Presentation is loading. Please wait.

Presentation is loading. Please wait.

Unit IV: GATE LEVEL DESIGN

Similar presentations


Presentation on theme: "Unit IV: GATE LEVEL DESIGN"— Presentation transcript:

1 Unit IV: GATE LEVEL DESIGN
Parasitic Elements Unit IV: GATE LEVEL DESIGN VLSI DESIGN

2 Out Line Sheet Resistance Capacitance Delay Models (Elmore Delay )
14/02/2009 VLSI Design

3 Introduction Wiring-Up of chip devices takes place through various conductors produced during processing Today, interconnects constitute the main source of delay in MOS circuits We will examine: Sheet Resistance – Resistance / Unit Area Area Capacitance Delay Units 14/02/2009 VLSI Design

4 The Wire schematics physical 14/02/2009 VLSI Design

5 Interconnect Impact on Chip
14/02/2009 VLSI Design

6 Wire Models Capacitance-only All-inclusive model 14/02/2009
VLSI Design

7 Impact of Interconnect Parasitics
Reduce reliability Affect performance and power consumption Classes of parasitics Resistive Capacitive Inductive 14/02/2009 VLSI Design

8 Transistor source/drain parasitics
Source/drain have significant capacitance, resistance. Measured same way as for wires. Source/drain R, C may be included in Spice model rather than as separate parasitics. 14/02/2009 VLSI Design

9 Resistance 14/02/2009 VLSI Design

10 Sheet Resistance A Resistance of a square slab of material RAB = ρL/A
=> R = ρL/(t*W) Let L = W (square slab) => RAB = ρ/t = Rs ohm / square t w L B RAB = ZRsh Z = L/W 14/02/2009 VLSI Design

11 Wire Resistance r = resistivity (W*m) R = sheet resistance (W/)
 is a dimensionless unit(!) Count number of squares R = R * (# of squares)  R is completely independent of the area of the square. 1μm per side square slab of material has exactly the same resistance as a 1 cm per side square slab of the same material if the thickness is the same. 14/02/2009 VLSI Design

12 Wire Resistance 14/02/2009 VLSI Design

13 Resistor Resistors are made by doped silicon or polysilicon on an IC chip Resistance is determined by Length, line Width, Height, and Dopant concentration 14/02/2009 VLSI Design

14 Dealing with Resistance
Selective Technology Scaling Use Better Interconnect Materials reduce average wire-length e.g. copper, silicides More Interconnect Layers A silicide is a compound that has silicon with more electropositive elements. Examples: nickel silicide, NiSi :sodium silicide, Na2Si ; magnesium silicide, Mg2Si Platinum silicide, PtSi ;Titanium silicide, TiSi2 ;Tungsten silicide, WSi2 14/02/2009 VLSI Design

15 Skin effect At low frequencies, most of copper conductor’s cross section carries current. As frequency increases, current moves to skin of conductor. Back EMF induces counter-current in body of conductor. Skin effect most important at gigahertz frequencies. 14/02/2009 VLSI Design

16 Skin effect, cont’d Isolated conductor: Conductor and ground:
Low frequency Low frequency High frequency High frequency 14/02/2009 VLSI Design

17 Skin depth Skin depth is depth at which conductor’s current is reduced to 1/3 = 37% of surface value: d = 1/sqrt(p f m s) f = signal frequency m = magnetic permeability s = wire conducitvity 14/02/2009 VLSI Design

18 Effect on resistance Low frequency resistance of wire:
Rdc = 1/ s wt High frequency resistance with skin effect: Rhf = 1/2 s d (w + t) Resistance per unit length: Rac = sqrt(Rdc 2 + k Rhf 2) Typically k = 1.2. 14/02/2009 VLSI Design

19 Contacts Resistance Contacts and vias also have 2-20 W
Use many contacts for lower R Many small contacts for current crowding around periphery 14/02/2009 VLSI Design

20 Layer Rs (Ohm / Sq) Aluminium 0.03 N Diffusion 10 – 50 Silicide 2 – 4
Typical sheet resistance values for materials are very well characterised Layer Rs (Ohm / Sq) Aluminium 0.03 N Diffusion 10 – 50 Silicide 2 – 4 Polysilicon N-transistor Channel 104 P-transistor Channel 2.5 x 104 14/02/2009 VLSI Design Typical Sheet Resistances for 5µm Technology

21 Sheet Resistance 14/02/2009 VLSI Design

22 Resistance Depends on resistivity of material r (Rho)
Sheet resistance Rs = r /t see Table 2-4, p. 80 Resistance R = Rs * L / W Corner approximation - count a corner as half a square Example: R = Rs(poly) * *(1/2) + 3*(1/2) squares R = 4Ω/sq * 15.5 squares = 62Ω Corner (1/2 Square) 1/2 Square 1/2 Square Corner (1/2 Square) Corner (1/2 Square) 14/02/2009 VLSI Design

23 Compute Resistance Partition long wire into rectangles
Count the number of squares ((l1/w1)+(l2/w2)+(l3/w3))Rs l1 w2 w1 l2 w3 l3 14/02/2009 VLSI Design

24 Modern Interconnect 14/02/2009 VLSI Design

25 14/02/2009 VLSI Design

26 Example: Intel 0.25 micron Process
5 metal layers Ti/Al - Cu/Ti/TiN Polysilicon dielectric 14/02/2009 VLSI Design

27 Layer Rs (Ohm / Sq) Aluminium 0.03 N Diffusion 10 – 50 Silicide 2 – 4
Typical sheet resistance values for materials are very well characterised Layer Rs (Ohm / Sq) Aluminium 0.03 N Diffusion 10 – 50 Silicide 2 – 4 Polysilicon N-transistor Channel 104 P-transistor Channel 2.5 x 104 14/02/2009 VLSI Design Typical Sheet Resistances for 5µm Technology

28 N-type Minimum Feature Device
Polysilicon L N - diffusion W Here ,Length to Width ratio ,denoted Z is (2/2)=1 R = 1square x Rs Ohm / Square = Rs = 104Ώ 14/02/2009 VLSI Design

29 Polysilicon L = 8λ W = 2λ N - diffusion R = Z Rs R = (L/W) * Rs
Here ,Length to Width ratio ,denoted Z is (8/2)=4 R = Z Rs R = (L/W) * Rs R = 4×104 Ώ 14/02/2009 VLSI Design

30 Exercise 1.Calculate ON resistance of the circuit shown from VDD to GND.If n-Channel sheet resistance Rsn=104 Ω per Square ,and P-channel sheet resistance Rsp=3.5×104 Ω per Square. Take CMOS Inverter as Circuit diagram.Given Zpu =1 and Zpd=1. . 14/02/2009 VLSI Design

31 Capacitance 14/02/2009 VLSI Design

32 Capacitors Charge storage device Memory Devices, esp. DRAM
Challenge: reduce capacitor size while keeping the capacitance High-k dielectric materials 14/02/2009 VLSI Design

33 Capacitors Parallel plate Stacked Deep Trench Dielectric Layer
Poly 2 Poly Si Si Poly Si Oxide Si Poly 1 Heavily Doped Si Parallel plate Stacked Deep Trench 14/02/2009 VLSI Design

34 Capacitance of Wire Interconnect
14/02/2009 VLSI Design

35 Interconnects - Capacitance
Parallel Plate Capacitance L W H Dielectric Substrate tdi Keep in mind: C is proportional to the overlap between conductors C is inversely proportional to their separation 14/02/2009 VLSI Design

36 Permittivity 14/02/2009 VLSI Design

37 Area Capacitance of Layers
Conducting layers are separated from each other by insulators (typically SiO2) This may constitute a parallel plate capacitor, C = є0єox A / D (farads) D = thickness of oxide, A = area, єox = 4 F/µm2 (SiO2) Area capacitance given in pF/µm2 14/02/2009 VLSI Design

38 Capacitance Value in pF×10-4 /µm2 (5μm Technology) Gate to Channel 4
Diffusion(Active) 1 Poly silicon to Substrate 0.4 Metal1 to substrate 0.3 Metal2 to substrate 0.2 Metal2 to metal 1 Metal2 to polysilicon 14/02/2009 VLSI Design

39 Capacitance Standard unit for a technology node is the gate - channel capacitance of the minimum sized transistor (2λ x 2λ), having W=L=feature Size ,given as Cg This is a ‘technology specific’ value 14/02/2009 VLSI Design

40 Cg may be evaluated for any MOS process
Cg may be evaluated for any MOS process. For example ,for 5μm MOS Circuits Area/Standard square= 5μm × 5μm =25 μm2(=area of minimum size transistor) Capacitance value (From table)=4 ×10-4 pF/µm2 Thus Standard value Cg = 25 μm2 × 4 ×10-4 pF/µm2= 0.01 pF 1.Calculate the gate capacitance value of 5μm technology minimum sized transistor with gate to channel capacitance value is 8 ×10-4 pF/µm2 2.Calculate the gate capacitance value of 2μm technology minimum sized transistor with gate to channel capacitance value is 8 ×10-4 pF/µm2 14/02/2009 VLSI Design

41 Interconnects - Capacitance
Fringing Capacitance Dielectric Substrate H Dielectric Substrate Cpp Cfr 14/02/2009 VLSI Design

42 Interconnects - Capacitance
Total Capacitance Dielectric Substrate + H w Keep in mind: C is proportional to the overlap between conductors C is inversely proportional to their separation 14/02/2009 VLSI Design

43 Fringing Capacitance 14/02/2009 VLSI Design

44 Interwire Capacitance
14/02/2009 VLSI Design

45 Wiring Capacitances (0.25 mm CMOS)
14/02/2009 VLSI Design

46 Note that τ is very similar to channel transit time τsd
Delay Unit For a feature size square gate, τ = Rs x Cg i.e for 5µm technology, τ = 104 ohm/sq x 0.01pF = 0.1ns Because of effects of parasitics which we have not considered in our model, delay is typically of the order of ns Note that τ is very similar to channel transit time τsd One standard (feature size square) gate area capacitance being charged through one feature size square of n-channel resistance (that is,through Rs for an nMOS pass transistor channel ) as in the figure Time constant τ=(1 Rs (n channel) × 1Cg ) seconds 14/02/2009 VLSI Design

47 Interwire Capacitance
Taken from “Digital Integrated Circuits”, 2nd Edition, Jan M. Rabaey, Anantha Chandrakasan, and Borivoje Nikolic Copyright 2002 J. Rabaey et al. 14/02/2009 VLSI Design

48 14/02/2009 VLSI Design

49 14/02/2009 VLSI Design

50 Interconnects - Inductance
It can be evaluated with the aid of its definition: v=Ldi/dt It is possible to calculate the inductance from its geometry and its environment A simpler approach relies on the fact that the capacitance c and inductance l (per unit length) are related: cl=em (e and m are the permittivity and permeability of the surrounding dielectric) Caution: conductor must be surrounded by a uniform dielectric 14/02/2009 VLSI Design

51 Simplifications Inductive effects can be ignored if the resistance is substantial (e.g. a long Al wire with small cross-section) or if the rise and fall times of the applied signals are slow When the wires are short, the cross-section is large or the material has low-resistivity, a capacitance-only model can be used When the separation between neighboring wires is large or when wires only run together for a short distance, inter-wire capacitance can be ignored, and all the parasitic capacitance can be modeled as capacitance to ground 14/02/2009 VLSI Design

52 Electrical Wire Model Ideal Wire:
Simplistic Useful for early phase of the design OK for small components, e.g. gates To study the effects of parasitics we need to model them 14/02/2009 VLSI Design

53 Electrical Wire Model Lumped vs Distributed
14/02/2009 VLSI Design

54 Electrical Wire Model – Lumped C
If resistive component is small and switching frequencies are in the low to medium range, it makes sense to consider only capacitive component Wire still represents an equipotential region Only impact on performance is the loading effect Popular model C 14/02/2009 VLSI Design

55 Electrical Wire Model – Lumped RC
If wire resistance is significant, a resistive-capacitive model is needed Lumped RC model is pessimistic and inaccurate for long interconnect wire Distributed rc-model is complex and no closed form solution exists Elmore delay formula: lumped RC comes to help R C 14/02/2009 VLSI Design

56 Electrical Wire Model – Lumped RC Elmore Delay
s R2 C2 R4 C4 C3 R3 Ci Ri 1 2 3 4 i Consider an RC-tree: The network has a single input node All capacitors between node and ground The network does not contain any resistive loop 14/02/2009 VLSI Design

57 Electrical Wire Model – Lumped RC Elmore Delay
s R2 C2 R4 C4 C3 R3 Ci Ri 1 2 3 4 i RC-tree property: Unique resistive path between the source node s and any other node i of the network  path resistance Rii Example: R44=R1+R3+R4 14/02/2009 VLSI Design

58 Electrical Wire Model – Lumped RC Elmore Delay
s R2 C2 R4 C4 C3 R3 Ci Ri 1 2 3 4 i RC-tree property: Extended to shared path resistance Rik: Example: Ri4=R1+R3 Ri2=R1 14/02/2009 VLSI Design

59 Electrical Wire Model – Lumped RC Elmore Delay
Assuming: Each node is initially discharged to ground A step input is applied at time t=0 at node s The Elmore delay at node i is: It is an approximation: it is equivalent to first-order time constant of the network Proven acceptable Powerful mechanism for a quick estimate 14/02/2009 VLSI Design

60 Electrical Wire Model – Lumped RC Elmore Delay
Special case: RC-chain (or ladder) Shared-path resistance path resistance R1 C1 R2 C2 RN CN Vin VN 14/02/2009 VLSI Design

61 Interconnects – Why do we care?
Technology scaling: miniaturization of devices (scale L, W, TOX, VTH) Device Scaling  Faster, smaller devices L[mm]=0.35, 0.25, 0.18, 0.12, etc  S0.7 Interconnect Scaling  Larger delays! Local interconnects: Almost constant Global interconnects:RC delay goes as 1/S or 1/S3 Also, parasitics give rise to a whole set of signal integrity issues  Design paradigm shift from device-centric to interconnect-centric 14/02/2009 VLSI Design

62 Interconnects – Why do we care?
14/02/2009 ITRS 2001 VLSI Design

63 New Interconnect Materials
Aluminium  Copper Lower resistivity Higher immunity to Electromigration SiO2  Low-k dielectrics 14/02/2009 VLSI Design

64 What causes delay? In MOS circuits capacitive loading is the main cause Due to: Device capacitance Interconnect capacitance 14/02/2009 VLSI Design

65 MOSFET capacitances MOS capacitances have three origins:
The basic MOS structure The channel charge The pn-junctions depletion regions 14/02/2009 VLSI Design

66 Channel capacitance The channel capacitance is nonlinear
Its value depends on the operation region Its formed of three components: Cgb - gate-to-bulk capacitance Cgs - gate-to-source capacitance Cgd - gate-to-drain capacitance Operation region C gb gs gd Cutoff ox W L Linear (1/2) C Saturation (2/3) C 14/02/2009 VLSI Design

67 MOS structure capacitances
Source/drain diffusion extend below the gate oxide by: xd - the lateral diffusion This gives origin to the source/drain overlap capacitances: Gate-bulk overlap capacitance: 14/02/2009 VLSI Design

68 Channel capacitance Cg = Weff  Leff  Cox Gate-to-bulk Gate-to-source
Cg + Cgbo 2/3 Cg + Cgso Gate-to-bulk 1/2 Cg + Cgbo Cgdo , Cgso Cgbo Gate-to-source Gate-to-drain Off Saturated Linear 14/02/2009 VLSI Design

69 Junction capacitances
Csb and Cdb and diffusion capacitances composed of: Bottom-plate capacitance: Side-wall capacitance: 14/02/2009 VLSI Design

70 Junction capacitances
0.24 mm process NMOS L(drawn) = 0.24 mm L(effective) = 0.18 mm W(drawn) = 2 mm Ls = 0.8 mm Cj (s, d) = 1.05 fF/mm2 Cjsw = 0.09 fF/mm Cbottom = 1.68 fF Csw = 0.32 fF Cg = 2.02 fF 14/02/2009 VLSI Design

71 Source/drain resistance
Scaled down devices  higher source/drain resistance: In sub- processes silicidation is used to reduce the source, drain and gate parasitic resistance 0.24 mm process R (P+) = 4 W/sq R (N-) = 4 W/sq 14/02/2009 VLSI Design

72 New Interconnect Materials
Aluminium  Copper Lower resistivity Higher immunity to Electromigration SiO2  Low-k dielectrics 14/02/2009 VLSI Design

73 --Take up one Idea. Make that Idea your life. Think of it. Dream of it
--Take up one Idea. Make that Idea your life. Think of it .Dream of it. Live on that Idea .Let the brain, muscles, nerves and every part of your body, be full of that idea and just leave every other idea alone. This is the way to success……. 14/02/2009 VLSI Design


Download ppt "Unit IV: GATE LEVEL DESIGN"

Similar presentations


Ads by Google