A Study in How Welding Parameters Affect the Porosity in Laser Welded High Pressure Die Cast AM50 Magnesium Alloy

(1)

INOM

EXAMENSARBETE MATERIALDESIGN, AVANCERAD NIVÅ, 30 HP

STOCKHOLM SVERIGE 2017,

A Study in How Welding

Parameters Affect the Porosity in

Laser Welded High Pressure Die

Cast AM50 Magnesium Alloy

EDWIN BERGSTEDT

(2)

Abstract

There are a need for reducing the weight of vehicles, one solution is to implement cast lightweight materials such as the high pressure die cast AM50 magnesium alloy. The

weldability of this cast alloy is poor and to implement the use of the alloy commercially a welding process is needed that limits the porosity of the weld. The aim of this thesis is to study the effect of the welding parameters on the porosity in the weld, for three laser welding methods. The welding methods examined are single spot and twin spot laser using either a beam splitter or separate optics. The microstructure of the base material are also examined in order to evaluate relations between the components of the microstructure and the porosity in the weld. It was concluded that the hydrogen in the base material was the main reason for the observed porosity in the weld and that the material contains high pressure gas. The welding parameters did not influence the porosity for the single beam laser process, however, for the dual beam processes the welding parameters could affect the amount of pores. It was found that a double weld reduced the amount of pores and that the size and distribution of the secondary phase particles would benefit from the treatment. The cleaning of the samples prior to welding increased the porosity, however, non-cleaned samples contained more oxide inclusions. The results indicate that a twin beam process could reduce the porosity in the weld of the AM50 alloy.

(3)

Sammanfattning

Det finns ett behov av att reducera vikten på fordon, en lösning är att implementera gjutna lätta material såsom formsprutad AM50-magnesiumlegering. Svetsbarheten hos denna gjutna legering är dålig och för att kommersiellt kunna använda legeringen krävs en

svetsprocess som begränsar svetsens porositet. Syftet med detta examensarbete är att studera svetsparametrarnas effekt på svetsens porositet för tre lasersvetsmetoder. De svetsmetoder som undersöks är enkelpunkts och dubbelpunktslaser där antingen en stråldelare eller separat optik använts. Basmaterialets mikrostruktur undersöks också för att utvärdera sambandet mellan mikrostrukturen och porositeten i svetsen. Man drog slutsatsen att väte i basmaterialet var huvudorsaken till den observerade porositeten i svetsen och att materialet innehåller gas under högt tryck. De undersökta svetsparametrarna påverkade inte porositeten för processen med en laserstråle, men för dubbelstråleprocesserna kan svetsparametrarna påverka mängden porer. Det visade sig att en svets utförd med två strålar minskade mängden porer och att storleken och fördelningen av sekundärfaspartiklarna gynnas av behandlingen. Prover som rengjordes före svetsning hade ökad porositet, men icke-rengjorda prover innehöll mer oxidinneslutningar. Resultaten indikerar att en dubbelstråleprocess kan minska porositeten då AM50-legeringen lasersvetsas.

(4)

List of abbreviations

HPDC High Pressure Die Casting

SEM Scanning Electron Microscopy

Nd:YAG Neodymium doped Yttrium Aluminium Garnet

TIG Tungsten Inert Gas

MAG Metal Active Gas

Nomenclature

pgas Absolute pressure inside the gas bubble p_atm Ambient pressure

ph Hydrostatic pressure (𝝆𝝆𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎× 𝒈𝒈 × 𝒉𝒉) σ Surface energy per unit area of the melt r Radius of the gas bubble

𝝆𝝆𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎 Density of the melt

h Distance from the pore to the melt surface

(7)

Introduction

1 1.1 Background

As a reaction to the imminent threat proposed by global warming the UN climate change conference COP21 were held in Paris in the end of year 2016. It was concluded that the emission of greenhouse gases needs to be heavily reduced to achieve the goal of

maximum 2°C temperature increase by the year 2100 [1]. The burning of fossil fuels produces greenhouse gases and in 2014, cars and heavy trucks in Sweden were responsible for 16,6 million tons of greenhouse gas emissions [2].

In order to reduce the emission of greenhouse gases and potentially halt the imminent global warming, laws and regulations such as the European emission standard Euro 6, stipulate the automotive industries to constantly reduce the emissions of new cars and trucks [3].

One way of reducing the emissions is to reduce the weight of the vehicle as the energy required or fuel consumption is proportional to the weight [4-6]. The automotive industries are facing challenges as the new generation of vehicles needs to be both lighter and at the same time safer, whilst maintaining or improving the comfort. In order to reduce the weight of the vehicle, new and light materials needs to be implemented in the production [7]. Alloys based on aluminium and magnesium are good candidates for replacing heavy parts as they are lightweight and can be cast into complex shapes with the high pressure die casting (HPDC) method [7-10].

The casting of large and complex details requires huge effort and large and heavy machines [8]. An alternative to this approach is to cast multiple simpler components and joining them together. Traditional mechanical joining techniques are in many cases limited, there may be need for pre drilling of holes, and the design could be restricted due to need of access during mounting. There is also a productivity limitation, as mechanical joining usually is slow. Therefore, high productive welding methods as laser welding show great promise in creating new and flexible designs. The fast and efficient laser welding would lead to increased productivity [11] and a substantial weight reduction [5].

Cast material is susceptive for a large number of internal and external defects [8], some of which may lead to porosity and other defects leading to loss of performance in and in the proximity to the weld joint [12]. Increased knowledge and understanding of these materials is requested from the industry in order for them to fully explore the potential of cast lightweight materials. It is by these premises that a VINNOVA (The Swedish Innovation agency) project was started to investigate the possibilities of lightweight metallic materials. The participants in the project are; Shiloh, Volvo PV, SAPA, Swerea KIMAB, Swerea SWECAST,

Jönköpings Tekniska Högskola and AB Lundbergs Pressgjuteri.

Conventional welding techniques for Magnesium are e.g. TIG and MAG. These welding methods are however limited as they subject the material for huge amounts of heat.

The heat generated by the conventional welding processes carries numerous disadvantages such as a large heat affected zone, high residual stress and distortion [5]. The high energy density and welding speed of the laser is a potent candidate for welding Magnesium alloys [5,

(8)

However, in HPDC Magnesium alloys there are still many questions unanswered related the excessive formation of gas pores in the weld fusion zone.

There are numerous explanation to the cause of porosity in laser welded HPDC

Magnesium. Keyhole instability, surface oxides or contamination and hydrogen trapped in the material since the casting are some of them [5, 6, 13, 14]. Zhao et al. claims that the keyhole collapse is unlikely in Magnesium alloys as vaporized Magnesium exerts a high gas pressure stabilizing the keyhole [13]. Zhao et al. were able to connect welding parameters to the porosity and found that the porosity tended to decrease with decreasing heat input [13]. This indicates that low beam power or fast welding speed should be beneficial for achieving low porosity fusion zones [13]. Harooni et al. also concludes that high welding speed is crucial to mitigate pore formation [6]. In comparison, the remedy Zhu et al. proposed for the wrought magnesium alloys are a larger melt pool and longer beam to metal interaction time. This would provide the hydrogen gas sufficient time to escape, hence reducing the porosity[14].

These statements are opposites of one another as Zhao et al. suggest decreasing power and increasing speed, whilst Zhu et al. claims increasing the power and reducing the speed is best for removing the porosity. Zhao et al. found that remelting of the fusion zone could significantly reduce the porosity [13].

There are several studies that conclude that the porosity is due to high hydrogen content and pressure in the base material and that a heat treatment at ~300°C could effectively remove the hydrogen from the material through diffusion, yielding a weldable material without

substantial porosity [6, 15].

It is obvious that the inconsistency among the present studies, especially regarding the HPDC magnesium alloys, requires further examination on the pore formation mechanisms.

1.2 Aim of the thesis

This thesis is a part of the VINNOVA project with the aim of finding suitable welding process parameters that allows for welding the AM50 magnesium alloy. Using highly productive laser welding methods while achieving good weld quality with limited porosity.

The main objective in the study is to examine how welding parameters affect the porosity in the fusion zone in AM50 HPDC Magnesium.

1.3 Laser welding

Light amplification by stimulated emission of radiation, or LASER, were invented in 1960 [16]. Laser light is special as it is monochromatic, coherent and directional [17]. This means that the light consists of photons with the same wavelength that are in phase, the light also travels in approximately the same and parallel direction [17, 18]. The first laser created was a ruby laser, where a synthetic ruby crystal rod were placed in between two mirrors, one of which were semi-transparent [16, 17]. As the ruby rod is energized the chromium starts emitting photons, these photons stimulate further photons to be emitted thus causing a cascade of emitted light [17].

Laser types

Since the development of the first solid state laser in 1960 many different techniques for

(9)

the light can be transported in fibre optic cables and focused by regular lenses [18]. The wavelength are also shorter than for the CO₂laser at 1.06 µm [16-18]. Fibre lasers are a different laser type and consist of an optical fibre with an outer pumping region and an inner active core that generates and propagates the laser light [17, 18]. By increasing the refractive index in each transition towards the centre of the fibre, the light cannot escape the inner core as the light is fully reflected at the boundary interface [17]. A source of light, usually a diode laser, is directed into the outer pumping region [17, 18]. The amplification is related to the fibre length and by increasing the length of the fibre higher power output is obtained [17].

Welding mechanisms

There are two types of mechanisms for laser welding, conduction mode or keyhole welding. The conduction mode welding is performed at lower power densities, usually less than 10⁶ W/cm² [11]. As the beam with lower intensity is directed at the surface, the

temperature rises in that point and transfers out into the material via conduction. At the point in contact with the beam the metal melts locally creating a weld pool [11]. The conduction mode welding results in a shallow and wide heat affected zone and the vaporization of elements is minimal [11].

Keyhole welding is contrary to the conduction welding performed using high power density. The high and localized energy of the laser evaporates the material in an instant and creates a cavity in the material that is sustained by the vapour and ablation pressure [11, 17].

The keyhole stability is dependent on constant movement of the laser beam to sustain a steady state condition [11]. As the beam moves the cavity follows and the pressure from the ablation of the material creates movement from the front to the back of the keyhole [11]. The ablation pressure must counter the forces collapsing the keyhole, such as surface tension and the hydrostatic pressure of the molten metal [11, 16]. Therefore, there is a certain minimum traversal speed of the beam at which the keyhole stabilizes [11]. The high speed and energy of keyhole welding produces a narrow and deep weld with minimal deformation in the material [18].

Multi beam

There are two principal techniques for multi beam welding: either the beam is split into two using a prism before or after the optics [17]. A beam splitter after the optics is usually favourable as the optics needed for splitting a beam is far less expensive than purchasing another complete set of optics. However, the setup using two separate optics can be far more optimized as more of the welding parameters can be altered independently.

1.4 Magnesium alloys

There are several commercial magnesium alloys available for casting, some of them are;

AM20, AM50, AM60, AZ81 and AZ91 [19]. The letter and number indicate the two main alloying elements e.g. the AZ91 alloy is a magnesium base alloy with around nine percent aluminium and one percent zinc or less [19]. The Aluminium-Manganese (AM) series of Magnesium alloys are increasing in popularity as they are more ductile in comparison to the Aluminium-Zinc alloys [20]. The AM50 alloy has a ductility of 6-10 percent in comparison to AZ91 which has a ductility below two percent [20].

The alloy investigated in this study is the AM50 alloy. The microstructure of AM50

(10)

solidify before being shot into the chamber. As the casting speed is high, the pre-solidified phase can become almost spherical due to the high shear force [22]. The melt solidifies dendritically, causing the remaining melt to be enriched in aluminium [21]. As the eutectic reaction starts, the remaining melt solidifies as α-Mg while the β-Mg17Al12 phase is

precipitated in the grain boundaries [21]. There can also be a small number of Al-Mn particles present in the AM50 such as the Al8Mn5 or Al4Mn [7].

It is believed that the strength of the AM50 alloy mainly is due to the β-Mg17Al₁₂ particles [21]. The melting point of the β-Mg17Al12 phase is low and due to the poor stability of the phase, the β-Mg17Al12 particles will start to coarsen and soften at temperatures as low as 125°C [21].

Another common defect present especially in Magnesium alloys is the band defects illustrated in Figure 1 [22, 23].

The band defect is usually a combination of a local composition difference in the eutectic, the defect band could also contain a band of pores [23]. The defect usually follows the shape of the casting and occurs near the walls of the chamber [12, 23]. The cause of this banded structure is thought to be the fast heat transfer from the die walls. The melt in contact with the wall solidifies, creating a shell and the solute concentration in the remaining melt increases [23]. The remaining melt also contains free flowing crystals formed previously in the process, these crystals migrate to the centre of the casting and forms a semi solid inner structure [12, 23]. The bands are created as the melt in between the semi solid centre and the solid shell is suspected to shear forces. This can create pores, ruptures and other changes in the microstructure along the segregation bands [12, 23].

1.5 High pressure die casting

The further development of the printing press in the early 19^th century sparked a Figure 1 Arrow showing a segregation band defect close to

the surface at 25.2x magnification. The scale bar corresponds to 1mm.

(11)

thin walled components whilst maintaining a good surface finish requires pressure and the first patent regarding pressure die casting was awarded Sturges in 1849 [24]. Sturges casting machine was simple and the operator manually pushed the melt into a mould cavity by pulling a lever [24].

The current generation high pressure die casting or HPDC machines are powered by hydraulics to achieve the high pressure needed during casting [8, 24]. There are two methods available, hot or cold chamber HPDC. The hot chamber technique is used mainly for zinc based alloys. The drawback of the hot chamber HPDC is that for alloys with relative high melting point, the melt is contaminated with iron from the mould [8]. The amount of air inclusions is also greater than for cold chamber HPDC [8].

The HPDC machine consists of a cylindrical shoot sleeve with a hole, were the melt is poured into. A hydraulic cylinder is attached to a plunger mounted inside the shoot sleeve, where the movement of the plunger forces the melt into the die cavity mould [24].

The HPDC process can be divided into three separate stages, the first stage is the filling where the correct amount of molten metal is poured into the shoot sleeve [8, 24]. The second stage is the injection, where the plunger is moved in a controlled manner, hence forcing the melt into the die cavity [8, 24]. The final stage is the ejection, when the part solidifies and cools to a temperature assuring structural integrity. The die halves are separated and the detail is ejected by ejector pins [8, 24]. After the final stage the mould is sprayed by a mixture of water and oil to cool and to add lubrication to the mould [24].

The pressure and velocity of the plunger in the injection stage varies. At first the velocity is relatively low in order to fill the gating system without contaminating the melt with air [8, 23]. When the gating system is filled, the speed increases and the main cavity is filled violently at high speed [8, 23]. As the mould fills the speed decreases and the pressure is increased to a maximum to completely fill the mould [23]. During the solidification the high pressure is maintained to minimize the effects of solidification shrinkage and to reduce the gas pore size [8]. The pressure is greater than 20 MPa [8] and can range from 20-160 MPa [23, 24] for aluminium alloys. The speed of the melt through the gate varies dependent on the alloy cast. In aluminium castings typical gating speed are in the range 20-50 m/s and for magnesium the gate speed could be as high as 120-150 m/s [24].

There are different beliefs regarding the behaviour of the metal when filling the mould.

Fredriksson et al. claims that the injected melt flows as a stream of metal from the mould cavity inlet until hitting the opposite wall [8]. From the back wall the melt spreads outwards along the wall and changes direction back towards the inlet. At a certain time the flow changes to turbulent and the rest of the mould fills turbulently [8].

Murray et al. are of another opinion. They suggest that the melt moving at high speed atomizes when passing the inlet and that the melt forms small droplets that freezes when hitting the mould wall and creates a skin on the surface [24]. When the initial skin has formed the mould fills turbulently [24]. These different takes on filling the cavity in HPDC process may both be correct, depending on the specific conditions of each experiment. Hans Ivar Laukli performed experiments by stopping the plunger mid action and investigating the solidification structure [23]. The results show that the mode of flow is dependent on both processing parameters and the geometry [23].

(12)

gas bubbles present in the melt to expand and if the change in volume is excessive, shrinkage pores could form [24].

1.6 Gas porosity in HPDC metal

There are many causes for porosity in cast metals, one of which is gas precipitation [8].

In most metals the solubility of gases is substantially lower in the solid phase than in the liquid [8]. This phenomenon can cause gas pores to precipitate during the solidification [8].

There are numerus factors affecting the solubility of gases in metals. Temperature, pressure and composition of the alloy are some of them [8]. As the melt solidifies the gaseous compounds are concentrated in the remaining melt [8]. When the gas concentration reaches the solubility limit gas bubbles usually precipitate, the bubbles can either form spontaneously through homogeneous nucleation or based on nuclei present in the melt through

heterogeneous nucleation [8, 25]. The former is highly unlikely as it requires very high partial pressure of gas [25]. The pressure inside the gas pore is controlled by the equilibrium

described in equation 1.1 [8].

The HPDC process utilises high pressure during the casting and solidification phase. As a result the volume of the gas pores inside the melt is reduced in accordance to the ideal gas law. The gas pores in HPDC is therefore small in size [8]. According to equation 1.1 the pressure inside the gas pore depends on the ambient pressure, the hydrostatic pressure and the relation between the surface energy of the melt and the pore radius [8]. The pressure inside the small gas pores formed in the HPDC process could therefore be high [8, 24].

(13)

Method

2

In total 130 welds were made as bead on plate, the first four were discarded as they were used for tuning purposes. Also welds 73-75 were removed from the study as they were used for tensile testing. The remaining 123 welds were used in this study and are as follows;

5-72 and 76-130. These welds were made with a variation of process parameters on 24 plates, the number of welds per plate varies from four to eight on each plate. The dimension and shape of each plate varied slightly, but all plates were 3 mm thick.

2.1 Cleaning

Before welding, the samples were treated according to one of five procedures. The methods used for cleaning included mechanical brushing and blasting and also chemical treatment with calcium-silicate and acetone. The cleaning were performed in one or two steps, where different methods were combined. Five of the samples were not cleaned in any way.

Table 1 shows the combinations of cleaning used and how many samples that were cleaned by each method.

Table 1 Cleaning procedure

Step 1 Step 2 Number of samples

None None 5

Acetone None 10

Calcium silicate None 1

Brushing Acetone 97

Blasting Acetone 10

2.2 Welding parameters

The 123 samples were welded using numerous combinations of the welding parameters, the parameters are; welding speed and power, spot size, focus position and optics. In Table 4 the welding parameters are listed, as well as their range and in how many discrete steps they were altered.

Table 2 The welding parameters range and the number of discrete steps for each range.

Range Discrete steps Speed 1 - 4.5 [m/min] 9

Power 1 - 3.4 [kW] 14

Spot size 0.2 - 0.66 [mm] 6 Focus position -10 - +20 [mm] 12

(14)

The samples 76-111 and 117-130 were welded using dual beam techniques, this

increases the level of complexity as distance between the spots and individual focus positions for each beam are added to the available parameters. Of all samples welded with dual beam, only two were welded with the beams in parallel position. This means that none of the beams is leading but are welding next to each other. The other dual beam samples were welded with the beams inline, where one beam leads and the other follows behind.

Four of the samples were welded by increasing the angle of the laser to 10° leading. All of the samples, except seven, were welded with argon gas at the flow rates; 40 l/min for the shielding gas and 5 l/min for the root gas. The remaining seven samples were welded using helium instead of argon at the same flow rates. The complete weld protocol can be found in the appendix.

2.3 Sample preparation

A strip with the width of a half centimetre, which contained the welds, was cut from each of the 24 plates. The strip were cut to separate the welds, and four or five weld samples were placed in a sample holder and was moulded into transparent plastic. In order to evaluate the validity of the pore content from the cross section, 14 of the weld samples were also cut along the centre of the weld in approximately two centimetre strips. These strips were placed in pairs and moulded into plastic.

2.3.1 Cutting and moulding

The initial and coarse cutting were performed on the AbrasiMatic 300 abrasive cutting machine. The samples were aligned and fastened on the cutting board. During the cutting the abrasive cutting disc was held in position and the sample was fed into the disc at a speed of 0.6 mm/s. The fine cutting was performed on the Struers Accutom-5 at a feeding rate of 0.05 mm/s.

The finished samples were placed in plastic sample holders and were cast into transparent plastic using the Buehler Simplimet 2000 automatic mounting press shown in Figure 2.

(15)

Each sample holder was placed in the casting machine and covered in 45 cm³ of Buhler TransOptic compression mounting compound. The settings used for casting were: 8 min of heating, 12 min of cooling, 290 bar of pressure and a temperature of 160°C. An example of the final results can be found in Figure 3.

2.3.2 Grinding and polishing

A first procedure for grinding and polishing were established experimentally by polishing one of the specimens until desired results were obtained. The following is the first procedure suggested for grinding and polishing the AM50 alloy. The grinding was performed in two steps at respective grit and at perpendicular angles, until all lines from the grit rubbing were in the same direction. The grits used were; 180, 340, 600, 1200, 2000 and 4000.

The polishing was performed on an automatic polishing machine consisting of the Struers RotoPol-11 and RotoForce-1 shown in Figure 4.

Figure 4 Struers automatic polishing machine Figure 3 The samples after casting, figure a) shows the weld cross section and b) shows the samples cut along the weld

(16)

Table 3 Polishing procedure 1

Step Diamond [µm] Time [min]

1 9 10

2 6 5

3 3 5

4 1 10

After inspection it was found that procedure 1 produced a shiny surface which in the microscope produced a high contrast image of the porosity.

It was later found that due to an inhomogeneous base material, the specimens reacted differently to procedure 1, thus yielding a great variety in shine amongst the samples. When polishing multiple samples another effect was noticed, during polishing with fine diamond suspension, new scratches emerged. The new scratches were too deep to be caused by the diamond particles, the scratches could be caused by the hard and brittle secondary β-Mg17Al₁₂ phase. Potentially, the particles of the secondary phase are ejected from the soft matrix during the polishing. Some results in section 3.1.5, indicates this scenario and is further discussed in section 4.1.5.

In order to achieve the desired polishing quality, procedure 2 were designed. To avoid the formation of the deep scratches the polishing disc were cleaned continuously with a brush every minute during the polishing. Also the time of the polishing steps were individually adapted for each specimen. The third procedure used for polishing was manual polishing using a suspension containing Al₂O₃ particles with the size of 0.05 μm. Procedure 3 was performed after procedure 2. Table 4 lists the polishing procedure and the corresponding samples.

Table 4 The procedures used and which samples subjected to each procedure

Procedure Samples Analysis

1 All Macroscopic

2 All Macroscopic

3 10-11, 44-45, 59-61, 71- 72, 101-106

Microscopic

4 7, 11, 35, 50, 76, 106, 125-126

SEM

The fourth and final polishing procedure is similar to the third procedure. The difference is that procedure four uses SiO2 particles with the size of 0.02 μm instead of alumina particles. Procedure four is used for SEM analysis as residual alumina particles, potentially, could interfere with the element analysis.

Vibration polishing

Vibration polishing was performed on one of the specimens to evaluate the procedure.

(17)

2.4 Analysis

The first and main analysis of the samples were macro analysis. All of the samples were polished and photographed in a microscope. Fifteen of the samples were also examined and photographed in a microscope to evaluate the polishing procedure. After analysing the macroscopic pictures, eight of the samples were chosen for SEM analysis.

2.4.1 Macro analysis

The macro analysis were performed on a Leica MZ6 microscope equipped with a Kappa PS 40 camera, the light source was a led ring light shown in Figure 5.

All images were taken using the same settings at 25.2x magnification. After the first polishing procedure the images in the microscope, from the cross section of the samples, were similar to that of Figure 6.

After polishing procedure two, the images were similar to the reference sample for the polish. The porosity was clearly visible as the ring led light on the microscope were reflected in the edges of the pores. This yielded a high contrast image suitable for image analysis shown in Figure 7.

Figure 6 Macroscopic images of the cross section after polishing procedure 1, a) shows sample 76 and b) sample 104 at 25.2x magnification. The scale bar corresponds to 1mm.

a) b)

Figure 5 The led ring light equipped on the Lecia MZ6

(18)

The samples cut along the weld were polished using the second polishing procedure and the samples were also photographed in the microscope. As the cut along the welds is

approximately two centimetres, each sample were photographed four times. The images were stitched together using the open source Gimp 2.8 software. An example of the stitched

together images is shown in Figure 8.

2.4.2 Micro analysis

The micro analysis were performed on the Leica DM IRM microscope equipped with the Kappa GigE vision Zelos-285C camera.

2.4.3 SEM analysis

As the TransOptic plastic is a nonconductive plastic, the samples needed to be prepared for the SEM analysis. In order to achieve a conductive surface, the specimen were covered in a conductive carbon tape. This tape was placed so that it was in contact with the sample to allow excess electrons to escape, preventing a charge building up inside the sample. Figure 9 shows the sample prepared for the SEM analysis.

Figure 8 Macroscopic images after polishing procedure 2 along the weld, a) shows sample 76 and b) shows 104. Four images taken at 25.2x magnification stitched together. The scale bar corresponds to 1 mm.

a)

b)

(19)

The SEM used for the analysis was the JEOL JSM-7001F field emission scanning electron microscope. The Oxford instruments X-Max 80 mm² detector were used for detecting the emitted x-rays used for element analysis.

2.4.4 Image analysis

The images were analysed using the open source Fiji [26] software, the Fiji software is based in ImageJ version 1.51f and contains numerous add-ons and packages pre-installed. The macro images were analysed with respect of porosity. A macro for analysing porosity in the weld was developed and can be found in the appendix. As the macro images contained artefacts outside the weld, the weld were cropped manually out of the images before running the macro. The general work process in the Fiji software can be described as follows; adjust the threshold thus highlighting the porosity, using the find edges function to draw circles around the pores, using the fill holes function to obtain homogenous pores on a white

background, running the watershed function to separate pores that were connected and finally running the analyse particle function to count and measure the pores.

Figure 10 shows the image before and after the image analysis. In figure b) each pore has been numbered and measured. The result files for each image analysis were saved as a csv file.

The images along the weld were also analysed in Fiji using the same macro. An example of the results can be found in Figure 11.

a)

Figure 10 Macro image analysis of porosity a) shows sample 104 before analysis and b) after image analysis at 25.2x magnification.

b) a)

(20)

The SEM images at 150x magnification from the SEM were also examined and analysed in the Fiji software. Each sample was examined in six positions: three in the weld and three in the base material. Three macros were created to analyse the images. The first one for large pores above 750 pixels, corresponding to 293 μm² in area. The second one for the pores or cavities less than this value. The reason for dividing the range of sizes in two is to make a coarse separation between the large and round gas pores and the smaller shrinkage cavities. The set area separation is based on observation of the size of shrinkage cavities. It should be noted that this method will count small gas pores as cavities and also large shrinkage cavities as gas pores.

The last macro created was to analyse the secondary phase. This macro is slightly different than the others as the phase, which is analysed, is white instead of the dark pores. All macros can be found in the appendix. Figure 12 contain examples on what the three Fiji macros measures.

2.5 Thermodynamic calculation

To investigate the solubility of hydrogen under the elevated external pressure of the HPDC process thermodynamic calculations were made in the software Thermo-Calc 2015a using the database TCAL4. The AM50 alloy were simplified to a system containing

magnesium with five weight percent aluminium. Also the secondary β-Mg17Al12 phase was neglected to further simplify the system. Two calculations were performed, one at ambient pressure and the other at a pressure of 80MPa. The selection of the high pressure was set arbitrary in the range of pressures used for HPDC of magnesium. Due to the simplifications

Figure 12 Shows a sample image from the weld containing; a) large pores, b) shrinkage cavities and c) the white secondary phase consisting of β-Mg17Al12

b) a)

c)

(21)

Results

3 3.1 AM50

3.1.1 Effects of the polishing

The result obtained after polishing procedure 1 indicated a great variety amongst the base material. The surface were sometimes extremely rugged, exposing what looked like a very porous material, while other samples seemed almost completely homogenous. In Figure 13 some examples are given on the variation of appearance among the samples after polishing procedure 1.

a)

c)

b)

055

081

105

(22)

Figure 14 shows the same samples as Figure 13, after polishing procedure 2. The samples look homogenous and the cavities in the base material are no longer visible.

a)

c)

b)

Figure 14 The base material after polishing procedure 2 are not showing a significant increase in the amount of pores. Figure a-c) corresponds to samples 55, 81 and 105. All images are taken at 25.2x magnification, the size of the scale bar is 1mm.

055

081

105

(23)

A question that arises is whether the porosity, shown in Figure 13, is the correct representation of the material. The microstructure could be an artefact introduced by the polishing procedure. To investigate the true nature of the material, some of the samples were examined and photographed in the microscope. Images were taken after polishing procedure 2, the samples were then processed according to procedure 3 and photographed again in the same location. Figure 15 shows the micro images after polishing procedures 2 and 3.

a)

c)

b)

059

104

105

Figure 15 Micrographs at 100x magnification in the base material, the left image is taken after polishing procedure 2 and the right image after procedure 3 for the samples 59, 104 and 105. The size of the scale bar is 0.2 mm.

(24)

To further examine the material, trials were made with vibration polishing. As in Figure 15, two polishing methods were compared. Figure 16 shows micrographs after polishing procedure 2 and vibration polishing.

a)

c)

b)

017

019

020

Figure 16 Micrographs at 100x magnification in the base material, the left image is taken after polishing procedure 2 and the right after vibration polishing for the samples 17, 19 and 20. The size of the scale bar is 0.2 mm.

(25)

3.1.2 Microstructure

The composition and components of the microstructure were evaluated in the SEM, the base material and also the weld were examined. It was found that the components in both the weld and the base material were alike. In Figure 17, the different main components of the microstructure are shown.

The spectrums were analysed multiple times on different locations in both the weld and the base material for the different components in Figure 17.

Figure 17 The typical components of the AM50 microstructure were a) is the magnesium matrix, b) is the secondary β-Mg17Al12 phase, c) is a Mg-Al oxide, d) is a Al-Mn particle and e) is a cavity

d) e)

c)

b)

a)

(26)

3.1.3 Secondary phase and size distribution

The secondary β-Mg17Al12 phase is known to precipitate mainly in the grain boundaries for the base material [21], Figure 18 shows the secondary phase in the base material and the weld at 150x magnification.

a)

c)

b)

007

035

050

Figure 18 The left figures show the secondary phase in the base material and the right figures show the secondary phase for the weld at 150x magnification. a) is sample 7, b) is sample 35 and c) is sample 50

(27)

Figure 19 shows the secondary β-Mg17Al12 phase in the base material and the weld at a magnification of 500x, it can be seen that the secondary phase precipitate in the grain

boundaries.

a)

c)

b)

007

035

050

Figure 19 The left figure shows the secondary phase in the base material and the right shows the secondary phase for the weld at 500x magnification. a) is sample 7, b) is sample 35 and c) is sample 50

(28)

Figure 20 shows the secondary β-Mg17Al12 phase in the base material and the weld at a magnification of 1000x.

a)

c)

b)

007

035

050

Figure 20The left figure shows the secondary phase in the base material and the right shows the secondary phase for the weld at 1000x magnification. a) is sample 7, b) is sample 35 and c) is sample 50

(29)

The histograms in Figure 21 show the size distribution of the secondary phase particles for six of the samples. In each figure the base material of one sample is compared to the weld.

It should be noted that the particle area for sample 35, shown in Figure 21 b), is substantially reduced after welding.

a) b)

c) d)

e) f)

(30)

3.1.4 Cavities in the base material

The amount of cavities in the base material varies in the different samples. Some samples have a low amount of cavities whilst others have a large amount of cavities. Figure 22 illustrates the differences in the amount of cavities among four of the samples at 150x magnification.

There may also be some cavities in the surface layer in the base material, shown in Figure 23. The lower part of the image is covered by the conductive carbon tape, thus it is difficult to realise the full extent of the cavities in the surface layer.

035 050

076 126

a) b)

c) d)

Figure 22 Different amount of cavities in the base material at 150x magnification were a) show a low amount of cavities, b) shows a large amount of cavities, c) shows a mixture of cavities and oxide inclusions and d) show a variation in the cavities size and numbers

(31)

In Figure 24 the cavities are presented at 500x magnification. In Figure 24 b) the vast amount of cavities are visible and in c) an example of the surface oxide is clearly visible.

035

Figure 23 Cavities along the edge of the base material in sample 35

035 050

a) b)

c) d)

(32)

The cavities are shown at 1000x magnification in Figure 25.

035 050

076 126

a) b)

c) d)

Figure 25 Different amount of cavities in the base material at 1000x magnification were a) show a low amount of cavities, b) shows a large amount of cavities and c-d) shows a mixture of cavities and oxide inclusions

(33)

3.1.5 Origin of the cavities

The cavities in the base material are thought to be mainly shrinkage pores created during solidification. However, the size and shape of the cavity coincide to some extent with that of the secondary phase as can be seen in Figure 26.

Also fractured parts of the secondary phase remnant in cavities are present in some of the SEM images, as can be seen in Figure 27.

Figure 26 The similarities of the shrinkage cavities and the secondary phase

(34)

At 5000x magnification shown in Figure 28, the secondary phase is present inside a cavity. There are some small white particles visible at this magnification that possibly are residue from the polishing.

Figure 28 The secondary phase in a cavity at 5000x magnification

(35)

3.2 Hydrogen and gas content in the base material

3.2.1 Solubility of hydrogen in Magnesium

The solubility of hydrogen is higher in the liquid phase than in the solid for the AM50 alloy. The effect of the external pressure in the HPDC process on the solubility of hydrogen is presented in Figure 29. The solubility of hydrogen, at the beginning of solidification,

increases almost 29 times. The mass percent hydrogen increases from 2.97 x 10^-3 at ambient pressure to 8.56 x 10^-2 mass percent at 80MPa. Also the solubility of hydrogen, as the last melt solidifies is increased 39 times. The mass percent hydrogen at complete solidification increases from 1.43 x 10^-3 at ambient pressure, to 5.60 x 10^-2 mass percent at 80MPa.

Figure 29 The solubility of hydrogen in the liquid and solid magnesium calculated in Thermo Calc at a) 0.1 MPa pressure and b) 80 MPa pressure

a) b)

(36)

3.2.2 Pressure inside the base material

During the polishing of sample 106, a remarkable phenomenon occurred. When

examining the sample post polishing an elevation was detected in the base material, shown in Figure 30. This is possibly due to a high internal gas pressure inside the material.

Figure 31 shows the elevation of sample 106 as seen in the SEM.

Figure 31 SEM image of the elevation in the base material of sample 106

Figure 30 The elevation in the base material of sample 106

(37)

3.3 Porosity

3.3.1 Porosity in the weld cross section

3.3.1.1 All of the samples combined

The effect of the welding speed on the percent porosity and number of pores per area for all of the samples combined, can be found in Figure 32.

The effect of the welding power on the percent porosity and number of pores per area for all of the samples combined, can be found in Figure 33.

Figure 32 The welding speed in relation to a) the number of pores per area and b) the percent pores in the weld for all of the samples

(38)

The effect of the spot size on the percent porosity and number of pores per area for all of the samples combined, can be found in Figure 34.

The focusing effect on the porosity can be viewed in Figure 35.

Figure 34 The laser spot size in relation to a) the number of pores per area and b) the percent pores in the weld for all of the samples

(39)

Figure 36 shows porosity in comparison to the three welding methods, were STD is the standard single beam laser welding, twin 50-50 is dual beam using beam splitter and twin head is dual beam using two optics.

Figure 36 The effect of the welding method on the a) amount of pores and b) the percent pores, STD is the standard single beam welding while twin and twin head uses dual beams

(40)

3.3.1.2 Single beam and dual beam methods

To narrow down the parameters affecting the porosity, the porosity were examined by separating the welding methods into single or dual beam processes. In Figure 37 and Figure 38 the welding speed is plotted against the porosity for single and twin spot welding

respectively.

Figure 38 The welding speed in relation to a) the number of pores per area and b) the percent pores in the weld for the twin spot welded samples

Figure 37 The welding speed in relation to a) the number of pores per area and b) the percent pores in the weld for the single beam welded samples

(41)

Figure 39 and Figure 40 show the power plotted against the porosity for single and dual beam respectively.

Figure 40 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the twin spot welded samples

Figure 39 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the single beam welded samples

(42)

The effect of the spot size on the percent porosity and number of pores per area for single and twin spot welding can be found in Figure 41 and Figure 42.

Figure 42 The laser spot size in relation to a) the number of pores per area and b) the percent pores in the weld for the twin spot welded samples

Figure 41 The laser spot size in relation to a) the number of pores per area and b) the percent pores in the weld for the single beam welded samples

(43)

To further reduce the available parameters affecting the porosity, the welding speed were evaluated at constant power. The power selected for the single beam and dual beam welded samples were 2200W and 1700W respectively, shown in Figure 43 and Figure 44.

These selections were based on the power group with the largest amount of samples.

Figure 44 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the twin spot welded samples at the power 1700W

Figure 43 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the single beam welded samples at the power 2200W

(44)

The welding power were also evaluated at constant speed. The speed was selected for the single beam and dual beam welded samples to 3 m/s, the results is shown in Figure 45 and Figure 46. The selection was based on the speed with the largest amount of samples.

Figure 46 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the twin spot welded samples at a welding speed of 3 m/s

Figure 45 The welding power in relation to a) the number of pores per area and b) the percent pores in the weld for the single beam welded samples at a welding speed of 3 m/s

(45)

3.3.2 Porosity along the weld

The porosity was also examined along the weld for 14 of the samples to validate the results obtained from the single cross sections, shown in Figure 47. The results show reasonable agreement for 11 out of the 14 samples. In the samples 105, 121 and 126 the porosity along the weld is substantially lower compared to the initial result from analysing the cross section.

Figure 47 The porosity along the weld compared to the porosity of the cross section for 14 samples were a) shows the number of pores per area and b) the percentage of porosity

(46)

3.3.3 Porosity related to the base material

The relation between the pore formation and the base material is represented by the Box-plots in Figure 48. The grouping in Figure 48 is purely by the base plate, hence, there may be large differences in the welding method and parameters in some plates while others have less diverse parameters. The plate that diverges the most from the others are plate number 23. This deviation may be an artefact caused by cutting the sample too close to the starting or end position of the weld. Samples 121 and 126 in plate 23 were cut and analysed along the weld. Both of them resulted in a significantly reduced amount of pores along the weld in comparison to the cross section as can be seen in Figure 47. The amount of pores in these samples are still high, but they are more in line with the results from the other plates.

Figure 48 Box-plots of a) the number of pores per area of the weld and b) the percentage of pores in the weld, both are grouped after the original plate number they were welded on

a) b)

(47)

3.3.4 Porosity related to cleaning before welding

The plates were cleaned before the welding using five methods. In Figure 50 and Figure 49 the porosity is plotted against the cleaning method for the single and dual beam method respectively. The number of samples in each group can be found in Table 1. It should be noted that the majority of the samples were performed using the brushing and acetone procedure.

Figure 50 The different methods of cleaning the samples before welding plotted against a) the number of pores per area and b) the percentage of porosity for the single beam welded samples

(48)

3.3.5 Porosity related to the welding parameters

In order to examine the effect of the welding parameters on the porosity, samples welded using similar parameters were grouped in pairs and evaluated. Table 5 shows the welding parameters, the porosity and the plate number for the samples selected.

Table 5 The welding parameters and porosity for samples being further examined

Sample Method Spot size Twin

distance Power Speed Cleaning Plate

number % Pores in weld

[mm²] [mm] [kW] [m/s] 1st 2nd Cross

section Along

7 STD 0.660 - 2.2 3 B A 1 8.7 10.6

21 STD 0.660 - 1.1 1.5 B A 6 9.3 10.6

35 STD 0.660 - 2.2 3 B A 6 2.9 4.3

50 STD 0.660 - 2.2 3 B - 8 5.4 6.4

76 Twin

50-50 0.625 1 2.2 3 - - 14 2.3 1.7

104 Twin

50-50 0.313 0.50 2.2 3 B A 19 18.0 13.5

105 Twin

50-50 0.313 0.50 1.1 3 B A 19 16.6 7.7

106 Twin

50-50 0.200 0.32 1.1 3 B A 14 4.8 8.1

121 Twin

head 0.625 1 2.0 2 B A 23 30.4 15.0

122 Twin

head 0.625 1 2.0 2 B A 24 4.3 5.3

125 Twin

head 0.625 1 2.0 3 B A 24 6.0 10.8

126 Twin

head 0.625 1 2.0 3 B A 23 31.5 19.2

B = Brushing, A = Acetone, - = No cleaning

(49)

The samples were grouped into nine pairs according to Table 6. One of the samples was chosen as a reference and the data in Table 6 shows the welding parameters and percent porosity in comparison to this reference. The grouping were mainly made to combine samples with similar welding parameters that differed in porosity. Four of the groups were also made with welds on the same base plate, this is to limit the effect of the base material when

comparing the welding parameters to the porosity.

Table 6 The groups created for comparing similar samples

Group

number Samples

Difference compared to

Reference

Comment Δ %Pores compared to Reference

# Reference Cross Along

1 7 50 Plate, Cleaning - Brushing -3 -4.2

2 21 35 Power, Speed

2x Speed and Power 35: Double

weld

-6.4 -6.3

3 76 104

Plate, Spot size, Distance, Cleaning

-0.5x Spot and Distance 76: No cleaning

+15.7 +11.8

4 105 106 Plate, Spot size,

Distance

-0.33x Spot and

Distance -11.8 +0.4

5 121 122 Plate, Focus +5 Pre laser

Focus -26.1 -9.7

6 125 126 Plate, Focus

125: +20 Pre laser Focus 126: +10 Post

laser Focus

+25.5 +8.4

7 104 105 Power -0.5x Power -1.4 -5.8

8 122 125 Speed, Focus

1.5x Speed 122: +5 Pre laser Focus 125: +20 Pre

laser Focus

+1.7 +5.5

9 121 126 Speed, Focus

1.5x Speed 126: +10 Post

laser Focus

+1.1 +4.2

(50)

Group 1

Group 1 consists of two samples welded with the same parameters, the differences are the base material and that sample 50 did not receive any brushing prior to welding. In Figure 51 the samples are presented after polishing procedure 1.

a)

b)

007

050

Figure 51 Samples a) 7 and b) 50 in group 1 after polishing procedure 1. All images are taken at 25.2x magnification, the size of the scale bar is 1mm.

(51)

As can be seen in Figure 51, sample 50 has a number of large cavities in the base material. The percent of pores is three percent less in sample 50 compared to sample 7, which can be seen in Figure 52

Figure 53 shows the cut along the weld for samples 7 and 50. The analysis of the cut

a)

b)

007

050

a)

b)

Figure 53 Samples a) 7 and b) 50 in group 1 after polishing procedure 2, cut along the weld. All

(52)

The base material is shown in Figure 54 for three positions in each sample. The base material of sample 7 looks generally more homogeneous and the secondary phase is less dominant.

Figure 55 shows three positions in the weld for each sample. The gas pores in sample 7 is spherical whilst the gas pores in sample 50 is somewhat deformed. Sample 50 also show more inclusions and smaller cavities than in sample 7.

007

a)

050

b)

Figure 54 The base material of samples a) 7 and b) 50 in group 1 at 150x magnification in the SEM taken at three positions in each sample

007

a)

b)

(53)

Figure 56 shows a histogram of the pore radius along the weld for samples 7 and 50.

The count is presented as a logarithmic scale as the large pores are few in number. They are, however, not negligible for the weld performance. In order to clarify, the pore radius of the histogram is divided into three regions, the first region is for radius below 0.05mm and is annotated as small pores. The second region in-between 0.05mm and up to 0.1mm is called medium sized pores and the pores larger than 0.1mm is called large pores. Sample 7 and sample 50 have similar amount of large pores, the smaller and medium sized pores is generally less in sample 50.

Figure 56 Histogram of the pore radius in counts per 100 mm² along the weld for the samples in group 1

0.00 ≤ S < 0.05 0.05 ≤ M < 0.1 0.10 ≤ L < ∞

(54)

Figure 57 and Figure 58 shows the area distribution of the secondary phase before and after the welding for the samples in group 1. The welding process reduces the size of the particles in both samples.

Figure 57 A histogram of the area distribution of the secondary phase for the base material in group 1

(55)

Group 2

Group 2 consists of samples 21 and 35 shown in Figure 59. Sample 35 is welded with twice the speed and power compared to sample 21. Sample 35 is also a double weld, meaning that after the first run the soot was removed and the sample were welded a second time with the same parameters.

a)

b)

021

035

(56)

The base material is the same for sample 21 and 35, the similarities can be seen in Figure 60. There are significantly less large pores in the weld in sample 35 than in sample 21, and the analysis show that sample 35 has 6.4 percent less pore area in the cross section.

In the cut along the weld shown in Figure 61 the differences are obvious, sample 21 shows a coalescence of large pores in the centre of the weld. The image analysis show that sample 35 has 6.3 percent less pore area in the weld, in agreement with the value from the cross section.

a)

b)

021

035

a)

(57)

Figure 62 a) shows the base material for sample 35, in b) the weld of sample 35 is presented. The gas pores in the weld of sample 35 is not perfectly spherical but slightly deformed.

Figure 63 shows the microstructure of sample 35 at 1000x magnification. It is

interesting to note that the precipitation of the secondary phase in the weld is very fine and in almost completely spherical particles.

035

a) BM

035

b) Weld

Figure 62 The base material a) and the weld b) is shown for sample 35 at 150x magnification in the SEM

BM Weld

Figure 63 Sample 35 at 1000x magnification in the SEM, figure a) shows the base material and figure b) shows the weld

(58)

A histogram is shown in Figure 64 of the pore radius along the weld for samples 21 and 35. It can be seen that sample 35 has no large pores and fewer medium and small size pores than sample 21.

0.00 ≤ S < 0.05 0.05 ≤ M < 0.1 0.10 ≤ L < ∞

(59)

Group 3

Group 3 consists of samples 76 and 104 shown in Figure 65. Both samples were welded using the dual beam method and the same parameters. The spot size and distance between the spots were halved in sample 104 in comparison to sample 76. Another difference is the base material and that sample 76 were not cleaned in any way before welding. The surface of the weld of sample 76 showed indications of an unstable welding process, there were surface blowouts and small defects.

a)

b)

076

104

(60)

The amount of porosity is about 16 percent higher in sample 104 than in 76, this can be seen in Figure 66.

The difference in porosity can easily be observed in the cut along the weld shown in Figure 67. Sample 104 has about 12 percent increased porosity along the weld in comparison to sample 76.

a)

b)

076

104

a)

b)

(61)

The base material and weld for sample 76 can be seen in Figure 68, both the base material and the weld consists of inclusions and cavities. The gas pores in the weld are few and irregular in shape.

The base material of sample 104 can be seen in Figure 69. As can be seen the base material is not homogenous and contains a large quantity of small pores.

076

a) BM

076

b) Weld

Figure 68 The base material a) and the weld b) is shown for sample 76 at 150x magnification in the SEM

Figure 69 The base material of sample 104 at three positions after polishing procedure 3, at 100x magnification. The size of the scale bar is 0.2 mm.

(62)

Figure 70 shows a histogram of the pore radius along the weld for samples 76 and 104.

It can be seen that sample 76 lacks large pores. Sample 76 show less porosity all over the size range.

0.00 ≤ S < 0.05 0.05 ≤ M < 0.1 0.10 ≤ L < ∞

A Study in How Welding Parameters Affect the Porosity in Laser Welded High Pressure Die Cast AM50 Magnesium Alloy

A Study in How Welding

Parameters Affect the Porosity in

Laser Welded High Pressure Die

Cast AM50 Magnesium Alloy

EDWIN BERGSTEDT

Abstract

Sammanfattning

Table of contents

List of abbreviations

Nomenclature

Introduction

1

1.1 Background

1.2 Aim of the thesis

1.3 Laser welding

1.4 Magnesium alloys

1.5 High pressure die casting

1.6 Gas porosity in HPDC metal

Method

2

2.1 Cleaning

2.2 Welding parameters

2.3 Sample preparation

2.4 Analysis

2.5 Thermodynamic calculation

Results

3

3.1 AM50

a)

c)

b)

055

081

105

a)

c)

b)

055

081

105

a)

c)

b)

059

104

105

a)

c)

b)

017

019

020

d) e)

c)

b)

a)

a)

c)

b)

007

035

050

a)

c)

b)

007

035

050

a)

c)

b)

007

035

050

035 050

076 126

a) b)

c) d)

035