See discussions, stats, and author profiles for this publication at: [Link]

net/publication/245308609

Wearable electronics self-powered by using human body heat: The state of the
art and the perspective

Article in Journal of Renewable and Sustainable Energy · November 2009

DOI: 10.1063/1.3255465

CITATIONS READS
173 6,325

2 authors:

V. Leonov Ruud Vullers

imec SMART Photonics
91 PUBLICATIONS 2,864 CITATIONS 176 PUBLICATIONS 4,087 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Wearable bioimpedance sensors for hydration monitoring and impedance spirometry View project

ULP Wireless - UWB View project

All content following this page was uploaded by V. Leonov on 16 May 2014.

The user has requested enhancement of the downloaded file.

 JOURNAL OF RENEWABLE AND SUSTAINABLE ENERGY 1, 062701 共2009兲

Wearable electronics self-powered by using human body

heat: The state of the art and the perspective
Vladimir Leonov1,a兲 and Ruud J. M. Vullers2,b兲
1
Smart Systems and Energy Technology Unit, Interuniversity Microelectronics Center
(IMEC), Kapeldreef 75, 3001 Leuven, Belgium
2
Smart Systems and Energy Technology Unit, IMEC/Holst Centre, High Tech Campus 31,
Eindhoven 5656 AE, The Netherlands
共Received 29 July 2009; accepted 6 October 2009; published online 6 November 2009兲

In this paper, we present our vision of what kind of wearable devices and how they
can be powered by the heat of human beings and by using ambient light. The basic
principles of designing body-powered devices and ways of their hybridizing with
photovoltaic cells are discussed. The mechanisms of thermoregulation in humans
and the laws of thermodynamics enable placing a distinct boarder between realistic
targets and the science fiction. These allow prediction of application areas for
wearable energy harvesters accounting for competitive batteries with long service
life. The existing family of body-powered wearable devices and new technologies
for thermopiles are discussed. The theory and practice point at the necessity of
using microelectronic and microelectromechanical system technologies for the tar-
get application area. These technologies for thermopiles offer the possibility of
reduced production cost. Therefore, autonomous systems powered thermoelectri-
cally could be successfully marketed. The related aspects of design and fabrication
are discussed. © 2009 American Institute of Physics. 关doi:10.1063/1.3255465兴

I. WHAT IS AN ENERGY HARVESTER

An energy harvester 共or an energy scavenger兲 is a small power generator that uses energy
available in the ambient, such as electromagnetic energy, wind, water flow, or a temperature
gradient. These sources are the same as those used in power plants or smaller power generators
such as the ones for powering houses in remote locations, light towers, spacecrafts, and on
transport 共except those based on fossil fuels兲. There are also microgenerators in the research stage,
which are to replace wiring and batteries in autonomous and stand-alone devices or in devices that
are difficult to access. Energy harvesters are being developed for the same purpose. An energy
harvester is typically from several to 1 cm size power microplant, which does not require fossil
fuel but converts into electricity any primary energy that is available in the ambient “for free.” The
reason to call them “harvesters” is a new application area: they are used for powering small
devices such as sensors or sensor nodes. This way of powering them eliminates the need for
cost-ineffective work such as wiring or either recharging or replacing batteries.
Despite the fact that plenty of microgenerators being developed worldwide are frequently
called scavengers or harvesters, not all of them are truly the ones. Probably, a miniature turbine in
a pipe cannot be called such as far as it introduces unwanted pressure drop. Therefore, to com-
pensate for this drop, the pressure at the pump station must be proportionally increased. Thereby,
such a turbine works actually as an ineffective converter of the energy spent by the pump station
into energy produced by the turbine.

a兲
Author to whom correspondence should be addressed. Electronic mail: leonov@[Link]. Tel.: ⫹32 16-288-367. FAX:
⫹32 16-281-576.
b兲
Electronic mail: vullersr@[Link].

1941-7012/2009/1共6兲/062701/14/$25.00 1, 062701-1 © 2009 American Institute of Physics

062701-2 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

The same is referred to a small dynamo or another converter of mechanical energy into
electricity. Such generator requires additional power, e.g., a cyclist spends more efforts if he/she
switches the bike’s dynamo on.
Converting the temperature differences available in the ambient can be safely called energy
harvesting unless it adversely affects humans themselves or the other objects built or used by
mankind. For example, the outer and inner wall surfaces of buildings have a temperature differ-
ence. A thermopile connected in between those two surfaces would generate energy; however,
such thermopile would cause additional heat exchange between the two surfaces. Therefore, this
way of power generation would require increasing the heating/cooling power of the home facilities
by a coefficient of 100–1000 more than the power gained.
Due to physical interaction of a microgenerator with ambient objects/subjects, it always af-
fects them to some extent. We arrive to the conclusion that a true energy harvester is not only a
microgenerator that uses energy available in the ambient. It also does not consume sensible
amount of primary energy required in a host/neighbor object/subject so there is no need to com-
pensate for any adverse effect caused by the harvester.

II. THE HISTORY OF ENERGY HARVESTING FOR POWERING DEVICES

It is a matter of fact that energy harvesters as we call them nowadays have been used for
centuries. They appeared before the age of electricity, therefore, their output was purely mechani-
cal. We can mention the invention of self-winding pocket watch by Swiss horologist Abraham-
Louis Perrelet before French revolution 共in about 1770兲 while not claiming that it was the first
known example of energy harvester in the history. The mainspring of this self-winding watch was
wound using the energy generated by the natural movements of the wearer.
Another source of power was used in the clock “Atmos” invented in 1927 by French engineer
Jean-Leon Reutter. The clock spring is wound using expansion or contraction of a sealed gas-filled
capsule due to natural day-night changes in ambient temperature and periodic variations in atmo-
spheric pressure.
Necessity of a wireless remote control for a television pushed American physicist Robert
Adler 共born in Austria兲 to invent in 1956 a device which was converting mechanical energy of
pushing the button on a remote control device into acoustic signals that controlled functions on the
television set 共i.e., it used the principle similar to that of tuning folks兲. The device called “Zenith
space command” stayed in use for over 2 decades. As far as we nowadays push buttons on our
battery-powered remote controls, there was no additional effort required from the user of ultra-
sonic remote control and it might be called energy harvester.
The age of electronics demands that the harvested energy must be converted into electricity.
Therefore, modern automatic watches generate electricity instead of winding the spring. The
photovoltaic 共PV兲 cells are also used in low-power handheld devices and watches. The first watch
powered by PV cells has been designed by American inventor Roger Riehl, which has been
marketed in 1972 under the brand name Synchronar. Some other modern watches are also pow-
ered using PV cells hidden behind a semitransparent dial. The simplest PV-powered calculators
consume nowadays only few microwatts like a watch; therefore, at daytime they do not use a
power from the backup primary battery.
Thermoelectric converter of the human body heat into electricity was the latest in the market
in the row of energy harvesters in handheld/wearable devices. The first device of this type, again
it was a watch 共Bulova Thermatron兲, has been marketed in 1980. However, the market success to
thermoelectrically powered watches came later, in 1990s, after perfecting the electronics of
watches.
The modern self-powered watches consume a power in between several and 1 ␮W, a very
low power. However, if we look back into the history, we can find out that even the first self-
winding watches consumed extremely low power, too. Indeed, walking with such a watch in the
pocket for just a few tens of minutes completely wound the mainspring. The same is related to the
“Atmos” clock. Before getting an idea of making such self-winding clock, J.-L. Reutter designed
the so-called 400-day Anniversary clock, wherein the power consumption has been reduced to
062701-3 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

extremely low level. It was so low that winding its spring once per 400 days was enough for
timekeeping. Only upon such dramatic reduction in power consumption it turned out possible for
him to design energy harvesting Atmos that consumes about 0.25 ␮W on estimate, i.e., about four
times less than the most economic modern electronic watches. Thus, the history gives actually a
very fruitful idea that before designing the devices based on harvesting ambient energy, 共i兲 the
research must be performed on the energy available in the ambient for a particular device and 共ii兲
the power consumption of the device itself must be minimized.

III. THEORETICAL BASES OF BODY-POWERED ELECTRONICS

A human body constantly generates heat as a useful side effect of metabolism. However, only
a part of this heat is dissipated into the ambient as a heat flow and infrared radiation, the rest of it
is rejected in a form of water vapor. Furthermore, only a small fraction of the heat flow can be
used in a wearer’s friendly and unobtrusive energy scavenger. 共For example, nobody would accept
a large device on his/her face. Therefore, the heat flow from it practically cannot be used.兲 At last,
due to the laws of thermodynamics, the heat flow cannot be effectively converted it into electricity.
However, a human being generates more than 100 W of heat; therefore, a quite useful electrical
power still can be obtained using a person as a heat generator. The tool for converting heat flow
into electricity is a thermoelectric generator 共TEG兲, the heart of which is a thermopile. Typically,
only a few watts of heat flow can be harvested unobtrusively on a person and thermoelectrically
converted into several milliwatts in a form of electricity. If we recall that watches consume 1000
times less, it is fairly good power. Moreover, PV cells of the same area typically generate much
less power because not much light is available indoors, where the authors and the reader of this
paper are resting now.
The human body is not a perfect heat generator as a heat supply for a wearable TEG. The
body has high thermal resistance; therefore, the heat flow is quite limited. This is explained by the
fact that warm-blooded animals have received in the process of evolution a very effective thermal
management. This includes a very high thermal resistance of the body at ambient temperatures
below 20– 25 ° C if the skin temperature decreases below thermal comfort.1,2 As a result, not much
heat is dissipated from the skin and only about 3 – 5 mW/ cm2 is available indoors, on average.
The measured performance characteristics of wearable TEGs 共Ref. 3兲 coincide with their
theoretical analysis.4,5 The practically confirmed conclusion is: a correctly designed unobtrusive
TEG in the right location on the human body can produce approximately 10– 30 ␮W / cm2 of
electrical power in moderate climate, on 24 h average. The produced power depends on the
thickness of a TEG and its size: the thicker the TEG, the better its power generation, while the
larger the TEG, the less power per unit area is produced.
At the system level, a short-term or long-term power reserve must be provided in the form of
rechargeable battery or a capacitor to avoid power shortages. Using such energy storage element,
the power gained by a TEG on occasional basis can be uniformly redistributed and consumed at
near-constant rate over a long period of time.
There are specific requirements to both the thermopile and TEG in energy scavengers includ-
ing wearable devices. First, the thermal resistance of a thermopile Rtp for maximum power gen-
eration must be equal to

R ppRTEG
Rtp = , 共1兲
R pp − RTEG

where R pp is the thermal resistance of air and holding elements interconnecting the cold and hot
plates of a TEG 共i.e., the elements connected thermally in parallel to the thermopile兲 and RTEG is
the optimal thermal resistance of a TEG, which maximizes power generation. The optimal thermal
resistance can be obtained from the equation of its thermal matching with the ambient4
062701-4 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 1. Body-powered wireless pulse oximeter.

共Rbody + Rsi兲Ret
RTEG = , 共2兲
2共Rbody + Rsi兲 + Ret
where Rbody is the thermal resistance of human body in location of a TEG 共this is the thermal
resistance between the body core and the chosen location on the skin兲, Rsi is the thermal resistance
of a heat sink, i.e., the thermal resistance due to convection and radiation on the outer side of TEG,
and Ret is the thermal resistance of TEG which would occur if the TEG becomes empty, with no
thermoelectric material in it. Equation 共2兲 is a thermal equivalent of electrical matching of a
generator with its load. The last requirement is that the ratio N,

N = Ret/共Rbody + Rsi兲, 共3兲

must preferably be more than 1. This ratio depends on the size and thickness of a TEG. Therefore,
the thinner TEG, the less power it regrettably produces.
Equations 共1兲–共3兲 call for a semiempty TEG, where the thermopile occupies only a small part
of the device volume; the rest must be occupied by air or a material with thermal conductivity less
than the thermal conductivity of air. This is a nice requirement because wearing something
“empty” is more pleasant than wearing heavy batteries.

IV. WEARABLE DEVICES POWERED BY THE USER’S BODY

As the first practical demonstration of body-powered medical devices, a wireless pulse oxime-
ter 共SpO2 sensor兲 has been designed, fabricated, and tested on people in 2006 共Fig. 1兲.6,7 The
device noninvasively measures oxygen content in arterial blood. A watch-size TEG is used in this
device as a power supply with a minimal power production of about 100 ␮W 共at night兲 and
typical variations during the day within the 100– 600 ␮W range. A goldlike coating of the radiator
provides nice appearance of the device. The coating has a high reflection coefficient in the visible
region of the spectrum 共to minimize heating by sunlight兲 and high emission coefficient in the
long-wave infrared region 共to enhance radiation losses from the TEG兲.
Typically, battery-powered pulse oximeters existing in the market have a power consumption
above 10 mW. Therefore, before making it powered from the human body, a power consumption
of electronics has been reduced by a factor of about 103. In addition, both a low-power radio used
and applied duty cycling enabled maintaining the overall power consumption at a record level of
62 ␮W 共Fig. 2兲. This battery-free device has a small supercapacitor 共22 mF兲 as a charge storage
element for buffering the radio transmission bursts. It is operational up to temperatures of
25– 26 ° C 共at daytime, and at a duty cycle of 26%兲. At higher ambient temperatures, the duty
062701-5 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 2. Power consumption pie of pulse oximeter at an update rate of 15 s.

cycle must be reprogramed 共decreased兲, otherwise the power generated could be less than the
power consumed. If this happens, the device switches into a sleeping regime for a while. During
sleep, its power consumption is extremely low and the device wakes up again upon collecting
enough charge in the supercapacitor, i.e., when the voltage on it exceeds 1.2 V. The disadvantage
of energy storage in a supercapacitor is that only about half of the energy produced over 24 h is
used for the application. 共When the storage supercapacitor is fully charged, there is no power
transfer into it from the TEG unless the next measurement of the oxygen content in a blood is
performed.兲 The performance characteristics of the pulse oximeter are listed in Table I.
Another example of battery-free body-powered devices is an electroencephalography 共EEG兲
system fabricated in 2007.8,9 The main challenges in creation of such complex system powered by
the wearer’s heat are 共i兲 lowering power consumption of biopotential readout while maintaining
the signal quality and 共ii兲 real-time data transmission. In the above example of pulse oximeter, the
signal processing is performed onboard so that power consumed by the transceiver is minimal
共Fig. 2兲. In EEG, on the contrary, real-time brain waves should be transmitted, so the wireless link
consumes large power.
Following historical examples of successful energy harvesters, this work has started with
effective reduction in power consumption of the EEG application. A low-power biopotential
readout application-specific integrated circuit 共ASIC兲 has been designed first, which provided
excellent signal quality despite very low-power consumption.10 Then, the battery-powered wire-
less EEG has been designed and its performance has been compared to the commercially available
wired EEG system in a hospital 共in application to polysomnography兲.11 Only after that, the system
has been turned into the self-powered one.
The set of experiments has shown that the forehead is a convenient location of a TEG on the
head. On one hand, the latter provides high heat flow on a large area. On the other hand, it is free
from thermally isolating hair. The power consumption of the EEG circuit 共see Table II兲 required a
large-sized TEG, i.e., larger than 50 cm2. To fit it to the patient’s head shape, it is designed as 10
sections with size of 1.6⫻ 4 cm2 on a stretchable headband 共Fig. 3兲. The radiators on the outer
side of thermopiles ensure effective heat dissipation into the ambient air and the thermal matching
of TEG with the environment. The TEG is designed for indoor use at typical ambient temperatures

TABLE I. Pulse oximeter features at 22 ° C, typically.

Power consumption 共␮W兲 62

Daytime power generation capability 共␮W兲 200
Output voltage, stabilized 共V兲 2
Autostart, within the 共min兲 15
Measured body parameters Heart pulse rate, oxygencontent in arterial blood
Generated voltage and power, voltage on a
Monitored system parameters supercapacitor
Carrier frequency 共GHz兲 2.4
Reprogrammable duty cycle; typically 共%兲 26
062701-6 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

TABLE II. EEG headband features.

Power consumption 共mW兲 0.8

Power generation capability at 22 ° C 共mW兲 2.5
Output voltage, stabilized 共V兲 2 and 2.8
Autostart at 22 ° C 共s兲 30–50
Real-time monitoring of brain waves 2 channels
Input-referred noise 共nV/ Hz1/2兲 ⬍60
Bandwidth and gain Selectable
CMRR 共dB兲 ⬎110
Carrier frequency 共GHz兲 2.4
Ambient temperatures; no headgear 共°C兲 21–26

maintained in hospital wards. At 22 ° C, it produces about 30 ␮W / cm2, i.e., close to the theoret-
ical limit of power generation on people at this temperature in a compact device. There is,
however, a drawback of such high power generation: at lower ambient temperatures, the heat flow
rapidly exceeds the sensation of discomfort and the device turns into uncomfortably cold object.
For example, at 19 ° C, the TEG already produces 3.7 mW, but the sensation of cold becomes too
annoying. This problem is solved in exactly the same way as someone solves it on the body level
in cold weather: a headgear should be worn on top of the system to limit the heat flow and make
it comfortable.
To improve EEG system described above, the TEG has been hybridized with PV cells in 2008.
Indeed, despite the fact that PV cells do not produce the same power as a TEG of the same area,
the outer surface of the latter is exposed to ambient light. Therefore, if the PV cells cover the TEG,
the joint power generation improves with no increase in the system weight and size. In addition,
this improved version of the device has been relocated from the forehead to the temples. For
thermal shunting the hair, its surface touching the skin has been provided with a set of fins. The
fins “comb” the person’s hair, while the device is being put on. As a result, a good thermal contact
with the skin is provided and the power generation becomes not worse than on the forehead. The
PV cells provide additional power; therefore, the power generated by the TEG can be proportion-
ally decreased. Therefore, smaller size of the TEG is required or its thickness can be decreased.
The EEG system with such hybrid power supply, in principle, has to be supplied with a recharge-
able battery because power generated by PV cells is much more unstable than the one generated
by the TEG. However, in this demonstrator used in the office, it was left with a supercapacitor to
show its self-start from the body heat. The headphonelike EEG diadem 共Fig. 4兲 is more comfort-
able, according to the user responses, in particular, at low ambient temperatures.9
The PV cells used in this device have been fabricated on a p-type float zone Si substrate. In
comparison to standard industrial process for monocrystalline PV cells, their surface has been
modified using random micropyramid formation. Together with the lithographically defined front-
side finger pattern, an efficiency of 17% has been obtained. The PV cells are interconnected in the

FIG. 3. Body-powered ECG headband: 共1兲 is a TEG and 共2兲 is an electronic module.
062701-7 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 4. Wireless EEG system with hybrid thermoelectric-PV power supply.

device in a chessboard way in two parallel circuits. This allowed obtaining an almost perfect
dependence of power versus direction of light: the power output varies only within 5%, while the
wearer makes a complete turn around him/herself in direct sunlight. The system has been tested at
ambient temperatures down to 7 ° C; however, no discomfort has been mentioned by its wearers.
Some features of this system are listed in Table III.
The power consumption of the EEG system is illustrated in Fig. 5. About half of the power is
consumed by the transceiver; therefore, more power-efficient radio would allow either shrinking
the generator size or powering more channels. However, the microcontroller and biopotential
readout consume the other half of the power. Taking this into account, the biopotential ASIC has
been redesigned and its latest eight-channel version12 consumes the same power as the two-
channel ASIC used in the battery-free EEG systems discussed above.

V. BODY-POWERED SYSTEMS IN CLOTHING

The battery-free systems discussed above have been designed to work at power consumption
near the minimum of expected power generation 共to avoid power shortages兲. Indeed, when com-
paring row 1 共the power consumption兲 with row 2 共power generation capability兲 in any table

TABLE III. Features of the battery-free electroencephalorgaphy diadem with hybrid power supply at
daytime, typically.

Power consumption 共mW兲 0.8

Power generation capability at 22 ° C 共mW兲 1.5
Power generation test 共mW兲
Thermoelectric generator 共no wind兲
at 9.5 ° C 5.5
at 21.8 ° C 1.5
Photovoltaic cells 共Belguim, March兲
Outdoors, sunny, at noon 45
In the office, cloudy 0.2
Autostart at 22 ° C 共s兲 10–30
Real-time monitoring of brain waves 2 channels
Ambient temperatures; tested 共°C兲 7–28
062701-8 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 5. Power consumption pie of self-powered battery-free EEG systems.

above, the difference by a coefficient of 2–3 is always the case. The excess energy, which might
be transferred into the charge storage supercapacitor most of the time, was not actually transferred
because the latter was already fully charged. Therefore, at least 50%–70% of energy was wasted.
If, however, a secondary battery is used as an energy storage element, it can provide the required
power for days or weeks even with no harvested energy.1 Therefore, using a battery enables
operating the device at the power produced, on average, so that much less power is required from
the TEG, on average. As a result, the TEG can be effectively miniaturized to be more comfortable
for its user.
To make a compact wearable self-powered device, the energy harvester is typically placed
close to the location on the body, wherein the body parameters are being measured. Therefore, it
seems logical that an electrocardiography 共ECG兲 system should be preferably integrated some-
where on the chest. That is why an ECG system described below is integrated into a shirt.
The ECG system in a shirt 共Fig. 6兲 is powered by a secondary battery. The power consumption
is 0.5 mW, and therefore if the system would have a supercapacitor instead of a battery, a power
of about 2 mW or much more would be required from the TEG at 22 ° C. 共This is to be sure that
if the wearer puts a business suit jacket on, the system still works.兲 With the battery, about 0.6 mW
from the TEG is enough. The battery is constantly recharged using the wearer’s body heat.
Fourteen 6.5-mm-thin thermoelectric modules with size of 3 ⫻ 4 cm2 have been integrated into
the front side of the shirt 共it is less than 1.5% of the shirt area兲. The radiators of TEG modules
have been painted like chameleon into the shirt colors, except one module that is to give impres-
sion about the module size. This actually illustrates two basic ways of module integration: one
way is to make them invisible and the other way is to use radiators as decoration elements. The
second way is even more interesting for the market, especially concerning the products for young
generation. The wiring and the other modules of ECG system are located on the inner side of the
shirt. PV cells are located on the shoulders to get maximum power both indoors and outdoors.

FIG. 6. Wireless ECG shirt powered by a hybrid thermoelectric-PV power supply: three thermoelectric modules, one of
which intentionally colored in pink 共1兲, and amorphous PV cells 共2兲.
062701-9 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 7. Electronic module of the EEG shirt upon its waterproof encapsulation.

In the office, the TEG typically generates the power within the 0.8–1 mW range at about 1 V
on the matched load. However, if the wearer walks indoors, the power production rapidly in-
creases to 2–3 mW. Because of high thermal resistance of TEG modules, they are never cold. 共In
this sense, the modules function in the same way as a textile.兲 In cold weather, the other pieces of
clothing are worn on top of the shirts. However, as measured at about 10 ° C outdoors while
wearing a thick jacket, the power generation is by 10%–20% better than indoors.
The power management module contains a fully integrated dc/dc upconverter fabricated in a
0.35 ␮m process.13 It works with an open-circuit input voltage of 0.5 V or higher from a TEG,
converts it up and charges a small 2.4 V NiMH battery. The converter ASIC provides up to 0.33
mW; therefore, three converters have been connected in parallel to enable input power up to 2 mW
to be efficiently transferred into the battery. The converter contains a charge pump with variable
number of stages. A control algorithm can be used to vary the number of stages and switching rate,
and therefore to operate always with near-maximum efficiency. The net power conversion effi-
ciency is 44% at an open-circuit TEG voltage of 2 V. It increases while the voltage decreases and
exceeds 70% at the voltage less than 0.9 V.
In parallel to the TEG power conversion circuit, there is a secondary parallel circuit that
allows charging the battery directly from PV cells. Two low-cost amorphous silicon solar cells,
with size of 2.5⫻ 4 cm2 each, are integrated into the shirt on its shoulders 共Fig. 6兲. PV cells are
added to the system because if the shirt is not worn for months, the battery can be discharged. 共In
such case, it can be charged again by placing the shirt in direct sunlight for a few days.兲 Therefore,
when the shirt is taken off and not used for a long time, it must be stored in an environment where
light is available periodically, e.g., in a wardrobe with windows. The small power provided by PV
cells is sufficient to compensate for the self-discharge of the battery that is by a factor of about 20
exceeds the standby power. In this way, even after months of nonuse, the electronics is maintained
in the ready-to-start state, waiting for the moment the shirt is used again. The open-circuit output
voltage of PV cells 共5.2 V in the office兲 is high enough; therefore, they are connected directly to
the battery through a simple rectifier circuit. The use of PV cells has other advantages, too. First,
it provides some additional power 共e.g., at least 100– 150 ␮W in the office兲, and, second, in
tropical countries, PV cells compensate for a lack of power generated by the TEG.
The necessity of laundry demands reliable waterproof encapsulation of the system compo-
nents: a TEG, PV cells, and electronics. The latter is integrated in a two-sided flex circuit and
encapsulated using a silicone molding 共Fig. 7兲. Smart power management is one of the key
062701-10 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

TABLE IV. Some features of the ECG shirt with hybrid power supply 共preliminary data兲.

Power generation capability at 22 ° C 共mW兲 1

Charge pump output range 共mW兲 0.03–1
Charge pump efficiency at 5 stages 共%兲 67⫾ 4
Power consumption in active regime 共mW兲 0.6
Power consumption in standby regime 共␮W兲 1
Autostart 共s兲 A few
Washing machine 共 at 1000 rpm兲 Passed
Iron pressing 共max. temp. for cotton兲 Passed

elements of self-powered systems. As discussed above, the system uses two complementary power
sources and its charge pump can track the optimal power conversion by changing the number of
its stages and switching frequency depending on input power. In addition, the microcontroller
keeps track of the input power from the TEG and of the available energy stored in the battery. If
these drop below a safe threshold, the system performs an autocompensation by temporarily
reducing the sampling rate and transmission duty cycle. This allows nonstop wireless acquisition
of an ECG signal on cost of its quality even under severe energy deficit, where the full-quality
signal acquisition 共that requires 0.5 mW兲 cannot be provided. If the voltage from a TEG drops to
near zero, which happens when the shirt is taken off, the system switches into a standby state with
low-power consumption 共Table IV兲.
Because of extreme fragility of thermopiles, the shock protection components enforce ther-
moelectric modules. As a result, its laundry with an acceleration of 250g is allowed. Despite using
a noniron shirt for the ECG system, its pressing can be performed safely because wiring sustains
all possible regimes of pressing.
The flex with electronics is divided into quasirigid islands with dense components. The islands
are interconnected through highly flexible zones. This gives a limited amount of flexibility to the
circuit and makes it compatible with both integration in clothes and laundry.
The production cost of TEGs still exceeds the cost of the batteries. This is because a very
small temperature drop can be obtained in wearable thermopiles, i.e., one to a few °C. As a result,
a large number of thermocouples, a few thousand, are required, thereby adversely affecting the
system cost. However, microelectronic technologies may help alleviate this problem.

VI. MODERN TECHNOLOGIES FOR WEARABLE THERMOPILES

It is quite interesting that the theory 共Sec. III兲 does not demand large-size thermopiles for the
maximum power generation in a wearable TEG. The only requirement for a thermopile is that it
must have high thermal resistance per cm2 of the harvester on the skin. If a small-size thermopile,
e.g., a micromachined one, replaces in such TEG, e.g., in a watch-size TEG, the thermopiles
purchased in the market, it can produce the same power. This does not mean that any small-size
thermopile is good for the target application. Still both the electrical contact resistance and thermal
conductance parallel to the thermopile in a TEG must be minimized because these are parasitic
factors that adversely affect its performance characteristics. In fact, both factors can render the
device worthless.
The modeling of micromachined thermopiles in a wearable TEG shows that due to scaling
laws, the smaller thermopile, the lower aspect ratio is required 共Fig. 8兲. 共An aspect ratio is the ratio
of the height of thermocouple legs to their lateral dimension.兲 Commercially available thermopiles
require high aspect ratio of about 30; therefore, the only feasible solution is to make multistage
thermopiles. Scaling the thermopile down results in proportional decrease in the aspect ratio that
is required for getting the same thermal resistance. As a result, at lateral dimensions below
10 ␮m, the optimal aspect ratio could be decreased to the values acceptable in microelectronics
062701-11 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 8. Scaling of a thermopile at its constant thermal resistance. While thermopiles on the market require high aspect ratio
l / w 共left兲, their miniaturized microelectronic counterparts need much lower aspect ratio 共right兲.

technologies.14 At larger lateral dimensions, i.e., several tens of micrometers, thick film and
microelectromechanical system 共MEMS兲 technologies could be used instead. However, the
smaller thermopile, the better is the chance of its market success.
As a proof of the theoretical concept, a micromachined thermopile with 0.5 ␮m tall poly-
SiGe thermocouples 关see Fig. 9共a兲兴 has been fabricated on a 4 ⫻ 4 mm2 die.15 Based on the
modeling results, in such a thermopile covered by a heat sink die 共Fig. 8, right兲, the air between
the dies thermally shunts the thermopile, thereby decreasing the useful temperature drop on it to a
few tens of millikelvins. Therefore, the thermopile 关Figs. 9共b兲 and 9共c兲兴 has been fabricated on top
of a 250 ␮m tall rim etched in a silicon wafer 关Fig. 9共d兲兴. This has been done for decoupling the
thermocouple height and parasitic thermal conductance through the air inside the TEG.16 In the
two versions of the thermopile, either 2350 or 4700 thermocouples have been fabricated close to
each other on top of the rim so that one thermocouple occupies an area of about 30⫻ 15 ␮m2.
After the ion etch of the substrate, the mean distance between the thermopile die and a bare silicon

FIG. 9. Micromachined polycrystalline SiGe thermopile: 共a兲 Design of a thermocouple with a 2.5 ␮m deep trench, 共a兲 and
共b兲 are hot and cold thermocouple junctions, respectively; 共b兲 SEM picture of three thermocouple bridges over a trench; 共c兲
a corner of the rim filled with thermocouples; 共d兲 the design of a thermopile on a silicon rim 共not in scale兲; 共e兲 the design
of a thermoelectric generator, wherein the thermopile die is flip-chip bonded to the heat sink 共top兲 die; and 关共f兲 and 共g兲兴 two
wrist TEGs with a micromachined thermopile inside.
062701-12 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

FIG. 10. Design of arcade thermocouples 共left: three thermocouples are shown兲 共modify from Ref. 17兲, and a SEM picture
of self-supported 6 ␮m tall released thermocouples with a critical dimension of 3 ␮m reported in Ref. 19.

die attached to its top as a heat sink die 关Fig. 9共e兲兴 increases by a factor of a few hundred. This
allows reaching N ⬎ 1 关Eq. 共3兲兴. An additional decrease in the parasitic thermal conductance
through the air between the hot plate and the radiator 关Fig. 9共e兲兴 has been reached by increasing
the distance between them and by using the thermally conducting aluminum spacer to interconnect
the top die with the radiator. As a result, despite the too low topography used in the device, a
temperature drop of 0.6– 1 ° C is obtained on the thermopile in a wrist device 关Figs. 9共f兲 and 9共g兲兴.
The temperature drop depends on the person’s activity and increases under the forced air convec-
tion, i.e., on a walking person. Two versions of a MEMS energy harvester have been fabricated. In
the first one 关Fig. 9共f兲兴, the radiator is supported by thermally isolating pillars. In the second one
关Fig. 9共g兲兴, the thermal resistance of the ambient air is further decreased by using pin-featured
radiator. An additional protection grid surrounds it and serves as a touch protection as well as an
outer shock protection. Two steps are still required to make such technology mature and competi-
tive to batteries in wearable devices: 共i兲 increase the aspect ratio and height of microthermocouples
and 共ii兲 develop a thin-film BiTe technology with better thermoelectric quality of the materials
than can be reached with poly-SiGe.
The required aspect ratio can be reached using projection lithography because a critical di-
mension of 1 – 3 ␮m is sufficient.17,18 The minimum target within the required 6 – 12 ␮m height
of thermocouples has been already reached 关Fig. 10兴,19 and the research is ongoing toward prac-
tical demonstration of poly-SiGe thermopiles with high aspect ratio. The second step, i.e., the film
technology for BiTe materials, is also in progress20–22 so that in the near future their thermoelectric
properties are expected to approach those of bulk materials.
The production cost of micromachined thermopiles can be by a factor of 100–1000 less than
the cost of today’s thermopiles in the market because only 1 – 2 mm2 of the wafer is required for
devices such as those shown in Fig. 9. Therefore, in mass production, wearable thermopiles can be
very competitive on cost with the batteries. The other technologies such as thick-film ones23,24
could also have a market success if a low-cost process is developed.

VII. CONCLUSION
The human body produces heat equivalent to the heat dissipated by a few laptops. Several tens
of watts out of the produced heat are dissipated as a heat flows from the skin, depending on
ambient conditions 共the temperature, humidity, and wind兲. Due to laws of thermodynamics, this
heat cannot be effectively converted into electricity. Within the 22– 30 ° C ambient temperature
range, the theoretical thermoelectric efficiency of converting human body heat into electricity
varies from 0.4% to 0.2%, on average. In a thin device, i.e., not thicker than a watch, the
obtainable power varies within 30− 7 ␮W / cm2 of the person’s skin, respectively. Taking into
account that only a few percent of the body area unobtrusively can be covered by thermoelectric
modules, the only few milliwatts of electrical power can be obtained on a person. The extensively
measured performance of wearable thermoelectric generators confirms the calculation result that,
in such applications, they outperform solar cells, on a typical person. Therefore, neither a mobile
062701-13 Wearable self-powered electronics J. Renewable Sustainable Energy 1, 062701 共2009兲

phone or a geographical positioning system device nor a personal digital assistant can be self-
powered. 共Only several minutes per day can be provided, which does not satisfy the market
needs.兲
Wearable sensors and medical devices, including wireless ones, rapidly evolve to very power-
efficient devices while maintaining high signal quality. This enables powering them using body
heat and ambient light. The market success to such self-powered systems can come only if the
production cost of thermopiles is dramatically decreased. Several parallel developments of micro-
electronic and MEMS technologies for a new film-based generation of thermopiles is ongoing, but
still these are not mature. Some additional information for the reader is provided in a power point
presentation on wearable electronics self-powered from the human body, its present state and
perspectives.25
The power-efficient self-powered smart electronics, in particular, wearable electronics, is
knocking on the doors of the market. However, the battery in such devices cannot be simply
replaced with an energy harvester. It requires system redesign, smart power management and,
frequently, a very-low-power radio. In clothes, the necessity of laundry and pressing gives a good
chance to self-powered systems to compete successively with batteries. Environmental aspects are
also in favor of energy harvesters that offer much greener garments for their postuse utilization.

ACKNOWLEDGMENTS
The authors gratefully acknowledge the designers of low-power electronic circuits and mod-
ules, T. Torfs, R. F. Yazicioglu, and I. Doms as well as the whole enthusiastic research team of
IMEC and Holst Centre working on energy harvesters, photovoltaic cells, and wireless systems.
Special thanks are due to C. Chan 共SEMICON Singapore兲 for the invitation to present our research
共Ref. 25兲, which has made it easier to summarize the work in this paper.
1
V. Leonov and R. Vullers, Proceedings of the Fifth European Conference on Thermoelectrics 共ECT ‘07兲, Odessa,
Ukraine, 10–12 September 2007, pp. 47–52.
2
Heat Loss from Animals and Man, edited by J. Monteith and L. Mount 共Butterworths, London, 1974兲.
3
V. Leonov, B. Gyselinckx, C. Van Hoof, T. Torfs, R. Yazicioglu, R. Vullers, and P. Fiorini, in Proceedings of the Second
European Conference on Smart Systems Integration (SSI ’08), Barcelona, Spain, 9 and 10 April 2008, edited by T.
Gessner 共VDE, Berlin, 2008兲, pp. 217–224.
4
V. Leonov and P. Fiorini, Proceedings of the Fifth European Conference on Thermoelectrics 共ICT ’07兲, Odessa, Ukraine,
10–12 September 2007, pp. 129–133.
5
V. Leonov and R. J. M. Vullers, J. Electron. Mater. 38, 1491 共2009兲.
6
T. Torfs, V. Leonov, and R. J. M. Vullers, Sensors & Transducers J. 80, 1230 共2007兲.
7
T. Torfs, V. Leonov, C. Van Hoof, and B. Gyselinckx, Proceedings of the Fifth IEEE International Conference on
Sensors, Daegu, Korea, 22–25 October 2006, pp. 427–430.
8
T. Torfs, V. Leonov, R. F. Yazicioglu, R. J. M. Vullers, P. Fiorini, B. Gyselinckx, and C. Van Hoof, Proceedings of the
Seventh IEEE International Conference on Sensors, Lecce, Italy, 26–29 October 2008, pp. 1269–1272.
9
M. Van Bavel, V. Leonov, R. F. Yazicioglu, T. Torfs, C. Van Hoof, N. E. Posthuma, and R. J. M. Vullers, Sensors &
Transducers J. 94, 103 共2008兲.
10
R. F. Yazicioglu, P. Merken, B. Puers, and C. Van Hoof, IEEE J. Solid-State Circuits 42, 1100 共2007兲.
11
N. de Vicq, F. Robert, J. Penders, B. Gyselinckx, and T. Torfs, Proceedings of IEEE Biomedical Circuits and Systems
Conference 共BIOCAS ’07兲, Montreal, Canada, 27–30 November 2007, pp. 163–166.
12
R. F. Yazicioglu, P. Merken, R. Puers, and C. Van Hoof, IEEE J. Solid-State Circuits 43, 3025 共2008兲.
13
I. Doms, P. Merken, R. Mertens, and C. Van Hoof, Digest of Technical Papers IEEE International Solid-State Circuits
Conference 共ISSCC兲, San Francisco, CA, 7–11 February 2009, pp. 300–301a.
14
V. Leonov, J. Electron. Mater. 38, 1483 共2009兲.
15
V. Leonov, Z. Wang, P. Fiorini, and C. Van Hoof, Sensors and Transducers J. 103, 29 共2009兲.
16
V. Z. Wang, V. Leonov, P. Fiorini, and C. Van Hoof, “Realization of wearable miniaturized thermoelectric generator for
human body applications,” Sens. Actuators, A 共in press兲.
17
V. Leonov, Z. Wang, R. Pellens, C. Gui, R. Vullers, and J. Su, Proceedings of the Fifth International Energy Conversion
Engineering Conference 共IECEC兲, St. Louis, MO, 25–27 June 2007, AIAA Paper No. 2007-4782-754.
18
J. Su, R. Vullers, M. Goedbloed, Y. van Andel, R. Pellens, C. Gui, V. Leonov, and Z. Wang, Proceedings of the Seventh
Power MEMS Workshop, Freiburg, Germany, 28 and 29 November 2007, pp. 153–156.
19
J. Su, M. Goedbloed, R. J. M. Vullers, Y. van Andel, V. Leonov, and Z. Wang, Proceedings of the 35th International
Conference on Micro & Nano Engineering, Ghent, Belgium, September 28–October 1 2009 关“Thermoelectric energy
harvester fabricated by stepper,” special issue of Microelectron. Eng. 共in press兲兴.
20
H. Böttner, Proceedings of the 21st International Conference on Thermoelectrics 共ICT ’02兲, Long Beach, CA, 25–29
August 2002, pp. 511–518.
21
G. Snyder, J. Lim, C.-K. Huang, and J.-P. Fleurial, Nature Mater. 2, 528 共2003兲.
22
S. Sedky, A. Kamal, M. Yomn, H. Bakr, R. Ghannam, V. Leonov, and P. Fiorini, Proceedings of the 15th IEEE
 062701-14 V. Leonov and R. J. Vullers J. Renewable Sustainable Energy 1, 062701 共2009兲

International Conference on Solid-State Sensors, Actuators and Microsystems 共Transducers ’09兲, Denver, CO, 21–25
June 2009, pp. 1035–1038.
23
I. Stark, Proceedings of the Third IEEE International Workshop on Wearable and Implantable Body Sensor Networks
共BSN ’06兲, Boston, MA, 3–5 April 2006.
24
V. Leonov, R. Vullers, M. Goedbloed, and Y. van Andel, Proceedings of the Sixth European Conference on Thermo-
electrics 共ECT ’08兲, Paris, 2–4 July 2008, pp. O17-1–4.
25
V. Leonov and R. J. M. Vullers, SEMICON Singapore, 20–22 May 2009, “Wearable electronics
self-powered from the human body: State of the art and perspectives,” [Link]
ProgrammesandEvents/cms/groups/public/documents/web_content/ctr_029926.pdf.

View publication stats