The thermal atomic battery is any device that converts the heat emitted by radioactive isotopes to electricity. Like nuclear reactor, the power generated by thermal atomic battery is ultimately derived from atomic energy. However, atomic battery relies solely on the spontaneous radioactive decay of atomic nucleus, rather than artificially-triggered nuclear fusion or fission in nuclear reactor. Although very costly, atomic battery has extremely long life span and high energy density compared to chemical battery. Therefore, they are usually used in situations requiring long operation without battery replacement or recharging, such as spacecraft, unmanned scientific facilities, pacemaker and etc. The atomic battery can be categorized by the form of energy converter and radioisotopes that it uses. The rest part of this report will elaborate the technical details of different forms of atomic battery.
The thermal atomic battery converts the atomic energy into heat first and then electricity. While thermal-to-electric conversion techniques has been studied extensively, there are many of them are available to be generalized to heat source powered by radioisotopes. To date, thermal converters include following forms, thermionic converter, radioisotope thermoelectric generator, thermophotovoltaic cell, alkali-metal thermal to electric converter and Stirling radioisotope generator.
The conversion of atomic energy to heat is quite simple. In thermal conversion atomic battery, the radioisotope, called fuel, is placed in a container. Alpha particles generated by alpha decay or beta particles generated by beta decay can easily interact with atoms of shielding materials and lose energy. This part of energy is dissipated in the form of heat. Therefore, the container and the radioisotope itself are used as heat source in thermal conversion atomic battery.
In thermionic converter, the heat generated from radioactive decay is used to heat a hot electrode to emit electron through thermionic emission at temperature 1500-2000 °K.  The emitted electron is collected by a cold electrode.  Plasma, usually consisting of Cs vapor, is maintained between the two electrodes to reduce the work needed for electron emission, magnify the currency, which is favorable to increase efficiency and modify the electron conducting property between electrodes. [2-4]. The efficiency can also be increased by lowering the potential difference between the top of the potential barrier in the interelectrode space and the Fermi level of the anode.  However, the efficiency of the thermionic converter cannot exceed 90% of Carnot efficiency.  In practice, the efficiency of thermionic converter can be close to 20%. 
Radioisotope thermoelectric generator makes use of Seebeck effect to directly transfer the temperature difference between heat source and heat sink to electricity. Its structure is quite simple. The hot end of a thermopile is attached to the heat source, and its cold end is attached to the heat sink, usually the ambient temperature. Thermopile for power generation is usually made of pairs of connected P-type and N-type semiconductors. In presence of temperature difference, the velocity of charge carrier in both semiconductors is different, which forms current. Due to its simplicity, radioisotope thermoelectric generator is very reliable and can be made very small, and thus widely used in spacecraft.  However, the efficiency of this converter is very low, about 7% in practical use.  One possible way to improve efficiency is to hybridize the system with other converter. The working temperature of a typical radioisotope thermoelectric generator is much lower than that of thermionic converter. The hot end temperature is 811 °K while the cold end temperature is 394 °K. 
Thermophotovoltaic cell is similar to other photovoltaic cell in that it converts the energy of photons generated by thermal emission of the heat source into electricity. Wien's Law gives out the relation between wavelength with peak light flux and the temperature as
where b is Wien's displacement constant and T is temperature. Considering that a reasonable temperature of radioisotope heater unit is around 1200K-1500 °K, the peak wavelength of the thermal emission is within the infrared light spectrum and the band gap of the photovoltaic cell should be smaller than solar cell that mainly uses visible light for energy conversion. Several thermophotovoltaic techniques have been proposed with reported efficiency up to 23%.  These pilot projects are encouraging to result in higher conversion efficiency than current state-of-the-art level.
The alkali-metal thermal to electric converter is a relatively young technique. This converter is basically a concentration cell with sodium vapor at 600-1000 °K and solid sodium at 100-500 °K as electrode as well as solid beta-alumina as electrolyte.  The reaction in the cell absorbs heat. Thus additional heat is required to maintain the cell at operating temperature. Some models have been reported with efficiency of 14-18% while the theoretical efficiency can reach 20%-40%. 
The Stirling radioisotope generator converts the heat from radioisotope to dynamic motion by a Stirling engine and then a generator converts the motion to electricity. A Stirling engine is a heat engine operating by cyclic compression and expansion of working fluid at different temperature levels. It is classified as an external combustion engine since the heat transfer take place through the engine wall and there is no mass exchange of working fluid with outside of engine, which is perfect for external heat source as radioisotope heater unit. NASA, DoE and Lockheed Martin have jointly launched project to develop this type of converter for the use of space craft.  In this project, helium is used as working fluid and efficiency as high as 23% has been achieved in experiments. The greatest challenge to date of Stirling generator is its reliability. Compared to other form of converter, it has moving parts and its vibration may harm other facilities within the spacecraft.  New technology is needed to overcome these disadvantages.
There are several criteria in selecting the radioisotope fuel. Firstly, the radioisotope should produce high energy decay. Alpha particle has a typical dynamic energy of 5 MeV while the figure of beta particle is typically 1 MeV. Secondly, the radiation must be easily absorbed and converted into heat. In general, alpha radiation has most significant heat effect; beta radiation can give off considerable amounts of gamma radiation through secondary radiation; gamma neutron radiation can easily penetrate common shielding material. Therefore, radioisotope of the type of alpha decay is favorable. Thirdly, the half-life of radioisotope must be long enough to release energy at a relatively constant rate for a reasonably long time. The life span of a thermal atomic battery is usually expected to be decades. Therefore, the half-life of radioisotope must be at least several decades. In the case of spacecraft, the energy density of radioisotope must be high to reduce the load of spacecraft.
No more than 30 radioisotopes satisfy the first two criteria. Currently, Pu-238, Cm-244 and Sr-90 are most widely used radioisotopes. Other isotopes such as Po-210, Pm-147, Cs-137, Ce-144, Ru-106, Co-60, Cm-242 have also been studied for this use.
Within these candidates, Pu-238 has the lowest shielding requirement and longest half-life. It needs less than 2.5mm of lead shielding to screen radiation. As a fuel for thermal atomic battery, it usually takes the form of PuO2 with a half-life of 87.7 years. Sr-90 has much lower energy density and produces gamma radiation. However, it is still used in some cases since it is relatively cheap. Am-241 has a half-life of 432 years but only 1/4 of energy density of Pu-238 and requires at least 18 mm lead shielding to screen the penetrating radiation it produced.
Thermal atomic battery is of great interests to the exploration of outer space. Thermalelectric converter has been widely used in many situations but suffers from low efficiency. Other techniques have been also proposed with much higher efficiency. Within them, Stirling converter is the closest one to practical usage. The isotopes that qualify for fuel in atomic battery are quite limited. Production of suitable radioisotope should be increased to meet the growing need to power the space missions.
© Wenshi Chen. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 G. Griaznov et al., "Thermoemission Reactor-Converters for Nuclear Power Units in Outer Space," Atomnaya Energiya, 66, 371 (1989).
 N. Rasor and C. Warner, "Correlation of Emission Processes for Adsorbed Alkali Films on Metal Surfaces," J. Appl. Phys. 35, 2589 (1964).
 J. Houston, "Theoretical Efficiency of the Thermionic Energy Converter," J. Appl. Phys. 30, 481 (1959)
 N. Rasor, "Thermionic Energy Conversion Plasmas," IEEE Trans. Plasma Sci. 19, 1191 (1991).
 J. Ingold, "Calculation of the Maximum Efficiency of the Thermionic Converter," J. Appl. Phys. 32, 769 (1961).
 V. Wilson, "Conversion of Heat to Electricity by Thermionic Emission," J. Appl. Phys. 30, 475 (1959).
 D. Anderson et al., "An Overview and Status of NASA's Radioisotope Power Conversion Technology NRA," U.S. National Aeronautics and Space Administration, NASA/TM-2005-213980, November 2005.
 D. J. Anderson, "NASA Radioisotope Power Conversion Technology NRA Overview," U.S. National Aeronautics and Space Administration, NASA/TM-2005-213981, November 2005.
 H. Nakata et al,"Ceramic Electrodes for an Alkali Metal Thermo-Electric Converter (AMTEC)" J. Appl. Electrochem. 23, 1251 (1993).
 S.-Y. Wu et al., "A Review on Advances in Alkalai Metal Thermal to Electric Converters," Int. J. Energy Res. 33 868 (2009).
 L. G. Thieme and J. G. Schreiber, "NASA GRC Stirling Technology Development Overview," AIP Conference Proc. 654, 613 (2003).