Solar cell technology is changing. New efficiency records are being set. Alta Devices has reached 28.8% efficiency in a thin film single-junction cell at 1-sun, and 30.8% efficiency in a thin-film dual junction cell at 1-sun.
Counter-intuitively, efficient external fluorescence is a necessity for approaching the ultimate limits. A great Solar Cell also needs to be a great Light Emitting Diode. Why would a solar cell, intended to absorb light, benefit from emitting light? Although it is tempting to equate light emission with loss, paradoxically, light emission actually improves the open-circuit voltage, and the efficiency.
The single-crystal thin film technology that achieved these high efficiencies, is created by epitaxial liftoff, and can be produced at cost well below the other less efficient thin film solar technologies. The path is now open to a 30% efficient photovoltaic technology, that can be produced at low cost.

Epitaxial lift off (ELO) of III-V semiconductor materials has often been used to make lightweight, ultrahigh efficiency solar cells. Unfortunately, a major promise - the ability to reuse the expensive substrate material following the removal of the epitaxial active region of the solar cell - has been elusive since the process results in irreversible damage to the parent wafer. Recently, we reported a method to avoid wafer damage caused by ELO using a combination of epitaxial protection layers and non-destructive cleaning procedures of both InP and GaAs substrates [1, 2]. This has resulted, for the first time, in the demonstration of multiple reuse of the parent wafer without incurring changes in either the chemical or morphological status of the original material. This suggests that the wafer itself may be used an unlimited number of times, employing a sequence of epitaxial growth, epitaxial removal using adhesive-free cold weld bonding to a metalized plastic foil, non-destructively cleaning the wafer surface, and then repeat. Indeed, multiple cycles of ELO followed my wafer reuse can be an effective means for achieving very low cost, flexible and extremely lightweight solar cells. In this work, we discuss the progress made in ELO growth of high efficiency p-n junction GaAs solar cells followed by adhesive-free cold-weld bonding to flexible substrates, and then once more reusing the parent wafer. We also discuss other high performance devices (e.g. LEDs, FETs, etc.) that have been grown on original and reused wafers. We will discuss the cost and complexity tradeoffs in this process multiple growths, ELO, and reuse of from a single parent wafer in achieving low cost, high power conversion efficiency III-V solar cells.
[1] K. Lee, K. T. Shiu, J. Zimmerman, and S. R. Forrest, "Multiple growths of epitaxial lift-off solar cells from a single InP substrate," Appl. Phys. Lett., vol. 97, p. 101107, 2010.
[2] K. Lee, J. D. Zimmerman, X. Xiao, K. Sun, and S. R. Forrest, "Reuse of GaAs substrates for epitaxial lift-off by employing protection layers," J. Appl. Phys., vol. 111, p. 033527, 2012.

Monolithically integrated III-V compound semiconductor layers on silicon substrates will enable semiconductor industry to fabricate new interesting device concepts in the near future. The device concepts that have received the most attention include CMOS compatible lasers that could be used in on-chip and chip-to-chip communication, n-channel high mobility transistors and silicon based high-efficiency solar cells. In order to realize these devices, many challenges existing in the polar-on-nonpolar epitaxy (e.g., anti-phase domains, stacking faults, lattice mismatch) needs to be overtaken. Despite the challenges however, it has recently been shown that an atomically smooth low defect density GaP buffer layer can be fabricated on silicon [1]. Therefore, an increasing amount of attention has been addressed on the properties of gallium phosphide based dilute nitride materials (E.g., GaPN, GaAsPN). The benefit of these materials is that the nitrogen incorporation enables strain compensation and the energy band structure engineering.
We have studied the growth of GaP and Ga(As)PN layers on silicon and GaP substrates by metalorganic vapor phase epitaxy. Properties of the grown layers have been examined by Raman scattering, photoreflectance, photoluminescence, IV, atomic force microscopy, Rutherford backscattering and XRD studies[2-5]. We discuss the properties of the fabricated Ga(As)PN layers and focus on the characteristics of the layers that could be utilized to realize a high-efficiency solar cells on silicon substrates. The solar cell device concepts that have drawn our attention include n-GaP heterojunction emitters, GaAsP tandem cells and GaAsPN based intermediate band solar cells (IBSC).
In this work, we present the main experimental results of the fabricated layers. The results show that defects in Ga(As)PN layers on silicon most likely arising from the III-V/Si interface degrade the quality of the layers and that high-quality layers are realized on GaP substrates. Anti-phase domain and stacking fault/threading dislocation type defects have been observed by transverse scan analysis performed by high-resolution XRD setup. Photoreflectance results of Ga(As)PN layers show the conduction band splitting which can then possibly in the future be used to realize IBSC device on silicon substrate. At the moment, we are fabricating our first IBSC devices. In addition, IV measurements performed under 0.54 SUN show that semiconductor devices such as n-GaP/p-Si heterojunction solar cell are realized.
[1]: K. Volz, et al., Journal of Crystal Growth, 315, 37, (2011).
[2]: H. Jussila, et al., J. Appl. Phys. 111, 043518 (2012).
[3]: H. Jussila, et al., Phys. Status Solidi C, 9, 1607 (2012).
[4]: S. Nagarajan, et al., J. Phys. D: Appl. Phys. 46, 165103 (2013).
[5]: H. Jussila, et al., Thin Solid Films, 534, 680-684, (2013).

Multiple batches of thin-film p-n junction GaAs thin-film photovoltaic cells were grown sequentially on a single parent wafer without loss in performance from growth to growth. Through the combination of cold-welding and a modified epitaxial lift-off (ELO) process that eliminates parent wafer damage, the device active region is transferred onto flexible plastic substrates [1]. A combination of alternating arsenide-based and phosphide-based lattice matched epitaxial layers with high etch selectivity grown on both sides of the AlAs sacrificial layer, along with thermal and plasma surface cleaning fully eliminates surface contaminants and damage, creating a pristine surface suitable for subsequent epitaxial layer growth without the use of costly and damaging wafer repolishing methods[2]. Using these methods, three iterations of lightweight and flexible GaAs thin film photovoltaic cells bonded to flexible plastic substrates were fabricated from a single GaAs substrate without any degradation in device performance after regrowth with power conversion efficiencies of ~18%. Wafer recycling without repolishing allows for an indefinite number of growth iterations on a single substrate without damage or loss in the original wafer thickness, providing a path towards cost-effective, high performance, flexible and lightweight thin-film compound semiconductor devices.
[1] K. Lee, K. T. Shiu, J. Zimmerman, and S. R. Forrest, "Multiple growths of epitaxial lift-off solar cells from a single InP substrate," Appl. Phys. Lett., vol. 97, p. 101107, 2010.
[2] K. Lee, J. D. Zimmerman, X. Xiao, K. Sun, and S. R. Forrest, "Reuse of GaAs substrates for epitaxial lift-off by employing protection layers," J. Appl. Phys., vol. 111, p. 033527, 2012

Mono-crystalline germanium (Ge) is the most widely used substrate in concentrated photovoltaics (CPV) high efficiency multijunction solar cells (MJSC). Increasing demand on MJSC would undoubtedly lead to a lack of this rare material, resulting in dramatic increase in the Ge wafer price, not to mention that today, the cost of the Ge substrate already represents a substantial share of the total cell cost. A typical MJSC process uses > 140 µm thick Ge wafers as substrates, whereas a few microns would be sufficient for the bottom cell to match the photogenerated currents at the top and middle subcells. Moreover, using a thin Ge film rather than a thick substrate would reduce electrical and thermal resistances, weight of the cell and increase efficiency of the Ge subcell. In order to separate the active region of the MJSC from its original substrate, a layer transfer process based on simple electrochemical porosification of Ge could be used. Similar processes have already been successfully applied to crystalline silicon thin film solar cells, yielding performances very similar to crystalline cells that are manufactured on standard thick Si wafer.
We have developed a process, using porous germanium, to separate a semiconductor structure from the Ge substrate (APL 102 (1), 011915 (2013)). First, a double porosity layer is formed on top of a p-type Ge substrate in HF based electrolyte; the topmost layer has a low porosity whereas the buried layer is highly porous. After ultra high vaccum (UHV) annealing at high temperature, the porous top layer transforms into quasi-monocristalline germanium (QMG) film and will serve later as seed for epitaxial deposition of III-V top and middle subcells. On the other hand, the buried layer forms a film with large lateral voids that weaken the interface between the QMG film and the substrate, creating a so called “separation layer”. After epitaxial deposition of the III-V MJSC device is done, the cell is bonded to a low cost host substrate, and separated from the original Ge substrate, which is then ready to be re-used in another manufacturing cycle and could potentially yield a large number of MJSC solar cells, as the active region of the Ge subcell is only a few microns thick.
The synthesis and transformation during annealing of mesoporous germanium double layer are key steps in this process. In this presentation, we show the results of electrochemical etching of Ge in HF electrolyte to form the double porosity layers, and its morphology transformation during UHV annealing at high temperature which is due to surface diffusion at constant volume. The monocrystallinity of the layers and their suitability as seed layers for epitaxy of GaAs is confirmed by X-ray diffraction rocking curves.

Upconversion of sub-bandgap photons is a promising approach to exceed the Shockley-Queisser limit in solar technologies. Calculations have indicated that ideal upconverter-enhanced cell efficiencies can exceed 44% for non-concentrated sunlight, but such improvements have yet to be observed experimentally. In this presentation, we develop both theoretical and experimental methods to understand and improve solar upconversion, considering electronic, photonic, and thermodynamic design constraints. First, we develop a thermodynamic model of an upconverter-cell considering a highly realistic narrow-band, non-unity-quantum-yield upconverter. As expected, solar cell efficiencies increase with increasing upconverter bandwidth and quantum yield, with maximum efficiency enhancements found for near-infrared upconverter absorption bands. Our model indicates that existing bimolecular and lanthanide-based upconverters will not improve cell efficiencies more than 1%, consistent with recent experiments. However, our calculations show that these upconverters can significantly increase cell efficiencies from 28% to over 34% with improved quantum yield, despite their narrow bandwidths. Then, we develop the experimental techniques to enhance upconversion efficiencies, tailoring both the optical density of states via plasmonics and the electronic density of states via pressure. Our results highlight the interplay of absorption and quantum-yield in upconversion, and provide a platform for optimizing future solar upconverter designs.

Current photovoltaic technologies harvest only a fraction of incoming solar energy since they are unable to utilize photons with energies below the cell band gap. Placed behind a solar cell, the upconverter converts transmitted low-energy photons to photons with energies higher than the cell band gap. The higher energy photons are absorbed by the solar cell and contribute to the photocurrent. We develop optical models of several state-of-the-art commercial and research thin-film solar cells incorporating the upconversion layer. We present both analytical models based on published EQE data as well as detailed finite difference time domain (FDTD) models that incorporate absorption in all cell layers. We model the improvement in absorption and overall cell performance of amorphous Si, CIGS, GaAs, CdTe, and Cu2O cells with upconverting layers. We incorporate and discuss the effect of interface texture and different cell layers on the absorption of upconverted photons and make suggestions for improving the overall cell design to get the maximum benefit from upconversion. We estimate that the cell efficiency enhancement can range from 0.5% to up to 7% absolute depending on the cell type and upconversion efficiency. This work connects to the fundamental efficiency limit analysis of narrow-bandwidth solar upconversion by our collaborators [1], but presents concrete optical models of current solar cells and discusses the promise of upconversion for particular applications.
[1] J.A. Briggs, A.C. Atre, J.A.Dionne.Narrow-bandwidth solar upconversion: Case studies of existing systems and generalized fundamental limits. Journal of Applied Physics 113, 124509 (2013).

Technological implementation of a high-efficiency quantum dot intermediate-band solar cell (QD-IBSC) must be accompanied with a sufficient photocurrent generation via IB states. The demonstration of QD-IBSCs is presently undergoing two stages. The first is to develop epitaxial growth or lithography technology to fabricate high density QD arrays or superlattice with high quality heterointerface, and the second stage is to realize partially-filled or ideally half-filled IB states in order to maximize the photocurrent generation by two-step absorption of infrared photons in solar spectrum.
For the former requirement, we have developed a strain-compensation or strain-balanced technique in order to grow multi-stacks of self-organized InAs QDs in GaNAs matrix on GaAs substrate by molecular beam epitaxy (MBE). For the latter, either doping of QDs or concentrated illumination of sunlight at around 1000 suns should in principle result in a sufficient population of carriers in IB and hence production of an additive photocurrent in QD-IBSC.
We show that sunlight concentration increases the optical generation rate and hence sufficient carriers are populated in IB, which in turn increases photocurrent production from IB to conduction band. Our InAs/GaNAs QD solar cell shows a conversion efficiency of 20.3% at 100 suns increasing to 21.2% at 1000 suns. The QD-IBSCs which commonly suffer from low absorption by QDs are thus expected to improve fast and perform better under concentrated sunlight illumination.

The advantages of converting solar energy directly into clean electricity have important consequences for realizing a sustainable future. However, the conversion efficiency needs to be improved, and new solar cell concepts are believed to play an important role in order to achieve this target.
For solar cells, infrared absorption and ultra-violet relaxation are considered to be the main factors for conversion efficiency limitation [1]. To reduce these losses, several ideas have been proposed. The intermediate band (IB) solar cell reduces the infrared absorption losses [2]. This concept proposes that by upconversion, the energetic addition of two low energy quanta resulting in one high energy quanta, the infrared light can be efficiently used and the energy conversion efficiency limit can be improved significantly compared to single junction solar cells.
Quantum structures are attractive candidates for design of IB states, and a number of IB solar cells based on InAs/GaAs have been reported. Important InAs quantum structures are quantum dots (QDs) and disklike quantum structures (nanodisks) formed in the early stages of InAs deposition [3]. While the InAs quantum structures have been the focus of IB research for over one decade, the detailed mechanisms of upconversion are still unclear due to the complexity. Upconversion in InAs QDs is generally accepted to occur via the two-step two-photon-absorption mechanism. The operation principle has been verified, but the realized efficiencies are too small for practical applications [4].
In this work we focus on a novel IB solar cell concept using both QDs and nanodisks. The major upconversion mechanisms in the InAs quantum structures were determined by macroscopic measurements of upconverted photoluminescence (PL) and photocurrent (PC) [5,6]. The results showed that the nanodisks with height of a few monolayers and lateral extension up to hundreds of nm show exceptionally high upconversion efficiencies due to the activation of the upconversion via Auger process. Further exploration of the InAs quantum structures with spectroscopic imaging of PL and PC revealed that, depending on the distribution of nanodisks and QDs in different sites, the local PC generation efficiency can change drastically. We discuss the energy transfer between the quantum structures and how this relates to the upconversion efficiency. Based on our finding we propose a novel design of an efficiency enhanced IB solar cell using both QDs and nanodisks.
This work was supported by JST-CREST and by the Strategic Research Infrastructure Project, MEXT, Japan.
[1] W. Shockley and H. J. Queisser, J. Appl. Phys. 32, 510 (1961).
[2] A. Luque and A. Marti Phys. Rev. Lett. 78, 5014 (1997).
[3] R. Heitz et al., Phys. Rev. Lett. 78, 4071 (1997).
[4] A. Marti et al., Phys. Rev. Lett. 97, 247701 (2006).
[5] D. M. Tex and I. Kamiya, Phys. Rev. B 83, 081309(R) (2011).
[6] D. M. Tex, I. Kamiya, and Y. Kanemitsu, Phys. Rev. B 87, 245305 (2013).

Quantum structures are promising candidates for next-generation solar cell materials because of their unique optical properties. Among them, InAs quantum structures attract much attention because InAs/GaAs system forms intermediate bands, which absorb infrared light due to the small band gap energy, and upconversion processes are expected to improve the light-energy conversion efficiency. As previously reported, two-dimensional disklike InAs quantum structures with heights of two and three monolayers (quantum well island: QWI) display efficient upconverted photoluminescence (PL) and photocurrent (PC), which occur through a multiparticle Auger process [1]. This novel application of the QWI may be more suitable than the quantum dots for enhancing the solar cell conversion efficiency. For the further understanding of the upconversion process in QWIs, it is significant to clarify the photocarrier decay and upconversion dynamics. At low temperatures, the photocarrier dynamics can be evaluated by upconverted PL decay dynamics. However, the room-temperature photocarrier dynamics remain unclear because of the very weak PL intensity. Recently, we developed a novel measurement method for femtosecond time-resolved PC dynamics based on femtosecond excitation correlation (FEC) spectroscopy. In this study, we investigated the photocarrier dynamics of QWIs using PC-FEC measurement.
The sample structures were prepared by molecular beam epitaxy. A nominally undoped structure was grown on top of a semi-insulating GaAs (001) substrate and a GaAs buffer. A single InAs layer and a GaAs/AlGaAs quantum well were grown in an AlGaAs matrix. In PC-FEC measurement, two femtosecond laser pulses with a time delay are focused on the sample surface, as is similar to the FEC measurement of PL [2]. We measured the correlation signal of the PC induced by the two pulses as a function of delay time, which approximately corresponds to the photocarrier population in the QWIs. The light source was the optical parametric amplifier based on the KGW:Yb regenerative amplifier (repetition rate: 200 kHz).
We observed the strong FEC signal under excitation at the resonance energy of InAs QWIs with heights of two monolayers. The decay time is around 100 ps, which is weakly dependent on the excitation photon energy. We attribute the FEC decay time to the intrinsic photocarrier lifetime in the InAs QWIs. We will present the principle of the PC-FEC measurement and discuss the temperature dependence of the photocarrier intrinsic lifetime in the InAs QWIs.
Part of this work was supported by The Sumitomo Industries Group CSR foundation, JST-CREST, JSPS KAKENHI (No. 25234567), and the Strategic Research Infrastructure Project of MEXT.
References:
[1] D. Tex, I. Kamiya,and Y. Kanemitsu, Phys. Rev. B 87, 245305 (2013).
[2] D. von der Linde, J. Kuhl, and E. Rosengart, J. Lumin. 24/25, 675 (1981).

Thin-film structures are considered to be essential to lower the cost of solar cells. Thinning the active regions, however, often reduces the optical absorption efficiency. Therefore, the enhancement of optical absorption in thin-films is inevitable to maintain and improve solar cell efficiency. Among the variety of absorption mechanisms in semiconductors, excitonic absorption is one of the most promising candidates to enhance the optical absorption, because it can be added to the normal band-to-band absorption. The excitonic absorption at room temperature occurs more efficiently in the materials with higher exciton binding energies. Unfortunately, semiconductors sensitive to the main part of the solar spectrum such as Si and GaAs exhibit rather lower exciton binding energies than thermal energy at room temperature. However, the excitonic absorption can be considerably enhanced even in these materials by constructing superlattice (SL) and multiple quantum well structures. In this study, we show the absorption characteristics of the solar cells consisting of an AlGaAs / GaAs SL p-i-n structure. Al0.5Ga0.5As (2nm) / GaAs (5nm) SL solar cells are grown on n-type GaAs (001) substrates by a molecular beam epitaxy method. The total active layer thickness is fixed at 2 micrometer or 1 micrometer. In order to compare the effect the excitonic absorption, we fabricate a bulk Al0.14GaAs pin junction. The Al composition of Al0.14GaAs bulk layer corresponds to the averaged one of the SL. For the SL solar cells, the open-circuit voltage is 1.08 V, and the short-circuit current density is 23.6 mA/cm2 under the solar radiation of AM 1.5 spectrum at 100 mW/cm2. The fill factor and the conversion efficiency are 0.83 and 21.0 %, respectively. On the external quantum efficiency of the SL, two distinct maxima appear at approximately 680 and 790 nm with a dip structure in between. The calculated absorption rate with the effect of excitons shows the strong and sharp peaks at the absorption edge around 800 nm. In contrast, the step-like feature is obtained for the electron-hole pair absorption without considering the Coulomb interaction due to the SL density of states. The experimental results clearly indicate that the excitonic absorption is occurred even at room temperature. Thinning the active layer thickness reduce the optical absorption efficiency. Especially, the efficiency is easily decreased in the range near the absorption edge because of the lower absorption coefficient. For the Al0.14GaAs bulk solar cell, the decrease rate of the EQE in the range of 100 nm-wide wavelength from the absorption edge is estimated to be 27 %. On the other hand, the decrease rate for the SL solar cell is estimated to be 15 %. This result suggests the SL solar cells have high absorption coefficient compared with the bulk solar cells because of the excitonic absorptions, and the high absorption efficiency should be maintained when the active layers thickness are reduced.

Quantum dots (QDs) are an exciting group of materials because their bandgap is dependent on their nanocrystal size. The use of QDs allows for precise control over the wavelengths of light which are emitted or absorbed in applications such as light emitting diodes or photovoltaic devices. To produce efficient devices made from QD films, it is necessary to control the rates of charge transport and exciton diffusion among the QDs which make up the film. One way to affect these rates is by changing the ligand present on the surfaces of the QDs, which alters the distance and electronic coupling between neighboring dots. Large, bulky ligands serve to isolate dots from their neighbors while small, compact ligands allow for close packing and enhance electronic coupling. Regardless of whether the goal is to increase or decrease charge transport / exciton diffusion, it is important to be able to quantify properties such as the interparticle spacing and center-to-center distances in QD thin films. In this study, we characterize the morphology of lead sulfide (PbS) thin films made from highly monodisperse PbS QDs (diameter standard deviation < 5%, absorbance peak FWHM ~ 60meV). The native oleic acid ligands are exchanged for alkane monothiols, alkane dithiols, and several other commonly used ligands as well as treated with hydrazine to strip the QDs of all surface ligands. The absorbance spectra of the thin films are measured in order to characterize the shift in the first absorbance peak - which gives an indication of how much the local environment around each QD has changed. To determine the interparticle spacing, the dots are assembled into a close packed monolayer on a transmission electron microscopy (TEM) grid and subsequently have their native ligands exchanged. The TEM images are automatically analyzed to identify the boundaries of each QD as well as its center point. This information is used to calculate the average diameter and average center-to-center distance of all QDs in the image. From this analysis we are able to quantify the change in interparticle spacing as a function of the surface ligand species. These results are compared with those from small angle X-ray scattering (SAXS) on thin films, which also determines the average diameter and average center-to-center distance. We discuss the advantages and disadvantages of the TEM and SAXS measurements as well as the ease of data collection and analysis. Altogether, this study aims to provide a better understanding of how ligand chemistry can be used to control the morphology and electronic properties of a QD film.

Split spectrum photovoltaics, where incident light is divided onto multiple solar cells on the basis of wavelength, are an exciting recent development in the solar energy field. This technology has the potential to exceed record conversion efficiencies by allowing systems to incorporate a large number of active p-n junctions while mitigating the constraints that plague monolithic cells: namely lattice matching and current matching. Each cell in a split spectrum system can be designed to have a different lattice constant (allowing for a larger range or more combinations of materials) and have different operating currents (allowing for more combinations of band spacing).
In this work, we examine a split spectrum system utilizing a single spectrum splitting device, here a dichroic filter, to divide the solar spectrum onto two cells. As opposed to many other split spectrum designs where numerous filters direct light onto numerous single junction cells, in this system each cell is then comprised of multiple active junctions. Each cell is then tailored to absorb its own portion of the solar spectrum at its own operating point. The combination of the two cells allows for the constructing of a photovoltaic system which effectively has four or five active junctions while maintaining lattice and current matching conditions in each cell.
A number of different cutoff frequencies for the dichroic filter are examined. Each cutoff frequency corresponds to its own combination of ideal band placements for both the shorter and longer wavelength cells. Designs which have the potential to have both cells meet the lattice matching and current matching conditions are then simulated using TCAD Sentaurus. Designs which show promise of delivering high system conversion efficien