Direct Patterning of Oxides by Pulsed Laser Stencil Deposition

DIRECT PATTERNING OF OXIDES BY

PULSED LASER STENCIL DEPOSITION

Ph.D. committee:

Chairman and secretary:

Prof. Dr. W.J. Briels (University of Twente) Supervisor:

Prof. Dr. Ing. D.H.A. Blank (University of Twente) Assistant supervisor:

Dr. Ing. A.J.H.M. Rijnders (University of Twente) Members:

Prof. Dr. D.B. Geohegan (Oak Ridge National Laboratory) Prof. J. Brügger (École Polytechnique Fédérale de Lausanne) Dr. B. Noheda (Rijksuniversiteit Groningen)

Prof. Dr. K.J. Boller (University of Twente) Prof. Dr. Ir. J. Huskens (University of Twente)

Cover: Artist impression of a pulsed laser generated plasma propagating through the

apertures of a stencil. On the front side, the material moves through the stencil in a shockwave regime as occurs at high pressures, on the back side the material moves through the stencil as a non-interacting plasma at low pressures.

The work described in this thesis was carried out at the Inorganic Material Science department at the Faculty of Science and Technology and the Mesa+ Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, the Netherlands. This work is financially supported by the EC funded FP6 program NaPa: Emerging Nanopatterning Methods (NMP4-CT-2003-500120).

P.M. te Riele

Direct Patterning of Oxides by Pulsed Laser Stencil Deposition Ph.D. thesis University of Twente, Enschede, the Netherlands ISBN: 978-90-365-2709-5

Printed by Wöhrmann Print Service, Zutphen.

DIRECT PATTERNING OF OXIDES

BY PULSED LASER STENCIL DEPOSITION

PROEFSCHRIFT

ter verkrijging van

de graad van doctor aan de Universiteit Twente op gezag van de rector magnificus

prof. dr. W.H.M. Zijm

volgens besluit van het College voor Promoties in het openbaar te verdedigen

op donderdag 11 september 2008 om 13.15

door

Paul Marie te Riele

geboren op 19 maart 1980 te Amsterdam

Dit proefschrift is goedgekeurd door: Prof. Dr. Ing. D.H.A. Blank (promotor) en Dr. Ing. A.J.H.M. Rijnders (assistent-promotor)

Contents

Chapter 1: General introduction and outline ……… 9

1. General introduction ………... 10

2. Outline ……….. 11

3. References ……….. 13

Chapter 2: Stencil deposition and fabrication ……….. 15

1. Introduction ……….. 16

2. Stencil deposition fundamentals ………... 17

3. Stencil fabrication ……… 19

1) Lithography stencil fabrication ……… 20

2) Direct patterning of stencils ……… 22

4. Characterization techniques ……….. 23

1) Atomic force microscopy ………. 23

2) Scanning electron microscopy ……….. 23

3) X-ray diffraction ………. 23

4) (Ferro)electrical analysis ………. 24

5. Stencil deposition for patterning oxides ……….. 24

6. Concluding remarks ……… 26

7. References ………... 27

Chapter 3: Pulsed laser deposition and plasma dynamics ……… 29

1. Introduction ………... 30

2. Thin film growth by PLD ………. 31

1) Fundamentals of PLD ……….. 31

2) Experimental set-up ………. 32

3. Plasma expansion ……….. 33

1) Low pressure free expansion regime ……… 34

2) High pressure shockwave regime ………. 35

4. Concluding remarks ……… 37

Chapter 4: Broadening mechanisms ………. 41 1. Introduction ……….. 42 2. Clogging ………... 43 3. Broadening effects ………. 45 1) Geometrical broadening ………. 46 2) Shockwave broadening ……….. 47

3) Surface diffusion broadening ………. 54

4. Minimization of broadening effects ……….. 56

5. Concluding remarks ………... 57

6. References ……….. 59

Chapter 5: Piezoelectricity in PbZrxTi1-xO3 thin films ……….. 61

1. Introduction ………... 62

2. Piezoelectricity in Pb(ZrxTi1-x)O3 (PZT) ……… 63

3. Target and Sample Preparation ………... 65

1) Target preparation ………... 65

2) Sample preparation ………. 67

3) Substrate treatment ………. 68

4. Piezoresponse AFM analysis of PZT thin films ………. 71

5. Epitaxial SrRuO3 top electrodes ………... 75

6. Piezoresponse STM analysis of PZT thin films ………. 77

7. Concluding remarks ………... 80

8. References ………... 82

Chapter 6: Ferroelectricity in PbZrxTi1-xO3 thin films ……….. 85

1. Introduction ………... 86

2. Electrical characterization of ferroelectric materials ………. 87

1) Electric fields inside ferroelectric matter ………... 87

2) Modeling of electrical characterization of ferroelectrics ……. 89

3. Preparation of ferroelectric thin films and devices ……… 93

4. Fatigue properties in all-oxide systems ……….. 94

5. Functional and structural properties ……… 98

6. Substrate induced strain ……… 100

1) Epitaxial growth of PZT thin films ……….………. 100

7. Stencil structured devices ………. 104

8. Concluding remarks ……… 105

9. References ……….. 107

Appendix 1: Technical milestones in stencil deposition ……… 109

1. Introduction ………... 110

2. Stencil-substrate separation control ……… 111

3. Micro-heating ………... 112

4. In-situ stencil docking ………. 113

5. Dynamic stencil deposition ……… 114

6. References ……….. 117

Summary……… 119

Samenvatting ……… 123

Chapter 1

General introduction and outline

Chapter 1: General introduction and outline 1. General introduction

2. Outline 3. References

Abstract

Stencil patterning is a powerful, additive patterning technique to standard polymer lift-off lithography. This single step patterning technique has high potential for patterning materials on fragile surfaces, non planar surfaces for MEMS technology but also for patterning oxides. These oxides come in a wide variety of properties and can often be grown heteroepitaxially. However, resist lift-off techniques fail in patterning oxides and etching techniques often induce defects when patterning these materials. This work will describe how ferroelectric PbZrxTi1-xO3 is processed into devices by stencil patterning.

1.1 General introduction

Thin films are deposited by humans for thousands of years [1]. The art to bring these thin films down to nanometer thicknesses is mastered by mankind for centuries. Only in the past century, the physical aspects of bringing thin films into the nanometer regime became an important field of research. With the nanometer being the length scale that connects atomic dimensions to the bulk sizes, the new field of research was called nanotechnology [2]. New definitions of nanotechnology require also lateral dimensions in the nanometer range. Nowadays, thin films consist of complex heterostructures of a broad range of materials [3]. Many lithography techniques that are able to create these lateral nanometer dimensions often show difficulty in patterning these complex material systems [4,5,6].

The main issue for patterning complex material systems is that these materials are often deposited at elevated temperatures. Polymer resist lift-off techniques can not be applied under these conditions. This problem can be solved by depositing material very locally through tiny apertures called stencil patterning. This technique is applied to create patterns since the Middle Ages on cloth and paper and was used into the twentieth century for copying large amounts of paperwork. In this work, the stencils are made of silicon nitride, a ceramic material that can withstand high temperatures. The apertures in these stencils can be made below 100 nanometer. These stencils are placed on the substrate and material is deposited through the apertures. The rest of the material is collected on the stencil and lifted off after the deposition process. Oxides material systems come in a wide variety of structures and properties. The properties range from insulators to (super-) conductors and from ferroelectric to ferromagnetic materials. Many oxides have a perovskite structure with comparable unit cell dimensions. This enables the epitaxial and coherent growth of heterostructures and the combination of the properties in crystalline devices.

In contrast to the characterization of ferromagnetic thin films, ferroelectric materials used in a capacitor device have to be contacted by wiring in order to study their electrical properties. In most cases, the bottom electrode and ferroelectric medium can be grown epitaxially at elevated temperatures but the top electrodes consist of metal patterns created by polymer lithography lift-off. This results in loss of the epitaxial relation in consecutive growth of multilayer devices. Also, the application of metal electrodes on oxide ferroelectrics gives rise to significant fatigue. These effects can be avoided if the top electrodes also consist of oxide materials which are deposited by local deposition through stencils.

Piezoelectric characterization of epitaxial ferroelectric thin films is often performed without a top-electrode by a technique called piezoresponse atomic force microscopy

comparable to the film thickness. This results in highly non-uniform electric fields inside the film and can therefore not be used to obtain the piezoelectric constants of ferroelectric materials. Also these non-uniformity effects are avoided when a top electrode is deposited on the ferroelectric thin film.

Another effect of electrical analysis of ferroelectric thin films by contact techniques is inclusion of the contribution of free charge carriers in a polarization measurement. This effect in ferroelectric characterization is often underestimated and can lead to misinterpretation of polarization measurements.

1.2 Outline of this thesis

In this work, stencil deposition technology is used to create all-oxide devices. The ferroelectric oxide PbZrxTi1-xO3 is deposited onto various substrates and sandwiched

between SrRuO3 electrodes. Pulsed laser deposition (PLD) is a technique that can

deposit a broad range of high quality oxide thin films.

Chapter 2 will start with a general description of the stencil deposition process. The basic aspects of material transport trough and the deposition of material onto the stencils will be discussed. The fabrication process by lithography and direct milling of apertures by focused ion beam milling are summarized. The chapter ends with a compatibility check of stencil patterning technology with high temperatures for oxide growth.

The basic concepts of the PLD process that are of interest when combined with stencil deposition are described in chapter 3. Stencil deposition has been combined with thermal evaporation techniques but the deposition dynamics are different in PLD. This chapter will address the expansion mechanisms of the ablated material flux which might influence successful stencil patterning.

Chapter 4 describes the results when stencil patterning technology is combined with PLD. Materials are deposited under different conditions and the influence of the process parameters on the material transport through apertures is mapped. The last paragraph presents the smallest obtainable feature sizes by PLD oxide stencil deposition at elevated temperatures.

Chapter 5 focuses on the piezoelectric characterization of PbZrxTi1-xO3 (PZT) thin films

in 2 different geometries. Piezoresponse AFM imaging is applied in an effort to obtain the piezoelectric constants of PZT. The highly non-uniform electric field is calculated and shows that this technique is not suited to obtain any values from. Therefore, a technique called piezoresponse scanning tunneling microscopy is applied. A symmetric all-oxide piezoelectric thin film device can be characterized by depositing an oxide top electrode.

The ferroelectric characterization of symmetrical all-oxide PZT devices is presented in chapter 6. The chapter starts with an overview of the implications when analyzing ferroelectric materials in a parallel electrode geometry. Subsequent, an electrical

model is proposed which can isolate dielectric, ferroelectric and charge transport effects. With this model, these different effects are studied during learning and fatigue measurements and at different measurement frequencies. Also the influence of strain on the electric properties is studied. This chapter ends with the analysis of symmetrical all-oxide devices deposited in a single stencil patterning step.

This thesis ends with an appendix that contains a brief description of some of the technical milestones that were reached during this research. Also a system for dynamic stencil deposition is presented in this appendix.

1.3 References

[1] R. Dale Guthrie, The nature of paleolithic art, University of Chicago Press, [ISBN 0226311260]

[2] N. Taniguchi, Proc. Intl. Conf. Prod. (1974) part 2

[3] M. Huijben, G. Rijnders, D.H.A. Blank, S. Bals, S. Van Aert, J. Verbeeck, G. Van Tendeloo, A. Brinkman, H. Hilgenkamp, Nature Materials 5 (2006) p. 556 [4] C. Hu, G.P. Li, P. Liu, E Worley, J. White, R. Kjar, IEEE Elec. Device Letters,

16 (2) (1995), p. 61

[5] M.G. Kang, K.T. Kim, C.I. Kim, Thin Solid Films, 398 (2001) p. 448

[6] W.J. Lee, C.R. Cho, S.H. Kim, I.K. You, B.W. Kim, Jpn. J. Appl. Phys. 38 (2) (1999) p. 1428

Chapter 2

Stencil deposition and fabrication

1. Introduction

2. Stencil deposition fundamentals 3. Stencil fabrication

1) Lithography stencil fabrication 2) Direct patterning of stencils 4. Characterization techniques

1) Atomic force microscopy 2) Scanning electron microscopy 3) X-ray diffraction

4) (Ferro)electrical analysis 5. Stencil deposition for patterning oxides 6. Concluding remarks

7. References

Abstract

Stencil technology can be combined with various physical vapor deposition techniques like thermal and e-beam evaporation to structure simple materials like metals. The stencil patterning can be applied to a broad range of substrates. Most stencils consist of a silicon support structure and a silicon nitride membrane which contains the patterning apertures. These stencils are created by standard silicon processing like (D)UV photolithography and laser interference lithography to obtain apertures down to 200 nm. By direct patterning of the stencil membranes by focused ion beam milling, apertures of 60 nm can be obtained. All of these stencils are chemically stable up to high temperatures (>850ºC) in an oxidizing environment and can therefore be used in patterning materials at elevated temperatures. This is a main requirement in the patterned growth of epitaxial complex oxide films.

2.1 Introduction

Stencil deposition, also called shadow mask deposition, is a resistless, lift off patterning technique which can be applied in patterning a broad range of materials deposited by various techniques and can be applied to a broad range of surfaces [1,2,3,4]. The stenciling technique is based on blocking the material flux at certain areas of the substrate. Some distinguishable examples are substrates with non-planar surfaces, fragile surfaces or even moving substrates. The latter is called dynamic stenciling technology [5].

Stencil deposition can be combined with many deposition techniques like thermal and e-beam evaporation and is applied in thin film patterning for several decades. The first reported techniques that resemble current stencil patterning techniques consist of underetching of positive photoresist [6]. By angular deposition it was possible to create patterns smaller than achievable by photolithography. This technique was used to reduce the size of the features by a factor of 4-5 and resulted in 300 nm lines while 1.3 micrometer was 1977 state-of-the-art photolithography line width. After the mid 90’s a lot of new interest in stenciling technology emerged with the development of stencils created by MEMS technology. The technique is not developed with the intention to compete with technologies like photolithography but is applied in fields where photolithography is deficient. Several advantages of stencil technology can be found in literature like:

• Less laborious; it is a one processing step technique compared to several steps in standard photolithography

• Deposition on non-planar surfaces; stencils can be brought into contact with 3D objects

• High selectivity; the patterning of material does not depend on the selectivity in the etching procedure

• Fast and parallel replication of patterns; multiple use of stencil results in a cost effective patterning technique especially when combined with stencil cleaning procedures [7]

• Deposition on fragile surfaces; direct patterning by local deposition on fragile surfaces that can not withstand solvents like self assembled monolayers and organic crystals [2,3]

• Environmentally friendly; no chemical etchants are needed for the processing [8]

In this work, advantage is taken from the fact that stenciling technology is compatible with depositions at high temperatures. This enables the growth of oxides [9] by local deposition. Photolithography lift-off techniques can not be applied because of the high

deposition temperatures. Stencil deposition of oxides can also overcome the patterning problems encountered in conventional photolithography etching techniques like chemical inertness, low chemical selectivity or plasma damage by dry etching techniques [10,11,12,13]. Although stencil patterning is a complementary technique with several advantages over photolithography, the fabrication of stencils is mostly performed by photolithography techniques.

This chapter will focus on the current status of stencil patterning, stencil fabrication and limiting effects. Paragraph 2.2 will discuss the basic aspects material transport through confined geometries, limitations of stencil patterning and consequent design rules for stencil fabrication. Paragraph 2.3 describes the stencil fabrication processes by lithography techniques. Lithography processing is best suited for future application of stencil technology due to parallel fabrication and low processing cost. Paragraph 2.3.2 describes the obtained apertures created by focused ion beam milling. In focused ion beam milling, the structures are directly written in the stencil without any resist patterning steps. This is therefore called a direct patterning technique and is needed to obtain sub-100 nanometer apertures. Even state of the art photolithography equipment cannot produce extremely small apertures in stencils. Paragraph 2.4 gives a brief summary of the used characterization techniques in this work. This chapter ends with the most important aspects of combining complex oxide deposition and stencil patterning. High substrate temperatures (400-800ºC) are required in order to deposit oxides epitaxially. The compatibility of stencils and substrates with these deposition conditions is checked and shown in paragraph 2.5.

2.2 Stencil deposition fundamentals

For stencil deposition, the stencil is clamped on top of the substrate. The situation in which the stencil is in contact with the substrate is shown schematically in figure 2.1.a. Material is only deposited on the areas of the substrate that are not shadowed from the particle flux by the stencil. Isolated patterns of the deposited material are revealed after

a) b)

c)

Figure 2.1: a) Schematic representation of an ideal stencil deposition. The deposited material is

lifted of from the substrate on areas shadowed by the stencil, b) stencils are clogged by deposition of material directly inside the aperture or by surface migration towards the aperture and c) stress induced bending of the stencil.

removal of the stencil.

Without any separation between the stencil and substrate, the deposited material would attach the stencil to the substrate which is occasionally observed for the deposition of thick layers of material. During lift-off, the fragile parts from the stencil will remain on the substrate surface and the stencil is destroyed. Although the stencil is kept in physical contact with the substrate, a gap between stencil and substrate is unavoidable. The exact gap is not known but can be determined by analysis techniques. When the gap is measured by optical microscopy, a gap <10 µm can be ensured. This gap is caused by misalignment and curvature of the substrate or stencil. After proper alignment of the stencil and substrate the gap is assumed to be ~1 µm. A drawback of the presence of this gap is the loss of resolution due to blurring of the patterns called broadening. These broadening effects will be discussed in detail in chapter 4.

The operating principle of stencil deposition is based on the shadowing effect of areas that do not have to be covered by the thin film. As a result the stencil is covered by the deposited material as well. Surface mobility of the deposited material on the stencil and direct deposition of material inside the apertures, results in a decrease of aperture dimension called clogging. This effect is schematically drawn in figure 2.1.b and shows the effect of clogging on the deposited pattern dimensions. The presence of clogging has an influence on the shape of the resulting patterns and the re-usability of the stencils. It is kept as a general rule that the amount of aperture reduction by clogging is equal to the thickness of the deposited film [14]. The exact amount of clogging depends on the aspect ratio of the deposition set-up and the kinetic energy of the deposited species. The problem of re-usability can be solved by chemically cleaning the stencils [15,16].

Another effect of the material being deposited on top of the stencil is the introduction of stress due to differences in thermal expansion coefficient, lattice mismatch between stencil and film and recrystallization processes. In the deposition of metals at room temperature, stress is mainly induced by temperature differences between the deposited material and the stencil. As the material cools down on the stencil, the differences in thermal expansion coefficient between the stencil and the deposited material induce stress. The induced strain during deposition can result in deformation of the stencil surface which cause an increase or decrease of the gap between stencil and substrate. This process is drawn in figure 2.1.c. The deformation can be reduced by corrugation of stencils [14,17] or by careful selection of stable structures in the stencil. Corrugation can be applied by locally reinforcing the fragile parts of the stencil. For stability reasons, the use of free standing structures inside the stencil should be avoided.

2.3 Stencil fabrication

The used stencils in this work are created by silicon processing techniques and consist of two parts; a silicon wafer and a silicon nitride membrane. The wafer is used as the support for mechanical stability; the membrane contains the well-defined small apertures. High material flux through the apertures can only be obtained if the membrane is thin. The aspect ratio between the aperture dimension and membrane thickness is required to be at least 1. However, the mechanical stability is negatively affected by the high aspect ratio with small apertures. Another effect is that the thickness has a significant influence on how well-defined the apertures can be after processing. The tapered sidewalls and loss of resolution are common for structures created by reactive ion etching (RIE, figure 2.2.a step 3) [18]. The mechanical support and membrane have to be flat on a micrometer scale to prevent gap variations over the stencil area, although in this work, this is a minor issue since only small 10 mm stencils are used. The apertures in the membrane can be created by lithography or direct writing techniques. The minimum size of the features in the membrane for resist techniques is defined by the structures dimensions in the resist although some techniques, like oxygen plasma etching [19], have shown aperture dimension reduction. Direct patterning techniques like focused ion beam (FIB) writing can be used to structure unpatterned membranes without any etching steps and without the consequently loss of resolution. The details of the stencil fabrication will be discussed in paragraph 2.3. The patterning technique is selected on the desired dimensions. The stencil fabrication process by photolithography and silicon processing is shown schematically in figure 2.2.a. This figure shows the different process steps needed to create a stencil. A [001] oriented silicon wafer is provided on both sides with a SiN thin film deposited by low pressure chemical vapor deposition (LPCVD). The advantage of depositing SiN by LPCVD is that the deposited film has low internal stress. Stress

20 40 60 80 100 0 1 2 3 4 2θ (°) lo g In te n s it y ( a .u .) Si b) a)

Figure 2.2: a) Process steps in stencil fabrication; 1. 0.2-1 µm low-pressure low-stress chemical

vapor deposition of silicon nitride on both sides of the silicon wafer, 2. resist patterning by photolithography, 3. transfer of resist pattern into SiN by reactive ion etching, 4. Si removal by KOH wet etch. b) shows θ-2θ scan of stencil with SiN on Si support, only Si peaks of the support

inside the SiN membrane resulting from the deposition process results in bending of the stencil or even rupture. In a θ-2θ scan shown in figure 2.2.b, no peaks can be identified of the SiN what demonstrates that the film is amorphous or nanocrystalline. The thickness of the SiN layer used for the membranes ranges from 200 nm to 1 µm. The 200 nm thick membranes have shown to be extremely fragile and therefore the 500 nm thick membranes are used standard in this work unless reported otherwise. The resist in step 2 is structured on the front side to define the stencil membrane apertures, on the back side it is used to define the stencil silicon support structures. The resist can be structured by several techniques but it is typically done by standard UV photolithography [20], DUV lithography [4], e-beam lithography [21] or interference lithography [22]. Patterns created by (D)UV and e-beam lithography can be of any desirable design while the interference lithography can only be used to obtain regular structures of circles, ellipses and stripes. On both sides, the pattern in the resist is transferred into the SiN by RIE. Only where the silicon support is etched in the wet etching step, the front side SiN layer becomes the membrane. The presence of the apertures in the stencil can also severely decrease the mechanical stability of the membrane. In order to prevent rupture during wet etching the Si by KOH, the RIE etching step can be performed after removing the Si.

The KOH wet etching step etches the Si away along the (111) planes which results in 53 degrees angle side walls of the support (step 4). These inclined sidewalls have to be taken into account in the design of the stencils. Typical dimensions of the membranes spanning between the silicon support are below 1 mm. The aspect ratio of the membrane (length/thickness ~2000) exclaims mechanical instability for larger membranes. Therefore, smart design of the silicon support or even corrugated membranes [23] are used to bring the stencil technique to a full wafer scale.

2.3.1 Photolithography stencil fabrication

In this work, stencils were used which were created by photolithography and direct writing techniques. The specifications of the stencils created by these techniques will be briefly discussed.

Photolithography stencils

The used stencils in this work are stencils mostly created by standard UV photolithography. These stencils contain a variety of apertures. The basic layout used in this work consists of 10x10mm2 stencil chip containing 4 1x1 mm2 membranes. The membranes are 500 nm thick. The apertures in this stencil have square and circular shapes. The square and circular apertures all have dimensions of 100, 50, 25, 10 and 5 µm. Although the resolution of the photolithography system is better, 5 µm is chosen as the smallest aperture size. Loss of resolution during the RIE pattern transfer results

Figure 2.3: a) stencil layout of the Mesa+ standard cleanroom stencils, each stencil consists of a

Si 10 mm square stencil support with 4 1 mm square membranes. The square and circular aperture widths in the membrane are 100, 50, 25, 10 and 5 µm. b) SEM image of a stencil created by LIL. The apertures are in a regular array with a 350 nm aperture size.

dimensions. The stencil layout is shown in figure 2.3.a and will be referred to as the standard stencil in the remainder of this thesis.

For stencils created by deep ultra violet (DUV) photolithography [4], the membrane thickness is 500 nm. The process is comparable to the stencils production described above except for the illumination. Although the feature sizes in the resist were obtained down to 100 nm, the etching process of the SiN resulted in minimal feature sizes of 200 nm.

Laser Interference Lithography Stencils

Another photolithography technique to structure the membranes of a stencil is Laser Interference Lithography (LIL) [24,25]. An interference pattern created by a so called Lloyd’s mirror configuration is projected onto a photoresist layer [26]. This interference pattern results in straight parallel lines in the photoresist. By two successive illumination steps and by rotating the sample, round or elliptical structures can be obtained. The major advantage of this technique is that by a simple interference set-up very small features can be obtained, even down to 100 nm [27]. The disadvantage is that only regular arrays of ellipsoid or circular apertures can be created. An example of 350 nm circular apertures is shown in the SEM image in figure 2.3.b. This technique is used in the creation of selective membranes for filtration purposes. Two types of LIL stencils were obtained commercially from Aquamarijn BV, the Netherlands.

Other resist techniques comprise of extreme UV, x-ray or e-beam illumination of resist and can result in sub 10 nm structures inside the resist layer. Although the resist patterns can be sub-100 nm, the resulting apertures in the stencil are often up to an order of magnitude larger. This loss of resolution occurs during pattern transfer into the membrane which is more than an order of magnitude thicker than the structure width. Direct patterning techniques are essential in order to obtain sub 100 nm structures inside stencil membranes.

b)

a)

Figure 2.4: a) Schematic drawing of the focused ion beam milling process by a 150 nm radius

ion beam. SEM image of 8 FIB milled holes in the unprocessed area of a LIL stencil 1 µm SiN membrane. The inset shows 100.000x magnification of a membrane with a 60 nm wide aperture.

2.3.2 Direct patterning of stencils

FIB techniques can directly write structures in the SiN membrane of a stencil. In the illing process. The incoming ion beam has

he FIB apertures are positioned on the unprocessed used FIB system, Ga ions are used in the m

several effects on surface [28]. The beam removes atoms from the substrate, causes electron emission, destroys or creates chemical bonds and generates heat inside the substrate. The first effect is used for the milling process and the intensity of the emitted electrons can be used to image the surface or to control the milling process. To control the milling process, the intensity of emitted electrons is monitored. A sudden drop in this intensity indicates that the ion beam pierces through the membrane and this drop is used to stop the milling process. The creation of chemical bonds, ion implantation and heat effects do not play any role in the patterning of stencils. In the used system, the beam radius is ~150 nm but smaller features can be obtained when using the Gaussian shape of the beam profile with a 50 nm full width at half maximum. The direct patterning method is schematically drawn in figure 2.4.a. Typically 4 to 10 seconds is needed to mill the beam through a 500nm thick membrane of the stencil with a 41 pA beam current and 0.04 mm3/C etch rate. An array of 9 holes inside the unpatterned area of a LIL stencil is shown in figure 2.4.b. The inset shows an aperture with a 60 nm width and a 90 nm height.

Two types of stencils have been patterned, the 500 nm thick standard stencil and the 1 µm thick commercial LIL stencil. T

membrane areas. Deposition of gold prior to the milling process is crucial for two reasons; first the membrane consists of insulating SiN where charging effects results in changing brightness when imaging. Secondly the trapped charge on the membrane induced by the milling process causes bending of the membrane due to electrostatic interaction with the grounded sample plate. The induced movement results in large and deformed apertures. In both cases, the charge can be removed from the membrane by depositing a conductive layer of gold. The gold can be removed by a potassium

iodide-iodine solution to prevent clogging of the apertures by the mobile gold when heating up the stencil during high temperature depositions.

2.4 Characterization techniques

Stencils, thin films and patterns created by sten range of characterization techniques in

cil deposition are characterized by a order to check surface morphology,

scopy (AFM)

n of thin films and acquiring cross-eposition. This characterization was

icroscopy (SEM)

ncil apertures but also for f the deposited material. The

For the characterization of the crystal structure of the deposited material, X-ray diffraction is performed on the deposited material and substrates. Different types of XRD equipment were used, all operated at CuKα1 wavelength (1.54056 Å). For the

crystallinity, composition and functionality of the deposited material. The specifications and commonly used settings of the most widely used techniques in this work are enumerated in this paragraph.

2.4.1 Atomic Force Micro

AFM was used for the morphological characterizatio sectional data of structures created by stencil d

done by a Nanoscope IV (Veeco Instruments, Digital Instruments, USA) and is capable of contact and tapping mode imaging as well as room temperature scanning tunneling microscopy (STM) imaging. Contact mode imaging was performed with a stiffness of 0.2 N/m and a resonance frequency of 30 kHz. For tapping mode imaging, the stiffness was 40 N/m with a resonance frequency of 300 kHz. The same AFM system was used for piezoresponse imaging and piezo-STM characterization (more detailed discussion in paragraph 5.6) of piezoelectric thin films and devices. For this type of characterization, the system was equipped with a signal access module (SAM) III. Via this module, direct access to the output signals is obtained for lock-in amplification but also the tip or sample bias can be controlled externally. Piezoresponse imaging was performed by PtCr coated tips with various spring constants. The spring constant depends on the mode of operation (writing or reading, paragraph 5.4). Also fully conductive Sb doped Si tips were used which are better wear resistant. STM imaging was done with PtIr wire cut tips.

2.4.2 Scanning Electron M

SEM imaging is mainly used for the characterization of ste the characterization of the structure and morphology o

imaging was done on a JEOL-5610 (JEOL Ltd., Tokyo, Japan) SEM equipped with a secondary electron detector and a backscatter detector. The microscope is operated between 0.5 and 30 kV with a best resulting lateral resolution of 3.5 nm.

characterization of powders, like grinded targets and target components, a

space vectors. For thin film and single crystal

, Aachen, Germany). The basic system consists of a TF analyzer 2000 operating system and 2 measuring heads, both for FE hysteresis investigation. The small devices, with low switching

rd to structure by wet chemical etching purpose, two-circle, X’pert APD (Panalytical, Almelo, the Netherlands) diffractometer with a sample changer is used. The sample is spun with 1 Hz during measurement to increase intensity. The irradiated length is fixed at 12 mm. An X’pert MPD (Panalytical, Almelo, the Netherlands) two-circle machine is used for the characterization of epitaxial single crystalline thin films and powders at elevated temperatures. This machine is capable of measuring epitaxial single crystalline thin films and substrates due to the variable offset between ω and 2θ. This unlocking of both axes enables the mapping of the reciprocal space for the out-of-plane reflections. This diffractometer is also equipped with a programmable Anton Paar high temperature stage for structural analysis at elevated temperatures, up to 1600 degrees. The irradiated length is variable and programmable from 0.5 and 20 mm. Both diffractometers are operated at 50 kV and 35 mA and have an angular resolution of 0.001º. Monochromaters are installed in the secondary beam path.

For full structural analysis of thin films and single crystals, a CAD4 (Bruker-Enraf Nonius, Delft, the Netherlands) four-circle diffractometer is used. The tube is operated at 40 kV and 20 mA and has a beam diameter of 1 mm. It is mainly used for obtaining reciprocal space maps along reciprocal

measurements, a Bruker D8 Discover (Bruker, Karlsruhe, Germany) is also used. This machine has higher beam quality and can also be used in combination with a 1D detector (VANTEC-1) for quick reciprocal space maps. The machine is operated at 40 kV and 40 mA. Both single crystal diffractometers have a monochromated primary beam.

2.4.4 (Ferro-) Electrical Characterization

For (ferro-)electrical characterization, ferroelectric test equipment of Aixacct is used (Aixacct

TFA-FE 409-2LC can be used for the analyses of

currents and high parasitic capacitance. This head contains a capacitance compensation circuit to level out high capacitances in the device and can measure up to 2 kHz. The TFA-FE 412-2HSIN head is used for measurements at high frequencies up to 10 kHz. The built in leakage current compensation in the system is not used in this research. The system is connected to the samples via a Karl Süss MP4 manual probe station with a tip resolution of 20-50 µm.

2.5 Stencil deposition for patterning oxides

For patterning materials, a lift-off technique is often wanted because unwanted effects of etching can be avoided. Some oxides are ha

a) b) c)

Figure 2.5: SEM images of the stencil a) before and b) after heat treatment of 800ºC for 2 hours

in vacuum, c) shows an AFM image of the substrate after heat treatment in contact with a stencil.

due to low selectivity and it is known that dry physical etching methods which show e to induced defects

TO) [32] substrate at

e mismatch induced high selectivity often alter the properties of the material du

[10,11,12,13,29]. These defects can sometimes be removed by annealing procedures but it is more beneficiary to structure these materials by local deposition through stencils [9]. Another used technique to pattern oxides by stencil deposition consists of low temperature metal deposition followed by an annealing step to create the correct oxide phase [30]. This technique does not work for all oxides and does not result in epitaxially or coherently grown patterns. Epitaxial patterns are shown by self-patterning methods but the patterns are located at random locations [31].

In order to stabilize the correct oxide phases during deposition and to obtain an epitaxial relation with the substrate, high substrate temperatures are needed. The compatibility of stencil technology and elevated substrate temperatures was tested by placing a LIL stencil in front of a single terminated SrTiO3 (S

800ºC for 2 hours inside a vacuum system. SEM images of the stencil and AFM images of the substrate are shown in figure 2.5. The SEM analysis shows no differences in the dimensions or shape of the apertures when comparing the stencils before and after heat treatment. In order to check whether any material is transferred from the stencil to the substrate at elevated temperatures, an atomically smooth STO surface is imaged by AFM after heat treatment. No material transport between stencil and STO substrate observed. This analysis cannot resolve any chemical changes of the surface. Changes in the crystal structure of the SiN membrane and consequent induced strain was checked by XRD. After 3 hours at 850ºC in ambient atmosphere, XRD measurements could not show any significant changes in the stencil which could indicate sintering of the SiN or the formation of any related oxide.

A drawback of the use of stencils at high temperatures is the induced strain due to recrystallization of the deposited material on the stencil and differences in thermal expansion coefficient. These effects lead to more limited re-use of stencils compared to room temperature depositions. At elevated temperatures, lattic

strain might play a role but will be of minor influence since the membrane is almost completely amorphous. The induced stress causes the membrane to bend which can induce locally increased gaps and more broadening. The deposited oxides can also prevent the chemical cleaning procedures which are typically applied to metal depositions [16] due to the chemical stability. Nevertheless, the experiments show that stencils are well suited for patterning materials at elevated temperatures in the vicinity of oxide substrates.

2.6 Concluding remarks

Stencil deposition has proven to be an effective technique for patterning inorganic aterials deposited by various physical vapor deposition technique. It has shown to be

surfaces, non-planar substrates or in materials The technique is already used in combination with

ransfer from resist to the SiN membrane. Smaller apertures, down to 60 nm,

ped. A second effect is bending of the

cal deposition through stencil apertures.

m

unique in patterning thin films on fragile that can not be etched otherwise.

many physical vapor deposition techniques like thermal evaporation and e-beam evaporation but also in combination with other soft-lithography chemical patterning routes.

In this work, stencils were created by (D)UV lithography, laser interference lithography or by direct patterning of blind stencils. The stencils created by lithography have shown dimensions down to 200 nm and the minimal dimensions are mostly limited by the pattern t

were created by direct patterning of the membrane by focused ion beam milling. This milling can result in sub-100nm structures even with beam dimensions of 150nm by making use of the Gaussian beam profile.

Several limitations are present during stencil deposition. First of all, the decrease of the aperture dimensions during deposition called clogging. This results in smaller structures and ultimately in completely clogged apertures. This can not be prevented but cleaning processes are being develo

membrane due to induced stress during deposition. This can lead to larger apertures or even rupture of the membrane. By placing reinforced structures or by selecting proper aperture geometries, this effect can be avoided. Another effect is broadening, which is a consequence of the presence of a gap. This gap can not be avoided and is even needed to prevent the stencil from being attached to the substrate surface. Clogging and broadening are discussed in detail in chapter 4.

Preliminary compatibility tests have shown that stencil deposition can also be used under the conditions needed for the growth of oxides on substrates at elevated temperatures. The stencils can withstand temperatures up to 850ºC under oxidizing conditions which enables structured growth of oxides by lo

2.7 References

[1] F. Vroegindeweij, E.A. Speets, J.A.J. Steen, J. Brugger and D.H.A. Blank, Appl. Phys. A 79 (2004) p. 743

[2] E.A. Speets, B.J, Ravoo, F.J.G Roesthuis, F. Vroegindeweij, D.H.A Blank and D.N. Reinhoudt, Nanoletters 4-5 (2004) p. 841

[3] E.A. Speets, P.M. te Riele, M.A.F. van den Boogaart, L.M. Doeswijk, B.J. Ravoo, A.J.H.M. Rijnders, J. Burger, D. Reinhoudt, Dave H.A. Blank, Adv. Funct. Mat. 16 (10) p. 1337 (cover)

[4] M.A.F. van den Boogaart, G.M. Kim, R. Pellens, J.P. van den Heuvel, J. Brugger, J. Vac. Sci. Technol. 22 (6) (2004) p. 3174

[5] S. Egger, A. Ilie, Y. Fu, J. Chongsathien, D.J. Kang, M.E. Welland, Nanoletters

5 (1) (2005) p. 15

[6] G.J. Dolan, Appl. Phys. Lett. 31 (5) (1977) p. 337

[7] O. Vazques Mena, M.A.F van den Boogaart, J. Brugger, Trans. & Eurosens. Conf. series 1B6.4

[8] M. Kolbel, W. Tjerkstra, G. Kim, J. Brugger, C.J.M. van Rijn, W. Nijdam, J. Huskens and D.N. Reinhoudt, Adv. Funct. Mater. 13-3 (2003) p. 219

[9] P.M. te Riele, J.A. Janssens, A.J.H.M. Rijnders, Dave H.A. Blank, J. of Phys.: Conf. series 59 (2007) p. 404

[10] C. Hu, G.P. Li, P. Liu, E Worley, J. White, R. Kjar, IEEE Elec. Device Letters,

16 (2) (1995) p. 61

[11] M.G. Kang, K.T. Kim, C.I. Kim, Thin Solid Films, 398 (2001) p. 448

[12] W.J. Lee, C.R. Cho, S.H. Kim, I.K. You, B.W. Kim, Jpn. J. Appl. Phys. 38 (2) (1999) p. 1428

[13] M.G. Kang, K. Tae, C.I. Kim, Thin Solid Films, 435 (1-2) (2003) p. 222

[14] M. Lishchynska, V. Bourenkov, M.A.F. van den Boogaart, L. Doeswijk, J. Brugger, J.C. Greer, Microelectr. Eng. 84 (2007) p. 42

[15] X.M. Yan, S.M. Contreras, M.M. Koebel, J.A. Liddle and G.A. Somorjai, Nano Lett. 5 (6), (2005), p. 1129

[16] O. Vazques Mena, M.A.F van den Boogaart, J. Brugger, Trans. & Eurosens. Conf. series 1B6.4

[17] M. van den Boogaart, M. Lishchynska, L. Doeswijk, J. Greer, J. Brugger, Sens. And act. 130-131 (2006) p. 568

[18] R.F. Fiueroa, S. Spiesshoefer, S.L. Burkett, L. Schaper, J. Vac. Sci. Technol.

22 (6) (2004) p. 3174

[19] M.M. Deshmuk, D.C. Ralph, M. Thomas and J. Silcox, Appl. Phys. Lett. 75 (11) (1999) p. 1631

[20] A. Tixier, Y. Mita, J.P. Gouy and H. Fujita, J. Micromech. Eng. 10 (2000) p. 157

[21] X.M. Yan, S.M. Contreras, M.M. Koebel, J.A. Liddle and G.A. Somorjai, Nano

[22] nd S. Kuiper, Nanotechnology 9 (1998), p. 343

[25] thuis, F. Vroegindeweij, D.H.A Blank and

[26] s, 10, ISBN:

0-444-[27] an Rijn, W. Nijdam, M.C. Elwenspoek, J. of Membrane Sci.

[30] oogaart

[31] Scholz, S. Bhattacharyya, D. Hesse, M. Alexe,

[32] pman, A.J.H.M. Rijnders, D.H.A Blank and H. Rogalla,

Lett. 5 (6), (2005) p. 1129 C.J.M. van Rijn, G.J. Veldhuis a

[23] M. van den Boogaart, M. Lishchynska, L. Doeswijk, J. Greer, J. Brugger, Sens. and act. 130-131 (2006) p. 568

[24] J. Brugger, J.W. Berenschot, S. Kuiper, W. Nijdam, B. Otter and M. Elwenspoek, Microelectron. Eng. 53 (2000) p. 403

E.A. Speets, B.J, Ravoo, F.J.G Roes

D.N. Reinhoudt, Nanoletters 4-5 (2004) p. 841

C.J.M. van Rijn, Membrane Science and Technology Serie 51489-9

S. Kuiper, C.J.M. v

150 (1) (1998) p. 1

[28] A.G.P. Troeman, Thesis, University of Twente, (2007) [ISBN 978-90-365-2572-5]

[29] J.K. Lee, T.Y. Kim, I. Chung, S.B. Desu, Appl. Phys. Lett. 75 (3) (1999) p. 334 C.V. Cojocaru, C. Harnagea, F. Rosei, A. Pignolet, M.A.F. van den B and J. Brugger, Appl. Phys. Lett. 86, (2005) p. 183107

I. Szafraniak, C. Harnagea, R.

Appl. Phys. Lett. 83 (11) (2003) p. 2211 G. Koster, B.L. Kro

Chapter 3

Pulsed laser deposition and plasma

dynamics

Chapter 3: Pulsed laser deposition and plasma dynamics 1. Introduction

2. Thin film growth by PLD 1) Fundamentals of PLD 2) Experimental set-up 3. Plasma expansion

1) Low pressure free expansion regime 2) High pressure shockwave regime 4. Concluding remarks

5. References

Abstract

Pulsed laser deposition is a powerful tool for growing thin films of complex materials on a broad range of substrates. When this technique is combined with stencil patterning techniques, new research evolves and application of complex materials is brought a step closer. Pulsed laser deposition is a deposition technique with a complex interaction between the ablated species which make the technique essentially different from many other deposition techniques. In order to combine stencil patterning technology successfully with pulsed laser deposition, observation and control of the material flux is needed. Especially in the field of oxides, where materials are often deposited at high process pressures, the formation of a shockwave can play a limiting role in stencil patterning.

3.1 Introduction

Stencil patterning technology has been applied to several deposition techniques. Most of these techniques are thermal evaporation techniques and are accompanied by low process pressures and low deposition rates. When using thermal evaporation techniques, the library of possible materials that can be deposited is fairly limited. For over two decades, pulsed laser deposition (PLD) developed into a powerful technique for growing thin films of complex materials. Although evaporation of materials by high power lasers originates from the mid sixties, it took until the eighties with the discovery of HTc superconductors for the technique to be applied extensively in research. PLD was able to grow high quality thin films of complex oxide materials like YBa2Cu3Ox [1,2]. Nowadays, PLD is used extensively in research for rapid

prototyping and even niche market production. High supersaturation ratios [3,4,5], high initial kinetic energy of the evaporated material and independent adjustment of deposition parameters make this technique unique.

By combining stencil patterning technology with PLD, many new materials can be patterned that are tough to pattern by other techniques. Especially in the field of oxides, the field where PLD excels other techniques, the combination could enable new research and incorporation of oxides in applications. The evaporation flux behavior and growth dynamics differ heavily from other thermal evaporation techniques and have their influence on stencil PLD. Predominantly, the pulsed character and the high amount of ablated material cause high density plasmas with a lot of interaction between the ablated species. The amount of interaction is also dependent on the type and pressure of the used background gas. This interaction causes the material flux to deviate from traveling in straight lines. Many developed models for stencil patterning are based on interaction-free material fluxes and cannot hold for PLD. Therefore, the fundamentals will be discussed extensively in this chapter.

First the different stages of the PLD process will be described together with a description of the experimental set-up. The most important effects when combining stencil technology with PLD have their roots in the plasma dynamics. These expansion dynamics in different regimes will be explained in paragraph 3.3. This paragraph will focus on the formation of a shockwave, which is typical for PLD, under different process conditions.

3.2 Thin film growth by pulsed laser deposition

3.2.1 Fundamentals of pulsed laser deposition

Pulsed Laser Deposition (PLD) is a thin film deposition technique where a pulsed laser beam is focused onto a pellet of material of which a thin film is desired. This pellet of source material, called the target, is heated locally by the high energy density of the laser light. The energy density on the target surface causes extreme heating of a small volume. The amount of heating depends on the absorption depth of the light, which is a function of the laser wavelength and the used target material. This local heating, results in the creation of a plasma without any melt phase which is called ablation. The amount of heat transported into the target is minimized due to the instantaneous character of the pulsed laser. The thermal diffusion length is defined as [6,7,8]:

l p th c l τ ρ κ 2 = (3.1)

with κ being the thermal diffusion coefficient, ρ the density of the target, cp the specific

heat and τl the convection time. This time for convection is often shorter than the laser

pulse, resulting in hardly any heating of the target. This holds particularly for oxides which often have a low thermal diffusion coefficient.

The evaporation process takes place in ~1 nanosecond. During this evaporation time a small volume of target material with a depth of the laser wavelength is evaporated. Typically this is only a fraction of the pulse duration of the lasers often used in PLD. This small evaporated volume is then heated by the remainder of the laser pulse resulting in a plasma of high temperature and pressure [5]. During the evaporation and heating process, hardly any expansion takes place. Expansion is driven by the high pressure in this small volume of the initial plasma. The process of plasma expansion dynamics is discussed in detail in section 3.3. The evaporated material is collected on the substrate material opposite to the target where a thin film is formed. This substrate material is mounted onto a heater for the depositions of thin films at elevated temperatures. A schematic drawing of a PLD set-up is given in figure 3.1.

PLD has several advantages over other physical vapor deposition techniques. Some of the advantages are;

• Stoichiometric ablation of multicomponent targets which enables stoichiometric transfer of material from target to substrate.

• High instant deposition rates (~µm/sec)

• The deposition parameters are almost independently adjustable over a wide range. This makes PLD a suited technique for the deposition of a wide range of materials.

• The energy source is placed outside the deposition system which enables adjustment of energy delivery without any constraints of the evaporation source. Also the evaporation without any melt phase gives room for experimental design freedom.

• No transfer of excessive process heat except for the energy transferred by the kinetic and thermal energy of the arriving species. This kinetic energy can even be controlled by pressure which enables the deposition on fragile and polymer substrates. High kinetic energies can be used for increased surface mobility or improved adhesion.

The high deposition rate in combination with the high pressure background gas can result in high supersaturation ratios. This high supersaturation together with stoichiometric transfer of material is ideal for the growth of complex oxide thin films.

3.2.2 Experimental set-up

The experimental set-up consists of a KrF excimer laser with a wavelength of 248 nm and a vacuum system with a base pressure of 10-7 mbar. The laser (Lambda Physic Compex 205) has a pulse duration of 25 ns, a maximum pulse energy of 700 mJ and a repetition rate between 1 and 50 Hz. A selection of the beam profile is made by a mask for better homogeneity and control over the plasma expansion dynamics. The laser beam is focused by a 453 mm focal length lens onto the rotating target. The beam hits the target under a 45º angle on a spotsize of 1 to 5 mm2 depending on the desired fluence and plasma expansion dynamics. The loss of power by the reflection on the mirrors and absorption in the UV window of the chamber is in the order of 20%. Typically, laser pulse energies after the mirrors, lens and window are in the order of only 100 mJ.

During deposition, the pressure inside the system can be controlled in the range of 10-5 and 10 mbar. As a background gas, oxygen, argon and nitrogen can be selected and their flow can be independently controlled by 40 mbar*l/min mass flow controllers. In

a) b)

the 10-3 to 1 mbar region, pressure is regulated by a control valve. Lower pressures are controlled by the inlet flow of the background gas (10-5-10-3 mbar), higher pressures are controlled reducing the pumping speed through a needle valve (1-10 mbar). For the deposition of oxides, oxygen pressures in the range of 10-1 are used. The target-substrate separation can be controlled between 35 and 90 mm. The sample is mounted onto a heater which can be PID controlled to temperatures from room temperature to 950ºC.

3.3 Plasma expansion

For metals, the absorption of the laser light takes place by stimulation of the free electrons in the metal after which the excited electrons interact with the atoms of the target and transfer their energy. In insulating materials, the light is absorbed directly by interband transitions. Both energy absorption processes take place with a heating rate in the order of 1012 K/sec which means that the material is evaporated within the first nanosecond. Most of the heating takes place when the material is in the vapor phase. The typical density of the initial vapour (1021cm-3) is dense enough to shield the target from the laser beam and the remainder of the laser pulse (~20 nsec) is used to heat up the vapor further to a plasma cloud of 10.000 to 20.000 K (figure 3.2). The dimensions of this initial plasma perpendicular to the target surface cloud are still of the order of micrometers and the accompanying pressure is 20-100 bar. Due to this high pressure, the plasma starts expanding in the direction of the highest pressure gradient. Since the pressure gradient is defined as:

(3.2) d

P P= ∆

with P being the pressure in the plasma and d the dimensions, the gradient is the largest in the z-direction perpendicular to the surface of the target. The z-direction pressure gradient is 3 orders of magnitude higher than the x- and y-direction gradient (y direction is pointing out of the paper surface). Due to these differences in pressure gradient, a highly forwarded particle flux is developed. As a result, the plasma is shaped into the typical full Maxwellian PLD ellipsoid plasma [9]. The process is sketched in figure 3.2.a.

A particle flux generated from a target is often described by a cosn(θ) distribution, where n is depending on the technique, source dimensions and the deposition parameters.

The angle θ is defined from the center of the evaporation spot and is zero around the spot normal (figure 3.2.a). The parameter n reduces to zero for material thermally evaporating from a point source. For PLD, n can reach values well above 10 and even values above 50 have been reported [10] measured by intensified gated CCD imaging or by measurement of the deposition rate as a function of θ. It is important to note that

1

)

(

=

n

σ

A

Figure 3.2: a) schematic representation of the first stages in the PLD ablation process. The

ablated material is heated into the initial plasma. The pressure gradient transforms this plasma into ellipsoid shape, b) calculated angular distribution of the particle flux over the angle θ for different values of n in a cosn(θ) distribution.

a) b) -1000 -50 0 50 100 0.2 0.4 0.6 0.8 1 θ (degrees) N o rm al iz ed depos iti on r at e n=50 n=1 n=0 n=10 n=4 n=2

this angular distribution and values for n in PLD are often obtained by measurements at the substrate position. This does not imply that the flux distribution function has the ablation spot as its origin but the distribution function is developing over time and distance. The value of n is not constant over the distance from the ablation spot. When the plasma is still in its initial phase, it has a homogeneous expansion in all directions (n≈0) and develops its directionality in the first 20 mm.

Figure 3.2.b shows the angular distribution of the material flux for different values of n. The parameter n is not only varying over the distance from the target as mentioned above but is also highly dependent on the deposition parameters. First of all, the spot size and spot geometry define the x and y dimensions of the initial plasma and consequently the pressure gradient and expansion in these directions. As a result, plasmas of small spot sizes are smaller and wider, indicating a lower value of n. The biggest influence on n is caused by the pressure of the gas where the plasma is expanding in. The next two sections will describe the influence of pressure on the plasma dynamics.

3.3.1 Low pressure expansion regime: Free expansion

In the used system the pressure can be controlled over 8 orders of magnitude between 10-7 and 10 mbar. The mean free path, the average distance over which a molecule or particle can travel between collisions, is directly related to the pressure via the relation:

(3.3) where l is the mean free path, n is the number density and σ is the cross sectional area for collision which is not only dependent on the type of molecules but also on the

Table 3.1: Number density and mean free path at different process pressures.

speed of the gas molecules. Table 3.1 shows the mean free path for different pressures [11,12] which can be obtained in the system during deposition.

For PLD, two pressure regimes have to be distinguished; one where the mean free path is much larger than the target-substrate separation and the regime where it is smaller than the target-substrate separation. In these regimes, the plasma expansion dynamics are different resulting in different growth dynamics.

For low process pressures, the ablated species in the fully developed plasma can expand into the chamber with hardly any interaction with the background gas since the mean free path is in the order of meters to kilometers. In this free expansion regime, the initial plasma starts expanding with a constant velocity of around 1 cm/µs [13] and within chamber dimensions, it is hardly decelerated by the background gas. The initial plasma is expanding adiabatically which converts the thermal energy of the particles in a high kinetic energy, up to 100 eV. The resulting plasma is wide and has low emission intensity in the visible light spectrum. In this pressure regime, the thin film is deposited on the substrate by direct bombardment of the ablated species. The surface mobility partially originates from the kinetic energy of the deposited material. The material flux has an forwardly peaked angular distribution with values of n typically between 5 and 15 [14,15,16].

3.3.2 High pressure expansion regime: Shock Wave

For pressures in the range of 0.1 mbar, the expanding plasma has a mean free path of several millimeters to centimeters and is gradually decelerated by collisions with the background gas. The leading edge of the plasma is colliding primarily into the static background gas and experiences the largest deceleration. The upcoming, non-decelerated plasma, collides again on the non-decelerated leading edge and results in a volume at the leading edge with number densities up to a factor of 4 higher than the background gas [17]. This deceleration causes a densification of the leading edge of the plasma, known as the shockwave [18]. Due to internal collisions and the collisions with the background gas, the power n is lowered locally and eventually is even reduced to zero [10]. At any pressure a shockwave will develop but the exact location of the shockwave is a function of the background pressure. A shockwave is formed at a distance of the typical target-substrate separation (50 mm) at around 0.1 mbar [19,20].

The exact pressure depends on the type of background gas (molecular weight and collision cross-sectional area) and the ablated material. Several models have been proposed to describe the expansion of a plasma in the shockwave regime [21,22,23]. The initial plasma does not differ from the free expansion case in the low pressure regime. The expansion velocity is therefore identical to the low pressure case and the deceleration starts after ~ 1 µs. The simplest model is that describes the propagation of the shock front is the phenomenological shock model which is described by R=atγ with 0<γ<1. This describes the position R of the leading edge with time and a and γ being fitting parameters. Since the shock model implies a plasma with an infinite velocity at t=0, the drag model is introduced to better describe an expanding shockwave front. The drag model reads R=α(1-e-βt_{) where β is the slowing coefficient}

and α being the stopping distance [13,24]. This slowing of the leading edge results in the densification.

Prediction of shockwave formation can also be performed by calculating the amount of background gas that is displaced and comparing it to the mass in the plasma plume:

p g SW M R

ρ

π

Mp is the mass of the material inside the plume. If this amount is comparable to the

ejected mass from the target, the plasma gets decelerated. This model links the formation of a shock wave to the used background pressure.

In the leading edge of the plasma, collisions between the species are taking place, resulting in the lowering of the value of the power n. Even at these high pressures, the material is still arriving at the substrate with a high n order flux distribution; the lowering of the value n only takes place inside the shock wave at the leading edge. This effect is depicted in figure 3.3.a.

Figure 3.3.b and c show optical images of a plasma expanding into the chamber where a shockwave has formed (b) at 0.13 mbar and without a shockwave (c) at 0.025 mbar. Figure 3.3.b shows that the material is more confined in the shockwave plasma with a high intensity as in the 0.025 case the material is spread with lower luminescence. The densification at the leading edge can not be observed due to interaction of this edge with the background gas what leads to excessive cooling and lower emission intensities. In order to observe this densification, absorption imaging should be applied [13,25,26].

a) _{b) c)}

Figure 3.3: a) the initial expansion of the plasma results in a high forwardly peaked velocity of

the total plasma (high n number); inside the shockwave the deceleration leads locally to the lowering of n, b) optical emission image in the visible light of a PZT plasma expanding into 0.13 mbar of oxygen and c) plasma expanding into 0.025 mbar of oxygen. Both images used the same time of illumination.

The excessive cooling induced by the high pressure background gas also results in a different thin film growth regime. The high pressure background gas causes the species inside plasma to loose their kinetic energy gained during the adiabatic expansion process. The species are thermalized and typically only have ~0.1 eV’s of energy left. Another effect of the high background pressure is the excessive cooling of the plasma, resulting in a high supersaturation. This high supersaturation is one of the key features for successfully growth of multicomponent oxide materials. In the low pressure case, material is deposited and growing onto the substrate as in the high pressure regime the material is growing from the supersaturated vapor followed by surface growth.

The expansion mechanisms for oxides differ heavily from thermal evaporation material fluxes since oxides are often deposited close to, or in the shockwave regime. Due to these differences, stencil deposition combined with PLD can result in altered material transport through the apertures of the stencils. Especially the lowering of the directionality in the leading edge of a shockwave can influence clogging and broadening mechanisms. The influence of the shockwave on clogging and broadening will be discussed in the next chapter.

3.4 Concluding remarks

Pulsed laser deposition is a thin film deposition technique which can be operated over a wide range of process parameters. A pulsed laser beam is used as an energy source which is placed outside the deposition system. The pulsed character of this deposition technique makes this physical vapor deposition technique unique. Key features like a low heat transfer, stoichiometric material transfer of multicomponent targets and high

degrees of supersaturation enables successful deposition and growth of a broad range of materials.

PLD can be operated in two major regimes; the free expansion deposition regime and the shockwave growth regime. In the free expansion regime, material can expand freely with little interaction with the background gas. PLD is operated in this regime where high kinetic energy of the arriving species is needed to obtain better adhesion or to enhance surface mobility. In the shockwave regime, the expanding plasma experiences a lot of interaction with the background gas and is decelerated in to a dense cloud. As a result, the expanding plasma interacts heavily with the background gas and is also cooled excessively. High supersaturation ratios are obtained in this regime and thin films are growing from this highly saturated plasma. This regime is used often in the growth of oxides.