• Journal of Infrared and Millimeter Waves
  • Vol. 40, Issue 1, 7 (2021)
Hong-Bo LU1、2、3, Xin-Yi LI2、*, Ge LI2, Wei ZHANG2, Shu-Hong HU1, Ning DAI1、*, and Gui-Ting YANG2
Author Affiliations
  • 1State Key Laboratory of Infrared Physics,Shanghai Institute of Technology Physics of the Chinese Academy of Sciences,Shanghai 200083,China
  • 2State Key Laboratory of Space Power-sources,Shanghai Institute of Space Power-sources,Shanghai 200245,China
  • 3University of Chinese Academy of Sciences,Beijing 100049,China
  • show less
    DOI: 10.11972/j.issn.1001-9014.2021.01.002 Cite this Article
    Hong-Bo LU, Xin-Yi LI, Ge LI, Wei ZHANG, Shu-Hong HU, Ning DAI, Gui-Ting YANG. Reducing Voc loss in InGaAsP/InGaAs dual-junction solar cells[J]. Journal of Infrared and Millimeter Waves, 2021, 40(1): 7 Copy Citation Text show less

    Abstract

    Smaller Voc of 1.0 eV/0.75 eV InGaAsP/InGaAs double-junction solar cell(DJSC) than the Voc sum of individual subcells has been observed, and there is little information of the origin of such Voc loss and how to minimize it. In this paper, it is disclosed that the dominant mechanism of minority-carrier transport at back-surface-field(BSF)/base interface of the bottom subcell is thermionic emission, instead of defect-induced recombination, which is in contrast to previous reports. It also shows that both InP and InAlAs cannot prevent the zinc diffusion effectively. In addition, intermixing of major III-V element occurs as a result of increasing thermal treatment. To suppress the above negative effects, an initial novel InP/InAlAs superlattice(SL) BSF layer is then proposed and employed in bottom InGaAs subcell. The Voc of fabricated cells reach 997.5 mV, and a reduction of 30 mV in Voc loss without lost of Jsc, compared with the results of conventional InP BSF configuration, is achieved. It would benefit the overall Voc for further four-junction solar cells.

    Introduction

    InGaAsP/InGaAs double-junction solar cells(DJSCs)with approximate bandgap combination of 1.0/0.75 eV are used in four-junction configuration to harvest 900~1 700 nm sunlight,and are crucially important for device performances 1. Previous reports Ref.[2-4]show that open-circuit voltage(Voc)of InGaAsP/InGaAs DJSC is smaller than the sum of individual subcells. To evaluate solar cells with different bandgaps,bandgap-voltage offset under open-circuit condition(Woc)is introduced 5. For multijunction solar cells,Woc can be described as

    ReferenceMethodBandgapIllumination

    Jsc

    (mA/cm2)

    Voc

    (mV)

    Woc

    (mV)

    Oshima [2]MBE1.0/0.71AM1.5G13.15701140
    Wu [3]MBE1.05/0.73AM1.5G16.1830950
    Zhao [4]MOVPE1.07/0.74AM1.5D10.2977833

    Table 1. Previous reported results for InGaAsP/InGaAs DJSC

    Woc(Jsc)=1qiEgi-VocJsc

    where Egi is the bandgap of each subcell,and VocJsc)stands for the open-circuit voltage when the device produces a given value of short-circuit current(Jsc)under illumination. And the Voc loss for multijunction solar cells would be defined as the gap between Wocm and the sum of Woci of subcells,as Wocm-iWociJsc.

    Wocm of InGaAsP/InGaAs DJSC at Jsc of conventional four-junction configuration(about 16.5 mA/cm2)is above 820 mV,higher than the Woc sum of InGaAsP(~330 mV)and InGaAs(~340 mV)individual subcells. Part results from previous reports are listed in Table 1,and there is no reference reporting the origin of such Voc loss and how to minimize it.

    Experience on III-V semiconductor devices reveals that the diffusion and intermixing at heterojunction interface of InP system always leads to device performance degradation6. Moreover,from the viewpoint of physics of solar cell device,the Voc of solar cells majorly depends on the heterojunction interface between base and back-surface field(BSF)layers. Therefore,considering the thermal history of DJSC structure,the bottom InGaAs subcell,especially the BSF/base interface,might be the key role to reduce Voc loss.

    In this paper,the evolution of dopant diffusion and recombination at BSF/base interface with increasing thermal treatment is studied. Based on experimental results,we propose a novel InP/InAlAs superlattice(SL)BSF layer for bottom InGaAs subcell. A reduction of 30 mV in Voc loss is achieved,compared with the results of conventional InP BSF configuration. It shows that such SL BSF would benefit the Voc enhancement for four-junction solar cells.

    1 Experiments

    Growth are done on n-type <100> InP substrates using MOVPE technique. The primary group III and group V precursors used are trimethylgallium(TMGa),trimethylindium(TMIn),trimethylgallium(TMAl),arsine(AsH3),and phosphine(PH3). The dopant precursors used are silane(SiH4)and diethylzinc(DEZn). V/III ratio of 200~300 and growth temperature of 650 °C are employed,as described previously 7.

    Three periods of isotype p+-barrier(100nm)/p--InGaAsP(500nm)/p+-barrier(100nm)/p++-In(Al0.1Ga0.9)As(100nm)double heterojunctions(DHs),separated by InGaAs spacer layers,are grown in the same stack of MOVPE layers,as illustrated in Fig. 1. Two types of barriers,InP and InAlAs respectively,are employed. After growth,individual DHs are exposed by a series of selective etches. Diluted HCl solution and H2SO4:H2O2:H2O solution are used for InP layers and arsenide layers,respectively. The overall element profiles in DHs are obtained through secondary ion mass spectra(SIMS)measurement,while the minority-carrier recombination process in DHs are evaluated using time-resolved photoluminescence(TRPL)technique. It should be pointed out that,the bandgap of p--InGaAsP in DHs is 0.83 eV,for TRPL measurement convenience.

    Cross-section of MOVPE stack containing three BSF/InGaAsP/BSF DHs.

    Figure 1.Cross-section of MOVPE stack containing three BSF/InGaAsP/BSF DHs.

    SIMS measurements are performed using Cs+ primary beam with a fixed 5 kV acceleration. The positive ions of the quasi-molecular cluster are collected and detected. TRPL measurements with a temporal resolution of ~200 ps are performed at room temperature. An H-10330-75 PMT is used to collect PL signals.

    The schematic cross-section of InGaAsP/InGaAs(1.0/0.75 eV)DJSC structure is shown in Fig. 2. The active region of each subcell consists of n-on-p junction(emitter/base)surrounded by n-type InP window layer and p-type BSF layer. In(Al0.1Ga0.9)As tunnel junction is used to connect subcells. The structures are then processed following the standard III-V solar cell device art. The cells are 1.0×1.0 cm2 in size.

    Cross-section of the InGaAsP/InGaAs double-junction solar cell structure.

    Figure 2.Cross-section of the InGaAsP/InGaAs double-junction solar cell structure.

    In-house photovoltaic current density-voltage (J-V)measurements are performed under AM0 solar simulator,without GaAs filter. External quantum efficiency(EQE)measurements are performed to give qualitative insight into the spectral response. Cells are placed on 25 °C cooled stages during measurements.

    2 Results and discussions

    Figure 3 shows the element profiles and PL decay curves for topmost DHs(DH1)in InP-barrier and InAlAs-barrier stacks. The zinc concentration around both DH center regions are of the same level about 5-6×1016 cm-3,which is similar to the typical doping level of base in solar cells. Sharp zinc diffusion profile near the interface between InGaAsP and barriers is observed. It should be pointed out that,the zinc doping level of both InP and InAlAs layers in DHs have been increased to 1-2×1018 cm-3,almost one order of magnitude higher than typical doping level used in BSF layer,to exacerbate such diffusion behavior. Moreover,the zinc concentration in the underneath In(Al0.1Ga0.9)As layer reaches 2×1019 cm-3. The nearly identical profiles of zinc atom suggest that both InP and InAlAs layer are of the similar effect as anti-diffusion barriers during the growth.

    Element profiles (a) and PL decay curves (b) for InP-barrier and InAlAs-barrier DH1s.

    Figure 3.Element profiles (a) and PL decay curves (b) for InP-barrier and InAlAs-barrier DH1s.

    It evidences in Fig. 3 that the minority-carrier decay time in InAlAs-barrier DH is much larger than that in InP-barrier DH. In symmetrical DHs,the effective lifetime τeff extracted from PL decay is related to both bulk carrier lifetime τbulk and surface recombination velocity S,as Ref.[8

    τeff (ns)DH1DH2DH3
    InP-barrier70.036.535.2
    InAlAs-barrier110.053.045.0

    Table 3. The effective minority-carrier lifetime of the DHs

    Barrierm2 (m0)m1 (m0)ΔEc (meV)S (cm/s)
    InP0.080.0472305793
    InAlAs0.0750.0474600.738

    Table 2. Calculated surface recombination velocity at barrier/InGaAsP interface using Eq.. (3)

    1τeff=1τbulk+2Sd

    where d is the thickness of confined layer in the DH. For high-quality materials,the τbulk approximately equals to the reciprocal product of spontaneous radiative recombination coefficient B and doping concentration N

    τbulk=BN-1 .

    The value of B could be calculated according to Ref.[9]. With the doping concentration acquired from SIMS,the τbulk in DH1 is about 200 ns. By mono-exponential fitting of decay curves,τeff of 70 ns and 110 ns are obtained for InP-barrier DH and InAlAs-barrier DH. According to Eq.2,the experimental S for InP/InGaAsP interface and InAlAs/InGaAsP interface are 232 cm/s and 103 cm/s,respectively. Such small recombination velocity suggests that the radiative process in the bulk dominates the carrier recombination,and the recombination at the interface is nearly neglectable.

    The primary mechanism of carrier transport at heterojunction interface includes thermionic emission and defect-induced recombination. When thermionic emission dominates,the recombination current writes 10

    J0=qSn1=qm2m12kBTπm112e-ΔEkBTn1

    where m1 and m2 are effective mass of confined material and barrier,and ΔE is band offset. It is quite obvious that S would exponentially decreases as the band offset increasing. Notice that electron is the minority-carrier in p-InGaAsP DHs. The values of S for DHs,in the scenario of thermionic emission dominating,are estimated using ΔEc and m1,2 from Ref.[11],as listed in Table 2. Surface recombination velocities S of 5793 cm/s and 0.738 cm/s are obtained for InP-barrier DH and InAlAs-barrier DH,respectively. Both calculated and experimental values show that,InAlAs-barrier DH presents smaller surface recombination velocity at heterojunction interface. Consider the complicated carrier transport mechanism at heterojunction interface,the gap between experimental S will not be so large. For example,the sharp diffusion profile of zinc would develop built-in field near the interface,it should reduce the population of minority-carrier reaching the interface and therefore,smaller theoretical value of S could be expected,especially for InP-barrier DH. For InAlAs-barrier DH,the larger experimental S than the calculated S implies the minor existence of trap-induced nonradiative recombination across the interface.

    For InGaAs,the conduction band offset ΔEc for InP barrier and InAlAs barrier are 0.25 eV and 0.52 eV,respectively. The larger offset in conduction band means smaller thermionic emission velocity. Meanwhile,the valence band offset ΔEv for InP barrier and InAlAs barrier are 0.35 eV and 0.17 eV11. The smaller offset in valence band indicates lower potential barrier for majority-carrier. Therefore,InAlAs should be more promising BSF layer in solar cells.

    Figure 4 displays the overall SIMS results for DH stacks,and Fig. 5 shows the PL decay curves for individual DHs. Extracted lifetimes are summarized in Table 3. It is obvious that zinc concentration in confined layers rises with increased thermal history from DH1 to DH3. The concentration in DH2 is about 1.0×1017 cm-3,while the concentration in DH3 is about 2.0×1017 cm-3,in spite of the type of barriers. This provides further evidence that both InP and InAlAs present similar ability to block zinc diffusion. Although there is a downward trend from DH1 to DH3,the effective lifetimes in InAlAs-barrier DHs are always longer than those in InP-barrier DHs. It suggests the surface recombination velocity is still dominated by thermionic emission process. With zinc concentration increasing to 1.0×1017 cm-3,the bulk carrier lifetime τbulk in DH2 decrease to approximately 100 ns,according to Eq.2. Therefore,the surface recombination velocity S increase to 434 cm/s and 221 cm/s,for InP-barrier DH2 and InAlAs-barrier DH2 respectively. Since the drop of band offset is the only cause for thermionic emission related increase of S,it is supposed that diffusion of major III-V element across the interface,which would lead to such shrink of band offset,occurs during the thermal treatment.

    Overall SIMS results of (a) InP-barrier DH stack and (b) InAlAs-barrier DH stack.

    Figure 4.Overall SIMS results of (a) InP-barrier DH stack and (b) InAlAs-barrier DH stack.

    PL decay curves of (a) InP-barrier DHs and (b) InAlAs-barrier DHs.

    Figure 5.PL decay curves of (a) InP-barrier DHs and (b) InAlAs-barrier DHs.

    As shown in Table 3,for DHs using the same type of barriers,the τeff in DH3 are quite close to the τeff in DH2. With zinc concentration of 2.0×1017 cm-3,the bulk carrier lifetime τbulk in DH3 is approximately 50 ns,and S are 210.2 cm/s and 55.6 cm/s,for InP-barrier DH3 and InAlAs-barrier DH3 respectively. The abnormal decrease of S is probably due to the photon recycling effect,which leads to the longer τeff and smaller S in DH3 than expected.

    The steady-state PL of DHs confirms the above hypothesis. As shown in Fig. 6,the PL peak for DH1 and DH2 are of similar intensity,while PL intensities of DH3 are nearly one order of magnitude stronger. Considering the optical configuration of DH stack,the photon recycling is the most effective in DH3,and is suppressed in DH1 and DH2 due to extra absorption from underlying narrow bandgap InGaAs spacers.

    Steady-state PL of (a) InP-barrier DHs and (b) InAlAs-barrier DHs. Weak peaks marked by asteroids in DH1 and DH2 are related to the spacers.

    Figure 6.Steady-state PL of (a) InP-barrier DHs and (b) InAlAs-barrier DHs. Weak peaks marked by asteroids in DH1 and DH2 are related to the spacers.

    The results of InGaAsP/InGaAs DJSCs using both InP and InAlAs BSF layers confirm the advantages and effectiveness of InAlAs BSF layer in practical device. Figure 7 shows light J-V and EQE measurements of the devices. Using InAlAs BSF layer,the cell presents an efficiency of 9.28% with a Voc of 983.2 mV,a Jsc of 15.6 mA/cm2 and an FF of 0.818. Meanwhile,the device using InP BSF present a Voc of 967.7 mV,a Jsc of 15.3 mA/cm2 and an FF of 0.819. An enhancement of Voc is obtained,without any cost of Jscand FF.

    (a) Light J-V and (b) spectra response curves for InGaAsP/InGaAs solar cells using InP and InAlAs BSF layers

    Figure 7.(a) Light J-V and (b) spectra response curves for InGaAsP/InGaAs solar cells using InP and InAlAs BSF layers

    It is well established that SL serve as effective barrier for element diffusion or intermixing,and dislocation threading,and it has been widely used in semiconductor devices such as high electron mobility transistors,laser diodes,electro absorption modulators 12-16. Also,the miniband in SL would not introduce extra potential barrier for carrier transport 17. An initial five-period InP(2nm)/InAlAs(2nm)SL BSF layer is designed and employed in bottom InGaAs subcell of DJSC. A Vocof 997.5 mV,a Jsc of 15.8 mA/cm2 and an FF of 0.824 are obtained as in Fig. 8. Both Voc and Jsc are boosted,as expected,in fabricated SL BSF device. The Voc approaches 1.0 V,resulting in a Woc of 752.5 mV. A reduction of 30 mV in Voc loss for DJSC is achieved,compared with the conventional InP BSF DJSC.

    Light J-V for InGaAsP/InGaAs DJSC using 5-period InP/InAlAs SL BSF layer.

    Figure 8.Light J-V for InGaAsP/InGaAs DJSC using 5-period InP/InAlAs SL BSF layer.

    3 Conclusions

    In general,the use of novel SL BSF layer in the bottom subcell reduces the Voc loss in InGaAsP/InGaAs DJSC.

    Experiments show that,the mechanism of minority-carrier transport at BSF/base interface of the bottom subcell of InGaAsP/InGaAs DJSCs is dominated by thermionic emission,instead of defect-induced recombination,which is in contrast to previous reports. It also shows that both InP and InAlAs cannot prevent the zinc diffusion effectively. In addition,intermixing of major III-V element occurs as a result of increasing thermal treatment.

    Based on the above results,an initial 5-period InP/InAlAs SL BSF layer is designed and employed in bottom InGaAs subcell of DJSC. A Voc of 997.5 mV,a Jsc of 15.8 mA/cm2 and an FF of 0.824 are obtained. The Vocapproaches 1.0 V,resulting in a Wocof 752.5 mV. A reduction of 30 mV in Voc loss for DJSC is achieved,compared with the results of conventional InP BSF configuration. It suggests that such SL BSF would benefit the Voc enhancement for four-junction solar cells.

    References

    [1] D. Frank, M. Grave, P. Beutel et al. Wafer bonded four‐junction GaInP/GaAs//GaInAsP/GaInAs concentrator solar cells with 44.7% efficiency. Prog. Photovolt: Res. Appl., 22, 277-282(2014).

    [2] R. Oshima, K. Makita, H. Mizuno et al. MBE-grown InGaAsP solar cells with 1.0 eV bandgap on InP (001) substrates for application to multijunction solar cells. Japanese Journal of Applied Physics, 14-14(2015).

    [3] Y. Wu, L. Ji, P. Dai et al. Effects of buffer layer and back-surface field on MBE-grown InGaAsP/InGaAs solar cells. Japanese Journal of Applied Physics, 55, 022301-4(2016).

    [4] Y. Zhao, J. Dong, K. Li et al. InGaAsP/InGaAs tandem photovoltaic devices for four-junction solar cells. Journal of Semiconductors, 36, 044011-4(2015).

    [5] R. R. King, D. Bhusari, A. Boca et al. Band gap-voltage offset and energy production in next-generation multijunction solar cells. Prog. Photovolt: Res. Appl., 19, 797-812(2011).

    [6] John Brice et al. Properties of INDIUM PHOSPHIDE, 6, 280-281(1991).

    [7] H Lu, X Li, W Zhang et al. MOVPE grown 1.0 eV InGaAsP solar cells with bandgap-voltage offset near to ideal radiative recombination limit. Solar Energy Materials and Solar Cells, 196, 65-69(2019).

    [8] M Boulou, D Bois. Cathodoluminescence measurements of the minority‐carrier lifetime in semiconductors. Journal of Applied Physics, 48, 4713-4721(1977).

    [9] R. K. Ahenkiel, M. S. Lundstrom. Minority Carriers in III-V Semiconductors: Physics and Applications, Semiconductors and Semimetals. Academic Press, Inc., 39, 57-59(1993).

    [10] D. Schroeder. Modelling of Interface Carrier Transport For Device Simulation, 166(1994).

    [11] S. Adachi. Physical Properties Of III-V Semiconductor Compound, 75-109(1992).

    [12] R L Ross, S P Svensson, P Lugli. Pseudomorphic HEMT Technology and Applications(1996).

    [13] D H Zhu, Z G Wang, J B Liang et al. 808 nm high-power laser grown by MBE through the control of Be diffusion and use of superlattice. JOURNAL OF CRYSTAL GROWTH, 175, 1004-1008(1997).

    [14] R V Chelakara, M R Islam, J G Neff et al. Growth of high-quality InAlP/InGaP quantum wells and InAlP/InGaP superlattice barrier cladding layers by metalorganic chemical vapor deposition. Journal of Crystal Growth, 145, 179-186(1994).

    [15] M E Givens, L J Mawst, C A Zmudzinski et al. Effect of compositionally graded and superlattice buffer layers on the device performance of graded barrier quantum well heterostructure laser diodes. Applied Physics Letters, 50, 301-303(1987).

    [16] J Zhang, X Y Chen, Y J Ma et al. Optimization of In0.6Ga0.4As/InAs electron barrier for In0.74Ga0.26As detectors grown by molecular beam epitaxy. Journal of Crystal Growth, 512, 84-89(2019).

    [17] John H Davis. The Physics of Low-Dimensional Semiconductors, 177-182(2005).

    Hong-Bo LU, Xin-Yi LI, Ge LI, Wei ZHANG, Shu-Hong HU, Ning DAI, Gui-Ting YANG. Reducing Voc loss in InGaAsP/InGaAs dual-junction solar cells[J]. Journal of Infrared and Millimeter Waves, 2021, 40(1): 7
    Download Citation