EFFICIENCY IMPROVEMENTS IN GaAs-on-Si SOLAR CELLS

Transcription

1 EFFICIENCY IMPROVEMENTS IN GaAs-on-Si SOLAR CELLS S.M. Vernon, S.P. Tobin, V.E. Haven, C. Bajgar and T.M. Dixon Spire Corporation, Patriots Park, Bedford, MA M. M. Al-Jassim, R.K. Ahrenkiel and K.A. Emery, SERI, Golden, CO ABSTRACT GaAs grown directly on silicon substrates has the potential for reducing the cost and weight of single-junction GaAs cells, as well as enabling a III-V/Si monolithic tandem technology. This paper reports the achievement of a GaAs-on-Si solar cell having an efficiency of 17.6% (one-sun, AM C, SERI measurement), which is the highest reported value far a cell of this structure. A GaAS-on-GAS cell recently fabricated has been measured at SERl to be 24.3% efficient (AM1.5), which is the highest one-sun value ever reported for any cell. in detail elsewhere [31 but basicaily consist of a three-step process: (I) a hydrogen bakeout at approximately IOOO C, (2) a low-temperature nucleation step at approximately YOO C, and (3) film growth at standard &As growth conditions including a temperature of approximately 700-C, a V-III ratio of 15, and a growth rate of 4 microns/hi-. The reactor pressure is maintained at 760 ton. h p METALLt*ATIOR CT/L \ /7)(1 1NTROD CTlON Silicon wafers have some definite advantages as substrates for the growth of thin-film GaAs solar cells. The properties of Si which make it an attractive choice for this application include its abundance, mechanical strength, large-diameter availability, low cost, proper bandgap for the bottom cell in a tandem, and its crystal structwe which permits the growth of single-crystal GaAs films. GaAs-on& growth technology has advanced rapidly in the past few years Ill, and very good majority-carrier devices (FETs) with performance equal to GaAs grown on GaAs substrates have been reported. Minority-carrier devices (solar cells, lasers, bipolar transistors) have aiso been reported, but usually with performance degraded relative to GaAs homoepitaxy. METALL,Z&TIOW FIGURE 1. GaAs-on-Si Cell Structure. 1 This paper reports the achievement of a GaAs-an-Si solar cell having an efficiency of 17.6% (one-sun, AMI.5, 25 C, measured a* SERI) which is the highest C&As-on-Si cell efficiency measured at, or verified by, a U.S. national laboratory. This represents a significant increase over our previously reported best efficiency value of 11.2% 121 The improvement is a direct result of the higher material quality achieved in these GaAs-on-Si films as determined by minority-carrier lifetime and defect-density measurements. EXPERIMENTAL The solar cell structure used in this work is shown in Figure 1. The epitaxial layers are grown by metalorganical chemical vapor deposition (MOCVO) in a commercially available system, the WI-MO cvd~m 450, which has a batch size of five two-inch wafers per run. The GaAs-on-Si growth conditions have been described In some of the growth runs, various additional high-temperature annealing steps were used. The effects of these growth parameters on cell performance and material quality are discussed below. The thermal annealing method used in our work is referred to as thermal cycle growth (TCG) and is based on the previous work of researchers at NTT in Japan [41. Figure 2 shows our typical heat schedule used to deposit the GaAs/Si cell structure studied. After depositing an initial I pm of GaAs by the three-step method described above, several more microns of n+ GaAs buffer are deposited by a repetition of high-temperature annealing, standard-conditions growth, and coaldown steps. The actual solar cell Layers are then deposited using our ~onvenfional recipe of growing GaAs at 700DC and GaAlAs window layers at SOOOC. The cell deviceprocessing steps are listed in Table 1. 0X0-8371/88/ $1.00 e 1988 IEEE

2 The double heterostructure (OH) show in Figure 3 ms also deposited onto TCG GaAsiSi buffer layers for the lifetime and dislocation density measurements. The minority-carrier lifetime was measured by the photoluminescence (PL) decay technique ISI and defect ifruct~ce was determined by cross-sectional and planview transmission electron microscopy (TEM) analysis. Oe,,*a,l&s Galls oa,pja* GaAr si FIGURE 2. Typical Thermal Cycle Growth Schedule. OilAS TABLE 1. Cell Process. I. ~vaoorate Back Contact (AtiCe) 2. All& Contact 3. Photolithography for Front Grid 4. Evaporate Front Contact (CrAu) 5. Evaporate Back Metal (Au) 6. Liftoff 7. Sinter 8. Photolithography for Mesa 9. Mesa Etch 10. Etch Cap Layer Ii. Evaporate AR Coating (Z S/MgF2) FIGURE 3. Structure for PL Lifetime and TEM Measurements. RESULTS Table 2 shows the results for unannealed GaAslSi and GaAs/GaAs structures. The improvement of GaAsiSi material quality with buffer layer thickness is clearly seen here. TABLE 2. Cell Performance and Crystal Quality vs. Growth Parameters Buffer Dislocation Open- Cdl Thickness PL Lifetime Density Circuit Efficiency Substrate A & (wd (nsec) (cme2) Voltage 09 (Xl GaAs NO 1 20 I:0 x si NO I x Si NO x Si NO x Solar Cell Test Conditions are 100 mw/cm2, AMI. global, total area, 25 C. All solar cells are 0.5 x 0.5 cm in area. 482

3 The effects of the TCG method were examined by first growing a GaAsiSi buffer (8 pm thick) which included one anneal cycle having a maximum temperature of 900X for 2 minutes. This Led to the fabrication of a cell with 15.1% efficiency and a PL lifetime of 1.75 nsec and a dislocation density of 4 x IO7 cm-2. Further improvements in the TCG method were studied by making buffer-layer runs of GaAs-on-Si using eight different TCG conditions. The parameters explored included maximum temperature values of 900 and 95O C and anneal times of two and twenty minutes. Then, solar cells and DH structures were grown, fabricated and tested. Figure 4 shows the typical surface morphologies of the three types of material. Figure 4c shows some crosshatch features which are indicative of the high structual quality of these lattice-mismatched layers. - GOAE coniroi 24.3% -- G&,Si 17.6% Wavelength (nm) FIGURE 5. Quantum Efficiencies of GaAslSi and GaAs Control Cells. GaAs = 6jm THICK FIGURE 4. GaAs-on-Si Surface Morphology vs. Growth Method. The eight different TCG conditions all yielded similar solar cell results. The best cell from each of the 8 wafers has an efficiency of between 17.5 and 18.1% AM1.5 measured at Spire. The SER! measurement of several of the very best cells shows a slightly lower efficiency of 17.6%. A GaAs-on-GaAs control cell in this batch has,been measured at SERI to be 24.3% efficiency. Figure 5 shows the spectral response of these two cells and Table 3 lists their performance values. The illuminated I-V curves from SERI for the best GaAslSi and GaAslGaAs cells are shown in Figures 6 and 7 respectively. Concentrator cells have also been fabricated and tested. For CaAs homoepitaxy, a maximum value of 25.4% at 200 suns has been achieved. For GaAs-on-Si, using material from an early, non-optimized TCG run, a cell efficiency of 18.5% at 370 suns has been measured by Sandia. The efficiency versus concentration for this cell is shown in Figure 8. The cell results discussed above are summarized in Table 4. TABLE 3. Recent GaAs Cell Results Substrate Comparison. On Si On GaAs Efficiency (%I (V) :,Fc~(c:%mplcm2) v,, SERI Data for best cell of &As-on-Si and GaAs-on-GaAs, grown and processed side by side. Conditions: AMI.5, 100 mwicm2, 25X, area = 0.25 cm2. 483

4 YOLTs3E wq OLT.mE ( w vqc = 0.8PN Vdr) ISi = A Jsc = wrd Fl,, *aim, = % I,= = 5.91 LA Emchwy = li.6 % v,, ; FIGURE 6. Efficirncy, Measurement of Best GaAs-on-Si Cell. Ym = dir ~& = 6,812 ma lsc I 4,c + Fll, ktor _ 86.46!z Ima = rd firdc cf = u.3 % - = Y FIGURE 7. Efficiency Measurement of Best CaAs-on-&As Cell. TABLE 0. Summary of Cell Efficiencies. &As-on&, one-sun lm,xoqed TCG: 17.6% Early TCC: 15.1% No TCG: 11.2%. GaAs-on&, at concentration ,000 Concentration (Suns) Improved TCG: >20% (predicted) Early TCG: 18.5% (370X) No TCG: 13.9% (500X). GaAs-on-GaAs controls One-s : X3% 200 suns: 25.9% FIGURE 8. GaAs/Si Concentrator Cells. All one-sun data measured at SERI; concentrator measurements from Sandia. For the matrix of 8 different TCG buffers, the DH growth runs had an unexplained problem which makes valid comparisons with ali the data impossible. This is evidence by the fact that the GaAsiGaAs DH structures lhad lifetimes as iow as 2.3 nsec and many defects were seen to be generated at the lower AI&As layer in the DH structure when viewed by cross-sectional TEM. Thus in these latest DH samples, improved GaAsiSi material quality is not directly observable even though significantly higher photovoltaic performance has been proven. A cross-sectional TEM view (see Figure 9) of the GaAs/Si heterointerface region does however indicate the improved crystal quality obtained by the TCG technique. Also, preliminary EBIC data on the best GaAs/Si cell wafer indicates a probable dislocation density of 8 x 106 cm2.

5 FIGURE 9. Cross-Section TEM s of GaAs-on&i. CONCLUSIONS In summary, the thermal cycle growth method has been shown to be effective in improving GaAs/Si photovoltaic performance. The TEM studies show that dislocation densities have been reduced by approximately an order of magnitude and minority-carrier lifetimes increased by more than a factor of two. The efficiency of &As-on-Si cells has been increased from 11.2% to 17.6% (one-sun) and from 13.9% to 18.5% (concentrated light) by use of the TCG~ technique. Improvements in bask GaAs cell growth and processing technology are also responsible for a portion of these increases, as GaAs/GaAs control cell efficiencies have climbed from 21.3 to 24.3% over the span of these experiments. The GaAs cell having an efficiency of 24.3% represents the highest one-sun efficiency ever reported. ACKNOWLEDGEMENTS This work was funded by the Solar Energy Research Institute whose support is gratefully acknowledged. The authors also express their appreciation to Mssrs. L. Geoffroy, and M. Sanfacon of Spire, K. Jones and D. Dunlavy of SERI, D. Ruby of Sand& and M. Lundstrom and students of Purdue for all their important technical contributions. REFERENCES D.W. Shaw, Epitaxial GaAs on Si: Progress and Potential Applications, Mat. Res. Sot. Symp. Proc., yl, 15 (1987). S.P. Tobin, SM. Vernon, V.E. Haven, C. Bajgar, M.M. Sanfacon, and S.J. Pearton, Factors Controlling the Efficiency of GaAs-on-Si Solar Cells, Proc. of the 19th IEEE PVSC, p. 113 (1987). S.M. Vernon, Y.E. Haven, S.P. Tobin, and R.G. Wolfson, Metalorganic Chemical Vapor Deposition of GaAs on Si for Solar Cell Applications, J. Crystal Growth, 77, 530 (1986). Y. Itoh, 1. Nishioka, A. Yamamoto, and M. Yamaeuchi. Heteroeoitauial Growth of GaAs on Si for S&r dell, Tech;. Dig. of the Intern. PVSEC-3, Tokyo, paper C-IV&, (1987). 5. R.K. Ahrenkiel, M.M. Al-Jassim, D.J. Dunlavy, K.M. Jones, S.M. Vernon, S.P. Tobin, and V.E. Haven, Minority-Carrier Properties of GaAs on Silicon, Appl. Phys. Let*. 53, 222 (1988). 465