090513 JohnCressler Slides

1John D. Cressler, 5/09

SiGe Technology:New Research Directions and

Emerging Application Opportunities

John D. CresslerKen Byers Professor

School of Electrical and Computer EngineeringGeorgia Tech, Atlanta, GA 30332-0250 USA

[email protected]

IEEE Southeastern Michigan Chapter IV, Ann Arbor, MI, May 2009IEEE Electron Devices Society Distinguished Lecture

This work was supported by NASA, DTRA, IBM, DARPA, JPL, TI, and NSC


A few monolayersof SiGe Goes Here Very Carefully!

A Level-Set on Miracles


Outline

Some Reminders on SiGe Scaling Trends and Performance Limits Using SiGe for Radar Systems Using SiGe for High-speed Analog Using SiGe in Extreme Environments Summary


Practice Bandgap Engineering but do it in Silicon!

SiGe Strained Layer Epi

EV

The Bright Idea!


Seamless Integration of SiGe into Si

When You Do It Right

No Evidenceof Deposition!100 nm


The SiGe HBT

The Idea: Put Graded Ge Layer into the Base of a Si BJT

Primary Consequences: smaller base bandgap increases electron injection ( ) field from graded base bandgap decreases base transit time (fT ) base bandgap grading produces higher Early voltage (VA ) decouples base profile from performance metrics

SiGe = III-V Speed + Si Manufacturing: Win-Win!


The SiGe HBT

E B C

SiGe50 nmSiGe = III-V Speed + Si ManufacturingWin-Win!

Conventional Shallow and Deep Trench Isolation + CMOS BEOL Unconditionally Stable, SiGe Epitaxial Base Profile 100% Si Manufacturing Compatibility SiGe HBT + Si CMOS on wafer


SiGe Success Story

1G

2G

3G

4G

Rapid Generational Evolution (full SiGe BiCMOS) Significant In-roads in Communications / Analog ICs

Important Point: 200 GHz @ 130 nm! (2G better than CMOS)

(130 nm)

(180 nm)

(500 nm)


SiGe Apps

Defense

Navigation

Automotive

Communications 1999 2000 2001 2002 2003 2004 2005 20060.00.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

W

o

r

l

d

w

i

d

e

C

h

i

p

S

a

l

e

s

In Billions of US $ SiGe, CAGR = 61.3% GaAs, CAGR = 8.7%

Years

Defense

Navigation

Automotive

Communications

SiGe Analog/MS ICs Are a Major Driver!


SiGe for Radar Systems- single chip T/R for phased arrays, space-based radar (2-10 GHz & up) - automotive radar (24, 77 GHz)

SiGe for Millimeter-wave Communications - Gb/s short range wireless links (60, 94 GHz)- cognitive radio / frequency-agile WLAN / 100 Gb Ethernet

SiGe for THz Sensing, Imaging, and Communications - imaging / radar systems, diagnostics, comm (94 GHz, 100-300 GHz)

SiGe for Analog Applications- the emerging role of C-SiGe (npn + pnp) + data conversion (ADC limits)

SiGe for Extreme Environment Electronics- extreme temperatures (4K to 300C) + radiation (e.g., space systems)

SiGe for Electronic Warfare - extreme wideband transceivers (20 MHz 20 GHz)- dynamic range enhanced receivers

Some New Opportunities


SiGe HBT vs CMOS

If Cost Were No Object, III-V Devices Would Always DominateSi-based Solutions in Terms of Raw Speed and Breakdown Voltage,BUT Integration and Cost Increasingly Decide Things

Clearly Both SiGe HBTs and (Strained-Si) CMOS Are Capable of Very Frequency Operation, BUT That Is Not the Whole Story

Some Key Advantages SiGe HBTs Have Over CMOS:- for equal performance, SiGe requires less aggressive lith (2 gen)- lith delta = cost delta (130 nm SiGe costs less than 65 nm CMOS) - SiGe HBTs give higher BVs at fixed performance (BVCEO vs BVCBO)- SiGe HBTs are better for generating CW power at fixed frequency- gm/m2 is MUCH larger for SiGe HBTs (need BIG nFETs for high fmax)- 1/f noise is 100x lower in SiGe HBTs; much better matching- SiGe HBTs have better output conductance at fixed scaling node- modeling of SiGe HBTs is easier for high-frequency design - SiGe HBTs have built-in total dose radiation hardness for space- SiGe HBTs are natural for mixed-signal & BiCMOS = HBTs + CMOS

Some Thoughts on the Matter

My Opinion: SiGe BiCMOS at a More Conservative Node Is Often Better Suited for MANY Applications Than Aggressively-Scaled CMOS


Strain Engineering in Si

Strained Si CMOS SiGe HBTs

SiGe MODFETs

All Are:Strain-Enhanced

Si-based Transistors

Close Cousins!


Some Morals to the Story:

1) CMOS is Cheaper Than SiGe at Fixed Scaling Node2) SiGe is Cheaper Than CMOS at Fixed Performance3) SiGe HBTs are More Natural for RF/Analog Design4) BiCMOS Often is Optimal for Integrated Solutions

Q: Where is SiGe the Best Fit?


Outline



TeraHertz Applications

Pic. avail. at http://www.advancedphotonix.com/

Imaging for Quality Control Detection of Non-Metallic Objects Medical Diagnostics Radio Astronomy Ultra-Wideband Communications (fast and secure) THz Active Circuits + ICs ??

300 GHz

Emerging THz Electronics Applications (100-300 GHz)


Cooling a SiGe HBT Emulates Vertical + Lateral Scaling- vsat , RB , and CCB as T better current drive, parasitics- and Ge grading (Eg,Ge(grade)/kT) more effective as T

Cryo-T as a Scaling Knob

Classical Scaling (300K):- vertical: NC , NB , WB

Pros: E, C, CSCL, and Cons: CCB and RB , BVCEO

- lateral:Pros: RB , CCB and CEBCons: RE and RC

- innovation needed - moral: scaling isnt free!

300 K


fT + fmax > 1 THz in SiGe Is Clearly Possible (at very modest lith) Both fT and fmax above 500 GHz at Cryo-T (T = scaling knob) Goal: Useful BV @ 500 GHz (BVCEO > 1.5 V + BVCBO > 5.5 V)

SiGe Performance Limits

8HP

200-500 GHz @ 130 nm Node!


A 300K Scaling Roadmap for SiGe HBTs

Can We Do This at 300K?

e.g., EU DOT5Project


Outline



DoD Radar T/R Modules

X-Band Phased Array Radar - realized with III-V MMICs, assembled into T/R MCMs ($1k each)- employs high power density approach (1 MW to run 25k HPD T/R modules)- costly to deploy and operate

25,000 high power multi-chip modules

1 MW diesel generator

liquid/air heat

exchanger

1990s

Courtesy of M. Mitchell

T/RTHAADRadar


A New DoD Radar Vision

A New Radar Paradigm- panel based + uses low power density approach (reduced prime power)- T/R = low-cost + efficient + highly integrated (with digital control)

The Future

SiGe T/R

Multi-ChipPanel

Multi-Panel Array

Courtesy of M. MitchellBegs for a SiGe BiCMOS Solution!


RF Manifold

In/Out

PA

From Radiator-A

T/R Module Topology

From Radiator-B

3-bitPhase Shifter

Pre-AmpPre-Gain

SP3TSwitch

LNA-B

LNA-ALNA

SwitchSwitch

1-bitPhase Shifter

4-bit Phase Shifterwith integrated LNA

RX Mode Challenges- need very aggressive LNA- must balance gain, linearity, and NF- need low power dissipation- need a good phase shifter

TX Mode Challenges- must deliver necessary output power- need high power efficiency- need high gain + stability

BiCMOS Design OptimumSiGe is an Ideal FitSingle chip solution

25John D. Cressler, 5/09VCC = 2.5 V, IC = 6 mA = 15 mW

X-band LNA

1.5 dB

Radar Spec Demands NF < 2.0 dB Across Band


Compare MOS Phase Shifter to HBT Phase Shifter Maintain Same Design Methodology

- target 4-bit phase shifter- use similar HP and LP filter sections- utilize meander layout for reasonable aspect ratio

MOS vs HBT Shifters

HBT MOS


SiGe PA Design Issues Understand Voltage Constraints / Instabilities Leverage CB-forced IE Mode for PA Voltage Swing Power Cell Layout Optimization For Thermal Coupling

125 mW / 40 dB gain 25% PAE X-band PAHP + HB Cascode

850 mW / 11 dB gain 18% PAE X-band PAHP + HB Cascode


SiGe T/R Results

Pout > 22 dBm P1dB > 20 dBm TX Gain > 17 dB PAE = 28%

NF < 3.5 dB RX Gain > 15 dB IIP3 = -10 dBm

RX TX

RX TX

13 mm2


SiGe Automotive Radar

24 GHz Short Range Radar + 77 GHz Long Range Radar - nice fit for SiGe technology (2G SiGe for SRR + 3G SiGe for LRR) - in the EU 24G SRR works until 2013, and then moves to 79G (LRR = 77G)

Courtesy of SARA

SRR LRR

Integration is Key


A Future SiGe Radar Path

Multilayer OrganicSubstrate (LCP)

Embedded SiGe T/RCircuits

Integrated RF MEMS Phase Shifters

IntegratedSiGe T/Ron Flexible Multilayer OrganicSubstrate

3D X-band to mm-wave SiGe Integrated Radars- flexible + low-cost + multilayer + wafer scale - RF MEMS phase shifters + antennae- for space-based radar & comm systems

85 90 95 1002

4

6

8

10

12

14

Measured Gain & Noise Figure

Gain

NF

Frequency (GHz)

W-band SiGe LNA

P1dB = -12.5 dBmIIP3 = -8 dBm

Collaboration with John Papapolymerou and Kevin Kornegay

G

a

i

n

a

n

d

N

o

i

s

e

(

d

B

)


Outline



C-SiGe Leverage

Q: Complementary SiGe HBT Technology (C-SiGe)?

Some Possibilities:- address issues analog C-Si BJTs have inherent trouble with:

- achieving high current gain for both npn + pnp- achieving high speed for both npn + pnp- achieving high output resistance for both npn + pnp- decoupling gain / speed / VA from base profile design

- enables a future performance scaling path for analog IC apps

MANY Intriguing Possibilities

Fact: C-Si BiCMOS (npn + pnp) is the Gold Standard for High-performance Analog ICs (high value add)


TIs BiCOM3X C-SiGe

Balanced 25 GHz fT / 60 GHz fmax, 6V BVCEO C-SiGe Platform Thick Film SOI for Isolation Being Used Leading-Edge Analog Parts (500 MS/s 14 bit ADC)

npn SiGe HBT

pnp SiGe HBT npn SiGe HBT


IHP C-SiGe Technology

IHPs C-SiGe HBT Technology- low RC and CCS collector construction (no STI between E and C)- reduced phosphorus diffusion in the C-doped base- npn SiGe HBT: peak fT / BVCEO of 170 GHz / 1.9 V- pnp SiGe HBT: peak fT / BVCEO of 90 GHz / 3.1 V (60 / 4.5)-

npn SiGe HBT

pnp SiGe HBT

[2] D Knoll et al, BCTM, pp. 30-33, 2007[1] B. Heinemann et al, IEDM, pp. 117-120, 2003


ADC Migration

Major Stressor on ADC Performance


40 GS/s Track/Hold Amp


Outline



Extreme Environments

Aerospace (aircraft, satellites ...) Space Exploration (Moon, Mars ...) Automotive (on-engine ) Drilling (oil, geothermal ...)

Exploration

DrillingCars

Aerospace

Extreme Environment Electronics:low-T, high-T, wide-T, radiation, shock, chemical


Moon Mars Outer Planets

Space Exploration

All Represent Extreme Environments!(Very Wide Temperature Swings + Radiation)


Temperature Ranges: +120C to -180C (300C swings!) 28 day cycles

Radiation: 100 krad over 10 years single event upset (SEU) solar events

Many Different Circuit Needs: digital building blocks analog building blocks data conversion (ADC/DAC) RF communications power conditioning actuation and control switches sensors / sensor interfaces

Requires Centralized Warm Box

Rovers / Robotics

The Moon:A Classic Extreme Environment!

Highly Mixed-Signal Flavor!


The Idea: Put Graded Ge Layer into the Base of a Si BJT

Primary Consequences: smaller base bandgap increases electron injection ( ) field from graded base bandgap decreases base transit time (fT ) base bandgap grading produces higher Early voltage (VA )

All kT Factors Are Arranged to Help at Cryo-T!

SiGe HBTs for Cryo-T


SiGe HBTs at Cryo-T

27C

-230C

dc ac

SiGe Exhibits Very High Speed at Very Low Power!

First Generation SiGe HBT


X-band LNA Operation at 15 K (Not Yet Optimized!)

Teff < 20 K (noise T) NF < 0.3 dB Gain > 20 dB dc power < 2 mW

Cryogenic SiGe LNAs

Collaboration with S. Weinreb, Cal Tech

NF = 0.3 dB!

This SiGe LNA is Also Rad-Hard!


SiGe at High-T? (200-300C)

Degradation, But Plenty of Performance Left! Device Reliability Looks Fine Just in: Robust Operation @ 300C for Selected Circuits

Gain Frequency


The Holy Grail of the Space Community- IC technology space-qualified without additional hardening (major cost adder)- high integration levels to support SoC / SiP (low cost)

SiGe For Space Systems

proton + electron belts

Major Question: Can SiGe Play a Major Role in Space?

Total Ionizing Dose (TID) ionizing radiation- TID is measured in rads (1 rad = 100 ergs per gram of energy absorbed)- 100-1000 krad(Si) over 10 years for typical orbit (300 rad(Si) is lethal to humans!)

Single Event Upset (SEU) high energy heavy ions- measure data upset cross-section () vs. Linear Energy Transfer (LET)- = # errors / particle fluence (ions/cm2): LET = charge deposition (pC/m)- Goals: low cross-section + high LET threshold


Total-Dose Response Multi-Mrad Total Dose Hardness (with no intentional hardening!)

- ionization + displacement damage very minimal over T; no ELDRS! Radiation Hardness Due to Epitaxial Base Structure (not Ge)

- thin emitter-base spacer + heavily doped extrinsic base + very thin base

63 MeV protons @ 5x1013 p/cm2 = 6.7 Mrad TID!

200 GHz SiGe HBT

3rd

2nd

1st

4th


Observed SEU Sensitivity in SiGe HBT Shift Registers- low LET threshold + high saturated cross-section (bad news!)

P. Marshall et al., IEEE TNS, 47, p. 2669, 2000

Goal

Single Event Effects

heavy ion


OUT

DATA

CLOCK

TCAD Ion Strike

Standard Master Slave Latch UPSETS

SEU: TCAD to Circuits

New RHBD SiGe Latch

SEU Soft


Reduce Tx-Tx Feedback Coupling Internal to the Latch Circuit Architecture Changes + Transistor Layout Changes

SiGe RHBD Success!

Future - Eliminate TMR & Be Faster!Path - Build a Rad-Hard System!

(no errors!)


Remote Electronics Unit

5 x 3 x 6.75 = 101 in3

11 kg 17 Watts -55oC to +125oC

1.5 x 1.5 x 0.5 = 1.1 in3 (100x) < 1 kg (10x) < 2 Watts (10x) -180oC to +125oC, rad tolerant

Conceptual integrated REU system-on-chip SiGe BiCMOS die

The X-33 Remote Health Unit, circa 1998

The ETDP Remote Electronics Unit, circa 2009

Specifications Goals

Analog front end die

Digital control die

Supports Many Sensor Types:Temperature, Strain, Pressure, Acceleration, Vibration, Heat Flux, Position, etc.

REU in connector housing!

Use This REU as a Remote Vehicle Health Monitoring Node


Temperature Controlled Environment

System Processor

(RAD750 or equivalent)

Solid State Data Recorder

Extreme Ambient Environment

Extreme Ambient Environment

Communications

Accelerometer (High sp)

Accelerometer (ch amp)

Thermocouple (Low sp)




Pressure Transducer

(Low sp)

Pressure Transducer

(Low sp)

REU Digital Control ASIC

REU Sensor

Interface ASIC

Ctrl

Data

Sensor Inputs

RS-485

REU in Connector Housing

Boot PROM


REU Sensor

Interface ASIC

Ctrl

Data

Sensor Inputs

RS-485


Boot PROM


REU Sensor

Interface ASIC

Ctrl

Data

Sensor Inputs

RS-485


Boot PROM


REU Sensor

Interface ASIC

Ctrl

Data

Sensor Inputs

RS-485


Boot PROM

REU Sensor

Interface ASIC


Data

CtrlSensor Inputs

RS-485


Boot PROM

REU Sensor

Interface ASIC


Data

CtrlSensor Inputs

RS-485


Boot PROM

REU Sensor

Interface ASIC


Data

CtrlSensor Inputs

RS-485


Boot PROM

REU Sensor

Interface ASIC


Data

CtrlSensor Inputs

RS-485


Boot PROM

Major Advantages: Eliminates Warm Box (size, weight, and power; allows de-centralized architecture) Significant Wiring Reduction (weight, reliability, simplifies testing & diagnostics) Commonality (easily adapted from one system to the next)

SiGe REU Architecture


MISSE-6 ISS Mission

Recent NASA photograph of MISSE-6 after deployment, taken by the Space Shuttle Crew

SiGe Circuits !


Summary

The Global Landscape: The Emerging Communications Infrastructure

- frequency bands pushing upward over time (stresses device design)- integration of RF + digital + analog + passives increasingly important- SiGe HBT BiCMOS is well-positioned to address this market

SiGe Technology is Here to Stay!

SiGe HBT BiCMOS Technology: The SiGe HBT is the First Practical Bandgap Engineered Device in Si Compared to Si BJTs, SiGe HBTs Offer Better:

- + VA + VA + fT + fmax + 1/f + NFmin + Compared to CMOS, SiGe HBTs Offer Better:

- fT/fmax/NF at fixed scaling node + matching + gm/area + 1/f noise, + Still Room for Lots of Performance Improvement (fT / fmax = 500 GHz) Still Lots to Learn About the Physics of These Interesting Devices MANY Interesting Application Possibilities and New Opportunities!


My Gang at Georgia Tech

Documents

090513 JohnCressler Slides