1. Introduction
To achieve better cost-effectiveness, several approaches have been explored. The approaches are mainly divided into two categories: improving energy conversion efficiencies of solar cells and reducing manufacturing cost. The energy conversion efficiency can be accomplished by either improving performance characteristics of solar cells such as short circuit current, open circuit voltage, and shunt resistance or extending the absorption spectrum of solar cells. Various efforts have been made to improve the spectral response of photovoltaic cells. The highest current conversion efficiency has achieved in III-V compound semiconductor multi-junction (MJ) photovoltaic (PV) cells in which several pn junctions with different bandgap semiconductor materials are stacked to cover different spectrum[1,2]. As a different approach, quantum wells (QWs)[3] or quantum dots (QDs)[4] having different bandgaps were inserted into the absorption layers. Solar cell structures such as strain-balanced InGaAsP/GaInP QW structure[5,6], InGaAs/GaAsP QW structure employed in GaInP/GaAs tandem solar cell[7] and GaSb/InGaAs QD and QW hybrid structure[8] have been explored. The open circuit voltage (Voc) is determined by smaller energy bandgap (Eg) material when two different bandgap materials are used. The Voc approximately amounts to 0.6 of the smaller energy bandgap (Eg) material. However, if QW or QD materials are inserted in the absorption layer of the PV cell, it is possible to increase the absorption spectrum without significant decrease in Voc. The number of QW needs to be optimized because the Voc decreases as the QW layers increase[9]. In our previous report, we demonstrated N-AlGaInP/p-GaInP single heterojunction structure solar cells[10]. Since the bandgap energy of the AlGaInP is 2.2 eV (λAlGaInP = 560 nm) and the bandgap energy of the GaInP layer is 1.85 eV (λGaInP = 670 nm), no light absorption occurs above the wavelength 670 nm. To extend the solar absorption spectrum edge, a semiconductor with Eg less than 1.85 eV needs to be employed. In this report, we have proposed a simple GaInP/GaAs QW heterojunction structure inserted in N-AlGaInP/p-GaInP heterojunction structure because the QW layers are easily lattice matched to GaAs substrate. The characteristics of the QW structure solar cells were compared with single heterojunction structure solar cells. A significant enhancement in both short circuit current density (Jsc) and external quantum efficiency (EQE) were observed without absorption spectrum extension while the Voc was reduced slightly from 1.42 V to 1.40 V. There have been reports to suppression of recombination to increase Jsc[11-13] This result is different with previous works in QW structures report because the previous reports by other groups achieved the enhancement in Jsc with a slight increase in absorption spectrum due to insertion of QW layers. This result suggests that Jsc can be improved significantly without extending absorption spectrum. The suppression of recombination. If it is verified that the insertion of the QW structures suppresses the carrier recombination and thus results in better performance, it can be used in in optical or electrical devices such as photodetectors, high speed transistors, and solar cells.
2. Experiment
The N-AlGaInP/p-GaInP single heterojunction solar cell structures were grown on p-GaAs substrates by metalorganic vapor phase epitaxy (MOCVD) in the Korea Advanced Nano-fabrication Center. The p-GaAs substrates were cut 6° off the (100) plane toward the <111>A direction. The epitaxial layer structures consist of a 200 nm p-GaAs contact layer, a 50 nm P-AlInP back surface field layer, a 590 nm p-GaInP (Eg = 1.85 eV) and 14 periods of a 10 nm p-GaInP/5 nm p-GaAs for QW structure or 800 nm p-GaInP for single heterojunction (HJ) structure (control cell), and an 400 nm N-AlGaInP (Eg = 2.2 eV) emitter layer followed by a N-AlInP window layer (Eg = 2.36 eV), and n-GaAs contact layer on top. All epilayers were lattice-matched to GaAs. The epitaxial layer structures were shown in Figure 1 (a) for QW structure and Figure 1 (b) for single HJ structure.
For backside p-metalization, Ti/Pt/Au were deposited on the back side of a GaAs substrate after a front side protective photoresist (PR) was applied. The thicknesses of the p-metals are 20 nm/20 nm/460 nm, respectively. For the front side n-metallization, grid patterns were formed by standard photolithography. n-metal contacts of AuGe/Ni/Au (80 nm/80 nm/400 nm thickness) were deposited and lifted off. The n-GaAs contact layer was etched to avoid unintended light absorption at the top contact layer and the mesas for active solar cell areas were formed by etching down to the GaAs substrate layer. Finally, ZnS/MgF2 layers for anti-reflection coating were deposited.
3. Results and Discussion
3.1. Current density-voltage characteristics
The current density-voltage (J-V) characteristics of PV cells under a 1-sun illumination condition are shown in Figure 2. The measurement results were obtained from 1 mm × 1 mm area devices with 20 μm width and 174 μm spacing line grids. The short-circuit current densities (Jscs) were 9.61 mA/cm2 for QW structure and 7.06 mA/cm2 for single HJ control cell, respectively. The Jsc was increased 36% in the solar cell employing QW structure. The open-circuit voltages (Vocs) were 1.40 V for QW structure and 1.42 V for single HJ structure. The Voc was slightly decreased and the reduction was less than 2% in the structure employing a 210 nm thick GaInP/GaAs QW (14 periods). Considering the fact that the bandgap of GaAs is 1.42 eV and the Voc of GaAs solar cell has the value around 1.02 V[14], the introducing GaAs in QW indicates a relatively small decrease of Voc as anticipated. In addition to the increased Jsc, the shunt resistance of QW structure solar cell is higher than that of single HJ solar cell. The efficiency of QW structure is 10 % and the efficiency of single HJ control cell is 5.2%. By introducing a QW structure into AlGaInP/GaInP HJ structure solar cell, the overall efficiency was almost doubled.
3.2. External quantum efficiency
External quantum efficiencies (EQEs) were measured to obtain the absorption spectral response and the measured EQEs are shown in Figure 3. The results were obtained from 5 mm × 5 mm area cells without ARC. The maximum EQE was measured 36.3% at 510 nm for single HJ control cell and 46% at 420 nm for QW structure cell, respectively. The absorption edges of both solar cells were 580 nm which corresponds to bandgap energy of 2.16 eV. The QW structure showed higher EQE than the single HJ structure, but no absorption occurred in GaInP/GaAs QW layer, Preliminarily, we speculated that the Jsc improvement was attributed to the absorption spectrum extension by GaInP/GaAs QW layer. However, the measured EQE data indicated that there is no enhancement of the absorption edge due to the QW layer. Since the cause of Jsc enhancement in QW structure was different from our assumption, it is imperative to find out the explanation for the improvement of the Jsc.
3.3. Secondary ion mass spectrometry depth profile
From EQE data, it was observed that the solar spectrum above the wavelength longer than 580 nm were not absorbed at all. Therefore, we need to verify whether p-GaInP/p-GaAs QW layers were grown in epitaxial layers as we designed. A secondary ion mass spectrometry (SIMS) was used for depth profiling of both epilayers. Figure 4 illustrates the SIMS depth profiles of the QW structure and HJ control structure. Two structures are identical except the 210 nm thick 14 periods of p-GaInP/p-GaAs QW layer. The most distinct difference in compositional element between both epitaxial layers is Arsenic in QW structure as shown in Figure 4 (a), because QW structure contains GaAs whereas HJ control structure consists of GaInP only. The arsenic content was clearly observed only in QW structure. For Al, both structures showed high contents in N-AlGaInP layers and abruptly decreased in p-GaInP layer in a depth of 800 nm. As the bandgap energies of the absorption edges were measured 2.16 eV in both structures, which corresponds to the bandgap of N-AlGaInP, we suspected that the p-GaInP layers (Eg = 1.85 eV) were not grown appropriately. However, the Al contents indicated that p-GaInP layers were formed correctly as designed. According to the depth profiles shown in Figure 4 (c)~(e), Gallium (Ga), phosphorus (P), and indium (In) contents also indicated that GaInP/GaAs QW layers were formed in the region from 1,200 nm to 1,450 nm depth. Oxygen atoms were known to be a nonradiative recombination center. The O concentration is almost same in both epitaxial layer structures.
From the EQE, we can see that light was absorbed only in the AlGaInP layers from 250 nm to 750 nm for both solar cells. SIMS depth profile indicated that the epitaxial layers of both solar cells are different only in the QW layers, which means that the thicknesses of epilayers, the material composition, and oxygen concentration cannot be a factor to generate the significant difference in the performance of the solar cells. However, the solar cell with QW structure showed 36% enhancement in Jsc. These results suggested that recombination during carrier transportation was suppressed by QW structure and thus the conversion efficiency was improved significantly. Increased efficiency through the suppression of recombination has been reported in nanostructured Si solar cells[15]. Our result is worth noting because the improvement in Jsc can be achieved without an extension of the response spectrum and the QW structure insertion in epilayer[16]. This indicate that inserting QW structure can improve carrier transport and thus results in conversion efficiency even without the absorption spectrum enhancement. There have been several theoretical report regarding current transport in QW structures[17,18]. Further investigation is required to explain more specific in photocurrent collection process in the QW structure.
4. Conclusion
The characteristics of solar cells employing GaInP/GaAs QW structure in a single AlGaInP/InGaP HJ structure were compared with solar cells having single AlGaInP/InGaP HJ structure without QW layer. The short-circuit current density was increased by 36% in the solar cell with the QW structure. In addition to the enhancement of Jsc, the shunt resistance improvement resulted in a significant increase in the conversion efficiency from 5.2% to 10%. The EQE data show that the absorption by the light above a wavelength of 580 nm did not occur in both solar cell structures. SIMS depth profile data indicated that the thicknesses of the both grown epilayers were identical except the QW layer. Since the p-GaInP/QW layers (QW structure) or the p-GaInP layer (single HJ structure) worked only for carrier transportation, it can be concluded that photogenerated carriers are collected more efficiently in QW structure. The high collection efficiency of photogenerated carriers is attributed to the suppression of recombination due to the existence of QW layers.