N94-11407

REVIEW OF BETAVOLTAIC ENERGY CONVERSION

Larry C. Olsen

Washington State University / Tri-Cities

100 Sprout Road
Richland, WA 99352

Betavoltaic energy conversion refers to the generation of power by coupling a beta

source to a semiconductor junction device. This paper briefly reviews the theory of betavoltaic
energy conversion and some past studies of the subject. Calculations of limiting efficiencies for
semiconductor Cells versus bandgap are presented along with specific studies for Pm-147 and Ni-
63 fueled devices. The approach used for fabricating Pm-147 fueled batteries by the author in
the early 1970's is reviewed. Finally, tt'ie potential performance of advanced betavoltaic power
sources is considered.

INTRODUCTION

Betavoltaic energy conversion refers to the generation of power by coupling a beta

source to a semiconductor junction device. Some interest haS been shown in this approach to
energy conversion at the last two SPRAT meetings. As a result, it seemed timely to review the
subject. This paper briefly reviews the theory of betavoltaic energy conversion, past studies in the
field and discusses the potential performance of betavoltaic _ystems based on the availability of
beta sources and currently available solar cell materials.

PAST BETAVOLTAIC STUDIES

The first report of an electron-voltaic effect was given by Ehrenberg, el al., in 1951 (ref.
1). They were primarily interested in the current magnification that resulted when selenium
photocells were bombarded by an electron beam. Rappaport was the first to describe betavoltaic
studies, that is, investigations involving beta sources coupled to semiconductor junction devices
(ref.2). He reported in 19S3 on characteristics of silicon alloy junctions coupled to a 50 milliecurie
Sr90-y 90 radioactive _0urce. One cell produced 0.8 microwatts with an overall efficiency of 0.2 %
being attained. The overall efficiency is based on the total power produced by the radioisotope
source. Plan and Roosbroeck reported on similar studies about the same time (ref. 3). They
discussed the general problem of betavoltaics and gave experimental results for Sr90-y 90
sources combined silicon and germanium devices.

A more detailed report on the work by Rappaport and coworkers at RCA was given in
1956 (ref. 4). Further results were described for silicon and germasnium alloy junctions coupled
to Sr90-y90beta sources. In addition, the theory of betavoltaic devices was formulated. The
interdependence of beta source parameters such as self absorption coefficient, beta energy

256
spectrum and actitivity, and semiconductor parameters such as energy gap and minority carrier
properties were emphasized. The RCA group also identified the potential of Pm-147 betavoltaics
in the 1956 paper. This was partially motivated by the negative results obtained with Sr90-y 90
sources coupled to Si devices. As a result of radiation damage, the maximum power produced by
a Si/Sr-90 system was found to decay to one-tenth of its initial value within one week of life. The
final contribution of the RCA group is contained in a 1964 paper by Flicker, et al. (ref. 5). Si and
GaAs diffused junction devices were coupled to Pm-147 sources. Beta sources were made by
precipitating Pm-147 as the hydrated oxide (Pm203,6H20) onto a substrate. Betavoltaic studies
with GaAs cells yielded very poor results. Studies with silicon cells included the fabrication of
prototype power sources consisting of a Pm-147 source combined with one or two silicon cells.
Overall efficiencies of 0.4 % and 0.77 % were achieved. Lifetime studies with these prototypes
showed only a slight effect due to radiation damage.

The most extensive effort concerning betavoltaic energy conversion appears to have
occurred in a program led by the author at Donald W. Douglas Laboratories, Richland, WA, from
1968 to 1974 (ref. 6 and 7). This effort was based on the use of Pm-147 beta sources combined
with Si n/p cells to produce nuclear batteries that were utilized as power sources for heart
pacemakers. A brief description of this effort is discussed in a subsequent section.

PRINCIPLES OF BETAVOLTAIC ENERGY CONVERSION

The basic entity in a betavoltaic power source consists of a beta-emitting material coupled
to a junction device as depicted in Figure 1. Some of the key aspects of betavoltaic energy
conversion are described by Figure 2. An equivalent circuit for a betavoltaic cell is essentially the
same as that for a solar cell, except that the current source is due to collection of electron-hole
pairs generated by high energy beta particles. The importance one places on the series
resistance Rs and the shunt resistance Rsh are reversed when comparing betavoltaics and
photovoltaics. The value of Rscan be relatively large in the case of betavoltaics since the value of
Jsc will typically be in the range of 1 i_A/sq.cm.to 100 izA/sq.cm., whereas in photovoltaic
applications Jsc is typically in the range of 10 to 40 mNsq.cm..Thus, Rs can be 100 ohms in a
betavoltaic cell and cause a ptoblem. On the other hand, it is important to minimize the shunt
conductance -- that is, maximize the shunt resistance. Since a loss current of 1 IJA may be
significant, it is necessary to utilize devices based on single crystal material. In the remainder of
this section, a synopsis of the theory of betavoltaics will be presented, and then utilized to
calculate the maximum efficiency of betavoltaic power sources versus semiconductor band gap.

The current supplied to a load by a betavoltaJc cell is given by

J = Jsc- JLoss(V) (1)

Jscis the short circuit current and JLoss is the loss current given by

JLoss = Jo exp(qV/kT) + Jor exp(qV/2kT) + Jot exp(BV) + V/Rsh (2)

257
where the loss terms refer to minority carder injection, depletion layer recombination, tunneling
and current loss through the effective shunt resistance. Since betavoltaic power sources typically
provide low current, the dominant loss mechanism is typically tunneling or depletion layer
recombination.

The short-circuit current is given by

Jsc = (l-r) O Jrmv (3)

where 'r' is the reflection coefficient for beta particles from the semionductor surface, Q is the
collection efficiency and JmaxiS the maximum possible current. The beta particle reflection
depends primarily on the atomic number of the semiconductor. The collection efficiency is the
fraction of electron-hole (EH) pairs collected as current relative to the total number of EH created
by the beta particle flux that enters the semiconductor device. Since the decrease of beta particle
flux within the semiconductor is proportional to exp(-ax), where a is the absorption coefficient,
analytical expressions for Q for a given device structure are essentially the same as those derived
for solar cells. In particular, the beta flux passing through a material can be written as

N(x) = (l-r) No exlM -ax ) (4)

If the beta particles penetrate only on the order of a minority carder diffusion length, then values of
Q can approach 1.0. For example, since betas from Pm-147 only penetrate silicon to a depth of
60 I_m, Q-values can approach 0.8 to 1.0. If Pm-147 is coupled to a direct bandgap material,
however, the collection efficiency will be significantly less since the diffusion length will be much
smaller. Thus the value of Q depends on properties of both the source and the semiconductor.

In order to calculate the maximum efficiency of a system, we must know the maximum
possible current. The key considerations concerning the calculation of Jmax are described in
Figure 3. One can define an effective ionization energy ¢ which is the average amount of energy
expended to create one electron-hole pair. An empirical relationship exists which relates ¢ to
semiconductor bandgap, namely,

= (2.8)Eg+ 0.5 eV (5-)

If N_ and E_ are the incident beta flux and the average beta particle energy, respectively, then
the maximum possible current that one can derive from a betavoltaic device is given by

Jmax = q NI3 ( E_ / ¢) (6)

The maximum power delivered by a cell can be written as

Pmax = JscVoc FF (7)

Once a beta source and device structure are defined, Jmax, r and Q can be calculated. Finally,
Voc and FF can be calculated if the dominant loss current term(s) is identified. The overall
efficiency is defined by

258
 11 = (Pmax/Pin) x 100 (8)
where
Pin = qNoE_
(9)

where No refers to the number of beta particles emitted by the source per second, per square cm.
of device area.

It is convenient to write the overall efficiency as a product of three terms, or efficiencies,

(lO)

(11)

nc = (i<)o (12)

9s = [VocFFI ¢]x100 % (13)

The term 11_ expresses the fraction of all betas created that are actually emitted from the
source and directed towards the device, and is therefore referred to as the beta source effidency.
11 c is a coupling efficiency since it involves properties of both beta source and the
semiconductor device. The term is designated as the semiconductor efficiency, since it
determines the maximum possible efficiency that can be attained with a given semiconductor
coupled to a particular beta source.

The maximum possible efficiency of a given betavoltaic system as a function of bandgap if

one assumes that the semiconductor device is an ideal hornojunction. In this case, the current
loss term is dictated by minority carrier injection. Following Green (ref.8), we estimate that Jo can
be written as

Jo = 1.5 x 105 exp (- F__.g

/ kT) Nsq.cm. (14)

The fill-factor can be accurately calculated as follows (ref.5)

FF = [ Voc - In (vcc + 0.72)] / [Voc + 1] , voc = Voc/kT (15)

Thus, once a given beta source is selected and an ideal device is assumed, the semiconductor
efficiency (11s) becomes a fuction only of bandgap. The potential efficiency of some systems will
be examined after possible beta sources are considered. One can calculate an upper limit to
betavoltaic device efficiency that is independent of the beta source, however. In particular, it can
be shown that

11s _; Eg / £(Eg) (16)

259
This limiting value of ns is plotted in Figure 4 versus bandgap. Due to the functional dependence
of the effective ionization energy, the limiting value of efficiency dses with bandgap and then
levels off at a value slightly over 30 %. Thus, in principle, it is advantageous to utilize large
bandgap devices. One must remember, however, that ideal cell behavior is being assumed.

BETA SOURCE CONSIDERATIONS

The process of selecting a beta emitter involves simultaneous consideration of isotope,

haft-life and the effects of radiation damage of semiconductor devices. To fabricate long lived
power sources, it is clearlly desirable to utilize isotopes with long haft-lives. On the other hand,
since the beta flux derivable from a source material is inversely proportional to the haft-life, the
value of Jmax and thus Pmax are inversely proportional to the haft-life. One must also consider the
beta particle energies relative to the semiconductor radiation damage threshold (Eth). In general,
it is preferable to have the maximum beta particle energy (Emax) less than Eth. Typically, Eth is on
the order of 200 keV to 400 keV. Other key considerations are the availability of the radioisotope
and the potential dose rate that might exist near the power source. Table 1 lists some possible
beta emitters that meet some of the criteria that have been identified. Availability has become a
key issue. The only isotopes that are readily available are tritium and Kr-85. Both are available in
gaseous form, and tritium can be obtained in the form of tritiated 13 foils. If one were interested in
one of the other isotopes, the Department of Energy would need to be consulted.

THEORETICAL EFFICIENCY OF Pm-147 AND NI-63 BETAVOLTAICS

Considerable attention has been given to the use of Pm-147 and Ni-63 in betavoltaic
systems. As noted above Pm-147 fueled batteries were actually reduced to practice. Ni-63 has
been considered in the past because of its long haft-life. The use of both of these isotopes is
hindered because of the complex processes required to generate the isotope. Calculated
efficiencles are considered here because of interest shown in these materials in the past, and for
the purpose of illustration.

Figure 5 gives a plot of theoretical efficiency of Pm-147 fueled devices versus bandgap
assuming ideal semiconductor junctions and bidirectional sources. By bidirectional sources, it is
implied that the beta flux from both sides of a slab of beta emitting material is utilized. Due to the
ideal cell assumption, the efficiency vs bandgap curve has a similar shape as the limiting efficiency
curve given in Figure 4.

Figure 6A abd 6B describe calculated results for Ni-63 fueled cells. The device
efficiencies are much lower in this case because of the beta source efficiency. As a result of the
low beta energy, Ni-63 sources would suffer from effects of self absorption. The low values of
current and power are results of the long half-life and low beta particle energy. Similar results are
obtained when one considers properties of tritium fueled betavoltaic devices.

260
 Pm-147 FUELED BETAVOLTAIC BATTERIES

The author led a program to develop Pm-147 fuelled betavoltaic batteries at the Donald
W. Douglas Laboratories, Richland, WA, from 1968 to 1974. Pm-147 was available in the form of
Pm203 from the U.S. Government. Custom made silicon cells were obtained from Heliotek ( now
Spectrolab) and from Centralab ( now ASEC). The cells had n/p junctions with a mesa around the
device periphery to minimize leakage currents at low voltages. The author benefited from
interactions with Gene Ralph at Heliotek and Peter Isles at Centralab.

The basic approach to battery construction Is illustrated by Figure 7. The n/p cells and
beta sources were stacked in tandem so that the devices were connected in series. The Pm-147
sources actually consisted of Pm203 deposited onto Ta sheet. Thus the sources were
unidirectional. Self-standing bidirectional Pm203 sources were under development when the
program was terminated. Properties of a typical silicon cell coupled to a unidirectional source are
described in Figures 8A and 8B. Batteries were typically designed with 5 mg/cm 2 Pm203
sources. Figure 9 shows a picture of three of the batteries that were made in reasonable
quantities. Their properties are summarized in Table 2. These batteries were referred to as
Betacel batteries.and were nominally 2 % efficient (overall efficiency). With bidirectional sources,
they would have had efficiencies of 4 %. The Model 400 Betacel was considered seriously for
powering heart pacemakers by companies in the United States and Germany. The short-circuit
current and maximum power versus time for a typical Model 400 Betacel are plotted in Figure 10.
Since the power required by the pacemaker circuitry was approximately 10 i_watts, the potential
lifetime was ten years. Over 100 people received Betacel powered pacemakers, and many of the
units lasted 10 years. Although the potential use of Betacel batteries for pacemakers appeared
very promising, the lithium battery was developed about the same time. Lithium batteries lasted
only 7 years, but since they were non-nuclear they were preferred by the pacemaker industry.
The Betacel batteries were also utilized to a limited extent for military purposes.

CONCLUSIONS

Interest in the use of betavoltaic energy conversion seems to 'pop up' every few years.
When the right application emerges, it may finally be utilized extensively. Many more choices are
now available for the semiconductor cell than were available when the Pm-147 fuelled batteries
were developed. Unfortunately, the choice of beta emitting material would appear to be more
limited. To place the potential use of betavoltaic power sources in perspective, it is useful to
estimate the power density versus time for some possible advanced systems. Figure 11
descnbes results of calculated properties of some advanced concepts. The thin film AIGaAs cells
are assumed to be self standing devices. There are many other devices that one could consider.
For example, GaP with an indirect bandgap, and thus potentially long diffusion length, would
certainly be of interest for coupling to Pm--147. inP with its radiation resistant properties could be
interesting for coupling to high energy beta emitters such as TI-204. Nevertheless, the power
density curves shown in Figure 11 can be used to make a few key points. Betavoltaic sytems
should only be considered for low power applications. For example, if one is intersted in power
levels on the order of one watt for ten years, it is clear that on the order of 1000 cm 3 of Pm-147
fueled devices must be considered. The size may not be a problem, but the cost might be

261
 prohibitive.If oneconsiders
anapplication
forwhich1 miliwattor 10rnicrowatts
arerequired,
tritiumor Pm-147fueledsystemsseemreasonable.

REFERENCES

1. W. Ehrenberg,
etal.,=The Electron Voltaic Effect," Proc. Roy. Soc. 64,424(1951).
2. P.Rappaport, "l'he Electron-Voltaic Effect in p-n Junctions Induced by Beta Particle
Bombardment," Phys. Rev. 93, 246(1953).
3. W.G. Plan and W. Van Roosbroeck, "Radioactive and Photoelectric p-n Junction Power
Sources," J.Appl. Phys. 25, 1422 (1954).
4. P. Rappaport, J. J. Loferski and E. G. linder, "The Electron-Voltaic Effect in Germanium and
Silicon P-N Junctions," RCA Rev. 17. 100 (1956).
5. H. Flicker, J. J. Loferski and T. S. Elleman, "Constructionof a Promethium-147 Atomic
Battery," IEEE Trans, ED-11.2 (1964).
6. L.C. Olsen, "Betavoltaic Energy Conversion," Energy Conversion 13, 117 (1973).
7. L.C. Olsen, "Advanced Betavoltaic Power Sources," Proc. 9th Intersociety Energy
Conversion Engineering Conference, page 754 (19741.
8. Martin A. Green,'Solar Cells," Prentice-Hall, Inc., 1982, Chapter 5.

TABLE 1 -- POSSIBLE BETA SOURCES TABLE 2 -- BETACEL CHARACTERISTICS

Model Model Model
Maximum
Characlerislic 50 200 400
Ema x r 1 /2 Jsc for $t
ISOTOPE (MeV) (yr) (arbltary units) Performance characteristicsi'
Maximum power (_W) 50 200 400
Volta8e at maximum power (3/) 113 3" 3 4- 0
H3 0.018 12,3 3 x 10 -3 Open circuit voltage ('V) I "7 4-7 4.9
Short circuit current (_A) 45 72 ! 12
Curies pm14V 12 73 66
N163 0.067 92 10 -3 El_ciency (_) 1 '0 0.7 i .7
Physical characteristics
Diameter (cm) I- 52 2.03 2 29
Pm147 0.230 2.62 1 (in) 0.60 0.80 0.90
Overall height (cm) 1-02 1.65 2.44
(in) 0.40 0.65 0.96
T1TM 0.765 3.75 17 Mass (i) 17 55 98
Radiation dose rate at 2' 5 cm
from battery center_
Kr 85 0.670 10.9 28
BOL (mrem/h) 2" 3 8.1 6 _I
AI end of 5 years (totem/h) 1.2 4-3 3 3
5 year time averai_:l (torero/h) 1' 7 5-9 4.4

p-LAYER

EMITTER

n-LAYER BETA

.V/_- BETAFLUX

Figure 1. Basic Approach To Betavoltaic Energy Conversion.

262
 I

BETA EMrFrER. ,_
_1////1111/1111//I/11_ JLOSS_
JSC

,/- V///////////_I/I/////////II////ZJAI
l
BETA EMrl-rER /_"

PMAX_
J = I/CELL AREA

J = Jsc - JLOSS(v)
PMAX = FF • JSC • VOC

Voc

CIRCUIT MODEL

..._.----,,_ I
• High energy electrons (beta rays) produce
electron/hole pairs in semiconductorcell

Isc l _'V
• Diode characteristics, JLOSS (V), of junction
determine the current. I. to external load

ILOSS • JLOSS (V) must be small in order to produce

useful power

Figure 2. Betavoltaic Principles.

HOMOGENEOUS SEMICONDUCTOR
Electron-hole pairs

NEH - (No electron-hole pairs/cm2/sec)

-__-,

E -- Effective ionization energy

= (2.8) Eg + 0.5 eV

N,B - Beta flux entering junction device

P/N JUNCTION

Maximum current
J
Jmax = Imax/cell area

EI

Figure 3. Considerations For Maximum Current.

263
 )iarnond

sic
GaP
e zo

10

o J ] , I ,
0 2 4 6 8

E9 (eV)

Figure 4. Limiting Betavoltaic EfficzencyVersus Semiconductor Bandgap.

401
' I 1 I ' I '

30 -

>-
z Q-I.O

IJ.. 20 - I//_///__ --
...I ond
._J

L,z.I

i0 L_aP SiC

Y Si
, I 1 J i 1 I
O0 2 4 6 8

Eg (eV)

Figure 5. Calculated Efficiency For Pm-147 Betavoltaic Systems Versus Bandgap.

264
 08 4.0 O.R

0.6
0.6
Jsc / 3.0 X

2 >"
U
Z

0.4 2.o _ _ o.4 IxJ

m
o
u.
0
f,_
f#}
GaAs U.
LLI
.-}

0.2 1• 0.2

S/i ! t

• I L I o.o oo 1 2 3 4
O00 I 2 3 4 O
Ee (oV)
EO(eV)

A_s,mptions: 1. Ideal Diode Characteristics

2. Two-Sided Beta Source With 100% Nickel 63

Figure 6. Calculated Properties Of Ni-63 Betavoltaic Systems Versus Bandgap.

// DIE'VIOE

STAINLESS STEEL
PROTECTIVE SHIELD POSITIVE TERMINAL r"
u : ........... _
/
• ........... I

/
_ LECTRON-BEAM

WELD
CERAMIC
_i INSULATOR
_rc_v/ALI_ __t STAINLESS-STEEL
_,_j_ OUTER HOUSING

F,_ _ " TANTALUM PRIMARY

_I_ _" HOUSING

ur ta_ n_-r

Figure 7. Approach Used For Betacel Construction.

265
 OVERALL EFFICIENCY AND POWER

I°°[____T_2s VS Pm203 THICKNESS

i i_,__

-I t _.
3/,, , I I ,

__ MODEL 400
I I , I''

_
I , i I I , 1,_60

50 N_

rl • 1% TIVI
/ '" 0 d:

o I , 0 0 '''
0.] 0.2 0.3 0.4 0.5 5 l0 15 20

V Ivoltsl Pm203 THICKNESS IMGICM2)

Figure 8. Betavoltaic Properties Of Silicon n/p Cells Coupled To Pm-147 Sources.

Figure 9. Betacel Batteries Developed At Donald W. Douglas Laboratories. Left to Right:

Model 200, Model 400 and Model 50.

266
 MODEL
400PMAXANDISCVSTIME ....

'°k""''"'l

l 1 L___L___ I I I l ____L__._
0 l 2 3 4 5 6 7 8 9 10

TIME (YR!

Figure10. Short Circuit Current and Maximum Power Versus Time For a Model 400 Battery.

2
• 10

tO0 I " | " | • ! • e - e " | " ! " ! " |

147
1
,Pm I 201xm
10 4 10
Two Cell GaAa Stack

147
Pm I 2 mll Silicon
E ¢,$

10 3 E

E
v
>.
I-

.'z 10 2 10 "1 r_
U)
LJl IH 31 Thin Film AIGcAs Z
¢3 uJ
& Two Sided Source O
rr
UJ
-2 rr
U/
101 10
o
o.
3 o
on SS. I GaAo on Ge a.

-3
10

-4
10
10"
0 2 4 $ e 10 t2 14 1E 18 2O

YEARS

Figure 11. Power Density Versus "l'ime For Advanced Betavoltaic Concepts.

267