Laser welding of aluminium alloys

(1)

O F T E C H N O L O G Y

D O C T O R A L THESIS

Laser Welding of Aluminium Alloys

TOMAS FORSMAN

(2)

L U L E A I T E K N I S K A

U N I V E R S I T E T

Laser welding o f aluminium alloys

Tomas Forsman

Department of Materials and Manufacturing Engineering Division of Manufacturing Systems Engineering

SE-97] 87 Luleå, Sweden

Doctoral thesis 2000:39

Akademisk avhandling

för avläggande av teknologie doktorsexamen, som med vederbörligt tillstånd av tekniska fakultetsnämnden vid Luleå tekniska univenitet kommer att offentligen fbnvaras i sal E 231, fredagen den 2 mars klockan 10.00.

Fakultetsopponent: Dr. William O'Neill, University of Liverpool, England Ordförande: Adj. Professor Claes Magnusson, Luleå tekniska universitet

Academic Thesis

for the degree ofDoctor of Philosophy, which with the due permission of the Board of Faculty of Engineering at Luleå University ofTechnology will be publicly defended in room E 231, Luleå University of

Technology, on Friday the 2°* of March at 10.00 a.m.

External examiner Dr. William O'Neill, University of Liverpool, England Chairman: Adj. Professor Claes Magnusson, Luleå University ofTechnology

(3)

Preface

This w o r k was carried out at the Division o f Manufacturing Systems Engineering at the Department o f Materials and Manufacturing Engineering at Luleå University o f Technology.

The experimental w o r k was mostly performed i n our excellent laser laboratory.

I w o u l d like to thank m y friends and colleagues at the division and the department for all their help during these years and wish you all the best for the future.

I w o u l d like to express m y gratitude to Professor Claes Magnusson, Luleå, Doctor John Powell, Nottingham, Docent Alexander Kaplan, Vienna and Doctor Conny Lampa, Gothenburg, for rewarding discussions and guiding through the course o f this research w o r k . Finally I w o u l d like to thank m y parents and m y sister for your support and Jenny for your love.

Luleå, December 2000

Tomas Forsman

(4)

Laser welding o f aluminium alloys

Abstract

This thesis treats laser welding o f aluminium alloys f r o m a practical perspective w i t h elements o f mathematical analysis. The theoretical w o r k has i n all cases been verified experimentally.

The aluminium alloys studied are from the 5xxx and 6xxx groups w h i c h are common for example i n the automotive industry.

A l u m i n i u m has many unique physical properties. The properties w h i c h more than others have been shown to influence the welding process is its high reflection, high thermal conductivity, l o w melting point, l o w viscosity and the alloying elements used. The most important physical properties have been described and studied experimentally as well as theoretically. The high surface reflectivity was shown to be o f little importance when welding has initiated because the deep and narrow gas/plasma filled 'keyhole' captures the incident light w h i c h leads to considerably higher total absorption. The l o w melting point and high boiling point was shown to yield a relatively large weld melt pool. A large melt pool together w i t h the l o w viscosity has been shown to result i n instability o f the melt during welding.

A number o f defects can occur when laser welding aluminium alloys. The defects considered to be most important to avoid include porosity, edge defects and process instability. Porosity has been shown to consist o f spherical (gas filled) and ellipsoidal voids w i t h different origins.

Edge defects were f o u n d to be o f different types when welding thin and thick material respectively. C o m m o n for both is that the heat distribution i n the workpiece determines the result and that it can be theoretically predicted. Process instability include random blowholes, smoke, spatter and w e l d depth variations. Natural variations exist but they can be controlled by the process parameters. The single most important reason for spatter from the melt, is the amount o f magnesium as an alloying element.

The strength o f the welded joints has been tested i n static as well as dynamic loading. In normal tensile tests the best joints reached 90% o f the strength o f the parent material. By using wire as a filler, a strength close to 100% was realised but this was due to a larger weld cross section. I n fatigue tests the laser welds have performed better than other types o f bonding such as T I G , resistance welding and riveted joints.

A fast and user friendly model predicting the weld dimensions was developed to cover laser welding w i t h C O , and N d : Y A G lasers o f a range o f common metals including aluminium alloys.

The conclusion o f this w o r k is that welding o f aluminium alloys is a challenge. I t is predicted that methods such as riveting and clinching i n the future i n many cases w i l l be replaced by the cost effective laser welding. Laser welding o f aluminium alloys w i l l be increasingly used in automated industry as knowledge o f the process continues to grow. I n cases where thin aluminium alloys are to be j o i n e d at high speed, lasers are expected to dominate.

(5)

Papers

Paper I T . Forsman, A . F. H . Kaplan, J. Powell & C. Magnusson. N d : Y A G laser welding o f aluminium; factors affecting absorptivity. Lasers i n Engineering. V o l 8, N o 4, 295-309 (1999).

Paper I I T . Forsman, A . F. H . Kaplan, J. Powell & C. Magnusson. Initiation and termination phenomena i n laser welding o f aluminium. Journal o f Laser Applications. V o l 12, N o 2, 81-84 (2000).

Paper I I I T . Forsman, C. Lampa, J. Powell & C . Magnusson. Prediction o f the cross- sectional geometry o f N d : Y A G laser welds in aluminium alloys. Submitted for publication i n Lasers i n Engineering.

Paper I V T . Forsman, K . Nilsson, J. Powell & C. Magnusson. Laser welding; the influence o f laser choice and material properties on weld dimensions. Submitted for publication i n Joumal o f Laser Applications.

Paper V T . Forsman, J. Powell & C. Magnusson. Process instability in laser welding of aluminium alloys at the boundary o f complete penetration. Submitted for publication i n Journal o f Laser Applications.

Paper V I T . Forsman, C. Lampa, J. Powell & C. Magnusson. Prediction o f laser weld dimensions i n various metals using computer simulation. Submitted for publication i n Intemational Journal o f H i g h Temperature Material Processes.

(6)

Laser welding of aluminium alloys

1 Introduction

The first w o r k i n g solid state ruby laser was invented by Theodore Maiman o f 1960 [1] in fierce competition w i t h other laboratories. Their w o r k was based on Einstein's paper i n 1917 showing that lasing action should be a possibility [2]. A laser consists o f a gas or crystal medium stimulated by an energy source w h i c h emits radiation o f one wavelength w h i c h is also i n phase. B y the use o f mirrors, part o f the emitted light is reflected back into the medium creating an amplifying effect. Today there are hundreds o f types o f commercial lasers but only C O , , N d : Y A G , Excimer and diode lasers are available at powers high enough to be used for material processing purposes.

C O , lasers emit radiation o f 10.64 Urn, i.e. i n the far infrared o f the electromagnetic field.

They f o r m the bulk o f industrial materials processing lasers because o f their excellent combination o f price, performance and power. C O , lasers perform hardening, cutting, welding and ablation o f most metals and are today commercially available i n powers o f up to 50 k W . The most recent type o f C O , laser, called a slab laser, makes more efficient use o f the lasing gas and is therefore cheaper to run.

N d : Y A G lasers emit light i n the infrared at 1.06 Llm just outside the range o f visible light (0,39-0.77 Llm). T h e y perform the same material processes as the C O , lasers but the maximum power commercially available is only 5 k W . The main reason for the success o f N d : Y A G lasers is the possibility to transfer the beam from the laser to the workpiece by an optical fibre w h i c h makes the laser very flexible. I n the most recent N d : Y A G lasers the stimulating energy source is a group o f diodes instead o f an arc lamp. This has increased the energy efficiency and the reliability o f the laser.

Excimer lasers produce radiation in the ultraviolet spectrum (typically 193, 248 and 308 nm) emitting pulsed light w i t h an average power up to 1 k W [3]. The short wavelength makes excimer lasers suitable for precision machining o f materials. Ceramics and many polymen are easily processed. Another recent application is i n thin film growth.

Diode (or semiconductor) lasers can be produced to emit i n a wide range o f wavelengths but the most common are based on AlGaAs and emit between 600 and 1100 nm. They are characterised by excellent electrical efficiency but maximum powers o f only a few hundred watts w h i c h so far limits their use i n material processing to the hardening o f metals. I t is predicted that i n the next few years diodes w i l l become cheaper and more densely packed enabling the manufacture o f lasers w i t h higher power.

Laser welding o f metals has been taking place the last 25 years since the intensity (power divided b y cross sectional area) o f the beam became high enough to create a so called 'keyhole', i . e. a vapour filled hole surrounded by melt. Laser keyhole welding is faster than other more traditional melt welding processes such as M I G and T I G and requires for example only 10% o f the heat input (defined as power divided by speed [J/m]) to create a w e l d i n 2 m m thick aluminium. Most laser welding w o r k has been, and is being performed i n m i l d steel

(8)

formerly were j o i n e d to a steel frame by riveting and clinching are n o w bonded using laser welded joints. The latest step is to exchange the steel automotive frame for a so called space frame made o f extruded aluminium profile as i n the Audi A 2 . Another exciting area is the use o f tailored blanks [4]. Sheets o f different alloy and/or thickness are welded together and the joined blank is then formed into its final shape. Steel tailored blanks are for example used i n all

the latest Volvo models for the side beams. W i t h tailored blanks the right sheet thickness or hardness can be used where i t is needed, saving material and weight.

Except f o r the automotive industry, laser welding is not yet common as a welding process. The main reason for this is likely to be the high investment costs involved for small companies. I t is important to remember that the more a laser works the more economical i t gets and for purposes where high precision and flexibility is needed there is no welding process to match it.

A n early problem w i t h laser welding o f aluminium was that a higher beam intensity is needed to initiate the keyhole than w h e n welding steel. Intensity values vary w i t h the alloying content but 1-2 M W / c n r is usually enough using N d Y A G light and 2-3 M W / c m using C O , . T o obtain 1 M W / c n r w i t h a typical beam diameter on the surface o f the workpiece o f 0.5 m m , a power o f just below 2 k W is needed. The reason w h y aluminium requires high intensity is explained by equation 1 [5]

where

P T.A

(1) d Ao

P = laser power [ W ]

d = beam diameter at the surface [m]

T^b = boiling temperature o f the material [K]

X = thermal conductivity o f the material [ W / m K ]

A⁰ = absorptivity o f laser light at normal incidence on the material [1],

Since aluminium has almost the same boiling temperature as steel, three times higher thermal conductivity and a lower absorptivity i t is obvious that a higher power or intensity is needed to obtain a keyhole weld.

W i t h modem N d : Y A G and C O , lasers there is sufficient power and the beam quality is good enough to reach a high intensity so the problem o f welding aluminium alloys has changed into one concerning the quality and strength o f the weld. The strength o f the weld is a function o f its shape and metallurgy. Some features such as cracks severely reduce the strength due to stress concentrations that lead to failure. The weld shape and metallurgy i n t u m are functions o f the aluminium alloy properties, the laser/material interaction and sometimes other factors such as the shroud gas. This text w i l l focus on these areas and describe the context into w h i c h the following papers fit.

(9)

2 Aluminium properties of interest for laser welding

A l u m i n i u m is the most abundant metal i n the crust o f the earth at 8% by weight compared to 5.8% o f iron [6]. It is obtained f r o m bauxite containing around 50% o f hydrated alumina (aluminium oxide) as well as iron oxides, silica and titania. Since the production o f virgin aluminium is very energy demanding, it is important that aluminium is reused as much as possible. A l u m i n i u m can easily be remelted and turned into a new product and is therefore an ideal material for recycling.

The l o w density o f aluminium is the principal reason w h y it is increasingly gaining interest by the automotive and aerospace industries [4, 7]. A beam made i n an aluminium alloy can, for instance, be 8 times as stiff as a steel beam o f the same weight. By using aluminium instead o f steel where appropriate, the weight o f a car can therefore be decreased by as much as 40%.

Decreased weight results i n decreased fuel consumption, w h i c h is important f o r cars and o f crucial significance i n the case o f aeroplanes.

Another property that makes aluminium very special is the hard aluminium oxide ( A f O , or alumina) w h i c h automatically builds up to a layer o f -10-20 n m (thicker i n a humid environment) on the surface o f an aluminium sheet. I f the layer is removed, f o r instance by grinding, it w i l l be recreated at once and continue to grow at a decreasing speed. The natural surface oxide efficiently protects the bulk material f r o m corrosion w h i c h means that aluminium can withstand humid and salty environments 100 times better than m i l d steel and 15 times better than zinc coatings [8]. The surface oxide has a melting point o f 2 0 5 0 ° C . Therefore, when the aluminium underneath the oxide melts during welding the oxide w i l l still be solid in the f o r m o f oxide fragments. I f the temperature o f welding is not high enough to melt the oxide this can cause defects such as inclusions i n the re-solidified weld.

The physical properties o f aluminium are i n many ways unique. The first property that might come to m i n d is its density w h i c h is about one third o f that o f steel. Another is its thermal conductivity which is about three times that o f steel. A third is its protective surface oxide.

O u t o f these and all the other physical properties some are important in the course o f laser welding and some are not. For instance the l o w density is important for the weight o f the construction at hand but the laser welding process is not necessarily influenced by this property. This section w i l l describe the most important properties influencing the welding process and the final result o f the weld. The properties w i l l be described i n the order i n which they appear when heating an aluminium workpiece w i t h a laser beam.

2.1 Absorption

In principal, light incident on a material surface w i l l be partly reflected, absorbed and transmitted. I n the case o f laser light i n the U V , visible and I R spectra incident on a metal workpiece no transmission w i l l take place. Part w i l l be absorbed w h e n the electromagnetic radiation interacts w i t h bound electrons creating structural vibration i n the first few atomic layers o f the surface [9] (the vibration is detected as heat) and the rest w i l l be reflected.

A l u m i n i u m is highly reflective to incoming light compared to other metals as shown i n figure 1.

(10)

Figure 1. Absorptivity as a junction oj wavelength jor NaCL, silicon, iron and aluminium [10].

The absorption f o r light incident on a metal surface can be calculated f r o m the optical constants called the refractive index and the extinction coefficient by

A = i 2

4 n c o s ö 4 n c o s # [n2 + £2) c o s20 + 2 « c o s # + l n + k2 + 2 n c o s 0 + c o s20 j (2)

where

A = absorption [1], n = refractive index [1],

8 = angle o f incidence from normal [°], k = extinction coefficient [1]

and the absorptivity (at normal incidence) is

A) = y ,⁴" ,² (3)

(n + l) +k

The value o f absorptivity is by definition the absorption at normal incidence at r o o m temperature on a smooth surface. The absorption is a function o f the incidence angle o f the light, the temperature o f the workpiece, the surface condition and the wavelength o f the light as shown i n figures 2 and 3.

(11)

Brewster

Angle of incidence (°)

Figure 2. Absorption qf CO, light in aluminium as a function qf temperature and angle qf incidence [3].

100

80

60

20

a Fe. boiling temp.

A A!, boiimg temp.

V — Fe. room temperature Al, room temperature

B v ^

0.1 1 ' 0 wavelength in um

Figure 3. Absorption qf light in aluminium and iron as a function qf wavelength and temperature [5].

The important parameters w h i c h influence absorption can be summarised as follows:

• A reduction i n wavelength generally yields higher absorption because the photons are more energetic and therefore can be absorbed by more electrons.

• Increased temperature yields higher absorption because the phonon population increases leading to more phonon/electron energy exchanges.

• The absorption also increases when changing f r o m solid to liquid and from liquid to gas as shown i n figure 3.

• Increased oxide thickness usually yields higher absorption as shown i n figure 4 [11].

• Increased surface roughness yields higher absorption because o f an increased number o f reflections i n the undulations. I f the roughness is less than the beam wavelength the surface

(12)

• Increased incidence angle f r o m the normal yields higher absorption up to the Brewster angle after w h i c h it decreases to zero (see figure 2).

Since the surface o f the workpiece is o f utmost importance for the absorption it is obvious that different alloy compositions w i l l influence the absorption greatly. So far only pure aluminium and iron have been discussed. Unfortunately optical constants or absorptivity values f o r aluminium alloys have not been recorded i n the hterature. Experimental results however have shown that the absorption increases as the aluminium content is diluted by the addition o f alloying elements.

T h i c k n e s s of A l²0³ ( micron )

Figure 4. Absorption of CO, light as a function of the thickness of an anodised (ALOJ surface [11].

The influence o f surface condition on the absorption was investigated in paper I o f this thesis.

Extruded sheets o f aluminium alloy AA6063 were prepared w i t h four different surface conditions - as received, sand blasted, anodised and bright anodised. These were then passed w i t h a defocused N d : Y A G laser beam to produce heating and a focused beam to produce welding. The energy absorbed by the sheets was measured by water calorimetry. I t was found that the sand blasted surface absorbed three times as m u c h energy as the bright anodised did o f the defocused beam (30% and 10% respectively). However w h e n welding there was no difference between the sheets; all o f them absorbed 60% o f the energy at l o w process speeds and 50% at high speeds as shown i n figure 5. The conclusion o f this finding was that even though the surface condition plays a major roll for the absorption o f l o w intensity radiation it is o f no importance w h e n keyhole welding is considered. Instead the main mechanism for absorption is by internal reflections o n the keyhole walls w h i c h are, i n all cases, liquid and close to the high absorption Brewster angle.

(13)

70 60

o ⁵⁰ 1 O (fl

•"•—As received

•"•"""Sand blasted Anodised 'natural' c

Anodised 'shiny' 0

4 5 6 7 8 9 10

Welding speed [m/min]

Figure 5. Absorption as a function of process speed in AA6063 with four different surfaces.

A difference i n weld absorption between N d : Y A G and C O , lasers was discovered i n paper I V . W h e n welding w i t h the t w o lasers using the same process parameters, weld cross sections o f different size and shape appeared. O n average the AA6016 alloy welded w i t h N d : Y A G had a 10% larger melt cross section than its C O , counterpart. As was shown earlier, the absorptivity o f N d Y A G radiation is higher than C O , i n aluminium (as well as i n most other metals) and this was considered to be responsible f o r some part o f the increased absorption. The main reason was however attributed to the great difference i n plasma absorption between the two laser wavelengths.

The absorption o f light i n the plasma, taking place above and inside the keyhole, is a function o f the fraction o f ionised gas, the distance the beam travels i n the plasma, the temperature and the wavelength. The plasma absorption can be determined using Saha's equation

n^e = number density o f electrons [ 1 / m³] n, = number density o f ions [ 1 / m3]

nⁿ = number density o f vapour atoms [ 1 / m³] g0 l = partial function o f ions [1]

g^{U n} = partial function o f atoms [1]

m^e = electron mass [kg]

k = Bolztmann's constant [J/K]

T = temperature [K]

h = Planck's constant [Js]

(4)

where

(14)

to determine n^n, and then the plasma absorption coefficient is calculated by

where

a = nn.

z V

' b-^hoy'm.2c0£03 InkT

1 - exp

"kT

a = absorption coefficient [ 1 / m ] Z = number o f protons [1]

e = electron charge [C]

CO = angular velocity [ l / s ]

c⁰ = speed o f light i n vacuum [m/s]

E^u = electrical permittivity i n vacuum [ F / m ] g = quantum mechanical Gaunt factor [1]

(5)

and [12]

g = — I n 4 e x p ( - C^£) ha>

kT

(6)

where

CE = Euler's constant [1]

w h i c h i n turn gives the absorption by A = l - e x p ( - Q z )

where

A = plasma absorption [1], z = distance travelled [ m ] .

(7)

Plasma absorption has been shown [13] to be proportional to the square o f the wavelength.

Therefore C O , radiation w i l l be absorbed 100 times as much as N d : Y A G and more heat w i l l be concentrated i n the upper part o f the weld and penetration w i l l be reduced (see paper I V ) .

(15)

2.2 Thermal conductivity

The f l o w o f heat in a medium goes f r o m hot to cold in order to minimise the entropy according to Fourier's first law

where

(8) dy

qy = heat f l o w in y direction [ W / m2

k = thermal conductivity [ W / m K ] , y = distance [m].

The thermal conductivity o f a material is a f u n c t i o n o f its structure and bonding and reflects the ease or difficulty involved i n the transfer o f energy through the material by elastic vibrations o f the lattice and by free electrons m o v i n g through the lattice [14]. Impurities, dislocations and second phases all restrain movement and therefore metal alloys have lower thermal conductivity than the pure metal. For example the thermal conductivity o f pure aluminium is 220 W / m K at r o o m temperature while most A A 6 x x x alloys have a value of around 160 W / m K . The thermal conductivity o f aluminium decreases w i t h increased temperature and the value for a pure aluminium melt is around 100 W / m K .

W h e n a laser heats the surface o f a workpiece the heat w i l l f l o w hemispherically (three- dimensionally) into the material. The f l o w o f heat into the surrounding air can be neglected since the thermal conductivity o f air is only 0.024 W / m K . D u r i n g keyhole welding the heat source changes from a surface point into something that could ideally be referred to as a vertical line source from w h i c h heat flows laterally (two-dimensionally). This is the way in w h i c h laser welding was traditionally modelled [15]. I f a vertical line source is applicable the thermal conductivity w i l l not greatly influence the weld w i d t h . This is shown in paper I V where welds i n AA6016 were compared to welds in m i l d steel and Inconel 718.

Convection (heat f l o w by liquid movement i n the weld melt) caused by temperature differences acts as an extra heat transfer mechanism at the top surface and, in the case o f a through weld, also at the bottom surface o f the sheet. This extra heat transfer mechanism has been simulated by a point source super imposed on the vertical line source by earlier workers [16]. Paper I I I and paper V I suggest the use o f an artificially high value o f thermal conductivity to compensate f o r the increased heat f l o w close to the sheet surface. Experimental welds have been shown to fit theoretical values for all eight combinations o f N d : Y A G and C O , welding o f m i l d steel, stainless steel, aluminium and titanium alloys.

(16)

Laser welding of aluminium alloys

2.3 Melting and boiling point

The melting point o f aluminium is l o w compared to other metals at 660 ° C . Alloying can decrease the melting point but usually by no more than ten degrees. The l o w melting point was shown i n paper I V to be the principal reason f o r the widening o f the weld as shown i n figure 6. Considering the relatively high boiling point o f 2470 ° C aluminium has a large molten temperature range. This large molten range means that aluminium melt pools are broader and more extensive than those o f other metals. This has been shown to make aluminium sensitive to keyhole instability [17].

0 "i H 0,0 0,5 1,0 1,5 2,0

Lateral distance [mm]

Figure 6. Peak temperature as a function of radial distance in three metals with varying properties.

2.4 Viscosity of melt

The viscosity o f pure aluminium melt is relatively l o w compared to other metals. A t 850 ° C the viscosity o f aluminium is for instance similar to that o f water at r o o m temperature [18].

W i t h the additions o f alloying elements and the presence o f surface oxides i t is however likely that the viscosity increases slightly w h e n aluminium alloys are being welded.

The viscosity o f the melt strongly influences its stability during welding. L o w viscosity results i n less damping o f irregular keyhole movements. I n addition the l o w viscosity o f aluminium increases the risk o f the melt sagging or an incompletely filled groove at f u l l penetration welding.

2.5 Surface tension of melt

The surface tension o f aluminium melt is roughly half that o f i r o n at their respective melting points. I t is however complicated to predict the surface tension o f the alloys because some alloying elements influence the surface tension strongly even at l o w concentrations.

(17)

J / m :

2 . 0 0 0

1.800

1.600

1.400

1.200

1.000

1 2 0 0 1 3 0 0 T^m 1 4 0 0 1 5 0 0 K

Figure 7. Surface tension as a function of temperature for copper.

The surface tension o f metals decreases w i t h increased temperature as shown i n figure 7.

D u r i n g welding this has the effect that the melt w h i c h is hottest, close to the laser heat source, is forced to the sides o f the melt pool, leading to stirring. This stirring, named after Marangoni, takes place as long as the laser is heating up more melt. Marangoni stirring increases the heat transfer by convection at the surface during welding leading to widening o f the weld. This widening effect was modelled i n paper I I I and V I by the implementation o f an artificially high thermal conductivity.

3 Welding defects in laser welding of aluminium

Welding defects are the principal reason f o r failure i n welded structures and must be avoided.

There is a final draft o f a European standard o n laser welding o f aluminium f r o m December 1999 [19] where quality levels for imperfections are described. For the highest quality level, called stringent, the following requirements must be met (t is the sheet thickness):

• no cracks permitted except micro cracks (less than 1 m m² crack area),

• no crater cracks permitted,

• m a x i m u m dimension o f pore 0.3t but no more than 4 m m ,

• m a x i m u m dimension o f the total projected porosity area 3% o f the welded area,

• m a x i m u m crater pipe 0.05t but no more than 1 m m ,

• no lack o f fusion permitted,

• no lack o f penetration permitted for f u l l penetration welds,

• m a x i m u m undercut 0.05t but no more than 1 m m ,

• m a x i m u m excess weld at the top 0.2 m m + 0.15t but no more than 5 m m ,

• m a x i m u m excessive penetration 0.2 m m + 0.15t but no more than 5 m m ,

• m a x i m u m linear misalignment 0.11 but no more than 1 m m ,

• m a x i m u m sagging O.lt + weld height but no more than 0.2t,

• m a x i m u m incomplete groove 0.05t but no more than 1 m m ,

• m a x i m u m root concavity O.lt but no more than 0.5 m m ,

• m a x i m u m shrinkage groove 0.05t but no more than 1 m m and

• m a x i m u m amount o f weld spatter depends o n the application.

(18)

3.1 Cracks

Cracks o f different types can develop when welding because o f shrinkage during cooling and residual stresses i n the weld. The most common type are hot cracks (also called solidification cracks) w h i c h start to f o r m w h e n the metal is above its solidus temperature and l o w melting point alloying elements accumulate at grain boundaries. D u r i n g solidification this area w i l l be less ductile and therefore cracks can develop w h e n the weld metal shrinks during cooling [20].

The susceptibility o f hot crack formation increases w i t h :

• more heat input (such as T I G and M I G instead o f laser) and

• larger interval o f solidification (such as highly alloyed aluminium).

Several studies have been performed i n order to rank the aluminium alloys i n cracking susceptibility [5, 21-22] and a general conclusion is that most o f the heat treatable 6xxx alloys are susceptible to hot cracks. These alloys can be welded w i t h o u t cracks by decreasing the power intensity and the speed simultaneously. B y this procedure the heat input w i l l increase but at the same time the cooling rate w i l l decrease.

3.2 Porosity

Porosity i n the weld is a problem i f the largest pore or the total amount o f pores is large enough to decrease the strength o f the weld as indicted earlier by the standard.

There are t w o principally different types o f pore that can develop w h e n welding aluminium as shown i n figure 8. One is formed by the precipitation o f hydrogen and has a spherical shape and the other is formed by the dynamic m o t i o n o f the keyhole and is ellipsoidal and situated close to the root o f the weld.

O

0 0

a) A

A - A

CD

S

o

0 0

Welding direction b)

Figure 8. Schematic sketch qf a) side-view and b) front cross-section qf an aluminium weld showing spherical and ellipsoidal pores.

Pores created by precipitated hydrogen (70-80% H [23]) are avoided or at least minimised i f the sheet surface is free f r o m hydrogen sources such as grease, o i l or paint and the shielding gas used is inert and pure. The excess hydrogen accumulates i n hot areas, w h i c h explains w h y the pores tend to end up i n the middle o f the w e l d where the cooling is the slowest.

(19)

Pores created by dynamic keyhole motion can occur when welding high depth-to-width ratios [24]. This type o f pore can be avoided by decreasing the ratio for example by decreasing the power intensity and the speed simultaneously. This is mainly a problem for non-penetrating welds.

3.3 Edge defects

Edge defects are anomalies occurring close to the end o f the weld such as close to the sheet edge (start, end or b o t t o m edge) or when the weld is finished inside the workpiece. Expressed more generally edge defects can occur when the welding process can not be considered as being steady state. The variation o f most interest is the heat input over time or distance which can lead to problems o f overheating or end craters. Edge defects are especially important to avoid since they are stress concentrations i n the structure.

3.3.1 Overheating

The problem o f overheating was studied i n paper I I where large AA5052 aluminium sheets o f 1 m m thickness were butt welded. The main problem was the initial 50 m m o f the weld where the melt was seen to increase i n size and repetitively fall out o f the weld leaving holes. It was theoretically shown by the use o f a mirror heat source that the reduced heat sink available close to the edge effectively leads to excessive heat-up during the first part o f the weld. By applying a varying power, starting at a l o w level and linearly increasing until reaching the steady state value, the heat-up was shown to decrease by a couple o f hundred degrees as shown in figure 9. The theoretical remedy was successfully employed i n practise, resulting i n welds free o f defects.

2500

2000 t

O a

£ 500 -

0 —

0 20 40 60 80 100 D i s t a n c e f r o m e d g e d=Vt [ m m ]

Figure 9. Temperature as a junction oj distance fiom the starting edge. The dashed line represents welding

(20)

W h e n welding a small workpiece using high heat input (e. g. welding w i t h a l o w speed at the power limit o f the laser) there is always the risk o f heat deformation but also o f a different type o f overheating. Since the workpiece is small the whole workpiece w i l l be heated up and because the speed is l o w this w i l l happen while the welding is performed. The effect is that the penetration becomes greater as the welding progresses.

I f the welding depth is nearly the same as the thickness o f the workpiece a f o r m o f overheating can take place. Heat can build up close to the bottom edge o f the sheet because i t is adjacent to l o w conducting air acting as insulation. I t has been shown that a f u l l penetration weld i n a 1 m m thick sheet produced w i t h a certain power/speed combination becomes less than 1 m m deep i n a 2 m m sheet. This effect is especially apparent when welding aluminium because o f its high thermal conductivity.

3.3.2 End craters

End craters are pits at the end o f a weld that was finished inside the workpiece. The crater is left when the power is suddenly terminated and the melt solidifies so fast that the excess melt at the sides of the keyhole does not have time to completely level out. H i g h conductivity materials such as aluminium naturally solidify fast w h i c h is w h y end craters are difficult to avoid. I t has however been found possible to reduce the size o f an end crater by the use o f power ramping as explained i n section 3.3.1 but by decreasing the power instead o f increasing it.

3.4 Instability effects

The keyhole has been experimentally shown to vary i n size and shape during welding [25].

The keyhole fluctuations are transferred to the surrounding melt except for high frequency oscillations w h i c h are viscously damped by the melt pool [26]. A number o f reasons for keyhole fluctuations have been proposed including [17, 27-29]:

• variation i n welding speed,

• variation i n laser intensity distribution,

• thermally induced material stress,

• thermal relaxation i n the material,

• changes i n the coupling rate caused by laser beam interaction w i t h a laser induced plasma,

• excess pressure produced by ablation interacting w i t h the surface tension and

• unevenly distributed front wall absorption.

The fluctuations o f the keyhole and melt pool have a certain frequency depending on the laser, workpiece material and process parameters. The effects o f the keyhole and melt fluctuations, including blowholes, spatter and depth variations, may appear to be random but they all depend on that frequency.

3.4.1 Blowholes

A blowhole is a pit i n or a hole through the workpiece created by an explosion w h e n welding.

Blowholes are thought to be caused by a sudden closure or collapse o f the keyhole leading to entrapped gas and plasma i n the melt pool. The entrapped gas w i l l leave the melt i n a violent manner due to the loss o f the former pressure equilibrium (which was discussed i n paper V ) .

(21)

A l l A A 5 x x x , A A 6 x x x and A A 7 x x x alloys w i t h M g or Z n (characterised by l o w boiling point and high vapour pressure) as an alloying element are potentially problematic [24] as far as blowholes are concerned, but large variations exist.

Blowholes i n aluminium alloys have a frequency o f between 0.01 and 1 per metre w h i c h has the consequence that long welds have to be performed i n order to measure the average. A few solutions have been proposed to l i m i t the problem w i t h blowholes such as:

• using a t w i n focus laser beam to weld w i t h one spot trailing another to create a prolonged and more open melt [30],

• using N d : Y A G instead o f C O , laser to decrease the interaction between the laser beam and the plasma [5] and

• an overlaid power w i t h changing frequency which counteracts the resonant frequency o f the keyhole [31].

3.4.2 Spatter and smoke

Spatter consist o f particles leaving the melt pool i n the f o r m o f melt, solid or a combination thereof. It is not totally clear what causes spatter but high-strength alloys or highly alloyed aluminium yield more spattering than purer alloys. Alloys containing elements w i t h a l o w boiling point such as M g , seem to generate the most spatter. Spatter f r o m the melt during welding can become a problem i f it sticks to the laser optics, damaging them as well as sticking to the workpiece and thereby demanding cleaning [32].

Smoke generated during welding can also stick to the lens shielding glass or the workpiece i n the f o r m o f soot. Deposited soot has been shown to consist o f oxidised alloying elements [33].

3.4.3 Depth variations

Variations i n the welding depth w h i c h appear to be cyclic are discussed i n paper V . N o r m a l depth variations, due to the possible reasons listed i n section 3.4, were found to have a frequency o f around 60 H z at a welding depth o f 1.7 m m . A n additional reason for normal variations was proposed to be instability caused by the conflict between surface tension and vapour pressure i n the keyhole. The surface tension tries to reduce the keyhole and even out the surface while the vapour pressure acts to widen the keyhole. Small variations i n these pressures can create points o f instability leading to a collapse of the keyhole.

A t the border o f complete penetration the welding process was unstable and the depth was seen to fluctuate more than before. This was attributed to sporadic penetration o f the melt (see figure 10b) and keyhole (see figure 10c) through the material as shown i n figure 10.

(22)

c) speed range: 8.0-4.5 m / m i n d) speed range: 4.5-1.9 m / m i n ' Figure 10. Photographs qf the root qf the welds in 2.0 mm AA6063.

4 Strength of laser welded aluminium

The strength o f a welded j o i n t depends on the metallurgy o f the weld metal, the shape o f the weld and the possible presence o f defects as discussed i n section 3. The ideal weld has the same metallurgy and shape as the original workpiece and is w i t h o u t defects. The reality o f aluminium welding is usually that the metallurgical properties are deteriorated, the shape might be better or worse than originally and defects w i l l always appear.

4.1 Metallurgical changes

The metallurgical conditions after welding depend on the alloy welded. Precipitation or strain hardened alloys are generally expected to lose part o f their strength during welding because the rapid heatmg and cooling destroys the hardening effect i n the weld fusion zone and i n part o f the heat affected zone ( H A Z ) . Precipitation hardened alloys w i l l have part o f their precipitates dtssolved while strain hardened alloys w i l l lose their dislocations. For precipitation hardened alloys the part o f the H A Z that reaches 500 ° C w i l l be fully dissolved and w i l l experience a lower strength than the surrounding material. Very close to the fusion boundary a large amount o f the alloying elements w i l l remain i n solid solution and experience natural ageing The result is a narrow softened zone w h i c h is a weak link i n the weld [34]. This softened area o f the H A Z is wider i n more traditional welding processes than i n laser welding because o f the slower heaung and cooling cycle. A typical w i d t h o f the softened part is 0.5 m m f o r laser welding and 5 m m for T I G welding. The H A Z is by definition the part o f the workpiece that

(23)

has not only been heated by the welding process but also been metallurgically affected b y the heat. Depending o n w h i c h alloy is discussed, the temperature l i m i t for affecting the material varies but a temperature o f 300 ° C is usually necessary for recrystallisation.

4.2 Weld shape

The shape o f the weld depends on the workpiece and j o i n t shape, heat input and the convection o f heat. The w e l d shape is defined by the appearance o f a vertical cross section normal to the welding direction and the shape o f the top o f the weld. The weld profile should usually be as identical as possible to the base material, e. i . w i t h a flat top and root. However a common way o f increasing the strength o f the weld is to ensure that the weld cross section is larger than the surrounding material by the use o f additional aluminium wire.

The shape o f the weld cross section due to conduction o f heat and the choice o f different lasers is discussed i n paper I V . As mentioned i n section 2.1 it was found that the N d : Y A G laser melted more material than its C O , counterpart. I n addition the reasons for weld widening were discussed. A t the top o f the weld plasma absorption, convective stirring due to surface tension forces and decreased heat sink all act to widen the melt. A t the bottom o f the weld, i n the case o f f u l l penetration, plasma absorption is l o w while convective stirring and decreased heat sink still exist. The principal reason for weld widening was f o u n d to be the melting point o f the material.

Figure 11. Weld cross sectional shapes at speeds of 3, 6 and 9 m/min a) calculated by the theoretical model and b)fiom experimental welding. Top shapes c) calculated by the model and d)for experiments.

(24)

Prediction o f the weld shape has been found to be o f great importance and was therefore the aim o f paper I I I and paper V I . I n paper I I I the cross sectional and top shapes o f N d : Y A G laser welds i n A A 6 0 6 3 alloy were predicted, as shown i n figure 11, by developing a model originated by Kaplan [35] and Lampa [36]. The model was simplified to make it fast to compute but still includes the beam intensity distribution, calculation o f the keyhole size and calculation o f the energy absorbed by plasma and Fresnel absorption.

The model was further developed to manage partial as w e l l as full penetration welding i n an increased range o f metals being welded by C O , and N d : Y A G lasers respectively. The complete model was incorporated on a C D w i t h a user friendly interface so that it can be o f use to engineers and laser users as discussed i n paper V I . One example o f an output w i n d o w f r o m the C D is shown i n figure 12.

Simulation output

Calculated depth as a function of welding speed

TU få

Laser rid YAG Power: 3D0O W

Alloy: AA60B3-T4 Speed [m/min]

" \ P r c v i Q U S i # 1 ! i Exit

Figure 12. Output window from the CD showing depth as a function of speed for an aluminium alloy.

4.3 Static testing

Static loading o f welds is usually performed by tensile tests perpendicular to the weld direction.

These tests are a convenient way o f categorising and comparing materials or i n this case welds.

A unique property o f aluminium is that both its strength and its ductility increase w i t h decreased temperature below 0 ° C as shown i n figure 13 [37].

- 2 0 0 0 200 4 0 0 T e m p e r a t u r e <~Ct

Figure 13. Strength and elongation as a function of temperature for an aluminium alloy.

(25)

The tensile strength o f N d : Y A G laser butt welds has been investigated where 1 m m thick A A 5 0 5 2 - O gave ultimate strengths o f 90% o f the base material value and A A 6 0 0 0 - T 4 resulted i n 85% o f the base material strength.

Earlier w o r k has reported the same ultimate strength as the base material for welds i n AA5182- O [4] but the ductility was reduced to 85%. Alloys subjected to considerable deformation hardening (H18) or heat treatment (T6) have shown reduced mechanical properties. I n this case the ultimate strength was reduced to between 70% and 90% o f that o f the base material [7].

4.4 Dynamic testing

Dynamic loading o f welds is usually performed by fatigue tests where the specimen is cyclically loaded perpendicular to the weld direction. The tests tell h o w many cycles a weld can take before failing at a specific applied stress. A l u m i n i u m has no endurance l i m i t like steel, i . e. a stress level at w h i c h there is a 50% probability that failure w i l l occur. Instead aluminium fails by fatigue even at l o w stress levels as shown i n figure 14 [37].

-

! 0 0 . 0 0 0 c y c l e f a t i g u e lite at 9 0 . 0 0 0 psi a p p l i e d sires s

-

! T o o l s t e e i ^ * * " ^ - ^

E n d u r a n c e l i m i t / = 6 0 . 0 0 0 psi

1

A l u m i n u m a l l o y

i i i i i i n i i . M i f i ii i i i i i i i n

N u m b e r o f cycles

Figure 14. Stress as a function of number of cycles to failure for a tool steel and an aluminium alloy.

Laser welds i n AA5052 sheet on top o f an AA6063 profile have been tested i n fatigue using a shear configuration [38]. The laser welds were shown to exhibit better fatigue properties than T I G welds, resistance welds and riveted joints and similar properties to f r i c t i o n stir welds.

(26)

5 Conclusions

It has been a trend i n the automotive industry during the last few years to replace sheet steel for aluminium alloys. Another trend is that the amount o f welded aluminium i n cars has increased f o r each model as a replacement o f riveting. This is clear i n the case o f the A u d i models where the aluminium A 2 model includes 30 m o f laser welded aluminium.

A number o f conclusions were made about laser welding o f aluminium during this w o r k including:

• The absorption during N d : Y A G laser keyhole welding was shown to be independent o f surface finish such as a rough surface or the presence o f surface oxide.

• The level o f absorption i n the keyhole was shown to be determined by the combined effect o f plasma absorption and the number o f internal keyhole reflections. I n the case o f N d : Y A G welding the plasma absorption can be neglected.

• The melting efficiency o f the welding process was found to decrease for higher speeds.

This effect was attributed to the development o f a more shallow keyhole giving fewer internal reflections.

• Overheating was shown to occur i n thin section aluminium resulting i n severe defects close to the starting edge o f the workpiece. The elimination o f these defects was shown, theoretically as well as experimentally, to be possible by employing linear power ramping.

• Keyhole stability was shown to be influenced by the conflict between surface tension trying to close the keyhole and the vapour pressure trying to keep it open.

• The welding process was shown to be unstable when the penetration o f the melt or the keyhole was o f the order o f the workpiece thickness. Welding faster resulted i n stable partial penetration while welding slower resulted i n stable f u l l penetration welding.

• D u r i n g the studies on weld shapes N d : Y A G lasers were shown to melt more material than did C O , lasers. The main reason for this difference was concluded to be the plasma absorption w h i c h i n the case o f C O , welding concentrates heat at the top o f the weld and re-radiates heat out o f the weld zone.

• The thermal conductivity o f the workpiece was shown to be o f little importance for the w i d t h o f the weld. Instead the main effect was found to be the melting point.

• The cross sectional and top geometries o f laser welds i n m i l d steel, stainless steel, aluminium and titanium alloys were successfully predicted by a theoretical model. This model was given a user friendly interface to be o f use for engineers and laser users.

The overall conclusion o f this w o r k is that welding o f aluminium alloys is a great challenge due to the special properties o f aluminium. I t is predicted that methods such as riveting and clinching i n the future w i l l be replaced by more cost effective laser welding. Laser welding o f aluminium alloys w i l l be increasingly used i n industry as knowledge o f the process continues to grow. I n cases where thin aluminium alloys are to be j o i n e d at high speed lasers are expected to dominate.

(27)

References

1. M A I M A N , T , H . Stimulated optical radiation in ruby. Namre. A u g 6'1' (1960).

2. E I N S T E I N , A . Zeitschrift f ü r Physik. 18, 121 (1917).

3. S C H U Ö C K E R , D . Handbook o f the eurolaser academy. V o l 1-2, Chapman & Hall (1998).

4. N A G E L , M , F I S C H E R , R, L Ö W E N & S T R A U B E , O . Production and application oj aluminum tailored blanks. Proceedings o f IBEC'97. 87-91 (1997).

5. R A P P , J, D A U S I N G E R , F & H Ü G E L , H . Laser beam welding of aluminium alloys.

Proceedings of E C L A T ' 9 6 . 97-106 (1996).

6. S T A N N E R , R, J, L. American scientist. V o l 64 258 (1976).

7. P O H L , T & S C H U L T Z , M . Laser beam welding oj aluminium alloys for light weight structures using CO^r and Nd:YAG-laser systems. Proceedings o f L A N E ' 9 7 . 181-192 (1997).

8. Handbok f ö r konstruktörer - hur man lyckas med aluminiumprofiler. SAP A (1995).

9. S T E E N , W , M . Laser material processing. Springer-Verlag, London (1991).

10. B E L F O R T E , D . The industrial laser annual handbook. Penwell publishing corporation (1986).

11. A R A T A , Y & M I Y A M O T O , I . Some jundamental properties oj high power laser beam as a heat source. Transactions o f the Japan welding society. V o l 3, N o 1, Apr, 152-162 (1972).

12. D O W D E N , J, P O S T A C I O G L U , N , D A V I S , M & K A P A D I A , P. Journal o f physics D : applied physics. V o l 20, 36-44 (1987).

13. M U L S E R , P. Hydrogen plasma production by giant pulse laser - a theoretical study.

Internal study, Institut f ü r Plasmaphysik, Garching bei M ü n c h e n (1995).

14. P O I R I E R , D , R & G E I G E R , G, H . Transport phenomena i n materials processing. T M S (1994).

15. R O S E N T H A L , D . Transactions o f A S M E . V o l 48, 848-866 (1946).

16. S T E E N , W , M , D O W D E N , J, D A V I S , M & K A P A D I A , P. A point and line source model oj laser keyhole welding. Journal o f physics D : applied physics. V o l 2 1 , 1255-1266 (1988).

17. D A U S I N G E R , F, RAPP, J, H O H E N B E R G E R , B & H Ü G E L , H . Laser beam welding oj aluminium alloys: state oj the art and recent developments. Advanced technologies and processes, Proceedings o f IBEC'97. 38-45 (1997).

18. H A T C H , J, E. A l u m i n i u m , properties and physical metallurgy. A S T M , USA (1984).

19. European standard, final draft, p r E N ISO 13919-2. Welding - electron and laser beam welded joints - part 2: aluminium and its weldable alloys. (1999).

20. B A R G E L - S C H U L T Z E . Material science. V D I - V e r l a g (1983).

21. K U T S U N A , M . Metallurgical aspects in laser welding oj steels and aluminium alloys. Proceedings o f I C A L E O ' 9 6 . Section D , 122-133 (1996).

22. K U T S U N A , M . Study on hot cracking in laser welding oj aluminium alloys. I I W Doc I V . 631- 95 (1995).

23. K U T S U N A , M , S U Z U K I , J, K I M U R A , S, S U G I Y A M A , S, Y U H K I , M & Y A M A O K A , H . C O , laser welding ojA2219, A5083 and A6063 aluminium alloys. Welding i n the world.

V o l 3 1 , N o 2, 126-135 (1993).

24. M A R T U K A N I T Z , R, P & S M I T H , D , J. Laser beam welding oj aluminum alloys. 6th International Conference on A l u m i n u m Weldments, Cleveland, Ohio, USA. American W e l d i n g Society, 309-323 (1995).

25. A R A T A , Y . Challenge oj laser advanced materials processing. Proceedings o f L A M P ' 8 7 . Osaka, May, 3-11 (1987).

(28)

26. K L E I N , T , V I C A N E K , M , K R O O S , J, D E C K E R , I & S I M O N , G. Oscillations of the keyhole in penetration laser beam welding. Journal o f physics D : applied physics. V o l 27, 2023- 2030 (1994).

27. K R I S T E N S E N , T . Studies o f dynamical process phenomena during laser keyhole welding.

Ph. D . thesis, Technical university o f Denmark (1993).

28. D O W D E N , J, K A P A D I A , P, C L U C A S , A D U C H A R M E , R & S T E E N , W , M . On the relation between fluid dynamic pressure and the formation of pores in laser keyhole welding. Journal o f laser applications. V o l 8, 183-190 (1996).

29. M A T S U N A W A , A , K I M , J - D , SETO, N , M I Z U T A N I , M & K A T A Y A M A , S. Dynamics of keyhole and molten pool in laser welding. Journal o f laser applications. V o l 10, N o 6, 247- 254 (1998).

30. G L U M A N N , C, R A P P , J, D A U S I N G E R , F & H Ü G E L H . Welding with combination of two CO,-lasers - advantages in processing and quality. Proceedings o f I C A L E O ' 9 3 . 672-681 (1993). "

3 1 . KLASSEN, M , S K U P I N , J, S C H U B E R T , E & SEPOLD, G. Study explores aluminium welds. Industrial laser solutions. Jan, 26-28 (1999).

32. F O R S M A N , T . Lasersvetsning av aluminiumplåt till aluminiumprofiler. Teknisk rapport 1998:26, Luleå tekniska universitet (1998).

33. M I Y A M O T O , I et al. Mechanism of soot deposition in laser welding. Proceedings o f I C A L E O ' 9 4 . 293-302 (1994).

34. G R O N G , 0 . Metallurgical modelling o f welding. The institute o f materials, University Press, Cambridge, U K (1994).

35. K A P L A N , A , F, H . A model of deep penetration laser welding based on calculation of the keyhole profile. Journal o f physics D : applied physics. V o l 27, 1805-1814 (1994).

36. L A M P A , C, K A P L A N , A , F, H , P O W E L L , J & M A G N U S S O N , C. An analytical thermodynamic model of laser welding. Joumal o f physics D : applied physics. V o l 30, 1293- 1299 (1997).

37. A S K E L A N D , D , R . The science and engineering o f materials. 3^{r d} ed. PWS publishing company, Boston (1994).

38. L E I J O N , G, L I N D E R , J, L A R S S O N , M , S T E N S I Ö , H , L A R S S O N , J, K , S T E N F O R S , S-E, P E T T E R S O N , I & B E N S O N , P. Fatigue behaviour o f joints between aluminium extruded profiles and aluminium sheets. Swedish Institute for Metals Research, I M - 1 9 9 9 - 539 (1999).

(29)

(30)

Paper I — N d : Y A G laser welding o f aluminium; factors affecting absorptivity

(31)

Paper I

Nd: Y A G laser welding of aluminium;

factors affecting absorptivity

(32)

Paper I - N d : Y A G laser welding o f aluminium; factors affecting absorptivity

(33)

Nd: Y A G laser welding of aluminium;

factors affecting absorptivity

T . F o r s m a n * , A . F . H . K a p l a n * * , J . P o w e l l * * * , C . Magnusson*

*Division o f Materials Processing, Luleå University ofTechnology, S-971 87 Luleå, Sweden Phone: +46 920 91771, E-mail: TomasEorsman@mb.luth.se

**Department o f Laser Technology, Vienna University ofTechnology, Arsenal, Objekt 207, A-1030 Vienna, Austria

***Laser Expertise L t d . , U n i t H , A c o m Park Industrial Estate, Harrimans Lane, D u n k i r k , Nottingham N G 7 2 T R , England

Phone: +44 115 985 1273

Abstract

This paper investigates the factors affecting the absorptivity during N d : Y A G laser keyhole welding o f a 6xxx aluminium alloy. The influence o f surface condition on absorption is shown to be negligible. Experimental absorption measurements by calorimetry are compared to analytical absorption values using a simple model based on Fresnel absorption during multiple reflections i n the keyhole.

1 Introduction

A material exposed to radiation can react i n three ways: by absorption, reflection and/or transmission. Since infrared light penetrates only up to two atomic diameters i n metals they can be considered opaque w h i c h means transmission can be ignored. Therefore an infrared laser beam directed onto a metal surface w i l l only be absorbed and/or reflected.

A l u m i n i u m is usually described as being very reflective to laser radiation. Typical absorption figures are presented i n table 1. This type o f absorption measurement usually shows h o w l o w intensity radiation at normal incidence is absorbed on a flat aluminium surface at r o o m temperature. The absorption has however been f o u n d to depend on the wavelength o f the incident light [1-3], the angle o f incidence to the surface [3], the surface temperature [2] and the surface roughness [4]. This makes it difficult to compare results and consequently they often differ.

Laser welding o f aluminium typically demands a power density exceeding 1 M W / c m² to heat the surface to its melting point where welding can begin and often to the boiling point to create a keyhole.

(34)

Paper I - N d : Y A G laser welding o f aluminium; factors affecting absorptivity

Table 1. Absorption in a l u m i n i u m ; data from the literature.

Absorption o f N d : Y A G Absorption o f C 0² light C o m m e n t Reference light (1.06 urn) (10.6 u m )

11% 5% Calculated HI

5% 3% Experimental [51

11% - Experimental [21

9% ^- Calculated [31

12% 4% Experimental [31

Extensive research has been performed on the absorption o f radiation but very few workers [6- 7] have mentioned absorption o f laser light o n aluminium surfaces using real process conditions, i.e. high power density, heating o f the work-piece and formation o f an absorbing keyhole.

This study o f energy absorption was performed to estimate h o w much o f the incident energy is absorbed by the work-piece during laser welding o f aluminium and to identify factors which can influence that absorption.

2 Experimental work

2.1 General

The experimental w o r k concentrated on N d : Y A G laser welding o f aluminium alloy AA6063 w i t h different surface finishes. The four surface finishes investigated are listed i n table 2. L o w intensity absorption investigations have already established that surface roughness [4] and surface oxidation [8] can have a profound effect on absorption f o r C O , laser radiation. For this reason it was decided to compare 'as received' aluminium w i t h a sand blasted (roughened) and anodised (oxidised) surfaces. Anodising effectively coats the aluminium w i t h a layer o f comparatively l o w conductivity oxide (A1,0³).

T a b l e 2. Surface condition o f a l u m i n i u m sheets.

Surface condition 1) As received

2) Sand blasted

3) Anodised; natural appearance

4) Anodised; polished or shiny appearance

(35)

Figure 1. Cross-sections of a) as received and b) anodised aluminium sheets.

A wide variety o f anodised surfaces are available commercially and two typical finishes were tested One o f these had the same visual appearance as the as received material and the other appeared polished or shiny. I n both cases the surface layer o f A L O , was 10 Urn thick. The presence o f the oxide coating is clear i n figure 1 w h i c h compares cross-sections o f as received and anodised samples.

The sample size in each case was 3 m m x 70 m m x 70 m m . The laser was a H A A S HL3006D N d Y A G w i t h a m a x i m u m power o f 4 k W . The absorption o f the sample surfaces was investigated at l o w intensity at ambient temperature and at high intensity during welding.

2.2 Absorption of low intensity energy

A l u m i n i u m sheet

Power meter

Figure 2. Set-up of low intensity absorption measurement.

(36)

A collimated N d : Y A G laser beam w i t h a diameter o f 25 m m and a power o f 500 W was reflected o f f the four different surfaces onto a power meter as shown i n figure 2. The angle o f incidence was kept inside the region where no increased reflection is to be expected for w e l l polished surfaces [2].

The energy absorbed by the sheets was measured by calorimetry and the results are shown i n table 3.

T a b l e 3. E n e r g y absorption o f N d : Y A G laser light. T h e total energy involved was 36 kJ i n each case.

Surface condition

E n e r g y absorbed by the work-piece fj]

E n e r g y absorbed by the p o w e r meter Q]

E n e r g y lost by diffuse reflection f j ]

As received 5216 (14%) 25200 (70%) 5584 (16%)

Sand blasted 11005 (31%) 4200 (12%) 20795 (58%)

Anodised 'natural'

4864 (14%) 20400 (57%) 10736 (30%)

Anodised 'shiny'

3659 (10%) 28500 (79%) 3841 (11%)

As can be seen f r o m table 3, the sand blasted surface absorbed more than twice as much energy as the other surfaces. The absorption for anodised surfaces were similar to the as received material. The table shows substantial energy losses w h i c h can be attributed to diffuse reflections from the aluminium sheet which miss the power meter. This type o f energy loss was naturally high for the sand blasted surface where the beam was scattered i n all directions by the multifaceted surface.

2.3 Absorption of high intensity energy during welding

Bead-on-plate welding was performed on the same materials as before w i t h the experimental parameters presented i n table 4.

T a b l e 4. E x p e r i m e n t a l parameters.

Parameter Value

Laser HL3006D N d Y A G

Beam guidance Optical fibre 0 0 . 6 m m Focal length o f lens 150 m m

W e l d i n g speed 4-10 m / m i n

Power 3000 W

Shielding gas None

The energy absorbed by the different surfaces was measured by calorimetry as before and the results are shown i n figure 3.

(37)

70

60 5

c o 50

absorpt

40

30

Energy

20

10

0

~ ^ " A s received

" " • " S a n d blasted Anodised 'natural' Anodised 'shiny'

Figure

4 5 6 7 8 Welding speed [m/min]

3. Energy absorption as a junction oj welding speed jor different surfaces.

10

9PS

a)

-

Figure 4. Cross-sections oj welds in sheet with anodised natural surface at speeds a) 5 m/min h) 6 m/min c) 7 m/mtn d) 8 m/min e) 9 m/min andj) 10 m/min.

(38)

Regardless o f surface condition the absorption was around 60% o f the incident energy for l o w speeds (-4 m / m i n ) and around 50% for the higher speeds investigated here (-10 m / m i n ) . Even though the four surfaces showed large differences i n energy absorption for l o w intensities there are minimal absorption differences when welding.

Figure 4 shows the change i n cross-sectional geometry o f a typical series o f welds as the welding speed is increased. There is a notable change f r o m a deep keyhole weld profile at the lowest speeds to a shallow weld at the highest speeds.

T o determine the relationship between the absorbed energy and the melting rate for each welding speed figure 5 was produced. I n this figure the weld cross-sectional area has been multiplied by the welding speed to give a melting rate i n m m Vs. This value drops o n average by - 3 0 % over the range o f welding speeds shown here but the reduction is concentrated i n the speed range f r o m 8 to 10 m / m i n .

3 0 0

2 5 0

nE 2 0 0 E

V

150

MM

i ^loo

5 0

0

As received Sand blasted Anodised 'natural' Anodised 'shiny'

7 8 Welding speed [m/min]

9 10

Figure 5. Cross-sectional area times speed as a junction of welding speed for different surfaces.

(39)

3 Theoretical work

3.1 General

The authors have constructed a simple theoretical model w i t h w h i c h to investigate the absorption o f N d : Y A G laser light during the welding o f aluminium. Previous w o r k [9-10] has demonstrated that during C O , laser welding the laser beam is absorbed by a combination o f t w o phenomena:

a) direct absorption during multiple reflections o f f the keyhole wall,

b) absorption by the plasma cloud inside the keyhole which then re-radiates energy to the keyhole wall.

I n the case o f N d Y A G laser welding the situation is simplified because plasma absorption o f the laser beam is minimal. This is because the wavelength o f N d Y A G laser light is 1.06 [lm rather than 10.6 \Xm for C O , lasers. A t this lower wavelength absorption by the plasma cloud becomes negligible [11]. The model can therefore concentrate purely upon multiple reflections and their related absorption events.

Figure 6. Keyhole shapes and typical multiple reflections in keyhole for speeds 5, 7 and 9 m/min for the corresponding experimental weld seam cross-sections from figure 4.

Laser welding of aluminium alloys

D O C T O R A L THESIS

Laser Welding of Aluminium Alloys

TOMAS FORSMAN

Laser welding o f aluminium alloys

Tomas Forsman

Preface

Abstract

Papers

Contents

Laser welding of aluminium alloys

1 Introduction

2 Aluminium properties of interest for laser welding

3 Welding defects in laser welding of aluminium

0 0

S

0 0

4 Strength of laser welded aluminium

Simulation output

5 Conclusions

References

Paper I

Nd: Y A G laser welding of aluminium;

factors affecting absorptivity

Nd: Y A G laser welding of aluminium;

factors affecting absorptivity

Abstract

1 Introduction

2 Experimental work

9PS

V

MM

i loo

3 Theoretical work

i ^loo