國立中山大學材料科學研究所 碩士論文
利用磁控濺鍍合成鎂銅金屬薄膜及其特性分析
Preparation and Characterization of Mg-Cu Binary Metallic Thin Film
研究生：周鴻昇 撰 指導教授：黃志青 博士
中華民國 九十六 年 六 月
i
Content
Content ... i
Tables List ... iv
Figures List ... vi
Abstract ... xii
Chapter 1 Introduction ... 1
1-1 Amorphous metallic alloys ... 1
1-2 Characteristics of bulk metallic glasses (BMGs) and thin film metallic glass (TFMGs) ... 2
1-3 The development of Mg-based thin film ... 5
1-4 The propose and motive of this research ... 6
Chapter 2 Background and Literature Review ... 8
2-1 Evolution of amorphous alloys ... 8
2-2 Systems of amorphous alloys ... 11
2-3 Evolution of fabrication methods for amorphous alloys ... 13
2-4 Glass-forming ability (GFA) ... 15
2-5 Theory and phenomena of sputter deposition process... 17
2-5-1 Introduction of sputtering ... 17
2-5-2 Systems of sputter deposition process ... 19
2-5-3 Thin film growth mechanism ... 21
2-5-4 Zone model for sputtered coatings ... 25
2-6 Fabrication of TFMGs ... 26
2-7 Properties of thin film metallic glass ... 30
2-7-1 Thermal properties ... 30
ii
2-7-2 Mechanical properties ... 31
2-7-3 Electric and magnetic properties ... 34
Chapter 3 Experimental Procedures ... 35
3-1 Materials ... 35
3-2 Sample preparation ... 35
3-2-1 Substrate preparation ... 35
3-2-2 Film preparation ... 36
3-2-3 Post-deposition treatments ... 38
3-3 Property measurements and analyses ... 39
3-3-1 X-ray diffraction ... 39
3-3-2 Preparation of TEM specimens of co-sputtered and multilayered thin films ... 40
3-3-2 Qualitative and quantitative constituent analysis ... 41
3-3-3 Thermal analysis using differential scanning calorimetry (DSC) ... 41
3-3-4 Nano-mechanical analysis using nanoindenter ... 41
Chapter 4 Results ... 43
4-1 EDS analysis of co-sputtered Mg-Cu thin films ... 43
4-2 X-ray diffraction analyses ... 43
4-3 TEM observation of co-sputtered Mg-Cu thin films ... 45
4-3-1 Microstructure of co-sputtered 100-150 (Mg17.7Cu82.3) and 100-100 (Mg23.5Cu76.5) thin films ... 45
4-3-2 High-resolution TEM observation of co-sputtered 100-100 thin films ... 46
4-4 Thermal analysis ... 46
4-4-1 DSC analysis of Mg-Cu co-sputtered thin films ... 46
4-4-2 Structural transformation of the 100 series thin films at 423 K ... 47 4-4-3 Formation of intermediate phase of Mg-Cu multilayered thin films at 413
iii
K and 363 K ... 48
4-4-4 Structural transformation of 20T32 by TEM observation ... 50
4-5 Mechanical analysis of co-sputtered 100-150 (Mg17.7Cu82.3), 100-100 (Mg23.5Cu76.5), and 100-50 (Mg40.4Cu59.6) using nanoindenter ... 50
Chapter 5 Discussion ... 52
5-1 Composition shift due to different powers ... 52
5-2 Oxidation of Mg-Cu co-sputtered film ... 55
5-3 Diffusion-induced phase transformation in multilayered thin films ... 56
5-4 Comparison between Mg-rich and Cu-rich amorphous alloy fabricated by sputtering and liquid-quenching process ... 59
5-5 Nano-mechanical properties of 100-150 (Mg17.7Cu82.3), 100-100 (Mg23.5Cu76.5), and 100-50 (Mg40.4Cu59.6) ... 62
Chapter 6 Conclusion ... 66
References ... 68
Tables ... 73
Figures... 90
iv
Tables List
Table 1-1 Functional properties and application fields of bulk amorphous and
nanocrystalline alloys [6]. ... 73
Table 2-1 The classification of amorphous alloy systems [7]. ... 74
Table 2-2 The classification of amorphous alloy systems [7]. ... 75
Table 2-3 Prediction and observation of metallic glass formation by ion mixing in binary metal systems [40]. ... 76
Table 2-4 Zone structures in thick evaporated and sputtered coating [51]. ... 77
Table 2-5 Structure phase and ductility of Pd-TFMG made by different types of target at different Ar pressures [13]. ... 78
Table 2-6 Tg, Tx and resistivity of Pd-TFMG made by arc-cast target at different Ar pressures [13]. ... 79
Table 2-7 Properties of Pd-TFMG and bulk metallic glass [13]. ... 80
Table 2-8 Electrical resistivity of the thin film metallic glasses and conventional electrical device materials [31]. ... 81
Table 3-1 The details of the co-deposition conditions. ... 82
Table 3-2 The details of multilayer sputtering conditions. ... 83
Table 3-3 The information of the Mg-Cu multilayered films. ... 84
Table 3-4 The information of the Mg-Cu co-sputtered films ... 85
Table 4-1 The composition of the Cu-Mg cosputtered films ... 86
Table 4-2 Structural and compositional comparison between Mg-Cu co-sputtered and multilayered thin films ... 87
Table 5-1 Comparison between thermal properties of Mg-based binary and ternary amorphous alloys. ... 88
v
Table 5-2 Comparison between Values of Young’s modulus and hardness from published nanoindentation works and this study. ... 89
vi
Figures List
Figure 1-1 A scheme of long-range-ordered structure [1]. ... 90
Figure 1-2 A scheme of short-range-ordered structure [1]. ... 90
Figure 1-3 The frame for the upscale models of the Vertu mobile phone is made of liquid-metal alloy because of its high strength, hardness, and scratch resistance[3]. ... 91
Figure 1-4 (a) A sketch of conical spring microactuator, and (b) a fundamental structure of micro-switch made by metallic glass thin films [5]. ... 92
Figure 1-5 FIB nanomold on completely glassy Zr-Al-Cu-Ni thin films [4]. ... 93
Figure 1-6 Schematic representation of the solid-state reaction at the interface of (a) Fe0.67Hf0.33 and (b) Fe0.50Hf0.50 films, showing the hypothetical iron profiles, considering a planar growth for every reacted layer [14]. ... 94
Figure 2-1 A schematic diagram of the splat quenching methos[43]. ... 95
Figure 2-2 A schematic diagram of the two roller quenching method[43]. ... 95
Figure 2-3 A schematic diagram of the chill block melt spinning [43]. ... 96
Figure 2-4 A schematic diagram of the planar flow casting process [43]. ... 96
Figure 2-5 Characteristics of metallic glasses [11]. ... 97
Figure 2-6 New approach for understanding GFA of amorphous materials [49]. ... 97
Figure 2-7 Events that occur on a surface being bombarded with energetic atomic-sized particles [51]. ... 98
Figure 2-8 A schematic illustration of a DC diode Sputtering System [53]. ... 98
Figure 2-9 A schematic illustration of the RF diode sputtering deposition [53]. ... 99
Figure 2-10 A schematic illusion of a planar magnetron sputtering system [53]. ... 99 Figure 2-11 The side view of the magnetic field configuration for circular planar
vii
magnetron cathode [50]. ... 100 Figure 2-12 The top view of the magnetic field configuration for a circular planar
magnetron cathode [50]. ... 100 Figure 2-13 Basic modes of thin-film growth [54]. ... 101 Figure 2-14 Coarsing of islands due to (a) Ostwald ripening, (b) sintering, and (c) cluster
migration [54]. ... 101 Figure 2-15 A schematic representation showing the superposition of physical process
which establishes structural zones [51]. ... 102 Figure 2-16 Structure zone model of sputtering deposited materials [51]. ... 103 Figure 2-17 Plane-view TEM micrographs and diffraction pattern of the films in (a)
as-deposited and annealed conditions at (b) 650, (c) 750, (d) 800, and (e) 850 K. The circled regions indicate the location for obtaining the diffraction patterns [12]. ... 104 Figure 2-18 The heat-flow rate as a function of temperature for a sputtered, multilayered
thin film of the average stoichiometry Ni68Zr32 [27]. ... 105 Figure 2-19 X-ray diffraction profile for the Ni-Zr thin film: (a) as deposited, (b) after DSC scan to 670 K and quench to room temperature, (c) after a DSC scan to 870 K [27]. ... 105 Figure 2-20 Cross-section bright-field TEM micrograph of a Ni/Zr bilayer annealed at
300oC for 60 min. Void may be seen at the Ni/NiZr interface as at V [39]. .. 106 Figure 2-21 Cross-section bright-field TEM micrographs of Ni/Zr bilayer annealed at
300oC for (a) 240 min and (b) 720 min. The growth of the voids can be noted [39]. ... 107 Figure 2-22 Correlations exist between vapor phase growth conditions and many of properties of the resultant thin film [53]. ... 108 Figure 2-23 The DSC curve of Pd-TFMG [13]. ... 108
viii
Figure 2-24 TTT diagram for the onset of crystallization of Pd-TFMG [13]. ... 109 Figure 2-25 SEM image of a Pd-thin-film metallic glass free-standing microbeam with a
notch fabricated by FIB [33]. ... 109 Figure 2-26 Variation of notch fracture toughness (KC) of Pd-based thin-film metallic glass as a function of annealing time in the supercooled liquid region at 640 K close to Tg = 637 K [33]. ... 110 Figure 2-27 SEM micrographs of the fracture behaviors ahead of the notch tips in Pd-based TFMG (a) as-deposited, an overview of the fractured sample, (b) as-deposited, a high magnification observation of the fracture surface, (c) annealed for 90 s and (d) annealed for 480 s [33]. ... 110 Figure 2-28 The SEM image of the deformation morphologies around the indents in the
Au/Cu multilayers with individual layer thicknesses [32]. ... 111 Figure 2-29 The pileup height (hpu) and the hardness (H) as a function of λ. Hardness (H)
was measured at the 200 nm indentation depth [32]. ... 111 Figure 2-30 FIB cross-sectional views of the indents in the Au/Cu multilayers with
individual layer thickneses (λ) of (a) λ=250 nm, (b) λ=100 nm, (c) λ= 50 nm, (d) λ= 25 nm. The bright and dark layers correspond to Au and Cu layers, respecrively. Inhomogeneous shear banding becomes prevalent with the decrease in λ[32]. ... 112 Figure 2-31 Electrical resistivity ρ as a function of annealing time Ta in the Pd76Cu6Si18
thin film metallic glasses annealed at various temperatures [31]. ... 112 Figure 2-32 Magnetic force microscopy images of films in the as-deposited and annealed
conditions [61]. ... 113 Figure 3-1 Mg-Cu binary phase diagram. ... 114 Figure 3-2 Flow chart of the experimental procedures. ... 115 Figure 3-3 Temperature profile of the isothermal heat-treatment at 423 K of the
ix
co-sputtered thin films. ... 116
Figure 3-4 Temperature profile of the isothermal heat-treatment at 413 K of the multilayered thin films. ... 117
Figure 3-5 Temperature profile of the isothermal heat-treatment at 363 K of the multilayered thin films. ... 118
Figure 3-6 Schematic illustrations of (a) Top-view and (b) Side-view of preparation of XTEM-specimen by focus ion beam (FIB). ... 119
Figure 3-7 The cross-section images of 20T32 and annealed 20T32 specimens during XTEM preparation via FIB technique. ... 120
Figure 4-1 XRD patterns of the 100 series. ... 121
Figure 4-2 XRD patterns of the 50 series. ... 122
Figure 4-3 XRD patterns of the as-deposited 20T32 and 40N32. ... 123
Figure 4-4 XRD patterns of the as-deposited 20T14 and 40N14. ... 124
Figure 4-5 XRD patterns of the pure Mg and Cu metallic films. ... 125
Figure 4-6 TEM plane-view bright-field image of 100-150. ... 126
Figure 4-7 Selected area diffraction pattern of 100-150. ... 127
Figure 4-8 TEM plane-view low-magnitude bright-field image of 100-100. ... 128
Figure 4-9 TEM plane-view low-magnitude dark-field image of 100-100. ... 129
Figure 4-10 TEM plane-view high-magnitude bright-field image of 100-100. In the circle, the typical form indirectly postulated the Mg-Cu amorphous phases in nature. ... 130
Figure 4-11 Selected area diffraction pattern of 100-100. ... 131
Figure 4-12 High-resolution TEM image of 100-100 with Mg-Cu amorphous/MgCu2 {110} crystalline structure. The marked region is the Mg-Cu amorphous phase. ... 132
Figure 4-13 High-resolution TEM image of the 100-100 specimen. MgCu2 particles in the {110} are around the Mg-Cu amorphous matrix in the marked region... 133
x
Figure 4-14 High-resolution TEM image of the structure of Mg-Cu amorphous/MgCu2 in
the {110} plane in the 100-100 specimen. ... 134
Figure 4-15 Modified non-isothermal DSC curve of 100-150. ... 135
Figure 4-16 The structural transformation of 100-150 at 423 K. ... 136
Figure 4-17 The structural transformation of 100-100 at 423 K. ... 137
Figure 4-18 The structural transformation of 100-50 at 423 K. ... 138
Figure 4-19 The structural transformation of 100-25 at 423 K. ... 139
Figure 4-20 The structural transformation of 20T32 at 413 K. ... 140
Figure 4-21 The structural transformation of 40N32 at 413 K. ... 141
Figure 4-22 The structural transformation of 20T14 at 363 K. ... 142
Figure 4-23 The structural transformation of 40N14 at 363 K. ... 143
Figure 4-24 (a) TEM bright-field image of the as-deposited 20T32 film with nominally Mg 150-nm-thick and Cu 50-nm-thick individual layers. (b) TEM bright-field image of the 20T32 film annealed at 413 K for 2 hours. ... 144
Figure 4-25 Modulus-displacement and hardness-displacement curves of 100-150, 100-100, and 100-50 at the strain rate of 5×10-3 s-1. ... 145
Figure 4-26 The compositional variation of Young’s modulus and hardness from the unloading regions compared among 100-150, 100-100, and 100-50. ... 146
Figure 4-27 Load-displacement curve of 100-150. ... 147
Figure 4-28 Load-displacement curve of 100-100. ... 148
Figure 4-29 Load-displacement curve of 100-50. ... 149
Figure 5-1 Cu content as a function with Cu power in the Mg-Cu co-sputtered films, where the 100 series means the Mg power is RF 100 W, and the 50 series means the Mg power is RF 50 W. ... 150
Figure 5-2 Probability of collision in Xe, Kr, Ar, and Ne [52]. ... 151 Figure 5-3 The structural transformation of Mg79Cu21 amorphous alloy annealed at 363 K
xi
[72]. ... 152 Figure 5-4 A transmission electron micrograph showing the lamellar structure in the
eutectic Cu-MgCu2 alloy, where white regions represent Cu crystalline phases, and black regions represent MgCu2 crystalline phases [73]. ... 153 Figure 5-5 The indentation depth increasing from one step to the next is the sum of the
elastic (Δhe) and the slow (Δhslow) and fast (Δhfast) plastic deformations, which can be determined by the slopes of the smoothed average curve, the corrected unloading curve and the slow regime [74]. ... 154 Figure 5-6 Load-contact depth curves of 100-150 and 100-100. ... 155
xii
Abstract
In this study, Mg-Cu thin film metallic thin films were fabricated via two ways, the co-deposition and post-annealing of the multilayered thin films. Amorphous Mg1-xCux, where x is from 38 to 82, thin films with nanocrystalline particles are able to be fabricated via co-sputtering. The mechanism of formation is different from the rapid quenching process.
For the Mg-Cu co-sputtering system, the mechanical properties of the Mg-Cu co-sputtered films were tested via MTS nanoindenter. Mg23.5Cu76.5 exhibits a higher Young’s modulus than Mg17.7Cu82.3 and Mg40.4Cu59.6 due to the partial amorphous structure. Moreover, the pop-in effects with a smaller size occurs of the Mg23.5Cu76.5 sample in a higher frequency than of the Mg17.7Cu82.3 and Mg40.4Cu59.6 samples. The small pop-in effects in the Mg23.5Cu76.5
sample approximate match the width of amorphous matrix via the HRTEM observation.
Another process to form the amorphous thin film is via the post isothermal annealing process of the multilayered thin films. However, for the specimens of 20T32 consisting of 150-nm Mg and 50-nm Cu individual layers, the Mg individual layers would react to the Cu individual layers during the annealing at a temperature of 413 K owning to the slight negative heat of mixing. Due to the localized diffusion near the interfaces, Mg2Cu gradually form during the isothermal annealing since Mg2Cu is the most stable phase below 548 K [62].
Localized interdiffusion near the interfaces between Mg and Cu individual layers induced the formation of Mg2Cu rapidly. For the 40T32 specimens consisting of 15 nm Mg and 5 nm Cu individual layers, Mg2Cu rapidly form at 413 K due to the high interface energy. Then, the similar result exhibits in the 20T14 and 40T14 specimens annealed at 363 K.
xiii
摘要
近幾十年來，具有獨特的物理特性及化學特性之非晶質合金受到眾多學者的研究及 探討。目前合成非晶質合金的方法為常見的為由液相至固相之液態激冷法。在 2000 年 後，藉由物理氣相層積製備非晶質合金薄膜由於其應用潛力，因此越來越受到廣泛注意。
本實驗分別利用共濺鍍及多層膜兩種不同之製程來製備鎂銅金屬薄膜，接著再利用
真空熱處理並藉由 XRD 觀察其結構之轉變。藉由 XRD 觀察可以發現，鎂銅共濺鍍薄
膜，銅含量從38~82 at%，呈現為奈米鎂銅介金屬化合物及鎂銅非晶質之奈米複材。在
423 K 真空熱處理後，發現其結晶相會隨著成分所在之共晶區域有所差異。在 MgCu2-Cu
區域中，熱處理後以奈米MgCu2為主；在Mg2Cu-MgCu2區域中則是以Mg2Cu 及 MgCu2
為主。然而，鎂銅多層膜中，在經由413 K 熱處理後發現其結晶相與其成分無關，均為
Mg2Cu，這是由於擴散僅發生在介面附近以及鎂銅之介金屬化合物在低溫時，其最穩定
相為Mg2Cu。
在此研究中，發現僅有利用共濺鍍製程有可能形成鎂銅非晶質合金薄膜，接著利用 奈米壓痕技術量測Mg17.7Cu82.3, Mg23.5Cu76.5, 及 Mg40.4Cu59.6之楊氏模數及硬度，可以發 現到Mg23.5Cu76.5具有最高之楊氏模數及硬度為118 GPa 及 4.6 GPa，再利用 TEM 觀察
Mg23.5Cu76.5共濺鍍薄膜，發現到其微結構呈現以 MgCu2為強化相之鎂銅非晶質合金複
材，因而表現出較高之楊氏模數及硬度。
1
Chapter 1 Introduction
1-1 Amorphous metallic alloys
The amorphous metallic alloy is a disordered material without the periodicity of crystalline structure. The structure of amorphous metallic alloys displays as a random structure which consists of a short-range order structure in contrast to the structure of crystalline materials which show a long-range order in repeating unit cells, as shown in Figure 1-1 [1]. In fact, the random atomic arrangement of the amorphous alloy is not completely disordered but exists a short-range order in a very small localized area, as shown in Figure 1-2 [1]. According to the appearance and thermodynamic characteristics of these materials, amorphous metallic alloys are also called as liquid metals, glassy metals, non-crystalline metals, or metallic glasses.
As people know, the rapidly quenching method from the liquid to solid phase is the most popular way to synthesize amorphous alloys in forms of ribbons, bulks, or powders. This principle is based on the solidification through a sufficiently high cooling rate in order to prevent from crystalline-phase nucleation. At the initial stage, it is considerably difficult to reach such a high cooling rate about 107 K/s. However, a better method that can reduces effectively the critical cooling rate (lower than 103 K/s) to fabricate bulk metallic glasses was discovered during 1980s by adding the third element into the original binary system. The amorphous alloys are very suitable for the increased demand to resist the severe environmental situations because of their particular properties.
2
The disordered structure is different from the crystalline structure with the same construction in repeating unit cell. The interval between atoms to each other is called the free volume. Due to a large mount of free volume in the disordered structure, amorphous alloys exhibit the shear band deformation mechanism, lower Young’s modulus, higher tensile strength, higher electric resistance, and excellent gas absorption ability, etc.
Recently, in order to achieve the ultimate goal of the light and tough engineering application, the amorphous alloy is considered to be a potential material. Inoue’s group [2]
produced ternary and multi-component bulk metallic glasses for functional applications, such as protective coating. The outshape of Vertu, the famous cellphone of Nokia, is coated by liquid-metal alloys, as shown in Figure 1-3 [3]. Owing to the trend of microminiaturization in the electronic industry, high strength and tough thin films metallic glasses with an excellent superplastic formation ability are very suitable for the enclosures of electrical parts. Up to now, the Pd-based and Zr-based metallic glass thin film is developed as a microactuator and nanopatterning, as shown in Figures 1-4 [4] and 1-5 [5], respectively. The functional properties and application fields are listed in Table 1-1 [6].
1-2 Characteristics of bulk metallic glasses (BMGs) and thin film metallic glass (TFMGs)
During the last decade, bulk metallic glasses (BMGs) were developed popularly.
Because metallic glasses exhibit many properties different from crystalline metals, research fields were focused on the fabrication of metallic glasses, as well as the rheological behavior, mechanical, thermodynamic, thermal, electric and magnetic properties, especially for the fabrication and mechanical properties. BMGs exhibit many excellent advantages for engineering applications, such as a high elastic energy, a high yield strength, a small elastic
3
limit, a low Young’s modulus, as well as good wear and corrosion resistances.
Besides, many metallic glasses have a large supercooled liquid region with the crystallization resistance. It implies BMGs have an excellent superplastic formation ability.
However, the biggest problem of the fabrication in using the rapidly quenching method is to achieve an enough cooling rate (higher than 106 K/s). However, Inoue [6] announced the empirical rules for the synthesis of amorphous alloys:
(1) Multicomponent systems consisting of more than three elements,
(2) Significant difference in atomic size ratios above 12% among the three main constituent elements,
(3) Negative heats of mixing among the three main constituent elements.
The third element replacement for partial or complete substitution by Gd, Ni, or Y in the Mg-Cu system is an important and valid way to solve the problem in fabrication and the diameter enlargement of BMGs [2,7]. In addition to the advantages mentioned above, adding the third or more elements into an alloy system is able to adjust the properties of BMGs for some particular applications.
It is well known that the thin film technology is widely applied in the development of the semiconductor industry. According to the types of interfaces, the five main interfaces and interface reactions (IRs) pointed by Zhu et al. [8] play important roles in this field. They are listed below:
(1) Oxide/silicon interfaces － metal-oxide-semiconductor (CMOS),
(2) Oxide/metal interfaces － gas sensor and giant magnetoresistence spin valve.
(3) Metal/metal interfaces － magnetic storage media,
(4) Metal/semiconductor interface － metal-semiconductor field-effect transistor,
4
(5) Semiconductor/semiconductor interface － short-wavelength laser diode.
In the films with several micrometers even nanometers in thickness, the interface reactions become very important in the thin-film field. To utilize interface reactions is able to produce different kinds of intermediate layers, such as the diffusion barrier. However, since thin film metallic glasses were discovered from 1950s, a lot of related properties were researched extensively.
Compared with TFMGs, the rapidly quenching method for bulk or ribbon of metallic glasses is mainly limited to the intrinsic nature of materials and the cooling rate in process.
Sputter and evaporation are the main methods for the fabrication of TFMGs. Two main ways for the fabrication of TFMGs are listed below:
(1) Direct methods: co-sputtering and alloy sputtering,
(2) Indirect methods: annealing-induced and stress-induced amorphization.
The direct methods are similar to the liquid quenching method for BMGs. In the co-deposition process, the TFMGs would be fabricated with an enough fast cooling rate (above 108 K/s), such as the success for the binary Al-Fe [9] and Au-La [10] systems.
However, TFMGs can be produced in using the alloy deposition technique with a lower cooling rate by adding the third or more elements. The Zr-Cu-Al [11], Zr-Cu-Al-Ni [12] and Pd-Cu-Si [13] TFMGs are successful cases in using the alloy sputter process. The annealing-induced process is a potential method for a large mount of the continuous fabrication with lower cost and high reliability. Although the stress-induced phase transformation (amorphization) is a well-known way for the modification of materials, the mechanism and phenomena will not be discussed particularly because this is not the emphasis of this study.
5
As mentioned above, the composition and microstructure of the films can be well controlled for any shape materials. Furthermore, the surface modification used by the metallic glass coating is able to adjust the properties of the substrate and promote the smooth and shiny surface, even enhance the mechanical, thermal, and electrical properties. An intermediate layer between a film and substrate frequently forms after a post-treatment, such as annealing and plastic deformation. The intermediate layer usually forms and exhibits completely different nature from the film and substrate, such as the Fe-Hf system as shown in Figure 1-6 [14].
1-3 The development of Mg-based thin film
Mg-based thin films, such as MgCl2, magnesia (magnesium oxide), MgF2, and MgB2, were studied popularly since 1980s. At the early stage of the development, most Mg-based thin films were synthesized by the chemical vapor deposition technique and the sol-gel method. In 1995, Magni and Somorjai [15] synthesized the magnesium chloride thin film which has widely been used as the Ziegler-Natta catalyst by chemical vapor deposition (CVD) in order to speed up the polymerization. Because the film exhibits good refractoriness, good corrosion resistance, high thermal conductivity, low electrical conductivity, and transparency to infrared, the magnesia thin film has been synthesized widely by the same method [16] and used as microelectronic devices, such as piezoelectric micro-actuators and sensors [17].
MgF2, another magnesium halide compound, and magnesia exhibit high transparency, low refractoriness, chemical and mechanical durability, so it has been used as the anti-reflective coating on glass. Recently, the physical vapor deposition technologies, like the sputter, evaporation, and pulsed laser deposition (PLD) techniques may be used to fabricate MgO [17], MgB2 [18], Pd-Mg [19,20,21] thin films.
6
Some magnesium alloys, such as pure magnesium, the Ln-Mg (La or Ce) alloys and the Mg-Ni metallic glass, were reported as suitable hydrogen-storage materials reported in a lot of references [22,23]. However, characteristics of Pd-Mg thin films have been popularly noticed. Higuchi et al. [20,21] reported a high-performance hydrogen-storage Pd-Mg alloys for future clean systems. Although the hydrogen-storage ability of magnesium is excellent, the high desorption temperature may decrease the value in its application. They introduced different thicknesses of the Pd layer with the opposite properties to magnesium in order to adjust the hydriding-dehydriding property.
Owing to the introduction of the Pd layer, the Pd-Mg thin film can not only store a large mount of hydrogen but also release them easily. In 2002, the tri-layer Pd-Mg-Pd thin film was fabricated. The tri-layer structure can control the hydriding/dehydriding property more and more easily.
1-4 The propose and motive of this research
The metallic glass is a potential material for the micro-electro-mechanical systems (MEMs) and other applications for some functional purposes, such as good wear and corrosion resistance, high hydrogen-storage ability, and antibiotic coating, etc. Inoue [7] have reported various BMG systems, such as Fe-, Co-, Zr-, Pd-, and Mg-based systems, and their applications since 1980. Reducing the critical cooling rate by the addition of alloy elements is able to modify the thermal, mechanical, electric or magnetic properties. Though BMGs have excellent properties, they might be hard to be made as micro- or even nano-components in MEMs. The thin-film technique for MEMs has been used in metallic glasses since 1980s. The large-area thin films with uniform-distribution of the composition and the structure are able to be fabricated. There are two kinds of thin films listed below:
7
(1) Bi- or multi-component monolayer thin film metallic glasses, (2) Bi- or multi-component multilayer thin film.
In this study, the above two kinds of Mg-Cu metallic thin films are fabricated by the magnetron sputter deposition and post-thermal treatment. The first method is to utilize the co-sputtering technique to fabricate the Mg-Cu monolayer thin film metallic glasses. The second is to fabricate the Mg-Cu multilayer thin film and induce the amorphization by annealing.
It is intended to investigate the amorphization nature and mechanism, caused by interdiffusion through an annealing process at different temperatures. The microstructure, thermal, and mechanical properties of the resulting thin films will be examined.
8
Chapter 2 Background and Literature Review
2-1 Evolution of amorphous alloys
During 1960s, Klement et al. [24] developed firstly the splat quenching method to fabricate the Au-Si amorphous alloy in avoiding crystallization during solidification. Owing to the unique characteristics of the amorphous structure, this method generated wide interests of the new type alloy among scientists and engineers. In 1969, Chen and Turnbull [25]
successfully synthesized an amorphous alloy of the Pd-Si-X (X=Ag, Cu or Au) ternary system. A few years later, Chen [26] fabricated the Pd-T-Si (T= Ni, Co or Fe) ternary amorphous alloy in using the die casting and roller-quenching method, and the maximum diameter of BMGs was 1 mm. They also analyzed effects of the alloy systems and compositions by the thermodynamic calculation. According to this report, as the replacement atoms increased, the arrangement of atoms was more random. The glassy metal was more easily to form.
Other potential processes for fabrication were published since early 1980s. In 1983, Schwarz and Johnson [10] fabricated first the La-Au TFMG by the solid-state reaction.
Furthermore, he found the broad hump from 25o to 45o by X-ray diffraction (XRD) with Cu Kα radiation, representing the formation of amorphous structure. After 1983, more alloy systems were fabricated by evaporation and sputtering. In 1986, Cotts et al. [27] successfully fabricated the Ni-Zr multilayer thin films by the magnetron sputter method, and observed the amorphization of crystalline metallic thin film by differential scanning calorimetry (DSC) at a constant temperature scanning rate of 10 K/min. The formation of amorphous phase means
9
that it is the intermediate state from crystalline phase to intermetallic compound. In the beginning of 1980s, Kui et al. [28] produced firstly the Pd-Ni-P bulk metallic glass with 5 mm diameter by the heating and cooling cycle. They used this method which may decrease the homogeneous and heterogeneous crystalline nucleation rate, and successfully improved the BMG maximum diameter from 5 mm to 10 mm by processing the Pd–Ni–P melt in a boron oxide flux.
Chen et al. [29] rapidly fabricated the Al-Fe, Bi-Fe and Bi-Ti binary thin films in 1988.
The thin film was quenched by liquid nitrogen while gas atoms were deposited on the NaCl substrate by evaporation or sputtering.
In the late 1980s, Inoue’s group [7] in Tohoku University of Japan developed many new multicomponent metallic glass systems with the lower cooling rate in Mg-, Ln-, Zr-, Fe-, Pd-, Cu-, Ti- and Ni- based systems. Amorphous alloys started to be popularly noted due to their particular properties which can be applied to MEMs parts, hydrogen storage medium, and etc.
In 1996, Dudonis et al. [30] produced the amorphous Zr-Cu TFMGs by co-sputtering. They first discussed the relation among deposition factors such as the working gas pressure, the substrate temperature and the microstructure of deposited thin films.
Near 2000, Akira’s group [11,13] at Precision and Intelligence Lab developed Zr-Cu-Al and Pd-Cu-Si ternary TFMGs, fabricated by alloy sputter deposition, for MEMs application.
Because of a large supercooled region, excellent three-dimension forming ability, good corrosion resistance and mechanical properties, the Zr- and Pd- based TFMG is an appropriate choice for the conical spring linear microactuator in MEMs. This report opened the TFMGs application in the micro-electro-mechanic industry.
10
Annealing-induced full amorphization in multi-component monolayer thin films was reported by Chu et al. [12] in 2004. The Zr-Al-Cu-Ni crystalline thin film was annealed under argon atmosphere at the heating rate of 40 K/min and the holding time of 60 s at the temperatures range from 550 to 950 K. The crystalline phase transferred to the meta-stable amorphous phase due to the atomic interdiffusion. Compared with direct fabrication methods of TFMGs, this method can save much energy and cost because this method does not need to maintain the very low temperature of the substrate. Thus, the post-treatment methods to fabricate TFMGs are potential processes with the convenience and low cost.
After one year, Inoue’s group [4] fabricated the nano-device by the nano-scale pattering technique. The Zr-Al-Cu-Ni TFMGs were deposited by magnetron sputtering. The nano-pattern on Zr-based TFMGs was produced by the focus ion beam (FIB) technology.
This nano-pattering technology may be applied to fabrication of nano-mold and high-density memories.
Many researchers have reported evolution and improvement of TFMGs fabrication by a wide margin since 1960s. Because of the application of TFMGs in MEMs, several important properties, such as the electric resistance [31], hardness [32], fracture phenomena [33], reliability [34] and mechanical properties [13,35], were noticed and researched. Due to the absence of defects, the size effect is the major factor which leads to different properties between bulk and thin-film specimens. When the dimension of a crystalline material becomes as small as they can in MEMs, the intrinsic length scales, such as a grain size and dislocation structure, are no longer small in comparison with the geometric dimensions of the material, which can cause reliability problems of the microcomponents. The research of Zhang et al.
[36] reported the deformation behavior of free-standing Pd-based TFMGs in 2005. As the results of the bending test, as-deposited films revealed ductility due to the existence of shear
11
Bands. After annealing, ductility might transfer to the brittleness because of the absence of shear bands.
In conclusion, it is ready to enter a more mature stage since amorphous alloys started to develop in 1960. Multi-component bulk metallic glasses produced by Inoue’s group in using the liquid-quench method have excellent mechanical properties, the superplastic formation ability and thermal stability. Due to these advantages, amorphous alloys can be used in bulk application easily. The outer shape of amorphous alloys is able to strength the resistance to severe circumstance. However, amorphous alloys made by deposition techniques are the potential materials as MEMs parts because it is fast, low-cost, and very easily to control the composition and morphology. Furthermore, all kinds and shapes of materials are easily to be coated with an amorphous thin layer for improvement of properties.
2-2 Systems of amorphous alloys
Up to now, many systems of amorphous alloys were discovered. Systems of amorphous alloys can be roughly divided into non-ferrous and ferrous alloy systems since 1980s. Table 2-1 [7] summarized the types of amorphous alloys. The non-ferrous systems are Mg-Ln-M (Ln = Lanthanide metal, M = Ni, Cu or Zn), Zr-Al-TM (TM = VI-VIII group transition metal), Zr-Al-TM, Zr-Ti-Al-TM, Ti-Zr-TM, Zr-Ti-TM-Be, Zr-(Nb, Pd)-Al-TM, Pd-Cu-Ni-P, Pd-Fe-Ni-P, Ti-Zr-Ni-Cu-Sn and Ti-Zr-Ni-Cu-(Si, B) systems. However, the ferrous alloy systems include Fe-(Al, Ga)-metalloid, (Fe, Co, Ni)-(Zr, Hf, Nb)-B, Fe-Co-Ln-B, Ni-Ti-P and Ni-Nb-(Cr,Mo)-(P, B) systems. The ferrous alloy systems have been developed during the last four years after the synthesis of the nonferrous alloy systems [7].
According to the features of the alloy components, systems of amorphous alloys are
12
divided into five groups listed in Table 2-2 [7]. The first group consists of ETM, IVB ~ VIB group transition metal, (or Ln), Al and LTM, VIIB ~ VIIIB group transition metal, such as Zr-Al-Ni. The second group is composed of LTM, ETM and metalloid as indicated by Fe-Zr-B. The third group is LTM (Fe)-(Al, Ga)-metalloid systems. The fourth group is indicated to Mg-Ln-LTM and ETM (Zr, Ti)-Be-LTM system. The fifth system is composed only of two kinds of group element (LTM and metalloid), such as Pd-Cu-Ni-P.
When the fabrication techniques for bulk metallic glasses were widely developed, thin film metallic glasses produced by gas deposition processes were noted as another potential material for MEMs. Thin film metallic glasses are usually divided to bi- or multi-component monolayer thin films, and bi- or multi- component multilayer thin films.
Physics researchers utilized binary metallic thin films to analyze the phase transformation between the crystalline, intermetallic and amorphous phases [27,37] and diffusivities in amorphous alloys [38] in 1980s. Many references reported the phenomena of amorphization through different solid-state reactions. It is particularly noted that the report of Newcomb and Tu [39] which shows the Ni-Zr bilayer thin film by sputter deposition exhibited the amorphization after annealing. However, Liu’s group [40] in Beijing Tsinghua University published a review paper in 1987 which is concerned with the binary systems of thin films by the ion mixing process. At least 54 alloy systems were found over the latest ten years, which is listed in Table 2-3 [40].
According to previous researches, binary equilibrium phase diagrams were usually used to predict the conditions under which the glassy phase can be formed. The formation enthalpy (Hf) and the alloy compositions are the two important parameters in consideration, reported by Liu [40]. Many systems with large negative Hf values frequently have a broad two-phase
13
region and/or complicated intermetallic compounds. It may imply that systems with negative Hf might form amorphous phase easier.
There are two types of the binary alloy systems. The first type of the binary systems has a negative value of Hf. Al-X (X=Au, Co, Fe, Mn, Mo, Nb, Ni, Pd, Pt, Ta, and Ti), Au-X (X=Co, Fe, Ni, Ru, and W), Co-X (X=Gd, Mo, Nb, Tb, Ti, and Zr), Fe-X (X=Gd, Mo, Nd, Tb, Ti, W, and Zr) systems belong to this type. However, Another type with positive value of Hf
reveals instability of amorphous phase implying the difficulty of the amorphous phase formation. Au-X (X=Co, Ru, Ti, and W), Ag-X (X=Co, Cr, Cu, Nb, and Ni), Cu-X (X=Cr, Fe, Os, Ta, and W) belong to this type. This empirical rule can roughly predict the formation of the amorphous phase or not.
At the same time of the expansion in the bulk metallic glass, Akira’s group [11] reported the synthesis of Zr-Cu-Al and Pd-Cu-Si systems of multi-component monolayer TFMGs and its applications as microactuator near 2000. Until now, new binary alloy system are still found, such as the Zr-Ta [41], Ni-Mn [42] and Au-Cu [32] systems.
2-3 Evolution of fabrication methods for amorphous alloys
Fabrication processes of metallic glasses may be divided into three types of reactions:
(1) Solid-solid reaction which occurs only in solid phase: ion implantation, ion beam mixing, mechanical alloying (MA), accumulative roll bonding (ARB), and annealing,
(2) Liquid-solid reaction which occurs from liquid to solid phase: splat quenching, chill block melt spinning (CBMS), planar flow casting (PFC), spray forming, conventional metallic mold casting, and high pressure die casting,
14
(3) Gas-solid reaction which is from gas to solid phase: sputter and evaporation deposition [43].
Generally speaking, it is easily to form amorphous alloy through the rapidly quenching methods, such as the melting spinning, and sputter deposition. The sputter deposition process is strongly involved in the phase transition from gas to solid. However, many researches reveal that it is possible to form amorphous thin film by controlling the factors in the deposition process or post-treatment.
As noted, the production of amorphous films requires very high deposition rates and low substrate temperatures, usually below their crystallization temperatures. A low temperature can freeze the atoms on the substrate and prevent them from diffusing into equilibrium lattice sites. In order to reach a sufficient cooling rate, liquid nitrogen or helium is usually used. By mid-1950s, Buckel [44] fabricated amorphous films of pure metals, such as Ga and Bi, by thermal evaporation onto the substrate maintained at the liquid helium temperature with the fast cooling rate above 1010 K/s. Because the cooling rate was too hard to reach, other methods, which form metallic glasses more easily such as liquid quenching, gradually became popular after 1960s.
The splat quenching method was first announced by Klement et al. [24] in order to raise the cooling rate, as shown in Figure 2-1 [43]. An amorphous irregular die (area ~ 0.2 mm2; thickness ~ 10 μm) can be fabricated under the cooling rate (~106 to 1010 K/s). Figure 2-2 [43]
shows the two roller quenching method announced by Chen and Miller in 1970. This method broke the limitation of amorphous alloys. Hence, it caused wide attention because uniform long ribbons (width ~ 2 mm; thickness ~ 50 μm) can be fabricated continuously by this method. This method brought a wide tide to raise the cooling rate for fabrications of more
15
tough, and light amorphous alloys.
However, the chill block melt-spinning process developed by Liebermann et al. in 1976 is to quench the melt jet on a high-speed rotating substrate wheel by inert gas, as shown in Figure 2-3 [43]. This method can largely promote the stability and quality of fabrication processes. Until now, other liquid quenching methods, such as the plannar flow casting process, are based on CBMS concept, as shown in Figure 2-4 [43].
Before 1980, the fabrication methods emphasize on how to increase the cooling rate in avoiding formation of the crystalline phase. However, Inoue et al. [2] develop the Mg-based ternary amorphous alloy by the copper mold casting method in 1991. Bulk metallic glasses of Mg-Cu-Y with 4 mm diameter were successfully synthesized. Addition of the third kind element cooperating with liquid quenching can effectively reduce the critical cooling rate (lower than 102 K/s). In 1992 [45], they successfully fabricate bulk metallic glasses with 7 mm in diameter by the high pressure die casting method.
The gas-solid reactions to form metallic glasses are well known as gas deposition techniques. The multi-component monolayer TFMGs can be fabricated by two methods: (1) co-sputter deposition process, and (2) alloy sputter deposition process, reported by Dudonis et al. [30] and Hata et al. [11], respectively. It is completely discussed in the later section 2-6.
2-4 Glass-forming ability (GFA)
As people know, the metallic glass can be synthesized by many methods, like ion beam mixing, melting spinning, die casting, and sputtering, etc. Since the first amorphous alloy fabricated in 1960, lots of alloy systems were reported by many researches. Due to the
16
complication of multi-component alloy systems, many criteria were announced to assume the glass-forming ability (GFA) thermal stability. By using these criteria, the design of alloy system, and evaluation and prediction of properties can become easier, effective, and more economic. Three criteria were reported:
(1) Supercooled liquid range, ΔTx ( = Tx - Tg, where Tx and Tg are the onset
crystallization temperature and the glass transition temperature, respectively) [45], (2) Reduced glass transition temperature, Trg ( = Tg /Tl, where Tl is the liquidus
temperature) [46], (3) γ ( = Tx/(Tg + Tl)) [47].
First, ΔTx is the initial criterion of GFA and popularly used as an index of the thermal stability. During a supercooled liquid range, as shown in Figure 2-5 [47], amorphous alloys exhibit the nature of ideal Newtonian fluid which is very suitable for forming. The large ΔTx
value indicates that the disordered structure can exist in a wide temperature range without crystallization. This implied the resistance for crystallization and growth of crystalline phase [48]. However, ΔTx is a rough parameter which does not consider any crystallization mechanism during cooling and heating process. Second, ratio of Tg/Tl was introduced for pure kinetic reactions associated with the need to avoid crystallization. Trg (= Tg /Tl) is typically assumed to be less dependent on compositions, while Tl often decreases more strongly. The interval between Tl and Tg thus generally decreases and the value of Trg
increases with increasing alloying concentration so that the probability of being able to cool through this ‘dangerous’ range without crystallization is enhanced.
Therefore, a new parameter γ based on Tx/(Tg + Tl) was announced by Lu and Liu [47]
in 2002. As Lu and Liu mentioned in the article, γ is more suitable and accurate to evaluate glass-forming abilities of alloy systems because of completely consideration of thermal
17
stability and resistance to crystalline, as shown in Figure 2-6 [49]. Besides, it also has a relation with the critical cooling rate (Rc) and critical specimen thickness (Zc), and the relationship has been formulated as follows:
) 19 . 117 exp(
10 1 .
5 × 21 − γ
=
Rc , (2-1) )
70 . 41 exp(
10 80 .
2 × −7 γ
=
Zc . (2-2)
2-5 Theory and phenomena of sputter deposition process
2-5-1 Introduction of sputtering
The sputter deposition, a physical fabrication process of thin films, utilizes ion bombardment and momentum transfer. The incident particles impact the surface or near-surface atoms of a target with enough energy which can break bonds and dislodge atoms by the bias between the anode and cathode. During this process, the collision between incident particles and atoms of a target can be viewed as elastic collision because the loss of momentum transfer is nearly zero. Sputtering occurs whenever any particle strikes a surface with enough energy to dislodge an atom from surface. The sputter yield is just the ratio of the number of emitted particles per incident particle [50]:
Y = (# of emitter particles/ # of incident particles). (2-3)
Sputtering can occur for virtually all incident species, including atoms, ions, electrons, photons, and neutrons as well as molecules and molecule ions. Generally speaking, the plasma includes inert gas ions, such as Ar+, Kr+, or small molecule ions such as N2+, O2+, etc.
18
Sputter yield is independent on the particle charge but is dependent on the physical momentum transfer and kinetic energy from incident the particles to surface atoms.
When energetic particles bombard on the target, many reactions occur between the ionized gas particles and target atoms near the surface, as shown in Figure 2-7 [51]. Most of the transferred energy (>95 %) appear as heat in the surface region and near-surface region.
Some of the bombarding particles are reflected as high energy neutrals and some are implanted into the surface. When an atomic sized energetic particle impinges on a surface, the particle bombardment effects can be classed as:
(1) Prompt effects (<10-12 s) – lattice collisions, physical sputtering, reflection from the surface,
(2) Cooling effects (>10-12 s to 10-10 s) – thermal spikes along collision cascades, (3) Delayed effects (>10-10 s to years) – diffusion, strain-induced diffusion and
segregation,
(4) Persistent effects – gas incorporation compressive stress due to recoil implantation.
Incident particles, also called as plasma, are produced by glow discharge. In the ultra-high vacuum chamber, dilute gas, such as argon, is introduced into the chamber. Gas molecules can be ionized when the bias between cathode and anode is large enough. At this moment, argon atoms are divided to Ar+ ions and electrons. Electrons produced at the initial of glow discharge are called primary electrons. However, plasma can not exist stably due to the unstable primary electron source. Actually, the stable plasma is maintained by stable secondary electrons source which are produced near the target surface by the collision between the ions and target atoms.
At the beginning of the discharge, the primary electrons from the cathode are accelerated by the electric field near the cathode to the anode. These energetic electrons collide with the
19
gas molecules and generate positive ions before they travel to the anode. The positive ions bombard the cathode surface, which result in the generation of secondary electrons from the cathode surface. The secondary electrons increase the ionization of the gas molecules and generate a self-sustained discharge [52]. The advantages for sputter deposition include:
(1) Excellent film uniformity, particularly over large areas, (2) Easy control of surface smoothness and uniform of films,
(3) Deposition of films with nearly bulk-like properties, which are predictable and stable,
(4) The sputtering deposition is essentially a kinetic process involving momentum exchange rather than a chemical and/or thermal process. Therefore, virtually any material can be introduced into a gas discharge or sputtered from solid,
(5) Good adhesion of films,
(6) Sputtering allows for the deposition of the films having the same composition as the target source.
2-5-2 Systems of sputter deposition process
According to types of power supply, systems of sputter deposition processes can be divided to four types: (1) direct current (DC) diode sputtering system, (2) radio frequency (RF) diode sputtering system, (3) magnetron sputtering, and (4) reactive sputtering.
The DC diode sputtering system is the simplest type of sputtering. Figure 2-8 [53] shows a pair of planar electrodes in a vacuum chamber. One electrode is a cold cathode, and the other is an anode. Glow discharge can occur with enough bias voltage. The top plasma-facing surface of the cathode is covered with a target material and the reverse side is water-cooled.
The substrates are placed on the anode. After the glow discharge starts, Ar ions in the glow
20
discharge are accelerated at the cathode fall and bombard the target resulting in the deposition of the film on the substrates. Near the cathode is a dark space or sheath in which may collect large mounts of primary and secondary electrons. The RF diode sputtering system is very similar to the DC sputtering system, but differ from power supply, as shown in Figure 2-9 [53]. The DC system is suitable for conductor targets, such as Fe, Cu, Zr, Ni, and Mg.
By substitution of an insulator target for a metal target in a DC diode sputtering system, the sputtering glow discharge can not be sustained because of the build-up of a surface charge of positive ions on the front side of the insulator target. To solve this problem, the RF power supply is substituted for the DC power supply. For a small part of the RF cycle, the cathode and anode are electrically reversed. This can eliminate the build-up charge on an insulating surface by providing equal number of ions, electrons, ions, and so on. The RF power supply is operated at high frequency about from 60 MHz to 80 MHz. However, the most common frequency is 13.56 MHz and its multiples (2× or 3×) [50].
A second key advantage of the RF diode system is that the oscillation of field in the plasma (at the driving frequency) resulting in additional electron motion with the plasma.
This has been described in several ways, but the most interesting point is an analogy to the electron ”surfing” on the electric field waves in the plasma. The end result of this enhanced electron movement is that the probability of an ionizing collision is increased for a given secondary electron, and this results in an increase in the plasma density compared to DC diode. The density increase results in higher ion currents and a faster sputtering rate.
Although DC and RF sputter deposition process is a convenient, high reliability and continuous process, a slow deposition rate and charging are still the major disadvantages of these processes. In 1935, Penning’s study [52] first revealed the low pressure sputtering under
21
a transverse magnetic field which was added on a DC glow discharge tube. A magnetron uses a static magnetic field configured at the cathode location, as shown in Figure 2-10 [53]. The magnetic field is located parallel to the cathode surface, as shown in Figure 2-11 [50].
Secondary electrons emitted from the cathode due to ion bombardment are constrained by this magnetic field to move in a direction perpendicular to both the electric field (normal to the surface) and the magnetic field. Figure 2-12 [50] shows an E×B drift, which is based on Hall Effect. This drift causes electrons to move parallel to the cathode surface in a direction 90 degree away from the magnetic field.
Basing on these issues mentioned, an additional magnetic field added can prolong the electron retaining time in the plasma and improve the probability of collisions. This enhances the efficiency of the simple diode sputtering, and also makes diode sputtering configuration operate under higher currents and pressures.
In reactive sputtering, thin films of compounds are deposited on substrates by the sputter deposition from metallic targets which react with reactive gas, such as oxygen, nitrogen with inert gas (invariably Ar). The types of compounds synthesized by reactive sputtering are oxides, nitrides, carbides, sulfides, and are briefly listed below [54]:
(1) Oxides (oxygen) – Al2O3, In2O3, SnO2, SiO2, and Ta2O5, (2) Nitrides (nitrogen, ammonia) – TaN, TiN, AlN, and Si3N4, (3) Carbides (methane, acetylene, propane) – TiC, WC, and SiC, (4) Sulfides (H2S) – CdS, CuS, and ZnS,
(5) Oxycarbides and oxynitrides of Ti, Ta, Al, and Si.
2-5-3 Thin film growth mechanism
22
The formation of thin films synthesized by the sputtering deposition process is related to gas-solid phase transformation which contains three stages:
(1) Nucleation, (2) Growth, (3) Coalescence.
At the initial stage of formation, energetic vapor atoms have some probability to condense on the substrate by forming bondings between vapor atoms and the substrate. An embryo composed of several adatoms forms on the substrate owing to stable bondings between atoms and the substrate. Based on thermodynamic theories, the form of embryos is usually assumed as sphere shape. Embryos with smaller than the critical size (r*) are unstable and they may maintain the unstable situation or re-evaporation. Contrarily, embryos can transfer to stable nuclei while the size of embryos is larger than r*. At this stage, nucleation is the main reaction, so this stage is also called the nucleation stage. The nucleation rate (N‧) can be obtained by some thermodynamic calculations and simplification, as in the following equation:
kT G E N E
MRT a PN
r
N A s des s
* 0
* exp
sin 2
2 − −Δ
= π θ π
& . (2-4)
The quantities Ns, Edes, Es and, ΔG* represent the total nucleation site density, the energy required to desorb it back into the vapor, the activation energy for surface diffusion, and critical free energy of nucleation, respectively.
After the nucleation stage, nuclei growth becomes the main reaction, which is called the growth stage. When a surface density of nuclei reaches a saturated situation, nuclei do not
23
form any more but start to grow up. In fact, there is no clear demarcation between the end of nucleation and the onset of growth. In this stage the prior nuclei incorporate impinging atoms and subcritical clusters, and then grow in size while the island density saturates rapidly. Many observation have pointed to three basic growth modes, as shown in Figure 2-13 [52]:
(1) Island (or Volmer-Weber) mode,
(2) Layer (or Frank-van der Merwe) mode, (3) Stranski-Krastanov mode.
The island growth occurs when the smallest stable clusters nucleate on the substrate and grow in three dimensions to form the inlands. This happens when bondings between molecules each other are more stable than those between molecules and the substrate. Some systems of metals on insulators, alkali halide, graphite, and mica substrates exhibit this mode of growth.
The layer growth is the opposite type to the island type. The extension of the smallest stable nuclei occurs overwhelmingly in two dimensions resulting in the formation of planar sheets. In this mode, the adatoms tend to form bonds with the substrate more likely than with each other. The most important example of the mode involves the single-crystal epitaxial growth of semiconductor films.
The layer plus island or Stranski-Krastanov (S.K.) growth mechanism is an intermediate combination of the aforementioned modes. In this case, after forming one or more monolayers, the subsequent layer growth becomes unfavorable. Although the transition from the two- or three- dimensional growth is not completely understood, the guessable cause that disturbs might be the binding energy characteristic of layer growth. The growth mode is fairly common and has been observed in metal-metal and metal-semiconductor systems.
24
At the last stage of thin film formation, it is called coalescence containing three basic mechanisms, as shown in Figure 2-14 [52]:
(1) Ostwald ripening, (2) Sintering,
(3) Cluster migration.
Prior to coalescence there is a collection of islands with varied sizes and with time. The larger ones grow or “ripen” at the expense of the smaller ones, with the assumption of no contacts between islands each other. In order to minimize the free energy of the island structure, islands have a tendency to grow up or ripen by inter-diffusion.
The sintering mechanism is similar to the ripening mechanism but with the contacts between two islands. The driving force of this mechanism for neck growth is simply the natural tendency to reduce the total surface (or area) of the system. This results in the observed mass transport into the neck controlled by surface diffusion. It is clearly that diffusion is a thermally activated reaction. The adatoms with higher energy not only move from the target to the substrate, but diffuse on the substrate. Hence, the thin film has a smooth surface and fewer defects.
However, excess kinetic energy is not only provided by adatoms but also substrates. An insufficient temperature can not support overall adatoms or clusters to migrate but a part of the adatoms to diffuse around to near islands. Ostwald ripening and sintering mechanisms predominate over the coalescence stage under a lower temperature environment. Cluster migration, a high-temperature coalescence mechanism, occurs as a result of collisions between separate island-like crystallites (or droplets) as they execute random motion. Cluster migration has been directly observed in many systems, such as Ag and Au on MoS2, Au and
25
Pd on MgO.
It is well known the three mechanisms play important roles in the formation of thin films.
Different types of mechanisms can lead to different microstructures which would immediately affect the properties of thin films.
2-5-4 Zone model for sputtered coatings
As described above, the microstructure of thin films is a critical key for its mechanical, electric, thermal, and magnetic properties. Film growth, especially the nucleation stage, determines the film natures such as film density, surface area, surface morphology and grain size. Important aspects of the film growth are listed below:
(1) Angle-of-incidence of the adatom flux effects – geometrical shadowing,
(2) Ratio of deposition temperature (degrees K) to the melting temperature (degree K) of the film material (T/Tm),
(3) Energy released on condensation,
(4) Adatom surface mobility on surfaces and different crystallographic planes, (5) Substrate surface roughness – initially and as the film develops,
(6) Deposition rate,
(7) Reaction and mass transport during deposition – segregation effects and void agglomeration,
(8) Mass transport and grain growth during deposition.
Figure 2-15 shows that different microstructures under different predominately processes, nearby, the shadow process, surface diffusion process, and bulk diffusion process.
Thornton [51] integrated these microstructures and announced structure zone model (SZM) to
26
describe the microstructure of sputtered and evaporated films, as shown in Figure 2-16.
Roughly speaking, the grain size varies with working gas pressure and Ts/Tm, where Ts
and Tm are the temperature of the substrate and melting temperature of the film, respectively, as listed in Table 2-4 [51]. A low temperature or working pressure atmosphere would make thin films with Zone 1 structure but porous. In Zone 1, the surface diffusion is insufficient to overcome the geometrical shadowing by the surface features. In this situation, the film composing of tapered grains with a high surface area exhibits a mossy appearance. Since the growth is strictly a function of the surface geometry, angle-of-incidence and adatom surface mobility, amorphous as well as crystalline materials show the columnar growth mode.
In Zone 2, the growth in dominated by the adatom surface diffusion. In this region, the surface diffusion allows the densification of the intercolumnar boundaries. However, the basic columnar morphology remains. The grain size increases and the surface features tend to be faceted. In Zone T, the coating exhibits a fibrous morphology and is considered to be a transition from Zone 1 to Zone 2. The formation of the Zone T is due to the energetic bombardment from reflected high energy neutrals from the sputtering target at low gas pressures. In Zone 3, the bulk diffusion allows the recrystallization, grain growth and densification. Often each grain in Zone 3 can be considered as a single crystal in the highly modified columnar morphology.
2-6 Fabrication of TFMGs
In previous chapters, the fabrication methods have been discussed. In this portion, the topic focuses on evaporation and sputtering.
27
The gas-solid reactions to form metallic glasses are well known as gas deposition process. The bi- or multi-component monolayer TFMGs can be fabricated by two methods, listed below:
(1) Co-sputter deposition process, (2) Alloy sputter deposition process.
In 1996, Dudonis et al. [30] successfully fabricated ZrxCu1-x TFMGs by the co-sputter deposition process. In their report, they discovered that the composition range of the formation is very wide (x = 0.05 ~ 0.95), and electrical conductance and reflectance vary with different compositions. Furthermore, Akira’s group [11,55] in Precision and Intelligence Laboratory at Tokyo Institute of Technology reported the Zr-Cu-Al and Pd-Cu-Si TFMGs, which were directly produced by the alloy sputter deposition process near 2000. Zr-Cu-Al and Pd-Cu-Si alloy targets were fabricated by the arc melting process.
Phase transformation through thermal treatment is a popular method for particular properties such as diffusion barriers [56] and magnetic storage devices [57]. According to the idea of phase transformation, Chu et al. [12] announced a concept for the application of TFMGs, fabricated using annealing-induced full amorphization, to fabricate thin film metallic glasses. In the report, Zr-Cu-Al-Ni thin films were first deposited on the glass substrate. Then, the films were amorphized via low-temperature annealing.
The TEM image and the associated diffraction of the as-deposited film as shown in Figure 2-17 (a) [12], clearly indicate a typical sputtered nanocrystalline structure with a grain size of about 10~30 nm. After annealing at 650 K, the microstructure has transformed into a two-phase nanophase composite of an amorphous matrix containing uniformly dispersed nanocrystallines, as shown in Figure 2-17 (b). When the film was annealed at 750
