Stress-free titanium-based thin films for inner ear microphones : The last missing part of a technology for totally implantable hearing aid implants

(1)

Linköpings universitet SE–581 83 Linköping

Linköping University | The Department of Physics, Chemistry and Biology

Master’s thesis, 30 ECTS | Applied Physics

2020 | LIU-IFM/LITH-EX-A--20/3821--SE

Stress-Free Titanium-based Thin

Films for Inner Ear Microphones

–

The last missing part of a technology for totally implantable

hearing aid implants

Spänningsfri och tunn titanﬁlm till en hörapparat för innerörat

Dina Ehsan

Supervisor : Naureen Ghafoor Examiner : Jens Birch

(2)

Upphovsrätt

Detta dokument hålls tillgängligt på Internet - eller dess framtida ersättare - under 25 år från publicer-ingsdatum under förutsättning att inga extraordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning. Över-föring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säkerheten och tillgängligheten ﬁnns lösningar av teknisk och administrativ art.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för upphovsman-nens litterära eller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/.

Copyright

The publishers will keep this document online on the Internet - or its possible replacement - for a period of 25 years starting from the date of publication barring exceptional circumstances.

The online availability of the document implies permanent permission for anyone to read, to down-load, or to print out single copies for his/hers own use and to use it unchanged for non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional upon the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement.

For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/.

(3)

Abstract

An implantable hearing aid device is being developed by a project group which is part of an EU initiative. This device contains a diaphragm consisting of a submicron thick freestanding titanium film, which should be free of internal stresses. Stress is the force exerted per unit cross-sectional area of the film and it can impair the functionality and performance of the device. The stress that evolves in a thin film during deposition at a substrate is compressive or/and tensile and affects the bending that occurs of the substrate due to the lateral force applied to the substrate by the stressed film.

The goal of this diploma work was to contribute to the understanding of in situ stress evolu-tion in a micron thick titanium film and thereby by tuning different physical parameters to obtain minimal residual stress in the films after growth. Titanium films were deposited on silicon sistrates using DC magnetron sputtering. The stress in the material varied, by tuning different physical parameters such as working pressure, power, distance between magnetron and the sample and substrate bias. For this thesis, firstly two different series were done; one where the changing parameter was the distance between the sample and the magnetron and one where it was the working pressure. Later a last series were done to see what effect the bias has on the stress. A multi-beam optical sensor system (MOS) was used to measure the stress in real-time during deposition. X-ray diffraction (XRD) was later used to make post-deposition stress measurements to verify the stress obtained from the MOS. However, the MOS shows the stress evolution in real time and XRD shows a ’final’ average value that can be compared with the stress obtained from MOS-data when the deposition is finished. The results showed that the stress goes from compressive to tensile as the working pressure and the distance between the magnetron and the sample increases. There are other factors, such as the temperature/heating in the main chamber, base pressure of the main chamber, cleaning of the sample and also where the argon gas is let in to the process chamber (in this project called the main chamber (MC)), that influences the results. This will in turn influence the repeatability of the data/measurements, since these effects can affect the process of nucle-ation and coalescence. The stress evolution can change if a bias is applied during the initial stage of the deposition process when the film has still not grown thick. This is could be due to the bias not having much of an impact on the stress evolution when the film is thicker and thereby more porous.

(4)

Populärvetenskaplig sammanfattning

En implanterbar hörapparat skall utvecklas av en projektgrupp, som är en del av ett EU-projekt. Apparaten innehåller ett membran som består av en fristående titanfilm som är cirka en mikrometer tunn och denna film ska vara fri från inre spänningar. Spänning är kraften per enhet tvärsnittsarea av den växande filmen och det kan påverka materialets funktionalitet och prestanda. Spänningen som bildas i titanfilmen kan antingen vara kompressiv spänning eller dragspänning. Detta kan leda till en böjning av substratet som filmen växer på som följd av de laterala krafter som appliceras av den växande filmen.

Målet med detta examensarbete var att bidra med förståelse av hur spänning kan bildas i tunna filmer och sedan genom att ändra på fysikaliska parametrar skapa en spänningsfri titanfilm på ett kiselsubstrat genom att använda DC magnetron sputtring. De fysikaliska parametrarna som kan påverka spänning som då bildas i ett material är argontryck i kam-maren, magnetronens effekt, avstånd mellan prov och magnetron samt elektrisk spänning som appliceras på substratet. Det var huvudsakligen två varierande fysikaliska parametrar som undersöktes i detta arbete; argontrycket och avståndet mellan magnetron och provet. Det gjordes två olika serier där påverkan av dessa parametrar utreddes och sedan gjordes ytterligare en serie för att undersöka hur elektrisk spänning påverkar spänningen. En ”multi-beam optical sensor” (MOS) användes för att mäta hur filmens spänning utvecklas under filmtillväxt. Sedan användes röntgendiffraktion (X-ray diffraction (XRD)) för att verifiera de slutgiltiga spänningsvärdena vid filmtillväxtens slut från MOS-data.

Resultatet visar att spänningen går från kompressiv till dragspänning när både argontrycket och avståndet mellan magnetronen och provet ökar. Det finns även andra faktorer som tem-peraturändringar och bastrycket i kammaren, rengöringsprocessen av substraten samt vart argongasen släpps in i sputtringskammaren, som kan ha en påverkan på resultatet. Detta gör det väldigt svårt att få repeterbarhet i data och mätningar, eftersom dessa faktorer kan ha olika påverkan på nukleations- och sammanväxningsprocessen av atomerna när spänning uppstår. Det är dock möjligt att minimera spänning i filmtillväxten genom att applicera en elektrisk spänning på substratet, dock skall detta göras i början av filmtillväxtsprocessen. Detta är på grund av att tjockleken på titanfilmen kan påverka den elektriska spänningens inverkan.

(5)

Acknowledgments

Foremost, I would like to express my sincere gratitude to my supervisor Naureen Ghafoor for giving me the opportunity to be part of this project, helping me throughout this work and giving valuable guidance. Her motivation has inspired me and she has taught me how to do research in a field that before this diploma work seemed too complicated, but now has made me more interested in research. Thank you for answering my never ending questions. I would also like to thank Per Sandström and Samiran Bairagi for their endless help and support every time I encountered problems with the systems I used. I will be forever grateful that, even though you were not part of this project and had hundred of other things to do, you were always around to help. This diploma work would nearly not have been possible to do without your help that encouraged me at times where things seemed impossible.

I would like to thank my examiner Jens Birch for his insightful comments and positive energy, Sjoerd Broekhuisen for his help regarding XRR-measurements and Babak Bakhit for teaching me how to carry out stress measurements with XRD. I also thank the EUREKA project group; SwissNeutronics, the University of Zurich and Cochlear Technology Centre in Belgium for a great collaboration. Your contribution has been valuable for me to carry out my diploma work.

Last but absolutely not the least, my forever and never ending gratitude to my parents and brother who have been my pillar of strength and source of energy. Thank you for your love, prayers, caring and sacrifices. Thank you for preparing me for my future and supporting me while I chase my dreams. Thank you for being my hope.

I would also like to dedicate a special thank you to my friends for being my cheerleaders and giving me mental support throughout the years. Your encouragement has helped me countless of times.

(6)

Chapter 1 Introduction

There is a large application range for the use of thin films and it is mostly used to modify or enhance the properties of a surface. The structure of thin films; morphology, grain size and boundaries, has an effect on the mechanical, optical and electrical properties of the material. This, in turn, will affect the way the material performs and for which applications it can be used for. Therefore, it is crucial to understand which physical parameters and processes can have an impact on the function and synthesis processes of the thin films [1].

By changing the physical parameters for the deposition process, it is possible to control properties of the thin film [2]. There are different types of deposition techniques that could be used to create thin films, but the outcome can be very much alike. Working pressure, sub-strate bias voltage, source power or temperature are a few physical parameters controlling the evolution of microstructure in thin films [3, 4] and it has mostly to do with the energy of the deposited atoms [5]. Deposited atoms will undergo certain stages when they meet each other, including nucleation, growth, and coalescence, where grain boundaries are formed [6, 2].

During and post-deposition, an internal stress can exist in the thin films. The origin of stress in thin films is associated with the evolution of structure and how energetic the deposited atoms are [5] when they arrive to the surface. Stress can cause the material to fail by either cracking or decohesion [7] which affects its functionality and performance. The stress can be either tensile or compressive [8].

Titanium is a material that is widely used in microelectromechanical systems (MEMS) be-cause of its physical properties. It has good conductivity, is thermally stable, has high hardness and do not have many crystallographic imperfections [4]. In MEMS applications Ti can be used for cochlear implants [9]. Thin films that are used for MEMS must be stress free or have minimal stress since stress can lead to the device to malfunction [4]. By doing in situ stress measurements during deposition and varying physical parameters affecting the deposition process, it is possible to control and minimise residual stress in thin films. Multi beam optical sensor (MOS) systems can be used to measure the stress during deposition and the plotted data will give information about the stress evolution in the material.

In this thesis thin films of titanium will be deposited onto a thin silicone substrate using DC magnetron sputtering. By varying different physical parameters, it is possible to see how the stress in the material will evolve. The goal is to create a stress-free titanium based thin film for an implantable cochlear implant. This thesis contributes to research for a product that can lead to good health and well being of people and thereby also contributes to sustainability.

(9)

1.1 Aim of the Thesis

The aim of this diploma work is to

• Contribute to the understanding of how in situ stress is evolved in a micron thick film and how different parameters controlling the deposition can affect the evolution of stress.

• Obtain minimal residual stress in a titanium based thin film by varying the physical pa-rameters working pressure and the position of the magnetron when doing deposition. • Combine different physical parameters and relate it to the theory in order to do the

measurements and learn the process of doing research.

This will be done by measuring stress in the films during growth and varying the physical parameters to receive a stress free titanium film that is 1 µm thick. In situ real-time stress measurements will be done followed by post-deposition stress measurements and the data will be analysed to contribute to the understanding of stress development

1.2 Delimitations

In order to simplify the project, the changing parameters will be working pressure and the position of the magnetron from the sample. If there will be time left until the diploma work comes to an end, also the effect of substrate bias will be investigated. For this thesis, the focus is only on the deposition of titanium and not other materials. The post-deposition stress measurements will be done on some of the samples and this is only to verify the values of stress obtained from the real-time measurements.

1.2.1 The Use of Related Work

Chapter 4 will contain related work done for this project before this diploma work was started. Images and data will be provided in order to obtain a better understanding of this project, but also why those changing parameters were chosen. However, in that chapter only parts of that related work highly connected to the measurements and results in this diploma work will be presented.

1.3 Outline

The thesis will as an introduction to the work contain the theory needed to understand the results and the related work, which is Chapter 2. The theory will mainly focus on the equip-ment used to conduct the measureequip-ments, the mathematical background of stress and how that is related to the equipment used and theoretical background of stress which will focus on what types of stress there are and how they evolve. Chapter 3 describes the method used to conduct the experiments/measurements, but also the physical parameters varied to see the effect on the stress. In Chapter 4, the related work will be presented. Chapter 5 is where the result from the experiments are shown and described. In the following two chapters the discussion and the conclusion made from this project will be presented. The first appendix gives data from the XRR-measurements and the second appendix includes some additional results to those who are presented in chapter 5.

(10)

Chapter 2 Theory

In the sections below, the theory necessary for the understanding of this project is described. Some project information will be presented at first, followed by the analysis and the math-ematical background of stress. Lastly some information regarding the techniques used to conduct this project is described briefly in order to understand the methods described in the next chapter.

2.1 Project Background

The University of Zurich and Cochlear Technology Centre in Belgium is developing an im-plantable hearing aid implant. One part of this device consists of a submicron thick titanium diaphragms which has a 0.5 mm diameter. The purpose of the diaphragms is to act as a sound receptor and pick up sound energy from fluid of the inner ear and transfer it to gas, from which a microphone can transform the energy to an electrical signal. Hence, the di-aphragms works as a sensor which has an impact on how well the microphone is performing. In order to integrate the diaphragms in the hearing aid implant, titanium is sputtered onto a polymer that is decomposable. The reason why titanium is used is because it is a biocompat-ible material. The diaphragms should be connected to the structure in a way that minimal residual stress, hermeticity and robustness is obtained.

The main focus of this project is to grow thin films of titanium onto silicon, which is the substrate, and during growth measure the stress. Later, parameters controlling the stress should be tuned to obtain a stress-free titanium film and a membrane that is a micron thick.

2.2 DC Magnetron Sputtering

DC magnetron sputtering is a PVD technique that is used for the deposition of thin films onto a substrate. This deposition method provide properties such as high hardness, resistance to corrosion, lower friction or other desired electrical/optical properties, which makes it often a good option for deposition of thin films. It can also provide deposition of metastable phases since materials can be created outside the thermodynamic equilibrium range [10].

The sputtering process starts by inserting the substrate into the main chamber and then evac-uating the air with a vacuum pump in order to decrease the pressure. The chamber will then be filled with a desired gas which will be ionized and create a plasma and later the ions will be accelerated towards the target. The target consists of the material that is going to be de-posited onto the substrate. When the gas-ions hits the target through collision, atoms will eject and deposit onto the substrate. The working gas used for this project is argon which produces Ar+, which is a positively charged ion. The reason why Ar+ ions are accelerated

(11)

towards the target is because a negative voltage is applied on the target. This principle is described by figure 2.1. Note that Ar+ is an inert gas, hence it will not interfere with the deposited material

Figure 2.1: This schematic figure shows the principle of DC magnetron sputtering. Sputtering oper-ates under vacuum conditions where target particles is sputtered onto the substrate when hit by ion gas.

The ultra high vacuum (UHV) magnetron sputtering machine used in this lab is from Mantis. Figure 2.2 shows a schematic of the system.

Figure 2.2: A schematic of the DC magnetron sputtering machine used for this project. The main parts of the system are named in the figure.

It was possible to change the position of the substrate holder and the magnetron (Z-shift), the tilt of the magnetron and the substrate holder and the rotation of the the substrate. The settings used for the depositions are explained in chapter 3.

The procedure of sample loading and unloading will be described in the following chapter. However, it is important to know some theory behind some of the steps. When unloading a

(12)

sample, the load-lock (LL) chamber needs to be vented, which is to relieve the low pressure in order to open the LL-door. After a sample is loaded in the LL, the LL needs to be pumped, which is to decrease the pressure to obtain vacuum. This is important since there is a valve between the main chamber (MC) and the LL. This valve needs to be opened when the sample is being transferred into the MC with the transfer-arm (which holds the sample holder with a clamp). In this process, the pressure in the MC will maintain constant since the vacuum inside the MC is high (low pressure). The optimal condition for when to open the valve is when the pressure in the LL is maximum one order of magnitude larger than than the pressure in the MC. The sample needs to be cleaned before being put into the system and the procedure for this is also explained in the next chapter.

As mentioned earlier, the target used for this project is titanium (75 mm in diameter and 3 mm thick), which will be deposited onto a thin silicon substrate. Titanium has a good corrosion resistance, which makes it an excellent metallic biomaterial to use for implants. How corrosive an implant is has an impact on how it performs and its durability, which in turn affects the biocompatibility of the medical device [11].

2.3 Stress Measurements with Multi-Beam Optical Sensor (MOS)

In order to obtain desired properties such as optical, electronic or mechanical, it is crucial to understand and control the stress in thin films during deposition. By controlling the stress during the deposition process, one is able to avoid the unwanted properties that could lead to the device performing poorly [7].

Stress in thin films can be measured by observing the change in the substrate curvature by using a laser system during deposition. The stress in the film induces a curvature in the substrate. In this project, a multi-beam optical sensor (MOS) from the brand k-Space Asso-ciates Inc. was used. With this device it was possible to real-time monitor the thin film stress by sending in an array of parallel laser beams through a collimator and monitor how the substrate curvature changes with a CCD camera [7]. This method of measuring stress is non-destructive [5]. Figure 2.3 shows how the laser system works and how 2 dimensional spots are created, which will help in the curvature measurements.

Figure 2.3: This figure shows how the detection of the laser works and how the 2D array spots are generated from the laser [12].

The optics that are used to control the laser are in a laser optics housing, with one lens and one mirror. As can be seen from the figure above, the lens is used to generate multiple paral-lel laser beams from one single laser. These paralparal-lel beams shines onto the substrate during

(13)

deposition as an array of spots which in turn is reflected onto the mirror that is positioned in such way that the reflected laser beams from the substrate is caught by the CCD camera. This creates a 2D array of spots (see figure 2.3) with a spacing∆d between the spots. During deposition, ∆d changes in two orthogonal directions which generates 2D curvature data. Hence this is used to determine curvature changes [12]. In the case of growing a film on the substrate, the curvature change is caused by internal stress in the film attached to the substrate. How the data is plotted and the mathematics behind this principle can be read in subsection 2.4.2.

The MOS-system is supplied with a laser diode and a temperature controller. The laser in-tensity can be controlled via the electronics connected to the laser optics housing box and via the software. The temperature controller controls the temperature of the laser and operates automatically whenever the laser power is on.

2.4 Stress - Analysis and Mathematical Background

Stress (σ) is the force per unit cross-sectional area ([N/m2]) of the film [8]. Stress in thin films can lead to the film not functioning as it should. It has an effect on the performance, reliability, durability [5] and the physical properties, such as sound velocity, band structure or the thermal properties, [1] of the device itself.

How the stress evolves in a film depends on the microstructural and the surface conditions of the film. This is strongly tied to the deposition process when the film grows. Stress could be extrinsic, which means that the stress appears when an external force is applied during film growth or if there is a thermal mismatch between the film and the substrate. However, in this thesis the main focus is the intrinsic stress which happens when the thin film is growing but is constrained compared to its in-plane dimension since it is bonded to a substrate. Structures will evolve in this process which could be grain growth, due to impurity, morphological evolution, formation of point defects etc [13].

2.4.1 Tensile and Compressive Stress

As mentioned earlier, the stress in a thin film can be determined by measuring the force applied on a certain area. A film is said to be undergoing a tensile stress when the surface of the substrate is contracting in a way that the radius of the curvature R is positive, R>0. However, when the stress induces a contraction of the film in such way that the R is negative, R<0, the film is under a compressive stress. This means that the curvature of the substrate can determine what type of stress is induced [8]. A schematic figure of this phenomena is shown below.

Figure 2.4: The difference between compressive and tensile stress, in both bending and how the radius of curvature differs.

(14)

When a material undergoes compressive stress, the size of it will reduce (compression). How-ever for tensile stress, the material will elongate. This can be seen in figure 2.4, where the ar-rows point to the film for the compressive stress and the arar-rows pointing away for the tensile stress. Hence, compressive stress can make a material peel off, buckle or blister and tensile stress can lead to the material to crack [5].

2.4.1.1 Origin of Tensile and Compressive Stress

When a film is growing on a substrate, it undergoes different stages in internal stress. It means that the stress evolution in thin films are not only compressive or tensile, it is a mix between both.

During the first step of film growth, target atoms will arrive to what is then a flat substrate with no stress. These atoms will diffuse on the surface of the substrate and encounter with other atoms [2] and start to form clusters. These clusters are called nucleation islands and are the origin of compressive stress due to a surface tension being created [6].

When these islands grow, coalescence will occur. Islands large enough will interact and grow together, which means that a grain boundary is created [6]. When the boundaries grow closer, the free energy of the thin film will decrease and this will yield a tensile stress [2]. This process is explained in figure 2.5 below.

Figure 2.5: In step 1, the nucleation islands are created by atoms encountering with each other which creates a compressive stress. The islands will grow large and closer to each other and form grain boundaries, which is seen in step 2. This creates a tensile stress. Compare to figure 2.4.

2.4.1.2 Parameters Affecting the Stress

There are some kinetic processes that happen during deposition that can have an impact on the stress. Deposition methods such as evaporation or electrodeposition are called non-energetic processes where the atoms being deposited have a lower kinetic energy. In this project, DC magnetron sputtering is used as a deposition method and that is an energetic deposition process. The atoms being deposited onto the substrate have high thermal energy compared to non-energetic processes [5].

The working pressure has an impact on the stress. Thin films tend to go from tensile to compressive stress when the argon pressure decreases. At low pressures of argon (typically around 1 mTorr), compressive stress is shown in the thin film. Tensile stress occurs when the pressure is increased. When the pressure is increased, there will be more collision between energetic sputtered atoms and the gas ions, hence they will lose more of their kinetic energy. For lower pressures, deposited atoms will not collide with as many gas ions, which means minimal kinetic energy will be lost during the path to the substrate surface. When arriv-ing to the surface, with more energy (than the case for higher pressures), they can induce displacement of atoms on the substrate surface. The displaced atom can relocate to a more

(15)

favourable site, but it can also create point defects. When point defects occur due to the involvement of energetic particles, it will lead to compressive stress developing which is a result of atomic peening [5]. In deposition processes, the atomic peening effect describes how energetic particles bombards the growing film and the effects it might have. This is related to the working pressure in the chamber as indicated [14].

By applying a bias voltage to the substrate during deposition will increase compressive stress or decrease tensile stress. By applying a voltage, there is an electric field created between the substrate and the anode, which makes the argon ions to accelerate towards the substrate. The bombardment of the gas ions will give higher energy to the target ions when a voltage is applied and as for the case for the pressure, target ions with high energy deposited on the surface can cause displacement of the other atoms and make the surface ’move in’. This will create a higher compressive stress in the thin film than it would for a film deposited onto a substrate without a bias voltage [3].

Stress is also dependent on the magnetron power and substrate temperature [4].

2.4.2 Stress Measurements with MOS

When the film grows on the substrate during deposition, a force is applied on the substrate causing it to bend, as can be seen in figure 2.4. Stoney derived an equation that determined the relationship between the curvature of the substrate [5] and the stress. The equation is called ”The Stoney formula” and is presented below:

1 R = 6 Yh2 s hfσ (2.1)

where hf is the thickness of the film, hs is the thickness of the substrate and Y is the biaxial modulus of the substrate [5], which in this case is silicon and has a value of 180.6 GPa. The product σhf is called the stress-thickness product, where the unit is force/length [N/m]. A film that is stressed applies a force on the substrate it is growing on as well as creating a bending. When a film is much thinner than the substrate, the bending of the substrate does not affect the film [13].

As mentioned in section 2.3, the stress evolution can be measured in real-time during film growth. The parallel laser beams generated from the MOS creates a pattern of laser spots on the film (see figure 2.3) and detects the changes of surface curvature. When the substrate is bending the spacing∆d between the spots will change. The mean differential spacing ∆d/d is the mean value of the difference between the original spots and the change of their distance. The relationship between the curvature 1/R and∆d/d is defined in equation 2.2 below as:

1 R = cosα 2L ¨ ∆d d (2.2)

where α is the angle of incidence (where the laser hits the substrate) and L is the distance between the substrate and the CCD camera [8]. This equation notes the mean vertical differ-ential spot spacing.

By combining equation 2.1 and 2.2, the stress σ of the film in Stoney’s formula can be ex-pressed in terms of the mean differential spacing∆d. Equation 2.3 below shows this relation-ship: σ ¨ hf = ∆d d ¨ Yh2 s¨cosα 12L (2.3)

(16)

The equation for the mean horizontal differential spot spacing is given below: 1 R = 1 2L ¨ cosα¨ ∆d d (2.4)

By combining equation 2.1 and 2.4, the stress σ of the film in Stoney’s formula can be ex-pressed as the following equation:

σ ¨ hf = ∆d d ¨

Yh2 s

12L ¨ cosα (2.5)

It is the CCD-camera that will measure how the laser spots are aligned. The information received will be plotted as the average horizontal or the average vertical mean differential spot spacing as a function of time. It is because the average of all differential spot spacing values are taken into account [8]. In this thesis, only the average vertical mean differential spacing is taken into consideration. However, a difference between how the stress looks like when comparing data from the vertical and the horizontal differential spot spacing will be provided in section 5.

The plotted data is however converted to the stress-thickness (σhf) product, which can be received from equation 2.3 when using the mean differential spot spacing data, as a function of thickness. It can also be plotted only as σ as a function of thickness. How the data is plotted as a function of thickness is described in section 2.5.

2.4.3 Important Aspects of the Stoney Equation

There are some important aspects to take into consideration when measuring the stress in thin films using the Stoney Equation and these are listed below:

1. Equation 2.1 is true if the substrate is flat before deposition. To obtain the stress in the film, a change in the curvature is necessary. When determining the stress of the film after deposition, the curvature must be measured before and after deposition. This is highly important when doing post-depostition stress measurements with XRD, which will be explained in section 2.5. For real-time measurements during deposition with MOS, the pre-deposition curvature is not needed, because a reference value is used [13].

2. Thin substrates should be used when making depositions since the curvature is in-versely proportional to the square of hs. This has to do with the sensitivity of the mea-surements. When producing thick films, it is however recommended to use thick sub-strates. This is mainly to avoid the fact that the substrate can deflect in a way that it may break during deposition. When making a thin film approximation which indicates that hf < < hs, the layers will contribute to the curvature independent of each other and additively. If the film has stress in it itself, it does not affect the layers that are below it. The thin film approximation is needed for Stoney’s formula but it can also become inaccurate. This happens when the ratio between the film and the substrate becomes noticeable [13]. The ratio between the film and the substrate hf/hsshould not be larger than 10%. When the Stoney equation fails, some modifications of the original formula needs to be done [15]. For this project, hs was approximately 155 µm and hf was never larger than 1 µm, hence fulfilling the condition.

3. It is important to know the thickness of the film when measuring the stress with curva-tures. The thickness and the stress will both develops during the deposition. Hence the rate of the deposition is necessary to measure the stress evolution. This is explained in previous subsection [13].

(17)

4. When doing stress measurements after deposition (with XRD), the stress-thickness product σhf should be replaced with integrated stress that measures the stress through-out the material. This is due to the fact that there are stress variations in the mate-rial and this is not determined only by curvature measurements after deposition. For stress-measurements during deposition, the curvature can be affected by multiple fac-tors. These factors could be the material that arrives from vapour, relaxation process in the film (dependent on the depth) and the stress from the surface. The stress of the surface is mostly because of the crystallographic orientation or the morphology [13]. 5. The Stoney equation is also highly dependent on the so called isotropy transverse

ar-gument. This implies that the substrate, related to the film, has transversal isotropic elastic property [15]. In this project, single crystalline Si(100) are used which also have anisotropic properties, hence fulfilling the condition. The thickness and the stress of the film should also be uniform and this is highly dependent on the deposition rate and process [15].

6. The mean stress can be measured, which is the case for this thesis. As mentioned earlier,

σhf is divided by hf [8].

2.5 X-Ray Diffraction (XRD)

Crystalline materials can create 3D diffraction pattern when they are exposed to a beam of incident X-rays. XRD is an analytical method used for the characterisation of crystalline materials. The information received could be about the structure, phase, crystallographic orientation, grain size, defects etc. This method is nondestructive [16].

For this project, XRD is mainly used for two different task; post-deposition stress mea-surements (wafer curvature) as a way to verify the stress obtained from MOS when doing real-time stress measurements, but also for the determination of the deposition rate which is a way to determine the thickness. The latter is because we want the data to be plotted as stress vs thickness.

The principle behind this technique is the use of an X-ray diffractometer that has an X-ray source, a sample stage and a detector. The source is a cathode tube that generates X-rays when a filament is heated where electrons are produced. These electrons accelerates to a target-material when a voltage is applied and a characteristic X-ray spectrum is generated when the electrons with enough energy can knock out the electrons from the inner shells of the material. Monochromatic X-rays are filtered from this spectra and directed towards the sample, as the detector is rotated and records diffracted rays [17]. When the incident X-rays interact with the film, constructive interference that satisfies Bragg’s law which shows the relationship between the wavelength of the X-rays to the angle of diffraction and the spacing between the lattices, see equation below [16].

nλ=2dsinθ (2.6)

where n is an integer, λ is the characteristic X-ray wavelength, θ is the angle of diffraction and d is the spacing between the diffracting crystal planes. Diffracted X-rays are detected by a detector, when the detector scan the sample in a pre-determined range of 2θ-angles. The data plotted will be diffraction peaks at certain angles and then converted into d-spacing which is unique for each compound.

(18)

2.5.1 X-ray Reflectivity (XRR)

A technique that can be used to study/characterise thin film surfaces is X-ray reflectivity. Suppose that there is a homogenous substrate, where on top of it a homogenous film is deposited. The incident X-ray beam will go through two interfaces where the the first one is the air/film and the second one is the film/substrate. The beam is partly reflected at each interface while some of the beam continues through the material. X-rays can be seen as waves and the difference in paths between the x-rays reflected at the first interface and the second interface results in a phase shift, that is dependent on the incident angle [18]. The surface has to be smooth in order for the reflected rays to not deviate from Fresnel’s reflectivity law. If a smooth thin film is present on a smooth substrate, the film thickness can be deduced from the angular spacings of oscillations in the reflected X-ray intensity. XRR is mostly used to measure the thickness of thin films [19].

By measuring the thickness, the rate of the deposition can be calculated by dividing the thickness with the deposition time. This ratio will later be multiplied with the time for each data point received from MOS in order to receive the data as a function of thickness. The thickness obtained from the XRR-measurements will be used in equation 2.3 to convert the vertical mean differential spot spacing data to stress. Thereby for this project, XRR was used to measure the thickness of the thin films.

When we receive data from the XRR, the plotted data will be a footprint followed by fringes. The distance between two fringe-maximas is proportional to the thickness of the film [18]. When analysing the data, a software is used to fit the data to a curve that is simulated. When fitting the plot received from measurements, parameters such as material density, layer thick-nesses and roughness of the interface are used. Since a simulation is used to fit the data, it is not an exact method. It however estimates the thickness roughly [20].

When doing XRR (and XRD), different scans are done which are dependent on certain angles and heights. Optics and scan parameters are important in order for the X-rays to be reflected on the sample and reliable data to be obtained. The parameters taken into consideration when doing XRR for this project are shown in the figure below.

Figure 2.6: This figure explains how the angles are related to the sample and the incident and reflected beam. It also shows the sample height Z.

ωis the angle between the incident beam and the sample surface, 2θ is the angle between the

incident and the reflected beam detected by the detector and Z is the sample height. ω has half the value of 2θ, ω = 2θ/2. The different scans are explained below:

• The 2θ scan is done in order to check the zero position and the intensity of the direct beam. A scan is made to see at which angle highest intensity of the detected x-rays are the highest, the position of the detector will be set to that angle.

(19)

• The Z-scan is the second scan done and this is in order to set the sample height, see figure 2.6. If the sample is not at the correct height, suppose that it is too far down, there is a risk that the incident X-rays will miss the sample and thereby not be detected by the detector, which will lead to no or very little signal will be obtained.

• When mounting the sample or when setting the sample height Z, the sample may tilt and not lay flat on the sample holder. This is why ω scan is done, in order to find the sample offset of the substrate plane with respect to the surface of the sample. 2θ will be set to a fixed value and the X-ray source will irradiate the sample in a ω-range.

• From the 2θ-ω scan, the data for which the thickness is determined from is obtained, where diffraction from planes parallel to the sample surface is collected. For this mea-surement, both 2θ and ω are varied, with ω being half the value of 2θ. The sample offsets will be set to zero for this scan [18].

In order to meet the analytical requirements, there are some optics that should be chosen when doing an XRR-measurement. When the X-ray beam hits the sample, the beam may be too wide, which means it can hit other areas that are not part of the sample. What can be done is to use some sort of collimator in order to narrow the beam and for XRD-measurements in general, slits can be used. The dimension of the chosen slits should be similar to the dimensions of the sample. If the slit is a divergence slit, it will be put in between the incident beam and the sample to match with the dimensions of the diffraction and the sample. If it is a receiving slit, it is put between the sample and the detector to improve resolution and eliminates scattered rays. A mask can also be used to correct the beam size so it matches the dimensions of the sample. The slits and the mask will provide maximum illumination on the sample. Filters can also be used to selectively reduce intensity of beams with certain wavelengths. For XRR-measurements the beam is lined-focused to illuminate more of the sample [21].

For the 2θ, ω, Z and 2θ-ω scans, some parameters such as scan range (°), step size and time per step (s) need to be determined. This will set a time for each measurement. The scan parameters together with the optics used to make the XRR-measurements are described in section 3.

2.5.2 Wafer Curvature Measurements for Film Stress

XRD can be used to measure the curvature of a substrate wafer post-deposition. The curva-ture of the substrate is 1/R in Stoney’s equation 2.1 and therefore this method is a way to determine stress in thin films. When a film that is growing on a substrate is strained, the substrate will bend with a radius of curvature R. The substrate wafer curvature can be mea-sured from ω-shifts of a substrate Bragg-peak when shifting the measurement position X on the sample, see figure 2.7. This can be done by doing high resolution rocking curve measure-ments (plot X-ray intensity as a function of ω) with XRD. The information that is obtained from the rocking curves are the direction in which the substrate lattice planes are oriented. The figure below shows a schematic of the wafer curvature measurement. Note that∆ω =

(20)

Figure 2.7: This figure is an illustration of a strained film attached to a substrate (the curve), with a radius R. ω is shifted at different positions X on the sample. It also shows a simplified schematic of the data received, which is the intensity as a function of ω that is shifted for each position X on the sample.

Figure 2.7 also shows a simplified schematic of how the plot obtained from the measurement looks like, where ω is shifted with the measurement position on the sample X. X will ap-proximately be the arc length of the curvature if assuming R > > X. Then X =∆ω*R, which means that if plotting ω as a function of X, 1/R is the slope of the curve. The stress of the thin film can thereby be calculated by using Stoney’s equation, see equation 2.1. For more ”exact” measurements, 1/R should be determined for the substrate both before and after deposition. Their difference will be equal to the right side of equation 2.1.

2.5.3 Transmission Electron Microscopy (TEM)

In order to analyse the samples, but to also learn how to use analytical techniques in research transmission electron microscopy (TEM) was used to examine materials on nanoscale level, since regular light microscopy is not an option. TEM is a powerful tool to study materials with dimensions smaller than 100 nm. It is possible to receive information regarding the morphology, crystal structure and fine features in the examined material [22]. An electron beam (energy > 100 kV) with a short wavelength is used, which provides images of the specimen with high resolution [23]. A schematic figure of transmission electron microscopy is shown below.

Figure 2.8: The electron beam interacts with atoms in the sample and transmit through it, The trans-mitted electrons are detected by a detector where information regarding morphology and/or crystallo-graphic orientation will be received.

(21)

The sample is put onto the TEM-grid where the position is controlled via a mechanical arm. The microscope consists of metal apertures and electromagnetic lenses (condensers and ob-jectives) to focus the electron beam coming from an electron gun. Electrons are negatively charged and behave like waves and later deflects when electric/magnetic field is applied. Only electrons with certain energy can pass through. The electrons that are transmitted is focused on the sample, which later provides information regarding the sample. When the electrons pass through the specimen, electromagnetic lenses focus, magnify and project them onto a screen as an image [22]. The image provided is in 2D.

2.5.3.1 Bright respectively Dark Field Mode

By changing the objective aperture to certain focal spots, it is possible to decide if either trans-mitted or diffracted electrons can pass through the aperture. If transtrans-mitted electrons passes through, a so called bright-field image is obtained where the dark parts of the image indi-cates thicker areas and heavy atoms. When the aperture blocks the transmitted electrons, the diffracted ones pass through. If the diffraction conditions are satisfied, information regarding the crystallinity is shown bright in the dark-field mode [22].

2.5.3.2 Sample Preparation

When studying samples with TEM, they must be electron transparent, which means that they must be thin enough for the electrons to transmit through the material. The thickness of the sample should be around 100 nm, which makes it easy for electrons to go through. The sample preparation starts by first cutting the sample with a diamond wheel saw in five to six pieces that are 2x1 mm in dimension. By observing the pieces under the microscope, the best two are chosen. The two pieces are flipped onto their sides with the film sides facing each other and then they are put into a TEM-grid and glued. The grid is glued onto a glass-plate and polished on both its sides with a series of diamond films where the grain size is decreased gradually, until a thickness of 50-60 µm is reaches. The sample is later put into a so called ion milling machine which makes it thinner by shooting ions on it. This is in order to make it electron transparent.

It is important to note that for this thesis work, TEM was only used once to examine a sample made prior to the start of this work. Hence, the results for that will be provided in chapter 4 and not 5.

(22)

Chapter 3 Method

In this chapter, the execution of the procedures done to conduct this project is described. The information and theory behind the machines is explained in the previous chapter, hence this chapter only focuses on the method. The settings for the machines are explained either in the text or in a table. The figures attached are used to simplify the understanding the principles for the reader.

3.1 The Substrate

As mentioned in section 2.2, silicon (Si(100)) substrates was used to make depositions on. The thickness of these substrates vary between 150-155 µm. A new flat substrate is used for every deposition. The thin film deposited onto the substrate is around 620-650 nm for most of samples and this is for the plotted data to be comparable. The thicknesses for the bias series might be larger than 650 nm.

3.2 Sample Cleaning

Surfaces usually contain contaminants that needs to be removed prior to sputtering. In order to remove particles and dust from the sample, there was a cleaning procedure done before the sample was mounted on the sample holder and was put on the rack of the transfer-arm. There was three different solvents used for cleaning; trichloroethylene, acetone and isopropanol.

The first step was to transfer the sample into a beaker and fill it with trichloroethylene to a level higher where the sample surface is covered with solvent. The beaker was put into an ultrasonic bath for three minutes. An ultrasonic bath is a small tank that is filled with water. An ultrasonic energy is applied which will make the water ”vibrate”, hence free the surface from particles [24]. The beaker was removed from the bath after three minutes and the liquid was poured into a jar with waste-trichloroethylene. Same procedure as done with the trichloroethylene was done with first the acetone and then isopropanol respectively, but for five minutes each. The reason why the liquids was used in that order is because acetone removes trichloroethylene and isopropanol removes acetone.

3.3 Deposition and Stress Measurement

All the magnetrons were lowered to their maximum level to prevent any collisions when loading the sample holder. The LL was vented and the clamp that fastens the sample holder was opened to release it. The sample holder was transferred to a fume hood where the sample was kept in a beaker with isopropanol. The sample was blown with N2and was put

(23)

onto the sample holder with its cleanest side facing down. The sample holder was reloaded in the LL and the clamp was fastened. Since the substrates that were used were thin, no clampers were used to fasten it in place on the sample holder in order to avoid the sample breaking and letting the sample to bend freely during deposition. The sample holder was pushed in almost 10 cm into the LL away from the valve opening with the transfer since the substrate has a light weight and has no clamps securing it. When the system was pumped, the substrate did not break or disappear in the chamber during the opening/closing of turbo or roughing pump valves. When the system was pumped and the pressure in the LL had decreased to a level where it was at least one order of magnitude larger than the pressure in the MC, the LL-MC gate valve was opened. The sample was transferred into the MC and the clamp that holds the sample holder was opened. The transfer arm was removed from the MC and the LL-MC gate valve was closed.

When the sample was loaded, the temperature controller and the laser were turned on. The intensity of the laser was set to maximum via the software and the CCD-camera and the optics were adjusted until a symmetric 2D array of spots could be seen, look at figure below.

Figure 3.1: A symmetric 2D array of laser spots prior to deposition.

For lower pressures, the optics were adjusted so the spots were closer to each other and for higher pressures the optics were adjusted so the spots where far away from each other (reason explained in section 6). When the laser setup was done, argon was opened to enter the main chamber. Before turning on the magnetron, 30 seconds of data was collected as a reference for the stress-measurements in the MOS-software. After, the magnetron and the measurement of the spots was turned on at the same time. The figure below shows how it looks like in the MC during deposition.

(24)

In order to understand which parameters/mechanisms control the stress, the stress of the film was measured under different conditions. The positions of the magnetron varied between 15, 30, 50 and 70 mm, where at 15 mm the magnetron was the closest to the sample. For each position, depositions were made for 1.5, 2.5, 3.5 and 4.5 mTorr. This was in order to see which effect the position of the magnetron and the pressures have on the stress evolution, look at table 3.1 below.

Table 3.1: These are the parameters that were varied during the depositions. Look at table A.1 for their rates.

70 mm 50 mm 30 mm 15 mm

1.5 mTorr 1.5 mTorr 1.5 mTorr 1.5 mTorr

Note that 70 mm, 50 mm, 30 mm and 15 mm are not the distances between the magnetron and the sample. These are the ’terms’/numbers used by the software. The distance between the magnetron and the sample when mentioning 70 mm is approximately 20 cm, for 50 mm it is 18 cm, for 30 mm it is 16 cm and for 15 mm it is 14.5 cm. In this thesis, the software numbers will be used.

When making the depositions, the power was kept constant at 300 W for all depositions. The sample stage was for all depositions kept at 142 mm position and pointed towards the transfer arm for successful loading and unloading of samples. No substrate heating or rotation of the sample stage were done. The angle of the magnetron was kept at -2.5° for all measurements. Table A.1 gives information regarding the rates.

First a series of rate depositions were done, where each deposition lasted for 10 minutes (see table A.1) and no stress measurements were done for these. The rate depositions were done in order to calculate the rates of the depositions. The rates were calculated by measuring the thickness of the film (with XRR - explained in section 3.5) and dividing it with the de-position time. For the stress-measurement dede-positions, a thickness of circa 620-650 µm was supposed to be reached. Therefore, rate depositions were done for when the magnetron were at different positions and the desired thickness was divided by the rate. Then, the time for the depositions where stress was measured could be calculated. The rates were also used to convert the plotted data from a function of time to a function of thickness. When using the MOS for the stress measurements, the distance between the sample and the CCD-camera was kept constant at 102 cm and the angle of incidence was 45°.

When the deposition was done, the magnetron was turned off, and the measurement of the mean differential spot spacing went on for another 5-10 minutes, with the argon running, to let it stabilise. When the measurement was done, the argon was turned off and after a few minutes where the pressure had stabilised, the LL-MC gate valve was opened and the sample was removed from the MC. When the gate-valve had closed, the LL was vented. In order to see what effect the a substrate bias has on the stress evolution and if it is possible to minimise the residual stress, a series where different biases were applied during the depo-sition process was made. This series was done for when the setting parameters were 30 mm and 2.5 mTorr (see table 3.1). For the first series, the argon gas was opened in the MC and the biases applied were 100 V and 130 V. For the second series, the argon was opened at the target and the biases applied were 0 V, 100 V, 150 V and 200 V.

(25)

3.4 TEM - Preparation and Analysis

The routine for the sample preparation is explained in section 2.5.3.2. The sample is first polished very roughly with a silicone film of rough grains. The diamond films that are used to polish the sample until it reaches a thickness of 50 µm, have the grain sizes 30, 15, 6, 3 and 1 µm respectively. The polish with the largest grain size was used first and then it was gradually decreased. When the desired thickness was obtained, the ion milling machine was used to make it thinner. When the sample was transparent enough, it was examined with the TEM following the principle described in the theory. Both dark and bright field images were taken in order to obtain information about the sample.

In this thesis, TEM was used for one sample (which was made prior to this diploma work). The result for that is provided in chapter 4

3.5 XRD - Preparation and Analysis

The XRD was used for two different purposes in this diploma work; reflectivity measure-ments in order to measure the thickness of the thin film and thereby the deposition rate and for post-deposition stress measurements to verify the stress from the laser measurements. In the subsections below, the method for both are explained.

3.5.1 Reflectivity Measurements

For the reflectivity measurements, the incident beam optics was parallel beam X-ray mirror for Cu radiation, a fixed beam mask of 10 mm and a slit of 1/32 (divergence slit). For the diffracted beam path the detector used was PIXcel-3D, a collimator slit of 0.27 mm (beam attenuator) and a large beta filter (Ni) was used. These optics were also set in the optics setup in the software. The generator was set to 40 kV and 40 mA. All the positions in the instrument settings tab was set to 0° or 0 mm and the sample was mounted as flat as possible on the sample stage. Since the sample was very thin (fragile), no tape was used to set it in place.

The first measurement was a 2θ-scan in order to check the zero position of the directed beam intensity. A manual scan was done with the following setting:

• Scan Mode Range: 1° • Step Size: 0.010° • Time per Step: 0.5 s

When the scan was done, ’Peak Mode’ was chosen by right mouse clicking and the cursor was moved to the peak position. When the right position of the detector was found, the sample offset for 2θ was set to 0.0° in order to correct the zero position.

The next scan was a Z-scan and this was done to set the sample height. In the instrument settings, Z was set to 9.5 mm in order to match the thickness of the sample. The settings for the manual scan was:

• Scan Mode Range: 1 mm • Step Size: 0.010 mm • Time per Step: 0.5 s

(26)

The Z-value in the instrument tab may need to be changed after this scan to a value less or higher than 9.5 mm. When the scan is done, the graph obtained has high intensity values at low Z and low intensity values at higher Z. The cursor was moved to half of the maximum intensity received. After this, an ω-scan was done in order to set the horizontal tilt. The setting for this measurement was as previous ones. When the scan was done, the cursor was moved to the peak with highest intensity. The Z-scan followed by the ω-scan should be repeated until there are no changes in both the positions, since as one of them are changed, the other will also change. When the last ω-scan is done, the sample offset for ω is set to zero. After this measurement, the scan where the thickness will be determined from, the 2θ-ω scan was done with the following setting:

• Scan Mode Range: 2° • Step Size: 0.005° • Time per Step: 0.5 s

With a software created in MATLAB, the plotted data was downloaded and the peaks of the fringes were analysed in order to measure the thickness of the film and thereby the deposition rate.

3.5.2 Wafer Curvature Measurements for Film Stress

For the wafer curvature measurements, XRD was used. The optics that were used to carry out the measurements were capillary X-Ray lens for the incident beam optics and a triple axis with open detector for the diffraction optics. For the incident beam optics a 2 mm mask was also used in order to confine the X-rays with a size of 2 mm by 2 mm. The reason why these optics were chosen was because of their sensitivity.

The same principle as described for previous subsection for the different scan where followed, but for this measurement also a χ scan was made which is for the tilt in the other direction as ω. For the stress measurements, a χ-ω and an ω-program was made, for nine different positions on the sample (-8 mm, -6 mm, -4 mm, ..., 8 mm) which is called X. A χ-ω program was made in the software to optimise ω (receive high intensity of ω scans) and after that ω scans were done for the nine different positions. This measurement was done for the nine different positions of the sample and for two different samples. One sample was where from the MOS-measurements, a more tensile stress could be observed and for the other sample a more compressive stress could be observed.

(27)

Chapter 4 Related Work

In this chapter, the work previously done in this project regarding stress measurements, vary-ing parameters and the analytical techniques will be presented. This will mostly include re-sults obtained from those measurements that was the main starting point to this diploma work. The method to conduct the experiments are similar to those who have been presented in chapter 3.

4.1 Varying Parameters

In this diploma work, the changing parameters are mainly the working pressure and the position of the magnetron from the sample. The power is however kept constant at 300 W. In previous work, the changing parameters were the power and the working pressure, but the distance between the magnetron and the sample was kept constant. The magnetron was at 70 mm position. What was investigated was at which powers and pressures minimum residual stress is obtained, but also when the quality of the thin films are the best.

The figure below shows some the deposition rates as a function of working pressure for three different powers; 200 W, 300 W and 400 W. It can be seen that for higher powers, the deposi-tion rate is also higher. The powers varied between 1.5 mTorr - 10 mTorr for each series, look at figure 4.1.

Figure 4.1: The difference in deposition rates between different powers for a series of working pressures. The powers chosen were 200 W, 300 W and 400 W, as indicated by the figure.

(28)

Figure 4.2 shows how the stress*thickness varies when the pressure is kept constant at 2.5 mTorr and the power is varied at 200 W, 300 W and 400 W. Figure

Figure 4.2: Difference between stress*thickness, for varying powers; 200 W, 300 W and 400 W. The pressure is kept constant at 2.5 mTorr. For the rates of these depositions, look at 2.5 mTorr in figure 4.1.

The graphs below are given as stress as a function of thickness. The deposition for 1.5/2.5 mTorr, see orange plot in figure 4.3, the pressures were changed between 1.5 mTorr and 2.5 mTorr every 25 nm.

(29)

Figure 4.3: The stress as a function of thickness for different pressures 1.5 mTorr, 1.5/2.5 mTorr, 2mTorr, 2.5 mTorr and 3 mTorr. The power was kept constant at 300 W.

Figure 4.4: The stress as a function of thickness for different pressures 3 mTorr, 5 mTorr and 8 mTorr. The power was kept constant at 300 W.

4.2 Analysis of the Samples

The residual stress was at a minimum for when the pressure was varied between 1.5 mTorr and 2.5 mTorr, at a power of 300 W (see figure 4.3. In order to see the film on a nanoscale level and examine its features such as morphology and crystal structure, TEM was done. The

(30)

principles follows the routine explained in section 2.5.3.2. The figure below shows the bright field image of the sample.

Figure 4.5: A bright field image of the sample where the working pressure is shifted between 1.5 and 2.5 mTorr every 25 nm.

4.3 Conclusions from the Related Work

The conclusion that were drawn from the related work is that a lower deposition rate will influence the quality of the film. For higher pressures, a more tensile stress is created in the film, which will also yield more voids and defects and make it easier for the film to fracture. Higher pressures will also give lower deposition rates, as indicated by figure ??. For this purpose, the range of pressure that was chosen to work in for this diploma work was between 1.5-4.5 mTorr and the power was always at a constant level of 300 W. The TEM-data shows, when the working pressure is shifted between 1.5/2.5 mTorr, the density of the film is high but as the film grows thicker, voids (defects) start to form. This was the main reason why it was chosen to make a series where the distance of the magnetron was moved closer to the sample. As the distance between the magnetron and the sample decreases, according to theory, there will be more energetic bombardment of particles which can make the film grow denser. Figure 4.3 also shows that for the 1.5/2.5 mTorr deposition, that the stress will be compressive from a beginning but as it starts going towards the tensile, it will be hard to reduce it towards lower values. It is also visible in figure 4.5 that for the first layers, there are minimum voids/defects in the structure (where the pressure probably is 1.5 mTorr), but the structure seem to change once the pressure is changed to 2.5 mTorr as voids start to form. Hence tensile stress might be hard to reverse.

(31)

Chapter 5 Result

5.1 The Stress Measurements

The figures in the subsections below show the stress evolution for the pressure and the dis-tance series, look at table 3.1 for the settings. The result is discussed in chapter 6.

5.1.1 Pressure Series

Below, the figures from the pressure series; 1.5 mTorr (figure 5.1), 2.5 mTorr (figure 5.2), 3.5 mTorr (figure 5.3) and 4.5 mTorr (figure 5.4) is presented.

Figure 5.1 shows how the stress evolves for different distances at a pressure of 1.5 mTorr. The distances are indicated both in table 3.1 and in the figure itself. Both the solid and dashed blue graphs shows the stress evolution in the thin film, using the same setting.

Figure 5.1: The stress evolution as a function of thickness for four distances; 15 mm, 30 mm, 50 mm and 70 mm. The pressure is kept constant at 1.5 mTorr for these measurements. There are two graphs for the distance 50 mm provided.

(32)

Both the solid and the dashed green line in figure 5.2 shows the stress evolution using the same settings.

Figure 5.2: The stress evolution as a function of thickness for four distances; 15 mm, 30 mm, 50 mm and 70 mm. The pressure is kept constant at 2.5 mTorr for these measurements. There are two different graphs for the distance 30 mm provided.

Figure 5.3: The stress evolution as a function of thickness for four distances; 15 mm, 30 mm, 50 mm and 70 mm. The pressure is kept constant at 3.5 mTorr for these measurements.

(33)

Figure 5.4: The stress evolution as a function of thickness for four distances; 15 mm, 30 mm, 50 mm and 70 mm. The pressure is kept constant at 4.5 mTorr for these measurements.

5.1.2 Distance Series

Below, the figures from the distance series; 15 mm (figure 5.5), 30 mm (figure 5.6), 50 mm (figure 5.7) and 70 mm (figure 5.8) is presented. For each figure, the distance is kept fixed and the pressure is varied (1.5 mTorr, 2.5 mTorr, 3.5 mTorr and 4.5 mTorr), look at table 3.1.

Figure 5.5: Stress evolution as a function of thickness for four different pressures, 1.5 mTorr, 2.5 mTorr, 3.5 mTorr and 4.5 mTorr, at a constant magnetron position of 15 mm. The noisy data for the 1.5 mTorr graph is due to the bad quality of the laser intensity.

(34)

Figure 5.6: Stress evolution as a function of thickness for four different pressures, 1.5 mTorr, 2.5 mTorr, 3.5 mTorr and 4.5 mTorr, at a constant magnetron position of 30 mm.

Note that for figure 5.7, there are two different plots for 1.5 mTorr. The both depositions were done with the same settings.

(35)

5.2 Other Relevant Results

The result in the subsections below were not mainly part of what was intended for investi-gation. These are however interesting findings which are relevant to the work. Note that the depositions for each subsection are done with the same sample to target distance and sput-tering gas pressure. The only difference is either the base pressure, the temperature of the heater in the MC and and the opening of argon gas inlet, either directly in the main chamber or closer to the target. The figures provided below are plots of the stress as a function of thickness. In Appendix B, the stress*thickness plots are provided.

5.2.1 Effect of Base Pressure in MC

Figure 5.9 below shows what effect a good and a bad base pressure in the MC prior to despo-sition could have on the stress evolution. The depodespo-sitions are made with the same settings, which is when the position of the magnetron was at 70 mm and the pressure of the argon was 1.5 mTorr. The only difference between these depositions is the base pressure, which for one deposition (blue graph) was high (9.9E-7 mTorr) and for the other (red graph) was low (3.2E-9 mTorr). High pressure indicates low vacuum and low pressure indicates high vacuum.

Stress-free titanium-based thin films for inner ear microphones : The last missing part of a technology for totally implantable hearing aid implants

Linköping University | The Department of Physics, Chemistry and Biology

Master’s thesis, 30 ECTS | Applied Physics

2020 | LIU-IFM/LITH-EX-A--20/3821--SE

Stress-Free Titanium-based Thin

Films for Inner Ear Microphones

The last missing part of a technology for totally implantable

hearing aid implants

Spänningsfri och tunn titanﬁlm till en hörapparat för innerörat

Dina Ehsan

Upphovsrätt

Copyright

Abstract

Populärvetenskaplig sammanfattning

Acknowledgments

Contents

Chapter 1

Introduction

1.1

Aim of the Thesis

1.2

Delimitations

1.2.1

The Use of Related Work

1.3

Outline

Chapter 2

Theory

2.1

Project Background

2.2

DC Magnetron Sputtering

2.3

Stress Measurements with Multi-Beam Optical Sensor (MOS)

2.4

Stress - Analysis and Mathematical Background

2.4.1

Tensile and Compressive Stress

2.4.2

Stress Measurements with MOS

2.4.3

Important Aspects of the Stoney Equation

2.5

X-Ray Diffraction (XRD)

2.5.1

X-ray Reflectivity (XRR)

2.5.2

Wafer Curvature Measurements for Film Stress

2.5.3

Transmission Electron Microscopy (TEM)

Chapter 3

Method

3.1

The Substrate

3.2

Sample Cleaning

3.3

Deposition and Stress Measurement

3.4

TEM - Preparation and Analysis

3.5

XRD - Preparation and Analysis

3.5.1

Reflectivity Measurements

3.5.2

Wafer Curvature Measurements for Film Stress

Chapter 4

Related Work

4.1

Varying Parameters

4.2

Analysis of the Samples

4.3

Conclusions from the Related Work

Chapter 5

Result

5.1

The Stress Measurements

5.1.1

Pressure Series