Search for content and authors

Strain Effect for Different Phosphorus Content of InGaAs/GaAsP Super-Lattice in GaAs p-i-n Single Junction Solar Cell.

Kentaroh Watanabe 1Yunpeng Wang 1Hassanet Sodabanlu 1Masakazu Sugiyama 2Yoshiaki Nakano 1,2

1. Research Center for Advanced Science and Technology, The University of Tokyo (RCAST), 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan
2. The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan


1.    Introduction

 A multiple quantum wells (MQWs) solar cell have an advantage of realizing higher open-circuit voltage (VOC) compared with a single junction solar cell composed by host barrier material alone [1]. Because the 1.2 eV effective band gap of optical absorber can be available with low degradation of PV performance, the InGaAs/GaAsP MQWs is one of the promised candidates as a middle cell of current-matched triple-junction solar cell [2]. Compensating a strain force on InGaAs well and GaAsP barrier layer each other during the epitaxial growth, totally minimized strain lead maximum PV performance [3]. A super-lattice (SL) consist of InGaAs well and extremely thin GaAsP barrier (~3nm) gives improved open-circuit voltage degradation at 1 SUN illumination compared to relatively thick (~10 nm) barrier thickness supported by improved carrier extraction efficiency assisted by tunneling effect through the thin barrier [4]. And also the InGaAs/GaAsP SL absorber in the i-region of GaAs p-i-n structure given better PV performance in concentrated sun irradiation by enhanced VOC maintaining the its short-circuit current (JSC) increment [5]. From the aspect of high-efficient correction of photo-generated carrier, both the crystal quality of SL and the carrier dynamics including a tunnel effect thorough the barrier layer is important issue for the SL solar cells. In this study, we prepared a different phosphorus content of InGaAs/GaAsP SL in GaAs p-i-n solar cells to evaluate the effect of barrier height for PV performance of SL cell.

2.    Experiment

2.1.MOVPE growth of SL in p-i-n structure of GaAs

Figure 1 shows the schematic structure of GaAs p-i-n single junction solar cell with 60 periods of InGaAs/GaAsP super-lattice in the i-region. All of our samples were grown by a planetary metal-organic vapor phase epitaxy (MOVPE) system (AIXTRON, AIX2000HT) on n-type GaAs (001) substrate. A strain-compensated SL play a role of narrow bandgap photon absorber in GaAs p-i-n solar cell, should be resulted in increased current density (JSC) compared to the GaAs bulk cell. During our MOVPE growth, substrate temperature was fixed to 610 oC and the inside pressure of growth chamber was controlled to 100 mbar. All of GaAs layer and the window In0.5Ga0.5P layer were grown under the 15 of V/III ratio. As a III groups of MO source, we used the tri-methyl-gallium (TMG) and the tri-methyl-indium (TMIn), respectively. And also the tertiary-butyl-arsine (TBA) and the tertiary-butyl-phosphine (TBP) were used as a MO sources for V groups. Highly purified hydrogen was used as a carrier gas. We fabricated the SL samples with three different partial pressure of TBP by changing the source flow late. For the SL structure, the thickness of well and barrier designed as 5 nm and 3 nm, respectively. And total thickness of i-region was 800 nm for all samples. We also fabricated bulk GaAs p-i-n single junction solar cell as a reference sample whose thickness of i-region was same with SL cells. As an in-situ strain sensor, we used multi-beam stress monitor method (MOSS using the Laytech, Epicurve System) for measuring the curvature of bowed wafer by accumulated strain during InGaAs and GaAsP growth. And during the SL growth, we applied the hydrogen purge for 1 s after InGaAs well gworth and subsequently also applied the 1sec TBP surface treatment before GaAsP barrier growth for improvement of interfacial abruptness between well and barrier [6].

After forming the backside planer and topside grid-pattern metal, mesa-etching by aqueous ammonia solution was applied for isolation each cell devices. Finally, the dual-layer of anti-reflection coatings (ARC) was applied by ZnS/SiO2 sputtering on the top surface.

2.2.Structure evaluation

After MOVPE growth of SL samples, the high resolution x-ray diffraction (HRXRD) (004) omega-2theta scan was attempted for estimating the thickness and atomic composition for each well and barrier in SL layer. Consequently continuous wave photoluminescence (CWPL) and time-resolved photoluminescence (TRPL) was measured for each SL sample to evaluate the effective band gap energy and to observe the dynamics of photo-generated carrier at room temperature. For the measurement of CWPL, wavelength of the excitation laser diode was 784nm and varied excitation power range was attempted in the range of 1.28-12.8 mW/mm2. The TRPL measurement was carried out by time-correlated single photon counting (TCSPC) method illuminating excitation pulsed laser with 12 pJ/pulse of energy. The excitation laser wavelength was also 784 nm and the pulse repetition frequency was 10 MHz. At the TRPL measurement, by using the monochromater, only 7nm of bandwidth of PL light was extracted at the peak wavelength obtained from the CWPL spectrum. The photon counting rate was set around 1% by regulating the number of PL photon with neutral density filter. Estimated time resolution of TRPL measurement was approximately 0.4 ns determined by instrumental response function (IRF).

2.3.Evaluation of solar cell

For obtained SL and reference bulk GaAs cells, PV performance was also evaluated. The current-voltage (I-V) characteristics was observed under AM1.5 G (100 mW/cm2) simulated illumination to determine their short-circuit current density (JSC), open-circuit voltage (VOC), fill factor (FF) and power conversion efficiency (η). Spectral response of the solar cell was measured by illuminating calibrated monochromatic light for estimating the external quantum efficiency (EQE). All of the measurement for PV characterization was carried out with sample temperature fixed to 25 OC as a standard condition.

3.    Result and Discussion

3.1. Detail sample structure from XRD and MOSS measurement

 Figure 2 shows the result of XRD measuremnt. Sufficiently sharp satellite peaks from InGaA/GaAsP SL layer was observed, indicating high crystal quality of sample. Fitting the model to the observed XRD rocking curve, we have estimated each well and barrier composition and thickness, respectively. Table.2 shows the fitting result of each sample. Approximately 5.1nm thickness of and 11% of indium content for well InGaAs layer was fixed for the series of samples. And the barrier thickness is also similar to 3.1 nm for each sample. The gradually different phosphorus content of barrier layer was obtained corresponding to the phosphorus partial pressure difference during SL growth. For the sample of No. 788, 794 and 900, the phosphorus content was estimated as 13%, 18% and 24%, respectively. The total accumulated strain situation was clearly shown in Fig. 3. Plotting the transient of wafer curvature during the growth, each SL samples showed significant wafer bowing indicated the accumulated compressive strain direction. Because the compressive strain should be provided by InGaAs well layer on GaAs substrate in this SL system, so compensating tensile strain was insufficient for compensating total strain. Comparing the curvature transition for the sample No. 788 and No. 900, total accumulated compressive strain was reduced by increasing the phosphorus content in GaAsP barrier.

3.2. PL result

 Figure 4 shows the measured CWPL spectrum of the SL cell samples of No.788, 794, 900. Each PL spectrum indicated the two individual peaks at 874 nm and around 915 nm, should be corresponding to the bulk GaAs bandgap and the SL effective bandgap energy, respectively. The PL peaks corresponding to the SL effective bandgap also shifted to shorter wavelength with increasing phosphorus content in GaAsP barrier. This result seems to be due to the increased confined energy level in the well accompanied with the higher potential barrier of increased phosphorus content in GaAsP. Relative intensity of PL peaks for each sample became stronger for the higher phosphorus content SL sample. Because the sample structure is very similar except to the barrier GaAsP composition in SL layer, this result may indicate that the reduced total compressive strain by increasing phosphorus content in barrier resulted in reduced defect density in SL layer. Figure 5 showed the excitation power dependent CWPL peak intensity at the wavelength of the bulk GaAs and the SL effective bandgap. In all SL cell samples, the SL related peak became rapidly stronger than the GaAs rerated peak. The photo-generated carriers in SL layer were forced to move to n and p GaAs region accelerated by built-in potential in the SL put in GaAs p-i-n structure. Because the case of larger number of carrier with high excitation rate resulted in the potential screening by escaped carriers in n and p region, therefore weaken effective built-in potential should provide increased carrier recombination rate in SL region. Figure 6 showed the TRPL decay curve at the GaAs and the SL effective bandgap peak of CWPL. Showing TRPL decay curve compared for two different strain condition of sample No788 and 900. The decay curve of SL related peak was almost similar to the decay curve provided from IRF. Therefore, the time-scale of photo carrier recombination in the SL layer is estimated shorter than 0.4 ns. On the other hand, the decay curve of GaAs related peak showed two different time scale; one is sufficiently faster than IRF (0.4 ns) and secondary tail of decay curve with ~1 ns of timescale. The sample structure include the flat potential region only allowing for n- or p- type GaAs layer, so the longer lifetime of PL decay is likely indicate the carrier diffusion timescale in flat-band potential of n- or p- type GaAs region.

3.3. PV performance

 Finally, all of the SL cells and the reference GaAs cell evaluated as a photovoltaic performance shown in Fig. 7 and Fig. 8, and summarized in Table. 3. The SL cell samples showed similar JSC around 30 mA/cm2, ~2 mA/cm2 larger than the reference bulk GaAs cell. Figure 7 shows the normalized EQE spectrum of different phosphorus content of SL cells and the reference bulk GaAs cell. All of samples of SL layer inserted in GaAs p-i-n structure absorbs longer wavelength photons beyond the GaAs cutoff wavelength λ=870 nm up to λ=930 nm, resulted in the increase in JSC. The shoulder of extended EQE spectrum in SL cells slightly changed by consistent to the effective bandgap energy examined by CWPL. The series of SL cells showed increased η with larger P content sample, and the best sample was No.900 with η=24.35%. The efficiency of SL cell depends on strongly on the VOC clearly different with phosphorus content in SL. Figure 8 shows the I-V characteristics of each cell, showed the improvement of VOC in the SL cell samples with reduced total accumulated strain. For the larger phosphorus content of barrier GaAsP should be resulted in the higher potential barrier for the electrons and holes in well InGaAs layer. In these series of SL samples, the thickness of indium content of the well is constant, so the result of larger η of SL cell with larger phosphorus content seems to be opposite tendency. Hence, we can conclude the around 3.1 nm of barrier thickness is sufficiently thin for realize the tunneling effect through the barrier transporting carrier to n- region for electrons and p- region for holes. And we also conclude the total strain accumulation in SL layer degrade VOC possibly due to the density of defects introduced not lattice-mismatch itself but by the strain accumulation in MOVPE growth.

4.    Conclusion

MOVPE gown super-lattice consist of the InGaAs well and GaAsP barrier inserted in the i-region of GaAs p-i-n structure was one of the useful approach to change the effective band gap of solar cell from the 1.42 eV for bulk GaAs toward smaller remaining high-crystal quality by strain compensating. We tried approximately 3.1 nm of very thin barrier thickness of SL with different phosphorus content of GaAsP barrier by changing different partial pressure of TBP during SL growth. Actually reduced total compressive strain accumulation was examined by in-situ curvature measurement for high-phosphorus content barrier of SL. CWPL and TRPL measurement was also revealed the compressive-strain-reduced SL sample relatively strong CWPL intensity and small carrier recombination timescale in SL than 0.4 ns. The series of solar cell samples with different strain accumulated SL was evaluated their performance. The smaller compressive accumulated SL cell showed recovery of open-circuit voltage and resulted in larger power conversion efficiency in spite of taller height of barrier potential in SL layer. This result indicated that the defect related PV degradation possibly recovered by reduced total accumulated strain. And also the enhanced carrier transportation assisted by tunneling effect through the thin-barrier apparently contributed to the PV performance.


A part of this study is supported by Research and Development of Innovative solar Cell program, New Energy and Industrial Technology Development Organization (NEDO), Japan.


[1] K. Barnham, et al., J. Appl. Phys., 80, p1201 (1996)

[2] N. J. Ekins-Daukes, et al., Appl. Phys. Lett., 75, p4195 (1999)

[3] H. Sodabanlu, et al., Jpn. J. Appl. Phys. 51, p10ND16 (2012)

[4] Y. Wang, et al., IEEE J. Photovoltaics, 2, p387 (2012)

[5] K. Watanabe, et al., AIP conf. Proc. of CPV8, 1477, p40 (2012)

[6] S. Ma, et al., J. Cryst. Growth, 352, p245 (2012)

Fig.1 Schematic sample structure of GaAs p-i-n solar cell with InGaAs/GaAsP SL in the intrinsic GaAs layer.


Fig.2 HRXRD (004) omega-2theta scans for each different phosphorus content of SL cells.


Fig.3 Time-dependent wafer curvature for the SL samples with different phosphorus partial pressure during SL growth.


Fig.4 CWPL spectrum for the SL cell samples of No.788, 794 and 900.


Fig.5 Excitation power dependency of the CWPL peak intensity of the SL cells at the wavelength corresponding to the GaAs bandgap (filled simbol) and the SL bandgap (blank simbol).



Fig.6 Time-dependent photoluminescence decay for the SL cell samples No.788 and No.900 at the wavelength corresponding to the GaAs bandgap (top) and the SL bandgap (bottom).


Fig.7 EQE spectrum of the SL cells and the reference GaAs bulk cell.



Fig.8 I-V characteristics of the SL cells and the reference GaAs bulk cell under AM 1.5 G illumination.

Table.1 Partial pressure condition of TBA and TBP for SL growth.

Sample No

Partial pressure for TBA

(×10-2 mbar)

Partial pressure for TBP

(×10-2 mbar)

V/III ratio













Table.2 Thickness and composition of well and barrier estimated from XRD result.



Sample No






















Table.3 PV performance of the different phosphorus content of SL cells and the reference cell.

Sample No








Reference GaAs





#788 SL





#794 SL





#900 SL






Legal notice
  • Legal notice:

Presentation: Oral at 17th International Conference on Crystal Growth and Epitaxy - ICCGE-17, Topical Session 4, by Kentaroh Watanabe
See On-line Journal of 17th International Conference on Crystal Growth and Epitaxy - ICCGE-17

Submitted: 2013-04-09 10:37
Revised:   2013-07-08 15:06