Electrolytic Extraction of Aluminium Bifilms

(1)

DEGREE PROJECT IN TECHNOLOGY, FIRST CYCLE, 15 CREDITS

STOCKHOLM, SWEDEN 2020

Electrolytic Extraction of

Aluminium Bifilms

SIMON BERGFORS

DAVIDA FLINK

(2)

Abstract

Bifilms is the oxide layer created between two surfaces in the melt of light metals that form an oxide layer. These become planar inclusions in the final casting and are problematic for the mechanical properties such as cracks and crack initiations. Bifilms are too thin to be viewed properly in two dimension cross-section method as they will only appear as thin lines. Because of this, it is relevant to test if it is possible to use electrolytic extraction (EE) as a alternative method to investigate bifilms. Both the deeply etched surface and the inclusions on a filter from the extraction are looked at in the scanning electron microscope (SEM) to get an understanding of the size and shape of the inclusions. With this, a greater understanding of these types of defects can be achieved.

After both the filtered inclusions and the surface are examined in SEM with images and Energy-dispersive X-ray spectroscopy (EDS), the images are measured in the software ImageJ. The measurements and analysis show that it is probably bifilms and that they can be relatively large, and not so circular.

However, the measurements with the filter have shown high levels of oxygen and carbon. Some levels of chlorine, nitrogen and iron have also been found. But if the surface is compared to the metal surface, it can be concluded that it is likely that bifilms have been found. There are sufficient levels of aluminum and oxygen present. Images in SEM also show the appearance of film-like inclusions.

If the method of electrolytic extraction is to be improved to investigate bifilms, optimizations such as filters of other compositions are recommended.

Keywords

Aluminium, Bifilms, Oxide films, Electrolytic extraction, Scanning electron microscope

(3)

Sammanfattning

Bifilms är det oxidskikt som skapas mellan två ytor i smältan hos lätta metaller som bildar ett oxidskikt. Dessa blir sedan plana inneslutningar i den slutliga gjutningen och är problematiskt för de mekaniska egenskaperna i form av sprickor och sprickinitieringar. Bifilms är för tunna för att de ska kunna ses korrekt i en tvärsnittsmetod, eftersom de bara kommer att visas som tunna linjer. På grund av detta är det relevant att testa om det är möjligt att använda elektrolytisk extraktion (EE) som en alternativ metod. Både den djupt etsade ytan och inneslutningarna på ett filter från extraktionen tittas på i svepelektronmikroskop för att få en förståelse för inneslutningarnas storlek och form. I och med det kan en högre förståelse uppnås för dessa typer av defekter.

Efter att både de filtrerade inneslutningarna och ytan granskats i SEM med bilder och Energy-dispersive X-ray spectroscopy (EDS), mäts bilderna i en mjukvara, ImageJ. De mätningarna och analyserna visar att det antagligen hittats bifilms och att de kan vara förhållandevis stora, samt inte så cirkulära.

Däremot har mätningarna med filtret visat höga halter av syre och kol. Även vissa halter av klor, kväve och järn har hittats. Men om ytan jämförs med metallytan så kan en slutsats dras om att det är troligt att bifilms har hittats. Där finns tillräckliga halter av aluminium och syre. Även bilder i SEM påvisar filmliknande inneslutningar till sitt utseende. Om metoden med elektrolytisk extraktion ska förbättras för att undersöka bifilms behövs optimeringar, som till exempel filter av en annan sammansättning.

Nyckelord

(4)

Authors

Simon Bergfors <sgvbe@kth.se> and Davida Flink <davidaf@kth.se> Department of Material Science and Engineering

School of Industrial Engineering and management (ITM) KTH Royal Institute of Technology

Place for Project

Department of Material Science and Engineering

School of Industrial Engineering and management (ITM) KTH Royal Institute of Technology Stockholm, Sweden

Examiner

Anders Eliasson

Department of Material Science and Engineering, Applied Process Metallurgy School of Industrial Engineering and management (ITM)

KTH Royal Institute of Technology

Supervisors

Andrey Karasev

KTH Royal Institute of Technology Hamid Doostmohammadi

(5)

1 Introduction

1.1 Purpose

Inclusions are a well-studied area. They are directly linked to the mechanical properties of alloys and there are many methods to investigate them, for example electrolytic extraction [1]. However, there are not as many studies of oxide films in light alloys, a kind of inclusion that also play a major role in the mechanical properties of the alloy [2]. The purpose of the report is partly to look at how well the method of electrolytic extraction works to investigate inclusions and double oxide films on a film filter and also on a aluminium sample surface. With the help of SEM we can then take high resolution images of the prepared samples and also see what kind of substances are present.

1.2 Aim

Can Electrolytic Extraction be used as a validated method for investigating bifilms in an aluminium matrix? What kind of inclusions and films have been found? What is their morphology, compositions, size, circularity and area?

1.3 Previous work

Similar works in this field, has among others, been about removing bifilms from aluminum melt with the help of stirring in a cylindrical rotor. This was than calculated using numerical analysis. It showed that one can change the amount of bifilm in a melt using different speeds and techniques and therefore provide a greater explanation for how to increase and decrease this type of entrapment [3]. Other similar work has been done on steel for non metallic inclusions by Du Hongying. In this paper, they have categorized inclusions by image analysis. First, the steel samples have been treated with or without a calcium treatment and then electrolytically extracted for viewing in a scanning electron microscope (SEM). The report shows that the inclusions vary slightly from the different zones on the samples and also with and without calcium treatment [1]. However, it must be remembered that this report deals with steel inclusions and not inclusions in an aluminum matrix.

(8)

1.4 Goal

A successful extraction where the inclusions can be viewed in SEM and Energy-dispersive X-ray spectroscopy (EDS) and the results can be converted into measurable data of the inclusions. Then the study can be evaluated and see if it is possible to measure bifilms with electrolytic extraction or not. If the method does not work, hopefully there will be enough data to explain why.

1.5 Benefits, Ethics and Sustainability

Bifilms are created in the melting process. This means that every time aluminum is recycled, more creation of bifilms occur. It results in a deterioration of the casting quality, hence also the end product. With good methods for investigating bifilms, understanding the phenomena, and how to take accurate countermeasures, this would result in a more durable and sustainable life cycle for aluminum and other metals with formation of bifilms. Longer life cycles of aluminum will have a major impact on the environment as less energy is required in recycling aluminum than the energy required to create aluminum from bauxite. With stronger aluminium, the demand and thus the production will increase. Primary production of aluminium is a great energy consumer, but with increased knowledge of bifilms, lighter and less material is needed because of stronger mechanical properties.

Some chemicals used in the electrolytic extraction process does have negative environmental effects if released into the normal sewage system, and must be addressed when used on industrial scale. However, from the perspective of this work, it is a very small quantity of electrolyte that makes it noticeably harmful to the environment.

(9)

2 Theoretical Studies

To comprehend the results from the method used, a theoretical study of bifilms and the procedure of extraction methodology is included. Also, some brief explanations of investigative technical aids is also included in this thesis.

2.1 Bifilms

Aluminum forms an oxide layer (Al2O3) that surrounds the metal after being

exposed to air. This layer will form immediately and protects the metal from corrosion [4]. However, this is also how bifilms will form. Bifilm is a thin, double oxide layer between two liquid surfaces in a metal matrix as seen in Figure 2.1 The thickness may vary somewhat between different films, but approximately 20nm is an approximate estimate according to Campbell in aluminum alloys. Thus, such a film consists of only a few atoms in thickness. It is also much thinner than studies first thought, so it has been chosen to categorize it into two different types of films, old and new. The old films found were approximately between 10-1000 µm and the new ones vary between 1nm-1µm. Because of their thin nature they are hard to detect but when the melt solidifies they will behave like cracks and significantly affect the mechanical properties in the casting [5].

Figure 2.1: Example image of the formation of a bifilm [5]

Bifilms have for a long time not been detected because of their thin nature. It is not until recently and with SEM technology we can find them. Even with this technology they can sometimes be hard to see. If seen in SEM they have a

(10)

cloud-like appearance and are very thin, 1nm-1µm in aluminum alloys. Earlier studies may have also shown bifilms in a study made by Rooney in 1992, Campbell explains. They have found narrow cracks that can even turn into round bubbles if the volume of gas is high. These gas bubbles are thus pores of hydrogen. Campbell means these can only exist with the help of bifilms [5].

2.1.1 Formation of oxide layer

The oxide layer covers the liquid melt and if the surface then folds together before solidification the oxides get encapsulated between the surfaces. Sometimes there may also be bubbles and inclusions of another phase together with the oxide layer, pores and non metallic inclusions. This happens specifically during the process when the metal is poured. Small droplets of metal end up on the surface when the melt is poured and the oxide layers get trapped in the melt and folded together, there for the name bifilms, as in a double layer oxide film. This process can also cause the film to fold several times and not necessarily a double fold as the name suggests. Although it is common that they are created when a melt is in motion, they can also emerge in other ways, for example through diffusion, interstitial diffusion or substitutional diffusion. Although both of these are a much slower process and the rate of diffusion varies between different metals. For this to happen, energy is needed so the surface can react with the environment and then diffuse into the melt. Campbell believes that most often this starts with small amorphous oxide layers (Al2O3) and then grows to subsequently become

a crystalline structure [5]. The reaction can be seen in Equation (1)

2 Al(s) +3

2O2(g) −−→ Al2O3(s) (1)

2.1.2 Mechanical proporties

It is important to understand and study bifilms because of their high impact on the mechanical properties as mentioned above. These effects can be for example cracks, hot tears, gas microporosity, leakage defects, grain boundary failures and grain boundary failures. [5] The size of the films are also important. For example, small and large inclusions in steel as seen in Figure 2.2 have different impact on

(11)

fatigue resistance as seen in a study made by Joseph Maciejewski 2015 [6]. One can only assume then that the same applies to aluminum inclusions.

Figure 2.2: Large and small inclusions in steel [6]

2.2 Extraction Methods

Electrolytic extraction (EE) is a potentiostatic extraction method based on running a small current through the sample to dissolve the metal matrix around the inclusions. The sample is first cleaned thoroughly from fats and residue with solvents and/or alcohol to get even conductual surfaces. The sample is then placed in an electrolyte as an anode connected to a static powersource. A platinum cathode, connected to the power source, is then lowered into the electrolyte, acting as a anode. Power is then switched on and carefully measured by volts, amperes and coulombs, to execute the process [7].

A study conducted by Y. Kanbe shows that EE is a further accurate method, on determining the amount and size of inclusions, then the well known 2D cross section method (CS-method) where the sample is cut and polished [8].

D. Janis writes about electrolytic extraction as one of two commonly used methods for extracting inclusions from metal. The other method is by acids or strong solvents, such as halogen-alcohols, is known as chemical extraction. Acids or strong solvents cause small inclusions to deteriorate and disappear or to be reduced in size, and will, as a result, affect the measured parameters on

(12)

the inclusions. The rate of dissolving aluminium oxide powder, Al2O3, was

the largest using a acid solution with 20% after 5 hours. Halogen alcohols, such as bromine-methanol or iodine-methanol, had a percentage of 0.3-0.5% of deterioration, after 10 hours in the solution. The method with least rate of dissolving the aluminium oxide powder, was the electrolytic extraction, which had a rate of 0.1% being dissolved after 11 hours in a 10%AA (10% acetylacetones–1% tetramethylammonium chloride–methanol) solution. Other non metallic inclusions such as magnesium- and calcium-oxide is also known to have a faster rate of deterioration in the chemical extraction methods mentioned above [9].

2.3 Etched metal surface as a result of EE

The electrolytic extraction results in non attached inclusions, but it also results in a significantly deep etched surface on the remaining metal sample involved in the process. This allows for a investigation of the inclusions still half stuck in the matrix.

2.4 Scanning Electron Microscope

Due to the demand for resolution where science need to see smaller and smaller objects in our microscopes, electron beams are being used instead of light beams. By shooting an electron beam you can get sharp images of objects in the scale of nanometers. To get these pictures correct, preparations of the samples are required, for example, placing the sample in a bath of ethanol. Afterwards, the sample must be dried completely. The surface must be relatively flat and can be covered with a layer of gold, only a few atoms in thick, if the material needs help with conductivity. If the images are successful, an x-ray spectroscopy analysis can be performed to find out which elements are present in the sample. In order to achieve as accurate results as possible, the method of electron beam to use, must be considered. Secondary electron detector and back-scattered electron detector is two methods. The data is then analyzed in a software and the results will be presented in atomic number and atomic weight [10].

(13)

3 Method

The aluminium samples used for this thesis is not a work of the authors. The samples were produced with the intent of a high content of bifilms per unit volume. The same principles as is explained in an report by A. Ahmadpour on the effects of stirring on double layer aluminium oxide films. [11] A commercially pure aluminium (99.9 % ) was used as a starting material, and was produced by our supervisor Hamid, and a student of his [12]. The samples was taken from, as can be seen schematically in Figure 3.1, different positions. Since the focus of this thesis is not the casting, but rather the method of getting information on the casted metal itself, the casting method will not be further disclosed or extensively discussed in this thesis.

Figure 3.1: The cutting of ingot and the samples original locations.

3.1 Electrolytic Extraction and Filtering

Electrolytic Extraction is a method of extracting inclusions from a metal sample. Dissolution of the metal matrix with the help of electric current and electrolyte, frees the inclusions which falls to the bottom of the extraction vessel. A schematic picture of this can be viewed in Figure 3.3a. After the extraction is done to the required amount, the solution with dissolved metal and undissolved inclusions can be filtered and examined in a SEM.

A pure aluminium sample, especially produced to have a high amount of bifilms per volume, was used for extraction. The 15.20x10.36x4.94(mm) sample was firstly cleaned with acetone and then petroleum benzene, to remove any residue, fats and dust. Then the sample was cleaned in ultrasonic bath of acetone for 2-3 minutes. All material used during filtering and extraction was cleaned with water, deionized

(14)

water and 3 times with methanol, to minimize contamination. Weighing of the aluminium sample was then conducted, and a pre-extraction and post-extraction weight was noted. To avoid extraction from all surfaces of the sample, the piece was covered in parafilm on all sides except one large area side. The parafilm was used as an insulator against the extraction fluid, and thus the constant current. To secure the parafilm and avoid penetration between the layers, a thin strip of tape around the edges was used. 250 (+-5) ml of 10%AA(10% acetylacetones–1% tetramethylammonium chloride–methanol) was transferred to a 500 ml beaker of cylindrical type. The sample was fastened in a platinum tweezer and this assembly was rinsed 3 times with methanol, to ensure cleanliness. This was then lowered into the beaker with the electrolyte through a glass lid with a hole in it, and fixing the depth about 5 mm under the surface of the electrolyte. A ring shaped platinum wire was fixed in the beaker, at the same depth as the middle point of the sample, and then connected with electrical wiring and tweezers. A schematic picture can be seen in Figure 3.3a. The wires were then connected to to a potentiostatic power source, a Vanaco VE-8, with the sample acting as an anode and the Pt-ring as an cathode. The power source was set to an electrical current and voltage, and the coulombs were measured using a coulomb meter and controller, setting it to cut power at 301 coulombs. During extraction the voltage, current and coulombs were noted to detect any fluctuations in the process. The numbers acquired can be partly seen in graphics in Figure 3.2, and in Table 3.1.

Figure 3.2: Time and charge shown on all measured points.

After the extraction stopped, the assembly of the sample was done in reverse order as disassembly. The solution was poured to a glass funnel with a ceramic filter and

(15)

Table 3.1: A table over measured parameters in the extraction process.

a 0.4 µm Whatman polycarbonate filter. This setup can be seen in 3.3b The beaker used for extraction, remaining sample, Pt-ring and tweezers was rinsed multiple times with methanol to ensure minimal loss of extracted inclusions. The fluid from rinsing was also poured into the funnel for filtering. A glass lid was placed on top of the funnel to lower the risk for contaminants. To help the process of filtration an aspirator was connected to the assembly of a flask with the funnel and filter attached. Minimal amounts of vacuum was applied to keep the film filter intact. Further drying of the film filter was done using the aspirator. The dry film filter was then carefully transferred to a closed sample container, making sure not to crease the filter or wrinkling from further drying.

(a) (b)

Figure 3.3: Setup of (a) potentiostatic setup and (b) filtering device. [13] The remaining aluminium sample was also transferred to a closed sample carrier. To prevent any bifilms or other inclusions being broken off from the etched

(16)

surface, the sample was not cleaned in ultrasonic bath, but only rinsed 3 times by methanol.

Dust being present in the lab environment, measures were taken in the entire process, in all steps, to minimize the amount of dust particles, and other contaminants on the finished filter, as this might affect the results of compositions.

All extractions and filtering conducted for this report had the same procedure. Diligence for contamination was also shared between all extractions and filtering.

3.2 Scanning Electron Microscope

Preparations for the SEM examinations of the extracted inclusions included cutting and mounting of a 1/16 part of the filter on a sample disc. Cutting was done with an exacto knife for a clean partial cake like piece of the filter. The film filter was then mounted with dual sided carbon tape to a sample disc. After the initial preparations the sample disc mounts on a rotational table for several discs, and the rotational table is mounted on the SEM work surface. Searching for inclusions was conducted in a somewhat systematic manner starting with the top part of the film filter section and scanning randomly for signs of inclusions downwards to the middle part of the filter using a SEM with energy-dispersive x-ray spectroscopy (EDS). Once an inclusion was found, pictures were taken, and compositions of several points was established if deemed necessary. A few reference points were also taken considering the film filter itself, to establish its composition and ratio of its elements. Another type of search was also done by reducing the magnification to get a larger filter area to view. This was done to accommodate a possible counting of the amount of films per area, and make calculations of the amount of bifilms per extracted volume of matrix.

Execution of the examination after EE, on the etched surface of the metal, was performed in a similar manner. But this was without the presence of a film filter, so no reference values of compositions was acquired, but only compositions at the objects of interest. The surface, being very uneven, gave a greater difficulty to finding inclusions in the somewhat topographical myriad of different lattice

(17)

directions and grain boundaries. This can be seen in Figure 3.4. When an object of interest was found, the same procedure as with the filter was conducted.

Figure 3.4: Picture taken with SEM of inclusion on the etched metal surface.

3.3 Measurements in ImageJ

The images are posted one at a time in ImageJ. The scale is firstly set hence ImageJ measures in pixels. The result of a marquee tool can be seen in Figure 3.5.

(a) Picture from SEM in original form. (b) Screenshot of inclusion beeing measured in ImageJ.

Figure 3.5: The method that most of the inclusion were measured by. The measurements were made by hand in the program with points clicked around the contours of selected objects of interest. When the entire object is marked,

(18)

ImageJ calculates the selected data, area, circularity, and Feret’s diameter. Some inclusions, with visual differences to films and with a high contrast ratio to the film filter, was measured by enhancing contrast further, find edges tool and tracing tool. It is important to know the size of the various inclusions in order to determine how they can affect the properties. For example, larger and more elongated, angular inclusions can be more harmful than small fully rounded ones. The measurements were then copied into excel files to be categorized and evaluated.

(19)

4 Results

The following results are what was received using the methods explained. The compositions are presented in atomic percentage due to the convenience of seeing direct ratios of elements in bifilms, intermetallic and non metallic inclusions.

4.1 Compositions of Bifilms

Compositions has been a vital part of identifying multiple layered aluminium oxide films that are mentioned as bifilms in this report. It is also a necessary part of analysing other inclusions that the authors found. The compositions will be presented with EDS spectrum, atomic percentage of elements in the composition recorded, by element and their fraction of a second element by numbers and by plotting these. Visual confirmation of bifilms is the first step and examples can be seen in Figure 4.1.

(a) Film like inclusion on the surface of film filter.

(b) Bifilm on the etched metal sample.

Figure 4.1: Examples of how the bifilms can look on their respective matrix. And the spectral analysis for the inclusions above can be seen in Figure 4.2 Looking at Figure 4.2c that is from the metallic surface sample, the spectral peaks of oxygen and aluminium are dominant. This high amount of oxygen originates either from a aluminium bifilm, or a oxide layer formed by oxygen in the air after electrolytic extraction. But a protective oxide layer that aluminium forms in air at normal atmosphere is only a few nanometers thick [5], and thus does not contain

(20)

(a) Point 3 in Figure 4.1, the surface of the film filter

(b) Point 4 in Figure 4.1, the surface of the film filter

(c) Point 3 in Figure 4.1b, the etched surface (d) Point 4 in Figure 4.1b, the etched surface

Figure 4.2: Spectrums of multiple points of interest, The upper row shows a

major part of carbon content, and the lower a majority of aluminium.

these high values of oxygen. It would barely be seen as a haze on the filter because of its thin nature. The possibility of this type of distribution of aluminium and oxygen to be a oxide layer formed post extraction is thereby ruled out. It is most probable to be an actual multiple layer of aluminium oxide and thus an aluminium bifilm. The high ratio of aluminium compared to the oxygen is to be considered to be an effect of the metal matrix because of small or thin inclusions [14].

In the filter samples, there is one picture that show bifilms of a thicker nature. This can be seen in Figure 4.3 where no filter is visible through the bifilm and no carbon is present in its composition. The ratio of aluminium to oxygen has a deviation of 5% with Al2O3. This is therefore a point to see how bifilms appear on

film filters.

(21)

clearly seen in the upper right corner around the inclusion. Viewing the results from analysis in the figure text, a substantial percentage of carbon and oxygen is seen, with a small percentage of aluminium. The effect of matrix or filter on the measuring points for small or thin inclusions is known from a report written 2010 [14]. Taking out the known ratio of carbon and oxygen that comes from the film filter, there is still aluminium and oxygen left in the composition to be a bifilm present.

Figure 4.3: A thicker bifilm on the film filter

Figure 4.4: A thin layer of bifilm on the film filter with point 15 showing 87.7%

carbon, 11.86% oxygen, and 0.45% aluminium.

Compiling the results of the spectral analysis, all the points were put in to tables that could be interpreted, as can be seen in Figure 4.5. The carbon content is divided by the oxygen content as the vertical axis, one can also see a fraction factor of the filter composition. The polycarbonate filter used has 89.2% carbon and 10.8% oxygen. This separates which points have a higher relation of carbon and

(22)

oxygen then the filter, and which ones has a lower relation. The lower relations has excess oxygen beyond the factor of the filter, hence it gives the room for stoichiometric compounds that contain oxygen, in the composition.

Figure 4.5: Factor of carbon% devided by oxygen% and the factor of the

polycarbonate film filter.

The graph in Figure 4.5 shows that the carbon content is low compared to the oxygen in the me1 sample. This is the etched metal surface sample that has no filter involved, and has a moderate amount of carbon percentage. This means that the me1 has a lot of additional oxygen compared to carbon, also meaning there is a high probability of oxides present. The b1 sample also shows a large number of points that has additional oxygen beyond the filter ratio, and by definition have a probability of containing oxides. Samples f1 and f2 has a more scattered appearance with about half of the points over and half of the points under the ratio of the filter. the conclusion drawn from this is that there are a lower ratio of films in this sample. Another point of view is that the points just above the line could have thin oxides present, but it is not part of the major content in these points. A few points are far from the line in Figure 4.5 and are considered to be carbides, or of at least non oxide nature.

In some of the EDS spectrums, chloride can be seen in various amounts. Using the same method as above and making a fraction of the theoretical composition of aluminium trichloride, the plotting of the points allows for interpretation of its

(23)

occurrence, and can be seen in Figure 4.6 A few points are beyond 0.5 in ratio but on average it can be seen that the chloride is of marginal amounts, and has little effect on the measuring points.

Figure 4.6: Factor of chloride% divided by aluminium% and the factor of the

aluminium chloride.

Nitrogen is also present in a few points and in substantial amounts. In some cases even a major element. The percentage of points with a high ratio of nitrogen in the compositions are approximately 18, and the origin of this is addressed in the discussions chapter.

4.2 Visual Results

Visual confirmation of the measuring points and how they correlate with the appearance of suspected bifilms, has been another main tool to interpret the results.

Taking the etched metal surface as a initial examination point, it is known from a previous study from 2010 , that the composition is heavily affected by the matrix if the inclusion is small or thin [14]. This could be seen in the compositions taken on the metal surface, and can also be seen in Figure 4.7a. There is almost no other content then aluminium and oxygen. The bifilm has 3 parts oxygen for every 2 parts aluminium. The composition for the shown inclusion has a possibility of this

(24)

ratio when also considering there is a strong effect of aluminium matrix.

Inclusion in Figure 4.7b comes from the film filter and shows a presumably thick and large bifilm which we can see from the values. There is virtually no carbon and no effect from the filter there of.

To show a representation of most found film like features, Figure 4.7c can be seen below. A seemingly thin inclusion on the filter with an array of multiple minor elements but with carbon and oxygen as major elements. Here the filter seem to have a great impact on the composition. 0.15 % aluminium and 11.0% oxygen was registered at the measuring point. The presumed film is a very thin sheet of something that resembles a multiple oxide layer. It is so thin and to some degree transparent, that the the film filter and its 0.4 um holes are just barely visible through it.

(a) Bifilm on the etched metal surface. (b) Bifilm on the film filter.

(c) Film like inclusion on the film filter.

Figure 4.7: Bifilms and Film like features on the etched metal surface, a), and the

(25)

4.3 General Inclusions

Multiple types of inclusions have been found. Compositions show possible intermetallic inclusions (IMI), non metallic inclusions (NMI), spinel, and bifilms. Since other inclusions then film like features has not been one of the aims and purposes of this thesis, the examination and analysis are not to a greater extent. Some has been found and photographed in SEM with points analysed with EDS. Two interesting ones are seen in Figure 4.8. Taken on sample f1, measuring point 6 and 5. Point 6 has carbon, aluminium and iron as major elements and a minor

Figure 4.8: Intermetallic inclusions at point 5 and 6.

part of oxygen. Using the same type of diagram and plotting the ratios of different combinations, there is no definite result that can be summarized more than that it is some kind of IMI. Point 5 has similar composition to point 6, but has what it appears, dissimilarities in the crystal growth. This is also some kind of IMI.

4.4 Size ranges

The film like inclusions were measured with the polygon marking tool in ImageJ to get measurements like area, circularity and Feret’s diameter. Circularity is the factor of how circular objects are, where a line is 0 and a circle is 1. Feret’s diameter is the longest straight line that can be accommodated inside an area or volume. In Figure 4.9 we see the results of those parameters. We see that the area is widely

(26)

distributed and that a number of film like inclusions has an area ranging from 100-40000 µm2_{in Figure 4.9a. As can be seen in Figure 4.9c, Feret’s is distributed to}

the lower ranges presented. Circularity shows that the films are not very circular, the distribution is pre-eminent to the 0.2-0.5 µm span. This can be seen in Figure 4.9b.

(a) Distribution of area. (b) Distribution of circularity.

(c) Distribution of Feret’s diameter.

Figure 4.9: Area, Circularity and Feret’s diameter.

In Table 4.1 it can be seen minimum, average, maximum and standard deviation on each and every measured parameter.

(27)

(28)

5 Discussion

5.1 The Findings on the Etched Metal Surface

On the etched, after EE, metal surface, it is most definitely bifilms because of multiple points of close ratios to aluminium oxide, and film-like visual confirmation. The effect of the metal matrix does have a large influence on the points measured compositions. But since no excess oxygen is present, as the case with the film filter, most certainly these are actual bifilms. Another fact is that the amount of points measured on the metal sample is few, but several of them show the same results. A greater number of points would give a higher grade of certainty.

5.2 The Findings on the Film Filter

It is impossible to know if the size or the amount of film like features found on the film filter is representative of an production piece of aluminium alloy, since no comparisons were made within the extent of this thesis. The film filter findings does not have many points with a composition that match up with the ratio of aluminium oxide. Disturbance of oxygen and carbon from the film filter has a high impact on the compositions measured. Despite these facts, with visual confirmation and the ratios of compositions combined, there is a high probability that most of them are bifilms. Several points on the film like features examined still do have an amount of aluminium that can not be ignored. As can be seen in Figure 4.7c there is 0.15% of aluminium and excess oxygen beyond the film filter ratio. This leaves a small fraction of the measured composition to be an actual aluminium bifilm.

The size of the films found on the filter is another concern with this method. With the methods used it is impossible of knowing with the small amount of quantitative data, if the film like features are reduced in size by being broken off during extraction. However, it is possible when large films are being extracted, but are still halfway stuck in the metal matrix, that they break in half or smaller fractions. This would happen by bubbles rising to the surface and breaking the films off, or by turbulence in the electrolyte due to these bubbles or difference in

(29)

temperature/density in the fluid.

5.3 Accuracy of SEM-EDS on bifilms

One problem that emerged, while trying to interpret the data from SEM, was the composition of the filter matrix having a substantial part of oxygen in it. The high carbon in the data, when looking on the filters, were a major element that could not be removed without affecting remaining elements with a large error. Thus affecting the the interpretation of inclusions. In 2010 a report on inclusions in tool steel, wrote about the major effect that the metal matrix or film filter has on smaller inclusions. This showed that compositions of small inclusions, with the diameter of smaller then 6 µm in diameter, had a great misreading from the matrix or filter behind it when using EDS method. [14] This is also probably the case with the compositions acquired during examinations of metal and film filter surface.

5.4 ImageJ

ImageJ is as mentioned earlier a software for image processing. It is an easy-to-use tool for obtaining necessary data from images but comes with some disadvantages. In these studies the images from SEM uses gray scale, and the films appearance blends into the film filter due to low contrasts. This created difficulties of using plugins or tools that measure differences in contrast. A polygon marking tool was used instead. This tool was used to highlight shapes in the images by hand. Points are placed around the film and then ImageJ calculates the desired data. ImageJ then converts pixels into the desired unit after the user sets the scale of the image. Even though the scale is accurate, it cannot be obtained precisely with the human factor. The accuracy of the result is dependent on the user, by how densely the marking points are placed, and how visual assumptions are made. This effects placement of the edges. Another factor is that the measurements are done on how the inclusions appear on the filter, and does not account for any creases or folds that may have happened during extraction or filtering. This is a factor that is hard to control and know the extent of. By these reasons, near exact results are impossible using this method.

(30)

5.5 Precipitation of Chlorides

If there is aluminum chlorides present, it does not come from the melting process from when the samples were produced, due to the boiling point(sublimes) of AlCl3

being approximately 453K [15], and the pure aluminium melt would have had over 933K [16]. With that said, precipitation of chloride compounds will have appeared during EE. There is somewhat conclusive results with 1 week old electrolyte, and 2 months old. The newer electrolyte did not produce the same amount of chlorides as the older. It is still a unknown origin of this result, but it might also be related to the lesser charge per minute that sample W1-1 and B1 received, and as a result the time span of the extraction. Though these are only theories and does not explain it.

(31)

6 Conclusions

Bifilms has been found on the etched metal surface, and it seems this method is superior to the method of electrolytic extraction that was used, when it comes to the task of confirming bifilms. But the information that can be acquired from this method is limited to partial area of the inclusion, and composition.

It is probable that the film-like inclusions on the filter are actual bifilms. But with a factor of uncertainty with light elements in SEM-EDS, we cannot confirm it. Concerning the amount of information that can be acquired through the the EE-method, area, size, circularity and Feret’s diameter gives essential information for determining the impact bifilms might have on the mechanical properties of aluminium alloys.

Is EE a viable method to examine bifilms? Not as of right now, but it has an advantage over some other standardized methods such as the cross section-method. With the ability for area measurements and detection of general inclusions, including plane inclusions. With optimisation of possibly both electrolyte and filter matrix the method will be a potent tool when examining bifilms and other inclusions in aluminium.

(32)

7 Future work

One part is optimization of the electrolytic extraction, such as other filter materials. For example polytetrafluoroethylene (PTFE), which is a synthetic fluoropolymer. Having fluoride in the filter instead of oxygen would give a great change in the compositions, looking for oxides. Possibly chloride and hydrogen-based polymer filters would also give an advantage when measuring compositions. Other types of filters that does not contain oxygen would also be a possible solution. It is also of interest to further examine how freshly mixed and a couple of months old electrolyte affect the precipitation of chloride on the bifilms. Both on the etched metal surface and the film filter there might have been results eventuated of these factors, so furthers investigations is advised.

Also, the accuracy of compositions could be improved with a more accurate technique. One suggested is wavelength-dispersive spectroscopy (WDS). Having more accurate method of determining compositions, especially considering light elements, would give a clear picture of intrinsic elements and compounds.

Quantitative analysis of size distribution and other size depending parameters on bifilms, is also recommended using electrolytic extraction. Having quantitative data of bifilms would yield further understanding and accuracy of their effect on mechanical properties and other material properties.

(33)

8 Acknowledgements

We would especially like to thank our two supervisors Andrey Karasev and Hamid Doostmohammadi at Department of Material Science and Engineering at the Royal Institute of Technology. A. Karasev whom helped us with electrolytic extractions, examinations with the scanning electron microscope, interpretations of the results of compositions, and discussions of results and theories with his extensive knowledge of the area of extraction methods. H. Doostmohammadi whom has given us feedback on calculations and interpretations of acquired results and has assisted with his great experience within the area of bifilms. Due to the prevailing situation in the world, this work has to a large extent been done from home. It has been a difficult situation for all of us and we are incredibly grateful for all the video meetings, and the email conversation we have had. We appreciate everything our supervisors has done for us. It would not have been possible without them.

(34)

References

[1] Du, Hongying. The evaluation of non-metallic inclusions in

calcium-treated steel by using electrolytic extraction. 2016.

[2] Campbell, John. “The consolidation of metals: the origin of bifilms”. In:

Journal of materials science 51.1 (2016), pp. 96–106.

[3] Bagherpour-Torghabeh,

Hamed, Raiszadeh, Ramin, and Doostmohammadi, Hamid. “Removal of Bifilms from Al Melts by Stirring in Unbaffled and Baffled Crucibles”. In:

Metallurgical and Materials Transactions B 49.6 (2018), pp. 3456–3469.

[4] Dong, Hanshan. Surface engineering of light alloys: Aluminium,

magnesium and titanium alloys. 2010.

[5] Campbell, John. Complete casting handbook: metal casting processes,

metallurgy, techniques and design. Butterworth-Heinemann, 2015.

[6] Maciejewski, Joseph. “The effects of sulfide inclusions on mechanical properties and failures of steel components”. In: Journal of Failure

Analysis and Prevention 15.2 (2015), pp. 169–178.

[7] Bergfors, S., Karasev, A., and Flink, D. Laboration one, sample W1-1. Feb. 2020.

[8] Kanbe, Yuichi et al. “Application of extreme value analysis for two-and three-dimensional determinations of the largest inclusion in metal samples”. In: ISIJ international 51.4 (2011), pp. 593–602.

[9] Janis, Diana et al. “Application of different extraction methods for investigation of nonmetallic inclusions and clusters in steels and alloys”. In: Advances in Materials Science and Engineering 2014 (2014), pp. 1–7. [10] Kazmiruk, Viacheslav. Scanning electron microscopy. InTech, 2012,

pp. 17–22. ISBN: 978-953-51-0092-8.

[11] Ahmadpour, A, Raiszadeh, R, and Doostmohammadi, Hamid. “Effect of stirring on behaviour of double oxide film defects in A356 aluminium melt”. In: International Journal of Cast Metals Research 27.4 (2014), pp. 221– 229.

(35)

[12] Bergfors, S., Doostmohammadi, H., and Flink, D. First meeting with

supervisors. Feb. 2020.

[13] Hussain, Sajjad. Application of Electrolytic Extraction 3-DTechnique for

Analysis of S... Jan. 2015. URL: https : / / www . slideshare . net /

SajjadHussain38 / application of electrolytic extraction -3dtechnique-for-analysis-of-stainless-steel.

[14] Doostmohammadi, H., Karasev, A., and Jönsson, P. G. “A Comparison of a Two-Dimensional and a Three-Dimensional Method for Inclusion Determinations in Tool Steel”. eng. In: steel research international 81.5 (2010), pp. 399–402. ISSN: 1611-3683.

[15] Aluminum chloride. URL: https : / / pubchem . ncbi . nlm . nih . gov /

compound/24012.

[16] Sundman, Bo et al. “Binary Element Systems”. In: Phase Diagrams. Department of Material Science and Engineering, KTH, 2006.

(36)

Electrolytic Extraction of Aluminium Bifilms