Al0.52In0.48P 55Fe x-ray-photovoltaic battery

An Al0.52In0.48P 55Fe radioisotope microbattery is demonstrated over the temperature range  −20 °C to 160 °C. Al0.52In0.48P p+-i-n+ mesa structures were used to collect the photons from a 238 MBq 55Fe radioisotope x-ray source. The effects of temperature on the key microbattery parameters were studied. Increasing the temperature, the saturation current increased; whilst the open circuit voltage, the maximum power and the conversion efficiency decreased. An open circuit voltage of 0.97 V and a conversion efficiency of 22% were measured in a single p+-i-n+ mesa structure at  −20 °C. The highest total microbattery maximum output power using two mesa structures was 1.2 pW at  −20 °C.


Introduction
The ability to supply small amounts of power over long periods of time is becoming increasingly important in many applications including microelectromechanical system technologies [1], biomedical [2] and aerospace applications [3]. Nuclear microbatteries, converting nuclear energy to electrical energy [4], could be used in these scenarios. In a nuclear microbattery, radioactive sources are coupled to conversion materials: the conversion material absorbs the high-energy particles emitted by the radioactive source generating electrical current. In contrast with traditional power supplies, such as chemical batteries, nuclear microbatteries offer important characteristics such as high energy density, stability, and long life. Wide bandgap semiconductors (e.g. diamond, SiC, GaAs) can be used as conversion materials in this type of system. Because of their wide bandgap, they present higher efficiency than alternative narrower bandgap materials such as silicon; this is due to the linear dependence of the conversion efficiency with bandgap [5]. Moreover, they could be advantageous in all the applications where the environment temperature can vary significantly during use as they are able to work over a broad range of temperature offering good electron mobility and low leakage current [6][7][8][9]. Recently, different alpha, beta and x-ray microbatteries have been proven [5,[10][11][12][13][14][15][16]. Alpha and beta emitters have received research interest because they have higher specific energy per Curie compared to x-ray emitters (e.g. 0.017 µW · g (Ci · cm 2 ) −1 for 55 Fe and 0.3 µW · g (Ci · cm 2 ) −1 for the beta emitter 147 Pm [4]), thus offering a higher output power. The advantages of alpha and beta microbatteries in future NASA mission are reported by Landis et al [14]; a critical issue for alphavoltaic converters is radiation damage by the particles, this is minimised in phosphorus-containing III-V compounds such as InGaP, 210 Po-InGaP alpha-voltaic microbattery have been successfully proven and analysed in [14]. Bormashov et al [10] demonstrated a 63 Ni-diamond beta-voltaic microbattery and a 238 Pu-diamond alpha-voltaic microbattery with conversion efficiencies as high as 0.6% and 3.6%, respectively, at room temperature. Chandrashekhar et al proved a 63 Ni-SiC microbattery with at least 6% efficiency [5], whilst Eiting et al a 33 P-SiC microbattery with 4.5% efficiency [17]. Cheng et al [15] reported a high open circuit voltage (1.64 V) 63 Ni-GaN beta-voltaic microbattery with conversion efficiency of 0.98%, at room temperature. Betavoltaic cells were also proven by Sharma et al [16] using P3HT semiconductive conjugated polymer; conversion efficiency as high as 0.78 was achieved at an e-beam energy of 10 keV. X-ray emitters have been also studied with particular attention; the electron capture x-ray emitter 55 Fe has the advantage of reduced device damage risk due to the low energy photons emitted, can be easily shielded providing safer working conditions, and is readily commercially available. Butera et al [13] reported a GaAs 55 Fe radioisotope x-ray microbattery working over the temperature range −20 °C to 70 °C, with a maximum output power of 1 pW (corresponding to 0.4 µW/Ci) and conversion efficiency of 9% at −20 °C. A III-V wide bandgap ternary compound that could be very useful as the conversion layer in a nuclear microbattery is Al 0.52 In 0.48 P. Al 0.52 In 0.48 P has an indirect bandgap of ~2.3 eV [18,19], and it is lattice matched with GaAs. Al 0.52 In 0.48 P is widely used in semiconductor optoelectronics and the crystalline quality of the lattice matched Al 0.52 In 0.48 P can be very high in comparison to III-V nitrides, IV and II-VI compounds of a similar bandgap. The doping in Al 0.52 In 0.48 P is also easier to control than in some II-VI semiconductors. In this paper for the first time a Al 0.52 In 0.48 P 55 Fe radioisotope microbattery is demonstrated. A p + -i-n + Al 0.52 In 0.48 P structure was used to collect the photons from a 238 MBq 55 Fe radioisotope x-ray source; the changes in the key microbattery parameters were studied over the temperature range −20 °C to 160 °C.

Device structure
Two 400 µm diameter unpassivated p + -i-n + Al 0.52 In 0.48 P mesa photodiodes (D1 and D2), located on the same chip, were illuminated by a 238 MBq 55 Fe radioisotope x-ray source (Mn Kα = 5.9 keV, Mn Kβ = 6.49 keV). The x-ray emitter was 5 mm away from top surface of the detectors. The Al 0.52 In 0.48 P epilayer of the devices was grown by metalorganic vapour phase epitaxy (MOVPE) on a commercial (1 0 0) n-GaAs: Si substrate with a misorientation of 10 degrees towards <1 1 1> A to suppress the CuPt-like ordered phase. The doping concentrations of the Al 0.52 In 0.48 P p + and n + layers were 5 × 10 17 cm −3 and 2 × 10 18 cm −3 , respectively. The layers' thicknesses were 0.2 µm for the p + -region, 2 µm for the i-region and 0.1 µm for the n + -region. After growth, the wafer was processed to form mesa structures using 1  Figure 1 shows a schematic cross sectional view of the Al 0.52 In 0.48 P mesa structure.
The x-ray quantum efficiency (QE) of a semiconductor detector is defined as the percentage of x-ray photons absorbed by the photodetector. Using the Beer-Lambert law and assuming complete charge collection in the active p-and i-layers, x-ray quantum efficiencies (QE) of 20% and 16% were calculated for the device for 5.9 keV and 6.49 keV photons, respectively, considering the GaAs dead region. The QE values were calculated, according equation (1), taking into account that 33% of the device surface was covered by the top metal contact.  [20,21]; whilst the GaAs, Ti and Au attenuation coefficients at 5.9 keV and 6.49 keV were taken after [20,22]. InGe 0.02

Results and discussion
The Al 0.52 In 0.48 P 55 Fe radioisotope microbattery was investigated over the temperature range −20 °C to 160 °C using a TAS Micro MT climatic cabinet with a dry nitrogen atmosphere (relative humidity <5%). At each analysed temperature, dark and illuminated current characteristics of each Al 0.52 In 0.48 P p + -i-n + photodiode (D1 and D2) were measured as functions of applied bias. Forward bias measurements from 0 V to 1.6 V were made in 0.01 V increments using a Keithley 6487 picoammeter/voltage source. The uncertainty associated with the current readings was 0.3% of their values plus 400 fA, while the uncertainty associated with the applied biases was 0.1% of their values plus 1 mV [23]. Figure 2 shows typical dark current characteristics as a function of forward bias at different temperatures.
In both the analysed Al 0.52 In 0.48 P photodiodes, the dark currents increased as a function of applied bias, this was due to the greater electric fields across the i-region in each Al 0.52 In 0.48 P structure. At high temperatures, the dark currents through the devices also increased due to the higher thermal energy available. At each temperature, a linear least square fit on the dark current data as a function of applied bias was used to extrapolate the saturation current (I 0 ) in both detectors [11,13]. Figure 3 shows the measured relationship between the logarithm of the saturation current and the temperature for D1. Similar results were obtained for D2.
The magnitude of the observed natural logarithm of the satur ation current decreased at higher temperatures. From −20 °C to 160 °C this decrease was 22.19 ± 0.13 for D1 and 22.18 ± 0.10 for D2. These values were in remarkable agreement with the expected decrease 22.27. The expected decrease was calculated as per [13], with using the assumption that the temperature dependence of the saturation current was proportional to exp(−E g /2kT) [24]. Figure 4 shows typical dark and illuminated current characteristics as functions of applied bias for D1 at 20 °C. Similar results were obtained for D2 at the same temperature.
The dark current increased exponentially as a function of applied bias (∝ exp{qV/nkT} where n is the ideality factor, k is the Boltzmann constant and T is the temperature) [6]. For both diodes, a linear least squares fit of the data showed that the natural logarithm of the dark current was linearly dependent on the applied forward bias. The gradients determined by the linear least squares fits were (22.23 ± 0.10) AV −1 and (22.36 ± 0.07) AV −1 for D1 and D2, respectively. Ideality factors as high as 1.779 ± 0.008 and 1.769 ± 0.006    were estimated for D1 and D2, respectively. These ideality factors (close to 2) indicated that generation-recombination currents were dominant over the diffusion currents in the devices. Under the illumination of the 55 Fe radioisotope x-ray source, the measured current through the devices (empty circles in figure 4) increased as photocurrent is added to the dark current. X-ray photons were absorbed by the detector creating electron-hole pairs as a consequence of the photoelectric effect. Electrons and holes in the device depletion region were, thus, swept to the p + -type and n + -type regions, respectively, generating in the device the observed photocurrent. In figure 4, the decrease in photocurrent, observed between 0 V and 0.8 V, was smaller than its uncertainty (0.08 ± 0.8) pA.
Under the illumination of the 55 Fe radioisotope x-ray source, the currents through both devices were measured over the temperature range −20 °C to 160 °C. Figure 5 shows typical current characteristics as a function of forward bias at different temperatures for D1. Similar results were obtained for D2.
At increased temperature, the softness in the knee of the measured current as a function of applied forward bias decreased in both photodiodes. The experimental values of the open circuit voltage (V OC ) were obtained as the interception point of the curves in figure 5 on the horizontal axis. Figure 6(a) shows the V OC as a function of temperature for D1; a similar behaviour was found also for D2. At increased temperature, the open circuit voltage decreased because of its dependent on the saturation current (I 0 ) [6]. V OC increased logarithmically with decreasing I 0 , according equation (2): where k is the Boltzmann constant, T is the temperature, I ph is the photocurrent through the device and I 0 the saturation current [6]. Figure 6   for D2. V OC as high as 0.97 V and 0.96 V were observed for D1 and D2, respectively, at −20 °C. These values are much higher than has been previously reported (0.3 V) using a GaAs 55 Fe radioisotope microbattery at the same temperature [13]. This is due to the higher bandgap of Al 0.52 In 0.48 P with respect to GaAs (at room temper ature, the bandgap of Al 0.52 In 0.48 P and GaAs are ~2.3 eV [19] and ~1.42 eV [6], respectively).
The experimental values of the short circuit current (I SC ) were obtained as the interception point of the curves in figure 5 on vertical axis. Figure 7 shows the I SC as a function of temperature for D1, similar results were found for D2.
An unclearly defined relationship was observed for the short circuit current with temperature in the Al 0.52 In 0.48 P 55 Fe radioisotope microbattery, this was due to the relatively high uncertainty of the picoammeter in measuring low currents through the photodiodes positioned in the climatic cabinet system.
The output power from each Al 0.52 In 0.48 P photodiodes was computed by multiplying the applied bias with the correspondent current measured through the device. When the bias was increased, the output power increased to a maximum, P m , and then decreased. Figure 8 shows the output power from D1, similar results were obtained for D2.
The magnitude of the measured maximum output power as a function of temperature is reported in figure 9(a). The magnitude of the measured maximum output power as a function of temperature is reported in figure 9(a). The conversion efficiency (η) was computed by dividing the maximum output power measured at each temperature by the maximum power (P th ) obtainable from the x-ray photons usefully absorbed by the device if the device conversion efficiency was 100%. P th was estimated knowing the activity of the source, the emission probabilities of Mn Kα and Mn Kβ x-rays from 55 Fe (0.245 and 0.0338, respectively [25]), the thickness of the radioisotope x-ray source's Be window (0.25 mm), the areas of the 55 Fe radioactive source and Al 0.52 In 0.48 P mesa device, the Al 0.52 In 0.48 P device QE (calculated in section 2) and assuming an electron-hole pair creation energy of 5.8 eV (2.5 times the bandgap). P th was found to be 2.8 pW. Figure 9(b) shows the efficiency as a function of temperature for D1. A conversion efficiency as high as 22% was observed at −20 °C for D1, similar results were found for D2 20%.
At increased temperature, the magnitude of the maximum output power decreased because of its dependent on the open circuit voltage [6]. A maximimum output power as high as 0.62 pW, corresponding to 0.3 µW/Ci, was observed from the best Al 0.52 In 0.48 P detector at −20 °C. This value is lower than the maximum output power observed by Butera et al [13] using a GaAs 55 Fe radioisotope microbattery, at the same temper ature. This is mainly due to the smaller number of useful photons per second absorbed in the Al 0.52 In 0.48 P detector (0.02 × 10 6 s −1 ) with respect to the GaAs device (0.05 × 10 6 s −1 ): the i-layer thickness of the Al 0.52 In 0.48 P photodiode was 2 µm, whilst the i-layer thickness of the GaAs device was 10 µm. An advantage of the reported Al 0.52 In 0.48 P 55 Fe radioisotope microbattery is that the output powers of the two 400 µm diameter Al 0.52 In 0.48 P devices can be combined, resulting in a total microbattery output power of 1.2 pW achieved at −20 °C. Table 2 reports the maximum output powers at some significant temperatures, for both the Al 0.52 In 0.48 P detectors studied.
The microbattery's maximum output power may be improved in the next prototype generation increasing the i-layer thickness of the Al 0.52 In 0.48 P photodetectors so that more photons are absorbed. A better microbattery system design could be also advantageous: in the reported design, only 0.3% of the emitted photons impinged on the surface of the Al 0.52 In 0.48 P devices and only 0.05% were actually absorbed in the devices. The number of photons per second emitted in any direction by the source was estimated knowing the activity of the source and the emission probabilities of Mn Kα and Mn Kβ x-rays   [25]); it was found that 6.6 × 10 7 photons s −1 are emitted by the 55 Fe radioactive source. Of these 6.6 × 10 7 photons s −1 , only half are emitted in the direction of the devices (we assumed that half of the photons were lost because emitted up). The number of photons per second on the devices (1.7 × 10 5 s −1 ) was estimated knowing the number of photons per second emitted by the source towards the devices (3.3 × 10 7 s −1 ), the thickness of the radioisotope x-ray source's Be window (0.25 mm) and the geometry of the source and detectors. Figure 10 shows schematically the geometry of the source and detector; the ratio between the area of the devices (0.25 mm 2 ) and the area of the radioactive 55 Fe source (28.27 mm 2 ) was calculated to be 0.0089. The number of photons on the detector was estimated by multiplying 0.0089 for the number of photons per seconds transmitted trough the x-ray source's Be window (1.9 × 10 7 s −1 ). The number of photons per seconds absorbed in the devices was calculated using the Beer-Lambert law and assuming complete (100%) charge collection in the active p-and i-layers: the number of photons per second on the devices was multiplied for the device QE (calculated according equation (1)) so that 3.4 × 10 4 photons s −1 were found to be absorbed in the devices.

Conclusion
In this paper for the first time an Al 0.52 In 0.48 P 55 Fe radioisotope microbattery is reported: a 238 MBq 55 Fe radioisotope x-ray source was coupled to p + -i-n + Al 0.52 In 0.48 P mesa photodiodes to achieve the conversion of nuclear energy into electrical energy. Al 0.52 In 0.48 P is lattice matched with GaAs and is commonly used in semiconductor optoelectronics. The crystalline quality of the lattice matched Al 0.52 In 0.48 P can be very high in comparison to III-V nitrides, IV and II-VI compounds of a similar bandgap. The doping in Al 0.52 In 0.48 P is also easier to control than in some II-VI semiconductors. These characteristics could provide benefits in the development of microbatteries in many applications. The reported Al 0.52 In 0.48 P 55 Fe radioisotope microbattery was characterised over the temperature range −20 °C to 160 °C. Increasing the temperature, the saturation current increased; whilst the open circuit voltage, the maximum power and the conversion efficiency decreased. An open circuit voltage as high as 0.97 V was measured in a single Al 0.52 In 0.48 P p + -i-n + mesa structure at −20 °C. A conversion efficiency of 22% was observed at −20 °C, taking into account attenuation from contacts and dead layer. The highest total microbattery maximum output power, obtained using the combined output powers from the two analysed mesa structures, was 1.2 pW at −20 °C; this value can be improved in the next prototype generation with a better system design. In the reported design, many of the photons emitted by the 55 Fe radioisotope x-ray source were lost: only 0.3% impinged the surface of the Al 0.52 In 0.48 P devices and only 0.05% were actually absorbed in the devices.