibrium position
and are evaporated. If G is the number of atoms evaporated per unit time and
per unit area, then:

G= ~[ nf(Vz)VzdVz (9)
Pmin

or

m Vmin (10)
"expt 2kr )
If the atoms are equally spaced within the lattice, a surface layer would consist
of n 2/3 with an evaporation time n2/aG. The average velocity of the surface Vs
is therefore:
 143

G-'-n(kT~'/2"exp( - L (11)

Anisimov et al. [9 ] derived, erroneously, velocities of liquid surfaces based on

the Debye frequency of the metal and a constant latent heat of vaporization.
Such data yielded surface velocities comparable to those of Ready [10] but
suggested a minimum velocity for low and a maximum for high-powers,
respectively.

2.2 Temperature rise

The heat transfer equation for a moving source has been studied previously
by Yilbas [6]. If instead of a source, a moving front is taken in the positive z-
direction, the resulting heat transfer equation is:
02T OT OT
g~x2 + peVs~z + ISexp(-Sz) =pC, Ot (12)

where K, C,, Vz and are the thermal conductivity, the density, the specific heat
at constant pressure, the surface velocity and the absorption coefficient,
respectively.
With boundary conditions:

T ( ~ , t ) =0, T(z,0) =0; 0_~ z=o =P

VL
K (13)

An analytical solution is extremely difficult because the surface velocity is a

function of time. However, the assumption that Vs is constant permits a so-
lution by standard techniques, followingwhich an iteration is required to con-
verge on the correct velocity at each point in time. After long-windedalgebra
and using the technique of Laplace transforms, the solution becomes:

Io 5V/-~ F2v/~.ierf( z + b ~ / ~
T(z,t) -2pC,(~5- V) L -\2x/~t ]
3b2+c 2 [ z _ F~
+ 2 2 erfc/~+ bx/tl
2b(b - c ) \2x/~t /
+
1 2zb z
z0 \ ~/~/ \2~/~t /

÷ ~ _ c e X p [ - (Sz+ (b2-c2)t]erfc - -{-cv/-t

144
pVL
b2ceXp(-~z)l-~[4b~ft'ierfc(~-~+b~t~
\2x/at /

_erfc(z + b x / t ) + e x p/| - - -2bz\

- ~ | e r f c |/~ - z bx/t )1 (14)
\2x/at } \ x/a/ \2~,/at
where b= V/2x/~ and c = b - ~ x / ~ and a and fi are the thermal diffusion and
absorption coefficient, respectively. Equation (14) can be simplified by assum-
ing that a~ >> V, which is true for all power intensities of interest. If, in addi-
tion, it can be assumed that t >> I/a~ 2, then eqn. (14) reduces at the surface
to:
T(O,t)- (I-pVL)K x/~'ierfc ~ +~erf ~ (15)

or on re-arranging,
T(O,t) V(t) (1-(VL/I))
T(O,oo) U(oo) (1-- (~=2xierfc(x)+erf(x) (16)

where
x= y/2 , / t / a
Introducing the fraction r of the steady-state temperature which has been
reached, allows eqn. (16) to be reduced to:
r3/2-exp(p)
. \1/2

Lo ' [((I-T~(oo)) -rU/2(I/re-T2(oo 'l/e)) .exp(p)J- (17)

I-~ CpTc Tr(oO)
= 2x ierfc (x) + erf (x)
where
mLo '
P= kT~ T , ( ~ )
[(
(I-
Tr2(Oo))l/2__((l/r2__T2r(OO)) 1/2] (18)

and
Tr(t) = T(O,t)/To (19)
Equation ( 17 ) has been used to estimate the times for the surface to reach 90%
of its steady-state temperature and these are given in Figs. 1 and 2; they were
obtained from eqn. (17) after setting r = 0.9. It should be pointed out here that
a dimensionless relationship for the surface temperature-time profile and the
 145

time to reach 90% of the steady-state temperature is possible if the terms Lo/
(Cp- Tc) and mLo/(kTc) are constants for all materials. While the variation is
only of the order of a few percent for the materials examined, the appearance
of the second term in the exponents or eqn. (17) means that errors of up to
15% might be expected.

10-3
[o(W/m2)
A 21010

Fe
N~

E m

~-:o.

-'t.
0.5 0.6 0.7
Reduced sfeady-sta~'esurface temperal"ure

Fig. 1. Time for the surface temperature to reach 90% of its steady-state surface temperature at
two different power intensities.

a
. / j -- Titanium
8"

~Z 2.

0 J I J I e
O O.OO1 O.O1 O.1 1 10
DISTANCE FROM SURFACE (mm)
Fig. 2. Time for the surface temperature to reach 90% of its steady-state value with absorbed
power intensity for different materials.
t46

3. Experimental

3.1 Experimental apparatus

A schematic view of the apparatus is given in Fig. 3. A N d YAG laser deliv-
ering 10-22 Joules with a pulse length of approximately 1.48 ms was used. An
energy-power meter was placed in the path of the laser beam to measure the
instantaneous laser output-power. The laser head was mounted vertically on
a drilling stand so that the drilling table could be used to hold and manipulate
the workpieces. The focusing lens, of 51 mm nominal focal length, was mounted
on a graduated barrel, providing a vertical movement by rotation of the barrel.
A nozzle of 0.6 mm outer diameter was placed 2 mm away and 1.5 mm above
the centre of the irradiated spot. The nozzle axis was set at an angle of 45 ° to
the axis of the laser beam. Argon and oxygen were used as assisting gases.
Three different materials were selected as workpieces, this selection being based
on the range of thermal properties of the materials, the latter including stain-
less steel En58, titanium and nickel.

3.2 Experimental procedure

The focal position and laser output-power were set to give a spot of about 1
mm diameter on the surface of the workpiece and a laser peak-power intensity
of about 101° W / m 2. Welding was then carried out using assisting gas pressures
of oxygen and argon.
After the welding process, tensile testing of resulting welds was carried out.
Also, the welds were m o u n t e d in perspex and ground down to a diameter in
order to show each weld in cross-section. The polished samples were etched
and then examined under the microscope.

~LENS
~y X /X,Y TABLE

Fig. 3. Experimental set-up

 147

4. Discussion

The behaviour of metals heated by a pulsed solid-statelasercan be predicted

from Fig. 1. Each pulse consists of a series of spikes, typically 1 Hs long, with
a peak-power intensityapproximately an order of magnitude greater than that
of the mean intensity for the pulse. At this level,it is clear that all materials
can respond to an individual spike, causing rapid initialheating. Cooling be-
tween spikes will be minimal, since they appear at approximately 1-5 M H z
[11 ], so that the overalleffectwillbe to integrate the spikes.
In Fig. 2, the temperature riseinside the stainlesssteeland titanium is given
for two differentpower-intensities.It can be seen that the depth of fusion is of
the order of I ram, which shows that welding is limited to samples thinner than
I ram. However, in the laserwelding process, the points below should be taken
into account [6 ]: (i) a vapour-pressure effect; (ii)nucleation within the mol-
ten zone; and (iii)convective influences.
The vapour-pressure effectis sufficientto force molten material up the sides
of the heated zone, though not necessarily to eject it, which gives rise to a
cratering effecton the weld pool. O n the other hand, criticalexplosions,caused
by vaporization below the material surface,m a y be ignored. Conduction effects
may occur, but diffusion is no longer the dominant mechanism. Yilbas [12]
has shown that nucleation is,in general,the dominant material ejectionmech-
anism. However, it m a y again be discounted in the context of welding due to
the low power-intensities involved.
In Fig. 4, the tensile-testresults for the welds obtained at different oxygen
pressure are given for different workpiece materials. It can be seen that the
breaking load decreases with increasing oxygen pressure, this effectbeing high
in the case of titanium and steel,which m a y indicate that increased oxygen-
pressure causes an increased rate of high-temperature oxidation in the weld
region. This results in increased mass-removal rate, which is consistent with

A - 02 - "--Ni
280. x .... Ti

. . . . . . . EN58

(23
'x.

0
0 15 30 ~5 70
PRESSURE (KPa}
Fig. 4. Breaking load for various oxygen pressures.
148

- 0 2 ° - - TI-Ni
90 × .... Sf-Ni

.... St-Ti
Z
v

60 ¸
\
\
u \\
LO \\
Z
30

. . \\

I I I I
15 30 /*5 70
PRESSURE (KPa}

Fig. 5. Breaking load of the resulting welds of different materials for various oxygen pressures.

-At-

120 /
/
/
/ 7
/
z

o
u

z
/ • -- Ni
~o- x
/ x .... Ti
.......... ENSB

I I I I
0 15 30 /+5 70
PRESSURE (KPa)
Fig. 6. Breaking load for various argon pressures.

-Ar- , --Ti-Ni
90- x . . . . S f - Ni
z -- S t - Ti

60 J
o
u

Z
30" ,

0 i l
1~5 30 /,5 7'0

PRESSURE (KPa }
Fig. 7. Breaking load of the resulting welds of different materials for various argon pressures.
 149

Fig. 8. Photographs of the weld cross-sections of stainless-steel samples: (a) 02 assisting gas:
Po2 = 45 kPa (Mag. X 10); (b) Ar assisting gas: PA,=45 kPa (Mag. X 12).

earlier work [ 12 ], i.e. leaving the thin layer of material in the weld pool. The
above argument is true also for welds of Ni-AI, St-Ti, S t - N i and T i - N i work-
pieces (Fig. 5). However, variation in breaking load with pressure occurs in
the case of T i - N i and Ni-A1 welds, which may be due to: (i) plasma appearing
on the workpiece surface initially being bent by the assisting gas (oxygen),
150

permitting increased laser power to reach the target, which in turn generates
a secondary plasma on the surface. This, of course, changes the plasma-cou-
pling effect of heating and absorption. In this case, this may vary with oxygen
pressure; (ii) the effect of the thermal properties of the workpieces, since A1
and Ni are two dissimilar materials.
The variation of breaking load with Ar pressure is given in Fig. 6 for different
workpieces. The general trend of the figures shows that the breaking-load in-
creases with increasing argon-pressure. This may be explained in terms of ini-
tial plasma generation and its effects on the interaction mechanism. Increased
argon pressure may bend the plasma occurring on the workpiece surface, as
discussed above. It also reduces the plasma temperature, since argon is an inert
gas which prevents high-temperature oxidation in the welding region. In this
case, less plasma appears on the surface, resulting in increased laser-power
reaching the workpiece. Since the heating effect of plasma is minimal, the laser
power enhances the melting and, as a consequence of this, mass removal from
the weld pool due to evaporation is reduced. This can also be seen from the
photographs of weld cross-sections (Fig. 8), i.e. the depth of the crater in the
workpiece is minimal. However, this is not true for all the welds of Ni-A1, St-
Ti, St-Ni and Ti-Ni workpieces (Fig. 7). The breaking-load decreases with
increasing argon pressure in the case of A1-Ni welds, the opposite being true
in the case of St-Ni welds. This contradiction may be caused by one or all of
the following: (i) plasma heating and absorption effects; (ii) the thermal prop-
erties of the workpieces; and (iii) variation of the laser power reaching the
workpiece due to plasma oscillation which occurs during its generation.
On comparing Figs. 4 and 6, it can be seen that the effect of assisting gas on
the breaking load is significant. Different assisting gases and their pressures
result in variation in the breaking load. In general, use of argon as assisting
gas improves the tensile strength whilst oxygen reduces this by a factor of
about 0.5.
Photographs of weld cross-sections are given in Fig. 8. It can be seen that
the crater generated in the surface region, due to evaporation, is deeper when
using oxygen than that obtained when using argon. This is probably due to
high temperature oxidation taking place during welding with oxygen ambients.

5. C o n c l u s i o n s

(1) The mathematical model presented in this work is only valid for the
cases where evaporation is the predominant mechanism of material removal.
A quasi-steady solution has been obtained for the temperature rise inside the
workpieces and results have been computed for the surface to reach 90% of its
steady-state temperature.
(2) Use of oxygen as an assisting gas in the weld region reduces the tensile
strength of the resulting welds, the reduction increasing with increasing oxy-
 151

gen pressure, which may be due to the high-temperature oxidation occuring

during welding.
(3) In the case of Ni-A1, St-Ti, St-Ni and Ti-Ni welds, use of oxygen re-
duces the tensile strength of the welds. On the other hand, the tensile strength
of the St-Ti welds increases with the use of argon, the contrary being true for
the A1-Ni, A1-Ti and St-Ni welds. This may be explained in terms of the ef-
fects of plasma and the thermal properties of the workpiece on the welding
mechanism.

References

1 Yilbas B.S. and Yilbas Z., Study into laser spot welding of multi and single core wires, Sym-
posium on High Power Lasers, The Hague, The Netherlands, 1987.
2 Tiziani A., Zambon A., Bonollo F. and Cantello M., Analytical model for thermal fields study
in laser welding, Proc. Laser5 Conf. on Laser Material Processing for Industry, Gournay-sur-
Marne, 1989, pp. 76-89.
3 Mazumder J. and Steen W.M., Heat transfer model for cw laser material processing, J. Appl.
Phys., 51 (2) (1980) 941-947.
4 Yilbas B.S., Heating of metals at a free surface by a laser irradiation - an electron kinetic
theory approach, Int. J. Eng. Sci., 24 (8) (1986) 1325-1334.
5 Hodget D.L., Pulsed laser drilling of engineering materials, Ph.D. Thesis, Mech. Eng. Dept.,
Birmingham University, UK, 1970.
6 Yilbas B.S., Non-conduction limited laser welding of sheet metals, Mech. Eng. Tech., 7 (1)
(1987) 23-25.
7 Stull D. and Sinke G., Thermodynamic Properties of the Elements, Advances in Chemistry
series, No. 18, Am. Chem. Soc. Publ., 1956.
8 Waldram J.R., The Theory of Thermodynamics, Cambridge Press, Cambridge, 1987.
9 Anisimov S.I., Bonch-Bruevich A.M., El'Yashevich M.A. and Imas Y.A., The action of pow-
erful light fluxes on metals, Soy. Phys. Tech. Phys., 11 (7) (1967) 945-952.
10 Ready J., Effects due to the absorption of lser radiation, J. Appl. Phys., 36 (1965) 462-473.
11 Yilbas B.S., Heat transfer mechanisms initiating the laser drilling of metals, Ph.D. Thesis
Mech. Eng. Dept. Birmingham University, 1982.
12 Yilbas B.S., Experimental study into ejected material during laser-metal interaction at su-
batmospheric pressures of drilling ambients, Doga TU, J. Eng. Environ., 12 (1) (1988) 68-
73.