Citation |

- Permanent Link:
- http://ufdc.ufl.edu/AA00014242/00001
## Material Information- Title:
- Scaling effects on metal-oxide-semiconductor device characteristics
- Creator:
- Walstra, Steven V., 1970-
- Publication Date:
- 1997
- Language:
- English
- Physical Description:
- vi, 142 leaves : ill. ; 29 cm.
## Subjects- Subjects / Keywords:
- Capacitance ( jstor )
Doping ( jstor ) Drains ( jstor ) Electric current ( jstor ) Electric potential ( jstor ) Electrons ( jstor ) Modeling ( jstor ) Oxides ( jstor ) Semiconductors ( jstor ) Transistors ( jstor ) Dissertations, Academic -- Electrical and Computer Engineering -- UF ( lcsh ) Electrical and Computer Engineering thesis, Ph.D ( lcsh ) Metal oxide semiconductors ( lcsh ) - Genre:
- bibliography ( marcgt )
non-fiction ( marcgt )
## Notes- Thesis:
- Thesis (Ph.D.)--University of Florida, 1997.
- Bibliography:
- Includes bibliographical references (leaves 132-141).
- General Note:
- Typescript.
- General Note:
- Vita.
- Statement of Responsibility:
- by Steven V. Walstra.
## Record Information- Source Institution:
- University of Florida
- Holding Location:
- University of Florida
- Rights Management:
- Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
- Resource Identifier:
- 028638920 ( ALEPH )
38746309 ( OCLC )
## UFDC Membership |

Downloads |

## This item has the following downloads: |

Full Text |

SCALING EFFECTS ON METAL-OXIDE-SEMICONDUCTOR DEVICE CHARACTERISTICS By STEVEN V. WALSTRA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA SCALING EFFECTS ON METAL-OXIDE-SEMICONDUCTOR DEVICE CHARACTERISTICS By STEVEN V. WALSTRA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1997 ACKNOWLEDGMENTS I would like to thank Prof. Chih-Tang Sah for his time and guidance as chairman of my supervisory committee, and Dr. Arnost Neugroschel, Dr. Toshikazu Nishida, Dr. Sheng Li, and Dr. Randy Chow for serving on my supervisory committee. Additional thanks go to K. Michael Han for many insightful discussions and debates concerning all aspects of device physics. I would also like to thank Dr. Changhong Dai, Dr. Shiuh-Wuu Lee, Mary Wesela, and Jerry Leon for providing the devices, measurement equipment, and technical expertise during my internship at Intel Corporation where the intrinsic capacitance data were taken. Financial support from a Semiconductor Research Corporation Fellowship is also gratefully acknowledged. TABLE OF CONTENTS Pag ACKNOWLEDGMENTS ii ABSTRACT v CHAPTERS 1 INTRODUCTION 1 2 EXTENDING THE ONE-DIMENSIONAL CURRENT MODEL 5 Introduction 5 Background 6 Long-Channel Theory 6 Pao-Sah Model 7 Bulk-Charge Model 14 Charge-Sheet Model 18 Comparison of Long-Channel Models 19 Two-Section Models 20 Beyond Two-Section Models 27 Examples Using Pao-Sah 27 Field-Matching Method 28 Saturation-Voltage Method 30 Surface-Potential Self-Saturation Method 31 In Search of the Match Point 31 Summary 38 3 POLYSILICON-GATE MOS LOW-FREQUENCY CAPACITANCE-VOLTAGE CHARACTERISTICS 41 Introduction 41 Metal-Gate CV 42 Polysilicon-Gate CV 46 Polysilicon-Gate Effects 51 Parameter Extraction Using the LFCV Model 57 3-Point Extraction Methodology 58 3-Region Extraction Methodology 63 Methodology Comparison 65 Convergence Speed-up Details 72 4 THE EFFECT OF INTRINSIC CAPACITANCE DEGRADATION ON CIRCUIT PERFORMANCE 75 Introduction 75 Background 75 Measurement of Intrinsic Capacitances 79 Measurement Configurations 82 Sample Measurements 85 Channel Hot-Carrier Stress Effects on Cgd and Cgs 94 Intrinsic Capacitance Degradation Model 97 Degraded Circuit Simulation 105 Conclusion 110 5 SUMMARY AND CONCLUSIONS 114 APPENDIX METAL-GATE LFCV MODEL DERIVATION 118 REFERENCES 132 BIOGRAPHICAL SKETCH 142 iv Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SCALING EFFECTS ON METAL-OXIDE-SEMICONDUCTOR DEVICE CHARACTERISTICS By Steven V. Walstra December 1997 Chairman: Chih-Tang Sah Major Department: Electrical and Computer Engineering As metal-oxide-semiconductor (MOS) transistor dimensions are decreased, channel-length modulation, polysilicon-gate depletion, and intrinsic-capacitance degradation have increasingly larger impacts on transistor performance. It is demonstrated that the Pao-Sah 1-D current model can be extended to include the channel- length modulation effect by use of a two-section model. This two-section model employs the normal long-channel Pao-Sah model in one region and adds a variable length depletion region in the other. Three methods for matching the boundary between the two regions are presented, with the best results coming from the most complex method of matching the longitudinal fields at the boundary point. The effect of polysilicon-gate depletion on the MOS low-frequency capacitance-voltage (LFCV) characteristics is demonstrated using a Fermi-Dirac-based model. It is shown that, as the oxide thickness decreases, the effect of poly silicon v depletion becomes increasingly pronounced. This depletion, in conjunction with the Fermi-Dirac carrier distribution, offset the current gain expected from thinning the MOS gate oxide. With this polysilicon-gate LFCV model, it is shown that the oxide thickness, flatband voltage, and gate and substrate doping concentrations can be extracted from experimental capacitance data. Two extraction methods, the 3-point and 3-region, are developed and are shown to work well with gate oxide thickness of 130A (2.7% RMS fit) and sub 30 (10% RMS fit). Voltage-accelerated stress is performed on state-of-the-art 0.24 pm effective- channel-length nMOS and pMOS devices to assess the impact on the most important intrinsic capacitances: Cgd and Cgs. The nMOS devices exhibit a Cgd reduction and Cgs enhancement with stress time, whereas the pMOS devices show negligible change. Because of Miller feedback, the nMOS Cgd reduction dominates the Cgs increase, resulting in an overall CMOS capacitive load reduction. Pre-stress and post-stress ID, Cgd, and Cgs data were fit using the BSIM3 device model. With the resulting parameter sets, a 31-stage ring oscillator was simulated for three situations: unstressed devices, stressed devices only including ID degradation, and stressed devices including ID, Cgd, and Cgs degradation. It is shown that the inclusion of the intrinsic capacitance degradation results in improved simulated circuit performance because the capacitive load reduction offsets the drain current reduction. This improved degradation methodology will result in looser guardbands and less reliability redesign. vi CHAPTER 1 INTRODUCTION The last three decades of production integrated circuits (IC) have seen two orders of magnitude decrease in device dimensions, from 25 pin in 1962 to 0.25 pm in 1997 [1-3]. This continual reduction, fueled by requirements for higher switching speeds, lower cost, and decreased power, has been sustained by improvements in lithography and has resulted in increased areal and chip densities (transistors/cm2 and transistors/chip). Compared to the -500 transistors/chip in the first experimental 64-bit static random- access memory (SRAM) in 1965, the -64M transistors/chip 64 Mbit dynamic random- access memory (DRAM) of 1997 and the -4G transistor/chip 4 Gbit DRAMs due from NEC in 2000 typify the strong push toward increased density. Increased areal density implies decreased dimensions. As transistor and capacitor dimensions decrease, previously negligible effects have become or are becoming increasingly important. Many of these effects were assumed avoidable through constant-field scaling [4], These scaling rules have been debated, amended, and improved [5-7] to account for noise-margin, hot-electron, and extrinsic-capacitance considerations, but present and future smaller dimensions have necessitated these effects be included in the design process. Several of these effects are discussed below. As channel lengths decrease, the thickness of the space-charge region at the drain of a metal-oxide-semiconductor (MOS) transistor becomes a significant fraction of 1 2 the total channel length. As the drain voltage is changed, the space-charge-region thickness also changes, resulting in an effective channel length which is drain-voltage dependent, an effect known as channel-length modulation (CLM). This problem can be tolerated for complementary MOS (CMOS) logic circuits, but needs to be properly modeled in order to predict the drive current of the MOS transistors in the circuit in order to estimate the speed of the resulting circuit. As the density of transistors increases, so does the power density. This requires a reduction in the operating voltage, since the active (switching) output power is proportional to the square of the operating voltage (PactÂ¡ve fciockCoV2). To obtain the same performance at lower voltages, the oxide thickness must be reduced. Simple MOS theory predicts the drain current is inversely proportional to the gate oxide thickness. However, for thin oxides (< 50 A), depletion of the polysilicon gate offsets the effects of thinner oxides, resulting in lower current and diminishing returns on oxide scaling. Additionally, the gate voltage cannot be reduced indefinitely, because a large enough margin is needed between the signal voltage and ground-plane noise to ensure that noise does not change the state of the device. The increased density of transistors also requires more closely-spaced interconnections between the transistors. Interconnect scaling has made delays due to the interconnection a limiter in process speed [8], and major efforts are currently underway to reduce the interconnect resistance and capacitance. Copper has recently been introduced into 1998 production by IBM to reduce the interconnect resistance. Additional efforts have been underway to lower the dielectric constant of the intermetal dielectrics in order 3 to reduce the interconnect capacitance. When the interconnect capacitance is reduced, the only remaining capacitance left to slow the CMOS circuitry is the intrinsic capacitance of the transistors, which cannot be easily reduced and will become the predominant speed limiter. There are many other issues concerning the perpetual reduction in transistor dimensions, the least of which is the brick wall of atomic dimensions. Clearly transistors cannot be scaled to less than ten or twenty atoms and still work in the traditional sense of transistors, yet this dissertation includes data from a transistor pushing the atomic limit with a gate insulator thickness of less than 30, or under six atomic layers of silicon and oxygen. The goal of this dissertation is to investigate the issues described in the previous paragraphs. Chapter 2 discusses the history of 1-dimensional drain current models and some of the methods which have been implemented to extend these models to include the CLM effect. The Pao-Sah model, the most accurate long-channel current models, will be extended to include the CLM effect using three different approaches. The CLM effect (as demonstrated in the new models) will be discussed, as well as the pros and cons of the approaches. Chapter 3 tackles the polysilicon depletion problem by deriving the Fermi- Dirac-statistics-based polysilicon-gate MOS low-frequency capacitance model, including the effect of dopant impurity deionziation. By comparing this with the traditional metal- gate model, the effect of polysilicon gate depletion will be shown to increase significantly as the oxide thins. With this model, a parameter extraction methodology is presented 4 which allows the extraction of substrate and gate doping concentrations as well as the oxide thickness and flatband voltage from experimental LFCV data. Two methodologies will be presented and compared, and data from thick (130) and thin (< 30) gate-oxide devices will be used. Additional oxide thickness issues, such as quantum effects, are also discussed. Chapter 4 considers the intrinsic capacitances, in particular, those most important in modern complementary MOS (CMOS) circuits: Cgd and Cgs. Compared to the drain current, which is also an intrinsic property of a MOS transistor, intrinsic capacitances have been relatively ignored because of measurement difficulty and relatively small impact compared to extrinsic capacitances. However, as processing and dielectric technology advances, the primary remaining capacitive load in CMOS circuits will be the intrinsic capacitances. The chapter presents an experimental investigation how these capacitances change with hot-carrier stress and, after modeling the stress- induced changes in the intrinsic capacitances, shows that part of the drain current degradation is offset by the intrinsic capacitance reduction, resulting in a slower degradation of overall circuit performance. CHAPTER 2 EXTENDING THE ONE-DIMENSIONAL CURRENT MODEL Introduction The simplest 1-D model is of crucial importance for applications in semiconductor physics. Although 3-D models will best match experimental data because of both inclusion of real effects and simply additional variables, they may be intractable as compact device models, where computational efficiency is critical. Conversely, these 3-D models are often validated by demonstrating their reduction to the rigorous 1-D forms for non-critical (wide and long channels with thick oxides) geometries. For back- of-the-envelope calculations, knowledge of the basic physics embodied in a good 1-D model is exceedingly useful. The required accuracy of a model is largely determined by the application. For predicting the drive current, such as might be required for a discrete-transistor specification sheet, a model need not worry about the linear or subthreshold regions of operation. Similarly, if modeling only the operating range (0 to power supply voltage), then the accumulation region of applied gate voltages can be ignored in the model. There are cases, particularly when attempting to predict the performance of new technology, where 3-D full-range MOSFET models are necessary, but they are a relative minority compared to the wide array of applications for 1-D models. 5 6 This chapter contains a brief history of one-dimensional (1-D) approaches to drain current models, including calculations and comparisons, followed by a new two- section model using the 1-D Pao-Sah long-channel IV model in conjunction with a variable-length depletion region. The goal is to extend the 1-D long-channel model to short-channel use. Background In 1926 Lilienfeld [9] submitted the patent for the first MOSFET device, an A1/A1203/Cu2S transistor. Thirty-two years later in 1960, Kahng and Atalla [10] fabricated the first silicon MOS transistor. A year later, the first MOST current-voltage (IV) papers were published internally at AT&T Bell Labs in 1961 by Kahng [11] and later at Stanford by Ihantola [12]. These were followed in 1964 by more complete (and widely released) 1-D theories by Sah [13] and Ihantola and Moll [14]. A comprehensive history of MOS developments was reviewed by Sah [1], In the subsequent years since the first MOST model, hundreds of papers and theses have been written about the modeling of various aspects of MOS transistors. This chapter will discuss the prevailing 1-D models including Pao-Sah, bulk-charge, charge-sheet, and the many two-section models. Long-Channel Theory "Long channel" is a term used to specify that short-channel effects can be neglected when modeling MOSTs, and the predominant short-channel effect is encroachment of the drain depletion region into the channel. The depletion region exists 7 due to the reverse-biased p/n junction between the substrate and the drain, and has nothing to do with the actual channel length. For long-channel devices, however, the amount of encroachment relative to the channel length is small, so the effective channel length is essentially constant (equal to the drawn gate length). For short channels, however, the effective channel length can be significantly reduced by the encroachment. Another short-channel effect neglected in long-channel theory is drain-induced barrier lowering [15], where the source barrier is lowered by the applied drain voltage. Pao-Sah Model The most accurate long-channel theory was published by Pao and Sah (PS) [16]. The PS model is the only one which correctly accounted for drift and diffusion. The PS theory, to be discussed below, contains a double integral, but can be reduced to a more efficient form containing only single integrals [17, 18], Although cumbersome to calculate, the PS double integral is extremely didactic and is a useful starting point for showing the approximations used to derive other long-channel IV models. The total current flowing in the channel is given by the integral [xi In = J(x,y)Z dx, (2.1) 0 where J(x,y) = JN + Jp = JN = qpnNEy + qDnVN = qDnNVÂ£ (2.2) JN and JP are the electron and hole current densities, respectively, and it is assumed that the current is dominated by electrons in an n-channel device in (2.2). The electron charge is q, pn and Dn are the electron mobility and diffusion respectively, and VN is the gradient of the electron concentration. The electron quasi-Fermi level,is measured relative the bulk Fermi level and normalized to kT/q. If d^/dx is assumed negligible (which is a fundamental assumption in the long-channel approximation and should be valid to a depth on the order of the drain junction depth), then ID can be found from summing up all the current from the surface down to some depth xÂ¡ below which the additional contribution is negligible: Id qDnZ(df/dy) xi N (x) dx 0 This can be transformed from physical space in the y direction to potential space as follows: (L ruD ID dy = qDnZ 0 0 (2.3) where UD=qVDS/(kT) is the normalized drain voltage at y=L and the lower limit 0 is the grounded source voltage at y=0. A similar transform in the x direction yields: Id = <2Dn~ Un Us N(U) dU (2.4) LJ0 JUF (-dU/dx) where (dU/dx) is the x-component of the electric field, which can easily derived from integrating Poissons equation by quadrature and is given below. The Boltzmann approximation to the carrier concentration is being used and the impurities are assumed completely ionized, but the Fermi-Dirac and deionized form can be used. Us is the normalized surface potential (where surface is at x=0), the total amount of surface band bending relative to the intrinsic Fermi level. It is a function of both the gate voltage and the drain voltage. UF is the normalized bulk Fermi level, below which the current 9 contribution is assumed negligible, and is analogous to the to physical point x=xÂ¡ in (2.3). The derivative (dU/dx) is found from (-dU/dx) = F(U,Â£,Uf)/Ld (2.5) where F (U, Â£, UF) = [exp (U-Â£-Up) + exp(UF-U) + (U-l)exp(UF) - (U+exp (-Â£) ) exp (UF) ]1,2 (2.6) After applying Einsteins relationship, Dn/pn = kT/q, (2.4) becomes -dUdÂ£ (2.7) The surface potential, Us(Â¡;), is needed in (2.7). The relationship between the surface potential and the gate voltage can be found by applying Gausss Law at the semiconductor/insulator interface. The resulting equation, given below, can be solved iteratively for Us for a given rkT 2 z K*b UD Us exp (UUF) q 2L Ld 0 UF F(U, f,UF) U = U* sign(Us) yF(Us,|,UF) (2.8) where U0 is the normalized gate voltage, q(VGS VFB)/kT; y is es/(LDC0); LD is the Debye length (V[eskT/(2nÂ¡)]/q); and F(US4,UP) is given by (2.6). Equation 2.7 is the traditional form of the PS integral, often called the Pao-Sah double integral. A more computationally friendly and accurate single-integral form [17] was used for the calculations in this dissertation. The mobility in (2.7) need not be taken out of the integrals. Instead, it can be a function of the vertical and lateral fields and moved inside of the integrals. In this chapter the mobility will be assumed independent of field. 10 A good way to understand Eq. 2.7 is to consider the three-dimensional band structure of a MOST under gate and drain bias, as shown in Figures 2.1-2.4, based on the original Pao-Sah paper [16], Figure 2.1 shows an idealized n-channel MOST. Figure 2.2 shows the corresponding energy band diagram with no applied terminal voltages except VGS=VFB. From the position of the Fermi level it is easily verified that the source and drain are n-type and the substrate is p-type (n-channel device). Electrons in the source and drain see a potential barrier toward the channel. Application of a positive voltage to the gate lowers the barrier near the surface, as shown in Figure 2.3. The applied gate voltage pulls electrons toward the surface (and pushes holes away from the surface), as can be seen from the position of the Fermi-level relative to the band edges. Farther into the substrate (away from the gate/substrate interface) there is no bending from the gate potential, so the region is identical to the unbiased case (Figure 2.2) and considered quasi-neutral. Applying a voltage to the drain (VDS < VDSsat) splits the Fermi level into quasi- Fermi levels (Fn for electrons and Fp for holes), as shown in Figure 2.4. One can imagine an electron in the conduction band surmounting the source barrier and then falling down the potential cliff until reaching the drain. This free fall is where the electron gains energy while moving across the channel. If the electron is not scattered while moving across the channel (losing energy to the lattice via phonons), it becomes increasingly energetic as it approaches the drain and may become hot enough to produce an e-h pair via impact, the resulting hole may generate interface traps via dehydrogenation of Si-H bonds near the Si/Si02 interface [19], This is only one of several mechanisms for interface trap generation. Fig. 2.1 / D s G Simplified view of two-dimensional MOS device. Fig. 2.2 Schematic 2-D energy band diagram of simple MOS device with source and drain grounded and VGS=VFB. Adapted from Pao and Sah [16]. 12 Fig. 2.3 Schematic 2-D energy band diagram of simple MOS device with VGS > VFB, drain and source grounded. Adapted from Pao and Sah [16]. 13 Fig. 2.4 Schematic 2-D energy band diagram of simple MOS device with VGS > VFB, 0 < VDS < VDSsat, and source grounded. Adapted from Pao and Sah [16]. 14 Figure 2.5 shows the result of applying a drain voltage in excess of VDSsat. As will be discussed in the two-section model section later, the drain depletion region becomes increasingly longer as the reverse-biased drain voltage increases. For this long-channel section of the dissertation, however, the change in length, AL, is assumed much less than the channel length L. The voltage drop across this thin depletion region often results in large fields which can greatly accelerate carriers, causing the interface damage mentioned above. Now that the effect of applied biases on the 2-D structure of the band has been discussed, it is easy to see the basis of the integral limits in Equation 2.7. The inner integral is integrating from the surface into the bulk (from Us to Up), which is a cross section of the channel as shown in Figure 2.6. The outer integral is integrating from drain to the source (UD to 0, source is grounded) along the channel. Thus, the double integral is summing up all the current contribution in the channel, exactly as would be expected. Since Us is a function of the drain voltage (or the channel potential), the order of the double integration is not trivially reversible. Bulk-Charge Model The first group of ID models, in order of complexity, were by Sah [13], Ihantola and Moll [14], and Sah and Pao [20]. These are all bulk charge models, taking increasingly more into account. As the name suggests, the bulk charge model takes the depleted region under the channel (in the bulk) into account. It assumes drift is the major component and so neglects the diffusion component. This greatly simplifies the problem and reduces (2.2) to J(x,y) = JN + Jp = JN = qpnNEy = qpnN(x) (dV/dy) (2.9) 15 Fig. 2.5 Schematic 2-D energy band diagram of simple MOS device with VGS > 0, Vds > ^DSsat an<^ source grounded. Adapted from Pao and Sah [16]. 16 Fig. 2.6 Schematic 2-D energy band diagram of simple MOS device with VGS > 0 and VDS < VDSsal. Cross-sections show the 1-D energy-band diagrams near the source and drain. G^Gate electrode, XsSubstrate electrode. 17 Id q/tnZ (dV/dy) i qN (x) dx 0 (2.10) ID = -nnZ(dV/dy)Qn (2.11) where Qn = -C0(VG v -Vs0) + (2qPxxÂ£s)1/2[Vso + V]1/2 (2.12) C0 is the oxide capacitance per unit area, V is Vos VFB, Vso is the surface potential at the source, and V is the channel potential (=VDS at the unsaturated drain). Pxx is the substrate impurity concentration and es is the dielectric constant of silicon. The first term is the charge accumulated in the channel and the second term is the uncompensated charge in the depletion region beneath the channel (i.e. bulk charge). Integrating (2.12) along the channel gives: ID = pn(Z/L)C0{ (VG Vs0)VDS V2s/2 (2.13) (1/C0) (2/3) (2qPxxÂ£s)1/2 [ (Vs0 + VDS)3/2 (Vs0)3/2]} This form is slightly different than the Sah-Pao and Ihantola-Moll forms because it is not assumed that VS0=2VF, where VF is the Fermi voltage. A more exact form [17] is: ID = pn(Z/L)C0{ Vg(Vsl Vs0) (1/2) where VSL is the surface potential at the drain. This differs from (2.13) in that the surface potential at the drain is calculated instead of assumed to be VSL=VS0 + VDS. When the drain current approaches or exceeds saturation (VDS > VDSSat), VSLsÂ£VS0 + VDS. Additionally, in subthreshold, VSL is typically closer to Vso than Vso + VDS [21]. As will 18 be shown later, the bulk charge formula should never be used for subthreshold calculations since it neglects diffusion, which is the primary subthreshold current contribution. The bulk charge form, compared to PS, is considerably easier to calculate, particularly when using (2.13) with Vso = 2VF, but is invalid in subthreshold. Equation 2.13 is also invalid in saturation as written, but that can be fixed somewhat by calculating the saturation voltage VDSsat and fixing the current for all drain voltages greater than VDSsat. This will make the first derivative (drain conductance) non-continuous at VDS=VDSsat. All saturation problems are solved in (2.14), where the calculation of VSL negates these problems. Iterative calculation of VSL is time consuming, particularly compared to assuming a constant, or pinned, surface potential value. Charge-Sheet Model While most of the interest centered on super-threshold operation of the MOST, some people became concerned with the lack of accurate modeling for subthreshold operation. Barron [21] and Van Overstaeten et al. [22] developed subthreshold formulae based on simplifications of the Pao-Sah integral, with results applicable only to the subthreshold region. Six years later, Brews [23] made a critical approximation which would allow both drift and diffusion components to be introduced simultaneously without the need for a double (or single) integral. When he proposed his "charge-sheet model," he introduced the following simplification: I = qZ/JnN(y) (df/dy) dÂ£/dy=d0s/dy 1/p dln(n)/dy (2.15) 19 This approximation for dZJdy was justified "based upon its success in producing correct I- V curves," although he added a footnote relating the formula to electrochemical potential. This wide-open statement resulted in several subsequent proofs which derived the same formula [17, 24, 25], Essentially, though, he decoupled the drift and diffusion components from the tight interdependency seen in the Pao-Sah form to the simple form of (2.15). Through a similar derivation to bulk-charge, ID is given by I=/in(Z/y) (1//3) {C0(l/j3+VG) (Vs(y) Vs0) (1/2) C0 (V2 (y) V|0) (2.16) - (3/2) (2qPxxes)1/2[ (0Vs(y) 1)3/2 (pVs0 1)3/2] + (2qPxxÂ£s)1/2[ (/JVs(y) 1)1/2 (0VSO 1)1/21) Eq. 2.16 reduces to bulk-charge form of Eq. 2.14 if VG, Vs(y), Vso 1/p and the square root terms are negligible. Unlike bulk-charge, this formula is valid in subthreshold and does not require a calculation of VDSsat (assuming VSL and Vso are calculated iteratively). Like bulk-charge, this is much easier to calculate than a double, or even single, integral. Brews, and many subsequent authors, validated the charge-sheet model by comparing it to the results of the Pao-Sah formula. It has been shown to be an excellent approximation, as will be discussed in the next section. Comparison of Long-Channel Models The Pao-Sah double-integral model has been heralded as the best long-channel model. Brews [23] went so far to say that "Comparison of the charge-sheet model with the Pao-Sah model has the force of comparison with experiment, since the Pao-Sah model is known to work well for long channel devices." Schrimpf et al. [26] agreed, saying Pao and Sah "produced a quantitative model so accurate that it is the standard by which other models 20 are judged." Since bulk-charge and charge-sheet are both approximations to Pao-Sah, it makes sense to compare them with Pao-Sah to see how accurate they are, taking into account that all the models are only valid for long-channel devices. Figure 2.7 shows all three methods simulated for Tox=500 , T=296 K, Pxx=1015 cm"3, W/L=10. These are typical parameters for LSI devices of the 1970s, and were chosen to match the data used in Pierret and Sheilds [17]. As can be seen, the bulk-charge and charge-sheet models underestimate the current. Figure 2.8 shows the percentage error for each model at the gate voltages shown in Fig. 2.7, demonstrating that the charge-sheet model maintains an error of less than 2.6% for all gate voltages, while the bulk charge model ranges from 2.5% for VCS=5.0V to 8.4% for Vos=2.0 V. This suggests that the much simpler charge sheet can be used in place of Pao-Sah incurring only about 2.5% error at low voltages. Figure 2.9 shows the subthreshold region for the same device with VDS=0.1 V. Clearly demonstrated in this figure is both the glaring inadequacy of the bulk-charge model for subthreshold modeling and the remarkable accuracy of the simple charge-sheet model. However, recall that this is charge-sheet with iteratively calculated surface potentials, so the numerical solution is not entirely trivial. Two-Section Models Up until now, only long-channel ID equations have been considered. For short- channel devices (<1 pm), the most prominant non-modeled effect on the drain current is finite drain conductance beyond saturation. The primary cause of this non-zero drain- conductance (gD) is channel shortening from the drain space-charge region (SCR) 21 ID versus VDS for different VGS values for the three 1-D ID models. Parameters are Tox=500 , T=296 K, Pxx=1015 cm'3, W/L=10, which were used to match data in Pierret and Shields [17]. 22 Fig. 2.8 Percentage error in ID for charge sheet and bulk charge relative to Pao- Sah versus VDS, from Fig. 2.7. Plots are Vos = 5, 4, 3, and 2 V, with higher errors for lower voltages. 23 VG (V) Fig. 2.9 ID versus VGS for Pao-Sah, charge sheet, and bulk charge using same data as Fig. 2.7 with VDS=0.1 V. Clearly bulk charge is not useful in subthreshold, whereas charge-sheet is almost coincident with Pao-Sah. 24 encroaching into the channel. This effect is often called channel-length modulation since the drain voltage modulates the effective channel length. The most logical approach is to divide the region between the source and drain into two sections: a source side and a drain side. The source side may contain any appropriate long-channel IV model, such as Pao-Sah, charge sheet, or bulk charge. The drain region is the depletion region, and can be modeled with or without mobile charge, 2- D effects, mobility differences, etc. The location of the boundary between these regions, and the voltages and fields at this boundary, are what make this a challenging problem. Figure 2.10 shows a diagram of a MOS transistor divided into two sections. There are essentially three things which differ among approaches to two-section theory: the the source-side IV model, the drain-side space-charge region (SCR) model, and the boundary conditions. Source Source Side (Long channel approx) Drain Side (SCR) Drain LeW=(L-AL) ' AL L y=o y=yM y=L v=o v=vM v=vD Fig. 2.10 Schematic diagram of two-section MOST for 1-D modeling. SCR means Space-Charge Region and Leff refers to the effective channel length. 25 The IV model can be one of the many already discussed. The SCR model can be assumed fully depleted, take mobile charge into account, or be a complete 2- or 3-D model. The boundary conditions are the most difficult and varied among approaches. Essentially, the potentials, fields, and charge at the boundary between the two regions need to be matched. The simplest two-section MOST model was introduced in 1965 by Reddi and Sah [27]. They used a source-side bulk-charge model for the current and a fully-depleted drain- side depletion model. From the first derivative of the bulk-charge model (Eq. 2.13 with Vs=2Vf), Reddi-Sah (and others) calculated the drain voltage where, for a constant gate voltage, the drain conductance drops to zero (VDSsat). They then assumed all voltage in excess of VDSsat falls across the SCR to form the drain region of the two-section model. By assuming complete depletion (no mobile charge) and no y-field at the boundary, the length can be calculated from simple p/n junction theory as: AL = [2es (VDS VDSsat + Vbi)/(qPxx)]1/2 (2.17) where Vbj = (kTq)ln(NdrainNsubslrate/nf) from standard abrupt-junction p/n theory. Replacing L by Leff=L-AL and Vso with 2VP in (2.13) yields the Reddi-Sah two-section current. The simplicity of this formula is extremely attractive, but the solution is dependent on the ID model. Specifically, it assumes that a VDSsat voltage can be found. If using Pao- Sah or charge-sheet, the surface potential is not constant and a VDSsat point does not actually exist. Even if VDSsat is found from extrapolation, the first derivatives of the drain current will be non-smooth at the point where the drain current switches from one model (Pao-Sah, 26 charge-sheet, bulk-charge) to another (constant ID), although this can be fixed with various smoothing transitional functions. Four years after Reddi and Sahs paper, Chiu and Sah [28] came out with a two- section model which solved Laplace equation in the oxide layer and matched values in four regions (source, drain, oxide, and bulk). The drain region was solved as a 2-D, fully- depleted region, and the solution required seven matching parameters. The complexity of the solution relinquished this model to an almost constant reference as "too complex. The following year (1969) Frohman-Bentchkowsky and Grove [29] developed a two-section model using bulk-charge model in the source region and an empirical model for the drain section. This simple model essentially added two additional fringe field contributions to the Reddi-Sah model and added two empirical variables to fit the data. Merckel, Borel, and Cupcea [30] added mobile charge to the drain region empirically by writing Poissons equation in the drain region as d2V/dy2 = q/Â£s(Pxx + IDS)/(qZa) (2.18) where a is essentially a fitting parameter related to the junction depth. This mobile charge is akin to the Kirk effect in bipolar devices, just as the drain-depletion encroachment is analogous to the Early effect. Using an iteratively determined VDSsaI, they were able to calculate the drain depletion width. Popa [31] devised a similar model and extended the drain depletion region to be of three types depending on the injected current. In both mobile-charge cases, fitting parameters were introduced either through (2.18) or mobility. Both used variations of the simple bulk charge model for the source side. 27 After Brews developed the charge-sheet model, all subsequent two-section models employed the charge-sheet model. Guebels and Van de Wiele [32] developed a three- section model to account for the x-field reversal near the drain. They employ the same trick as the previous papers by fitting the a in (2.18), using VDSsat (or IDsat) and adding some empiricism to their field calculations. Beyond Two-Section Models The charge-sheet model (and Pao-Sah, as will be shown) does not lend itself well to analytical two-section models due to the greater complexity of the drain current model relative to bulk charge. As noted above, fitting parameters and empirical formulae were required to be introduced to satisfy some of the boundary conditions. The newer compact models, such as BSIM [33,34] and Siemens [35-37] model, are based loosely on one-section bulk-charge and charge-sheet models, respectively, sometimes dividing the model into different sections based on operation (separate subthreshold and superthreshold formulae). They both model short-channel effects by adding semi-empirical additions to the threshold voltage, which makes for a considerably faster calculation speed at the expense of a less-physical model. Examples Using Pao-Sah The goal was to develop a two-section model which employs the Pao-Sah integral as the source-side current formula. The following is a description of the methodology and results of the exercise. 28 Field-Matching Method The Pao-Sah current has already been discussed, as have been models for the depletion region. Let us consider the matching boundary of the two section model to occur at the point y=yM where the channel voltage is VM with a lateral field EM and electric field gradient d2Us/dy2=dEM/dy. A simple way to look at this problem is from the Poissons equation in the drain region while considering the boundary conditions. Within the drain region, which extends from y=yM to y=L, the boundary conditions are (see Fig 2.10): V(L)=Vds v(yM)=vM dV(yM)/dy=EM (field at the match point) d2V(yM)/dy2= (l/es)[qPXx + (mobile charge terms)] = C It is possible from Pao-Sah to calculate dV(yM)/dy=EMps [38]. This gives us the following equations after integrating the Poissons equation twice with the above boundary conditions: The ideal additional equation would be d2V(yM)/dy2 on the Pao-Sah side, but this quantity is incalculable from the Pao-Sah integral. If it is assumed that assume Em=0 (as was done in Reddi-Sah), the depletion length into the channel can be easily found. It is reasonable to assume that the lateral field at the matching point (EM) is much less than the field right at the drain (ED), so ED Em- making the difference in yM small. This gives (from 2.20, also 2.17) 29 yM = L (2 (VDS VM + Vbi)/C)1/2 (2.21) Where Vbi accounts for the pre-existing depletion region originating from the abrupt p/n junction. Since the yM approximation has already been made, it will be assumed that the field throughout the drain region is a constant at the boundary and is given by EMdep = (VDS Vm)/(L yM) (2.22) Clearly there are conflicting assumptions (EM = 0, and now EM 0). One might wonder why Em is not (VDS VM + Vbi)/(L yM) to be consistent with 2.21. This comes from the subtlety of the boundary conditions. Looking back to Figure 2.2, note that the integration is actually from Vs + Vbj to VDS + Vbj, which excludes the p/n depletion layers. The Vbis cancel out for symmetrical devices, so this is no problem. At VDS=0 (and source grounded), no current or field is expected, which would make VM correctly equal to 0 in (2.22). However, if Vbi were added to (2.22), then VM would have to equal VbÂ¡, which would incorrectly cause a field (and possibly current flow depending on Vos). Essentially, (2.22) gives the excess field. However, Vbi does contribute to the depletion width, so it is included in (2.21). The normalized field on the Pao-Sah side at the boundary is given by [32] [ exp (Us) -1 ] exp (-UM-UF) 'Mps 2 r - F(Us,Um,Uf) + exp (Us-UM-UF)-exp (UF-Us)+exp(UF)-exp (-UF) V L m(us exp(U-f-UF) -dUdf 0 UF F(U,f,UF) x dU (2.23) 30 where UM is the normalized matching voltage, VM*(q/kT). Figure 2.11 shows the results of this approach, with mobile charge terms neglected (C=qPxx/esi) for Pxx=5xl017 cm-3, T=300 K, and Tox=50 . The data cover a wide range of channel lengths from 'U pm to , and for all cases the width is equal to the length (square devices). The saturation current predicted by long-channel theory for these square devices would be the same for all channel lengths, so the deviation from this is due to channel- length modulation, which clearly becomes more important and the channel length decreases. Figure 2.12 shows that the drain conductance (gD=dID/dVDS) is smooth, which is important for circuit simulator applications. Although not shown, the derivative of the drain conductance is also smooth. Thus, this field-matching model successfully extends the 1-D Pao-Sah model to short-channels, at least with regards to including the effective channel shortening effect. Saturation-Voltage Method Reddi and Sah [27] assumed VM=VDSsat, which simplified things considerably. VDssat *s easy to calculate when using the bulk-charge formula assuming VS0=2VF since the derivative of the surface potential with respect to the drain voltage is zero. The Pao-Sah current, however, does not technically saturate (numerically there will be a point where the current does not increase, but it will be at a drain voltage well in excess of the normal VDSsat point). This problem is solved by extrapolating VDSsat from dID/dVDS versus VDS without channel shortening. Figures 2.13 and 2.14 show the results of employing this method with the same device as used in the previous section (Pxx=5xl017 cm-3, T=300 K, Tox=50 ), using Eq. 2.21 for yM with VM=VDSsat. Clearly the channel-length modulation is being 31 accounted for, but the transition is slightly abrupt. A look at the resulting drain conductance (Fig. 2.14) shows a drastic discontinuity near the calculated VDSsat point. Use of a fitting function could rectify this derivative problem, and is a common practice for compact models. Surface-Potential Self-Saturation Method Another possible way to circumvent finding the VM point was posed by Katto and Itoh [39], Instead of finding VDSsat, they used the fact that the surface potential itself will saturate when solved iteratively from (2.8). Thus replacing the matching voltage, VM, with the surface potential at the drain, VSL (solved iteratively) gives another decoupled way to solve for yM. Using the surface potential to find the depletion thickness was also used by Sah [2], This is better than the VDSsat method since there will not be an immediate point where saturation occurs. However, as shown in Figs. 2.15 and 2.16, the current still has a slight jump resulting in discontinuities in gD. In Search of the Match Point Sah [2] showed pictorially that in saturation, the energy band near the drain edge will actually be bent upward, or in other words, the surface will be accumulated rather than inverted (actually, the surface will still be depleted, but now accumulation refers only to the shape of the band bending). This must be the case since the potential along the channel is actually higher than VGS VGT = VDSsat. This means that there must be a point along the channel at which the band bending is zero at the surface, and this point would be an excellent candidate for the yM point. Like the methods above, however, this point has some 32 Fig. 2.11 ID versus VDS plots for different channel lengths (square devices) using field matching at the match point. Vos = 5 V, Pxx=5xl017 cm-3, T=300K, Tox=50 . gD (mA/V) 33 Fig. 2.12 gD versus VDS plots for different channel length (square devices) using field matching at the match point. Same parameters as Fig. 2.11. 34 Fig. 2, 13 ID versus VDS plots for different channel lengths (square devices) using VM=iterative surface potential at drain. VGS = 5 V, Pxx=5xl017 cm-3, T=300K, Tox=50 . gD (mAA/) 35 Fig. 2.14 gD versus VDS plots for different channel length (square devices) using VM=iterative surface potential at drain. Same parameters as Fig. 2.13. 36 Fig. 2.15 ID versus VDS plots for different channel lengths (square devices) using VM=VDSsac vcs = 5 V, Pxx=5xlOn cnr3, T=300K, Tox=50 . gD (mA/V) 37 Fig. 2.16 gD versus VDS plots for different channel length (square devices) using VM=VDSsaf Same parameters as Fig. 2.15. 38 logical flaws. For instance, the field in the x-direction is zero by definition, which means that using the channel potential at this point to calculate the current from the long-channel model will clearly invalidate the gradual channel approximation (Ex EY), a basis for the Pao-Sah ID derivation. A simple approximation for this point would be to use VM = VGS VFB when VDS > VGS VFB, which is akin to setting VDSsat = VGS VFB. This ends up resulting in the same sort of problem seen in the VDSsat method. It is interesting to verify the existence of this turn-around region in the channel near the drain, however. This was done recently using the MINIMOS device simulator [40]. MINIMOS was modified to use a constant mobility model so as to be comparable to the 1- D model cases above. Figure 2.17 shows the resulting electrostatic potentials into the substrate at different points near the drain edge of a 50, 100x100 pm (corrected for subdiffusion) nMOST with Pxx=5xl017 cm-3 at VGS=1.5 V and VDS=3.0 V. VFB was fixed at zero for this case. What is clear is that the band moves from inversion (top) through flatband into accumulation (bottom) at the surface (x=0.0). The flatband point occurs when the channel potential is equal to VGS VFB = 1.5 V, as expected. Summary This chapter reviewed the history of 1-D long-channel drain-current models and discussed the pros and cons of their derivation and applications. From this, the importance of a non-pinned surface potential was shown, as demonstrated by the excellent approximation of the simple charge-sheet model to the Pao-Sah double integral-the best of the 1-D long-channel models. 39 Next, methods to extend the 1-D into two 1-D sections to create the best full-range 1-D model. It was discovered that, no matter what, the depletion region is strictly 2-D, and obtaining a 1-D approximation requires rather substantial assumptions. One model, the field-matching approach, was seen to give reasonably good characteristics, while all the other approximations (VDSsat and surface-potential self-saturation) resulted in discontinuities in the first (and higher) derivatives. A 2-D simulation was used to verify that there is a point in the saturated channel where the (x-directed) field reverses and the surface band bending is, thus, zero. This point has been suggested many times before in our group, but never verified two-dimensionally. Attempting to use this point to demark the boundary of the source region and drain region of the two-section model results in the same poor results as the VDSsat method. 40 SiOo/Si interface X (Â¡im) Fig. 2.17 Electrostatic potentials into the substrate at different points near the drain edge of a 50, 100x100 pm (corrected for subdiffusion) nMOST with Pxx=5xl017 cm3 at Vcs=1.5 V and VDS=3.0 V. The Y=1.321 pm (near source) and Y=50.000 pm (middle of the channel) curves are indistinguishable. The band is flat at the Si02/Si surface when the channel electrostatic potential equals Vos -VFB = 1.5 V (VFB = 0 for this data). CHAPTER 3 POLYSILICON-GATE MOS LOW-FREQUENCY CAPACITANCE-VOLT AGE CHARACTERISTICS Introduction For modern ULSI technology, polysilicon gates are universally used on MOS devices. With respect to MOS device characteristics, there is no advantage to substituting metal gates with heavily-doped polysilicon (poly) gates. In fact, poly gates, as will be shown in this chapter, greatly reduce the effectiveness of thinning the oxide layer to increase the drain current. The use of poly gates is a question of cost as well as performance, however, and poly gates have some tremendous processing and density benefits over metal gates. Polysilicon gates can withstand high temperature steps that would cause most deposited metal gates to evaporate, particularly the source/drain drive- in step. Polysilicon gates also allow for self-alignment of the gate over the oxide between the source and drain, removing what would be the most difficult (and costly) alignment step in the process flow [41-42], This chapter covers the derivation of a Fermi-Dirac-based polysilicon-gate MOS low-frequency capacitance-voltage model. This model will be used to illustrate the effects of polysilicon gates on MOS low-frequency (LF) capacitance-voltage (CV) characteristics compared to metal-gate LFCV characteristics. A useful application for the model is physical parameter extraction, which is demonstrated in this chapter using two 41 42 different methodologies: 3-point fit and 3-region fit. Sample parameter extractions for thick (130) and thin (20) gate oxides are shown, and discussion about limitations of the model are presented. Quantum effects are purposely ignored, and the reasoning behind this decision is discussed. Important details related to fast convergence of the parameter-extraction routines are also given. Metal-Gate CV Ideal metal-gate CV theory using Boltzmann statistics has been extensively discussed [3,43], as well the extension to include Fermi-Dirac carrier distribution and deionization effects [44-46]. The appendix contains the full metal-gate LFCV model derivation, taking Fermi-Dirac statistics and deionization into account. The relevant solutions are given below. Figure 3.1 shows a schematic diagram of an ideal metal-gate MOS device and the corresponding band diagram. From Figure 3.1 (b), as explained in the appendix, Kirchkoffs voltage law around the loop gives: m + V0 = *s VIX + (Ec EjJ/q + VF + VG (3.1) where is the electron affinity of the substrate, Ec and E| are the conduction-band edge and intrinsic energies, respectively, in the substrate, and VF is the Fermi voltage, which is equivalent to (EÂ¡ Fp)/q for p-type material, where FP is the quasi-Fermi-level for holes and q is the electron charge. Collecting these terms in cleaner form gives VG = V0 + VIX + MS (3.2) 43 (A) VG + 1 Gate V//////////////A Substrate (B) Fig. 3.i MOS capacitor schematic and corresponding energy-band diagram. (A) Schematic diagram of a MOS capacitor and (B) corresponding energy- band diagram depicting the potential drops. Shown is a positive voltage VG applied at the gate, resulting in the Si02/Si surface entering inversion. 44 where 0MS = 0M s = M Us + (Ec Ex)/q + VF) is the work- function difference between the metal and the semiconductor. As will be shown in the next section, the work function difference for a polysilicon-gate MOS device is much simpler than metal-gate MOS since the substrate and gate materials are the same. The drop across the oxide can be found from Gausss Law requirements as Vg = CgEjj^/Cg (Qot Qit) /Cg. (3-3) where es is dielectric constant of the semiconductor (~11.7x8.85xlO-l4F/cm2 for Si), Elx is the field across the oxide, C0 is the oxide capacitance, and Q0T and QIT represent fixed and interface trapped oxide charge respectively. With this relation, (3.2) can be rewritten as " VFB + VIX + Â£SEIX/C0, (3.4) where Vpg, the flat-band voltage, is given by VFB = MS (Qot + Qit) /C0. (3*5) For metal-gates, there is no capacitive contribution from the metal, so the gate capacitance is simply the series equivalent of the fixed oxide capacitance, C0, and the variable substrate capacitance, Cjx. Cg = CixC0/(Cix + Cg). (3.6) The field going into the substrate, EIX, and the substrate capacitance, Cix, are given by 45 2kT nv[53/2(-u: 7) ^3 / 2 ( Uy+Up) ] a + nc[^3/.2( uIX+ucuF) 7^/2 ( ucuF) ] pxx< UIX + ln 1 + gAexp (Up UA UIX) ^1 + gAexp (Up UA) J Nxx<-Uix + ln 1 + gDexp (UD Up + UIX) ll + gDexp(UD Up) Civ = ~~N\j7\/2 (~UIX Uv+Up) + NcjFVj (Ulx+uc Up) 1 + gAexp (UF UA UIX) (3.8) L11 + gDexp(UD Up + UIX) where all U values are potentials normalized to kT/q and referenced to the intrinsic Fermi level. For example, Up is the normalized Fermi level, qVF/(kT). Pxx is acceptor substrate doping concentration and Nxx is the donor substrate doping concentration, and gA and gD are the corresponding degeneracy factors for the trap levels UA (acceptor energy level) and UD (donor energy level, not to be confused with the normalized drain voltage of an MOS transistor). Nv is the valance band density of states and Nc is the conduction band density of states. Those familiar with MOS capacitance equations might find these far more complex than they recall; a perusal of the appendix should clear up any questions about 46 this form. However, it is instructive to show how this reduces to a more familiar Boltzmann form. First, all of the Fermi-Dirac integrals [F,/2(t|) and Fj/jCn) terms] reduce to exponentials in the Boltzmann range of applied gate voltages (T| < -4). Second, there is typically only one dominant dopant, so one of the last two terms in (3.7) and (3.8) can be neglected (the first can be neglected for n-type substrate, and the second for p-type substrate). Furthermore, if deionization is neglected (UF UA UIX < -3 for p-type or UD Up + UIX < -3) for n-type), then the last two terms of (3.7) reduce to PXXUIX NXXU[X. Likewise, the two lines of (3.8) reduce to Pxx Nxx when deionization is neglected. As an example of the simplified form, let us consider a p-type substrate in strong accumulation. In this case, it can be assumed that only the accumulated surface carrier term is dominant (UIX is large and negative). Noting also that, in the Boltzmann case, Up Uv = ln(Pxx/Nv), (3.7) and (3.8) would reduce to EIX = i| (2kTPxx/Â£s) exp (UIX/2) Clx = qi| tpxxÂ£s/ (2kT) ] exp (UIX/2) These are the more tractable strong-accumulation forms found in undergraduate textbooks [3, 43] and which form the basis for one well-known oxide-thickness extrapolation algorithm [47], Polvsilicon-Gate CV Implicit in the derivation of the metal-gate CV theory above was that the capacitance of the gate is infinite and that the voltage drop across the gate is zero. With 47 metal gates, this is a reasonable assumption for the ideal isolated device. However, with polysilicon gates, there is a finite polysilicon gate capacitance as well as a voltage drop [3,49]. Indeed, the capacitor is now a semiconductor-oxide-semiconductor device, so it will have a corresponding surface potential for the gate, as well as an associated gate capacitance with a form exactly like the substrate capacitance. This requires only minor additional derivation to arrive at the poly-gate MOS capacitor (MOSC) ideal device characteristics. Figure 3.2 shows the band diagram for an n+-polysilicon gate MOS capacitor with a p-type substrate (a schematic of the device would be identical to 3.1 (a), with a metal gate replaced by a polysilicon gate). From this figure it is clear that the potential drop across the device can be given similarly to (3.1) as ^F-poiy + Assuming the energy gap has not narrowed due to the higher gate doping, the (Ec - terms are identical and cancel because the materials are both silicon. The electron affinity is the same for both the gate and substrate for the same reason. This reduces (3.9) to VG = v0 + VIX + VIG + VF + vF.poly. (3.10) Thus, for the poly-gate case, <|>MS (more aptly called <|>os, where G represents the gate, but still traditionally referred to as M for metal) is simply given by *ms = VF + VF_poly. (3.11) For Figure 3.2, <|>MS is given by ln(PxxNGG/nf), where Pxx is the substrate doping (P implying p-type) and NGG is the gate doping (N implying n-type). This simple formula assumes a Boltzmann carrier distribution in the substrate and gate, which is invalid in the 48 Fig. 3.2 Band diagram of n+ polysilicon-gate MOS capacitor with all the potential drops labeled. The band diagram shown depicts a positive voltage VG applied at the gate, with the Si02/Si surface entering inversion and the poly-Si/Si02 surface depleting. 49 gate due to the high doping and likely invalid substrate for modern ULSI devices. A more appropriate formula using inverse Fermi-Dirac integrals can be used using the examples in the appendix. The extra potential drop from the poly gate is easily taken into account via Kirkoffs law with VIG: vG = vFB + VIX + VIG + esEIX/C0, (3.12) where VFB from (3.5) still holds assuming negligible contribution from the polysilicon/oxide interface, using (3.11) for (|>MS. Finally, the gate capacitance formula needs to be extended for three capacitors in parallel. This changes (3.6) to Cg = CixC0Cig/(CixC0 + CixCig + CigC0), (3.13) where Cjg, the capacitance from the polysilicon gate, is given by -NvÂ£/;!(-UIG-Uv+UF) + NcK/2 (Uig+Uc-Uf) (3.14) This is simply (3.8) re-written with the band notation for the gate. Thus, UIG is the normalized surface potential in the gate, EIC is the field in the gate (defined below), and PGG> ngg- Uv. Uc, Nc, Nv, gA, gD, UD, UA are precisely as defined before, except that they apply now to the gate rather than the substrate. UF above was called UF poly PGG 1 + gAexp(UF UA UIG) ^GG 1 + gGxp (UB UF + U-G) 50 elsewhereit is left as UF in (3.14) to maintain the symmetry of the equation. The gate field is given by 2kT Nv[73/2<-Uig-Uv+Uf) 53/2(-Uv+UF) ] + Nct^S/2^ UIG+UC UF) 1^/2 i Uc UF) ] ^GG ( ^IX + ^-n 1 + gAexp(UF UA UIG) L 1 + gAexp(UF UA) Nqg (Ujx + In 1 + gDexp(UD UF + UIG) 1 + gDexp(UD UF) which is identical to (3.7) with the surface potentials changed. Again, the same caveat applies to (3.15)all the terms refer to the gate now, not the substrate. Things like trap levels and band edges are nearly, if not exactly, the same in the substrate and polysilicon gate. However, UF is clearly quite different (assuming the gate and substrate are not doped identically, which would make a poor capacitor or transistor). An additional equation, which was not needed in the metal-gate case, is required to relate the gate and substrate. This equation equates the charge density at the gate/oxide interface with the charge density at the oxide/substrate interface: Â£s^ix + Qitx + Â£s^ig + Qitg = 0 Qitx is the interface charge at the substrate/insulator interface and QITG is the charge at the gate/insulator interface. It is assumed that these values are negligible, and that the dielectric constant for the silicon substrate and the silicon gate are identical (already implicitly assumed in the equation). This gives the following 51 EIX -EIG' which allows the surface potential in the gate to be related to the surface potential in the substrate. The iterative solution of the above equation requires many calculations of (3.7) and (3.15), and is the most time-consuming part of the LFCV solver as well as any software using the routine (such as a parameter extractor which works by comparing the data to the theoretical curve, as discussed later in the application section). Polvsilicon-Gate Effects The effect of polysilicon gates, compared to metal gates, is a reduction of the gate capacitance, Cg, when the gate is in depletion. This is arises when the value of CÂ¡g falls below that of C0 and Cix, which only occurs during gate depletion and substrate inversion or accumulation, and only then to a significant degree for thin oxides. This is easily visualized from the three series capacitancesthe one which dominates is the smallest, and the capacitance due to the substrate and gate are both minimized during depletion (and maximized during accumulation, as well as inversion for the LF case). As oxides thin, the oxide capacitance increases, which causes the effect of the substrate and gate depletion to have more control over the characteristics of the Cg-VG curve. Figure 3.3 shows the difference between metal-gate and polysilicon-gate data, normalized to C0, for two different technologies. The higher pair of curves for a 1000 A oxide (thick oxide means low oxide capacitance) shows little difference between polysilicon gates and metal gates. The lower pair of curves for a 50 oxide (thin oxide means large oxide capacitance) shows a large decrease in Cg for all values of VG, particularly for VG > 1V, where the gate is still in depletion and the substrate is inverted. 52 VQ / (1 V) Fig. 3.3 Comparison of metal-gate and n+ poly-gate MOSC curves for two different technologies. One set has 1000 oxide with Pxx=3xl016 cm-3 and the second set has 50 oxide with Pxx=2xl017 cm-3. In each case, the VFB is adjusted to be -1.0V and the gate doping is 3xl017 cm-3. Clearly shown is the dramatic difference between poly-gate (dotted line) and metal-gate (solid line) for the 50 case, and the negligible impact on the 1000 A casethe polysilicon gate effects increase as the oxide scales thinner. 53 This continual decrease in Cg for increasing VG (in this n+ poly-gate on p-Si substrate) is often referred to as poly depletion, since the polysilicon gate is still depleting. Eventually the gate itself will invert, and the characteristics will be much improved. However, resulting field caused by the gate voltage required to invert the gate is typically beyond the reliability limit of 4MV/cm in properly scaled devices. In fact, the only way to make the gate invert sooner is to lower the gate doping, which exaggerates the poly depletion effect even more until the gate inverts. It might seem, as it did to this author, that the ultimate solution would be to use undoped gates, as they would invert much sooner and behave just like metal gates at reasonably low applied gate voltages. This works well in simulation, but the question then becomes: where is the supply of minority carriers to invert the gate? In particular, for an n+gate in a rapidly switching MOST, what would supply the holes? It has been shown that, for at least one technology, the holes are likely supplied via thermal generation (rather than ion impact) [50]. Thermal generation, then, could not supply the holes fast enough for practical use of an undoped gate. However, it might be possible to design in a minority carrier source nearby to supply minority carriers (similar to how the source and drain supply minority carriers in the substrate). The reduction in the gate capacitance due to poly depletion causes a reduction in the drive current, which degrades circuit performance [51-54], since the amount of current supplied by the transistor directly relates to the switching speed of the device. In a complementary-MOS (CMOS) circuit, the current charged up the interconnect and intrinsic capacitances of the next transistors in the line, as discussed in detail in Chapter 54 4. Because of this poly-gate ID reduction, there may eventually be a move back to metal gate (or silicides) once the processing issues of gate alignment are solved. It is instructive to look at the individual capacitance components to see how the complex poly LFCV curve forms. Figure 3.4 shows such a curve for a theoretical 50.0 oxide with an n+ gate doped (rather lowly) to 9xl018 cm-3 and a substrate doped to 5xl017 cm-3. The gate area is lxlO-4 cm2 and the flat band voltage is 1.0 V. The Cg curve, being the serial sum of C0, Cix, and Cig (Eq. 3.17), is always lower than the component curves. It can be clearly seen how each of these three components influences the overall structure of the resulting gate capacitance. In fact, this regional effect will be used to help speed up parameter extraction in the next section. Also of interest is a breakdown of potentials across the MOSC device as a function of V0. Figure 3.5 shows the four components of VG, namely VIX, VIG, Vox, and Vpg (see Eq. 3.12) as a function of VG using the same parameters as the example in the last paragraph. To show show these are related to the resulting gate capacitance, the Cg-VG curve is also plotted. What is most relevant in this figure is that as the primary dip in the CV curve occurs as the surface potential in the substrate, Vlx, sweeps from accumulation to inversion (i.e. moves from a small negative number to about one volt), and ends sharply as the surface potential approaches its maximum (strong inversion). Similarly, the secondary polydepletion dip occurs as the gate surface potential, V1G, moves from accumulation to inversion (again, moves from a small negative voltage to around a volt). Note that the final surface potential in the gate is higher than that in the substrate (VIG > VIX when VG > 4V). This agrees with the common approximation that 55 VQ/(1 V) Fig. 3.4 Individual capacitance values for a theoretical 100x100 |lm nMOSC with a 50 gate oxide, Pxx=5.0xl017 cm-3, NGG=9xl018 cm-3, T=300 K, and VFB=-1.0 V. This figure demonstrates how the three parallel capacitances (Cix, Cjg, and C0) add to give the overall gate capacitance. See Fig. 3.5 for the corresponding potential breakdown. 56 VG / (1 V) Fig. 3.5 Individual potential breakdown for a theoretical 100x100 pm nMOSC with a 50 gate oxide, Pxx=5.0xl017 cm'3, NOG=9xl018 cm"3, T=300 K, and VFB=-1.0 V, along with the corresponding LFCV curve. Note how the surface potential in the substrate, V!X, increases rapidly in the range VG = -1 to 0 V as Cg increases (substrate inversion) and the similar increase in VIG in the range VG=1 to 3 V (gate inversion). See Fig. 3.4 for the corresponding capacitance breakdown. (dd 0/ 60 57 the surface potential pins to a little over 2VF, since the Fermi voltage in the gate will be larger than that of the substrate due to the greater gate doping. Parameter Extraction Using the LFCV Model Of the multitude of variables in the LFCV equations, most of them are known to a reasonable degree of accuracy (such as the dielectric constant, energy gap, conduction- band density, etc.), can be measured easily (temperature), or need not be known very accurately (acceptor and donor trap level) due to their small effect. This leaves the gate and substrate doping, the oxide thickness, and the flatband voltage as the unknown parameters. These parameters may be extracted from experimental data by comparing experimental data to the theoretical model presented in this chapter. This may appear to be an easy task, since the equation need only be used, along with some data, in conjunction with a non-linear least-squares-solver. However, one will note that the polysilicon gate LFCV formula is doubly parametric (that is, is related through two parametersthe surface potentials UIX and UIG), neither of which are known from the data. Thus, solving this problem is non trivial. The first step toward a solution, then, is to write a program which will calculate Cg given VG. This requires intensive calculations to find U|X and U|G for each VG, but can be done since there is only one unique solution. Thus, with a Cg(VG) routine written, a nonlinear least-squares-fit program can be used. The code written for this dissertation took advantage of the fact that, as the solution converges to values of the unknown parameters, the values of the surface potentials at each experimental data point could be 58 used for initial guesses for each subsequent iteration of VG to find each Cg (since the parameters {substrate and gate doping, oxide thickness, and flatband voltage)) should not be changing too rapidly). This greatly increased the convergence rate over estimating UIX and U|G on each call, at the expense of additional code complexity and memory usage. 3-Point Extraction Methodology If the model were perfect, then it would require only three points to match the experimental data to the model. Why only three data points for four parameters? Because the additional constraint that one of the points should be the minimum of the experimental LFCV curve can be used. From this information, the flatband can be found by comparing the VG of the theoretical minimum with the VG of the experimental minimum. The other three parameters can be found directly from the Cg values of the three points. Figure 3.6 shows the three points, labeled Cg_acc, Cg_depl, and Cg_dep2, as they relate to the whole LFCV curve. Only Cg_dep| is uniquethe other two points can be anywhere within their region. The Cg.acc point is a point from the LFCV gate accumulation region. From this, a good estimate of the oxide thickness can be found, since the other parameters have very little influence over this point (see Figs. 3.7 and 3.8). Cg.acc asymptotically approaches C0, which is inversely proportional to the oxide thickness, Tox, via the parallel plate formula. There has been much research in obtaining Tox and/or C0 from (substrate) accumulation CV data [48,55-58]. 59 Fig. 3.6 Example of a general polysilicon-gate (n-i- gate, p substrate for this case) showing how all the important regions can be labeled in terms of the gate state rather than the typical substrate state. This regional breakdown is used to improve the speed and accuracy of the parameter extraction routine. 60 The Cg_dep| point is the minimum of the LFCV curve, and allows us to find the substrate doping, since the substrate depletion region is strongly dependent on the substrate doping concentration. In fact, depletion CV data can also be used to determine the substrate doping profile [59-61], Figure 3.7 shows LFCV data for several different constant substrate doping concentrations, clearly demonstrating the strong dependence of substrate doping on the location of Cg_depi. This was also demonstrated in Fig. 3.4, since the main influence in this depletion region is CIX, which itself is strongly dependent on UF (see Eq. 3.8), which is directly related to the the inverse Fermi-Dirac integral (natural logarithm if assuming a Boltzmann distribution) of the substrate doping. The position of the minimum along the VG axis also allows us to estimate the flatband voltage by comparing the VG of the minimum of the theoretical curve to the VG of the data. The Cg.dep2 point is from the gate depletion region. Figure 3.8 shows that the gate doping has the most affect on this part of the curve, whereas Figure 3.7 shows that the substrate doping has very little effect in this region. For the n+ gate on p-substrate example in Figure 3.8, the substrate is in inversion. However, even if the substrate were n-type (and the substrate thus accumulated), Cg depletion would still occur because the gate would still be in depletion (of course, the entire curve would be shifted due to the flatband difference). Hence, this point is called Cg_dep2, with the dep in reference to the depleted state of the gate. By varying the parameters in the appropriate regions to match these three points, a unique parameter set will be obtained which will describe a theoretical LFCV curve passing through the three points. 61 Fig. 3.7 Effect of substrate doping changes (Pxx) on LFCV characteristics. The depletion-1 region (see Fig. 3.6) is the region of largest impact. 62 Fig. 3.8 Effect of gate doping changes (NG0) on LFCV characteristics. The depletion-2 region (see Fig. 3.6) is the region of largest impact. 63 3-Region Extraction Methodology As good as our model is, there are still several effects which are not being considered. These include retrograde doping in the substrate and quantum effects in the substrate inversion channel. Retrograde doping is commonly used for sub-'/2-micron design to maintain a high sub-surface doping concentration to prevent punchthrough, while still maintaining a low VT for low-V0 operation (to accommodate the thin oxides) [7]. Figure 3.9 shows an example of a retrograde profile from our internally-modified MINIMOS. The LFCV model assumes a constant doping profile in both the substrate and gate, and so deviation from this assumption will cause changes in the experimental LFCV curve relative to the theoretical model. Charge-carrier layer push-out due to quantum effects in the inversion and accumulation layers has been an area of much research [62-65]. Experimental verification of these quantum effects are invariably at low temperatures, where phonons will not broaden the quantum bands into a continuum. Although some amount of quantum effect is likely present, it is probably impossible to model correctly when one considers thermal broadening, Si02/Si interface roughness and transitional regions, non- random dopant distribution, and other non-idealities. These will all tend to broaden the electron levels into a more classical continuum. It has been noted that electrical and optical oxide thicknesses do not often agree, and the difference has been attributed to quantum effects. As will be discussed later, the effect is likely overestimated. More important, if there is a difference, it is the electrically effective oxide thickness (as determined from electrical experiments, such as CV) 64 Fig. 3.9 Sample of retrograde doping profile, showing low surface concentration (5xl016 cm-3) and higher bulk concentration (lxlO18 cm"3). \525aVSSÂ£k '5RKrYiKa,\ 65 which is most important compared to the optical thickness (which is not what affects device performance). Due to these two main non-idealities (non-constant doping and quantum effects), there could be some dependence on the extracted parameters using only three points. That is, extracted parameters might be dependent on which points we choose for Cg_acc and Cg_dep2. To overcome this, the entire curve could be fit to the model. This would result in extremely long convergence times, as a partial derivative must be calculated for each variable at every point for every iteration. However, Figures 3.7 and 3.8 show that some parameters have no influence on the LFCV curve in certain gate-voltage regions. Thus, the information provided from their partial derivatives does not help convergence, and will actually slow down the convergence, not to mention waste time during the calculation. Instead of fitting all the data to the model, the data can be broken up into the same three regions suggested in Fig. 3.6 for the three-point fit. Then the model can be fit using only the parameter (or parameters) dominant in the specific region, thus greatly improving the convergence rate (since the data being used is most relevant). This adds complication to the coding, as the data partitioning into each region (discussed later) must be automated, and a different fitting routines must be created for each of the three regions (same model, but separate partial derivative calculations). Methodology Comparison Figure 3.10 shows a comparison of the fit using the 3-point and 3-region methods to experimental data for a 130 oxide from an industrial 100x100 |xm MOST transistor. 66 VG / (1 V) Fig. 3.10 Theoretically generated curves compared to original LFCV data using the (A) three-point and (B) full-curve extractions to on an n+polysilicon gate, 100x100 ftm nMOST. Extracted parameters were: (A) Xox=130, Pxx=9.0xl016 cm-3, Ngg=5.8x1019 cm'3, and VFB=-1.06V. (B) Xox=130, Pxx=8.7xl016 cm'3, NGG=3.0xl019 cm-3, and V^-1.0 IV. 67 The solid-line, of course, is the theoretical curve using the extracted parameters, and the x marks are the data used for the extraction. The top (A) curve uses only the three points marked to fit the data while the lower (B) curve uses the full set of data. The RMS error (for the second curve), calculated from the square root of the sum of the squares of the difference between theoretical and experimental capacitance, divided by the square root of n 5 (5 = degrees of freedom = 1 + number of fitting parameters), was 2.7%, an excellent fit for only four parameters. There is not much in literature to compare the goodness of these results, as there are not many capacitance models available (most compact models, such as BSIM3 [34], fit IV characteristics only, and do not consider capacitance extraction). Aggressively scaled MOSC device data is given in Figure 3.11. This shows two LFCV curves from a -20 oxide from an industrial MOST transistor, where the 20 was determined from some optical method, most likely ellipsometry. This is a rather complex figure, and requires some explanation. There are two experimental curves: one p+ poly gate and one n+ poly-gate, both on a p-well. Thus, in the V0 > 1V region, the n+ gate is in depletion, while the p+gate is in accumulation (clearly the curves were shifted to align them, as the flatband should differ by about a volt between the two curves, although a threshold adjustment implant would offset this somewhat). From looking at the data at 1.3V and assuming Cg = C0, they conclude that the effective oxide thickness is 33 for the depleted n+ gate, and 28.5A for the accumulated p+ gate device. This is a poor approximation, since the oxide thickness value would vary greatly at different points along the depleted curve. However, they correctly state that the difference between the 68 two is due to poly depletion, which is a reasonable statement when applied to that specific gate voltage only. Thus, they attribute a 4.5 reduction in effective oxide thickness due to polysilicon depletion at 1.3V, which is the operating voltage for that technology. They next assert that the difference between the accumulated-gate curve (28.5 based on assuming Cg = C0) and the optically measured oxide (20) must be entirely due to quantum effects. This conclusion is wrong, as the quantum model most likely does not take all the effects mentioned previously into account (such as thermal broadening, interface roughness and transitional region, not to mention retrograde doping). More importantly, they are completely ignoring the severe error of using Cg = C0! Figure 3.12 shows the fit (using the three-point method) to the data in Figure 3.11. From this fit, the extracted oxide thickness is 24.4. Compared to the industry-stated results, the same effective oxide thickness reduction due to poly (4.7 here versus 4.5) is seen. However, by correctly accounting for the distribution (instead of assuming Cg = C0), an additional 4.1 reduction from the Fermi-Dirac distribution is also found (that is, from the fact that Cg < C0)! This leaves 4.4 of difference between the extracted oxide thickness of 24.4 and the optically measured thickness of 20. This 4.4 difference may include some quantum effects, but it may also be due to optical errors, such as not accounting for the transitional layer properly [66-67] and/or some other effects (i.e. doping profile). Ellipsometry and other optical methods can not be used on the actual device (since the gate electrode is not transparent), so it does not measure the oxide thickness in the active part of the device, which may differ slightly due to the additional 69 C0 calculated from optical Xo=20.0A Fig. 3.11 P+poly/p-well and n+poly/p-well 20 industrial data. Data is shifted to align minimums, and labelling refers to industrial interpretation of the two curves. Please see text for explanation of this breakdown. Compare to Fig. 3.12, which is the authors interpretation of the same data after quantumless parameter extraction. 70 VG / (1 V) Fig. 3.12 Theoretically generated LFCV data three-point parameter extraction using data in Fig. 3.11. Extracted parameters are shown. Instead of attributing the difference between the optical thickness of 20.0 and the extrapolated thickness of 28.5 at VG=1.3V to quantum effects, we find that 4.1 A is due to the distribution function used (Fermi-Dirac), with the remaining 4.4 possibly due to error (in optical measurement and/or other factors). 71 processing. As far as parameter extraction is concerned, the most important factor should be agreement with electrical results, not optical. Although a three-point method was used here (to improve the match at the 1.3V point), a full fit to the data in Fig. 3.12 has about a 10% RMS error, which, considering the thin oxide and the fact that the substrate gate doping is not constant, is extremely good. An interesting side note, brought up during the proposal for this project, concerned fitting a Boltzmann model to the data instead of a Fermi-Dirac. The surprising result of this was a better fit (6.6% RMS error), but, as would be expected, a thicker extracted oxide thickness of 27.5. This is an interesting result, as it shows that using the wrong model can appear to give better results (in terms of fit), even though the resulting parameters actually have greater error (due to the incorrect carrier distribution). Philosophically, the issue of oxide thickness is an interesting topic. Many people would argue that TEM is the only way to measure the true thickness. However, ignoring that this is a destructive and time-consuming technique, it only yields the thickness of that particular cross section. What is desired is the average oxide thickness, as it is the average oxide thickness which affects the amount of charge accumulated by an applied voltage in a MOSC. This is why Fowler Nordheim (FN) tunnelling is also not a particularly good methodit will always underestimate the oxide thickness since the tunnelling will occur in the thinner spots on the gate. Additionally, one would rather not stress the devices while trying to find the oxide thickness. One recent technique for ultra- thin oxide thickness determination is using quantum oscillations in the tunnelling gate current, which are caused by quantum interference of electrons in the oxide conduction 72 band [68]. This method potentially suffers from the same problems as FN, and additionally requires knowledge of the effective mass and oxide barrier height, the latter two of which add about 2.5 of error to the results [69], assuming they are known to within 5%. Convergence Speed-up Details Iteration stops when none of the fitting parameters (i.e. the extracted parameters) changes by more than 5x 1 (T4% between successive serial cycles (i.e. each parameter was fit, and none changed by more than 5xlCT4%). Below are several of the methods employed to speed up this convergence. The calculations involved in the extraction are extremely complex. Of greatest importance is the convergence speed of the poly LFCV model, which itself must converge on the two surface potentials just to give one Cg data point for a given VG. This one data point is used in the numerical partial derivatives for the non-linear least-squares- fit, which means the Cg(VG) is called twice for each experimental data point for each iteration! Because this model is called so frequently, it is important to keep track of all the converged surface potentials for each experimental data point so that the LFCV model has a good estimate for subsequent calls. The delta used for the numerical partial derivatives, as it turns out, has a major influence on the correctness of the fit. Since two of the parameters vary logarithmically (the substrate and gate doping), the actual parameters used during the fit are the logs of these parameters. Thus, the deltas used for the derivatives must be calculated differently. Another problem is that the delta used for the oxide thickness is dependant upon the 73 actual thickness of the oxide (that is, it should be different for a 50 oxide compared to a 1000 ). The empirical results for the best deltas, as determined from analysis of the numerical derivatives, were A=I07 for ln(Nxx), A=0.1 for ln(N0G), A= 10-4 for VFB- and 10~7xTox for Tox. For example, the partial derivative of Cg with respect to Tox is calculated from (Cg(VG)! Cg(VG)2)/2A, where Cg(VG), is the gate capacitance calculated at some VG with an oxide thickness of Tox + A and Cg(VG)2 is the gate capacitance calculated at the same VG with an oxide thickness of Tox A. The reason arbitrarily small values cannot be used, of course, is because the LFCV model itself is only accurate to about eight digits (less near flatband) due to its own internal convergence criteria [46]. Finally, to start the extraction, a reasonable initial guess must be made. The initial guess of the oxide thickness is simply Aeox/Cgmax, where Cgmax is the maximum gate capacitance in the dataset and A is the gate area. This is the standard first-order approximation based on Cg = C0. For the substrate doping initial guess, the asymptotic high-frequency CV formula for Cg [43] is solved iteratively using the minimum and maximum Cg values from the data set. The gate doping is simply set to 3xl019 cm-3. With these three parameters approximately known, the flatband voltage is estimated from ^Grain_data ^Gmiiuheory- Since the minimum of the CV data is not necessarily given, but is needed internally to estimate the flatband, the minimum three data points (in terms of Cg) are used to estimate the true Cg minimum based on the parabolic minimum formula [70]. This slightly improves the convergence, but not as much as would be expected, largely because the minimum of the CV curve is not very parabolic. 74 Because it was clear that convergence was slower as the results approached the final values, a trick was developed to improve this end case. Whenever a trend was visible during a fit, the routine doubled the amount of the parameter increase. A trend, in this case, is defined as three successive moves of a parameter in the same direction for all the parameters (possibly different directions for different parameters). This cut down the number of iterations by about 20% in most cases. One thing which would have improved the speed of convergence greatly would be to use a simpler model for the Fermi-Dirac integral. The Cody-Thatcher model [71] is extremely accurate, but requires the quotient of ten exponentials from a Chebyshev approximation. This approximation was used instead of some other simpler (though less accurate) approximations [72-76] because it was desired to add as little error as possible from the Fermi-Dirac integral calculation. CHAPTER4 THE EFFECT OF INTRINSIC CAPACITANCE DEGRADATION ON CIRCUIT PERFORMANCE Introduction In this chapter, the relatively obscure subject of intrinsic capacitances will be discussed. The area of MOS intrinsic capacitance has received little attention over the years due to the difficulty of measurement and small impact relative to extrinsic capacitances such as interconnect and packaging. However, as the push toward higher density continues, the extrinsic capacitance is being reduced as much as possible to improve performance. This will eventually leave the intrinsic capacitance as the primary load in CMOS circuits, thus making this a topic worth studying now. After a discussion of the intrinsic capacitances which most effect CMOS circuits (Cgd and Cgs), direct experimental measurements of the effect of hot-carrier degradation on intrinsic capacitance will be discussed, and the results modeled. The impact of this degradation on circuit performance will be evaluated and shown to offset some of the losses due to ID degradation. Background The effect of hot-carrier degradation on the drain current, ID, has been studied intensely since Abbass initial observation in 1975 [77]. Another intrinsic property of a 75 76 MOS transistor, the intrinsic capacitance, has a much shorter history of study with regard to hot-carrier degradation. The first systematic study of intrinsic capacitances was done by Sah [13] in 1964, which was used by Meyer in 1971 [78] in his widely referenced work. In his paper he defines the intrinsic capacitance between terminals as: dQx That is, the change in the charge at terminal x due to a change in voltage at terminal y. This definition applies to any two-or-more terminal device, but from now on will be used with respect to a 4-terminal MOS transistor. Thus, it is clear that there are 16 possible intrinsic capacitance terms for a 4-terminal MOS transistor. Please note that in this small-signal defination, all of the non-y terminals are virtual ground. Thus, dVy is referenced to ground (i.e. it is essentially relative to all the other terminals). At first thought, one might assert that there are only 8 possible capacitances since Cxy=Cyx. However, this is not true because our definition of intrinsic capacitances does not represent static capacitive values and are not reciprocal. Consider the two intrinsic capacitances Cgd and Cdg. Neglecting overlap capacitance, when the applied gate voltage is less than the gate threshold voltage, VGT, both of these capacitances should be zero (since both Cgd=dQg/dVd and Cdg=dQd/dVg are zero due to no existing channel). Once VGS > VGT (and VDS < VDSsat), Cgd and Cdg will both have some finite positive value when the channel forms. The interesting case is when VGS > VGT and VDS > VDSsat. Now there is a channel, but it is pinched off near the drain end. Cdg (dQd/dVg) is non-zero since a change in the gate voltage still affects the charge associated 77 with the drain (Qd); Cgd (dQg/dVd) is zero since a change in the drain voltage has no affect on the gate charge since the drain is not connected to the channel due to the pinch- off. As clear as this seems now, both Meyer [78] and others [79] assumed that the capacitances should be reciprocal. These should not be confused with the small-signal circuit element terms, which are named the same way but actually are reciprocal by definition. Ward and Dutton [80] were the first to argue that the intrinsic capacitances were, in fact, non-reciprocal. The paper also stressed the importance of including all the capacitances, particularly the gate to bulk (Cgb) capacitance, which had been omitted by Sah, and hence Meyer. Ward and Duttons charge-based model was a huge improvement at the time, as Meyers model does not guarantee charge-conservation in circuit simulators (due to omitting Cgb), resulting in erroneous results for the simplest of circuits. Papers predating Meyers work largely used discrete devices, and so authors logically argued that modeling the intrinsic capacitances would be useless since the capacitance from packaging and external circuitry would be vastly larger [81], Furthermore, there was no direct method to measure the data to verify the models. With the advent of integrated circuits, the primary capacitive load between CMOS circuit cells (i.e. an nMOS and pMOS inverter pair) became dominated by the intrinsic and interconnect capacitances, rather than the packaging and external circuitry. Thus, modeling the intrinsic capacitance (as well as interconnect) became important. Integrated circuits also hailed the need for compact models to simulate large numbers of transistors. One of the first compact models was CSIM [82] from AT&T Bell 78 labs. Surprisingly, the authors of this model stayed with the simple Meyer model, although argued that including the intrinsic capacitances was critical, particularly Cgd. Cgd accounts for most of the intrinsic capacitance load in CMOS circuits due to the Miller feedback effect [82]. Berkeleys BSIM [33] built upon CS1M, also retained the Meyer model. BSIM2, however, corrected this deficiency by including a non-reciprocal intrinsic capacitance model. The BSIM3 [34] model moved from an strongly empirical d.c. model to a more physically-based model, but retained the unaltered a.c. model (including intrinsic capacitances) from BSIM2, suggesting a lag in a.c. model development. Current a.c. models are extremely poor. A great deal of additional research is needed before a.c. models become nearly as sophisticated as d.c. MOS current models. There are two reasons the a.c. models are so far behind the d.c. models. First, intrinsic capacitance data have only been available since the early 1980s, over twenty years after the first MOS transistor ID data. Second, until recently, external capacitances and interconnect capacitances dominated the total capacitive load, making the intrinsic capacitance fairly unimportant. However, as the transistor dimensions have decreased and substantial improvements in drain current density become difficult due to physical limitations, major efforts have been implemented to reduce the interconnect capacitances, such as low-k dielectrics. This has increased the impact of intrinsic capacitances in overall circuit performance and, with improved interconnect, could become the predominant capacitive load in the circuit. It is interesting to note that publications on intrinsic capacitance modeling have been increasing year-to-year since Ward and Duttons work [83-90]. 79 Measurement of Intrinsic Capacitances Because direct measurement of the intrinsic capacitance is difficult, many of the first measurements were done with on-chip circuitry using reference capacitors [92- 93] or op-amps circuits configured as coulombers [94]. Eventually, external circuitry was used, including a lock-in amplifier connected to an HP 4145 (as a voltage source) [95] and later an off-the-rack LCR meter [96], such as the HP 4275 A. In this section, the measurement of a few of the intrinsic capacitances will be described. These can be done using an HP 4275 or HP 4276 (same equipment with different a.c. frequency ranges), or the newer HP 4284. The first discussion of using an LCR meter for the direct measurement of intrinsic capacitances was written by K. C.-K. Weng and P. Yang in 1985 [96]. In this letter, many of the important problems with measuring the intrinsic capacitances were discussed. The main problem is that LCR meters are not designed to measure intrinsic capacitances. There are two sets of terminals on the LCR meter: High and Low. The high port applies the d.c. bias as well as the superimposed a.c. test signal. The low port measures the resulting small-signal current. From the magnitude and phase difference of the current relative to the applied small-signal test voltage, the capacitance can be found. Unfortunately, the low port is a virtual a.c. and d.c. ground, so no d.c. bias may be applied to it. To measure Cgd (dQg/dVd), the high port is attached to the drain (to apply the dVd) while the low port is attached to the gate (to measure the dQg via the small signal-current, ig times dt). If Cgd is desired as a function of VGS, the problem becomes apparent: How can VGS be ramped if the gate is grounded? 80 The only solution, of course, is to independently bias the three terminals not connected to the low port, as shown in Figure 4.1 for aCgd measurement. Thus, two additional power supplies are required, along with the internal d.c. power supply in the LCR meter. These power supplies must be well calibrated with one-another to ensurethat no potential difference exists between them when the same voltage is programmed. The burden of negotiating the polarities of the theee power supplies, once worked out, can be easily programmed into an automated station. As an example of the polarity problem, consider the following: if Cgd at VGS=2 V, VDS=3V, and Vxs=0 (note: the device is active, with a current flowing from the drain to the source, unlike standard CV measurements, where the source, drain, and substrate are tied together) is desired, the source and substrate can be biased at -2 V and the drain can be biased at 1 V. Since the gate is virtual ground (VGS=0), it is easy to verify that the above applied voltages give the desired potential differences (VGS, VDS, and Vxs). There is nothing particularly odd about this configuration except that it differs from the traditional C-V measurements where the substrate is the ground reference instead of the gate. In the above case of Cgd, the source and substrate may be tied together to forego one of the power supplies in Figure 4.1. If Vxs not equal to zero was required, however, all terminals must be biased independently. Thus, if one is designing a measurement station where any of the possible intrinsic capacitances can be measured, three power supplies (including the internal one of the LCR meter) are necessary. 81 LCR Meter Fig. 4.1 Measurement configuration for Cgd. Requires LCR meter with internal d.c. power supply, as well as two additional external d.c. power supplies. 82 Measurement Configurations Although the standard textbook MOS device is symmetric with respect to interchanging the source and drain, production devices may be asymmetric. This asymmetry may be the result implant shadowing, drain and/or source engineering, or hot- carrier-induced degradation, among other possibilities. Implant shadowing is an interesting case, as it may result in the gate/source and gate/drain overlap regions being different lengths, as shown in Figure 4.2. While the resulting ID characteristics are symmetric (that is, the ID versus VDS characteristics are the same if the source and drain leads are swapped), the measured Cgd characteristics (as well as Cgs, Cdg, and Cds) are asymmetric. This occurs because the measured characteristics include the constant overlap component, as shown in the following simple equation: c c + r ^-gtLmeasured ^ov_drain ^gd- The Cov drain term is composed of the constant overlap of the gate with the drain, as well as an inner and outer fringe component. These fringe components have been calculated theoretically [97], and assuming they are constant as a function of gate voltage introduces negligible error [96]. The value of the measured Cgd in subthreshold (where Cgd measured = Cov drain) has been used to estimate the length of the gate-to-drain overlap region [98], and with the drawn channel length know, these overlap values could be used to extract the effective channel length. When necessary, the normal and reverse configurations of Cgd and Cgs measurements will be specified. These are shown in Figure 4.3. qjf1" or Cgdnorm refers to the normal measurement mode, where the high port is applied to the drain for a Cgd 83 gate-^ Fig. 4.2 Simplified schematic of asymmetric gate overlap, which results in ''o v_drai ir'-'ov_source 84 Fig. 4.3 (A) C, gd_norm (B) C (C)C (D) C Measurement configurations for (A) Cgd in normal configuration mode; (B) Cgs in normal configuration mode; (C) Cgd in reverse configuration mode; and (D) Cgs in reverse configuration mode. 85 measurement. Cgdv or Cgd rev refers to the reverse measurement mode, where the high port is applied to the source for a Cgd measurement. This is necessary because, for short- channel devices, the resulting Cov value (where Cov is Covdrain or Cov sourcc) can become a significant fraction of the total effective intrinsic capacitance. Although perhaps not obvious now, Cgs = Cgd when VDS=0. However, due to the difference in Cov, Cgsmeasured may not equal Cgd measured. Figure 4.4 shows the Cgs and Cgd measurements in thenormal and reverse modes for a 20 x 20 pm device. Figure 4.5 shows the same measurements on a 20 x 0.40 pm device (effective channel length is 0.24 pm). Comparing the two figures clearly shows the negligible impact of Cov on the long-channel device Cgd and Cgs characteristics and the large impact on the short-channel device. In both cases, the Cgd and Cgs values are almost identical, as is the overlap-induced difference of about 3 fF (This 3 fF offset is not visible on the Ldrawn=20 pm device because it contributes less than 2% to the maximum capacitance, whereas the overlap contributes about 60% of the total measured maximum capacitance for the Ldrawn=0.40 pm device). Later in this chapter, the results of channel hot-carrier stress on Cgd and Cgs will be shown. Because channel hot-carrier stress is inherently asymmetric (since the damage occurs near the drain edge), it is necessary to lay down the above notation for later use. Sample Measurements For all capacitance measurements in this chapter, an HP 4828A LCR meter was used with a small-signal voltage was 400MHz at 60 mV peak-to-peak. These number were chosen after testing a wide range of a.c. signal voltages and frequencies 86 Fig. 4.4 Cgd_norm Cgd_rev CgSI10rm, and CgS_rev versus VqÂ§ for a 20x20 pm MOST with VDS=0.0. Although it appears that all four curves are the same, there are actually two sets of curves, Cgdnorm/Cgsrev and C-gd_re\/CgS_norm separated by 3 fF. Very little difference is seen because the overlap capacitances shift is much less than then the peak Cgd and Cgs values. Compare this with Fig. 4.5. 87 Fig. 4.5 ^'gd_:iorm' ^'gd_rev' ^'gs norm' ^"gs rev Versus Vq^ for a 20x0.40 401 MOST with VDS=0.0. Roughly 3 fF parallel shift of Cgs_norm/Cgdrev and Cgs rev/Cgd norm is due to a difference in constant overlap capacitance between the source and drain. 88 to obtain the most accurate results. The 60 mV signal may seem a little large to those familiar with common C-V measurements, where 25 mV is typically used, but is actually on the low end of the 23 mV to 400 mV found in most intrinsic capacitance papers [95- 96,98-106]. Frequencies below 100 MHz result in extremely poor-resolution (noisy) intrinsic capacitance data, while frequencies above 500 MHz begin to show markedreduction due to series resistance. LCR-specific settings on the HP 4284A were a medium integration time with 8-cycle averaging. So far the measurement procedures and naming conventions of intrinsic capacitance have been discussed. Figures 4.4 and 4.5 showed sample measurements with VDs=0. Although this is the typical way capacitances are measured, the ability to measure the capacitance of active devices, where VDS > 0 when VGS > VGT (where VGT is the threshold voltage at which an inversion channel form between the source and drain), is important. Why is this capability important? Because in a real circuit, this will commonly occur. If a correct model for the behavior of an operating transistor is desired, then data from an active device is required. Indeed, without this data, it would be like trying to verify an IDsat model with data only taken in subthreshold! Examples of Cgd measurements on active devices are shown in Figures 4.6 and 4.7 for 20 x 20 pm and 20 X 0.40 pm as a function of VGS for VDS = 0.0, 0.5, and 1.0 V (vsx = 0.0V). Cgd transitions from Cov drain to a larger value once VDS < VDSsat, or the channel is no longer pinched-off. From a charge perspective, this means changes in VDS (dVd) cause changes in Qchanne|, which in turn cause changes in Qg (dQg), resulting in a (dd l) / p60 89 Fig. 4.6 VGS /(1 V) Cgd versus VGS for a 20 x 20 |im MOST with VDS=0.0,0.5, and 1.0 V. Cgd / (1 fF) 90 Fig. 4.7 Cgd versus VGS for a 20 x 0.40 |im MOST with VDS=0.0, 0.5, and 1.0 V. 91 Cgd. Thus, as VDS increases, the point at which this transition occurs also increases, as can be seen in the figures. As mentioned previously, Cgd is the most important intrinsic capacitance because, in a common-source configuration (which is the configuration for all CMOS circuits), the effective load is 2(Cgs + Cgd( 1 Av)), where Av is the gain between the gate input and drain output (a large negative number). The next most important capacitance, based on the above load formula, is Cgs. Figures 4.8 and 4.9 show both Cgs and Cgd for a 20 x 20 pm and 20 x 0.40 pm as a function of VGS for VDS = 0.0, 0.5, and 1.0 V. Unlike Cgd, Cgs will have a finite value as long as VGS is greater than VGT, since the channel will always be connected to the source. At VDS=0, Cgs=Cgd since the channel charge is equally controlled by the source and drain. However, if VGS > VGT (channel forms) and VDS > VDSsat (drain pinched off), then the source terminal will actually control more than half of the channel charge, resulting in a rise in Cgs above the value at VDS=0. However, once VGS increases to a point that VDS < VDSsat, the drain is no longer pinched off, and the Cgs value begins to decline with increasing VGS as Cgd increases rapidly. This is clearly demonstrated in Figure 4.8 (and to a lesser extent in 4.9), where the decline in Cgs corresponds to the increase in Cgd. The model for Cgd and Cgs will be discussed later. Recalling the discussion about the overlap-capacitance shifting in the previous section, the capacitances shows in 4.8 and 4.9 are actually Cgdrra and Cgav in order to offset the effects of the overlap capacitance. (Fig. 4.5 shows why this was necessary) 92 VGS /(1 V) Fig. 4.8 Cgd and Cgs versus VGS for a 20 x 20 |im MOST with VDS=0.0,0.5, and 1.0 V. 93 Cgd versus VGS for a 20 x 0.40 (im MOST with VDS=0.0, 0.5, and 1.0 V. Fig. 4.9 |

Full Text |

LD
1780 199J UNIVERSITY OF FLORIDA 3 1262 08556 6049 xml version 1.0 encoding UTF-8 REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd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f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f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b 506 ILWf DQG VXE Âƒ b 506 ILWf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f KDYH VHHQ WZR RUGHUV RI PDJQLWXGH GHFUHDVH LQ GHYLFH GLPHQVLRQV IURP SLQ LQ WR SP LQ >@ 7KLV FRQWLQXDO UHGXFWLRQ IXHOHG E\ UHTXLUHPHQWV IRU KLJKHU VZLWFKLQJ VSHHGV ORZHU FRVW DQG GHFUHDVHG SRZHU KDV EHHQ VXVWDLQHG E\ LPSURYHPHQWV LQ OLWKRJUDSK\ DQG KDV UHVXOWHG LQ LQFUHDVHG DUHDO DQG FKLS GHQVLWLHV WUDQVLVWRUVFP DQG WUDQVLVWRUVFKLSf &RPSDUHG WR WKH WUDQVLVWRUVFKLS LQ WKH ILUVW H[SHULPHQWDO ELW VWDWLF UDQGRP DFFHVV PHPRU\ 65$0f LQ WKH 0 WUDQVLVWRUVFKLS 0ELW G\QDPLF UDQGRP DFFHVV PHPRU\ '5$0f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f WUDQVLVWRU EHFRPHV D VLJQLILFDQW IUDFWLRQ RI PAGE 8 WKH WRWDO FKDQQHO OHQJWK $V WKH GUDLQ YROWDJH LV FKDQJHG WKH VSDFHFKDUJHUHJLRQ WKLFNQHVV DOVR FKDQJHV UHVXOWLQJ LQ DQ HIIHFWLYH FKDQQHO OHQJWK ZKLFK LV GUDLQYROWDJH GHSHQGHQW DQ HIIHFW NQRZQ DV FKDQQHOOHQJWK PRGXODWLRQ &/0f 7KLV SUREOHP FDQ EH WROHUDWHG IRU FRPSOHPHQWDU\ 026 &026f ORJLF FLUFXLWV EXW QHHGV WR EH SURSHUO\ PRGHOHG LQ RUGHU WR SUHGLFW WKH GULYH FXUUHQW RI WKH 026 WUDQVLVWRUV LQ WKH FLUFXLW LQ RUGHU WR HVWLPDWH WKH VSHHG RI WKH UHVXOWLQJ FLUFXLW $V WKH GHQVLW\ RI WUDQVLVWRUV LQFUHDVHV VR GRHV WKH SRZHU GHQVLW\ 7KLV UHTXLUHV D UHGXFWLRQ LQ WKH RSHUDWLQJ YROWDJH VLQFH WKH DFWLYH VZLWFKLQJf RXWSXW SRZHU LV SURSRUWLRQDO WR WKH VTXDUH RI WKH RSHUDWLQJ YROWDJH 3DFWcYH f IFLRFN&R9f 7R REWDLQ WKH VDPH SHUIRUPDQFH DW ORZHU YROWDJHV WKH R[LGH WKLFNQHVV PXVW EH UHGXFHG 6LPSOH 026 WKHRU\ SUHGLFWV WKH GUDLQ FXUUHQW LV LQYHUVHO\ SURSRUWLRQDO WR WKH JDWH R[LGH WKLFNQHVV +RZHYHU IRU WKLQ R[LGHV $f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Âƒ RU XQGHU VL[ DWRPLF OD\HUV RI VLOLFRQ DQG R[\JHQ 7KH JRDO RI WKLV GLVVHUWDWLRQ LV WR LQYHVWLJDWH WKH LVVXHV GHVFULEHG LQ WKH SUHYLRXV SDUDJUDSKV &KDSWHU GLVFXVVHV WKH KLVWRU\ RI GLPHQVLRQDO GUDLQ FXUUHQW PRGHOV DQG VRPH RI WKH PHWKRGV ZKLFK KDYH EHHQ LPSOHPHQWHG WR H[WHQG WKHVH PRGHOV WR LQFOXGH WKH &/0 HIIHFW 7KH 3DR6DK PRGHO WKH PRVW DFFXUDWH ORQJFKDQQHO FXUUHQW PRGHOV ZLOO EH H[WHQGHG WR LQFOXGH WKH &/0 HIIHFW XVLQJ WKUHH GLIIHUHQW DSSURDFKHV 7KH &/0 HIIHFW DV GHPRQVWUDWHG LQ WKH QHZ PRGHOVf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Âƒf DQG WKLQ Âƒf JDWHR[LGH GHYLFHV ZLOO EH XVHG $GGLWLRQDO R[LGH WKLFNQHVV LVVXHV VXFK DV TXDQWXP HIIHFWV DUH DOVR GLVFXVVHG &KDSWHU FRQVLGHUV WKH LQWULQVLF FDSDFLWDQFHV LQ SDUWLFXODU WKRVH PRVW LPSRUWDQW LQ PRGHUQ FRPSOHPHQWDU\ 026 &026f FLUFXLWV &JG DQG &JV &RPSDUHG WR WKH GUDLQ FXUUHQW ZKLFK LV DOVR DQ LQWULQVLF SURSHUW\ RI D 026 WUDQVLVWRU LQWULQVLF FDSDFLWDQFHV KDYH EHHQ UHODWLYHO\ LJQRUHG EHFDXVH RI PHDVXUHPHQW GLIILFXOW\ DQG UHODWLYHO\ VPDOO LPSDFW FRPSDUHG WR H[WULQVLF FDSDFLWDQFHV +RZHYHU DV SURFHVVLQJ DQG GLHOHFWULF WHFKQRORJ\ DGYDQFHV WKH SULPDU\ UHPDLQLQJ FDSDFLWLYH ORDG LQ &026 FLUFXLWV ZLOO EH WKH LQWULQVLF FDSDFLWDQFHV 7KH FKDSWHU SUHVHQWV DQ H[SHULPHQWDO LQYHVWLJDWLRQ KRZ WKHVH FDSDFLWDQFHV FKDQJH ZLWK KRWFDUULHU VWUHVV DQG DIWHU PRGHOLQJ WKH VWUHVV LQGXFHG FKDQJHV LQ WKH LQWULQVLF FDSDFLWDQFHV VKRZV WKDW SDUW RI WKH GUDLQ FXUUHQW GHJUDGDWLRQ LV RIIVHW E\ WKH LQWULQVLF FDSDFLWDQFH UHGXFWLRQ UHVXOWLQJ LQ D VORZHU GHJUDGDWLRQ RI RYHUDOO FLUFXLW SHUIRUPDQFH PAGE 11 &+$37(5 (;7(1',1* 7+( 21(',0(16,21$/ &855(17 02'(/ ,QWURGXFWLRQ 7KH VLPSOHVW PRGHO LV RI FUXFLDO LPSRUWDQFH IRU DSSOLFDWLRQV LQ VHPLFRQGXFWRU SK\VLFV $OWKRXJK PRGHOV ZLOO EHVW PDWFK H[SHULPHQWDO GDWD EHFDXVH RI ERWK LQFOXVLRQ RI UHDO HIIHFWV DQG VLPSO\ DGGLWLRQDO YDULDEOHV WKH\ PD\ EH LQWUDFWDEOH DV FRPSDFW GHYLFH PRGHOV ZKHUH FRPSXWDWLRQDO HIILFLHQF\ LV FULWLFDO &RQYHUVHO\ WKHVH PRGHOV DUH RIWHQ YDOLGDWHG E\ GHPRQVWUDWLQJ WKHLU UHGXFWLRQ WR WKH ULJRURXV IRUPV IRU QRQFULWLFDO ZLGH DQG ORQJ FKDQQHOV ZLWK WKLFN R[LGHVf JHRPHWULHV )RU EDFN RIWKHHQYHORSH FDOFXODWLRQV NQRZOHGJH RI WKH EDVLF SK\VLFV HPERGLHG LQ D JRRG PRGHO LV H[FHHGLQJO\ XVHIXO 7KH UHTXLUHG DFFXUDF\ RI D PRGHO LV ODUJHO\ GHWHUPLQHG E\ WKH DSSOLFDWLRQ )RU SUHGLFWLQJ WKH GULYH FXUUHQW VXFK DV PLJKW EH UHTXLUHG IRU D GLVFUHWHWUDQVLVWRU VSHFLILFDWLRQ VKHHW D PRGHO QHHG QRW ZRUU\ DERXW WKH OLQHDU RU VXEWKUHVKROG UHJLRQV RI RSHUDWLRQ 6LPLODUO\ LI PRGHOLQJ RQO\ WKH RSHUDWLQJ UDQJH WR SRZHU VXSSO\ YROWDJHf WKHQ WKH DFFXPXODWLRQ UHJLRQ RI DSSOLHG JDWH YROWDJHV FDQ EH LJQRUHG LQ WKH PRGHO 7KHUH DUH FDVHV SDUWLFXODUO\ ZKHQ DWWHPSWLQJ WR SUHGLFW WKH SHUIRUPDQFH RI QHZ WHFKQRORJ\ ZKHUH IXOOUDQJH 026)(7 PRGHOV DUH QHFHVVDU\ EXW WKH\ DUH D UHODWLYH PLQRULW\ FRPSDUHG WR WKH ZLGH DUUD\ RI DSSOLFDWLRQV IRU PRGHOV PAGE 12 7KLV FKDSWHU FRQWDLQV D EULHI KLVWRU\ RI RQHGLPHQVLRQDO 'f DSSURDFKHV WR GUDLQ FXUUHQW PRGHOV LQFOXGLQJ FDOFXODWLRQV DQG FRPSDULVRQV IROORZHG E\ D QHZ WZR VHFWLRQ PRGHO XVLQJ WKH 3DR6DK ORQJFKDQQHO ,9 PRGHO LQ FRQMXQFWLRQ ZLWK D YDULDEOHOHQJWK GHSOHWLRQ UHJLRQ 7KH JRDO LV WR H[WHQG WKH ORQJFKDQQHO PRGHO WR VKRUWFKDQQHO XVH %DFNJURXQG ,Q /LOLHQIHOG >@ VXEPLWWHG WKH SDWHQW IRU WKH ILUVW 026)(7 GHYLFH DQ $$&X6 WUDQVLVWRU 7KLUW\WZR \HDUV ODWHU LQ .DKQJ DQG $WDOOD >@ IDEULFDWHG WKH ILUVW VLOLFRQ 026 WUDQVLVWRU $ \HDU ODWHU WKH ILUVW 0267 FXUUHQWYROWDJH ,9f SDSHUV ZHUH SXEOLVKHG LQWHUQDOO\ DW $7t7 %HOO /DEV LQ E\ .DKQJ >@ DQG ODWHU DW 6WDQIRUG E\ ,KDQWROD >@ 7KHVH ZHUH IROORZHG LQ E\ PRUH FRPSOHWH DQG ZLGHO\ UHOHDVHGf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f )RU VKRUW FKDQQHOV KRZHYHU WKH HIIHFWLYH FKDQQHO OHQJWK FDQ EH VLJQLILFDQWO\ UHGXFHG E\ WKH HQFURDFKPHQW $QRWKHU VKRUWFKDQQHO HIIHFW QHJOHFWHG LQ ORQJFKDQQHO WKHRU\ LV GUDLQLQGXFHG EDUULHU ORZHULQJ >@ ZKHUH WKH VRXUFH EDUULHU LV ORZHUHG E\ WKH DSSOLHG GUDLQ YROWDJH 3DR6DK 0RGHO 7KH PRVW DFFXUDWH ORQJFKDQQHO WKHRU\ ZDV SXEOLVKHG E\ 3DR DQG 6DK 36f >@ 7KH 36 PRGHO LV WKH RQO\ RQH ZKLFK FRUUHFWO\ DFFRXQWHG IRU GULIW DQG GLIIXVLRQ 7KH 36 WKHRU\ WR EH GLVFXVVHG EHORZ FRQWDLQV D GRXEOH LQWHJUDO EXW FDQ EH UHGXFHG WR D PRUH HIILFLHQW IRUP FRQWDLQLQJ RQO\ VLQJOH LQWHJUDOV > @ $OWKRXJK FXPEHUVRPH WR FDOFXODWH WKH 36 GRXEOH LQWHJUDO LV H[WUHPHO\ GLGDFWLF DQG LV D XVHIXO VWDUWLQJ SRLQW IRU VKRZLQJ WKH DSSUR[LPDWLRQV XVHG WR GHULYH RWKHU ORQJFKDQQHO ,9 PRGHOV 7KH WRWDO FXUUHQW IORZLQJ LQ WKH FKDQQHO LV JLYHQ E\ WKH LQWHJUDO >[L ,Q -[\f= G[ f ZKHUH -[\f -1 -S -1 TSQ1(\ T'Q91 T'Q19e f -1 DQG -3 DUH WKH HOHFWURQ DQG KROH FXUUHQW GHQVLWLHV UHVSHFWLYHO\ DQG LW LV DVVXPHG WKDW WKH FXUUHQW LV GRPLQDWHG E\ HOHFWURQV LQ DQ QFKDQQHO GHYLFH LQ f 7KH HOHFWURQ FKDUJH LV T SQ DQG 'Q DUH WKH HOHFWURQ PRELOLW\ DQG GLIIXVLRQ UHVSHFWLYHO\ DQG 91 LV WKH PAGE 14 JUDGLHQW RI WKH HOHFWURQ FRQFHQWUDWLRQ 7KH HOHFWURQ TXDVL)HUPL OHYHOLV PHDVXUHG UHODWLYH WKH EXON )HUPL OHYHO DQG QRUPDOL]HG WR N7T ,I GAG[ LV DVVXPHG QHJOLJLEOH ZKLFK LV D IXQGDPHQWDO DVVXPSWLRQ LQ WKH ORQJFKDQQHO DSSUR[LPDWLRQ DQG VKRXOG EH YDOLG WR D GHSWK RQ WKH RUGHU RI WKH GUDLQ MXQFWLRQ GHSWKf WKHQ ,' FDQ EH IRXQG IURP VXPPLQJ XS DOO WKH FXUUHQW IURP WKH VXUIDFH GRZQ WR VRPH GHSWK [c EHORZ ZKLFK WKH DGGLWLRQDO FRQWULEXWLRQ LV QHJOLJLEOH ,G T'Q=GIG\f [L 1 [f G[ 7KLV FDQ EH WUDQVIRUPHG IURP SK\VLFDO VSDFH LQ WKH \ GLUHFWLRQ WR SRWHQWLDO VSDFH DV IROORZV / UX' ,' G\ T'Q= f ZKHUH 8' T9'6N7f LV WKH QRUPDOL]HG GUDLQ YROWDJH DW \ / DQG WKH ORZHU OLPLW LV WKH JURXQGHG VRXUFH YROWDJH DW \ $ VLPLODU WUDQVIRUP LQ WKH [ GLUHFWLRQ \LHOGV ,G 'Qa 8Q 8V 18f G8 f /-8) G8G[f ZKHUH G8G[f LV WKH [FRPSRQHQW RI WKH HOHFWULF ILHOG ZKLFK FDQ HDVLO\ GHULYHG IURP LQWHJUDWLQJ 3RLVVRQfV HTXDWLRQ E\ TXDGUDWXUH DQG LV JLYHQ EHORZ 7KH %ROW]PDQQ DSSUR[LPDWLRQ WR WKH FDUULHU FRQFHQWUDWLRQ LV EHLQJ XVHG DQG WKH LPSXULWLHV DUH DVVXPHG FRPSOHWHO\ LRQL]HG EXW WKH )HUPL'LUDF DQG GHLRQL]HG IRUP FDQ EH XVHG 8V LV WKH QRUPDOL]HG VXUIDFH SRWHQWLDO ZKHUH VXUIDFH LV DW [ f WKH WRWDO DPRXQW RI VXUIDFH EDQG EHQGLQJ UHODWLYH WR WKH LQWULQVLF )HUPL OHYHO ,W LV D IXQFWLRQ RI ERWK WKH JDWH YROWDJH DQG WKH GUDLQ YROWDJH 8) LV WKH QRUPDOL]HG EXON )HUPL OHYHO EHORZ ZKLFK WKH FXUUHQW PAGE 15 FRQWULEXWLRQ LV DVVXPHG QHJOLJLEOH DQG LV DQDORJRXV WR WKH WR SK\VLFDO SRLQW [ [c LQ f 7KH GHULYDWLYH G8G[f LV IRXQG IURP G8G[f )8e8If/G f ZKHUH ) 8 e 8)f >H[S 8e8Sf H[S8)8f 8OfH[S8)f 8H[S ef f H[S fÂ§8)f @ f $IWHU DSSO\LQJ (LQVWHLQfV UHODWLRQVKLS 'QSQ N7T f EHFRPHV G8Ge f 7KH VXUIDFH SRWHQWLDO 8Vcf LV QHHGHG LQ f 7KH UHODWLRQVKLS EHWZHHQ WKH VXUIDFH SRWHQWLDO DQG WKH JDWH YROWDJH FDQ EH IRXQG E\ DSSO\LQJ *DXVVfV /DZ DW WKH VHPLFRQGXFWRULQVXODWRU LQWHUIDFH 7KH UHVXOWLQJ HTXDWLRQ JLYHQ EHORZ FDQ EH VROYHG LWHUDWLYHO\ IRU 8V IRU D JLYHQ UN7 ] .rE 8' 8V H[S 8fÂ§8)f T / /G 8) )8 I8)f 8f 8r VLJQ8Vf \)8V_8)f f ZKHUH 8 LV WKH QRUPDOL]HG JDWH YROWDJH T9*6 9)%fN7 \ LV HV/'m&f /' LV WKH 'HE\H OHQJWK 9>HVN7Qcf@Tf DQG )8683f LV JLYHQ E\ f (TXDWLRQ LV WKH WUDGLWLRQDO IRUP RI WKH 36 LQWHJUDO RIWHQ FDOOHG WKH 3DR6DK GRXEOH LQWHJUDO $ PRUH FRPSXWDWLRQDOO\ IULHQGO\ DQG DFFXUDWH VLQJOHLQWHJUDO IRUP >@ ZDV XVHG IRU WKH FDOFXODWLRQV LQ WKLV GLVVHUWDWLRQ 7KH PRELOLW\ LQ f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f (OHFWURQV LQ WKH VRXUFH DQG GUDLQ VHH D SRWHQWLDO EDUULHU WRZDUG WKH FKDQQHO $SSOLFDWLRQ RI D SRVLWLYH YROWDJH WR WKH JDWH ORZHUV WKH EDUULHU QHDU WKH VXUIDFH DV VKRZQ LQ )LJXUH 7KH DSSOLHG JDWH YROWDJH SXOOV HOHFWURQV WRZDUG WKH VXUIDFH DQG SXVKHV KROHV DZD\ IURP WKH VXUIDFHf DV FDQ EH VHHQ IURP WKH SRVLWLRQ RI WKH )HUPLOHYHO UHODWLYH WR WKH EDQG HGJHV )DUWKHU LQWR WKH VXEVWUDWH DZD\ IURP WKH JDWHVXEVWUDWH LQWHUIDFHf WKHUH LV QR EHQGLQJ IURP WKH JDWH SRWHQWLDO VR WKH UHJLRQ LV LGHQWLFDO WR WKH XQELDVHG FDVH )LJXUH f DQG FRQVLGHUHG TXDVLQHXWUDO $SSO\LQJ D YROWDJH WR WKH GUDLQ 9'6 9'6VDWf VSOLWV WKH )HUPL OHYHO LQWR TXDVL )HUPL OHYHOV )Q IRU HOHFWURQV DQG )S IRU KROHVf DV VKRZQ LQ )LJXUH 2QH FDQ LPDJLQH DQ HOHFWURQ LQ WKH FRQGXFWLRQ EDQG VXUPRXQWLQJ WKH VRXUFH EDUULHU DQG WKHQ IDOOLQJ GRZQ WKH SRWHQWLDO fFOLIIf XQWLO UHDFKLQJ WKH GUDLQ 7KLV fIUHH IDOOf LV ZKHUH WKH HOHFWURQ JDLQV HQHUJ\ ZKLOH PRYLQJ DFURVV WKH FKDQQHO ,I WKH HOHFWURQ LV QRW VFDWWHUHG ZKLOH PRYLQJ DFURVV WKH FKDQQHO ORVLQJ HQHUJ\ WR WKH ODWWLFH YLD SKRQRQVf LW EHFRPHV LQFUHDVLQJO\ HQHUJHWLF DV LW DSSURDFKHV WKH GUDLQ DQG PD\ EHFRPH fKRWf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f ZKLFK LV D FURVV VHFWLRQ RI WKH FKDQQHO DV VKRZQ LQ )LJXUH 7KH RXWHU LQWHJUDO LV LQWHJUDWLQJ IURP GUDLQ WR WKH VRXUFH 8' WR VRXUFH LV JURXQGHGf DORQJ WKH FKDQQHO 7KXV WKH GRXEOH LQWHJUDO LV VXPPLQJ XS DOO WKH FXUUHQW FRQWULEXWLRQ LQ WKH FKDQQHO H[DFWO\ DV ZRXOG EH H[SHFWHG 6LQFH 8V LV D IXQFWLRQ RI WKH GUDLQ YROWDJH RU WKH FKDQQHO SRWHQWLDOf WKH RUGHU RI WKH GRXEOH LQWHJUDWLRQ LV QRW WULYLDOO\ UHYHUVLEOH %XON&KDUJH 0RGHO 7KH ILUVW JURXS RI ,' PRGHOV LQ RUGHU RI FRPSOH[LW\ ZHUH E\ 6DK >@ ,KDQWROD DQG 0ROO >@ DQG 6DK DQG 3DR >@ 7KHVH DUH DOO EXON FKDUJH PRGHOV WDNLQJ LQFUHDVLQJO\ PRUH LQWR DFFRXQW $V WKH QDPH VXJJHVWV WKH EXON FKDUJH PRGHO WDNHV WKH GHSOHWHG UHJLRQ XQGHU WKH FKDQQHO LQ WKH EXONf LQWR DFFRXQW ,W DVVXPHV GULIW LV WKH PDMRU FRPSRQHQW DQG VR QHJOHFWV WKH GLIIXVLRQ FRPSRQHQW 7KLV JUHDWO\ VLPSOLILHV WKH SUREOHP DQG UHGXFHV f WR -[\f -1 -S -1 TSQ1(\ TSQ1[f G9G\f f PAGE 21 )LJ 6FKHPDWLF HQHUJ\ EDQG GLDJUDP RI VLPSOH 026 GHYLFH ZLWK 9*6 9GV A'6VDWf DQA VRXUFH JURXQGHG $GDSWHG IURP 3DR DQG 6DK >@ PAGE 22 )LJ 6FKHPDWLF HQHUJ\ EDQG GLDJUDP RI VLPSOH 026 GHYLFH ZLWK 9*6 DQG 9'6 9'6VDO &URVVVHFWLRQV VKRZ WKH HQHUJ\EDQG GLDJUDPV QHDU WKH VRXUFH DQG GUDLQ *A*DWH HOHFWURGH ;V6XEVWUDWH HOHFWURGH PAGE 23 ,G TWQ= G9G\f L T1 [f G[ f ,' QQ=G9G\f4Q f ZKHUH 4Q &9* Y 9Vf T3[[eVf>9VR 9@ f & LV WKH R[LGH FDSDFLWDQFH SHU XQLW DUHD 9Â LV 9RV 9)% 9VR LV WKH VXUIDFH SRWHQWLDO DW WKH VRXUFH DQG 9 LV WKH FKDQQHO SRWHQWLDO 9'6 DW WKH XQVDWXUDWHG GUDLQf 3[[ LV WKH VXEVWUDWH LPSXULW\ FRQFHQWUDWLRQ DQG HV LV WKH GLHOHFWULF FRQVWDQW RI VLOLFRQ 7KH ILUVW WHUP LV WKH FKDUJH DFFXPXODWHG LQ WKH FKDQQHO DQG WKH VHFRQG WHUP LV WKH XQFRPSHQVDWHG FKDUJH LQ WKH GHSOHWLRQ UHJLRQ EHQHDWK WKH FKDQQHO LH EXON FKDUJHf ,QWHJUDWLQJ f DORQJ WKH FKDQQHO JLYHV ,' SQ=/f&^ 9* 9Vf9'6 9V f &f f T3[[eVf > 9V 9'6f 9Vf@` 7KLV IRUP LV VOLJKWO\ GLIIHUHQW WKDQ WKH 6DK3DR DQG ,KDQWROD0ROO IRUPV EHFDXVH LW LV QRW DVVXPHG WKDW 96 9) ZKHUH 9) LV WKH )HUPL YROWDJH $ PRUH H[DFW IRUP >@ LV ,' SQ=/f&^ 9J9VO 9Vf f 9O 9f f f &f T3[[eVf>96/f 96f@` ZKHUH 96/ LV WKH VXUIDFH SRWHQWLDO DW WKH GUDLQ 7KLV GLIIHUV IURP f LQ WKDW WKH VXUIDFH SRWHQWLDO DW WKH GUDLQ LV FDOFXODWHG LQVWHDG RI DVVXPHG WR EH 96/ 96 9'6 :KHQ WKH GUDLQ FXUUHQW DSSURDFKHV RU H[FHHGV VDWXUDWLRQ 9'6 9'66DWf 96/Ve96 9'6 $GGLWLRQDOO\ LQ VXEWKUHVKROG 96/ LV W\SLFDOO\ FORVHU WR 9VR WKDQ 9VR 9'6 >@ $V ZLOO PAGE 24 EH VKRZQ ODWHU WKH EXON FKDUJH IRUPXOD VKRXOG QHYHU EH XVHG IRU VXEWKUHVKROG FDOFXODWLRQV VLQFH LW QHJOHFWV GLIIXVLRQ ZKLFK LV WKH SULPDU\ VXEWKUHVKROG FXUUHQW FRQWULEXWLRQ 7KH EXON FKDUJH IRUP FRPSDUHG WR 36 LV FRQVLGHUDEO\ HDVLHU WR FDOFXODWH SDUWLFXODUO\ ZKHQ XVLQJ f ZLWK 9VR 9) EXW LV LQYDOLG LQ VXEWKUHVKROG (TXDWLRQ LV DOVR LQYDOLG LQ VDWXUDWLRQ DV ZULWWHQ EXW WKDW FDQ EH IL[HG VRPHZKDW E\ FDOFXODWLQJ WKH VDWXUDWLRQ YROWDJH 9'6VDW DQG IL[LQJ WKH FXUUHQW IRU DOO GUDLQ YROWDJHV JUHDWHU WKDQ 9'6VDW 7KLV ZLOO PDNH WKH ILUVW GHULYDWLYH GUDLQ FRQGXFWDQFHf QRQFRQWLQXRXV DW 9'6 9'6VDW $OO VDWXUDWLRQ SUREOHPV DUH VROYHG LQ f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f LQWHJUDO :KHQ KH SURSRVHG KLV FKDUJHVKHHW PRGHO KH LQWURGXFHG WKH IROORZLQJ VLPSOLILFDWLRQ T=-Q1\f GIG\f GeG\ GVG\ S GOQQfG\ f PAGE 25 7KLV DSSUR[LPDWLRQ IRU G=-G\ ZDV MXVWLILHG EDVHG XSRQ LWV VXFFHVV LQ SURGXFLQJ fFRUUHFWf 9 FXUYHV DOWKRXJK KH DGGHG D IRRWQRWH UHODWLQJ WKH IRUPXOD WR HOHFWURFKHPLFDO SRWHQWLDO 7KLV ZLGHRSHQ VWDWHPHQW UHVXOWHG LQ VHYHUDO VXEVHTXHQW fSURRIVf ZKLFK GHULYHG WKH VDPH IRUPXOD > @ (VVHQWLDOO\ WKRXJK KH GHFRXSOHG WKH GULIW DQG GLIIXVLRQ FRPSRQHQWV IURP WKH WLJKW LQWHUGHSHQGHQF\ VHHQ LQ WKH 3DR6DK IRUP WR WKH VLPSOH IRUP RI f 7KURXJK D VLPLODU GHULYDWLRQ WR EXONFKDUJH ,' LV JLYHQ E\ LQ=\f f ^&OM9*f 9V\f 9Vf f & 9 \f 9_f f f T3[[HVf> 9V\f f S9V f@ T3[[eVf> -9V\f f 962 ff (T UHGXFHV WR EXONFKDUJH IRUP RI (T LI 9* 9V\f 9VR } S DQG WKH VTXDUH URRW WHUPV DUH QHJOLJLEOH 8QOLNH EXONFKDUJH WKLV IRUPXOD LV YDOLG LQ VXEWKUHVKROG DQG GRHV QRW UHTXLUH D FDOFXODWLRQ RI 9'6VDW DVVXPLQJ 96/ DQG 9VR DUH FDOFXODWHG LWHUDWLYHO\f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Âƒ 7 3[[ FP :/ 7KHVH DUH W\SLFDO SDUDPHWHUV IRU /6, GHYLFHV RI WKH V DQG ZHUH FKRVHQ WR PDWFK WKH GDWD XVHG LQ 3LHUUHW DQG 6KHLOGV >@ $V FDQ EH VHHQ WKH EXONFKDUJH DQG FKDUJHVKHHW PRGHOV XQGHUHVWLPDWH WKH FXUUHQW )LJXUH VKRZV WKH SHUFHQWDJH HUURU IRU HDFK PRGHO DW WKH JDWH YROWDJHV VKRZQ LQ )LJ GHPRQVWUDWLQJ WKDW WKH FKDUJHVKHHW PRGHO PDLQWDLQV DQ HUURU RI OHVV WKDQ b IRU DOO JDWH YROWDJHV ZKLOH WKH EXON FKDUJH PRGHO UDQJHV IURP b IRU 9&6 9 WR b IRU 9RV 9 7KLV VXJJHVWV WKDW WKH PXFK VLPSOHU FKDUJH VKHHW FDQ EH XVHG LQ SODFH RI 3DR6DK LQFXUULQJ RQO\ DERXW b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f WKH PRVW SURPLQDQW QRQPRGHOHG HIIHFW RQ WKH GUDLQ FXUUHQW LV ILQLWH GUDLQ FRQGXFWDQFH EH\RQG VDWXUDWLRQ 7KH SULPDU\ FDXVH RI WKLV QRQ]HUR GUDLQ FRQGXFWDQFH J'f LV FKDQQHO VKRUWHQLQJ IURP WKH GUDLQ VSDFHFKDUJH UHJLRQ 6&5f PAGE 27 ,' YHUVXV 9'6 IRU GLIIHUHQW 9*6 YDOXHV IRU WKH WKUHH ,' PRGHOV 3DUDPHWHUV DUH 7R[ Âƒ 7 3[[ FPn :/ ZKLFK ZHUH XVHG WR PDWFK GDWD LQ 3LHUUHW DQG 6KLHOGV >@ PAGE 28 )LJ 3HUFHQWDJH HUURU LQ ,' IRU FKDUJH VKHHW DQG EXON FKDUJH UHODWLYH WR 3DR 6DK YHUVXV 9'6 IURP )LJ 3ORWV DUH 9RV DQG 9 ZLWK KLJKHU HUURUV IRU ORZHU YROWDJHV PAGE 29 9* 9f )LJ ,' YHUVXV 9*6 IRU 3DR6DK FKDUJH VKHHW DQG EXON FKDUJH XVLQJ VDPH GDWD DV )LJ ZLWK 9'6 9 &OHDUO\ EXON FKDUJH LV QRW XVHIXO LQ VXEWKUHVKROG ZKHUHDV FKDUJHVKHHW LV DOPRVW FRLQFLGHQW ZLWK 3DR6DK PAGE 30 HQFURDFKLQJ LQWR WKH FKDQQHO 7KLV HIIHFW LV RIWHQ FDOOHG FKDQQHOOHQJWK PRGXODWLRQ VLQFH WKH GUDLQ YROWDJH PRGXODWHV WKH HIIHFWLYH FKDQQHO OHQJWK 7KH PRVW ORJLFDO DSSURDFK LV WR GLYLGH WKH UHJLRQ EHWZHHQ WKH VRXUFH DQG GUDLQ LQWR WZR VHFWLRQV D fVRXUFH VLGHf DQG D fGUDLQ VLGHf 7KH fVRXUFH VLGHf PD\ FRQWDLQ DQ\ DSSURSULDWH ORQJFKDQQHO ,9 PRGHO VXFK DV 3DR6DK FKDUJH VKHHW RU EXON FKDUJH 7KH fGUDLQ UHJLRQf LV WKH GHSOHWLRQ UHJLRQ DQG FDQ EH PRGHOHG ZLWK RU ZLWKRXW PRELOH FKDUJH HIIHFWV PRELOLW\ GLIIHUHQFHV HWF 7KH ORFDWLRQ RI WKH ERXQGDU\ EHWZHHQ WKHVH UHJLRQV DQG WKH YROWDJHV DQG ILHOGV DW WKLV ERXQGDU\ DUH ZKDW PDNH WKLV D FKDOOHQJLQJ SUREOHP )LJXUH VKRZV D GLDJUDP RI D 026 WUDQVLVWRU GLYLGHG LQWR WZR VHFWLRQV 7KHUH DUH HVVHQWLDOO\ WKUHH WKLQJV ZKLFK GLIIHU DPRQJ DSSURDFKHV WR WZRVHFWLRQ WKHRU\ WKH WKH VRXUFHVLGH ,9 PRGHO WKH GUDLQVLGH VSDFHFKDUJH UHJLRQ 6&5f PRGHO DQG WKH ERXQGDU\ FRQGLWLRQV 6RXUFH 6RXUFH 6LGH /RQJ FKDQQHO DSSUR[f 'UDLQ 6LGH 6&5f 'UDLQ /H: /$/f n $/ / \ R \ \0 \ / Y R Y Y0 Y Y' )LJ 6FKHPDWLF GLDJUDP RI WZRVHFWLRQ 0267 IRU PRGHOLQJ 6&5 PHDQV f6SDFH&KDUJH 5HJLRQf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f 5HGGL6DK DQG RWKHUVf FDOFXODWHG WKH GUDLQ YROWDJH ZKHUH IRU D FRQVWDQW JDWH YROWDJH WKH GUDLQ FRQGXFWDQFH GURSV WR ]HUR 9'6VDWf 7KH\ WKHQ DVVXPHG DOO YROWDJH LQ H[FHVV RI 9'6VDW IDOOV DFURVV WKH 6&5 WR IRUP WKH GUDLQ UHJLRQ RI WKH WZRVHFWLRQ PRGHO %\ DVVXPLQJ FRPSOHWH GHSOHWLRQ QR PRELOH FKDUJHf DQG QR \ILHOG DW WKH ERXQGDU\ WKH OHQJWK FDQ EH FDOFXODWHG IURP VLPSOH SQ MXQFWLRQ WKHRU\ DV $/ >HV 9'6 9'6VDW 9ELfT3[[f@ f ZKHUH 9EM N7TfOQ1GUDLQ1VXEVOUDWHQIf IURP VWDQGDUG DEUXSWMXQFWLRQ SQ WKHRU\ 5HSODFLQJ / E\ /HII /$/ DQG 9VR ZLWK 93 LQ f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f WR DQRWKHU FRQVWDQW ,'f DOWKRXJK WKLV FDQ EH IL[HG ZLWK YDULRXV VPRRWKLQJ WUDQVLWLRQDO IXQFWLRQV )RXU \HDUV DIWHU 5HGGL DQG 6DKfV SDSHU &KLX DQG 6DK >@ FDPH RXW ZLWK D WZR VHFWLRQ PRGHO ZKLFK VROYHG /DSODFH HTXDWLRQ LQ WKH R[LGH OD\HU DQG PDWFKHG YDOXHV LQ IRXU UHJLRQV VRXUFH GUDLQ R[LGH DQG EXONf 7KH GUDLQ UHJLRQ ZDV VROYHG DV D IXOO\ GHSOHWHG UHJLRQ DQG WKH VROXWLRQ UHTXLUHG VHYHQ PDWFKLQJ SDUDPHWHUV 7KH FRPSOH[LW\ RI WKH VROXWLRQ UHOLQTXLVKHG WKLV PRGHO WR DQ DOPRVW FRQVWDQW UHIHUHQFH DV WRR FRPSOH[f 7KH IROORZLQJ \HDU f )URKPDQ%HQWFKNRZVN\ DQG *URYH >@ GHYHORSHG D WZRVHFWLRQ PRGHO XVLQJ EXONFKDUJH PRGHO LQ WKH VRXUFH UHJLRQ DQG DQ HPSLULFDO PRGHO IRU WKH GUDLQ VHFWLRQ 7KLV VLPSOH PRGHO HVVHQWLDOO\ DGGHG WZR DGGLWLRQDO IULQJH ILHOG FRQWULEXWLRQV WR WKH 5HGGL6DK PRGHO DQG DGGHG WZR HPSLULFDO YDULDEOHV WR ILW WKH GDWD 0HUFNHO %RUHO DQG &XSFHD >@ DGGHG PRELOH FKDUJH WR WKH GUDLQ UHJLRQ HPSLULFDOO\ E\ ZULWLQJ 3RLVVRQfV HTXDWLRQ LQ WKH GUDLQ UHJLRQ DV G9G\ TeV3[[ ,'6fT=Df f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f RU PRELOLW\ %RWK XVHG YDULDWLRQV RI WKH VLPSOH EXON FKDUJH PRGHO IRU WKH VRXUFH VLGH PAGE 33 $IWHU %UHZV GHYHORSHG WKH FKDUJHVKHHW PRGHO DOO VXEVHTXHQW WZRVHFWLRQ PRGHOV HPSOR\HG WKH FKDUJHVKHHW PRGHO *XHEHOV DQG 9DQ GH :LHOH >@ GHYHORSHG D WKUHH VHFWLRQ PRGHO WR DFFRXQW IRU WKH [ILHOG UHYHUVDO QHDU WKH GUDLQ 7KH\ HPSOR\ WKH VDPH WULFN DV WKH SUHYLRXV SDSHUV E\ ILWWLQJ WKH D LQ f XVLQJ 9'6VDW RU ,'VDWf DQG DGGLQJ VRPH HPSLULFLVP WR WKHLU ILHOG FDOFXODWLRQV %H\RQG 7ZR6HFWLRQ 0RGHOV 7KH FKDUJHVKHHW PRGHO DQG 3DR6DK DV ZLOO EH VKRZQf GRHV QRW OHQG LWVHOI ZHOO WR DQDO\WLFDO WZRVHFWLRQ PRGHOV GXH WR WKH JUHDWHU FRPSOH[LW\ RI WKH GUDLQ FXUUHQW PRGHO UHODWLYH WR EXON FKDUJH $V QRWHG DERYH ILWWLQJ SDUDPHWHUV DQG HPSLULFDO IRUPXODH ZHUH UHTXLUHG WR EH LQWURGXFHG WR VDWLVI\ VRPH RI WKH ERXQGDU\ FRQGLWLRQV 7KH QHZHU FRPSDFW PRGHOV VXFK DV %6,0 >@ DQG 6LHPHQfV >@ PRGHO DUH EDVHG ORRVHO\ RQ RQHVHFWLRQ EXONFKDUJH DQG FKDUJHVKHHW PRGHOV UHVSHFWLYHO\ VRPHWLPHV GLYLGLQJ WKH PRGHO LQWR GLIIHUHQW VHFWLRQV EDVHG RQ RSHUDWLRQ VHSDUDWH VXEWKUHVKROG DQG VXSHUWKUHVKROG IRUPXODHf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fV HTXDWLRQ LQ WKH GUDLQ UHJLRQ ZKLOH FRQVLGHULQJ WKH ERXQGDU\ FRQGLWLRQV :LWKLQ WKH GUDLQ UHJLRQ ZKLFK H[WHQGV IURP \ \0 WR \ / WKH ERXQGDU\ FRQGLWLRQV DUH VHH )LJ f 9/f 9GV Y\0f Y0 G9\0fG\ (0 ILHOG DW WKH PDWFK SRLQWf G9\0fG\ OHVf>T3;[ PRELOH FKDUJH WHUPVf@ & ,W LV SRVVLEOH IURP 3DR6DK WR FDOFXODWH G9\0fG\ (0SV >@ 7KLV JLYHV XV WKH IROORZLQJ HTXDWLRQV DIWHU LQWHJUDWLQJ WKH 3RLVVRQfV HTXDWLRQ WZLFH ZLWK WKH DERYH ERXQGDU\ FRQGLWLRQV 9'6 9 &f / \0f (P/ \0f f 7KLV UHGXFHV DOO WKH ERXQGDU\ FRQGLWLRQV WR RQH HTXDWLRQ ZLWK WZR XQNQRZQV \0 DQG 90f 7KH LGHDO DGGLWLRQDO HTXDWLRQ ZRXOG EH G9\0fG\ RQ WKH 3DR6DK VLGH EXW WKLV TXDQWLW\ LV LQFDOFXODEOH IURP WKH 3DR6DK LQWHJUDO ,I LW LV DVVXPHG WKDW DVVXPH (P DV ZDV GRQH LQ 5HGGL6DKf WKH GHSOHWLRQ OHQJWK LQWR WKH FKDQQHO FDQ EH HDVLO\ IRXQG ,W LV UHDVRQDEOH WR DVVXPH WKDW WKH ODWHUDO ILHOG DW WKH PDWFKLQJ SRLQW (0f LV PXFK OHVV WKDQ WKH ILHOG ULJKW DW WKH GUDLQ ('f VR (' } (P PDNLQJ WKH GLIIHUHQFH LQ \0 VPDOO 7KLV JLYHV IURP DOVR f PAGE 35 \0 / 9'6 90 9ELf&f f :KHUH 9EL DFFRXQWV IRU WKH SUHH[LVWLQJ GHSOHWLRQ UHJLRQ RULJLQDWLQJ IURP WKH DEUXSW SQ MXQFWLRQ 6LQFH WKH \0 DSSUR[LPDWLRQ KDV DOUHDG\ EHHQ PDGH LW ZLOO EH DVVXPHG WKDW WKH ILHOG WKURXJKRXW WKH GUDLQ UHJLRQ LV D FRQVWDQW DW WKH ERXQGDU\ DQG LV JLYHQ E\ (0GHS 9'6 9Pf/ \0f f &OHDUO\ WKHUH DUH FRQIOLFWLQJ DVVXPSWLRQV (0 DQG QRZ (0 s f 2QH PLJKW ZRQGHU ZK\ (P LV QRW 9'6 90 9ELf/ \0f WR EH FRQVLVWHQW ZLWK 7KLV FRPHV IURP WKH VXEWOHW\ RI WKH ERXQGDU\ FRQGLWLRQV /RRNLQJ EDFN WR )LJXUH QRWH WKDW WKH LQWHJUDWLRQ LV DFWXDOO\ IURP 9V 9EM WR 9'6 9EM ZKLFK H[FOXGHV WKH SQ GHSOHWLRQ OD\HUV 7KH 9ELfV FDQFHO RXW IRU V\PPHWULFDO GHYLFHV VR WKLV LV QR SUREOHP $W 9'6 DQG VRXUFH JURXQGHGf QR FXUUHQW RU ILHOG LV H[SHFWHG ZKLFK ZRXOG PDNH 90 FRUUHFWO\ HTXDO WR LQ f +RZHYHU LI 9EL ZHUH DGGHG WR f WKHQ 90 ZRXOG KDYH WR HTXDO 9Ec ZKLFK ZRXOG LQFRUUHFWO\ FDXVH D ILHOG DQG SRVVLEO\ FXUUHQW IORZ GHSHQGLQJ RQ 9RVf (VVHQWLDOO\ f JLYHV WKH H[FHVV ILHOG +RZHYHU 9EL GRHV FRQWULEXWH WR WKH GHSOHWLRQ ZLGWK VR LW LV LQFOXGHG LQ f 7KH QRUPDOL]HG ILHOG RQ WKH 3DR6DK VLGH DW WKH ERXQGDU\ LV JLYHQ E\ >@ > H[S 8Vf @ H[S 808)f n0SV U )8V8P8If H[S 8V808)fH[S 8)8VfH[S8)fH[S 8)f 9 / rPXV H[S8I8)f G8GI 8) )8I8)f [ Â‘G8 f PAGE 36 ZKHUH 80 LV WKH QRUPDOL]HG PDWFKLQJ YROWDJH 90rTN7f )LJXUH VKRZV WKH UHVXOWV RI WKLV DSSURDFK ZLWK PRELOH FKDUJH WHUPV QHJOHFWHG & T3[[HVLf IRU 3[[ [O FP 7 DQG 7R[ Âƒ 7KH GDWD FRYHU D ZLGH UDQJH RI FKDQQHO OHQJWKV IURP n8 SP WR rr DQG IRU DOO FDVHV WKH ZLGWK LV HTXDO WR WKH OHQJWK VTXDUH GHYLFHVf 7KH VDWXUDWLRQ FXUUHQW SUHGLFWHG E\ ORQJFKDQQHO WKHRU\ IRU WKHVH VTXDUH GHYLFHV ZRXOG EH WKH VDPH IRU DOO FKDQQHO OHQJWKV VR WKH GHYLDWLRQ IURP WKLV LV GXH WR FKDQQHO OHQJWK PRGXODWLRQ ZKLFK FOHDUO\ EHFRPHV PRUH LPSRUWDQW DQG WKH FKDQQHO OHQJWK GHFUHDVHV )LJXUH VKRZV WKDW WKH GUDLQ FRQGXFWDQFH J' G,'G9'6f LV VPRRWK ZKLFK LV LPSRUWDQW IRU FLUFXLW VLPXODWRU DSSOLFDWLRQV $OWKRXJK QRW VKRZQ WKH GHULYDWLYH RI WKH GUDLQ FRQGXFWDQFH LV DOVR VPRRWK 7KXV WKLV ILHOGPDWFKLQJ PRGHO VXFFHVVIXOO\ H[WHQGV WKH 3DR6DK PRGHO WR VKRUWFKDQQHOV DW OHDVW ZLWK UHJDUGV WR LQFOXGLQJ WKH HIIHFWLYH FKDQQHOn VKRUWHQLQJ HIIHFW 6DWXUDWLRQ9ROWDJH 0HWKRG 5HGGL DQG 6DK >@ DVVXPHG 90 9'6VDW ZKLFK VLPSOLILHG WKLQJV FRQVLGHUDEO\ 9'VVDW rV HDV\ WR FDOFXODWH ZKHQ XVLQJ WKH EXONFKDUJH IRUPXOD DVVXPLQJ 96 9) VLQFH WKH GHULYDWLYH RI WKH VXUIDFH SRWHQWLDO ZLWK UHVSHFW WR WKH GUDLQ YROWDJH LV ]HUR 7KH 3DR6DK FXUUHQW KRZHYHU GRHV QRW WHFKQLFDOO\ VDWXUDWH QXPHULFDOO\ WKHUH ZLOO EH D SRLQW ZKHUH WKH FXUUHQW GRHV QRW LQFUHDVH EXW LW ZLOO EH DW D GUDLQ YROWDJH ZHOO LQ H[FHVV RI WKH QRUPDO 9'6VDW SRLQWf 7KLV SUREOHP LV VROYHG E\ H[WUDSRODWLQJ 9'6VDW IURP G,'G9'6 YHUVXV 9'6 ZLWKRXW FKDQQHO VKRUWHQLQJ )LJXUHV DQG VKRZ WKH UHVXOWV RI HPSOR\LQJ WKLV PHWKRG ZLWK WKH VDPH GHYLFH DV XVHG LQ WKH SUHYLRXV VHFWLRQ 3[[ [O FP 7 7R[ Âƒf XVLQJ (T IRU \0 ZLWK 90 9'6VDW &OHDUO\ WKH FKDQQHOOHQJWK PRGXODWLRQ LV EHLQJ PAGE 37 DFFRXQWHG IRU EXW WKH WUDQVLWLRQ LV VOLJKWO\ DEUXSW $ ORRN DW WKH UHVXOWLQJ GUDLQ FRQGXFWDQFH )LJ f VKRZV D GUDVWLF GLVFRQWLQXLW\ QHDU WKH FDOFXODWHG 9'6VDW SRLQW 8VH RI D ILWWLQJ IXQFWLRQ FRXOG UHFWLI\ WKLV GHULYDWLYH SUREOHP DQG LV D FRPPRQ SUDFWLFH IRU FRPSDFW PRGHOV 6XUIDFH3RWHQWLDO 6HOI6DWXUDWLRQ 0HWKRG $QRWKHU SRVVLEOH ZD\ WR FLUFXPYHQW ILQGLQJ WKH 90 SRLQW ZDV SRVHG E\ .DWWR DQG ,WRK >@ ,QVWHDG RI ILQGLQJ 9'6VDW WKH\ XVHG WKH IDFW WKDW WKH VXUIDFH SRWHQWLDO LWVHOI ZLOO VDWXUDWH ZKHQ VROYHG LWHUDWLYHO\ IURP f 7KXV UHSODFLQJ WKH PDWFKLQJ YROWDJH 90 ZLWK WKH VXUIDFH SRWHQWLDO DW WKH GUDLQ 96/ VROYHG LWHUDWLYHO\f JLYHV DQRWKHU GHFRXSOHG ZD\ WR VROYH IRU \0 8VLQJ WKH VXUIDFH SRWHQWLDO WR ILQG WKH GHSOHWLRQ WKLFNQHVV ZDV DOVR XVHG E\ 6DK >@ 7KLV LV EHWWHU WKDQ WKH 9'6VDW PHWKRG VLQFH WKHUH ZLOO QRW EH DQ LPPHGLDWH SRLQW ZKHUH VDWXUDWLRQ RFFXUV +RZHYHU DV VKRZQ LQ )LJV DQG WKH FXUUHQW VWLOO KDV D VOLJKW fMXPSf UHVXOWLQJ LQ GLVFRQWLQXLWLHV LQ J' ,Q 6HDUFK RI WKH 0DWFK 3RLQW 6DK >@ VKRZHG SLFWRULDOO\ WKDW LQ VDWXUDWLRQ WKH HQHUJ\ EDQG QHDU WKH GUDLQ HGJH ZLOO DFWXDOO\ EH EHQW XSZDUG RU LQ RWKHU ZRUGV WKH VXUIDFH ZLOO EH DFFXPXODWHG UDWKHU WKDQ LQYHUWHG DFWXDOO\ WKH VXUIDFH ZLOO VWLOO EH GHSOHWHG EXW QRZ DFFXPXODWLRQ UHIHUV RQO\ WR WKH VKDSH RI WKH EDQG EHQGLQJf 7KLV PXVW EH WKH FDVH VLQFH WKH SRWHQWLDO DORQJ WKH FKDQQHO LV DFWXDOO\ KLJKHU WKDQ 9*6 9*7 9'6VDW 7KLV PHDQV WKDW WKHUH PXVW EH D SRLQW DORQJ WKH FKDQQHO DW ZKLFK WKH EDQG EHQGLQJ LV ]HUR DW WKH VXUIDFH DQG WKLV SRLQW ZRXOG EH DQ H[FHOOHQW FDQGLGDWH IRU WKH \0 SRLQW /LNH WKH PHWKRGV DERYH KRZHYHU WKLV SRLQW KDV VRPH PAGE 38 )LJ ,' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWKV VTXDUH GHYLFHVf XVLQJ ILHOG PDWFKLQJ DW WKH PDWFK SRLQW 9RV 9 3[[ [O FP 7 7R[ Âƒ PAGE 39 J' P$9f )LJ J' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWK VTXDUH GHYLFHVf XVLQJ ILHOG PDWFKLQJ DW WKH PDWFK SRLQW 6DPH SDUDPHWHUV DV )LJ PAGE 40 )LJ ,' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWKV VTXDUH GHYLFHVf XVLQJ 90 LWHUDWLYH VXUIDFH SRWHQWLDO DW GUDLQ 9*6 9 3[[ [O FP 7 7R[ Âƒ PAGE 41 J' P$$f )LJ J' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWK VTXDUH GHYLFHVf XVLQJ 90 LWHUDWLYH VXUIDFH SRWHQWLDO DW GUDLQ 6DPH SDUDPHWHUV DV )LJ PAGE 42 )LJ ,' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWKV VTXDUH GHYLFHVf XVLQJ 90 9'6VDF YFV 9 3[[ [O2Q FQU 7 7R[ Âƒ PAGE 43 J' P$9f )LJ J' YHUVXV 9'6 SORWV IRU GLIIHUHQW FKDQQHO OHQJWK VTXDUH GHYLFHVf XVLQJ 90 9'6VDI 6DPH SDUDPHWHUV DV )LJ PAGE 44 ORJLFDO IODZV )RU LQVWDQFH WKH ILHOG LQ WKH [GLUHFWLRQ LV ]HUR E\ GHILQLWLRQ ZKLFK PHDQV WKDW XVLQJ WKH FKDQQHO SRWHQWLDO DW WKLV SRLQW WR FDOFXODWH WKH FXUUHQW IURP WKH ORQJFKDQQHO PRGHO ZLOO FOHDUO\ LQYDOLGDWH WKH JUDGXDO FKDQQHO DSSUR[LPDWLRQ ([ } ( PAGE 45 1H[W PHWKRGV WR H[WHQG WKH LQWR WZR VHFWLRQV WR FUHDWH WKH EHVW IXOOUDQJH PRGHO ,W ZDV GLVFRYHUHG WKDW QR PDWWHU ZKDW WKH GHSOHWLRQ UHJLRQ LV VWULFWO\ DQG REWDLQLQJ D DSSUR[LPDWLRQ UHTXLUHV UDWKHU VXEVWDQWLDO DVVXPSWLRQV 2QH PRGHO WKH ILHOGPDWFKLQJ DSSURDFK ZDV VHHQ WR JLYH UHDVRQDEO\ JRRG FKDUDFWHULVWLFV ZKLOH DOO WKH RWKHU DSSUR[LPDWLRQV 9'6VDW DQG VXUIDFHSRWHQWLDO VHOIVDWXUDWLRQf UHVXOWHG LQ GLVFRQWLQXLWLHV LQ WKH ILUVW DQG KLJKHUf GHULYDWLYHV $ VLPXODWLRQ ZDV XVHG WR YHULI\ WKDW WKHUH LV D SRLQW LQ WKH VDWXUDWHG FKDQQHO ZKHUH WKH [GLUHFWHGf ILHOG UHYHUVHV DQG WKH VXUIDFH EDQG EHQGLQJ LV WKXV ]HUR 7KLV SRLQW KDV EHHQ VXJJHVWHG PDQ\ WLPHV EHIRUH LQ RXU JURXS EXW QHYHU YHULILHG WZRGLPHQVLRQDOO\ $WWHPSWLQJ WR XVH WKLV SRLQW WR GHPDUN WKH ERXQGDU\ RI WKH VRXUFH UHJLRQ DQG GUDLQ UHJLRQ RI WKH WZRVHFWLRQ PRGHO UHVXOWV LQ WKH VDPH SRRU UHVXOWV DV WKH 9'6VDW PHWKRG PAGE 46 6L2R6L LQWHUIDFH ; cLPf )LJ (OHFWURVWDWLF SRWHQWLDOV LQWR WKH VXEVWUDWH DW GLIIHUHQW SRLQWV QHDU WKH GUDLQ HGJH RI D Âƒ [ SP FRUUHFWHG IRU VXEGLIIXVLRQf Q0267 ZLWK 3[[ [O FPf DW 9FV 9 DQG 9'6 9 7KH < SP QHDU VRXUFHf DQG < SP PLGGOH RI WKH FKDQQHOf FXUYHV DUH LQGLVWLQJXLVKDEOH 7KH EDQG LV IODW DW WKH 6L6L VXUIDFH ZKHQ WKH FKDQQHO HOHFWURVWDWLF SRWHQWLDO HTXDOV 9RV 9)% 9 9)% IRU WKLV GDWDf PAGE 47 &+$37(5 32/<6,/,&21*$7( 026 /2:)5(48(1&< &$3$&,7$1&(92/7 $*( &+$5$&7(5,67,&6 ,QWURGXFWLRQ )RU PRGHUQ 8/6, WHFKQRORJ\ SRO\VLOLFRQ JDWHV DUH XQLYHUVDOO\ XVHG RQ 026 GHYLFHV :LWK UHVSHFW WR 026 GHYLFH FKDUDFWHULVWLFV WKHUH LV QR DGYDQWDJH WR VXEVWLWXWLQJ PHWDO JDWHV ZLWK KHDYLO\GRSHG SRO\VLOLFRQ SRO\f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f DOLJQPHQW VWHS LQ WKH SURFHVV IORZ >@ 7KLV FKDSWHU FRYHUV WKH GHULYDWLRQ RI D )HUPL'LUDFEDVHG SRO\VLOLFRQJDWH 026 ORZIUHTXHQF\ FDSDFLWDQFHYROWDJH PRGHO 7KLV PRGHO ZLOO EH XVHG WR LOOXVWUDWH WKH HIIHFWV RI SRO\VLOLFRQ JDWHV RQ 026 ORZIUHTXHQF\ /)f FDSDFLWDQFHYROWDJH &9f FKDUDFWHULVWLFV FRPSDUHG WR PHWDOJDWH /)&9 FKDUDFWHULVWLFV $ XVHIXO DSSOLFDWLRQ IRU WKH PRGHO LV SK\VLFDO SDUDPHWHU H[WUDFWLRQ ZKLFK LV GHPRQVWUDWHG LQ WKLV FKDSWHU XVLQJ WZR PAGE 48 GLIIHUHQW PHWKRGRORJLHV SRLQW ILW DQG UHJLRQ ILW 6DPSOH SDUDPHWHU H[WUDFWLRQV IRU WKLFN Âƒf DQG WKLQ Âƒf JDWH R[LGHV DUH VKRZQ DQG GLVFXVVLRQ DERXW OLPLWDWLRQV RI WKH PRGHO DUH SUHVHQWHG 4XDQWXP HIIHFWV DUH SXUSRVHO\ LJQRUHG DQG WKH UHDVRQLQJ EHKLQG WKLV GHFLVLRQ LV GLVFXVVHG ,PSRUWDQW GHWDLOV UHODWHG WR IDVW FRQYHUJHQFH RI WKH SDUDPHWHUH[WUDFWLRQ URXWLQHV DUH DOVR JLYHQ 0HWDO*DWH &9 ,GHDO PHWDOJDWH &9 WKHRU\ XVLQJ %ROW]PDQQ VWDWLVWLFV KDV EHHQ H[WHQVLYHO\ GLVFXVVHG >@ DV ZHOO WKH H[WHQVLRQ WR LQFOXGH )HUPL'LUDF FDUULHU GLVWULEXWLRQ DQG GHLRQL]DWLRQ HIIHFWV >@ 7KH DSSHQGL[ FRQWDLQV WKH IXOO PHWDOJDWH /)&9 PRGHO GHULYDWLRQ WDNLQJ )HUPL'LUDF VWDWLVWLFV DQG GHLRQL]DWLRQ LQWR DFFRXQW 7KH UHOHYDQW VROXWLRQV DUH JLYHQ EHORZ )LJXUH VKRZV D VFKHPDWLF GLDJUDP RI DQ LGHDO PHWDOJDWH 026 GHYLFH DQG WKH FRUUHVSRQGLQJ EDQG GLDJUDP )URP )LJXUH Ef DV H[SODLQHG LQ WKH DSSHQGL[ .LUFKNRIIfV YROWDJH ODZ DURXQG WKH ORRS JLYHV pP 9 rV 9,; (F (M-T 9) 9* f ZKHUH W!0 LV WKH ZRUNIXQFWLRQ IRU WKH PHWDO 9 LV WKH SRWHQWLDO GURS DFURVV WKH R[LGH bV LV WKH HOHFWURQ DIILQLW\ RI WKH VXEVWUDWH (F DQG (_ DUH WKH FRQGXFWLRQEDQG HGJH DQG LQWULQVLF HQHUJLHV UHVSHFWLYHO\ LQ WKH VXEVWUDWH DQG 9) LV WKH )HUPL YROWDJH ZKLFK LV HTXLYDOHQW WR (c )SfT IRU SW\SH PDWHULDO ZKHUH )3 LV WKH TXDVL)HUPLOHYHO IRU KROHV DQG T LV WKH HOHFWURQ FKDUJH &ROOHFWLQJ WKHVH WHUPV LQ FOHDQHU IRUP JLYHV 9* 9 9,; p06 f PAGE 49 $f 9* *DWH 9$ 6XEVWUDWH %f )LJ L 026 FDSDFLWRU VFKHPDWLF DQG FRUUHVSRQGLQJ HQHUJ\EDQG GLDJUDP $f 6FKHPDWLF GLDJUDP RI D 026 FDSDFLWRU DQG %f FRUUHVSRQGLQJ HQHUJ\ EDQG GLDJUDP GHSLFWLQJ WKH SRWHQWLDO GURSV 6KRZQ LV D SRVLWLYH YROWDJH 9* DSSOLHG DW WKH JDWH UHVXOWLQJ LQ WKH 6L6L VXUIDFH HQWHULQJ LQYHUVLRQ PAGE 50 ZKHUH 06 0 pV p0 8V (F ([fT 9)f LV WKH ZRUN IXQFWLRQ GLIIHUHQFH EHWZHHQ WKH PHWDO DQG WKH VHPLFRQGXFWRU $V ZLOO EH VKRZQ LQ WKH QH[W VHFWLRQ WKH ZRUN IXQFWLRQ GLIIHUHQFH IRU D SRO\VLOLFRQJDWH 026 GHYLFH LV PXFK VLPSOHU WKDQ PHWDOJDWH 026 VLQFH WKH VXEVWUDWH DQG JDWH PDWHULDOV DUH WKH VDPH 7KH GURS DFURVV WKH R[LGH FDQ EH IRXQG IURP *DXVVfV /DZ UHTXLUHPHQWV DV 9J &J(MMA&J 4RW 4LWf &J f ZKHUH HV LV GLHOHFWULF FRQVWDQW RI WKH VHPLFRQGXFWRU a[[O2O)FP IRU 6Lf (O[ LV WKH ILHOG DFURVV WKH R[LGH & LV WKH R[LGH FDSDFLWDQFH DQG 47 DQG 4,7 UHSUHVHQW IL[HG DQG LQWHUIDFH WUDSSHG R[LGH FKDUJH UHVSHFWLYHO\ :LWK WKLV UHODWLRQ f FDQ EH UHZULWWHQ DV 9)% 9,; e6(,;& f ZKHUH 9SJ WKH IODWEDQG YROWDJH LV JLYHQ E\ 9)% p06 fÂ§ 4RW 4LWf & rf )RU PHWDOJDWHV WKHUH LV QR FDSDFLWLYH FRQWULEXWLRQ IURP WKH PHWDO VR WKH JDWH FDSDFLWDQFH LV VLPSO\ WKH VHULHV HTXLYDOHQW RI WKH IL[HG R[LGH FDSDFLWDQFH & DQG WKH YDULDEOH VXEVWUDWH FDSDFLWDQFH &M[ &J &L[&&L[ &Jf f 7KH ILHOG JRLQJ LQWR WKH VXEVWUDWH (,; DQG WKH VXEVWUDWH FDSDFLWDQFH &L[ DUH JLYHQ E\ PAGE 51 N7 QY>X f fÂ§ A 8\8Sf @ mD QF>A X,;XFfÂ§X)f fÂ§ A XFfÂ§X)f @ S[[ 8,; OQ J$H[S 8S 8$ 8,;f A J$H[S 8S 8$f 1[[8L[ OQ J'H[S 8' 8S 8,;f OO J'H[S8' 8Sf &LY aa1?M? a8,; 8Y8Sf 1FM)9M 8O[XF 8Sf J$H[S 8) 8$ 8,;f f / J'H[S8' 8S 8,;f ZKHUH DOO f8f YDOXHV DUH SRWHQWLDOV QRUPDOL]HG WR N7T DQG UHIHUHQFHG WR WKH LQWULQVLF )HUPL OHYHO )RU H[DPSOH 8S LV WKH QRUPDOL]HG )HUPL OHYHO T9)N7f 3[[ LV DFFHSWRU VXEVWUDWH GRSLQJ FRQFHQWUDWLRQ DQG 1[[ LV WKH GRQRU VXEVWUDWH GRSLQJ FRQFHQWUDWLRQ DQG J$ DQG J' DUH WKH FRUUHVSRQGLQJ GHJHQHUDF\ IDFWRUV IRU WKH WUDS OHYHOV 8$ DFFHSWRU HQHUJ\ OHYHOf DQG 8' GRQRU HQHUJ\ OHYHO QRW WR EH FRQIXVHG ZLWK WKH QRUPDOL]HG GUDLQ YROWDJH RI DQ 026 WUDQVLVWRUf 1Y LV WKH YDODQFH EDQG GHQVLW\ RI VWDWHV DQG 1F LV WKH FRQGXFWLRQ EDQG GHQVLW\ RI VWDWHV 7KRVH IDPLOLDU ZLWK 026 FDSDFLWDQFH HTXDWLRQV PLJKW ILQG WKHVH IDU PRUH FRPSOH[ WKDQ WKH\ UHFDOO D SHUXVDO RI WKH DSSHQGL[ VKRXOG FOHDU XS DQ\ TXHVWLRQV DERXW PAGE 52 WKLV IRUP +RZHYHU LW LV LQVWUXFWLYH WR VKRZ KRZ WKLV UHGXFHV WR D PRUH IDPLOLDU %ROW]PDQQ IRUP )LUVW DOO RI WKH )HUPL'LUDF LQWHJUDOV >)W_f DQG )MM&Qf WHUPV@ UHGXFH WR H[SRQHQWLDOV LQ WKH %ROW]PDQQ UDQJH RI DSSOLHG JDWH YROWDJHV 7_ f 6HFRQG WKHUH LV W\SLFDOO\ RQO\ RQH GRPLQDQW GRSDQW VR RQH RI WKH ODVW WZR WHUPV LQ f DQG f FDQ EH QHJOHFWHG WKH ILUVW FDQ EH QHJOHFWHG IRU QW\SH VXEVWUDWH DQG WKH VHFRQG IRU SW\SH VXEVWUDWHf )XUWKHUPRUH LI GHLRQL]DWLRQ LV QHJOHFWHG 8) 8$ 8,; IRU SW\SH RU 8' 8S 8,; f IRU QW\SHf WKHQ WKH ODVW WZR WHUPV RI f UHGXFH WR 3;;8,; 1;;8>; /LNHZLVH WKH WZR OLQHV RI f UHGXFH WR 3[[ 1[[ ZKHQ GHLRQL]DWLRQ LV QHJOHFWHG $V DQ H[DPSOH RI WKH VLPSOLILHG IRUP OHW XV FRQVLGHU D SW\SH VXEVWUDWH LQ VWURQJ DFFXPXODWLRQ ,Q WKLV FDVH LW FDQ EH DVVXPHG WKDW RQO\ WKH DFFXPXODWHG VXUIDFH FDUULHU WHUP LV GRPLQDQW 8,; LV ODUJH DQG QHJDWLYHf 1RWLQJ DOVR WKDW LQ WKH %ROW]PDQQ FDVH 8S 8Y OQ3[[1Yf f DQG f ZRXOG UHGXFH WR (,; L_ N73[[eVf H[S 8,;f &O[ TL_ WS[[eV N7f @ H[S 8,;f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f LGHDO GHYLFH FKDUDFWHULVWLFV )LJXUH VKRZV WKH EDQG GLDJUDP IRU DQ QSRO\VLOLFRQ JDWH 026 FDSDFLWRU ZLWK D SW\SH VXEVWUDWH D VFKHPDWLF RI WKH GHYLFH ZRXOG EH LGHQWLFDO WR Df ZLWK D PHWDO JDWH UHSODFHG E\ D SRO\VLOLFRQ JDWHf )URP WKLV ILJXUH LW LV FOHDU WKDW WKH SRWHQWLDO GURS DFURVV WKH GHYLFH FDQ EH JLYHQ VLPLODUO\ WR f DV fÂ§A)SRL\ (F fÂ§ (Lf r ALJ 9 ;V a 9; (F (Mf T 9) 9* f $VVXPLQJ WKH HQHUJ\ JDS KDV QRW QDUURZHG GXH WR WKH KLJKHU JDWH GRSLQJ WKH (F WHUPV DUH LGHQWLFDO DQG FDQFHO EHFDXVH WKH PDWHULDOV DUH ERWK VLOLFRQ 7KH HOHFWURQ DIILQLW\ LV WKH VDPH IRU ERWK WKH JDWH DQG VXEVWUDWH IRU WKH VDPH UHDVRQ 7KLV UHGXFHV f WR 9* Y 9,; 9,* 9) Y)SRO\ f 7KXV IRU WKH SRO\JDWH FDVH _!06 PRUH DSWO\ FDOOHG _!RV ZKHUH f*f UHSUHVHQWV WKH JDWH EXW VWLOO WUDGLWLRQDOO\ UHIHUUHG WR DV f0f IRU PHWDOf LV VLPSO\ JLYHQ E\ rPV 9) 9)BSRO\ f )RU )LJXUH _!06 LV JLYHQ E\ OQ3[[1**QIf ZKHUH 3[[ LV WKH VXEVWUDWH GRSLQJ f3f LPSO\LQJ SW\SHf DQG 1** LV WKH JDWH GRSLQJ f1f LPSO\LQJ QW\SHf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fV ODZ ZLWK 9,* Y* Y)% 9,; 9,* HV(,;& f ZKHUH 9)% IURP f VWLOO KROGV DVVXPLQJ QHJOLJLEOH FRQWULEXWLRQ IURP WKH SRO\VLOLFRQR[LGH LQWHUIDFH XVLQJ f IRU _!06 )LQDOO\ WKH JDWH FDSDFLWDQFH IRUPXOD QHHGV WR EH H[WHQGHG IRU WKUHH FDSDFLWRUV LQ SDUDOOHO 7KLV FKDQJHV f WR &J &L[&&LJ&L[& &L[&LJ &LJ&f f ZKHUH &MJ WKH FDSDFLWDQFH IURP WKH SRO\VLOLFRQ JDWH LV JLYHQ E\ 1Ye8,*8Y8)f 1F. 8LJ8F8If f 7KLV LV VLPSO\ f UHZULWWHQ ZLWK WKH EDQG QRWDWLRQ IRU WKH JDWH 7KXV 8,* LV WKH QRUPDOL]HG VXUIDFH SRWHQWLDO LQ WKH JDWH (,& LV WKH ILHOG LQ WKH JDWH GHILQHG EHORZf DQG 3**! QJJ 8Y 8F 1F 1Y J$ J' 8' 8$ DUH SUHFLVHO\ DV GHILQHG EHIRUH H[FHSW WKDW WKH\ DSSO\ QRZ WR WKH JDWH UDWKHU WKDQ WKH VXEVWUDWH 8) DERYH ZDV FDOOHG 8) SRO\ 3** J$H[S8) 8$ 8,*f A** J*k[S 8% 8) 8*f PAGE 56 HOVHZKHUHfÂ§LW LV OHIW DV 8) LQ f WR PDLQWDLQ WKH V\PPHWU\ RI WKH HTXDWLRQ 7KH JDWH ILHOG LV JLYHQ E\ N7 1Y>8LJ8Y8If 8Y8)f @ 1FWA6A 8,*8& 8)f A L 8F 8)f @ A** A,; AQ J$H[S8) 8$ 8,*f / J$H[S8) 8$f 1TJ fÂ§8M[ ,Q J'H[S8' 8) 8,*f J'H[S8' 8)f ZKLFK LV LGHQWLFDO WR f ZLWK WKH VXUIDFH SRWHQWLDOV FKDQJHG $JDLQ WKH VDPH FDYHDW DSSOLHV WR ffÂ§DOO WKH WHUPV UHIHU WR WKH JDWH QRZ QRW WKH VXEVWUDWH 7KLQJV OLNH WUDS OHYHOV DQG EDQG HGJHV DUH QHDUO\ LI QRW H[DFWO\ WKH VDPH LQ WKH VXEVWUDWH DQG SRO\VLOLFRQ JDWH +RZHYHU 8) LV FOHDUO\ TXLWH GLIIHUHQW DVVXPLQJ WKH JDWH DQG VXEVWUDWH DUH QRW GRSHG LGHQWLFDOO\ ZKLFK ZRXOG PDNH D SRRU FDSDFLWRU RU WUDQVLVWRUf $Q DGGLWLRQDO HTXDWLRQ ZKLFK ZDV QRW QHHGHG LQ WKH PHWDOJDWH FDVH LV UHTXLUHG WR UHODWH WKH JDWH DQG VXEVWUDWH 7KLV HTXDWLRQ HTXDWHV WKH FKDUJH GHQVLW\ DW WKH JDWHR[LGH LQWHUIDFH ZLWK WKH FKDUJH GHQVLW\ DW WKH R[LGHVXEVWUDWH LQWHUIDFH eVAL[ 4LW[ eVALJ 4LWJ Â‘ 4LW[ LV WKH LQWHUIDFH FKDUJH DW WKH VXEVWUDWHLQVXODWRU LQWHUIDFH DQG 4,7* LV WKH FKDUJH DW WKH JDWHLQVXODWRU LQWHUIDFH ,W LV DVVXPHG WKDW WKHVH YDOXHV DUH QHJOLJLEOH DQG WKDW WKH GLHOHFWULF FRQVWDQW IRU WKH VLOLFRQ VXEVWUDWH DQG WKH VLOLFRQ JDWH DUH LGHQWLFDO DOUHDG\ LPSOLFLWO\ DVVXPHG LQ WKH HTXDWLRQf 7KLV JLYHV WKH IROORZLQJ PAGE 57 (,; (,*n ZKLFK DOORZV WKH VXUIDFH SRWHQWLDO LQ WKH JDWH WR EH UHODWHG WR WKH VXUIDFH SRWHQWLDO LQ WKH VXEVWUDWH 7KH LWHUDWLYH VROXWLRQ RI WKH DERYH HTXDWLRQ UHTXLUHV PDQ\ FDOFXODWLRQV RI f DQG f DQG LV WKH PRVW WLPHFRQVXPLQJ SDUW RI WKH /)&9 VROYHU DV ZHOO DV DQ\ VRIWZDUH XVLQJ WKH URXWLQH VXFK DV D SDUDPHWHU H[WUDFWRU ZKLFK ZRUNV E\ FRPSDULQJ WKH GDWD WR WKH WKHRUHWLFDO FXUYH DV GLVFXVVHG ODWHU LQ WKH DSSOLFDWLRQ VHFWLRQf 3ROYVLOLFRQ*DWH (IIHFWV 7KH HIIHFW RI SRO\VLOLFRQ JDWHV FRPSDUHG WR PHWDO JDWHV LV D UHGXFWLRQ RI WKH JDWH FDSDFLWDQFH &J ZKHQ WKH JDWH LV LQ GHSOHWLRQ 7KLV LV DULVHV ZKHQ WKH YDOXH RI &cJ IDOOV EHORZ WKDW RI & DQG &L[ ZKLFK RQO\ RFFXUV GXULQJ JDWH GHSOHWLRQ DQG VXEVWUDWH LQYHUVLRQ RU DFFXPXODWLRQ DQG RQO\ WKHQ WR D VLJQLILFDQW GHJUHH IRU WKLQ R[LGHV 7KLV LV HDVLO\ YLVXDOL]HG IURP WKH WKUHH VHULHV FDSDFLWDQFHVfÂ§WKH RQH ZKLFK GRPLQDWHV LV WKH VPDOOHVW DQG WKH FDSDFLWDQFH GXH WR WKH VXEVWUDWH DQG JDWH DUH ERWK PLQLPL]HG GXULQJ GHSOHWLRQ DQG PD[LPL]HG GXULQJ DFFXPXODWLRQ DV ZHOO DV LQYHUVLRQ IRU WKH /) FDVHf $V R[LGHV WKLQ WKH R[LGH FDSDFLWDQFH LQFUHDVHV ZKLFK FDXVHV WKH HIIHFW RI WKH VXEVWUDWH DQG JDWH GHSOHWLRQ WR KDYH PRUH FRQWURO RYHU WKH FKDUDFWHULVWLFV RI WKH &J9* FXUYH )LJXUH VKRZV WKH GLIIHUHQFH EHWZHHQ PHWDOJDWH DQG SRO\VLOLFRQJDWH GDWD QRUPDOL]HG WR & IRU WZR GLIIHUHQW WHFKQRORJLHV 7KH fKLJKHUf SDLU RI FXUYHV IRU D $ R[LGH WKLFN R[LGH PHDQV ORZ R[LGH FDSDFLWDQFHf VKRZV OLWWOH GLIIHUHQFH EHWZHHQ SRO\VLOLFRQ JDWHV DQG PHWDO JDWHV 7KH ORZHU SDLU RI FXUYHV IRU D Âƒ R[LGH WKLQ R[LGH PHDQV ODUJH R[LGH FDSDFLWDQFHf VKRZV D ODUJH GHFUHDVH LQ &J IRU DOO YDOXHV RI 9* SDUWLFXODUO\ IRU 9* 9 ZKHUH WKH JDWH LV VWLOO LQ GHSOHWLRQ DQG WKH VXEVWUDWH LV LQYHUWHG PAGE 58 94 9f )LJ &RPSDULVRQ RI PHWDOJDWH DQG Q SRO\JDWH 026& FXUYHV IRU WZR GLIIHUHQW WHFKQRORJLHV 2QH VHW KDV Âƒ R[LGH ZLWK 3[[ [O FP DQG WKH VHFRQG VHW KDV Âƒ R[LGH ZLWK 3[[ [O FP ,Q HDFK FDVH WKH 9)% LV DGMXVWHG WR EH 9 DQG WKH JDWH GRSLQJ LV [O FP &OHDUO\ VKRZQ LV WKH GUDPDWLF GLIIHUHQFH EHWZHHQ SRO\JDWH GRWWHG OLQHf DQG PHWDOJDWH VROLG OLQHf IRU WKH Âƒ FDVH DQG WKH QHJOLJLEOH LPSDFW RQ WKH $ FDVHfÂ§WKH SRO\VLOLFRQ JDWH HIIHFWV LQFUHDVH DV WKH R[LGH VFDOHV WKLQQHU PAGE 59 7KLV FRQWLQXDO GHFUHDVH LQ &J IRU LQFUHDVLQJ 9* LQ WKLV Q SRO\JDWH RQ S6L VXEVWUDWHf LV RIWHQ UHIHUUHG WR DV fSRO\ GHSOHWLRQf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f >@ 7KHUPDO JHQHUDWLRQ WKHQ FRXOG QRW VXSSO\ WKH KROHV IDVW HQRXJK IRU SUDFWLFDO XVH RI DQ XQGRSHG JDWH +RZHYHU LW PLJKW EH SRVVLEOH WR GHVLJQ LQ D PLQRULW\ FDUULHU VRXUFH QHDUE\ WR VXSSO\ PLQRULW\ FDUULHUV VLPLODU WR KRZ WKH VRXUFH DQG GUDLQ VXSSO\ PLQRULW\ FDUULHUV LQ WKH VXEVWUDWHf 7KH UHGXFWLRQ LQ WKH JDWH FDSDFLWDQFH GXH WR SRO\ GHSOHWLRQ FDXVHV D UHGXFWLRQ LQ WKH GULYH FXUUHQW ZKLFK GHJUDGHV FLUFXLW SHUIRUPDQFH >@ VLQFH WKH DPRXQW RI FXUUHQW VXSSOLHG E\ WKH WUDQVLVWRU GLUHFWO\ UHODWHV WR WKH VZLWFKLQJ VSHHG RI WKH GHYLFH ,Q D FRPSOHPHQWDU\026 &026f FLUFXLW WKH FXUUHQW FKDUJHG XS WKH LQWHUFRQQHFW DQG LQWULQVLF FDSDFLWDQFHV RI WKH QH[W WUDQVLVWRUV LQ WKH OLQH DV GLVFXVVHG LQ GHWDLO LQ &KDSWHU PAGE 60 %HFDXVH RI WKLV SRO\JDWH ,' UHGXFWLRQ WKHUH PD\ HYHQWXDOO\ EH D PRYH EDFN WR PHWDO JDWH RU VLOLFLGHVf RQFH WKH SURFHVVLQJ LVVXHV RI JDWH DOLJQPHQW DUH VROYHG ,W LV LQVWUXFWLYH WR ORRN DW WKH LQGLYLGXDO FDSDFLWDQFH FRPSRQHQWV WR VHH KRZ WKH fFRPSOH[f SRO\ /)&9 FXUYH IRUPV )LJXUH VKRZV VXFK D FXUYH IRU D WKHRUHWLFDO Âƒ R[LGH ZLWK DQ Q JDWH GRSHG UDWKHU ORZO\f WR [O FP DQG D VXEVWUDWH GRSHG WR [O FP 7KH JDWH DUHD LV O[O2 FP DQG WKH IODW EDQG YROWDJH LV fÂ§ 9 7KH &J FXUYH EHLQJ WKH VHULDO VXP RI & &L[ DQG &LJ (T f LV DOZD\V ORZHU WKDQ WKH FRPSRQHQW FXUYHV ,W FDQ EH FOHDUO\ VHHQ KRZ HDFK RI WKHVH WKUHH FRPSRQHQWV LQIOXHQFHV WKH RYHUDOO VWUXFWXUH RI WKH UHVXOWLQJ JDWH FDSDFLWDQFH ,Q IDFW WKLV fUHJLRQDOf HIIHFW ZLOO EH XVHG WR KHOS VSHHG XS SDUDPHWHU H[WUDFWLRQ LQ WKH QH[W VHFWLRQ $OVR RI LQWHUHVW LV D EUHDNGRZQ RI SRWHQWLDOV DFURVV WKH 026& GHYLFH DV D IXQFWLRQ RI 9 )LJXUH VKRZV WKH IRXU FRPSRQHQWV RI 9* QDPHO\ 9,; 9,* 9R[ DQG 9SJ VHH (T f DV D IXQFWLRQ RI 9* XVLQJ WKH VDPH SDUDPHWHUV DV WKH H[DPSOH LQ WKH ODVW SDUDJUDSK 7R VKRZ VKRZ WKHVH DUH UHODWHG WR WKH UHVXOWLQJ JDWH FDSDFLWDQFH WKH &J9* FXUYH LV DOVR SORWWHG :KDW LV PRVW UHOHYDQW LQ WKLV ILJXUH LV WKDW DV WKH SULPDU\ fGLSf LQ WKH &9 FXUYH RFFXUV DV WKH VXUIDFH SRWHQWLDO LQ WKH VXEVWUDWH 9O[ VZHHSV IURP DFFXPXODWLRQ WR LQYHUVLRQ LH PRYHV IURP D VPDOO QHJDWLYH QXPEHU WR DERXW RQH YROWf DQG HQGV VKDUSO\ DV WKH VXUIDFH SRWHQWLDO DSSURDFKHV LWV PD[LPXP VWURQJ LQYHUVLRQf 6LPLODUO\ WKH VHFRQGDU\ SRO\GHSOHWLRQ fGLSf RFFXUV DV WKH JDWH VXUIDFH SRWHQWLDO 9* PRYHV IURP DFFXPXODWLRQ WR LQYHUVLRQ DJDLQ PRYHV IURP D VPDOO QHJDWLYH YROWDJH WR DURXQG D YROWf 1RWH WKDW WKH ILQDO VXUIDFH SRWHQWLDO LQ WKH JDWH LV KLJKHU WKDQ WKDW LQ WKH VXEVWUDWH 9,* 9,; ZKHQ 9* 9f 7KLV DJUHHV ZLWK WKH FRPPRQ DSSUR[LPDWLRQ WKDW PAGE 61 94 9f )LJ ,QGLYLGXDO FDSDFLWDQFH YDOXHV IRU D WKHRUHWLFDO [ _OP Q026& ZLWK D Âƒ JDWH R[LGH 3[[ [O FP 1** [O FP 7 DQG 9)% 9 7KLV ILJXUH GHPRQVWUDWHV KRZ WKH WKUHH SDUDOOHO FDSDFLWDQFHV &L[ &MJ DQG &f DGG WR JLYH WKH RYHUDOO JDWH FDSDFLWDQFH 6HH )LJ IRU WKH FRUUHVSRQGLQJ SRWHQWLDO EUHDNGRZQ PAGE 62 9* 9f )LJ ,QGLYLGXDO SRWHQWLDO EUHDNGRZQ IRU D WKHRUHWLFDO [ SP Q026& ZLWK D Âƒ JDWH R[LGH 3[[ [O FPn 12* [O FP 7 DQG 9)% 9 DORQJ ZLWK WKH FRUUHVSRQGLQJ /)&9 FXUYH 1RWH KRZ WKH VXUIDFH SRWHQWLDO LQ WKH VXEVWUDWH 9; LQFUHDVHV UDSLGO\ LQ WKH UDQJH 9* WR 9 DV &J LQFUHDVHV VXEVWUDWH LQYHUVLRQf DQG WKH VLPLODU LQFUHDVH LQ 9,* LQ WKH UDQJH 9* WR 9 JDWH LQYHUVLRQf 6HH )LJ IRU WKH FRUUHVSRQGLQJ FDSDFLWDQFH EUHDNGRZQ GG PAGE 63 WKH VXUIDFH SRWHQWLDO SLQV WR D OLWWOH RYHU 9) VLQFH WKH )HUPL YROWDJH LQ WKH JDWH ZLOO EH ODUJHU WKDQ WKDW RI WKH VXEVWUDWH GXH WR WKH JUHDWHU JDWH GRSLQJ 3DUDPHWHU ([WUDFWLRQ 8VLQJ WKH /)&9 0RGHO 2I WKH PXOWLWXGH RI YDULDEOHV LQ WKH /)&9 HTXDWLRQV PRVW RI WKHP DUH NQRZQ WR D UHDVRQDEOH GHJUHH RI DFFXUDF\ VXFK DV WKH GLHOHFWULF FRQVWDQW HQHUJ\ JDS FRQGXFWLRQ EDQG GHQVLW\ HWFf FDQ EH PHDVXUHG HDVLO\ WHPSHUDWXUHf RU QHHG QRW EH NQRZQ YHU\ DFFXUDWHO\ DFFHSWRU DQG GRQRU WUDS OHYHOf GXH WR WKHLU VPDOO HIIHFW 7KLV OHDYHV WKH JDWH DQG VXEVWUDWH GRSLQJ WKH R[LGH WKLFNQHVV DQG WKH IODWEDQG YROWDJH DV WKH fXQNQRZQf SDUDPHWHUV 7KHVH SDUDPHWHUV PD\ EH H[WUDFWHG IURP H[SHULPHQWDO GDWD E\ FRPSDULQJ H[SHULPHQWDO GDWD WR WKH WKHRUHWLFDO PRGHO SUHVHQWHG LQ WKLV FKDSWHU 7KLV PD\ DSSHDU WR EH DQ HDV\ WDVN VLQFH WKH HTXDWLRQ QHHG RQO\ EH XVHG DORQJ ZLWK VRPH GDWD LQ FRQMXQFWLRQ ZLWK D QRQOLQHDU OHDVWVTXDUHVVROYHU +RZHYHU RQH ZLOO QRWH WKDW WKH SRO\VLOLFRQ JDWH /)&9 IRUPXOD LV GRXEO\ SDUDPHWULF WKDW LV LV UHODWHG WKURXJK WZR SDUDPHWHUVfÂ§WKH VXUIDFH SRWHQWLDOV 8,; DQG 8,*f QHLWKHU RI ZKLFK DUH NQRZQ IURP WKH GDWD 7KXV VROYLQJ WKLV SUREOHP LV QRQ WULYLDO 7KH ILUVW VWHS WRZDUG D VROXWLRQ WKHQ LV WR ZULWH D SURJUDP ZKLFK ZLOO FDOFXODWH &J JLYHQ 9* 7KLV UHTXLUHV LQWHQVLYH FDOFXODWLRQV WR ILQG 8_; DQG 8_* IRU HDFK 9* EXW FDQ EH GRQH VLQFH WKHUH LV RQO\ RQH XQLTXH VROXWLRQ 7KXV ZLWK D &J9*f URXWLQH ZULWWHQ D QRQOLQHDU OHDVWVTXDUHVILW SURJUDP FDQ EH XVHG 7KH FRGH ZULWWHQ IRU WKLV GLVVHUWDWLRQ WRRN DGYDQWDJH RI WKH IDFW WKDW DV WKH VROXWLRQ FRQYHUJHV WR YDOXHV RI WKH XQNQRZQ SDUDPHWHUV WKH YDOXHV RI WKH VXUIDFH SRWHQWLDOV DW HDFK H[SHULPHQWDO GDWD SRLQW FRXOG EH PAGE 64 XVHG IRU LQLWLDO JXHVVHV IRU HDFK VXEVHTXHQW LWHUDWLRQ RI 9* WR ILQG HDFK &J VLQFH WKH SDUDPHWHUV ^VXEVWUDWH DQG JDWH GRSLQJ R[LGH WKLFNQHVV DQG IODWEDQG YROWDJHff VKRXOG QRW EH FKDQJLQJ WRR UDSLGO\f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fÂ§WKH RWKHU WZR SRLQWV FDQ EH DQ\ZKHUH ZLWKLQ WKHLU UHJLRQ 7KH &JDFF SRLQW LV D SRLQW IURP WKH /)&9 JDWH DFFXPXODWLRQ UHJLRQ )URP WKLV D JRRG HVWLPDWH RI WKH R[LGH WKLFNQHVV FDQ EH IRXQG VLQFH WKH RWKHU SDUDPHWHUV KDYH YHU\ OLWWOH LQIOXHQFH RYHU WKLV SRLQW VHH )LJV DQG f &JDFF DV\PSWRWLFDOO\ DSSURDFKHV & ZKLFK LV LQYHUVHO\ SURSRUWLRQDO WR WKH R[LGH WKLFNQHVV 7R[ YLD WKH SDUDOOHO SODWH IRUPXOD 7KHUH KDV EHHQ PXFK UHVHDUFK LQ REWDLQLQJ 7R[ DQGRU & IURP VXEVWUDWHf DFFXPXODWLRQ &9 GDWD >@ PAGE 65 )LJ ([DPSOH RI D JHQHUDO SRO\VLOLFRQJDWH QL JDWH S VXEVWUDWH IRU WKLV FDVHf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f ZKLFK LV GLUHFWO\ UHODWHG WR WKH WKH LQYHUVH )HUPL'LUDF LQWHJUDO QDWXUDO ORJDULWKP LI DVVXPLQJ D %ROW]PDQQ GLVWULEXWLRQf RI WKH VXEVWUDWH GRSLQJ 7KH SRVLWLRQ RI WKH PLQLPXP DORQJ WKH 9* D[LV DOVR DOORZV XV WR HVWLPDWH WKH IODWEDQG YROWDJH E\ FRPSDULQJ WKH 9* RI WKH PLQLPXP RI WKH WKHRUHWLFDO FXUYH WR WKH 9* RI WKH GDWD 7KH &JGHS SRLQW LV IURP WKH JDWH GHSOHWLRQ UHJLRQ )LJXUH VKRZV WKDW WKH JDWH GRSLQJ KDV WKH PRVW DIIHFW RQ WKLV SDUW RI WKH FXUYH ZKHUHDV )LJXUH VKRZV WKDW WKH VXEVWUDWH GRSLQJ KDV YHU\ OLWWOH HIIHFW LQ WKLV UHJLRQ )RU WKH Q JDWH RQ SVXEVWUDWH H[DPSOH LQ )LJXUH WKH VXEVWUDWH LV LQ LQYHUVLRQ +RZHYHU HYHQ LI WKH VXEVWUDWH ZHUH QW\SH DQG WKH VXEVWUDWH WKXV DFFXPXODWHGf &J GHSOHWLRQ ZRXOG VWLOO RFFXU EHFDXVH WKH JDWH ZRXOG VWLOO EH LQ GHSOHWLRQ RI FRXUVH WKH HQWLUH FXUYH ZRXOG EH VKLIWHG GXH WR WKH IODWEDQG GLIIHUHQFHf +HQFH WKLV SRLQW LV FDOOHG &JBGHS ZLWK WKH fGHSf LQ UHIHUHQFH WR WKH GHSOHWHG VWDWH RI WKH JDWH %\ YDU\LQJ WKH SDUDPHWHUV LQ WKH DSSURSULDWH UHJLRQV WR PDWFK WKHVH WKUHH SRLQWV D XQLTXH SDUDPHWHU VHW ZLOO EH REWDLQHG ZKLFK ZLOO GHVFULEH D WKHRUHWLFDO /)&9 FXUYH SDVVLQJ WKURXJK WKH WKUHH SRLQWV PAGE 67 )LJ (IIHFW RI VXEVWUDWH GRSLQJ FKDQJHV 3[[f RQ /)&9 FKDUDFWHULVWLFV 7KH fGHSOHWLRQf UHJLRQ VHH )LJ f LV WKH UHJLRQ RI ODUJHVW LPSDFW PAGE 68 )LJ (IIHFW RI JDWH GRSLQJ FKDQJHV 1*f RQ /)&9 FKDUDFWHULVWLFV 7KH fGHSOHWLRQf UHJLRQ VHH )LJ f LV WKH UHJLRQ RI ODUJHVW LPSDFW PAGE 69 5HJLRQ ([WUDFWLRQ 0HWKRGRORJ\ $V JRRG DV RXU PRGHO LV WKHUH DUH VWLOO VHYHUDO HIIHFWV ZKLFK DUH QRW EHLQJ FRQVLGHUHG 7KHVH LQFOXGH UHWURJUDGH GRSLQJ LQ WKH VXEVWUDWH DQG TXDQWXP HIIHFWV LQ WKH VXEVWUDWH LQYHUVLRQ FKDQQHO 5HWURJUDGH GRSLQJ LV FRPPRQO\ XVHG IRU VXEnPLFURQ GHVLJQ WR PDLQWDLQ D KLJK VXEVXUIDFH GRSLQJ FRQFHQWUDWLRQ WR SUHYHQW SXQFKWKURXJK ZKLOH VWLOO PDLQWDLQLQJ D ORZ 97 IRU ORZ9 RSHUDWLRQ WR DFFRPPRGDWH WKH WKLQ R[LGHVf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f PAGE 70 )LJ 6DPSOH RI UHWURJUDGH GRSLQJ SURILOH VKRZLQJ ORZ VXUIDFH FRQFHQWUDWLRQ [O FPf DQG KLJKHU EXON FRQFHQWUDWLRQ O[O2 FPf ?D966eN n5.U PAGE 71 ZKLFK LV PRVW LPSRUWDQW FRPSDUHG WR WKH RSWLFDO WKLFNQHVV ZKLFK LV QRW ZKDW DIIHFWV GHYLFH SHUIRUPDQFHf 'XH WR WKHVH WZR PDLQ QRQLGHDOLWLHV QRQFRQVWDQW GRSLQJ DQG TXDQWXP HIIHFWVf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f GRPLQDQW LQ WKH VSHFLILF UHJLRQ WKXV JUHDWO\ LPSURYLQJ WKH FRQYHUJHQFH UDWH VLQFH WKH GDWD EHLQJ XVHG LV PRVW UHOHYDQWf 7KLV DGGV FRPSOLFDWLRQ WR WKH FRGLQJ DV WKH GDWD SDUWLWLRQLQJ LQWR HDFK UHJLRQ GLVFXVVHG ODWHUf PXVW EH DXWRPDWHG DQG D GLIIHUHQW ILWWLQJ URXWLQHV PXVW EH FUHDWHG IRU HDFK RI WKH WKUHH UHJLRQV VDPH PRGHO EXW VHSDUDWH SDUWLDO GHULYDWLYH FDOFXODWLRQVf 0HWKRGRORJ\ &RPSDULVRQ )LJXUH VKRZV D FRPSDULVRQ RI WKH ILW XVLQJ WKH SRLQW DQG UHJLRQ PHWKRGV WR H[SHULPHQWDO GDWD IRU D Âƒ R[LGH IURP DQ LQGXVWULDO [ _[P 0267 WUDQVLVWRU PAGE 72 9* 9f )LJ 7KHRUHWLFDOO\ JHQHUDWHG FXUYHV FRPSDUHG WR RULJLQDO /)&9 GDWD XVLQJ WKH $f WKUHHSRLQW DQG %f IXOOFXUYH H[WUDFWLRQV WR RQ DQ QSRO\VLOLFRQ JDWH [ IWP Q0267 ([WUDFWHG SDUDPHWHUV ZHUH $f ;R[ Âƒ 3[[ [O FP 1JJ [ FPn DQG 9)% 9 %f ;R[ Âƒ 3[[ [O FPn 1** [O FP DQG 9A ,9 PAGE 73 7KH VROLGOLQH RI FRXUVH LV WKH WKHRUHWLFDO FXUYH XVLQJ WKH H[WUDFWHG SDUDPHWHUV DQG WKH f[f PDUNV DUH WKH GDWD XVHG IRU WKH H[WUDFWLRQ 7KH WRS $f FXUYH XVHV RQO\ WKH WKUHH SRLQWV PDUNHG WR ILW WKH GDWD ZKLOH WKH ORZHU %f FXUYH XVHV WKH IXOO VHW RI GDWD 7KH 506 HUURU IRU WKH VHFRQG FXUYHf FDOFXODWHG IURP WKH VTXDUH URRW RI WKH VXP RI WKH VTXDUHV RI WKH GLIIHUHQFH EHWZHHQ WKHRUHWLFDO DQG H[SHULPHQWDO FDSDFLWDQFH GLYLGHG E\ WKH VTXDUH URRW RI Q fÂ§ GHJUHHV RI IUHHGRP QXPEHU RI ILWWLQJ SDUDPHWHUVf ZDV b DQ H[FHOOHQW ILW IRU RQO\ IRXU SDUDPHWHUV 7KHUH LV QRW PXFK LQ OLWHUDWXUH WR FRPSDUH WKH fJRRGQHVVf RI WKHVH UHVXOWV DV WKHUH DUH QRW PDQ\ FDSDFLWDQFH PRGHOV DYDLODEOH PRVW FRPSDFW PRGHOV VXFK DV %6,0 >@ ILW ,9 FKDUDFWHULVWLFV RQO\ DQG GR QRW FRQVLGHU FDSDFLWDQFH H[WUDFWLRQf $JJUHVVLYHO\ VFDOHG 026& GHYLFH GDWD LV JLYHQ LQ )LJXUH 7KLV VKRZV WZR /)&9 FXUYHV IURP D Âƒ R[LGH IURP DQ LQGXVWULDO 0267 WUDQVLVWRU ZKHUH WKH Âƒ ZDV GHWHUPLQHG IURP VRPH RSWLFDO PHWKRG PRVW OLNHO\ HOOLSVRPHWU\ 7KLV LV D UDWKHU FRPSOH[ ILJXUH DQG UHTXLUHV VRPH H[SODQDWLRQ 7KHUH DUH WZR H[SHULPHQWDO FXUYHV RQH S SRO\n JDWH DQG RQH Q SRO\JDWH ERWK RQ D SZHOO 7KXV LQ WKH 9 9 UHJLRQ WKH Q JDWH LV LQ GHSOHWLRQ ZKLOH WKH SJDWH LV LQ DFFXPXODWLRQ FOHDUO\ WKH FXUYHV ZHUH VKLIWHG WR DOLJQ WKHP DV WKH IODWEDQG VKRXOG GLIIHU E\ DERXW D YROW EHWZHHQ WKH WZR FXUYHV DOWKRXJK D WKUHVKROG DGMXVWPHQW LPSODQW ZRXOG RIIVHW WKLV VRPHZKDWf )URP ORRNLQJ DW WKH GDWD DW 9 DQG DVVXPLQJ &J & WKH\ FRQFOXGH WKDW WKH HIIHFWLYH R[LGH WKLFNQHVV LV Âƒ IRU WKH GHSOHWHG Q JDWH DQG $ IRU WKH DFFXPXODWHG S JDWH GHYLFH 7KLV LV D SRRU DSSUR[LPDWLRQ VLQFH WKH R[LGH WKLFNQHVV YDOXH ZRXOG YDU\ JUHDWO\ DW GLIIHUHQW SRLQWV DORQJ WKH GHSOHWHG FXUYH +RZHYHU WKH\ FRUUHFWO\ VWDWH WKDW WKH GLIIHUHQFH EHWZHHQ WKH PAGE 74 WZR LV GXH WR SRO\ GHSOHWLRQ ZKLFK LV D UHDVRQDEOH VWDWHPHQW ZKHQ DSSOLHG WR WKDW VSHFLILF JDWH YROWDJH RQO\ 7KXV WKH\ DWWULEXWH D Âƒ UHGXFWLRQ LQ HIIHFWLYH R[LGH WKLFNQHVV GXH WR SRO\VLOLFRQ GHSOHWLRQ DW 9 ZKLFK LV WKH RSHUDWLQJ YROWDJH IRU WKDW WHFKQRORJ\ 7KH\ QH[W DVVHUW WKDW WKH GLIIHUHQFH EHWZHHQ WKH DFFXPXODWHGJDWH FXUYH Âƒ EDVHG RQ DVVXPLQJ &J &f DQG WKH RSWLFDOO\ PHDVXUHG R[LGH Âƒf PXVW EH HQWLUHO\ GXH WR TXDQWXP HIIHFWV 7KLV FRQFOXVLRQ LV ZURQJ DV WKH TXDQWXP PRGHO PRVW OLNHO\ GRHV QRW WDNH DOO WKH HIIHFWV PHQWLRQHG SUHYLRXVO\ LQWR DFFRXQW VXFK DV WKHUPDO EURDGHQLQJ LQWHUIDFH URXJKQHVV DQG WUDQVLWLRQDO UHJLRQ QRW WR PHQWLRQ UHWURJUDGH GRSLQJf 0RUH LPSRUWDQWO\ WKH\ DUH FRPSOHWHO\ LJQRULQJ WKH VHYHUH HUURU RI XVLQJ &J & )LJXUH VKRZV WKH ILW XVLQJ WKH WKUHHSRLQW PHWKRGf WR WKH GDWD LQ )LJXUH )URP WKLV ILW WKH H[WUDFWHG R[LGH WKLFNQHVV LV Âƒ &RPSDUHG WR WKH LQGXVWU\VWDWHG UHVXOWV WKH VDPH HIIHFWLYH R[LGH WKLFNQHVV UHGXFWLRQ GXH WR SRO\ Âƒ KHUH YHUVXV Âƒf LV VHHQ +RZHYHU E\ FRUUHFWO\ DFFRXQWLQJ IRU WKH GLVWULEXWLRQ LQVWHDG RI DVVXPLQJ &J &f DQ DGGLWLRQDO Âƒ UHGXFWLRQ IURP WKH )HUPL'LUDF GLVWULEXWLRQ LV DOVR IRXQG WKDW LV IURP WKH IDFW WKDW &J &f 7KLV OHDYHV Âƒ RI GLIIHUHQFH EHWZHHQ WKH H[WUDFWHG R[LGH WKLFNQHVV RI Âƒ DQG WKH RSWLFDOO\ PHDVXUHG WKLFNQHVV RI Âƒ 7KLV Âƒ GLIIHUHQFH PD\ LQFOXGH VRPH TXDQWXP HIIHFWV EXW LW PD\ DOVR EH GXH WR RSWLFDO HUURUV VXFK DV QRW DFFRXQWLQJ IRU WKH WUDQVLWLRQDO OD\HU SURSHUO\ >@ DQGRU VRPH RWKHU HIIHFWV LH GRSLQJ SURILOHf (OOLSVRPHWU\ DQG RWKHU RSWLFDO PHWKRGV FDQ QRW EH XVHG RQ WKH DFWXDO GHYLFH VLQFH WKH JDWH HOHFWURGH LV QRW WUDQVSDUHQWf VR LW GRHV QRW PHDVXUH WKH R[LGH WKLFNQHVV LQ WKH DFWLYH SDUW RI WKH GHYLFH ZKLFK PD\ GLIIHU VOLJKWO\ GXH WR WKH DGGLWLRQDO PAGE 75 & FDOFXODWHG IURP RSWLFDO ;R $ )LJ 3SRO\SZHOO DQG QSRO\SZHOO fÂƒf LQGXVWULDO GDWD 'DWD LV VKLIWHG WR DOLJQ PLQLPXPV DQG ODEHOOLQJ UHIHUV WR LQGXVWULDO LQWHUSUHWDWLRQ RI WKH WZR FXUYHV 3OHDVH VHH WH[W IRU H[SODQDWLRQ RI WKLV EUHDNGRZQ &RPSDUH WR )LJ ZKLFK LV WKH DXWKRUfV LQWHUSUHWDWLRQ RI WKH VDPH GDWD DIWHU TXDQWXPOHVV SDUDPHWHU H[WUDFWLRQ PAGE 76 9* 9f )LJ 7KHRUHWLFDOO\ JHQHUDWHG /)&9 GDWD WKUHHSRLQW SDUDPHWHU H[WUDFWLRQ XVLQJ GDWD LQ )LJ ([WUDFWHG SDUDPHWHUV DUH VKRZQ ,QVWHDG RI DWWULEXWLQJ WKH GLIIHUHQFH EHWZHHQ WKH RSWLFDO WKLFNQHVV RI Âƒ DQG WKH fH[WUDSRODWHGf WKLFNQHVV RI Âƒ DW 9* 9 WR TXDQWXP HIIHFWV ZH ILQG WKDW $ LV GXH WR WKH GLVWULEXWLRQ IXQFWLRQ XVHG )HUPL'LUDFf ZLWK WKH UHPDLQLQJ Âƒ SRVVLEO\ GXH WR HUURU LQ RSWLFDO PHDVXUHPHQW DQGRU RWKHU IDFWRUVf PAGE 77 SURFHVVLQJ $V IDU DV SDUDPHWHU H[WUDFWLRQ LV FRQFHUQHG WKH PRVW LPSRUWDQW IDFWRU VKRXOG EH DJUHHPHQW ZLWK HOHFWULFDO UHVXOWV QRW RSWLFDO $OWKRXJK D WKUHHSRLQW PHWKRG ZDV XVHG KHUH WR LPSURYH WKH PDWFK DW WKH 9 SRLQWf D IXOO ILW WR WKH GDWD LQ )LJ KDV DERXW D b 506 HUURU ZKLFK FRQVLGHULQJ WKH WKLQ R[LGH DQG WKH IDFW WKDW WKH VXEVWUDWH JDWH GRSLQJ LV QRW FRQVWDQW LV H[WUHPHO\ JRRG $Q LQWHUHVWLQJ VLGH QRWH EURXJKW XS GXULQJ WKH SURSRVDO IRU WKLV SURMHFW FRQFHUQHG ILWWLQJ D %ROW]PDQQ PRGHO WR WKH GDWD LQVWHDG RI D )HUPL'LUDF 7KH VXUSULVLQJ UHVXOW RI WKLV ZDV D EHWWHU ILW b 506 HUURUf EXW DV ZRXOG EH H[SHFWHG D WKLFNHU H[WUDFWHG R[LGH WKLFNQHVV RI Âƒ 7KLV LV DQ LQWHUHVWLQJ UHVXOW DV LW VKRZV WKDW XVLQJ WKH ZURQJ PRGHO FDQ DSSHDU WR JLYH fEHWWHUf UHVXOWV LQ WHUPV RI ILWf HYHQ WKRXJK WKH UHVXOWLQJ SDUDPHWHUV DFWXDOO\ KDYH JUHDWHU HUURU GXH WR WKH LQFRUUHFW FDUULHU GLVWULEXWLRQf 3KLORVRSKLFDOO\ WKH LVVXH RI R[LGH WKLFNQHVV LV DQ LQWHUHVWLQJ WRSLF 0DQ\ SHRSOH ZRXOG DUJXH WKDW 7(0 LV WKH RQO\ ZD\ WR PHDVXUH WKH fWUXHf WKLFNQHVV +RZHYHU LJQRULQJ WKDW WKLV LV D GHVWUXFWLYH DQG WLPHFRQVXPLQJ WHFKQLTXH LW RQO\ \LHOGV WKH WKLFNQHVV RI WKDW SDUWLFXODU FURVV VHFWLRQ :KDW LV GHVLUHG LV WKH DYHUDJH R[LGH WKLFNQHVV DV LW LV WKH DYHUDJH R[LGH WKLFNQHVV ZKLFK DIIHFWV WKH DPRXQW RI FKDUJH DFFXPXODWHG E\ DQ DSSOLHG YROWDJH LQ D 026& 7KLV LV ZK\ )RZOHU 1RUGKHLP )1f WXQQHOOLQJ LV DOVR QRW D SDUWLFXODUO\ JRRG PHWKRGfÂ§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Âƒ RI HUURU WR WKH UHVXOWV >@ DVVXPLQJ WKH\ DUH NQRZQ WR ZLWKLQ b &RQYHUJHQFH 6SHHGXS 'HWDLOV ,WHUDWLRQ VWRSV ZKHQ QRQH RI WKH ILWWLQJ SDUDPHWHUV LH WKH H[WUDFWHG SDUDPHWHUVf FKDQJHV E\ PRUH WKDQ [ 7b EHWZHHQ VXFFHVVLYH VHULDO F\FOHV LH HDFK SDUDPHWHU ZDV ILW DQG QRQH FKDQJHG E\ PRUH WKDQ [O&7bf %HORZ DUH VHYHUDO RI WKH PHWKRGV HPSOR\HG WR VSHHG XS WKLV FRQYHUJHQFH 7KH FDOFXODWLRQV LQYROYHG LQ WKH H[WUDFWLRQ DUH H[WUHPHO\ FRPSOH[ 2I JUHDWHVW LPSRUWDQFH LV WKH FRQYHUJHQFH VSHHG RI WKH SRO\ /)&9 PRGHO ZKLFK LWVHOI PXVW FRQYHUJH RQ WKH WZR VXUIDFH SRWHQWLDOV MXVW WR JLYH RQH &J GDWD SRLQW IRU D JLYHQ 9* 7KLV RQH GDWD SRLQW LV XVHG LQ WKH QXPHULFDO SDUWLDO GHULYDWLYHV IRU WKH QRQOLQHDU OHDVWVTXDUHV ILW ZKLFK PHDQV WKH &J9*f LV FDOOHG WZLFH IRU HDFK H[SHULPHQWDO GDWD SRLQW IRU HDFK LWHUDWLRQ %HFDXVH WKLV PRGHO LV FDOOHG VR IUHTXHQWO\ LW LV LPSRUWDQW WR NHHS WUDFN RI DOO WKH FRQYHUJHG VXUIDFH SRWHQWLDOV IRU HDFK H[SHULPHQWDO GDWD SRLQW VR WKDW WKH /)&9 PRGHO KDV D JRRG HVWLPDWH IRU VXEVHTXHQW FDOOV 7KH GHOWD XVHG IRU WKH QXPHULFDO SDUWLDO GHULYDWLYHV DV LW WXUQV RXW KDV D PDMRU LQIOXHQFH RQ WKH FRUUHFWQHVV RI WKH ILW 6LQFH WZR RI WKH SDUDPHWHUV YDU\ ORJDULWKPLFDOO\ WKH VXEVWUDWH DQG JDWH GRSLQJf WKH DFWXDO SDUDPHWHUV XVHG GXULQJ WKH ILW DUH WKH ORJV RI WKHVH SDUDPHWHUV 7KXV WKH GHOWDV XVHG IRU WKH GHULYDWLYHV PXVW EH FDOFXODWHG GLIIHUHQWO\ $QRWKHU SUREOHP LV WKDW WKH GHOWD XVHG IRU WKH R[LGH WKLFNQHVV LV GHSHQGDQW XSRQ WKH PAGE 79 DFWXDO WKLFNQHVV RI WKH R[LGH WKDW LV LW VKRXOG EH GLIIHUHQW IRU D Âƒ R[LGH FRPSDUHG WR D Âƒf 7KH HPSLULFDO UHVXOWV IRU WKH EHVW GHOWDV DV GHWHUPLQHG IURP DQDO\VLV RI WKH QXPHULFDO GHULYDWLYHV ZHUH $ ,f IRU OQ1[[f $ IRU OQ1*f $ IRU 9)% DQG a[7R[ IRU 7R[ )RU H[DPSOH WKH SDUWLDO GHULYDWLYH RI &J ZLWK UHVSHFW WR 7R[ LV FDOFXODWHG IURP &J9*f &J9*ff$ ZKHUH &J9*f LV WKH JDWH FDSDFLWDQFH FDOFXODWHG DW VRPH 9* ZLWK DQ R[LGH WKLFNQHVV RI 7R[ $ DQG &J9*f LV WKH JDWH FDSDFLWDQFH FDOFXODWHG DW WKH VDPH 9* ZLWK DQ R[LGH WKLFNQHVV RI 7R[ $ 7KH UHDVRQ DUELWUDULO\ VPDOO YDOXHV FDQQRW EH XVHG RI FRXUVH LV EHFDXVH WKH /)&9 PRGHO LWVHOI LV RQO\ DFFXUDWH WR DERXW HLJKW GLJLWV OHVV QHDU IODWEDQGf GXH WR LWV RZQ LQWHUQDO FRQYHUJHQFH FULWHULD >@ )LQDOO\ WR VWDUW WKH H[WUDFWLRQ D UHDVRQDEOH LQLWLDO JXHVV PXVW EH PDGH 7KH LQLWLDO JXHVV RI WKH R[LGH WKLFNQHVV LV VLPSO\ $HR[&JPD[ ZKHUH &JPD[ LV WKH PD[LPXP JDWH FDSDFLWDQFH LQ WKH GDWDVHW DQG $ LV WKH JDWH DUHD 7KLV LV WKH VWDQGDUG ILUVWRUGHU DSSUR[LPDWLRQ EDVHG RQ &J & )RU WKH VXEVWUDWH GRSLQJ LQLWLDO JXHVV WKH DV\PSWRWLF KLJKIUHTXHQF\ &9 IRUPXOD IRU &Jf >@ LV VROYHG LWHUDWLYHO\ XVLQJ WKH PLQLPXP DQG PD[LPXP &J YDOXHV IURP WKH GDWD VHW 7KH JDWH GRSLQJ LV VLPSO\ VHW WR [O FP :LWK WKHVH WKUHH SDUDPHWHUV DSSUR[LPDWHO\ NQRZQ WKH IODWEDQG YROWDJH LV HVWLPDWHG IURP A*UDLQBGDWD A*PLLXKHRU\ 6LQFH WKH PLQLPXP RI WKH &9 GDWD LV QRW QHFHVVDULO\ JLYHQ EXW LV QHHGHG LQWHUQDOO\ WR HVWLPDWH WKH IODWEDQG WKH PLQLPXP WKUHH GDWD SRLQWV LQ WHUPV RI &Jf DUH XVHG WR HVWLPDWH WKH WUXH &J PLQLPXP EDVHG RQ WKH SDUDEROLF PLQLPXP IRUPXOD >@ 7KLV VOLJKWO\ LPSURYHV WKH FRQYHUJHQFH EXW QRW DV PXFK DV ZRXOG EH H[SHFWHG ODUJHO\ EHFDXVH WKH PLQLPXP RI WKH &9 FXUYH LV QRW YHU\ SDUDEROLF PAGE 80 %HFDXVH LW ZDV FOHDU WKDW FRQYHUJHQFH ZDV VORZHU DV WKH UHVXOWV DSSURDFKHG WKH ILQDO YDOXHV D fWULFNf ZDV GHYHORSHG WR LPSURYH WKLV HQG FDVH :KHQHYHU D WUHQG ZDV YLVLEOH GXULQJ D ILW WKH URXWLQH GRXEOHG WKH DPRXQW RI WKH SDUDPHWHU LQFUHDVH $ WUHQG LQ WKLV FDVH LV GHILQHG DV WKUHH VXFFHVVLYH PRYHV RI D SDUDPHWHU LQ WKH VDPH GLUHFWLRQ IRU DOO WKH SDUDPHWHUV SRVVLEO\ GLIIHUHQW GLUHFWLRQV IRU GLIIHUHQW SDUDPHWHUVf 7KLV FXW GRZQ WKH QXPEHU RI LWHUDWLRQV E\ DERXW b LQ PRVW FDVHV 2QH WKLQJ ZKLFK ZRXOG KDYH LPSURYHG WKH VSHHG RI FRQYHUJHQFH JUHDWO\ ZRXOG EH WR XVH D VLPSOHU PRGHO IRU WKH )HUPL'LUDF LQWHJUDO 7KH &RG\7KDWFKHU PRGHO >@ LV H[WUHPHO\ DFFXUDWH EXW UHTXLUHV WKH TXRWLHQW RI WHQ H[SRQHQWLDOV IURP D &KHE\VKHY DSSUR[LPDWLRQ 7KLV DSSUR[LPDWLRQ ZDV XVHG LQVWHDG RI VRPH RWKHU VLPSOHU WKRXJK OHVV DFFXUDWHf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f GLUHFW H[SHULPHQWDO PHDVXUHPHQWV RI WKH HIIHFW RI KRWFDUULHU GHJUDGDWLRQ RQ LQWULQVLF FDSDFLWDQFH ZLOO EH GLVFXVVHG DQG WKH UHVXOWV PRGHOHG 7KH LPSDFW RI WKLV GHJUDGDWLRQ RQ FLUFXLW SHUIRUPDQFH ZLOO EH HYDOXDWHG DQG VKRZQ WR RIIVHW VRPH RI WKH ORVVHV GXH WR ,' GHJUDGDWLRQ %DFNJURXQG 7KH HIIHFW RI KRWFDUULHU GHJUDGDWLRQ RQ WKH GUDLQ FXUUHQW ,' KDV EHHQ VWXGLHG LQWHQVHO\ VLQFH $EEDVf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f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f 2QFH 9*6 9*7 DQG 9'6 9'6VDWf &JG DQG &GJ ZLOO ERWK KDYH VRPH ILQLWH SRVLWLYH YDOXH ZKHQ WKH FKDQQHO IRUPV 7KH LQWHUHVWLQJ FDVH LV ZKHQ 9*6 9*7 DQG 9'6 9'6VDW 1RZ WKHUH LV D FKDQQHO EXW LW LV fSLQFKHG RIIf QHDU WKH GUDLQ HQG &GJ G4GG9Jf LV QRQ]HUR VLQFH D FKDQJH LQ WKH JDWH YROWDJH VWLOO DIIHFWV WKH FKDUJH DVVRFLDWHG PAGE 83 ZLWK WKH GUDLQ 4Gf &JG G4JG9Gf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f FDSDFLWDQFH ZKLFK KDG EHHQ RPLWWHG E\ 6DK DQG KHQFH 0H\HU :DUG DQG 'XWWRQfV FKDUJHEDVHG PRGHO ZDV D KXJH LPSURYHPHQW DW WKH WLPH DV 0H\HUfV PRGHO GRHV QRW JXDUDQWHH FKDUJHFRQVHUYDWLRQ LQ FLUFXLW VLPXODWRUV GXH WR RPLWWLQJ &JEf UHVXOWLQJ LQ HUURQHRXV UHVXOWV IRU WKH VLPSOHVW RI FLUFXLWV 3DSHUV SUHGDWLQJ 0H\HUfV ZRUN ODUJHO\ XVHG GLVFUHWH GHYLFHV DQG VR DXWKRUV ORJLFDOO\ DUJXHG WKDW PRGHOLQJ WKH LQWULQVLF FDSDFLWDQFHV ZRXOG EH XVHOHVV VLQFH WKH FDSDFLWDQFH IURP SDFNDJLQJ DQG H[WHUQDO FLUFXLWU\ ZRXOG EH YDVWO\ ODUJHU >@ )XUWKHUPRUH WKHUH ZDV QR GLUHFW PHWKRG WR PHDVXUH WKH GDWD WR YHULI\ WKH PRGHOV :LWK WKH DGYHQW RI LQWHJUDWHG FLUFXLWV WKH SULPDU\ FDSDFLWLYH ORDG EHWZHHQ &026 FLUFXLW FHOOV LH DQ Q026 DQG S026 LQYHUWHU SDLUf EHFDPH GRPLQDWHG E\ WKH LQWULQVLF DQG LQWHUFRQQHFW FDSDFLWDQFHV UDWKHU WKDQ WKH SDFNDJLQJ DQG H[WHUQDO FLUFXLWU\ 7KXV PRGHOLQJ WKH LQWULQVLF FDSDFLWDQFH DV ZHOO DV LQWHUFRQQHFWf EHFDPH LPSRUWDQW ,QWHJUDWHG FLUFXLWV DOVR KDLOHG WKH QHHG IRU FRPSDFW PRGHOV WR VLPXODWH ODUJH QXPEHUV RI WUDQVLVWRUV 2QH RI WKH ILUVW FRPSDFW PRGHOV ZDV &6,0 >@ IURP $7t7 %HOO PAGE 84 ODEV 6XUSULVLQJO\ WKH DXWKRUV RI WKLV PRGHO VWD\HG ZLWK WKH VLPSOH 0H\HU PRGHO DOWKRXJK DUJXHG WKDW LQFOXGLQJ WKH LQWULQVLF FDSDFLWDQFHV ZDV FULWLFDO SDUWLFXODUO\ &JG &JG DFFRXQWV IRU PRVW RI WKH LQWULQVLF FDSDFLWDQFH ORDG LQ &026 FLUFXLWV GXH WR WKH 0LOOHU IHHGEDFN HIIHFW >@ %HUNHOH\fV %6,0 >@ EXLOW XSRQ &60 DOVR UHWDLQHG WKH 0H\HU PRGHO %6,0 KRZHYHU FRUUHFWHG WKLV GHILFLHQF\ E\ LQFOXGLQJ D QRQUHFLSURFDO LQWULQVLF FDSDFLWDQFH PRGHO 7KH %6,0 >@ PRGHO PRYHG IURP DQ VWURQJO\ HPSLULFDO GF PRGHO WR D PRUH SK\VLFDOO\EDVHG PRGHO EXW UHWDLQHG WKH XQDOWHUHG DF PRGHO LQFOXGLQJ LQWULQVLF FDSDFLWDQFHVf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fV ZRUN >@ PAGE 85 0HDVXUHPHQW RI ,QWULQVLF &DSDFLWDQFHV %HFDXVH GLUHFW PHDVXUHPHQW RI WKH LQWULQVLF FDSDFLWDQFH LV GLIILFXOW PDQ\ RI WKH ILUVW PHDVXUHPHQWV ZHUH GRQH ZLWK RQFKLS FLUFXLWU\ XVLQJ UHIHUHQFH FDSDFLWRUV > @ RU RSDPSV FLUFXLWV FRQILJXUHG DV FRXORPEHUV >@ (YHQWXDOO\ H[WHUQDO FLUFXLWU\ ZDV XVHG LQFOXGLQJ D ORFNLQ DPSOLILHU FRQQHFWHG WR DQ +3 DV D YROWDJH VRXUFHf >@ DQG ODWHU DQ RIIWKHUDFN /&5 PHWHU >@ VXFK DV WKH +3 $ ,Q WKLV VHFWLRQ WKH PHDVXUHPHQW RI D IHZ RI WKH LQWULQVLF FDSDFLWDQFHV ZLOO EH GHVFULEHG 7KHVH FDQ EH GRQH XVLQJ DQ +3 RU +3 VDPH HTXLSPHQW ZLWK GLIIHUHQW DF IUHTXHQF\ UDQJHVf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f LV GHVLUHG WKH VRXUFH DQG VXEVWUDWH FDQ EH ELDVHG DW 9 DQG WKH GUDLQ FDQ EH ELDVHG DW 9 6LQFH WKH JDWH LV YLUWXDO JURXQG 9*6 f LW LV HDV\ WR YHULI\ WKDW WKH DERYH DSSOLHG YROWDJHV JLYH WKH GHVLUHG SRWHQWLDO GLIIHUHQFHV 9*6 9'6 DQG 9[Vf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f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f WKH PHDVXUHG &JG FKDUDFWHULVWLFV DV ZHOO DV &JV &GJ DQG &GVf DUH DV\PPHWULF 7KLV RFFXUV EHFDXVH WKH PHDVXUHG FKDUDFWHULVWLFV LQFOXGH WKH FRQVWDQW RYHUODS FRPSRQHQW DV VKRZQ LQ WKH IROORZLQJ VLPSOH HTXDWLRQ F fÂ§ F U AJW/PHDVXUHG ARYBGUDLQ AJG 7KH &RY GUDLQ WHUP LV FRPSRVHG RI WKH FRQVWDQW RYHUODS RI WKH JDWH ZLWK WKH GUDLQ DV ZHOO DV DQ LQQHU DQG RXWHU IULQJH FRPSRQHQW 7KHVH IULQJH FRPSRQHQWV KDYH EHHQ FDOFXODWHG WKHRUHWLFDOO\ >@ DQG DVVXPLQJ WKH\ DUH FRQVWDQW DV D IXQFWLRQ RI JDWH YROWDJH LQWURGXFHV QHJOLJLEOH HUURU >@ 7KH YDOXH RI WKH PHDVXUHG &JG LQ VXEWKUHVKROG ZKHUH &JG PHDVXUHG &RY GUDLQf KDV EHHQ XVHG WR HVWLPDWH WKH OHQJWK RI WKH JDWHWRGUDLQ RYHUODS UHJLRQ >@ DQG ZLWK WKH GUDZQ FKDQQHO OHQJWK NQRZ WKHVH RYHUODS YDOXHV FRXOG EH XVHG WR H[WUDFW WKH HIIHFWLYH FKDQQHO OHQJWK :KHQ QHFHVVDU\ WKH fQRUPDOf DQG fUHYHUVHf FRQILJXUDWLRQV RI &JG DQG &JV PHDVXUHPHQWV ZLOO EH VSHFLILHG 7KHVH DUH VKRZQ LQ )LJXUH TMI RU &JGQRUP UHIHUV WR WKH fQRUPDOf PHDVXUHPHQW PRGH ZKHUH WKH KLJK SRUW LV DSSOLHG WR WKH GUDLQ IRU D &JG PAGE 89 JDWHA )LJ 6LPSOLILHG VFKHPDWLF RI DV\PPHWULF JDWH RYHUODS ZKLFK UHVXOWV LQ nfnR YBGUDL LUnnRYBVRXUFH f PAGE 90 )LJ $f & JGBQRUP %f & &f& 'f & 0HDVXUHPHQW FRQILJXUDWLRQV IRU $f &JG LQ QRUPDO FRQILJXUDWLRQ PRGH %f &JV LQ QRUPDO FRQILJXUDWLRQ PRGH &f &JG LQ UHYHUVH FRQILJXUDWLRQ PRGH DQG 'f &JV LQ UHYHUVH FRQILJXUDWLRQ PRGH PAGE 91 PHDVXUHPHQW &JGY RU &JG UHY UHIHUV WR WKH fUHYHUVHf PHDVXUHPHQW PRGH ZKHUH WKH KLJK SRUW LV DSSOLHG WR WKH VRXUFH IRU D &JG PHDVXUHPHQW 7KLV LV QHFHVVDU\ EHFDXVH IRU VKRUW FKDQQHO GHYLFHV WKH UHVXOWLQJ &RY YDOXH ZKHUH &RY LV &RYGUDLQ RU &RY VRXUFFf FDQ EHFRPH D VLJQLILFDQW IUDFWLRQ RI WKH WRWDO HIIHFWLYH LQWULQVLF FDSDFLWDQFH $OWKRXJK SHUKDSV QRW REYLRXV QRZ &JV &JG ZKHQ 9'6 +RZHYHU GXH WR WKH GLIIHUHQFH LQ &RY &JVPHDVXUHG PD\ QRW HTXDO &JG PHDVXUHG )LJXUH VKRZV WKH &JV DQG &JG PHDVXUHPHQWV LQ WKHQRUPDO DQG UHYHUVH PRGHV IRU D [ SP GHYLFH )LJXUH VKRZV WKH VDPH PHDVXUHPHQWV RQ D [ SP GHYLFH HIIHFWLYH FKDQQHO OHQJWK LV SPf &RPSDULQJ WKH WZR ILJXUHV FOHDUO\ VKRZV WKH QHJOLJLEOH LPSDFW RI &RY RQ WKH ORQJFKDQQHO GHYLFH &JG DQG &JV FKDUDFWHULVWLFV DQG WKH ODUJH LPSDFW RQ WKH VKRUWFKDQQHO GHYLFH ,Q ERWK FDVHV WKH &JG DQG &JV YDOXHV DUH DOPRVW LGHQWLFDO DV LV WKH RYHUODSLQGXFHG GLIIHUHQFH RI DERXW I) 7KLV I) RIIVHW LV QRW YLVLEOH RQ WKH /GUDZQ SP GHYLFH EHFDXVH LW FRQWULEXWHV OHVV WKDQ b WR WKH PD[LPXP FDSDFLWDQFH ZKHUHDV WKH RYHUODS FRQWULEXWHV DERXW b RI WKH WRWDO PHDVXUHG PD[LPXP FDSDFLWDQFH IRU WKH /GUDZQ SP GHYLFHf /DWHU LQ WKLV FKDSWHU WKH UHVXOWV RI FKDQQHO KRWFDUULHU VWUHVV RQ &JG DQG &JV ZLOO EH VKRZQ %HFDXVH FKDQQHO KRWFDUULHU VWUHVV LV LQKHUHQWO\ DV\PPHWULF VLQFH WKH GDPDJH RFFXUV QHDU WKH GUDLQ HGJHf LW LV QHFHVVDU\ WR OD\ GRZQ WKH DERYH QRWDWLRQ IRU ODWHU XVH 6DPSOH 0HDVXUHPHQWV )RU DOO FDSDFLWDQFH PHDVXUHPHQWV LQ WKLV FKDSWHU DQ +3 $ /&5 PHWHU ZDV XVHG ZLWK D VPDOOVLJQDO YROWDJH ZDV 0+] DW P9 SHDNWRSHDN 7KHVH QXPEHU ZHUH FKRVHQ DIWHU WHVWLQJ D ZLGH UDQJH RI DF VLJQDO YROWDJHV DQG IUHTXHQFLHV PAGE 92 )LJ &JGBQRUPf &JGBUHYf &J6,UP DQG &J6BUHY YHUVXV 9Ti IRU D [ SP 0267 ZLWK 9'6 $OWKRXJK LW DSSHDUV WKDW DOO IRXU FXUYHV DUH WKH VDPH WKHUH DUH DFWXDOO\ WZR VHWV RI FXUYHV &JGQRUP&JVUHY DQG &JGBUH?&J6BQRUP VHSDUDWHG E\ I) 9HU\ OLWWOH GLIIHUHQFH LV VHHQ EHFDXVH WKH RYHUODS FDSDFLWDQFHV VKLIW LV PXFK OHVV WKDQ WKHQ WKH SHDN &JG DQG &JV YDOXHV &RPSDUH WKLV ZLWK )LJ PAGE 93 )LJ AnJGBLRUPn AnJGBUHYn AnJV QRUPn AJV UHY 9HUVXV 9TA IRU D [ 0267 ZLWK 9'6 5RXJKO\ I) SDUDOOHO VKLIW RI &JVBQRUP&JGUHY DQG &JV UHY&JG QRUP LV GXH WR D GLIIHUHQFH LQ FRQVWDQW RYHUODS FDSDFLWDQFH EHWZHHQ WKH VRXUFH DQG GUDLQ PAGE 94 WR REWDLQ WKH PRVW DFFXUDWH UHVXOWV 7KH P9 VLJQDO PD\ VHHP D OLWWOH ODUJH WR WKRVH IDPLOLDU ZLWK FRPPRQ &9 PHDVXUHPHQWV ZKHUH P9 LV W\SLFDOO\ XVHG EXW LV DFWXDOO\ RQ WKH ORZ HQG RI WKH P9 WR P9 IRXQG LQ PRVW LQWULQVLF FDSDFLWDQFH SDSHUV > @ )UHTXHQFLHV EHORZ 0+] UHVXOW LQ H[WUHPHO\ SRRUUHVROXWLRQ QRLV\f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f LV LPSRUWDQW :K\ LV WKLV FDSDELOLW\ LPSRUWDQW" %HFDXVH LQ D UHDO FLUFXLW WKLV ZLOO FRPPRQO\ RFFXU ,I D FRUUHFW PRGHO IRU WKH EHKDYLRU RI DQ RSHUDWLQJ WUDQVLVWRU LV GHVLUHG WKHQ GDWD IURP DQ DFWLYH GHYLFH LV UHTXLUHG ,QGHHG ZLWKRXW WKLV GDWD LW ZRXOG EH OLNH WU\LQJ WR YHULI\ DQ ,'VDW PRGHO ZLWK GDWD RQO\ WDNHQ LQ VXEWKUHVKROG ([DPSOHV RI &JG PHDVXUHPHQWV RQ DFWLYH GHYLFHV DUH VKRZQ LQ )LJXUHV DQG IRU [ SP DQG ; SP DV D IXQFWLRQ RI 9*6 IRU 9'6 DQG 9 YV[ 9f &JG WUDQVLWLRQV IURP &RY GUDLQ WR D ODUJHU YDOXH RQFH 9'6 9'6VDW RU WKH FKDQQHO LV QR ORQJHU SLQFKHGRII )URP D FKDUJH SHUVSHFWLYH WKLV PHDQV FKDQJHV LQ 9'6 G9Gf FDXVH FKDQJHV LQ 4FKDQQH_ ZKLFK LQ WXUQ FDXVH FKDQJHV LQ 4J G4Jf UHVXOWLQJ LQ D PAGE 95 GG Of S )LJ 9*6 9f &JG YHUVXV 9*6 IRU D [ _LP 0267 ZLWK 9'6 DQG 9 PAGE 96 &JG I)f )LJ &JG YHUVXV 9*6 IRU D [ _LP 0267 ZLWK 9'6 DQG 9 PAGE 97 &JG 7KXV DV 9'6 LQFUHDVHV WKH SRLQW DW ZKLFK WKLV WUDQVLWLRQ RFFXUV DOVR LQFUHDVHV DV FDQ EH VHHQ LQ WKH ILJXUHV $V PHQWLRQHG SUHYLRXVO\ &JG LV WKH PRVW LPSRUWDQW LQWULQVLF FDSDFLWDQFH EHFDXVH LQ D FRPPRQVRXUFH FRQILJXUDWLRQ ZKLFK LV WKH FRQILJXUDWLRQ IRU DOO &026 FLUFXLWVf WKH HIIHFWLYH ORDG LV &JV &JG $Yff ZKHUH $Y LV WKH JDLQ EHWZHHQ WKH JDWH LQSXW DQG GUDLQ RXWSXW D ODUJH QHJDWLYH QXPEHUf 7KH QH[W PRVW LPSRUWDQW FDSDFLWDQFH EDVHG RQ WKH DERYH ORDG IRUPXOD LV &JV )LJXUHV DQG VKRZ ERWK &JV DQG &JG IRU D [ SP DQG [ SP DV D IXQFWLRQ RI 9*6 IRU 9'6 DQG 9 8QOLNH &JG &JV ZLOO KDYH D ILQLWH YDOXH DV ORQJ DV 9*6 LV JUHDWHU WKDQ 9*7 VLQFH WKH FKDQQHO ZLOO DOZD\V EH FRQQHFWHG WR WKH VRXUFH $W 9'6 &JV &JG VLQFH WKH FKDQQHO FKDUJH LV HTXDOO\ FRQWUROOHG E\ WKH VRXUFH DQG GUDLQ +RZHYHU LI 9*6 9*7 FKDQQHO IRUPVf DQG 9'6 9'6VDW GUDLQ SLQFKHG RIIf WKHQ WKH VRXUFH WHUPLQDO ZLOO DFWXDOO\ FRQWURO PRUH WKDQ KDOI RI WKH FKDQQHO FKDUJH UHVXOWLQJ LQ D ULVH LQ &JV DERYH WKH YDOXH DW 9'6 +RZHYHU RQFH 9*6 LQFUHDVHV WR D SRLQW WKDW 9'6 9'6VDW WKH GUDLQ LV QR ORQJHU SLQFKHG RII DQG WKH &JV YDOXH EHJLQV WR GHFOLQH ZLWK LQFUHDVLQJ 9*6 DV &JG LQFUHDVHV UDSLGO\ 7KLV LV FOHDUO\ GHPRQVWUDWHG LQ )LJXUH DQG WR D OHVVHU H[WHQW LQ f ZKHUH WKH GHFOLQH LQ &JV FRUUHVSRQGV WR WKH LQFUHDVH LQ &JG 7KH PRGHO IRU &JG DQG &JV ZLOO EH GLVFXVVHG ODWHU 5HFDOOLQJ WKH GLVFXVVLRQ DERXW WKH RYHUODSFDSDFLWDQFH VKLIWLQJ LQ WKH SUHYLRXV VHFWLRQ WKH FDSDFLWDQFHV VKRZV LQ DQG DUH DFWXDOO\ &JGUUD DQG &JDY LQ RUGHU WR RIIVHW WKH HIIHFWV RI WKH RYHUODS FDSDFLWDQFH )LJ VKRZV ZK\ WKLV ZDV QHFHVVDU\f PAGE 98 9*6 9f )LJ &JG DQG &JV YHUVXV 9*6 IRU D [ _LP 0267 ZLWK 9'6 DQG 9 PAGE 99 &JG YHUVXV 9*6 IRU D [ LP 0267 ZLWK 9'6 DQG 9 )LJ PAGE 100 &KDQQHO +RW&DUULHU 6WUHVV (IIHFWV RQ DQG &f %HFDXVH WKH LQWULQVLF FDSDFLWDQFHV DUH VRPHZKDW GLIILFXOW WR PHDVXUH DV ZHOO DV WKH UHODWLYHO\ VPDOO FRQWULEXWLRQ RI LQWULQVLF FDSDFLWDQFH RQ FLUFXLW SHUIRUPDQFH LQ SDVW JHQHUDWLRQV YHU\ OLWWOH ZRUN KDV EHHQ GRQH WR LQYHVWLJDWH WKH LPSDFW RI KRWFDUULHU VWUHVV RQ LQWULQVLF FDSDFLWDQFH $OWKRXJK WKH ILUVW UHSRUW RI KRWFDUULHU GHJUDGDWLRQ RQ ,' ZDV SXEOLVKHG LQ E\ $EEDV DQG 'RFNHUW\ >@ WKH ILUVW LQYHVWLJDWLRQ RI &JG DQG &JV GHJUDGDWLRQ ZDV QRW SXEOLVKHG XQWLO E\ PAGE 101 f $IWHU WKH ILQDO VWUHVVHG &JG PHDVXUHPHQW WDNH &JV PHDVXUHPHQW QRUPDO PRGHf DQG WKHQ PHDVXUH WKH ILQDO fVWUHVVHGff ,' YHUVXV 9'6 DV LQ f 7KH DFFHOHUDWHG VWUHVV FRQGLWLRQV ZHUH 9*6 9 9'6 9 IRU Q026 GHYLFHV DQG 9*6 9 9'6 9 IRU S026 GHYLFHV IRU D WRWDO VWUHVV WLPH RI KU WZHOYH &JG&JV PHDVXUHPHQWV WRWDOf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f DQG f WKH ,' GDWD ZHUH PHDVXUHG RQ D GLIIHUHQW DSSDUDWXV DQ DXWRPDWHG SUREHUf 7KLV LV DFFHSWDEOH VLQFH WKHUH ZLOO EH QHJOLJLEOH PHDVXUHPHQW HUURU IURP UHSURELQJ DQG PHDVXULQJ WKH ,' FXUYHV +RZHYHU LQ VWHSV f DQG f LW LV YHU\ LPSRUWDQW WKDW WKH VWUHVV EH SHUIRUPHG ZLWKRXW UHSURELQJ LQ D VKLHOGHG SUREH ER[ SUHIHUDEO\ ZLWK WKH FDSDFLWDQFHPHDVXULQJ SUREHV DOORZLQJ WKH IRUFH DQG VHQVH OLQHV IURP WKH /&5 PHWHU KLJK DQG ORZ SRUWV WR FRQQHFW ULJKW DW WKH SUREH WLS LH DW WKH WUDQVLVWRU SDGVf $IWHU FDOLEUDWLQJ ]HURLQJf WKH /&5 PHWHU WR DFFRXQW IRU WKH SUREH FRQILJXUDWLRQ FDSDFLWDQFH DQ\ DGGLWLRQDO UHSURELQJ FDQ HDVLO\ DGG VHYHUDO IHPWRIDUDG WR WKH PHDVXUHG FDSDFLWDQFH ZKLFK LV RQ WKH RUGHU RI WKH GHJUDGDWLRQ DPRXQWV VKRZQ ODWHUf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r1Y DOWKRXJK LQWHUHVWLQJ LV QRW D FRPSRQHQW ZKLFK FRPHV LQWR SOD\ LQ DFWXDO FLUFXLW RSHUDWLRQ ,QVWHDG WKH PXFK PRUH LPSRUWDQW LQWULQVLF FDSDFLWDQFH &JrUP LV PHDVXUHG EHIRUH DQG DIWHU WKH VWUHVV EXW QRW LQ VLWX EHFDXVH WKDW ZRXOG UHTXLUH HLWKHU D PDQXDO UHSURELQJ ZKLFK LV SURKLELWLYHO\ ORQJ RU D VZLWFKLQJ PDWUL[ ZKLFK ZRXOG FRXOG QRW EH ]HURHG RXW SURSHUO\ ZLWK WKH /&5 PHWHU $Q LQWHUHVWLQJ LGHD IRU D QHZ SLHFH RI HTXLSPHQW ZRXOG EH DQ /&5 PHWHU ZKLFK DOORZV VHYHUDO VKRUW DQG VKXQW ]HURV WR EH VWRUHG LQ WKH /&5 PHWHUfV PHPRU\ 7KLV ZD\ WKH HTXLSPHQW FRXOG EH ]HURHG WKURXJK GLIIHUHQW FRQILJXUDWLRQV RI WKH FLUFXLW DQG WKH VRIWZDUH FRXOG WKHQ WHOO WKH /&5 ZKLFK SDUWLFXODU ]HUR fVHWf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f GHYLFH H[KLELWV OHVV GHJUDGDWLRQ VLPSO\ EHFDXVH WKH GUDLQ FXUUHQW GXULQJ VWUHVV LV DOVR VPDOOHU GXH WR WKH ORQJHU FKDQQHO OHQJWK VLQFH WR WKH ILUVW RUGHU WKH GUDLQ FXUUHQW LV SURSRUWLRQDO WR /f %HFDXVH WKH GHJUDGDWLRQ LV GXH WR LQWHUIDFH WUDS JHQHUDWLRQ DV GLVFXVVHG LQ WKH QH[W VHFWLRQ WKH VPDOOHU WKH FXUUHQW UHVXOWV LQ D VPDOOHU IOXHQFH DQG WKXV IHZHU JHQHUDWHG KROHV UHVXOWLQJ LQ IHZHU LQWHUIDFH WUDSV $OVR QRWDEOH LQ )LJXUHV DQG LV DQ LQFUHDVH LQ &JG IRU 9*6 9 ,QWULQVLF &DSDFLWDQFH 'HJUDGDWLRQ 0RGHO %RWK WKH UHGXFWLRQ LQ &JG IRU 9*6 9 DQG WKH LQFUHDVH IRU 9*6 9 FDQ EH H[SODLQHG IURP E\ VLPSOH PRGHO >@ 7KH JDWHWRGUDLQ FDSDFLWDQFH LV JLYHQ E\ WKH IROORZLQJ LQWHJUDO :}& U/ &JG YDF[fG[ f YDF ZKHUH YA LV WKH DSSOLHG DF WHVW VLJQDO ,Q WKH DEVHQFH RI D QRQXQLIRUP FKDUJH GHQVLW\ LQ WKH JDWH RU DW WKH JDWH LQWHUIDFH LH ZLWK D VSDWLDOO\ FRQVWDQW 427 DQG 4UUf WKH DSSOLHG WHVW VLJQDO XVHG WR PHDVXUH &JG VKRXOG IDOO XQLIRUPO\ DFURVV WKH FKDQQHO DV VKRZQ LQ WKH VWUDLJKW XQVWUHVVHG FXUYH RI )LJXUH 3XW VLPSO\ WKH DSSOLHG VLJQDO FRQWUROV DOO WKH FKDUJH DW WKH GUDLQ HGJH DQG SURJUHVVLYHO\ OHVV DV WKH VLJQDO GURSV DFURVV WKH FKDQQHO %\ *DXVVfV ODZ WKH FKDUJH LQ WKH FKDQQHO PXVW EH EDODQFHG RXW E\ FKDUJH RQ WKH JDWH PAGE 104 &JG I)f 9*6 9f )LJ &JG YHUVXV 9*6 IRU D [ SP 0267 ZLWK 9'6 DIWHU KRW FDUULHU VWUHVV DW 9*6 9 DQG 9'6 9 7KH GLIIHUHQW FXUYHV UHSUHVHQW PHDVXUHPHQWV WDNHQ GXULQJ WKH VWUHVV DQG GHPRQVWUDWH D UHGXFWLRQ LQ &JG DV D IXQFWLRQ RI VWUHVV WLPH /DVW FXUYH LV VHFRQGV RU KU RI VWUHVV WLPH PAGE 105 &JG I)f 9*6 9f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f[f EHLQJ YDF[fr&r:}G[ 7KLV JLYHV WKH FRQYHQLHQW DV\PSWRWLF H[SUHVVLRQ RI &JG n:/f& IRU &JG LQ VWURQJ LQYHUVLRQ ZLWK QR WUDSSHG FKDUJHf 'XULQJ VWUHVV LQWHUIDFH WUDSV DUH JHQHUDWHG E\ KRWKROH GHK\GURQL]DWLRQ RI LQWHUIDFH 6L+ ERQGV QHDU WKH 026 GUDLQ HGJH DV GLVFXVVHG E\ 6DK >@ :KHQ WKH FKDQQHO LV VWURQJO\ LQYHUWHG DQG DOO WKHVH LQWHUIDFH WUDSV DUH ILOOHG WKH ORFDO WKUHVKROG YROWDJH QHDU WKH GUDLQ ZLOO DOZD\V EH KLJKHU WKDQ WKH UHVW RI WKH FKDQQHO GXH WR WKH ILOOHG LQWHUIDFH WUDSV RIIVHWWLQJ WKH DSSOLHG JDWH YROWDJH 7KXV WKH FRQGXFWDQFH QHDU WKH GUDLQ HGJH LV ORZHU IRU VWUHVVHG GHYLFHV 7KLV PHDQV WKH YDF[f LQ f ZLOO GURS PRUH UDSLGO\ LQ WKH GDPDJHG GUDLQ UHJLRQ WKDQ LQ WKH UHVW RI WKH FKDQQHO UHVXOWLQJ LQ OHVV WRWDO FKDUJH FRQWUROOHG E\ WKH GUDLQ DQG D ORZHU &JG YDOXH 7KLV LV FOHDUO\ GHSLFWHG LQ WKH SRVWVWUHVV LQYHUVLRQ FXUYH RI )LJXUH ZKHUH WKH DUHD XQGHU WKH FXUYH WKH YDOXH RI WKH LQWHJUDOf LV HDVLO\ VHHQ WR EH OHVV WKDQ WKH XQVWUHVVHG FDVH 7KH RQO\ UHPDLQLQJ TXHVWLRQ LV WKH VOLJKW LQFUHDVH LQ &JG DW WKH ORZHU JDWH YROWDJHV $W WKHVH YROWDJHV PRVW RI WKH FDSDFLWDQFH LV GXH WR RYHUODS 'XULQJ VWUHVV LQ DGGLWLRQDO WR WKH LQWHUIDFH WUDS JHQHUDWLRQ WKHUH PD\ EH VRPH KROH WUDSSLQJ LQ WKH R[LGH IURP KRW KROHV JHQHUDWHG LQ WKH GHSOHWLRQ OD\HU LQMHFWHG RYHU WKH 6L EDUULHU LQWR WKH R[LGHf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f 7KH IRUPXOD LV LGHQWLFDO WR f ZLWK WKH OLPLWV VZDSSHG RI FRXUVH YDF LV QRZ PD[LPXP DW WKH VRXUFH HQG DQG ]HUR DQG WKH GUDLQf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f 9*6 9f )LJ &JV YHUVXV 9*6 DW 9'6 DQG 9 EHIRUH DQG DIWHU &+( VWUHVV RQ D [ _OP Q0267 DW VWUHVVHG DW 9*6 9 DQG 9'6 9 IRU KU PAGE 110 FJG I)f )LJ , , , , , O O O O O O O O L L L O O L 9*6 9f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f LV WKH FXUUHQW DFFHSWHG 6(0$7(&+ LQGXVWU\ VWDQGDUG 7KH LQLWLDO DQG ILQDO ,'9'6 GDWD VKRZQ LQ )LJXUH ZHUH ILW WR WKH %6,0 GF PRGHO XVLQJ DSSURSULDWH SDUDPHWHUV 7KH UHVXOWLQJ SDUDPHWHUV ZHUH VDYHG LQ D SDUDPHWHU VHW WKHVH SDUDPHWHU VHWV DUH ODWHU XVHG ZKHQ VLPXODWLQJ WKH FLUFXLW 7KXV WKH LQLWLDO DQG ILQDO ,9 FXUYHV ZHUH LQGLYLGXDOO\ ILW DQG WKH UHVXOWLQJ SDUDPHWHU VHWV ZHUH VDYHG 7KHVH ZLOO EH UHIHUUHG WR DV WKH fIUHVKf DQG fVWUHVVHGf ,9 VHWV UHVSHFWLYHO\ 7KH DF PRGHO XVHG LQ %6,0 LV PXFK OHVV VRSKLVWLFDWHG WKDQ WKH GF PRGHO DQG WKHUH LV QR DXWRPDWHG H[WUDFWLRQ PHWKRGRORJ\ IRU LW 7KXV ILWWLQJ &JG DQG &JV KDG WR EH GRQH PDQXDOO\ 7KH GDWD ILW ZDV WKH 9'6 GDWD LQ )LJXUHV DQG DJDLQ PAGE 112 UHVXOWLQJ LQ fIUHVKf DQG fVWUHVVHGf &9 VHWV IRU WKH W XQVWUHVVHGf GDWD DQG WKH W KU VWUHVVHGf GDWD UHVSHFWLYHO\ 7KLV DOORZV XV WR FRPSDUH WKH IROORZLQJ WKUHH VFHQDULRV f &LUFXLW SHUIRUPDQFH XVLQJ IUHVK ,9 DQG IUHVK &9 SDUDPHWHU VHWV 7KLV ZLOO JLYH WKH SHUIRUPDQFH RI DQ XQVWUHVVHG FLUFXLW f &LUFXLW SHUIRUPDQFH XVLQJ VWUHVVHG ,9 DQG IUHVK &9 SDUDPHWHU VHWV DV GRQH FXUUHQWO\ LQ LQGXVWU\ 7KLV ZLOO JLYH WKH SHUIRUPDQFH RI D VWUHVVHG FLUFXLW ZLWKRXW LQFOXGLQJ LQWULQVLF FDSDFLWDQFH FKDQJHV f &LUFXLW SHUIRUPDQFH XVLQJ VWUHVVHG ,9 DQG VWUHVVHG &9 SDUDPHWHUV VHWV ZKLFK FRPSDUHG WR f ZLOO VKRZ XV WKH HIIHFW RI LQFOXGLQJ WKH &JG GHJUDGDWLRQ DQG &JV HQKDQFHPHQWf 7R FRPSDUH WKHVH D VLPSOH ULQJ RVFLOODWRU FLUFXLW ZLOO EH XVHG ZKLFK LV FRPSULVHG RI VRPH RGGQXPEHU RI &026 LQYHUWHUV FKDLQHG WRJHWKHU )LJXUH VKRZV DQ H[DPSOH RI D VWDJH ULQJ RVFLOODWRU DORQJ ZLWK WKH LQGLYLGXDO &026 LQYHUWHU SDLU FLUFXLW &RQVLGHU D YROWDJH 9FF DSSOLHG WR Q026 DQG S026 JDWHVfÂ§WKLV ZLOO FDXVH WKH Q026 GHYLFH WR WXUQ RQ DQG WKH S026 GHYLFH WR WXUQ RII ZKLFK UHVXOWV LQ JURXQG 9f DSSHDULQJ DW WKH GUDLQ OHDG RI WKH Q026 GHYLFH &RQVLGHU D YROWDJH RI JURXQG 9f DSSOLHG WR WKH Q026 DQG S026 JDWHVfÂ§WKLV ZLOO FDXVH WKH S026 GHYLFH WR WXUQ RQ DQG WKH Q026 GHYLFH WR WXUQ RII ZKLFK UHVXOWV LQ 9FF DSSHDULQJ DW WKH GUDLQ OHDG RI WKH S026 GHYLFH 7KXV WKH &026 FLUFXLW LV DQ LQYHUWHU VLQFH WKH YROWDJH DSSOLHG WR WKH JDWH LV LQYHUWHG LQ WKH ORJLF VHQVH RI WKH WHUPf DW WKH RXWSXW $Q\ RGG FRQQHFWLRQ RI LQYHUWHUV ZLOO UHVXOW LQ WKH YROWDJH RVFLOODWLQJ IURP KLJK WR ORZ DURXQG WKH FKDLQ 7KLV RVFLOODWLRQ IUHTXHQF\ LQ RIWHQ XVHG WR GHPRQVWUDWH WKH LPSDFW RI ,' GHJUDGDWLRQ RQ FLUFXLW PAGE 113 9'6 9f )LJ ,QLWLDO DQG ILQDO ,' YHUVXV 9'6 FKDUDFWHULVWLFV IRU D [ SP 0267 6WUHVV FRQGLWLRQ ZDV 9*6 9 DQG 9'6 9 IRU KU )URP WKHVH GDWD FRPH WKH fIUHVKf LQLWLDOf DQG fVWUHVVHGf ILQDOf SDUDPHWHUV VHWV PAGE 114 )LJ 5LQJ RVFLOODWRU $f &026 LQYHUWHU DV S0267 DQG Q0267 FLUFXLW DEEUHYLDWHG E\ WKH ORJLF V\PERO %f VWDJH ULQJ RVFLOODWRU ZLWK IDQRXW RI WKUHH FRPSRVHG RI &026 LQYHUWHUV DQG LGHDO QR FDSDFLWDQFH RU UHVLVWDQFHf LQWHUFRQQHFW PAGE 115 SHUIRUPDQFH >@ EHFDXVH DV ,' GURSV GXH WR GHJUDGDWLRQ LW WDNHV ORQJHU IRU D VZLWFKLQJ WUDQVLVWRU WR FKDUJH XS WKH LQWULQVLF FDSDFLWDQFH RI WKH QH[W LQYHUWHU DV ZHOO DV WKH LQWHUFRQQHFWf ZKLFK VORZV GRZQ WKH RYHUDOO ULQJ RVFLOODWLRQ IUHTXHQF\ 8VLQJ WKH WKUHH SDUDPHWHU VHWV GLVFXVVHG DERYH ZKLFK ZHUH GHULYHG IURP WKH GDWD DOUHDG\ SUHVHQWHG WKUHH VHWV RI ULQJ RVFLOODWRU ZDYHIRUPV DV VKRZQ LQ )LJXUH FDQ EH IRXQG 7KHVH UHSUHVHQW WKH RXWSXW QRGH YROWDJH RI DQ\ JLYHQ &026 LQYHUWHU LQ D VWDJH ULQJ ZLWK D IDQRXW RI HDFK RXWSXW QRGH FRQQHFWHG WR WKUHH &026 LQYHUWHU LQSXWVf DQG DUH VLPXODWHG XVLQJ ,QWHOfV 63,&(OLNH FLUFXLW VLPXODWRU 7KH IDQRXW LV XVHG WR VLPXODWH D W\SLFDO FLUFXLW ZKHUH RQH WUDQVLVWRU GULYHV PXOWLSOH GRZQVWUHDP WUDQVLVWRUV 7KHVH DGGLWLRQ WUDQVLVWRUV REYLRXVO\ DGG WR WKH ORDG ,QWHUFRQQHFW FDSDFLWDQFH LV QHJOHFWHG VLQFH LWV LPSDFW LV OD\RXWGHSHQGHQW 7KH UHVXOWLQJ RVFLOODWLRQ IUHTXHQFLHV IRU WKH /GUDZQ SP ULQJ ZHUH DV IROORZV IRU HDFK SDUDPHWHU VHW )UHVK ,9 )UHVK &9 0+] 6WUHVVHG ,9 )UHVK &9 0+] 6WUHVVHG ,9 6WUHVV &9 0+] 7KH DERYH GDWD FOHDUO\ GHPRQVWUDWH WKDW E\ LQFOXGLQJ WKH VWUHVVHG LQWULQVLF FDSDFLWDQFH VRPH RI WKH ,' GHJUDGDWLRQ LV RIIVHW UHVXOWLQJ LQ D KLJKHU SRVWVWUHVV RSHUDWLQJ IUHTXHQF\ 7KLV LV VLPSO\ EHFDXVH WKH FDSDFLWLYH ORDG ZKLFK WKH ,' PXVW GULYH LV GHJUDGLQJ VLPXOWDQHRXVO\ ZLWK ,' 1RWH WKDW WKH ,' GHJUDGDWLRQ LQ RQH GHYLFH LV EHLQJ RIIVHW E\ WKH &JG GHJUDGDWLRQ LQ RWKHU GHYLFHV LQ WKH QH[W VWDJH 7KH GLIIHUHQFH EHWZHHQ WKH DERYH H[DPSOHV PD\ ORRN TXLWH VPDOO 0+] IRU WKH QRUPDO ,9RQO\ GHJUDGDWLRQ VHW YHUVXV 0+] IRU RXU ,9 DQG &9 GHJUDGDWLRQ PAGE 116 VHWf 7R SXW LW LQWR SHUVSHFWLYH FRQVLGHU WKH IROORZLQJ H[DPSOH 6XSSRVH RQH LV GHVLJQLQJ DQ 0+] SURFHVVRU ZLWK D FULWLFDO WUDQVLVWRU SDWK ZKLFK OLPLWV WKH PD[LPXP IUHTXHQF\ )XUWKHUPRUH DVVXPH WKDW WKH DFFHOHUDWHG VWUHVV DW KU UHSUHVHQWV H[DFWO\ \HDUV RI QRUPDO RSHUDWLRQ ZKLFK LV WKH VSHFLILFDWLRQ UHTXLUHG IRU WKH 0+] SURFHVVRU )LQDOO\ DVVXPH WKDW WKLV SURFHVVRU LV UHTXLUHG WR UHPDLQ ZLWKLQ b RI 0+] GXULQJ LWV \HDU OLIH 0+] WR 0+]f )LJXUH VKRZV SLFWRULDOO\ ZKDW FDQ EH GHGXFHG IURP WKH WDEOH DERYH QDPHO\ WKDW WKH 6WUHVVHG ,9 )UHVK &9 VHW LQGXVWU\ QRUPDO PHWKRGRORJ\f ZLOO UHVXOW LQ D SUHGLFWHG IDLOXUH VLQFH WKH UHVXOWLQJ VLPXODWHG IUHTXHQF\ RI 0+] LV OHVV WKDW WKH 0+] JXDUGEDQG +RZHYHU ZKHQ WKH 6WUHVVHG ,9 6WUHVVHG &9 LV XVHG WKH UHVXOWLQJ 0+] VLPXODWHG IUHTXHQF\ LV ZHOO ZLWKLQ WKH JXDUGEDQG UDQJH SUHYHQWLQJ XQQHFHVVDU\ UHGHVLJQ DQGRU VFUDS DFWXDOO\ WKH SURFHVVRU ZRXOG SUREDEO\ EH VROG DV D VORZHU YHUVLRQ DW ORZHU PDUJLQf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f W QVf ([SDQGHG YLHZ 6WUHVVHG ,9 6WUHVVHG &9 )UHVK ,9 6WUHVVHG ,9 )UHVK &9 )UHVK &9 )LJ %HJLQQLQJ IHZ F\FOHV RI D VWDJH ULQJ RVFLOODWRU FLUFXLW XVLQJ WKUHH GLIIHUHQW SDUDPHWHU VHWV 7KH KLJKHVW IUHTXHQF\ FXUYH XQVWUHVVHG ,9 XQVWUHVVHG &9f H[SHFWHGO\ FRPHV IURP WKH VLPXODWLRQ XVLQJ WKH WZR XQVWUHVVHG ,9 DQG &9 SDUDPHWHU VHWV 7KH VORZHVW IUHTXHQF\ FXUYH VWUHVVHG ,9 XQVWUHVVHG &9f FRPHV IURP WKH VLPXODWLRQ XVLQJ RQO\ WKH VWUHVVHG ,9 ZKLOH WKH PLGGOH IUHTXHQF\ FXUYH VWUHVVHG ,9 VWUHVVHG &9f FRPHV IURP XVLQJ ERWK WKH VWUHVVHG ,9 DQG &9 SDUDPHWHU VHWV 7KLV GHPRQVWUDWHV WKDW LQFOXVLRQ RI &JG GHJUDGDWLRQ UHVXOWV LQ FLUFXLW SHUIRUPDQFH LPSURYHPHQW GXH WR RIIVHWWLQJ VRPH RI WKH ,' GHJUDGDWLRQ PAGE 119 )UHTXHQF\ 0+]f )LJ ([DPSOH GHPRQVWUDWLQJ KRZ LQFOXGLQJ LQWULQVLF FDSDFLWDQFH FDQ UHVXOW LQ VXEVWDQWLDO EHQHILWV +HUH LW LV DVVXPHG WKDW WKH DFFHOHUDWHG VWUHVV RI KU LV HTXLYDOHQW WR \U RI RSHUDWLRQ IRU D ILFWLRQDO 0+] SURFHVVRU ZLWK D s b DOORZHG IUHTXHQF\ GHYLDWLRQ %\ LQFOXGLQJ &JG GHJUDGDWLRQ WKH SURFHVVRU SHUIRUPV ZLWKLQ VSHFLILFDWLRQ ZKLOH E\ QRW LQFOXGLQJ WKH LQWULQVLF FDSDFLWDQFH GHJUDGDWLRQ DV LV QRUPDOO\ GRQHf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f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fVDWXUDWLRQ YROWDJHf DQG fVXUIDFH SRWHQWLDO VHOIVDWXUDWLRQf ERWK VKRZHG WKH H[SHFWHG FKDQQHOOHQJWKPRGXODWLRQLQGXFHG VDWXUDWLRQ FXUUHQW LQFUHDVH DV WKH FKDQQHO OHQJWK IRU D VTXDUH GHYLFHf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f :LWK WKLV SRO\VLOLFRQJDWH /)&9 PRGHO LW ZDV VKRZQ WKDW WKH R[LGH WKLFNQHVV IODWEDQG YROWDJH DQG JDWH DQG VXEVWUDWH GRSLQJ FRQFHQWUDWLRQV FRXOG EH H[WUDFWHG IURP H[SHULPHQWDO FDSDFLWDQFH GDWD 7ZR H[WUDFWLRQ PHWKRGV WKH SRLQW DQG UHJLRQ ZHUH GHYHORSHG DQG ZHUH VKRZQ WR ZRUN ZHOO ZLWK Âƒ b 506 ILWf DQG VXE Âƒ b 506 ILWf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f ZLWK VWUHVV 7KH SUHVWUHVV DQG SRVW VWUHVV ,' &JG DQG &JV GDWD ZHUH ILW XVLQJ WKH %6,0 GHYLFH PRGHO VR WKDW D VWUHVVHG FLUFXLW FRXOG EH VLPXODWHG $ VLPXODWLRQ RI D VWDJH ULQJ RVFLOODWRU YHULILHG WKDW WKH GHFUHDVH LQ WKH FDSDFLWLYH ORDG IURP &JG UHGXFWLRQ SDUWLDOO\ RIIVHW WKH ,' UHGXFWLRQ UHVXOWLQJ LQ LPSURYHG VLPXODWHG SHUIRUPDQFH FRPSDUHG WR VLPXODWLRQV RQO\ WDNLQJ WKH ,' UHGXFWLRQ LQWR DFFRXQW ZKLFK LV WKH VWDQGDUG LQGXVWU\ SUDFWLFHf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f FKDUDFWHULVWLFV RI D PHWDOR[LGH VHPLFRQGXFWRU 026f FDSDFLWRU DUH IDLUO\ VWUDLJKWIRUZDUG )LJXUH $ $f VKRZV WKDW WKH WKUHH OD\HUV ZKLFK FRPSULVH WKH 026 QDPH WKH PHWDO JDWHf WKH LQVXODWRU 6L RU R[LGHf DQG WKH VHPLFRQGXFWRU VXEVWUDWHf &KDSWHU VKRZV WKH H[WHQVLRQ RI WKLV PRGHO WR SRO\VLOLFRQ JDWHV DV ZHOO DV WKH HIIHFWV RI WKH SRO\VLOLFRQ JDWH RQ WKH GHYLFH FKDUDFWHULVWLFV DV ZHOO DV WKH LPSOLFDWLRQV RQ GHYLFH SHUIRUPDQFH ,GHDOO\ WKH &9 FXUYH PRGHO ZRXOG KDYH RQH HTXDWLRQfÂ§WKH JDWH FDSDFLWDQFH &J DV D IXQFWLRQ RI WKH JDWH YROWDJH 9* $W YHU\ OHDVW LW ZRXOG EH JRRG WR KDYH D VHW RI SDUDPHWULF HTXDWLRQV ZLWK 9* DQG &J DV IXQFWLRQV RI VRPH RWKHU SDUDPHWHU VXFK DV WKH VXEVWUDWH VXUIDFH SRWHQWLDO 9,;f 'XH WR WKH FRPSOH[LWLHV RI WKH PDWKHPDWLFV WKH ODWWHU LV WKH EHVW WKDW FDQ EH GRQH PAGE 125 *DWH 3RWHQWLDO 7ZR VLPSOH HTXDWLRQV DUH UHTXLUHG RQH IRU WKH JDWH YROWDJH DQG RQH IRU WKH JDWH FDSDFLWDQFH /RRNLQJ DJDLQ DW )LJ $ $f .LUNRII V YROWDJH ODZ UHTXLUHV 9* 90 9 9,; $ODf 7KDW LV WKH YROWDJH DSSOLHG DW WKH JDWH PXVW EH HTXDO WR WKH GURSV DFURVV WKH PHWDO WKH R[LGH DQG WKH VHPLFRQGXFWRU UHVSHFWLYHO\ %HFDXVH WKH )HUPL OHYHO RI WKH PHWDO GRHV QRW QHFHVVDU\ FRLQFLGH ZLWK WKH )HUPL OHYHO RI WKH VXEVWUDWH ZKLFK LV FRQWUROOHG E\ WKH GRSDQWf WKHUH LV DQ DGGLWLRQDO WHUP G!06 WKH PHWDOWRVHPLFRQGXFWRU ZRUNIXQFWLRQ GLIIHUHQFHWR DFFRXQW IRU WKLV RIIVHW 7KLV LV EHVW YLVXDOL]HG RQ D EDQG GLDJUDP DV VKRZQ LQ )LJ $ %f IRU DQ DUELWUDU\ SRVLWLYH DSSOLHG YROWDJH ZLWK D SVXEVWUDWH 026 FDSDFLWRU 026&f )URP )LJXUH $ %f WKH YROWDJH GURSV DFURVV WKH GHYLFH DUH FOHDUO\ !P 9 ;V a 9,; (& (M-T 9) 9* $OEf ZKHUH 2P LV WKH ZRUNIXQFWLRQ IRU WKH PHWDO LV WKH HOHFWURQ DIILQLW\ IRU WKH VXEVWUDWH (F DQG ( DUH WKH FRQGXFWLRQ EDQG HGJH DQG LQWULQVLF HQHUJLHV UHVSHFWLYHO\ DQG 9) LV WKH )HUPL YROWDJH ZKLFK LV HTXLYDOHQW WR (> )SfT &ROOHFWLQJ WKHVH WHUPV LQ D IRUP PRUH OLNH WKH SUHYLRXV HTXDWLRQ JLYHV 9R 9 9,; Wf06 $OFf ZKHUH G!06 p0 2V W!0 \V T(F (>f 9)f ,Q (T $ OD D YROWDJH GURS DFURVV WKH PHWDO ZDV LQFOXGHG ZKLFK ZDV SXUSRVHO\ QHJOHFWHG LQ WKH EDQG GLDJUDP DQG KHQFH LQ (T $OE ,GHDOO\ WKH YROWDJH GURS DFURVV WKH JDWH ZLOO EH 29 :KHQ D PHWDO LV XVHG WKH GURS LV HIIHFWLYHO\ 29 3RO\VLOLFRQ JDWHV LQWURGXFH D GHSOHWLRQ OD\HU ZKLFK FDXVHV PAGE 126 9* )LJ $O 026 FDSDFLWRU VFKHPDWLF DQG FRUUHVSRQGLQJ HQHUJ\EDQG GLDJUDP $f 6FKHPDWLF GLDJUDP RI D 026 FDSDFLWRU DQG %f FRUUHVSRQGLQJ HQHUJ\EDQG GLDJUDP GHSLFWLQJ WKH SRWHQWLDO GURSV 6KRZQ LV D SRVLWLYH YROWDJH 9 DSSOLHG DW WKH JDWH UHVXOWLQJ LQ WKH 6L6L VXUIDFH HQWHULQJ LQYHUVLRQ PAGE 127 D YROWDJH GURS DV ZHOO DV H[WUD FDSDFLWDQFH &KDSWHU GHDOV ZLWK WKLV LPSRUWDQW HIIHFW ZKLOH LQ WKLV DSSHQGL[ LW ZLOO VLPSO\ EH DVVXPHG WKDW WKH PHWDO LV D SHUIHFW FRQGXFWRU WKXV 90 9 DQG LV QHJOHFWHG LQ (T $OF 7KH GURS DFURVV WKH LQVXODWRU 9 ZKHUH WKH f2f LV IRU R[LGH VLQFH 6L LV WKH SUHYDOHQW LQVXODWRU IRU VLOLFRQ GHYLFHVf ZLOO EH IRXQG LQ WKH QH[W VHFWLRQ 7KH SRWHQWLDO DFURVV WKH VHPLFRQGXFWRU WKH VXUIDFH SRWHQWLDO 9O[ FDQQRW EH HDVLO\ PHDVXUHG RU IRXQG 7KLV ZLOO EH WKH XQNQRZQ YDULDEOH ZKLFK UHODWHV $ &f DQG &J IRUPXOD )URP D VWULFWO\ PDWKHPDWLFDO SRLQW RI YLHZ 9,; RU WKH QRUPDOL]HG HTXLYDOHQW 8[f LV WKH SDUDPHWULF YDULDEOH IRU WKH WZR HTXDWLRQV 9* DQG &Jf 2[LGH 3RWHQWLDO &KDUJH QHXWUDOLW\ JXDUDQWHHV WKDW WKH VXP RI WKH FKDUJHV WKURXJK WKH FLUFXLW LQ )LJ $ $f LV ]HUR 7KXV 4J 4RW 4LW 4V $f 4J WKH JDWH FKDUJH LV HTXDO WR WKH FKDUJH DW WKH JDWHLQVXODWRU LQWHUIDFH E\ *DXVVfV WKHRUHPf VR 4* OH(O e97R[ &*9 6LPLODUO\ 4V HV(,; DW WKH LQVXODWRUVLOLFRQ LQWHUIDFH ZKHUH ( LV WKH HOHFWULF ILHOG DW WKH JDWHR[LGH LQWHUIDFH ([ LV WKH HOHFWULF ILHOG DW WKH R[LGHVXEVWUDWH LQWHUIDFH 7R[ LV WKH R[LGH WKLFNQHVV DQG & LV WKH FRQVWDQW LQVXODWRU R[LGHf FDSDFLWDQFH 4R[ DQG 4,7 UHSUHVHQW WKH WUDSSHG FKDUJH DQG LQWHUIDFH FKDUJH UHVSHFWLYHO\ DQG H LV WKH GLHOHFWULF FRQVWDQW RI WKH LQVXODWRU 6XEVWLWXWLQJ WKH DERYH WZR UHODWLRQVKLSV LQWR $f DQG VROYLQJ IRU WKH R[LGH LQVXODWRUf SRWHQWLDO JLYHV Y mV(L[&R f 4RW 4LWf4T $f PAGE 128 :LWK WKLV UHODWLRQ $OFf FDQ EH UHZULWWHQ DV 9T a >pPV 4TW 4LWA&@ YL[ eVpL[AR $f 9)E 9L[ HV(O[& $f ZKHUH 9)% WKH IODWEDQG YROWDJH LV JLYHQ E\ 9)% !PV 4TW 4LWf & $f 7KH WHUPV RI $f DUH DOPRVW DOO NQRZQ HV FDQ EH IRXQG LQ D KDQGERRN & LV D FRQVWDQW DW D FRQVWDQW WHPSHUDWXUHf DQG LQ WKH FDVH RI GHULYLQJ WKLV IRUPXOD FDQ EH DVVXPHG NQRZQ 9_; LV WKH SDUDPHWULF YDULDEOH GLVFXVVHG HDUOLHU VR WKHUH LV RQO\ RQH XQNQRZQ YDULDEOH (,; WKH ILHOG DFURVV WKH VHPLFRQGXFWRU %HIRUH WKLV SUREOHP LV UHFWLILHG WKH FDSDFLWDQFH DVSHFW RI WKH &9 GHULYDWLRQ ZLOO EH LQYHVWLJDWHG *DWH &DSDFLWDQFH /RRNLQJ DJDLQ DW )LJ $O $f WKH JDWH FDSDFLWDQFH &J VHHPV DOPRVW WULYLDO ,W LV VLPSO\ WKH VHULDO FRPELQDWLRQ RI WKH PHWDO JDWHFRQWDFWf FDSDFLWDQFH DVVXPHG WR EH LQILQLWH VLQFH D FRQGXFWRU KDV QR VSDFH FKDUJH OD\HU ZLGWKf WKH LQVXODWRU FDSDFLWDQFH &f DQG WKH VHPLFRQGXFWRU FDSDFLWDQFH &L[f DV VKRZQ LQ )LJ $ ,% 7KH fL[f VXEVFULSW GHVLJQDWHV WKH EDQG EHQGLQJ IURP WKH LQWHUIDFH Lf WR WKH VXEVWUDWH [f ZKLFK LV DQ LPSRUWDQW GHVLJQDWLRQ ZKHQ SRO\VLOLFRQ JDWHV DUH XVHG +HQFH WKH JDWH FDSDFLWDQFH LV &J &T &L[ VHULDO FDSDFLWDQFH VXPPDWLRQf $f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f VWDWLVWLFV 7R ILQG WKH FDUULHU FRQFHQWUDWLRQ LQ D VHPLFRQGXFWRU DV D IXQFWLRQ RI HQHUJ\ WZR WKLQJV DUH QHHGHG WKH WKUHHGLPHQVLRQDO GHQVLW\ RI VWDWHV '(f DQG WKH GLVWULEXWLRQ IXQFWLRQ (f 7KH UHODWLRQVKLS EHWZHHQ WKH GHQVLW\ RI VWDWHV DQG HQHUJ\ LV DSSUR[LPDWHO\ SDUDEROLF QHDU WKH ERWWRP LQ (N VSDFHf RI WKH FRQGXFWLRQ EDQG 7KH FRQGXFWLRQ HOHFWURQV ZLOO WHQG WR EH QHDU WKLV PLQLPD DOORZLQJ XV WR XVH WKH IROORZLQJ SDUDEROLF IRUPXOD >@ '(f >7PrKf@M ( (FfG( $f 7KH ZHOONQRZQ )HUPL'LUDF RFFXSDWLRQ IXQFWLRQ LV JLYHQ E\ (f ^ H[S > ()(fN7@ ` $f ZKHUH () LV WKH )HUPL OHYHO ([DPLQDWLRQ RI $f VKRZV WKDW ( ()f 7KXV WKH )HUPL OHYHO LV WKH HQHUJ\ ZKHUH KDOI RI WKH WRWDO HOHFWURQV DUH FRQWDLQHG LQ WKH OHYHOV EHORZ () $ IRUPXOD IRU WKH )HUPL OHYHO LV JLYHQ ODWHU >$f DQG $f@ PAGE 130 (OHFWURQ &RQFHQWUDWLRQ %\ LQWHJUDWLQJ WKH SURGXFW RI $f DQG $f RYHU WKH HQHUJ\ IURP WKH ERWWRP RI WKH FRQGXFWLRQ EDQG (Ff WR IUHH VSDFH (9/ >YDFXXP OHYHO@f WKH QXPEHU RI FRQGXFWLRQ HOHFWURQV FDQ EH IRXQG Q A9/ I(f'(fG( (F $ f PH HYO ( (F K (F ^ H[S>( (IfN7@` 6XEVWLWXWLQJ H ( (FfN7 G( N7GHf U () (FfN7 DQG QRWLQJ WKDW H9/ } ZKLFK PHDQV WKH XSSHU LQWHJUDO UDQJH FDQ EH DSSUR[LPDWHG E\ LQILQLW\f UHVXOWV LQ Q Q PHN7 RR eV Ge K R 8 H[S > e U@ @ ` f7LQHN7 e Ge K ^ H[S >e WM@ ` 1f $ f 1F NQRZQ DV WKH HIIHFWLYH GHQVLW\ RI FRQGXFWLRQ EDQG VWDWHV LV JLYHQ E\ 7QUXN7 [OPHPf f FP LV WKH )HUPL'LUDF LQWHJUDO RI WKH nK RUGHU ZKLFK LV VKRUWKDQG IRU rnI! fÂ§ ? Q GH ^ H[S > H WM @` $ f $ f $ Df $Ef $ f PAGE 131 :K\ LV WKHUH D ? Q WHUP LQ IURQW RI WKH LQWHJUDO" 7KLV QRWDWLRQ RI WKH )HUPL'LUDF )'f LQWHJUDO SURSRVHG E\ 'LQJOH >@ LV IURP D IDPLO\ RI )' LQWHJUDOV RI WKH IRUP Uf (M Ge GL"f f $ f U MOf ^ H[S>H IM@ ` 7KXV WKH O? Q WHUP FRPHV IURP Ufr 7KH 'LQJOH QRWDWLRQ KDV D QXPEHU RI EHQHILFLDO SURSHUWLHV FRPSDUHG ZLWK WKH RWKHU )' QRWDWLRQDO IDPLO\ FDOOHG 6RPPHUIHOG QRWDWLRQ ZKLFK GLIIHUV RQO\ E\ WKH IDFWRU RI 7MOf 7KH PRVW XVHIXO 'LQJOHQRWDWLRQf SURSHUW\ ZKHQ LW FRPHV WR GHYLFH SK\VLFV LV G fÂ§L: cU+LLf D f G Q ZKLFK PDNHV GLIIHUHQWLDWLQJ DQG LQWHJUDWLQJ )' LQWHJUDOV TXLWH VLPSOH DV ZLOO EH VHHQ ODWHU :KHQ WKLV 'LQJOH QRWDWLRQ LV XVHG ZRUNLQJ ZLWK )' LQWHJUDOV EHFRPHV DOPRVW DV HDV\ DV XVLQJ H[SRQHQWLDOV LH WKH RXWFRPH RI XVLQJ WKH %ROW]PDQQ GLVWULEXWLRQ IXQFWLRQ LQVWHDG RI WKH )HUPL'LUDF GLVWULEXWLRQ IXQFWLRQ ZKHQ GHULYLQJ WKH FDUULHU FRQFHQWUDWLRQf :KHQ 6RPPHUIHOG QRWDWLRQ LV XVHG LQVWHDG RI 'LQJOH GLIIHUHQWLDWLRQ UHVXOWV LQ D UMOfUMf PXOWLSOLFDWLYH WHUP LQ $f +ROH &RQFHQWUDWLRQ 7KH GHULYDWLRQ IRU WKH KROH FRQFHQWUDWLRQ LV LGHQWLFDO WR WKDW RI WKH HOHFWURQ FRQFHQWUDWLRQ H[FHSW WKDW WKH GHQVLW\ RI VWDWHV HTXDWLRQ LV UHIHUHQFHG IURP WKH YDOHQFH EDQG DQG XVHV WKH HIIHFWLYH KROH PDVV 7KH RFFXSDWLRQ IXQFWLRQ IRU KROHV LV (f@ RU LQ RWKHU ZRUGV WKH KROHV DUH ZKHUH WKH HOHFWURQV DUH QRW '(f >7PKKf @ W_ (\ (fG( $ f PAGE 132 (\ ^ (f ` (f G( $f (Yn (Y DQG (Yf DUH WKH WRS DQG ERWWRP RI WKH YDODQFH EDQG UHVSHFWLYHO\ 8VLQJ WKH VDPH PHWKRGV DV DERYH DQG PDNLQJ WKH VDPH ZLGHEDQG DSSUR[LPDWLRQ WKH IROORZLQJ HTXDWLRQ UHVXOWV WUPKN7 Hn GV K f8 R ^ H[S>en Tf@` $f 1Ye8fnf $f H n LV (Y (fN7 DQG Un LV (Y ()fN7 ^ (JN7 UMf` 7KLV LV D GLIIHUHQW DOWKRXJK HTXLYDOHQW SUHVHQWDWLRQ WKDQ RWKHUV KDYH XVH >W\SLFDOO\ S 1Y@L Ue*f@ DQG EHWWHU VKRZV WKH V\PPHWU\ RI WKH HOHFWURQ DQG KROH GLVWULEXWLRQV 7KH 1Y WHUP NQRZQ DV WKH HIIHFWLYH GHQVLW\ RI YDODQFH EDQG VWDWHV LV JLYHQ E\ WPKN7 L 1Y $Df K [OPKPf7f FPa $Ef &RPSDULQJ $Ef WR $ Ef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f &KDUJH 'HQVLW\ 7KH FKDUJH GHQVLW\ LQ WKH VHPLFRQGXFWRU LV JLYHQ E\ WKH HTXDWLRQ S T1 3 1D 3' Q7f $f 7KH 1 DQG 3 WHUPV DUH DV JLYHQ DERYH $ DQG $f DQG WKH 1$ DQG 3' WHUPV DUH WKH LRQL]HG DFFHSWRUV DQG GRQRUV UHVSHFWLYHO\ 7KH Q7 WHUP UHSUHVHQWV WKH FRQWULEXWLRQ IURP WUDSSHG FKDUJH *HQHUDOO\ LW LV DVVXPHG WKDW DOO RI WKH LPSXULWLHV DUH FRPSOHWHO\ LRQL]HG LQ GRSHG VLOLFRQ EHFDXVH VKDOORZ OHYHO LPSXULWLHV DUH XVHG +RZHYHU DW ORZ WHPSHUDWXUHV DQGRU KLJK GRSLQJ FPnf LQFRPSOHWH LPSXULW\ LRQL]DWLRQ RFFXUV DQG WKH DSSUR[LPDWLRQ LV QR ORQJHU YDOLG )RU GHHS OHYHO LPSXULWLHV GHLRQL]DWLRQ ZLOO EHFRPH VLJQLILFDQW HYHQ DW PRGHUDWH GRSLQJ DQG URRP WHPSHUDWXUH )RU SW\SH LPSXULWLHV WKH UDWLR RI HPSW\ DFFHSWRUV 1$f WR ILOOHG DFFHSWRUV 1$f LV 1$1r OJ$fH[S >(S ($@ N7f $f ZKHUH J$ LV WKH GHJHQHUDF\ IDFWRU DQG ($ LV WKH DFFHSWRU HQHUJ\ OHYHO 1RWLQJ WKDW 1$$ 1D 1D $f FDQ EH WUDQVIRUPHG LQWR QDD 1D 1 $f J$H[S>($ ()@ N7f $ VLPLODU HTXDWLRQ FDQ EH GHULYHG IRU QW\SH LPSXULWLHV DQG LV JLYHQ E\ PAGE 134 1GG 1G 1e $f J'H[S>() (G@ N7f 7KHVH WZR HTXDWLRQV WDNH GHLRQL]DWLRQ LQWR DFFRXQW *HQHUDOO\ LW LV DVVXPHG WKDW LPSXULWLHV DUH FRPSOHWHO\ LRQL]HG ZKLFK LQ SW\SH PDWHULDO LPSOLHV 1$ 1$$ 7KLV LV D JRRG DSSUR[LPDWLRQ ZKHQ 7 LV ODUJH RU () } ($ $W YHU\ ORZ WHPSHUDWXUHV J$H[S>($()@N7f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f ,PSXULW\ EDQGLQJ FDQ DOVR RFFXU EXW VLQFH DQ XQGHUO\LQJ DVVXPSWLRQ RI WKHVH GHULYDWLRQV LV XQLIRUP GRSLQJ LPSXULW\ EDQGLQJ ZLOO EH QHJOHFWHG $VVXPLQJ WKDW WKH VHPLFRQGXFWRU KDV QHJOLJLEOH WUDSSLQJ WKHQ $f FKDUJH GHQVLW\f EHFRPHV S T3 1 1D 3'f $f 6HPLFRQGXFWRU (OHFWULF )LHOG :LWK $f FDUULHU FRQFHQWUDWLRQ IRUPXODH $ f DQG $f DQG LRQL]HG LPSXULW\ FRQFHQWUDWLRQV $f DQG $f WKH FKDUJH GHQVLW\ FDQ EH ZULWWHQ DV D IXQFWLRQ PAGE 135 RI HQHUJ\ 8VLQJ WKLV DQ HTXDWLRQ IRU WKH VHPLFRQGXFWRU ILHOG (,; FDQ EH IRXQG YLD 3RLVVRQfV HTXDWLRQ 6WDUWLQJ IURP WKH GF VWHDG\VWDWH 3RLVVRQ HTXDWLRQ LQ RQH GLPHQVLRQ eVG(G[ S $f ZKHUH eV LV WKH GLHOHFWULF fFRQVWDQWf RI VLOLFRQ ( LV WKH HOHFWULF ILHOG DQG S LV DV JLYHQ LQ $f ,QWHJUDWLQJ E\ TXDGUDWXUH QRWLQJ ( G9G[f JLYHV >@ eVG(G[ eVGG[fG9G[f HD>G9G[fGG9f@G9G[f eVf GG9f G9G[f eVf G(G9f $f 7KXV IURP $f DQG $f G( eVf SGY $f (DFK VLGH RI WKH HTXDWLRQ FDQ EH LQWHJUDWHG IURP WKH VXUIDFH WR ]HUR ([ WR IRU WKH ILHOG WHUP DQG 8,; WR IRU WKH FKDUJH GHQVLW\ WHUPf QRWLQJ WKDW 9 N7Tf8 >DQG G9 N7TfG8@ N7 ,,,; 8Y8)f L 8\8Sf @ AFLAA 8L[W8FfÂ§8Sf f f" 8Sf @ J$H[S 8) fÂ§ 8$ fÂ§ 8,;f 1DD 8L[ ,Q 8W[ ,Q J$H[S 8) 8$f J'H[S8' 8S 8,;f J'H[S8' 8)f $f PAGE 136 8VH ZDV PDGH RI $ f WR LQWHJUDWH WKH )' IXQFWLRQV 2QH PLJKW ZRQGHU KRZ WKH 8f[ WHUP HQGHG XS LQ $f ZKHQ WKHUH ZDV QR IUHHYDULDEOH 8 RU 9f LQ WKH RULJLQDO HTXDWLRQV IRU WKH FDUULHU FRQFHQWUDWLRQV RU WKH LRQL]HG LPSXULWLHV 7KH VXUIDFH SRWHQWLDO 8,; UHSUHVHQWV WKH DPRXQW RI DGGLWLRQDO EDQG EHQGLQJ RI WKH VLOLFRQ EDQG DW WKH 6L6L LQWHUIDFH FDXVHG E\ WKH DSSOLHG ILHOG ,W ZDV ]HUR LQ WKH HTXLOLEULXP ]HURILHOG VWDWH DQG QRW LQFOXGHG LQ WKH HTXDWLRQV 6HPLFRQGXFWRU &DSDFLWDQFH )LQDOO\ WKH VHPLFRQGXFWRU FDSDFLWDQFH &L[ LV QHHGHG )RU ORZIUHTXHQF\ WKLV LV JLYHQ E\ &M[ 38,;f(,; $f 7KH ORZ IUHTXHQF\ DVVXPSWLRQ PHDQV WKDW WKH PLQRULW\ FDUULHUV FDQ EH JHQHUDWHG TXLFNO\ HQRXJK WR IROORZ WKH VPDOOVLJQDO YROWDJH )RU WUDQVLVWRUV 0267Vf /)&9 FXUYHV UHVXOW HYHQ ZKHQ KLJK 0+]f IUHTXHQFLHV DUH XVHG LQ &9 PHDVXUHPHQWV EHFDXVH WKH KLJKO\ GRSHGf VRXUFH DQG GUDLQ ZLOO VXSSO\ WKH QHFHVVDU\ PLQRULW\ FDUULHUV WR IROORZ WKH DF VLJQDO DV ORQJ DV WKH FKDQQHO LV VKRUW HQRXJK 6XEVWLWXWLQJ $ f $f $f DQG $f LQWR $Qf RU $Sf DQG WKHQ SODFLQJ WKH UHVXOW LQ $f JLYHV T L[ $f PAGE 137 $ GHJHQHUDWH GHLRQL]HG /) &9 FXUYH FDQ QRZ EH JHQHUDWHG E\ VXEVWLWXWLQJ $f LQWR $f >WKH JDWH YROWDJH IRUPXOD@ DQG $f LQWR $f >WKH JDWH FDSDFLWDQFH IRUPXOD@ ZKHUH 9* DQG &J DUH JHQHUDWHG LQ SDLUV DV D IXQFWLRQ RI 9,; RU 8,;f )HUPL /HYHO 7KURXJKRXW WKLV GHULYDWLRQ LW KDV EHHQ DVVXPHG WKDW WKH )HUPL OHYHO LV NQRZQ &RQVLGHULQJ WKH GHJHQHUDWH LRQL]HG SW\SH FDVH ILUVW $f JLYH DVVXPLQJ 1$$}QM QDD ()@N7f $f )RU WKH LQWULQVLF FDVH QL 1\AD8(Y (-N7f $f 6ROYLQJ WKHVH WZR FDVHV IRU WKH )HUPL OHYHO UHODWLYH WR WKH LQWULQVLF OHYHO JLYHV () (M N7 > +L1Yf \ 1A1Yf @ $f ZKHUH \A LV WKH LQYHUVH )HUPL'LUDF LQWHJUDO RI WKH nK RUGHU 7KH LQYHUVH )HUPL'LUDF LQWHJUDO LV DQDORJRXV WR D QDWXUDO ORJDULWKP LQ WKH %ROW]PDQQ UHJLPH ,W LV LPSRUWDQW WR QRWH WKDW XQOLNH ORJDULWKPV \A ; PAGE 138 5()(5(1&(6 >@ & 7 6DK (YROXWLRQ RI WKH 026 7UDQVLVWRUfÂ§)URP &RQFHSWLRQ WR 9/6, 3URF _((( f >@ & 7 6DK )XQGDPHQWDOV RI 6ROLG6WDWH (OHFWURQLFV 6WXG\ *XLGH 6LQJDSRUH :RUOG 6FLHQWLILF f 6HH 6HFWLRQ DQG $SSHQGL[ % >@ 6 6]H 3K\VLFV RI 6HPLFRQGXFWRU 'HYLFHV 1HZ PAGE 139 >@ + ,KDQWROD 6WDQIRUG (OHFWURQLFV /DERUDWRU\ 5HSW 1R 6HSWHPEHU f >@ & 7 6DK &KDUDFWHULVWLFV RI WKH 0267 ,((( 7UDQV (OHFWURQ 'HY (' f >@ + ,KDQWROD DQG / 0ROO 'HVLJQ 7KHRU\ RI D 6XUIDFH )LHOG(IIHFW 7UDQVLVWRU 6ROLG6WDWH (OHFWURQ f >@ 5 7URXWPDQ 9/6, /LPLWDWLRQ IURP 'UDLQ,QGXFHG %DUULHU /RZHULQJ ,((( 7UDQV (OHFWURQ 'HY (' f >@ + & 3DR DQG & 7 6DK (IIHFWV RI 'LIIXVLRQ &XUUHQW RQ &KDUDFWHULVWLFV RI 0267V 6ROLG6WDWH (OHFWURQ f >@ 5 ) 3LHUUHW DQG $ 6KLHOGV 6LPSOLILHG /RQJ&KDQQHO 026)(7 7KHRU\ 6ROLG6WDWH (OHFWURQ f >@ 0 3HUVL DQG *LOGHQEODW &RPSXWDWLRQDOO\ (IILFLHQW 9HUVLRQ RI WKH 3DR6DK 0RGHO ZLWK 9DULDEOH 0RELOLW\ 6ROLG6WDWH (OHFWURQ f >@ & 7 6DK )XQGDPHQWDOV RI 6ROLG6WDWH (OHFWURQLFV 6ROXWLRQ 0DQXDO 6LQJDSRUH :RUOG 6FLHQWLILF f 6HH $SSHQGL[ >@ & 7 6DK DQG + & 3DR 7KH (IIHFWV RI )L[HG %XON &KDUJH RQ WKH &KDUDFWHULVWLFV RI 0267V ,((( 7UDQV (OHFWURQ 'HY (' f >@ 0 % %DUURQ /RZ/HYHO &XUUHQWV LQ ,QVXODWHG *DWH )LHOG (IIHFW 7UDQVLVWRUV 6ROLG6WDWH (OHFWURQ f >@ 5 9DQ 2YHUVWUDHWHQ 'HFOHUFN DQG / %URX[ ,QDGHTXDF\ RI WKH &ODVVLFDO 7KHRU\ RI WKH 0267 2SHUDWLQJ LQ :HDN ,QYHUVLRQ ,((( 7UDQV (OHFWURQ 'HY (' f >@ 5 %UHZV $ &KDUJH6KHHW 0RGHO RI WKH 026)(7 6ROLG6WDWH (OHFWURQ f >@ ) 9DQ GH :LHOH $ /RQJ&KDQQHO 026)(7 0RGHO 6ROLG6WDWH (OHFWURQ f PAGE 140 >@ $ 1XVVEDXP 5 6LQKD 'RNRV 7KH 7KHRU\ RI WKH /RQJ&KDQQHO 026)(7 6ROLG6WDWH (OHFWURQ f >@ 5 6FKULPSI + -X DQG 5 0 :DUQHU -U 6XEWKUHVKROG 7UDQVFRQGXFWDQFH LQ WKH /RQJ&KDQQHO 026)(7 6ROLG6WDWH (OHFWURQ f >@ 9 5HGGL DQG & 7 6DK 6RXUFH WR 'UDLQ 5HVLVWDQFH %H\RQG 3LQFK2II LQ 0267V ,((( 7UDQV (OHFWURQ 'HY (' f >@ 7 / &KLX DQG & 7 6DK &RUUHODWLRQ RI ([SHULPHQWV :LWK D 7ZR6HFWLRQ 0RGHO 7KHRU\ RI WKH WKH 6DWXUDWLRQ 'UDLQ &RQGXFWDQFH RI 0267V 6ROLG6WDWH (OHFWURQ f >@ )URKPDQ%HQWFKNRZVN\ DQG $ 6 *URYH &RQGXFWDQFH RI 026 7UDQVLVWRUV LQ 6DWXUDWLRQ ,((( 7UDQV (OHFWURQ 'HY (' f >@ 0HUFNHO %RUHO DQG 1 = &XSFHD $Q $FFXUDWH /DUJH6LJQDO 026 7UDQVLVWRU 0RGHO IRU 8VH LQ &$' ,((( 7UDQV (OHFWURQ 'HY (' f >@ $ 3RSD $Q ,QMHFWLRQ /HYHO 'HSHQGHQW 7KHRU\ RI WKH 026 7UDQVLVWRU LQ 6DWXUDWLRQ ,((( 7UDQV (OHFWURQ 'HY (' f >@ 3 3 *XHEHOV DQG ) 9DQ GH :LHOH $ 6PDOO *HRPHWU\ 026)(7 0RGHO IRU &$' $SSOLFDWLRQV 6ROLG6WDWH (OHFWURQ f >@ % 6KHX / 6FKDUIHWWHU 3 .R 0 & -HQJ %6,0 %HUNHOH\ 6KRUW &KDQQHO ,*)(7 0RGHO IRU 026 7UDQVLVWRUV ,((( 6ROLG6WDWH &LUFXLWV 6& f >@ -+ +XDQJ = + /LX 0 & -HQJ +XL 0 &KDQ 3 .R DQG & +X %6,0 0DQXDO YHUVLRQ f 8& %HUNHOH\ f >@ 0 0LXUD0DWWDXVFK DQG + -DFREV $QDO\WLFDO 0RGHO IRU &LUFXLW 6LPXODWLRQ ZLWK 4XDUWHU 0LFURQ 026)(7 6XEWKUHVKROG &KDUDFWHULVWLFV -DSDQHVH $SSO 3K\VLFV / f >@ 0 0LXUD0DWWDXVFK $QDO\WLFDO 026)(7 0RGHO IRU 4XDUWHU 0LFURQ 7HFKQRORJLHV ,((( 7UDQV &$' f PAGE 141 >@ 0 0LXUD0DWWDXVFK 8 )HOGPDQQ $ 5DKP 0 %ROOX DQG 6DYLJQDF 8QLILHG &RPSOHWH 026)(7 0RGHO IRU $QDO\VLV RI 'LJLWDO DQG $QDORJ &LUFXLWVf ,))) 7UDQV &$' f >@ + & + 3DR 7KHRUHWLFDO DQG ([SHULPHQWDO ,QYHVWLJDWLRQV RI 0267Vf 6ROLG 6WDWH (OHFWURQLFV /DERUDWRU\ 7HFKQLFDO 5HSRUW 8QLYHUVLW\ RI ,OOLQRLV f >@ + .DWWR DQG < ,WRK $QDO\WLFDO ([SUHVVLRQV IRU WKH 6WDWLF 026 7UDQVLVWRU &KDUDFWHULVWLFV %DVHG RQ WKH *UDGXDO &KDQQHO 0RGHO 6ROLG6WDWH (OHFWURQ f >@ 6 6HOEHUKHUU $ 6FKW] DQG + 3W]O 0,1,026 $ 7ZR'LPHQVLRQDO 026 7UDQVLVWRU $QDO\]HU ,((( 7UDQV (OHFWURQ 'HY (' f >@ 6DUDFH 5 .HUZLQ .OHLQ DQG 5 (GZDUGV 0HWDO1LWULGH2[LGH6LOLFRQ )LHOG(IIHFW 7UDQVLVWRUV ZLWK 6HOI$OLJQHG *DWHV 6ROLG6WDWH (OHFWURQ f >@ 5 %RZHU DQG 5 'LOO ,QVXODWHG *DWH )LHOG (IIHFW 7UDQVLVWRUV )DEULFDWHG 8VLQJ WKH *DWH DV 6RXUFH'UDLQ 0DVN ,((( ,('0 f >@ & 7 6DK )XQGDPHQWDOV RI 6ROLG 6WDWH (OHFWURQLFV 6LQJDSRUH :RUOG 6FLHQWLILF f >@ 5 6HLZDW] DQG 0 *UHHQ 6SDFH &KDUJH &DOFXODWLRQV IRU 6HPLFRQGXFWRUV / $SSO 3KYV f >@ : +XQWHU 7KLQ 2[LGH 026 &DSDFLWRU 6WXGLHV RI )DVW 6XUIDFH 6WDWHV 7HFKQLFDO 5HSRUW 6ROLG6WDWH (OHFWURQLFV /DERUDWRU\ 8QLYHUVLW\ RI ,OOLQRLV 8UEDQD f >@ 6 :DOVWUD &9%DVHG 2[LGH 7KLFNQHVV ([WUDSRODWLRQ 3URFHGXUH IRU 7KLQ2[LGH 026 'HYLFHV 06(( 7KHVLV 8QLYHUVLW\ RI )ORULGD *DLQHVYLOOH $XJXVW f >@ 0 0F1XWW DQG & 7 6DK 'HWHUPLQDWLRQ RI WKH 026 2[LGH &DSDFLWDQFH $SSO 3KYV f >@ PAGE 142 >@ 6: /HH & /LDQJ &6 3DQ : /LQ DQG 0DUN $ 6WXG\ RQ WKH 3K\VLFDO 0HFKDQLVP LQ WKH 5HFRYHU\ RI *DWH &DSDFLWDQFH WR &R[ LQ ,PSODQWHG 3RO\VLOLFRQ 026 6WUXFWXUHV ,((( (OHFWURQ 'HYLFH /HWWHUV f >@ 5 &KDSPDQ & :HL %HOO 6 $XU %URZQ 5 +DNHQ 0LFURQ &026 IRU +LJK 3HUIRUPDQFH DW 9 ,((( ,('0 f >@ & :RQJ 6XQ < 7DXU & 2K 5 $QJHOXFFL DQG % 'DYDUL 'RSLQJ RI 1 DQG 3 3RO\VLOLFRQ LQ D 'XDO*DWH &026 3URFHVV ,((( ,('0 f >@ &< /X 6XQJ + .LUVFK 6 +LOOHQLXV 7 6PLWK / 0DQFKDQGD $QRPDORXV &9 &KDUDFWHULVWLFV RI ,PSODQWHG 3RO\ 026 6WUXFWXUH LQ QS 'XDO*DWH &026 7HFKQRORJ\ ,((( (OHFWURQ 'HYLFH /HWWHUV f >@ &KHQ 0 &KDQ 3 .R & +X -+ +XDQJ 3RO\VLOLFRQ *DWH 'HSOHWLRQ (IIHFW RQ ,& 3HUIRUPDQFH 6ROLG6WDWH (OHFWURQ f >@ 0DVHUMLDQ 3HWHUVVRQ DQG & 6YHQVVRQ 6DWXUDWLRQ FDSDFLWDQFH RI WKLQ R[LGH 026 VWUXFWXUHV DQG WKH HIIHFWLYH VXUIDFH GHQVLW\ RI VWDWHV RI VLOLFRQ 6ROLG6WDWH (OHFWURQ f >@ % 5LFF 3 2OLYR 7 1JX\HQ 7 6 .XDQ DQG )HUULDQL 2[LGH7KLFNQHVV 'HWHUPLQDWLRQ LQ 7KLQ,QVXODWRU 026 6WUXFWXUHVf ,((( 7UDQV (OHFWURQ 'HY f >@ 6 :DOVWUD DQG &7 6DK 7KLQ 2[LGH 7KLFNQHVV ([WUDSRODWLRQ IURP &DSDFLWDQFH 9ROWDJH 0HDVXUHPHQWV ,((( 7UDQV (OHFWURQ 'HY f >@ 6 :DOVWUD DQG &7 6DK ([WHQVLRQ RI WKH 0F1XWW6DK 0HWKRG IRU 0HDVXULQJ 7KLQ 2[LGH 7KLFNQHVVHV RQ 026 'HYLFHV DFFHSWHG IRU SXEOLFDWLRQ LQ 6ROLG 6WDWH (OHFWURQ f >@ & 2 7KRPDV .DKQJ DQG 5 & 0DQ] ,PSXULW\ 'LVWULEXWLRQ LQ (SLWD[LDO 6LOLFRQ )LOPV (OHFWURFKHP 6RF f >@ : -RKQVRQ DQG 3 3DQRXVLV 7KH ,QIOXHQFH RI 'HE\H /HQJWK RQ WKH &9 0HDVXUHPHQW RI 'RSLQJ 3URILOHV ,((( 7UDQV (OHFWURQ 'HY f >@ /HKRYHF &9 3URILOLQJ RI 6WHHS 'RSDQW 'LVWULEXWLRQV 6ROLG6WDWH (OHFWURQ f PAGE 143 >@ ) 6WHUQ DQG : +RZDUG 3URSHUWLHV RI 6HPLFRQGXFWRU 6XUIDFH ,QYHUVLRQ /D\HUV LQ WKH (OHFWULF 4XDQWXP /LPLWf 3K\VLFDO 5HYLHZ f >@ 7 $QGR $ )RZOHU DQG ) 6WHUQ (OHFWURQLF 3URSHUWLHV RI 7ZR'LPHQVLRQDO 6\VWHPV 5HY 0RGHUQ 3K\VLFV f >@ 6X 3 2OLYR DQG % 5LFF 4XDQWXP0HFKDQLFDO 0RGHOLQJ RI $FFXPXODWLRQ /D\HUV LQ 026 6WUXFWXUH ,((( 7UDQV (OHFWURQ 'HY f >@ 0 9DQ 'RUW 3 :RHUOHH DQG $ :DONHU $ 6LPSOH 0RGHO IRU 4XDQWLVDWLRQ (IIHFWV LQ +HDYLO\'RSHG 6LOLFRQ 026)(7V DW ,QYHUVLRQ &RQGLWLRQV 6ROLG6W (OHFWURQ f >@ 6 &KRQJVDZDQJYLURG ( ,UHQH $ .DOQLWVN\ 6 7D\ DQG (OOXO 5HIUDFWLYH ,QGH[ 3URILOHV RI 7KHUPDOO\ *URZQ DQG &KHPLFDOO\ 9DSRU 'HSRVLWHG )LOPV RQ 6LOLFRQ (OHFWURFKHP 6RF f >@ 6& .DR DQG 5 'RUHPXV 6LOLFD )LOPV RQ 6LOLFRQ 7KLFNQHVV 0HDVXUHPHQW E\ (OHFWURQ 0LFURVFRS\ DQG (OOLSVRPHWU\ (OHFWURFKHP 6RF f >@ ( :X 6 /R : $EDGHHU $ $FRYLF %LXFKDQDQ 7 )XUXNDZD %URFKX DQG 5 'XIUHVQH 'HWHUPLQDWLRQ RI 8OWUD7KLQ 2[LGH 9ROWDJHV DQG 7KLFNQHVV DQG WKH ,PSDFW RQ 5HOLDELOLW\ 3URMHFWLRQ ,((( ,536 f >@ 6 =DIDU &RQUDG 4 /LX ( ,UHQH +DPHV 5 .XHKQ DQG :RUWKPDQ 7KLFNQHVV DQG (IIHFWLYH (OHFWURQ 0DVV 0HDVXUHPHQWV IRU 7KLQ 6LOLFRQ 'LR[LGH )LOPV 8VLQJ 7XQQHOLQJ &XUUHQW 2VFLOODWLRQV $SSO 3KYV /HWW f >@ 3 %HYLQJWRQ 'DWD 5HGXFWLRQ DQG (UURU $QDO\VLV IRU WKH 3K\VLFDO 6FLHQFHV 1HZ PAGE 144 >@ 3 9DQ +DOHQ DQG 3XOIUH\ $FFXUDWH 6KRUW 6HULHV $SSUR[LPDWLRQV WR )HUPL 'LUDF ,QWHJHUV RI 2UGHU DQG f $SSO 3KYV f >@ < &KDQJ DQG $ ,]DEHOOH )XOO 5DQJH $QDO\WLF $SSUR[LPDWLRQV IRU )HUPL (QHUJ\ DQG )HUPL'LUDF ,QWHJUDO BLK LQ WHUPV RI $SSO 3KYV f >@ 6 $ :RQJ 6 3 0F$OLVWHU DQG = 0 /L $ &RPSDULVRQ RI 6RPH $SSUR[LPDWLRQV IRU WKH )HUPL'LUDF ,QWHJUDO RI 2UGHU 9 6ROLG6W (OHFWURQ f >@ 6 $EEDV DQG 5 'RFNHUW\ +RW&DUULHU ,QVWDELOLW\ LQ ,**(7fVf $SSOLHG 3K\VLFV /HWWHUV f >@ 0H\HU 026 0RGHO DQG &LUFXLW 6LPXODWLRQ 5&$ 5HYLHZ f >@ & 9LVZDQDWKDQ DQG 0 /HY\ 0RGHOOLQJ ,QWHU(OHFWURGH &DSDFLWDQFHV LQ D 026 7UDQVLVWRUf ,('0 f >@ :DUG DQG 5 'XWWRQ $ &KDUJH2ULHQWHG 0RGHO IRU 026 7UDQVLVWRU &DSDFLWDQFHV ,((( 6ROLG6WDWH &LUFXLWV 6& f >@ + 6FKLFKPDQ DQG +RGJHV 0RGHOLQJ DQG 6LPXODWLRQ RI ,QVXODWHG*DWH )LHOG (IIHFW 7UDQVLVWRU 6ZLWFKLQJ &LUFXLWV 6ROLG6WDWH &LUFXLWV 6& f >@ 6 /LX DQG / 1DJHO 6PDOO6LJQDO 026)(7 0RGHOV IRU $QDORJ &LUFXLW 'HVLJQ ,))( 6ROLG6WDWH &LUFXLWV 6& f >@ 5RELQVRQ < (O0DQV\ DQG $ 5 %RRWKUR\G $ *HQHUDO )RXU7HUPLQDO &KDUJLQJ&XUUHQW 0RGHO IRU WKH ,QVXODWHG*DWH )LHOG(IIHFW 7UDQVLVWRU, 6ROLG 6WDWH (OHFWURQ f >@ < (O0DQV\ DQG $ 5 %RRWKUR\G $ *HQHUDO )RUPXODWLRQ RI WKH 3DUDPHWHUV RI WKH (TXLYDOHQW &LUFXLW 0RGHO RI WKH ,*)(7,, 6ROLG6WDWH (OHFWURQ f >@ + 3DUN 3 .R DQG & +X $ 0HDVXUHPHQW%DVHG &KDUJH 6KHHW &DSDFLWDQFH 0RGHO RI 6KRUW&KDQQHO 026)(7V IRU 63,&( ,('0 f PAGE 145 >@ 6 &KXQJ $ &KDUJH%DVHG &DSDFLWDQFH 0RGHO RI 6KRUW&KDQQHO 026)(7V ,((( 7UDQV &$' f >@ : %XGGH DQG : /DPIUOHG $ &KDUJH6KHHW &DSDFLWDQFH 0RGHO %DVHG RQ 'UDLQ &XUUHQW 0RGHOLQJ ,((( 7UDQV (OHFWURQ 'HY f >@ + 3DUN 3 .R DQG & +X $ &KDUJH 6KHHW &DSDFLWDQFH 0RGHO RI 6KRUW &KDQQHO 026)(7V IRU 63,&( ,((( &$' f >@ 0 (O1RNDOL DQG $ $I]DOL.XVKDD $ 6XEWKUHVKROG 0RGHO IRU WKH $QDO\VLV RI 026 ,&V 3URF 1LQWK %LHQQLDO 8QLYHUVLWY*RYHUQPHQW,QGXVWUY 0LFURHOHFWURQLFV 6YPS f >@ 5KR /HH 0 6KXU 7 )MHOGO\ 8QLILHG 4XDVL6WDWLF 026)(7 &DSDFLWDQFH 0RGHO ,((( 7UDQV (OHFWURQ 'HY f >@ &KR 6 6 .DQJ .+ .LP 6+ /HH $Q $FFXUDWH ,QWULQVLF &DSDFLWDQFH 0RGHOLQJ IRU 'HHS 6XEPLFURPHWHU 026)(7V ,((( 7UDQV (OHFWURQ 'HY f >@ + ,ZDL DQG 6 .RK\DPD &DSDFLWDQFH 0HDVXUHPHQW 7HFKQLTXH LQ +LJK 'HQVLW\ 026 6WUXFWXUHV ,('0 f >@ + ,ZDL 2ULVWLDQ :DONHU DQG 5 'XWWRQ 6PDOO *HRPHWU\ 026 7UDQVLVWRU 0HDVXUHPHQWV DQG 2EVHUYHG 6KRUW DQG 1DUURZ &KDQQHO (IIHFWV 9/6, 6YPS 7HFKQRO f >@ 3DXORV $QWRQLDGLV DQG < 7VLYLGLV 0HDVXUHPHQW RI ,QWULQVLF &DSDFLWDQFHV RI 026 7UDQVLVWRUV ,((( ,QWHUQDWLRQDO 6ROLG6WDWH &LUFXLWV &RQI f >@ % 6KHX 3 .R DQG ) & +VX &KDUDFWHUL]DWLRQ RI ,QWULQVLF &DSDFLWDQFHV RI 6PDOO*HRPHWU\ 026)(7V ,('0 f >@ :HQJ DQG 3 PAGE 146 >@ 7 &KDQ $ :X 3 .R DQG & +X $ &DSDFLWDQFH 0HWKRG WR 'HWHUPLQH WKH *DWHWR'UDLQ6RXUFH 2YHUODS /HQJWK RI 026)(7V ,((( (OHFWURQ 'HY /HWWHUV ('/ f >@ 6FKPLWW/DQGVLHGHO DQG 'RUGD 1RYHO +RW(OHFWURQ (IIHFWV LQ WKH &KDQQHO RI 026)(7fV 2EVHUYHG E\ &DSDFLWDQFH 0HDVXUHPHQWV ,((( 7UDQV (OHFWURQ 'HY (' f >@ + ,VKLXFKL < 0DWVXPRWR 6 6DZDGD DQG 2]DZD 0HDVXUHPHQW RI ,QWULQVLF &DSDFLWDQFH RI /LJKWO\ 'RSHG 'UDLQ /''f 026)(7fV ,((( 7UDQV (OHFWURQ 'HY ('f >@ < PAGE 147 >@ 5 *KRGVL < PAGE 148 %,2*5$3+,&$/ 6.(7&+ 6WHYHQ 9 :DOVWUD ZDV ERUQ LQ &DOXPHQW &LW\ ,QGLDQD LQ +H UHFHLYHG WKH %DFKHORU RI 6FLHQFH GHJUHH LQ HOHFWULFDO HQJLQHHULQJ KLJK KRQRUVf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f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y STEVEN V. WALSTRA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1997 ACKNOWLEDGMENTS I would like to thank Prof. Chih-Tang Sah for his time and guidance as chairman of my supervisory committee, and Dr. Arnost Neugroschel, Dr. Toshikazu Nishida, Dr. Sheng Li, and Dr. Randy Chow for serving on my supervisory committee. Additional thanks go to K. Michael Han for many insightful discussions and debates concerning all aspects of device physics. I would also like to thank Dr. Changhong Dai, Dr. Shiuh-Wuu Lee, Mary Wesela, and Jerry Leon for providing the devices, measurement equipment, and technical expertise during my internship at Intel Corporation where the intrinsic capacitance data were taken. Financial support from a Semiconductor Research Corporation Fellowship is also gratefully acknowledged. TABLE OF CONTENTS Pag ACKNOWLEDGMENTS ii ABSTRACT v CHAPTERS 1 INTRODUCTION 1 2 EXTENDING THE ONE-DIMENSIONAL CURRENT MODEL 5 Introduction 5 Background 6 Long-Channel Theory 6 Pao-Sah Model 7 Bulk-Charge Model 14 Charge-Sheet Model 18 Comparison of Long-Channel Models 19 Two-Section Models 20 Beyond Two-Section Models 27 Examples Using Pao-Sah 27 Field-Matching Method 28 Saturation-Voltage Method 30 Surface-Potential Self-Saturation Method 31 In Search of the Match Point 31 Summary 38 3 POLYSILICON-GATE MOS LOW-FREQUENCY CAPACITANCE-VOLTAGE CHARACTERISTICS 41 Introduction 41 Metal-Gate CV 42 Polysilicon-Gate CV 46 Polysilicon-Gate Effects 51 Parameter Extraction Using the LFCV Model 57 3-Point Extraction Methodology 58 3-Region Extraction Methodology 63 Methodology Comparison 65 Convergence Speed-up Details 72 4 THE EFFECT OF INTRINSIC CAPACITANCE DEGRADATION ON CIRCUIT PERFORMANCE 75 Introduction 75 Background 75 Measurement of Intrinsic Capacitances 79 Measurement Configurations 82 Sample Measurements 85 Channel Hot-Carrier Stress Effects on Cgd and Cgs 94 Intrinsic Capacitance Degradation Model 97 Degraded Circuit Simulation 105 Conclusion 110 5 SUMMARY AND CONCLUSIONS 114 APPENDIX METAL-GATE LFCV MODEL DERIVATION 118 REFERENCES 132 BIOGRAPHICAL SKETCH 142 iv Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SCALING EFFECTS ON METAL-OXIDE-SEMICONDUCTOR DEVICE CHARACTERISTICS By Steven V. Walstra December 1997 Chairman: Chih-Tang Sah Major Department: Electrical and Computer Engineering As metal-oxide-semiconductor (MOS) transistor dimensions are decreased, channel-length modulation, polysilicon-gate depletion, and intrinsic-capacitance degradation have increasingly larger impacts on transistor performance. It is demonstrated that the Pao-Sah 1-D current model can be extended to include the channel- length modulation effect by use of a two-section model. This two-section model employs the normal long-channel Pao-Sah model in one region and adds a variable length depletion region in the other. Three methods for matching the boundary between the two regions are presented, with the best results coming from the most complex method of matching the longitudinal fields at the boundary point. The effect of polysilicon-gate depletion on the MOS low-frequency capacitance-voltage (LFCV) characteristics is demonstrated using a Fermi-Dirac-based model. It is shown that, as the oxide thickness decreases, the effect of poly silicon v depletion becomes increasingly pronounced. This depletion, in conjunction with the Fermi-Dirac carrier distribution, offset the current gain expected from thinning the MOS gate oxide. With this polysilicon-gate LFCV model, it is shown that the oxide thickness, flatband voltage, and gate and substrate doping concentrations can be extracted from experimental capacitance data. Two extraction methods, the 3-point and 3-region, are developed and are shown to work well with gate oxide thickness of 130A (2.7% RMS fit) and sub 30Ã (10% RMS fit). Voltage-accelerated stress is performed on state-of-the-art 0.24 (Xm effective- channel-length nMOS and pMOS devices to assess the impact on the most important intrinsic capacitances: Cgd and Cgs. The nMOS devices exhibit a Cgd reduction and Cgs enhancement with stress time, whereas the pMOS devices show negligible change. Because of Miller feedback, the nMOS Cgd reduction dominates the Cgs increase, resulting in an overall CMOS capacitive load reduction. Pre-stress and post-stress ID, Cgd, and Cgs data were fit using the BSIM3 device model. With the resulting parameter sets, a 31-stage ring oscillator was simulated for three situations: unstressed devices, stressed devices only including ID degradation, and stressed devices including ID, Cgd, and Cgs degradation. It is shown that the inclusion of the intrinsic capacitance degradation results in improved simulated circuit performance because the capacitive load reduction offsets the drain current reduction. This improved degradation methodology will result in looser guardbands and less reliability redesign. vi CHAPTER 1 INTRODUCTION The last three decades of production integrated circuits (IC) have seen two orders of magnitude decrease in device dimensions, from 25 pin in 1962 to 0.25 pm in 1997 [1-3]. This continual reduction, fueled by requirements for higher switching speeds, lower cost, and decreased power, has been sustained by improvements in lithography and has resulted in increased areal and chip densities (transistors/cm2 and transistors/chip). Compared to the -500 transistors/chip in the first experimental 64-bit static random- access memory (SRAM) in 1965, the -64M transistors/chip 64 Mbit dynamic random- access memory (DRAM) of 1997 and the -4G transistor/chip 4 Gbit DRAMs due from NEC in 2000 typify the strong push toward increased density. Increased areal density implies decreased dimensions. As transistor and capacitor dimensions decrease, previously negligible effects have become or are becoming increasingly important. Many of these effects were assumed avoidable through constant-field scaling [4], These scaling rules have been debated, amended, and improved [5-7] to account for noise-margin, hot-electron, and extrinsic-capacitance considerations, but present and future smaller dimensions have necessitated these effects be included in the design process. Several of these effects are discussed below. As channel lengths decrease, the thickness of the space-charge region at the drain of a metal-oxide-semiconductor (MOS) transistor becomes a significant fraction of 1 2 the total channel length. As the drain voltage is changed, the space-charge-region thickness also changes, resulting in an effective channel length which is drain-voltage dependent, an effect known as channel-length modulation (CLM). This problem can be tolerated for complementary MOS (CMOS) logic circuits, but needs to be properly modeled in order to predict the drive current of the MOS transistors in the circuit in order to estimate the speed of the resulting circuit. As the density of transistors increases, so does the power density. This requires a reduction in the operating voltage, since the active (switching) output power is proportional to the square of the operating voltage (PactÂ¡ve â€œ fciockCoV2). To obtain the same performance at lower voltages, the oxide thickness must be reduced. Simple MOS theory predicts the drain current is inversely proportional to the gate oxide thickness. However, for thin oxides (< 50 A), depletion of the polysilicon gate offsets the effects of thinner oxides, resulting in lower current and diminishing returns on oxide scaling. Additionally, the gate voltage cannot be reduced indefinitely, because a large enough margin is needed between the signal voltage and ground-plane noise to ensure that noise does not change the state of the device. The increased density of transistors also requires more closely-spaced interconnections between the transistors. Interconnect scaling has made delays due to the interconnection a limiter in process speed [8], and major efforts are currently underway to reduce the interconnect resistance and capacitance. Copper has recently been introduced into 1998 production by IBM to reduce the interconnect resistance. Additional efforts have been underway to lower the dielectric constant of the intermetal dielectrics in order 3 to reduce the interconnect capacitance. When the interconnect capacitance is reduced, the only remaining capacitance left to slow the CMOS circuitry is the intrinsic capacitance of the transistors, which cannot be easily reduced and will become the predominant speed limiter. There are many other issues concerning the perpetual reduction in transistor dimensions, the least of which is the brick wall of atomic dimensions. Clearly transistors cannot be scaled to less than ten or twenty atoms and still work in the traditional sense of transistors, yet this dissertation includes data from a transistor pushing the atomic limit with a gate insulator thickness of less than 30A, or under six atomic layers of silicon and oxygen. The goal of this dissertation is to investigate the issues described in the previous paragraphs. Chapter 2 discusses the history of 1-dimensional drain current models and some of the methods which have been implemented to extend these models to include the CLM effect. The Pao-Sah model, the most accurate long-channel current models, will be extended to include the CLM effect using three different approaches. The CLM effect (as demonstrated in the new models) will be discussed, as well as the pros and cons of the approaches. Chapter 3 tackles the polysilicon depletion problem by deriving the Fermi- Dirac-statistics-based polysilicon-gate MOS low-frequency capacitance model, including the effect of dopant impurity deionziation. By comparing this with the traditional metal- gate model, the effect of polysilicon gate depletion will be shown to increase significantly as the oxide thins. With this model, a parameter extraction methodology is presented 4 which allows the extraction of substrate and gate doping concentrations as well as the oxide thickness and flatband voltage from experimental LFCV data. Two methodologies will be presented and compared, and data from thick (130Ã) and thin (< 30Ã) gate-oxide devices will be used. Additional oxide thickness issues, such as quantum effects, are also discussed. Chapter 4 considers the intrinsic capacitances, in particular, those most important in modern complementary MOS (CMOS) circuits: Cgd and Cgs. Compared to the drain current, which is also an intrinsic property of a MOS transistor, intrinsic capacitances have been relatively ignored because of measurement difficulty and relatively small impact compared to extrinsic capacitances. However, as processing and dielectric technology advances, the primary remaining capacitive load in CMOS circuits will be the intrinsic capacitances. The chapter presents an experimental investigation how these capacitances change with hot-carrier stress and, after modeling the stress- induced changes in the intrinsic capacitances, shows that part of the drain current degradation is offset by the intrinsic capacitance reduction, resulting in a slower degradation of overall circuit performance. CHAPTER 2 EXTENDING THE ONE-DIMENSIONAL CURRENT MODEL Introduction The simplest 1-D model is of crucial importance for applications in semiconductor physics. Although 3-D models will best match experimental data because of both inclusion of real effects and simply additional variables, they may be intractable as compact device models, where computational efficiency is critical. Conversely, these 3-D models are often validated by demonstrating their reduction to the rigorous 1-D forms for non-critical (wide and long channels with thick oxides) geometries. For back- of-the-envelope calculations, knowledge of the basic physics embodied in a good 1-D model is exceedingly useful. The required accuracy of a model is largely determined by the application. For predicting the drive current, such as might be required for a discrete-transistor specification sheet, a model need not worry about the linear or subthreshold regions of operation. Similarly, if modeling only the operating range (0 to power supply voltage), then the accumulation region of applied gate voltages can be ignored in the model. There are cases, particularly when attempting to predict the performance of new technology, where 3-D full-range MOSFET models are necessary, but they are a relative minority compared to the wide array of applications for 1-D models. 5 6 This chapter contains a brief history of one-dimensional (1-D) approaches to drain current models, including calculations and comparisons, followed by a new two- section model using the 1-D Pao-Sah long-channel IV model in conjunction with a variable-length depletion region. The goal is to extend the 1-D long-channel model to short-channel use. Background In 1926 Lilienfeld [9] submitted the patent for the first MOSFET device, an A1/A1203/Cu2S transistor. Thirty-two years later in 1960, Kahng and Atalla [10] fabricated the first silicon MOS transistor. A year later, the first MOST current-voltage (IV) papers were published internally at AT&T Bell Labs in 1961 by Kahng [11] and later at Stanford by Ihantola [12], These were followed in 1964 by more complete (and widely released) 1-D theories by Sah [13] and Ihantola and Moll [14]. A comprehensive history of MOS developments was reviewed by Sah [1], In the subsequent years since the first MOST model, hundreds of papers and theses have been written about the modeling of various aspects of MOS transistors. This chapter will discuss the prevailing 1-D models including Pao-Sah, bulk-charge, charge-sheet, and the many two-section models. Long-Channel Theory "Long channel" is a term used to specify that short-channel effects can be neglected when modeling MOSTs, and the predominant short-channel effect is encroachment of the drain depletion region into the channel. The depletion region exists 7 due to the reverse-biased p/n junction between the substrate and the drain, and has nothing to do with the actual channel length. For long-channel devices, however, the amount of encroachment relative to the channel length is small, so the effective channel length is essentially constant (equal to the drawn gate length). For short channels, however, the effective channel length can be significantly reduced by the encroachment. Another short-channel effect neglected in long-channel theory is drain-induced barrier lowering [15], where the source barrier is lowered by the applied drain voltage. Pao-Sah Model The most accurate long-channel theory was published by Pao and Sah (PS) [16]. The PS model is the only one which correctly accounted for drift and diffusion. The PS theory, to be discussed below, contains a double integral, but can be reduced to a more efficient form containing only single integrals [17, 18]. Although cumbersome to calculate, the PS double integral is extremely didactic and is a useful starting point for showing the approximations used to derive other long-channel IV models. The total current flowing in the channel is given by the integral J(x,y)Z dx, (2.1) 0 where J(x,y) = JN + Jp = JN = qpnNEy + qDnVN = qDnNVÂ£ . (2.2) JN and JP are the electron and hole current densities, respectively, and it is assumed that the current is dominated by electrons in an n-channel device in (2.2). The electron charge is q, pn and Dn are the electron mobility and diffusion respectively, and VN is the gradient of the electron concentration. The electron quasi-Fermi level,is measured relative the bulk Fermi level and normalized to kT/q. If d^/dx is assumed negligible (which is a fundamental assumption in the long-channel approximation and should be valid to a depth on the order of the drain junction depth), then ID can be found from summing up all the current from the surface down to some depth xÂ¡ below which the additional contribution is negligible: Id qDnZ(df/dy) xi N (x) dx 0 This can be transformed from physical space in the y direction to potential space as follows: (L ruD ID dy = qDnZ 0 0 (2.3) where UD=qVDS/(kT) is the normalized drain voltage at y=L and the lower limit 0 is the grounded source voltage at y=0. A similar transform in the x direction yields: Id = <2Dn~ Un Us N(U) dU (2.4) LJ0 JUF (-dU/dx) where (dU/dx) is the x-component of the electric field, which can easily derived from integrating Poissonâ€™s equation by quadrature and is given below. The Boltzmann approximation to the carrier concentration is being used and the impurities are assumed completely ionized, but the Fermi-Dirac and deionized form can be used. Us is the normalized surface potential (where surface is at x=0), the total amount of surface band bending relative to the intrinsic Fermi level. It is a function of both the gate voltage and the drain voltage. UF is the normalized bulk Fermi level, below which the current 9 contribution is assumed negligible, and is analogous to the to physical point x=xÂ¡ in (2.3). The derivative (dU/dx) is found from (-dU/dx) = F(U,Â£,Uf)/Ld (2.5) where F (U, Â£, UF) = [exp (U-Â£-Up) + exp(UF-U) + (U-l)exp(UF) - (U+exp (-Â£) ) exp (-UF) ]1,2 (2.6) After applying Einsteinâ€™s relationship, Dn/pn = kT/q, (2.4) becomes -dUdÂ£ (2.7) The surface potential, Us(Â¡;), is needed in (2.7). The relationship between the surface potential and the gate voltage can be found by applying Gaussâ€™s Law at the semiconductor/insulator interface. The resulting equation, given below, can be solved iteratively for Us for a given q. rkT 2 z uD Us exp (Uâ€”UF) q 2L Ld 0 UF F(U, f,UF) Uâ€ž = U* sign(Us) yF(Us,|,UF) (2.8) where UG is the normalized gate voltage, q(VGS - VFB)/kT; y is es/(LDÂ«C0); LD is the Debye length (V[eskT/(2nÂ¡)]/q); and F(US4,UP) is given by (2.6). Equation 2.7 is the traditional form of the PS integral, often called the Pao-Sah double integral. A more computationally friendly and accurate single-integral form [17] was used for the calculations in this dissertation. The mobility in (2.7) need not be taken out of the integrals. Instead, it can be a function of the vertical and lateral fields and moved inside of the integrals. In this chapter the mobility will be assumed independent of field. 10 A good way to understand Eq. 2.7 is to consider the three-dimensional band structure of a MOST under gate and drain bias, as shown in Figures 2.1-2.4, based on the original Pao-Sah paper [16], Figure 2.1 shows an idealized n-channel MOST. Figure 2.2 shows the corresponding energy band diagram with no applied terminal voltages except VGS=VFB. From the position of the Fermi level it is easily verified that the source and drain are n-type and the substrate is p-type (n-channel device). Electrons in the source and drain see a potential barrier toward the channel. Application of a positive voltage to the gate lowers the barrier near the surface, as shown in Figure 2.3. The applied gate voltage pulls electrons toward the surface (and pushes holes away from the surface), as can be seen from the position of the Fermi-level relative to the band edges. Farther into the substrate (away from the gate/substrate interface) there is no bending from the gate potential, so the region is identical to the unbiased case (Figure 2.2) and considered quasi-neutral. Applying a voltage to the drain (VDS < VDSsat) splits the Fermi level into quasi- Fermi levels (Fn for electrons and Fp for holes), as shown in Figure 2.4. One can imagine an electron in the conduction band surmounting the source barrier and then falling down the potential â€˜cliffâ€™ until reaching the drain. This â€˜free fallâ€™ is where the electron gains energy while moving across the channel. If the electron is not scattered while moving across the channel (losing energy to the lattice via phonons), it becomes increasingly energetic as it approaches the drain and may become â€˜hotâ€™ enough to produce an e-h pair via impact, the resulting hole may generate interface traps via dehydrogenation of Si-H bonds near the Si/Si02 interface [19], This is only one of several mechanisms for interface trap generation. Fig. 2.1 / D s G Simplified view of two-dimensional MOS device. Fig. 2.2 Schematic 2-D energy band diagram of simple MOS device with source and drain grounded and VGS=VFB. Adapted from Pao and Sah [16]. 12 Fig. 2.3 Schematic 2-D energy band diagram of simple MOS device with VGS > VFB, drain and source grounded. Adapted from Pao and Sah [16]. 13 Fig. 2.4 Schematic 2-D energy band diagram of simple MOS device with VGS > VFB, 0 < VDS < VDSsat, and source grounded. Adapted from Pao and Sah [16]. 14 Figure 2.5 shows the result of applying a drain voltage in excess of VDSsat. As will be discussed in the two-section model section later, the drain depletion region becomes increasingly longer as the reverse-biased drain voltage increases. For this long-channel section of the dissertation, however, the change in length, AL, is assumed much less than the channel length L. The voltage drop across this thin depletion region often results in large fields which can greatly accelerate carriers, causing the interface damage mentioned above. Now that the effect of applied biases on the 2-D structure of the band has been discussed, it is easy to see the basis of the integral limits in Equation 2.7. The inner integral is integrating from the surface into the bulk (from Us to Up), which is a cross section of the channel as shown in Figure 2.6. The outer integral is integrating from drain to the source (UD to 0, source is grounded) along the channel. Thus, the double integral is summing up all the current contribution in the channel, exactly as would be expected. Since Us is a function of the drain voltage (or the channel potential), the order of the double integration is not trivially reversible. Bulk-Charge Model The first group of ID models, in order of complexity, were by Sah [13], Ihantola and Moll [14], and Sah and Pao [20]. These are all bulk charge models, taking increasingly more into account. As the name suggests, the bulk charge model takes the depleted region under the channel (in the bulk) into account. It assumes drift is the major component and so neglects the diffusion component. This greatly simplifies the problem and reduces (2.2) to J(x,y) = JN + Jp = JN = qi/nNEy = qpnN(x) (dV/dy) (2.9) 15 Fig. 2.5 Schematic 2-D energy band diagram of simple MOS device with VGS > 0, Vds > ^DSsatâ€™ an<^ source grounded. Adapted from Pao and Sah [16]. 16 Fig. 2.6 Schematic 2-D energy band diagram of simple MOS device with VGS > 0 and VDS < VDSsat. Cross-sections show the 1-D energy-band diagrams near the source and drain. G^Gate electrode, XsSubstrate electrode. 17 Id q/tnZ (dV/dy) i qN (x) dx 0 (2.10) ID = -nnZ(dV/dy)Qn (2.11) where Qn = -C0(VG - v -Vs0) + (2qPxxÂ£s)1/2[Vso + V]1/2 (2.12) C0 is the oxide capacitance per unit area, VÂ¿ is Vos - VFB, Vso is the surface potential at the source, and V is the channel potential (=VDS at the unsaturated drain). Pxx is the substrate impurity concentration and es is the dielectric constant of silicon. The first term is the charge accumulated in the channel and the second term is the uncompensated charge in the depletion region beneath the channel (i.e. bulk charge). Integrating (2.12) along the channel gives: ID = pn(Z/L)C0{ (VG - VS0)VDS - V2s/2 - (2.13) (1/C0) (2/3) (2qPxxÂ£s)1/2 [ (Vs0 + VDS)3/2 - (Vs0)3/2]} This form is slightly different than the Sah-Pao and Ihantola-Moll forms because it is not assumed that VS0=2VF, where VF is the Fermi voltage. A more exact form [17] is: ID = pn(Z/L)C0{ Vg(Vsl - Vs0) - (1/2) where VSL is the surface potential at the drain. This differs from (2.13) in that the surface potential at the drain is calculated instead of assumed to be VSL=VS0 + VDS. When the drain current approaches or exceeds saturation (VDS > VDSSat), VSLsÂ£VS0 + VDS. Additionally, in subthreshold, VSL is typically closer to Vso than Vso + VDS [21]. As will 18 be shown later, the bulk charge formula should never be used for subthreshold calculations since it neglects diffusion, which is the primary subthreshold current contribution. The bulk charge form, compared to PS, is considerably easier to calculate, particularly when using (2.13) with Vso = 2VF, but is invalid in subthreshold. Equation 2.13 is also invalid in saturation as written, but that can be fixed somewhat by calculating the saturation voltage VDSsat and fixing the current for all drain voltages greater than VDSsat. This will make the first derivative (drain conductance) non-continuous at VDS=VDSsat. All saturation problems are solved in (2.14), where the calculation of VSL negates these problems. Iterative calculation of VSL is time consuming, particularly compared to assuming a constant, or pinned, surface potential value. Charge-Sheet Model While most of the interest centered on super-threshold operation of the MOST, some people became concerned with the lack of accurate modeling for subthreshold operation. Barron [21] and Van Overstaeten et al. [22] developed subthreshold formulae based on simplifications of the Pao-Sah integral, with results applicable only to the subthreshold region. Six years later, Brews [23] made a critical approximation which would allow both drift and diffusion components to be introduced simultaneously without the need for a double (or single) integral. When he proposed his "charge-sheet model," he introduced the following simplification: I = qZpnN(y)(df/dy) dÂ£/dy=d0s/dy - 1/p dln(n)/dy (2.15) 19 This approximation for dZJdy was justified "based upon its success in producing â€˜correctâ€™ I- V curves," although he added a footnote relating the formula to electrochemical potential. This wide-open statement resulted in several subsequent â€˜proofsâ€™ which derived the same formula [17, 24, 25], Essentially, though, he decoupled the drift and diffusion components from the tight interdependency seen in the Pao-Sah form to the simple form of (2.15). Through a similar derivation to bulk-charge, ID is given by I=/in(Z/y) (1//3) {C0(l/j3+VG) (Vs(y) - Vs0) - (1/2) C0 (V2 (y) - V|0) (2.16) - (3/2) (2qPxxes)1/2[ (0Vs(y) - 1)3/2 - (pVs0 - 1)3/2] + (2qPxxÂ£s)1/2[ (/JVs(y) - 1)1/2 - (]3VS0 - 1)1/21) Eq. 2.16 reduces to bulk-charge form of Eq. 2.14 if VG, Vs(y), Vso Â» 1/p and the square root terms are negligible. Unlike bulk-charge, this formula is valid in subthreshold and does not require a calculation of VDSsat (assuming VSL and Vso are calculated iteratively). Like bulk-charge, this is much easier to calculate than a double, or even single, integral. Brews, and many subsequent authors, validated the charge-sheet model by comparing it to the results of the Pao-Sah formula. It has been shown to be an excellent approximation, as will be discussed in the next section. Comparison of Long-Channel Models The Pao-Sah double-integral model has been heralded as the best long-channel model. Brews [23] went so far to say that "Comparison of the charge-sheet model with the Pao-Sah model has the force of comparison with experiment, since the Pao-Sah model is known to work well for long channel devices." Schrimpf et al. [26] agreed, saying Pao and Sah "produced a quantitative model so accurate that it is the standard by which other models 20 are judged." Since bulk-charge and charge-sheet are both approximations to Pao-Sah, it makes sense to compare them with Pao-Sah to see how accurate they are, taking into account that all the models are only valid for long-channel devices. Figure 2.7 shows all three methods simulated for Tox=500 Ã, T=296 K, Pxx=1015 cm"3, W/L=10. These are typical parameters for LSI devices of the 1970s, and were chosen to match the data used in Pierret and Sheilds [17]. As can be seen, the bulk-charge and charge-sheet models underestimate the current. Figure 2.8 shows the percentage error for each model at the gate voltages shown in Fig. 2.7, demonstrating that the charge-sheet model maintains an error of less than 2.6% for all gate voltages, while the bulk charge model ranges from 2.5% for VCS=5.0V to 8.4% for Vos=2.0 V. This suggests that the much simpler charge sheet can be used in place of Pao-Sah incurring only about 2.5% error at low voltages. Figure 2.9 shows the subthreshold region for the same device with VDS=0.1 V. Clearly demonstrated in this figure is both the glaring inadequacy of the bulk-charge model for subthreshold modeling and the remarkable accuracy of the simple charge-sheet model. However, recall that this is charge-sheet with iteratively calculated surface potentials, so the numerical solution is not entirely trivial. Two-Section Models Up until now, only long-channel ID equations have been considered. For short- channel devices (<1 pm), the most prominant non-modeled effect on the drain current is finite drain conductance beyond saturation. The primary cause of this non-zero drain- conductance (gD) is channel shortening from the drain space-charge region (SCR) 21 ID versus VDS for different VGS values for the three 1-D ID models. Parameters are Tox=500 Ã, T=296 K, Pxx=1015 cm'3, W/L=10, which were used to match data in Pierret and Shields [17]. 22 Fig. 2.8 Percentage error in ID for charge sheet and bulk charge relative to Pao- Sah versus VDS, from Fig. 2.7. Plots are Vos = 5, 4, 3, and 2 V, with higher errors for lower voltages. 23 VG (V) Fig. 2.9 ID versus VGS for Pao-Sah, charge sheet, and bulk charge using same data as Fig. 2.7 with VDS=0.1 V. Clearly bulk charge is not useful in subthreshold, whereas charge-sheet is almost coincident with Pao-Sah. 24 encroaching into the channel. This effect is often called channel-length modulation since the drain voltage modulates the effective channel length. The most logical approach is to divide the region between the source and drain into two sections: a â€˜source sideâ€™ and a â€˜drain sideâ€™. The â€˜source sideâ€™ may contain any appropriate long-channel IV model, such as Pao-Sah, charge sheet, or bulk charge. The â€˜drain regionâ€™ is the depletion region, and can be modeled with or without mobile charge, 2- D effects, mobility differences, etc. The location of the boundary between these regions, and the voltages and fields at this boundary, are what make this a challenging problem. Figure 2.10 shows a diagram of a MOS transistor divided into two sections. There are essentially three things which differ among approaches to two-section theory: the the source-side IV model, the drain-side space-charge region (SCR) model, and the boundary conditions. Source Source Side (Long channel approx) Drain Side (SCR) Drain LeW=(L-AL) ' AL L y=o y=yM y=L v=o v=vM v=vD Fig. 2.10 Schematic diagram of two-section MOST for 1-D modeling. SCR means â€™Space-Charge Regionâ€™ and Leff refers to the effective channel length. 25 The IV model can be one of the many already discussed. The SCR model can be assumed fully depleted, take mobile charge into account, or be a complete 2- or 3-D model. The boundary conditions are the most difficult and varied among approaches. Essentially, the potentials, fields, and charge at the boundary between the two regions need to be matched. The simplest two-section MOST model was introduced in 1965 by Reddi and Sah [27]. They used a source-side bulk-charge model for the current and a fully-depleted drain- side depletion model. From the first derivative of the bulk-charge model (Eq. 2.13 with Vs=2Vf), Reddi-Sah (and others) calculated the drain voltage where, for a constant gate voltage, the drain conductance drops to zero (VDSsat). They then assumed all voltage in excess of VDSsat falls across the SCR to form the drain region of the two-section model. By assuming complete depletion (no mobile charge) and no y-field at the boundary, the length can be calculated from simple p/n junction theory as: AL = [2es (VDS - VDSsat + Vbi)/(qPxx)]1/2 (2.17) where Vbj = (kTq)ln(NdrainNsubslrate/nf) from standard abrupt-junction p/n theory. Replacing L by Leff=L-AL and Vso with 2VP in (2.13) yields the Reddi-Sah two-section current. The simplicity of this formula is extremely attractive, but the solution is dependent on the ID model. Specifically, it assumes that a VDSsat voltage can be found. If using Pao- Sah or charge-sheet, the surface potential is not constant and a VDSsat point does not actually exist. Even if VDSsat is found from extrapolation, the first derivatives of the drain current will be non-smooth at the point where the drain current switches from one model (Pao-Sah, 26 charge-sheet, bulk-charge) to another (constant ID), although this can be fixed with various smoothing transitional functions. Four years after Reddi and Sahâ€™s paper, Chiu and Sah [28] came out with a two- section model which solved Laplace equation in the oxide layer and matched values in four regions (source, drain, oxide, and bulk). The drain region was solved as a 2-D, fully- depleted region, and the solution required seven matching parameters. The complexity of the solution relinquished this model to an almost constant reference as "too complex." The following year (1969) Frohman-Bentchkowsky and Grove [29] developed a two-section model using bulk-charge model in the source region and an empirical model for the drain section. This simple model essentially added two additional fringe field contributions to the Reddi-Sah model and added two empirical variables to fit the data. Merckel, Borel, and Cupcea [30] added mobile charge to the drain region empirically by writing Poissonâ€™s equation in the drain region as d2V/dy2 = q/Â£s(Pxx + IDS)/(qZa) (2.18) where a is essentially a fitting parameter related to the junction depth. This mobile charge is akin to the Kirk effect in bipolar devices, just as the drain-depletion encroachment is analogous to the Early effect. Using an iteratively determined VDSsaI, they were able to calculate the drain depletion width. Popa [31] devised a similar model and extended the drain depletion region to be of three types depending on the injected current. In both mobile-charge cases, fitting parameters were introduced either through (2.18) or mobility. Both used variations of the simple bulk charge model for the source side. 27 After Brews developed the charge-sheet model, all subsequent two-section models employed the charge-sheet model. Guebels and Van de Wiele [32] developed a three- section model to account for the x-field reversal near the drain. They employ the same trick as the previous papers by fitting the a in (2.18), using VDSsat (or IDsat) and adding some empiricism to their field calculations. Beyond Two-Section Models The charge-sheet model (and Pao-Sah, as will be shown) does not lend itself well to analytical two-section models due to the greater complexity of the drain current model relative to bulk charge. As noted above, fitting parameters and empirical formulae were required to be introduced to satisfy some of the boundary conditions. The newer compact models, such as BSIM [33,34] and Siemenâ€™s [35-37] model, are based loosely on one-section bulk-charge and charge-sheet models, respectively, sometimes dividing the model into different sections based on operation (separate subthreshold and superthreshold formulae). They both model short-channel effects by adding semi-empirical additions to the threshold voltage, which makes for a considerably faster calculation speed at the expense of a less-physical model. Examples Using Pao-Sah The goal was to develop a two-section model which employs the Pao-Sah integral as the source-side current formula. The following is a description of the methodology and results of the exercise. 28 Field-Matching Method The Pao-Sah current has already been discussed, as have been models for the depletion region. Let us consider the matching boundary of the two section model to occur at the point y=yM where the channel voltage is VM with a lateral field EM and electric field gradient d2Us/dy2=dEM/dy. A simple way to look at this problem is from the Poissonâ€™s equation in the drain region while considering the boundary conditions. Within the drain region, which extends from y=yM to y=L, the boundary conditions are (see Fig 2.10): V(L)=Vds v(yM)=vM dV(yM)/dy=EM (field at the match point) d2V(yM)/dy2= (l/es)[qPXx + (mobile charge terms)] = C It is possible from Pao-Sah to calculate dV(yM)/dy=EMps [38]. This gives us the following equations after integrating the Poissonâ€™s equation twice with the above boundary conditions: The ideal additional equation would be d2V(yM)/dy2 on the Pao-Sah side, but this quantity is incalculable from the Pao-Sah integral. If it is assumed that assume Em=0 (as was done in Reddi-Sah), the depletion length into the channel can be easily found. It is reasonable to assume that the lateral field at the matching point (EM) is much less than the field right at the drain (ED), so ED Â» Em. making the difference in yM small. This gives (from 2.20, also 2.17) 29 yM = L - (2 (VDS - VM + Vbi)/C)1/2 (2.21) Where Vbi accounts for the pre-existing depletion region originating from the abrupt p/n junction. Since the yM approximation has already been made, it will be assumed that the field throughout the drain region is a constant at the boundary and is given by EMdep = (VDS - Vm)/(L - yM) (2.22) Clearly there are conflicting assumptions (EM = 0, and now EM * 0). One might wonder why Em is not (VDS - VM + Vbi)/(L - yM) to be consistent with 2.21. This comes from the subtlety of the boundary conditions. Looking back to Figure 2.2, note that the integration is actually from Vs + Vbj to VDS + Vbj, which excludes the p/n depletion layers. The Vbiâ€™s cancel out for symmetrical devices, so this is no problem. At VDS=0 (and source grounded), no current or field is expected, which would make VM correctly equal to 0 in (2.22). However, if Vbi were added to (2.22), then VM would have to equal VbÂ¡, which would incorrectly cause a field (and possibly current flow depending on Vos). Essentially, (2.22) gives the excess field. However, Vbi does contribute to the depletion width, so it is included in (2.21). The normalized field on the Pao-Sah side at the boundary is given by [32] [ exp (Us) -1 ] exp (-UM-UF) 'Mps 2 r - F(Us,Um,Uf) + exp (Us-UM-UF)-exp (UF-Us)+exp(UF)-exp (-UF) V L Â°m(us exp(U-f-UF) -dUdf 0 UF F(U,f,UF) x â– dU (2.23) 30 where UM is the normalized matching voltage, VM*(q/kT). Figure 2.11 shows the results of this approach, with mobile charge terms neglected (C=qPxx/esi) for Pxx=5xl017 cm-3, T=300 K, and Tox=50 Ã. The data cover a wide range of channel lengths from 'U pm to Â°Â°, and for all cases the width is equal to the length (square devices). The saturation current predicted by long-channel theory for these square devices would be the same for all channel lengths, so the deviation from this is due to channel- length modulation, which clearly becomes more important and the channel length decreases. Figure 2.12 shows that the drain conductance (g^dlp/dVpj) is smooth, which is important for circuit simulator applications. Although not shown, the derivative of the drain conductance is also smooth. Thus, this field-matching model successfully extends the 1-D Pao-Sah model to short-channels, at least with regards to including the effective channelÂ¬ shortening effect. Saturation-Voltage Method Reddi and Sah [27] assumed VM=VDSsat, which simplified things considerably. VDssat *s easy to calculate when using the bulk-charge formula assuming VS0=2VF since the derivative of the surface potential with respect to the drain voltage is zero. The Pao-Sah current, however, does not technically saturate (numerically there will be a point where the current does not increase, but it will be at a drain voltage well in excess of the normal VDSsat point). This problem is solved by extrapolating VDSsat from dID/dVDS versus VDS without channel shortening. Figures 2.13 and 2.14 show the results of employing this method with the same device as used in the previous section (Pxx=5xl017 cm-3, T=300 K, Tox=50 Ã), using Eq. 2.21 for yM with VM=VDSsat. Clearly the channel-length modulation is being 31 accounted for, but the transition is slightly abrupt. A look at the resulting drain conductance (Fig. 2.14) shows a drastic discontinuity near the calculated VDSsat point. Use of a fitting function could rectify this derivative problem, and is a common practice for compact models. Surface-Potential Self-Saturation Method Another possible way to circumvent finding the VM point was posed by Katto and Itoh [39], Instead of finding VDSsat, they used the fact that the surface potential itself will saturate when solved iteratively from (2.8). Thus replacing the matching voltage, VM, with the surface potential at the drain, VSL (solved iteratively) gives another decoupled way to solve for yM. Using the surface potential to find the depletion thickness was also used by Sah [2], This is better than the VDSsat method since there will not be an immediate point where saturation occurs. However, as shown in Figs. 2.15 and 2.16, the current still has a slight â€˜jumpâ€™ resulting in discontinuities in gD. In Search of the Match Point Sah [2] showed pictorially that in saturation, the energy band near the drain edge will actually be bent upward, or in other words, the surface will be accumulated rather than inverted (actually, the surface will still be depleted, but now accumulation refers only to the shape of the band bending). This must be the case since the potential along the channel is actually higher than VGS - VGT = VDSsat. This means that there must be a point along the channel at which the band bending is zero at the surface, and this point would be an excellent candidate for the yM point. Like the methods above, however, this point has some 32 Fig. 2.11 ID versus VDS plots for different channel lengths (square devices) using field matching at the match point. Vos = 5 V, Pxx=5xl017 cm"3, T=300K, Tox=50 Ã. gD (mA/V) 33 Fig. 2.12 gD versus VDS plots for different channel length (square devices) using field matching at the match point. Same parameters as Fig. 2.11. 34 Fig. 2, 13 ID versus VDS plots for different channel lengths (square devices) using VM=iterative surface potential at drain. VGS = 5 V, Pxx=5xl017 cm-3, T=300K, Tox=50 Ã. gD (mAA/) 35 Fig. 2.14 gD versus VDS plots for different channel length (square devices) using VM=iterative surface potential at drain. Same parameters as Fig. 2.13. 36 Fig. 2.15 ID versus VDS plots for different channel lengths (square devices) using VM=VDSsac vgs = 5 V, Pxx=5xl017 cnr3, T=300K, Tox=50 Ã. gD (mA/V) 37 Fig. 2.16 gD versus VDS plots for different channel length (square devices) using VM=VDSsaf Same parameters as Fig. 2.15. 38 logical flaws. For instance, the field in the x-direction is zero by definition, which means that using the channel potential at this point to calculate the current from the long-channel model will clearly invalidate the gradual channel approximation (Ex Â» EY), a basis for the Pao-Sah ID derivation. A simple approximation for this point would be to use VM = VGS - VFB when VDS > VGS - VFB, which is akin to setting VDSsat = VGS - VFB. This ends up resulting in the same sort of problem seen in the VDSsat method. It is interesting to verify the existence of this turn-around region in the channel near the drain, however. This was done recently using the MINIMOS device simulator [40]. MINIMOS was modified to use a constant mobility model so as to be comparable to the 1- D model cases above. Figure 2.17 shows the resulting electrostatic potentials into the substrate at different points near the drain edge of a 50Ã, 100x100 pm (corrected for subdiffusion) nMOST with Pxx=5xl017 cm-3 at VGS=1.5 V and VDS=3.0 V. VFB was fixed at zero for this case. What is clear is that the band moves from inversion (top) through flatband into accumulation (bottom) at the surface (x=0.0). The flatband point occurs when the channel potential is equal to VGS - VFB = 1.5 V, as expected. Summary This chapter reviewed the history of 1-D long-channel drain-current models and discussed the pros and cons of their derivation and applications. From this, the importance of a non-pinned surface potential was shown, as demonstrated by the excellent approximation of the simple charge-sheet model to the Pao-Sah double integral-the best of the 1-D long-channel models. 39 Next, methods to extend the 1-D into two 1-D sections to create the best full-range 1-D model. It was discovered that, no matter what, the depletion region is strictly 2-D, and obtaining a 1-D approximation requires rather substantial assumptions. One model, the field-matching approach, was seen to give reasonably good characteristics, while all the other approximations (VDSsat and surface-potential self-saturation) resulted in discontinuities in the first (and higher) derivatives. A 2-D simulation was used to verify that there is a point in the saturated channel where the (x-directed) field reverses and the surface band bending is, thus, zero. This point has been suggested many times before in our group, but never verified two-dimensionally. Attempting to use this point to demark the boundary of the source region and drain region of the two-section model results in the same poor results as the VDSsat method. 40 SiOo/Si interface X (|im) Fig. 2.17 Electrostatic potentials into the substrate at different points near the drain edge of a 50Ã, 100x100 pm (corrected for subdiffusion) nMOST with Pxx=5xl017 cm-3 at Vcs=1.5 V and VDS=3.0 V. The Y=1.321 pm (near source) and Y=50.000 pm (middle of the channel) curves are indistinguishable. The band is flat at the Si02/Si surface when the channel electrostatic potential equals VGS -VFB = 1.5 V (VFB = 0 for this data). CHAPTER 3 POLYSILICON-GATE MOS LOW-FREQUENCY CAPACITANCE-VOLT AGE CHARACTERISTICS Introduction For modern ULSI technology, polysilicon gates are universally used on MOS devices. With respect to MOS device characteristics, there is no advantage to substituting metal gates with heavily-doped polysilicon (poly) gates. In fact, poly gates, as will be shown in this chapter, greatly reduce the effectiveness of thinning the oxide layer to increase the drain current. The use of poly gates is a question of cost as well as performance, however, and poly gates have some tremendous processing and density benefits over metal gates. Polysilicon gates can withstand high temperature steps that would cause most deposited metal gates to evaporate, particularly the source/drain drive- in step. Polysilicon gates also allow for self-alignment of the gate over the oxide between the source and drain, removing what would be the most difficult (and costly) alignment step in the process flow [41-42], This chapter covers the derivation of a Fermi-Dirac-based polysilicon-gate MOS low-frequency capacitance-voltage model. This model will be used to illustrate the effects of polysilicon gates on MOS low-frequency (LF) capacitance-voltage (CV) characteristics compared to metal-gate LFCV characteristics. A useful application for the model is physical parameter extraction, which is demonstrated in this chapter using two 41 42 different methodologies: 3-point fit and 3-region fit. Sample parameter extractions for thick (130Ã) and thin (20Ã) gate oxides are shown, and discussion about limitations of the model are presented. Quantum effects are purposely ignored, and the reasoning behind this decision is discussed. Important details related to fast convergence of the parameter-extraction routines are also given. Metal-Gate CV Ideal metal-gate CV theory using Boltzmann statistics has been extensively discussed [3,43], as well the extension to include Fermi-Dirac carrier distribution and deionization effects [44-46]. The appendix contains the full metal-gate LFCV model derivation, taking Fermi-Dirac statistics and deionization into account. The relevant solutions are given below. Figure 3.1 shows a schematic diagram of an ideal metal-gate MOS device and the corresponding band diagram. From Figure 3.1 (b), as explained in the appendix, Kirchkoffâ€™s voltage law around the loop gives: Â®m + V0 = *s - VIX + (Ec - EjJ/q + VF + VG (3.1) where is the electron affinity of the substrate, Ec and E| are the conduction-band edge and intrinsic energies, respectively, in the substrate, and VF is the Fermi voltage, which is equivalent to (EÂ¡ - Fp)/q for p-type material, where FP is the quasi-Fermi-level for holes and q is the electron charge. Collecting these terms in cleaner form gives VG = V0 + VIX + Â®MS (3.2) 43 (A) VG + 1 Gate V//////////////A Substrate (B) Fig. 3.i MOS capacitor schematic and corresponding energy-band diagram. (A) Schematic diagram of a MOS capacitor and (B) corresponding energy- band diagram depicting the potential drops. Shown is a positive voltage VG applied at the gate, resulting in the Si02/Si surface entering inversion. 44 where 0MS = Â®M - Â®s = Â®m ~ Us + function difference between the metal and the semiconductor. As will be shown in the next section, the work function difference for a polysilicon-gate MOS device is much simpler than metal-gate MOS since the substrate and gate materials are the same. The drop across the oxide can be found from Gaussâ€™s Law requirements as Vg = CgEjj^/Cg (Qot Qit) /C0. (3-3) where es is dielectric constant of the semiconductor (~11.7x8.85xlO-l4F/cm2 for Si), Elx is the field across the oxide, C0 is the oxide capacitance, and Q0T and QIT represent fixed and interface trapped oxide charge respectively. With this relation, (3.2) can be rewritten as ~ VFB + VIX + Â£SEIX/C0, (3.4) where Vpg, the flat-band voltage, is given by VFB = 0MS â€” ÃQot + /C0- (3.5) For metal-gates, there is no capacitive contribution from the metal, so the gate capacitance is simply the series equivalent of the fixed oxide capacitance, C0, and the variable substrate capacitance, Cjx. Cg = CixCg/(Cix + Cg). (3.6) The field going into the substrate, EIX, and the substrate capacitance, Cix, are given by 45 2kT Nv[J3/2(-U: 7) â€” ^3 / 2 (â€”Uy+Up) ] Â«3 + nc[^3/.2( uIX+ucâ€”uF) â€” 7^/2 ( ucâ€”Up) ] pxx< UIX + ln 1 + gAexp (Up - UA - UIX) ll + gAexp (Up - UA) J Nxx<-Uix + ln 1 + gDexp(UD - Up + UIX) ll + gDexp(UD - Up) Civ = ~~Nv!E/2 ( â€”UIX Uv+Up) + NcX/2 (Uix+Uc Up) 1 + gAexp (Up - UA - UIX) (3.8) Lll + gDexp(UD - Up + UIX) where all â€˜Uâ€™ values are potentials normalized to kT/q and referenced to the intrinsic Fermi level. For example, Up is the normalized Fermi level, qVF/(kT). Pxx is acceptor substrate doping concentration and Nxx is the donor substrate doping concentration, and gA and gD are the corresponding degeneracy factors for the trap levels UA (acceptor energy level) and UD (donor energy level, not to be confused with the normalized drain voltage of an MOS transistor). Nv is the valance band density of states and Nc is the conduction band density of states. Those familiar with MOS capacitance equations might find these far more complex than they recall; a perusal of the appendix should clear up any questions about 46 this form. However, it is instructive to show how this reduces to a more familiar Boltzmann form. First, all of the Fermi-Dirac integrals [Fy2(r|) and Fj/jCn) terms] reduce to exponentials in the Boltzmann range of applied gate voltages (T| < -4). Second, there is typically only one dominant dopant, so one of the last two terms in (3.7) and (3.8) can be neglected (the first can be neglected for n-type substrate, and the second for p-type substrate). Furthermore, if deionization is neglected (UF - UA - UIX < -3 for p-type or UD - Up + UIX < -3) for n-type), then the last two terms of (3.7) reduce to PxxUiX - NXXU]X. Likewise, the two lines of (3.8) reduce to Pxx - Nxx when deionization is neglected. As an example of the simplified form, let us consider a p-type substrate in strong accumulation. In this case, it can be assumed that only the accumulated surface carrier term is dominant (UIX is large and negative). Noting also that, in the Boltzmann case, Up - Uv = ln(Pxx/Nv), (3.7) and (3.8) would reduce to = i| (2kTPxx/Â£s) exp (UIX/2) Clx = qj [PxxÂ£s/ (2kT) ] exp (UIX/2) These are the more tractable strong-accumulation forms found in undergraduate textbooks [3, 43] and which form the basis for one well-known oxide-thickness extrapolation algorithm [47], Polvsilicon-Gate CV Implicit in the derivation of the metal-gate CV theory above was that the capacitance of the gate is infinite and that the voltage drop across the gate is zero. With 47 metal gates, this is a reasonable assumption for the ideal isolated device. However, with polysilicon gates, there is a finite polysilicon gate capacitance as well as a voltage drop [3,49]. Indeed, the capacitor is now a semiconductor-oxide-semiconductor device, so it will have a corresponding surface potential for the gate, as well as an associated gate capacitance with a form exactly like the substrate capacitance. This requires only minor additional derivation to arrive at the poly-gate MOS capacitor (MOSC) ideal device characteristics. Figure 3.2 shows the band diagram for an n+-polysilicon gate MOS capacitor with a p-type substrate (a schematic of the device would be identical to 3.1 (a), with a metal gate replaced by a polysilicon gate). From this figure it is clear that the potential drop across the device can be given similarly to (3.1) as â€”^F-poiy + Assuming the energy gap has not narrowed due to the higher gate doping, the (Ec - terms are identical and cancel because the materials are both silicon. The electron affinity is the same for both the gate and substrate for the same reason. This reduces (3.9) to VG = v0 + vIX + VIG + VF + vF.poly. (3.10) Thus, for the poly-gate case, <|>MS (more aptly called <|>os, where â€˜Gâ€™ represents the gate, but still traditionally referred to as â€˜Mâ€™ for metal) is simply given by *ms = VF + VF_poly. (3.11) For Figure 3.2, <|>MS is given by ln(PxxNGG/nf), where Pxx is the substrate doping (â€˜Pâ€™ implying p-type) and N00 is the gate doping (â€˜Nâ€™ implying n-type). This simple formula assumes a Boltzmann carrier distribution in the substrate and gate, which is invalid in the 48 Fig. 3.2 Band diagram of n+ polysilicon-gate MOS capacitor with all the potential drops labeled. The band diagram shown depicts a positive voltage VG applied at the gate, with the Si02/Si surface entering inversion and the poly-Si/Si02 surface depleting. 49 gate due to the high doping and likely invalid substrate for modern ULSI devices. A more appropriate formula using inverse Fermi-Dirac integrals can be used using the examples in the appendix. The extra potential drop from the poly gate is easily taken into account via Kirkoffâ€™s law with VIG: VG = VFB + VIX + VIG + esEIX/C0, (3.12) where VFB from (3.5) still holds assuming negligible contribution from the polysilicon/oxide interface, using (3.11) for (|>MS. Finally, the gate capacitance formula needs to be extended for three capacitors in parallel. This changes (3.6) to Cg = CixC0Cig/(CixC0 + CixCig + CigC0), (3.13) where Cjg, the capacitance from the polysilicon gate, is given by -NvÂ£/;!(-UIG-Uv+UF) + NcK/2 (Uig+Uc-Uf) (3.14) This is simply (3.8) re-written with the band notation for the gate. Thus, UIG is the normalized surface potential in the gate, EIC is the field in the gate (defined below), and PGG> ngg- Uv. Uc, Nc, Nv, gA, gD, UD, UA are precisely as defined before, except that they apply now to the gate rather than the substrate. UF above was called UF poly PGG 1 + gAexp(UF - UA - UIG) Ngg 1 + gGÂ©xp (UB UF + U-G) 50 elsewhereâ€”it is left as UF in (3.14) to maintain the symmetry of the equation. The gate field is given by 2kT Nv[73/2<-Uig-Uv+Uf) - 53/2(-Uv+UF) ] + Nct^S/2^ UIG+UC UF) 1^/2 i Uc Up)] ^GG ( ^IX + ^-n 1 + gAexp (UF - UA - UIG) L 1 + gAexp (uF - UA) NGG (â€”Uix + In 1 + gDexp(UD - Up + UIG) 1 + gDexp(UD - Up) which is identical to (3.7) with the surface potentials changed. Again, the same caveat applies to (3.15)â€”all the terms refer to the gate now, not the substrate. Things like trap levels and band edges are nearly, if not exactly, the same in the substrate and polysilicon gate. However, UF is clearly quite different (assuming the gate and substrate are not doped identically, which would make a poor capacitor or transistor). An additional equation, which was not needed in the metal-gate case, is required to relate the gate and substrate. This equation equates the charge density at the gate/oxide interface with the charge density at the oxide/substrate interface: Â£s^ix + Qitx + Â£s^ig + Qitg = 0 â– Qitx is the interface charge at the substrate/insulator interface and QITG is the charge at the gate/insulator interface. It is assumed that these values are negligible, and that the dielectric constant for the silicon substrate and the silicon gate are identical (already implicitly assumed in the equation). This gives the following 51 EIX - -EIG' which allows the surface potential in the gate to be related to the surface potential in the substrate. The iterative solution of the above equation requires many calculations of (3.7) and (3.15), and is the most time-consuming part of the LFCV solver as well as any software using the routine (such as a parameter extractor which works by comparing the data to the theoretical curve, as discussed later in the application section). Polvsilicon-Gate Effects The effect of polysilicon gates, compared to metal gates, is a reduction of the gate capacitance, Cg, when the gate is in depletion. This is arises when the value of CÂ¡g falls below that of C0 and Cix, which only occurs during gate depletion and substrate inversion or accumulation, and only then to a significant degree for thin oxides. This is easily visualized from the three series capacitancesâ€”the one which dominates is the smallest, and the capacitance due to the substrate and gate are both minimized during depletion (and maximized during accumulation, as well as inversion for the LF case). As oxides thin, the oxide capacitance increases, which causes the effect of the substrate and gate depletion to have more control over the characteristics of the Cg-VG curve. Figure 3.3 shows the difference between metal-gate and polysilicon-gate data, normalized to C0, for two different technologies. The â€˜higherâ€™ pair of curves for a 1000 A oxide (thick oxide means low oxide capacitance) shows little difference between polysilicon gates and metal gates. The lower pair of curves for a 50 Ã oxide (thin oxide means large oxide capacitance) shows a large decrease in Cg for all values of VG, particularly for VG > 1V, where the gate is still in depletion and the substrate is inverted. 52 VQ / (1 V) Fig. 3.3 Comparison of metal-gate and n+ poly-gate MOSC curves for two different technologies. One set has 1000 Ã oxide with Pxx=3xl016 cm-3 and the second set has 50 Ã oxide with Pxx=2xl017 cm-3. In each case, the VFB is adjusted to be -1.0V and the gate doping is 3xl017 cm-3. Clearly shown is the dramatic difference between poly-gate (dotted line) and metal-gate (solid line) for the 50Ã case, and the negligible impact on the 1000 A caseâ€”the polysilicon gate effects increase as the oxide scales thinner. 53 This continual decrease in Cg for increasing VG (in this n+ poly-gate on p-Si substrate) is often referred to as â€˜poly depletion,â€™ since the polysilicon gate is still depleting. Eventually the gate itself will invert, and the characteristics will be much improved. However, resulting field caused by the gate voltage required to invert the gate is typically beyond the reliability limit of 4MV/cm in properly scaled devices. In fact, the only way to make the gate invert sooner is to lower the gate doping, which exaggerates the poly depletion effect even more until the gate inverts. It might seem, as it did to this author, that the ultimate solution would be to use undoped gates, as they would invert much sooner and behave just like metal gates at reasonably low applied gate voltages. This works well in simulation, but the question then becomes: where is the supply of minority carriers to invert the gate? In particular, for an n+gate in a rapidly switching MOST, what would supply the holes? It has been shown that, for at least one technology, the holes are likely supplied via thermal generation (rather than ion impact) [50]. Thermal generation, then, could not supply the holes fast enough for practical use of an undoped gate. However, it might be possible to design in a minority carrier source nearby to supply minority carriers (similar to how the source and drain supply minority carriers in the substrate). The reduction in the gate capacitance due to poly depletion causes a reduction in the drive current, which degrades circuit performance [51-54], since the amount of current supplied by the transistor directly relates to the switching speed of the device. In a complementary-MOS (CMOS) circuit, the current charged up the interconnect and intrinsic capacitances of the next transistors in the line, as discussed in detail in Chapter 54 4. Because of this poly-gate ID reduction, there may eventually be a move back to metal gate (or silicides) once the processing issues of gate alignment are solved. It is instructive to look at the individual capacitance components to see how the â€˜complexâ€™ poly LFCV curve forms. Figure 3.4 shows such a curve for a theoretical 50.0 Ã oxide with an n+ gate doped (rather lowly) to 9xl018 cm-3 and a substrate doped to 5xl017 cm-3. The gate area is lxlO-4 cm2 and the flat band voltage is â€”1.0 V. The Cg curve, being the serial sum of C0, Cix, and Cig (Eq. 3.17), is always lower than the component curves. It can be clearly seen how each of these three components influences the overall structure of the resulting gate capacitance. In fact, this â€˜regionalâ€™ effect will be used to help speed up parameter extraction in the next section. Also of interest is a breakdown of potentials across the MOSC device as a function of V0. Figure 3.5 shows the four components of VG, namely VIX, VIG, Vox, and Vpg (see Eq. 3.12) as a function of VG using the same parameters as the example in the last paragraph. To show show these are related to the resulting gate capacitance, the Cg-VG curve is also plotted. What is most relevant in this figure is that as the primary â€˜dipâ€™ in the CV curve occurs as the surface potential in the substrate, Vlx, sweeps from accumulation to inversion (i.e. moves from a small negative number to about one volt), and ends sharply as the surface potential approaches its maximum (strong inversion). Similarly, the secondary polydepletion â€˜dipâ€™ occurs as the gate surface potential, VIG, moves from accumulation to inversion (again, moves from a small negative voltage to around a volt). Note that the final surface potential in the gate is higher than that in the substrate (VIG > VIX when VG > 4V). This agrees with the common approximation that 55 VQ/(1 V) Fig. 3.4 Individual capacitance values for a theoretical 100x100 |lm nMOSC with a 50 Ã gate oxide, Pxx=5.0xl017 cm-3, NGG=9xl018 cm-3, T=300 K, and VFB=-1.0 V. This figure demonstrates how the three parallel capacitances (Cix, Cjg, and C0) add to give the overall gate capacitance. See Fig. 3.5 for the corresponding potential breakdown. 56 VG / (1 V) Fig. 3.5 Individual potential breakdown for a theoretical 100x100 pm nMOSC with a 50 Ã gate oxide, Pxx=5.0xl017 cm'3, NGG=9xl018 cm-3, T=300 K, and VFB=-1.0 V, along with the corresponding LFCV curve. Note how the surface potential in the substrate, VIX, increases rapidly in the range VG = -1 to 0 V as Cg increases (substrate inversion) and the similar increase in VIG in the range VG=1 to 3 V (gate inversion). See Fig. 3.4 for the corresponding capacitance breakdown. (dd 0/ 60 57 the surface potential pins to a little over 2VF, since the Fermi voltage in the gate will be larger than that of the substrate due to the greater gate doping. Parameter Extraction Using the LFCV Model Of the multitude of variables in the LFCV equations, most of them are known to a reasonable degree of accuracy (such as the dielectric constant, energy gap, conduction- band density, etc.), can be measured easily (temperature), or need not be known very accurately (acceptor and donor trap level) due to their small effect. This leaves the gate and substrate doping, the oxide thickness, and the flatband voltage as the â€˜unknownâ€™ parameters. These parameters may be extracted from experimental data by comparing experimental data to the theoretical model presented in this chapter. This may appear to be an easy task, since the equation need only be used, along with some data, in conjunction with a non-linear least-squares-solver. However, one will note that the polysilicon gate LFCV formula is doubly parametric (that is, is related through two parametersâ€”the surface potentials UIX and UIG), neither of which are known from the data. Thus, solving this problem is non trivial. The first step toward a solution, then, is to write a program which will calculate Cg given VG. This requires intensive calculations to find U|X and U|G for each VG, but can be done since there is only one unique solution. Thus, with a Cg(VG) routine written, a nonlinear least-squares-fit program can be used. The code written for this dissertation took advantage of the fact that, as the solution converges to values of the unknown parameters, the values of the surface potentials at each experimental data point could be 58 used for initial guesses for each subsequent iteration of VG to find each Cg (since the parameters {substrate and gate doping, oxide thickness, and flatband voltage)) should not be changing too rapidly). This greatly increased the convergence rate over estimating UIX and UjG on each call, at the expense of additional code complexity and memory usage. 3-Point Extraction Methodology If the model were perfect, then it would require only three points to match the experimental data to the model. Why only three data points for four parameters? Because the additional constraint that one of the points should be the minimum of the experimental LFCV curve can be used. From this information, the flatband can be found by comparing the VG of the theoretical minimum with the VG of the experimental minimum. The other three parameters can be found directly from the Cg values of the three points. Figure 3.6 shows the three points, labeled Cg_acc, Cg_depl, and Cg_dep2, as they relate to the whole LFCV curve. Only Cg_dep| is uniqueâ€”the other two points can be anywhere within their region. The Cg.acc point is a point from the LFCV gate accumulation region. From this, a good estimate of the oxide thickness can be found, since the other parameters have very little influence over this point (see Figs. 3.7 and 3.8). Cg.acc asymptotically approaches C0, which is inversely proportional to the oxide thickness, Tox, via the parallel plate formula. There has been much research in obtaining Tox and/or C0 from (substrate) accumulation CV data [48,55-58]. 59 Fig. 3.6 Example of a general polysilicon-gate (n-i- gate, p substrate for this case) showing how all the important regions can be labeled in terms of the gate state rather than the typical substrate state. This regional breakdown is used to improve the speed and accuracy of the parameter extraction routine. 60 The Cg_dep| point is the minimum of the LFCV curve, and allows us to find the substrate doping, since the substrate depletion region is strongly dependent on the substrate doping concentration. In fact, depletion CV data can also be used to determine the substrate doping profile [59-61], Figure 3.7 shows LFCV data for several different constant substrate doping concentrations, clearly demonstrating the strong dependence of substrate doping on the location of Cg_depi. This was also demonstrated in Fig. 3.4, since the main influence in this depletion region is CIX, which itself is strongly dependent on UF (see Eq. 3.8), which is directly related to the the inverse Fermi-Dirac integral (natural logarithm if assuming a Boltzmann distribution) of the substrate doping. The position of the minimum along the VG axis also allows us to estimate the flatband voltage by comparing the VG of the minimum of the theoretical curve to the VG of the data. The Cg.dep2 point is from the gate depletion region. Figure 3.8 shows that the gate doping has the most affect on this part of the curve, whereas Figure 3.7 shows that the substrate doping has very little effect in this region. For the n+ gate on p-substrate example in Figure 3.8, the substrate is in inversion. However, even if the substrate were n-type (and the substrate thus accumulated), Cg depletion would still occur because the gate would still be in depletion (of course, the entire curve would be shifted due to the flatband difference). Hence, this point is called Cg_dep2, with the â€˜depâ€™ in reference to the depleted state of the gate. By varying the parameters in the appropriate regions to match these three points, a unique parameter set will be obtained which will describe a theoretical LFCV curve passing through the three points. 61 Fig. 3.7 Effect of substrate doping changes (Pxx) on LFCV characteristics. The â€˜depletion-1â€™ region (see Fig. 3.6) is the region of largest impact. 62 Fig. 3.8 Effect of gate doping changes (NG0) on LFCV characteristics. The â€˜depletion-2â€™ region (see Fig. 3.6) is the region of largest impact. 63 3-Region Extraction Methodology As good as our model is, there are still several effects which are not being considered. These include retrograde doping in the substrate and quantum effects in the substrate inversion channel. Retrograde doping is commonly used for sub-'/2-micron design to maintain a high sub-surface doping concentration to prevent punchthrough, while still maintaining a low VT for low-V0 operation (to accommodate the thin oxides) [7]. Figure 3.9 shows an example of a retrograde profile from our internally-modified MINIMOS. The LFCV model assumes a constant doping profile in both the substrate and gate, and so deviation from this assumption will cause changes in the experimental LFCV curve relative to the theoretical model. Charge-carrier layer push-out due to quantum effects in the inversion and accumulation layers has been an area of much research [62-65]. Experimental verification of these quantum effects are invariably at low temperatures, where phonons will not broaden the quantum bands into a continuum. Although some amount of quantum effect is likely present, it is probably impossible to model correctly when one considers thermal broadening, Si02/Si interface roughness and transitional regions, non- random dopant distribution, and other non-idealities. These will all tend to broaden the electron levels into a more classical continuum. It has been noted that electrical and optical oxide thicknesses do not often agree, and the difference has been attributed to quantum effects. As will be discussed later, the effect is likely overestimated. More important, if there is a difference, it is the electrically effective oxide thickness (as determined from electrical experiments, such as CV) 64 Fig. 3.9 Sample of retrograde doping profile, showing low surface concentration (5xl016 cm-3) and higher bulk concentration (lxlO18 cm"3). 65 which is most important compared to the optical thickness (which is not what affects device performance). Due to these two main non-idealities (non-constant doping and quantum effects), there could be some dependence on the extracted parameters using only three points. That is, extracted parameters might be dependent on which points we choose for Cg_acc and Cg_dep2. To overcome this, the entire curve could be fit to the model. This would result in extremely long convergence times, as a partial derivative must be calculated for each variable at every point for every iteration. However, Figures 3.7 and 3.8 show that some parameters have no influence on the LFCV curve in certain gate-voltage regions. Thus, the information provided from their partial derivatives does not help convergence, and will actually slow down the convergence, not to mention waste time during the calculation. Instead of fitting all the data to the model, the data can be broken up into the same three regions suggested in Fig. 3.6 for the three-point fit. Then the model can be fit using only the parameter (or parameters) dominant in the specific region, thus greatly improving the convergence rate (since the data being used is most relevant). This adds complication to the coding, as the data partitioning into each region (discussed later) must be automated, and a different fitting routines must be created for each of the three regions (same model, but separate partial derivative calculations). Methodology Comparison Figure 3.10 shows a comparison of the fit using the 3-point and 3-region methods to experimental data for a 130Ã oxide from an industrial 100x100 |im MOST transistor. 66 VG / (1 V) Fig. 3.10 Theoretically generated curves compared to original LFCV data using the (A) three-point and (B) full-curve extractions to on an n+polysilicon gate, 100x100 |Xm nMOST. Extracted parameters were: (A) Xox=130Ã, Pxx=9.0xl016 cm-3, Ngg=5.8x1019 cm-3, and VFB=-1.06V. (B) Xox= 130Ã, Pxx=8.7xl016 cm'3, NGG=3.0xl019 cmâ€™3, and VFB=-1.01V. 67 The solid-line, of course, is the theoretical curve using the extracted parameters, and the â€˜xâ€™ marks are the data used for the extraction. The top (A) curve uses only the three points marked to fit the data while the lower (B) curve uses the full set of data. The RMS error (for the second curve), calculated from the square root of the sum of the squares of the difference between theoretical and experimental capacitance, divided by the square root of n â€” 5 (5 = degrees of freedom = 1 + number of fitting parameters), was 2.7%, an excellent fit for only four parameters. There is not much in literature to compare the â€˜goodnessâ€™ of these results, as there are not many capacitance models available (most compact models, such as BSIM3 [34], fit IV characteristics only, and do not consider capacitance extraction). Aggressively scaled MOSC device data is given in Figure 3.11. This shows two LFCV curves from a -20Ã oxide from an industrial MOST transistor, where the 20Ã was determined from some optical method, most likely ellipsometry. This is a rather complex figure, and requires some explanation. There are two experimental curves: one p+ polyÂ¬ gate and one n+ poly-gate, both on a p-well. Thus, in the V0 > 1V region, the n+ gate is in depletion, while the p+gate is in accumulation (clearly the curves were shifted to align them, as the flatband should differ by about a volt between the two curves, although a threshold adjustment implant would offset this somewhat). From looking at the data at 1.3V and assuming Cg = C0, they conclude that the effective oxide thickness is 33Ã for the depleted n+ gate, and 28.5A for the accumulated p+ gate device. This is a poor approximation, since the oxide thickness value would vary greatly at different points along the depleted curve. However, they correctly state that the difference between the 68 two is due to poly depletion, which is a reasonable statement when applied to that specific gate voltage only. Thus, they attribute a 4.5Ã reduction in effective oxide thickness due to polysilicon depletion at 1.3V, which is the operating voltage for that technology. They next assert that the difference between the accumulated-gate curve (28.5Ã based on assuming Cg = C0) and the optically measured oxide (20Ã) must be entirely due to quantum effects. This conclusion is wrong, as the quantum model most likely does not take all the effects mentioned previously into account (such as thermal broadening, interface roughness and transitional region, not to mention retrograde doping). More importantly, they are completely ignoring the severe error of using Cg = C0! Figure 3.12 shows the fit (using the three-point method) to the data in Figure 3.11. From this fit, the extracted oxide thickness is 24.4Ã. Compared to the industry-stated results, the same effective oxide thickness reduction due to poly (4.7Ã here versus 4.5Ã) is seen. However, by correctly accounting for the distribution (instead of assuming Cg = C0), an additional 4.1Ã reduction from the Fermi-Dirac distribution is also found (that is, from the fact that Cg < C0)! This leaves 4.4Ã of difference between the extracted oxide thickness of 24.4Ã and the optically measured thickness of 20Ã. This 4.4Ã difference may include some quantum effects, but it may also be due to optical errors, such as not accounting for the transitional layer properly [66-67] and/or some other effects (i.e. doping profile). Ellipsometry and other optical methods can not be used on the actual device (since the gate electrode is not transparent), so it does not measure the oxide thickness in the active part of the device, which may differ slightly due to the additional 69 C0 calculated from optical Xo=20.0Ã Fig. 3.11 P+poly/p-well and n+poly/p-well â€˜20Ãâ€™ industrial data. Data is shifted to align mÃnimums, and labelling refers to industrial interpretation of the two curves. Please see text for explanation of this breakdown. Compare to Fig. 3.12, which is the authorâ€™s interpretation of the same data after quantumless parameter extraction. 70 VG / (1 V) Fig. 3.12 Theoretically generated LFCV data three-point parameter extraction using data in Fig. 3.11. Extracted parameters are shown. Instead of attributing the difference between the optical thickness of 20.0Ã and the â€˜extrapolatedâ€™ thickness of 28.5Ã at VG=1.3V to quantum effects, we find that 4.1 A is due to the distribution function used (Fermi-Dirac), with the remaining 4.4Ã possibly due to error (in optical measurement and/or other factors). 71 processing. As far as parameter extraction is concerned, the most important factor should be agreement with electrical results, not optical. Although a three-point method was used here (to improve the match at the 1.3V point), a full fit to the data in Fig. 3.12 has about a 10% RMS error, which, considering the thin oxide and the fact that the substrate gate doping is not constant, is extremely good. An interesting side note, brought up during the proposal for this project, concerned fitting a Boltzmann model to the data instead of a Fermi-Dirac. The surprising result of this was a better fit (6.6% RMS error), but, as would be expected, a thicker extracted oxide thickness of 27.5Ã. This is an interesting result, as it shows that using the wrong model can appear to give â€˜betterâ€™ results (in terms of fit), even though the resulting parameters actually have greater error (due to the incorrect carrier distribution). Philosophically, the issue of oxide thickness is an interesting topic. Many people would argue that TEM is the only way to measure the â€˜trueâ€™ thickness. However, ignoring that this is a destructive and time-consuming technique, it only yields the thickness of that particular cross section. What is desired is the average oxide thickness, as it is the average oxide thickness which affects the amount of charge accumulated by an applied voltage in a MOSC. This is why Fowler Nordheim (FN) tunnelling is also not a particularly good methodâ€”it will always underestimate the oxide thickness since the tunnelling will occur in the thinner spots on the gate. Additionally, one would rather not stress the devices while trying to find the oxide thickness. One recent technique for ultra- thin oxide thickness determination is using quantum oscillations in the tunnelling gate current, which are caused by quantum interference of electrons in the oxide conduction 72 band [68]. This method potentially suffers from the same problems as FN, and additionally requires knowledge of the effective mass and oxide barrier height, the latter two of which add about 2.5Ã of error to the results [69], assuming they are known to within 5%. Convergence Speed-up Details Iteration stops when none of the fitting parameters (i.e. the extracted parameters) changes by more than 5x 1 (T4% between successive serial cycles (i.e. each parameter was fit, and none changed by more than 5xl0'4%). Below are several of the methods employed to speed up this convergence. The calculations involved in the extraction are extremely complex. Of greatest importance is the convergence speed of the poly LFCV model, which itself must converge on the two surface potentials just to give one Cg data point for a given VG. This one data point is used in the numerical partial derivatives for the non-linear least-squares- fit, which means the Cg(VG) is called twice for each experimental data point for each iteration! Because this model is called so frequently, it is important to keep track of all the converged surface potentials for each experimental data point so that the LFCV model has a good estimate for subsequent calls. The delta used for the numerical partial derivatives, as it turns out, has a major influence on the correctness of the fit. Since two of the parameters vary logarithmically (the substrate and gate doping), the actual parameters used during the fit are the logs of these parameters. Thus, the deltas used for the derivatives must be calculated differently. Another problem is that the delta used for the oxide thickness is dependant upon the 73 actual thickness of the oxide (that is, it should be different for a 50 Ã oxide compared to a 1000 Ã). The empirical results for the best deltas, as determined from analysis of the numerical derivatives, were A=I0â€œ7 for ln(Nxx), A=0.1 for ln(N0G), A= 10-4 for VFB- and 10~7xTox for Tox. For example, the partial derivative of Cg with respect to Tox is calculated from (Cg(VG)! - Cg(VG)2)/2A, where Cg(VG), is the gate capacitance calculated at some VG with an oxide thickness of Tox + A and Cg(VG)2 is the gate capacitance calculated at the same VG with an oxide thickness of Tox - A. The reason arbitrarily small values cannot be used, of course, is because the LFCV model itself is only accurate to about eight digits (less near flatband) due to its own internal convergence criteria [46]. Finally, to start the extraction, a reasonable initial guess must be made. The initial guess of the oxide thickness is simply Aeox/Cgmax, where Cgmax is the maximum gate capacitance in the dataset and A is the gate area. This is the standard first-order approximation based on Cg = C0. For the substrate doping initial guess, the asymptotic high-frequency CV formula for Cgâ€ž [43] is solved iteratively using the minimum and maximum Cg values from the data set. The gate doping is simply set to 3xl019 cm-3. With these three parameters approximately known, the flatband voltage is estimated from ^Grain_data " ^Gmiiuheory- Since the minimum of the CV data is not necessarily given, but is needed internally to estimate the flatband, the minimum three data points (in terms of Cg) are used to estimate the true Cg minimum based on the parabolic minimum formula [70]. This slightly improves the convergence, but not as much as would be expected, largely because the minimum of the CV curve is not very parabolic. 74 Because it was clear that convergence was slower as the results approached the final values, a â€˜trickâ€™ was developed to improve this end case. Whenever a trend was visible during a fit, the routine doubled the amount of the parameter increase. A trend, in this case, is defined as three successive moves of a parameter in the same direction for all the parameters (possibly different directions for different parameters). This cut down the number of iterations by about 20% in most cases. One thing which would have improved the speed of convergence greatly would be to use a simpler model for the Fermi-Dirac integral. The Cody-Thatcher model [71] is extremely accurate, but requires the quotient of ten exponentials from a Chebyshev approximation. This approximation was used instead of some other simpler (though less accurate) approximations [72-76] because it was desired to add as little error as possible from the Fermi-Dirac integral calculation. CHAPTER4 THE EFFECT OF INTRINSIC CAPACITANCE DEGRADATION ON CIRCUIT PERFORMANCE Introduction In this chapter, the relatively obscure subject of intrinsic capacitances will be discussed. The area of MOS intrinsic capacitance has received little attention over the years due to the difficulty of measurement and small impact relative to extrinsic capacitances such as interconnect and packaging. However, as the push toward higher density continues, the extrinsic capacitance is being reduced as much as possible to improve performance. This will eventually leave the intrinsic capacitance as the primary load in CMOS circuits, thus making this a topic worth studying now. After a discussion of the intrinsic capacitances which most effect CMOS circuits (Cgd and Cgs), direct experimental measurements of the effect of hot-carrier degradation on intrinsic capacitance will be discussed, and the results modeled. The impact of this degradation on circuit performance will be evaluated and shown to offset some of the losses due to ID degradation. Background The effect of hot-carrier degradation on the drain current, ID, has been studied intensely since Abbasâ€™s initial observation in 1975 [77]. Another intrinsic property of a 75 76 MOS transistor, the intrinsic capacitance, has a much shorter history of study with regard to hot-carrier degradation. The first systematic study of intrinsic capacitances was done by Sah [13] in 1964, which was used by Meyer in 1971 [78] in his widely referenced work. In his paper he defines the intrinsic capacitance between terminals as: dQx That is, the change in the charge at terminal x due to a change in voltage at terminal y. This definition applies to any two-or-more terminal device, but from now on will be used with respect to a 4-terminal MOS transistor. Thus, it is clear that there are 16 possible intrinsic capacitance terms for a 4-terminal MOS transistor. Please note that in this small-signal defination, all of the non-y terminals are virtual ground. Thus, dVy is referenced to ground (i.e. it is essentially relative to all the other terminals). At first thought, one might assert that there are only 8 possible capacitances since Cxy=Cyx. However, this is not true because our definition of intrinsic capacitances does not represent static capacitive values and are not reciprocal. Consider the two intrinsic capacitances Cgd and Cdg. Neglecting overlap capacitance, when the applied gate voltage is less than the gate threshold voltage, VGT, both of these capacitances should be zero (since both Cgd=dQg/dVd and Cdg=dQd/dVg are zero due to no existing channel). Once VGS > VGT (and VDS < VDSsat), Cgd and Cdg will both have some finite positive value when the channel forms. The interesting case is when VGS > VGT and VDS > VDSsat- Now there is a channel, but it is â€˜pinched offâ€™ near the drain end. Cdg (dQd/dVg) is non-zero since a change in the gate voltage still affects the charge associated 77 with the drain (Qd); Cgd (dQg/dVd) is zero since a change in the drain voltage has no affect on the gate charge since the drain is not connected to the channel due to the pinch- off. As clear as this seems now, both Meyer [78] and others [79] assumed that the capacitances should be reciprocal. These should not be confused with the small-signal circuit element terms, which are named the same way but actually are reciprocal by definition. Ward and Dutton [80] were the first to argue that the intrinsic capacitances were, in fact, non-reciprocal. The paper also stressed the importance of including all the capacitances, particularly the gate to bulk (Cgb) capacitance, which had been omitted by Sah, and hence Meyer. Ward and Duttonâ€™s charge-based model was a huge improvement at the time, as Meyerâ€™s model does not guarantee charge-conservation in circuit simulators (due to omitting Cgb), resulting in erroneous results for the simplest of circuits. Papers predating Meyerâ€™s work largely used discrete devices, and so authors logically argued that modeling the intrinsic capacitances would be useless since the capacitance from packaging and external circuitry would be vastly larger [81], Furthermore, there was no direct method to measure the data to verify the models. With the advent of integrated circuits, the primary capacitive load between CMOS circuit cells (i.e. an nMOS and pMOS inverter pair) became dominated by the intrinsic and interconnect capacitances, rather than the packaging and external circuitry. Thus, modeling the intrinsic capacitance (as well as interconnect) became important. Integrated circuits also hailed the need for compact models to simulate large numbers of transistors. One of the first compact models was CSIM [82] from AT&T Bell 78 labs. Surprisingly, the authors of this model stayed with the simple Meyer model, although argued that including the intrinsic capacitances was critical, particularly Cgd. Cgd accounts for most of the intrinsic capacitance load in CMOS circuits due to the Miller feedback effect [82]. Berkeleyâ€™s BSIM [33] built upon CS1M, also retained the Meyer model. BSIM2, however, corrected this deficiency by including a non-reciprocal intrinsic capacitance model. The BSIM3 [34] model moved from an strongly empirical d.c. model to a more physically-based model, but retained the unaltered a.c. model (including intrinsic capacitances) from BSIM2, suggesting a lag in a.c. model development. Current a.c. models are extremely poor. A great deal of additional research is needed before a.c. models become nearly as sophisticated as d.c. MOS current models. There are two reasons the a.c. models are so far behind the d.c. models. First, intrinsic capacitance data have only been available since the early 1980s, over twenty years after the first MOS transistor ID data. Second, until recently, external capacitances and interconnect capacitances dominated the total capacitive load, making the intrinsic capacitance fairly unimportant. However, as the transistor dimensions have decreased and substantial improvements in drain current density become difficult due to physical limitations, major efforts have been implemented to reduce the interconnect capacitances, such as low-k dielectrics. This has increased the impact of intrinsic capacitances in overall circuit performance and, with improved interconnect, could become the predominant capacitive load in the circuit. It is interesting to note that publications on intrinsic capacitance modeling have been increasing year-to-year since Ward and Duttonâ€™s work [83-90]. 79 Measurement of Intrinsic Capacitances Because direct measurement of the intrinsic capacitance is difficult, many of the first measurements were done with on-chip circuitry using reference capacitors [92- 93] or op-amps circuits configured as coulombers [94]. Eventually, external circuitry was used, including a lock-in amplifier connected to an HP 4145 (as a voltage source) [95] and later an off-the-rack LCR meter [96], such as the HP 4275 A. In this section, the measurement of a few of the intrinsic capacitances will be described. These can be done using an HP 4275 or HP 4276 (same equipment with different a.c. frequency ranges), or the newer HP 4284. The first discussion of using an LCR meter for the direct measurement of intrinsic capacitances was written by K. C.-K. Weng and P. Yang in 1985 [96]. In this letter, many of the important problems with measuring the intrinsic capacitances were discussed. The main problem is that LCR meters are not designed to measure intrinsic capacitances. There are two sets of terminals on the LCR meter: High and Low. The high port applies the d.c. bias as well as the superimposed a.c. test signal. The low port measures the resulting small-signal current. From the magnitude and phase difference of the current relative to the applied small-signal test voltage, the capacitance can be found. Unfortunately, the low port is a virtual a.c. and d.c. ground, so no d.c. bias may be applied to it. To measure Cgd (dQg/dVd), the high port is attached to the drain (to apply the dVd) while the low port is attached to the gate (to measure the dQg via the small signal-current, ig times dt). If Cgd is desired as a function of VGS, the problem becomes apparent: How can VGS be ramped if the gate is grounded? 80 The only solution, of course, is to independently bias the three terminals not connected to the low port, as shown in Figure 4.1 for aCgd measurement. Thus, two additional power supplies are required, along with the internal d.c. power supply in the LCR meter. These power supplies must be well calibrated with one-another to ensurethat no potential difference exists between them when the same voltage is programmed. The burden of negotiating the polarities of the theee power supplies, once worked out, can be easily programmed into an automated station. As an example of the polarity problem, consider the following: if Cgd at VGS=2 V, VDS=3V, and Vxs=0 (note: the device is active, with a current flowing from the drain to the source, unlike standard CV measurements, where the source, drain, and substrate are tied together) is desired, the source and substrate can be biased at -2 V and the drain can be biased at 1 V. Since the gate is virtual ground (VGS=0), it is easy to verify that the above applied voltages give the desired potential differences (VGS, VDS, and Vxs). There is nothing particularly odd about this configuration except that it differs from the traditional C-V measurements where the substrate is the ground reference instead of the gate. In the above case of Cgd, the source and substrate may be tied together to forego one of the power supplies in Figure 4.1. If Vxs not equal to zero was required, however, all terminals must be biased independently. Thus, if one is designing a measurement station where any of the possible intrinsic capacitances can be measured, three power supplies (including the internal one of the LCR meter) are necessary. 81 LCR Meter Fig. 4.1 Measurement configuration for Cgd. Requires LCR meter with internal d.c. power supply, as well as two additional external d.c. power supplies. 82 Measurement Configurations Although the standard textbook MOS device is symmetric with respect to interchanging the source and drain, production devices may be asymmetric. This asymmetry may be the result implant shadowing, drain and/or source engineering, or hot- carrier-induced degradation, among other possibilities. Implant shadowing is an interesting case, as it may result in the gate/source and gate/drain overlap regions being different lengths, as shown in Figure 4.2. While the resulting ID characteristics are symmetric (that is, the ID versus VDS characteristics are the same if the source and drain leads are swapped), the measured Cgd characteristics (as well as Cgs, Cdg, and Cds) are asymmetric. This occurs because the measured characteristics include the constant overlap component, as shown in the following simple equation: c â€” c + r ^-gtLmeasured ^ov_drain ^gd- The Cov drain term is composed of the constant overlap of the gate with the drain, as well as an inner and outer fringe component. These fringe components have been calculated theoretically [97], and assuming they are constant as a function of gate voltage introduces negligible error [96]. The value of the measured Cgd in subthreshold (where Cgd measured = Cov drain) has been used to estimate the length of the gate-to-drain overlap region [98], and with the drawn channel length know, these overlap values could be used to extract the effective channel length. When necessary, the â€˜normalâ€™ and â€˜reverseâ€™ configurations of Cgd and Cgs measurements will be specified. These are shown in Figure 4.3. qr1 or Cgdnorm refers to the â€˜normalâ€™ measurement mode, where the high port is applied to the drain for a Cgd 83 gate-^ Fig. 4.2 Simplified schematic of asymmetric gate overlap, which results in 'â€œ'o v_drai ir'-'ov_source â€¢ 84 Fig. 4.3 (A) C, gd_norm (B) C (C)C (D) C Measurement configurations for (A) Cgd in normal configuration mode; (B) Cgs in normal configuration mode; (C) Cgd in reverse configuration mode; and (D) Cgs in reverse configuration mode. 85 measurement. Cgdv or Cgd rev refers to the â€˜reverseâ€™ measurement mode, where the high port is applied to the source for a Cgd measurement. This is necessary because, for short- channel devices, the resulting Cov value (where Cov is Covdrain or Cov source) can become a significant fraction of the total effective intrinsic capacitance. Although perhaps not obvious now, Cgs = Cgd when VDS=0. However, due to the difference in Cov, Cgsmeasured may not equal Cgd measured. Figure 4.4 shows the Cgs and Cgd measurements in thenormal and reverse modes for a 20 x 20 pm device. Figure 4.5 shows the same measurements on a 20 x 0.40 pm device (effective channel length is 0.24 pm). Comparing the two figures clearly shows the negligible impact of Cov on the long-channel device Cgd and Cgs characteristics and the large impact on the short-channel device. In both cases, the Cgd and Cgs values are almost identical, as is the overlap-induced difference of about 3 fF (This 3 fF offset is not visible on the Ldrawn=20 pm device because it contributes less than 2% to the maximum capacitance, whereas the overlap contributes about 60% of the total measured maximum capacitance for the Ldrawn=0.40 pm device). Later in this chapter, the results of channel hot-carrier stress on Cgd and Cgs will be shown. Because channel hot-carrier stress is inherently asymmetric (since the damage occurs near the drain edge), it is necessary to lay down the above notation for later use. Sample Measurements For all capacitance measurements in this chapter, an HP 4828A LCR meter was used with a small-signal voltage was 400MHz at 60 mV peak-to-peak. These number were chosen after testing a wide range of a.c. signal voltages and frequencies 86 Fig. 4.4 Cgd_normâ€™ Cgd_revâ€™ CgSnorm, and CgS_rev versus VqÂ§ for a 20x20 pm MOST with VDS=0.0. Although it appears that all four curves are the same, there are actually two sets of curves, Cgdnorm/Cgsrev and C-gd_re\/Cgs_norm separated by 3 fF. Very little difference is seen because the overlap capacitances shift is much less than then the peak Cgd and Cgs values. Compare this with Fig. 4.5. 87 Fig. 4.5 ^'gÃº_:iorm' ^gd_revâ€™ ^gs_normâ€™ ^"gs_rev Versus V( .g for a 20x0.40 U111 MOST with VDS=0.0. Roughly 3 fF parallel shift of Cgs norm/Cgd rev and Cgs rev/Cgd norm is due to a difference in constant overlap capacitance between the source and drain. 88 to obtain the most accurate results. The 60 mV signal may seem a little large to those familiar with common C-V measurements, where 25 mV is typically used, but is actually on the low end of the 23 mV to 400 mV found in most intrinsic capacitance papers [95- 96,98-106]. Frequencies below 100 MHz result in extremely poor-resolution (noisy) intrinsic capacitance data, while frequencies above 500 MHz begin to show markedreduction due to series resistance. LCR-specific settings on the HP 4284A were a medium integration time with 8-cycle averaging. So far the measurement procedures and naming conventions of intrinsic capacitance have been discussed. Figures 4.4 and 4.5 showed sample measurements with VDs=0. Although this is the typical way capacitances are measured, the ability to measure the capacitance of active devices, where VDS > 0 when VGS > VGT (where VGT is the threshold voltage at which an inversion channel form between the source and drain), is important. Why is this capability important? Because in a real circuit, this will commonly occur. If a correct model for the behavior of an operating transistor is desired, then data from an active device is required. Indeed, without this data, it would be like trying to verify an IDsat model with data only taken in subthreshold! Examples of Cgd measurements on active devices are shown in Figures 4.6 and 4.7 for 20 x 20 pm and 20 X 0.40 pm as a function of VGS for VDS = 0.0, 0.5, and 1.0 V (vsx = 0.0V). Cgd transitions from Cov drain to a larger value once VDS < VDSsat, or the channel is no longer pinched-off. From a charge perspective, this means changes in VDS (dVd) cause changes in Qchanne|, which in turn cause changes in Qg (dQg), resulting in a (dd l) / p60 89 Fig. 4.6 VGS /(1 V) Cgd versus VGS for a 20 x 20 |im MOST with VDS=0.0,0.5, and 1.0 V. Cgd / (1 fF) 90 Fig. 4.7 Cgd versus VGS for a 20 x 0.40 |im MOST with VDS=0.0, 0.5, and 1.0 V. 91 Cgd. Thus, as VDS increases, the point at which this transition occurs also increases, as can be seen in the figures. As mentioned previously, Cgd is the most important intrinsic capacitance because, in a common-source configuration (which is the configuration for all CMOS circuits), the effective load is 2(Cgs + Cgd( 1 - Av)), where Av is the gain between the gate input and drain output (a large negative number). The next most important capacitance, based on the above load formula, is Cgs. Figures 4.8 and 4.9 show both Cgs and Cgd for a 20 x 20 pm and 20 x 0.40 pm as a function of VGS for VDS = 0.0, 0.5, and 1.0 V. Unlike Cgd, Cgs will have a finite value as long as VGS is greater than VGT, since the channel will always be connected to the source. At VDS=0, Cgs=Cgd since the channel charge is equally controlled by the source and drain. However, if VGS > VGT (channel forms) and VDS > VDSsat (drain pinched off), then the source terminal will actually control more than half of the channel charge, resulting in a rise in Cgs above the value at VDS=0. However, once VGS increases to a point that VDS < VDSsat, the drain is no longer pinched off, and the Cgs value begins to decline with increasing VGS as Cgd increases rapidly. This is clearly demonstrated in Figure 4.8 (and to a lesser extent in 4.9), where the decline in Cgs corresponds to the increase in Cgd. The model for Cgd and Cgs will be discussed later. Recalling the discussion about the overlap-capacitance shifting in the previous section, the capacitances shows in 4.8 and 4.9 are actually Cgdrra and Cgav in order to offset the effects of the overlap capacitance. (Fig. 4.5 shows why this was necessary) 92 Fig. 4.8 Cgd and Cgs versus VGS for a 20 x 20 Jim MOST with VDS=0.0,0.5, and 1.0 V. 93 Cgd versus VGS for a 20 x 0.40 (im MOST with VDS=0.0, 0.5, and 1.0 V. Fig. 4.9 94 Channel Hot-Carrier Stress Effects on and C,â€ž Because the intrinsic capacitances are somewhat difficult to measure, as well as the relatively small contribution of intrinsic capacitance on circuit performance in past generations, very little work has been done to investigate the impact of hot-carrier stress on intrinsic capacitance. Although the first report of hot-carrier degradation on ID was published in 1975 by Abbas and Dockerty [1], the first investigation of Cgd and Cgs degradation was not published until 1988 by Yao, Peckerar, Friedman, and Hughes [107]. Since then there have been several papers [102-106] by two research groups showing Cgd and Cgs degradation for various stress conditions. Only one paper, by Dai, Walstra, and Lee, [108] showed the impact of Cgd and Cgs degradation on circuit performance. This section will present those data, a model for the degradation [109], and additional supplementary information not released in that short paper. Transistors from a 0.35 pm CMOS technology for 2.5 V operation were used; the same devices shown throughout this chapter. Drawn channel lengths were 0.40 pm and 0.48 pm, with effective channel lengths of 0.24pm and 0.32pm respectively. Accelerated stress was performed using the following procedure: 1) Take unstressed (â€˜freshâ€™) ID versus VDS data from 0 to 2.5 V at Vos=2.5, 2.0, 1.5, and 1.2 V. 2) Take â€˜freshâ€™ Cgs (normal mode) for reference. 3) Take Cgd (normal mode) versus VGS from 0 to 2.5V at VDS=0.0,0.5, and 1.0 V. 4) Without re-probing, stress for exponentially longer times (see next paragraph for stress conditions), followed by capacitance measurements as in (3). 95 5) After the final stressed Cgd measurement, take Cgs measurement (normal mode) and then measure the final (â€˜stressedâ€™) ID versus VDS as in (1). The accelerated stress conditions were VGS = 1 V, VDS = 4 V for nMOS devices and VGS = -1 V, VDS = -4 V for pMOS devices, for a total stress time of 14.6 hr (twelve Cgd/Cgs measurements total). Hot carrier stressing is a complicated and much- debated topic. Although it is possible that some degradation mechanisms may occur at these high-voltage stress conditions which could never occur during normal operation, it is believed that this accelerated degradation of the Cgd data will still be indicative of what will occur over long-time operation at normal operating voltages. Forward-biasing the source to increase the drain current without greatly changing the drain-field profile could be used [110], but was not considered at the time the measurements were made and is not yet accepted practice at Intel Corporation, where these measurements were taken. In steps (1) and (5), the ID data were measured on a different apparatus (an automated prober). This is acceptable since there will be negligible measurement error from re-probing and measuring the ID curves. However, in steps (3) and (4), it is very important that the stress be performed without reprobing in a shielded probe box, preferably with the capacitance-measuring probes allowing the force and sense lines from the LCR meter high and low ports to connect right at the probe tip (i.e. at the transistor pads). After calibrating (zeroing) the LCR meter to account for the probe configuration capacitance, any additional reprobing can easily add several femtofarad to the measured capacitance, which is on the order of the degradation amounts (shown later). Although it is possible to integrate the IV measurements into the circuit, it is advisable to add as little 96 additional circuitry as possible due to the exceptionally low capacitance values being measured. Figure 4.1 shows the equipment set-up to measure the Cgd during stress. The previous section discussed the a.c. signal and frequency settings used for the measurements. Although Cgfv could be monitored, the actual time to measure the intrinsic capacitance curves is about five minutes, which would add an extra hour over the whole stress time. Furthermore, Câ„¢v, although interesting, is not a component which comes into play in actual circuit operation. Instead, the much more important intrinsic capacitance, CgÂ°rm, is measured before and after the stress, but not in situ because that would require either a manual reprobing, which is prohibitively long, or a switching matrix, which would could not be zeroed out properly with the LCR meter. An interesting idea for a new piece of equipment would be an LCR meter which allows several short and shunt zeros to be stored in the LCR meterâ€™s memory. This way the equipment could be zeroed through different configurations of the circuit and the software could then tell the LCR which particular zero â€˜setâ€™ to use before switching the circuit over. Using the methodology outlined above, nMOS and pMOS devices were stressed for exponentially increasing time spans between Cgd measurements. The total stress time was 14.6 hr, which when added to the measurement time of the in situ Cgd measurements, is the length of time between the end of the work day and the beginning of the next. 97 Figure 4.10 shows the results of this hot-carrier stress on Cgd at VDS = 0.0 V for a 20 x 0.40 |im device at each time time interval. Figure 4.11 shows the same situation for a 20 x 0.48 pm device. In both cases, it is clear that for VGs > 0.4 V, Cgd decreases with increasing stress time. The longer-channel (0.48 pm drawn) device exhibits less degradation simply because the drain current during stress is also smaller due to the longer channel length (since, to the first order, the drain current is proportional to 1/L). Because the degradation is due to interface trap generation, as discussed in the next section, the smaller the current results in a smaller fluence, and thus, fewer generated holes resulting in fewer interface traps. Also notable in Figures 4.10 and 4.11 is an increase in Cgd for VGS < 0.3 V. Intrinsic Capacitance Degradation Model Both the reduction in Cgd for VGS > 0.4 V and the increase for VGS < 0.2 V can be explained from by simple model [109]. The gate-to-drain capacitance is given by the following integral: WÂ»C0 rL Cgd = vac(x)dx, (4.1) vac J0 where v^ is the applied a.c. test signal. In the absence of a non-uniform charge density in the gate or at the gate interface (i.e. with a spatially constant QOT and Qrr), the applied test signal used to measure Cgd should fall uniformly across the channel, as shown in the straight "unstressed" curve of Figure 4.12. Put simply, the applied signal controls all the charge at the drain edge and progressively less as the signal drops across the channel. By Gaussâ€™s law, the charge in the channel must be balanced out by charge on the gate, Cgd / (1 fF) 98 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 VGS /(1 V) Fig. 4.10 Cgd versus VGS for a 20 x 0.40 |lm MOST with VDS=0.0 after hot- carrier stress at VGS=1.0 V and VDS=4.0 V. The different curves represent measurements taken during the stress and demonstrate a reduction in Cgd as a function of stress time. Last curve is 52440 seconds, or 14.6 hr of stress time. Cgd / (1 fF) 99 VGS /(1 V) Fig. 4.11 Cgd versus VGS for a 20 x 0.48 pm nMOST with VDS=0.0 after hot- carrier stress at VGS=1.0 V and VDS=4.0 V. The different curves represent measurements taken during the stress and demonstrate a reduction in Cgd as a function of stress time. Last curve is 52440 seconds, or 14.6 hours of stress time. Less degradation is seen for this longer-channel device compared to Fig. 4.10 simply because the current is lower; hence the fluence is lower for the same amount of time, resulting in less interface damage. 100 Fig. 4.12 Idealized picture of a MOS transistor with trapped positive charge and interface traps near the drain edge. Diagram below shows the drop of the a.c. test signal across the channel, and how it is affected by the trapped charge near the drain for weak-inversion and inversion regions. The area under the curve is proportional to Cgd, as shown in Eq. 4.1. 101 with the amount of charge at any location â€˜xâ€™ being vac(x)*C0*WÂ»dx. This gives the convenient asymptotic expression of Cgd='/2(WL)C0 for Cgd in strong inversion (with no trapped charge). During stress, interface traps are generated by hot-hole dehydronization of interface Si-H bonds near the MOS drain edge, as discussed by Sah [2], When the channel is strongly inverted, and all these interface traps are filled, the local threshold voltage near the drain will always be higher than the rest of the channel due to the filled interface traps offsetting the applied gate voltage. Thus, the conductance near the drain edge is lower for stressed devices. This means the vac(x) in (4.1) will drop more rapidly in the damaged drain region than in the rest of the channel, resulting in less total charge controlled by the drain and a lower Cgd value. This is clearly depicted in the "post-stress, inversion" curve of Figure 4.12, where the area under the curve (the value of the integral) is easily seen to be less than the unstressed case. The only remaining question is the slight increase in Cgd at the lower gate voltages. At these voltages, most of the capacitance is due to overlap. During stress, in additional to the interface trap generation, there may be some hole trapping in the oxide (from hot holes, generated in the depletion layer, injected over the Si02 barrier into the oxide). This adds a small amount of positive charge near the drain, which increases the conductivity near the drain, and hence increases the overall Cgd value, as shown pictorially in Figure 4.12 as "post-stress, weak inversion." Once Vos increases into inversion, however, the negative interface charges compensate and then exceed the small effect of trapped positive charge. In the case of the Cgd increase due to trapped holes, it is 102 likely that a weak channel forms near the drain, but does not extend to the source. This channel, which responds to changes in the a.c. test signal, results in the small Cgd increase with stress. As mentioned previously, Cgs is the second most important capacitance in terms of circuit performance after Cgd. Figure 4.13 shows the initial and final Cgs values for a 20 x 0.40 |im device. Cgs increases with stress, for precisely the same reason as Cgd decreases. That is, the localized negative charge caused by the interface traps results in decreased conductivity near the drain edge, which causes a larger overall value of vac across the channel. In the charge-control sense, more of the channel charge is controlled by the source (and, consequently, less is controlled by the drain, as has been already seen). The formula is identical to (4.1), with the limits swapped (of course, vac is now maximum at the source end and zero and the drain). Because of Miller multiplication, the effect on the total capacitive load due to the Cgs increase after stress is much less than the Cgd decrease, so the overall load decreases due to stress. Figure 4.14 shows the results of stress on a pMOS device. As would be expected, there is considerably less degradation due to the decreased current caused by the lower hole mobility. This lower mobility essentially results in fewer hot-holes, and consequently, less interface trap generation. It would appear that after some initial weak bond breaking, very little additional degradation occurs. This is also seen in pMOS drain current degradation, which is always less than the nMOS equivalent. Due to the results shown in Figure 4.14, the affect of degradation on p-channel devices will be ignored. /(1 fF) 103 VGS /(1 V) Fig. 4.13 Cgs versus VGS at VDS=0.0, 0.5, and 1.0 V before and after CHE stress on a 20 x 0.40 |lm nMOST at stressed at VGS=-1.0 V and VDS=-4.0 V for 14.6 hr. cgd / (1 fF) 104 Fig. 4.14 I I 1 I I I I I I I I l l l I l l l l I l i i i I l l i I -2.5 -2.0 -1.5 -1.0 -0.5 -0.0 0.5 VGS /(1 V) Cgd degradation for 0.40 Jim pMOST after stress at VGS=-1.0 V and Vd$=-4.0 V for 3.0 hr. Negligible change after the first 10 s of stress. 105 Degraded Circuit Simulation The degradation of Cgd has been demonstrated before [102-107]. This information on its own is interesting, but it is more important to assess the impact of this degradation on circuit performance. Otherwise, all that has been done is measurement without analysis. In this section, the effect of intrinsic capacitance circuit performance will be discussed by simulating degraded transistors in a ring oscillator. This will be compared to the standard methodology of only modeling the IV degradation without accounting for the intrinsic capacitance degradation. Before circuits can be simulated, individual transistors must be simulated. This is done by fitting experimental data to a device model. For the work in this dissertation, the BSIM3 model [34] was used, with some slight extensions made to the a.c. model to improve convergence. The BSIM3 model (version 3.3) is the current accepted SEMATECH industry standard. The initial and final ID-VDS data, shown in Figure 4.15, were fit to the BSIM3 d.c. model using appropriate parameters. The resulting parameters were saved in a parameter set; these parameter sets are later used when simulating the circuit. Thus, the initial and final IV curves were individually fit and the resulting parameter sets were saved. These will be referred to as the â€˜freshâ€™ and â€˜stressedâ€™ IV sets, respectively. The a.c. model used in BSIM3 is much less sophisticated than the d.c. model, and there is no automated extraction methodology for it. Thus, fitting Cgd and Cgs had to be done manually. The data fit was the VDS=0.0 data in Figures 4.10 and 4.13, again 106 resulting in â€˜freshâ€™ and â€˜stressedâ€™ CV sets for the t=0 (unstressed) data and the t=14.6 hr (stressed) data respectively. This allows us to compare the following three scenarios: (1) Circuit performance using fresh IV and fresh CV parameter sets. This will give the performance of an unstressed circuit. (2) Circuit performance using stressed IV and fresh CV parameter sets, as done currently in industry. This will give the performance of a stressed circuit without including intrinsic capacitance changes. (3) Circuit performance using stressed IV and stressed CV parameters sets, which compared to (2), will show us the effect of including the Cgd degradation (and Cgs enhancement). To compare these, a simple ring oscillator circuit will be used, which is comprised of some odd-number of CMOS inverters chained together. Figure 4.16 shows an example of a 31-stage ring oscillator, along with the individual CMOS inverter pair circuit. Consider a voltage Vcc applied to nMOS and pMOS gatesâ€”this will cause the nMOS device to turn on and the pMOS device to turn off, which results in ground (0 V) appearing at the drain lead of the nMOS device. Consider a voltage of ground (0 V) applied to the nMOS and pMOS gatesâ€”this will cause the pMOS device to turn on and the nMOS device to turn off, which results in Vcc appearing at the drain lead of the pMOS device. Thus, the CMOS circuit is an inverter since the voltage applied to the gate is inverted (in the logic sense of the term) at the output. Any odd connection of inverters will result in the voltage oscillating from high to low around the chain. This oscillation frequency in often used to demonstrate the impact of ID degradation on circuit (vw 0/al 107 VDS /(1 V) Fig. 4.15 Initial and final ID versus VDS characteristics for a 20 x 0.40 pm MOST. Stress condition was VGS=1,0V and VDS=4.0V for 14.6 hr. From these data come the â€˜freshâ€™ (initial) and â€˜stressedâ€™ (final) parameters sets. 108 Fig. 4.16 Ring oscillator. (A) CMOS inverter as pMOST and nMOST circuit, abbreviated by the logic symbol; (B) 31 -stage ring oscillator with fanout of three, composed of CMOS inverters and ideal (no capacitance or resistance) interconnect. 109 performance [43], because as ID drops due to degradation, it takes longer for a switching transistor to charge up the intrinsic capacitance of the next inverter (as well as the interconnect), which slows down the overall ring oscillation frequency. Using the three parameter sets discussed above, which were derived from the data already presented, three sets of ring oscillator waveforms, as shown in Figure 4.17, can be found. These represent the output node voltage of any given CMOS inverter in a 31-stage ring with a fanout of 3 (each output node connected to three CMOS inverter inputs), and are simulated using Intelâ€™s SPICE-like circuit simulator. The fanout is used to simulate a typical circuit, where one transistor drives multiple down-stream transistors. These addition transistors obviously add to the load. Interconnect capacitance is neglected since its impact is layout-dependent. The resulting oscillation frequencies for the Ldrawn=0.40 p.m ring were as follows for each parameter set: Fresh IV, Fresh CV 85.6 MHz Stressed IV, Fresh CV 80.2 MHz Stressed IV, Stress CV 82.1 MHz The above data clearly demonstrate that by including the stressed intrinsic capacitance, some of the ID degradation is offset, resulting in a higher post-stress operating frequency. This is simply because the capacitive load, which the ID must drive, is degrading simultaneously with ID. Note that the ID degradation in one device is being offset by the Cgd degradation in other devices in the next stage. The difference between the above examples may look quite small (80.2 MHz for the normal IV-only degradation set versus 82.1 MHz for our IV and CV degradation 110 set). To put it into perspective, consider the following example. Suppose one is designing an 85 MHz processor with a critical 31-transistor path which limits the maximum frequency. Furthermore, assume that the accelerated stress at 14.6 hr represents exactly 10 years of normal operation, which is the specification required for the 85 MHz processor. Finally, assume that this processor is required to remain within 5% of 85 MHz during its 10-year life (80.75 MHz to 89.25 MHz). Figure 4.18 shows pictorially what can be deduced from the table above, namely that the "Stressed IV, Fresh CV" set (industry normal methodology) will result in a predicted failure, since the resulting simulated frequency of 80.2 MHz is less that the 80.75 MHz guardband. However, when the "Stressed IV, Stressed CV" is used, the resulting 82.1 MHz simulated frequency is well within the guardband range, preventing unnecessary redesign and/or scrap (actually, the processor would probably be sold as a slower version at lower margin). Conclusion This chapter examined the effect of channel hot-carrier stress on the two main intrinsic capacitances in a common-source MOST CMOS circuit: Cgd and Cgs. From measurement of these curves before and after stress, along with the ID characteristics, fresh and stressed CMOS inverters were simulated, and the effect of stress on a CMOS- based ring oscillator was demonstrated. It was clearly shown that the inclusion of Cgd degradation offsets the well-known ID degradation by reducing the capacitive load the drain current must drive. It is important to note that the interconnect capacitance was ignored in this case, although it is quite large in reality. As transistors scale smaller, however, it is predicted by the SI A roadmap [111] that low-k dielectrics and lower resistivity interconnects will be used in an effort to reduce the RC delay from interconnect. As the interconnect capacitance is reduced, the intrinsic capacitance becomes more significant. The exact important of one over the other is a function of layout, and cannot be easily assessed. However, it is clear that the importance of intrinsic capacitances on circuit performance will only increase as efforts are made to reduce all extrinsic capacitance factors. / (1 V) 112 t/(1 ns) Expanded view Stressed IV Stressed CV Fresh IV I Stressed IV Fresh CV â–¼ Fresh CV Fig. 4.17 Beginning few cycles of a 31-stage ring oscillator circuit using three different parameter sets. The highest frequency curve (unstressed IV, unstressed CV) expectedly comes from the simulation using the two unstressed IV and CV parameter sets. The slowest frequency curve (stressed IV, unstressed CV) comes from the simulation using only the stressed IV, while the middle frequency curve (stressed IV, stressed CV) comes from using both the stressed IV and CV parameter sets. This demonstrates that inclusion of Cgd degradation results in circuit performance improvement due to offsetting some of the ID degradation. 113 Frequency (MHz) Fig. 4.18 Example demonstrating how including intrinsic capacitance can result in substantial benefits. Here it is assumed that the accelerated stress of 14.6 hr is equivalent to 10 yr of operation for a fictional 85 MHz processor with a Â± 5% allowed frequency deviation. By including Cgd degradation, the processor performs with-in specification, while by not including the intrinsic capacitance degradation (as is normally done), the processor is estimated to fail, possibly resulting unnecessary redesign/scrap. CHAPTER 5 SUMMARY AND CONCLUSION As MOS transistors scale smaller, previously unimportant or avoidable problems such as channel-length modulation, polysilicon gate depletion, and intrinsic capacitance degradation become significant. In the previous chapters, each of these scaling-related issues was discussed and the ramifications of the problems were demonstrated. In the first two cases, the problem was accounted for by extending previous theory to accommodate it. In the later case, the previous degraded-circuit simulation methodology was extended to show the unexpected benefits of including intrinsic capacitance degradation in circuit simulations. The history and derivations of the prominent long-channel current model was introduced so that the pros and cons of each could be discussed. The Pao-Sah current, by explicitly taking drift and diffusion into account, was shown to be the most accurate long- channel model, while the simpler charge-sheet model was shown to be nearly as good. Because these long-channel models do not take the drain encroachment into account, they need to be extended to be useful in todayâ€™s short-channel regime. Although the depletion region near the drain is 2-dimensional, the 1-D Pao-Sah model was extended to include the channel-shortening effect by dividing the MOS channel into two sections: an ideal 1- D long-channel portion and a 1-D drain depletion region to account for the channel-length modulation effect. The ideal long-channel portion used the Pao-Sah current model to 114 115 calculate the current, and three methods were proposed and implemented to find the boundary potential between the two sections. The most complicated method of matching the longitudinal fields was shown to be the only one capable of demonstrating smooth transitions in both the drain current and drain conductance at the point where the drain voltage exceeds VDSsat and the drain space-charge layer thickens. The other two methods, â€™saturation voltageâ€™ and â€™surface potential self-saturation,â€™ both showed the expected channel-length-modulation-induced saturation current increase as the channel length (for a square device) decreases, but the transition point near VDSsat in the ID versus VDS plot was abrupt enough to cause discontinuities in the drain conductance. The effect of polysilicon gate depletion on the MOS LFCV characteristics was demonstrated using a Fermi-Dirac based model. It was shown that, as the oxide thickness decreases, the effect of polysilicon depletion become increasingly pronounced. The purpose of thinning the gate oxide is to increase the carrier concentration for a given applied voltage, but polydepletion offsets an increasingly large portion of this gain, as does the Fermi-Dirac distribution (compared to the Boltzmann distribution). With this polysilicon-gate LFCV model, it was shown that the oxide thickness, flatband voltage, and gate and substrate doping concentrations could be extracted from experimental capacitance data. Two extraction methods, the 3-point and 3-region, were developed and were shown to work well with 130Ã (2.7% RMS fit) and sub 30Ã (10% RMS fit) data. Quantum effects were neglected because it is believed that thermal broadening, surface roughness, and non-random dopant distributions will all cause the localized states to broaden into a continuum. Details about the poly LFCV model were also discussed. 116 Intrinsic capacitance was predicted to become an increasingly large part of the capacitive load as the extrinsic capacitances, predominantly interconnect, are reduced in order to improve circuit performance. Measurements of the two most importance intrinsic capacitances, Cgd and Cgs, were performed on state-of-the-art 0.24 pm effective- channel-length nMOS and pMOS devices. Voltage accelerated stress of the nMOS devices via channel-hot electrons showed that Cgd decreases and Cgs increases with stress time, whereas the pMOS devices saw negligible change. Because of Miller feedback, however, the nMOS Cgd reduction dominates the Cgs increase, resulting in an overall CMOS load reduction. This load reduction offsets the drop in the drain current, both of which are caused the the degradation mechanism: interface traps. A model was given to qualitatively explain the Cgd reduction (and Cgs increase) with stress. The prestress and post stress ID, Cgd, and Cgs data were fit using the BSIM3 device model so that a stressed circuit could be simulated. A simulation of a 31-stage ring oscillator verified that the decrease in the capacitive load from Cgd reduction partially offset the ID reduction, resulting in improved simulated performance compared to simulations only taking the ID reduction into account (which is the standard industry practice). Although these simulations ignored interconnect capacitance, it is clear that the current ID-only method is conservative and, by taking Cgd degradation into account, the design guardbands can be loosened, resulting in less costly redesign and scrap. This dissertation has examined several of the current and future problems associated with scaling transistor characteristics and presented extended models and new methodology to account for their effects. It is clear that, for the short term, 1-D models 117 can be used to model the most important short-channel/thin-oxide deviations from simple theory. There is no reason why 1-D models cannot be used, with appropriate extensions and partitions, until transistors are replaced with a completely new technology, which is unlikely to happen in the next twenty years. Even as the 1-D models become less accurate, they will always retain importance for providing initial guesses for more complete 2- and 3-dimensional models, since the 1-D model will always embody a majority of the first-order device effects. APPENDIX METAL-GATE LFCV MODEL DERIVATION This appendix contains a complete derivation of the low-frequency, degenerate, deionized, Fermi-Dirac-based, metal-gate CV model [46], as used as a starting point for the polysilicon-gate model in Chapter 3. Similar derivations were made by Seiwatz and Green [44] and Hunter [45], Basic CV Equations The low-frequency CV (LFCV) characteristics of a metal-oxide- semiconductor (MOS) capacitor are fairly straightforward. Figure A. 1 (A) shows that the three layers which comprise the MOS name: the metal (gate), the insulator (Si02, or oxide), and the semiconductor (substrate). Chapter 3 shows the extension of this model to polysilicon gates, as well as the effects of the polysilicon gate on the device characteristics as well as the implications on device performance. Ideally, the CV curve model would have one equationâ€”the gate capacitance, Cg, as a function of the gate voltage, VG. At very least, it would be good to have a set of parametric equations, with VG and Cg as functions of some other parameter (such as the substrate surface potential, VIX). Due to the complexities of the mathematics, the latter is the best that can be done. 118 119 Gate Potential Two simple equations are required: one for the gate voltage, and one for the gate capacitance. Looking again at Fig. A. 1 (A), Kirkoff s voltage law requires VG = VM + V0 + VIX (A.la) That is, the voltage applied at the gate must be equal to the drops across the metal, the oxide, and the semiconductor respectively. Because the Fermi level of the metal does not necessary coincide with the Fermi level of the substrate (which is controlled by the dopant), there is an additional term, d>MS, the metal-to-semiconductor work-function difference,to account for this offset. This is best visualized on a band diagram, as shown in Fig. A. 1 (B) for an arbitrary positive applied voltage with a p-substrate MOS capacitor (MOSC). From Figure A. 1 (B), the voltage drops across the device are clearly: 4>m + V0 = Xs ~ VIX + Ec and E, are the conduction band edge and intrinsic energies, respectively, and VF is the Fermi voltage, which is equivalent to (E[ - Fp)/q. Collecting these terms in a form more like the previous equation gives Vo = V0 + VIX + the metal was included, which was purposely neglected in the band diagram and, hence, in Eq. A.lb. Ideally, the voltage drop across the gate will be OV. When a metal is used, the drop is effectively OV. Polysilicon gates introduce a depletion layer which causes 120 VG Fig. A.l MOS capacitor schematic and corresponding energy-band diagram. (A) Schematic diagram of a MOS capacitor and (B) corresponding energy-band diagram depicting the potential drops. Shown is a positive voltage V0 applied at the gate, resulting in the Si02/Si surface entering inversion. 121 a voltage drop, as well as extra capacitance. Chapter 3 deals with this important effect, while in this appendix it will simply be assumed that the metal is a perfect conductor; thus VM=0V, and is neglected in Eq. A.lc. The drop across the insulator, V0 (where the â€™Oâ€™ is for oxide, since Si02 is the prevalent insulator for silicon devices) will be found in the next section. The potential across the semiconductor, the surface potential Vlx, cannot be easily measured or found. This will be the unknown variable which relates (A. 1C) and Cg formula. From a strictly mathematical point of view, VIX (or the normalized equivalent, U,x) is the parametric variable for the two equations (VG and Cg). Oxide Potential Charge neutrality guarantees that the sum of the charges through the circuit in Fig A. 1 (A) is zero. Thus, Qg + Qot + Qit + Qs = 0- (A.2) Qg, the gate charge, is equal to the charge at the gate/insulator interface (by Gaussâ€™s theorem), so QG = le0E0l = Â£0V0/Tox = CGV0. Similarly, Qs = -Â£SEIX at the insulator/silicon interface, where E0 is the electric field at the gate/oxide interface, EIX is the electric field at the oxide/substrate interface, Tox is the oxide thickness, and C0 is the constant insulator (oxide) capacitance. Qox and QIT represent the trapped charge and interface charge, respectively, and e0 is the dielectric constant of the insulator. Substituting the above two relationships into (A.2) and solving for the oxide (insulator) potential gives Vo = Â«sEix/Co ~ (Qqt + Qit)/Qq- (A.3) 122 With this relation, (A.lc) can be rewritten as ÃQqt + Qit^ /Cq] + VIX + Â£SEIX/C0 (A.4) VFb + ^ix + Â«sEix/Cq, (A.5) where VFB, the flat-band voltage, is given by VFB - 4>ms (Qqt + Qit) /C0- (A.6) The terms of (A.5) are almost all known. es can be found in a handbook. C0 is a constant (at a constant temperature), and, in the case of deriving this formula, can be assumed known. V|X is the parametric variable discussed earlier, so there is only one unknown variable: EIX, the field across the semiconductor. Before this problem is rectified, the capacitance aspect of the CV derivation will be investigated. Gate Capacitance Looking again at Fig. A.l (A), the gate capacitance, Cg, seems almost trivial. It is simply the serial combination of the metal (gate-contact) capacitance (assumed to be infinite since a conductor has no space charge layer width), the insulator capacitance (C0), and the semiconductor capacitance (Cix), as shown in Fig. A. IB. The â€™ixâ€™ subscript designates the band bending from the interface (i) to the substrate (x), which is an important designation when polysilicon gates are used. Hence, the gate capacitance is 1/Cg = 1/Cq + 1/Cix (serial capacitance summation) (A.7) The capacitive contribution from a polysilicon-gate space charge layer adds another serial term, and is discussed in detail in Chapter 3. 123 Again there is only one unknown: the semiconductor space-charge capacitance. Equations for Cix and EIX will be derived in a following section, but first a formulae for electron and hole concentrations must be derived, since Cix and EIX will be functions of the carrier concentrations. Semiconductor Carrier Concentration Formulae Easily variable carrier concentration is what differentiates a semiconductor from an insulator or a conductor. In this section, the electron and hole concentrations will be derived using Fermi-Dirac (degenerate) "statistics." To find the carrier concentration in a semiconductor as a function of energy, two things are needed: the three-dimensional density of states, D3(E), and the distribution function, /(E). The relationship between the density of states and energy is approximately parabolic near the bottom (in E-k space) of the conduction band. The conduction electrons will tend to be near this minima, allowing us to use the following parabolic formula: [43] D3(E) = [47T(2m*/h2)3/2]j (E - Ec)dE. (A.8) The well-known Fermi-Dirac occupation function is given by /(E) = {1 + exp [ (ef-E)/kT]}-1, (A.9) where EF is the Fermi level. Examination of (A.9) shows that /(E=EF)=0.5. Thus, the Fermi level is the energy where half of the total electrons are contained in the levels below EF. A formula for the Fermi level is given later [(A.37) and (A.38)]. 124 Electron Concentration By integrating the product of (A.8) and (A.9) over the energy from the bottom of the conduction band (Ec) to free space (EVL [vacuum level]), the number of conduction electrons can be found. n ^VL f(E)D(E)dE Ec (A. 10) 2me 3/2 evl 1 E - Ec h2 Ec {1 + exp[(E - Ef)/kT]} Substituting e=(E - Ec)/kT (=> dE=kTde), rj=(EF - Ec)/kT, and noting that eVL Â» 1, (which means the upper integral range can be approximated by infinity), results in n = 4 n 2mekT 3/2 oo Â£1/s de h2 o U + exp [ e - r] ]} 2)TinekT 3/2 2 " e1/2 de h2 0 {1 + exp [e - tj] } = NcJ1/a(tJ) â€¢ (A. 14) Nc, known as the effective density of conduction band states, is given by 2 TnrukT = 2.51xl019(me/m)3/2(T7300)3/2 cm-3, is the Fermi-Dirac integral of the 'h order, which is shorthand for */a(n)= â€” \ n de {1 + exp [ e - tj ]} (A. 12) (A. 13) (A. 15a) (A. 15b) (A. 16) 125 Why is there a 2/\ n term in front of the integral? This notation of the Fermi-Dirac (FD) integral, proposed by Dingle [112], is from a family of FD integrals of the form 1 râ€ Ej dÂ£ = . (A. 17) r (j+l) â€˜ 0 {1 + exp [ Â£ - fj] } Thus, the 2l\ n term comes from T( 1.5)'1. The Dingle notation has a number of beneficial properties compared with the other FD notational family, called Sommerfeld notation, which differs only by the factor of T(j+l)-1. The most useful (Dingle-notation) property, when it comes to device physics, is d â€”7iW = Jj-iOi). (A. 18) d n which makes differentiating and integrating FD integrals quite simple, as will be seen later. When this Dingle notation is used, working with FD integrals becomes almost as easy as using exponentials (i.e., the outcome of using the Boltzmann distribution function instead of the Fermi-Dirac distribution function when deriving the carrier concentration). When Sommerfeld notation is used instead of Dingle, differentiation results in a r(j+l)/r(j) multiplicative term in (A. 18). Hole Concentration The derivation for the hole concentration is identical to that of the electron concentration, except that the density of states equation is referenced from the valence band and uses the effective hole mass. The occupation function for holes is (1 - /(E)], or, in other words, the holes are where the electrons are not. D3(E) = [47T(2mh/h2) 3/2] t| (Ey - E)dE (A. 19) 126 Ev {1- / (E) } D (E) dE (A.20) Ev' Ev and Evâ€™ are the top and bottom of the valance band, respectively. Using the same methods as above, and making the same wide-band approximation, the following equation results: 2trmhkT 3/2 2 e'1/! ds h2 o {1 + exp [ Â£' - tj â€™ ] } (A.21) = NvÂ£/2U)') - (A.22) e ' is (Ev - E)/kT, and r/' is (Ev - EF)/kT {= (Eg/kT - rj)}. This is a different, although equivalent, presentation than others have use [typically, p = Nv^/2 ( â€”rjâ€”Â£G)], and better shows the symmetry of the electron and hole distributions. The Nv term, known as the effective density of valance band states, is given by 27tmhkT i3/2 Nv = 2 (A.23a) h2 = 2.51xl019(mh/m)3/2(T/300)3/2 cm~3 . (A.23b) Comparing (A.23b) to (A. 15b) shows that the difference between the density of states for electrons in the valance band and the density of states for holes in the conduction band stems from the difference in the effective masses for electrons and holes. Semiconductor Relations Formulae for the electron and hole concentration were derived so that an equation for the field and capacitance across the semiconducting material could be found. 127 Both of these equations will be functions of the carrier concentrations, which, in turn, will be functions of the potential across the semiconductor, VIX (surface potential). Charge Density The charge density in the semiconductor is given by the equation p = q(-N + P - NA + PD - nT) . (A.24) The N and P terms are as given above (A. 14 and A.22), and the NA and PD terms are the ionized acceptors and donors, respectively. The nT term represents the contribution from trapped charge. Generally, it is assumed that all of the impurities are completely ionized in doped silicon because shallow level impurities are used. However, at low temperatures and/or high doping (> 1018 cm'3), incomplete impurity ionization occurs, and the approximation is no longer valid. For deep level impurities, deionization will become significant even at moderate doping and room temperature. For p-type impurities, the ratio of empty acceptors (NA) to filled acceptors (NA) is NA/NÂ° = (l/gA)exp( [Ep - EA]/kT) , (A.25) where gA is the degeneracy factor and EA is the acceptor energy level. Noting that NAA = Na + Na, (A.25) can be transformed into naa Na = N- = . (A.26) 1 + gAexp([EA - EF] /kT) A similar equation can be derived for n-type impurities, and is given by 128 Ndd Nd = NÂ£ = . (A.27) 1 + gDexp ( [EF - ED]/kT) These two equations take deionization into account. Generally, it is assumed that impurities are completely ionized, which, in p-type material, implies NA = NAA. This is a good approximation when T is large or Ep Â» EA. At very low temperatures, gAexp([EA-EF]/kT) will not be significantly less than 1, and deionization will occur even when the Fermi level is above the impurity level. For very high doping with even a shallow-level acceptor, EF will still lie below EA, causing deionization. For deep-level donors, Ep can easily fall below EA. Of course, this is mathematically apparent, but it is also physically intuitive. At very low temperatures, there will not be enough thermal energy to ionize the impurities, so deionization is expected. At high impurity concentrations, the dopant becomes a significant part of the composition and the impurity level becomes a non- negligible part of the band structure (also, the energy gap narrows, but that is a completely different problem). Impurity banding can also occur, but since an underlying assumption of these derivations is uniform doping, impurity banding will be neglected. Assuming that the semiconductor has negligible trapping, then (A.24) (charge density) becomes p = q(P - N - Na + PD) (A.28) Semiconductor Electric Field With (A.28), carrier concentration formulae (A. 14) and (A.22), and ionized impurity concentrations (A.26) and (A.27), the charge density can be written as a function 129 of energy. Using this, an equation for the semiconductor field, EIX, can be found via Poissonâ€™s equation. Starting from the d.c. steady-state Poisson equation in one dimension, Â£sdE/dx = -p, (A.29) where Â£s is the dielectric â€™constantâ€™ of silicon, E is the electric field, and p is as given in (A.28). Integrating by quadrature, noting E=-(dV/dx), gives [43] Â£sdE/dx = -Â£s(d/dx)(dV/dx) = -ea[(dV/dx)(d/dV)](dV/dx) = - (Â£s/2) (d/dV) (dV/dx)2 = - (Â£s/2) (dE2/dV) . (A.30) Thus, from (A.29) and (A.30) dE2 = (2/Â£s) pdv (A.31) Each side of the equation can be integrated from the surface to zero (E(x to 0 for the field term and UIX to 0 for the charge density term), noting that V=(kT/q)U [and dV=(kT/q)dU], 2kT f -^3/2 ^ Uv+UF) l?3/2 ( UV+UF) ] ^ci-^3/2^ Uix"tUcâ€”Up) - )?3/2 ( Uc-Up) ] 1 + gAexp (UF â€” UA â€” UIX) Naa( Uix + In 1 Utx + In 1 + gAexp (UF - UA) 1 + gDexp(UD - Up + UIX) 1 + gDexp(UD - UF) .(A.32) 130 Use was made of (A. 18) to integrate the FD functions. One might wonder how the U)x term ended up in (A.32) when there was no free-variable U (or V) in the original equations for the carrier concentrations or the ionized impurities. The surface potential, UIX, represents the amount of additional band bending of the silicon band at the Si/Si02 interface caused by the applied field. It was zero in the equilibrium, zero-field state, and not included in the equations. Semiconductor Capacitance Finally the semiconductor capacitance, Cix, is needed. For low-frequency, this is given by Cix = -P(UIX)/EIX. (A.33) The low frequency assumption means that the minority carriers can be generated quickly enough to follow the small-signal voltage. For transistors (MOSTs), LFCV curves result even when high (1MHz) frequencies are used in CV measurements because the (highly doped) source and drain will supply the necessary minority carriers to follow the a.c. signal, as long as the channel is short enough. Substituting (A. 14), (A.22), (A.26), and (A.27) into (A.28n) or (A.28p), and then placing the result in (A.33), gives -NvJ1/2(-UIX-Uv+UF) + NCK/S(UIX+UC-UF) r Naa - L 1 + gAexp(UF - UA - UIX) - r ^DD -i L 1 + gDexp(UD - UF + UIX) -1 (A.34) 131 A degenerate, deionized LF CV curve can now be generated by substituting (A.32) into (A.5) [the gate voltage formula] and (A.34) into (A.7) [the gate capacitance formula], where VG and Cg are generated in pairs as a function of VIX (or UIX). Fermi Level Throughout this derivation, it has been assumed that the Fermi level is known. Considering the degenerate, ionized, p-type case first, (A.22) give, assuming NAAÂ»nj, naa = NvJ^ÃœEv - EF]/kT). (A.35) For the intrinsic case ni = Nv^UEv - EJ/kT). (A.36) Solving these two cases for the Fermi level relative to the intrinsic level gives EF - Ej = kT [ (Hi/Nv) - (N^/Nv) ] , (A.37) where is the inverse Fermi-Dirac integral of the 'h order. The inverse Fermi-Dirac integral is analogous to a natural logarithm in the Boltzmann regime. It is important to note that, unlike logarithms, Jy,1 (X/Y) * ^/(X) - JC/.} (Y). Furthermore, (A.37) can be roughly approximated by kTyi^1 (/NA), but not -kT^^1 (NA/nÂ±). Analogous to p-type, the Fermi level the n-type, degenerate (Fermi-Dirac), fully-ionized case is given by Ep - Ex = kT[^7/(NDD/Nc) - ITi/,1 (rXi/Nj.) ] . (A.38) In the deionized region, there is no analytical formula for EF (or the normalized UF)â€”EF must be found iteratively. REFERENCES [1] C. T. Sah, "Evolution of the MOS Transistorâ€”From Conception to VLSI," Proc. |EEE, 76, 1280(1988), [2] C. T. Sah, Fundamentals of Solid-State Electronics - Study Guide. Singapore: World Scientific (1993). See Section 412 and Appendix B2. [3] S. Sze. Physics of Semiconductor Devices. New York: Wiley (1981). [4] R. Dennard, F. Gaensslen, H.-N. Yu, V. L. Rideout, E. Bassous, and A. LeBlanc, "Design of Ion-Implanted MOSFETs with Very Small Physical Dimensions," IF.F.F. J. Solid-State Circuits. SC-9, 256 (1974). [5] P. Chatterjee, W. Hunter, T. Holloway, and Y. Lin, "The Impact of Scaling Laws on the Choice of n-Channel or p-Channel for MOS VLSI," IEEE Electron Dev. Letters. EDL-1, 220 (1980). [6] B. Davari, R. Dennard, and G. Shahidi, "CMOS Scaling for High Performance and Low Power-The Next Ten Years," Proc. of the IEEE, 83, 595 (1995). [7] Y. Taur, D. Buchanan, W. Chen, D. Frank, K. Ismail, S.-H. Lo, G. Sai-Halasz, R. Viswanathan, H.-J. Wann, S. Wind, and H.-S. Wong, "CMOS Scaling into the Nanometer Regime," Proc. of the IEEE. 85, 486 (1997). [8] M. Bohr, "Interconnect Scalingâ€”The Real Limiter to High Performance ULSI," IEDM. 241 (1995). [9] J. E. Lilienfeld, "Method and Apparatus for Controlling Electric Currents," U.S. Patent 1,745,175. Filed Oct. 8, 1926, granted Jan. 18 (1930). [10] D. Kahng and M. M. Atalla, IRE-IEEE Solid State Devices Research Conference. Pittsburgh, (1960). [11] D. Kahng, "Silicon-Silicon Dioxide Surface Device," Bell Tele. Lab. Tech. Memorandum. Jan. 16, (1961). 132 133 [12] H. K. J. Ihantola, Stanford Electronics Laboratory. Rept. No. 1661-1, September (1961). [13] C. T. Sah, "Characteristics of the MOST," IEEE Trans. Electron Dev.. ED-11, 324(1964). [14] H. K. J. Ihantola and J. L. Moll, "Design Theory of a Surface Field-Effect Transistor," Solid-State Electron.. 7, 423 (1964). [15] R. Troutman, "VLSI Limitation from Drain-Induced Barrier Lowering," IEEE Trans. Electron Dev.. ED-26, 461 (1979). [16] H. C. Pao and C. T. Sah, "Effects of Diffusion Current on Characteristics of MOSTs," Solid-State Electron., 9, 927 (1966). [17] R. F. Pierret and J. A. Shields, "Simplified Long-Channel MOSFET Theory," Solid-State Electron.. 26, 143 (1983). [18] M. Persi and G. Gildenblat, "Computationally Efficient Version of the Pao-Sah Model with Variable Mobility," Solid-State Electron., 38, 1461 (1995). [19] C. T. Sah, Fundamentals of Solid-State Electronics - Solution Manual. Singapore: World Scientific (1996). See Appendix. [20] C. T. Sah and H. C. Pao, "The Effects of Fixed Bulk Charge on the Characteristics of MOSTs," IEEE Trans. Electron Dev.. ED-13, 393 (1966). [21] M. B. Barron, "Low-Level Currents in Insulated Gate Field Effect Transistors," Solid-State Electron.. 15, 293 (1972). [22] R. J. Van Overstraeten, G. Declerck, and G. L. Broux, "Inadequacy of the Classical Theory of the MOST Operating in Weak Inversion," IEEE Trans. Electron Dev.. ED-20, 1150 (1973). [23] J. R. Brews, "A Charge-Sheet Model of the MOSFET," Solid-State Electron.. 21, 345 (1978). [24] F. Van de Wiele, "A Long-Channel MOSFET Model," Solid-State Electron.. 22, 991 (1979). 134 [25] A. Nussbaum, R. Sinha, D. Dokos, "The Theory of the Long-Channel MOSFET," Solid-State Electron.. 27, 97 (1984). [26] R. D. Schrimpf, D. H. Ju, and R. M Warner Jr, "Subthreshold Transconductance in the Long-Channel MOSFET," Solid-State Electron.. 30, 1043 (1987). [27] V. G. K. Reddi and C. T. Sah, "Source to Drain Resistance Beyond Pinch-Off in MOSTs," IEEE Trans. Electron Dev., ED-12, 139 (1965). [28] T. L. Chiu and C. T. Sah, "Correlation of Experiments With a Two-Section- Model Theory of the the Saturation Drain Conductance of MOSTs," Solid-State Electron., 11, 1149(1968). [29] D. Frohman-Bentchkowsky and A. S. Grove, "Conductance of MOS Transistors in Saturation," IEEE Trans. Electron Dev., ED-16, 108 (1969). [30] G. Merckel, J. Borel, and N. Z. Cupcea, "An Accurate Large-Signal MOS Transistor Model for Use in CAD," IEEE Trans. Electron Dev., ED-19, 681 (1972). [31] A. Popa, "An Injection Level Dependent Theory of the MOS Transistor in Saturation," IEEE Trans. Electron Dev., ED-19, 774 (1972). [32] P. P. Guebels and F. Van de Wiele, "A Small Geometry MOSFET Model for CAD Applications," Solid-State Electron.. 26,2677, (1983). [33] B. J. Sheu, D. L. Scharfetter, P. K. Ko, M. C. Jeng, "BSIM: Berkeley Short- Channel IGFET Model for MOS Transistors," IEEE Solid-State Circuits. SC-22, 558(1987). [34] J.H. Huang, Z. H. Liu, M. C. Jeng, K. Hui, M. Chan, P. K. Ko, and C. Hu, "BSIM3 Manual (version 2.0)," UC Berkeley (1994). [35] M. Miura-Mattausch and H. Jacobs, "Analytical Model for Circuit Simulation with Quarter Micron MOSFET: Subthreshold Characteristics," Japanese J. Appl. Physics. 29, L2270 (1990). [36] M. Miura-Mattausch, "Analytical MOSFET Model for Quarter Micron Technologies," IEEE Trans. CAD. 13, 610, (1994). 135 [37] M. Miura-Mattausch, U. Feldmann, A. Rahm, M. Bollu, and D. Savignac, "Unified Complete MOSFET Model for Analysis of Digital and Analog Circuits,â€ IF.F.F. Trans. CAD. 15, 1, (1996). [38] H. C. H. Pao, "Theoretical and Experimental Investigations of MOSTs,â€ Solid- State Electronics Laboratory, Technical Report #5, University of Illinois (1966). [39] H. Katto and Y. Itoh, "Analytical Expressions for the Static MOS Transistor Characteristics Based on the Gradual Channel Model," Solid-State Electron.. 17, 1283 (1974). [40] S. Selberherr, A. SchÃ¼tz, and H. Potzl, "MINIMOS - A Two-Dimensional MOS Transistor Analyzer," IEEE Trans. Electron Dev., ED-27, 1540 (1980). [41] J. Sarace, R. Kerwin, D. Klein, and R. Edwards, "Metal-Nitride-Oxide-Silicon Field-Effect Transistors, with Self-Aligned Gates," Solid-State Electron,. 11, 653 (1968). [42] R. Bower and R. Dill, "Insulated Gate Field Effect Transistors Fabricated Using the Gate as Source-Drain Mask," IEEE IEDM, 102 (1966). [43] C. T. Sah, Fundamentals of Solid State Electronics. Singapore: World Scientific (1991). [44] R. Seiwatz and M. Green, "Space Charge Calculations for Semiconductors," L Appl. Phvs.. 29, 1034 (1958). [45] W. Hunter, "Thin Oxide MOS Capacitor Studies of Fast Surface States," Technical Report #15. Solid-State Electronics Laboratory, University of Illinois, Urbana (1971). [46] S. Walstra, "CV-Based Oxide Thickness Extrapolation Procedure for Thin-Oxide MOS Devices," M.S.E.E. Thesis, University of Florida, Gainesville, August (1994). [48] M. J. McNutt and C. T. Sah, "Determination of the MOS Oxide Capacitance," J. Appl. Phvs.. 46, 3909 (1975). [49] G. Yaron, D. Frohman-Bentchkowsky, "Capacitance Voltage Characterization of Poly Si-Si02-Si Structures," Solid-State Electron.. 23, 433 (1988). 136 [50] S-W Lee, C. Liang, C-S. Pan, W. Lin, and J. Mark, "A Study on the Physical Mechanism in the Recovery of Gate Capacitance to Cox in Implanted Polysilicon MOS Structures," IEEE Electron Device Letters. 13, 2 (1992). [51] R. Chapman, C. Wei, D. Bell, S. Aur, G. Brown, R. Haken, "0.5 Micron CMOS for High Performance at 3.3V," IEEE IEDM, 52 (1988). [52] C. Wong, J. Sun, Y. Taur, C. Oh, R. Angelucci, and B. Davari, "Doping of N+ and P+ Polysilicon in a Dual-Gate CMOS Process," IEEE IEDM, 238 (1988). [53] C-Y Lu, J. Sung, H. Kirsch, S. Hillenius, T. Smith, L. Manchanda, "Anomalous C-V Characteristics of Implanted Poly MOS Structure in n+/p+ Dual-Gate CMOS Technology," IEEE Electron Device Letters. 10, 192 (1989). [54] K. Chen, M. Chan, P. Ko, C. Hu, J-H Huang, "Polysilicon Gate Depletion Effect on IC Performance," Solid-State Electron,. 38, 1975 (1995). [55] J. Maserjian, G. Petersson, and C. Svensson, "Saturation capacitance of thin oxide MOS structures and the effective surface density of states of silicon," Solid-State Electron.. 17, 335(1974). [56] B. RiccÃ³, P. Olivo, T. Nguyen, T. S. Kuan, and G. Ferriani, "Oxide-Thickness Determination in Thin-Insulator MOS Structures,â€ IEEE Trans. Electron Dev.. 35, 432 (1988). [57] S. Walstra and C-T Sah, "Thin Oxide Thickness Extrapolation from Capacitance- Voltage Measurements," IEEE Trans. Electron Dev., 44, 1136 (1997). [58] S. Walstra and C-T Sah, "Extension of the McNutt-Sah Method for Measuring Thin Oxide Thicknesses on MOS Devices," accepted for publication in Solid- State Electron. (1997). [59] C. O. Thomas, D. Kahng, and R. C. Manz, "Impurity Distribution in Epitaxial Silicon Films," J. Electrochem. Soc.. 11, 1055 (1962). [60] W. Johnson and P. Panousis, "The Influence of Debye Length on the C-V Measurement of Doping Profiles," IEEE Trans. Electron Dev,. 18, 965 (1971). [61] K. Lehovec, "C-V Profiling of Steep Dopant Distributions," Solid-State Electron,. 27, 1097 (1984). 137 [62] F. Stern and W. Howard, "Properties of Semiconductor Surface Inversion Layers in the Electric Quantum Limit,â€ Physical Review. 163, 816 (1967). [63] T. Ando, A. Fowler, and F. Stern, "Electronic Properties of Two-Dimensional Systems" Rev. Modern Physics, 54, 437, (1982). [64] J. SuÃ±Ã©, P. Olivo, and B. RiccÃ³, "Quantum-Mechanical Modeling of Accumulation Layers in MOS Structure," IEEE Trans. Electron Dev.. 39, 1732 (1992). [65] M. J. Van Dort, P. Woerlee, and A. Walker, "A Simple Model for Quantisation Effects in Heavily-Doped Silicon MOSFETs at Inversion Conditions," Solid-St. Electron., 37, 411 (1994). [66] S. Chongsawangvirod, E. Irene, A. Kalnitsky, S. Tay, and J. Ellul, "Refractive Index Profiles of Thermally Grown and Chemically Vapor Deposited Films on Silicon," J. Electrochem. Soc.. 137, 3536 (1990). [67] S-C Kao and R. Doremus, "Silica Films on Silicon: Thickness Measurement by Electron Microscopy and Ellipsometry," J. Electrochem. Soc.. 141, 1832 (1994). [68] E. Wu, S. Lo, W. Abadeer, A. Acovic, D. Biuchanan, T. Furukawa, D. Brochu, and R. Dufresne, "Determination of Ultra-Thin Oxide Voltages and Thickness and the Impact on Reliability Projection," IEEE IRPS, 184 (1997). [69] S. Zafar, K. Conrad, Q. Liu, E. Irene, G. Hames, R. Kuehn, and J. Worthman, "Thickness and Effective Electron Mass Measurements for Thin Silicon Dioxide Films Using Tunneling Current Oscillations," Appl. Phvs, Lett.. 67, 1032 (1995). [70] P. Bevington, Data Reduction and Error Analysis for the Physical Sciences. New York:McGraw-Hill (1969). See Figure 11-4. [71] W. J. Cody and H. C. Thacher, Jr., "Rational Chebyshev Approximations for Fermi-Dirac Integrals of Orders -1/2, 1/2, and 3/2," Math, Comput.. 21, 30 (1967). [72] N. G. Nilsson, "An Accurate Approximation for the Generalized Einstein Relation for Degenerate Semiconductors," Phvs. Stat. Sol.. 19, K75 (1973). [73] J. S. Blakemore, "Approximations for Fermi-Dirac Integrals, Especially the Function Used to Describe Electron Density in a Semiconductor," Solid-St. Electron,. 25, 1067 (1982). 138 [74] P. Van Halen and D. Pulfrey, "Accurate, Short Series Approximations to Fermi- Dirac Integers of Order -1/2, 1/2, 1, 3/2, 2, 5/2, 3, and 7/2,â€ J. Appl. Phvs.. 57, 5271 (1985). [75] Y. Chang and A. Izabelle, "Full Range Analytic Approximations for Fermi Energy and Fermi-Dirac Integral Â£j/a in terms of %h" J. Appl. Phvs.. 65, 2162, (1989). [76] S. A. Wong, S. P. McAlister, and Z. M. Li, "A Comparison of Some Approximations for the Fermi-Dirac Integral of Order V2," Solid-St. Electron.. 37, 61 (1994). [77] S. Abbas and R. Dockerty, "Hot-Carrier Instability in IGGETâ€™s,â€ Applied Physics Letters, 27. 147(1975). [78] J. Meyer, "MOS Model and Circuit Simulation," RCA Review. 32,42 (1971). [79] C. Viswanathan and M. Levy, "Modelling Inter-Electrode Capacitances in a MOS Transistor,â€ IEDM. 38 (1979). [80] D. Ward and R. Dutton, "A Charge-Oriented Model for MOS Transistor Capacitances," IEEE J. Solid-State Circuits. SC-13, 703 (1978). [81] H. Schichman and D. Hodges, "Modeling and Simulation of Insulated-Gate Field- Effect Transistor Switching Circuits," J, Solid-State Circuits. SC-3, 285 (1968). [82] S. Liu and L. Nagel, "Small-Signal MOSFET Models for Analog Circuit Design," IEEE .1. Solid-State Circuits. SC-17, 983 (1982). [83] J. Robinson, Y. El-Mansy, and A. R. Boothroyd, "A General Four-Terminal Charging-Current Model for the Insulated-Gate Field-Effect Transistor-I," Solid- State Electron.. 23,405 (1980). [84] Y. El-Mansy, and A. R. Boothroyd, "A General Formulation of the Parameters of the Equivalent Circuit Model of the IGFET-II," Solid-State Electron,. 23, 411 (1980). [85] H. Park, P. Ko, and C. Hu, "A Measurement-Based Charge Sheet Capacitance Model of Short-Channel MOSFETs for SPICE," IEDM. 40 (1986). 139 [86] S. Chung, "A Charge-Based Capacitance Model of Short-Channel MOSFETs," IEEE Trans. CAD. 8, 1 (1989). [87] W. Budde and W. Lamfrled, "A Charge-Sheet Capacitance Model Based on Drain Current Modeling," IEEE Trans. Electron Dev.. 37, 1678 (1990). [88] H. Park, P. Ko, and C. Hu, "A Charge Sheet Capacitance Model of Short Channel MOSFETs for SPICE," IEEE CAD. 10, 376 (1991). [89] M. El-Nokali and A. Afzali-Kushaa, "A Subthreshold Model for the Analysis of MOS ICs," Proc. Ninth Biennial Universitv/Government/Industrv Microelectronics Svmp., 169(1991). [90] K. Rho, K. Lee, M. Shur, T. Fjeldly, "Unified Quasi-Static MOSFET Capacitance Model." IEEE Trans. Electron Dev., 40, 131 (1993). [91] D. Cho, S. S. Kang, K-H Kim, S-H Lee, "An Accurate Intrinsic Capacitance Modeling for Deep Submicrometer MOSFETs," IEEE Trans. Electron Dev,. 42, 540(1995). [92] H. Iwai and S. Kohyama, "Capacitance Measurement Technique in High Density MOS Structures," IEDM. 235, (1980). [93] H. Iwai, J. Oristian, J. Walker, and R. Dutton, "Small Geometry MOS Transistor Measurements and Observed Short and Narrow Channel Effects," VLSI Svmp. Technol.. 78(1984). [94] J. Paulos, D. Antoniadis, and Y. Tsividis, "Measurement of Intrinsic Capacitances of MOS Transistors," IEEE International Solid-State Circuits Conf.. 238 (1982). [95] B. Sheu, P. Ko, and F. C. Hsu, "Characterization of Intrinsic Capacitances of Small-Geometry MOSFETs," IEDM. 80 (1984). [96] K. Weng and P. Yang, "A Direct Measurement Technique for Small Geometry MOS Transistor Capacitances," IEEE Electron Dev. Letters. EDL-6.40 (1985). [97] E. Greeneich, "An Analytical Model for the Gate Capacitance of Small-Geometry MOS Structures," IEEE Trans. Electron Dev.. ED-30, 1838 (1983). 140 [98] T. Chan, A. Wu, P. Ko, and C. Hu, "A Capacitance Method to Determine the Gate-to-Drain/Source Overlap Length of MOSFETs," IEEE Electron Dev Letters.. EDL-8, 269(1987). [99] D. Schmitt-Landsiedel and G. Dorda, "Novel Hot-Electron Effects in the Channel of MOSFETâ€™s Observed by Capacitance Measurements," IEEE Trans. Electron Dev.. ED-32. 1294(1985). [100] H. Ishiuchi, Y. Matsumoto, S. Sawada, and 0. Ozawa, "Measurement of Intrinsic Capacitance of Lightly Doped Drain (LDD) MOSFETâ€™s," IEEE Trans. Electron Dev.. ED-32,2238(1985). [101] Y. Yeow, "Measurement and Numerical Modeling of Short-Channel MOSFET Gate Capacitances," IEEE Trans. Electron Dev.. ED-35, 2510 (1987). [102] Y. Yeow, C. Ling, and L. Ah, "Observation of MOSFET Degradation Due to Electrical Stressing Through Gate-to-Source and Gate-to-Drain Capacitance Measurement," IEEE Trans. Electron Dev. Letters. 12, 366 (1991). [103] C. Ling, Y. Yeow, and L. Ah, "Characterization of Charge Trapping in Submicrometer NMOSFETs by Gate Capacitance Measurements," IEEE Trans. Electron Dev. Letters. 13, 587 (1992). [104] C. Ling, S. Tan, and D. Ang, "A Study of Hot Carrier Degradation in NMOSFETs by Gate Capacitance and Charge Pumping Current," IEEE Trans. Electron Dev., 42, 1321 (1995). [105] C. Ling, D. Ang, S. Tan, "Effects of Measurement Frequency and Temperature Anneal on Differential Gate Capacitance Spectra Observed in Hot Carrier Stressed MOSFETs," IEEE Trans. Electron Dev., 42, 1528 (1995). [106] R. Ghodsi, Y-T Yeow, C. Ling, and M. Alam, "Arriving at a Unified Model for Hot-Carrier Degradation in MOSFETs Through Gate-to-Drain Capacitance Measurement," IEEE Trans. Electron Dev.. 41,2423 (1994). [107] C. Yao, M. Peckerar, D. Friedman, and H. Hughes, "On the Effect of Hot-Carrier Stressing on MOSFET Terminal Capacitances," IEEE Trans. Electron Dev.. 35, 384(1988). [108] C. Dai, S. Walstra, S-W Lee, "The Effect of Intrinsic Capacitance Degradation on Circuit Performance," Svmp. VLSI Technology. 196 (1996). 141 [109] R. Ghodsi, Y. Yeow, M. Alam, "Numerical Simulation of MOSFETâ€™s Gate Capacitances for the Evaluation of Hot-Carrier Generated Interface States and Trapped Carriers," Simulation of Semiconductor Devices and Processes: Vol. 5 (1993). [110] C. T. Sah, A. Neugroschel, and K. M. Han, "Current-Accelerated Channel Hot Carrier Stress of MOS Transistors," submitted to Electronic Letters (1997). [111] Semiconductor Industry Association, The National Technology Roadmap for Semiconductors (1994). [112] R. B. Dingle, "The Fermi-Dirac integrals" Appl. Sci. Res. B, 6, 225 (1957). BIOGRAPHICAL SKETCH Steven V. Walstra was born in Calument City, Indiana, in 1970. He received the Bachelor of Science degree in electrical engineering (high honors) in 1992 and the Masters of Science degree in electrical engineering in 1994, both from the University of Florida. His masters thesis work concerned thin oxide-thickness extrapolation from capacitance-voltage measurements. He will receive the Doctor of Philosophy degree in electrical engineering from the University of Florida in 1997 under the guidance of Dr. Chih-Tang Sah. During his doctoral studies, which included polysilicon-gate LFCV modeling and parameter extraction, two-section MOS transistor drain current modeling, and intrinsic capacitance degradation, he was awarded a Semiconductor Research Corporation fellowship. He spent the summers of 1995 and 1996 interning in Intel Corporationâ€™s TCAD department in Santa Clara, California, where he investigated intrinsic capacitance degradation and circuit aging simulation. 142 I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy Chih-Tang Sah, Chairman Robert C. Pittman Eminent Scholar and Graduate Research Professor of Electrical and Computer Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Professor of Electrical and Computer Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Toshikazu Nishida Associate Professor of Electrical and Computer Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. SKeng Li Professor of Electrical and Computer Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. r VW|A.J n Clin, Randy Yhd lan-bjiieh Cho W Professor of Computer and Information Science and Engineering f This dissertation was submitted to the Graduate Faculty of the College of Engineering and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1997 Karen A. Holbrook Dean, Graduate School LO 1780 199J UNIVERSITY OF FLORIDA 3 1262 08556 6049 |