Performance assessment of multijunction solar cells incorporating GaInNAsSb
© Aho et al.; licensee Springer. 2014
Received: 18 November 2013
Accepted: 24 January 2014
Published: 5 February 2014
We have measured the characteristics of molecular beam epitaxy grown GaInNAsSb solar cells with different bandgaps using AM1.5G real sun illumination. Based on the solar cell diode characteristics and known parameters for state-of-the-art GaInP/GaAs and GaInP/GaAs/Ge cells, we have calculated the realistic potential efficiency increase for GaInP/GaAs/GaInNAsSb and GaInP/GaAs/GaInNAsSb/Ge multijunction solar cells for different current matching conditions. The analyses reveal that realistic GaInNAsSb solar cell parameters, render possible an extraction efficiency of over 36% at 1-sun AM1.5D illumination.
88.40.hj; 88.40.jm; 88.40.jp; 81.15.Hi
KeywordsIII-V Semiconductor multijunction solar cells GaInNAsSb solar cells Molecular beam epitaxy High efficiency solar cells Dilute nitrides
Multijunction solar cells (MJSC) are instrumental in concentrated (CPV) and space photovoltaic systems. The driving force for the material and technological development of MJSCs is the need for higher conversion efficiency. In CPV systems, the conversion efficiency is further increased owing to the use of concentrated light and therefore any efficiency gain that can be made by using more suitable materials and advanced design would lead to significant gain in overall system efficiency. The record CPV efficiency for lattice-matched GaInP/GaAs/GaInNAsSb SC is 44% . On the other hand, the best lattice-matched GaInP/GaAs/Ge exhibit a peak efficiency of 43.3% under concentration  and 34.1% at 1 sun . Efficiencies as high as 50% have been predicted for cells with a larger number of junctions and high concentration . To this end, a promising approach is to integrate dilute nitrides and standard GaInP/GaAs/Ge. Yet, so far, such heterostructures have exhibited low current generation .
The GaInNAs and GaInNAsSb solar cells reported in the literature have typically high bandgap voltage offsets (Woc), indicating poor junction properties [6, 7]. The offsets can be higher than 0.6 V, which is a rather high value when compared to GaInAs materials exhibiting a Woc of 0.4 V or even lower . Recent studies on GaInNAs grown by molecular beam epitaxy (MBE) have demonstrated that by employing proper fabrication parameters [8–10], the Woc can be reduced below 0.5 V . Another peculiar feature of GaInNAs solar cells is their shunt-like junction operation [6, 12]. This feature has been associated with clustering in GaInNAs due to phase separation of GaInNAs. Phase separation and shunt-like operation can also be avoided in MBE by the optimizing of the growth parameters . In this paper, we focus on GaInNAsSb-based multijunction SCs, in particular on evaluating the practical bandgap and thickness limitations set by the subjunctions. Using realistic solar cell parameters for GaInNAsSb, based on the diode model and Kirchhoff's laws, we estimate the efficiency of GaInP/GaAs/GaInNAsSb and GaInP/GaAs/GaInNAsSb/Ge solar cells.
Experimental details and models
The experimental set consisted of single-junction GaInNAsSb p-i-n SCs with bandgaps ranging from 0.84 to 1.0 eV. The structures were grown by solid source MBE, equipped with SUMO cells for group III atoms, thermal crackers for group V elements and RF plasma source for atomic N flux generation. The N composition (y) of Ga1−xIn x N y As1−y was 0.035 while the In composition (x) was approximately 2.7 times the N composition to ensure lattice matching to GaAs. The GaInNAsSb samples were also closely lattice-matched to GaAs using Sb compositions of up to 0.04. For all structures, the lattice matching was verified by X-ray diffraction measurements.
We also fabricated a GaInP/GaAs/GaInNAs triple-junction test SC structure including a GaInNAs subjunction with a bandgap of 0.9 eV. The triple-junction solar cell and the fabrication details are described elsewhere . After the MBE process, the samples were processed to solar cells having TiAu contact metals on p-side and NiGeAu for the n-side. Then the surface was coated with a two-layer TiO/SiO antireflection (AR) coating. The current–voltage (I-V) characteristics of single and multijunction solar cells were measured at the real sun (AM1.5G). The real sun intensity level was measured with a Kipp&Zonen CM11 pyranometer (Delft, the Netherlands). The external quantum efficiency (EQE) of the GaInNAs SC was also measured. Our EQE system was calibrated using NIST-calibrated Si and Ge detectors. Moreover, we measured the room-temperature photoluminescence (PL) spectra to determine the bandgaps of GaInNAsSb subjunction materials. The solar cell measurements and calculations are performed for one sun illumination unless otherwise stated when data is presented.
Here, I is the current of the multijunction device which contains one to four junctions inside, I i is the current through an individual solar cell, V i (I) is the voltage of single-junction device, n i is the quality factor of the i th diode, kB is the Boltzmann coefficient, T is the device temperature (T = 300 K), ILi is the current generated by the junction i, Egi is the bandgap (300 K) of the i th junction, I0i is the reverse saturation current of the i th junction at 300 K, Rs is the device total series resistance, and V is the device total voltage. We have neglected the shunt resistance for simplicity, which is a good approximation for most of the high-quality SC devices. Here, we have also approximated the tunnel junctions as ideal lossless contacts between the solar cell junctions. One should keep in mind that this approximation is not valid at extremely high concentrations; the concentration limit depends strongly on the quality of the tunnel junctions.
From the data presented in Figures 1 and 2b, we have calculated the Woc values for selected GaInNAs and GaInNAsSb single-junction SCs. For GaInNAs SC with Eg = 1 eV the Woc was 0.58 V and for GaInNAsSb with Eg = 0.90 eV, the Woc was 0.59 V. The best Woc we have achieved so far from GaInNAs single-junction SCs is 0.49 V . The observations made here are in accordance with previously published reports which indicate that the Sb-based solar cells have a slightly higher Woc values compared to GaInNAs SCs [6, 9].
Characteristics of GaInNAsSb p-i-n diodes at different illumination conditions
GaInNAs (1 eV)
AM1.5G (900-nm LP)
GaInNAs (1 eV)
GaInNAsSb (0.9 eV)
Theoretical and practical limits for current generation in GaInNAsSb SC
The theoretical upper limit for the bandgap of GaInNAsSb in GaInP/GaAs/GaInNAsSb solar cell operating at AM1.5G is 1.04 eV. In practice, the bandgap needs to be slightly smaller than this because the EQEav target of approximately 100% is impractical for GaInNAsSb. EQEav values of approximately 90% have been achieved for GaInP, GaAs, and Ge junctions [12, 15], and thus, we set the EQEav = 90% as a practical upper limit for GaInNAs subjunction operation which sets the upper limit for the GaInNAsSb bandgap to 1.02 eV. The current matching limits for different bandgaps of GaInNAsSb are presented in Figure 3b, where N compositions were calculated using the Vegard law and the band anti-crossing model .
Ideal and practical J sc v alues for GaInP/GaAs/GaInNAsSb and GaInP/GaAs/GaInNAsSb/Ge SCs
Jsc(GaInP) + Jsc(GaAs)(mA/cm2)
Jsc(GaInNAsSb) + Jsc(Ge)(mA/cm2)
Jsc-current matched 3J(mA/cm2)
Jsc-current matched 4J(mA/cm2)
The optimal bandgap for GaInNAsSb junction of the triple- and four-junction SCs depends on the target spectrum and the performance of the subjunctions [12, 15]. In a four-junction cell, it would be beneficial to have slightly larger bandgaps for the top junctions, especially for the AM1.5G spectrum. The GaInP/GaAs top cells have already been well optimized and that is the reason why the bandgap shifting is probably not the best practical step to start with, especially because the Woc values of top junctions with larger bandgaps increase easily .
For the efficiency simulation of MJSCs, we used the measured results for GaInNAsSb and parameters for state-of-the-art GaInP/GaAs  and GaInP/Ga(In)As/Ge  SCs with standard bandgaps of 1.9/1.4/0.70 eV. The calculated multijunction SC characteristics with GaInNAsSb subjunctions are based on the data presented in Tables 1 and 2 and the diode Equations 1 to 3.
Estimated 1-sun efficiencies for GaInNAsSb multijunction solar cells at AM1.5G
This work, 
This work, 
3 J-GaInP/GaAs/GaInNAs (15.5 mA/cm2)
This work, 
3 J-GaInP/GaAs/GaInNAs (15.5 mA/cm2)
This work, 
This work, 
This work, 
Estimated 1-sun efficiencies for GaInNAsSb multijunction solar cells at AM1.5D
3 J-GaInP/GaAs/GaInNAsSb (0.90 eV)
3 J-GaInP/GaAs/GaInNAs (15.5 mA/cm2)
15.15 (Ideal 3 J)
14.68 (Ideal 4 J)
Results and discussion
According to our measurements and calculations, it would be beneficial to design the GaInNAs junction to overproduce current (see Figure 4a). Our calculations show that when GaInNAs junction generates more current than other junctions one would get approximately 1 percentage points higher efficiency compared to exactly current-matched triple-junction device. This is in line with reported data for GaInP/GaAs/GaInNAsSb triple-junction cells .
The efficiency improvement upon adding GaInNAsSb junction to a double- or triple-junction cell shows clear dependence on the illumination spectrum. When GaInP/GaAs/Ge triple-junction cells are compared with GaInP/GaAs/GaInNAs, one observes that at AM1.5G, the efficiency is 0.4 to 1.4 percentage points better when GaInNAs subjunction is used, depending of the design and the GaInNAs subjunction performance. However, it turns out that a four-junction SC with 1 eV GaInNAs, does not perform well at AM1.5G illumination. The added Ge junction does not improve the efficiency when compared to its triple junction reference (GaInP/GaAs/GaInNAs cell). This is simply due to the fact that the subjunctions of GaInP/GaAs/GaInNAs (Eg = 1 eV)/Ge SCs do not have the optimum bandgaps for current matching at AM1.5G conditions. When such a device is measured at AM1.5D, the situation changes and due to less blue rich spectrum, the multijunction device has better current matching between the subjunctions . The studied four-junction device can have 1.6- to 1.7-percentage point higher efficiency at 1-sun than its GaInNAs triple-junction reference depending on the current matching. We have also compared the effect of bandgap on the efficiency of triple-junction devices. When a GaInNAsSb subjunction with Eg = 0.9 eV instead of GaInNAs with Eg = 1.0 eV is used at AM1.5D, the obtainable efficiency drops a 1.4 percentage points but since a device would be easier to realize with generation of excess current, the drop in practice would be smaller (see Figure 4a).
We have made a preliminary estimate for the performance of GaInP/GaAs/GaInNAs/Ge SC under concentrated sunlight at AM1.5D using GaInP/GaAs/Ge parameters from reference . When compared to 1-sun results, the benefit of using a GaInNAs junction starts to be significant at concentrated sunlight. We estimate that GaInP/GaAs/GaInNAs triple-junction SCs operated at a concentration of 300 times have up to 3- to 6-percentage point higher efficiencies than GaInP/GaAs/Ge SCs. The situation gets even more favorable for using GaInNAs when four-junction devices are considered. Our calculations show that the efficiency can be further improved by approximately 3.5 percentage points compared with a GaInP/GaAs/GaInNAs triple-junction device by adding the fourth junction.
Another important aspect that needs to be addressed to make sure of these advantages is the AR coating. The four-junction devices are already very demanding from the AR coating point of view since even the lowest short circuit current density of 13.79 mA/cm2 used in the calculations requires an EQEav of 91%. Commonly used AR coatings on GaInP/GaAs/Ge should be improved since the reflectance has traditionally been optimized for GaInP and GaAs subjunction current generation. This can be done in GaInP/GaAs/Ge SCs with almost no additional loss as Ge produces excess current that is able to accommodate the loss due to inappropriate AR coating. This leads to the fact that many Ge-based multijunction devices have EQEav less than 90%. To improve the AR coating, one needs to adopt new schemes. One potential candidate is the moth eye pattern fabricated onto window layers of multijunction SCs. Such AR coatings are able to provide low reflectivity throughout the entire absorption spectrum of multijunction SCs .
Four-junction SCs are also sensitive to changes in spectral conditions since the photons need to be shared more equally than in Ge-based triple-junction devices. However, calculations have proved that inserting the fourth junction [12, 15] or even more junctions would in fact be beneficial from the total yearly produced energy point of view, even if the changing spectral conditions were considered. Another positive factor for the GaInP/GaAs/GaInNAsSb/Ge SCs is the fact that the top cells can be made thin to obtain current matching. This will bring clear savings in fabrication costs, especially for CPV cells. There are indications that by using thin subjunctions, the epitaxial costs could be even cut by half . The multijunction SC approach easily gets cost limited by the substrate costs and thus substrate recycling would be obvious companion to this approach. Therefore, the optimal GaInP/GaAs/GaInNAsSb/Ge structure would depend on the device efficiency, the cost of epitaxy and the cost of substrate and environment where the SC would be operated.
The efficiency improvements to GaInP/GaAs/GaInNAsSb SC after adding the Ge junction calculated in this paper may seem small but when calculating the SC system costs and generated energy factor, the grid-connected systems would provide better values since the total system costs do not increase too much . In this paper, we have not estimated the effect of the lower Ge junction current generation on Voc of Ge junction in the four-junction device. It was dropped out because of the lack of information on Ge subjunction performance in high-quality GaInP/GaAs/Ge SC. This might bias our results towards slightly overestimated Voc and FF values for the four-junction SCs. On the other hand, in four-junction SCs, the quantum defect is lower in the Ge subjunction and the overall temperature of the whole SC will be lower, especially in CPV operation. In practice, this makes higher efficiencies and higher Voc possible at high concentrations.
We have presented our GaInNAsSb diode characteristics with different N and Sb compositions and estimated the efficiency of GaInP/GaAs/GaInNAsSb and GaInP/GaAs/GaInNAsSb/Ge solar cells. Our calculations based on measurements and a diode model reveal that at AM1.5G and at current matching condition, the use of GaInNAsSb junction as the bottom junction of a triple junction SC can increase the efficiency by approximately 4 percentage points compared to GaInP/GaAs double junction SC and have 1.4 percentage points higher efficiency than a GaInP/GaAs/Ge SC. At AM1.5D, the GaInNAsSb-based four-junction cell has a potential to show 1.7 percentage points higher efficiency than the GaInP/GaAs/GaInNAsSb triple-junction device. The achievable efficiencies for GaInNAsSb four-junction solar cells at AM1.5D 1-sun illumination are estimated to be over 36%. Our future target is to increase the GaInNAsSb EQE close to 100%, minimize the losses in front surface reflection and develop low-loss tunnel junctions.
the bandgap (300 K) of the i th junction
external quantum efficiency
current of the two-terminal solar cell
reverse saturation current of the i th junction at 300 K
current through an individual solar cell
current generated by the i th junction
molecular beam epitaxy
multijunction solar cell
quality factor of the i th diode
device total series resistance
device total voltage
open circuit voltage
voltage of a single-junction device
bandgap open circuit voltage offset
solar cell efficiency.
The authors acknowledge the Finnish Funding Agency for Technology and Innovation, Tekes, via projects ‘Solar III-V’ #40120/09 and ‘Nextsolar’ #40239/12, and European Space Agency via project Contract N.: 4000108058/13/NL/FE. A. Aho acknowledges Graduate School in Electronics, Telecommunications and Automation, Wärtsilä Foundation, Finnish Foundation for Technology Promotion, and Ulla Tuominen Foundation for financial support. The authors also acknowledge MSc Ville-Markus Korpijärvi, DSc Juha Tommila, Wenxin Zhang, BSc Joel Salmi and BSc Pekka Malinen for their technical support.
- Green MA, Emery K, Hishikawa Y, Warta W, Dunlop ED: Solar cell efficiency tables (version 41). Prog Photovolt Res Appl 2013, 21(1):1–11. 10.1002/pip.2352View ArticleGoogle Scholar
- CPV World Record by AZUR SPACE Solar Power: 43.3 Percent Efficiency. http://www.azurspace.com/images/pdfs/pi-2012-AzurSpace-Rekord_EN.pdf
- Bett AW, Dimroth F, Guter W, Hoheisel R, Oliva E, Phillips SP, Schöne J, Siefer G, Steiner M, Wekkeli A, Welser E, Meusel M, Köstler W, Strobl G: Highest efficiency multi-junction solar cell for terrestrial and space applications. In 24th European Photovoltaic Solar Energy Conference and Exhibition. Hamburg; 2009:1–6.Google Scholar
- King RR, Bhusari D, Boca A, Larrabee D, Liu XQ, Hong W, Fetzer CM, Law DC, Karam NH: Band gap-voltage offset and energy production in next-generation Multijucntion Solar Cells. Prog Photovol: Res Appl 2011, 19: 797–812. doi:10.1002/pip.1044 doi:10.1002/pip.1044 10.1002/pip.1044View ArticleGoogle Scholar
- King RR, Sherif RA, Kinsey GS, Kurtz S, Fetzer CM, Edmondson KM, Law DC, Cotal HL, Krut DD, Karam NH: Bandgap engineering in high-efficiency multijunction concentrator cells. In International Conference on Solar Concentrators for the Generation of Electricity or Hydrogen May 1–5, 2005. Scottsdale, Arizona; 2005. NREL/CD-520–38172 NREL/CD-520-38172Google Scholar
- Jackrell DB, Bank SR, Yuen HB, Wistey MA, Harris JS Jr: Dilute nitride GaInNAs and GaInNAsSb solar cells by molecular beam epitaxy. J Appl Phys 2007, 101: 114916. 10.1063/1.2744490View ArticleGoogle Scholar
- Khan A, Kurtz SR, Prasad S, Johnson SW, Gou J: Correlation of nitrogen related traps in InGaAsN with solar cell properties. Appl Phys Lett 2007, 90: 243509. 10.1063/1.2747664View ArticleGoogle Scholar
- Aho A, Tukiainen A, Polojärvi V, Salmi J, Guina M: High current generation in dilute nitride solar cells grown by molecular beam epitaxy. In Proceedings of the SPIE 2013 volume 8620. San Francisco; 2013. doi:10.1117/12.2002972 doi:10.1117/12.2002972Google Scholar
- Aho A, Tukiainen A, Korpijärvi VM, Polojärvi V, Salmi J, Guina M: Comparison of GaInNAs and GaInNAsSb solar cells grown by plasma-assisted molecular beam epitaxy. In AIP Conference Proceedings. Toledo; 2012:49–52. Volume 1477. http://dx.doi.org/10.1063/1.4753831 Volume 1477.Google Scholar
- Aho A, Tukiainen A, Polojärvi V, Salmi J, Guina M: MBE growth of high current dilute III-V-N single and triple junction solar cells. In EU Pvsec 2012 27th European Photovoltaic Solar Energy Conference and Exhibition. Frankfurt; 2012:290–292. doi:10.4229/27thEUPVSEC2012–1BV.7.13 doi:10.4229/27thEUPVSEC2012-1BV.7.13Google Scholar
- Tommila J, Aho A, Tukiainen A, Polojärvi V, Salmi J, Niemi T, Guina M: Moth-eye antireflection coating fabricated by nanoimprint lithography on 1 eV dilute nitride solar cell. Prog Photovolt Res Appl 2013, 21: 1158–1162. doi:10.1002/pip.2191 doi:10.1002/pip.2191Google Scholar
- Friedman DJ, Kurtz SR: Breakeven criteria for the GaInNAs junction in GaInP/GaAs/GaInNAs/Ge four-junction solar cells. Prog Photovolt Res Appl 2002, 10: 331–344. doi:10.1002/pip.430 doi:10.1002/pip.430 10.1002/pip.430View ArticleGoogle Scholar
- Aho A, Tukiainen A, Polojärvi V, Gubanov A, Salmi J, Guina M, Laukkanen P: Lattice matched dilute nitride materials for III-V high-efficiency multi-junction solar cells: growth parameter optimization in molecular beam epitaxy. In EU Pvsec 2011 26th European Photovoltaic Solar Energy Conference and Exhibition. Hamburg; 2011:58–61. doi:10.4229/26thEUPVSEC2011–1AO.8.3 doi:10.4229/26thEUPVSEC2011-1AO.8.3Google Scholar
- ASTM G 173–03: Standard tables for reference solar spectral irradiances: direct normal and hemispherical on 37° tilted surface. West Conshohoken, PA: ASTM International; 2003. doi:10.1520/G0173–03R12 doi:10.1520/G0173-03R12Google Scholar
- Kurtz SR, Myers D, Olson JM: Projected performance of three- and four-junction devices using GaAs and GaInP. In 26th IEEE, Photovoltaic specialists conference September 29- October 3, 1997. Anaheim: IEEE; 1997. doi:10.1109/PVSC.1997.654226 doi:10.1109/PVSC.1997.654226Google Scholar
- Vurgaftman I, Meyer JR: Band parameters for nitrogen-containing semiconductors. J Appl Phys 2003, 94: 3675. 10.1063/1.1600519View ArticleGoogle Scholar
- Takamoto T, Ikeda E, Kurita H, Ohmori M: Over 30% efficient InGaP/GaAs tandem solar cell. Appl Phys Lett 1997, 70: 381. doi:10.1060/1.118419 doi:10.1060/1.118419 10.1063/1.118419View ArticleGoogle Scholar
- Kirk AP: High efficacy thinned four-junction solar cell. Semicond Sci Technol 2011, 26: 155013. doi:10.1088/0268–1242/26/12/125013 doi:10.1088/0268-1242/26/12/125013View ArticleGoogle Scholar
- Wiemer M, Sabnis V, Yuen H: 43.5% efficient lattice matched solar cells. In Proceedings of SPIE 8108 High and Low Concentrator Systems for Solar Electric Applications VI. San Diego, CA; 2011. doi:10.1117/12.897769 doi:10.1117/12.897769Google Scholar
- Azur space CPV triple junction solar cell - Type 3C40C (5.5*5.5mm2). http://www.azurspace.com/images/pdfs/CPV%20TJ%20Solar%20Cell%203C40C%205.5x5.5mm.pdf
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.