Optimal emitter-collector gap for thermionic energy converters

Jae-Hyung Lee, Igor Bargatin, Nicholas A. Melosh, and Roger T. Howe

Citation: Appl. Phys. Lett. 100, 173904 (2012); doi: 10.1063/1.4707379

View online: http://dx.doi.org/10.1063/1.4707379
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v100/i17
Published by the American Institute of Physics.

Additional information on Appl. Phys. Lett.

Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors

Downloaded 17 Apr 2013 to 142.104.240.194. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
 APPLIED PHYSICS LETTERS 100, 173904 (2012)

Optimal emitter-collector gap for thermionic energy converters

Jae-Hyung Lee,1 Igor Bargatin,1 Nicholas A. Melosh,2 and Roger T. Howe1,a)
1
Department of Electrical Engineering, Stanford University, Stanford, California 94305, USA
2
Department of Materials Science and Engineering, Stanford University, Stanford, California 94305, USA

(Received 21 February 2012; accepted 11 April 2012; published online 27 April 2012)
In this letter, we calculate numerically the emitter-collector gaps that maximize the power
conversion efficiency of vacuum thermionic energy converters (TECs). The optimum arises
because efficiency drops both at very large gaps, due to space-charge limitations on the TEC
current, and at very small gaps, due to the increased parasitic heat loss via near-field radiative heat
transfer. For typical TECs made with cesiated tungsten electrodes, the optimal gaps range from
900 nm to 3 lm and are approximately equal to the characteristic wavelength of the emitter thermal
radiation, as given by Wien’s displacement law. V C 2012 American Institute of Physics.

[http://dx.doi.org/10.1063/1.4707379]

Thermionic energy converters (TECs) are heat engines than plasma TECs if micron-scale gaps are used (<10 lm).2
that convert heat directly to electricity at very high tempera- This fact has been a motivation for recent efforts to develop
tures, typically >1000  C. This energy conversion process is microfabricated thermionic energy converters.7–9
based on thermionic emission—the evaporation of electrons However, as we demonstrate in this letter, the gap can
from conductors at high temperatures—and was first pro- also be too small. At gaps of a few microns or less, the dis-
posed in 1915.1 In the 1950 s–1960 s, the first practical TECs tance between the electrodes becomes smaller than the char-
with conversion efficiencies of 10%–15% were created, and acteristic wavelength of thermal radiation, given by Wien’s
in the subsequent two decades, both NASA and the Soviet displacement law. As a result, the traditional Stefan-
space programs developed TECs as power sources for space Boltzmann formula, valid only for far-field propagating-
missions.2–6 These devices were fabricated using vacuum- wave radiative heat transfer, can no longer be used. Instead,
tube technology and had electrode gaps on the order of near-field evanescent-wave radiative heat transfer, which is
100 lm, formed by precision machining. sometimes referred to as photon tunneling, starts to domi-
The energy diagram of a TEC is shown in Fig. 1. As the nate. The total radiative heat transfer can then be increased
emitter is heated to high temperatures, the thermal distribu- by many orders of magnitude, as recently demonstrated both
tion develops a long high-energy tail so that some electrons theoretically and experimentally.10,11
begin to overcome the work function barrier and evaporate
from the hot emitter. The electrons can then cross the vac-
uum inter-electrode gap, condense at the relatively cool col-
lector, and finally return to the emitter through an external
load using the difference between the Fermi levels of the two
electrodes to perform useful work. The thermionic currents,
emitted from both emitter and collector, are governed by the
Richardson-Dushman law.
Space charge between the electrodes can dramatically
reduce the output power and efficiency of TECs because the
electrons traversing the inter-electrode gap repel each other.
For macroscopic gaps (>100 lm), the resulting additional
energy barrier (Fig. 1) can reduce the output power and the
conversion efficiency by many orders of magnitude. As a
result, early TECs in the 1950s-1980s used ignited cesium
plasma to neutralize space charge between the electrodes.
Such plasma TECs achieved high output powers, but only at
the cost of greatly increased complexity and decreased maxi-
mum efficiency.4
As an alternative to using plasma, the deleterious effects FIG. 1. The energy diagram of the thermionic energy conversion process.
of space charge can also be mitigated by making the intere- EF;E and EF;C are the Fermi levels of the emitter and the collector, respec-
lectrode gap small enough so that there is not enough space tively. EVac is the position-dependent vacuum level. uE and uC are the work
functions of the emitter and the collector. uME and uMC are the additional
to develop a significant additional energy barrier. It has long barriers due to space charge as seen from the two electrodes. V0 is the volt-
been known that such vacuum TECs can be more efficient age difference between two electrodes and –q is the electron charge. Qprop
and QEvan are the heat fluxes due to propagating-wave and evanescent-wave
photons. Qlead is the heat flux due to conduction in the leads connecting the
a)
E-mail address: [email protected]. electrodes to the load.

0003-6951/2012/100(17)/173904/4/$30.00 100, 173904-1 C 2012 American Institute of Physics

V

Downloaded 17 Apr 2013 to 142.104.240.194. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
 173904-2 Lee et al. Appl. Phys. Lett. 100, 173904 (2012)

For given temperatures of the emitter and the collector, The net energy carried by thermionically emitted electrons
this increase in radiative heat transfer does not affect the out- from emitter to collector is given by2
put power of the TEC. However, it increases the necessary
 
heat input to the emitter and therefore reduces its power con- uE þ uME 2kB
version efficiency. The interplay between the effects of space QThermionic ¼ Jnet þ ðJE TE  JC TC Þ :
e e
charge and near-field radiative heat transfer means that there
is an optimal gap (or a range of gaps) on the order of a few A part of this thermionic energy flux, namely, Jnet V0 ; is con-
microns, which maximizes the energy conversion efficiency. verted to electricity with the rest deposited in the collector as
We note that at much smaller distances, on the order of a heat when thermionic electrons are absorbed into the collec-
few nanometers, electrons begin to tunnel across the gap, tor. The propagating-wave heat transfer is described by the
potentially resulting in another peak in efficiency.12 How- familiar Stefan–Boltzmann law at large inter-electrode gaps.
ever, in this letter, we focus on the currently experimentally However, when two surfaces approach close enough such
feasible9 micron-range gaps. In particular, we calculate the that the gap is on the order of the characteristic wavelength
optimal conditions for electrodes made of cesiated tungsten, of thermal radiation given by Wien’s displacement law,
which have traditionally been used in TECs.2–4 kch  2:9  103 m  K=TE , the propagating-wave heat flux
The energy conversion efficiency of a TEC can be writ- gets modified, but even more importantly, the evanescent
ten as follows: waves from each surface begin to couple to each other and
exchange energy. If the optical properties of both electrodes
Pout Jnet ðV0  Jnet Rlead Þ
g¼ ¼ ; are known, the two components of the radiation heat flux can
Pin Qtotal be calculated as follows:10
where V0 is the voltage difference between two electrodes, ð1
1
Jnet is the net electron current density, Rlead is the total resist- QProp ¼ dx½hðx; TE Þ  hðx; TC Þ
ance of the leads between the electrodes and the external p2 0
ðx
load (Fig. 1), normalized to unit area, and Qtotal is the total c

energy flux between the emitter and the collector.  sProp ðx; b; e1 ðxÞ; e2 ðxÞÞdb;
0
According to the Richardson-Dushmann equation, the ð1
net current, Jnet, is the difference between the thermionic 1
QEvan ¼ dx½hðx; TE Þ  hðx; TC Þ
currents from the emitter and the collector p2 0
ð1
Jnet ¼ JE  JC  sEvan ðx; b; e1 ðxÞ; e2 ðxÞÞdb;
x
c
   
u þ uME u þ uMC
¼ AT 2E exp  E  AT 2C exp  C ; where hðx; TÞ is the mean energy of a Planck oscillator at
kB TE kB TC
the angular frequency x; b is the parallel wavevector, which
must be equal for both surfaces, e1 ðxÞ and e2 ðxÞ are the rela-
where JE and JC are the thermionic current densities from tive permittivities of the emitter and the collector, respec-
emitter and collector, respectively, and A is a materials- tively, and sProp and sEvan are the coupling coefficients for
specific Richardson-Dushmann constant, which we assume propagating and evanescent waves, which are defined in
equal to the ideal value of 1:2  106 Am2 K2 for simplic- Ref. 10.
ity. TE and TC are the temperatures of the emitter and the col- Historically, many TECs have been made of tungsten,
lector, respectively. The additional barriers, uME and uMC ; whose optical properties in the infrared range, where most
are due to the electrostatic field created by the electrons radiative heat transfer occurs, are described reasonably well
between the electrodes and can be calculated using Lang- by the Drude model14
muir’s space-charge theory,13 as described in detail in Sec.
10.3 of Ref. 3. The output voltage is then given by r0 =s
e¼1 ;
V0 ¼ ðuE þ uME Þ  ðuC þ uMC Þ: e0 ðx2 þ ix=sÞ
The thermionic current originating from the collector,
JC , corresponds to electrons flowing in the wrong direction, where r0 is dc conductivity and s is electron relaxation
i.e., from the collector to emitter. The collector current time. Since the dc conductivity depends on temperature, we
becomes negligible (JC  JE ) if the collector temperature is used a 6th order polynomial to fit the published data of tung-
small enough; however, for some regimes of operation, the sten’s resisitivity versus temperature.15 The electron relaxa-
reverse current from the collector can become significant tion time was estimated as s ¼ 1=ðaT 2 þ bT 3 Þ with
mechanism of energy loss, reducing the output power and a ¼ 107 s1 K2 and b ¼ 2  106 s1 K3 (Ref. 14).
conversion efficiency. Since TECs are high-current devices, the electrical re-
The net energy input into the emitter is equal to the total sistance of the leads that connect the electrodes to the exter-
energy flux from emitter to collector, Qtotal ¼ Qthermionic nal load should be optimized.2 If the electrical resistance of
þ Qprop þ Qevan þ Qlead , due to a number of fundamental the lead is too large, most of the potentially available output
energy carriers: thermionically emitted electrons (Qthermionic), power will be wasted on Joule heating in the leads. Con-
propagating-wave photons (Qprop), evanescent-wave photons versely, if the electrical resistance of the lead is too small,
(Qevan), and electron heat conduction in the leads (Qlead). the heat conduction through the leads will be large, and

Downloaded 17 Apr 2013 to 142.104.240.194. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
 173904-3 Lee et al. Appl. Phys. Lett. 100, 173904 (2012)

maintaining the temperature of the emitter will require ex-

cessive input power. This is due to the Wiedemann–Franz
law, which relates the minimum heat conduction through
metal leads to the electrical resistance of the leads
L
QLead ¼ ðT 2  TC2 Þ;
2Rlead E

where the Lorenz number L ¼ 2:24  108 WXK2 . The

optimal lead resistance that maximizes the conversion effi-
ciency can be easily calculated analytically if all other pa-
rameters are known.2
Using the formulas above, we numerically calculated
the I-V curves, the maximum output power, and the other
energy fluxes as a function of the interelectrode gap for any
given values of TE , TC , uE , and uC . Fig. 2 shows one exam-
ple, where TE ¼ 2000 K, TC ¼ 900 K, uE ¼ 2.67 eV, and FIG. 3. Maximum energy conversion efficiency versus gaps for emitter tem-
uC ¼ 1.5 eV. We chose uC ¼ 1.5 eV because it is approxi- peratures of 1500, 2000, and 2500 K, collector temperature of 900 K, and
mately the lowest work function obtainable for cesiated collector work function of 1.5 eV. The emitter work functions were assumed
to be TE ½K=750 eV. The optimal gap ranges were from 300 nm to 6 lm for
tungsten and uE ¼ 2000/750 ¼ 2.67 eV because it is approxi- TE ¼ 1500 K, from 200 nm to 4 lm for TE ¼ 2000 K, and from 150 nm to
mately optimal for this emitter temperature.2 3 lm for TE ¼ 2500 K.
For gaps above a few microns, space charge becomes
significant, and the output power falls off rapidly. Con- generally range from a few tenths of a micron to a few
versely, for gaps below a micron, the total radiative heat microns. Similar graphs were then generated by varying the
transfer increases rapidly due to the evanescent-wave heat emitter temperature from 400 K to 2600 K and assuming the
transfer. The thermionic energy conversion process is gener- emitter work function of TE ½K=750 eV, which is approxi-
ally most efficient when the total energy flux is dominated mately optimal for vacuum TECs (see Sec. 2.8 of Ref. 2). In
by the thermionic component, Qthermionic, so that the ratio of addition, the collector work function was varied from 0.5 to
output power to total energy flux is roughly constant. 1.5 eV to compare the efficiencies of tungsten micron-scale
The energy conversion efficiency can also be plotted thermionic converters that could potentially be achieved in
versus gap for any given values of TE , TC , uE , and uC , as the future with more effective work-function-lowering coat-
illustrated in Fig. 3. As expected, the efficiency is small at ings. For all work functions, we assumed that the collector
both very large gaps and very small gaps and reaches the temperature is uC ½eV  600 K, again because it is approxi-
maximum at gaps on the order of a micron. In Fig. 3, we also mately optimal for TECs (Sec. 2.8 of Ref. 2). Such work
defined the optimal gap range, where the energy conversion functions are feasible for cesiated tungsten electrodes if the
efficiency is at least 0.9 of the maximum. The optimal gaps liquid cesium reservoir is maintained at a temperature of
600–700 K (Sec. 4.2 of Ref. 2).
Figs. 4 and 5 summarize the results of these simulations
by showing the maximum energy conversion efficiencies

FIG. 2. The total energy flux, the heat transfer through propagating and
evanescent waves, the maximum output power per unit area (corresponding
to Rlead ¼ 0), and the heat transferred by thermionically emitted electrons
(for the same maximum power bias conditions) as a function of interelec- FIG. 4. The maximum energy conversion efficiency versus the emitter tem-
trode gap for TE ¼ 2000 K, TC ¼ 900 K, uE ¼ 2000/750 ¼ 2.67 eV, and perature for collector work functions of 0.5, 1.0, and 1.5 eV. We assumed
uC ¼ 900/600 ¼ 1.5 eV. The inset shows I-V curves for several gaps. The the emitter work function, uE , to be TE ½K=750eV, and the collector temper-
circles signify the bias conditions corresponding to maximum output power. ature uC ½eV  600 K. The inset shows the exergy efficiency (also known as
At large gaps, propagating-wave heat flux is approximately constant and can the second-law efficiency), i.e., the energy conversion efficiency divided by
be approximated by the Stefan-Boltzmann formula, QProp  eff ðTE4  TC4 Þ the Carnot efficiency, 1  TC =TE , for the corresponding emitter and collec-
with the effective emissivity eff  0.07. tor temperatures.

Downloaded 17 Apr 2013 to 142.104.240.194. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
 173904-4 Lee et al. Appl. Phys. Lett. 100, 173904 (2012)

there is an optimal inter-electrode gap that maximizes the

energy conversion efficiency of vacuum TECs. The effi-
ciency is typically maximized for gaps similar to the charac-
teristic wavelength of thermal radiation of the emitter. For
the particular case of cesiated tungsten electrodes, the effi-
ciency is nearly optimal for gaps between 900 nm and 3 lm
over a wide range of operating conditions. Microfabricated
vacuum TECs therefore show great promise for achieving ef-
ficient conversion of heat directly to electricity and have
potential application in topping cycles for concentrated solar
thermal power plants.9,18,19
This work was supported in part by DARPA seed grant
no. W91CRB-10-1-0001, DARPA and Robert Bosch North
America through the Center on Interfacial Engineering for
FIG. 5. The optimal gap range versus the emitter temperature at the same MEMS (CIEMS), and BASF through the Stanford Energy
condition as Fig. 4. The thick black line shows Wien’s displacement law, and Environment Affiliates Program.
kch  2:9  103 m  K=TE ; to illustrate the relationship between optimal
gaps and the characteristic wavelength of thermal radiation from the emitter. 1
W. Schlichter, Ann. Phys. 47(15), 573–640 (1915).
2
G. N. Hatsopoulos and E. P. Gyftopoulos, Thermionic Energy Conversion
(Fig. 4) and optimal gap ranges (Fig. 5) versus the emitter (MIT, Cambridge, 1973), Vol. 1.
3
G. N. Hatsopoulos and E. P. Gyftopoulos, Thermionic Energy Conversion
temperature. The inset of Fig. 4 shows the second-law (MIT, Cambridge, 1979), Vol. 2.
(exergy) efficiency, relative to the Carnot limit. Interestingly, 4
P. Shefsiek, IEEE Trans. Plasma Sci. 38, 8 (2010).
5
a large overlap between the optimal gap ranges for all three National Research Council, An Assessment of the DTRA’s Advanced
different collector work functions could be found, corre- Thermionics Research and Development Program, National Academy
Press (2001)
sponding to gaps from 900 nm to 3 lm (Fig. 5). We note also 6
S. F. Adams, AIP Conf. Proc. 813, 590–597 (2006).
that the middle of the optimal gap range is similar to the 7
C. Zhang, Ph.D. dissertation, University of Michigan, 2003.
8
characteristic thermal radiation wavelength, kch . D. King, K. Zavadil, D. Jennison, C. Battaile, and A. Marshall, Low Work
TECs made from materials other than tungsten will gen- Function Material Development for the Microminiature Thermionic
Converter, Sandia Report SAND2004-0555, 2004.
erally have different ranges of optimal gaps as the details of 9
J.-H. Lee, I. Bargatin, T. O. Gwinn, M. Vincent, K. A. Littau, R. Mabou-
evanescent heat transfer depend on the optical properties of dian, Z.-X. Shen, N. A. Melosh, and R. T. Howe, in IEEE MEMS
the electrodes. In fact, some materials manifest additional 10
Conference-Technical Digest, Paris, France, Jan 28-Feb 5, 2012.
mechanisms of near-field heat transfer due to resonant sur- S. Basu, Z. M. Zhang, and C. J. Fu, Int. J. Energy Res. 33, 1203–1232
(2009).
face waves.16,17 However, the optimal gaps are still likely to 11
A. Narayanaswamy, S. Shen, L. Hu, X. Chen, and G. Chen, Appl. Phys. A
be on the order of a few microns for all materials because 96, 357–362 (2009).
12
space charge generally becomes significant at gaps above J.-I. Lee, Y.-H. Jeong, H.-C. Ho, R. Hannebauer, and S.-K. Yoo, Appl.
Phys. Lett. 95, 223107 (2009).
10 lm, and all mechanisms of near-field radiative heat trans- 13
I. Langmuir, Phys. Rev. 21, 419–435 (1923).
fer begins to dominate at distances of a few microns or less 14
S. Roberts, Phys. Rev. 114, 104–115 (1959).
at the temperatures of interest to TECs. The significance of 15
J. Maas, R. Huguenin, and V. A. Gasparov, J. Phys. F: Met. Phys. 15,
these results is that the optimal gaps of vacuum TECs can be 271–278 (1985).
16
K. Joulain, J.-P. Mulet, F. Marquier, R. Carminati, and J.-J. Greffet, Surf.
conveniently achieved using MEMS fabrication technology. Sci. Rep. 57, 59–112 (2005).
Moreover, at least for tungsten electrodes, microfabricated 17
S. Shen, A. Narayanaswamy, and G. Chen, Nano Lett. 9, 2909–2913
TECs have a relatively wide range of optimal gaps (more (2009).
18
than an order of magnitude), indicating that they should be A. Ogino, T. Muramatsu, and M. Kando, Jpn. J. Appl. Phys. 43, 309–314
(2004).
robust to changes in the gap during operation. 19
J. W. Schwede, I. Bargatin, D. C. Riley, B. E. Hardin, S. J. Rosenthal, Y.
In summary, we have shown that due to the interplay Sun, F. Schmitt, P. Pianetta, R. T. Howe, Z.-X. Shen, and N. A. Melosh,
between space charge and near-filed radiative heat transfer, Nat. Mater. 10, 762–767 (2010).

Downloaded 17 Apr 2013 to 142.104.240.194. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions