⇦Chinese Physics Letters, 2017, Vol. 34, No. 2, Article code 028802 Application of AlGaInP with Sb Incorporation in Lattice-Matched 5-Junction Tandem Solar Cells Yang Zhang(张杨)1, Qing Wang(王青)1**, Xiao-Bin Zhang(张小宾)2, Na Peng(彭娜)2, Zhen-Qi Liu(刘振奇)2, Bing-Zhen Chen(陈丙振)2, Shan-Shan Huang(黄珊珊)2, Zhi-Yong Wang(王智勇)1 Affiliations 1Institute of Laser Engineering, Beijing University of Technology, Beijing 100022 2Redsolar New Energy Technology Co. Ltd., Zhongshan 528437 Received 25 November 2016 **Corresponding author. Email: wangqing@bjut.edu.cn Citation Text: Zhang Y, Wang Q, Zhang X B, Peng N and Liu Z Q et al 2017 Chin. Phys. Lett. 34 028802 Abstract It is well known that conventional GaInP/GaInAs/Ge three-junction (3J) solar cells are difficult to continue to ascend when the efficiencies reach 32% and 42% under AM0 and AM1.5D concentrated, respectively. In AlGaInP/AlGaInAs/GaInAs/GaInNAs/Ge five-junction (5J) solar cells, the performance of the AlGaInP, AlGaInAs and GaInNAs sub cell is the key factor for conversion efficiency of the 5J solar cell. We investigate the AlGaInP/AlGaInAs/Ge 3J solar cell. By incorporating surfactant trimthylantimony into the AlGaInP material, the crystal quality of AlGaInP is improved and the spectrum absorption range of AlGaInAs is extended. The current density of each sub cell exceeds 11.3 mA/cm$^{2}$ as is desired. Then we apply this 3J structure to grow the lattice-matched 5J solar cell and obtain the short circuit current of 134.96 mA, open circuit voltage of 4399.6 mV, fill factor of 81.7% and conversion efficiency of 29.87%. DOI:10.1088/0256-307X/34/2/028802 PACS:88.40.jp, 88.40.hj, 81.15.Gh © 2017 Chinese Physics Society Article Text As we all know, with the development of the high concentrator photovoltaic technology (HCPV), many researchers have developed 3J metamorphic (MM) structure and inverted metamorphic (IMM) structure, and their conversion efficiency has exceeded 42% (AM1.5D, 500 sun).[1,2] However, it is difficult to further improve the efficiency because of the band gap limitation in normal GaInP/GaInAs/Ge 3J cells. Hence, the united research team of the Fraunhofer institute and the SOITEC company developed a new four-junction (4J) bonding solar cells with 44.7% efficiency in 2014. The solar cells adopted the GaInP/GaAs/GaInAsP/GaInAs (1.9 eV/1.4 eV/1.0 eV/0.7 eV) structure.[3] According to the latest reports, the conversion efficiency of this 4J solar cell has been raised to 46%, so far the highest in the world.[4] Spectrolab developed direct bonding 5J solar cells with a 2.2 eV/1.7 eV/2.2 eV/1.7 eV/0.73 eV band gap combination. Under 1 sun AM0 spectrum, it obtained an open circuit voltage $V_{\rm oc}$ of 4.76 V, a short circuit current density $J_{\rm sc}$ of 12.12 mA/cm$^{2}$, a fill factor (FF) of 84.6% and a conversion efficiency (Eff) of 35.8%.[5] We can obtain higher conversion efficiency multi-junction solar cells by the directly bonding method. However, sub cells need to be grown separately on two different substrates, and to be stripped from their own substrates, respectively. This method has a high request for the process technology, and it affects the yield. To avoid the technology challenge mentioned above, we directly grow AlGaInP/AlGaInAs/GaInAs/ GaInNAs/Ge lattice-matched 5J solar cells on Ge substrates by the MOCVD technology. Except for the work which describes AlGaInP/GaInP/AlGaInAs/GaInAs/Ge 5J solar cells produced by Dimiroth et al.,[6] there are hardly any other works about the lattice-matched 5J solar cell structure. As for the 5J solar cell discussed in this study, compared with GaInP/GaInAs/GaInNAs/Ge lattice-matched 4J solar cells, the major difference is that the 5J solar cells adopt AlGaInP, AlGaInAs and GaInAs sub cells to further subdivide the 300–900 nm spectrum, whose photons are absorbed by the GaInP and GaInAs sub cells in the 4J solar cell. Actually, the upright lattice matched 5J solar cell also has its own problems. First, the insufficient quality of the GaInNAs material results in lower current density, but for the 5J solar cell, the further waveband subdivision could alleviate the requirement of the GaInNAs subcell current density, and could obtain higher open circuit voltage and conversion efficiency. Secondly, the oxygen incorporation produces deep level traps in the high-Al-composition AlGaInP material due to the high dissociation energy of the bond between aluminum and oxygen. The serious oxygen contamination will result in low current density. Thirdly, the narrow photon absorption region and high Al composition (20%) of the AlGaInAs subcell will also lead to low current density. Both of them may be a current limitation junction in the 5J solar cell. Therefore, based on the research of GaInNAs and GaInAs sub cell reported by another team, we carry out a research work on an AlGaInP/AlGaInAs/Ge 3J solar cell. Similar work has not yet been seen in reports. In this work, 3J and 5J solar cells are fabricated by Veeco K475 MOCVD equipment. Trimethylgallium (TMGa), trimethylaluminum (TMAl), trimethylindium (TMIn) and dimethylhydrazine (DMHy) are used as group III sources. Arsine (AsH$_{3}$) and phosphine (PH$_{3}$) are used as group V sources. Zn (from dimethylzinc) is applied as p-type dopants for each sub cell's base regions and back field layers. In addition, Si (from disilane) and Te (from diethyl telluride) are employed as n-type dopants for window layers and emitter regions. The AlGaInP/AlGaInAs/Ge 3J structure shown in Fig. 1(a) is directly fabricated by the MOCVD on p-type Ge (100) substrates in orientation of 9$^{\circ}$ towards [111]. Before its growth, PH$_{3}$ is introduced into the chamber to form a Ge junction with a band gap of 0.67 eV. We adopt the AlGaAs/GaAs structure with high doped concentration as the tunnel junction; it can connect AlGaInP, AlGaInAs and Ge sub cells to form a 3J solar cell structure.
cpl-34-2-028802-fig1.png
Fig. 1. (a) The AlGaInP/AlGaInAs/Ge 3J solar cell schematic structure. (b) The AlGaInP/AlGaInAs/GaInAs/GaInNAs/Ge 5J solar cell schematic structure.
The electron beam evaporation is used to achieve Ti/Au deposition on the Ge substrate and AuGeNi/Au deposition on the GaInAs contact layer. Moreover, the SiO$_{2}$/TiO$_{2}$ film is deposited as the anti-reflection coating by the assisted ion deposition (AID) technology. Then, the wafer is cut into 3 cm $\times$ 4 cm solar cell chips. External quantum efficiencies (EQE) are measured by a spectral response measurement system (Bentham: PVE300). The $I$–$V$ characteristics are measured by a class AAA solar simulator (Spectrolab: X225) under 1 sun AM0 spectrum (25$^{\circ}\!$C). After the discussion of AlGaInP/AlGaInAs/Ge 3J solar cell, the 5J solar cell can be built by inserting GaInNAs and GaInAs sub cells between the Ge sub cell and the AlGaInAs sub cell as shown in Fig. 1(b). We select 1.98 eV band gap AlGaInP materials whose Al composition is 8% as the top cell of the 3J solar cell. According to the calculation, if we intend to obtain more than 32% conversion efficiency in the 5J solar cell under 1 sun AM0 spectrum, the open circuit voltage $V_{\rm oc}$ should not be lower than 4700 mV, the fill factor (FF) should not be less than 85%, and the current density needs to reach 11.3 mA/cm$^{2}$. Thus, we design three AlGaInP sub cells with different base region thicknesses of 200 nm, 350 nm and 500 nm. The $I$–$V$ curve and relevant electrical parameters by standard AM0 spectrum test are shown in Fig. 2(a).
cpl-34-2-028802-fig2.png
Fig. 2. (a) The $I$–$V$ curves of the AlGaInP/AlGaInAs/Ge 3J solar cell with the 1.98 eV band gap AlGaInP material. For the AlGaInP base region of 500 nm, $I_{\rm sc}$=97.6 mA, $V_{\rm oc}$=2.926 V, FF=84.3%, Eff=14.86%. For the AlGaInP base region of 350 nm, $I_{\rm sc}$=113.47 mA, $V_{\rm oc}$=2.931 V, FF=84.7%, Eff=17.28%. For the AlGaInP base region of 200 nm, $I_{\rm sc}$=122.68 mA, $V_{\rm oc}$=2.932 V, FF=83.5%, Eff=18.54%. (b) EQE curves of the AlGaInP/AlGaInAs/Ge 3J solar cell with the 1.98 eV band gap AlGaInP material.
When the thickness of the AlGaInP sub cell base region is 500 nm, the short circuit current is 97.60 mA and the conversion efficiency is 14.86%. With the decrease of the thickness, the short circuit current increases. When the thickness decreases to 200 nm, we obtain the short circuit current of 122.68 mA and the conversion efficiency of 18.54%. These results indicate that the AlGaInAs middle cell is the current limiting junction in the 3J solar cell. More incident light will transmit into the AlGaInAs sub cell when reducing the thickness of the AlGaInP sub cell. Therefore, the short circuit current of the 3J solar cell is improved.
Table 1. Current density of each sub cell with different base region thicknesses.
Thickness of the AlGaInP base (nm) AlGaInP sub cell $J^{\rm C}_{\rm sc}$ (mA/cm$^2$) AlGaInAs sub cell $J^{\rm C}_{\rm sc}$ (mA/cm$^2$)
200 11.82 10.05
350 13.23 9.59
500 15.67 8.33
The EQE curves of the 3J solar cell measured by the PVE300 are shown in Fig. 2(b). We neglect curves of the Ge bottom cells because the current of the 3J cell stays the same whatever the thickness of the bottom cell changes. The current intensity of each sub cell is calculated under the AM0 standard spectrum as listed in Table 1. First, when the AlGaInP (top cell) base region is 200 nm, its EQE in 300–625 nm has a serious decline to only 70%. A potential reason may be that the top cell is too thin to absorb 300–625 nm photons effectively. There is an EQE response in the spectrum from 425 nm to 625 nm shown in Fig. 2(b). The photons which should be absorbed by the AlGaInP materials reach the AlGaInAs sub cell and are absorbed by the AlGaInAs materials. With the increase of the top cell base region thickness, the abnormal response reduces gradually. Secondly, when the thickness of the AlGaInP base region is 350 nm, the EQE curve improves in the short wave direction compared with the 250 nm AlGaInP base region. In the long wave direction, however there is still insufficient photon absorption. The absorption coefficient of long wavelength photons is low and needs thicker materials for sufficient absorption.[8-10] Thus the EQE of the top cell in the long wave direction increases obviously until the thickness increases up to 500 nm. However, the current density of the middle cell is only 8.33 mA/cm$^{2}$. Thirdly, the current density of the AlGaInAs sub cell increases significantly with the decrease of the top cell thickness. The increase does not come from the photon absorption between 625 nm and 740 nm in the spectrum, but the photons in the range of 300–625 nm which are absorbed insufficiently by the AlGaInP top cell. With the thickness of the top cell reducing from 500 nm to 200 nm, the current density of the middle cell only increases from 8.33 mA/cm$^{2}$ to 10.05 mA/cm$^{2}$, much less than 11.3 mA/cm$^{2}$ as is expected. This is because there are many layers between the top cell base and the middle cell, such as the back field layer, the tunnel junction, the middle cell's window layer and the emitter layer. These epitaxial layers will absorb a part of the photons which leak from the top cell. Therefore, this part of the photons cannot all be converted into current of the middle cell. According to the above analysis, the AlGaInAs subcell is a current limitation junction. To solve this problem, we could extend the spectrum response range of the AlGaInAs subcell by increasing the band gap of the AlGaInP top cell.
cpl-34-2-028802-fig3.png
Fig. 3. Relationship between Al composition and EQE peak value and band gap of AlGaInP subcell (the thickness of the base region is 200 nm).
With the increasing Al composition, the band gap of AlGaInP also increases, but the concentration of oxygen (O) impurity elevates with the increase of the aluminum (Al) component.[7] The oxygen impurities affect on a minority of carrier lifetime results in a decline of the EQE peak value shown in Fig. 3. Therefore, a new method is adopted to increase the band gap of AlGaInP. The surfactant trimthylantimony (TMSb) is introduced when growing the AlGaInP emitter and base. The surfactant Sb not only can decrease the roughness of some materials, but also can increase the material band gap. There are four AlGaInP samples prepared with an Al composition of 10% and the TMSb-to-PH$_{3}$ mole ratios (Sb/P) of 0, $0.75\times10^{-4}$, $1.5\times10^{-4}$ and $3.0\times10^{-4}$, respectively.
cpl-34-2-028802-fig4.png
Fig. 4. Relationship between the band gap of AlGaInP and the Sb/P ratio.
Figure 4 shows that the band gaps of the four samples are 2.005 eV, 2.018 eV, 2.029 eV and 2.043 eV, respectively, and the band gap of the AlGaInP materials increases with the TMSb incorporation. A similar description can be found in growth of the GaInP material. The [$\bar{1}$$\bar{1}$0] P dimers produce the surface thermodynamic driving force for formation of the CuPt ordering structure during the GaInP growth. Sb atoms as a donor are added during the MOCVD growth of GaInP. The solubility of Sb is small in GaInP due to the large size of the Sb atom relative to P. In addition, the volatility of Sb is much less than that of P. Sb atoms may accumulate on the surface during growth due to these two causes. The P dimers are removed by the addition of Sb, and the GaInP ordering is eliminated too.[13,14] It is noticed that CuPt-B ordering in the AlGaInP material is observed. Therefore, incorporation of Sb can also eliminate ordering of AlGaInP, and can increase the band gap of AlGaInP. We grow another 3J solar cell by using Al$_{0.1}$GaInP with a $3.0\times10^{-4}$ Sb/P ratio where the thickness of the AlGaInP base is also 200 nm, and compare its EQE curves with Al$_{0.08}$GaInP without Sb incorporation shown in Fig. 5(a). First, the absorption edge of the AlGaInP sub cell with Sb incorporation appears to blueshift, and the AlGaInAs subcell exhibits a greater spectrum absorption range and higher EQE peak value. However, due to the thinner base thickness of the top cell, the AlGaInAs subcell still has an abnormal response from 500 nm to 600 nm. According to the calculation, the current density of the top cell is 11.56 mA/cm$^{2}$, and it is 11.31 mA/cm$^{2}$ for the middle cell which reaches our desired current density for the 5J solar cell. Secondly, under the condition of the same base thickness, Sb incorporation could increase EQE, and the peak value reaches 81.2%, which is higher than that of the Al$_{0.08}$GaInP without Sb. As mentioned above, the EQE of the AlGaInP sub cell decreases with the increasing Al composition, Sb incorporation also avoids this problem. We deduce that incorporation of Sb improves the crystal quality of the AlGaInP materials by surface reconstruction. The lower crystal defects lead to higher minority carrier lifetime. However, the EQE value of 81.2% is not high enough. There are two potential reasons as follows: Firstly, the thickness of the top cell base region is too thin. Secondly, AlGaInP alloys are sensitive to oxygen contaminant, and Sb incorporation could not hold back oxygen impurity incorporation in AlGaInP materials. The oxygen impurities as deep level traps also have an effect on minority carrier lifetime. Thirdly, in comparison of the $I$–$V$ curves of the two samples shown in Fig. 5(b), Sb incorporation not only improves the short circuit current of the 3J solar cell, but also increases its open circuit voltage. The value of 2.979 V is 47 mV higher than that of the Al$_{0.08}$GaInP without the Sb sample. Finally, we can see the constant current region in Fig. 5(b), which indicates that the shunt resistance of Al$_{0.1}$GaInP with the Sb sample (red color) is larger than that of another sample. The FF increases from 83.5% to 85.3%, which also verifies our prediction about the improvement of crystal quality. Based on the above discussion, we apply Al$_{0.1}$GaInP with the Sb material as the top cell into the 5J solar cell, and its base region is 200 nm. The GaInNAs sub cell is researched by other colleagues. Meanwhile, the 4J and 3J solar cells are grown as a comparison. The parameters of these structures are listed in Table 2.
cpl-34-2-028802-fig5.png
Fig. 5. (a) EQE curves of the AlGaInP/AlGaInAs/Ge 3J solar cell with and without Sb incorporation. (b) The $I$–$V$ curve of AlGaInP/AlGaInAs/Ge 3J solar cell with and without Sb incorporation: for Al$_{0.08}$GaInP without Sb, $I_{\rm sc}$=122.68 mA, $V_{\rm oc}$=2.932 V, FF=83.5%, Eff=18.54%; for Al$_{0.1}$GaInP with Sb addition, $I_{\rm sc}$=137.49 mA, $V_{\rm oc}$=2.979 V, FF=85.3%, Eff=21.56%.
Table 2. The 3J, 4J and 5J solar cell structures.
Spectrum range 3J sample 4J sample 5J sample
300–600/650 nm GaInP 1.89 eV 600 nm base GaInP 1.89 eV 600 nm base AlGaInP 2.043 eV 200 nm base
600/650–740 nm GaInAs 1.42 eV 3 μm base GaInAs 1.42 eV 3 μm base AlGaInAs 1.67 eV 1.7 μm base
740–900 nm GaInAs 1.42 eV 3 μm base
900–1150 nm GaInNAs 1.05 eV 2 μm base GaInNAs 1.05 eV 2 μm base
1150–1800 nm Ge 0.67 eV Ge 0.67 eV Ge 0.67 eV
We calculate the current density of each sub cell by the result of the EQE curve shown in Fig. 6. The values are 11.74 mA/cm$^{2}$, 11.21 mA/cm$^{2}$, 11.53 mA/cm$^{2}$, 11.07 mA/cm$^{2}$ and 18.67 mA/cm$^{2}$, respectively. From the data, we can see that the current densities of sub cells match with each other except the Ge bottom cell. The GaInNAs sub cell is a current limiting junction. As shown in Fig. 7, the short circuit current of the 5J solar cell is 134.96 mA, less than 203.76 mA of the 3J solar cell, and is basically equal to the short circuit current of the 4J solar cell. For the 4J solar cell, the current density of the top two-sub cell is usually 16–17 mA/cm$^{2}$, while the GaInNAs sub cell is about 11 mA/cm$^{2}$. There is a key influence on the conversion efficiency of the solar cell because of the serious current limitation in the GaInNAs sub cell. However, the 5J solar cell does not have such a problem. Further subdivision of the spectrum contributes to current matching of the sub cell. Thus we can obtain the maximum photo-current output with the low heating effect. According to the experience from other studies, a factor of 420 mV which is a gross rule of thumb should be subtracted from the band gap energy to obtain an upper limit of open circuit voltage in the solar cell. The difference value between actual $V_{\rm oc}$ and the upper limit represents the crystal quality. The smaller the difference is, the better the crystal quality is.[12] From Table 2, we can obtain the 5J solar cell $V_{\rm oc}$ upper limit of 4760 mV. Figure 7 shows that the actual $V_{\rm oc}$ of the 5J solar cell is 4399.66 mV, which is much higher than that of the 3J and 4J solar cells. However, it is still lower than the upper limit. This indicates that the crystal quality of the 5J solar cell is not very good. In addition, the fill factor of the 5J sample is 81.7%, which is lower than that of the 3J and 4J samples. The reasons are as follows: on the one hand, the poor crystal quality results in low shunt resistance. On the other hand, when the currents of sub cells match with each other, the $I$–$V$ curve's knee point of the solar cell is affected by both of them, which makes the FF decrease. When the currents of sub cells do not match with each other, the $I$–$V$ curve knee's point of the entire solar cell is determined by only one sub cell.[11]
cpl-34-2-028802-fig6.png
Fig. 6. EQE of the AlGaInP/AlGaInAs/GaInAs/ GaInNAs/Ge 5J solar cell.
cpl-34-2-028802-fig7.png
Fig. 7. The $I$–$V$ curves of the 3J, 4J and 5J solar cells. For 3J, $I_{\rm sc}=203.76$ mA, $V_{\rm oc}=2765.2$ mV, FF=85.9%, Eff=29.83%; for 4J, $I_{\rm sc}=135.56$ mA, $V_{\rm oc}=3144.69$ mV, FF=85.4%, Eff=22.3%; for 5J, $I_{\rm sc}=134.96$ mA, $V_{\rm oc}=4399.66$ mV, FF=81.7%, Eff=29.87%.
Finally, the conversion efficiency of the 5J solar cell is 29.87% under 1 sun AM0 standard spectrum, which is higher than that of the 3J and 4J solar cell samples. However, there is still a large gap compared with the best conversion efficiency in the world. We will continue to optimize the thickness of each sub cell and the doping concentration, and focus on improving the crystal quality of AlGaInP and GaInNAs sub cells to further improve the 5J solar cell's conversion efficiency. For the 5J solar cell, the current density of 11.3 mA/cm$^{2}$ is our objective. On account of the narrow spectrum absorption range (620–740 nm) and high Al composition, the current density of the AlGaInAs sub cell will be a current limitation junction. This work studies the $I$–$V$ curve and the EQE of the AlGaInP/AlGaInAs/Ge 3J solar cell whose top cell adopts the 1.98 eV AlGaInP material. The current density of the AlGaInAs sub cell cannot reach 11.31 mA/cm$^{2}$ whatever the thickness of the top cell is adjusted to. A straightforward way to solve this problem is to increase the band gap of AlGaInP. With the increasing Al composition, the band gap of AlGaInP increases, but the EQE of the AlGaInP sub cell drops obviously. Therefore, Sb incorporation when growing AlGaInP material is introduced to increase the band gap. In conclusion, the addition of Sb in the AlGaInP material could increase the band gap and could improve the crystal quality. Based on these advantages, firstly the spectrum absorption range of the AlGaInAs is extended from 600 nm to 740 nm, and the current density of this sub cell reaches 11.31 mA/cm$^{2}$. Secondly, the EQE peak value of the AlGaInP sub cell exceeds 80% at a higher Al composition of 10%. Thirdly, the open circuit voltage and the fill factor of the AlGaInP/AlGaInAs/Ge 3J solar cell are also improved. The AlGaInP material with the Sb incorporation is introduced into the 5J solar cell. Compared with 3J and 4J solar cells, we obtain a higher open circuit current and avoid current limitation of the GaInNAs sub cell in the 5J solar cell. Finally, we obtain a 5J upright lattice matched solar cell. The short circuit current is 134.96 mA, the open circuit voltage is 4399.66 mV, the FF is 81.7% and the conversion efficiency is 29.87%. References Status of C3MJ+ and C4MJ Production Concentrator Solar Cells at SpectrolabHigh Efficiency Multijunction Photovoltaic DevelopmentWafer bonded four-junction GaInP/GaAs//GaInAsP/GaInAs concentrator solar cells with 44.7% efficiencyAlInP benchmarks for growth of AlGaInP compounds by organometallic vapor-phase epitaxyGaInNAs/Ge (1.10/0.67 eV) double-junction solar cell grown by metalorganic chemical vapor deposition for high efficiency four-junction solar cell applicationInGaAsN solar cells with 1.0 eV band gap, lattice matched to GaAsBreakeven criteria for the GaInNAs junction in GaInP/GaAs/GaInNAs/Ge four-junction solar cellsFill factor as a probe of current-matching for GaInP 2 /GaAs tandem cells in a concentrator system during outdoor operationAdvances in High-Efficiency III-V Multijunction Solar CellsBand-gap control of GaInP using Sb as a surfactantSurfactant controlled growth of GaInP by organometallic vapor phase epitaxy
[1] Ermer J H, Jones R K, Hebert P, Pien P, King R R, Bhusari D, Brandt R, Al O T, Fetzer C M, Kinsey G S and Karam N 2012 IEEE J. Photovoltaics 2 209
[2] Wilt D M and Stan M 2012 Industrial & Engineering Chemistry Research 51 11931
[3] Dimroth F, Grave M, Beutel P, Fiedeler U, Karcher C, Tibbits T N D, Oliva E, Siefer G, Schachtner M, Wekkeli A, Bett A W, Krause R, Piccin M, Blanc N, Drazek C, Guiot E, Ghyselen B, Salvetat T, Tauzin A, Signamarcheix T, Dobrich A, Hannappel T and Schwarzburg K 2014 Prog. Photovoltaics 22 277
[4]https://www.ise.fraunhofer.de/en/press and media/pdfs zu presseinfos englisch/2014/press release new world record for solar cell efficiency at 46 percent.pdf
[5]Chiu P T, Law D C, Woo R L, Singer S B, Bhusari D, Hong W D, Zakaria A, Boisvert J, Mesropian S, King R R and Karam N H 2014 IEEE 40th Photovoltaic Specialist Conference (PVSC) 0011
[6]Dimroth F, Baur C, Meusel M, Riesen S V and Belt A W 2003 The 3rd World Confrence Photovoltaic Energy Conversion 1 p 616
[7] Bertness K A, Kurtz S R, Asher S E and Reedy R C 1999 J. Cryst. Growth 196 13
[8] Zhang X B, Chen B Z, Pan X, Wang L, Ma D F, Zhang Y, Yang C B and Wang Z Y 2015 J. Phys. D 48 475106
[9] Kurtz S R, Allerman A, Jones E D, Gee J M, Banas J and Hammons B E1999 Appl. Phys. Lett. 74 729
[10] Friedman D J and Kurtz S R 2002 Prog. Photovoltaics 10 331
[11] McMahon W E, Emery K E, Friedman D J, Ottoson L, Young M S, Ward J S, Kramer C M, Duda A and Kurtz S 2008 Prog. Photovoltaics 16 213
[12] King R R, Law D C, Edmondson K M, Fetzer C M, Kinsey G S, Yoon H, Krut D D, Ermer J H, Sherif R A and Karam N H 2007 Adv. OptoElectron. 2007 29523
[13] Shurtleff J K, Lee R T, Fetzer C M and Stringfellow G B 1999 Appl. Phys. Lett. 75 1914
[14] Lee R T, Shurtlef J K, Fetzer C M, Stringfellow G B, Lee S and Seong T Y 2000 J. Appl. Phys. 87 3730