538

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008

Pyroelectric Energy Conversion: Optimization Principles

Gael Sebald, Elie Lefeuvre, and Daniel Guyomar
AbstractIn the framework of microgenerators, we present in this paper the key points for energy harvesting from temperature using ferroelectric materials. Thermoelectric devices prot from temperature spatial gradients, whereas ferroelectric materials require temporal uctuation of temperature, thus leading to dierent applications targets. Ferroelectric materials may harvest perfectly the available thermal energy whatever the materials properties (limited by Carnot conversion eciency) whereas thermoelectric materials eciency is limited by materials properties (ZT gure of merit). However, it is shown that the necessary electric elds for Carnot cycles are far beyond the breakdown limit of bulk ferroelectric materials. Thin lms may be an excellent solution for rising up to ultra-high electric elds and outstanding eciency. Dierent thermodynamic cycles are presented in the paper: principles, advantages, and drawbacks. Using the Carnot cycle, the harvested energy would be independent of materials properties. However, using more realistic cycles, the energy conversion eectiveness remains dependent on the materials properties as discussed in the paper. A particular coupling factor is dened to quantify and check the eectiveness of pyroelectric energy harvesting. It is dened similarly to an electromechanical coupling factor as k2 = p2 0 =("33 cE ), where p, 0 , "33 , cE are pyroelectric coecient, maximum working temperature, dielectric permittivity, and specic heat, respectively. The importance of the electrothermal coupling factor is shown and discussed as an energy harvesting gure of merit. It gives the eectiveness of all techniques of energy harvesting (except the Carnot cycle). It is nally shown that we could reach very high eciency using h111i0.75Pb(Mg1=3Nb2=3 )-0.25PbTiO3 single crystals and synchronized switch harvesting on inductor (almost 50% of Carnot eciency). Finally, practical implementation key points of pyroelectric energy harvesting are presented showing that the dierent thermodynamic cycles are feasible and potentially eective, even compared to thermoelectric devices.

I. Introduction onstant advances in electronics push past boundaries of integration and functional density toward completely autonomous microchips embedding their own energy source. In this eld, research continues to develop higher energy-density batteries, but the amount of energy available is nite and remains relatively weak, limiting the

Manuscript received June 6, 2007; accepted November 26, 2007. This work was supported by the Agence Nationale pour la Recherche from the French government, under grant #ANR-06-JCJC-0137. The authors are with INSA-Lyon, Laboratoire de G enie Electrique et de Ferro electricit e, Villeurbanne, France (e-mail: gael.sebald@insalyon.fr). Digital Object Identier 10.1109/TUFFC.2008.680

systems lifespan, which is paramount in portable electronics. Extended life is also particularly advantageous in systems with limited accessibility, such as biomedical implants, structure-embedded microsensors, or safety monitoring devices in harsh environments and contaminated areas. The ultimate long-lasting solution should therefore be independent of the limited energy available in batteries by recycling ambient energies and continually replenishing the energy consumed by the system. Some possible ambient energy sources are thermal energy, light energy, or mechanical energy. Harvesting energy from such renewable sources has stimulated important research eorts over the past years. Several devices from millimeter scale down to microscale have been presented, with average powers in the 10 W to 10 mW range [1]. Work on vibration-powered piezoelectric electrical generators has led to new energy conversion techniques, such as synchronized switching harvesting (SSH) techniques, based on nonlinear processing of the piezoelectric voltage [2][7]. As a result, the mechanical-to-electrical energy conversion capability of active materials is signicantly increased: typically by factors of 4 to 15, depending on the considered technique. From the eciency point of view it has been shown that SSH techniques may be implemented with electronic circuits consuming less than 5% of the energy produced by the piezoelectric element. This novel approach is very promising for improving the eectiveness and power density of piezoelectric microgenerators. But it can also be theoretically extended to most other energy conversion processes (for example, strain/stress variation, temperature variation, and other processes). Thermoelectric modules are the main way for energy harvesting from temperature. It is now possible to nd commercial thermoelectric generators from W to kW electric output energy. These are based on temperature gradients leading to heat ow through the thermoelectric generator, and a small percentage of the heat ow is converted to electric energy. Materials properties are the key parameter for improving both the output power (increase of the thermal heat ow, thus making it dicult to keep the temperature gradient) and the eciency (improving the Seebeck coecient and gure of merit). However, the possibility of harvesting thermal energy is limited in the case of microgenerators because the temperature dierentials across a chip are typically low. Pyroelectric materials may be used for thermal energy to electric energy conversion. The pyroelectric eect was discovered before the piezoelectric eect and is mainly used for pyroelectric infrared temperature detectors. Contrary

08853010/$25.00 c 2008 IEEE

sebald et al.: pyroelectric energy conversion: optimization principles

539 TABLE I Coefficients Used in the Simulations. Coecient 33 p cE h c Unit Cm2 K1 Jm3 K1 K K Fm1 Value 10000 103 2.5 106 301 300

to thermoelectric generators, pyroelectric materials do not need a temperature gradient (spatial gradient), but temporal temperature changes. This opens dierent applications elds, where temperature gradients are not possible and where temperature is not static. Small-scale microgenerators with dimensions smaller than the temperature spatial uctuation length may with diculty be subjected to temperature gradients. Natural temperature time variations occur due to convection process, and this thermal energy is dicult to transform in a stable temperature gradient. On the other hand, it is possible to transform a temperature gradient into a temperature variable in time using a caloric uid pumping between hot and cold reservoirs. The pumping unit may require much less energy than the total produced energy (depending on the scale of the device) and may produce temperature variations of 1 to 20 C at 2 Hz for example. To optimize energy harvesting from temperature, the rst step should be the optimization of energy conversion. Then, the problems of electric loading (modifying the cycles shape) should be addressed. The aim of this paper is to present methods for optimizing energy conversion from temperature variations using pyroelectric materials and to describe the most important parameters in materials choice and device design. The rst part is devoted to thermodynamic cycles that could be used for energy conversion and the second part deals with a pyroelectric materials survey. Finally the practical application problems of thermodynamic cycles are discussed in the last part of the article.

For Carnot cycle, = 100 for the sake of clarity on the gure (to 0 33 obtain larger dierence between adiabatic and isothermal dielectric permittivity).

33 =

dD dE

, p=

dD d dU = , cE = d dE d

.
E

(3)

In the following part, we present four dierent energy harvesting cycles. For each cycle, we give PE cycle (polarization vs. electric eld) and cycle (entropy vs. temperature). In the two cycles, the area of the cycle is the converted energy. It is the same area in PE cycle and in cycle. In the PE cycle, the cycle is clockwise, meaning a negative energy (i.e., energy given to the outer medium). In the cycle, the cycle counter-clockwise, meaning a positive energy (i.e., energy given by the outer medium to the material). The coecients dened in (3) are assumed to be constant for the electric eld and temperature ranges considered here. Coecients used in simulations are given in Table I. A. Carnot Cycle

II. Thermodynamic Cycles When talking about energy harvesting from heat, one should rst consider classical thermodynamic cycles. We aim to answer here several questions: What cycles could be imagined to harvest energy from heat? What is their eciency (dened as electric work divided by the heat transferred from a hot reservoir to a cold reservoir)? What are the important parameters of the pyroelectric materials for optimizing the eciency? Are those cycles realistic?

The Carnot cycle is dened as two adiabatic and two isothermal curves on the (PE) cycle (see Fig. 1). It is considered as the optimal energy harvesting cycle whose eciency is Carnot = 1 c h (4)

where c and h are cold and hot temperatures, respectively. The demonstration of that result is very interesting to understand the underlying limitations of such cycle. In the rst adiabatic increase of the electric eld (path A-B) p d = dE cE p h = EM c cE (5) (6)

For a given temperature variation, it is possible to consider it as a static problem involving two temperature reservoirs, which is a common interpretation in thermodynamics. We need rst to establish the equations of pyroelectric materials [8]. dD = 33 dE + pd d d = pdE + cE (1) (2)

ln

where EM is the maximum amplitude of the applied electric eld. In practical applications, this means that one should know the maximum temperature variation to know what the necessary electric eld is. In the isothermal decrease of the electric eld (path B-C) d = pdE and dQ = dS Qh = pEM h (7)

where D, E , , and are electric displacement, electric eld, temperature, and entropy, respectively. The coecients are dened as:

540

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008

eld is limited to 2 K for bulk ceramics [12][18] and could reach 512 K for thin lms [19], [20]. A too-large temperature variation will result in a degraded Carnot cycle, because it is impossible to get enough electrocaloric eect. Moreover, it is hardly realistic to force the electric eld at a given value without paying a lot of wasted energy (see Section IV). As a consequence, the Carnot cycle is not feasible at all in practical applications. B. SECE Cycle SECE stands for synchronized electric charge extraction. We use that acronym because of numerous papers concerning the nonlinear switching of the piezoelectric voltage for energy harvesting from vibrations and damping [3], [6], [7]. From the thermodynamics point of view, this technique is one of the most natural cycles. It consists of extracting the electric charge stored on the active material when the maximum temperature is reached, i.e., when the stored electric energy is maximum, and doing it again when the temperature is minimum (Fig. 2). This energy extracted may be then transferred to an electrical energy storage cell such as a capacitor or to an electrochemical battery for future needs, using, for example, the circuit described in [3] or the power converter detailed in Section IV. The theoretical description of this cycle is as follows. Along the path (C-D), the temperature is decreased resulting in a decrease of the open-circuit electric eld. p (9) Em = (c h ) 33 where Em is the minimum electric eld on the sample. During that temperature variation
Fig. 1. Thermodynamic cycles for Carnot cycle. (a) PE cycle and (b) cycle.

dQ = cE d

p2 d 33 p2 2 2 (c h ) 2 33

(10) (11)

where Qh is the heat taken from the hot reservoir. The two following steps are very similar and are not detailed here. The resulting energy conversion ratio gives We = (h c ) = pEM h c . h (8)

Qc1 = cE (c h )

The resulting conversion ratio was already given in (4). It is very interesting to notice that this conversion ratio does not depend on material properties. The only restrictionand main drawbackis that one should know rst the temperature variation before starting any cycle. Furthermore, (6) links electric eld amplitude to the temperatures ratio. Using realistic coecients values (see, for example, [9][11], p = 600 106 Cm2 K1 , and cE = 2.5 106 Jm3 K1 for a 0.75Pb[Mg1/3 Nb2/3 ]O3 0.25PbTiO3 ceramic), and for a temperature dierence of 1 C around room temperature, we need an electric eld of 14 kVmm1 , which is far beyond the electric breakdown of bulk ceramics. The maximum temperature variation induced in ferroelectric materials when applying an electric

where Qc1 is the heat given to the cold source during the cooling. Then the electric eld is decreased to 0 in isothermal condition (by connecting the sample to a resistance for example, path D-E). Due to electrocaloric activity in ferroelectric materials, heat is transferred from the sample to the cold source Qc2 = pEm c . (12)

As a result, using (9), total heat Qc transferred to the cold source is Qc = cE (c h ) + p2 (h c )2 . 2 33 (13)

The two other segments of the cycle are very similar. Total heat transferred to the hot source is Qh = cE (h c ) + p2 (h c )2 . 2 33 (14)

sebald et al.: pyroelectric energy conversion: optimization principles

541

divided by the product of noncoupled ones). For weakly 1 (most common case as shown in coupled case, i.e., k 2 Section III) SECE = k 2 Carnot . (19)

For a perfect coupled material (k 2 = 1), we obtain a conversion ratio that tends to the Carnots one provided that this latter is much smaller than unity. The advantages of such energy harvesting technique are: No control of the voltage. No special attention to be paid to the temperature variation; do not need to know the temperature in advance. Possible whatever the material (only pyroelectric activity is important, whatever the electrocaloric activity).

The main drawback is the poor conversion ratio compared to Carnot cycles. In fact, the k 2 for common materials (PZT ceramics) is around 2 103 and may reach 4.7 102 for some single crystals (see Section III for details about materials). C. SSDI Cycle SSDI stands for synchronized switch damping on inductor. This technique was developed prior to the SSH techniques for dissipating the mechanical energy of vibrating structures with piezoelectric inserts to damp the structural resonance modes [2]. Synchronized switch means that the voltage of the ferroelectric material is switched on an inductor at every maximum or minimum of the temperature, so that the electric eld polarity is quasi-instantaneously reversed (Fig. 3). From a thermodynamics point of view, the only dierence with SECE is that the electric eld is not reduced to 0, but nearly to its opposite value. The use of resonant circuit including an inductor is in ne an ingenious way to perform that operation at minimized energy cost (due to the inductor imperfections, a small amount of energy is lost during the electric eld polarity reversal process). Let us start the cycle explanation from point A. The temperature is increased in open-circuit condition. Due to pyroelectric activity, a positive electric eld appears on the ferroelectric material. Reaching the maximum temperature, the electric eld is inversed from EM 0 to Em0 with a lossy inversion ratio Em0 = EM 0 (20)

Fig. 2. Thermodynamic cycles for SECE cycle. (a) PE cycle and (b) cycle.

The total electric work is found assuming that the internal energy does not change at the end of one cycle Qh + Qc = WE WE = p (h c )2 33
2

(15) (16)

where WE is the electric energy. Finally, the conversion ratio gives SECE = with k2 = p2 h . 33 cE (18) |WE | k2 = Carnot Qh 1 + 0.5k 2 Carnot (17)

Variable k 2 is a dimensionless number giving the electrothermal coupling factor (at temperature h ), similar to the electromechanical coupling factor (coupled coecient

where is the inversion quality. = 1 is a perfect inversion, and = 0 is the SECE case. Then the temperature is decreased to its minimum. The absolute value of the electric eld is increased, and then the inversion process is repeated. This cycle is repeated indenitely. The maximum value of the electric eld is thus increased for every cycle and would tend to an innite value for a perfect inversion process. It is assumed

542

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008

Using (20) and (21) Qh = cE (h c ) p2 2 2 (h c ) 2 33 p2 1 + + 33 1

2 (h h c ). (24)

Similarly for the heat transferred to the cold source (path C-D and D-E) Qc = cE (c h ) p2 2 2 (c h ) 2 33 p2 1 + + 33 1

2 h c ). (25) (c

Using (15), WE = p2 33 1+ 1 (h c )2 . (26)

Electric energy is very similar to the SECE example. The term (1 + )/(1 ) shows the energy conversion magnication using the SSDI energy conversion cycle. Finally, the conversion ratio is SSDI = 1 + k2 k2 2 Carnot + 1 2 1+ Carnot . 1 1), we obtain (28)

(27)

For weakly coupled materials, (k 2 SSDI = k 2

1+ Carnot . 1

Fig. 3. Thermodynamic cycles for SSDI cycle. (a) PE cycle and (b) cycle.

It is noticeable that the SSDI process may be seen as a coupling magnication, because the apparent coupling factor compared to SECE becomes
2 kapp = k2

1+ . 1

(29)

that the second principle of thermodynamics guarantees an irreversible process due to losses during inversion. The cycle area will increase until the electric eld gain due to temperature variation equals exactly losses. EM Em p = (h c ). 33 (21)

This latter coupling may become much larger than 1. Indeed, for excellent inversion ((1 + )/(1 ) 1), the expression may be simplied SSDI =
2 kapp Carnot . 2 1 + kapp

(30)

Calculation of heat transferred to hot and cold sources is very similar to the SECE example. During temperature increase (path E-F) Qh1 = cE (h c ) p2 2 2 (h c ). 2 33

(22)

During isothermal voltage inversion (path F-C) Qh1 = ph (Em EM ). (23)

Thus, as for SECE example, the conversion ratio may tend to Carnots one for excellent coupling factor (k 2 1) and for excellent inversion quality ( 1). As noted in the beginning of this section, the SSDI technique was not designed for energy harvesting but for mechanical vibration damping. The so-called SSHI technique [6] was derived from SSDI for the purpose of energy harvesting. Energy conversion cycle of SSHI is relatively near to that of SSDI. However, understanding of its analysis requires a description of the associated electronic circuit behavior. So for clarity of this paper we have chosen to detail the SSHI cycles in Section III.

sebald et al.: pyroelectric energy conversion: optimization principles

543

Fig. 5. Eciency of the conversion for dierent techniques as a function of the coupling factor squared. Dashed line is for Carnot cycle, solid line is for SSDI with dierent inversion factors , and dasheddotted line is for pure resistive load.

The external boundary condition is E = D (33)

where is the resistive load connected to the piezoelectric material. When the temperature variation is sinusoidal, there exists an optimal load depending on the frequency, OPT =
Fig. 4. Thermodynamic cycles for resistive cycle. (a) PE cycle and (b) cycle.

1 33

(34)

where is the pulsation of the temperature variation. Electrical energy dissipated per cycle is WE = p2 (h c )2 . 4 33 (35)

D. Resistive Cycle When wondering how to consume the electricity converted from heat energy, one could think just connect a resistor to the active material electrodes! This is indeed the simplest way to perform energy conversion cycles on ferroelectric materials (which is known as Standard AC in some references [5], [6]). For the sake of presenting a comprehensive argument, we study here the corresponding cycles (Fig. 4). Lefeuvre et al. [6] gave detailed calculations for electromechanical conversion using a single resistive load. The development presented below is an adaptation to electrothermal conversion. The starting equations become = D 33 E + p = pE + cE Q where dotted variables are time derivatives. (31) (32)

If we neglect the electrocaloric coupling, Qh = cE (h c ). And nally, Resistive = E. Discussion of Cycles We show here four dierent cycles with dierent eciencies and dierent principles. Table II summarizes the results. Fig. 5 illustrates the eciency for all techniques as a function of the coupling factor squared. It is clear that SSDI is much more ecient than the others (except Carnot) and increases the overall eciency to 50% of that of Carnot for an inversion quality of 0.8 and a coupling factor squared of only 10%, whereas it is limited to 5% for other techniques. 2 k Carnot . 4 (37) (36)

544

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008 TABLE II Comparison Between Four Different Cycles for Energy Harvesting.

Cycle Carnot

Eciency ref c =1 h

Maximum electric eld EM EM cE = ln p h c

Diculty to implement High voltage amplier and diculty to really harvest energy Simple and ecient electronic circuit

Necessary information Full temperature prole known in advance. No prerequisite info.

SECE

k2 ref 1 + 0.5k 2 ref k2 2 ref + 1 2 2 k ref 4 1+ ref 1

EM =

p (h c ) 33

SSDI

SSHI = 1 + k2

EM =

p (h c ) 33 (1 )

Simple and ecient electronic circuit

No prerequisite info.

Resistive

Resistive =

p EM = (h c ) 2 233

Very simple circuitry

Frequency information

Fig. 6. Eciency of the conversion as a function of temperature variation = h c , with c = 300 K. Dashed line is for Carnot cycle and solid line is for SECE cycle with dierent coupling factors.

Fig. 6 shows the eciency of conversion as a function of temperature dierence between hot and cold reservoirs. Compared to thermoelectric energy harvesting (using Seebeck eect), where eciency is limited by materials properties, eciency for pyroelectric materials may tend to Carnots one. Thermoelectric conversion eciency may be expressed as [21], [22] h c ZT + 1 1 = . (38) h ZT + 1 + h /c For the best thermoelectric materials, the gure of merit ZT reaches 1 around room temperature with Bi2 Te3 materials for example [23]. As a result, the best eciency reaches 17% of Carnot eciency (considering low temperature dierences to maximize eciency). To get 50% of Carnot eciency, one should nd a material having a gure of merit of 9, which is ten times higher than the best known thermoelectric materials, and we have not considered here the large temperature dierences case, which re-

sults in degrading the eciency. Consequently, the Carnot cycle is most interesting for energy harvesting, but full temperature prole information is necessary before any temperature variation occurs. This process is indeed possible for a controlled temperature variation (as for fuel engines). One could imagine a controlled gas heater inducing temperature variations and pyroelectric energy harvesting. Another limitation is the necessary electric eld to be applied to the pyroelectric material. As described in Section III (for example, 1 C temperature variation), using realistic materials properties, the electric eld should be in the 14 kVmm to 1K1 range. When not broken with electric arcs, most bulk ferroelectric materials are highly nonlinear, either for dynamics nonlinearities, or for static nonlinearities [24][27]. For noncontrolled temperature variationsi.e., imagine a temperature perturbation in the vicinity of a pyroelectric material, such as going outside by 20 C or opening a doorimplementing the Carnot cycle is not possible, except if one can predict the future temperatures values. In such cases, the resistive case is not realistic (a temperature variation is rarely a sinus), since the resistor is adapted on the frequency of excitation. On the contrary, SECE, SSDI, and SSHI could be used. With a sinusoidal excitation, the voltage of the ferroelectric element should be reversed (SSDI and SSHI) or short-circuited (SECE) at every maximum or minimum of the voltage signal. Calculations given in sections B and C are suitable for any periodic temperature signal with constant maximum amplitude (even if this diers from sinus). When this diers from sinus, the necessary choice is this: when should the switch occur? To solve this problem, a probabilistic approach [28] and similar techniques adapted to random signals are necessary to maximize energy harvesting.

III. Pyroelectric Materials We aim in this part of the article to present a survey on existing materials. What are the important parame-

sebald et al.: pyroelectric energy conversion: optimization principles

545

Fig. 7. Coupling factor squared as a function of voltage response for dierent materials.

ters? For standard pyroelectric materials, dierent gures of merit exist. The two following ones are dedicated to pyroelectric sensors and are dened as [29]: Current responsivity gure of merit: Fi = p . cE (39)

Voltage responsivity gure of merit: Fv = p . 33 cE (40)

In Section II, we showed the great importance of another parameter for energy harvestingthe electrothermal coupling factor dened in (18). This parameter may be also named energy gure of merit FE . For every material presented in Table III, we give two dierent gures of merit at room temperature (unless otherwise specied). The rst one is the electrothermal coupling factor. The electric eld sensitivity to temperature variation is also given (= p/ 33 ). Indeed, for a given temperature variation, the obtained voltage that appears on the ferroelectric material is important. Voltages that are too high result in inherent losses of the electronic circuitry of the harvesting devices. It is also dicult to harvest energy in the case of very small pyroelectric voltages eciently because of the voltage drop of semiconductors. One may object that whatever the electric eld sensitivity, we can adjust the thickness of ferroelectric material to keep the voltage in a given useful range of variation. However, bulk ceramics below 80 m are nearly impossible to handle. Then, the technology of thick and thin lms may be used to lower thickness, but ferroelectric properties usually decrease quickly [30], [31]. Inversely, a thick ferroelectric material may generate high voltages for a given temperature variation, but its high thermal mass opposes quick temperature variations. Fig. 7 shows the electrothermal coupling factor as a function of voltage response to a temperature variation. Among all materials, a few exhibit a coupling factor squared above 1%. Simple harvesting devices, such as SECE, require a large coupling factor to ensure an eective energy harvesting. To perform the Carnot

cycle with a pyroelectric material, one should apply an electric eld proportional to cE /p (see Table II for details). Assuming that all the materials have very similar heat capacity, minimizing the electric eld is the same as maximizing the pyroelectric coecient. From that point of view, it is highly unrealistic to think about Carnot cycles using polymers. Composites could be interesting because they exhibit a very high breakdown electric eld while keeping a quite high pyroelectric coecient. Bulk materials exhibit a very high pyroelectric coecient, but usually break above 46 kVmm1 . Additional data on electric breakdown resistance of materials is required to get more precise information about the feasibility of Carnot cycles. The best materials are (1-x)Pb(Mg1/3 Nb2/3 )-xPbTiO3 (PMN-PT) single crystals, with a coupling factor as high as 4.7%. Moreover, voltage response is quite high. However, those materials are expensive and fragile. Single crystals are grown by Bridgman technique [43][46]. It should thus be dicult to really think of industrial use, unless performance is the main priority. It should be noted that the PVDF exhibits a coupling factor as high as bulk ceramics but with a voltage response much larger (six times greater than PZT). This material seems to be very interesting since it is low cost, not fragile, and stable under large electric eld or temperature variations. Another excellent material is the PZT/PVDF-HFP composite, with very high coupling factor. This kind of material is easy to use, because of its exibility, but suers large dependence on temperature. Moreover, the value given here is only for 70 C, and it decreases quickly when changing the temperature. Most probably, the future of pyroelectric energy harvesting is related to composites investigations.

IV. Energy Harvesting Devices Orders of magnitude of the powers consumed by various CMOS electronic devices that could be powered by miniature energy harvesting devices are presented in Fig. 8. As explained in previous sections, implementation of Carnot, SECE, SSH, or SSDI cycles require controlling the pyroelectric voltage with the temperature variations. Practical solutions for voltage control in the cases of SSH and SECE techniques are presented in [2][7]. The following subsections concern the eective energy harvesting compared to optimized energy conversion presented in Section II. Indeed, except the pure resistive case, where the resistance connected to the active material electrodes may be considered as the simulation of an electric load, the converted energythat is to say, the useful or usable energy actually diers from harvested energy as shown below. A. Ideal Power Interface Implementation of energy harvesters using the presented energy conversion cycles requires power interfaces for achieving the desired energy exchanges between the active material and the electrical energy storage cell. Reversible voltage ampliers could be used for this purpose,

546

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008 TABLE III Pyroelectric Properties for Different Class of Materials. Coecient Unit p Cm2 K1 33 0 cE (106 ) Jm3 K1
3 p/ 33 (10 ) Vm1 K1

k2 %

Ref.

Note

PZN-PT and PMN-PT single crystals 111 110 001 001 011 111 001 011 111 001 011 111 PMN-0,25PT PMN-0.25PT PMN-0.25PT PMN-0,33PT PMN-0,33PT PMN-0,33PT PMN-0,28PT PMN-0,28PT PMN-0,28PT PZN-0,08PT PZN-0,08PT PZN-0,08PT 1790 1187 603 568 883 979 550 926 1071 520 744 800 961 2500 3000 5820 2940 650 5750 2680 660 3820 1280 950 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 210 54 23 11 34 170 11 39 183 15 66 95 4.79 0.81 0.17 0.08 0.38 2.12 0.08 0.46 2.50 0.10 0.62 0.96 [9,32] [9] [9] [33] [33] [33] [33] [33] [33] [33] [33] [33]

Bulk ceramics PZT PMN-0.25PT ceramic (BaSrCa)O3 PLZT 0.5/53/47 PLZT 8/53/47 PLZT 14/53/47 533 746 4000 360 97 19 1116 2100 16000 854 238 296 2.5 2.5 2.5 2.5 2.5 2.5 54 40 28 48 46 7 0.37 0.38 1.44 0.22 0.06 0.002 [34] [9] [35] [36] [36] [36]

@22 C 2 C

Thin lms (PZT)/PZT composite PbCaTiO3 PZT 700 nm PMZT 700 nm 180 220 211 352 1200 253 372 255 2.5 2.5 2.5 2.5 17 98 64 156 0.039 0.28 0.17 0.70 [37] [38] [39] [39]

Polymers and composites PVDF PZT/P(VDF-TrFE) 50% PZT/PVCD-HFP 50/50 vol% PZT0.3/PU0.7 vol%
Values

33 33.1 450 90

9 69.2 85 23

1.8 2 2 2

314 54 598 442

0.14 0.028 4.28 0.63

[40] [34] [41] [42]

@70 C 2 C

given here may not be accurate due to the lack of precisions in references (temperatures of measurement of pyroelectric coecient especially). Moreover, the heat capacitance is most of the time estimated using similar materials. However, the range of variation of CE being quite small and its inuence is weak.

Fig. 8. Powers consumed by CMOS electronic devices.

but a critical parameter for these power interfaces is their eciency. The working principle of so-called linear voltage ampliers used in audio applications limits their eciency to 50% in theory and to lower values in practice. Another possibility is to use switching mode power converters. Contrary to linear ampliers, switching mode ampliers may theoretically reach an eciency of nearly 100%. Indeed, by principle, they are exclusively made up with electronic switches (ON state = very small resistance, OFF state = quasi-innite resistance) associated with quasilossless passive elements such as inductors, transformers, and capacitors. The circuit presented on Fig. 9 could be the ideal voltage amplier for controlling the energy harvesting voltage cycles because it may deliver a perfectly controlled ac voltage and may have very weak energy losses. This H-bridge switching mode power converter is a wellknown structure used for controlling the power exchanges between ac and dc electrical sources [49]. The four electronic switches are controlled through a pulse width modulator (PWM) that turns ON and OFF the switches at high frequency (much higher than the temperature varia-

sebald et al.: pyroelectric energy conversion: optimization principles

547

made up with bipolar or MOSFET transistors. Indeed, the OFF stage resistance of such transistors is typically below 100 M, so the leakage currents of the switches may reach several microamperes, and they may dissipate most of the power produced by the pyroelectric material. This point becomes even more critical as the frequency of the temperature cycle is low. As a conclusion, although not developed yet, it is highly feasible to think about high-eciency ampliers for eective regenerative thermodynamic cycles on ferroelectric materials. B. Case of SSHI
Fig. 9. Circuit diagram of the energy harvesting device including the H-bridge switching mode power interface.

tion frequency) with variable duty cycle. The duty cycle D is the key parameter for controlling the average ac voltage Vac , respectively, to the dc voltage Vdc (0 D 1): Vac = Vdc (2D 1). (41)

The PWM may be also disabled. In this case, all the switches are simultaneously opened, leaving the electrodes of the active material on an open circuit. This particular state is necessary, for instance, in some stages of SECE or SSDI cycles. The inductor L connected between the active material and the electronic circuit forms a low-pass lter with the ferroelectric material capacitor C . Thus, the voltage V across the active material electrodes can be considered as perfectly smoothed as long as the switching frequency of the PWM remains much higher than 2 LC . In other words, if this condition is veried, the voltage ripple due to switching across the active material is negligible, and thus: V = Vdc (2D 1). (42)

Theoretical waveforms in the cases of SECE and SSDI are presented on Fig. 10(a) and (b), respectively. In practice, energy losses due to imperfections of the real components aect the eciency. For instance, high power (1 kW to 100 kW), high voltage (500 V to 5 kV) industrial switching mode ampliers have typical eciencies between 85% and 95%. Low power (1 mW to 10 W), low voltage (1 V to 20 V) switching mode ampliers commonly used in wearable electronic devices also have high eciencies, typically between 70% and 90%. However, it is important to mention here that the considered energy harvesting devices are in the microwatt range, but with relatively high voltages (50 V to 1 kV), so the characteristics of the required power amplier are out of usual application domains and require a specic design. Such switching mode interfaces have been successfully demonstrated in the cases of vibrationpowered piezoelectric and electrostatic energy harvesting devices. Their eciency is typically above 80% for output power levels in the 50 W to 1 mW range and with mechanical frequencies between 10 Hz and 100 Hz [50]. One of the critical points in this ultra low power domain is the OFF state resistance of the electronic switches, which are

Section II explored theoretical developments for SECE and SSDI cycles in the case of pure energy conversion without eective energy harvesting. It was shown that the SECE cycle is a special case of SSDI cycle where = 0. For the sake of presenting a comprehensive argument, we will develop here the SSHI case. The SSHI energy conversion cycle may be performed with the ideal switching mode power interface previously presented. However, understanding of this technique is simpler considering the circuit that was rst proposed [2]: the pyroelectric element is connected to a switched inductor in parallel to the ac side of a rectier bridge, the dc side of the rectier being connected to an energy storage cell, as shown in Fig. 11(a). Typical waveforms are shown in Fig. 11(b). After reaching a minimum temperature c , the switch S 1 is turned ON. An oscillating discharge of the pyroelectric capacitor C occurs then through the inductor L. The switch is turned OFF after a time LC corresponding to half a period of the electrical oscillations, so that the pyroelectric voltage polarity is reversed. A small amount of energy is dissipated in the inductor during this operation, so the reversed voltage absolute value is reduced by a factor compared to its value just before the switch is turned ON (0 < 1). Then the temperature increases, resulting in an increase of the open circuit voltage. When reaching voltage Vdc , the rectier bridge conducting and the current owing out the pyroelectric element directly supply the energy storage cell until the temperature reaches its maximum value h . We will consider here that when the diode bridge conducts, the total electric charge owing out the pyroelectric element exactly equals the charge owing through the energy storage cell. In such case, we have = CV + p Q (43)

where Q, V , , C , are electric charge, voltage, temperature, capacitance of the pyroelectric element, and the surface of its electrodes, respectively. Note that it is easier to formulate the problem using voltage and electric charge variables instead of electric induction and electric eld variables when the circuit used for achieving the energy conversion cycles is included in the analysis. When temperature reaches value cond , the rectier bridge starts to conduct. Considering that the ON stage duration of the

548

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008

Fig. 10. Typical waveforms of (a) SECE and (b) SSDI with the H-bridge switching mode power interface.

Fig. 11. (a) Circuit diagram of SSHI circuit and (b) typical waveforms.

switch S 1 is much smaller than the temperature cycle period, and this temperature cond is given by p(Cond c ) = C (Vdc Vdc ). (44)

According to (47), there is an optimal value of voltage Vdc that maximizes the harvested energy (Vdc )opt = p(h c ) . 2C (1 ) (48)

Then we calculate the electric charge generated by the pyroelement between time tcond and T /2 Q = p(h cond ). (45)

Thus, maximum energy harvested per temperature cycle is given by WMAX = (p(h c ))2 . 2C (1 ) (49)

The energy received by the storage cell between time tcond and T /2 is given by W = Q Vdc . (46)

And the maximum harvested energy per cycle and per volume unit of active material is given by WMAX (J m3 ) = p2 (h c )2 . 2 33 (1 ) (50)

Symmetry of the cycle and combination of (44) with (45) and (46) lead to the expression of the energy harvested per cycle Wcycle = 2Vdc (p(h c ) CVdc (1 )). (47)

To illustrate the order of magnitude of power, energy, and optimal dc voltage, consider a 2 C temperature variation at 1 Hz (M = 1 C), with an inversion ratio of 0.8 and a pyroelectric element of 1 nF ( 111 PMN-PT single crystal,

sebald et al.: pyroelectric energy conversion: optimization principles TABLE IV Energy Densities for 300 to 310 K Cyclic Temperature Variations and Number of Cycles Per Hour for Producing 30 W/CM3. Energy density (J/cm3 ) 0.149 0.0118 0.00855 0.00540 Cycles per hour for producing 30 W/cm3 0.725 9.12 12.6 20.0

549

Material 111 PMN-0,25PT Single crystal PMN-0.25PT Ceramic PbCaTiO3 Thin lm PVDF

area of 1 cm2 and thickness of 850 m). Those parameters give (Vdc )OPT = 890 V, harvested power of 0.32 mW, and harvested energy per cycle of 0.32 mJ. In addition, the results presented above show that optimizing energy conversion in no-load cases is the same as optimizing the energy harvesting, especially in terms of materials properties and inversion ratio. Finally, the question that comes when speaking about pyroelectric energy harvesting is the performance comparison with thermoelectric eect. Conditions for energy conversion are not the same because energy harvesting using pyroelectric eect requires temperature variations in time whereas thermoelectric eect needs temperature variations in space (temperature gradients). The proposed comparison is done considering temperature variations between 300 K and 310 K at frequency F for thermoelectric energy conversion, and pyroelectric energy conversion in the case of 300 K and 310 K for the cold and hot sources, respectively. Typical energy density of miniature thermoelectric modules in this case of operation is near 30 W/cm3 power density. As a comparison, Table IV gives the number of temperatures cycles per hour needed to get 30 W/cm3 power density with pyroelectric eect in the case of SSHI cycles ( = 0.6) for several materials.

The simplest devices would require a very high electrothermal coupling factor (k 2 = p2 0 /( 33 cE )), and we focused the investigation on pyroelectric materials comparing that coupling factor. We found that using 0.75Pb(Mg1/3Nb2/3 )O3 -0.25PbTiO3 single crystals oriented 111 and SSHI with an inversion factor of 0.8, it should be possible to reach a conversion of more than 50% of the Carnot cycle ratio (14% of the Carnot cycle for effective energy harvesting). Some important problems were pointed out that can interfere with the technical implementation of such cycles, such as the frequency problems and eciency optimization. Nevertheless, we can expect to get very high harvested energies using realistic materials compared to standard thermoelectric devices. Finally, ferroelectric materials are both pyroelectric and piezoelectric. When designing an electrothermal energy harvester, it should be possible to get high sensitivity to vibrations (when bonding a ferroelectric material on a host structure). However, the solution will be two-fold. The frequencies may be very dierent between vibration and temperature vibration. In such cases, electronics may easily be adapted and optimized to address only one frequency range. On the other hand, frequencies may be close to each other. In such cases, the resulting voltage on the active element will be the sum of both contributions. A smart controller is then necessary to optimize the energy conversion, similar to the situation illustrated by the random signals case in electromechanical energy harvesting [18].

References
[1] S. Roundy, E. S. Leland, J. Baker, E. Carleton, E. Reilly, E. Lai, B. Otis, J. M. Rabaey, P. K. Wright, and V. Sundararajan, Improving power output for vibration-based energy scavengers, IEEE Pervasive Computing, no. 1, pp. 2836, 2005. [2] D. Guyomar, A. Badel, E. Lefeuvre, and C. Richard, Towards energy harvesting using active materials and conversion improvement by nonlinear processing, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 52, no. 4, pp. 584595, 2005. [3] E. Lefeuvre, A. Badel, C. Richard, and D. Guyomar, Piezoelectric energy harvesting device optimization by synchronous electric charge extraction, J. Intell. Mater. Syst. Struct., vol. 16, no. 10, pp. 865876, 2005. [4] A. Badel, E. Lefeuvre, C. Richard, and D. Guyomar, Eciency enhancement of a piezoelectric energy harvesting device in pulsed operation by synchronous charge inversion, J. Intell. Mater. Syst. Struct., vol. 16, no. 10, pp. 889901, 2005. [5] A. Badel, D. Guyomar, E. Lefeuvre, and C. Richard, Piezoelectric energy harvesting using a synchronized switch technique, J. Intell. Mater. Syst. Struct., vol. 17, no. 8-9, pp. 831839, 2006. [6] E. Lefeuvre, A. Badel, C. Richard, and D. Guyomar, A comparison between several vibration-powered piezoelectric generators for standalone systems, Sens. Actuators A, vol. 126, no. 2, pp. 405416, 2006. [7] A. Badel, A. Benayad, E. Lefeuvre, L. Lebrun, C. Richard, and D. Guyomar, Single crystals and nonlinear process for outstanding vibration powered electrical generators, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 53, no. 4, pp. 673684, 2006. [8] J. Grindlay, An Introduction to the Phenomenological Theory of Ferroelectricity. Oxford: Pergamon Press, 1970. [9] D. Guyomar, G. Sebald, B. Guiard, and L. Seveyrat, Ferroelectric electrocaloric conversion in 0.75(PbMg1/3 Nb2/3 O3 )0.25(PbTiO3 ) ceramics, J. Phys. D. Appl. Phys., vol. 39, no. 20, pp. 44914496, 2006.

V. Conclusion Energy harvesting from heat is possible using pyroelectric materials and may be of great interest compared to thermoelectric conversion. Pyroelectric energy harvesting requires temporal temperature variationswe may say time gradients of temperatureswhereas thermoelectric energy harvesting requires spatial gradients of temperatures. Usual wasted heat more likely creates spatial gradients rather than time gradients. However, the conversion ratio (dened as the ratio of net harvested energy divided by the heat taken from the hot reservoir) could be much larger for pyroelectric energy harvesting. In theory, it could reach the conversion ratio of the Carnot cycle, whatever the materials properties. However, the conversion ratio of thermoelectric conversion is highly limited by the materials properties. We showed four dierent pyroelectric energy harvesting cycles, having dierent eectiveness and advantages.

550

ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 55, no. 3, march 2008 [32] A. Benayad, G. Sebald, L. Lebrun, B. Guiard, S. Pruvost, D. Guyomar, and L. Beylat, Segregation study and modeling of Ti in Pb[(Mn1/3 Nb2/3 )0.60 Ti0.40 ]O3 single crystal growth by Bridgman method, Mater. Res. Bull., vol. 41, no. 6, pp. 1069 1076, 2006. [33] G. Sebald, L. Lebrun, and D. Guyomar, Stability of morphotropic PMN-PT single crystals, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 51, no. 11, pp. 14911498, 2004. [34] G. Sebald, L. Lebrun, B. Guiard, and D. Guyomar, Morphotropic PMN-PT system investigated through comparison between ceramic and crystal, J. Eur. Ceram. Soc., vol. 25, no. 12, pp. 25092513, 2005. [35] K. Zawilski, C. Custodio, R. Demattei, S. G. Lee, R. Monteiro, H. Odagawa, and R. Feigelson, Segregation during the vertical Bridgman growth of lead magnesium niobate-lead titanate single crystals, J. Cryst. Growth, vol. 258, no. 3, pp. 353367, 2003. [36] J. Han, A. von Jouanne, T. Le, K. Mayaram, and T. S. Fiez, Novel power conditioning circuits for piezoelectric micropower generators, in Proc. 19th Annu. IEEE Appl. Power Electron. Conf. Expo., vol. 3, 2004, pp. 15411546. [37] X. Shengwen, K. D. T. Ngo, T. Nishida, C. Gyo-Bum, and A. Sharma, Converter and controller for micro-power energy harvesting, in Proc. 20th Annu. IEEE Appl. Power Electron. Conf. Expo., vol. 1, 2005, pp. 226230. [38] Y. Tang, X. Zhao, X. Wan, X. Feng, W. Jin, and H. Luoa, Composition, dc bias and temperature dependence of pyroelectric properties of 1 1 1-oriented (1-x)Pb(Mg1/3 Nb2/3 )O3 -xPbTiO3 crystals, Mater. Sci. Eng. B, vol. 119, no. 1, pp. 7174, 2005. [39] M. Davis, D. Damjanovic, and N. Setter, Pyroelectric properties of (1-x)Pb(Mg1/3 Nb2/3 )O3 -xPbTiO3 and (1x)Pb(Zn1/3 Nb2/3 )O3 -xPbTiO3 single crystals measured using a dynamic method, J. Appl. Phys., vol. 96, no. 5, pp. 2811 2815, 2004. [40] W. Y. Ng, B. Ploss, H. L. W. Chan, F. G. Shin, and C. L. Choy, Pyroelectric properties of PZT/P(VDF-TrFE) 0-3 composites, in Proc. 12th IEEE Int. Symp. Applications Ferroelectrics, vol. 2, 2000, pp. 767770. [41] D. S. Kang, M. S. Han, S. G. Lee, and S. H. Song, Dielectric and pyroelectric properties of barium strontium calcium titanate ceramics, J. Eur. Ceram. Soc., vol. 23, no. 3, pp. 515158, 2003. [42] A. Pelaiz Barranco, F. Calderon Pinar, and O. Perez Martinez, PLZT ferroelectric ceramics on the morphotropic boundary phase. Study as possible pyroelectric sensors, Phys. Stat. Sol. (a), vol. 186, no. 3, pp. 479485, 2001. [43] M. K. Cheung, K. W. Kwok, H. L. W. Chan, and C. L. Choy, Dielectric and pyroelectric properties of lead zirconate titanate composite lms, Integrated Ferroelectrics, vol. 54, no. 1, pp. 713719, 2003. [44] E. Yamaka, H. Watanabe, H. Kimura, H. Kanaya, and H. Ohkuma, Structural, ferroelectric, and pyroelectric of highly caxis oriented Pb1-xCaxTiO3 thin lm grown by radio-frequency magnetron sputtering, J. Vacuum Sci. Technol. A: Vacuum, Surfaces, Films, vol. 6, pp. 29212928, 1988. [45] Q. Zhang and R. W. Whatmore, Improved ferroelectric and pyroelectric properties in Mn-doped lead zirconate titanate thin lms, J. Appl. Phys., vol. 94, no. 8, pp. 52285233, 2003. [46] S. B. Lang and S. Muensit, Review of some lesser-known applications of piezoelectric and pyroelectric polymers, Appl. Phys. A: Mater. Sci. Proc., vol. 85, no. 2, pp. 125134, 2006. [47] L. F. Malmonge, J. A. Malmonge, and W. K. Sakamoto, Study of pyroelectric activity of PZT/PVCD-HFP composite, Materials Research, vol. 6, no. 4, pp. 469473, 2003. [48] K. S. Lam, Y. W. Wong, L. S. Tai, Y. M. Poon, and F. G. Shin, Dielectric and pyroelectric properties of lead zirconate/polyurethane composites, J. Appl. Phys., vol. 96, no. 7, pp. 38963899, 2004. [49] R. W. Erickson and D. Maksimovic, Fundamentals of Power Electronics. Berlin: Science+Business Media Inc., 2001. [50] E. Lefeuvre, D. Audigier, C. Richard, and D. Guyomar, Buckboost converter for sensorless power optimization of piezoelectric energy harvester, IEEE Trans. Power Electron., vol. 22, no. 5, pp. 20182025, 2007.

[10] G. Sebald, L. Seveyrat, D. Guyomar, L. Lebrun, B. Guiard, and S. Pruvost, Electrocaloric and pyroelectric properties of 0.75Pb(Mg1/3 Nb2/3 )O3 -0.25PbTiO3 single crystals, J. Appl. Phys., vol. 100, no. 12, art. no. 124112, 2006. [11] G. Sebald, S. Pruvost, L. Seveyrat, L. Lebrun, D. Guyomar, and B. Guiard, Study of electrocaloric properties of high dielectric constant ferroelectric ceramics and single crystals, J. Eur. Ceramic Society, vol. 27, no. 13-15, pp. 40214024, 2007. [12] R. Radebaugh, W. N. Lawless, J. D. Siegwarth, and A. J. Morrow, Feasibility of electrocaloric refrigeration for the 4K15K temperature range, Cryogenics, vol. 19, no. 4, pp. 187208, 1979. [13] R. Zhang, S. Peng, D. Xiao, Y. Wang, B. Yang, J. Zhu, P. Yu, and W. Zhang, Preparation and characterization of (1x)Pb(Mg1/3 Nb2/3 )O3 -xPbTiO3 electrocaloric ceramics, Cryst. Res. Technol., vol. 33, no. 5, pp. 827832, 1998. [14] L. Shaobo and L. Yanqiu, Research on the electrocaloric effect of PMN/PT solid solution for ferroelectrics MEMS microcooler, Materials Science Engineering B, vol. 113, no. 1, pp. 4649, 2004. [15] E. Birks, L. Shebanov, and A. Sternerg, Electrocaloric eect in PLZT ceramics, Ferroelectrics, vol. 69, pp. 125129, 1986. [16] P. D. Thacher, Electrocaloric eects in some ferroelectric and antiferroelectric Pb(Zr,Ti)O3 compounds, J. Appl. Phys., vol. 39, no. 4, pp. 19962002, 1968. [17] L. Shebanovs, K. Borman, W. N. Lawless, and A. Kalvane, Electrocaloric eect in some perovskite ferroelectric ceramics and multilayer capacitors, Ferroelectrics, vol. 273, pp. 137142, 2002. [18] L. Shebanovs and K. Borman, On lead-scandium tantalate solid solutions with high electrocaloric eects, Ferroelectrics, vol. 127, pp. 143148, 1992. [19] A. S. Mischenko, Q. Zhang, J. F. Scott, R. W. Whatmore, and N. D. Mathur, Giant electrocaloric eect in thin-lm PbZr0.95 Ti0.05 O3 , Science, vol. 311, no. 5765, pp. 12701271, 2006. [20] A. S. Mischenko, Q. Zhang, R. W. Whatmore, J. F. Scott, and N. D. Mathur, Giant electrocaloric eect in the thin lm relaxor ferroelectric 0.9PbMg1/3 Nb2/3 O3 -0.1PbTiO3 near room temperature, Appl. Phys. Lett., vol. 89, no. 24, art. no. 242912, 2006. [21] G. Min, D. M. Rowe, and K. Kontostavlakis, Thermoelectric gure-of-merit under large temperature dierences, J. Phys. D. Appl. Phys., vol. 37, no. 8, pp. 13011304, 2004. [22] I. B. Cado and E. Miller, Thermoelectric Materials and Devices. New York: Reinhold Publishing Corporation, 1960. [23] H. Beyer, J. Nurnus, H. B ottner, A. Lambrecht, E. Wagner, and G. Bauer, High thermoelectric gure of merit ZT in PbTe and Bi2 Te3 -based superlattices by a reduction of the thermal conductivity, Physica E, vol. 13, no. 2-4, pp. 965968, 2002. [24] B. Ducharne, D. Guyomar, and G. Sebald, Low frequency modeling of hysteresis behavior and dielectric permittivity in ferroelectric ceramics under electric eld, J. Phys. D. Appl. Phys., vol. 40, no. 2, pp. 551555, 2007. [25] G. Sebald, E. Boucher, and D. Guyomar, Dry-friction based model for hysteresis related behaviors in ferroelectric materials, J. Appl. Phys., vol. 96, no. 5, pp. 27852791, 2004. [26] G. Sebald, L. Lebrun, and D. Guyomar, Modeling of elastic nonlinearities in ferroelectric materials including nonlinear losses: Application to relaxors single crystals, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 52, no. 4, pp. 596603, 2005. [27] R. C. Smith, S. Seelecke, Z. Ounaies, and J. Smith, A free energy model for hysteresis in ferroelectric materials, J. Intell. Mater. Syst. Struct., vol. 11, no. 1, pp. 6279, 2000. [28] D. Guyomar and A. Badel, Nonlinear semi-passive multimodal vibration damping: An ecient probabilistic approach, J. Sound Vibration, vol. 294, no. 1-2, pp. 249268, 2006. [29] R. W. Whatmore, Pyroelectric arrays: Ceramics and thin lms, J. Electroceram., vol. 13, no. 1-3, pp. 139147, 2004. [30] R. A. Dorey and R. W. Whatmore, Electroceramic thick lm fabrication for MEMS, J. Electroceram., vol. 12, no. 1-2, pp. 1932, 2004. [31] D. Damjanovic, Ferroelectric, dielectric and piezoelectric properties of ferroelectric thin lms and ceramics, Rep. Prog. Phys., vol. 61, no. 9, pp. 12671324, 1998.

sebald et al.: pyroelectric energy conversion: optimization principles Gael Sebald was born in 1978. He graduated from INSA Lyon in Electrical Engineering in 2001 (M.S. degree) and received a masters degree in acoustics the same year. He received a Ph.D. degree in acoustics in 2004. He was then a Japan Society for Promotion of Science Awardee (20042005) for a post-doctoral position in the Institute of Fluid Science of Tohoku University, Sendai, Japan, where he worked on ferroelectric bers and vibration control. Gael Sebald is now associate professor at INSA Lyon, Lyon, France. His main research interests are now materials characterization, hysteresis modeling, multiphysics coupling in smart materials, and energy harvesting on vibration and heat.

551

Polytechnique de Toulouse, France. He received the diploma from both universities in 2001 for his work on power electronics converters topologies. In 2002 he got a position of assistant professor at Institut National des Sciences Appliqu ees (INSA) de Lyon, Lyon, France, and he joined the Laboratoire de G enie Electrique et Ferro electricit e. His current research activities include piezoelectric systems, energy harvesting, vibration control, and noise reduction.

Elie Lefeuvre was born in France in 1971. He received the B.S. and M.S. degrees in electrical engineering respectively from Paris-XI University, Paris, France, in 1994 and from Institut National Polytechnique de Toulouse, Toulouse, France, in 1996. At the same time, he was a student at the electrical engineering department of Ecole Normale Sup erieure de Cachan, Cachan, France. He prepared his Ph.D. degree at Laval University of Qu ebec, Qu ebec, Canada, and at the Institut National

Daniel Guyomar received a degree in mechanical engineering, a Doctor-engineer degree in acoustics from Compi` egne University, and a Ph.D. degree in physics from Paris VII University, Paris, France. From 1982 to 1983 he worked as a research associate in uid dynamics at the University of Southern California, Los Angeles, CA. He was a National Research Council Awardee (19831984) detached at the Naval Postgraduate School to develop transient wave propagation modeling. He was hired in 1984 by Schlumberger to lead several research projects dealing with ultrasonic imaging, then he moved to Thomson Submarine activities in 1987 to manage the research activities in the eld of underwater acoustics. Pr. Daniel Guyomar is presently a full-time University Professor at INSA Lyon (Lyon, France), director of the INSA-LGEF laboratory. His present research interests are in the eld of semi-passive vibration control, energy harvesting on vibration and heat, ferroelectric modeling, electrostrictive polymers, and piezoelectric devices.