High internal and external quantum efficiency InGaN/GaN solar cells

APPLIED PHYSICS LETTERS 98, 021102 共2011兲

High internal and external quantum efficiency InGaN/GaN solar cells Elison Matioli,1,a兲 Carl Neufeld,2 Michael Iza,1 Samantha C. Cruz,1 Ali A. Al-Heji,1 Xu Chen,2 Robert M. Farrell,2 Stacia Keller,2 Steven DenBaars,1 Umesh Mishra,2 Shuji Nakamura,1 James Speck,1 and Claude Weisbuch1,3 1

Department of Materials, University of California, Santa Barbara, California 93106-5050, USA Department of Electrical and Computer Engineering, University of California, Santa Barbara, California 93106, USA 3 Laboratoire de Physique de la Matière Condensée, CNRS, Ecole Polytechnique, 91128 Palaiseau, France 2

共Received 13 November 2010; accepted 22 December 2010; published online 10 January 2011兲 High internal and external quantum efficiency GaN/InGaN solar cells are demonstrated. The internal quantum efficiency was assessed through the combination of absorption and external quantum efficiency measurements. The measured internal quantum efficiency, as high as 97%, revealed an efficient conversion of absorbed photons into electrons and holes and an efficient transport of these carriers outside the device. Improved light incoupling into the solar cells was achieved by texturing the surface. A peak external quantum efficiency of 72%, a fill factor of 79%, a short-circuit current density of 1.06 mA/ cm2, and an open circuit voltage of 1.89 V were achieved under 1 sun air-mass 1.5 global spectrum illumination conditions. © 2011 American Institute of Physics. 关doi:10.1063/1.3540501兴 III-nitrides have been proven as an excellent material system for optoelectronic devices. While high efficiency light-emitting diodes and lasers1 have been demonstrated, many of these material system properties are also of interest for photovoltaics.2 InxGa1−xN alloys have a direct band gap that covers nearly the entire solar spectrum 共0.7–3.4 eV兲. Their high absorption coefficient 共⬃105 cm−1兲 共Refs. 3–5兲 yields strong light absorption for relatively thin InGaN layers. While high quantum efficiency III-nitride solar cells with band gaps larger than 2.4 eV have been demonstrated,6,7 a broader spectral response could be realized by applying them as one component in multijunction solar cells to convert the high energy part of the solar spectrum. Alternatively, this material system could be envisaged for a pure nitridebased solar cell. Several challenges need to be overcome in the IIInitrides to take full advantage of its potential. The lattice mismatch between InN and GaN and low temperature growth of InGaN induce impurity incorporation, point, extended, and morphological defects 共e.g., v-defects8兲, generating nonradiative recombination centers 共NRCs兲.9 These NRCs reduce carrier lifetimes and reduce the solar cell shortcircuit current.10 An efficient conversion of lower energy photons requires high In content InGaN, which limits the maximum thickness of this layer, reducing its light absorption. Moreover, there are significant polarization-related charges at the InGaN/GaN interfaces which result in large electric fields in the InGaN layer.11 For Ga-polar p-GaN/iInGaN/n-GaN structures, the polarization-related electric field is in opposite sense to the depletion field of the diode,12,13 and forward biasing the device further increases the electric field in the InGaN layer. In this letter we report on the demonstration of high internal quantum efficiency 共IQE兲 GaN/InGaN solar cells a兲

Electronic mail: [email protected].

0003-6951/2011/98共2兲/021102/3/$30.00

and an improved light incoupling by textured surface, yielding high external quantum efficiency 共EQE兲. A combination of light absorption and EQE measurements allowed the independent assessment of the light absorbed in the InGaN and of the solar cell IQE. The high measured IQE reveals an efficient conversion of absorbed photons into electrons and holes and an efficient transport of these carriers outside the device. The degrading effect of the polarization fields was avoided with a high doping of the n- and p-GaN, which helps to screen the polarization-related charges at the heterointerfaces. An improved light incoupling into the solar cells was achieved by rough surface, induced during the p-GaN growth, which reduced the reflection of the incident light at the device surface and increased the path length of the light inside the device active region. The devices reported in this work are GaN/InGaN p-i-n solar cells, grown by metal-organic chemical vapor deposition 共MOCVD兲 on 共0001兲 sapphire substrates. The active region was formed by a 60 nm thick InGaN grown at 880 ° C, with In content ⬃12%, below a 300 nm thick p-type doped GaN layer. The n- and p-type GaN doping was 关Si兴 ⬃ 6 ⫻ 1018 cm−3 and 关Mg兴 ⬃ 8 ⫻ 1019 cm−3. The schematic of the device is shown in Fig. 1共a兲. Two solar cell structures with different surface roughness, induced during growth by the p-GaN growth temperature, are presented. The p-GaN in the smoother sample was grown at 955 ° C 共rms roughness of ⬃7 nm兲 and at 880 ° C in the rougher sample 共rms roughness of ⬃41 nm兲. The 12.5⫻ 12.5 ␮m2 atomic force microscope 共AFM兲 scans of both device surfaces are shown in Figs. 1共b兲 and 1共c兲. A 30/300 nm thick Pd/Au grid was deposited on top of the devices with a spacing of 100 ␮m between the centers of adjacent grid lines. The IQE of the solar cells was assessed from the ratio between the EQE and the light absorption, which were determined by independent measurements. The EQE was measured from the processed solar cells, under 1 sun air-mass 1.5 global spectrum 共AM1.5G兲 illumination, using an Oriel solar simulator with a

98, 021102-1

© 2011 American Institute of Physics

021102-2

Matioli et al.

FIG. 1. 共Color online兲 共a兲 Schematic of the InGaN/GaN solar cell structure. 12.5⫻ 12.5 ␮m2 AFM scans of the surface of the 共b兲 smoother sample with rms roughness of ⬃7 nm and 共c兲 rough sample with rms roughness of ⬃41 nm.

Xe lamp. An Oriel 260 monochromator was used to assess the spectral dependence of the EQE. The light absorption was measured from the combination of the transmission and reflection measurements, performed in the unprocessed wafer, using a Shimadzu UV3600 UV-VIS-NIR spectrophotometer. This tool uses a white light source and a monochromator for the spectral measurements. The transmitted and reflected beams from the sample are collected in an integrating sphere and measured by a photomultiplier detector. The use of integrating sphere is of fundamental importance to thoroughly measure diffuse transmission T共␭兲 and reflection R共␭兲—this technique is especially useful for samples with rough or textured surfaces. The absorption was then determined as A共␭兲 = 1 − T共␭兲 − R共␭兲. The measurement of T共␭兲 and R共␭兲 was performed over a surface of 1 cm diameter of the sample; therefore, A共␭兲 corresponds to the average light absorption over this surface. For a fair determination of the IQE, the EQE was measured on all devices covering this same surface where the light absorption was measured. A total of 12 devices, with 0.5 ⫻ 0.5, 1 ⫻ 1, and 2 ⫻ 2 mm2 sizes, were measured and used to determine the average EQE and standard deviation. Figures 2共a兲 and 2共b兲 show the absorption and average EQE for

Appl. Phys. Lett. 98, 021102 共2011兲

both smoother and rough samples, as well as the maximum and minimum EQE measured curves. The comparison between the absorption and EQE measurements yields useful insights about the photoconversion efficiency in the InGaN layer. In the GaN spectral absorbing region 共␭ ⬍ 365 nm兲, the generated carriers from the strong light absorption do not contribute to current generation 共EQE⬃ 0兲; instead, they mostly recombine due to the short carrier diffusion in the neutral p-GaN region. This comparison is more interesting in the InGaN spectral region 共370 ⬍ ␭ ⬍ 410 nm兲, where the spectral behavior of the light absorption follows closely the EQE curve, both in shape and magnitude. This means that nearly all absorbed photons in this layer were converted into electrons and holes, and these charges were efficiently separated and transported out of the device. The region above 410 nm corresponds to the absorption tail of InGaN and to the absorption by the sapphire substrate, which is negligible. The nonzero absorption observed in this region in the smoother sample 关Fig. 2共a兲兴 is an artifact of the measurement. It corresponds to light propagating laterally in this nonabsorbing medium that was converted from the vertically incoming light by scattering at the surface roughness. The AFM image of the smoother sample surface shown in Fig. 1共b兲 reveals a rms roughness of ⬃7 nm. While this surface roughness creates laterally propagating light, it is not rough enough to scatter this portion of the light back within the integrating sphere aperture; therefore, it is not collected by the integrating sphere, appearing as absorbed light. This measurement artifact was verified by changing the size of the integrating sphere aperture, which consequently changes the amount of collected scattered light. This caused a change in the absorption curve only in the nonabsorbing region 共␭ ⬎ 410 nm兲, remaining unchanged in the InGaN and GaN spectral regions. The IQE for both the smooth and rough surface solar cells was determined from the ratio between the EQE and the light absorption curves 共Fig. 3兲. The IQE is higher than 90% in both samples for ␭ from 380 to 410 nm. The smoother sample had a slightly higher IQE, reaching values up to 97%, while the rough sample had an IQE up to 93%. The reason for the higher IQE observed in the smoother solar cell is not fully understood and is attributed to run to run variations in the MOCVD growth. The high measured IQE revealed an efficient electron-hole generation and carrier transport out of the active region.

FIG. 2. 共Color online兲 Measured absorption 共solid兲, average 共solid with error bars兲, maximum and minimum 共dashed curves兲 EQE curves for 共a兲 smoother and 共b兲 rough samples. The average EQE and its standard deviation 共error bars兲 were determined from measurements of a total of 12 devices located in the same area of the sample where the absorption was measured.

021102-3

Appl. Phys. Lett. 98, 021102 共2011兲

Matioli et al.

TABLE I. Solar cell performance metrics.

FIG. 3. 共Color online兲 IQE for the smoother 共dashed兲 and rough 共solid兲 surface solar cells, determined from the ratio between the EQE and the light absorption curves.

Although these solar cells present high IQEs, notably the smoother sample, the absorption measurement 关Fig. 2共a兲兴 shows that at most only ⬃60% 共at 380 nm兲 of the incoming light is being absorbed by the InGaN in this sample. Therefore, although the conversion of photons into electron-holes is very efficient, the light absorption needs to be improved. One initial attempt to improve the light absorption was to increase the surface roughness of the solar cells. This reduces the reflection of the incoming light in the top surface and increases the optical path length of the light inside the solar cell due to a change in the incidence angle.14 Figure 2共b兲 shows the enhanced light absorption in the InGaN, which increased to ⬃80% 共at 380 nm兲 due to the rougher surface. The peak EQE 共average EQE curve兲 increased from 56% to 72%, at 380 nm, even though the IQE at this same wavelength was slightly reduced from 94% to 90%. Figure 4 shows the current density versus voltage for 1 ⫻ 1 mm2 devices, with rough and smooth surfaces, under illumination conditions of 1 sun-AM1.5G. The illumination intensity was calibrated at 1 sun by matching the measured short-circuit 0 , calculated from the excurrent to the theoretical value Jsc 0 perimental EQE共␭兲 and theoretical sun spectrum s共␭兲 as Jsc = q兰EQE共␭兲s共␭兲d␭. The solar cell with rough surface presented a short-circuit current density of 1.06 mA/ cm2, which was 27% higher than the smoother sample 共0.83 mA/ cm2兲. The open circuit voltages were 1.89 and

current density (mA/cm2)

1.2 1 0.8 0.6 0.4 0.2 0

0

rough smooth 0.5

1

1.5

2

voltage (V) FIG. 4. 共Color online兲 Current density vs voltage for 1 ⫻ 1 mm2 solar cells, with rough 共solid兲 and smooth 共dashed兲 surfaces, under 1 sun-AM1.5G conditions.

Device

Voc 共V兲

Jsc 共mA/ cm2兲

FF 共%兲

Pmax 共mW/ cm2兲

EQEpeak 共%兲

IQEpeak 共%兲

Smooth Rough

1.83 1.89

0.83 1.06

76.6 78.6

1.16 1.57

56 72

97 93

1.83 V for the rough and smoother samples, respectively. These devices presented high fill factors 共FF兲, 78.6% 共rough兲 and 76.6% 共smoother兲, revealing excellent solar cell performances. The better light coupling of the solar cell with rough surface to the device active region resulted in an enhancement of 35% in maximum output power. Table I summarizes the performance metrics for both devices. To conclude, the IQE of InGaN/GaN solar cells was assessed through the combination of absorption and EQE measurements. This revealed an efficient conversion of absorbed photons into electrons and holes and an efficient transport of these carriers outside the device, yielding IQE as high as 97%. An enhanced light incoupling into the InGaN was achieved by a rough surface which slightly reduced the solar cell IQE but enhanced its light absorption by 33%. However, there is still at least 20% of the incoming light that is either reflected off the solar cell surface or directly transmitted through the device. An enhanced light coupling scheme is required to improve the InGaN/GaN solar cell performances, which can be achieved by an improved surface texture, by antireflecting and high reflecting coatings in the top and bottom surfaces, respectively, or by coupling the incoming light to guided light using photonic crystals, which can be then fully absorbed by the active region. This study is based upon work partially supported as part of the “Center for Energy Efficient Materials” at the University of California, Santa Barbara 共UCSB兲, an Energy Frontier Research Center funded by the U.S. Department of Energy 共DOE兲, Office of Science, Office of Basic Energy Sciences under Award No. DE-SC0001009, by the DOE under Project No. DE-FC26-06NT42857, and by the Solid State Lighting and Energy Center 共SSLEC兲 at the UCSB. S. Nakamura and G. Fasol, The Blue Laser Diode 共Springer, Berlin, 1997兲. O. Jani, I. Ferguson, C. Honsberg, and S. Kurtz, Appl. Phys. Lett. 91, 132117 共2007兲. 3 J. Muth, J. Lee, I. Shmagin, R. Kolbas, Jr., H. Casey, B. Keller, U. Mishra, and S. DenBaars., Appl. Phys. Lett. 71, 2572 共1997兲. 4 R. Singh, D. Doppalapudi, T. Moustakas, and L. Romano, Appl. Phys. Lett. 70, 1089 共1997兲. 5 A. David and M. Grundmann, Appl. Phys. Lett. 97, 033501 共2010兲. 6 C. Neufeld, N. Toledo, S. Cruz, M. Iza, S. DenBaars, and U. Mishra, Appl. Phys. Lett. 93, 143502 共2008兲. 7 X. Zheng, R. Horng, D. Wuu, M. Chu, W. Liao, M. Wu, R. Lin, and Y. Lu, Appl. Phys. Lett. 93, 261108 共2008兲. 8 X. Wu, C. Elsass, A. Abare, M. Mack, S. Keller, P. Petroff, S. DenBaars, J. Speck, and S. Rosner, Appl. Phys. Lett. 72, 692 共1998兲. 9 D. Cherns, S. Henley, and F. Ponce, Appl. Phys. Lett. 78, 2691 共2001兲. 10 J. Wierer, A. Fischer, and D. Koleske, Appl. Phys. Lett. 96, 051107 共2010兲. 11 F. Bernardini and V. Fiorentini, Phys. Rev. B 57, R9427 共1998兲. 12 S. Chichibu, A. Uedono, T. Onuma, B. Haskell, A. Chakraborty, T. Koyama, P. Fini, S. Keller, S. DenBaars, J. Speck, U. Mishra, S. Nakamura, S. Yamaguchi, S. Kamiyama, H. Amano, I. Akasaki, J. Han, and T. Sota, Nature Mater. 5, 810 共2006兲. 13 T. Takeuchi, S. Sota, M. Katsuragawa, M. Komori, H. Takeuchi, H. Amano, and I. Akasaki, Jpn. J. Appl. Phys., Part 2 36, L382 共1997兲. 14 E. Yablonovitch, J. Opt. Soc. Am. 72, 899 共1982兲. 1 2