Laser energy transmission as a potential technology of wireless energy transmission has attracted the attention of researchers in recent years. The laser power converter (LPC) is the core component of a laser energy transmission system and its conversion efficiency restricts the overall efficiency of the wireless energy transmission system. However, most of the research efforts on LPCs have focused on GaAs LPCs with a cut-off wavelength of 875 nm^[1-4], and small-area GaAs cells with conversion efficiencies of more than 60% have been reported^[4-7]. Due to the availability of high power and small beam divergence of 1064 nm lasers, the development of high efficiency 1064 nm LPC becomes meaningful to wireless energy transfer over long distances for both moving and stationary targets. A 1064 nm laser power conversion efficiency of 39.4% based on Si cells with a complicated pyramid-textured surface was achieved at a power density of 58 mW/cm², and it was reduced to 38.8% at 1.3 W/cm^2[8]. A conversion efficiency of 34.5% (1.2 W, λ = 1064 nm) has been demonstrated for the GaInAsP/InP LPCs with an active area of 3.5 × 3.5 mm² at uniform radiation conditions^[9]. AlGaInAs is another material could be used to build LPCs, LPCs of AlGaInAs lattice matched to InP substrate and converting 1070 nm laser energy showed a conversion efficiency of 37.2% at a power density of 0.51 W/cm² with an open-circuit voltage of 0.73 V. However, a large series resistance resulted in a decrease of conversion efficiency with increasing power density^[10]. Meanwhile, metamorphic materials are another alternative for fabricating LPCs, and 1064 nm LPCs of In_0.24Ga_0.76As grown on GaAs substrates were developed with an efficiency of 38.5% based on the previously developed metamorphic growth technology^[11-13]. The conversion efficiency decreases monotonously due to the thermal effect with the increase of input laser power^[14]. A MOCVD-grown 1080 nm In_0.24Ga_0.76As LPCs with an aperture of 5.5 × 5.5 mm² has been demonstrated with a conversion efficiency of 37.87%, an open-circuit voltage of 0.63 V, and a fill factor of 0.7571 at a laser power density of 0.51 W/cm^2[15].

The open circuit voltage of a single-junction 1064 nm LPC is some 0.6–0.7 V, while several volts are usually required for power supplies in electronic systems. To make the LPCs practically useful for real situations, two schemes could be used to increase output voltage: the first is to connect multiple single-junction LPCs placed side-by-side in series^[16-18] and the second is to vertically stack sub-cells connected by tunnel junctions to form multi-junction LPCs^[19-23]. In this paper, vertically-stacked triple-junction InGaAsP LPCs for converting the power of a 1064 nm laser are developed. So far, there have been no reports on multi-junction 1064 nm LPCs.

The structures of InGaAsP triple-junction LPCs were grown by MOCVD, and LPCs with an aperture of 1 × 1 cm² were fabricated. A conversion efficiency of 32.6% was achieved at a power density of 1.1 W/cm² under 1064 nm laser irradiation at room temperature with an open-circuit voltage of 2.16 V and a fill factor of 0.74.

The absorbing layer thicknesses of each sub-cell of the triple-junction InGaAsP LPCs are designed based on the rule of current matching and the absorption coefficient of InGaAsP material at 1064 nm. The thicknesses of the sub-cells can be determined using the following equation:

$ f/N = {\rm{exp}}\left( {{{ - }}\alpha \sum\limits_{m - 1}^{n - 1} {{x_{{m}}}} } \right) \times \left[ {1 - \exp \left( { - \alpha {x_{{n}}}} \right)} \right], $  (1)

where α is the absorption coefficient, x_n is the thickness of the absorbing layer of the nth sub-cell from the top, f is the percentage of the light being absorbed by the LPC, N is the number of the sub-cells. The ideal tunnel junction (TJ) connecting the sub-cells is transparent for the incident light. However, achieving both n-type and p-type doping concentration of more than 1 × 10¹⁹ cm^–3 for InP and Al_0.48In_0.52As are difficult. Therefore, InGaAs is chosen to build the tunnel junctions for the triple-junction InGaAsP LPC. The depletion layer thickness of InGaAs TJ with a doping concentration of 1 × 10¹⁹ cm^–3 on both sides is about 16 nm, and the thicknesses on two sides of the InGaAs pn junction are set to 20 nm considering the possible Zn diffusion. Fig. 1 shows the schematic structure of a vertically-stacked triple-junction InGaAsP LPC. Using 1.08 eV InGaAsP lattice matched to InP substrate as the absorption layer to obtain as high as possible voltage, while maintaining reasonable absorption coefficient at 1064 nm, the layer thicknesses of the triple-junction 1064 nm LPC are calculated to be 264, 456, and 2109 nm for cell 1, 2, and 3, respectively, based on the absorption coefficient of 1.4 × 10⁴ cm^–1 at 1064 nm for 1.08 eV InGaAsP^[24] and taking into account of the absorption of incident 1064 nm light by the InGaAs tunnel junctions.

The current–voltage (I–V) characteristic of the LPC was calculated and the loss mechanisms (e.g., series resistance, reflection of surface, and surface recombination) were taken into consideration during the calculation. The I–V characteristic of the multi-junction InGaAsP LPC can be described by the following equation:

$ I = {I_{\rm{ph}}} - {I_{\rm s}}\left[ {\exp \left( {\frac{{V/N + I {R_{\rm s}}}}{{n k T}} \times q} \right) - 1} \right] - \frac{{V/N + I {R_{\rm s}}}}{{{R_{\rm{sh}}}}}, $  (2)

where I is the output current, I_ph the photogenerated current of sub-cell, I_s the reverse saturation current, q the electron charge, V the output voltage, k the Boltzmann’s constant, T the temperature, n the ideality factor of the diode, N the number of sub-cells, R_s the series resistance, and R_sh the shunt resistance of the sub-cell.

Table 1 shows the parameters used for the calculation of the characteristics of the triple-junction InGaAsP LPCs with an aperture of 1 × 1 cm².

Fig. 2 shows calculated I–V and P–V characteristics of the triple-junction InGaAsP LPC at an incident 1064 nm laser power density of 1 W/cm² taking the absorption of the tunnel junctions into account, the short circuit current is 237.5 mA/cm², the open circuit voltage is 2.174 V, the fill factor is 81.9%, and the conversion efficiency is 42.3%.

Prior to the growth of the triple-junction InGaAsP LPC structures, InGaAsP epitaxial layers were grown to calibrate the composition and growth rate. Room temperature photoluminescence (PL) peak at 1153 nm confirms the band gap of InGaAsP of 1.08 eV.

The triple-junction LPC epitaxial structure was grown on p-type InP substrates by an AIXTRON-200/4 MOCVD system using trimethylgallium (TMGa) and trimethylindium (TMIn), as well as arsine (AsH₃) and phosphine (PH₃) as the group III and group V precursors, respectively, and ultra-high purity hydrogen (H₂) as the carrier gas. Diethylzinc (DEZn) and silane (SiH₄) were used as the p-type and n-type doping sources, respectively. Starting from the InP substrate, a 300-nm InP buffer with a Zn concentration of 2 × 10¹⁸ cm⁻³ was grown followed by InGaAsP sub-cells of 2310, 456, and 264 nm thick, respectively, separated by 20-nm n⁺-InGaAs (Si: 2 × 10¹⁹ cm⁻³) /20-nm p⁺-InGaAs (Zn: 2 × 10¹⁹ cm⁻³) tunnel junctions, and finally a 500-nm InP current spreading layer and a 100-nm InGaAs contact layer with a Si concentration of 1 × 10¹⁸ and 2 × 10¹⁹ cm⁻³, respectively, were grown to finish the whole LPC structure. Each pn junction subcell comprised of a p-type and n-type layer doped with Zn and Si to 4 × 10¹⁷ and 2 × 10¹⁸ cm⁻³, respectively, and was sandwiched between a 50-nm InP back field layer doped with Zn of 2 × 10¹⁸ cm⁻³ and a 50-nm InP window layer doped with Si of 2 × 10¹⁸ cm⁻³.

Triple-junction InGaAsP LPC was fabricated by evaporating front grid and back ohmic contact metals, etching of InGaAs contact layer, and finally depositing 85-nm SiO₂/100-nm TiO₂ antireflection dielectric films. The reflectivity of the antireflection coating was measured using a reference sample and shown in Fig. 3, which confirmed the low reflection of 1.8% at 1064 nm. Subsequently, the LPC wafers were sawn into chips of 1 × 1 cm². Silver paste was applied between LPC chips and Cu-plated ceramic heat sinks to improve thermal conductivity during the measurements.

The measurement of laser LPC conversion efficiency was performed at room temperature, and the fiber guided 1064 nm optical beam with a divergence of about 25° perpendicularly illuminated the LPC chip surface with the end of the fiber at a distance to the surface to ensure that the whole light spot is received by the aperture. I–V curves of the triple-junction InGaAsP LPC under different 1064 nm monochromatic power densities of 610, 989, 1100, and 1233 mW/cm² are shown in Fig. 4 and the extracted parameters are shown in Table 2.

It can be seen from the Table 2 that the conversion efficiency of 32.13% at the optical power density of 989 mW/cm² is much lower than the expected conversion efficiency of around 42.3% based on the design. A little bit higher conversion efficiency of 32.6% and higher open circuit voltage of 2.16 V are achieved at 1.1 W/cm². The series resistance and shunt resistance are determined to be 0.22 and 43 Ω of the sub-cell, respectively, by fitting the I–V curve obtained at 1100 mW/cm² using Eq. (2). The low shunt resistance indicates a significant leakage, probably via surface states of the side walls of the cell, which are not passivated by deposition of dielectrics.

To find the reasons why the conversion efficiency is lower than the expected efficiency, the external quantum efficiency (EQE) was measured using a xenon light source passing a monochromator and a standard Si cell as the reference. As shown in Fig. 5, the EQE at 1064 nm is 71.7% and is consistent with the short-circuit current of the triple-junction laser photovoltaic power converter. However, the EQE is much lower than the calculated EQE 84% assuming that the sub-cells are current-matched and the minority carrier lifetime is long enough for them to transport to the depletion layer and be separated by the built-in electric field, and finally collected by the electrodes to contribute to the output current.

There are two main causes that lower the conversion efficiency. The first is that the thicknesses of the sub-cells deviate from the optimal current-matching values due to inappropriate absorption coefficient is taken for the design of triple-junction InGaAsP cells, which leads to a current mismatch between sub-cells and output current is constrained by the lowest one and much smaller than the expected value. The second is that the internal quantum efficiency is lower than the assumed 95% resulting from a shorter carrier diffusion length. For the reference single junction InGaAsP cells with an absorbing layer thickness 3114 nm, the same as the total active layer thickness of the triple-junction InGaAsP cell, the EQE at 1064 nm is more than 80%. This means the diffusion length of the photogenerated minority carriers is larger than the thicknesses of the absorbing layers of the top and middle sub-cells, which are smaller than 0.5 μm. If the thicknesses of the sub-cells were designed to enable current-matching between the sub-cells, then the EQE of the triple-junction LPC should be larger than the single junction LPC because the thickness of the bottom cell is much smaller than that of the single junction LPC and carrier collection efficiency of the bottom cell is not less than that of the single junction LPC, which supports the current mismatch in our triple-junction LPC design. Therefore, while the minority carrier lifetime may not be long enough, it is not a bottleneck for the conversion efficiency of the triple-junction cells. Meanwhile, the low shunt resistance 132 Ω and high series resistance 0.66 Ω of the triple-junction LPC also contributes to the reduction of the efficiency.

The conversion efficiency of the LPC is 32.13% lower than the expected 42.3% at 1 W/cm² by 10.17% (absolute value), of which 6.11% is due to the decrease of the short-circuit current, about 3.41% is attributed to the reduction of the fill factor, and the open-circuit voltage 2.14 V lower than the calculated 2.174 V caused an efficiency decrease of roughly 0.645%. Therefore, it is necessary to use the proper absorption coefficient to optimize the thicknesses of the absorbing layers of the sub-cells for current matching, to improve the quality of the material to increase the minority carrier lifetime, carrier diffusion length, and hence the internal quantum efficiency, and consequently to achieve a maximized short circuit current. In addition, because 20 nm p⁺-InGaAs/20 nm n⁺-InGaAs tunnel junction was used, the two tunnel junctions absorb about 4.7% of the incident light. The conversion efficiency of the InGaAsP LPC could be increased by some 2% (absolute value) by using non-absorbing p⁺-InP/n⁺-InP or p⁺-InGaAlAs/n⁺-InGaAlAs tunnel junction or the combination of the two.

We designed triple-junction InGaAsP photovoltaic devices to convert 1064 nm laser power into electric power with high output voltage, which are grown on InP substrates by MOCVD and processed into LPC chips. The performance of the LPCs are characterized at room temperature, and a conversion efficiency of 32.6% have been achieved for triple-junction InGaAsP LPCs of 1 × 1 cm² under 1064 nm light illumination at a power density of 1.1 W/cm² with a V_oc of 2.16 V, J_sc of 225 mA/cm² and an FF of 0.74. It is concluded that optimizing the thickness of absorbing layer and adopting non-absorption tunnel junction may significantly improve the conversion efficiency of InGaAsP triple-junction laser power converters.

This work was partially supported by the Jiangsu Province Science Foundation for Youths (No. BK20170431) and the National Natural Science Foundation of China (No. 61604171).