Carriers transport properties in GaInP solar cells grown by molecular beam epitaxy

Solid State Communications 200 (2014) 9–13

Contents lists available at ScienceDirect

Solid State Communications journal homepage: www.elsevier.com/locate/ssc

Carriers transport properties in GaInP solar cells grown by molecular beam epitaxy P. Dai a,b, S.L. Lu a,n, M. Arimochi c, S. Uchida c, T. Watanabe c, X.D. Luo d, H. Yang a a Key Laboratory of Nanodevices and Applications, Suzhou Institute of Nano-Tech and Nano-Bionics, Chinese Academy of Sciences, Suzhou Industrial Park, Ruoshui Road 398, Suzhou, PR China*Corresponding author. b Graduate University of Chinese Academy of Sciences, Beijing 100049, PR China c Advanced Material Laboratories, Sony Corporation, Atsugi Tec. 4-14-1 Asahi-cho, Atsugi-shi, Kanagawa 243-0014, Japan d Jiangsu Key Laboratory of ASIC Design, Nantong University, 226019, PR China

art ic l e i nf o

a b s t r a c t

Article history: Received 18 June 2014 Received in revised form 30 August 2014 Accepted 10 September 2014 by Michael Manfra Available online 18 September 2014

The transport properties of carriers in GaInP solar cells grown by molecular beam epitaxy are investigated by temperature-dependent current–voltage (I–V) measurements. In contrast to GaInP/ AlGaInP heterostructure, a long PL decay time is observed in GaInP/AlInP, which is ascribed to a lower interface recombination due to an improved carriers' confinement in the case of the high-energy barrier. However, the series resistance induced by the high potential barrier at GaInP/AlInP interface due to a big valence band offset prevents the improvement of solar cell's performance. An S-shape like I–V characteristic observed at low temperatures indicates that the transport of major carriers is limited by the barrier. A calculation based on the combination of a normal photovoltaic device with a barrieraffected thermal carriers transport explicitly explains this abnormal I–V characteristic. Our study demonstrates the critical role of the barrier-induced series resistance in the determination of solar cell's performance. & 2014 Elsevier Ltd. All rights reserved.

Keywords: A. Semiconductor B. Molecular beam epitaxy C. Solar cell D. Transport and optical properties E. Temperature-dependent I–V measurement

1. Introduction The efficiencies of III–V compound semiconductor multi-junction solar cells (SCs) continue to rise with optimization of the device design and improvements in material quality [1–3]. The transport of majority carriers through potential barriers together with the recombination of minority carriers at the interfaces plays an important role in the determination of the SC's performance. An efficient back surface field (BSF) layer can confine the photo-generated minority carriers and at the same time, ensure the transportation of majority carriers to be efficiently collected [4–7]. Numerical simulations about the transport characteristics of majority carriers have been done [8,9], however, the experimental study of combination of the transportation of majority carriers and the recombination of minority carriers in the function of BSF is scarce. Especially, for low-temperature space application and for high-intensity concentrating photovoltaic application, the effect of the resistance resulting from the high potential barrier on the SC's performance will become more serious. Furthermore, the physical property of a heterostructure was greatly affected by different growth methods [10]. Metal-organic

n

Corresponding author. E-mail address: [email protected] (S.L. Lu).

http://dx.doi.org/10.1016/j.ssc.2014.09.012 0038-1098/& 2014 Elsevier Ltd. All rights reserved.

chemical vapor deposition (MOCVD) technique is generally used for the epitaxial growth for the SCs. As one of the most important epitaxial techniques, the research of SCs grown by molecular-beamepitaxy (MBE) has not been extensively investigated. And the performance of the earlier GaAs SCs grown by MBE is worse than those obtained by MOCVD growth due to the low growth temperature and the presence of isolated defects [11,12]. However, a highly efficient MBE-grown GaInP/GaAs/GaInAsN triple-junction cell was recently reported by Solar Junction [13]. The experimental results of our group also demonstrated that MBE-grown phosphorus-containing III–V compound semiconductor solar cells are comparable to the case of MOCVD growth [14,15]. A comparative study of MBE-grown photovoltaic device is necessary to improve SC's efficiency as well as optimize the device's performance [16,17]. In this paper, we studied the effect of carriers' transport property with different BSF layers on the performance of the GaInP SC by using time-resolved photoluminescence (TRPL) and temperaturedependent current–voltage (I–V) measurements. It is found that a high-energy AlInP barrier could confine the photo-generated minority carriers effectively. However, in contrast to AlGaInP BSF, the highenergy barrier also results in a large series resistance in GaInP SC. An S-shape like I–V characteristic which is observed at low temperatures indicates that the limitation is induced by the barrier to the majority

10

P. Dai et al. / Solid State Communications 200 (2014) 9–13

3. Results and discussion Fig. 2 shows the PL decay curves of the two heterostructures. In the case of AlInP barrier, the PL decay time of 500 ps is obtained while the decay time of only 120 ps is observed for the case of AlGaInP barrier. Since the open voltage (Voc) of a photovoltaic device is mainly determined by the recombination loss, therefore, a long PL decay time indicates that the AlInP barrier is more promising as the BSF for SCs as soon as only the recombination of minority carriers are

PL int. (a.u.)

0.9

2

Energy (eV)

Energy (eV)

1.8

0

-2 0.4

0.0 GaInP

0.8

1.2

Depth ( μm)

Al(Ga)InP

2.05 2.10 Phot. Ene. (eV) GaAs

1.15 1.16 1.17 1.18 1.19 1.20 1.21 Fig. 1. Calculated energy band diagram at the BSF interface. The inset (left) shows the energy band diagram of the whole p–n junction, the right inset shows the PL spectrum of AlGaInP at room temperature.

τ=500 ps

AlInP barrier

τ=120 ps AlGaInP barrier

0

100

200

300

Time delay (ps) Fig. 2. PL decay curves of the two heterostructures.

)

The growth was performed by Veeco GEN20A dual-chamber all solid-state MBE machine equipped with a valved phosphorous cracker cell and a valved arsenic cracker cell. The structures of GaInP SCs grown on p-type GaAs substrate are composed of a p-GaInP base layer, an n þ -GaInP emitter layer and an AlInP as the window layer to decrease the surface recombination. The investigated two SCs have the same doping densities of GaInP base layer of 5  1016 cm  3 and BSF layer of 2  1018 cm  3, respectively. The only variation of the structures lies in the BSF layer material. For the first cell (hereafter referred as A), the BSF layer was p þ AlGaInP with the band gap energy of 2.08 eV and the second cell (hereafter referred as B) was p þ -AlInP layer. Fig. 1 presents the calculated energy band diagram at the BSF interface. The inset (left) shows the energy band diagram of the whole p–n junction. The inset in the right shows the PL spectrum of AlGaInP at room temperature (RT). The ΔEc and ΔEv between InGaP and AlInP used in the calculation are 0.25 eV and 0.13 eV, while in AlInGaP and InGaP are 0.134 eV and 0.066 eV, respectively [18,19]. The large conduction and valence band offsets are obviously observed in the case of AlInP/GaInP interface. The photovoltaic devices were processed following the standard III–V SC device art. The detailed growth and device fabrication art were all listed in our recent report [14]. In order to study the optical properties at the interface of GaInP base layer and BSF, two similar heterostructures of p-AlGaInP/p-GaInP/p-AlGaInP and p-AlInP/p-GaInP/p-AlInP which have the same doping density of respective layer as the corresponding layer within GaInP SCs were grown. The transient PL evolution was measured by using a synchroscan streak camera with a time resolution of 15 ps. The I–V characteristics were recorded using a voltage source and a current meter type (Keithley 2440) in a two-terminal configuration under the standard air mass 1.5 global (AM1.5G) illumination.

(

2. Experimental details

PL Intensity (arb. units)

carriers' transport. A comparative study demonstrates the critical role of the barrier-induced series resistance in the SC's performance.

Fig. 3. The J–V characteristics of two GaInP SCs (A and B) under AM1.5G at 1 sun, inset shows the EQE of the two SCs.

concerned. Fig. 3 shows the current density–voltage (J–V) characteristics of two GaInP SCs (A and B) under AM1.5G at 1 sun. For sample A, an efficiency of 13.8% with the Voc of 1.27 V, a short-circuit current (Jsc) of 12.9 mA/cm2 and an fill factor (FF) of 84% is obtained. While for sample B, an efficiency of 13.5% with the Voc of 1.29 V, a Jsc of 12.6 mA/cm2, and an FF of 83% is obtained. As is expected, the open voltage of sample B with AlInP barrier is larger than that of sample A. Voc is greatly affected by the recombination of minority carriers in the solar cell, the larger Voc in sample B is attributed to the reduced surface-recombination. Considering the different band gap energies of AlGaInP and AlInP, a larger conduction band offset with respect to the GaInP base layer provides a potential barrier for minority electrons and increases the electrons confinement. The four times longer PL decay time due to reduced surface- recombination results in about 20 mV increase in open voltage.[20] However, because the operation of the photovoltaic device includes the process of photon absorption, carriers transport and collection, a high performance of a photovoltaic device relies on many parameters. As can be seen from Fig. 3, with increasing applied voltage, the series resistance (Rs) becomes larger for sample B. The calculated total Rs of sample A is 15.8 Ω, much smaller than sample B of 29.1 Ω. Because of the same device fabrication method, the difference can be regarded as from the intrinsic structures of the different devices. The long minority carriers related PL decay time in the case of the AlInP barrier results in large Rs, which prevents the improvement of Jsc and FF and therefore the conversion efficiency. The inset in Fig. 3 shows the external quantum efficiency (EQE) comparison between the

P. Dai et al. / Solid State Communications 200 (2014) 9–13

Current (mA)

two GaInP cells. Compared to sample B with AlInP barrier, the EQE of sample A with AlGaInP barrier is enhanced at the longer wavelength part, benefiting from the better current collection. The calculated current densities by integrating the measured EQE with AM1.5G solar spectrums are 12.92 mA/cm2 for sample A and 12.58 mA/cm2 for sample B, which excellently match the measured values of Jsc in J  V measurement for the two cells. Moreover, for sample B, J–V curve exhibits a kind of curvature, which is similar to the numerical simulations by Gudovskikh et al. [9], resulting from the limitation of the majority carriers passing through the high potential barrier. To further investigate the effect of AlGaInP and AlInP BSFs on the performance of GaInP SC, temperature-dependent I–V measurements were performed at the dark condition and under the light illumination, respectively. The I–V characteristics of the SCs are significantly influenced by the temperature and the optical carrier injection as can be seen from Fig. 4. Fig. 4(a) and (b) shows

Current (mA)

the I–V curves of samples A and B at RT, 150 K, 50 K and 5 K without light illumination. With decreasing temperature, an increased series resistance-affected I–V characteristic is exhibited and even more serious for sample B with AlInP barrier. Fig. 4(c) and (d) shows the I–V curves under light illumination. For sample A, at 150 K, the shape of I–V curve is still normal. While, for sample B, an S-shape like I–V characteristic is observed at 150 K and the series resistance is increased greatly. At RT, the J–V curve recovered to the normal behavior. It is reasonable to ascribe the different behaviors of I–V curves of the two devices to the effect of the different BSF layers. The I–V characteristics shown in Figs. 3 and 4 clearly exhibit a kind of curvature shape at RT and an S-like shape at low temperatures for sample B. This behavior is typically thought to be affected by the high potential barrier for the majority carriers passing through the BSF. The formation of this barrier is dominated by the large value of ΔEv at the GaInP/AlInP interface [9]. In the case of an existing barrier, the

0

0

-1

-1

-2

0

5K

5K

50 K

50 K

150 K

150 K

RT

RT

1 Voltage (V)

2

11

-2

9

9

6

6

0

1 Voltage (V)

2

3

150 K RT

3

3

150 K

0

0

RT 0

1 Voltage (V)

2

0

1 Voltage (V)

2

3

Fig. 4. (a) and (b) show the I–V curves of samples A and B at RT, 150 K, 50 K and 5 K without light illumination; (c) and (d) show the I–V curves under light illumination for different temperatures.

P. Dai et al. / Solid State Communications 200 (2014) 9–13

transport through the barrier is accomplished by tunneling and thermal assistant carriers transport. The tunneling process is temperature less independent, while the latter one is temperature dependent. The total I–V characteristic can be described as the superposition of an illuminated p–n junction (I1 ) and a Schottky barrier affected thermal carriers transport (I2) [21]: I ¼ I1 þ I2


  qðV þ IRs Þ ðV þ IRs Þ 1  I 1 ¼ I L  I s exp kT Rsh pffiffiffiffi!   qφ  α V B exp I 2 ¼ AT 2 exp kT kT

Calculation Exp.

Current (mA)

0

-5

-10

1

B, AlInP BSF

15 14 13

A D

12 11

C

12 14 16 18 20 22 24 26 28 30

5

0

Ref. 13

16

10

where IL is the short-circuit current when there is no parasitic resistance, Rsh is the shunt resistance, Rs is series resistance, φΒ is the height of the barrier, α ¼ sqrt(q/4πεd) is a constant and d is the effective width of the barrier. Fig. 5 presents the calculated and experimental results with φΒ of 14 meV and effective barrier width of 30 nm. The good agreement indicates that the limitation to the majority carriers' transport results in a great influence on the I–V characteristic of SC. As a result, an additional series resistance induced by the barrier will increase. Since this additional series resistance is exponentially proportional to the height of the barrier, a small change in the barrier height might result in a great influence on the I–V characteristic of SCs. The BSF-induced majority carriers' transport also can be affected by different doping densities of base layer within the SC because the distance of the valence band to the Fermi energy level is determined by the doping density. A low doping density in the base layer will reduce the offset of the effective barrier height. We studied the J–V characteristics of GaInP SCs with different doping densities of the base layer. Fig. 6 shows the J–V curves of another two GaInP SCs (C and D). These two SCs have the different doping densities of GaInP base layer

-15

17

Efficiency (%)

12

2

3

Voltage (V)

Seires resistance (Ω ) Fig. 7. The photovoltaic efficiency of GaInP cells as a function of series resistance for both AlGaInP and AlInP barriers.

but the same BSF layer (AlGaInP). The doping densities of the BSF layers are the same as sample A. For the base layer, sample C has the doping density of 4  1017 cm  3 and sample D has the doping density of 1  1017 cm  3. In case of sample C, an efficiency of 10.5% with the Voc of 1.277 eV, a Jsc of 11.9 mA/cm2, and an FF of 69% is obtained. While for sample D, the efficiency is 12.9% with the Voc of 1.27 eV, Jsc of 12.8 mA/cm2, and FF of 79.7%. The notable difference is the series resistance in the two structures, Rs of sample C is 23.1 Ω while 17.7 Ω for sample D. Because the resistivity decreases with increased doping density, therefore, the base layer itself will not result in this difference. The lager series resistance might be in part ascribed to the increased barrier height in the case of the low base doping density [21,22]. Fig. 7 summarized the photovoltaic efficiency of GaInP cells as a function of series resistance for both AlGaInP and AlInP barriers. With increasing resistance, the conversion efficiency linearly decreased, which is correlated with the energy loss. However, for AlInP BSF, it follows the different way and shows a larger tolerance of series resistance. The larger band offset between AlInP and GaInP provides a barrier to effectively confine the minority electrons and also to limit the majority carriers' transport. The combination of these two functions together determines the conversion efficiency. The higher doping density of the AlInP BSF leads to the reduction of the effective height of the majority carrier barrier. In our growth system, beryllium (Be) is used as the p type dopant. During the AlGaInP and AlInP growth, a typical temperature of approximately 530 1C is used to ensure the material quality. The heavily doped p-AlInP is a serious problem for MBE growth. Because the highest doping density we can achieve is limited to 2  1018 cm  3, comparable to the hole concentration of 3  1018 cm  3 obtained by gas source molecular beam epitaxy (GSMBE) [23]. A high quality p-AlInP is desirable to improve the performance of the photovoltaic device for AlInP as the BSF.

Fig. 5. Calculated and experimental results of I–V curves for sample B at 150 K.

15

2

Jsc ( mA/cm )

10

4. Conclusions

C D

5 0 -5 -10 -15 0.0

0.4 0.8 Voltage (V)

1.2

Fig. 6. The J–V curves of GaInP SCs with different doping densities of GaInP base.

In summary, the correlation between the transport of majority carrier and the optical property of minority carrier in GaInP solar cells with different barrier heights grown by MBE is investigated by TRPL and temperature-dependent I–V measurements. The larger band offset between AlInP and GaInP provides a barrier to effectively confine the minority electrons and also to limit the transport of majority carriers. The critical role of the barrierinduced series resistance in the SC's performance is comparatively studied. Our research demonstrates that for low-temperature space application and for high-intensity concentrating photovoltaic application, the effect of the resistance resulting from the high potential barrier on the SC's performance becomes more serious.

P. Dai et al. / Solid State Communications 200 (2014) 9–13

Acknowledgments The authors would like to thank Prof. Jiannong Wang of HKUST for their helpful time-resolved PL measurement. This work is supported in part by NSFC (61176128) and the SINANO-SONY joint program (Grant nos. Y1AAQ11002 and Y2AAQ11004). References [1] R.R. King, D.C. Law, K.M. Edmondson, C.M. Fetzer, G.S. Kinsey, H. Yoon, R.A. Sherif, N.H. Karam, Appl. Phys. Lett. 90 (2007) 183516. [2] J.F. Geisz, D.J. Friedman, J.S. Ward, A. Duda, W.J. Olavarria, T.E. Moriarty, J.T. Kiehl, M.J. Romero, A.G. Norman, K.M. Jones, Appl. Phys. Lett. 91 (2008) 023502. [3] W. Guter, J. Schoene, S.P. Philipps, M. Sterner, G. Siefer, A. Wekkeli, E. Welser, E. Oliva, A.W. Bett, F. Dimroth, Appl. Phys. Lett. 94 (2009) 223504. [4] Beatriz Galiana, Ignacio Rey-Stolle, Mathieu Baudrit, Iván García, Carlos Algora, Semicond. Sci. Technol. 21 (2006) 1387–1392. [5] D.J. Friedman, S.R. Kurtz, A.E. Kibbler, J.M. Olson, in: Proceedings of the 22nd IEEE Photovoltaic Specialists Conference, 1991, p. 358. [6] N.H. Rafat, S.M. Bedair, in: IEEE Proceeding of the 4th World Conference on Photovoltaic Energy Conversion, 1994, p. 1906. [7] T. Takamoto, E. Ikeda, H. Kurita, Appl. Phys. Lett. 70 (1997) 381. [8] A. Kanevce, J.M. Olson, W.K. Metzger, in: Proceedings of the 35th IEEE Photovoltaic Specialists Conference, 2010, p. 2066.

13

[9] A.S. Gudovskikh, N.A. Kaluzhniy, V.M. Lantratov, S.A. Mintairov, M.Z. Shvarts, V.M. Andreev, Thin Solid Films 516 (2008) 6739. [10] C. Ghezzi, R. Magnanini, A. Parisini, L. Tarricone Enos Gombia, M. Longo, Phys. Rev. B 77 (2008) 125317. [11] S.P. Tobin, C.B. Vernon, S.J. Wojtczuk, M.R. Melloch, A. Keshavarzi, T.B. Stellwag, S. Venkatensan, M.S. Lundstrom, K.A. Emery, IEEE Trans. Electron Device 37 (1990) 469. [12] P. Leinonan, M. Pessa, J. Haapamaa, K. Rakennus, IEEE Xplore-Photovolt. Specialists Conf. (2000), http://dx.doi.org/10.1109/PVSC.2000.916098. [13] M.A. Green, K. Emery, Y. Hishikawa, W. Warta, E.D. Dunlop, Prog. Photovolt.: Res. Appl. 21 (2013) 1–11. [14] S.L. Lu, L. Ji, W. He, P. Dai, H. Yang, M. Arimochi, H. Yoshida, S. Uchida, M. Ikeda, Nanoscale. Res. Lett. 6 (2011) 576. [15] L. Ji, S.L. Lu, A. Tackeuchi, L.F. Bian, Y.Y. Wu, P. Dai, M. Arimochi, S. Uchida, T. Watanabe, H. Yang, Solar Energy Mater. Solar Cells 127 (2014) 1–5. [16] P. Dai, S.L. Lu, Y.Q. Zhu, L. Ji, W. He, M. Tan, H. Yang, M. Arimochi, H. Yoshida, S. Uchida, M. Ikeda, J. Cryst. Growth 378 (2013) 604. [17] P. Dai, S.L. Lu, L. Ji, W. He, L.F. Bian, H. Yang, M. Arimochi, H. Yoshida, S. Uchida, M. Ikeda, J. Semicond. 34 (2013) 104006. [18] V. Meyer, J.R. Meyer, L.R. Ram-Mohan, J. Appl. Phys. 89 (2001) 5815. [19] Y. Ishitani, et al., J. Appl. Phys. 80 (1996) 4592. [20] Handbook of Photovoltaic Science and Engineering, in: A. Luque, S. Hegedus (Eds.). [21] R. Hoheisel, A.W. Bett, IEEE J. Photovolt. 2 (2012) 398. [22] D.A. Clugston, P.A. Basore, in: Proceedings of the 26th IEEE Photovoltaic Specialists Conference, 1997, p. 207. [23] T. Yokosuka, A. Takamori, M. Nakajima, Appl. Phys. Lett. 58 (1991) 1521.