Nanoenabled Self-Powered Sensor
Systems for Personal Health and
Personal Environmental Monitoring
Veena Misra
Director, NSF ASSIST Nanosytems Center
Professor of ECE
North Carolina State University
Korea-U.S. Forum on Nanotechnology
September 29-30, 2014
1960
1970
1980
1990
2000
2010
2020
2
Lung Pollution in China
Thingprogress.org
3
Environmental
Factors
Air Quality
Particulates
Gases
Heavy Metals
Chemicals
Noise
Organic Solvents
4
Altered
Pathways
Inflammation
Oxidative Stress
Insulin Signal Disruption
Biological Impact of Environmental
Triggers
Health + Environmental Link =
Exposure
Chronic
Disease
Heart Disease
Arrhythmia
Lung Disease
Asthma
Brain
Alzheimers
Stress
5
Medtronic, IEEE Spectrum 2014
ASSIST Vision
•Direct correlation of personal health and
personal environment
•Correlation of multimodal health sensing to
produce a systemic picture of wellness
6
The Wearable Health Market
Vital Connect
Piix Heart
Monitor
Metria Patch by Vancive
MC10 Hydration Sensor
Nike
Fuel
Valencell
Proteus Digital Health
Sotera Wireless
FDA Approved
What are the gaps in these products?
FitBit
Consumer Products
MISFIT
ASSIST technologies can provide disruptive advances
in existing wearable products and influence the
future generation of wearable systems
7
Effectiveness of Wearable Health
Products is lacking
Most wearable's are not being used by
65+ older citizens
Usage of sustained utilization drops
below 50% in less then 15 months
A third of the users stop wearing the
wearable device within 6 months
Endeavor Partners, May 2014
Why is Effectiveness Low?
Human and Social
Factors/User
Feedback
Form
Factor/Design/A
esthetics
Battery
Life/Hassle
Open
Platforms
Actionable
Information/Multiple
Sensors
Quality/Robustness/
Medically Validation
Comfortable and
biocompatible
9
Potential Solutions?
• The right multimodal, unobtrusive sensors for health and environment
• The data backbone supporting these sensors
• Medically validated / reliable sensor data
AND
• Ultra Low Power Components to enable ultra long battery lifetime
AND
• Energy harvesting energy to realize infinite lifetimes
10
Ultra Low Power Components
Computation
Communication
Sensors
Power Management
Nanotechnologies
Towards
continuous
monitoring
Interfaces
11
Optimizing the Power for
Computation and Communication
Biocompatibility
COTS Add-on
User
Health & Environmental
Sensing
Platform
Energy Storage
Antenna
Software
Body Energy
Harvesting
SoC
Gas Sensor
Data Aggregator
Power Management
Digital Control /
Processing /
Management
R
A
D
I
O
Signal Processing
Software
Analog Front End
Health Sensors
Physiological
Platform
12
Battery-free
Energy Harvesting ExG with MICS-Band Radio
1.5
Voltage (V)
2.5 mm
0.13 µm CMOS
1
RF Pulse, -10dBm
VBoost (on storage
cap)
Boost Converter
turns on
0.5
3.3 mm
•
•
•
•
•
•
•
•
4 Channel ExG
MICS Radio
0
5 DC-DC converters
0
2
MCU + Dig Accelerators
Integrated Power Management
MUCH more integration than any other wireless BSN SoC
Wireless RF pulse provides one-time kick-start
Runs indefinitely thereafter on thermal energy
VThermoElectricGenerator
4
6
time (s)
Source: Calhoun et al., ISSCC 2012
8
10
Ultra Low Power SoC
ADC IN
(V)
1
655 ms
19µW total chip
power from 30
mV input supply
VBoost
sample
0.8
0.6
0.4
TX EN
1
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
650 µs
0.5
Source: Calhoun et al.,
ISSCC 2012
1
TX DATA
0
1
Time (s) Header
Data
0.5
0.2
0.4
00.6
1
0.5
0.8
1
CRC
1.2
1.4
1.798
1.7981
1.7982
1.7983
1.7984
1.7985
1.7986
1.798
1.7981
1.7982
1.7983
1.7984
1.7985
1.7986
1.6
1.8
0.5
0
1.7979
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.7987
1.6
1.8
EKG < 20mW relying only on energy harvesting and storage capacitors
Boost converters < 10mV
14
Highlight: Measured Radio
Performance
ASSIST Ultra wideband (UWB)
Transmit
• ULP TX for system level
energy savings
960µm
ASSIST Receivers
• Best in class power vs
sensitivity
• ~100nW Wakeup RX
UWB
1250µm
10k
Power [μW]
1k
100
10
Spec
ASSIST
1
Low Power Radios
Clock Harvesting Rx
0.1
-100
UWB Rx
Wake Up Rx
-80
-60
Sensitivity [dBm]
15
Wentzloff, UMich
-40
-20
Power
Data Rate
Center Frequency
Bandwidth
Output power
Value
7.44
187.5
3.8
490
---
Unit
µW
kbps
GHz
MHz
dBm
How low is ULTRA low when it comes to Power
ASSIST SoC / processor
EnOcean STM 31xC
IMEC ISSCC’14
Semtech SX1282
Supply voltage
0.5V
2.1V to 5.0V
1.2V
1.0V to 1.6V
Processor
16b MSP430
custom
32b ARM Cortex M0
8b CoolRISC
Processor perf.
<1µW @ 200 kHz
5.1mA @3-5V
Not reported
1.2µW @ 32 kHz, 600 µW
typical
Power harvesting
Yes. RF, Solar, TEG
Yes.
No.
No.
ASSIST Radios
EnOcean STM 31xC
IMEC ISSCC’14
Semtech SX1282
Supply voltage
0.5V to 1.0V
2.1V to 5.0V
1.2V
1.0V to 1.6V
TX power consump.
6µW @200kbps
30mW @ 125 kbps
4.6 mW
~40mW
RX power consump.
200µW WBAN RX
120nW @12.5 kbps WU
40mW
3.8 mW
~12mW
ASSIST Antennas
State of the Art
Form factor
~50mm x 50mm
~ twice as big
Front to back ratio
~19 dB (3X – 4X more range)
~ 5 dB
Signal suppression
3-5 dB (>50X less power)
20-30 dB
Out of band rejection
30 dB
5-10 dB
16
Requires a systems driven, nanotechnology enabled
approach to realize low-power wearable sensors for
environmental and physiological monitoring
Gas/particulate sensing
Biochemical sensing
Ozone, NOx, H2S, VOCs,
particulate matter
Cortisol, epinephrine
electrolytes/hydration
Nano-structured
MOx
Polymer coated
MEMS
Hydrogel skin
interface
Environmen
Platform
Stress, cardiovascular health
Optical particulate sensing
Asthma, cardiovascular health
Bioelectrical sensing
Biophotonic sensing
Pulse oximetry
Acoustic sensing
Body sounds
Optical
TX/RX
Low-power
controller/radio
Cardiovascular health
Polymer coated
sdAb
COTS microphone
Asthma
ExG, hydration
AgNW
electrodes
Conductive
hydrogels
Hydration, cardiovascular health
17
Metal-oxide nanowire based gas sensors are
amenable to integration with CMOS
CMOS chip designed
for MOx nanosensor
integration
18
Sensors 2012, 12, 5517-5550
Metal-oxide nanowire based gas sensors amenable
to integration with CMOS
MOx coated Si NWs self assembled on CMOS
Anatase TiO2 chemiresistor
Rutile SnO2 chemiresistor
Crystalline metal-oxide-coated Si nanowires were fabricated using DRIE
and ALD, and self-assembled on electrode arrays
Self heating to 175°C with <20 μW power consumption is possible by
thermal
insulation
Sensors 2012, 12, 5517-5550
19
Silver nanowire electrodes on soft materials for
bioelectrical sensing
20
Hydrogel-based noninvasive passive skin interfaces
for sweat sampling
Sweat Methyl Red
PAAm
Color Spectrum
pH <4.4
4.4<pH<6.2 pH > 6.2
Used pH-color indicators to visually monitor diffusion
of acidic fluids (sweat mimics) through hydrogels
Diffusion
Capillary Tube
SPH Gel
Hydrated (Body)
Gel
Time lapse
over
roughly 12
hours
Peach pH Sensor
Diffusion
Sweat Diffusion
Fluid transfer from superporous hydrogel to a capillary
Establish a skin mimic for
interfacing and sensor testing
Hydrophilic Channels – Capillary Action
Superporous
Hydrogel –
Evaporation
SPH
Membrane
Sweat
Solution
Future Work: Create a CapillaryOsmotic Pump for continualpassive sweat intake
Superporous Hydrogel – Capillary
and Osmotic Sweat Intake
Usually used for drug
penetration; we will use for
sweat collection with various
skin (membrane) materials 21
Building the basic building blocks for a skin-coupled
biochemical sensor to monitor cortisol levels in
sweat
Liquid metal electrode
Optional evaporation opening
Intake rates via
osmotic uptake were
measured as ~15
nL/cm2min, much
greater than human
sweat rates (0.01 –
1 nL/cm2min)
Microfluidic channel
interfaced to sensor
sites
Ionic hydrogel
Potentiodynamic
electrochemical
measurements
using
functionalized
NW-enhanced
IDTs
Hydrogel for sweat
collection
sdAb
sdAb+antigen @ RT
sdAb+antigen @ 40C_7 Days
-6
6.0x10
-6
4.0x10
Graded protective coating for timed release
Current (Amps)
-6
OH
HO
OAc
OH
AcO
HO
O
O
O
O
O
O
AcO
OAc
OAc
HO
O
O
CN
OH
O
O
OH
OH
HO
OH
AcO
HO
OH
OAc
O
O
O
HO
OH
OH
n
Trehalose
Polymer-1
Encapsulation of antibody with polymer/trehalose enables time
delayed dissolution of coating to expose one sensor at a time for
long term use
sdAb-antigen
stability at 400C
for 7 days was
demonstrated.
2.0x10
0.0
-6
-2.0x10
-6
-4.0x10
-6
-6.0x10
-0.6
-0.4
-0.2
0.0
0.2
Potential (Volts)
0.4
0.6
22
Looking at both sides of the power equation 
Towards Self-Powered Operation
Energy Harvesting
One harvester does not fit all?
iMEC
24
This is a Very Difficult Problem to
Address for Human Worn Devices!
• Thermal: Small DT from human to ambient
• Mechanical: Weak base excitations at low frequencies for human
motion
• Indoor solar: 1 – 10 mW/cm2 available
• RF scavenging: << 1 mW from ambient rf. Much higher powers can
be achieved if directed rf is utilized.
• Inductive coupling: Tens mW available at close proximity
State of the Art Flexible
Thermoelectric Generators
18.8 mW/cm2 K2
Needs for flexible thermoelectric
generators
• Good materials (higher ZT)
• Longer length (100 – 300 mm) than
typical film thickness for Bi2Te3
• Nanostructuring to decrease
thermal conductivity
• Thermal resistance of harvester
should be on order hundreds cm2 K
/W
Insulation area/(active material area)
Glatz et al., J MEMS 18, 763 (2009) Leonov, J. Micromech. Microeng. 21 (2011) 125013
26
Flexible Thermoelectric Harvesters
Maximum'Power'(uW)'
NCSU TEG System Model
Approach:
2
• Flexible, open-platform TEG package
enabling integration of thermoelectrics
with excellent performance from many
sources
• Flexible, high performance heat sinks
• New material & process approaches
To Increase Power
High ZT
Reduced Parasitics
Heat
Collection
Heat
Rejection
Area = 1.5 cm x 1.5 cm
50 legs
w/o spreader
Barrier: Parasitics (esp. skin
resistance) have a huge impact
on TEG performance; output
voltages ~ 5 – 30 mV
• ~10 – 20 mW on a flexible hand-built package
with COTS components
• Better Flexible package
• Reducing skin resistance is essential
• “Nano” in PbTe/PeTe1-xSex heterostructures,
CNT-loaded polymers, ECD Bi2Te3 in alumina
templates, hierarchical wicking structures
Maximum Power (uW)
Thermoelectrics
400
350
300
250
200
150
100
50
0
lkin
Wa
TEG: 1.5 x 1.5 cm
50 legs
K = 0.8 – 1.5 W/mK
ZT=0.8
ZT = 1
ZT=1.2
ZT=1.5
kin
kin
kin
kin
S
S
S
t
t
t
es
ors
ors
,B
W
W
y
,
,
r
a
ing
ary
on
k
n
l
o
a
W
Sta
Sta
S
est
B
g,
zT of PbTe/PbSe0.2Te0.8
High zT
in short-period SLs
Approach zT
of bulk PbTe
28
Non-Resonant Human Worn
Harvesters
Piezoelectric
Plate
Measured prototype: 57 μW/cm2
peak power exceeds that of endloaded cantilever on Si by > 35 times
Rotor
Magnets
Shad Roundy
29
Activity
Walking
Jogging
Running
Power Available
95 mW
360 mW
700 mW
Power Density
10 mW/gram
37 mW/gram
74 mW/gram
Housing
Future Areas of Collaboration to Enable
Self-Powered Miniaturized Sensors
• Ultra low power electronics can change the sensing paradigm by
enabling long term and continuous sensing
• Sensors innovation lies in lowering the power consumption and
biochemical sensing
• Energy harvesting from the body is hard but is a worthwhile game
changing technology
• Thermoelectrics: Higher ZT, Flexible packaging, larger areas
• Mechanical: need new designs including novel form factors (shoes, knees,
chest straps)
• Nano is creating advancements in sensor system components
• Samsung and KAIST (low power electronics, energy harvesting,
sensors, flexible materials, manufacturing)
30
Thank you
31