Page 1
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
From MEMS to NEMS: Smaller Is Still Better
Clark T.-C. Nguyen
Dept. of Electrical Engineering & Computer ScienceUniversity of Michigan
Ann Arbor, Michigan 48105-2122
(Last Month: Program Manager, DARPA/MTO)
MARC’06 MeetingJan. 25-26, 2006
Page 2
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Outline
• Introduction:MEMS technologyintegration with transistors: an early driver for MEMS
• Benefits of Scalingsize reductionspeed, energy conservation, complexity, economy
• DARPA/MTO Program ExamplesNano Mechanical Array Signal Processors (NMASP)Chip-Scale Atomic Clock (CSAC)Micro Gas Analyzers (MGA)
• Conclusions (What’s Next?)
Page 3
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Angle set by mechanical meansto control the path of light
MEMS:MEMS: MMicro EElectro MMechanical SSystems
• A device constructed using micromachining (MEMS) tech.• A micro-scale or smaller device/system that operates mainly
via a mechanical or electromechanical means• At least some of the signals flowing through a MEMS device
are best described in terms of mechanical variables, e.g., displacement, velocity, acceleration, temperature, flow
MEMS
Input:voltage, current
acceleration, velocitylight, heat …
Output:voltage, current
acceleration, velocitylight, heat, …
Control:voltage, current
acceleration velocity
light, heat, …
[Wu, UCLA]Transducer to
Convert Controlto a Mechanical Variable (e.g., displacement,
velocity, stress, heat, …)
Transducer to Convert Controlto a Mechanical Variable (e.g., displacement,
velocity, stress, heat, …)
Page 4
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Other Common Attributes of MEMS
• Feature sizes measured in microns or less
• Merges computation with sensing and actuation to change the way we perceive and control the physical world
• Planar lithographic technology often used for fabricationcan use fab equipment identical to those needed for IC’showever, some fabrication steps transcend those of conventional IC processing
MEMSTechnology
Gimballed, SpinningMacro-Gyroscope
MicromechanicalVibrating Ring Gyroscope
Signal Conditioning Circuits
80 mm
1 mm
(for 80X sizeReduction)
[Najafi, Michigan]
Page 5
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Silicon Substrate
Glass Substrate
Bulk Micromachining and Bonding
• Use the wafer itself as the structural material
• Adv: very large aspect ratios, thick structures
• Example: deep etching and wafer bonding
Silicon SubstrateSilicon Substrate
Glass Substrate
Silicon Substrate
Metal InterconnectAnchor
MovableStructure Electrode
MicromechanicalVibrating Ring Gyroscope
1 mm
Microrotor(for a microengine)
[Najafi, Michigan] [Pisano, UC Berkeley][Pisano, UC Berkeley]
Page 6
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
• Fabrication steps compatible with planar IC processing
Surface Micromachining
Page 7
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
• Completely monolithic, low phase noise, high-Q oscillator (effectively, an integrated crystal oscillator)
• To allow the use of >600oC processing temperatures, tungsten (instead of aluminum) is used for metallization
OscilloscopeOutput
Waveform
Single-Chip Ckt/MEMS Integration
[Nguyen, Howe 1993][Nguyen, Howe 1993]
Page 8
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Technology Trend and Roadmap for MEMSin
crea
sing
abi
lity
to c
ompu
te
increasing ability to sense and act
Num
ber o
f Tra
nsis
tors
Number of Mechanical Components
100
101
102
103
104
105
106
107
108
109
100 101 102 103 104 105 106 107 108 109
AdaptiveOptics
Integrated FluidicSystems
DistributedStructural
ControlTerabit/cm2
Data Storage
Optical Switches& Aligners
InertialNavigationOn a Chip
Displays
Weapons,Safing, Arming,
and Fusing
Majority ofEarly MEMS
Devices(mostly sensors)
ADXL-50
Digital MicromirrorDevice (DMD)
ADXL-278
Future MEMSIntegration Levels
Enabled Applications
OMM 32x32
ADXL-78
CPU’s
Pentium 4
ADXRS
i-STAT 1Caliper
Phased-ArrayAntenna
Page 9
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Benefits of Size Reduction: IC’s
• Numerous benefits attained by scaling of transistors:
FasterSpeed
Higher Current DriveLower Capacitance
Higher Integration DensityLower Supply Voltage
LowerPower
Higher Circuit Complexity & Economy of Scale
• But … some drawbacks:poorer reliability (e.g., hot e- effects)lower dynamic range (analog ckts suffer)
Page 10
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Example: Micromechanical Accelerometer
• The MEMS Advantage:>30X size reduction for accelerometer mechanical elementallows integration with IC’s
xo
x
a
Acceleration
Inertial Force
Spring
Proof Mass
Basic Operation Principle
400
μm
Analog Devices ADXL 78
Displacement
maFx i =∝
Tiny mass means small output need integrated transistor
circuits to compensate
Tiny mass means small output need integrated transistor
circuits to compensate
Page 11
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
DisplaysPhased-ArrayAntenna
Technology Trend and Roadmap for MEMSin
crea
sing
abi
lity
to c
ompu
te
increasing ability to sense and act
Num
ber o
f Tra
nsis
tors
Number of Mechanical Components
100
101
102
103
104
105
106
107
108
109
100 101 102 103 104 105 106 107 108 109
AdaptiveOptics
Integrated FluidicSystems
DistributedStructural
ControlTerabit/cm2
Data Storage
Optical Switches& Aligners
InertialNavigationOn a Chip
Weapons,Safing, Arming,
and Fusing
Majority ofEarly MEMS
Devices(mostly sensors)
ADXL-50
Digital MicromirrorDevice (DMD)
ADXL-278
Future MEMSIntegration Levels
Enabled Applications
OMM 32x32
ADXL-78
CPU’s
Pentium 4
i-STAT 1
ADXRSCaliper Microfluidic Chip
Adv.: small size, small sample, fast analysis speed
Adv.: small size, small sample, fast analysis speed
Caliper
Analog Devices ADXRSIntegrated Gyroscope
Adv.: small sizeAdv.: small size
OMM 8x8 OpticalCross-Connect Switch
Adv.: faster switching, low loss, larger networks
Adv.: faster switching, low loss, larger networks
Adv.: low loss, fast switching, high fill factor
Adv.: low loss, fast switching, high fill factor
TI Digital Micromirror Device
Page 12
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Benefits of Size Reduction: MEMS
• Benefits of size reduction clear for IC’s in elect. domainsize reduction speed, low power, complexity, economy
• MEMS: enables a similar concept, but …
MEMS extends the benefits of size reductionbeyond the electrical domain
Performance enhancements for applicationdomains beyond those satisfied by electronics
in the same general categories Speed
Power ConsumptionComplexity
Economy
Frequency , Thermal Time Const. Actuation Energy , Heating Power Integration Density , Functionality Batch Fab. Pot. (esp. for packaging)
Page 13
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Nano Mechanical Array Signal Processors (NMASP)
Page 14
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Basic Concept: Scaling Guitar StringsGuitar String
Guitar
Vibrating “A”String (110 Hz)Vibrating “A”
String (110 Hz)
High Q
110 Hz Freq.
Vib.
Am
plitu
de
Low Q
r
ro m
kfπ21
=
Freq. Equation:
Freq.
Stiffness
Mass
fo=8.5MHzQvac =8,000
Qair ~50
μMechanical Resonator
Performance:Lr=40.8μm
mr ~ 10-13 kgWr=8μm, hr=2μmd=1000Å, VP=5VPress.=70mTorr
[Bannon 1996]
Page 15
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
-60
-50
-40
-30
-20
-10
0
8.7 8.9 9.1 9.3Frequency [MHz]
Tran
smis
sion
[dB
]
Pin=-20dBm
In Out
VP
Sharper roll-off
Sharper roll-off
Loss PoleLoss Pole
Performance:fo=9MHz, BW=20kHz, PBW=0.2%
I.L.=2.79dB, Stop. Rej.=51dB20dB S.F.=1.95, 40dB S.F.=6.45
Performance:fo=9MHz, BW=20kHz, PBW=0.2%
I.L.=2.79dB, Stop. Rej.=51dB20dB S.F.=1.95, 40dB S.F.=6.45
Design:Lr=40μm
Wr=6.5μm hr=2μm
Lc=3.5μmLb=1.6μm VP=10.47VP=-5dBm
RQi=RQo=12kΩ
[S.-S. Li, Nguyen, FCS’05]
3CC 3λ/4 Bridged μMechanical Filter
[Li, et al., UFFCS’04]
Page 16
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
• Constructed in SiC material w/ 30 nm Al metallization for magnetomotive pickup
Lr
Wr
h
Frequency [GHz]
Mag
neto
mot
ive
Res
p. [n
V]
Design/Performance:Lr =1.1 μm, Wr =120 nm, h= 75 nm
fo=1.029 GHz, Q =500 @ 4K, vacuum
Design/Performance:Lr =1.1 μm, Wr =120 nm, h= 75 nm
fo=1.029 GHz, Q =500 @ 4K, vacuum
[Roukes, Zorman 2002]
Nanomechanical Vibrating Resonator
Page 17
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Scaling-Induced Performance Limitations
Mass Loading Noise
• Differences in rates of adsorption and desorption of contaminant molecules
mass fluctuationsfrequency fluctuations
Temperature Fluctuation Noise
• Absorption/emission of photons
temperature fluctuationsfrequency fluctuations
ContaminantMolecules
NanoresonatorMass ~10-17 kg
mk
2π1
of =
Photons
NanoresonatorVolume ~10-21 m3
• Problem: if dimensions too small phase noise significant!• Solution: operate under optimum pressure and temperature
[J. R. Vig, 1999]
Page 18
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
-100-98-96-94-92-90-88-86-84
1507.4 1507.6 1507.8 1508 1508.2
1.51-GHz, Q=11,555 Nanocrystalline Diamond Disk μMechanical Resonator
• Impedance-mismatched stem for reduced anchor dissipation
• Operated in the 2nd radial-contour mode• Q ~11,555 (vacuum); Q ~10,100 (air)• Below: 20 μm diameter disk
PolysiliconElectrode R
Polysilicon Stem(Impedance Mismatched
to Diamond Disk)
GroundPlane
CVD DiamondμMechanical Disk
Resonator Frequency [MHz]
Mix
ed A
mpl
itude
[dB
]
Design/Performance:R=10μm, t=2.2μm, d=800Å, VP=7Vfo=1.51 GHz (2nd mode), Q=11,555
fo = 1.51 GHzQ = 11,555 (vac)Q = 10,100 (air)
[Wang, Butler, Nguyen MEMS’04]
Q = 10,100 (air)
Page 19
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Wireless Phone
Miniaturization of RF Front Ends
26-MHz XstalOscillator
26-MHz XstalOscillator
DiplexerDiplexer
925-960MHz RF SAW Filter925-960MHz
RF SAW Filter
1805-1880MHz RF SAW Filter
1805-1880MHz RF SAW Filter
897.5±17.5MHz RF SAW Filter
897.5±17.5MHz RF SAW Filter
RF Power Amplifier
RF Power Amplifier
Dual-Band Zero-IF Transistor Chip
Dual-Band Zero-IF Transistor Chip
3420-3840MHz VCO
3420-3840MHz VCO
90o0o
A/D
A/D
RF PLL
Diplexer
From TX
RF BPF
Mixer I
Mixer Q
LPF
LPF
RXRF LO
XstalOsc
I
Q
AGC
AGC
LNA
Antenna
Problem: high-Q passives pose a bottleneck against miniaturizationProblem: high-Q passives pose a bottleneck against miniaturization
Page 20
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Wireless Phone
Raised Inductor Q ~30-70
Raised Inductor Q ~30-70
Planar Spiral Inductor
Planar Spiral Inductor
Q <10 too small
Q <10 too small
Vibrating Resonator1.5-GHz, Q~12,000
Vibrating Resonator1.5-GHz, Q~12,000
Vibrating Resonator72-MHz, Q~146,000
Vibrating Resonator72-MHz, Q~146,000
Single-ChipRealization
Miniaturization of RF Front Ends
Page 21
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Chip-Scale Atomic Clocks (CSAC)
Page 22
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA NIST F1 Fountain Atomic Clock
VolVol: ~3.7 m: ~3.7 m33
Power: ~500 WPower: ~500 WAcc: Acc: 11××1010––1515
Stab: 3.3x10Stab: 3.3x10--1515/hr/hr
Physics PackagePhysics Package
After 1 sec Error: 10-15 secAfter 1 sec
Error: 10-15 sec
Loses 1 sec every 30 million years!
Loses 1 sec every 30 million years!
Page 23
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Benefits of Accurate Portable Timing
Secure Communications
Networked Sensors
Faster frequency hop ratesFaster frequency hop rates
Faster acquire of pseudorandom signals
Faster acquire of pseudorandom signals
Superior resilience against jamming or
interception
Superior resilience against jamming or
interception
More efficient spectrum utilization
More efficient spectrum utilization
Longer autonomy periodsLonger autonomy periods GPS
Faster GPS acquireFaster GPS acquire
Higher jamming margin
Higher jamming margin
Fewer satellites needed
Fewer satellites needed
Larger networks with longer autonomy
Larger networks with longer autonomy
Better TimingBetter Timing
Page 24
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Accurate Portable Timekeepers
• Stability quite good at carrier offset freqs. around 1 kHz
• L{fm=1kHz} = -140dBc/Hz
High-Q Oscillators• Tank Q ~ tens of thousands• Example: crystal oscillator
• Quartz: vibrating mechanical resonance high Q, stable
10MHz
Shear Mode
Excellent stability at offset
frequencies ~1kHz
Excellent stability at offset
frequencies ~1kHz
Accuracy (@ tiny offset freqs.) good,
but not good enough for some applications
Accuracy (@ tiny offset freqs.) good,
but not good enough for some applications
Page 25
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Atomic Clock Fundamentals
133Cs
m = 0f = 4
m = 0f = 3
m = 1
• Frequency determined by an atomic transition energy
Energy Band Diagram
Excite e- to the next orbital
Excite e- to the next orbital
Opposite e- spins
Opposite e- spins
ΔE = 0.000038 eV
ΔE = 1.46 eV
ν = ΔE/h= 352 THz852.11 nm
ν = ΔE/ħ= 9 192 631 770 Hz
Page 26
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Miniature Atomic Clock Design
ν = ΔE/ħ= 9 192 631 770 Hz
HyperfineSplitting Freq.
HyperfineSplitting Freq.
Sidebands
ModulatedLaser
PhotoDetector
133Cs vapor at 10–7 torr
Mod f
μwave osc
VCXO4.6 GHz
9.2GHz
4.6GHz
Atoms become transparent to light at 852 nm
Atoms become transparent to light at 852 nm
Carrier(852 nm)
λ
Close feedback loop to lock
Close feedback loop to lock
vo
Page 27
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
State-of-Practice
State-of-Research
Datum R2000Vol: 9,050 cm3
Power: 60 WAcc: 5×10–11
Temex RMOVol: 230 cm3
Power: 10 WAcc: 1×10–11
NISTNIST-- F1F1
CSACCSAC
VolVol: 1 cm: 1 cm33
Power: 30 Power: 30 mWmWAcc: 1x10Acc: 1x10--1111
Stab: Stab: 11××1010––1111/hr/hr
Miniaturizing Atomic Clocks
VolVol: ~3.7 m: ~3.7 m33
Power: ~500 WPower: ~500 WAcc: Acc: 11××1010––1515
Stab: 3.3x10Stab: 3.3x10--1515/hr/hr
μs/dayμs/day
ps/dayps/day
Page 28
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Chip-Scale Atomic Clock
• Key Challenges:thermal isolation for low powercell design for maximum Qlow power μwave oscillator
Atomic Clock Concept Cs or Cs or RbRbGlassGlassDetectorDetector
VCSELVCSEL
SubstrateSubstrate
GHzGHzResonatorResonatorin Vacuumin Vacuum
MEMS andMEMS andPhotonic Photonic
TechnologiesTechnologies
VolVol: 1 cm: 1 cm33
Power: 30 Power: 30 mWmWStab: Stab: 11××1010––1111
Chip-ScaleAtomic Clock
Laser 133Cs vapor at 10–7 torr
Mod f
μwave oscVCXO
4.6 GHzvo PhotoDetector
Page 29
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Pros and Cons of Miniaturization
Page 30
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Rth= 18 K/WCth= 40 J/K
Rth= 18 K/WCth= 40 J/K Rth= 61,000 K/W
Cth= 4.8x10-5 J/KRth= 61,000 K/WCth= 4.8x10-5 J/K
P RthCth
T = P x Rth
Cth ~ volume
Rth ~ support lengthX-section area
Macro-Scale Micro-Scale
P (@ 80oC) = 1 mWP (@ 80oC) = 1 mW
Warm Up, τ = 3 sWarm Up, τ = 3 s
P (@ 80oC) = 3 WP (@ 80oC) = 3 W
Warm Up, τ = 12 min.Warm Up, τ = 12 min.
3,000x lower power3,000x lower power
240x faster warm up240x faster warm up
300x300x300 μm3
Atomic Cell @ 80oC
Long, Thin Nitride
Tethers w/ Metal Leads
T Sensor(underneath)
Heater
LaserInsulation
Macro-Oven(containing heater
and T sensor)2 cm3
Atomic Cell @ 80oC
Thermally Isolating Feet
Laser25oC
2 cm
Micro-Scale Oven-Control Advantages
Page 31
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Challenge: Miniature Atomic CellLarge Vapor Cell Tiny Vapor Cell
1,000XVolumeScaling
Wall collision dephasesatoms lose coherent state
Wall collision dephasesatoms lose coherent state
Inte
nsity
Mod f9.2 GHz
SurfaceVolume
More wall collisions stability gets worse
More wall collisions stability gets worse
lower Qlower Q
lowest Qlowest QAtomic
Resonance
Page 32
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Challenge: Miniature Atomic CellLarge Vapor Cell Tiny Vapor Cell
1,000XVolumeScaling
Inte
nsity
Mod f9.2 GHz
Atomic Resonance
Soln: Add a buffer gas
Soln: Add a buffer gas
Lower the mean free path of the atomic vapor
Lower the mean free path of the atomic vapor
Return to higher Q
Return to higher Q
Buffer Gas
Page 33
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
1.5 mm
4.2 mm
1.5 mm
Laser
Optics
Cell
Photodiode
1 mm
Total Volume: 9.5 mm3 Stability: 2.4 x 10-10 @ 1sCell Interior Vol: 0.6 mm3 Power Cons: 75 mW
Total Volume: 9.5 mm3 Stability: 2.4 x 10-10 @ 1sCell Interior Vol: 0.6 mm3 Power Cons: 75 mW
1st Chip-Scale Atomic Physics Package
GlassND
SiQuartz
ND
Lens
Alumina
VCSEL
Page 34
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Tiny Physics Package Performance
NIST’sChip-Scale
Atomic Physics Package
NIST’sChip-Scale
Atomic Physics Package
40 50 60 70 80 905.65
5.66
5.67
PD S
igna
l [V]
Frequency Detuning, Δ [kHz] from 9,192,631,770 Hz
7.1 kHz
Contrast: 0.91% 2.4e-10 Allandeviation @ 1 s2.4e-10 Allan
deviation @ 1 s
• Experimental Conditions:Cs D2 ExcitationExternal (large) Magnetic ShieldingExternal Electronics & LO Cell Temperature: ~80 ºCCell Heater Power: 69 mWLaser Current/Voltage: 2mA / 2VRF Laser Mod Power: 70μW
DimeDime
Open Loop Resonance:Drift to Be Removed in Phase 3
Drift to Be Removed in Phase 3
Sufficient to meet CSAC
program goals
Sufficient to meet CSAC
program goals
100 101 102 103 104 10510-12
10-11
10-10
10-9
Alla
n D
evia
tion,
σy
Integration Time, τ [s]
Stability Measurement:
Drift IssueDrift Issue
Rb (D1)Rb (D1) 1 day1 day1 hour1 hour
Cs (D2)Cs (D2)
Q =1.3x106Q =1.3x106
CSAC Goal
Page 35
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Physics Package Power Diss. < 10 mW
Heater/Sensor SuspensionCesium cell
Frame Spacer
VCSEL Suspension
VCSEL / Photodiode 20 pin LCC
7 mm
0
2
4
6
8
10
12
0 20 40 60 80 100 120 140Temperature [oC]
Pow
er [m
W] Measured
Model
Only ~5 mWheating power
needed to achieve 80oC
cell temperature
Only ~5 mWheating power
needed to achieve 80oC
cell temperature
• Achieved via MEMS-based thermal isolation
Symmetricom / Draper Physics
Package Assembly
Symmetricom / Draper Physics
Package Assembly
Page 36
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA A 9.95 cc, 153 mW Atomic Clock w/ Chip-Scale Physics Package
3.94 cm
3.53
cm
Physics PackagePhysics Package
Symmetricom’smeasured Allan deviation easily satisfies CSAC Phase 2 Goal
Symmetricom’smeasured Allan deviation easily satisfies CSAC Phase 2 Goal
0.47 cm
Packaged CSAC
153 mWTotal:
10 mWC-field
51 mWHeater Power (air)
2 mWVCSEL drive
75 mWRF
6 mWSignal Processing
8 mWMicroprocessor
1 mWPower Regulation
Physics PackagePhysics Package
1.43 cmPower Budget < 200 mW0.64 cm
9.95 cm3 Total Package Volume9.95 cm3 Total
Package Volume
MEMS-based thermal isolation allows low physics package power consumption
MEMS-based thermal isolation allows low physics package power consumption
5x10-10
@ 1 sec5x10-10
@ 1 sec
CSAC Phase 2 Goal
Page 37
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Atomic Clock Technology Progression
State-of-Practice
State-of-Research
HP 5071AVol: 29,700 cm3
Power: 50 WAcc: 5×10–13
Datum R2000Vol: 9,050 cm3
Power: 60 WAcc: 5×10–11
Temex RMOVol: 230 cm3
Power: 10 WAcc: 1×10–11
NISTPP Vol: 9.5 mm3
Power: 75 mW+elect.Stab: 10–11/hr
Symmetricom CSACVol: 9.95 cm3
Power: 153 mWStab: 5×10–11/100s
NISTNIST-- F1F1
VolVol: ~3.7 m: ~3.7 m33
Power: ~500 WPower: ~500 WAcc: Acc: 3.83.8××1010––1515
Stab: 3.3x10Stab: 3.3x10--1515/hr/hr CSACCSAC
VolVol: 1 cm: 1 cm33
Power: 30 Power: 30 mWmWAcc: 1x10Acc: 1x10--1111
Stab: Stab: 11××1010––1111/hr/hr
Stab = Allan deviation/integration time
Stab = Allan deviation/integration time
Physics PackagePhysics Package
Page 38
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Micro Gas Analyzers (MGA)
Page 39
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
Gas SensitivePolymer
CapacitorPlates
Conventional Sensor Separation Analyzer
Micro Gas Analyzers• Objective: enable remote detection of chemical agents via tiny,
ultra-low power, fast, chip-scale gas analyzers that greatly reduce the incidence of false positives
• Approach: use micromachining technologies to implement separation-based analyzers (e.g., gas chromatographs, mass spectrometers) at the micro-scale to enhance gas selectivity
ΔC ~gas conc.
Species A Species B
• Problem: polymer has finite sensitivity to both A & B
Species A
B
A
B
• Result: species A & B now separated can identify and analyze individually
• Problem: too big, too slow, power hungry
Page 40
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Advantages of Miniaturization
Portable Gas Chromatograph Chip-Scale Gas Chromatograph
Reduction FactorsReduction FactorsReduction Factors
13″
19″ Depth = 10″Preconcentrator Detector Array
5 m
m
SeparationColumn
1-2 cmMicropump
Size 40,500 cm3
Sensitivity 1 ppb
Analysis Time 15 min.
Energy Per Analysis 10,000 J
Size 2 cm3
Sensitivity 1 ppt
Analysis Time 4 sec
Energy Per Analysis 1 J
225X225X225X
10,000X10,000X10,000X
20,000X20,000X20,000X
1,000X1,000X1,000X
Page 41
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Basic Approach: Separation Analyzer
Input GasMixture Pre-Concentrator Separator Detector
ElectronicProcessor
ThreeAnalytes
CompactedSlice of
Analytes
SeparatedAnalytes
Pump
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Min
iatu
rizat
ion
Page 42
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Multi-Stage Pre-Concentration
ToSeparator
[Honeywell]
AbsorbentFilmHeater
Incoming GasMixture
Release Analytein Phase
Release Analytein Phase
Thin Concentrated PlugThin Concentrated Plug
Heat ~2mW to Release
Analyte
Heat ~2mW to Release
Analyte
10 ppb10 ppb
1,000 ppb1,000 ppb
2,000 ppb2,000 ppb
3,000 ppb3,000 ppb 5,000 ppb5,000 ppb
Page 43
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Fast PHASED Pre-Concentration
• Below: 20-stage PHASED preconc./sep. analysis of a 720 ppmHexane-in-Air sample at 60 cm/s sample velocity
PHASED chip with 20-stage
preconcentratorand separator
PHASED chip with 20-stage
preconcentratorand separator
Only 300ms needed for sample
absorption @ 20oC!
Only 300ms needed for sample
absorption @ 20oC!300 ms300 ms
Sample Expulsion @ 120oC
Sample Expulsion @ 120oC
Page 44
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Basic Approach: Separation Analyzer
Input GasMixture Pre-Concentrator Separator Detector
ElectronicProcessor
ThreeAnalytes
CompactedSlice of
Analytes
SeparatedAnalytes
Pump
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Tiny Dimensionsfaster separationlower power
Tiny Dimensionsfaster separationlower power
Min
iatu
rizat
ion
Page 45
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Scaling Leads to Faster Separation
• Example: gas chromatograph separation column
unique analyte interactions with the column wallsdifferent analyte velocitiesresult: separation after a finite distance
WideChannel
ThinChannel
240 μm
150 μm
StationaryPhase
Carrier Gas (Mobile Phase)Miniaturize
x
Conc. Conc.
x
PeakBroadens
PeakStays Thin
LessSeparationNeeded toResolve
Page 46
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Scaling Leads to Faster Separation
• Example: gas chromatograph separation column
unique analyte interactions with the column wallsdifferent analyte velocitiesresult: separation after a finite distance
WideChannel
ThinChannel
240 μm
150 μm
StationaryPhase
Carrier Gas (Mobile Phase)Miniaturize
• Result of Scaling: shorter column length; faster analysis time
Surface-to-Volume Ratio
PeakSpreading
SeparationDistance
ColumnWidth
Page 47
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Gas Chromatography in Less Than 4s!
Solv
ent
Tolu
ene
DM
MP
DEM
P
DIM
P n-do
deca
ne
1-de
cano
l
3-m
ethy
lhex
ane
0 1.2 2.4 3.6 4.8
16000
32000
46000
64000
Elution time [s]
80000
Rel
ativ
e In
tens
ity
Peak capacity >40, in 4 sec
Peak capacity >40, in 4 sec
Sandia’s micro-GC ColumnSandia’s micro-GC Column
Design/Measurement Data:0.75m x 100μ column
0.1μ DB-5 stationary phaseHeart-cut 275 msec peak injection
Temperature: ~30 deg C/secH2 carrier: 35-39 psi at 1 psi/sec
Design/Measurement Data:0.75m x 100μ column
0.1μ DB-5 stationary phaseHeart-cut 275 msec peak injection
Temperature: ~30 deg C/secH2 carrier: 35-39 psi at 1 psi/sec
1,6-
dich
loro
hexa
ne Green = AnalyteBlue = Inteferent
Green = AnalyteBlue = Inteferent
Page 48
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Basic Approach: Separation Analyzer
Input GasMixture Pre-Concentrator Separator Detector
ElectronicProcessor
ThreeAnalytes
CompactedSlice of
Analytes
SeparatedAnalytes
Pump
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Tiny Dimensionsfast time constants10,000X gain factor via multi-stagingenhanced sensitivitylower power
Tiny Dimensionsfaster separationlower power
Tiny Dimensionsfaster separationlower power
Tiny Dimensionshigher sensitivityfaster refresh ratelower powerarrays for specificity
Tiny Dimensionshigher sensitivityfaster refresh ratelower powerarrays for specificity
Min
iatu
rizat
ion
Page 49
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Zeptogram Mass Sensors
0 50 100 150 200 250 300 350
-1000
-800
-600
-400
-200
0
Freq
uenc
y S
hift
(Hz)
Time (sec)
~100 zg
Nozzle
ShutterNanomechanicalResonator
0 1000 2000 3000 4000-3000
-2500
-2000
-1500
-1000
-500
0
133 MHz190 MHz
Freq
uenc
y S
hift
(Hz)
Mass (zeptograms)
0 2000 40000.1
1
10
100
δ m (
zg)
Time (s)
~7 zg
>1Hz/zg100 zg Au atom
clumps resolved!100 zg Au atom
clumps resolved!
Nanomechanical ResonatorNanomechanical Resonator
Measurement noise level indicates ~7 zg of resolutionMeasurement noise level
indicates ~7 zg of resolution
Au
Page 50
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA
• Example: ion trap mass spectrometer (ITMS)separate analytes by molecular weight
• Advantages of Miniaturization:can support smaller mean free path relaxed vacuum req.result: substantially lower power requirement
ITMS: Scaling Leads to Lower Power
vRF
Detector
DetectorVRF
time
time
DetectorOutput
100 ms TrappedIons
Stability Thresholds(dependent on mass)
Page 51
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Operation at 1.7 Torr!
80 90 100 110 120
0.04
0.08
0.12
0.16
Det
ecto
r Cur
rent
[pA
]
m/z
Mass spectrum of DMMP by a single 1-mm ion trap @ 1.7 TorrMass spectrum of DMMP by a
single 1-mm ion trap @ 1.7 Torr
Resistive Glass Drift Tube DetectorResistive Glass
Drift Tube Detector
Test Jig (vacuum chamber insert)
Test Jig (vacuum chamber insert)
ChanneltronDetector
ChanneltronDetector
Highest pressure ever demonstrated!Highest pressure
ever demonstrated!
1mm Cylindrical Ion Trap1mm Cylindrical Ion Trap
Page 52
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA 40μm-Ion Traps Functional!
• Should allow much higher pressure operation (~76 Torr)
Xe ion peakXe ion peak
Array of 40μm Si Ion TrapsArray of 40μm Si Ion Traps
Silicon
Heavily-Doped Polysilicon Oxide Aluminum
Xenon signal obtained at low He pressure (10-4 Torr)Xenon signal obtained at
low He pressure (10-4 Torr)
Page 53
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Gas Analyzer Technology Progression
LLNLVol: 40,500 cm3
Power: 11.5 WEnergy/Analysis: 10 kJAnalysis Time: 15 min.
MGA ObjectiveMGA ObjectiveVolVol: 2 cm: 2 cm33
Power: <200 Power: <200 mWmWEnergy/Analysis: 1 JEnergy/Analysis: 1 J
Analysis Time: 4 sAnalysis Time: 4 s
Agilent 6852AVol: 60,000 cm3
Power: 20 WEnergy/Analysis: 18 kJAnalysis Time: 15 min.
Sandia μChem LabVol: 1,050 cm3
Power: 4.5 WEnergy/Analysis: 540 JAnalysis Time: 2 min.
Gas Chromatograph/Mass Spectrometer (GC/MS) is
a “gold standard” in chemical gas detection with excellent immunity
to false alarms
Gas Chromatograph/Mass Spectrometer (GC/MS) is
a “gold standard” in chemical gas detection with excellent immunity
to false alarms
Problems: too big, too slow, power hungry
Problems: too big, too slow, power hungry
Solution: use MEMS technology to miniaturize the GC/MS, which in turn makes it faster and more
energy efficient
Solution: use MEMS technology to miniaturize the GC/MS, which in turn makes it faster and more
energy efficient
small enough for projectile delivery1 ppt det. limitvery fastbattery operable
small enough for projectile delivery1 ppt det. limitvery fastbattery operable
Page 54
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Example: Micromechanical Accelerometer
• The MEMS Advantage:>30X size reduction for accelerometer mechanical elementallows integration with IC’s
xo
x
a
Acceleration
Inertial Force
Spring
Proof Mass
Basic Operation Principle
400
μm
Analog Devices ADXL 78
Displacement
maFx i =∝
Tiny mass means small output need integrated transistor
circuits to compensate
Tiny mass means small output need integrated transistor
circuits to compensate
Page 55
C. T.-C. Nguyen, “From MEMS to NEMS: Smaller Is Still Better,” MARC’06 Meeting, 1/25-26/05
DARPADARPA Conclusions
• MEMS are micro-scale or smaller devices/systems that operate mainly via a mechanical or electromechanical means
• MEMS NEMS offer the same scaling advantages that IC technology offers (e.g., speed, low power, complexity, cost), but they do so for domains beyond electronics:
• Micro … nano … it’s all good• Just as important: MEMS or NEMS have brought together
people from diverse disciplines this is the key to growth!• What’s next? Nano-nuclear fusion? Chip-scale atomic
sensors? … limitless possibilities …
resonant frequency (faster speed)actuation force (lower power)
# mechanical elements (higher complexity)integration level (lower cost)
Size