Linköpings universitet SE–581 83 Linköping
Linköping University | Department of Physics, Chemistry and Biology
Master’s thesis, 30 ECTS | Chemistry
2021 | LIU-IFM/LITH-EX-A--2021/3940--SE
Electrically Modified Quartz
Crystal Microbalance to Study
Surface Chemistry Using Plasma
Electrons as Reducing Agents
Pentti Niiranen
Supervisor : Doctoral candidate Hama Nadhom Examiner : Professor Henrik Pedersen
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 detta oförändrat för ickekommersiell forskning och för undervisning. Överföring av upphovs-rätten vid en senare tidpunkt inskränker inte sådant tillstånt. 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 omnämnd 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 så-dant sammanhang som är kränkande för upphovsmannens 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 download, 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/.
Abstract
Metallic films are important in various applications, such as electric devices where they can act as contacts. In electrical devices, the substrate typically consists of silicon dioxide (SiO2) which is a temperature-sensitive substrate. Therefore, plasma enhanced
chemical vapor deposition (PECVD) are better suited than thermally activated chemical vapor deposition (CVD). Depositing metallic films with PECVD demands co-reactants that act as reducing agents. However, these are not well-studied and do not always have the power enough to perform the reduction reaction for metals. Recently it has been concluded that electrons can act as reducing agents in the deposition of first row transition metallic films in a PECVD process. By supplying a positive bias to the substrate, the electrons got attracted to the surface of the substrate, which facilitated metal growth. The study concluded that metal growth only occurred at conductive -and semiconductive substrates and that the substrate bias and plasma power affected the metal growth. The process is however not well understood, which causes a knowledge gap, signifying that studies of the surface chemistry are needed.
Here a new modified analytical method to study the surface chemistry in the newly developed process mentioned above is presented. The analytical method consists of an electrically modified quartz crystal microbalance (QCM) with gold electrodes as a con-ductive substrate. This allows the electron current to run through the QCM during the measurement. By supplying a DC-voltage to the front electrode it gets readily biased (negative and positive) and by placing a capacitor in the circuit, it connects the AC-circuit (oscillator circuit) and the DC-circuit (DC-voltage bias circuit). At the same time, it blocks the DC-current from going back to the oscillator but allows the high-frequency signal to pass from the QCM.
The results in this thesis concluded that the QCM can be electrically modified to al-low an electron flux to the QCM while using it as a substrate when electrons are used as reducing agents. Scanning electron microscopy (SEM) of a QCM crystal revealed that a 2 µm film had been deposited while SEM coupled with energy dispersive X-ray spec-troscopy (EDS) showed that the film indeed contained iron. Further analysis was made by high-resolution X-ray photoelectron spectroscopy (HR-XPS) to find the elemental compo-sition of the film, which revealed that the thin film contained 41 at.% iron. In addition, this study investigated if the QCM could be used to study CVD processes where electrons were used as reducing agents. The results indeed revealed that it is possible to study the surface chemistry where electrons are used as reducing agents with the electrically modified QCM in order to gain knowledge concerning film deposition. Initial results of the QCM showed that film growth could be studied when varying the plasma power between 5 W to 15 W and the QCM bias between -40 V to +40 V. The method generated easily accessible data concerning the process where electrons are used as reducing agents, which gained insight to the method that never has been disclosed before.
Acknowledgments
First and foremost, I want to express my sincere gratitude to professor Henrik Pedersen for giving me the opportunity to contribute to this thrilling and educative research. Also, for opening my eyes to an intriguing field of chemistry. In addition, I would like to acknowledge the, soon to be, Dr. Hama "Hamamuir" Nadhom. While finishing his own thesis, he was still helpful during the writing of this thesis. I would also want to acknowledge guest professor Daniel Lundin, whom I respect for his deep knowledge in plasma physics. I want to thank you Daniel, for making plasmas more understandable. Also, i would like to acknowledge Polla Rouf and Robert Boyd for the XPS -and SEM analysis. In addition, I would like to thank M.Sc. student Pamburayi "Pambu" Mpofu for his office companionship throughout this thesis. Lastly, i want to address a special and big acknowledgment to my Roosa, for being supportive for my wishes and goals, olet tärkeä.
"The more i learn, the more i realize how much i do not know" Albert Einstein
Contents
Abstract iii
Acknowledgments iv
Contents v
List of Figures vii
1 Introduction 1
1.1 Thin films . . . 2
1.2 Thin film deposition . . . 3
1.3 Purpose . . . 4
2 Fundamentals of chemical vapor deposition 5 2.1 Historical development of chemical vapor deposition . . . 5
2.2 Chemical vapor deposition . . . 6
2.2.1 Atomic layer deposition . . . 10
2.3 Plasma enhanced chemical vapor deposition . . . 11
2.3.1 Plasma basics . . . 11
2.3.2 Fundamentals of plasma enhanced chemical vapor deposition . . . 13
3 Quartz Crystal Microbalance 15 3.1 The Working principle of the QCM . . . 15
3.2 The Sauerbrey equation . . . 17
3.3 State of knowledge of the biased QCM . . . 20
4 Characterization techniques of thin films used in this thesis 21 4.1 Scanning electron microscopy . . . 21
4.2 X-ray photoelectron spectroscopy . . . 22
5 Experimental details 25 5.1 System set-up . . . 25
5.2 QCM apparatus . . . 25
5.4 QCM measurements . . . 27
6 Result and discussion 28
7 Conclusion 36
7.1 Outlook . . . 37
List of Figures
2.1 Schematical illustration over a CVD set-up, including (1) carrier gas, (2) evapora-tion chamber that accommodates the precursor and (3) a mass flow controller that controls the flow of gases. When the precursor enters the (4) deposition chamber, it can undergo chemical reactions, eventually forming a film on the substrate that is placed on the (5) substrate holder, to which the heat supply is controlled by (6) a temperature control system. Lastly, (7) a vacuum pump that is used to acquire a desired pressure and (8), a scrubber to which toxic by-products are evacuated to. . 6 2.2 Schematical illustration of the mass transport during film deposition with CVD. . 8 2.3 Classification over different types of CVD. Remade with inspiration from ref [44]. . 10 2.4 Schematic description of an ALD cycle, including sequential introduction of
re-spective precursor to the deposition chamber, where the precursors are allowed to adsorb followed by a purging step with an inert gas in between the introduction of precursor A and B, respectively. . . 11
3.1 (a) Randomly orientated dipole moments and (b) directional dipole moments in-duced by an electrical field. . . 16 3.2 A schematical representation of the cutting angle of the quartz crystal, for AT-cut
quartz crystals ϕ = 35.25° . . . . 17 3.3 Illustration of the effect of the frequency when a rigid film is deposited on the
crystal. The wavelength of the wave increases and consequently, the frequency decreases. . . 18
5.1 The deposition system that was used in this thesis. The figure displays (1) the QCM apparatus and (2), the different parts used in the deposition chamber. . . 26
6.1 Depositions of iron on the QCM crystal when the QCM were biased with -40 V (yellow line), 0 V (red line) and +40 V (black line). By increasing the electrical bias to the QCM, the growth rate increases. The measurement of the deposition shows that the QCM can be electrically biased during the deposition of iron when electrons interact with the QCM. . . 29
6.2 A film that was deposited with iron on a QCM crystal at 15 W and 0 V bias was subjected to SEM -and XPS analysis. In (a), a cross-section obtained by SEM cou-pled with a focused ion beam (FIB) to obtain the thickness of the deposited film while (b) shows an elemental composition mapping (EDS mapping) of the film. In (a), a thickness of approximately 2 µm was obtained while (b) shows that the film indeed contained iron (red area). The green area represents the quartz crystal while the blue area is platinum, which was used to avoid charging effect of the thin film. . . 29 6.3 High-resolution X-ray photoelectron spectroscopy (HR-XPS) spectrum of a quartz
crystal, where an iron film had been deposited on. The deposition parameters was 10 W plasma power and +20 V QCM bias. . . 30 6.4 QCM results of film deposition of iron on a QCM crystal at 0 W, 5 W, 10 W and
15W at a fixed QCM bias of +10 V showing the growth characteristics. For 5 W, the highest growth rate could be monitored but a steep increase was seen after 40 s of the deposition. Depositing at 10 W exhibits a linear growth throughout the whole deposition. When depositing at 15 W, the lowest growth rate was obtained and generated a decrease of growth rate during the deposition. Note that no growth occurred when the plasma was not ignited (0 W). . . 31 6.5 Deposition of iron done on a QCM crystal at 10 W while varying the QCM bias
be-tween 0 V and +40 V. By increasing the bias supplied to the substrate, an increased growth can be seen. . . 32 6.6 Deposition of iron done on a QCM crystal at 15 W while varying the QCM bias
be-tween 0 V and +40 V. By increasing the bias supplied to the QCM, a lower growth rate can be seen. Also, a decrease of growth rate is seen at +30 V and +40 V. . . 33 6.7 By time-resolving the film deposition of iron on a QCM crystal at 10W, +40 V, it is
meant to divide the deposition into sequences instead of depositing the film con-tinuously. The deposition consisted of individual sequences of FeCp2and plasma ignition with 5 s sequences. Figure 6.7 shows that the mass adsorbed to the QCM crystal gradually increases during the sequence when only the precursor is intro-duced to the deposition chamber (white area) and finally saturates after approxi-mately 5 s. In the next sequence when the plasma is introduced and the precursor are discriminated, a decrease of mass can be seen, which is interpreted to be re-moval of cyclopentadienyl ligands from the precursor. . . 34 6.8 Recorded current densities produced by the plasma along with how much current
that is drawn through the QCM. The measurements shows that the QCM draws approximately 100 % of the current produced in the plasma. . . 35
Chapter 1
Introduction
Thin metal films can be found in a broad range of important applications, such as solar cells, mirrors, magnetoresistive access memories and also in microelectronics.[1, 2, 3, 4] In microelectronics, metal films can serve as interconnections between transistors and also as contact material in transistors.[5] Metal films are typically synthesized with either physical vapor deposition (PVD) or chemical vapor deposition (CVD). However, when depositing films on substrates with inherently topographically complex geometries, CVD is more well-suited due to the volume deposition nature of CVD. Employing CVD implies that vaporized molecules undergo chemical reactions in the gas phase and at the gas-solid interface of a substrate. Therefore, when depositing for an example iron films, iron containing molecules are needed. The iron center in these molecules typically inherits a positive oxidation state, meaning that the film formation must proceed through a reduction reaction, from a positive oxidation state to a metallic oxidation state of zero. To emphasize, a contribution of electrons is needed to the iron center. This is normally supplied through a co-reactant that act as a reducing agent. However, these co-reactants are poorly documented and knowledge of molecules with the ability to reduce metals, such as iron which has a negative standard reduction potential are limited.[6] This is due to the thermodynamical disadvantage, which can be interpreted through a positive Gibbs free energy [7]:
∆G=´nFE˝ (1.1)
Where
∆G = The change in Gibbs free energy
n = Number of electrons participating in the reaction F = Faraday’s constant
1.1. Thin films
Considering this, a newly published CVD method revealed that electrons from a plasma discharge can act as reducing agents when depositing cobalt, iron and nickel films.[8] The reduction was induced by a positive bias voltage supplied to the substrate which attracted the plasma electrons to the substrate and consequently led to the growth of a metal film. The results in the study [8] disclosed that the metal content of the film depended on the polarity of the supplied bias and the conductivity of the substrate. With a negative bias voltage to the substrate (i.e. repelling the electrons), no film growth was obtained. In contrast, when a positive bias to the substrate was delivered, film growth was obtained. In the films, a higher metal content was achieved when the deposition was made on an electrically conductive substrate in comparison to a semi-conductive substrate, while no metal content was obtained on a non-conductive substrate. This suggested that the electron current from the plasma had to be conducted away from the substrate for metal growth to occur. However, no clear model concerning how the process worked was presented.
1.1
Thin films
A thin film is an exceedingly thin layer of atoms (monolayer). These monolayers can be stacked on top of each other (multilayer) and ranges between a few micrometers (1 µm = 10´6 m) down to a few nanometers (1 nm = 10´9 m) in thickness and can nowadays be controlled down to an atomic level.[9, 10] The fabrication of thin films is not a new but rather prehistoric concept, which dates back to approximately 5000 BC when the Egyptians coated different artifacts, such as statues, stone vases and nails with layers of gold for decorative purposes. At the time being, thin layers of gold could be as thin as 100-300 nm, roughly equivalent to 0.1 % of the thickness of a human hair.[11, 12, 13] Later, around 1000 BC, mercury was explored in China as an adhesive medium for coating metals on metals (cold mercury gilding), which in 1836 were exchanged to a less toxic production method, elec-trolytic gilding.[14]
Since then, thin films have been synthesized for additional purposes than decorative, such as corrosion resistance for tools that operates in rough environments, increased hardness of cutting tools or to alter the electrical conductivity for the use in microelectronics to name a few.[15, 16] By combining the properties of a thin film that is deposited on a substrate, the surface chemistry can be altered while maintaining the bulk property.[17] One example is to form a metallic film on a silicon wafer, consequently increasing the conductivity of the substrate. Another example can be to coat a corrosion resistant thin film on top of a substrate in order to protect the bulk material from corrosion. Hand-in-hand with the development of thin films, several techniques have emerged and are highly refined and exceptionally sophisticated. To-date, there are a range of different techniques that can be employed to coat thin films, such as CVD and PVD. Both these are deposition techniques that the modern society fully relies on due to the incorporation of thin (metal) films in modern electronics,
1.2. Thin film deposition
such as computers and mobile phones that accommodates transistors. Transistors are the fundamental component of many electrical devices beyond computers and mobile phones and can be found in essentially all electronics that are used on a daily basis. Because of this, without the use of thin films, life would look completely different, and the electronics would probably perform substantially inferior.
1.2
Thin film deposition
Thin films can be deposited by different type of techniques, most commonly electroplating, PVD or CVD. The employed method typically depends on the desired properties, such as thickness, structure and production cost of the thin film. Common to both PVD and CVD is that they are carried out in a vacuum chamber but with the difference of how the film form.
Depositing a film with PVD implies that the film is formed by vaporizing a target mate-rial that carries the desired atoms, such as evaporation or sputtering where high-energetic ions bombard the target material, resulting in vaporized atoms. The evaporated atoms then diffuse towards the substrate where the atoms condense, consequently forming a film on the substrate.[17] A wide range of films can be formed by PVD, where two examples are nitrides and carbides.[18] One advantage with PVD is the absence of toxic waste from the process. However, one clear disadvantage of PVD is the low growth rates in comparison to e.g. CVD. In addition, PVD is a line-of-sight deposition technique, meaning that the deposition is directional and has issues with coating complex geometries, such as deep voids.[17]
On the other hand, CVD relies on chemical reactions for film formation to occur. By al-lowing volatile molecules in a vapor phase which contain the desired atom (referred to as precursors) to react both in the gas phase and at the substrate, it eventually forms a film.[16] This implies that a steady supply of energy (for an example thermal energy, plasma or radi-ation) is needed for the chemical reactions to proceed. However, supplying thermal energy at elevated temperatures poses a problem for temperature sensitive substrate, such as silicon dioxide (SiO2) [19], which is frequently used in the electronics industry. To circumvent the problem with elevated temperatures, the energy stored in a plasma can be used to lower the thermal energy needed for the deposition. When a plasma is used, the technique is often referred to as plasma enhanced chemical vapor deposition (PECVD).
Among the biggest advantages with CVD is the possibility of achieving high deposition rates, meaning that thick coatings can be deposited in a shorter time. In addition, in contrast to PVD, deep voids do not pose a problem for CVD, which can coat very complex geometries. Also, CVD can coat films even in conditions that do not require a high vacuum.[20] However, a downside with CVD is the toxic by-products that is produced and that high temperatures are needed. There are however techniques that can bypass this, such as plasma enhanced
1.3. Purpose
chemical vapor deposition (PECVD).[20] In this thesis, CVD is the technique of interest and is thoroughly described in Chapter 2. For an elaboration of PVD processes, the reader is referred to ref [17].
1.3
Purpose
Information concerning CVD processes can be obtained through a wide range of analyt-ical techniques which are possible to use for in-situ studies, where several methods has been reported successfully. Depending on what kind of information that is sought (for example reaction dynamics or growth rate), different techniques can be employed, such as Fourier-transform infrared spectroscopy, secondary-ion mass spectrometry, ellipsometry and reflectance spectroscopy.[21, 22, 23, 24] However, these techniques are not always possible to install in each and every vacuum chamber. There are alternative analytical techniques that can be easily installed in essentially any vacuum chamber, such as QCM, which has been successfully engaged in several CVD studies.[25, 26] The simplicity of the vacuum compatible QCM, producing easily-accessible data, making it a viable candidate for the above-mentioned process.[8]
For the QCM to be viable in a process where electrons are used as reducing agents, it must be electrically biased to attract the plasma electrons to the surface. To my knowledge, no study has presented a biased QCM where an electron flux is allowed to pass through the crystal while at the same time monitor film growth. At normal operational mode, the QCM is typically not electrically biased and to my understanding, there are at least a couple of known modifications of the QCM that has been used in thin film monitoring, although in PVD and not in CVD.[27, 28, 29]
This thesis aims to develop a new analytical method to study the surface chemistry in the process where electrons as reducing agents.[8] What is proposed in this thesis is an electrically biased QCM that allows an electron flux from a plasma to pass through the QCM crystal during film deposition. This has to my knowledge never been reported. Therefore, the purpose in this master thesis is two-fold:
1. Can the QCM be modified so that it is compatible with a positive bias which allows an electron flux to the QCM which acts as a substrate?
2. Can the modified QCM in (1) be used to understand the process where electrons are used as reducing agent?
Chapter 2
Fundamentals of chemical vapor
deposition
2.1
Historical development of chemical vapor deposition
In 1960, the term “CVD” was coined by the well-known, and sometimes referred to as the father of CVD, J.M. Blocher. Blocher wanted to distinguish between techniques where a solid thin film was obtained through chemical reactions (CVD) to those who were obtained by physical means (PVD).[30]
The history of CVD does however extend further than 1960. Two of the early pioneers in CVD, Sawyer and Man patented a method in 1880 on coating pyrolytic carbon on a lamp filament to increase its luminescence.[31] Succeeding this, several successful results were obtained to make light sources more efficient.[32, 33] In addition to the development of light sources, metal purification was a highly prominent research field at the time, resulting in sev-eral historically important inventions. Whereas one is the Mond process that was developed in 1890 and used for purification of nickel by the use of carbon dioxide. By supplying carbon dioxide to nickel, nickel tetra carbonyl, Ni(CO)4 was formed, a volatile compound from which pure nickel could be obtained by decomposing Ni(CO)4at elevated temperatures.[34] Additional important inventions were the purification of silicon, which was reported 1909, a frequently used element in electronics and was obtained by the reduction of silicon tetrachlo-ride (SiCl4), a process that is still used today.[35] Beyond coating lamp filaments, purifying metals and utilizing CVD for microelectronics (conductive,-semiconductive and insulating films), CVD has been used in a wide range of areas. Examples are optical films (reflective and antireflective), magnetic films (memory), chemical films (sensors, protective barrier, anti-corrosion and catalyst) and mechanical films (wear resistance and hardness).[36, 37, 38, 15]
2.2. Chemical vapor deposition
Owing to the development of vacuum technology, thin film fabrication by CVD has emerged to the technology we know today. Vacuum, from the Latin “vacuus”, meaning empty space, hence evacuating air from the deposition chamber, making deposition possible at lower pressure and with less impurities, such as oxygen. Another factor that assisted CVD to thrive is the constantly increasing demand of semiconductor devices, which emerged in the late 60s.[39, 40, 41]
2.2
Chemical vapor deposition
The abbreviation “CVD” is an umbrella term that comprises several types of CVD techniques. The common denominator for these techniques is that a solid film is produced through chem-ical reactions between vaporized precursors.[42] Baschem-ically, a CVD deposition system consists of the following components: (I) a carrier gas (II) an evaporation chamber that accommo-dates the precursor, (III) a control system for the flow of precursors, (IV) a deposition cham-ber where the film formation proceeds, (V) a substrate holder where the substrate is placed on, (VI) a temperature control system, (VII) a vacuum pump to pump down the system to a desired pressure and (VIII) a scrubber for the toxic by-products. For a schematic illustration, see Figure 2.1.
Figure 2.1: Schematical illustration over a CVD set-up, including (1) carrier gas, (2) evapora-tion chamber that accommodates the precursor and (3) a mass flow controller that controls the flow of gases. When the precursor enters the (4) deposition chamber, it can undergo chem-ical reactions, eventually forming a film on the substrate that is placed on the (5) substrate holder, to which the heat supply is controlled by (6) a temperature control system. Lastly, (7) a vacuum pump that is used to acquire a desired pressure and (8), a scrubber to which toxic by-products are evacuated to.
2.2. Chemical vapor deposition
Before the deposition is performed, the evaporation chamber that accommodates the (typically) solid precursor (which contain the desired atom(s)) are supplied with thermal energy, allowing the precursor to evaporate. However, note that the precursor also can be obtained by flowing a gas over a solid [42] such as aluminum trichloride (AlCl3):
2Al(s) +6HCl(g)Ñ2AlCl3(g) +3H2(g) (2.1)
When the precursor is vaporized, a carrier gas can be allowed to flow through the evapo-ration chamber, carrying the precursor to the deposition chamber. The supplied volume of carrier gas is typically given in units of standard cubic centimeter per minute (sccm) and are controlled with a mass flow controller. In the deposition chamber, the precursor are allowed to react when sufficient energy has been supplied (or other activated manner, such as a plasma) where the temperature is typically controlled by a temperature control system. Inside the deposition chamber, different reactions can occur and are separated into gas phase reactions and surface reactions. The gas phase reactions refer to the decomposition and/or dissociation of the precursors in the gas phase which activates the precursors. The species formed in the gas phase are then transported across a boundary layer in the vicinity of the substrate, which is a horizontal region that gets established above the substrate due to the flow of gases. In the boundary layer, the velocity and concentrations of the species are not the same as in the main gas stream.[43] The mass transport of the precursors across the boundary layer is given by the product of the mass transfer and the concentration gradient of precursors in the gas phase and at the surface, whereas the mass transfer is the quotient of the diffusivity of a species and the thickness of the boundary layer:
F= D
δ ˆ(Cg´Cs) (2.2)
Where
F = Flux of mass transport from the gas phase to the substrate D = Diffusion constant of the precursor
δ= Thickness of boundary layer
Cg= Concentration of precursor in the gas phase CS= Concentration of precursor at the substrate
Well at the surface, the activated precursors adsorbs to the substrate, which is placed on the substrate holder. At the surface, the species can interact with the surface, where it either chemisorb or physisorb which is followed by diffusing along the surface for latter chemisorp-tion. Eventually, the strength of the bond between the physisorbed species and the substrate becomes more prominent in strength and film growth starts. From the reactions that occur at
2.2. Chemical vapor deposition
the surface, volatile by-products might form, which are eventually evacuated to a scrubber due to the typically toxic nature of the by-products.[16] For a scheme over gas phase -and surface reactions, see Figure 2.2.
Thus, CVD can be described as a six-way process: (I) generation of reactive species, (II) transport of the reactive species to the substrate, (III) adsorption of the reactive species to the substrate, (IV) surface reactions followed by (V) desorption and (VI) transport of the by-products away from the substrate.[16]
Figure 2.2: Schematical illustration of the mass transport during film deposition with CVD.
The nature of chemical reactions, typically being thermodynamically endothermic (∆H > 0) implies that a steady supply of energy is necessary in order to activate the chemical reaction (∆ G < 0) during the deposition, according to Gibbs free energy [7]:
∆G=∆H ´ T∆S (2.3)
Where
∆G = The change in Gibbs free energy ∆H = The change in enthalpy
T = Temperature
2.2. Chemical vapor deposition
Consequently, deposition with CVD requires activation of the reactions. The type of en-ergy used to activate the reactions generates different types of CVD, such as thermal enen-ergy (e.g. high-temperature CVD, HTCVD), radiation (e.g. laser CVD, LCVD) or electrically (e.g. plasma enhanced CVD, PECVD).[16] Normally, if the source of energy is not defined, the activation energy originates from thermal energy supplied to the system.
In addition, CVD can be operated at different pressures, of which CVD are classified ac-cording to. The types of CVD range from atmospheric pressure (atmospheric pressure CVD, APCVD) and sub-atmospheric pressure (low pressure CVD, LPCVD) to ultra-high vacuum (ultra-high vacuum CVD, UHVCVD).[16] The pressure has a big impact on the collision frequency during the deposition process in CVD. This can be observed when calculating the mean free path (λm f p, Equation 2.4), a lower pressure during the deposition increases the mean free path, which is the distance a particle needs to travel before colliding with another particle.[44] By managing the pressure, the collision frequency can be controlled, which might reduce the growth rate but increase the quality of the thin film.
λm f p= kT ?
2πPd2 (2.4)
Where
λm f p= Mean free path k=Boltsmann’s constant T=Temperature
P=Pressure
d=molecule diameter
Lastly, CVD can be classified according to the type of precursors that are used. Com-mon examples of precursors are metalorganic precursors (metal organic CVD, MOCVD) and halides (halide CVD, HCVD).[16] For a scheme of different classifications, see Figure 2.3.
2.2. Chemical vapor deposition
Figure 2.3: Classification over different types of CVD. Remade with inspiration from ref [44].
2.2.1
Atomic layer deposition
Atomic layer deposition (ALD) is a type of CVD that is a younger technique in comparison to conventional CVD. The difference between ALD and CVD is that ALD is based on self-limiting surface reactions that occurs in sequences. ALD was first described in the 1960s in the former Soviet Union but under the name “molecular layering” and later developed by the well-known Finnish researcher, Tuomo Suntola under the name “atomic layer epitaxy, ALE” which got patented in 1972.[45] However, the working name for this technique are now commonly referred to as ALD. The ALD is a highly sophisticated technique to deposit thin films that can be controlled down to an Ångström level (1 Å = 10´10 m). There are several advantages when depositing films with ALD, such as high good step coverage and confor-mal coatings, even in high aspect ratio areas.[46] However, one of the biggest bottlenecks for ALD is the slow growth rate that are achieved, meaning that it takes much longer time to deposit a film with ALD in comparison to conventional CVD. For an example, to deposit an Al2O3with ALD can generate a growth rate of 1 Å per cycle meanwhile the same coating can generate a growth rate of 180 Å per minute with conventional CVD.[47, 48]
When depositing material with CVD, the film formation occurs through both chemical reactions in the gas phase and also at the surface as described above. In ALD, the film formation proceeds through surface reactions, solely. By introducing the precursors in in-dividual sequences to the deposition chamber, the precursors are allowed to adsorb to the surface while excess precursors are purged away. These different steps where precursors are
2.3. Plasma enhanced chemical vapor deposition
introduced to the deposition chamber and purged away together forms an ALD cycle.[46] These ALD cycles can be described according to following. Precursor A gets introduced to the deposition chamber where it is allowed to chemisorb on the surface of the substrate, any excess of precursor A is then purged away from the deposition chamber. This enables one monolayer to chemisorb to the surface of the substrate. In the next sequence, precursor B is introduced to the deposition chamber, which reacts with the chemisorbed precursor A through ligand exchange. After this, any excess of precursor B and also the reaction by-products between precursor A and precursor B are evacuated from the deposition chamber. These ALD cycles are repeated until the desired thickness of the film is obtained. For a schematic description of an ALD cycle, see Figure 2.4.
Figure 2.4: Schematic description of an ALD cycle, including sequential introduction of re-spective precursor to the deposition chamber, where the precursors are allowed to adsorb followed by a purging step with an inert gas in between the introduction of precursor A and B, respectively.
2.3
Plasma enhanced chemical vapor deposition
2.3.1
Plasma basics
Astoundingly, 99 % of our universe constitutes of plasma, making the plasma by far the most common state of matter.[9] In space, the plasma can be found in various interstellar matter while it on earth mostly is artificially obtained. The phenomenon of plasmas is old but the term ”plasma” is a relatively new and dates back to a study from 1928, when Irving Lang-muir, a well-known American physicist and chemist investigated the oscillations in ionized
2.3. Plasma enhanced chemical vapor deposition
gases.[49] By supplying sufficient energy to any solid, it gradually gets more disordered, eventually entering a gas phase. If additional energy is supplied to the gas, electrons orbiting the nucleus get stripped from the species and becomes ionized. Unlike a gas, a plasma is a medium where electrons, ions but also neutrals co-exist and are therefore regarded as the 4th state of matter. In the plasma, the density of electrons (ne) and ions (ni) are roughly balanced (ni«ne).[50] This implies that, on a macroscopic view, the sum of the ion densities is equal to the density of the electrons and is referred to as the plasma density (n). The total plasma density is given by the sum of different ion densities (n1, n2, n3. . . ):
ne=n1+n2+n3+..=n (2.5)
As a whole, the plasma volume is electrically neutral but when looking at individual particles (microscopic level), the plasma is not neutral, i.e. it is quasi-neutral. Additionally, in contrast to a gas, the plasma can conduct electricity since it consists of freely moving charged particles. A plasma can be either thermal or non-thermal, implying that the ions and electrons are either in thermal equilibrium or not. When a plasma is in thermal equilibrium, the temperature of the electrons is roughly in the neighborhood of the ion temperature and are in addition almost fully ionized (α « 1), meaning that almost no density of neutral species (nn) can be found in the volume of the plasma [50]:
α= ni
ni+nn (2.6)
However, thermal plasmas are mostly found in interstellar matter and because of this not interesting from a film deposition point-of-view. Non-thermal and weakly ionized (α < 1) plasmas (also known as cold plasma and non-equilibrium plasma), such as the glow discharge in film deposition is however of interest. In these plasmas, the electrons and ions are not in thermal equilibrium, implying that the temperature between these are not equal. In processes such as PECVD, the electron temperature is in the range of 1 eV ď kTe ď10 eV while the ion temperature is merely « 0.04 eV (1 eV = 11600 K).[16] At these conditions, the electron density and the ion density are typically in the range of approximately 108to 1012 ˆcm´3, respectively, while the density of neutrals are « 1015 ˆcm´3.[16] The generation of a plasma can be realized through several means, but in thin film deposition it is typically done by exposing a gas to an electrical field. By supplying an electrical field to the gas (for an example argon) which is stored in a hollow cathode at a pressure (p) of 1.3 Pa ď p ď 133.3
2.3. Plasma enhanced chemical vapor deposition
Pa , several species are obtained during the discharge, such as ionized atoms (Equation 2.7) and excited atoms (Equation. 2.8) [16]
Ar+e´ÑAr++2e´ (2.7)
Ar+e´ÑAr˚+e´ (2.8)
2.3.2
Fundamentals of plasma enhanced chemical vapor deposition
The plasma has various applications in material science, where it can be used in different ways. In PECVD, the plasma is used in a reactive mode fashion by generating reactive species through fragmentation of the precursor.[16] The production of reactive species is obtained by inelastic electron-precursor collisions, also known as electron impact, which generates reactive species, such as radicals and ions. Radicals, which possesses unpaired electrons, making the radical very eager to form new bonds. Also, in addition to reactive radicals and ions, UV-photons can be found in the plasma that is due to the recombination of ions and electrons and can be used to dissociate the precursors.[51]
Even though the PECVD process is even further from thermal equilibrium in compari-son to thermally activated CVD, meaning that the film formation would not normally occur. However, by taking advantage of the reactive species formed in the plasma a film can be allowed to form. The inelastic collisions that occur in the plasma activate the gas phase chem-istry, enabling a thin film to form at lower temperatures even though it is thermodynamically unfavorable. Common examples of inelastic electron-precursor collisions are the dissociation (Equation 2.9), ionization (Equation 2.10) and dissociative ionization (Equation 2.11), where X = precursor [16]:
e´+X2Ñ2X+e´ (2.9)
e´+X2ÑX+2 +2e´ (2.10)
e´+X
2ÑX++X+2e´ (2.11)
Therefore, instead of supplying thermal energy to activate the reactions as in thermal CVD, PECVD utilizes the highly energetic electrons that reside in the plasma. For example, when depositing silicon nitride, temperatures can reach 850 °C for thermal CVD processes while in PECVD, a temperature of merely 350 °C is sufficient for the film formation.[52] Because
2.3. Plasma enhanced chemical vapor deposition
of this, PECVD is a commonly used technique in, especially, the electronics industry where temperature sensitive substrates, such as SiO2 is frequently used. By supplying too much energy to a temperature sensitive substrate, film stress might arise.
PECVD is similar to the thermally activated CVD processes (with the exception of the reaction activation) and can be explained by the same six-step process: (I) generation of reactive species, (II) transport to the substrate, (III) adsorption to the substrate, (IV) surface reactions followed by (V) desorption and (VI) transport of the by-products away from the substrate.[16] The plasma is highly complex, and a multitude of reactive species can be found which makes the PECVD process complex to study. However, advantages with using PECVD is the low deposition temperatures that can be carried, the high deposition rates that can be achieved and the complex geometries that can be coated.
Chapter 3
Quartz Crystal Microbalance
Quartz crystal microbalance (QCM) is a sensor that has an inherent ability to measure mass changes down to the nanogram regime. The name “quartz crystal nanobalance” would there-fore be a more suitable name. The ability to measure small and rapid changes in adsorbed mass has been explored in various technical fields where the application can be in biosens-ing, corrosion studies or in film deposition where it is used to study film growth, to name a few.[53, 54] One strength with the QCM is that it is vacuum-compatible, making it a power-ful tool when measuring adsorption at a gas-solid interface that proceeds in vacuum, such as in CVD. Multiple studies have proven the effectiveness of employing the CVD in surface studies.[55]
3.1
The Working principle of the QCM
Basically, a QCM consist of a piezoelectric quartz crystal that is sandwiched in between two metallic electrodes (typically gold). To the electrodes an AC voltage is supplied, which induces an oscillation of the quartz crystal when the frequency of the AC voltage matches the resonance frequency of the crystal. With an AC current, the polarity of the current changes and thus, makes the crystal oscillate back and forth (compressed and stretched crystal). During the oscillations, an acoustic wave travels forth and back between the electrodes, where the wavelength of the wave is inversely proportionate to the frequency of the wave which decreases when the surface gets thicker.[56] So, when mass adsorbs to the crystal, the frequency of the oscillations decreases. Initially, the quartz crystal oscillates at its resonance frequency, when mass adsorbs to the QCM, the oscillation frequency of the QCM decreases. The frequency change between these values can then be probed to generate information concerning the mass that absorbs at the gas-crystal interface.[56]
The basis of the QCM is the converse piezoelectric effect (sometimes referred to as inverse piezoelectric effect). The name piezoelectricity derives from the Greece words “piezein” and “elektron”, meaning “to press” and “amber”, respectively and can be translated to “press
3.1. The Working principle of the QCM
electricity”. So, piezoelectricity refers to the phenomenon when a material has the inherent ability to produce electricity in response to a mechanical stress. The discovery of piezoelec-tricity was made in 1880 by the brothers Jacques and the more famous Pierre Curie.[57] The year after the discovery of piezoelectricity, the French Physicist Gabriel Lippman anticipated that an inverse piezoelectric effect would exist, meaning that a mechanical stress can be induced by electricity. The converse piezoelectric effect was later verified and used for the first piezoelectric oscillator, patented 1918 by Bell Telephone Laboratories.[58, 59] The crystal used in this oscillator was Rochelle salt, a material that readily dissolve in water. For that reason, the Rochelle salt was exchanged to the quartz crystal (SiO2) which is still used today. Both these crystals, and piezoelectric materials in general, are non-centrosymmetric, meaning that they lack an inversion center. Consequently, when an external force is applied to these materials, the overall polarization (ÝÑP ) is non-zero, which is crucial for the application of mass sensing in the QCM.
On a molecular level, the dipole moments in a non-centrosymmetric crystal (such as quartz) tend to be randomly distributed at equilibrium (Figure 3.1(a)). By applying an electrical field to a non-centrosymmetric crystal, the equilibrium gets disturbed, which generates a net dipole moment (ÝÑP ‰ 0). The consequence of this is that the dipole moments of each unit cell order themself accordingly to the polarity of the electrical field (Figure 3.1(b)).[56] This phenomenon enables the quartz crystal to oscillate forth and back. Note that, by using an AC current, the polarity of the current changes and thus, makes the crystal oscillate forth and back. The (horizontal) shear oscillations make the crystal compress and stretch. When oscillating, an acoustic wave travels forth and back between the electrodes, where the wave-length of the wave is inversely proportionate to the frequency of the wave which decreases when the surface gets thicker.[56]
Figure 3.1: (a) Randomly orientated dipole moments and (b) directional dipole moments induced by an electrical field.
3.2. The Sauerbrey equation
The quartz crystal can be cut in different orientations, generating specific features of the crystal, such as temperature sensitivity to name one. The most common cut that is employed is the AT-cut, which is cut at a 35.25° angle in respect to its optical axis (z axis), see Figure 3.2. Generally, when the temperature supplied to a given material increases, the stiffness of the material decreases and is due to distortion of atoms, resulting in weaker interactions between the individual atoms in the crystal. The stiffness of a material is given by the stiffness tensor and for AT-cut crystals, certain components in the stiffness tensor increases with increased temperature. This implies that the stiffness increases and consequently, the AT-cut quartz is referred to as having a temperature-compensated cut.[56]
Figure 3.2: A schematical representation of the cutting angle of the quartz crystal, for AT-cut quartz crystals ϕ = 35.25°
3.2
The Sauerbrey equation
Günter Sauerbrey discovered that the change in resonance frequency of a quartz crystal was directly proportionate to the adsorbed mass on a piezoelectric crystal during his doctoral studies 1959.[60] As stated above, a sound wave (also known as transverse wave) between the electrodes is generated when the crystal is set to shear oscillation. When the thickness of the crystal increases (i.e. the thickness of an adsorbed layer increases), the wavelength gets longer before reflecting back to the back electrode, see Fig. 3.3. An increased wavelength of a wave corresponds to a decreased frequency of the same wave according to
f0= Cs λ =
Cs 2(ds+df)
3.2. The Sauerbrey equation
Where
f0is the fundamental resonance frequency Csis the speed of sound wave through the crystal λis the wavelength of the sound wave
dsare the thickness of the electrodes df is the thickness of the deposited film
Figure 3.3: Illustration of the effect of the frequency when a rigid film is deposited on the crystal. The wavelength of the wave increases and consequently, the frequency decreases.
From Equation 3.1 it can be concluded that f09 λ´1, implying that a decrease of frequency can be observed when the film thickness increases. However, note that Equation 3.1. only holds if the thin film is rigid (firmly attached) and evenly distributed over the front electrode. Whereas the film thickness can be obtained by:
df =´ 1 pf ˆ Zq 2 f02ˆ ∆ f n (3.2) Where
df is the thickness of the deposited film Pf is the density of the thin film
Zqis the acoustic wave impedance in the AT-cut quartz (8.8 ˆ 106ˆkg ˆ m´2ˆs´1) f0is the fundamental resonance frequency
∆f is the shift of the frequency n is the overtone order
3.2. The Sauerbrey equation
The reader should note that Equation 3.2. can be rearranged in order to express it as mass density adsorbed to the front electrode and in addition, expressed in more simplified terms, where the equation is commonly known as the Sauerbrey equation.[56]
mf =´ a pqGq 2 f02 ˆ ∆ f n =´Cmsˆ∆ f n (3.3) Where
mf is the mass density
pqis the density of quartz (2.648 g ˆ cm´3)
Gqis shear modulus of quartz (2.947 ˆ 1011ˆg ˆ cm´1ˆs´2) f0is the fundamental resonance frequency
∆ f is the shift of frequency n is the overtone order
Cmsis the mass sensitivity constant
The Sauerbrey equation is often expressed in terms of Cms (mass sensitivity), which is a constant that, typically, depends on the resonance frequency of the crystal. The mass sen-sitivity informs what mass density (ng ˆ cm´2) that is needed to shift the frequency 1 Hz. The mass sensitivity is proportionate to the inverse square root of the resonance frequency (C 9 f´20 ), implying that crystals with a higher fundamental resonance frequencies have a higher mass sensitivity. However, the noise (N) that is obtained during the measurements is proportionate to the square root of the resonance frequency (N 9 f2
0).[56] Choosing a crystal with a higher resonance frequency do because of this not necessarily imply that it performs better than a crystal with lower resonance frequency.
When monitoring thin film growth, it is necessary to choose a suitable crystal that can be employed in the environment it is supposed to work in, such as temperature during the deposition. Also, the density of the thin film highly affects the read-out value, which can be seen in Equation. 3.2. To be clear, when depositing a thin film that consists of multiple species with all-over unknown density it affects the read-out thickness. This implies that each crystal needs to be subjected to post deposition thickness analysis when there exist an uncertainty of the film density. However, the QCM can still be employed to appreciate the growth rate during the deposition processes.
3.3. State of knowledge of the biased QCM
3.3
State of knowledge of the biased QCM
Typically, the QCM are employed when it is not electrically modified through a voltage bias. However, at least a couple of studies has reported a modified QCM which were electrically modified.[27, 28, 29] The aim in these three studies was to measure the metal flux ionization in PVD deposition and were carried out through two different modifications. In one of the modifications (1) [27, 29], the QCM was combined with a gridded energy analyzer (GEA). By placing the GEA on top of the QCM and bias all the grids with different polarities, it discriminated the cationic -and neutral species. To emphasize, a negative bias to the grids allowed cationic -and neutral species to be deposited on the QCM crystal. In contrast, when the grids were biased with a positive polarity, only the neutral species where allowed to be deposited on the QCM crystal. Note that the different grids where biased in such way that electrons were not allowed to reach the crystal. In another study (2) [28] , a strong magnetic field was used in combination with a directly biased QCM. The magnetic field was used in the study to deflect the electrons from reaching the QCM crystal, while the bias of the crystal allowed for discrimination between neutrals and cations. The metal flux (Fm) in (1) and (2) could be calculated from the total flux rate (ions and neutrals, Rt) and the neutral flux rate (Rn).
Fm= Rt ´Rn
Chapter 4
Characterization techniques of thin films
used in this thesis
4.1
Scanning electron microscopy
Scanning electron microscopy (SEM) is a widely used characterization technique for thin films to obtain information concerning e.g. elemental composition and topography. The word microscope descends from the Greek words “mikros” and “scope”, meaning “small” and “look”, respectively. While the human eye can only distinguish two different objects that are bigger than 0.2 mm (1 mm = 106nm), a modern SEM can distinguish between two objects that is bigger than 1 nm (1 nm = 10´6 mm).[61] This is what is referred to as resolution, the ability to distinguish two different objects. With a resolution down to 1 nm, the SEM has the ability to generate important information concerning a material, such as surface structure.[61] Because of this, SEM is widely used in multiple disciplines, such as material science, forensic science, mining, life science and is used in both academia and in the industry.
Basically, a SEM consists of (1) an electron source, (2) electromagnetic lenses, (3) sample chamber, (4) detector(s) and (5) a vacuum chamber. The electron gun ejects primary electrons (1-30 keV, [62]) which are focused to a narrower diameter by the electromagnetic lenses. When the primary electrons are ejected from the electron gun, the diameter of the beam is initially typically « 50 µm. When the beam passes the electromagnetic lenses (condenser lens, objective lens and scan coils), the diameter of the electron beam gets reduced to the range of 0.5 µm to 10 nm.[61] To resolve small features, a narrow diameter of the electron beam is crucial. When the beam of primary electrons has passed the electromagnetic lenses, it interacts with the sample. The electron-sample interaction can generate a wide range of information, which depend on what interaction that is recorded. However, herein we focus on secondary electrons (SE) and X-ray photons.
4.2. X-ray photoelectron spectroscopy
If the electrons inelastically interact with the surface atoms, it causes electrons to be emitted from an outer shell. These electrons are SE and are commonly used to imaging the sample, generating information concerning the topography of the sample. The intensity of the SE that are emitted from the sample highly depends on the structure of the surface. At specific features, such as edges, a higher number of electrons can be emitted in comparison to a void, where there is a higher probability for SE to be adsorbed by the sample. This means that surface features can be obtained by measuring SE intensity across the sample surface.[61]
An additional source of information are the X-ray photons, which carries information about the elemental composition of a sample. The X-rays are collected by an EDS detector, which is an abbreviation for energy dispersive X-ray spectroscopy detector. The EDS measures the en-ergy and intensity distribution of the emitted X-rays. By the interaction between the primary electrons and the sample, it causes a hole in an inner shell of the atom when the primary electrons have a high enough energy to ionize the atoms. To fill the hole that is created in the shell, an electron from an outer hole relaxes to the inner hole (i.e. a lower energy state). Consequently, the energy difference between these energy levels are then emitted as an X-ray photon, which are characteristic for each element in the periodic table due to the defined electron structure of each element. These X-rays enables elemental composition analysis of the sample.[61]
In this thesis, a focused ion beam (FIB, Crossbeam 1540 EsD, Zeiss) in combination with a SEM (LEO 1550 Gemini microscope (Zeiss)) equipped with an EDS detector (Oxford Instru-ments) was used to determine the thickness of the deposited film along with an elemental composition of the thin film.
4.2
X-ray photoelectron spectroscopy
In 1981, the famous Swedish physicist, Kai Siegbahn, at the time active at Uppsala University received the Nobel prize for his work concerning the X-ray photoelectron spectroscopy (XPS). The XPS is a powerful and highly sensitive tool in material analysis to analyze surfaces (ď 10 nm) of a sample down to a detection limit of 0.1 to 1 at.%.[63] The XPS can acquire a wide range of information concerning a sample, such as elemental composition, chemical state (oxidation state) and bonding state in the sample. The scientific principle behind the XPS is the photoelectric effect, a physical phenomenon that Albert Einstein was awarded the Nobel prize for in 1924.[64] By exposing a material with light, electrons can be emitted (i.e. photoelectrons) if the energy of the light exceeds the ionization energy of the atoms in the material.
Principally, an XPS spectrometer consists of (1) a vacuum chamber, (2) an X-ray source, and (3) a detector. In a vacuum chamber, a sample gets irradiated with X-ray photons which
4.2. X-ray photoelectron spectroscopy
originates from the X-ray source. The X-ray photons ionizes the atoms if the X-ray photon energy (hvphoton) exceeds the ionization energy of the atom, consequently making a core electron to get ejected. The kinetic energy of the ejected electron (Ekin) is what the detector measures and is the energy difference between the X-ray photon and the binding energy of the electron. This is expressed as [65]:
EB=hvphoton´Ekin´ φ (4.1)
Where
EB= Binding energy
hvphoton= The X-ray photon energy
Ekin= The kinetic energy of the emitted electron φ= The working potential of the spectrometer
In the periodic table, each element contains an individual set-up of electron configura-tion, meaning that the binding energy of these electrons are specific for each element. This is why XPS can enable elemental analysis of the sample. Furthermore, assume an atom (e.g. iron) with an oxidation state of 0, having a pre-determined bonding energy. Now assume that iron loses an electron, i.e. generates an oxidation state of +I. At this stage, to remove one additional electron from the iron atom would require a higher energy due to the stronger attraction to the nucleus of the remaining electrons. Generally, for higher oxidation states, the bonding energy for the specie increases.[65] Due to this, the XPS can detect the chemical state of the atoms within a sample. Lastly, assume iron being bonded to a more electronegative element, such as oxygen. In this configuration, oxygen pulls the electrons more strongly than iron, meaning that a higher density of electrons resides at the oxygen. Consequently, this implies that it needs more energy to remove an electron from iron. Thus, the bonding energy in this configuration increases. The XPS can because of this also detect the bonding state within the sample. To depict the bonding energies in different chemical states such as Iron [66], see Table 4.2.
Table 4.2:Binding energy for different oxidation states of iron.[66] Compound Binding energy Fe2p3/2(eV)
Fe 706.3
FeO 709.7
Fe2O3 710.6
In this thesis, high-resolution X-ray photoelectron spectroscopy (HR-XPS) was used for ele-mental analysis along with detecting the oxidation state of iron. It was performed by scanning
4.2. X-ray photoelectron spectroscopy
the thin film in the energy range of 0-1200 eV with a pass energy of 160 eV and step size of 0.1 eV. A binding energy in the range of 20-40 eV was used for high resolution spectra with a pass energy of 20 eV. The sputtering source consisted of argon (0.5 keV). A C 1s peak with a value of 285 eV was used for calibration in all the spectra.
Chapter 5
Experimental details
5.1
System set-up
The deposition system consisted of a custom-built 150 mm diameter vertical stainless-steel deposition chamber. To the deposition chamber, a mechanical pump and a stainless-steel evaporation chamber was connected through metal pipes. The pipe connecting the stainless-steel evaporation chamber and the deposition chamber was heated using heating tape (Hemi Heating® Type S) to eliminate possible recrystallization of the precursor, resulting from any temperature gradient between the evaporation chamber and the deposition chamber. The evaporation chamber was connected to the carrier gas flask via metal pipes, whereas the flow of carrier gas was controlled by a mass flow controlled. In the lid of the deposition chamber, a custom-build titanium hollow cathode plasma system (diameter = 7 mm, length = 53 mm) was mounted. To the hollow cathode, the plasma gas (Ar, 70 sccm) was supplied through a PTFE (polytetrafluoroethylene) hose, whereas the flow was controlled by a mass flow controller. In addition, to the hollow cathode, two PTFE hoses was connected, which supplied circulating cooling water to the hollow cathode. The ignition of the plasma was enabled by supplying a DC-current from a plasma power supply.
5.2
QCM apparatus
The QCM sensor (Allectra) was connected to an oscillator circuit (McVac), which in turn was connected to a thin film monitor unit (BeamTec TFM260). The obtained frequency change from the measurements was analysed using SQM160 software (Inficon). Both the oscilla-tor circuit and the bias voltage was connected to the QCM through a BNC (Bayonet Neill-Concelman) feedthrough flange, with a capacitor (100 nF) placed in the connection outside the BNC feedthrough flange. The capacitor connected the two circuits (AC -and DC-circuit) and in addition allowed probing for the frequency change while at the same time blocking the DC-voltage from entering the oscillator circuit. Through the BNC feedthrough flange, the bias voltage was supplied through a coaxial cable from a power supply (EA Ps 200 B), which readily biased the front electrode of the QCM. From the bottom of the QCM sensor,
5.3. Film deposition
a coaxial cable connected the backside of the QCM housing to the oscillator circuit, which in turn was connected to the thin film monitor unit. To electrically isolate the grounded chamber, an electrically insulating PTFE vacuum break was connected in between the BNC feedthrough flange and the CF (conflat) flange. Additionally, the QCM was connected to a water-cooling system through electrically non-conductive PTFE hoses. However, the pur-pose of the water-cooling system was to protect the crystal from thermal breakage, but this was found to not happened. Therefore, the water-cooling system was not used during the measurements. Lastly, a multimeter was connected to the set-up to allow recordings of the plasma current passing through the QCM during the experiments. For a scheme of the con-nections, see Figure 5.1.
Figure 5.1: The deposition system that was used in this thesis. The figure displays (1) the QCM apparatus and (2), the different parts used in the deposition chamber.
5.3
Film deposition
The deposition was done by depositing iron films, using ferrocene, bis(cyclopentadienyl)Fe(II), (FeCp2) as the precursor, plasma electrons as a reducing agent and utilizing a gold coated QCM crystal as a substrate. Both the carrier gas for FeCp2and also the plasma gas consisted of argon (99.995 %). During all film depositions, the flow of carrier gas was 2 sccm while 70 sccm was supplied to the hollow cathode. The evaporation of the FeCp2 was achieved by heating the stainless-steel evaporation chamber with a custom-made heating jacket, making
5.4. QCM measurements
FeCp2sublime at 70 °C, consistent with the determined sublimation temperature of FeCp2 prior to this study.[8] Before all film depositions, the evaporation chamber was purged with argon gas (35 sccm, « 1.5 h), while pumping down the pressure (ď 6 Pa) of the deposition chamber. The deposition consisted of igniting the plasma simultaneously as FeCp2was intro-duced to the deposition chamber for a time frame of 60 s. The film deposition was performed as a function of plasma parameters, where a plasma power between 0 and 15 W was used, the QCM was biased between -40 V to +40 V. In addition, time-resolved depositions were performed, which consisted of introducing the precursor to the chamber for 5 s (plasma not ignited) followed by igniting the plasma for 5 s (precursor not introduced).
5.4
QCM measurements
The QCM measurements was conducted with gold coated AT-cut quartz crystal (d = 14 mm, f0= 6 MHz) that was bought from ColnaTec and was used as a substrate in the depositions. Each measurement consisted of placing the QCM sensor in the gas stream upstream, 70 mm diagonally from the cavity of the hollow cathode. To ensure similar surface chemistry dur-ing each deposition, the quartz crystal was replaced with a fresh uncoated crystal before each measurement. In addition, before each deposition, the QCM housing was freshly masked (see Figure 14) with Kapton tape to ensure maximum deposition on the quartz crystal and thus minimizing deposition on the QCM housing. In addition, the lid of the deposition chamber was masked with Kapton tape to ensure no leakage of plasma electrons during the deposi-tion. Each deposition was started 9 s after the QCM measurement was started to generate repeatable data acquisition.
Chapter 6
Result and discussion
The thesis had two-fold purpose, (1) to investigate if the QCM can be modified so that it is compatible with a positive bias which allows an electron flux from the plasma to the QCM where the QCM crystal acts as a substrate. The second goal (2) was to investigate if the mod-ified QCM could be used to understand the process where electrons are used as reducing agents.[8]
Firstly, to answer the question tied to the first purpose, that the QCM could be modified with a positive bias which allowed an electron flux from a plasma to the QCM, film de-position with FeCp2 as a precursor was carried out. The dede-positions were carried out at a substrate bias of -40 V, 0 V and +40 V, respectively. During the depositions (-40 V, 0 V and +40 V), the plasma power was kept constant at 10 W. The results showed that when depositing iron on the QCM crystal at a substrate bias of -40 V, no detectable film growth occurred. When setting a substrate bias of 0 V, film growth starts and that the growth rate (time derivative of the thickness) drastically increases when increasing the substrate bias to +40 V, see Figure 6.1 for results. This agrees with earlier reported results when electrons were used as a reducing agents for the deposition of iron films, which concluded that the growth rate of iron films was highest at +40 V when depositing under the same substrate bias range.[8] Also, note that a delay of the film growth is seen for +40 V, which increases after 17 s, this is believed to be a nucleation delay. This has earlier been reported to appear in metal deposition of CVD processes.[67]
Figure 6.1: Depositions of iron on the QCM crystal when the QCM were biased with -40 V (yellow line), 0 V (red line) and +40 V (black line). By increasing the electrical bias to the QCM, the growth rate increases. The measurement of the deposition shows that the QCM can be electrically biased during the deposition of iron when electrons interact with the QCM.
To verify that a film had been deposited to the QCM and that it contained iron, a QCM crystal that was deposited at 15 W and 0 V substrate bias was subjected to cross-section SEM and EDS mapping, see Fig 6.2. The cross-section SEM analysis revealed that a film with a thickness of approximately 2 µm had been deposited, Figure 6.2. To verify that iron was present, an elemental composition mapping was done in EDS mode during the SEM analysis, which indeed revealed that iron had been deposited (Figure 6.2(b), red area). In Figure 6.2(b), the green area corresponds to the quartz crystal, while the blue area corresponds to platinum that was coated on the surface of the crystal to avoid charging effect during the SEM analysis.
Figure 6.2: A film that was deposited with iron on a QCM crystal at 15 W and 0 V bias was subjected to SEM -and XPS analysis. In (a), a cross-section obtained by SEM coupled with a focused ion beam (FIB) to obtain the thickness of the deposited film while (b) shows an ele-mental composition mapping (EDS mapping) of the film. In (a), a thickness of approximately 2 µm was obtained while (b) shows that the film indeed contained iron (red area). The green area represents the quartz crystal while the blue area is platinum, which was used to avoid charging effect of the thin film.
To verify the relative elemental composition of a deposited iron film, a QCM crystal that was deposited at 10 W plasma power and +20 V substrate bias was subjected to HR-XPS analysis. The HR-XPS spectrum (Figure 6.3) revealed that the deposited film indeed con-tained iron and that metallic state iron was observed. The relative elemental composition of the iron film was as follows: 41 at. % Fe, 14 at. % C, 42 at. % O, and 3 at. % N. The film had a high content of oxygen which was expected due to the long period (several weeks) post deposition that the crystal was subjected to air. Iron is well-known to be an oxyphilic element that readily reacts with oxygen. In addition, the carbon impurities are expected to originate from the cyclopentadienyl (Cp) ligands, which remained in the film while the nitrogen impurity is likely to originate from the (relatively) low vacuum that was used under the deposition (51 pa). Thus, the results reveals that the QCM can be modified to allow an electron flux to the surface of the QCM during iron deposition.
Figure 6.3: High-resolution X-ray photoelectron spectroscopy (HR-XPS) spectrum of a quartz crystal, where an iron film had been deposited on. The deposition parameters was 10 W plasma power and +20 V QCM bias.
Secondly, to answer the next question tied to the two-fold purpose of the thesis, if the modified QCM could be used to understand the deposition process where electrons are used as reducing agents, multiple of measurements were done at a plasma power of 5 W to +15 W and a QCM bias of 0V to +40 V whereas measurements are presented in Figure 6.4-6.6.
To monitor the effect that the plasma power had on the deposition, different characteris-tics, such as sudden decrease or increase of growth rate was obtained depending on whether 0 W, 5 W, 10 W or 15 W was used, where a representative graph can be seen in Figure 6.4,
where a constant QCM bias of +40 V was used. According to the acquired data, the highest growth rate was obtained when the lowest plasma power was used, and that the growth rate decreased with increased plasma power. In the previous study where electrons were used as a reducing agents, it showed that an optimum of growth rate seemed to exist in the process and that the growth rate did not scale with the plasma power.[8] This study cannot draw the conclusion that an optimum exists due to the limited range of plasma power that was investigated. However, this study suggests that the number of electrons participating in the reactions is highly important in the deposition of iron films when electrons are used as reducing agents. Note that 0 W was recorded to show that no deposition was made when the plasma was not ignited (0 W).
Figure 6.4: QCM results of film deposition of iron on a QCM crystal at 0 W, 5 W, 10 W and 15W at a fixed QCM bias of +10 V showing the growth characteristics. For 5 W, the highest growth rate could be monitored but a steep increase was seen after 40 s of the deposition. Depositing at 10 W exhibits a linear growth throughout the whole deposition. When depositing at 15 W, the lowest growth rate was obtained and generated a decrease of growth rate during the deposition. Note that no growth occurred when the plasma was not ignited (0 W).
The QCM was further employed to study the growth rate as a function of QCM bias in the range of 0 V to +40 V for depositions done at plasma power of 10 W (Figure 6.5) and 15 W (Figure 6.6). By depositing iron films on the QCM crystal with varying bias supplied to the QCM crystal, the trend seems to be that an increased film growth scales with the magnitude of the supplied bias at 10 W. By shifting the bias to +30 V and +40 V, a significant shift of film growth is achieved in comparison with a QCM bias of 0 V to 20 V can be observed. However, the linearity of the growth decreases when utilizing a higher bias supplied to the QCM. However, the quality and thickness of the iron films are still unknown and needs to be subjected to thickness analysis and elemental composition analysis.
