<%BANNER%>

Leakage Current Behavior of Reactive RF Sputtered HfO2 Thin Films


PAGE 1

LEAKAGE CURRENT BEHAVIOR OF REACTIVE RF SPUTTERED HfO2 THIN FILMS By M I C HA E L N. J ONES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 By Michael N. Jones

PAGE 3

This document is dedicated to the memory of my grandfather, Louis R. Jones. His support for me encouraged me to complete this work. I am glad I was able to make my grandfather proud.

PAGE 4

iv ACKNOWLEDGEMENTS First, I would like to thank my parents for encouraging me to attend the University of Florida and to continue in pursuit of a higher education. I feel that I have made the most of my college years, which can be attributed in part to my parental support. Meanwhile, my sister, Kayla, has been a large part of my life and I am grateful for that. I want to thank all of my friends for enriching my experience while at school, and I thank Kathryn for her compassion and support. I am very grateful to my research advisor, Dr. David Norton. Dr. Norton guided and encouraged me throughout this project. I would like to express gratitude to Dr. Amelia Dempere at the Major Analytical Instrumentation Center (MAIC) for providing me with monetary support and the opportunity to work with a great staff. I would like to thank all of my group members, especially Byoung-Seong Jeon and Yongwook Kwon. Byoung-Seong taught me how to use the sputtering system and worked with me to accommodate both of our schedules. Yongwook helped me make electrical measurements and interpret data. I would also like to thank Joshua Howard and Dr. Singh’s group for allowing me to use their CV/IV measurement system.

PAGE 5

v TABLE OF CONTENTS Page ACKNOWLEDGMENTS ……………………………………………………………… iv LIST OF TABLES ……………………………………………………………………... vii LIST OF FIGURES …………………………………………………………………….viii ABSTRACT ……………………………………………………………………………..xi CHAPTER 1 INTRODUCTION...........................................................................................................1 1.1 Need for Alternate High-k Dielectric Material on Silicon.......................................1 1.2 Deposition on ITO....................................................................................................2 2 BACKGROUND.............................................................................................................4 2.1 Introduction to High-k Gate Dielectrics on Silicon.................................................4 2.2 Review of Candidate Materials................................................................................8 2.3 Literature Review of HfO2 Thin Film....................................................................10 2.3.1 Structure..........................................................................................................10 2.3.2 Processing and Properties...............................................................................12 2.3.3 MOSFET Device Compatibility.....................................................................15 2.4 Sputtering...............................................................................................................1 5 2.4.1 DC Sputtering..................................................................................................15 2.4.2 RF Reactive Magnetron Sputtering.................................................................18

PAGE 6

vi3 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON INDIUMTIN-OXIDE (ITO).........................................................................................................19 3.1 Introduction............................................................................................................19 3.2 Experimental Procedure.........................................................................................19 3.2.1 Substrate Preparation.......................................................................................19 3.2.2 Film Growth....................................................................................................20 3.2.3 Measurements..................................................................................................20 3.3 Results and Discussion...........................................................................................22 3.3.1 Dielectric Constant..........................................................................................22 3.3.2 Conduction Mechanism...................................................................................23 3.3.3 Leakage Current..............................................................................................29 3.3.4 Microstructure.................................................................................................34 3.4 Conclusions............................................................................................................40 4 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON SILICON...42 4.1 Introduction............................................................................................................42 4.2 Experimental Procedure.........................................................................................43 4.2.1 Substrate Preparation.......................................................................................43 4.2.2 Film Growth....................................................................................................43 4.2.3 Measurements..................................................................................................44 4.3 Results and Discussion...........................................................................................46 4.3.1 Microstructure.................................................................................................46 4.3.2 Dielectric Constant..........................................................................................53 4.3.3 Leakage Current Behavior..............................................................................55 4.3.4 Current Mechanism.........................................................................................60 4.4 Conclusions............................................................................................................62 5 SUMMARY ................................................................................................ ................ 64 REFERENCES..................................................................................................................66 BIOGRAPHICAL SKETCH............................................................................................70

PAGE 7

vii LIST OF TABLES Table page 1 Band gaps and band offsets on silicon for candidate high-k dielectrics [3,4].............9 2 Film thicknesses of HfO2 samples on ITO determined by stylus profilometry........21 3 Film thicknesses of HfO2 samples on p-type silicon determined by stylus profilometry and ellipsometry...................................................................................44

PAGE 8

viii LIST OF FIGURES Figure page 1 Band diagram of semiconductor/insulator interface at flat band................................7 2 High-k gate dielectric candidates................................................................................7 3 Schematic of a DC sputtering system.......................................................................17 4 Interactions with incident ions at the target surface..................................................17 5 Schematic cross-section of the metal-insulator-ITO structure used in this work. The aluminum electrode was 150-200 nm thick; the thickness of the HfO2 film, as measured by profilometry, is shown in Table 2........................................................21 6 Schottky emission conduction mechanism. !B is the schottky barrier height. Note the magnitude of !M, the energy required for thermionic emission from the metal.25 7 Leakage current of HfO2 on ITO measured at various temperatures. The sample was sputtered with 5 mTorr O2 and 100C substrate temperature...................................27 8 Schottky emission plots of HfO2 on ITO measured at different temperatures. The samples was sputtered with different deposition conditions: (a) 5 mTorr O2 and 100C, and (b) 5 mTorr O2 and 25C........................................................................28 9 Room temperature leakage current plots of HfO2 on ITO showing the effect of substrate temperature. Each plot shows a different set of deposition oxygen pressures: (a) 1 mTorr O2, (b) 5 mTorr O2, (c) 10 mTorr O2....................................30 10 Room temperature leakage current plots of HfO2 on ITO, showing the effect of oxygen pressure during film deposition. Each graph displays a different substrate temperature: (a) 25C, (b) 100C, (c) 200C...........................................................32 11 Leakage current of HfO2 on ITO measured at room temperature and 2 V versus growth conditions......................................................................................................33 12 XRD diffractogram of HfO2 on ITO grown with 10 mTorr O2 and varying substrate temperature................................................................................................................34

PAGE 9

ix 13 RMS roughness increase measured by AFM as oxygen pressure was varied during 200C substrate temperature deposition....................................................................35 14 RMS roughness increase measured by AFM. The increase corresponds to increasing substrate temperature when the films were grown with 10 mTorr O2......................36 15 AFM derivative image of HfO2 grown on ITO at 5 mTorr O2 and 200C..............37 16 AFM image of HfO2 grown on ITO at 10 mTorr O2 and 100C with highpass filter applied to enhance grain edges..................................................................................38 17 AFM image of HfO2 grown on ITO at 10 mTorr O2 and 200C with highpass filter applied to enhance grain edges..................................................................................39 18 Drawing of MIS structure with interface layer.........................................................45 19 XRD diffractogram of HfO2 films grown on silicon at 200C substrate temperature, showing effect of oxygen pressure on crystallization...............................................46 20 XRD diffractogram of HfO2 films grown on silicon at 25C substrate temperature, showing the effect of oxygen pressure on crystallization.........................................47 21 XRD diffractogram of HfO2 films grown on silicon with 5 mTorr O2 pressure, showing the effect of substrate temperature on crystallization.................................47 22 RMS roughness as a function of substrate temperature for HfO2 on Si deposited at48 23 RMS roughness as a function of oxygen pressure for HfO2 on Si deposited at 200C. ............................................................................................................................... ....48 24 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 200C..................................................................................................................49 25 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 100C..................................................................................................................50 26 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 25C....................................................................................................................51 27 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 5 mTorr O2 and 200C..................................................................................................................52 28 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 1 mTorr O2 and 200C..................................................................................................................53

PAGE 10

x 29 Dielectric constant (calculated from accumulation capacitance) as a function of growth conditions for HfO2 grown on silicon...........................................................54 30 Dielectric constant (calculated from accumulation capacitance) as a function of growth conditions for HfO2 grown on silicon...........................................................55 31 Room temperature leakage current measurement of HfO2 on silicon showing the decrease in leakage current that corresponds with increasing substrate temperature during deposition. The graphs show the trend for films grown with (a) 5 mTorr and (b) 10 mTorr oxygen pressure...................................................................................56 32 Room temperature leakage current measurement of HfO2 on silicon showing the decrease in leakage current that corresponds with increasing oxygen pressure during deposition. The graphs show the trend for films grown at (a) 100C and (b) 200C substrate temperature.................................................................................................57 33 Room temperature leakage current of HfO2 on silicon showing (a) effect of substrate temperature when films are grown with 1 mTorr O2 and (b) effect of oxygen pressure when films are grown at 25C....................................................................59 34 Schematic E vs. x band diagram of the Al/HfO2/SiO2/p-type Si structure in this experiment.................................................................................................................61

PAGE 11

xi Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science LEAKAGE CURRENT BEHAVIOR OF REACTIVE RF SPUTTERED HfO2 THIN FILMS By Michael N. Jones December 2003 Chairman: David P. Norton Major Department: Materials Science and Engineering HfO2 is currently under consideration as a potential high-k material to replace conventional SiO2. HfO2 thin films were deposited onto indium-tin-oxide (ITO) and ptype silicon (100) substrates using reactive RF sputter deposition. Substrate temperature and the amount of oxygen used during deposition were varied to determine their effect on leakage current and microstructure. X-ray diffraction (XRD) and atomic force microscopy (AFM) were both used for microstructural characterization. Al/HfO2/ITO structures were used to measure leakage current and dielectric constant as a function of deposition conditions. Temperature-dependent leakage current measurements allowed for the extraction of the HfO2/ITO Schottky barrier height. In addition, metal-oxidesemiconductor (MOS) structures were fabricated on silicon with aluminum electrodes. The MOS structures were used to measure current-voltage and capacitance-voltage characteristics as a function of deposition conditions.

PAGE 12

1 CHAPTER 1 INTRODUCTION 1.1 Need for Alternate High-k Dielectric Material on Silicon The scaling of the metal-oxide-semiconductor field-effect transistor (MOSFET) feature size to sub-100 nm dimensions requires a decrease in thickness of the current thermal oxide SiO2 gate dielectric to less than 1 nm. At this thickness, direct tunneling leakage current is significant for SiO2. A solution to the problem is to find materials with high dielectric constants relative to SiO2. A material with a higher dielectric constant (high-k material) can be made thick enough to avoid direct tunneling and still maintain the capacitance of the thinner SiO2 film. An alternative gate dielectric material should have a high dielectric constant, thermal and chemical stability in contact with silicon, high quality interfaces, and low leakage current. Some of the materials that researchers have considered are Y2O3, Al2O3, SrTiO3 (STO), BaxSr1-xTiO3 (BST), Ta2O5, and TiO2. Materials like Y2O3, and Al2O3 do not provide a significant advantage over SiO2 because their dielectric constants are not significantly higher. In contrast, ultrahigh-k materials like STO and BST cause poor short channel effects due to the fringing field induced barrier lowering effect. It has also been reported that some high-k dielectrics such as Ta2O5, TiO2, and strontium titanate (STO) are unstable in direct contact with silicon and have a high leakage current. HfO2, however, with a dielectric constant on the order of 20-25, is a promising alternative gate dielectric. HfO2 reportedly has good thermal stability on silicon and can

PAGE 13

2 consume native oxide to form HfO2. It is the first high-k material to show compatibility with the present complementary metal-oxide-semiconductor (CMOS) polysilicon gate process. HfO2 has a high dielectric constant and relatively low leakage current. In this study, HfO2 was investigated as an alternative gate dielectric. HfO2 was deposited by reactive RF magnetron sputtering under varying oxygen partial pressures and substrate temperatures onto silicon and indium-tin-oxide (ITO)-coated glass substrates. Dielectric constant and leakage current properties were studied on ITO. Metal-oxide-semiconductor structures were fabricated on silicon to measure capacitancevoltage and current-voltage properties. The microstructures of these films were also investigated as a function of deposition conditions. 1.2 Deposition on ITO Most of the research effort to date on HfO2 has focused on the dielectric and leakage current properties of films deposited directly on silicon. In this case, the measured properties of HfO2 on silicon reflect both the intrinsic properties of the HfO2 film as well as the effect of a SiO2 interface layer. While consistent with the gate dielectric application, this structure makes it difficult to extract fundamental properties on the HfO2 film itself. To this end, we have investigated the dielectric and leakage current properties of HfO2 films deposited by reactive RF sputtering onto ITO-coated glass substrates. The objective of this study is to understand the electrical properties of the HfO2 without influence from an interface layer. ITO is a conducting oxide, so the dielectric constant of the HfO2 can be measured without accounting for the effect of interfacial layers. The

PAGE 14

3 effects of substrate temperature and oxygen pressure on dielectric constant, leakage current, and film microstructure are also investigated.

PAGE 15

4 CHAPTER 2 BACKGROUND 2.1 Introduction to High-k Gate Dielectrics on Silicon Alternative, high dielectric constant materials are needed to replace SiO2 as the gate oxide in metal-oxide-semiconductor (MOS) devices. The current device scaling trend requires SiO2 film thickness thinner than 1 nm. SiO2 films thinner than 1 nm would have direct tunneling leakage current greater than 1 A/cm2, which is unacceptable. A gate dielectric forms a parallel-plate capacitor, where the capacitance is given by d A Cr 0 (2.1) where A is the area of the capacitor, "o is the permittivity of free space, "r is the dielectric constant, and d is the dielectric thickness, respectively. The use of an alternative high-k dielectric would allow for a thicker gate dielectric layer to be used, reducing the tunneling current while maintaining a given capacitance. The thickness of SiO2 film that would give a capacitance-voltage curve equivalent to the new gate oxide is called the equivalent oxide thickness (EOT). A new gate dielectric should be scalable to 1 nm EOT. There are several additional conditions that an alternative gate dielectric must meet to replace thermally grown SiO2 on silicon. They are summarized in the list below. 1. The dielectric constant of the new material should be significantly higher than the dielectric constant of SiO2 film (~3.9); 2. The new gate dielectric should be thermodynamically stable in contact with silicon;

PAGE 16

5 3. The band gap of the material should be large, and the band offsets on silicon need to be greater than 1 eV to act as a tunneling barrier for both electrons and holes; 4. The interface trap defect density should be on par with current SiO2; 5. The defect density within the oxide layer should be minimal; 6. A stable amorphous phase material is preferred to avoid grain boundary leakage current paths; 7. The new material should require minimal change from the existing oxide fabrication process; 8. The dielectric should be thermodynamically stable in contact with poly-silicon; 9. There should be a low diffusion constant for dopant atoms of poly-silicon; [1] The last three conditions refer to the compatibility of the new dielectric oxide with the current MOSFET device processing. After the oxide film is grown on a semiconducting substrate, the gate material is subsequently deposited on top of the gate dielectric material. Heavily-doped polycrystalline silicon (polysilicon) is currently used as the gate material. Therefore, to meet the requirements for current MOSFET processing, the material must meet requirements dictated by the oxide/polysilicon interface in addition to the oxide/silicon interface [1] If these conditions are not met, the use of a metal gate could be used to solve this compatibility problem [2] The band offset requirement is very important to minimize leakage current. The barrier at both the conduction and valence band needs to be high to act as a barrier for electrons and holes, respectively. A schematic band diagram of the semiconductor/insulator interface is presented in Figure 1 One of the problems with the band offset condition is that materials that have large band gaps tend to have low dielectric constants. Figure 2 illustrates this trend. Therefore, it is important to choose a material that has a relatively high dielectric constant and a band gap that is large enough

PAGE 17

6 to give low leakage currents. Fortunately, the large band gap materials are usually favored by the stability criterion [3, 4] If the new material is not stable in contact with silicon, an interface layer (most likely SiO2) will form. In some cases the interface layers are shown to improve the density of interface traps, but they provide a serious obstacle for device scaling. Two capacitors in series result in the overall capacitance given in equation 2.1. 2 11 1 1 C C Ctot (2.1) Since the interface layer will have a lower dielectric constant than the high-k dielectric, it will limit the maximum achievable capacitance – or the minimum achievable EOT. The interface thickness must be strictly controlled so that the dielectric layer can be scaled to EOT less than 1 nm. In summary, it is desirable to select a material with reasonably high dielectric constant, large band offsets, and stability in contact with silicon. In the next section, several candidate high-k dielectrics are reviewed in respect to the criteria that has just been discussed.

PAGE 18

7 Figure 1: Band diagram of semiconductor/insulator interface at flat band 3 4 5 6 7 8 9 10 010203040506070Dielectric Constant, kBandgap (eV)SiO2HfO2ZrO2ZrSiO4HfSiO4TiO2La2O3Ta2O5Y2O3Si3N4Al2O3 Figure 2: High-k gate dielectric candidates [3,4,5] Semiconductor Insulator Band gap Valence Band Offset V alence Band Conduction Band Conduction Band Offset

PAGE 19

8 2.2 Review of Candidate Materials Presently, many researchers are searching for alternative gate dielectric materials to replace silicon dioxide. Since a higher dielectric constant means we can grow thicker films to reduce leakage current, why not use dielectrics with the highest dielectric constants? Many metal oxides and ferroelectric materials have been investigated as candidate materials, but most of them are not stable in contact with silicon. Furthermore, the dielectric constant of materials generally tends to increase as the band gap decreases, making it difficult to select a material with a large band gap and dielectric constant [1] For example, SrTiO3 thin films have a dielectric constant of approximately 300 at room temperature but a band offset of nearly zero. Therefore, this material cannot insulate against tunneling current [1] Robertson calculates that the barrier heights for Ta2O5, SrTiO3, and PbTiO3 on silicon are very small as shown in Table 1 Schottky emission for these films figures to be quite leaky [3,4] Accordingly, schottky emission has been shown to be the primary mechanism involved in the current transport in Ta2O5 films [6,7] Lee et al. report that annealing at 500-800 C for 30 minutes is required to lower the leakage current of Ta2O5 films, which roughens the film surface and increases EOT [8]

PAGE 20

9 Table 1: Band gaps and band offsets on silicon for candidate high-k dielectrics [3,4] Dielectric Material Band gap (eV) Conduction Band Offset (eV) Valence Band Offset (eV) SiO2 9 3.5 4.4 Si3N4 5.3 2.4 1.8 SrTiO3 3.3 -0.1 PbTiO3 3.4 0.6 Ta2O5 4.4 0.3 3 TiO2 3.1 0 ZrO2 5.8 1.4 3.3 HfO2 6 1.5 3.4 Al2O3 8.8 2.8 4.9 Y2O3 6 2.3 3.6 La2O3 6 2.3 ZrSiO4 6 1.5 3.4 HfSiO4 6 1.5 On the other hand, some materials have reasonable leakage current properties but do not offer a significant dielectric constant advantage over SiO2. Al2O3 (K~7.6) has been shown to have good leakage current characteristics [9,10] but does not have a significantly higher dielectric constant than SiO2. The same can be said for Si3N4 and most silicates. A promising gate dielectric should offer high dielectric constants, reasonable barrier heights, and stability on silicon. In addition to the tradeoff between dielectric constant and band gap, researchers have shown that materials with dielectric constant greater then 25 show degradation in MOSFET turn-off/on characteristics [11] This is caused by fringing fields from the gate

PAGE 21

10 to source/drain regions which further induce fields from the source/drain to the channel. Furthermore, gate control is weakened by this so-called fringing field induced barrier lowering [2] Materials like TiO2, Ta2O5, BST, and STO, have high dielectric constants, but require the use of barrier layer to suppress barrier lowering [2,11] and to prevent reaction and interdiffusion with the silicon substrate [12] As discussed in the previous section, the low dielectric constant interfacial layer will dominate the capacitance-voltage characteristics and limit the achievable EOT. ZrO2 and HfO2 promise to be useful as alternative gate dielectrics. Amorphous ZrO2 has been grown, and has thin EOT, low leakage current, negligible frequency dispersion, interface density on par with SiO2, small hysteresis, and excellent reliability [12] ZrO2 and HfO2 have a dielectric constant of about 25 and a large band gap. The band gap of HfO2 is about 6 eV. ZrO2 and HfO2 are theoretically more stable against the formation of SiO2 on silicon than other oxides [12] Robertson predicts ZrO2 and HfO2 to have large band offsets on silicon of about 1.5 eV [3,4] Both materials have been shown to have good properties. In fact, successful transistors with ZrO2 and HfO2 have been fabricated with the present polysilicon gate material [13,14,15,16,17] 2.3 Literature Review of HfO2 Thin Film 2.3.1 Structure Most of the work on HfO2 has been focused on amorphous films to replace SiO2. Several factors favor an amorphous dielectric. Amorphous materials do not contain grain boundaries or dislocations that can trap charge and offer fast diffusion pathways for leakage current. In addition, stresses in amorphous materials can be taken up by small variations in the random network, where they may likewise be taken up by misfit

PAGE 22

11 dislocations in a polycrystalline material. Amorphous structures also tend to minimize electronically active defects, but may give rise to shallow traps. Epitaxial films can be free of grain boundaries and defects, but they require more difficult and expensive methods to process [18] Therefore, amorphous films are usually preferred for gate dielectrics. Therefore, the crystallization of HfO2 films is an important issue. If an amorphous structure is desired, the problem of growing films that remain amorphous during device processing is not trivial. In fact, this is important for HfO2. Crystallization often occurs during growth and post-growth annealing [13,19,20,21,22] The monoclinic phase is the stable HfO2 phase, while metastable phases like orthorhombic, tetragonal, and cubic are occasionally observed. Sputtering and chemical vapor deposition (CVD) give the films with the best amorphous character. Additionally, thin HfO2 films seem to crystallize at higher temperatures than thick HfO2 films. 20-30 nm reactive sputtered films show monoclinic crystal structure even before they are annealed. Meanwhile, 20-30 nm films sputtered from a HfO2 target were amorphous until 500C annealing. The films were a mixture of monoclinic and orthorhombic phases up to 700C annealing where only the monoclinic phase dominated [19] However, thinner sputtered films of 18.5 nm [20] and 0.5 nm thickness [23] remained amorphous up to 700C annealing temperature. A hafnium metal layer was sputtered for both films. The first proceeded with subsequent reactive sputtering of the hafnium target on top of the hafnium seed layer, and the second ended with the rapid thermal oxidation of the hafnium film. In another case, thin DC reactive sputtered HfO2 remained amorphous until 650C annealing [21] The highest

PAGE 23

12 crystallization temperature reported was for CVD HfO2 that reportedly remained amorphous after gate activation annealing at 950C for 60 s [16 ]. Currently, there is a lot of interest in atomic layer deposition (ALD) of HfO2 films. There have been some promising results using ALD as the deposition method, including the processing of a MOSFET device [17] However, the nanocrystallinity of HfO2 films continues to be a concern [13] 2.3.2 Processing and Properties Several deposition processes have been used to grow HfO2 films: atomic layer deposition, pulsed laser deposition, jet vapor deposition, chemical vapor deposition, sputtering of pure HfO2 targets, reactive sputtering of hafnium targets, and sputter deposition of hafnium metal and subsequent oxidation, among others. Most of the literature has been directed toward the sputtering techniques. Sputtering is a fairly simple and trusted thin film deposition process, especially for the growth of amorphous films. The process is discussed further in section 2.4 Even though the dielectric constant of evaporated films is reported to be about 25 [24] most of the reviewed literature gave effective dielectric constant values between 12 and 21. The discrepancy between values of 12 and 21 is because of the interfacial layers and is greatly influenced by deposition and annealing procedure. One of the biggest factors in high-k gate dielectric growth is interface thickness and quality. It greatly influences properties such as dielectric constant, minimum achievable EOT, leakage current, and interface trap density. Fortunately, results of HfO2 on silicon show that high quality interfaces can be formed without special silicon surface treatment, and with the use of surface treatment an EOT of 7.1 angstroms has been achieved [25]

PAGE 24

13 Gutowski et al. stated through theoretical and experimental evidence that the HfO2/Si interface is even more stable with respect to silicides than the ZrO2/Si interface [26] However, in most cases reviewed, the growth of an interface layer, usually hafnium silicate, is observed. Several factors including deposition method, oxygen pressure, deposition temperature, annealing time/temperature, and annealing ambient – determine the thickness and quality of the interface. Even though some researchers have shown that the hafnium silicate interface layer is beneficial to reduce the interface trap density [27,28] interface thickness control is a key issue in gate dielectric processing because it limits EOT scaling. Several researchers have deposited HfO2 by reactive sputtering [15,20,21,29,30,31,32] In each case, the thickness of the interface layer increases with increasing oxygen pressure and post-deposition annealing temperature. Callegari et al. annealed HfO2 films in O2 at 600C, then measured the thickness of the interface with transmission electron microscopy (TEM). After correcting for the interface thickness, the dielectric constant of HfO2 was 21 [29] For thin HfO2 films, annealing lowers the leakage current [19,32] It is also used to fully oxidize and densify films that are deposited with hafnium metal targets instead of reactive sputtering [23,25] Fully oxidized films have better leakage current, and denser films have higher dielectric constants [19]. For this reason, reactive sputtered films and those sputtered from HfO2 targets tend to have better electrical properties than the films that are oxidized after hafnium metal deposition. However, if the annealing conditions are optimized, it is possible to scale post-deposition oxidized HfO2 below 1 nm more easily than reactive sputtered films [28]

PAGE 25

14 However, while leakage current and dielectric properties have been shown to improve with annealing of thin oxide films, the opposite may be true with thicker films. In the case of Ta2O5, where the main conduction mechanisms were the poole-frenkel effect and the schottky effect, 20 nm films were improved by oxygen annealing while 80 nm films were severely degraded by oxygen annealing [33] This phenomenon may be caused by the large effect an oxide/Si interfacial layer can have on leakage current, particularly when the oxide film is relatively thin. The major drawbacks to annealing thin films are that excessive annealing also causes more growth of the interfacial layer and may lead to crystallization of the HfO2 film. The scaling will become more difficult due to the increase in interface thickness [22,31,32,34] ; the crystallization can lead to increased leakage current. There are a few steps to curb this interfacial effect. Annealing in N2 instead of O2 shows a more drastic increase in leakage current and a smaller effect on the interface [15,21,32] For example, two similar HfO2 films were annealed in N2 and O2. For the film annealed in N2, EOT is approximately 1.9 nm, while annealing in O2 results in EOT of 2.8 nm [21] Additionally, surface preparation of the silicon substrate by annealing in NH3 or N2 suppresses the growth of an interface layer [25,27] Nitrogen passivates the surface of the silicon substrate, allowing an EOT as low as 7.1 angstroms to be achieved [25] All of the studies reviewed by this author reported very low leakage current values after annealing in either O2 or N2. There has not been much study into the leakage current mechanism of HfO2. In one study, the Fowler-Nordheim tunneling mechanism was observed in sputtered films 20-30 nm thick after annealing [19] In another study, Zhu et al. used a Fowler-Nordheim experiment at low temperature to extract band offsets.

PAGE 26

15 The HfO2/Si conduction band offset was calculated to be approximately 1.13 0.13 eV. The calculated Pt/HfO2 offset is 2.48 eV, and the Al/HfO2 offset is 1.28 eV. A trap level 1.5 eV below the HfO2 conduction band was calculated as well, which would contribute to poole-frenkel conduction. In the Al/HfO2/Si structure, schottky emission dominated over poole-frenkel since the Al/HfO2 offset is smaller than the trap level [35] Outside of this, there is little reference to the current transport mechanism of HfO2. 2.3.3 MOSFET Device Compatibility In addition to low leakage current, high dielectric constant, good reliability, thin EOT, negligible frequency dispersion, interface density on par with SiO2, and small hysteresis (as mentioned at the end of section 2.2), HfO2 is also compatible with current MOSFET processing, including the use of polysilicon gate material. Polysilicon gate MOSFETs with HfO2 gate dielectrics have been fabricated [16,17,23,25] and the performance of these devices is good. The HfO2 devices have been better than ZrO2 MOSFET devices [17] Using reactive sputtered HfO2 with a polysilicon gate, thermal stability up to 1000 C was reported along with EOT of 1.20 nm and leakage current of 1 x 10-3 A/cm2 at 1 V [23] By annealing the silicon wafer in NH3 and using a TaN gate, the EOT was ~ 7.1 angstroms and the leakage current was 1 x 10-2 A/cm2 at -1.5 V [25] 2.4 Sputtering 2.4.1 DC sputtering Sputtering has become a widely accepted process for the deposition of thin films. Mirrors were first coated by sputter deposition as early as 1877. Until the late 1960s, evaporation was preferred over sputtering. Then, the need for alloys with stringent stoichiometric limits, conformal coverage, and better adherence, for magnetic and

PAGE 27

16 microelectronic materials, increased the demand for sputtering deposition. The addition of RF (which allows the sputtering of insulators) and magnetron sputtering extended the capability of sputtering. Today, the capabilities of sputtering and the availability of highpurity targets and gases make sputtering a popular choice for the deposition of thin films [36] Some of the benefits of sputtering include: 1. High uniformity of thickness; 2. Good adhesion of film to substrate; 3. Reproducibility of films; 4. Ability to deposit and maintain the stoichiometry of the target material; 5. Relative simplicity of thickness control. [37] The sputtering mechanism is described in the following paragraphs. Sputter deposition uses a momentum transfer process to deposit thin films of target material onto substrates. A schematic of a typical DC sputtering system is shown in Figure 3 In DC sputtering, a sputtering gas (usually argon) is introduced into a vacuum chamber where an applied voltage ionizes the argon gas. The newly formed Ar+ ions accelerate toward the negatively charged cathode, which is at the target material. When an ion approaches the target, there are five major interactions that can occur. They are listed below and illustrated in Figure 4 1. Reflection : The ion may be reflected; probably neutralized in the process. 2. Secondary Electron Emission: The impact of the ion may cause the target material to eject an electron. 3. Ion implantation: The ion may become implanted in the target material. 4. Rearrangements: The impact may rearrange the structure of the target material. This could result in point defects or even lattice defects.

PAGE 28

17 5. Sputtering: The impact leads to a series of collisions between atoms, possibly leading to the ejection of target atoms [38] Figure 3: Schematic of a DC sputtering system Figure 4: Interactions with incident ions at the target surface [38] Power Supply Anode Substrate Target / Cathode Gas Inlet Vacuum Chamber

PAGE 29

18 A plasma, or glow discharge, consisting mostly of ions, electrons, and neutral atoms, is maintained between the target and substrate. During the glow discharge, it is observed that a current flows and a film condenses on the substrate [36] The major drawback to DC sputtering is that it cannot be used for insulating targets. 2.4.2 RF Reactive Magnetron Sputtering The thin film deposition process used in this thesis, RF reactive magnetron sputtering is a hybrid of several other types of sputtering. What follows is a description of these sputtering types. RF sputtering. DC sputtering cannot be used for insulating films because the surface on the target becomes positively charged. The use of a RF power source with an impedance matching network solves this problem and allows the sputtering of insulating materials. The voltage is applied at a frequency of 13.56 MHz; this allows for charging and discharging of the insulating target [38] Magnetron sputtering. Magnetron sputtering utilizes magnetic fields to increase deposition rates and allow for low operating pressures and temperatures. For example, the application of a planar magnetron will cause electrons in the glow discharge to follow a helical path, increasing the rate of collisions and ionization. Therefore, magnetron sputtering grows high quality films at lower operating pressures [37] Reactive sputtering. Reactive sputtering can be used to deposit films of such materials like oxides or nitrides through the use of pure metal targets. In addition to the sputtering gas (argon), a reactive species–oxygen or nitrogen, for example–is introduced into the growth chamber. The sputtered target atoms react with the gas to form the new material.

PAGE 30

19 CHAPTER 3 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON INDIUM-TIN-OXIDE (ITO) 3.1 Introduction Much of the research effort to date on HfO2 films as a gate dielectric has focused on the dielectric and leakage current properties of films deposited directly on silicon. While relevant for silicon MOSFET technology, the measured properties of HfO2 on silicon reflect both the intrinsic properties of the HfO2 film as well as the effect of a SiO2 interface layer. This structure makes it difficult to extract fundamental properties on the HfO2 film itself. To this end, we have investigated the dielectric and leakage current properties of HfO2 films deposited by RF reactive sputtering onto indium-tin-oxide(ITO)-coated glass substrates. The objective of this study is to understand the electrical properties of the HfO2 without influence from an interface layer. ITO is a conducting oxide, so the dielectric constant of the HfO2 can be measured without accounting for the effect of interfacial layers. The effects of substrate temperature and oxygen pressure on microstructure, leakage current, and dielectric constant are investigated. 3.2 Experimental Procedure 3.2.1 Substrate preparation Reactive RF sputter deposition was used to deposit HfO2 thin films onto ITOcoated glass substrates. Substrates were cut to a size of approximately 1 cm x 1 cm. Then, they were ultrasonically cleaned in trichloroethylene, acetone, methanol, and

PAGE 31

20 deionized water for five minutes each. They were blown dry with nitrogen gas. All substrates were then mounted onto a platen with silver paint and loaded into the sputtering system. The chamber was pumped down to a base pressure between 5 x 10-8 and 5 x 10-7 Torr. 3.2.2 Film growth HfO2 films were grown by reactive RF magnetron sputtering from a 2 inch diameter 99.99% pure hafnium target. The RF power was maintained at 100 W. Oxygen partial pressure was varied from 1 mTorr to 10 mTorr, while the total pressure (Ar + O2) was kept at 30 mTorr. Substrate temperature was varied from room temperature to 200C. The deposition conditions as used in this study are shown in Table 2 Film thickness was maintained at 60-80 nm for all samples. After loading the substrates into the deposition chamber, the temperature was ramped up to the appropriate setting. Meanwhile, the hafnium target was pre-sputtered with the shutter closed to remove any oxidation from the surface. The shutter was opened, and oxygen was slowly flowed into the system. 3.2.3 Measurements Before deposition, a portion of each substrate surface was masked in silver paint. When the paint is removed by sonication in acetone, it leaves behind a bare ITO surface. The thickness of the step was then measured by stylus profilometry. Table 2 shows the results of the profilometry measurements. For electrical property measurements, circular aluminum electrodes with area # *10-4 cm2 were sputtered as the top electrode using a shadow mask with a 200 micron diameter pattern. The thickness of these electrodes was approximately 150-200 nm. The

PAGE 32

21 ITO left bare by the silver paint mask served as the bottom electrode. The currentvoltage properties and dielectric constant of HfO2 was measured on ITO between the aluminum and ITO electrode. Current-voltage curves were measured at various temperatures to study the conduction mechanism, and capacitance was measured in order to calculate the dielectric constant. Figure 5 is a schematic cross-section if the metalinsulator-ITO structure used to make electrical measurements. Table 2: Film thicknesses of HfO2 samples on ITO determined by stylus profilometry Substrate Temperature 25C 100C 200C 1 mTorr 67.5 nm 70.0 nm 60.0 nm 5 mTorr 63.0 nm 59.0 nm 76.5 nm Oxygen Pressure 10 mTorr 65.0 nm 73.5 nm 64.0 nm Figure 5: Schematic cross-section of the metal-insulator-ITO structure used in this work. The aluminum electrode was 150-200 nm thick; the thickness of the HfO2 film, as measured by profilometry, is shown in Table 2 ITO HfO2 Al Glass Thickness measured by profilometry

PAGE 33

22 In addition to electrical measurements, x-ray diffraction (XRD) and atomic force microscopy (AFM) were utilized to examine the film microstructure. Films without deposited electrodes were scanned with the Philips APD 3720 diffractometer, and then examined by tapping mode AFM on the Digital Instruments Dimension 3100. The roughness, crystallinity, and grain structure were observed. 3.3 Results and Discussion 3.3.1 Dielectric Constant For a parallel-plate capacitor, the dielectric constant can be calculated by A d C k 0 (3.1) where C is the capacitance; d is the dielectric thickness; "0 is the permittivity of free space; and A is the area of the capacitor. For Al-HfO2-ITO structures, the capacitance was measured at 1 MHz. Note that there is little previous work on the dielectric constant of HfO2 thin films on substrates other than silicon. In the case of HfO2 on silicon, the effective dielectric constant is reduced due to the HfxSiyOz layer that forms between the oxide and the silicon substrate. The interfacial layer acts as a capacitor in series with the dielectric film. Accordingly, it is difficult to extract the dielectric constant of the HfO2 films in this way. Even if the Si/HfO2 interface thickness is measured by cross-section transmission electron microscopy (TEM), for example, the exact dielectric constant of the interface remains unknown. In the present structure, no significant interface is formed between the oxide and ITO substrate. We can directly calculate the dielectric constant from capacitance and film thickness measurements. Dielectric constant values were calculated for all the

PAGE 34

23 samples listed in Table 2 The dielectric constant of HfO2 at 1 MHz was between 20 and 24 for all but one sample. The variation was not systematic with deposition conditions. The sample grown at 100C and 5 mTorr O2 had the only outlying data point, with a dielectric constant of approximately 29. Most of the samples had dielectric constant approximately 20-21. This value is in agreement with what was determined on silicon by Callegari et al. after correcting for interface thickness [29] 3.3.2 Conduction Mechanism A variety of conduction phenomena occurs when insulating films are sandwiched between two electrodes. The identification of the dominant (rate-limiting) mechanism is important in understanding the current-voltage characteristics of the structure being studied. Two broad categories describe these mechanisms: barrier-limited and bulklimited. Barrier-limited mechanisms operate in the vicinity of the interface between insulator and the contacts. The transport of charge into the insulator limits the conduction. Schottky emission and tunneling are the most prominent examples of these types of mechanisms. In the bulk-limited case, the current is limited by the transport of carriers through the insulator. In other words, sufficient numbers of carriers are injected into the insulator, but they experience difficulty in reaching the other electrode due to bulk transport limitations. Examples are intrinsic and Poole-Frenkel conduction mechanisms [36] Schottky emission The schottky mechanism resembles the thermionic emission of electrons from a heated metal to a vacuum. Thermionic emission is described by the following equation: T k T C JM RDexp2 (3.2)

PAGE 35

24 !M is the work function of the metal and CRD is the Richardson constant. For thermionic emission, electrons need to acquire !M, typically 4-5 eV. Schottky barriers heights ( !B), however, are smaller (in the range of 1 eV) because the electrons only need to acquire enough energy to access the empty conduction band of an insulator. The value of the schottky barrier height depends on the particular interface between the metal and insulator/semiconductor. Figure 6 co m p a r es t h e t h er mi on i c e mi ss i on of an e l ec t ron from a metal to vacuum with the schottky emission at the interface of a metal and insulator [36] The current density governed by schottky emission is given by the Richardson-Dushman equation, bt nnf bt nf 2/1 0 3 24 1 expdEq Tk TAJ (3.3) where Tk CAB RDexp (3.4) The variables T and E are temperature and electric field, respectively. Constants k, q, "o, and "d represent Boltzmans constant, electron charge, permittivity of free space, and the dynamic dielectric constant, respectively [33]

PAGE 36

25 Figure 6: Schottky emission conduction mechanism. !B is the schottky barrier height. Note the magnitude of !M, the energy required for thermionic emission from the metal. The pre-exponential term A is used to extract the schottky barrier height, $B. Equation 3.4 can be written as 2 / 1 2 / 1 0 3 24 1 ) ln( ln E q T k A T Jd (3.5) Therefore the plot of ln(J/T2) versus E0.5 should give a line with y-intercept equal to ln(A). CRD is the Richardson constant and is given by 3 2* 4 h k m e CRD (3.6) where m* is the effective mass [39] For the free electron approximation CRD = 120 Acm-2K-2. The effective mass of an electron in HfO2 was measured to be 0.1m0 [35] ; accordingly CRD = 12 Acm-2K-2. Vacuum Conduction Band METAL INSULATOR !B !M e

PAGE 37

26 Current-voltage measurements were made at temperatures ranging from 25C to 250C. The measurements were made for the substrate injection case (i.e. positive gate voltage), where electrons were injected from the ITO and transported through the HfO2 film to the aluminum electrode. Since the injection of carriers into the insulator conduction band concerns the interface between the first electrode and the insulator, this allowed us to extract the height of the barrier at the ITO/HfO2 interface. The schottky height of the aluminum/HfO2 has been studied before and was calculated to be ~1.28 eV [35] Figure 7 shows the temperature-dependent current-voltage curves for a film deposited at 100C in 5 mTorr oxygen. Figure 8 shows the corresponding schottky fits for films grown at (a) 100C and (b) 25C. The ln(J/T2) versus E1/2 graphs for both illustrate a good fit to the schottky model. However, only a self-consistent dynamic dielectric constant can ensure that leakage current conduction mechanism in the film is the schottky mechanism. This means that the dynamic dielectric constant, calculated from the slope of the lines in the schottky plot, should be between the optical dielectric constant and the static dielectric constant or very close to the optical dielectric constant [8] The square of the refractive index (n) gives the optical dielectric constant. For HfO2, n is approximately 2 [40] so the optical dielectric constant is around 4. In the previous section, we estimated the static dielectric constant to be 20~21 at 1 MHz. The dynamic constant for the data in the graphs below is approximately 4-5. From these linear fits, we extracted the barrier height at the HfO2/ITO interface. The range of data is approximately 1.1 0.2 eV.

PAGE 38

27 Despite what appears to be a reasonable schottky fit, some of the samples examined had dynamic dielectric constants less than 4, which are not consistent with the schottky process. However, this phenomenon was observed in the films deposited at higher substrate temperature and/or high oxygen pressure, thus yielding high leakage current. For these films the conduction mechanism may be influenced by the schottky emission process, another mechanism is also significant. For example, the contribution from current through grain boundaries, if crystallization takes place, would cause the measured dynamic dielectric constant to be small [33] It is probable that films deposited at the higher temperature and/or oxygen pressure have higher degrees of crystallinity that would contribute to the overall leakage current and may cause deviation from the schottky model. The data in following sections supports this explanation. It will be seen that crystallinity is evident and increases with higher substrate temperature and oxygen pressure. 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0246Gate Voltage (V)Current Density (A/cm2) 100 C 150 C 200 C 250 C Figure 7: Leakage current of HfO2 on ITO measured at various temperatures. The sample was sputtered with 5 mTorr O2 and 100C substrate temperature.

PAGE 39

28 -30 -29 -28 -27 -26 -25 -24 -23 -22 -21 -20 2000400060008000E0.5 (V/m)0.5ln(J/T2) [ln(AK-2cm-2)] 100 C 150 C 200 C 250 C(a) -34 -33 -32 -31 -30 -29 -28 -27 -26 50007000900011000E0.5 (V/m)0.5ln(J/T2) [ln(Acm-2K-2)] 25 C 150 C 200 C 250 C ( b ) Figure 8: Schottky emission plots of HfO2 on ITO measured at different temperatures. The samples was sputtered with different deposition conditions: (a) 5 mTorr O2 and 100C, and (b) 5 mTorr O2 and 25C

PAGE 40

293.3.3 Leakage Current The effects of substrate temperature and oxygen partial pressure on the leakage current behavior of HfO2 on ITO were also examined. Leakage currents as low as 8 x 1010 A/cm2 were obtained at 1 V for HfO2 films grown at 5 mTorr O2 and 25C. Other films displayed values on the order of 10-7 A/cm2. Figure 9 shows the effect of varying substrate temperature (25C, 100C, and 200C), while the oxygen pressure remains the same. As a general trend, increasing substrate temperature caused an increase in leakage current. In the 1 mTorr O2 case, it is difficult to differentiate whether the current increase between the 100C and 200C films is due to the effect of substrate temperature or the reduction of film thickness. However, those films grown with 5 mTorr and 10 mTorr O2 show a clear degradation in leakage current properties as the substrate temperature is increased. In fact, this increase in leakage current between the films exists despite an increase in film thickness. We believe that the increasing leakage current results from increasing crystallinity as the films are grown under higher substrate temperature.

PAGE 41

30 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 1E-1 1E+0 0246Gate Voltage (V)Current Density (A/cm2)25oC 100oC 200oC 1E-11 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 024Gate Voltage (V)Current Density (A/cm2) 25oC 200oC 100oC 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 1E-1 1E+0 0246Gate Voltage (V)Current Density (A/cm2)25oC 100oC 200oC Figure 9: Room temperature leakage current plots of HfO2 on ITO showing the effect of substrate temperature. Each plot shows a different set of deposition oxygen pressures: (a) 1 mTorr O2, (b) 5 mTorr O2, (c) 10 mTorr O2. (a) (b) (c)

PAGE 42

31 Figure 10 shows the effect of varying oxygen pressure (1 mTorr, 5 mTorr, 10 mTorr), while the substrate temperature remains the same. For films grown with room temperature or 100C substrates, leakage current is seen to decrease when the oxygen pressure is increased from 1 mTorr to 5 mTorr. However, the leakage current then increased when the oxygen pressure is further increased to 10 mTorr. One possible explanation for the observed oxygen pressure dependence is that the films are not fully oxidized at 1 mTorr O2. The increase of oxygen to 5 mTorr O2 oxidizes the films, resulting in a leakage current decrease. However, when the oxygen pressure is increased further, the leakage current increases due to the contribution of film crystallization. The trend is more obvious for films grown at 200C substrate temperature. The leakage current is monotonically increased by additional oxygen pressure due to increasing crystallization. Figure 11 shows the room temperature leakage current for each of the nine samples studied measured at 2 V. It is easy to see the trend in leakage current as a function of growth conditions.

PAGE 43

32 1E-11 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 02468Gate Voltage (V)Current Density (A/cm2)1 mTorr O2 10 mTorr O2 5 mTorr O2 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 02468Gate Voltage (V)Current Density (A/cm2)10 mTorr O25 mTorr O21 mTorr O2 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 1E-2 1E-1 1E+0 0246Gate Voltage (V)Current Density (A/cm2)10 mTorr O25 mTorr O21 mTorr O2 Figure 10: Room temperature leakage current plots of HfO2 on ITO, showing the effect of oxygen pressure during film deposition. Each graph displays a different substrate temperature: (a) 25C, (b) 100C, (c) 200C. (a) (b) (c)

PAGE 44

33 1 5 10 25 100 200 1.00E-09 1.00E-08 1.00E-07 1.00E-06 1.00E-05 1.00E-04Leakage Current Density (A/cm2) Oxygen Pressure (mTorr)Substrate Temperature (deg. C) Figure 11: Leakage current of HfO2 on ITO measured at room temperature and 2 V versus growth conditions.

PAGE 45

343.3.4 Microstructure In addition to leakage current, we have also examined the microstructure of HfO2 films using X-ray diffraction (XRD). Figure 12 shows XRD results for HfO2 films grown on ITO with 10 mTorr O2 pressure. The emergence of the broad hump at 2 % 28.3 indicates the existence of crystallinity in the film deposited at 200C. The peak is the (-111) plane of the monoclinic HfO2 structure which is the most common HfO2 phase [19,21,22,41] The emergence of this peak supports the idea that enhanced crystallinity at higher oxygen pressure and substrate temperature yields the observed increase in leakage current. 20304050Intensity (arb.)ITO (222) ITO (400) ITO (441) HfO2M-(111)2 (deg.) 200C 25C 100C Figure 12: XRD diffractogram of HfO2 on ITO grown with 10 mTorr O2 and varying substrate temperature.

PAGE 46

35 In order to gain further information about the film microstructure, tapping mode atomic force microscopy (AFM) was used. The topography of the HfO2 films grown on ITO strongly resembles the topography of the ITO surface on the several micron scale. The polycrystalline grain structure of the ITO film is visible. The grains are about 0.5 micron in diameter and the rms roughness is about 3.416 nm. All of the samples examined by AFM showed the ITO surface structure on the HfO2 film surface. However, roughness differences in the films correlated with increasing oxygen pressure and substrate temperature. Figure 13 and Figure 14 plot this trend. The increase in roughness is consistent with an increase in crystallinity with high temperature and/or oxygen pressure during deposition. 0 2 4 6 8 10 12 0510Oxygen Pressure (mTorr)RMS Roughness (nm) Figure 13: RMS roughness increase measured by AFM as oxygen pressure was varied during 200C substrate temperature deposition.

PAGE 47

36 02468 1012 0100200Substrate Temperature (deg. C)RMS Roughness (nm) Figure 14: RMS roughness increase measured by AFM. The increase corresponds to increasing substrate temperature when the films were grown with 10 mTorr O2. AFM images at a smaller scale show grain structure in the films. F i g u r e 15 shows an A F M i m a g e of t h e f i l m g r o wn at 5 m Torr O2 and 200C. The figure is actually an image of the derivative of the height. It represents what the surface might look like to the eye or through a microscope. Figure 16 and Figure 17 have been enhanced to show the grain boundaries better. This is done by the application of a highpass filter. Each pixel is weighted according to the average of the surrounding pixels. There is a noticeable grain size increase from Figure 16 to Figure 17 which corresponds to the increase in substrate temperature from 100C to 200C. Since the highest substrate temperature gives the maximum surface mobility, this is consistent with island growth and coalescence [42]

PAGE 48

37 Figure 15: AFM derivative image of HfO2 grown on ITO at 5 mTorr O2 and 200C

PAGE 49

38 Figure 16: AFM image of HfO2 grown on ITO at 10 mTorr O2 and 100C with highpass filter applied to enhance grain edges.

PAGE 50

39 Figure 17: AFM image of HfO2 grown on ITO at 10 mTorr O2 and 200C with highpass filter applied to enhance grain edges.

PAGE 51

403.4 Conclusions In conclusion, the leakage current behavior for HfO2 films deposited by RF reactive sputtering onto ITO substrates with an Ar + O2 gas flow was examined. Oxygen pressure and substrate temperature were varied to optimize films for high dielectric constant and low leakage current. The dielectric constant of HfO2 films was calculated to be ~20-21 from 1 MHz capacitance measurements. There was no systematic variation of the dielectric constant according to deposition conditions. A low leakage current of 8 x 10-10 A/cm2 was obtained for a Al/HfO2/ITO structure sputtered at 25C substrate temperature and 5 mTorr O2 pressure when the film thickness was ~ 63 nm. Current-voltage curves were measured at various temperatures to study the leakage current mechanism under substrate electron injection. For most films deposited below 200C, the data fit the schottky emission mechanism well. The schottky barrier height at the ITO/HfO2 interface was extracted to be ~ 1.1 0.2 eV. Leakage current was increased by raising the substrate temperature; the lowest leakage currents were obtained with room temperature depositions. Increasing the oxygen amount from 1 mTorr O2 to 10 mTorr O2 also increased the leakage current. The increase in leakage current with increasing oxygen pressure and substrate temperature during deposition was attributed to increasing crystallinity. This assertion of increasing crystallinity is supported by XRD data as well as AFM imaging and roughness data obtained from the AFM study. However, for films grown at temperature less than 200C, the leakage current decreased when the oxygen pressure increased from 1 mTorr to 5 mTorr. To explain this, we asserted that films grown with 1 mTorr O2 were not fully oxidized. The increase in oxygen pressure from 1 mTorr to 5 mTorr resulted in further

PAGE 52

41 oxidation of HfO2 film, reducing the leakage current. The leakage current increased when the oxygen pressure was increased further because the increase in crystallinity then became the dominating factor. The crystallization of the films was supported by increasing rms roughness values as substrate temperature and oxygen pressure were increased. XRD seemed to indicate the existence of small grains as well. Likewise, AFM imaging shows the appearance of the grains in several samples. The crystallization of the films was also reflected by schottky emission fitting. The films grown at high oxygen pressure and temperature did not have self-consistent dynamic dielectric constants in the schottky model. Since the most crystalline films were not consistent with schottky dominated current transport, we inferred that the crystallinity of the films contributes significantly to the conduction transport mechanism of these HfO2 films.

PAGE 53

42 CHAPTER 4 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON SILICON 4.1 Introduction The scaling of the metal-oxide-semiconductor field-effect transistor (MOSFET) feature size to sub-100nm dimensions requires a decrease in thickness of the current thermal oxide SiO2 gate dielectric to less than 1 nm. At this thickness, direct tunneling leakage current is significant for SiO2. A solution to the problem is to find materials with high dielectric constants relative to SiO2. A material with a higher dielectric constant (high-k material) can be made thick enough to avoid direct tunneling and still maintain the capacitance of the thinner SiO2 film. HfO2, with a dielectric constant on the order of 20-25, is a promising alternative gate dielectric. HfO2 has good thermal stability on silicon and can consume native oxide to form HfO2. It is the first high-k material to show compatibility with the present complimentary metal-oxide-semiconductor (CMOS) polysilicon gate process. HfO2 has a high dielectric constant and relatively low leakage current. In this study, the properties HfO2 thin films were investigated as an alternative gate dielectric. HfO2 was deposited by reactive RF magnetron sputtering under varying oxygen partial pressures and substrate temperatures onto silicon substrates. Metal-oxidesemiconductor structures were fabricated on silicon to measure capacitance-voltage and current-voltage properties.

PAGE 54

434.2 Experimental Procedure 4.2.1 Substrate Preparation Reactive RF sputter deposition was used to deposit HfO2 thin films onto p-type Si (100) substrates. Substrates were cleaved to a size of approximately 1 cm x 1 cm. Each substrate was ultrasonically cleaned in trichloroethylene, acetone, methanol, and deionized water for five minutes each, then blown dry with nitrogen gas. Next, the silicon substrates were immersed in a 10:1 H2O:HF solution for 45 seconds, then rinsed twice successively in deionized water to remove native oxide and to hydrogen-terminate the substrate surface. The substrates were again blown dry with nitrogen. All substrates were mounted onto a platen with silver paint and loaded into the sputtering system. A portion of each substrate surface was covered in silver paint for thickness measurement after growth. The chamber was pumped down to a base pressure between 5 x 10-8 and 5 x 10-7 Torr. 4.2.2 Film Growth HfO2 films were grown by reactive RF magnetron sputtering from a 2 inch diameter 99.99% pure hafnium target. The RF power was maintained at 100 W. Oxygen partial pressure was varied from 1 mTorr to 10 mTorr, while the total pressure (Ar + O2) was kept at 30 mTorr. Substrate temperature was varied from room temperature to 200C. There was a total of nine combinations of deposition conditions, as shown in Table 3 Film thickness was 60-85 nm for all samples. After loading the substrates into the deposition chamber, the temperature was ramped up to the appropriate setting. Meanwhile, the hafnium target was pre-sputtered

PAGE 55

44 with the shutter closed to remove any oxide from the surface. The shutter was opened, and oxygen was slowly flowed into the system. Table 3: Film thicknesses of HfO2 samples on p-type silicon determined by stylus profilometry and ellipsometry. Substrate Temperature 25C 100C 200C 1 mTorr 62.5 nm 65.0 nm 63.0 nm 5 mTorr 63.0 nm 59.0 nm 83.0 nm Oxygen Pressure 10 mTorr 62.0 nm 65.0 nm 71.5 nm 4.2.3 Measurements The thickness of the films was measured by stylus profilometry and ellipsometry using the Filmetrics F20 thin-film measurement system. After the thickness was determined, samples were prepared for electrical property measurement. Aluminum electrodes with area # *10-4 cm2 were sputtered as the top electrode using a shadow mask with a 200 micron diameter circle pattern. The thickness of these electrodes was approximately 150-200 nm. The backside of the silicon substrate was mounted onto a glass slide with conductive silver paint, the bottom electrode. The current-voltage properties and capacitance-voltage properties of HfO2 were measured on silicon. The current was measured under substrate electron injection. The capacitance-voltage curve was obtained at 1 MHz. The voltage was swept from depletion to accumulation and then back to depletion at sweep rate of 1.6 V/s. Figure 18 is a schematic cross-section of the metal-insulator-semiconductor structure used to make electrical measurements in this work.

PAGE 56

45 In addition to electrical measurements, x-ray diffraction (XRD) and atomic force microscopy (AFM) were utilized to examine the film microstructure. Films without deposited electrodes were scanned with the Philips APD 3720 diffractometer, and then examined by tapping mode AFM on the Digital Instruments Dimension 3100. The roughness, crystallinity, and grain structure were obtained. Figure 18: Drawing of MIS structure with interface layer Silicon HfO2 Al Ag HfxSiyOz

PAGE 57

464.3 Results and Discussion 4.3.1 Microstructure X-ray diffraction (XRD) was used to investigate the formation of polycrystalline grains as a function of processing conditions. The XRD results shown in Figure 19 through Figure 21 show that crystallinity is evident in the HfO2 films. The monoclinic peak appears at 2 % = 28.3 for most films. The intensity of this peak increases for higher oxygen pressure and substrate temperature used during deposition. This is undesirable for gate dielectric applications as grain boundaries yield paths for enhanced leakage current. 1020304050Intensity (arb.)2 (deg.)Substrate M-(-111)10 mTorr O25 mTorr O21 mTorr O2 Figure 19: XRD diffractogram of HfO2 films grown on silicon at 200C substrate temperature, showing effect of oxygen pressure on crystallization.

PAGE 58

47 1020304050Intensity (arb.)2 (deg.)Substrate10 mTorr O25 mTorr O21 mTorr O2M-(-111) Figure 20: XRD diffractogram of HfO2 films grown on silicon at 25C substrate temperature, showing the effect of oxygen pressure on crystallization. 1020304050Intensity (arb.)2 (deg.)Substrate M-(-111)200C 100C 25C Figure 21: XRD diffractogram of HfO2 films grown on silicon with 5 mTorr O2 pressure, showing the effect of substrate temperature on crystallization.

PAGE 59

48 AFM studies were also performed for several films grown on silicon. Figure 22 and Figure 23 show the trend of increases roughness as oxygen pressure and substrate temperature is increased. This is indicative of increasing crystallization. 0 2 4 6 8 0100200Substrate Temperature (deg. C)RMS Roughness (nm) Figure 22: RMS roughness as a function of substrate temperature for HfO2 on Si deposited at 10 mTorr O2. 0 2 4 6 8 0510Oxygen Pressure (mTorr)RMS Roughness (nm) Figure 23: RMS roughness as a function of oxygen pressure for HfO2 on Si deposited at 200C.

PAGE 60

49 Figure 24 through Figure 28 show three-dimensional views of the HfO2 film surfaces grown at different temperatures and oxygen pressures. At 10 mTorr O2 and 200C, the grains are largest and completely cover the scan surface. The figures show that the grain size decreases and the films get smoother with lowering temperature and oxygen pressure. This is because surface diffusivity is the highest at high substrate temperatures. Accordingly, the growth rate of the films decreases 10-fold when the oxygen pressure changes from 1 mTorr O2 to 5 or 10 mTorr O2. This is because the target is oxidized at the high pressures. The slow deposition rate combined with the high substrate temperature allows islands to grow larger since the surface diffusivity is high. The result of this is large grains at high temperature and oxygen pressure. Figure 24: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 200C.

PAGE 61

50 Figure 25: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 100C.

PAGE 62

51 Figure 26: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10 mTorr O2 and 25C.

PAGE 63

52 Figure 27: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 5 mTorr O2 and 200C.

PAGE 64

53 Figure 28: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 1 mTorr O2 and 200C. 4.3.2 Dielectric Constant The dielectric properties of the HfO2 films were determined using capacitance measurement with an applied voltage sufficient for accumulation. Overall, the dielectric constant slightly increased with increasing substrate temperature and oxygen pressure. This probably reflects the increase in crystallinity. The behavior is complicated by the combined effects of increasing crystallinity, interface growth and composition, and the presence of mobile ions. The dielectric constant decrease in the 1 mTorr series going from 100 to 200C in Figure 29 may be attributed to the growth of the interface layer. The decrease in dielectric constant in Figure 30 going from 1 mTorr to 5 mTorr O2 at 25C could result from the oxidation of the film. However, the effect of the

PAGE 65

54 crystallization dominates most of the data as the dielectric constant increases with increasing crystallinity. 10 15 20 25 30 35 0100200Dielectric Constant 1 mTorr Oxygen 5 mTorr Oxygen 10 mTorr OxygenSubstrate Temperature (degrees C) Figure 29: Dielectric constant (calculated from accumulation capacitance) as a function of growth conditions for HfO2 grown on silicon

PAGE 66

55 10 15 20 25 30 35 0510Oxygen Pressure (mTorr)Dielectric Constant 25 C 100 C 200 C Figure 30: Dielectric constant (calculated from accumulation capacitance) as a function of growth conditions for HfO2 grown on silicon 4.3.3 Leakage Current Behavior The leakage current properties are affected by the crystallinity and stoichiometry of the HfO2 film, as well as the quality and thickness of the interfacial layer at the Si/HfO2 boundary. The behavior of leakage current with increasing substrate temperature and oxygen pressure is shown in Figure 31 and Figure 32 Note that the interface layer is thickest when the films are grown at high oxygen pressure and substrate temperature. When the interface thickness is increased, leakage current of these films decreases. Some small deviations in the trend can be explained by changes in overall film thickness.

PAGE 67

56 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0123456Gate Voltage (V)Current Density (A/cm2)25C 200C 100C Figure 31: Room temperature leakage current measurement of HfO2 on silicon showing the decrease in leakage current that corresponds with increasing substrate temperature during deposition. The graphs show the trend for films grown with (a) 5 mTorr and (b) 10 mTorr oxygen pressure. 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0246810Gate Voltage (V)Current Density (A/cm2)25C 200C 100C(a) (b)

PAGE 68

57 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0123Gate Voltage (V)Current Density (A/cm2)1 mTorr O25 mTorr O210 mTorr O2 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 012345Gate Voltage (V)Current Desnity (A/cm2)1 mTorr O25 mTorr O210 mTorr O2 Figure 32: Room temperature leakage current measurement of HfO2 on silicon showing the decrease in leakage current that corresponds with increasing oxygen pressure during deposition. The graphs show the trend for films grown at (a) 100C and (b) 200C substrate temperature. (a) (b)

PAGE 69

58 The leakage current for the films grown at 25C more generally reflect the properties of the HfO2 film as minimal interfacial Si-Hf-O is formed. When the films were grown with 1 mTorr O2, leakage current increased when substrate temperature increased. When the films were grown with 25C substrate temperature, the leakage current increased when the oxygen pressure was increased. These trends are shown in Figure 33 The lowest leakage current at 2 V was measured for the film grown with 1 mTorr O2 at 25C. This is probably because of the increase in crystallinity, which has a larger impact on leakage current than the interface growth in this particular case.

PAGE 70

59 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0123Gate Voltage (V)Current Density (A/cm2)25C 200C 100C 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 01234Gate Voltage (V)Current Density (A/cm2)1 mTorr O25 mTorr O210 mTorr O2 Figure 33: Room temperature leakage current of HfO2 on silicon showing (a) effect of substrate temperature when films are grown with 1 mTorr O2 and (b) effect of oxygen pressure when films are grown at 25C (a) (b)

PAGE 71

604.3.4 Current Mechanism In order to begin discussion of possible current mechanism combinations involved in the complicated current transport through our structure, one needs to review several transport mechanisms. In schottky emission, carriers need to obtain enough energy to be transported over the schottky barrier into the next material. Current governed by schottky emission follows the form of the equation below [33] 2 / 1 0 3 24 1 expdE q T k T A J (4.1) Tunneling occurs when charge transport occurs through an insulating layer by the penetration of carriers horizontally (at constant energy) through a barrier, rather than surmounting enough energy to be transported over the barrier (as in schottky emission). Tunneling is a quantum mechanical effect and requires a sufficient electric field. Tunneling is described by equation 4.3 [36] 2 / 3 2 / 1 2 2) ( 3 ) 2 ( 8 exp 8B B Tq E q h m h E q J (4.2) Third, Poole-Frenkel emission, is a bulk transport mechanism. Charge is trapped in the insulator by impurity levels. The charge can be transported to the insulator conduction band by internal emission processes. When a field is applied, the potential well associated with the trap is distorted, increases the chances for the trapped electron to escape. Equation 4.4 below describes the poole-frenkel conduction mechanism [36] 2 / 1 0 31 exp expd t PFE q T k T k q E c J (4.3)

PAGE 72

61 The three mechanisms listed above are commonly seen in oxides, including HfO2. The leakage current behavior of HfO2 grown on silicon can be explained by combining the contributions of several mechanisms. Figure 34: Schematic E vs. x band diagram of the Al/HfO2/SiO2/p-type Si structure in this experiment The leakage current properties of HfO2 in direct contact with ITO has been reported in Chapter 3 We concluded that the current transport was governed by schottky emission at the ITO/HfO2 interface under substrate carrier injection. On the contrary, HfO2 film does not grow in direct contact with silicon. An interface layer is formed between the HfO2 film and the silicon substrate, most likely SiO2 or some HfxSiyOz. Now, as seen by the band diagram in Figure 34 above, the barrier height makes it less likely for schottky emission to occur from silicon to the interface layer than it would be for schottky emission to occur from silicon to HfO2 in the absence of an interface. Our data shows that schottky emission is not the dominant mechanism for the films on silicon. Si SiO2 HfO2 Al EV EV EV EC EC EC EF

PAGE 73

62 In fact, there is probably a complicated assortment of mechanisms that control the current transport in the Al/HfO2/SiO2/Si structure. We would like to suggest a conduction mechanism for our samples, when electrons were injected from the substrate. The injected electrons tunnel through the interfacial layer into bulk trap levels inside the HfO2. Therefore, the conduction in these films combines the tunneling through the interface and the poole-frenkel type in the HfO2. The rapid discontinuous leakage current increase observed at an applied voltage of 3-5 V (not shown) in some of the samples grown with 1 mTorr O2 may be caused by the breakdown of the thin HfxSiyOz layer at the interface with silicon. This rapid increase is not observed in the films grown on ITO (in chapter 3 ). The mechanism described above was also believed to occur in Ta2O5 films on silicon [33] 4.4 Conclusions HfO2 films were reactive RF sputtered onto p-type Si (100) substrates with an Ar + O2 gas flow. Oxygen pressure and substrate temperature were varied to determine their effect on the dielectric and microstructural properties of HfO2 on silicon. AFM and XRD were used to characterize film microstructure. XRD peaks grew more intense with higher substrate temperature and oxygen pressure. In addition, AFM shows that grains were largest and most prevalent in films grown at high oxygen pressure and substrate temperature. The dielectric constant was calculated for each sample based on the measure accumulation capacitance and film thickness. Dielectric constant generally increased with increasing substrate temperature and oxygen pressure due to the crystallinity. The exception is for films grown at low temperature where the dielectric constant decreases

PAGE 74

63 when the oxygen amount is increased from 1 to 5 mTorr. Apparently, the films are not fully oxidized under low oxygen pressure at low temperature. At low temperature, the oxidation of the film and/or the growth of an interface layer had a larger effect on the dielectric constant than the film crystallization. Low leakage currents in high temperature and/or high oxygen pressure films is due to the effect of the HfO2/Si interfacial layer. The current transport mechanism in these films was speculated to be a combination of tunneling through the interface layer, and then poole-frenkel and grain boundary transport through the film.

PAGE 75

64 CHAPTER 5 SUMMARY HfO2 thin films were deposited onto indium-tin-oxide (ITO) and p-type silicon (100) substrates using reactive RF sputter deposition. Substrate temperature and the amount of oxygen used during deposition were varied to determine their effect on leakage current and microstructure. X-ray diffraction (XRD) and atomic force microscopy (AFM) were both used for microstructural characterization. Crystallinity was determined to be present in films on both substrates and increases with increasing oxygen pressure and substrate temperature. Al/HfO2/ITO structures were used to measure leakage current and dielectric constant as a function of deposition conditions. Temperature-dependent leakage current measurements allowed for the extraction of the HfO2/ITO schottky barrier height. The schottky barrier height at the ITO/HfO2 interface was extracted to be ~ 1.1 0.2 eV. However, the films grown at high oxygen pressure and substrate temperature were not consistent with the schottky mechanism because of the grain boundary contribution. In general, leakage current on ITO was increased by increasing the oxygen pressure and substrate temperature because of enhanced crystallization. However, at low temperatures the films may not have been fully oxidized at 1 mTorr oxygen pressure. Therefore, the leakage current was improved by more oxygen (5 mTorr) and then subsequently degraded by the addition of excessive oxygen (10 mTorr). The dielectric constant of HfO2 films was calculated to be ~20-21 from 1 MHz capacitance measurements on ITO.

PAGE 76

65 In addition, metal-oxide-semiconductor (MOS) structures were fabricated on silicon with aluminum electrodes. The MOS structures were used to measure currentvoltage and capacitance-voltage characteristics as a function of deposition conditions. The dielectric constant was calculated for each sample based on the measure accumulation capacitance and film thickness. Dielectric constant generally increased with increasing substrate temperature and oxygen pressure due to the crystallinity. The exception is for films grown at low temperature where the dielectric constant decreases when the oxygen amount is increased from 1 to 5 mTorr. Apparently, the films are not fully oxidized under low oxygen pressure at low temperature as was seen from leakage current measurements on ITO as well. The leakage current for the Al/HfO2/Si structures decreased for high substrate temperature and/or oxygen pressure during depositon. This effect is attributed to the growth of the Si/HfO2 interfacial layer. The current transport mechanism in these structures on silicon was speculated to be a combination of tunneling through the interface layer, and then poole-frenkel and grain boundary transport through the film.

PAGE 77

66 REFERENCES 1. K. Cho, Computational Materials Science 23, 43-7 (2002). 2. B. Cheng, M. Cao, R. Rao, A. Inani, P. V. Voorde, W. M. Greene, J. M. C. Stork, Z. Yu, P. M. Zeitzoff, and J. C. S. Woo, IEEE Transactions on Electron Devices, 46, 1537 (1999). 3. J. Robertson, J. Vac. Sci. Technol. B 18, 1785 (2000). 4. J. Robertson, J. of Non-Crystalline Solids 303, 94-100 (2002). 5. J. S. Suehle, E.M. Vogel, M. D. Edelstein, C. Richter, N. V. Nguyen, I. Levin, D. L. Kaiser, H. Wu, and J. B. Bernstein, Amer. Vac. Soc. 2001 6th International Symposium on Plasma Process-Incuced Damage, Monterey, CA, USA 90 (2001). 6. A. Paskaleva, E. Atanassova, T. Dimitrova, Vacuum 58, 470-7 (2000). 7. T. P. Ma, IEEE, 6th International Conference on Solid-State and IntegratedCircuit Technology, 2001. Proceedings., 1, 297 (2001). 8. J. S. Lee, S. J. Chang, J. F. Chen, S. C. Sun, C. H. Liu, U. H. Liaw, Materials Chemistry and Physics 77, 242-7 (2002). 9. W. H. Ha, M. H. Choo, S. Im, J. of Non-Crystalline Solids 303, 78-82 (2002). 10. M. D. Groner, J. W. Elam, F. H. Rabreguette, S. M. George, Thin Solid Films 413, 186-197 (2002). 11. G. C. Yeap, S. Krishnan, and Ming-Ren Lin, Electronics Letteres, 34, 1150 (1998). 12. W.-J. Qi, R. Nieh, B. H. Lee, L. Kang, Y. Jeon, K. Onishi, T. Ngai, S. Banerjee, and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest., 145-8 (1999). 13. P. S. Lysaught, P. J. Chen, R. Bergmann, T. Messina, R. W. Murto, H. R. Huff, J of Non-Crystalline Solids 303, 54-63 (2002).

PAGE 78

67 14. L. Kang, Y. Jeon, K. Onishi, B. H. Lee, W.-J. Qi, R. Nieh, S. Gopalan, and J. C. Lee, 2000 Symp. On VLSI Technology Digest of Technical Papers, 44-45 (2000). 15. L. Kang, K. Onishi, Y. Jeon, B. H. Lee, C. Kang, W.-J. Qi, R. Nieh, S. Gopalan, R. Choi, and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest., 358 (2000). 16. S. J. Lee, H. F. Luan, W. P. Bai, C. H. Lee, T. S. Jeon, Y. Senzaki, D. Roberts, and D. L. Kwong, Int. Electr. Devices Meeting. IEDM Technical Digest, 31-4 (2000). 17. Y. Kim, G. Gebara, M. Freiler, J. Barnett, D. Riley, J. Chen, K. Torres, J. Lim, B. Foran, F. Shaapur, A. Agarwal, P. Lysagth, G. A. Brown, C. Young, S. Borthakur, H.-J. Li, B. Nguyen, P. Zeitzoff, G. Bersuker, D. Derro, R. Bergmann, R. W. Murto, A. Hou, H. R. Huff, E. Shero, C. Pomarede, M. Givens, M. Mazanec, and C. Werkhoven, Int. Electr. Devices Meeting. IEDM Technical Digest, 455-8 (2001). 18. A. M. Stoneham, J. of Non-Crystalline Solids 303, 114-122 (2002). 19. C.T. Kuo and R. Kwor, Thin Solid Films 213, 257-64 (1992). 20. B.-H. Lee, L. Kang, W.-J. Qi, R. Nieh, Y. Jeon, K. Onishi and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest, 133-6 (1999). 21. S.-W. Nam, J.-H. Yoo, S. Nam, H.-J. Choi, D. Lee, D.-H. Ko, J. H. Moon, J.-H. Ku, S. Choi, J. of Non-Crystalline Solids 303, 139-143 (2002). 22. S. Xing, N. Zhang, Z. Song, Q. Shen, C. Lin, Microelectronic Engineering 1 66, 451-6 (2002). 23. B. H. Lee, R. Choi, L. Kang, S. Gopalan, R. Nieh, K. Onishi, Y. Jeon, W.-J. Qi, C. Kang and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest, 3942 (2000). 24. C. T. Hsu, Y. K. Su and M. Yokoyama, Jpn. J. Appl. Phys. 31, 2501-4 (1992). 25. R. Choi, C. S. Kang, B. H. Lee, K. Onishi, R. Nieh, S. Gopalan, E. Dharmarajan, and J. C. Lee, Symposium on VLSI Technology Digest of Technical Papers, 15-6 (2001). 26. M. Gutowski, J. E. Jaffe, C.-L. Liu, M. Stoker, R. I. Hegde, R. S. Rai, and P. J. Tobin, Applied Physics Letters 80, 11 (2002).

PAGE 79

68 27. P. D. Kirsch, C. S. Kang, J. Lozano, J. C. Lee, J. G. Ekerdt, J. Applied Physics 91, 7 (2002). 28. B. H. Lee, L. Kang, R. Nieh, W.-J. Qi, and J. C. Lee, Applied Physics Letters 76, 14 (2000). 29. A. Callegari, E. Cartier, M. Gribelyuk, H. F. Okorn-Schmidt, and T. Zabel, J. Applied Physics 90, 12 (2001). 30. L. Kang, B. H. Lee, W.-J. Qi, Y. Jeon, R. Nieh, S. Gopalan, K. Onishi, and J. C. Lee, IEEE Electron Device Letters 21, 181 (2000). 31. K. L. Ng, N. Zhan, M.C. Poon, C. W. Kok, M. Chan, and H. Wong, Electron Devices Meeting, Hong Kong. Proceedings. 51-4 (2002). 32. N. Zhan, K. L. Ng, M. C. Poon, C. W. Kok, M. Chan, and H. Wong, Electron Devices Meeting, Hong Kong. Proceedings. 43-6 (2002). 33. E. Atanassova, N. Novkovski, A. Paskaleva, M. Pecovska-Gjorgjevich, SolidState Electronics 46, 1887-98 (2002). 34. B. K. Park, J. Park, M. Cho, C. S. Hwang, Applied Physics Letters 80, 13 (2002). 35. W. J. Zhu, T.-P. Ma, T. Tamagawa, J. Kim, and Y. Di, IEEE Electron Device Letters 23, 97 (2002). 36. M. Ohring, The Materials Science of Thin Films Academic Press, San Diego, CA (1992). 37. J. George, Preparation of Thin Films Marcel Dekker, Inc., New York, NY (1992). 38. B. Chapman, Glow Discharge Processes John Wiley & Sons, New York, NY (1980). 39. C. K. Maiti, S. Maikap, S. Chatterjee, S. K. Nandi, S. K. Samata, Solid-State Electronics 47, 1995-00 (2003). 40. M. Fadel, O. A. Azim M., O. A. Omer, R. R. Basily, Appl. Phys. A 66, 335-43 (1998). 41. H. Ikeda, S. Goto, K. Honda, M. Sakashita, A. Sakai, S. Aima, and Y. Yasuda, Jp. J. Appl. Phys. 41, 2476-9 (2002).

PAGE 80

69 42. K. Wasa and S. Hayakawa, Handbook of Sputter Deposition Technology Noyes Publications, Westwood, NJ (1992).

PAGE 81

70 BIOGRAPHICAL SKETCH The author was born in Erie, Pennsylvania, and raised in Orlando, Florida, since the age of two. In high school, the author was an accomplished trumpet player, performing in ensembles such as the Florida All-State band and the Florida Symphony Youth Orchestra. The latter included a performance tour of Australia that culminated in concerts at Sydney Town Hall and the famous Sydney Opera House. After graduation from Lake Howell High School in Winter Park, Florida, he was accepted to the University of Florida in Fall 1998. He received a Bachelor of Science degree in materials science and engineering in June 2003. As an undergraduate, ceramics was his subject of specialization. Through the 3/2 degree program, the author was able to enroll in graduate courses while still in pursuit of his undergraduate degree. Thus, he graduated with a Master of Science degree in materials science and engineering in December 2003, only a few months after earning his Bachelor of Science. During the transition from undergraduate to graduate studies, he found a home in Dr. Norton’s electronic oxides group. The author will remain in Dr. Norton’s group to study copper diffusion barriers for his Ph.D. dissertation.


Permanent Link: http://ufdc.ufl.edu/UFE0001740/00001

Material Information

Title: Leakage Current Behavior of Reactive RF Sputtered HfO2 Thin Films
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0001740:00001

Permanent Link: http://ufdc.ufl.edu/UFE0001740/00001

Material Information

Title: Leakage Current Behavior of Reactive RF Sputtered HfO2 Thin Films
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0001740:00001


This item has the following downloads:


Full Text












LEAKAGE CURRENT BEHAVIOR OF REACTIVE RF SPUTTERED
HfO2 THIN FILMS













By

MICHAEL N. JONES


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003






























Copyright 2003

By

Michael N. Jones






























This document is dedicated to the memory of my grandfather, Louis R. Jones.
His support for me encouraged me to complete this work.
I am glad I was able to make my grandfather proud.















ACKNOWLEDGEMENTS

First, I would like to thank my parents for encouraging me to attend the

University of Florida and to continue in pursuit of a higher education. I feel that I have

made the most of my college years, which can be attributed in part to my parental

support. Meanwhile, my sister, Kayla, has been a large part of my life and I am grateful

for that. I want to thank all of my friends for enriching my experience while at school,

and I thank Kathryn for her compassion and support.

I am very grateful to my research advisor, Dr. David Norton. Dr. Norton guided

and encouraged me throughout this project. I would like to express gratitude to Dr.

Amelia Dempere at the Major Analytical Instrumentation Center (MAIC) for providing

me with monetary support and the opportunity to work with a great staff.

I would like to thank all of my group members, especially Byoung-Seong Jeon

and Yongwook Kwon. Byoung-Seong taught me how to use the sputtering system and

worked with me to accommodate both of our schedules. Yongwook helped me make

electrical measurements and interpret data. I would also like to thank Joshua Howard and

Dr. Singh's group for allowing me to use their CV/IV measurement system.















TABLE OF CONTENTS


Page

A CKN OW LED GM EN TS ........................ ..... .................. ............. iv

LIST OF TABLES .......... ......... .......... .......... ........vii

LIST OF FIGURES .......... .... .......... ......... .... ................ viii

ABSTRACT ......... .................... .............. ............xi

CHAPTER

1 IN T R O D U C T IO N .......... ................................................................ .. .. ...... ....

1.1 Need for Alternate High-k Dielectric Material on Silicon..................................... 1
1.2 D position on ITO ......................................... ............................... ............ 2

2 B A CK G R O U N D ........................................................ ............ .................. .. 4

2.1 Introduction to High-k Gate Dielectrics on Silicon .............................................. 4
2.2 Review of Candidate M materials ....................................................... ........ .. .... 8
2.3 Literature Review of HfO2 Thin Film ........... ............................... .............. 10
2.3.1 Structure ... ...................... ... .. .............. ............. 10
2.3.2 Processing and Properties ........................................ ................... ...... 12
2.3.3 MOSFET Device Compatibility ....................................................... 15
2.4 Sputtering ............ .............................. ............... 15
2.4.1 D C Sputtering ............................. ................. ...... ........... 15
2.4.2 RF Reactive M agnetron Sputtering.............................................................. 18









3 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON INDIUM-
T IN -O X ID E (IT O ) ................................................................................ .............. .. 19

3.1 Introduction ...................................................................... ........ 19
3.2 Experim ental Procedure ........................................................... .............. 19
3.2.1 Substrate Preparation.................................................. ........................... 19
3.2.2 F ilm G row th .................................................................. .. ........ .... 20
3 .2 .3 M easurem ents.................................................. ............. ...................... 20
3 .3 R results and D iscu ssion .................................................................................. 22
3.3.1 D electric C onstant......................................................... .......................... 22
3.3.2 C onduction M echanism ................................................................ ............. 23
3.3.3 Leakage Current ......................................... ...... .. .. .... ............. 29
3 .3 .4 M icro stru ctu re ............................................................... ..................... 3 4
3.4 Conclusions ................................... ................................ ........ 40

4 LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2 ON SILICON... 42

4 .1 In tro du ctio n ................................................................................ 4 2
4.2 E xperim ental Procedure .......................................................... ............... 43
4.2.1 Substrate Preparation........................................................... .............. 43
4 .2 .2 F ilm G row th ................................................................ .. ...... ........ .. 43
4 .2 .3 M easurem ents.................................................. ............. ...................... 44
4 .3 R results and D iscu ssion ............................................................ .............. 46
4.3.1 M icrostructure .... .............. .......................................... .. ....... .............. 46
4.3.2 D electric C onstant......................................................... .......................... 53
4.3.3 Leakage Current B behavior ........................................... ......................... 55
4.3.4 Current M echanism ........................................................................ 60
4.4 C conclusions ...................................................................... ........ 62

5 SU M M ARY .................................... .......................... ..... ..... ........ 64

REFERENCES ................................. ............... 66

BIOGRAPH ICAL SKETCH ....................................................................................... 70















LIST OF TABLES


Table pae

1 Band gaps and band offsets on silicon for candidate high-k dielectrics [3,4]............. 9

2 Film thicknesses of HfO2 samples on ITO determined by stylus profilometry ........ 21

3 Film thicknesses of HfO2 samples on p-type silicon determined by stylus
profilom etry and ellipsom etry .......................................................................... ..... 44















LIST OF FIGURES


Figure page

1 Band diagram of semiconductor/insulator interface at flat band ............................ 7

2 H igh-k gate dielectric candidates ..................................................................... .. .. .

3 Schematic of a DC sputtering system .............. ..... ......................... ........... 17

4 Interactions with incident ions at the target surface............................................... 17

5 Schematic cross-section of the metal-insulator-ITO structure used in this work. The
aluminum electrode was 150-200 nm thick; the thickness of the HfO2 film, as
measured by profilometry, is shown in Table 2. ..................................................21

6 Schottky emission conduction mechanism. OB is the schottky barrier height. Note
the magnitude of (M, the energy required for thermionic emission from the metal. 25

7 Leakage current of HfO2 on ITO measured at various temperatures. The sample was
sputtered with 5 mTorr 02 and 1000C substrate temperature. .................................27

8 Schottky emission plots of HfO2 on ITO measured at different temperatures. The
samples was sputtered with different deposition conditions: (a) 5 mTorr 02 and
1000C, and (b) 5 mTorr 02 and 25 C.............................. ....................... .............. 28

9 Room temperature leakage current plots of HfO2 on ITO showing the effect of
substrate temperature. Each plot shows a different set of deposition oxygen
pressures: (a) 1 mTorr 02, (b) 5 mTorr 02, (c) 10 mTorr 02............................... 30

10 Room temperature leakage current plots of HfO2 on ITO, showing the effect of
oxygen pressure during film deposition. Each graph displays a different substrate
temperature: (a) 25C, (b) 1000C, (c) 200C. .............................................. 32

11 Leakage current of HfO2 on ITO measured at room temperature and 2 V versus
grow th conditions ......... .......................................... ................. .. 33

12 XRD diffractogram of HfO2 on ITO grown with 10 mTorr 02 and varying substrate
temperature....................... .......... ... ................................. 34









13 RMS roughness increase measured by AFM as oxygen pressure was varied during
200C substrate temperature deposition......... .... ................................. ............. 35

14 RMS roughness increase measured by AFM. The increase corresponds to increasing
substrate temperature when the films were grown with 10 mTorr 02 .................... 36

15 AFM derivative image of Hf02 grown on ITO at 5 mTorr 02 and 2000C .............. 37

16 AFM image of Hf02 grown on ITO at 10 mTorr 02 and 1000C with highpass filter
applied to enhance grain edges........................................................ .................... 38

17 AFM image of Hf02 grown on ITO at 10 mTorr 02 and 2000C with highpass filter
applied to enhance grain edges........................................................ .................... 39

18 Drawing of MIS structure with interface layer ............................... .............. 45

19 XRD diffractogram of Hf02 films grown on silicon at 2000C substrate temperature,
showing effect of oxygen pressure on crystallization..................... ...............46

20 XRD diffractogram of Hf02 films grown on silicon at 250C substrate temperature,
showing the effect of oxygen pressure on crystallization. .................................... .. 47

21 XRD diffractogram of Hf02 films grown on silicon with 5 mTorr 02 pressure,
showing the effect of substrate temperature on crystallization.............................47

22 RMS roughness as a function of substrate temperature for Hf02 on Si deposited at48

23 RMS roughness as a function of oxygen pressure for Hf02 on Si deposited at 2000C.
................................................ ......... .......... .............. .. ................. 4 8

24 Three dimensional AFM view of Hf02 surface grown on p-Si(100) at 10 mTorr 02
an d 2 0 0 0C ................... .......... .......... ......... .................................. 4 9

25 Three dimensional AFM view of Hf02 surface grown on p-Si(100) at 10 mTorr 02
and 100 C. ................... ......... ..................... ................................. 50

26 Three dimensional AFM view of Hf02 surface grown on p-Si(100) at 10 mTorr 02
an d 2 5 0C ................... .......... .......... ......... ................................... 5 1

27 Three dimensional AFM view of Hf02 surface grown on p-Si(100) at 5 mTorr 02
an d 2 0 0 0C ................... .......... .......... ......... .................................. 5 2

28 Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 1 mTorr 02
an d 2 0 0 0C ................... .......... .......... ......... .................................. 5 3









29 Dielectric constant (calculated from accumulation capacitance) as a function of
growth conditions for HfO2 grown on silicon............... ......................................... 54

30 Dielectric constant (calculated from accumulation capacitance) as a function of
growth conditions for HfO2 grown on silicon............... ......................................... 55

31 Room temperature leakage current measurement of HfO2 on silicon showing the
decrease in leakage current that corresponds with increasing substrate temperature
during deposition. The graphs show the trend for films grown with (a) 5 mTorr and
(b) 10 mTorr oxygen pressure .............. ..... ......... .............. 56

32 Room temperature leakage current measurement of Hf02 on silicon showing the
decrease in leakage current that corresponds with increasing oxygen pressure during
deposition. The graphs show the trend for films grown at (a) 1000C and (b) 200C
substrate tem perature.................................... ......... .. .............. .. ........ .... 57

33 Room temperature leakage current of HfO2 on silicon showing (a) effect of substrate
temperature when films are grown with 1 mTorr 02 and (b) effect of oxygen
pressure when films are grown at 25C .............. ..................................... ......... 59

34 Schematic E vs. x band diagram of the Al/HfO2/Si02/p-type Si structure in this
ex p erim en t ...................................... ................................. ............... 6 1
















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

LEAKAGE CURRENT BEHAVIOR OF REACTIVE RF SPUTTERED
HfO2 THIN FILMS

By

Michael N. Jones

December 2003

Chairman: David P. Norton
Major Department: Materials Science and Engineering

HfO2 is currently under consideration as a potential high-k material to replace

conventional SiO2. HfO2 thin films were deposited onto indium-tin-oxide (ITO) and p-

type silicon (100) substrates using reactive RF sputter deposition. Substrate temperature

and the amount of oxygen used during deposition were varied to determine their effect on

leakage current and microstructure. X-ray diffraction (XRD) and atomic force

microscopy (AFM) were both used for microstructural characterization. Al/HfO2/ITO

structures were used to measure leakage current and dielectric constant as a function of

deposition conditions. Temperature-dependent leakage current measurements allowed

for the extraction of the HfO2/ITO Schottky barrier height. In addition, metal-oxide-

semiconductor (MOS) structures were fabricated on silicon with aluminum electrodes.

The MOS structures were used to measure current-voltage and capacitance-voltage

characteristics as a function of deposition conditions.














CHAPTER 1
INTRODUCTION

1.1 Need for Alternate High-k Dielectric Material on Silicon

The scaling of the metal-oxide-semiconductor field-effect transistor (MOSFET)

feature size to sub-100 nm dimensions requires a decrease in thickness of the current

thermal oxide SiO2 gate dielectric to less than 1 nm. At this thickness, direct tunneling

leakage current is significant for SiO2. A solution to the problem is to find materials with

high dielectric constants relative to SiO2. A material with a higher dielectric constant

(high-k material) can be made thick enough to avoid direct tunneling and still maintain

the capacitance of the thinner SiO2 film. An alternative gate dielectric material should

have a high dielectric constant, thermal and chemical stability in contact with silicon,

high quality interfaces, and low leakage current. Some of the materials that researchers

have considered are Y203, A1203, SrTiO3 (STO), BaxSri-xTiO3 (BST), Ta205, and TiO2.

Materials like Y203, and A1203 do not provide a significant advantage over SiO2 because

their dielectric constants are not significantly higher. In contrast, ultrahigh-k materials

like STO and BST cause poor short channel effects due to the fringing field induced

barrier lowering effect. It has also been reported that some high-k dielectrics such as

Ta205, TiO2, and strontium titanate (STO) are unstable in direct contact with silicon and

have a high leakage current.

HfO2, however, with a dielectric constant on the order of 20-25, is a promising

alternative gate dielectric. HfO2 reportedly has good thermal stability on silicon and can









consume native oxide to form HfO2. It is the first high-k material to show compatibility

with the present complementary metal-oxide-semiconductor (CMOS) polysilicon gate

process. HfO2 has a high dielectric constant and relatively low leakage current.

In this study, HfO2 was investigated as an alternative gate dielectric. HfO2 was

deposited by reactive RF magnetron sputtering under varying oxygen partial pressures

and substrate temperatures onto silicon and indium-tin-oxide (ITO)-coated glass

substrates. Dielectric constant and leakage current properties were studied on ITO.

Metal-oxide-semiconductor structures were fabricated on silicon to measure capacitance-

voltage and current-voltage properties. The microstructures of these films were also

investigated as a function of deposition conditions.

1.2 Deposition on ITO

Most of the research effort to date on HfO2 has focused on the dielectric and

leakage current properties of films deposited directly on silicon. In this case, the

measured properties of HfO2 on silicon reflect both the intrinsic properties of the HfO2

film as well as the effect of a SiO2 interface layer. While consistent with the gate

dielectric application, this structure makes it difficult to extract fundamental properties on

the HfO2 film itself.

To this end, we have investigated the dielectric and leakage current properties of

HfO2 films deposited by reactive RF sputtering onto ITO-coated glass substrates. The

objective of this study is to understand the electrical properties of the HfO2 without

influence from an interface layer. ITO is a conducting oxide, so the dielectric constant of

the HfO2 can be measured without accounting for the effect of interfacial layers. The









effects of substrate temperature and oxygen pressure on dielectric constant, leakage

current, and film microstructure are also investigated.














CHAPTER 2
BACKGROUND

2.1 Introduction to High-k Gate Dielectrics on Silicon

Alternative, high dielectric constant materials are needed to replace SiO2 as the

gate oxide in metal-oxide-semiconductor (MOS) devices. The current device scaling

trend requires SiO2 film thickness thinner than 1 nm. SiO2 films thinner than 1 nm would

have direct tunneling leakage current greater than 1 A/cm2, which is unacceptable. A

gate dielectric forms a parallel-plate capacitor, where the capacitance is given by


C (2.1)


where A is the area of the capacitor, so is the permittivity of free space, &r is the dielectric

constant, and d is the dielectric thickness, respectively. The use of an alternative high-k

dielectric would allow for a thicker gate dielectric layer to be used, reducing the

tunneling current while maintaining a given capacitance. The thickness of SiO2 film that

would give a capacitance-voltage curve equivalent to the new gate oxide is called the

equivalent oxide thickness (EOT). A new gate dielectric should be scalable to 1 nm

EOT. There are several additional conditions that an alternative gate dielectric must meet

to replace thermally grown SiO2 on silicon. They are summarized in the list below.

1. The dielectric constant of the new material should be significantly higher than the
dielectric constant of SiO2 film (-3.9);

2. The new gate dielectric should be thermodynamically stable in contact with silicon;









3. The band gap of the material should be large, and the band offsets on silicon need to
be greater than 1 eV to act as a tunneling barrier for both electrons and holes;

4. The interface trap defect density should be on par with current SiO2;

5. The defect density within the oxide layer should be minimal;

6. A stable amorphous phase material is preferred to avoid grain boundary leakage
current paths;

7. The new material should require minimal change from the existing oxide fabrication
process;

8. The dielectric should be thermodynamically stable in contact with poly-silicon;

9. There should be a low diffusion constant for dopant atoms of poly-silicon; [1]

The last three conditions refer to the compatibility of the new dielectric oxide

with the current MOSFET device processing. After the oxide film is grown on a

semiconducting substrate, the gate material is subsequently deposited on top of the gate

dielectric material. Heavily-doped polycrystalline silicon (polysilicon) is currently used

as the gate material. Therefore, to meet the requirements for current MOSFET

processing, the material must meet requirements dictated by the oxide/polysilicon

interface in addition to the oxide/silicon interface [1]. If these conditions are not met, the

use of a metal gate could be used to solve this compatibility problem [2].

The band offset requirement is very important to minimize leakage current. The

barrier at both the conduction and valence band needs to be high to act as a barrier for

electrons and holes, respectively. A schematic band diagram of the

semiconductor/insulator interface is presented in Figure 1. One of the problems with the

band offset condition is that materials that have large band gaps tend to have low

dielectric constants. Figure 2 illustrates this trend. Therefore, it is important to choose a

material that has a relatively high dielectric constant and a band gap that is large enough









to give low leakage currents. Fortunately, the large band gap materials are usually

favored by the stability criterion [3, 4].

If the new material is not stable in contact with silicon, an interface layer (most

likely SiO2) will form. In some cases the interface layers are shown to improve the

density of interface traps, but they provide a serious obstacle for device scaling. Two

capacitors in series result in the overall capacitance given in equation 2.1.

1 1 1
S= + (2.1)
Ctot C1 C2

Since the interface layer will have a lower dielectric constant than the high-k dielectric, it

will limit the maximum achievable capacitance or the minimum achievable EOT. The

interface thickness must be strictly controlled so that the dielectric layer can be scaled to

EOT less than 1 nm. In summary, it is desirable to select a material with reasonably high

dielectric constant, large band offsets, and stability in contact with silicon. In the next

section, several candidate high-k dielectrics are reviewed in respect to the criteria that has

just been discussed.











Conduction Band


Conduction Band


Valence Band Offset


Band gap


Valence Band


Semiconductor


Insulator


Figure 1: Band diagram of semiconductor/insulator interface at flat band


SiO2
A*203




HfSi,4
ZrSi04
ZrO2
HfO2
S Y203
SiN4 Ta205

*La203
TiO2

0 10 20 30 40 50 60 70

Dielectric Constant, k


Figure 2: High-k gate dielectric candidates [3,4,5]









2.2 Review of Candidate Materials

Presently, many researchers are searching for alternative gate dielectric materials

to replace silicon dioxide. Since a higher dielectric constant means we can grow thicker

films to reduce leakage current, why not use dielectrics with the highest dielectric

constants? Many metal oxides and ferroelectric materials have been investigated as

candidate materials, but most of them are not stable in contact with silicon. Furthermore,

the dielectric constant of materials generally tends to increase as the band gap decreases,

making it difficult to select a material with a large band gap and dielectric constant [1].

For example, SrTiO3 thin films have a dielectric constant of approximately 300 at room

temperature but a band offset of nearly zero. Therefore, this material cannot insulate

against tunneling current [1].

Robertson calculates that the barrier heights for Ta205, SrTiO3, and PbTiO3 on

silicon are very small as shown in Table 1. Schottky emission for these films figures to

be quite leaky [3,4]. Accordingly, schottky emission has been shown to be the primary

mechanism involved in the current transport in Ta205 films [6,7]. Lee et al. report that

annealing at 500-800 C for 30 minutes is required to lower the leakage current of Ta205

films, which roughens the film surface and increases EOT [8].









Table 1: Band gaps and band offsets on silicon for candidate high-k dielectrics [3,4]

Conduction Band Valence Band

Dielectric Material Band gap (eV) Offset (eV) Offset (eV)

SiO2 9 3.5 4.4

Si3N4 5.3 2.4 1.8

SrTiO3 3.3 -0.1

PbTiO3 3.4 0.6

Ta205 4.4 0.3 3

TiO2 3.1 0

ZrO2 5.8 1.4 3.3

HfO2 6 1.5 3.4

A1203 8.8 2.8 4.9

Y203 6 2.3 3.6

La203 6 2.3

ZrSiO4 6 1.5 3.4

HfSiO4 6 1.5



On the other hand, some materials have reasonable leakage current properties but

do not offer a significant dielectric constant advantage over SiO2. A1203 (K-7.6) has

been shown to have good leakage current characteristics [9,10], but does not have a

significantly higher dielectric constant than SiO2. The same can be said for Si3N4 and

most silicates. A promising gate dielectric should offer high dielectric constants,

reasonable barrier heights, and stability on silicon.

In addition to the tradeoff between dielectric constant and band gap, researchers

have shown that materials with dielectric constant greater then 25 show degradation in

MOSFET turn-off/on characteristics [11]. This is caused by fringing fields from the gate









to source/drain regions which further induce fields from the source/drain to the channel.

Furthermore, gate control is weakened by this so-called fringing field induced barrier

lowering [2]. Materials like TiO2, Ta205, BST, and STO, have high dielectric constants,

but require the use of barrier layer to suppress barrier lowering [2,11], and to prevent

reaction and interdiffusion with the silicon substrate [12]. As discussed in the previous

section, the low dielectric constant interfacial layer will dominate the capacitance-voltage

characteristics and limit the achievable EOT.

ZrO2 and HfO2 promise to be useful as alternative gate dielectrics. Amorphous

ZrO2 has been grown, and has thin EOT, low leakage current, negligible frequency

dispersion, interface density on par with SiO2, small hysteresis, and excellent reliability

[12]. ZrO2 and HfO2 have a dielectric constant of about 25 and a large band gap. The

band gap of HfO2 is about 6 eV. ZrO2 and HfO2 are theoretically more stable against the

formation of SiO2 on silicon than other oxides [12]. Robertson predicts Zr02 and HfO2 to

have large band offsets on silicon of about 1.5 eV [3,4]. Both materials have been shown

to have good properties. In fact, successful transistors with ZrO2 and HfO2 have been

fabricated with the present polysilicon gate material [13,14,15,16,17].

2.3 Literature Review of HfO2 Thin Film

2.3.1 Structure

Most of the work on HfO2 has been focused on amorphous films to replace SiO2.

Several factors favor an amorphous dielectric. Amorphous materials do not contain grain

boundaries or dislocations that can trap charge and offer fast diffusion pathways for

leakage current. In addition, stresses in amorphous materials can be taken up by small

variations in the random network, where they may likewise be taken up by misfit









dislocations in a polycrystalline material. Amorphous structures also tend to minimize

electronically active defects, but may give rise to shallow traps. Epitaxial films can be

free of grain boundaries and defects, but they require more difficult and expensive

methods to process [18]. Therefore, amorphous films are usually preferred for gate

dielectrics. Therefore, the crystallization of HfO2 films is an important issue.

If an amorphous structure is desired, the problem of growing films that remain

amorphous during device processing is not trivial. In fact, this is important for HfO2.

Crystallization often occurs during growth and post-growth annealing [13,19,20,21,22].

The monoclinic phase is the stable HfO2 phase, while metastable phases like

orthorhombic, tetragonal, and cubic are occasionally observed.

Sputtering and chemical vapor deposition (CVD) give the films with the best

amorphous character. Additionally, thin HfO2 films seem to crystallize at higher

temperatures than thick HfO2 films. 20-30 nm reactive sputtered films show monoclinic

crystal structure even before they are annealed. Meanwhile, 20-30 nm films sputtered

from a HfO2 target were amorphous until 5000C annealing. The films were a mixture of

monoclinic and orthorhombic phases up to 7000C annealing where only the monoclinic

phase dominated [19]. However, thinner sputtered films of 18.5 nm [20] and 0.5 nm

thickness [23] remained amorphous up to 7000C annealing temperature. A hafnium

metal layer was sputtered for both films. The first proceeded with subsequent reactive

sputtering of the hafnium target on top of the hafnium seed layer, and the second ended

with the rapid thermal oxidation of the hafnium film. In another case, thin DC reactive

sputtered HfO2 remained amorphous until 6500C annealing [21]. The highest









crystallization temperature reported was for CVD HfO2 that reportedly remained

amorphous after gate activation annealing at 9500C for 60 s [16].

Currently, there is a lot of interest in atomic layer deposition (ALD) of HfO2

films. There have been some promising results using ALD as the deposition method,

including the processing of a MOSFET device [17]. However, the nanocrystallinity of

HfO2 films continues to be a concern [13].

2.3.2 Processing and Properties

Several deposition processes have been used to grow HfO2 films: atomic layer

deposition, pulsed laser deposition, jet vapor deposition, chemical vapor deposition,

sputtering of pure HfO2 targets, reactive sputtering of hafnium targets, and sputter

deposition of hafnium metal and subsequent oxidation, among others. Most of the

literature has been directed toward the sputtering techniques. Sputtering is a fairly simple

and trusted thin film deposition process, especially for the growth of amorphous films.

The process is discussed further in section 2.4.

Even though the dielectric constant of evaporated films is reported to be about 25

[24], most of the reviewed literature gave effective dielectric constant values between 12

and 21. The discrepancy between values of 12 and 21 is because of the interfacial layers

and is greatly influenced by deposition and annealing procedure. One of the biggest

factors in high-k gate dielectric growth is interface thickness and quality. It greatly

influences properties such as dielectric constant, minimum achievable EOT, leakage

current, and interface trap density. Fortunately, results of HfO2 on silicon show that high

quality interfaces can be formed without special silicon surface treatment, and with the

use of surface treatment an EOT of 7.1 angstroms has been achieved [25].









Gutowski et al. stated through theoretical and experimental evidence that the

HfO2/Si interface is even more stable with respect to silicides than the ZrO2/Si interface

[26]. However, in most cases reviewed, the growth of an interface layer, usually hafnium

silicate, is observed. Several factors including deposition method, oxygen pressure,

deposition temperature, annealing time/temperature, and annealing ambient determine

the thickness and quality of the interface. Even though some researchers have shown that

the hafnium silicate interface layer is beneficial to reduce the interface trap density

[27,28], interface thickness control is a key issue in gate dielectric processing because it

limits EOT scaling. Several researchers have deposited HfO2 by reactive sputtering

[15,20,21,29,30,31,32]. In each case, the thickness of the interface layer increases with

increasing oxygen pressure and post-deposition annealing temperature. Callegari et al.

annealed HfO2 films in 02 at 6000C, then measured the thickness of the interface with

transmission electron microscopy (TEM). After correcting for the interface thickness, the

dielectric constant of Hf02 was 21 [29].

For thin HfO2 films, annealing lowers the leakage current [19,32]. It is also used

to fully oxidize and densify films that are deposited with hafnium metal targets instead of

reactive sputtering [23,25]. Fully oxidized films have better leakage current, and denser

films have higher dielectric constants [19]. For this reason, reactive sputtered films and

those sputtered from HfO2 targets tend to have better electrical properties than the films

that are oxidized after hafnium metal deposition. However, if the annealing conditions

are optimized, it is possible to scale post-deposition oxidized HfO2 below 1 nm more

easily than reactive sputtered films [28].









However, while leakage current and dielectric properties have been shown to

improve with annealing of thin oxide films, the opposite may be true with thicker films.

In the case of Ta205, where the main conduction mechanisms were the poole-frenkel

effect and the schottky effect, 20 nm films were improved by oxygen annealing while 80

nm films were severely degraded by oxygen annealing [33]. This phenomenon may be

caused by the large effect an oxide/Si interfacial layer can have on leakage current,

particularly when the oxide film is relatively thin.

The major drawbacks to annealing thin films are that excessive annealing also

causes more growth of the interfacial layer and may lead to crystallization of the HfO2

film. The scaling will become more difficult due to the increase in interface thickness

[22,31,32,34]; the crystallization can lead to increased leakage current. There are a few

steps to curb this interfacial effect. Annealing in N2 instead of 02 shows a more drastic

increase in leakage current and a smaller effect on the interface [15,21,32]. For example,

two similar HfO2 films were annealed in N2 and 02. For the film annealed in N2, EOT is

approximately 1.9 nm, while annealing in 02 results in EOT of 2.8 nm [21].

Additionally, surface preparation of the silicon substrate by annealing in NH3 or N2

suppresses the growth of an interface layer [25,27]. Nitrogen passivates the surface of

the silicon substrate, allowing an EOT as low as 7.1 angstroms to be achieved [25].

All of the studies reviewed by this author reported very low leakage current

values after annealing in either 02 or N2. There has not been much study into the leakage

current mechanism of HfO2. In one study, the Fowler-Nordheim tunneling mechanism

was observed in sputtered films 20-30 nm thick after annealing [19]. In another study,

Zhu et al. used a Fowler-Nordheim experiment at low temperature to extract band offsets.









The HfO2/Si conduction band offset was calculated to be approximately 1.13 + 0.13 eV.

The calculated Pt/HfO2 offset is 2.48 eV, and the Al/HfO2 offset is 1.28 eV. A trap level

1.5 eV below the HfO2 conduction band was calculated as well, which would contribute

to poole-frenkel conduction. In the Al/HfO2/Si structure, schottky emission dominated

over poole-frenkel since the Al/HfO2 offset is smaller than the trap level [35]. Outside of

this, there is little reference to the current transport mechanism of HfO2.

2.3.3 MOSFET Device Compatibility

In addition to low leakage current, high dielectric constant, good reliability, thin

EOT, negligible frequency dispersion, interface density on par with SiO2, and small

hysteresis (as mentioned at the end of section 2.2), HfO2 is also compatible with current

MOSFET processing, including the use of polysilicon gate material. Polysilicon gate

MOSFETs with HfO2 gate dielectrics have been fabricated [16,17,23,25], and the

performance of these devices is good. The HfO2 devices have been better than ZrO2

MOSFET devices [17]. Using reactive sputtered HfO2 with a polysilicon gate, thermal

stability up to 1000 C was reported along with EOT of 1.20 nm and leakage current of 1

x 10- A/cm2 at 1 V [23]. By annealing the silicon wafer in NH3 and using a TaN gate,

the EOT was 7.1 angstroms and the leakage current was 1 x 10-2 A/cm2 at -1.5 V [25].

2.4 Sputtering

2.4.1 DC sputtering

Sputtering has become a widely accepted process for the deposition of thin films.

Mirrors were first coated by sputter deposition as early as 1877. Until the late 1960s,

evaporation was preferred over sputtering. Then, the need for alloys with stringent

stoichiometric limits, conformal coverage, and better adherence, for magnetic and









microelectronic materials, increased the demand for sputtering deposition. The addition

of RF (which allows the sputtering of insulators) and magnetron sputtering extended the

capability of sputtering. Today, the capabilities of sputtering and the availability of high-

purity targets and gases make sputtering a popular choice for the deposition of thin films

[36]. Some of the benefits of sputtering include:

1. High uniformity of thickness;

2. Good adhesion of film to substrate;

3. Reproducibility of films;

4. Ability to deposit and maintain the stoichiometry of the target material;

5. Relative simplicity of thickness control. [37]

The sputtering mechanism is described in the following paragraphs.

Sputter deposition uses a momentum transfer process to deposit thin films of

target material onto substrates. A schematic of a typical DC sputtering system is shown

in Figure 3. In DC sputtering, a sputtering gas (usually argon) is introduced into a

vacuum chamber where an applied voltage ionizes the argon gas. The newly formed Ar+

ions accelerate toward the negatively charged cathode, which is at the target material.

When an ion approaches the target, there are five major interactions that can occur. They

are listed below and illustrated in Figure 4.

1. Reflection: The ion may be reflected; probably neutralized in the process.

2. Secondary Electron Emission: The impact of the ion may cause the target material to
eject an electron.

3. Ion implantation: The ion may become implanted in the target material.

4. Rearrangements: The impact may rearrange the structure of the target material. This
could result in point defects or even lattice defects.






17


5. Sputtering: The impact leads to a series of collisions between atoms, possibly leading
to the ejection of target atoms [38].


Vacuum Chamber


Anode

- Substrate


Figure 3: Schematic of a DC sputtering system


Incident Ion


I Reflected Ions & Neatrals


/


Implanted lon


Collison cascades
may terminate inside or
the material


S G- Secondary Electrons


Sputtered Atoms








Collison cascades
may result in sputtered
atoms


Figure 4: Interactions with incident ions at the target surface [38]


Surface.


I


Target / Cathode


Power Supply

Gas Inlet









A plasma, or glow discharge, consisting mostly of ions, electrons, and neutral

atoms, is maintained between the target and substrate. During the glow discharge, it is

observed that a current flows and a film condenses on the substrate [36]. The major

drawback to DC sputtering is that it cannot be used for insulating targets.

2.4.2 RF Reactive Magnetron Sputtering

The thin film deposition process used in this thesis, RF reactive magnetron

sputtering, is a hybrid of several other types of sputtering. What follows is a

description of these sputtering types.

RF sputtering. DC sputtering cannot be used for insulating films because the

surface on the target becomes positively charged. The use of a RF power source with an

impedance matching network solves this problem and allows the sputtering of insulating

materials. The voltage is applied at a frequency of 13.56 MHz; this allows for charging

and discharging of the insulating target [38].

Magnetron sputtering. Magnetron sputtering utilizes magnetic fields to increase

deposition rates and allow for low operating pressures and temperatures. For example,

the application of a planar magnetron will cause electrons in the glow discharge to

follow a helical path, increasing the rate of collisions and ionization. Therefore,

magnetron sputtering grows high quality films at lower operating pressures [37].

Reactive sputtering. Reactive sputtering can be used to deposit films of such

materials like oxides or nitrides through the use of pure metal targets. In addition to the

sputtering gas (argon), a reactive species-oxygen or nitrogen, for example-is introduced

into the growth chamber. The sputtered target atoms react with the gas to form the new

material.















CHAPTER 3
LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED HfO2
ON INDIUM-TIN-OXIDE (ITO)

3.1 Introduction

Much of the research effort to date on HfO2 films as a gate dielectric has focused

on the dielectric and leakage current properties of films deposited directly on silicon.

While relevant for silicon MOSFET technology, the measured properties of HfO2 on

silicon reflect both the intrinsic properties of the HfO2 film as well as the effect of a SiO2

interface layer. This structure makes it difficult to extract fundamental properties on the

HfO2 film itself.

To this end, we have investigated the dielectric and leakage current properties of

HfO2 films deposited by RF reactive sputtering onto indium-tin-oxide(ITO)-coated glass

substrates. The objective of this study is to understand the electrical properties of the

HfO2 without influence from an interface layer. ITO is a conducting oxide, so the

dielectric constant of the HfO2 can be measured without accounting for the effect of

interfacial layers. The effects of substrate temperature and oxygen pressure on

microstructure, leakage current, and dielectric constant are investigated.

3.2 Experimental Procedure

3.2.1 Substrate preparation

Reactive RF sputter deposition was used to deposit HfO2 thin films onto ITO-

coated glass substrates. Substrates were cut to a size of approximately 1 cm x 1 cm.

Then, they were ultrasonically cleaned in trichloroethylene, acetone, methanol, and









deionized water for five minutes each. They were blown dry with nitrogen gas. All

substrates were then mounted onto a platen with silver paint and loaded into the

sputtering system. The chamber was pumped down to a base pressure between 5 x 10-

and 5 x 10-7 Torr.

3.2.2 Film growth

HfO2 films were grown by reactive RF magnetron sputtering from a 2 inch

diameter 99.99% pure hafnium target. The RF power was maintained at 100 W. Oxygen

partial pressure was varied from 1 mTorr to 10 mTorr, while the total pressure (Ar + 02)

was kept at 30 mTorr. Substrate temperature was varied from room temperature to

2000C. The deposition conditions as used in this study are shown in Table 2. Film

thickness was maintained at 60-80 nm for all samples.

After loading the substrates into the deposition chamber, the temperature was

ramped up to the appropriate setting. Meanwhile, the hafnium target was pre-sputtered

with the shutter closed to remove any oxidation from the surface. The shutter was

opened, and oxygen was slowly flowed into the system.

3.2.3 Measurements

Before deposition, a portion of each substrate surface was masked in silver paint.

When the paint is removed by sonication in acetone, it leaves behind a bare ITO surface.

The thickness of the step was then measured by stylus profilometry. Table 2 shows the

results of the profilometry measurements.

For electrical property measurements, circular aluminum electrodes with area nt

*10-4 cm2 were sputtered as the top electrode using a shadow mask with a 200 micron

diameter pattern. The thickness of these electrodes was approximately 150-200 nm. The









ITO left bare by the silver paint mask served as the bottom electrode. The current-

voltage properties and dielectric constant of HfO2 was measured on ITO between the

aluminum and ITO electrode. Current-voltage curves were measured at various

temperatures to study the conduction mechanism, and capacitance was measured in order

to calculate the dielectric constant. Figure 5 is a schematic cross-section if the metal-

insulator-ITO structure used to make electrical measurements.

Table 2: Film thicknesses of Hf02 samples on ITO determined by stylus profilometry
Substrate Temperature
250C 1000C 2000C
Oxygen 1 mTorr 67.5 nm 70.0 nm 60.0 nm
Pressure 5 mTorr 63.0 nm 59.0 nm 76.5 nm
10 mTorr 65.0 nm 73.5 nm 64.0 nm


HAl

HfO2


Thickness measured
by profilometry


ITO


Glass


Figure 5: Schematic cross-section of the metal-insulator-ITO structure used in this work.
The aluminum electrode was 150-200 nm thick; the thickness of the HfO2 film, as
measured by profilometry, is shown in Table 2.









In addition to electrical measurements, x-ray diffraction (XRD) and atomic force

microscopy (AFM) were utilized to examine the film microstructure. Films without

deposited electrodes were scanned with the Philips APD 3720 diffractometer, and then

examined by tapping mode AFM on the Digital Instruments Dimension 3100. The

roughness, crystallinity, and grain structure were observed.

3.3 Results and Discussion

3.3.1 Dielectric Constant

For a parallel-plate capacitor, the dielectric constant can be calculated by

C-d
k =-- (3.1)


where C is the capacitance; d is the dielectric thickness; so is the permittivity of free

space; and A is the area of the capacitor. For Al-Hf02-ITO structures, the capacitance

was measured at 1 MHz.

Note that there is little previous work on the dielectric constant of HfO2 thin films

on substrates other than silicon. In the case of HfO2 on silicon, the effective dielectric

constant is reduced due to the HfxSiyOz layer that forms between the oxide and the silicon

substrate. The interfacial layer acts as a capacitor in series with the dielectric film.

Accordingly, it is difficult to extract the dielectric constant of the HfO2 films in this way.

Even if the Si/HfO2 interface thickness is measured by cross-section transmission

electron microscopy (TEM), for example, the exact dielectric constant of the interface

remains unknown.

In the present structure, no significant interface is formed between the oxide and

ITO substrate. We can directly calculate the dielectric constant from capacitance and

film thickness measurements. Dielectric constant values were calculated for all the









samples listed in Table 2. The dielectric constant of HfO2 at 1 MHz was between 20 and

24 for all but one sample. The variation was not systematic with deposition conditions.

The sample grown at 1000C and 5 mTorr 02 had the only outlying data point, with a

dielectric constant of approximately 29. Most of the samples had dielectric constant

approximately 20-21. This value is in agreement with what was determined on silicon by

Callegari et al. after correcting for interface thickness [29].

3.3.2 Conduction Mechanism

A variety of conduction phenomena occurs when insulating films are sandwiched

between two electrodes. The identification of the dominant (rate-limiting) mechanism is

important in understanding the current-voltage characteristics of the structure being

studied. Two broad categories describe these mechanisms: barrier-limited and bulk-

limited. Barrier-limited mechanisms operate in the vicinity of the interface between

insulator and the contacts. The transport of charge into the insulator limits the

conduction. Schottky emission and tunneling are the most prominent examples of these

types of mechanisms. In the bulk-limited case, the current is limited by the transport of

carriers through the insulator. In other words, sufficient numbers of carriers are injected

into the insulator, but they experience difficulty in reaching the other electrode due to

bulk transport limitations. Examples are intrinsic and Poole-Frenkel conduction

mechanisms [36].

Schottky emission. The schottky mechanism resembles the thermionic emission

of electrons from a heated metal to a vacuum. Thermionic emission is described by the

following equation:


J = CRD T2 ep(3.2)
k-Ts









DM is the work function of the metal and CRD is the Richardson constant. For

thermionic emission, electrons need to acquire DM, typically 4-5 eV. Schottky barriers

heights (DB), however, are smaller (in the range of 1 eV) because the electrons only need

to acquire enough energy to access the empty conduction band of an insulator. The value

of the schottky barrier height depends on the particular interface between the metal and

insulator/semiconductor. Figure 6 compares the thermionic emission of an electron from

a metal to vacuum with the schottky emission at the interface of a metal and insulator

[36].

The current density governed by schottky emission is given by the Richardson-

Dushman equation,

1/2
J=Aq. 2expq- E 1 (3.3)
{k-T) 4-e-eo Ed


where A = CRD exp B (3.4)


The variables T and E are temperature and electric field, respectively. Constants k, q, So,

and Ed represent Boltzman's constant, electron charge, permittivity of free space, and the

dynamic dielectric constant, respectively [33].











Vacuum


Conduction Band

c ed


METAL


INSULATOR


Figure 6: Schottky emission conduction mechanism. OB is the schottky barrier height.
Note the magnitude of (M, the energy required for thermionic emission from the metal.


The pre-exponential term A is used to extract the schottky barrier height, DB.

Equation 3.4 can be written as


(3.5)


In 4 In(A)+ 1 q3 1 E /2
Ir'J It~J^.^.^.f


Therefore the plot of ln(J/T2) versus E.5 should give a line with y-intercept equal to

ln(A). CRD is the Richardson constant and is given by


4 .-e m *. k


(3.6)


CRD


where m* is the effective mass [39]. For the free electron approximation CRD = 120

Acm- K-2. The effective mass of an electron in HfO2 was measured to be 0.1mo [35];


accordingly CRD = 12 Acm-2K-2









Current-voltage measurements were made at temperatures ranging from 250C to

2500C. The measurements were made for the substrate injection case (i.e. positive gate

voltage), where electrons were injected from the ITO and transported through the HfO2

film to the aluminum electrode. Since the injection of carriers into the insulator

conduction band concerns the interface between the first electrode and the insulator, this

allowed us to extract the height of the barrier at the ITO/HfO2 interface. The schottky

height of the aluminum/HfO2 has been studied before and was calculated to be -1.28 eV

[35].

Figure 7 shows the temperature-dependent current-voltage curves for a film

deposited at 1000C in 5 mTorr oxygen. Figure 8 shows the corresponding schottky fits

for films grown at (a) 1000C and (b) 250C. The ln(J/T2) versus E1/2 graphs for both

illustrate a good fit to the schottky model. However, only a self-consistent dynamic

dielectric constant can ensure that leakage current conduction mechanism in the film is

the schottky mechanism. This means that the dynamic dielectric constant, calculated

from the slope of the lines in the schottky plot, should be between the optical dielectric

constant and the static dielectric constant or very close to the optical dielectric constant

[8]. The square of the refractive index (n) gives the optical dielectric constant. For HfO2,

n is approximately 2 [40], so the optical dielectric constant is around 4. In the previous

section, we estimated the static dielectric constant to be 20-21 at 1 MHz. The dynamic

constant for the data in the graphs below is approximately 4-5. From these linear fits, we

extracted the barrier height at the HfO2/ITO interface. The range of data is

approximately 1.1 + 0.2 eV.









Despite what appears to be a reasonable schottky fit, some of the samples

examined had dynamic dielectric constants less than 4, which are not consistent with the

schottky process. However, this phenomenon was observed in the films deposited at

higher substrate temperature and/or high oxygen pressure, thus yielding high leakage

current. For these films the conduction mechanism may be influenced by the schottky

emission process, another mechanism is also significant. For example, the contribution

from current through grain boundaries, if crystallization takes place, would cause the

measured dynamic dielectric constant to be small [33]. It is probable that films deposited

at the higher temperature and/or oxygen pressure have higher degrees of crystallinity that

would contribute to the overall leakage current and may cause deviation from the

schottky model. The data in following sections supports this explanation. It will be seen

that crystallinity is evident and increases with higher substrate temperature and oxygen

pressure.



1E-3

1 E-4
<1E-3--------------------------
-oo m .. .. ..
1 E-5 akoi m ....... )K--K ),<--- --- ^ ^
Io KKKXIOK X 100 C
Q 1 E-6 --
w K -150 C
1E-7 200 C
S250 C
0 1E-8
0 2 4 6
Gate Voltage (V)


Figure 7: Leakage current of Hf02 on ITO measured at various temperatures. The
sample was sputtered with 5 mTorr 02 and 1000C substrate temperature.













-20
-21
-22
-23
-24
-25
-26
-27
-28
-29
-30
2000


4000 6000 8000


E0. (V/m)0.5


-26

-27

-28
E
0 -29

- -30

-31
I--
S-32

-33


-34 '
5000

(b)


7000
E0.5


9000

(V/m)0.5


11000


Figure 8: Schottky emission plots of HfO2 on ITO measured at different temperatures.
The samples was sputtered with different deposition conditions: (a) 5 mTorr 02 and
1000C, and (b) 5 mTorr 02 and 25C


C?
E


C?




I-

ra









3.3.3 Leakage Current

The effects of substrate temperature and oxygen partial pressure on the leakage

current behavior of HfO2 on ITO were also examined. Leakage currents as low as 8 x 10-

10 A/cm2 were obtained at 1 V for HfO2 films grown at 5 mTorr 02 and 250C. Other

films displayed values on the order of 10- A/cm2. Figure 9 shows the effect of varying

substrate temperature (250C, 1000C, and 2000C), while the oxygen pressure remains the

same. As a general trend, increasing substrate temperature caused an increase in leakage

current. In the 1 mTorr 02 case, it is difficult to differentiate whether the current increase

between the 1000C and 2000C films is due to the effect of substrate temperature or the

reduction of film thickness. However, those films grown with 5 mTorr and 10 mTorr 02

show a clear degradation in leakage current properties as the substrate temperature is

increased. In fact, this increase in leakage current between the films exists despite an

increase in film thickness. We believe that the increasing leakage current results from

increasing crystallinity as the films are grown under higher substrate temperature.














- 1E-3
E 1E-4
^ 1E-5
' 1E-6
c 1E-7
o 1E-8
C 1E-9
1E-10
1 E-11


0 2 4 6
(a) Gate Voltage (V)


1 E+O
1 E-1
1E-2
1E-3
1 E-4
1E-5
1E-6
1 E-7
1 E-8
1E-9


I2000C








0 2 4
(b) Gate Voltage (V)


(c) Gate Voltage (V)



Figure 9: Room temperature leakage current plots of Hf02 on ITO showing the effect of
substrate temperature. Each plot shows a different set of deposition oxygen pressures: (a)
1 mTorr 02, (b) 5 mTorr 02, (c) 10 mTorr 02.


1 E+0
E 1E-1
S1E-2
S1E-3
- 1E-4
c 1E-5
o 1E-6
S1E-7
S1E-8
S 1E-9
1E-10


c
E






0
C,

U)
13
,-

0


-2000C


"i-O0C
_- 1000C


lI 250C









Figure 10 shows the effect of varying oxygen pressure (1 mTorr, 5 mTorr, 10

mTorr), while the substrate temperature remains the same. For films grown with room

temperature or 1000C substrates, leakage current is seen to decrease when the oxygen

pressure is increased from 1 mTorr to 5 mTorr. However, the leakage current then

increased when the oxygen pressure is further increased to 10 mTorr. One possible

explanation for the observed oxygen pressure dependence is that the films are not fully

oxidized at 1 mTorr 02. The increase of oxygen to 5 mTorr 02 oxidizes the films,

resulting in a leakage current decrease. However, when the oxygen pressure is increased

further, the leakage current increases due to the contribution of film crystallization. The

trend is more obvious for films grown at 2000C substrate temperature. The leakage

current is monotonically increased by additional oxygen pressure due to increasing

crystallization. Figure 11 shows the room temperature leakage current for each of the

nine samples studied measured at 2 V. It is easy to see the trend in leakage current as a

function of growth conditions.











1 E-4
1 E-5
1 E-6
1 E-7
1 E-8
1 E-9
1E-10
1E-11


2 4 6
Gate Voltage (V)


1 E-2
1E-3
1 E-4
1E-5
1 E-6
1 E-7
1E-8
1E-9


0

(b)


1E+0
1E-1
1E-2
1E-3
1E-4
1E-5
1E-6
1E-7
1E-8
1E-9


2 4 6
Gate Voltage (V)


2 4 6

Gate Voltage (V)


Figure 10: Room temperature leakage current plots of HfO2 on ITO, showing the effect
of oxygen pressure during film deposition. Each graph displays a different substrate
temperature: (a) 250C, (b) 1000C, (c) 2000C.



















1.00E-09

1.00E-08

1.00E-07

1.00E-06

1.00E-05


Leakage
Current
Density
(A/cm 2)


1.00E-04
25 1
100 5
Substrate 200 10 Oxygen
Temperature Pressure
(deg. C) (mTorr)


Figure 11: Leakage current of HfO2 on ITO measured at room temperature and 2 V
versus growth conditions.









3.3.4 Microstructure

In addition to leakage current, we have also examined the microstructure of HfO2

films using X-ray diffraction (XRD). Figure 12 shows XRD results for HfO2 films

grown on ITO with 10 mTorr 02 pressure. The emergence of the broad hump at 20 28.30

indicates the existence of crystallinity in the film deposited at 2000C. The peak is the

(-111) plane of the monoclinic HfO2 structure which is the most common HfO2 phase

[19,21,22,41]. The emergence of this peak supports the idea that enhanced crystallinity

at higher oxygen pressure and substrate temperature yields the observed increase in

leakage current.





Hf02 ITO (222)
M-( 111
O (400)
rIO (441)




C1000C


250C

20 30 40 50
2e (deg.)

Figure 12: XRD diffractogram of Hf02 on ITO grown with 10 mTorr 02 and varying
substrate temperature.









In order to gain further information about the film microstructure, tapping mode

atomic force microscopy (AFM) was used. The topography of the HfO2 films grown on

ITO strongly resembles the topography of the ITO surface on the several micron scale.

The polycrystalline grain structure of the ITO film is visible. The grains are about 0.5

micron in diameter and the rms roughness is about 3.416 nm. All of the samples

examined by AFM showed the ITO surface structure on the HfO2 film surface. However,

roughness differences in the films correlated with increasing oxygen pressure and

substrate temperature. Figure 13 and Figure 14 plot this trend. The increase in roughness

is consistent with an increase in crystallinity with high temperature and/or oxygen

pressure during deposition.


E 12
( 10
d) 8
Ss 6

0 4
S2


0 5 10

Oxygen Pressure (mTorr)

Figure 13: RMS roughness increase measured by AFM as oxygen pressure was varied
during 2000C substrate temperature deposition.










i12----- -

E 2
10
8


0 -4



0 100 200

Substrate Temperature (deg. C)

Figure 14: RMS roughness increase measured by AFM. The increase corresponds to
increasing substrate temperature when the films were grown with 10 mTorr 02.

AFM images at a smaller scale show grain structure in the films.

Figure 15 shows an AFM image of the film grown at 5 mTorr 02 and 2000C. The figure

is actually an image of the derivative of the height. It represents what the surface might

look like to the eye or through a microscope. Figure 16 and Figure 17 have been

enhanced to show the grain boundaries better. This is done by the application of a

highpass filter. Each pixel is weighted according to the average of the surrounding

pixels. There is a noticeable grain size increase from Figure 16 to Figure 17 which

corresponds to the increase in substrate temperature from 1000C to 2000C. Since the

highest substrate temperature gives the maximum surface mobility, this is consistent with

island growth and coalescence [42].
















-500


250


500


Figure 15: AFM derivative image of Hf02 grown on ITO at 5 mTorr 02 and 200C



















500














250














0
0 250 500

nm

Figure 16: AFM image of HfO2 grown on ITO at 10 mTorr 02 and 1000C with highpass
filter applied to enhance grain edges.


































250














0
0 250 500

nm

Figure 17: AFM image of Hf02 grown on ITO at 10 mTorr 02 and 2000C with highpass
filter applied to enhance grain edges.









3.4 Conclusions

In conclusion, the leakage current behavior for HfO2 films deposited by RF

reactive sputtering onto ITO substrates with an Ar + 02 gas flow was examined. Oxygen

pressure and substrate temperature were varied to optimize films for high dielectric

constant and low leakage current. The dielectric constant of HfO2 films was calculated to

be -20-21 from 1 MHz capacitance measurements. There was no systematic variation of

the dielectric constant according to deposition conditions.

A low leakage current of 8 x 10-10 A/cm2 was obtained for a Al/HfO2/ITO

structure sputtered at 250C substrate temperature and 5 mTorr 02 pressure when the film

thickness was 63 nm. Current-voltage curves were measured at various temperatures to

study the leakage current mechanism under substrate electron injection. For most films

deposited below 2000C, the data fit the schottky emission mechanism well. The schottky

barrier height at the ITO/HfO2 interface was extracted to be 1.1 0.2 eV.

Leakage current was increased by raising the substrate temperature; the lowest

leakage currents were obtained with room temperature depositions. Increasing the

oxygen amount from 1 mTorr 02 to 10 mTorr 02 also increased the leakage current. The

increase in leakage current with increasing oxygen pressure and substrate temperature

during deposition was attributed to increasing crystallinity. This assertion of increasing

crystallinity is supported by XRD data as well as AFM imaging and roughness data

obtained from the AFM study. However, for films grown at temperature less than 2000C,

the leakage current decreased when the oxygen pressure increased from 1 mTorr to 5

mTorr. To explain this, we asserted that films grown with 1 mTorr 02 were not fully

oxidized. The increase in oxygen pressure from 1 mTorr to 5 mTorr resulted in further









oxidation of HfO2 film, reducing the leakage current. The leakage current increased

when the oxygen pressure was increased further because the increase in crystallinity then

became the dominating factor.

The crystallization of the films was supported by increasing rms roughness values

as substrate temperature and oxygen pressure were increased. XRD seemed to indicate

the existence of small grains as well. Likewise, AFM imaging shows the appearance of

the grains in several samples.

The crystallization of the films was also reflected by schottky emission fitting.

The films grown at high oxygen pressure and temperature did not have self-consistent

dynamic dielectric constants in the schottky model. Since the most crystalline films were

not consistent with schottky dominated current transport, we inferred that the crystallinity

of the films contributes significantly to the conduction transport mechanism of these

HfO2 films.















CHAPTER 4
LEAKAGE CURRENT PROPERTIES OF RF SPUTTERED
HfO2 ON SILICON

4.1 Introduction

The scaling of the metal-oxide-semiconductor field-effect transistor (MOSFET)

feature size to sub-100nm dimensions requires a decrease in thickness of the current

thermal oxide SiO2 gate dielectric to less than 1 nm. At this thickness, direct tunneling

leakage current is significant for SiO2. A solution to the problem is to find materials with

high dielectric constants relative to SiO2. A material with a higher dielectric constant

(high-k material) can be made thick enough to avoid direct tunneling and still maintain

the capacitance of the thinner SiO2 film. HfO2, with a dielectric constant on the order of

20-25, is a promising alternative gate dielectric. HfO2 has good thermal stability on

silicon and can consume native oxide to form HfO2. It is the first high-k material to show

compatibility with the present complimentary metal-oxide-semiconductor (CMOS)

polysilicon gate process. HfO2 has a high dielectric constant and relatively low leakage

current.

In this study, the properties HfO2 thin films were investigated as an alternative

gate dielectric. HfO2 was deposited by reactive RF magnetron sputtering under varying

oxygen partial pressures and substrate temperatures onto silicon substrates. Metal-oxide-

semiconductor structures were fabricated on silicon to measure capacitance-voltage and

current-voltage properties.









4.2 Experimental Procedure

4.2.1 Substrate Preparation

Reactive RF sputter deposition was used to deposit HfO2 thin films onto p-type Si

(100) substrates. Substrates were cleaved to a size of approximately 1 cm x 1 cm. Each

substrate was ultrasonically cleaned in trichloroethylene, acetone, methanol, and

deionized water for five minutes each, then blown dry with nitrogen gas. Next, the

silicon substrates were immersed in a 10:1 H20:HF solution for 45 seconds, then rinsed

twice successively in deionized water to remove native oxide and to hydrogen-terminate

the substrate surface. The substrates were again blown dry with nitrogen. All substrates

were mounted onto a platen with silver paint and loaded into the sputtering system. A

portion of each substrate surface was covered in silver paint for thickness measurement

after growth. The chamber was pumped down to a base pressure between 5 x 10-s and 5

x 10-7 Torr.

4.2.2 Film Growth

HfO2 films were grown by reactive RF magnetron sputtering from a 2 inch

diameter 99.99% pure hafnium target. The RF power was maintained at 100 W. Oxygen

partial pressure was varied from 1 mTorr to 10 mTorr, while the total pressure (Ar + 02)

was kept at 30 mTorr. Substrate temperature was varied from room temperature to

2000C. There was a total of nine combinations of deposition conditions, as shown in

Table 3. Film thickness was 60-85 nm for all samples.

After loading the substrates into the deposition chamber, the temperature was

ramped up to the appropriate setting. Meanwhile, the hafnium target was pre-sputtered









with the shutter closed to remove any oxide from the surface. The shutter was opened,

and oxygen was slowly flowed into the system.

Table 3: Film thicknesses of Hf02 samples on p-type silicon determined by stylus
profilometry and ellipsometry.
Substrate Temperature

250C 1000C 2000C

Oxygen 1 mTorr 62.5 nm 65.0 nm 63.0 nm
Pressure
5 mTorr 63.0 nm 59.0 nm 83.0 nm

10 mTorr 62.0 nm 65.0 nm 71.5 nm



4.2.3 Measurements

The thickness of the films was measured by stylus profilometry and ellipsometry

using the Filmetrics F20 thin-film measurement system. After the thickness was

determined, samples were prepared for electrical property measurement. Aluminum

electrodes with area n 10-4 cm2 were sputtered as the top electrode using a shadow mask

with a 200 micron diameter circle pattern. The thickness of these electrodes was

approximately 150-200 nm. The backside of the silicon substrate was mounted onto a

glass slide with conductive silver paint, the bottom electrode. The current-voltage

properties and capacitance-voltage properties of HfO2 were measured on silicon. The

current was measured under substrate electron injection. The capacitance-voltage curve

was obtained at 1 MHz. The voltage was swept from depletion to accumulation and then

back to depletion at sweep rate of 1.6 V/s. Figure 18 is a schematic cross-section of the

metal-insulator-semiconductor structure used to make electrical measurements in this

work.









In addition to electrical measurements, x-ray diffraction (XRD) and atomic force

microscopy (AFM) were utilized to examine the film microstructure. Films without

deposited electrodes were scanned with the Philips APD 3720 diffractometer, and then

examined by tapping mode AFM on the Digital Instruments Dimension 3100. The

roughness, crystallinity, and grain structure were obtained.















HfxSiyOz


Figure 18: Drawing of MIS structure with interface layer


Silicon


Ag






46


4.3 Results and Discussion

4.3.1 Microstructure

X-ray diffraction (XRD) was used to investigate the formation of polycrystalline

grains as a function of processing conditions. The XRD results shown in Figure 19

through Figure 21 show that crystallinity is evident in the HfO2 films. The monoclinic

peak appears at 20 = 28.3 for most films. The intensity of this peak increases for higher

oxygen pressure and substrate temperature used during deposition. This is undesirable

for gate dielectric applications as grain boundaries yield paths for enhanced leakage

current.


M-(-111)
Substrate


S10 mTorr 02




WW I W, '1W. ALI6lQ6l,





10 20 30 40 50
29 (deg.)

Figure 19: XRD diffractogram of HfO2 films grown on silicon at 2000C substrate
temperature, showing effect of oxygen pressure on crystallization.






















20 A (deg.) 40 50


Figure 20: XRD diffractogram of Hf02 films grown on silicon at 250C substrate
temperature, showing the effect of oxygen pressure on crystallization.


200C

100C

25C


20 30 2(deg.) 40 50
28 (deg.)


Figure 21: XRD diffractogram of Hf02 films grown on silicon with
showing the effect of substrate temperature on crystallization.


5 mTorr 02 pressure,


u,

'-S
m


A
-^wW
y^
.^.^r- A
itMm'^ Aj






48

AFM studies were also performed for several films grown on silicon. Figure 22

and Figure 23 show the trend of increases roughness as oxygen pressure and substrate

temperature is increased. This is indicative of increasing crystallization.


E 8






Co



0 100 200

Substrate Temperature (deg. C)

Figure 22: RMS roughness as a function of substrate temperature for HfO2 on Si
deposited at 10 mTorr 02.







Co
W 2



0 5 10

Oxygen Pressure (mTorr)

Figure 23: RMS roughness as a function of oxygen pressure for HfO2 on Si deposited at
2000C.











Figure 24 through Figure 28 show three-dimensional views of the HfO2 film


surfaces grown at different temperatures and oxygen pressures. At 10 mTorr 02 and


2000C, the grains are largest and completely cover the scan surface. The figures show


that the grain size decreases and the films get smoother with lowering temperature and


oxygen pressure. This is because surface diffusivity is the highest at high substrate


temperatures. Accordingly, the growth rate of the films decreases 10-fold when the


oxygen pressure changes from 1 mTorr 02 to 5 or 10 mTorr 02. This is because the


target is oxidized at the high pressures. The slow deposition rate combined with the high


substrate temperature allows islands to grow larger since the surface diffusivity is high.


The result of this is large grains at high temperature and oxygen pressure.



Dig t,4 Ir"nt rinnm Nt nr/5!::o|r
Scan s1z i.1.0530 ar
SLjri rtOv 1. 9~Q Il
iutnbtr oF sirap'le 512
Jafo caul N5 fIm
:ut !.ccle 25 Ot 'a


I. 'IN" U1fi.4 ra i
:rrq. :Ci IIpAr -.0 OC0l rn
..rTq, :MN r )J $.S4 rn
:rr;. rrC O r ro

view l.AQ I
VI:ght 3ng 1!










l7 A.4 G. r, 3.8 MA


2 2~.UCO ni/Jiv


Figure 24: Three dimensional AFM view of Hf02 surface grown on p-Si(100) at 10
mTorr 02 and 200C.


































Li AlII t [ ; IrNn Tui-iCC t

Scar rate L 96 -I
rLCmk-I ur ,,4 mpi k.
Irra.;e Data. 'ei pht


TH R-Aim ivio.ri rNT



TnTrr.. Frr. (P.0) 067 ni
Irr. Pa 49; nm
Irr. P" ner


I I I I r
0.2 0.4 u.6 0.8 O'






Figure 25: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10

mTorr 02 and 1000C.


iew argI I!
liqh[ arylgi





































Oigritr'l InbtrkIMnn" NI-5CrA-kr
Scar size L-000 -M

Nr~ter of s12pl :
:11.c' L-r1 Uaigll
Oal:4. scal ,, n


Iryj Z r~anr.F
IF!, P,3 W TO,
I rI. pris (pq,
Irn. P3r
IIS. R~2~


\ 0, .2 00 [..-. ;d ki'.
2 25-UUC W



Figure 26: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 10

mTorr 02 and 250C.


271.<01 -a
-XOx rn

1.$ rh~
~2.'!1 r.~



~blight argli!


A. 7 .4 A.6 Rl2


Im.?pe Ststistic;
































Ligl tl ItIs[ri ur T'lts Nlrt.Sc.iP
*CAr 1i ; I. O( ,mn
Scar rate l.S 6 -I=
Nranrwr ur rt,-lrfP. 512
Irra.e^ Data. -eight
Dai. r al q' 2 5. 1111
engsgM P.,)S I4CL(CO.O .


Iry. ranpI7P -.L76 nMi
]rrg, ;;4u riear 0 (00iQCiOO rn
:rrg, Ars (=q ) '2 504 rrm
:rrg, a I BS nir
Irrg. Arrha 26.7?6 nfn



11h. Irylt


U.I 0.4 U.6 0.


x C.20O i.iir'd I
: '5 '000 i-r.ivc


Figure 27: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 5 mTorr

02 and 2000C.












Ttinj nr ii ght


















C'.? 0. 0 .6 C 3 n



Figure 28: Three dimensional AFM view of HfO2 surface grown on p-Si(100) at 1 mTorr
02 and 2000C.


4.3.2 Dielectric Constant

The dielectric properties of the HfO2 films were determined using capacitance

measurement with an applied voltage sufficient for accumulation. Overall, the dielectric

constant slightly increased with increasing substrate temperature and oxygen pressure.

This probably reflects the increase in crystallinity. The behavior is complicated by the


combined effects of increasing crystallinity, interface growth and composition, and the

presence of mobile ions. The dielectric constant decrease in the 1 mTorr series going


from 100 to 200C in Figure 29 may be attributed to the growth of the interface layer.

The decrease in dielectric constant in Figure 30 going from 1 mTorr to 5 mTorr O at

25C could result from the oxidation of the film. However, the effect of the
25'C could result from the oxidation of the film. However, the effect of the






54

crystallization dominates most of the data as the dielectric constant increases with

increasing crystallinity.











35

a 30
U -*- 1 mTorr Oxygen
0 25
S-m- 5 mTorr Oxygen
20
10 m orr Oxygen
U Ac- 10 mrTorr Oxygen


0 100


200


Substrate Temperature (degrees C)


Figure 29: Dielectric constant (calculated from accumulation capacitance) as a function
of growth conditions for HfO2 grown on silicon










35

C 30

o 25 25 C
0
o .-m- 100 C
20 200 C

15

10
0 5 10

Oxygen Pressure (mTorr)

Figure 30: Dielectric constant (calculated from accumulation capacitance) as a function
of growth conditions for HfO2 grown on silicon

4.3.3 Leakage Current Behavior

The leakage current properties are affected by the crystallinity and stoichiometry

of the HfO2 film, as well as the quality and thickness of the interfacial layer at the

Si/HfO2 boundary. The behavior of leakage current with increasing substrate temperature

and oxygen pressure is shown in Figure 31 and Figure 32. Note that the interface layer is

thickest when the films are grown at high oxygen pressure and substrate temperature.

When the interface thickness is increased, leakage current of these films decreases. Some

small deviations in the trend can be explained by changes in overall film thickness.










C4J
E
0


<,

a

c-


0
0


1 E-3

1 E-4

1 E-5

1 E-6

1 E-7

1 E-8

1 E-9

1E-10


2 4 6 8


Gate Voltage (V)


1 E-3

1 E-4

1 E-5

1 E-6

1 E-7

1 E-8

1 E-9

1E-10


0 1 2 3 4 5 6


Gate Voltage (V)


Figure 31: Room temperature leakage current measurement of HfO2 on silicon showing
the decrease in leakage current that corresponds with increasing substrate temperature
during deposition. The graphs show the trend for films grown with (a) 5 mTorr and (b)
10 mTorr oxygen pressure.


G)
..
0
4-


3
C










1 E-3

1 E-4

1 E-5

1E-6

1 E-7

1E-8

1 E-9

1E-10


1 E-3

1 E-4

1E-5

1 E-6

1 E-7

1E-8

1 E-9

1E-10


5 mTorr 02
m n


1 mTorr 02 ,,* *


A 10 mTorr 02


* -


**


C(

)
0






c


0
0


3 1 2
(a) Gate Voltage (V)










A ** AAAAAAA AAA
.* 10 mT5 morr 02AAA


S5 m-m--*-orr 02""
**mmmm 5 mTorr 02


Gate Voltage (V)


Figure 32: Room temperature leakage current measurement of HfO2 on silicon showing
the decrease in leakage current that corresponds with increasing oxygen pressure during
deposition. The graphs show the trend for films grown at (a) 1000C and (b) 200C
substrate temperature.


AAA AAA A+


C4




E-











The leakage current for the films grown at 250C more generally reflect the

properties of the HfO2 film as minimal interfacial Si-Hf-O is formed. When the films

were grown with 1 mTorr 02, leakage current increased when substrate temperature

increased. When the films were grown with 25C substrate temperature, the leakage

current increased when the oxygen pressure was increased. These trends are shown in

Figure 33. The lowest leakage current at 2 V was measured for the film grown with 1

mTorr 02 at 250C. This is probably because of the increase in crystallinity, which has a

larger impact on leakage current than the interface growth in this particular case.














200C A
AAA


A AA


(I)
*4-
E









0
cl


3
o


1 E-3

1 E-4

1E-5

1 E-6

1 E-7

1E-8

1E-9

1E-10


AA AAA AA At


A A AA


100 C


0 0 0


oo0 0 0 0 0


1 2

(a) Gate Voltage (V)










10 mTorr 2 AA &

S" 5 mTorr 02

1** 0****** .
1 mTorr 02


Gate Voltage (V)


Figure 33: Room temperature leakage current of HfO2 on silicon showing (a) effect of
substrate temperature when films are grown with 1 mTorr 02 and (b) effect of oxygen
pressure when films are grown at 250C


S0 0

00


25 C
o oo oo o oo


i


>,



0



3
O


0 00 0 00
O


1 E-3


1 E-4

1 E-5

1E-6

1 E-7

1E-8

1 E-9

1E-10









4.3.4 Current Mechanism

In order to begin discussion of possible current mechanism combinations involved

in the complicated current transport through our structure, one needs to review several

transport mechanisms.

In schottky emission, carriers need to obtain enough energy to be transported over

the schottky barrier into the next material. Current governed by schottky emission

follows the form of the equation below [33].


1. q3 -E 1 (4.1)
{k-TJ 4-z-eo- JEd


Tunneling occurs when charge transport occurs through an insulating layer by the

penetration of carriers horizontally (at constant energy) through a barrier, rather than

surmounting enough energy to be transported over the barrier (as in schottky emission).

Tunneling is a quantum mechanical effect and requires a sufficient electric field.

Tunneling is described by equation 4.3 [36].

q2 "E2 8-z -(2m)1/2 ( )3/2
JT = exp -- -(q 2 (4.2)
8-Tt.h.(& 3-h-q-E

Third, Poole-Frenkel emission, is a bulk transport mechanism. Charge is trapped

in the insulator by impurity levels. The charge can be transported to the insulator

conduction band by internal emission processes. When a field is applied, the potential

well associated with the trap is distorted, increases the chances for the trapped electron to

escape. Equation 4.4 below describes the poole-frenkel conduction mechanism [36].


JPF =c. .exp q- .exp I (4.3)
k-T k-T z-o- Ed









The three mechanisms listed above are commonly seen in oxides, including HfO2.

The leakage current behavior of HfO2 grown on silicon can be explained by combining

the contributions of several mechanisms.



Ec

Ec Ec


Ec

....................................................................... ............ ....... ............ ................................................................ I -I- E F










Figure 34: Schematic E vs. x band diagram of the Al/HfO2/SiO2/p-type Si structure in
this experiment



The leakage current properties of HfO2 in direct contact with ITO has been

reported in Chapter 3. We concluded that the current transport was governed by schottky

emission at the ITO/HfO2 interface under substrate carrier injection. On the contrary,

HfO2 film does not grow in direct contact with silicon. An interface layer is formed

between the HfO2 film and the silicon substrate, most likely SiO2 or some HfxSiyOz.

Now, as seen by the band diagram in Figure 34 above, the barrier height makes it less

likely for schottky emission to occur from silicon to the interface layer than it would be

for schottky emission to occur from silicon to HfO2 in the absence of an interface. Our

data shows that schottky emission is not the dominant mechanism for the films on silicon.









In fact, there is probably a complicated assortment of mechanisms that control the current

transport in the Al/HfO2/SiO2/Si structure.

We would like to suggest a conduction mechanism for our samples, when

electrons were injected from the substrate. The injected electrons tunnel through the

interfacial layer into bulk trap levels inside the HfO2. Therefore, the conduction in these

films combines the tunneling through the interface and the poole-frenkel type in the

HfO2. The rapid discontinuous leakage current increase observed at an applied voltage of

3-5 V (not shown) in some of the samples grown with 1 mTorr 02 may be caused by the

breakdown of the thin HfxSiyOz layer at the interface with silicon. This rapid increase is

not observed in the films grown on ITO (in chapter 3). The mechanism described above

was also believed to occur in Ta205 films on silicon [33].

4.4 Conclusions

HfO2 films were reactive RF sputtered onto p-type Si (100) substrates with an Ar

+ 02 gas flow. Oxygen pressure and substrate temperature were varied to determine their

effect on the dielectric and microstructural properties of HfO2 on silicon. AFM and XRD

were used to characterize film microstructure. XRD peaks grew more intense with higher

substrate temperature and oxygen pressure. In addition, AFM shows that grains were

largest and most prevalent in films grown at high oxygen pressure and substrate

temperature.

The dielectric constant was calculated for each sample based on the measure

accumulation capacitance and film thickness. Dielectric constant generally increased

with increasing substrate temperature and oxygen pressure due to the crystallinity. The

exception is for films grown at low temperature where the dielectric constant decreases









when the oxygen amount is increased from 1 to 5 mTorr. Apparently, the films are not

fully oxidized under low oxygen pressure at low temperature. At low temperature, the

oxidation of the film and/or the growth of an interface layer had a larger effect on the

dielectric constant than the film crystallization. Low leakage currents in high

temperature and/or high oxygen pressure films is due to the effect of the HfO2/Si

interfacial layer. The current transport mechanism in these films was speculated to be a

combination of tunneling through the interface layer, and then poole-frenkel and grain

boundary transport through the film.















CHAPTER 5
SUMMARY

HfO2 thin films were deposited onto indium-tin-oxide (ITO) and p-type silicon

(100) substrates using reactive RF sputter deposition. Substrate temperature and the

amount of oxygen used during deposition were varied to determine their effect on

leakage current and microstructure. X-ray diffraction (XRD) and atomic force

microscopy (AFM) were both used for microstructural characterization. Crystallinity

was determined to be present in films on both substrates and increases with increasing

oxygen pressure and substrate temperature.

Al/HfO2/ITO structures were used to measure leakage current and dielectric

constant as a function of deposition conditions. Temperature-dependent leakage current

measurements allowed for the extraction of the HfO2/ITO schottky barrier height. The

schottky barrier height at the ITO/HfO2 interface was extracted to be 1.1 + 0.2 eV.

However, the films grown at high oxygen pressure and substrate temperature were not

consistent with the schottky mechanism because of the grain boundary contribution. In

general, leakage current on ITO was increased by increasing the oxygen pressure and

substrate temperature because of enhanced crystallization. However, at low temperatures

the films may not have been fully oxidized at 1 mTorr oxygen pressure. Therefore, the

leakage current was improved by more oxygen (5 mTorr) and then subsequently

degraded by the addition of excessive oxygen (10 mTorr). The dielectric constant of

HfO2 films was calculated to be -20-21 from 1 MHz capacitance measurements on ITO.









In addition, metal-oxide-semiconductor (MOS) structures were fabricated on

silicon with aluminum electrodes. The MOS structures were used to measure current-

voltage and capacitance-voltage characteristics as a function of deposition conditions.

The dielectric constant was calculated for each sample based on the measure

accumulation capacitance and film thickness. Dielectric constant generally increased

with increasing substrate temperature and oxygen pressure due to the crystallinity. The

exception is for films grown at low temperature where the dielectric constant decreases

when the oxygen amount is increased from 1 to 5 mTorr. Apparently, the films are not

fully oxidized under low oxygen pressure at low temperature as was seen from leakage

current measurements on ITO as well. The leakage current for the Al/HfO2/Si structures

decreased for high substrate temperature and/or oxygen pressure during depositon. This

effect is attributed to the growth of the Si/HfO2 interfacial layer. The current transport

mechanism in these structures on silicon was speculated to be a combination of tunneling

through the interface layer, and then poole-frenkel and grain boundary transport through

the film.















REFERENCES


1. K. Cho, Computational Materials Science 23, 43-7 (2002).

2. B. Cheng, M. Cao, R. Rao, A. Inani, P. V. Voorde, W. M. Greene, J. M. C.
Stork, Z. Yu, P. M. Zeitzoff, and J. C. S. Woo, IEEE Transactions on Electron
Devices, 46, 1537 (1999).

3. J. Robertson, J. Vac. Sci. Technol. B 18, 1785 (2000).

4. J. Robertson, J. of Non-Crystalline Solids 303, 94-100 (2002).

5. J. S. Suehle, E.M. Vogel, M. D. Edelstein, C.. Richter, N. V. Nguyen, I. Levin,
D. L. Kaiser, H. Wu, and J. B. Bernstein, Amer. Vac. Soc. 2001 6 International
Symposium on Plasma Process-Incuced Damage, Monterey, CA, USA, 90
(2001).

6. A. Paskaleva, E. Atanassova, T. Dimitrova, Vacuum 58, 470-7 (2000).

7. T. P. Ma, IEEE, 6th International Conference on Solid-State and Integrated-
Circuit Technology, 2001. Proceedings., 1, 297 (2001).

8. J. S. Lee, S. J. Chang, J. F. Chen, S. C. Sun, C. H. Liu, U. H. Liaw, Materials
Chemistry and Physics 77, 242-7 (2002).

9. W. H. Ha, M. H. Choo, S. Im, J. of Non-Crystalline Solids 303, 78-82 (2002).

10. M. D. Groner, J. W. Elam, F. H. Rabreguette, S. M. George, Thin Solid Films
413, 186-197 (2002).

11. G. C. Yeap, S. Krishnan, and Ming-Ren Lin, Electronics Letteres, 34, 1150
(1998).

12. W.-J. Qi, R. Nieh, B. H. Lee, L. Kang, Y. Jeon, K. Onishi, T. Ngai, S. Banerjee,
and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest., 145-8
(1999).

13. P. S. Lysaught, P. J. Chen, R. Bergmann, T. Messina, R. W. Murto, H. R. Huff, J
of Non-Crystalline Solids 303, 54-63 (2002).










14. L. Kang, Y. Jeon, K. Onishi, B. H. Lee, W.-J. Qi, R. Nieh, S. Gopalan, and J. C.
Lee, 2000 Symp. On VLSI Technology Digest of Technical Papers, 44-45 (2000).

15. L. Kang, K. Onishi, Y. Jeon, B. H. Lee, C. Kang, W.-J. Qi, R. Nieh, S. Gopalan,
R. Choi, and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest., 35-
8 (2000).

16. S. J. Lee, H. F. Luan, W. P. Bai, C. H. Lee, T. S. Jeon, Y. Senzaki, D. Roberts,
and D. L. Kwong, Int. Electr. Devices Meeting. IEDM Technical Digest, 31-4
(2000).

17. Y. Kim, G. Gebara, M. Freiler, J. Barnett, D. Riley, J. Chen, K. Torres, J. Lim, B.
Foran, F. Shaapur, A. Agarwal, P. Lysagth, G. A. Brown, C. Young, S.
Borthakur, H.-J. Li, B. Nguyen, P. Zeitzoff, G. Bersuker, D. Derro, R. Bergmann,
R. W. Murto, A. Hou, H. R. Huff, E. Shero, C. Pomarede, M. Givens, M.
Mazanec, and C. Werkhoven, Int. Electr. Devices Meeting. IEDM Technical
Digest, 455-8 (2001).

18. A. M. Stoneham, J. of Non-Crystalline Solids 303, 114-122 (2002).

19. C.T. Kuo and R. Kwor, Thin Solid Films 213, 257-64 (1992).

20. B.-H. Lee, L. Kang, W.-J. Qi, R. Nieh, Y. Jeon, K. Onishi and J. C. Lee, Int.
Electr. Devices Meeting. IEDM Technical Digest, 133-6 (1999).

21. S.-W. Nam, J.-H. Yoo, S. Nam, H.-J. Choi, D. Lee, D.-H. Ko, J. H. Moon, J.-H.
Ku, S. Choi, J. of Non-Crystalline Solids 303, 139-143 (2002).

22. S. Xing, N. Zhang, Z. Song, Q. Shen, C. Lin, Microelectronic Engineering 1 66,
451-6 (2002).

23. B. H. Lee, R. Choi, L. Kang, S. Gopalan, R. Nieh, K. Onishi, Y. Jeon, W.-J. Qi,
C. Kang and J. C. Lee, Int. Electr. Devices Meeting. IEDM Technical Digest, 39-
42 (2000).

24. C. T. Hsu, Y. K. Su and M. Yokoyama, Jpn. J. Appl. Phys. 31, 2501-4 (1992).

25. R. Choi, C. S. Kang, B. H. Lee, K. Onishi, R. Nieh, S. Gopalan, E. Dharmarajan,
and J. C. Lee, Symposium on VLSI Technology Digest of Technical Papers, 15-6
(2001).

26. M. Gutowski, J. E. Jaffe, C.-L. Liu, M. Stoker, R. I. Hegde, R. S. Rai, and P. J.
Tobin, Applied Physics Letters 80, 11 (2002).










27. P. D. Kirsch, C. S. Kang, J. Lozano, J. C. Lee, J. G. Ekerdt, J. Applied Physics 91,
7 (2002).

28. B. H. Lee, L. Kang, R. Nieh, W.-J. Qi, and J. C. Lee, Applied Physics Letters 76,
14 (2000).

29. A. Callegari, E. Cartier, M. Gribelyuk, H. F. Okorn-Schmidt, and T. Zabel, J.
Applied Physics 90, 12 (2001).

30. L. Kang, B. H. Lee, W.-J. Qi, Y. Jeon, R. Nieh, S. Gopalan, K. Onishi, and J. C.
Lee, IEEE Electron Device Letters 21, 181 (2000).

31. K. L. Ng, N. Zhan, M.C. Poon, C. W. Kok, M. Chan, and H. Wong, Electron
Devices Meeting, Hong Kong. Proceedings. 51-4 (2002).

32. N. Zhan, K. L. Ng, M. C. Poon, C. W. Kok, M. Chan, and H. Wong, Electron
Devices Meeting, Hong Kong. Proceedings. 43-6 (2002).

33. E. Atanassova, N. Novkovski, A. Paskaleva, M. Pecovska-Gjorgjevich, Solid-
State Electronics 46, 1887-98 (2002).

34. B. K. Park, J. Park, M. Cho, C. S. Hwang, Applied Physics Letters 80, 13 (2002).

35. W. J. Zhu, T.-P. Ma, T. Tamagawa, J. Kim, and Y. Di, IEEE Electron Device
Letters 23, 97 (2002).

36. M. Ohring, The Materials Science of Thin Films, Academic Press, San Diego,
CA (1992).

37. J. George, Preparation of Thin Films, Marcel Dekker, Inc., New York, NY
(1992).

38. B. Chapman, Glow Discharge Processes, John Wiley & Sons, New York, NY
(1980).

39. C. K. Maiti, S. Maikap, S. Chatterjee, S. K. Nandi, S. K. Samata, Solid-State
Electronics 47, 1995-00 (2003).

40. M. Fadel, O. A. Azim M., O. A. Omer, R. R. Basily, Appl. Phys. A 66, 335-43
(1998).

41. H. Ikeda, S. Goto, K. Honda, M. Sakashita, A. Sakai, S. Aima, and Y. Yasuda, Jp.
J. Appl. Phys. 41, 2476-9 (2002).






69



42. K. Wasa and S. Hayakawa, Handbook of Sputter Deposition Technology, Noyes
Publications, Westwood, NJ (1992).















BIOGRAPHICAL SKETCH

The author was born in Erie, Pennsylvania, and raised in Orlando, Florida, since

the age of two. In high school, the author was an accomplished trumpet player,

performing in ensembles such as the Florida All-State band and the Florida Symphony

Youth Orchestra. The latter included a performance tour of Australia that culminated in

concerts at Sydney Town Hall and the famous Sydney Opera House.

After graduation from Lake Howell High School in Winter Park, Florida, he was

accepted to the University of Florida in Fall 1998. He received a Bachelor of Science

degree in materials science and engineering in June 2003. As an undergraduate, ceramics

was his subject of specialization. Through the 3/2 degree program, the author was able to

enroll in graduate courses while still in pursuit of his undergraduate degree. Thus, he

graduated with a Master of Science degree in materials science and engineering in

December 2003, only a few months after earning his Bachelor of Science. During the

transition from undergraduate to graduate studies, he found a home in Dr. Norton's

electronic oxides group. The author will remain in Dr. Norton's group to study copper

diffusion barriers for his Ph.D. dissertation.