Symposium EN11 : Silicon for Photovoltaics

Silicon heterojunction (SHJ) solar cells [1], composed of a stack of thin intrinsic/doped amorphous layers, have been extensively developed towards high conversion efficiency photovoltaic property, which has been already employed in industrial production. Though a theoretical calculation demonstrates that the intrinsic limit of its efficiency reaches ~29% [2], the layer structures in actual SHJ devices still leave to be improved. The fabrication processing, such as doping methods, is considered for achieving ideal layer formation controlled with precise dopant concentration. In general, SIMS is used as a reliable analysis technique for obtaining the elemental profiles along the depth direction of the layers grown on flat substrates, which enables to determine an optimal doping condition. However, the elemental profiles on the pyramid surfaces in SHJ is unknown due to well-known dimensional limitation in the conventional analytical experiment. Therefore, it is a challenging issue to clarify how we effectively introduce the dopants in the pyramid surfaces determined by accurate characterization in commercial SHJ devices.
Laser-assisted atom probe tomography (APT) has been proven useful for visualizing elemental distributions in Si-based device structures [3]. APT has a potential to directly detect the dopants and draw the concentration profiles perpendicular to the pyramid surface thanks to a site-specific lift-out by focused ion beam (FIB) technique. In this work, we established an experimental success in analyzing pyramid surface area less than 2×2 μm² dimension. In order to introduce B atoms into amorphous layers, we compared the elemental distributions of p-type layer doped by two kinds of doping molecules; (CH₃)₃B and B₂H₆.
After forming textured surface by alkali solution, intrinsic amorphous Si thin layer was grown by using plasma chemical vapor deposition, followed by B doped layer formation using (CH₃)₃B or B₂H₆. Finally, ~100 nm-thick of indium tin oxide as a transparent conducting layer was deposited on top. For APT specimen preparation, FIB apparatus equipped with SEM was used. A local electrode atom probe (LEAP4000XHR, CAMECA) equipped with a 355 nm wavelength pulsed laser was employed for APT analysis. Based on the atom maps including thin intrinsic/doped amorphous layers, we found that the amount of carbon, as a residual contaminant, doped by (CH₃)₃B is higher than that by B₂H₆. Furthermore, a large amount of hydrogen was directly observed in the amorphous layers, which may affect the photovoltaic property. The atom maps around the intrinsic/doped amorphous layers will be also presented.
This work was supported in part by NEDO (New Energy and Industrial Technology Development Organization) under METI (Ministry of Economy, Trade and Industry).
[1] M. Taguchi et al., Prog. Photovoltaics 13, 481 (2005).
[2] L. C. Andreani et al., Adv. Phys.: X 4, 1548305 (2019).
[3] A. D. Giddings et al., Scripta Mater. 148, 82 (2018).

As an emerging technology, the surface passivation with an ultrathin silicon oxide and highly doped poly-Si stack has attracted a great attention because of the outstanding passivation quality. The carrier transport through the ultrathin SiO_x was originally proposed through quantum tunneling, and therefore it was named as tunnel oxide passivated contact (TOPCon)^[1] although the transport through pin-holes has been identified as the transport path in some cases^[2]. The highest laboratory solar cell efficiencies of 25.7% and 24.4% were achieved with n-type and p-type FZ c-Si wafers, respectively^[3.4], and a large-area cell efficiency of 22.8% was achieved with n-type CZ c-Si wafer using an industrial capable fabrication process^[5], demonstrating the potential in the efficiency improvement of solar panels. However, most of the poly-Si passivation layers are doped with P to form an n-TOPCon structure for high quality passivation. From the industrial application point of view, p-type poly-Si as the passivation and contact layer (p-TOPCon) is highly desired in the solar cell manufacturing, especially for upgrading the current PERC lines.
In in this contribution, we report our recent progresses in the research and development of high quality p-TOPCon, where the ultrathin SiOx layer is made by chemical oxidization in high temperature HNO₃, plasma assisted N₂O oxidization, and thermal oxidization; and the p-poly-Si layer is made by PECVD with in-situ doping. We find that the SiO_x made by the thermal oxidization performs better than those made by the other methods, with which a supper passivation using the p-TOPCon on n-type solar grade CZ c-Si wafers is achieved with iV_oc of 729 mV, effective carrier lifetime (t_eff) of 2.5 ms at 1×10¹⁵ cm^-3 carrier density, and saturated recombination current density (J₀) of 13 pA/cm². XPS analysis shows that the thermally oxidized film has a higher oxidization quality with a ratio of Si⁴⁺ chemical bonding configuration than those made by the other methods. In addition, the thermally oxidized SiO_x layer reduces the B aggregation in the SiO_x layer and the B diffusion into the c-Si wafer effectively as measured by ECV. We also investigated the PECVD a-Si:H deposition as well as various hydrogenation processes after the high temperature crystallization and their influences on the passivation quality. We find a wide PECVD process window for achieving high-quality passivation and an innovative hydrogenation method using water vapor annealing for improving the passivation quality effectively. We will also report the solar cell results with the p-TOPCon contact.
[1]. F. Feldmann, M. Bivour, C. Reichel, et al., Sol. Ener. Mater. Sol. Cel 131 (2014) 46–50.
[2] R. Peibst, U. Römer, Y. Larionova, et al., Sol. Ener. Mater. Sol. Cel. 158 (2016) 60-67.
[3] A. Richter, J. Benick, F. Feldmann, et al., Sol. Ener. Mater. Sol. Cel., 173 (2017) 96-105.
[4] A. Richter, J. Benick, R. Müller, et al., Prog. Photovolt. Res Appl. 23 (2017) 1-8.
[5] N. Nandakumar, J. Rodriguez, T. Kluge, et al., Prog Photovolt Res Appl. (2018) 1-6.

Silicon solar cells based on PERC architecture [1][2] have become the new industrial standard beyond the Al-BSF architecture [3]. To aim at higher efficiencies, research is on innovative architectures based on carrier-selective passivating contacts (CSPCs) for quenching recombination mechanisms at contact interfaces. Focussing on Si-based CSPCs, low thermal budget heterojunction (HTJ) cells and high thermal budget TOPCon-like cells have achieved record efficiency values ~26% in both front/back-contacted (FBC) [4] [5] and interdigitated back-contacted (IBC) [6] [7] configurations. However, these architectures are challenging from upscaling stand point; thus, leaner and cost-effective processes that can enable efficiencies between 24% and 25% on large area are industrially appealing.
In this contribution, we first review the opto-electrical properties and passivation quality of in-house developed semi-transparent CSPCs based on silicon alloyed with oxygen or carbon and manufactured at different thermal budgets. Then, starting from standard FBC cell configuration and recognizing the inherent parasitic optical losses due to Si-based CSPCs, we monitor the evolution of short-circuit current density in simple-process FBC and IBC devices endowed with our semi-transparent CSPCs.
In case of low thermal budget CSPCs compatible with HTJ architecture, we have developed and optimized n-type and p-type nc-SiO_x:H layers, which concurrently exhibit wide band gap, optimal crystallinity fraction and low activation energy (E_a-n = 46.6 meV and E_a-p = 89.6 meV). These properties make our nc-SiO_x:H layers excellent candidates for ideal carrier transport [8]. When employed in a double-side textured cell precursor based on 280-μm thick n-type FZ c-Si wafer, our nc-SiO_x:H layers enable iV_OC = 735 mV and iFF = 86.85% with τ_eff > 10 ms at also Δn < 10¹⁵ cm^-3. Rear junction devices based on such type of precursors, completed with sputtered ITO and Ag-based screen-printed contacts, present efficiencies in excess of 21% with J_SC = 39.1 mA/cm² for cell area of 7.84 cm² [9]. While further back-end optimization is needed to demonstrate higher efficiencies, this material system can boost the efficiency of FBC HTJ devices well beyond the current 25.1% efficiency record [4].
Passing to high thermal budget CSPCs, we have studied, as function of thickness, a wide range of poly-Si(O_x)(C_x) material systems by means of LPCVD (i.e. double-sided) or PECVD (i.e. single-sided) routes with ex-situ (implantation, diffusion) or in-situ doping and on flat or textured interfaces. Even though we have realized textured poly-Si CSPCs with thickness down to 6 nm (n-type poly-Si, V_OC = 728 mV, J₀ = 6.5 fA/cm²), our best n-type (p-type) CSPCs are flat 35-nm (20-nm) thick poly-SiO_x layers, which exhibit iV_OC up to 740 (716 mV) and J₀ down to 3 (11) fA/cm². These layers, similarly to poly-SiC_x layers [10], can be more transparent than poly-Si counterparts. As such, we have introduced polyO_x-polyO_x FBC devices [11][12], which could yield 21.5% for cell area of 1 cm², by slightly improving the V_OC of our poly-poly FBC standard devices (V_OC = 691 mV vs V_OC = 682 mV) and massively enhancing the J_SC (J_SC = 40.7 mA/cm² vs J_SC = 38.1 mA/cm²) without the use of any dual anti-reflective coating (DARC). To further augment the J_SC, we have devised a simple-process IBC poly-Si architecture, that exhibits J_SC-DARC = 42.2 mA/cm² [13] with 23% for cell area of 2 cm² [14].

[1] F. Fertig, et al., SiliconPV (2019)
[2] Press release Longi (2019)
[3] S. Gatz, et al., RRL (2011)
[4] D. Adachi, et al., APL (2015)
[5] A. Richter, et al., SOLMAT (2017)
[6] K. Yoshikawa, et al., Nat. En. (2017)
[7] C. Holleman, et al., PiP (2019)
[8] P. Procel, et al., SOLMAT (2018)
[9] Y. Zhao, et al., in preparation (2019)
[10] A. Ingenito, et al., IEEE JPV (2019)
[11] O. Isabella, et al., patent 2017E00057 NL (2017)
[12] G. Yang, et al., APL (2018)
[13] G. Yang, et al., patent 017880 NL-PD (2018)
[14] G. Yang, et al., SOLMAT (2018)

Carrier-selective passivating contacts (CSPCs) based on poly-Si [1][2], poly-SiO_x [3][4], and poly-SiC_x [5][6] have been developed and have enabled record-high efficiency in front/back-contacted (FBC > 25%) [5] and interdigitated back-contacted (IBC > 26%) [1] cell architectures. Such CSPCs consist of heavily doped poly-Si alloys deposited on an ultra-thin SiO_x layer, prepared by thermal oxidation [1], wet-chemical process [3], UV/O₃ process [7], or low temperature plasma oxide [8].
In this work, the ultra-thin SiO_x layers are prepared by oxidizing c-Si surface with N₂O plasma in the same Plasma Enhanced Chemical Vapor Deposition (PECVD) reactor that is used for the subsequent deposition of poly-Si(O_x)(C_x) layers. In this way, the deposition of both layers in a single reactor is enabled without breaking the vacuum. We optimized the conditions of N₂O plasma in terms of SiO_x layer thickness and uniformity. These PECVD N₂O plasma SiO_x layers (PO-SiO_x) are then tested in combination with our in-house developed poly-SiO_x CSPCs [3], by replacing the wet-chemical SiO_x layer with the PO-SiO_x in the role of tunnelling layer. The standard thickness of the poly-SiO_x CSPCs is 35 nm for n-type and 20 nm for p-type.
By varying the N₂O plasma exposure time, from 3 min to 18 min, the PO-SiO_x layer thickness can be well controlled between 1 nm and 2 nm. A longer-than 18 minutes N₂O plasma exposure does not further increase the PO-SiO_x layer thickness. Compared to the wet-chemical SiO_x formed by HNO₃, which is a self-limiting process growing up to around 1.4-nm thick SiO_x layer [2], our thickness-tunable PO-SiO_x is more favourable. In fact, according to rigorous TCAD modelling, for obtaining the highest possible FF in solar cells with high-thermal budget poly-Si(O_x)(C_x) layers, the thin SiO_x layer should be around 1-nm thick [9].
The poly-SiO_x CSPC processes based on wet-chemical SiO_x yield iVoc equal to 740 mV or 718 mV for flat or textured n-type, respectively, and 716 mV for flat p-type [3]. For n-type poly-SiO_x CSPC prepared on 1.3-nm thick PO-SiO_x on flat c-Si, an iVoc equal to 711 mV is obtained. By increasing the N₂O plasma treatment time, the iVoc increases to 723 mV. With the same process conditions, n-type poly-SiO_x CSPC on textured c-Si gives 710 mV. The p-type poly-SiO_x CSPC with PO-SiO_x is also under development by optimizing the N₂O plasma condition and the high-temperature annealing process. In this case, preliminary results indicate an iVoc equal to 668 mV for flat layers.
Poly-SiO_x – poly-SiO_x FBC cell, featuring wet-chemical SiO_x on both sides, textured n-type poly-SiO_x at the front side and flat p-type poly-SiO_x at the rear side, performs an efficiency of 21.5% with Voc = 691 mV, FF = 76.4%. The replacement of wet-chemical SiO_x with PO-SiO_x, is ongoing, for which a higher FF and similar Voc at cell level are expected, leading to a cell efficiency well above 21.5%. Results will be presented at the conference.

[1] F. Haase, et al. SiliconPV, 2018
[2] G. Yang, et al. APL, 2016
[3] G. Yang, et al. APL, 2018
[4] I. Mack, et al. SOLMAT, 2017
[5] A. Richter, et al. SOLMAT, 2017
[6] G. Nogay, et al. SOLMAT, 2017
[7] R. van der Vossen, et al. 7^th SiliconPV, 2017
[8] W. Lerch, et al., ECS Transactions, 2012
[9] P. Procel, et al. IEEE JPV, 2018

We summarize our recent findings on the properties of doped poly-Si structures for solar cell applications in terms of their impurity gettering effects (iron); through analysis of 3D elemental distribution profiles of, phosphorus /and boron (diffused dopants), undoped poly-Si (control) and implanted iron (impurity).
Doped poly-Si films with an underlying interfacial oxide layer have been shown to enable high efficiency silicon solar cells, due to their excellent electrical contact and surface passivating effects , and also because they possess very strong gettering properties (>99.9% metal (Fe- impurity) removal from the bulk) without compromising its passivating qualities.
Our study uses atom probe microscopy to determine the 3D distribution of elements to explore the phenomenon of gettering by poly-Si based layers. We investigate the role of dopants, oxygen, grain-boundaries and, interfaces, and observe if trends in the elemental distributions play a role in impurity gettering. This study is performed in conjunction with transmission electron microcopy (TEM) to facilitate atom probe reconstruction and support atom probe tomography (APT) analysis.
Our results indicated the gettering properties of poly-Si layers differ based on the dopants present (phosphorus vs boron) and are enhanced 1. in the presence of dopants (compared to undoped contacts), 2. at interfaces and 3. in the presence of oxygen.
The potential implications of these findings are in designing passivating contacts with enhanced gettering and passivating qualities. This will support the development of new solar cell processes for higher efficiency silicon solar cells at lower costs.

Defects in semiconductors are typically involved in performance degradation of photovoltaic devices in terms of recombination mechanisms and light instability [1-2]. Defects basically act as available energy states that can capture or emit charge inside the layer itself or to other materials/systems. In case of hydrogenated amorphous silicon (a-Si:H), these defects are related to non-periodic variations of the lattice potential, known as band tail energy states. These defects feature two state conditions: (i) neutral when empty or (ii) charged when occupied. Similarly, sub-gap electronic states close to mid-gap energy are defects in the form of dangling bond states [1]. The latter defects are dominant for recombination mechanisms at room temperature and they show amphoteric nature, which means that they exhibit three different states of charge. Altogether, these energy states build a complex configuration for charge collection that is not fully understood yet.
By means of rigorous TCAD modelling [3], we hereby discuss the role of the energy states for charge collection in silicon heterojunction (HTJ) solar cells inside contact stacks, namely intrinsic a-Si:H/doped n-type or p-type a-Si:H / transparent conductive oxide (TCO). Specifically, we investigate the different nature and properties of defects which affect recombination mechanisms for charge transfer, also known as trap-assisted tunneling (TAT). Furthermore, we correlate these mechanisms to solar cell external parameters for pursuing more efficient device.
We observe that TAT processes are not dominant in case of energy alignment of occupied states and available states within contact stack system. In presence of a proper energy alignment, field/thermionic emission and band-to-band processes from and to hetero-interfaces are dominant. In fact, such events are more efficient, as they do not exhibit any energy loss in terms of elastic transitions between energy states. On the other hand, TAT transitions are evident in the absence of energy states alignment and imply energy variation. Accordingly, we analyze the individual material properties that evidence TAT processes. Interestingly, such conditions are fulfilled only for positive charge collection while TAT is not apparent for negative charge collection. The electronic properties, ruling the aforementioned energy alignment, are associated to the Fermi-energy of each individual layer forming the heterojunction contact: p-type a-Si:H activation energy (E_a-p) and TCO carrier concentration (N_TCO). Therefore, we performed numerical calculations to evaluate charge transport in terms of fill factor (FF) within reasonable ranges of E_a and N_TCO values. From our modelling results, we observed four different intervals featuring different charge collection processes:
a) 35< E_a-p<260meV and 1x10²⁰< N_TCO< x10²¹cm^-3: Elastic charge transitions (FF>85%)
b) 260< E_a-p<430meV and 1x10²⁰< N_TCO<1x10²¹cm^-3: TAT based on dangling bond states (80%<FF<85%)
c) 35< E_a-p<260meV and 1x10¹⁹< N_TCO<1x10²⁰cm^-3: TAT based on valence band tail states (55%<FF< 80%)
d) 260< E_a-p<430meV and 1x10¹⁹< N_TCO<1x10²⁰cm^-3: TAT based on dangling bond states (60%<FF<80%). Note that here the energy variation within TAT intervals is larger than b).
We identified that defects featuring energy close to Fermi-energy (i) work more effectively on TAT processes and (ii) are closer than 10nm to the p-type a-Si:H/TCO interface (tunneling distance). In particular, TAT based on dangling bond states is more efficient than TAT based on tail states. Such a difference is ascribed to the initial dynamics (from neutral state) of recombination based on amphoteric states, that is able to capture positive charge from c-Si or negative charge from TCO, whereas tail states only allow to capture positive charge from c-Si.
[1] J. Melskens et al., Phys. Rev. B, 91, 245207 (2015)
[2] S. Olibet et al., IEEE J. Photovoltaics, 4, 1331 (2014)
[3] P. Procel et al., Sol. Energy Mater. Sol. Cells, 186, 66 (2018)

We demonstrate a new, purely mechanical mask alignment technique for dopant patterning of interdigitated back contact solar cells (IBCs) monocrystalline silicon (c-Si) solar cells. This technique uses ceramic pins to hold the mask and substrate tightly together for dopant deposition using plasma- enhanced chemical vapor deposition (PECVD). Problems associated with this dopant patterning step are investigated and solutions are developed for each. IBC silicon solar cells are the current record holders for silicon photovoltaic efficiencies based on various contacting structures [1]. Additionally, passivated contact schemes involving heavily doped polycrystalline silicon on silicon oxide (poly-Si/SiO_x) have been demonstrated on both two-sided architectures [2][3] and IBC architectures [4][5][6], achieving efficiencies over 25%. These poly-Si/SiO_x IBCs have been demonstrated with oxides of ~2.2 nm thickness to great success. Many groups have achieved high efficiencies on IBCs with these 2.2 nm passivating oxides by patterning the back-side dopants using photolithographic techniques, ion implantation, or laser doping, but these processes are undesirable due to their complexity, potential damage to the cell or cost. However, IBC solar cells have not yet been demonstrated using the ~1.5 nm tunneling oxide passivated contact structure fully contacted with doped polycrystalline silicon (poly-Si) fingers separated by intrinsic poly-Si. We investigate factors that are critical for the performance of IBC solar cells based on these tunneling oxide poly-Si passivated contacts. Because many of the above-mentioned dopant patterning techniques have led to fabrication complications, we have chosen to pattern the back side using masked PECVD of doped hydrogenated amorphous silicon (a-Si:H). Although this is a simpler process, there are a few troubles which can occur. During patterning of doped lines using direct deposition through a shadow mask, we show using time-of-flight secondary ion mass spectrometry (TOF-SIMS) that the intrinsic poly-Si gap becomes contaminated with dopants, leading to shunting. Two possible contamination mechanisms during high-temperature crystallization annealing are investigated. Using controlled experiments in a crystallization tube furnace, it is shown that dopants can desorb from the surface of the doped a-Si:H and transfer through the gas phase to be readsorbed onto the intrinsic gap and diffuse inward to contaminate the intrinsic gap. Mitigation strategies are developed for each of these contamination mechanisms to minimize shunt losses, confirmed using conductivity measurements. Dopant migration during PECVD is mitigated through a partial etch back of the a-Si:H surface to clear the intrinsic gap of dopants before crystallization. Gas phase transfer is mitigated by addition of a barrier between the doped a-Si:H and the gas phase, either with a capping layer or by annealing in an oxygen ambient which forms a surface layer of SiO_x, etched away by further processing. Complete IBC cells fabricated using these processing strategies will be presented at the conference. Improved IBC manufacture techniques will reduce the cost of such high efficiency cells and modules, promoting the expanded use of solar energy.
References
[1] Yoshikawa et al., Solar Energy Matrerials and Solar Cells, 173, 37-42, (2017)
[2] Feldmann et al., Solar Energy Matrerials and Solar Cells, 131, 46-50, (2014)
[3] Zhang et al., Solar Energy Matrerials and Solar Cells, 187, 13-122, (2018)
[4] Haase et al., Jpn. J. Appl. Phys., 56, 08MB15, (2017)
[5] Reinäcker et al., Energy Procedia 92, 412-418 (2016)
[6] Young et al., IEEE J. Photovolt. 6, 41 (2016)

Silicon wafers constitute a significant cost factor in solar cell and module manufacturing. The porous silicon (PSI) layer transfer process is a kerfless wafering technique that allows for a drastically reduced material and energy consumption per wafer and thus has great potential to reduce the wafer cost. In this process, a thick, highly p-doped substrate is electrochemically porosified, reorganized at high temperature and used as a growth substrate for silicon epitaxy. The epitaxial layer can subsequently be lifted off and the substrate wafer is reused. Key requirements for the realization of this potential in PV industry are, amongst others, a high lift-off yield of the epitaxial wafers from the porosified growth substrate, a robust process sequence, and a high electronic quality at the level of wire-cut Cz wafers. Furthermore, the kerfless wafers should be readily usable in standard solar cell manufacturing equipment, providing drop-in replacement wafers for PV industry. Here, we summarize the recent progress demonstrating several of these key aspects: first, we show that the PSI process is robust with respect to a moderate variation of processing parameters. In high-volume production, it is desirable to use e.g. a typical range of growth substrate resistivities and to use an electrolyte for the porosification process for an extended time. Both conditions would lead to a varying porosity of the growth substrate. We simulate this by a controlled variation of etching parameters, which also results in a variation of porosities in the growth substrate. We find a) that the lifetime of the PSI wafers is independent of the growth substrate separation layer porosities and b) that a reasonable process window in terms of the etching current density can be found. Second, we discuss the statistical evaluation of a large set of epitaxial runs with respect to the process yield: we are able to achieve lift-off of 59 out of 62 PSI wafers, which demonstrates a lift-off yield of the PSI process within our rigorously defined process window of at least 88 % with an error probability of 5 %. Finally, we show that high minority carrier lifetimes of up to 4.3 ms on n-type PSI wafers are achievable right after the epitaxy process, and that these lifetimes can be increased either by phosphorous diffusion gettering or by gettering with a n-type polysilicon on oxide (POLO) junction to up to 8 ms. With this, the PSI wafers are on par with standard PV wafers regarding the electronic quality.

Different silicon solar cells technologies have surpassed or are close to surpassing 26% efficiency. Dielectric oxides and amorphous silicon-based layers combined with minimal metal/silicon contact areas were responsible for reducing the surface saturation current density below 3 fAcm^-2. At open circuit, for passivating contact solar cells, the recombination is mostly fundamental (Auger and radiative), representing near 80% of the total recombination. At the maximum power injection, the fundamental recombination fraction drops to less than 50% as the surface recombination and SRH bulk step in. As a result, to further increase the performance is paramount to reduce the bulk dependence and secure proper surface passivation. Bulk dependence can be mitigated either by improving the bulk quality or by reducing the wafer thickness. We demonstrate for commercial relevant high-quality CZ wafers, thinner wafers and surface saturation current densities 10 times lower than the present state-of-the-art, are required to further increase the efficiency and narrow the gap towards the fundamental limit. For an n-type SRH bulk lifetime of 10 ms and a resistivity of 3 Ohm.cm, the optimum substrate thickness range is between 40 to 60 µm. By optimizing the thin layer of a-Si:H (<7nm) we have accomplished surface saturation currents below 1 fAcm^-2across multiple substrates thicknesses, reaching effective minority carrier lifetimes over 2.2 ms and implied open-circuit voltages of 765 mV on 40 µm-thin n-type CZ substrates after ip/in a-Si:H stack deposition. We use thinner wafers as testbed to optimize the thin intrinsic a-Si:H layer, as the effective minority carrier lifetime response to surface passivation increases inversely with the substrate thickness, and bulk recombination becomes less predominant in the final make-up of the lifetime, particularly at maximum power injection. We manufactured screen-printed silicon heterojunction solar cells on free-standing 40 µm-thin CZ-wafers with efficiencies close to 21%. In these cells the screen-printing layout needs to be optimized, and further gains in current and fill factor are expected in the coming months. These results were a direct consequence of an extreme good surface passivation and improved generation current provided by an antireflective bi-layer thin-ITO/SiO₂. The antireflective bi-layer increased the current density in 1.2 mAcm^-2 from our baseline antireflecting coating, mostly by mitigating the parasitic absorption via thinner ITO front layer.

Today' highest-efficiency solar cells typically operate near the threshold between low-level and high-level injection. It is not well understood if pushing further into a regime in which the cell operating point is solidly in high level injection at all times of the day has further benefits for initial solar cell performance, for reducing degradation rates, or for controlling the charge state and diffusivity of impurities that determine recombination in the bulk. In this work we explore the potential advantages of using very high resistivity n- and p-type, to manufacture high performance solar cells. Analytical modeling indicates that high resistivity substrates (10 Ωcm - >10k Ωcm) are required to have bulk Shockley-Read-Hall lifetimes in the millisecond range to outperform wafers with standard resistivities (< 10 Ωcm). Additionally, for resistivities over 10 Ω.cm, efficiencies show to be weakly dependent of the bulk resistivity. These results if experimental verified, can lead to more affordable ingot manufacturing, by lessening the requirements of dopants homogeneity along the ingot. We successfully passivated both n- and p- type substrates using i-a-Si:H, obtaining surface saturation current densities comfortable below 10 fAcm^-2and effective minority-carrier lifetimes over 2 ms at maximum power injection over the entire range of bulk resistivities (3 Ωcm- >10k Ωcm). For very high resistivity wafers (15k Ωcm) effective minority-carrier lifetimes over 10 ms were measured for both n- and p-type substrates. At moment we are manufacturing and characterizing solar cells and passivated samples, both p- and n-type over a wide range of temperatures (up to 80 ^oC) and injections. Preliminary results on silicon heterojunction solar cells indicate similar performance behavior for different light intensities regardless the base resistivity of the substrate.

Liquid phase crystallization (LPC) technology offers an alternative solution to the established wafer-based solar cells. It is a kerfless fabrication method that promises high efficiency solar cells on large-sized glass substrates. Recently, a conversion efficiency of 14.2% for back contacted 13 µm thick silicon on glass was demonstrated [1]. In this technology, amorphous or nano-crystalline silicon films prepared by various fabrication methods are used. In our work, we focus on films deposited by plasma-enhanced chemical vapor deposition (PECVD) due to its industrial relevance and easy integration into large-scale manufacturing. The standard layer stack for PECVD-based LPC solar cell consists of a multifunctional SiOx/SiN/SiOx (ONO) interlayer directly deposited on the glass followed by a 5-10 μm thick amorphous silicon (a-Si:H) precursor layer and finally a SiOx capping layer to promote the wettability of the film during crystallization. To drive out the hydrogen from the standard a-Si:H layer, the whole stack goes through a 2-day oven annealing (RT to 550 °C) process before crystallization by an 808 nm line-shaped continuous wave (CW) laser source. Without this hydrogen removal, the material delaminates during crystallization [2]. Unfortunately, this slow annealing process increases the manufacturing time.
In our contribution, we sum up our recent progress in the fabrication of the LPC layer stack and present an alternative solution to further reduce the manufacturing time of LPC cells. To reduce the annealing time, a porous material with interconnected voids was realized [3]. This allowed to rapidly anneal the a-Si:H precursor layer within 30 minutes (under ambient or nitrogen atmosphere) up to 550°C and drive out the hydrogen without material delamination, thus decreasing the manufacturing time and heating cost drastically. We examined this hypothesis by introducing artificial vents into dense films for the H effusion, which usually delaminate during the 30-minute rapid annealing process. With these vents, the film stabilizes depending on the distance between the vents. Furthermore, the LPC process induces strong thermal-induced stress which may lead to glass cracking. Profilometry was implemented to evaluate the radius of surface curvature and Stoney's equation was used to quantify the mechanical stress before and after laser crystallization. To lower the thermal-induced stress, laser scan speed was optimized to decrease the thermal gradient between layers of mismatching thermal expansion coefficients [4]. As a result, we were able to achieve (using porous precursor material) crack-free crystallized poly-crystalline silicon with grain diameters of several µm and grain lengths of several mm. Also, the crystallization was made possible without the need for a capping layer.
In conclusion, we were able to get rid of the capping layer fabrication step and significantly decrease the annealing time. Hence, we present a high-throughput approach for future up-scaling of liquid phase crystallization technology.

Keywords: Liquid phase crystallization, Amorphous silicon, Rapid annealing, Hydrogen effusion, Thermal-induced stress

References

[1] C. T. Trinh et al., SOL ENERG MAT SOL C 174, 187-195 (2018).
[2] O. Gabriel et al., IEEE J PHOTOVOLT 4, 1343-1348 (2014).
[3] W. Beyer, SOL ENERG MAT SOL C 78, 235-267 (2003).
[4] T. Pliewischkies et al., PHYS. STATUS SOLIDI A 212, 317-322 (2015).

Developing strategies towards Si-based tandem solar cells has gained considerable attention in the photovoltaic research community to prepare for the future demand of higher-efficiency solar cells. In this respect, we have proposed “smart stack” technology, an approach using metal nanoparticle arrays to bond dissimilar semiconductor materials both electrically and optically. Such functional metal nanoparticle arrays can be readily available by a method known as block copolymer nanolithography. Using Pd or Cu nanoparticle arrays, for example, we have fabricated triple-junction tandem cells comprising of InGaP/GaAs top cells and crystalline Si (c-Si) bottom cells. Through the fine tuning of the InGaP/GaAs cells as well as the introduction of advanced c-Si cell technologies (such as tunneling oxide passivated contact or heterojunction), efficiencies of over 30% have been recently realized. Other than InGaP/GaAs top cells, lead halide perovskites are of great interest, and efforts to demonstrate perovskite/c-Si tandem by the smart stack technology is also actively ongoing. This presentation overviews the concept, process, and device performances of the smart stack technology, including some of recent attempts.

Silicon is the dominating solar cell material, therefore add-ons on the silicon solar cell that can improve the power conversion efficiency are urgently needed. In certain organic materials singlet fission generates two triplet (spin 1) excitons from one singlet (spin 0) exciton. If the triplet excitons are harvested in the silicon solar cell the efficiency could be dramatically increased, as we show. There are different transfer pathways between the organic singlet fission material and silicon. We have simulated the achievable efficiency for each transfer path with realistic assumptions such as a singlet fission quantum efficiency of 1.7 (1.7 e-h pairs per high energy photon), a transmission loss of 5%, and different entropy gains of the Singlet Fission process.
Even with these realistic assumptions, the efficiency of a silicon/singlet fission solar cell can be as high as 34% when combined with the current record silicon solar cell of 27%. We found that dissociating the triplet excitons at the interface leads to a large potential efficiency gain because a triplet energy lower than the silicon bandgap still leads to charge generation, and allows for high current generation. We also find that current singlet fission materials do not absorb light strongly enough, motivating sensitization schemes. A direct triplet exciton transfer shows lower overall efficiencies because the energy level requirements are more strict, however the solar cell architecture is more elegant since there are no additional contacts needed. Finally, we compare the singlet fission/silicon solar cells to the efficiency potential of perovskite/silicon tandem solar cells. We find that tandem cells are particularly beneficial for a silicon base cell with low efficiency, while a highly efficient silicon solar cells benefits less from the perovskite top cell. In contrast, the efficiency gain from the singlet fission layer is almost constant for all silicon base cells, and for highly efficient silicon cells would clearly outperform a high-efficiency perovskite top cell.
We also fabricated Silicon solar cells with a top layer of tetracene. The silicon base cell are back-contacted, so we can HF-etch the silicon solar cell from one side to have direct access to the Silicon <111> sides of pyramidally textured silicon. Through measureing the photocurrent under a magnetic field we can differentiate between the photocurrent contribution of singlet and triplet excitons. A newly improved magnetic field dependent photocurrent setup allows us to measure current changes in the order of 0.01% and is a vital tool for a precise attribution of the origin of the photocurrent. We find that after deposition of the tetracene layer we see an injection of singlets or photons into silicon, but after aging the solar cell we see evidence for triplet transfer. The characteristic Merrifield curve (photocurrent as function of applied magnetic field ) inverts, which suggest the injection of triplet excitons from tetracene into silicon. This behavior can be observed for and tetracene-silicon solar cells that have been aged for five days in air or six weeks encapsulated in nitrogen atmosphere. We discuss a changing orientation of the tetracene molecules over time and a thin layer of silicon dioxide growing between tetracene and silicon. A better understanding of the energy transfer processes at the interface will be importante for future device applications.

Si clathrates are cage-like, crystalline Si allotropes with potentially exciting optoelectronic properties. Room temperature and atmospheric pressure metastable clathrates are synthesized in the presence of alkali guest atoms, such as Na, which occupy the interstitial sites in the cages. Na guests, however, degenerately dope the crystals. Realizing the potential of these materials as new, crystalline Si-based, earth abundant electronic materials requires fundamental understanding of guest properties and how to control their occupancy and diffusion in the cage-like host crystal structure. Elucidation of these properties is complicated by complex interactions among the Na donors. For this reason, a detailed understanding of isolated, or nearly isolated, Na donors is critical to eventually realizing the potential of these semiconductors for electronic and opto-electronic devices.
Here, we present an EPR study of Na guests in type II Si clathrate as a function of temperature and Na concentration. Since at low temperatures a Na donor is paramagnetic, EPR hyperfine interactions allow low concentrations of Na ions to be detected while probing the interaction between ions and the Si cages. In low temperature measurements, the EPR spectrum with ≤1 at. % of Na exhibits four hyperfine lines due to the interaction of the Na valence electron with the nuclear spin (I=3/2) of Na²³. The hyperfine coupling constant (13.3 mT) together with the atomic coupling constant for an isolated Na atom (31.6 mT) indicates that about 42% of the wave function of the paramagnetic electron resides on the Na atom and the rest resides on the surrounding Si sublattice. The type II crystal structure includes both small and large cages. At this low Na concentration, the Na is almost entirely in the large cages and a small anisotropy in the four hyperfine lines indicates it is slightly off center in the cage, as suggested by other measurements. We also identify an EPR contribution at g~2.005 which is consistent with the presence of Si dangling bonds in a highly disordered or amorphous Si phase. Identifying this line as due to dangling bonds allows a clearer identification of the other features in the clathrate spectrum.
The observation of structure from "super hyperfine" interactions of the Na valence electron with more than one magnetic nucleus provides additional insight into the properties Na as a donor in type II clathrate. Doublets surrounding each of the four Na hyperfine lines are identified, using EasySpin spectral simulations, as arising from naturally occurring Si²⁹ nuclei in the surrounding Si cage. Based on the magnitude of the Si²⁹ interaction, it is estimated that about half of the spin density on the Si sub-lattice extends past the confining cage. Additional fine structure is observed halfway between the strong hyperfine lines. Simulations suggest they arise from the interaction of either one or two electrons on Na²³ nuclei (I=3) in adjacent cages. This spectrum consists of 7 lines, four of which are hidden by the four strong hyperfine lines for isolated Na²³ atom cages, with the others positioned between the hyperfine lines as observed. Interactions of electrons associated with Na²³ atoms in adjacent cages is consistent with spin density extending beyond the confining cage. In addition to probing the nature of the Na donor in the clathrate structure, EPR provides other useful information. For example, using the integrated intensity of the Na hyperfine lines we estimate the concentration of isolated Na in samples specifically prepared to have low Na content to be ~3 x 10¹⁷/cm³ which is approaching a regime where the doping is not degenerate. This work was supported by National Science Foundation Award #1810463.

Intermediate band photovoltaics have been proposed to better utilize the solar spectrum and ‘hyperdoping’ of silicon is an effective method to realize such a material [1]. We investigate various hyperdoping methods and dopant concentration profile tailoring, to increase the charge carrier lifetime in intermediate band semiconductor materials. Hyperdoping is a well-established technique to introduce large concentrations of deep-level dopants such as heavy chalcogens and transition metals for intermediate band formation [1,2]. Various hyperdoping and annealing methods — namely, dopant ion implantation followed by pulsed laser melting (PLM) [3], or dopant ion implantation followed by flash lamp annealing (FLA)[4], or dopant thin film deposition followed by PLM [5] can produce impurity-supersaturated, single crystalline layers, which absorb sub-bandgap light. A drawback of the non-equilibrium PLM technique is inherent dopant segregation to the silicon surface [6], which leads to short charge carrier lifetimes without contributing to device performance. In this study, we investigate the extent of the surface segregation on the photocarrier lifetime using controlled etching methods as a way of removing the surface-segregated silicon surface.
We investigate the influence of the etch processing on charge carrier lifetime using terahertz (THz) spectroscopy. Optical pump THz probe spectroscopy is a non-contact photoconductivity measurement for measuring carrier lifetime with picosecond temporal resolution [7]. We focus our study of the recombination dynamics on the gold-hyperdoped silicon (Si:Au) system, prepared via thin film deposition and PLM. We find that whereas unprocessed Si:Au layers exhibit rapid recombination with a lifetime of 16 ps, CF₄-based reactive ion etching (RIE) increases the lifetime by a factor of 2, contrary to the widely-known role of RIE for increasing surface recombination. We also characterize the structural and absorption properties of the material. Additionally, we simulate the lifetime dependence on dopant concentration profile. We use these THz lifetime measurements to verify results of dopant profile engineering, to optimize materials for intermediate band photovoltaics. This study provides a material processing-based approach towards overcoming the historically-low external quantum efficiencies reported for hyperdoped Si optoelectronic devices.

[1] Warrender J. M., “Laser hyperdoping silicon for enhanced infrared optoelectronic properties”, Applied Physics Review 3, 031104 (2016)
[2] Mailoa J. P., et al., “Room-temperature sub-band gap optoelectronic response of hyperdoped silicon”, Nature Communication 5, 3011 (2014)
[3] Winkler M. T., et al, “Insulator-to-metal transition in sulfur-doped silicon”, Physics Review Letters 106, 178701 (2011)
[4] Liu F., et al., “On the insulator-to-metal transition in titanium-implanted silicon” Sci Rep. 8, 4164 (2018)
[5] Warrender J. M., et al., “Incorporation of gold into silicon by thin film deposition and pulsed laser melting” Applied Physics Letters 109, 231104 (2016)
[6] Recht D., et al., “Supersaturating silicon with transition metals by ion implantation and pulsed laser melting” Journal of Applied. Physics. 114, 124903 (2013)
[7] Sher M. -J., et al., “Picosecond carrier recombination dynamics in chalcogen-hyperdoped silicon,” Applied Physics Letters 105, 053905 (2014)

Following the recent success of monolithically integrated Perovskite/Si tandem solar cells, there has been a surge of interest in alternative wide bandgap top-cell materials with prospects of a fully earth-abundant, stable and efficient tandem solar cell. Thin film chalcogenides such as Cu₂ZnSnS₄ (CZTS, 1.6 eV band gap) or Cu₂BaSnS₄ (CBTS, 2.0 eV band gap) could be suitable candidates. However, this class of materials has the disadvantage that generally at least one high temperature step (> 500 C) is needed during the synthesis, which could contaminate the Si bottom cell. Here, we systematically investigate the monolithic integration of CZTS and CBTS on a Si bottom solar cell. A simple double-sided Tunnel Oxide Passivated Contact (TOPCon) structure is used as bottom cell, and a thin TiN layer is selected as both a diffusion barrier and a recombination layer between the two sub-cells. We show that TiN successfully mitigates in-diffusion of CZTS elements into the c-Si bulk during the high temperature sulfurization process, and no evidence of electrically active deep Si bulk defects in samples protected by just 10 nm TiN. Post-process minority carrier lifetime in Si exceeded 1.5 ms, i.e., a promising implied open-circuit voltage (i-V_oc) of 715 mV after the high temperature sulfurization. Based on these results, we demonstrate the first proof-of-concept two-terminal CZTS/Si and CBTS/Si tandem devices with a efficiencies up to 3.3%. A general implication of this study is that the growth of complex semiconductors on Si using high temperature steps is technically feasible, and can potentially lead to efficient monolithically integrated two-terminal tandem solar cells.

Supersaturated solutions of transition metal impurities in Si have been shown to create intermediate bands (IBs) in between the valence and conduction bands in Si. This new IB induces sub-band gap absorption, thereby increasing the range of optical absorption in silicon further in the infrared. The extended response of this hyperdoped Si make it a promising material for photovoltaic devices. Ion implantation followed by pulsed laser melting has been demonstrated as a method to produce concentrations of impurities in Si that are well above the solid solubility limit. Recently, photodetectors fabricated from Si hyperdoped with Au or Ti have been shown to have sub bandgap responsivity, demonstrating the material’s promise for increasing the efficiency of Si solar cells beyond conventional limits.
To achieve devices that could be commercialized for PV cells or other demanding applications, hyperdoped Si must show significant optical absorption, and high quality Ohmic contacts for carrier extraction must be achieved. In this work, we fabricated Si layers hyperdoped with Au or Ti at varying thickness, measured the optical absorption enhancement relative to Si, and attempted to form Ohmic contacts to the layers. The results show significant enhancement of optical absorption by increasing the implant energy. For making Ohmic contacts to hyperdoped materials, we tried several treatments, including boron or phosphorus shallow doping, rapid thermal annealing (RTA) of contact, etching off the top metallic layer, and modifying the PLM process to suppress dopant segregation. Recipes for Ohmic contacts to each layer were demonstrated. And the low ohmic contact resistivity around 0.1 ohmcm2 can be achieved.
Prototype photodetectors were fabricated out of Au or Ti hyperdoped Si materials produced using the II-PLM method. The devices were fabricated using the optimized Ohmic contact process to maximize the carrier extraction. The IV characterization shows a rectifying effect for the Si based junction. The detection range is extended to 2um, well beyond the limit of typical Si photodetectors. More importantly, the external quantum efficiency in the sub-band gap region was increased by nearly two orders of magnitude over previously published work. These results represent a significant improvement in the development of a hyperdoped Si solar cell. The enhancement in optical absorption and carrier extraction should translate to a significant increase in solar cell efficiency.

ZnS_xSe_1-x films are promising materials for front carrier-selective contacts in silicon photovoltaics given their wide bandgaps and low resistivities compared to amorphous silicon, with the potential to capture more photo-generated current than a traditional heterojunction with intrinsic thin layer (HIT) solar cell. We have synthesized ZnS_xSe_1-x films (x ranging from 0 to 1) grown on Si by molecular beam epitaxy and employed X-ray photoelectron spectroscopy and spectroscopic ellipsometry to measure band offsets of ZnS_xSe_1-x with respect to Si and have compared the results to Sentaurus optoelectronic simulations of photovoltaic devices incorporating ZnS_xSe_1-x carrier-selective contacts. Further experimentally-determined parameters, including complex refractive index and resistivity, were also included in simulations of a HIT-style cell to determine the ZnS_xSe_1-x top contact mole fraction x, doping level, and thickness for optimal device performance. In simulations, the highest performance ZnS_xSe_1-x contact is found to outperform the simulated and experimental performance of a comparable HIT cell with an a-Si/ITO front contact. Experimental transport measurements for ZnS_xSe_1-x top contacts will be discussed.

This material is based upon work supported by the NSF and the DOE under NSF CA No. EEC-1041895, by the DOE under Award Nos. DE-EE0006335 and DE-EE0004946, and by the NSF Graduate Research Fellowship under Grant No. 1144469.

The low temperature budget, availability of various deposition techniques, and wide range of adjustability in physical and electrical properties makes poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) a promising candidate as an hole-transport layer (HTL) for low-cost silicon based heterojunction solar cells. However, majority of works utilizing a PEDOT:PSS layer for silicon solar cells incorporate a spin-coating process for the deposition, which in fact is not a suitable technique for mass production as it inherently requires batch processing and a large portion of the dispensed solution is wasted in the process. In this work, we spray coat PEDOT:PSS layers on silicon surfaces as an alternative and use two different equipment: an in-house built airbrush system and a commercial ultrasonic coating system. For both systems, we investigate the coating quality on mirror-polished and random pyramid textured monocrystalline wafers. We analyze the effect of sample temperature and sample-to-nozzle distance on morphological properties and on electrical performance (contact resistivity, surface saturation current density) of the deposited films. Since the optimal deposition parameters and chemical content of the PEDOT:PSS films are heavily influenced by the hydrophobicity of the silicon surface, we perform the depositions on (1) surfaces having an ultra-thin chemical oxide, and (2) surfaces that are subjected to an HF dip prior to the deposition for oxide removal. Under optimized conditions, we achieve surface saturation current densities below 30 fA/cm² and contact resistivities below 100 mΩ.cm². Finally, to demonstrate the performance on a device level, the films are used as HTLs at the rear side of silicon solar cells where an n-type amorphous silicon layer serves as the electron transport layer at the front side of the cell.

Si-based near-and mid-infrared photodetectors have been demonstrated through the introduction of an intermediate band in the bandgap of Si by hyperdoping with deep level impurities. The potential CMOS compatibility of such materials makes it attractive to the ICT and military sectors, and could advance the realization of cheap and high-resolution infrared imaging. These materials are typically fabricated by ion implantation followed by pulsed laser melting (PLM) to achieve hyperdoping concentrations beyond the Mott limit (~10²⁰ cm^-3).

Studies involving Te-, Se-, and S-hyperdoped Si have demonstrated optical activity up to the mid-infrared [1-2]. However, at the highest doping concentrations the material undergoes an insulator-to-metal transition, making it unfavourable for a near-infrared photodetector. On the other hand, detectors made from Si hyperdoped with transition metals that introduce deep level states closer to mid-gap of Si, such as Ti and Au, exhibit a near-infrared photoresponse with an activation energy between 0.72—0.78 eV [3-4]. However, these diodes have very low quantum efficiencies, and the reason for this is currently unclear.

In a recent study of Au-hyperdoped Si diodes [5], we observed a deep level defect at E_C-0.35 eV originating from the interaction between implantation and PLM-induced defects at concentrations as high as 10¹⁴ cm^-3 within the device depletion layer. In this presentation we present photocurrent data from these diodes and show that the E_C-0.35 eV defect and the Au-related deep level give comparable contributions to photocurrent. Hence, our experiments suggest that the near-infrared spectral kink previously observed in both Au- and Ti-hyperdoped Si diodes is almost certainly related to the E_C-0.35 eV defect and not hyperdoping, indicating that there is an underlying problem with present photodiode structures for hyperdoped materials. In a further series of experiments, we show that the hyperdoping of Ti in Si (required for optical activity) cannot be achieved by PLM in the nanosecond regime, providing further evidence for the contribution of processing defects to the observed near-infrared absorption and photoresponse in the Si(Ti) system.

[1] T. G. Kim, et al., Appl. Phys. Lett. 88, 241902 (2006).
[2] M. Tabbal et al., Appl. Phys. A 98, 589 (2010).
[3] J. Olea, et al., J. Appl. Phys. 104, 016105 (2008).
[4] P. Mailoa, et al., Nat. Commun., 5:3011 (2014).
[5] S. Q. Lim, et al., to be published, (2019).

Silicon microwires (MWs) have been widely investigated for realizing the high-efficiency silicon solar cells thanks to the outstanding light absorption property and ease of the formation of the radial junction. To achieve the high-efficiency MW based radial junction solar cells, it is necessary to control the structure of MW arrays such as diameter and length of the MWs and spacing between MWs. The structure of MW arrays is generally determined by the process conditions of the optical lithography, which is the method of forming the pattern. However, the patterning process based on optical lithography reaches the fundamental diffraction limit of about 200 nm. Also, the optical lithography process uses photoresist that is harmful to the human body and requires expensive equipment and a specific place, such as a clean room.
In this presentation, we introduced a novel fabrication process of MW arrays through microsphere lithography. In general, the microsphere lithography is the technique to form hexagonally packed ordered arrays of micrometer-sized latex or silica spheres as lithography masks. The spacing between the spheres can be controlled simply by etching the uniformly arranged the microsphere arrays. Also, microsphere lithography has the advantages of low cost, large scale fabrication, and high throughput performance.
We formed a monolayer of polystyrene beads using microsphere lithography on a silicon wafer and controlled the spacing between the beads through the oxygen plasma etching up to 100 nm. As a result, we successfully fabricated MW arrays with a dense spacing of 100 nm using polystyrene beads monolayer as an etching mask. The fabricated MWs via the microsphere lithography process exhibited an excellent light absorption of 97% at the entire wavelength region (300-1100 nm) even without anti-reflection film layer. The MW arrays with dense spacing enable the use of MWs that are shorter than that of MWs with micro-scale spacing due to the reducing of a reflective surface. Also, the reduction of the MW length has the effect of reducing the surface recombination of the carriers separated by the radial junction of the MW. Therefore, we expect the proposed microsphere lithography would be the essential process for the formation of optimized MW arrays for the high-efficiency radial junction solar cells.

BaSi₂ is regarded as a promising absorber material in thin-film solar cell applications, due to its excellent optical and electrical properties, including a suitable band gap (~1.3 eV), a high light absorption coefficient reaching 10⁵ cm⁻¹ for photon energy > 1.5 eV, a long minority carrier lifetime τ (~10-27 μs), and essentially elemental abundance and nontoxicity [1]. Besides molecular beam epitaxy and thermal evaporation, sputtering is an industrially applicable technique to fabricate BaSi₂ thin films. However, structure transformation and property degradation due to thermal annealing process have become major concerns for obtaining high-quality BaSi₂ films [2]. Herein, we present a facile annealing method to suppress the surface oxidation of sputtered BaSi₂ process by employing a capping BaSi₂ film, which further improves surface uniformity of BaSi₂ films.
Depositions of BaSi₂ thin films were done in a radio-frequency magnetron sputtering setup. Detailed deposition parameter can be found in our previous research [2]. Glass substrates were used. After the sputtering process, deposited BaSi₂ films were subsequently annealed at 600 °C for 30 min in vacuo. During the annealing process, the sample was covered by the other BaSi₂ film on the glass substrate, with a film facing film configuration. Raman spectra were recorded on a Renishaw InVia Raman microscope with a 633-nm laser. Raman mapping was done with a scan area of 40 × 40 μm².
Three distinct regions can be observed from the Raman mapping of the BaSi₂ sample, which was partially covered by the other BaSi₂ film during the annealing, including (i) a uniform region (covered by BaSi₂), (ii) a transition region and (iii) a non-uniform region (exposed to the annealing environment). Five Raman peaks of BaSi₂ (bands at 276 cm⁻¹, 293 cm⁻¹, 355 cm⁻¹, 376 cm⁻¹, and 486 cm⁻¹) can be found in all these three regions [3]. In addition to those peaks, a peak at 245 cm⁻¹, which might be related to the surface oxidation, and a peak at 519 cm⁻¹, which indicates the formation of Si nanocrystals (NCs), can be observed in the transition and non-uniform regions. Raman intensity maps of peak 245 cm⁻¹ show that surface oxidation mainly occurred in the exposed (non-uniform) region, and more Si NCs are also detected in this region due to the oxidation. Contrarily, the covered region presents nearly no Si NCs and oxide peaks. This suggests that surface oxidation could be effectively suppressed by covering BaSi₂ films during annealing. Also, BaSi₂ distributes more uniformly in the covered region according to the Raman maps of peak intensity of 486 cm⁻¹ (BaSi₂). In the exposed region, a smaller width of peak at 486 cm⁻¹ can be observed, uncovering a better BaSi₂ crystalline quality in the exposed region comparing to the covered region. Because of the existence of capping BaSi₂ layer, less heat can reach to the covered BaSi₂, thus leading to a less effective crystallization.
In this annealing method, the capping BaSi₂ layer might function in effect as an oxygen-eliminating agent, so that less oxygen exists between those two BaSi₂ films. In our configuration, the capping BaSi₂ layer is closer to the heater, which heats to a higher temperature the capping BaSi₂ compared to the covered BaSi₂. Hence, the residual oxygen in the annealing environment is inclined to react with the capping BaSi₂ film and this suppresses the oxidation of the covered BaSi₂. However, the lower temperature of covered BaSi₂ in turn slightly slows down the crystallization of BaSi₂. Currently, we are working on the optimization of annealing condition to improve the crystalline quality, such as increasing the temperature, prolonging the duration and adjusting the annealing atmosphere. Further characterizations are being conducted as well.

References
[1] T. Suemasu, and N. Usami, J Phys D Appl Phys, 2 (2017) 023001.
[2] Y. Tian, et al., ACS Appl Energy Mater, 1 (2018) 3267.
[3] Y. Terai, et al., Jpn J Appl Phys, 5S1 (2017) 05DD02.

Although semiconductors pn junctions and metal semiconductor Schottky junctions are considered to be constituent elements of photovoltaic and electronic devices. But after the discovery of graphene, immense interest has been developed to make graphene silicon Schottky junction solar cells. The ultrathin 0.33 nm layer of graphene has capability to absorb 2.3% of visible light. These ultrathin layers are not just used as an absorbance material, but due to its essential properties such as tunable work function it increases the Schottky barrier height and built in potential for efficient transport of charge carriers in photovoltaic devices. Recently, graphene/Si Schottky junction solar cells are extensively used to harvest solar energy but high efficiency is limited due to huge surface recombination at the interface. Moreover, during the wet transfer process; wrinkles, surface defects and impurities deteriorate the performance of device. Thus, we propose an ideal approach to directly grow the graphene on planar and textured silicon by using plasma-enhanced CVD, compatible with industrial level applications. Thus, by controlling the thickness of interfacial layer Al₂O₃ and graphene, we achieved highest efficiency. The high-k dielectric layer of Al₂O₃ blocks the electron and eventually reduces the surface recombination at the interface. However, with the optimization of graphene thickness layer, there was an improvement in the work function of graphene, likewise an enhancement in open circuit voltage. Resultantly significantly improving overall device efficiency. Additionally, with control of thickness of graphene on bare silicon, efficiency was increased from 3.6% to 5.51%. Thus, so far with interfacial layer optimized thickness, we achieved highest efficiency of 8.4%. Interestingly, with doping of nitric acid and anti-reflection polymer layer coating of PMMA efficiency is remarkably improved from 5.51% to 9.18%. However, with doping and polymer layer coating the cell stability for longer time was compromised, while, with Al₂O₃ interfacial layer solar cells have excellent stability for more than one year.

The photovoltaic industry is mainly dominated by crystalline silicon (c-Si) based solar cells where, the contact selectivity is usually achieved by doping the wafer surfaces with phosphorus and boron atoms.
Several alternatives are used in order to avoid the high temperature, furnace-based, diffusion process. Examples include silicon heterojunction using both intrinsic and doped amorphous silicon films, or the formation of p⁺ and n⁺ regions by laser-firing of doped dielectric films. Nevertheless, in both cases the use of toxic and flammable gases is required.
Recently, the use of dopant-free materials based on transition metal oxides like MoO_x and V₂O_x have shown excellent hole and electron selectivity. The use of TiO₂ is an attractive option to form electron-selective contacts, due to its small conduction- and large valence-band offsets (ΔE_c ~0.05 eV, ΔE_v ~2.0 eV), allowing an easy electrons transport through the c-Si/TiO₂ interface while blocking the holes. The introduction of a SiO₂ interlayer at the c-Si/TiO₂ interface improves the quality of the selective contact reaching efficiencies up to 21.6%. The replacement of this high temperature (700 C) SiO₂ layer by other dielectric films deposited a low temperatures is an interesting objective. [1]
In this work we study the properties of Al₂O₃/TiO₂ stacks deposited by Atomic Layer Deposition at low temperature (125 C) as electron transport layers. The goal is to use the optimized Al₂O₃/TiO₂ stacks as selective contacts in interdigitated back-contacted (IBC) c-Si(n) solar cells. Results have confirmed surface recombination velocities below 40 cm/s with implied open circuit voltage values of 675 mV in symmetrical Al₂O₃/TiO₂ test samples. Specific contact resistance values below 3 mΩcm² are also measured. These excellent results pave the way to use these stacks as electron selective contacts on IBC solar cells, in combination with V₂O_x hole selective contacts. Experimental and technological details as well as first IBC solar cell results will be presented.
[1] X. Yang, et al., Advanced Materials 2016, 28, 5891.

To produce high-quality silicon ingots for solar cells, good control of growth conditions such as the gas flow and the temperature distribution is required. This last has a direct effect on the growth rate, the shape of solid/liquid interface, and thermal stresses. For this concern, it is important to measure the temperatures inside the furnace and around the crucible. However, it is difficult to implement infrared thermometers in directional solidification (DS) furnaces nor installing many thermocouples (TCs) because of the limited space and the movement during the growth. The current study focuses on finding the appropriate positions to install TCs around the crucible that allow us to get the temperature distribution. Due to the presence of abundant variables such as the crucible movement and the temperature of multiple heaters, we decided to solve this issue by a combination of crystal growth simulation and machine learning. Crystal growth experiments with typical conditions were simulated using CGSim software, and the results were used as training data for machine learning. To predict temperature distribution from the limited number of TCs, we have made ₂₁C₄ = 5985 neural network (NN) models by picking 4 TC positions out of 21 candidate positions on the crucible wall. The NN models consist of 8 inputs (3 heater temperatures: upper, middle and bottom, 4 measured TC temperatures: TC₁, TC₂, TC₃ and TC₄, and crucible position) and 21 outputs that correspond to the temperatures distribution on the crucible wall. The minimum loss was obtained at the NN model with the TC positions at 0, 1, 2 and 9 cm from the bottom of the crucible to its top. All the NN models with the ten lowest losses have the TC positions 0, 1 and 2 cm from the bottom of the crucible. This suggests the importance of measuring the temperature around the bottom of the crucible to obtain the precise temperature distribution.

Mono- and multi-crystalline silicon-based solar cells form the bulk of global installed photovoltaic capacity (~403 GW in 2018) for commercial power generation, with a 27% cumulative annual growth rate achieved in 2018, with grid parity being achieved in an increasing number of countries. The industry has matured over the last 15 years, going from silicon shortage to solar overcapacity, which has resulted in anti-dumping duties being imposed in several countries. Power conversion efficiencies in silicon are impacted by bulk defects, contacts, solar spectrum mismatch, etc. Specifically, the absorption coefficient in silicon starts declining below 10³ cm^-1 above around 800nm, declining to one-tenth that value near 1000nm. Such material limitations also manifest themselves in a peak in the responsivity of silicon-based detectors. In this work, we report on the development of an upconversion-based absorber (NaYF₄: Yb(18%), Er(2%), Gd(15%)) operating near 980 nm used for increasing the efficiency of a commercial solar cell. The absorber was synthesized using thermal decomposition, followed by powder X-ray diffraction (PXRD) measurements which reveal a mostly hexagonal phase closely associated with high upconversion photoluminescence (PL) efficiencies. An ink composed of the product with polystyrene (PS) was formed with a loading ratio of 23% w/w in chloroform. Micro glass slides were cleaned using stepwise solvent clean, followed by a plasma ashing step, prior to spin coating with the ink to deposit a thin film. Scanning electron microscopy (SEM) scans of the deposited reveal a thickness of ~158±33 nm. Atomic force microscopy (AFM) scans reveal that the roughness of the films is approximately 31 nm. When excited with 785 nm laser light (intensity ~560 W/cm²), upconversion emission at green (539 nm) and red (665 nm) is observed using a Horiba/Labram spectrometer. Upon excitation with 980 nm, however, green emission is observed with the naked eye. To test the viability of the idea, we used the coated glass slide in two measurement setups with a low-cost commercial solar cell (the slide placed on top of the solar cell): a) a 980 nm laser diode (Thorlabs, intensity ~ 450 mW/cm²), b) under AM1.5G conditions with a solar simulator (Newport LCS-100). Under direct irradiance of 980 nm laser diode, we observed an enhancement of the short-circuit current (J_sc) of ~ 25%. The control device in each case was a blank uncoated glass slide placed on top of the solar cell. As expected, the enhancements were found to be somewhat lower with the solar simulator (~7.2% in J_sc). These increases are preliminary results with unoptimized films in a measurement geometry that is unlike one we envisage for deployed solar cells. We are currently engaged in verifying these initial measurements, and tuning the precise loading ratio with these upconversion absorbers to obtain higher increases in short-circuit currents, and developing films that can be deposited directly on top of the commercial solar cell to minimize incoupling losses.

The growing energy supply demanded by modern society poses challenges to scientists and engineers in that many of the currently used technologies and prospective innovations must be directed in a more sustainable way. The reduction in the use of fossil fuels and replacement with new environmentally friendly energy sources, especially renewable energy resources represents one of the present grand energy challenges. Sunlight is the most abundant free renewable energy source and crystalline silicon photovoltaics are among the leading technologies for solar power conversion. The efficiency of the solar cell mainly depends on the band gap and is limited to a maximum theoretical power conversion of ~ 31% [1]. One possible strategy to improve efficiency is to optimize the response of the photoactive materials at shorter wavelengths. The combination of high-efficiency solar cells in a tandem configuration on a Si platform using thin film III–V materials as well as perovskites has recently attracted great attention [2]. Silicon clathrates are inclusion compounds that consist of a Si cage-like framework with a direct 1.9 eV band gap and can be considered for photovoltaic applications including thin-film single-junction devices such as Si clathrate on c-Si for all-Si tandem cells.
The structural optimization and electronic properties of Si clathrates with cubic structures I (CS-I) and II (CS-II), silicon surface and Si clathrates on silicon surface have been performed using first-principles calculations within the framework of density functional theory (DFT).
The atomic optimization of Si clathrates with and without Na atoms as well as Si (001) surface were carried out. The obtained results showed that the band gap of pure Si clathrate is larger in comparison of band gap of Si surface. The present of sodium shifts the Fermi level to conduction band which indicates the importance of guest removing from clathrate framework. The reconstruction Si clathrate on c-Si surface was performed using ab-initio molecular dynamic simulations during 5 ps at 300 K. The clathrate structure remains cage-like framework after deposition of Si surface and the electronic structure of tandem configuration is changed due to removing dangling bonds on the surface. Thus, the present study revealed the possibilities for modifying the electronic and optical properties of crystalline silicon photovoltaics by deposition of thin-film Si clathrate.

REFERENCES
[1] D. Ginley et al. MRS Bull. 33 (2008) 355.
[2] M. T. Borgström et al., IEEE J. Photovolt. 8 (2018) 733.

Since the conversion efficiency of single junction silicon (Si) solar cells is close to the theoretical limit, the application of tandem solar cells with stacking order in different band gap energy (E_g) has attracted the attention. It is expected that the solar cell performance can reach around 40% efficiency by combining the silicon solar cell (E_g~1.12 eV) as the bottom layer and dye-sensitized solar cell (DSSC) based on the N719 dye (E_g~1.7 eV) as the top layer for tandem solar cell structure. In that cell, poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate)/PEDOT:PSS can be used as the buffer layer between the top cell and bottom cell which involved the silver (Ag) grid and passivation layer component. Although the DSSC/Si tandem solar cells have been investigated so far, there is no detailed information near the PEDOT:PSS of this tandem solar cell. Therefore, here, we studied the effect of changing in property of buffer layer containing PEDOT:PSS on the solar cell characteristics.
To make the solar cell device, firstly, we prepared the electron transport layer by depositing titaniumdiisopropoxidebis(acetylacetonate)/TiO_x via spin-coating which was followed by nanoporous TiO₂ (P25) deposition via doctor blade method onto FTO glass substrate. Then, this layer was annealed at 450°C for 30 minutes and soaked in N719 dye for 18 hours to fabricate the DSSC electrode. Secondly, for Si solar cell preparation, the 80nm-thick ITO was sputtered on a pn junction Si wafer. Thereafter, Al electrode was deposited on the back surface of Si wafer by thermal evaporation method. Also, PEDOT:PSS (PH1000) layer with dimethylsulfoxide (DMSO) addition was coated on ITO by spin-coating. Finally, the performance of DSSC/Si tandem solar cell was measured by AM 1.5 solar simulator after an iodine electrolyte injection.
As a Si solar cell, a pn junction Si wafer (pn-Si) and a pn junction Si wafer with Ag grid and a passivation layer (Ag-pn-Si) were prepared. When the only PEDOT:PSS layer was deposited between the top cell and the pn-Si bottom cell, the open-circuit voltage (V_oc) of tandem cell was more than 1 V, whereas the V_oc of DSSC and Si solar cell before stacking were 0.6-0.7 V and 0.5 V, respectively. On the other hand, if Ag-pn-Si is introduced as a bottom cell, the device stability became poor since it is difficult to protect the electrode corrosion due to the iodine electrolyte usage. Although the only deposited ITO layer in-between the top cell and bottom cell could greatly reduce the V_oc up to 0.2 V, interestingly, the introduction of ITO layer under PEDOT:PSS has succeeded in electrode protection from corrosion which has improved the photovoltaic properties. Based on these results, it is noted that the improvement of buffer layer in the DSSC/Si tandem solar cell can increase the conversion efficiency of photovoltaic device. Furthermore, PEDOT:PSS film conditions on the photovoltaic properties of DSSC/Si tandem solar cell will be discussed later.

In the case of the shingled solar module, the power decrease according to the shading is large because of the reduction of the cell area due to the division. Therefore, it is necessary to develop a simulation method with high accuracy according to shading ratio and type to analyze the power loss and to design a module structure that shows stable power even when shading.
Based on the measured photo I-V data, the cell strips circuit is modeled using a double diode model, and the shingled string circuit is modeled by serial connection of the electrical equivalent circuits of the cell strips according to the number of interconnection. Since the cell strips are bonded by a conductive adhesive, the cell strip circuits are connected in series via the RECA corresponding to the ECA resistance component. Next, circuit modeling for shading loss analysis of shingled strings is performed. Finally, the accuracy of shading loss analysis circuit modeling is verified by comparing simulation results with actual measured values.
Based on the accurate shading loss analysis method, the shading loss according to the module structure is analyzed by applying the bypass diode according to the serial, parallel, and parallel / parallel arrangement of the string. We then design the module structure with low shading loss by calculating the power rate in shading.

New technologies for fabricating high-power modules include a cell dividing and bonding technique. This technique divides and joins cells into a string arranged in series and parallel to produce a module. Since the bus bar structure is not visible, the effective area for absorbing light increases, and since there is no space between cells, modules with high density can be manufactured. Therefore, modules with higher output than existing modules can be manufactured in the same area. The module manufactured by the dividing and bonding technique is called a shingled PV module, and this module requires a new electrode structure for dividing and bonding. Therefore, we designed the front electrode structure suitable for the shingled PV module by calculating the power loss according to the number of cell divisions and the number of fingers. Simulation results show that the number of fingers at maximum efficiency decreases as the number of divisions increases. This is because as the number of divisions increases, the distance between the busbars becomes narrow, so that the current collection of one finger between the bus bars becomes smaller. Therefore, in order to collect the same amount of current, the interval between the fingers must be widened, which means a decrease in the number of fingers. Therefore, as the number of divisions increases, the maximum efficiency is obtained in the case where the number of fingers is small. In addition, the actual solar cell of the electrode pattern having the optimum number of fingers for each type of divided cell is manufactured and analyzed. In order to investigate the applicability of the solar cell with the designed electrode structure to the shingled PV module, we investigated the characteristics change after bonding the cells divided by the laser with the conductive adhesive.

Silicon carbide (SiC) has attracted considerable attention for photonic and electronic device applications due to inherent electrical and high thermal conductivity, indirect wide band gap, large critical breakdown electric field, high saturation electron drift velocity and high chemical stability in the use of metal-semiconductor (MS) Schottky barrier diodes (SBDs) with and without interfacial layer. It has also superior characteristics that present alternatives to the inherent problems in the use of oxide and carbon based catalysts. In this study, a novel metal-insulator-semiconductor (MIS) diode including a silicon nitride (Si₃N₄) layer was investigated to improve SiC based MS diode. Although there are many reported work on the electrical characteristics of this kind of wide band gap semiconductor diodes under the effect of insulator layer, it is still an open research area to develop a complete understanding on contribution to the electronic characteristics. Recently, high dielectric constant materials have attracted considerable attention as an alternative interfacial layers for their photonic and electronic device applications at metal/semiconductor interface. Under this aim, the current-voltage, capacitance-voltage and conductance-voltage characteristics of MIS 4H-SiC diodes were analyzed and compared to the MS diodes. The MS heterojunction diode was fabricated by thermal evaporation of Au metal onto the n-4H SiC wafer substrate and Si₃N₄ layer was deposited into the metal/semiconductor interface by magnetron sputtering method. Forward and reverse biasing behavior of diodes were discussed according to thermionic emission with Gaussian distribution of barrier height and Schottky emission models, respectively. Device parameters as, barrier height, ideality factor, interface states, and series resistance were also calculated comparatively among these two diodes. The capacitance values for MS diode was found in decreasing behavior from ideality with crossing the certain forward bias voltage point whereas rapid increase in conductance values with the increasing voltage was observed for all fabricated diodes. Under all these works, high device performance was considerably observed from the Au/n-4H SiC diodes with Si₃N₄ interfacial insulator layer.

The electronically-coupled upconverter (ECUC) is a novel concept for harvesting sub-band gap photons in silicon solar cells through the impurity photovoltaic effect [1]. Unlike conventional intermediate band approaches [2] in which mid-gap defects/impurities are introduced to the bulk of the device, this approach utilises a wider-band gap upconverter layer (e.g. a-Si) situated at the rear of a conventional Si solar cell in which the defects are contained. The band offset between the upconverter and the Si-bulk sweeps sub-band gap photocarriers generated in the upconverter into the base whilst isolating carriers generated in the base from entering the upconverter, allowing a net gain in carriers.
While a-Si ECUC layers containing ion-implantation-induced defects have been shown to absorb at sub-band gap wavelengths [3], it remains to be seen whether sub-band gap photocarriers can indeed be generated. In this work, we use a range of different implantation and post-implantation annealing conditions to control the type and concentration of defects in the a-Si. We investigate the generation of carriers in a-Si ECUC layers by measuring the band-to-band photoluminescence from both the Si substrate and the a-Si layer itself, under sub-band gap laser excitation of a few kW/cm² . We also use time-resolved photoluminescence to study the recombination dynamics within the a-Si layer. Finally, we explore the possibility of incorporating other impurities into the ECUC layer to enhance its sub-band gap photoresponse.

References:
[1] N.-P. Harder and D. Macdonald, “Electronic up-conversion: a combination of the advantages of impurity photovoltaics and (optical) up-conversion,” 2005.
[2] A. Luque and A. Martí, “Increasing the Efficiency of Ideal Solar Cells by Photon Induced Transitions at Intermediate Levels,” Phys. Rev. Lett., vol. 78, no. 26, pp. 5014–5017, 1997.
[3] D. MacDonald, K. McLean, P. N. K. Deenapanray, S. De Wolf, and J. Schmidt, “Electronically-coupled up-conversion: An alternative approach to impurity photovoltaics in crystalline silicon,” Semicond. Sci. Technol., 2008.

Recently facrication of heterostructure membranes using sacrificial layers has been widely investigeted as it can be used to understand the effect of interfacial strain inside the bottom-up heterostructures. In addition, similar to 2D materials like graphene, a transfer method of a thin hetetostructure membrane into other devices is simulating industrial interest. However, most of the previous works report only limited size of the membranes. In this study, we show a cost-effective way to fabricate large scale poly-Si membranes grown ona water-soluble Sr₃Al₂O₆ (SAO) sacrificial buffer layer. SrCO₃ and α-Al₂O₃ are mixed at a molar ratio of 3:1 and calcined at 1200°C to synthesize SAO. The SAO slurry is prepared by mixing the synthesized SAO powder with PVB and ethanol, and a 10 μm-thick SAO layer is deposited on a Si wafer with a micrometer film applicator and then sintered 650°C. After the deposition of Si on top of Si on top of the SAO layer using various deposition method, the dissolution of SAO into water. XRD analysis reveals that poly-Si membranes are obtained. In order to explore thr possible application to solar cells, photovoltaic properties of the poly-Si membranes such as light absorption, current density, open circuit voltage and minority carrier were examined

Al₂O₃ is one of the widely investigated materials for preventing moisture permeation of electronic devices. However, a phase transition occurs when it is exposed to a harsh environment making a channel inside through which water can permeate. In this work, a silane-based polymer (Silamer) layer is introduced to solve this problem as Al₂O₃ is deposited on a back sheet (BS) of silicon solar cells. Al₂O₃ was deposited at 70 °C by atomic layer deposition (ALD) and Silamer was spin-coated followed by curing in low vaccum at the same temperature. During this process, the Silamer layer forms a three-dimensional SiO₂-embedded organic layer and planarizes the BS covering pin-hole defects, which reduces the moisture permeation. By adjusting the thickness and the structure of the deposited layers, Silamer (400 nm) / Al₂O₃ (40 nm) / BS is found to be the most effective for waterproofing with the lowest WVTR value through Ca tests. As it is applied to silicon solar cells, this organic/inorganic hybrid coating shows to reduce potential-induced degradation effectively which is mostly caused by ion migration through the channels formed by the moisture permeation. In addition, alternative deposition methods of Al₂O₃ other than ALD for scaling up will be discussed.

Silicon (Si) based solar energy conversion systems have a large variety of applications, including photovoltaics, photodiodes and photodetectors. The conversion efficiency depends on the absorption of incoming solar energy. Unfortunately, photoresponse of Si is limited by its band gap (1.12 eV) that does not allow absorption below visible wavelength of sunlight. For effective solar energy conversion to electrical signal, enhanced photoresponse has been achieved by doping silicon beyond its solid solubility limit. The highly non-equilibrium supersaturation process is known as hyperdoping that includes ion implantation followed by pulsed laser melting and rapid resolidification. The high concentration of dopants modifies the optoelectronic properties of Si by forming defect levels within the forbidden band gap, but it comes with a price of metastability. Most recently gold (Au) has gained renewed interest as an impurity in Si that is intentionally introduced to extend optical absorption towards infrared region. However, experimental evidence showed that upon thermal relaxation, the absorption suddenly drops down. This suggests that the solubility and stability of dopants in Si are equally important as infrared absorption. Therefore, a systematic study is essential to identify possible dopants that have either high solubility or low diffusivity so that it becomes kinetically trapped during thermal relaxation and preserves the high optical absorption. In this work, we use first-principles density functional theory (DFT) to obtain the possible defect configurations of candidate dopants in Si. Primarily late 3d transition metals are chosen as they provide defect levels in the middle of the band gap of Si and are promised to provide infrared absorption. Among them, manganese (Mn), cobalt (Co) and silver (Ag) are chosen from optical point of view. On the other hand, from the solubility point of view, nickel (Ni) and copper (Cu) shows better solubility (1 and 2 orders of magnitude higher) than Au, which are group IB materials. Further, we analyze the optical absorption of possible equilibrium defect configurations in conjunction with their solubility. Finally, we discuss how diffusivity can play a role during thermal relaxation and consequently on optical absorption.

As the PV industry strives toward terawatt scale, addressing the last remaining cost centers of the crystalline silicon value chain is essential to achieving the extremely low levelized cost of electricity (LCOE) required to drive solar's adoption. To date, improvements within the value chain have focused primarily on cost reduction, with most technical contributions happening on the surface of the material, at the cell level. New fabrication methods have emerged that change this dynamic and allow for significant strides in both cost reduction and technical advancements with the absorber materials themselves.

Alternative absorber fabrication methods such as the Direct Wafer process are synergistic with lower cost and higher absorption, as well as more efficient energy usage. In Direct Wafer manufacturing, which has reached high-volume production, wafers are grown one-by-one directly from a small reservoir of molten silicon in a high-purity growth environment. Direct access to the wafer-forming surface and the molten silicon reservoir allow modifications of individual wafer properties, customized for a given cell architecture. Efficiency parity with high-performance cast multi-crystalline wafers has been demonstrated on a PERC production line at 20.5% average efficiency. This discussion will also demonstrate that Direct Wafer technology is synergistic for use with future cell architectures such as TOPCon that are insensitive to bulk resistivity and require high minority carrier lifetime or device architectures such as tandem which necessitate a low-cost base material with bandgap suited for long wavelength absorption.

The high-purity growth environment of the Direct Wafer process limits the introduction of dopant impurities into the feedstock and allows high bulk resistivity beyond 100 Ω-cm. Since wafers are also grown one-by-one, bulk resistivity can also be easily tuned and targeted around virtually any resistivity. This allows for a much tighter distribution than what is possible in ingot-based wafers where resistivity distribution is governed by solid-liquid segregation leading to a wide range of resistivities along the ingot height.

The low impurity concentration in Direct Wafer product is also reflected in the high effective lifetime that can be achieved. Electron lifetime in Direct Wafer product increases with increasing resistivity and effective lifetime >1 ms have been validated. Advanced cell architectures like PERT and TOPCon cells require lifetimes of milliseconds to achieve their efficiency entitlement and leverage potential LCOE advantages despite increased cell production cost. Currently, mostly high-cost n-type mono-crystalline wafers are used for advanced cell architectures because of higher lifetime in n-type silicon. With Direct Wafer product, there is considerable potential to accelerate the adoption of higher efficiency device architectures as low-cost, high-resistivity Direct Wafer product replaces the more expensive n-type mono wafer.

Advanced, low-cost wafer manufacturing also provides a course to cost effective tandem cell fabrication. The technical roadmap for the Direct Wafer process provides for a cost of just $0.12 per wafer through a combination of ~1.5g/W silicon utilization and lower non-silicon conversion costs than current ingot-sawn wafers. This base, combined with low-cost cell processing tailored for long wavelength absorption at lower current, makes an economically favorable candidate for realizing commercial tandem modules that exceeds the performance of today’s best single-junction silicon modules at a lower total cost/watt.

An inherent high-purity growth environment and unique features distinguish Direct Wafer technology from standard sawn wafers. The benefits of a more efficient approach to fabrication coupled with the ability to access the raw material at the melt level present an opportunity to leverage low-cost wafer technology with numerous future advanced cell architectures.

Currently, the photovoltaic (PV) market is dominated by boron-doped, or p-type, silicon (Si) solar cells. More specifically, the Si PV market is moving from a more traditional architecture, the aluminum back surface field (Al-BSF) solar cell to a more advance structure, the passivated emitter rear contact (PERC) solar cell. The PERC cell features exceptional surface passivation and therefore, higher efficiencies compared to the Al-BSF. However, with the surface no longer limiting the PERC cell, bulk lifetime becomes increasingly more important to raise efficiencies. One source of recombination within the PERC Si bulk is light and elevated temperature induced degradation (LeTID) where cells and modules experience a decline in minority carrier lifetime when exposed to illumination at elevated temperatures. LeTID was discovered only ~5 years ago and therefore, the degradation and regeneration mechanism, along with the defect structure, are unknown. A preliminary solution to LeTID has been discovered [1] but requires intense illumination and high temperatures and can take ~15 hours to complete. In order for p-PERC cells to continue to dominate the PV market, we must ensure their efficiencies rise and remain stable throughout their time in the field.

LeTID can result in a ~10% relative efficiency loss of p-PERC solar modules over their lifetime in the field [2]. The defect which causes LeTID is thought to be hydrogen, and more specifically, related to the over hydrogenation of cells during the firing process to create contacts. Many empirical studies have been conducted to link hydrogen to LeTID, but no atomistic studies have been brought forward to solidify this hypothesis. In this work, we aim to study the role of hydrogen in LeTID by intentionally inducing defects in boron-doped float zone (p-FZ) silicon (Si) and therefore, manipulating the amount of hydrogen that is held in the Si bulk after fast firing. We hypothesize that by allowing the Si bulk to hold more hydrogen, samples will experience increased degradation caused by LeTID compared to samples without intentionally induced defects.

References
[1] A. Herguth, C. Derricks, P. Keller, B. Terheiden, Energy Procedia, 124, 740-744, (2017)
[2] F. Kersten, P. Englehart, H. Ploigt, A. Stekolnikov, T. Linder, F. Stenzel, M. Bartzsch, A. Szpeth, K. Petter, J. Heitmann, J. Muller, Solar Energy Materials and Solar Cells, 142, 83-86, (2015)

Solar energy is the most abundant renewable energy source that can be harvested by various approaches such as photovoltaics (PV), photo-thermal heating, photo-thermoelectricity, and artificial photo-synthesis. Among these technologies, silicon (Si)-based solar cells have been developed and successfully commercialized due to the abundance of Si materials on earth. In Si-based solar cells, the separation of photo-excited electron-hole (e-h) pairs is limited by the built-in electric field (E, ~10⁴ V/m) of p-n junctions. Tribo-induced potential at a metal-insulator-semiconductor (MIS) sliding junction results in very high interfacial electric field E (~10⁷-10⁸ V/m). Such fields can be used to separate photo-generated electron-hole pairs that transport through the junction via quantum tunneling. We show that this phenomenon can generate direct-current (d.c.) with high current density J. The strong synergetic effect between the photo-excitation and the interfacial electronic excitation at the MIS sliding contact boosts the photocurrent output in Si-based Schottky solar cell by ~50 times and the triboelectric power output by ~28 times. Charge relaxation dynamics calculations agrees well with conductive-atomic force microscope (C-AFM) results and show the coupling effect between the photo- and the tribo-induced electric fields. Experimental observation of the tribo-photovoltaic effect provides new fundamental understandings of electromechanical coupling, and thus offers a new direction for future solar-mechanical energy co-harvesting.

Semi-transparent solar cells (STSCs) are an attractive energy conversion device because it can be used in various applications in our daily life such as building-integrated photovoltaics, and vehicle-integrated photovoltaics. Currently, organic photovoltaics (OPV), and dye-sensitized solar cells (DSSC) based STSCs have been developed so far. However, the developed STSCs up to now have limitations in power conversion efficiency (PCE), and stability. Furthermore, they have shown specific colors owing to the colors of polymers or dyes. To address this issue, some researchers have been developed near infrared (NIR)-absorbing STSCs to fabricate neutral-color STSCs like that of glass. However, NIR-absorbing STSCs have been shown very low PCE under 1% until now. Accordingly, a different approach is required to develop STSCs with neutral color, high PCE, and long-term stability.
As a method to develop new-concept neutral-color STSCs with high PCE and stability, crystalline silicon (c-Si) based STSCs can be considered. c-Si would be one of the best candidates to develop STSCs because conventional c-Si solar cells are known to exhibit high PCE and long-term stability compared to other solar cells. However, the development of c-Si based STSCs is extremely challenging due to the opaque characteristic of c-Si wafers with thicknesses of 200 µm. To fabricate STSCs using c-Si, thinning of c-Si has been considered as a method to increase transparency. However, as the absorption of long-wavelength light is extremely limited in thin films, their application to solar cells is not suitable. In addition, a thin c-Si film has a particular color because of absorption spectrum cutoffs in the visible light wavelength range.
In this presentation, we show a novel approach to develop neutral-color transparent c-Si solar cells. First of all, we have developed a neutral-color transparent c-Si substrate using a 200-µm thick c-Si wafer, which is known to be opaque. Transparent c-Si substrates were fabricated by placing hole-shaped light transmission windows on a bare c-Si wafer. These windows were designed to enable the transmission of all incident visible light through the substrate, resulting in a colorless substrate. In addition, the spacing between the holes was appropriately selected by considering the minimum angle of resolution for humans to ensure that the individual transmission windows are not visible to the human eye. A light absorption area was also designed on the substrate to efficiently absorb incident light of the spectral range between 300 nm and 1100 nm. The transmittance of the transparent c-Si substrate was systematically tuned from 20% to 50 % under the full solar spectrum. The STSCs fabricated with the substrate exhibit a PCE of up to 12.2%, with J_sc = 29.2 mA/cm²; V_oc = 588 mV; and FF = 71.1% with a transmittance of 20 %. The cell performance of the transparent c-Si solar cells is higher than those of other neutral-color semi-transparent solar cells reported thus far. Hence, our novel c-Si STSCs presents a unique opportunity to develop next-generation neutral-color STSCs which would satisfy high efficiency as well as high stability. Furthermore, we believe that this study makes a significant contribution to not only photovoltaics fields but also various transparent optoelectronics.

The route to the highest performance and industry-relevant photovoltaic cell and module is getting narrow and tortious, an inspiring technology challenge. One of the most promising path on this quest is the combination of single silicon heterojunction interdigitated back-contact (SHJ-IBC) ells and innovative 3D woven interconnection technology. Furthermore, we demonstrate that stacking IBC cells in a 4-terminal (4T) tandem with perovskite top cells based on CsFAPbIBr passivated by atomic layer deposited Al₂O₃ can lead to a record efficiency of 27.1% at cell level, and 25.3% efficiency for mini-modules (higher values are expected by the conference).

First, we present for SHJ-IBC devices a litho-free and simplified process flow using laser ablation of a distributed Bragg reflector hard mask composed of bilayers of SiO_x/SiN_x for patterning the different contact polarities on the rear. The consecutive partial dry etching step enables an in-situ switch from hole to electron contact by the removal of the superficial p+ a-Si:H layer and its replacement with n+ a-Si:H layer. An additional advantage of this partial etch process is the protection of the passivation layer during the patterning, which eliminates a wet cleaning step and an additional deposition of the intrinsic a-Si:H layer. We report for this simplified process flow the best device with 22.7% efficiency (22.4% average efficiency and area of 2x2 cm₂) and a V_OC of 728 mV demonstrating high-quality passivation.

Second, we discuss the challenges of interconnecting IBC cells, notably connecting opposite cell polarities in the same plane with selective isolation and the necessity to use low-temperature processing. To develop an industrially relevant technology, low-cost and simplicity are essential but unmet requirements by current solutions. To overcome this challenge, we introduce the 3D woven interconnection concept based on standard module materials. Polymer encapsulant is interwoven with low-temperature solder-coated metal ribbons to form a three-dimensional woven fabric. The encapsulant provides the electrical insulation between ribbons of opposite polarity. Part of the ribbons electrically interconnect neighboring cells while ribbons in a perpendicular direction connect the cell metallization to the cell-to-cell ribbon, hence creating a two-level metallization. Additionally, the co-development of the materials and the lamination process enables a combined soldering and module lamination in one process step and within the limits of industrial laminators. We will demonstrate the adaptation of this technology to bifacial IBC cells as well as to busbarless cell design. We report on the properties of the interconnect in single-cell and mini-modules demonstrating < rel. 2% FF loss, Rsh> 7MΩ in isolation tests and showcase its stability in over 400 thermal cycles.

Third, we present our latest development in perovskite device stacks focusing on Cs_0.15(CH₅N₂)_0.85Pb(I_0.71Br_0.29)₃ (CsFAPbIBr) as a photo-stable mixed-halide with a wide bandgap (1.72 eV). We show that the key to achieving high V_OC is the careful control of perovskite-hole transport layer interface. Through effective interface passivation, we fabricate 1.72 eV mixed-halide perovskite solar cells with a V_OC deficit as low as 0.5 V. We demonstrate in single junction semi-transparent perovskite devices with efficiencies of 11.7% without passivation and 13.8% with Al₂O₃passivation layer in small area devices. This device also maintains an average transmittance of 90% in the wavelength range of 700 – 1200 nm making it an ideal top cell in 2 or 4-terminal tandem devices, which consequently contributes to a record 27.1% efficient perovskite/IBC c-Si tandem cell (area: 0.13 cm²), and 25.3% efficiency for 4 cm² perovskite minimodule on 4 cm² c-Si IBC cell. In the end, we will also report on upscaling of perovskite modules up to 30x30 cm² area.

The cost of photovoltaic electricity hinges on the metallization of the Si solar cells, which dominate today’s commercial cells. In a bid to further reduce the cost of the metallization, one of the feasible methods is to use cheaper metals to replace the dominant screen-printed Ag metal contacts, which account for the most cost of the metallization. Among all the metals, Cu is considered as an excellent potential alternative due to its similar conductivity (1.7 μΩ-cm for copper, 1.6 μΩ-cm for silver) but ~90 times less cost. However, Cu is known to diffuse into Si at high temperature, leading to shunt the junction and reduce the minority carrier lifetime in the cell. Thus, researchers have used plated Ni/Cu in which Ni acts as a barrier through NiSi formation to block Cu from diffusing into Si. But, because of the involving processing steps and reliability issues, this technology is not adopted. Since screen-printing technology is high-throughput and simple to manufacture solar cells, the most cost-effective method is, therefore, replacing the Ag or part of the Ag in the paste by Cu. But this method puts a strict requirement to the nature of the paste, in particular, that prevents the Cu from diffusing into Si during the high-temperature sintering of the contacts. Thus, it is important to investigate the microstructural (SEM, TEM), optical (Raman Spectroscopy) and electrical (conducting AFM) properties of the Ag/Cu/Si interface through one-step atmospherical firing process and thus to understand the contact mechanism of the Ag/Cu paste. This work, therefore, reports on the microstructural analyses (SEM, TEM) of the Ag/Cu/Si interface, supported by the Raman Spectroscopy and the conductive AFM. The preliminary studies showed that there was no evidence of Cu diffusion into the Si and Cu was found only in the reformed glass after high-temperature sintering. The conductive AFM revealed the contact mechanism was through the Ag crystallites grown in the Si. The electrical output parameters of the solar cell corroborated the microstructural, optical and electrical analyses.

Passivated emitter and rear contact (PERC) silicon photovoltaic cells comprise over 50% of all solar cells manufactured in 2019, and are expected to completely replace traditional aluminum back surface field (Al-BSF) cells by 2023. In addition to higher power yield due to reduced recombination at the rear contact, PERC cells can easily be manufactured with monofacial or bifacial architectures. However, the additional passivating dielectric layer in PERC cells, as well as the reduced rear-side metallization in bifacial cells, calls into question the long-term durability of this technology versus the established Al-BSF cell.
Here we compare the performance of monofacial and bifacial PERC cells, benchmarking against Al-BSF, both as bare cells, and in minimodules through damp heat exposure. Cells are characterized with illuminated I-V curve tracing, Suns-V_oc, and electroluminescence, as well as photoluminescence and spatially resolved external quantum efficiency. Cell fragments are also investigated with photoconductive atomic force microscopy.
In this work, we show that PERC cells have greater stability through damp heat exposure than Al-BSF. Bifacial PERC cells also show surprisingly high mechanical durability compared with full rear side aluminized cells. Al-BSF cells show greater corrosion and resulting resistive losses than PERC cells when encapsulated with EVA. Performance loss due to additional degradation modes are also quantified. Comparison of electroluminescence and photoluminescence images of mini-modules, as well as external quantum efficiency measurements, reveal the underlying degradation mechanisms for each cell type. These results are confirmed with photoconductive atomic force microscopy, with which electrical properties of the cell including V_oc and I_sc are mapped cross-sectionally to study operational changes of the degraded cell.

A laser-fired contact (LFC) process is one of the techniques for making local electrical contacts at the rear side of PERC (passivated emitter and rear cell) solar cells. In the LFC process, opening of the passivated dielectric layers and alloying of Si and Al need to be made in a single step laser process. For this reason, the LFC process is accompanied by the loss of Al and the laser damage to the Si wafer. In this study, we present a novel multi-step LFC process combining the conventional LFC and laser induced forward transfer (LIFT) processes. The modified LFC scheme we proposed consists of three steps: (1) opening of the passivation layers and partial alloying of Al-Si, (2) additional deposition of Al on the local contact holes, (3) post laser firing of the transferred Al. Applying the modified LFC process to the PERC cells, we demonstrate the effective recombination velocity of the laser processed wafers can be remarkably reduced while maintaining the low contact resistance. The best of the PERC solar cell with the modified LFC process exhibited 20.5 % of an efficiency while the conventional LFC cell showed 18.6 %. The efficiency gains of the modified LFC PERC cells was largely contributed by the enhanced open circuit voltage (V_oc) and fill factor (FF).

Significant growth of n-type solar cell with passivated contact technology is expected in the coming few years. However, whether it will become the mainstream technology to replace the dominant position of p-type in the market will be subject to the total cost of ownership improvement throughout the whole value chain engagement. Today p-PERC cell efficiency is approaching 22.5%; n-type needs to demonstrate >23% efficiency with good yield and cost to be competitive. Passivated contact is a key technology for n-type in the drive for higher efficiency, however, metallization pastes specially developed for passivated contact is needed to achieve performance and cost competitiveness in mass production. DuPont^TM, a leading innovator in metallization technology for Si solar cells, is developing a total package metallization solution for n-type passivated contact structure to help our customers stay ahead in this next wave of high efficiency technology upgrade after PERC. A new generation Solamet® PV3Nx with improved performance on contacting ultra-lightly doped boron emitter and an upgraded PVD2x busbar paste were developed to be suitable for both double printing and dual printing applications. In addition, an innovative formulation was developed to minimize the recombination loss on thin polycrystalline silicon film for rear side metallization. Overall, we believe innovations in metallization technology will lead to a potential efficiency of >23% for n-type in mass production.

Preparing chemical fuels from natural resources by using renewable energies and thereby replacing the fossil fuel exploitation with all its detrimental consequences is and will be a major research activity and challenge now and in the near future.

One approach which already receives considerable attention is the generation of hydrogen from water via electrolysis using solar cells as a power source. We have been concerned with this concept: an integrated photovoltaic (PV) - electrochemical (EC) device feasible for stand-alone operation under sunlight illumination, yielding high solar-to-hydrogen (STH) conversion efficiencies, using preferably natural abundant and non-toxic materials and having an up-scalable technology.

As solar cell material we have chosen thin film silicon (hydrogenated amorphous and microcrystalline silicon). The thin film silicon approach has the unique features of i) an easy vertical and lateral integration of multi-junction cells which allows for tuning of the required output voltage while making very efficient use of the solar spectrum, ii) an already established scalable production technology, and iii) earth-abundant and non-toxic photovoltaic base materials.

Alternatively, we have investigated series connected crystalline silicon wafer based solar cells.

We will show and compare the performance of PV-EC coupled systems with STH conversion efficiencies of close to 10% and 14% for thin film and crystalline silicon solar cells, respectively, and demonstrate the development and up-scaling of related PV-EC cassette systems including the preparation of earth abundant based catalyst materials.

The results can be used to discuss the concept of a closely coupled PV-EC system vs. “wired” systems where a photovoltaic module is connected to a possibly locally remote electrolyzer, and to estimate annual hydrogen production for typical outdoor illumination conditions.

Development of high-efficiency solar cell modules and new application fields are significant for further development of PV (photovoltaics) and the creation of new clean energy infrastructure based on PV. Especially, Si tandem solar cells such as III-V/Si, chalcogenide/Si, and perovskite/Si tandem cells are desirable for high-efficiency and low-cost cells. We surveyed the progress of heterogeneous Si tandem solar cells and analyzed the prospects of their conversion efficiencies by estimating ERE (external radiative efficiency) as a measure of non-radiative recombination loss. Although 35.9 % efficiency under 1-sun has been demonstrated with InGaP/GaAs/Si 3-junction tandem cell, 41% efficiency will be realized with the 3-junction Si tandem solar cells by improving ERE from 3% to 20%. Regarding 2-junction Si tandem cells, 32.8% and 28% have been achieved with GaAs/Si and perovskite/Si tandem cells, respectively. The 2-junction Si tandem cells have an efficiency potential of 35% by realizing 20% ERE. The authors have demonstrated 28.2% with mechanically stacked InGaP/GaAs/Si tandem solar cell. Most recently, Sharp has also achieved 33.0% efficiency with 3.604cm² InGaP/GaAs/Si 3-junction solar cells by using the mechanical stack.
High-efficiency and low-cost solar cells such as Si tandem cells have great potential for PV as a vital energy source in mobility application where the installation area is limited. The NEDO established a “PV-Powered Vehicle Strategy Committee” for investigating the potential for contributing to reducing CO₂ emissions in the transport sector by installing high-efficiency solar cells on automobiles. According to the IRENA’s and NEDO’s reports, new broader PV market with more than 10GW and 50GW in 2030 and 2050, respectively are expected to be established. Cumulative PV capacity for PV-powered vehicles will be 50GW and 0.4TW in 2030 and 2050, respectively. According to statistics by the Ministry of Japan, 2/3 of the family car runs less than 30 km per day in Japan. Namely, the average annual energy that is needed to the light-weight family car powered by sunlight will be 642 kWh/year. 642 kWh/year is not an incredible value as well as a promising because the use of more than 30 % of high-efficiency PV enables the society that majority of the family cars run by the sun and without supplying fuel. Thus, we are developing high-efficiency and low-cost solar cells and modules for automobile applications.
Approaches on automobile application by using III-V/Si tandem, partial concentration, and static low concentration and so forth are also presented. Because of the space limitation, it should be a high-efficiency panel. III-V/Si tandem cells are one candidate. The CPV (concentrator PV) is very attractive for saving the cost of the cell. Cars move quickly, and appearance is essential. Trackers were thought to challenge to implement. One of our choices is a static concentrator customized to the automobile. In addition to high-efficiency, low-cost, 3-dimensional curved surface, color variation and so forth are needed and thus further R&D of solar cells, and modules for automobile applications are necessary.
Although III-V cells have an extremely high conversion efficiency of up to 46.0%, which is suitable for this application, cost reduction is necessary to realize the concept. The Si tandem solar cells by combining Si and other materials such as III-V compound, II-VI compound, chalcopyrite, perovskite, and so forth are desirable for realizing super high-efficiency and low cost. Recently, Si tandem solar cells have paid considerable attention because of high-efficiency and low-cost potential.
New static low concentrators were proposed for the vehicle application. Besides, a III–V/Si partial CPV module with high diffuse sunlight transmission is studied and low concentration static III-V/Si partial CPV module with 27.3% annual efficiency for car-roof application has been develpoed. .

Advancements in photovoltaic technologies are hindered by degradation mechanisms such as potential induced degradation (PID) and current-induced degradation (CID). In this work, impedance spectroscopy is used to examine passivated emitter and rear cell (PERC) silicon modules with PID and CID. A comparison between control and degraded modules is done to identify key differences in the impedance spectra and determine the extent of the degradation. CID was also examined at the cell level, where reductions in minority carrier lifetimes could be more accurately measured. The PID and CID mechanisms studied are found to induce unique changes in the impedance spectroscopy results, making them distinguishable and quantifiable. Finally, the ability to mitigate CID through the use of different silicon wafers and a current induced regeneration process was characterized by impedance spectroscopy.

Tandem solar cells consisting of Si and III-V compounds are one of the promising candidates for next-generation terrestrial photovoltaic systems. Surface-activated bonding (SAB) at room temperature (RT), in which surfaces of substrates are activated before bonding by creating dangling bonds under an energetic particle bombardment in a high vacuum, is applied to form Si/GaAs heterointerfaces with a low interface electrical resistance [1], and high-efficiency InGaP/GaAs/Si triple-junction cells are demonstrated [2]. Even though the interface electrical resistance (~10^-1 Ωcm²) is low enough for solar cells, it is still higher than the ideal one at defect-free heterointerfaces (~10^-4 Ωcm²), presumably due to the defects introduced during SAB processes. The interface resistance varies depending on SAB conditions including energetic atom irradiation and post-bonding annealing [1]. In order to understand the origin of the resistance, atomistic structure of the hetero-interface, depending on SAB conditions, have been examined by cross-sectional transmission electron microscopy (X-TEM).
In general, X-TEM specimens with heterointerfaces, in which the bonding materials are different in etching rate, are fabricated with milling techniques using energetic ions such as focused ion beam (FIB). We have clarified that the structural and compositional properties of semiconductor homointerfaces fabricated by SAB are modified during FIB processes operated at RT, especially for wide-gap materials, and such a modification can be suppressed by FIB processes operated at -150 ^oC [3]. In the present work, we have therefore examined the atomic arrangement and composition at Si/GaAs heterointerfaces fabricated by SAB using X-TEM specimens fabricated by FIB milling operated at -150 ^oC.
Si/GaAs heterointerfaces were fabricated at RT under a SAB condition [4], with the substrates of B-doped (100) p-Si (with a carrier concentration of 2x10¹⁴ cm^-3) and Si-doped (100) n-GaAs (2x10¹⁶ cm^-3). X-TEM specimens with an as-bonded heterointerface were prepared at -150 ^oC by using a FIB system (FEI, Helios NanoLab600i) with a cold stage customized for the FIB system (IZUMI-TECH, IZU-TSCS004) [3]. The specimens were examined by high-angle annular dark-field (HAADF) and energy dispersive x-ray spectroscopy (EDX) analyses under scanning TEM (STEM) with a JEOL JEM-ARM200F analytical microscope.
HAADF-STEM-EDX revealed that, an amorphous layer, about 4 nm in thickness, is introduced along the as-bonded heterointerface. Most of the amorphous layer are composed of Si, and no amorphous GaAs is apparently observed, as proposed [4]. Atomic intermixing across the heterointerface, within a range of a few nm, is observed. Both Ga and As atoms penetrate into the amorphous Si layer, and the amount of Ga atoms in the layer is about two times larger in comparison with As atoms. Those excess Ga atoms, acting as p-type dopant, can improve the electrical property of the amorphous p-Si layer, that is damaged during the SAB processes. Similarly, the electrical property of the n-GaAs bonding surface can be improved via the penetration of Si atoms. Those self-restoration processes may assist the formation of low-resistance Si/GaAs heterointerfaces by SAB. We will also discuss the effect of post-bonding annealing.
[1] J. Liang, L. Chai, S. Nishida, M. Morimoto, and N. Shigekawa, Jpn. J. Appl. Phys. 54 (2015) 030211
[2] N. Shigekawa, J. Liang, R. Onitsuka, T. Agui, H. Juso, and T. Takamoto, Jpn. J. Appl. Phys. 54 (2015) 08KE03.
[3] Y. Ohno, H. Yoshida, N. Kamiuchi, R. Aso, S. Takeda, Y. Shimizu, Y. Nagai, J. Liang, and N. Shigekawa, submitted to Jpn. J. Appl. Phys.
[4] Y. Ohno, H. Yoshida, S. Takeda, J. Liang, and N. Shigekawa, Jpn. J. Appl. Phys. 57, 02BA01 (2018).

The silicon heterojunction (SHJ) solar cell is promising for mass production of high efficiency solar cells. To date, the average SHJ solar cell conversion efficiency in mass production is well above 23%. The improvement of transparent conducting oxide (TCO) in SHJ solar cells in terms of transparency and conductivity is one of the key issues to achieve higher conversion efficiencies. The most commonly used TCO is Sn-doped indium oxide (ITO), due to its high transparency and low resistivity. Compared with ITO, W-doped indium oxide (IWO) shows higher charge carrier mobility resulting in high transparency in the near infrared (NIR) spectral range while maintaining low resistivity.¹ The carrier mobility can be further increased when hydrogen is introduced during the deposition leading to hydrogenated IWO:H.² Therefore, IWO and IWO:H could be advantageous candidates for a high-efficiency SHJ solar cells.

Until now, the application of IWO and IWO:H in SHJ solar cells has been studied using soft deposition methods such as reactive plasma deposition (RPD)¹ or radio-frequency (rf) mode sputtering.² But for the mass production of SHJ solar cells, direct current (DC) mode sputtering is commonly used as it offers high throughputs and is an established deposition technology in the optoelectronic industry. To successfully introduce the novel, high mobility TCOs into a SHJ solar cell production line the DC sputtering of IWO and IWO:H needs to be developed and investigated with respect to possible sputtering damages on the thin silicon layers of SHJ solar cells.

Here, we present our most recent work on DC sputtered IWO on SHJ solar cells. In preliminary experiments to this study we have optimized the deposition process for single layers of IWO on glass from Corning type Eagle. The resulting films exhibited high transparency (average absorption of < 2% from 400 - 1100 nm), low sheet resistance (105 Ω/sq), and carrier mobility of up to 55 cm²/Vs. The IWO films were also applied to 19 × 19 mm² SHJ solar cells increasing the short circuit current density J_sc from 38.65 mA/cm² to 39.09 mA/cm² compared to ITO without introducing sputtering damage to the underlying solar cell. In this contribution, we extend the study of DC sputtered IWO further by introducing hydrogen to fabricate IWO:H and also apply the TCOs on full size M2 solar cells. The novel TCO layers deposited by DC sputtering will be compared to ITO with respect to contact resistance and sputtering damage to the underlying silicon layers. To our knowledge, this is the first study of DC sputtered IWO and IWO:H using an industrially relevant technique for SHJ solar cells.

1. Meng, F. et al. Jpn. J. Appl. Phys., 2017, 09, 2–6.
2. Koida, T., Ueno, Y. & Shibata, H. Phys. Status Solidi Appl. Mater. Sci., 2018, 215, 1–14.

What is the value of improving the efficiency of a solar cell? In this contribution we introduce a tool to assess the value of efficiency (VOE) by analyzing the dynamic co-dependence of levelized cost of electricity (LCOE) on manufacturing cost and efficiency. Looking at historic trends we find that the allowable cost for innovation has dropped rapidly in the past and is projected to further decline in the future, but at a slower pace. This is good news for researchers, as it will ease the difficulties of hitting a moving cost target. Additionally, we find that VOE values for residential- and utility scale systems converge at different levels, indicating a stronger future diversification of these PV market segments. Similarly, we find local differences for VOE suggesting a further possible market diversification based on region. Efficiency is a much stronger driver for LCOE reductions in a handful of highly industrialized countries like the U.S., Germany, Australia or Japan than in the rest of the world. Advanced solar cell concepts like n-type monocrystalline PERC silicon solar cells could, hence, be economically preferable to multicrystalline silicon solar cells in these regions, but may not be the better choice in countries like China or India. Extending the analysis to more advanced concepts including tandem solar cells, we confirm previous finding that these concepts have great potential to improve LCOE for residential applications, but will likely struggle to compete on the utility market. Finally, we explore the detrimental impact of degradation, and show that competing on the basis of LCOE necessitates matching the stability of established technologies.

One method to improve silicon solar cell efficiency is to stack a silicon cell beneath a wide-bandgap top cell. To improve the overall tandem cell efficiency from that of silicon alone, the top cell material should have a higher spectral efficiency than silicon for the wavelengths of light within the top cell’s bandgap. GaAs is currently the material with the highest recorded cell efficiency for a single junction cell, at 29.1% efficiency. GaAs also has a theoretical and realized spectral efficiency higher than that of silicon between the wavelengths of 360-860 nm. Our team was able to exploit these GaAs attributes to obtain a record efficiency of 32.8% for a mechanically-stacked rear heterojunction (RHJ) GaAs-on-Si (GaAs//Si) tandem cell designed to operate in the 4-terminal (4T) (i.e. optically coupled but electrically independent) configuration.

The purpose of this study is to further optimize the GaAs emitter layer thickness to maximize the efficiency of the 4T GaAs//Si tandem cells. Our record 4T GaAs//Si tandem solar cell was achieved using a 2 µm GaAs emitter layer but subsequent optical modeling of the 4T GaAs//Si tandem solar cells suggests that a slightly thicker emitter could potentially lead to even better top cell efficiencies. In addition, NREL began a collaboration with ISFH that provides high-efficiency interdigitated back contact (IBC) solar cells that could further improve the tandem cell efficiencies, when used as the Si bottom cell in the 4T GaAs//Si cell.

Our un-certified current-voltage (I-V) results show an efficiency increase of 0.43% (absolute) by increasing the emitter thickness from 2.0 µm to 2.8 µm. The tandem cells are currently pending cell certification but the results for the GaAs//Si tandem cell with an emitter thickness of 2.8 µm is expected to be >32% and could potentially exceed the existing record efficiency. We will present the results of modeling the GaAs emitter thickness using modified Hovel equations that estimates the J_sc, based upon light absorption and carrier recombination. We will compare these modeled predictions to the NREL-certified I-V and external quantum efficiency (EQE) results obtained from 4T GaAs//Si tandem cells fabricated with absorber layer thicknesses varying from 1.5 to 3.5 µm. We will also compare the performance of our 4T GaAs//Si tandem cells to the record 4T GaAs//Si tandem cell efficiency of 32.8%.

High efficiency crystalline Si (c-Si) solar cells can be realized by employing carrier selective contacts to efficiently collect one type of photogenerated carriers by providing excellent passivation performance and low contact resistance at the heterointerface. Atomic layer deposited TiO_x (ALD-TiO_x) is known as one of the promising candidates for electron selective contacts owing to the small conduction band offset (<0.05 eV) and large valence band offset (>2.0 eV), which can selectively transport photogenerated electrons in c-Si while repelling holes. Passivation performance could be modified by both surface treatment of the c-Si substrate prior to deposition and post-deposition process. Therefore, it is important to understand underlying physics during the process to establish the guideline how to improve passivating contacts.
In this contribution, we report on the role of the interlayer in improving passivating contact with ALD-TiO_x/c-Si, and show that formation of low-density SiO_x prior to depositing TiO_x is of crucial importance to improve passivation performance.

All experiments were carried out on double-side mirror-polished float-zone (Fz) grown c-Si(100) substrates. After degreasing and removing a native oxide layer, various chemical treatments were done to form SiO_x interlayer with different densities. TiO_x layers were deposited at 150 ^oC by a thermal ALD (GEMStar-6, Arradiance). The forming gas annealing were carried out in a mixed gas (3% H₂ and 97% Ar) to activate the surface passivation performance. The effective minority carrier lifetime of the samples was measured by a WCT-120 lifetime tester (Sinton Instruments). For a part of samples, EEL spectra were obtained using an EEL spectrometer attached to a Cs-corrected STEM (JEM-ARM200FC, JEOL Ltd.), which was operated at 200 kV. The desorption of hydrogens was characterized by thermal desorption spectroscopy (TDS) using a quadrupole mass spectrometer with different heating rates to analyze Si-H₂ bonding energies.

The effective carrier lifetime of about 1.4 ms was achieved by inserting ultra-thin SiO_x interlayer formed by nitric acid at room temperature. The TEM images and EEL spectra revealed that non-stoichiometric SiO_x existed for as-deposited samples and Ti contained, near-stoichiometric SiO₂ was formed after post deposition annealing. This suggests that diffusion of Ti and O atoms took place during the post deposition annealing. To enhance this process, we considered that introduction of SiO_x interlayer with lower density is effective. As a result, employment of so-called SC-2 solution (HCl:H₂O₂:H₂O=1:1:5) prior to deposition of TiO_x led realization of the effective lifetime of 1.7 ms. The enhanced diffusion of O from TiO_x would also contribute to the reduction of the contact resistance by increasing O vacancies in the ALD-TiO_x layer. Furthremore, higher Si-H₂ bonding energy was confirmed by TDS, suggesting that hydrogens also play a role in improving passivation performance.

An outstanding issue in photovoltaic modules is the accelerated degradation caused by the presence of moisture, which leads to interfacial instability, encapsulant decomposition and corrosion at contacts. Currently, experimental observation and characterization of moisture in PV modules is not trivial, which presents a major obstacle to designing high-reliability modules. First principles calculations provide a suitable way to study the ingress of water and its detrimental effect on the polymer encapsulant and on the interface between the semiconductor and the polymer. In this work, we use density functional theory (DFT) computations to model the structure, degradation mechanisms, and adhesive strength on metal surfaces of ethylene vinyl acetate (EVA), the most common polymer encapsulant used in Si PV modules. The molecular structure of EVA is modeled both as an isolated single chain and in a crystalline arrangement with closely packed chains, using a structure prediction algorithm that includes Van der Waals corrections applied to the standard DFT functional. Infrared active modes computed for the low energy EVA structures using density functional perturbation theory match well with reported experiments. The Nudged Elastic Band (NEB) method is applied to model the decomposition mechanism of EVA, with and without the presence of water, following the Norrish I and Norrish II mechanisms, leading to the formation of acetaldehyde and acetic acid, respectively. Computed energy barriers show a preference for acetic acid formation, are lower in the presence of a water solvent than in vacuum, and match well with reported experimental activation energies. The NEB computed energy barrier is further seen to be dependent on the percentage of deacetylation and the presence or absence of a catalyst, such as a proton or a hydroxyl ion. This systematic study leads to a clear picture of the hydrolysys-driven decomposition of EVA in terms of energetically favorable mechanisms, possible intermediate structures, and IR signatures of reaction products. We further model the adhesion of EVA and related polymers (including decomposition products) on surfaces of metals such as Cu and Ag, in order to determine the sticking power of the polymer encapsulant and thus the strength of the interface with and without the presence of water.

While standard design of Silicon heterojunction (SHJ) uses amorphous silicon (a-Si) doped layer, switching to nanocrystalline silicon (nc-Si) is a promising route to improve efficiency [1]. One challenge to ensure a good contact is the fast nucleation of the nc-Si on the underneath intrinsic a-Si passivating layer. To promote the crystallization without damaging the passivation, a short SiH₄/H₂/CO₂ plasma treatment was introduced prior to the p-layer deposition [2], [3] which enabled efficiencies up to 23.45% [4]. As shown by the results of p-i-n solar cells [5], clarifying the role of the boron source in SHJ is crucial for further improvement.

In this work, we compare the use of BF₃ and TMB for nc-Si hole contacts. Efficiency values over 23% are obtained for both gases for the same deposition parameters, however cells using TMB-based layers yield a better FF (81.7% vs. 80.8%), whereas BF₃ cells exhibit a better J_sc (40.0 mA/cm² vs. 39.2 mA/cm²).

Symmetric stacks of c-Si(n) / a-Si(i) / nc-Si(p) / ITO address the question of the stability of the passivation along the process. We see that the implied open-circuit voltage (iV_oc) is strongly reduced by the p-layer deposition for BF₃, while it is not systematically changed for TMB. The ITO deposition reduces also more strongly the passivation when deposited on BF₃ p-layer, reducing iV_oc by ~ 30mV while it is not the case for the samples with TMB p-layer. However, annealing enables full recovery of the sputtering damage with even slight improvement (~735 mV) compared to the value before ITO deposition.

Using Spectroscopic ellipsometry, similar thicknesses were found for both dopant gases, yet with a smaller refractive index for BF₃-based layers (n=2.6 vs. n=3.5 at 600 nm) and a higher optical bandgap for BF₃ (E₀₄=2.23 eV vs. 1.79 eV), which suggests a more porous layer. UV-Raman spectroscopy yields a higher crystallinity for BF₃ layers than TMB (Χ_c = 60% v.s. Χ_c = 40%), confirming the structural difference of both materials. Both results are consistent with the difference in short-circuit current when used in a solar cell.

The marked difference between these materials indicates distinct optimization path that need to be unraveled. We plan to further investigate the nucleation and the doping efficiency differences by varying the dopant gas flux and the surface pretreatment and relate them to the past results obtain from p-i-n solar cells. We will also investigate the contribution of carbon and fluorine via Secondary Ion Mass Spectroscopy (SIMS) to understand the effect of the dopant gas on the device performance to enable further efficiency improvement.

References:
[1] J. P. Seif et al., vol. 6, no. 5, pp. 1132–1140, 2016.
[2] L. Mazzarella et al., Phys. Status Solidi Appl. Mater. Sci., vol. 214, no. 2, 2017.
[3] M. Boccard et al., Proc. SiliconPV2018, AIP, 2018.
[4] J. Haschke et al., in PVSC-46 Chicago, 2019.
[5] J. Koh et al., J. Appl. Phys., vol. 85, no. 8, pp. 4141–4153, 1999.

The efficiencies of small-pixel perovskite photovoltaics have increased to well above 20%, while the question is whether fabrication methods can be transferred to scalable manufacturing process. Here we report a method of fast blading large area perovskite films at an unprecedented speed of 99 millimeter-per-second or higher in ambient condition by tailoring solvent coordination capability. Combing volatile non-coordinating solvents to Pb²⁺ and low-volatile, coordinating solvents achieves both fast drying and large perovskite grains at room temperature. The reproducible fabrication yields a record certified module efficiency with aperture area of 63.7 cm². The perovskite modules also show a small temperature coefficient of -0.13%/°C and nearly fully recoverable efficiency after 58 cycles of shading, much better than commercial silicon and thin film solar modules. The application of the coating method to perovskite/silicon tandem cells and will also be presented. We will answer the question whether the perovskite layers can be fabricated at the speed of silicon cells are produced in the regular production lines.

Crystalline Si (c-Si) solar cells are driving the progression of renewable electricity generation technologies thanks to lowering costs and increasing efficiencies. One solution to maintain this cost-competitiveness on the long-term involves increasing efficiencies beyond the limit of c-Si by stacking a perovskite solar cell on a commercial c-Si one to form a photovoltaic tandem device. The tunable bandgap, soft processing conditions and high single-junction performance of perovskites indicate that this approach could upgrade c-Si solar cells to efficiencies >30% through a few extra process steps with low additional process costs.

For maximum photocurrent and compatibility with existing c-Si process flows, the perovskite solar cell should be deposited directly on the textured front side of the c-Si solar cell, a texture that improves light management in the c-Si. But this pyramidal texture imposes several microfabrication challenges as the perovskite absorber is typically deposited via solution processing, and is about one order of magnitude thinner than the height of the pyramids it needs to cover. Achieving a conformal deposition of all the layers of the top cell on this pyramidal texture and hence maximum performance requires a fine control over the layer formation to avoid pinhole formation.

In that regard, electron microscopy techniques, notably analytical transmission electron microscopy (TEM), can shed some light on the device nanostructure and its dependence on processing/operation conditions, guiding the development of devices. However, the fragile nature of perovskite solar cells complicates their preparation into thin cross-sections necessary for TEM observations and their analysis with high-energy electrons. This presentation will review artifacts that may occur during TEM sample preparation and observation, elaborate several strategies to identify and mitigate them, before discussing several topics correlating nanostructure and performance of perovskite single-junction and tandem solar cells. This contribution will present how electron microscopy data coupled with other techniques provide valuable inputs to guide the development of high-efficiency (>25%) tandems featuring textured n- and p-type c-Si solar cells^1,2, notably by i) identifying optimal bottom cell contact nanostructures, ii) isolating crystallographic and chemical features enabling the recombination junction to quench shunts,³ iii) guiding the removal of shunts running through the perovskite absorber on textured c-Si by adapting process conditions and iv) visualizing the dewetting of charge-selective layers during the crystallization of the perovskite solar cell on certain recombination junctions. In addition, degradation pathways triggered by reverse voltages (also investigated through TEM in situ biasing experiments)⁴, during long-term operation at maximum power point at various temperatures, or during damp heat tests (85°C/85% relative humidity) will be examined. These results highlight the dynamic nature of the perovskite nanostructure (ionic migration within the absorber and into the contacts, volatilization of species, crystallographic phase change/decomposition) depending on the external stimuli and its influence on the solar cell performance.

1 Sahli, F. et al. Nat. Mater. 17, 820–826 (2018)
2 Nogay, G. et al. ACS Energy Lett. 4, 844–845 (2019)
3 Sahli, F. et al. Adv. Energy Mater. 8, 1701609 (2018)
4 Jeangros, Q. et al. Nano Lett. 16, 7013–7018 (2016)

Integrating metal halide perovskite top cells with crystalline silicon or CIGS bottom cells into monolithic tandem devices has recently attracted increased attention due to the high efficiency potential of these cell architectures. To further increase the tandem device performance to a level well above the best single junctions, optical and electrical optimizations as well as a detailed device understanding of this advanced tandem architecture need to be developed. Here we present our recent results on monolithic tandem combinations of perovskite with crystalline silicon and CIGS, as well as tandem relevant aspects of perovskite single junction solar cells.
By selecting a front contact layer stack with less parasitic absorption and utilizing the p-i-n perovskite top cell polarity, a certified conversion efficiency of 25.0% for a monolithic perovskite/silicon tandem solar cells was enabled. Further fine-tuning of the stack optics as well as contact layers improved the efficiency to 26.0% (0.8 cm² area) and we present how especially the fill factor of highly efficient tandem solar cells behaves under current-mismatch conditions. In strong mismatch the FF of the tandem cell is enhanced which reduces the sensitivity of efficiency to spectral mismatch. ^[1] Additionally, the introduction of light trapping foils with textured surfaces is presented together with the influence on texture position on lab performance and outdoor energy yield.^[2]
The monolithic combination of perovskite and CIGS was highly challenging up to now as the CIGS surface is rather rough. By implementing a conformal hole transport layer, an 21.6% efficient monolithic perovskite/CIGS tandem (0.8 cm² area) was realised. Absolute photoluminescence of the perovskite and CIGS sub-cells gives insights into the contributions to the tandem open-circuit voltage (Voc). To further improve the tandem efficiency, the Voc of perovskite top cells needs to be enhanced via reduction of non-radiative recombination at the interface between perovskite and the charge selective layers. This can either be done via proper interlayers or via fine-tuned charge selective contacts.
Recently we have shown that self-assembled monolayers (SAM) could be implemented as appropriate hole selective contacts.^[4] The implementation of new generation SAM molecules enabled further reduction of non-radiative recombination losses with Voc’s up to 1.19 V and efficiency of 21.2% for p-i-n perovskite single junctions with band gaps of 1.63 eV and 1.55 eV, respectively.
References:
[1] Köhnen, Jost, Stannowski, Albrecht et al., Sustainable Energy and Fuels, doi: 10.1039/C9SE00120D
[2] Jost, Topic, Stannowski, Albrecht et al., Energy and Environmental Science 2018, 11, 3511
[3] Jost, Bertram, Koushik, Albrecht et al., ACS Energy Letters 2019, 4 , 583
[4] Magomedov, Al-Ashouri, Albrecht, Getautis et al., Advanced Energy Materials 2018, 8, 1801892

Wide-bandgap (~1.7-1.8 eV) perovskite solar cells have attracted substantial research interest in recent years due to their great potential to fabricate efficient tandem solar cells via combining with a lower bandgap (1.1-1.3 eV) absorber (e.g., Si, copper indium gallium diselenide, or low-bandgap perovskite). However, wide-bandgap perovskite solar cells usually suffer from large open circuit voltage (V_oc) deficits caused by small grain sizes and photoinduced phase segregation. We show that in addition to large grain sizes and passivated grain boundaries, controlling interface properties is critical for achieving high V_oc’s in the inverted wide-bandgap perovskite solar cells. We adopt guanidinium bromide solution to tune the effective doping and electronic properties of the surface layer of perovskite thin films, leading to the formation of a graded perovskite homojunction. The enhanced electric field at the perovskite homojunction is revealed by Kelvin probe force microscopy measurements. This advance enables an increase in the V_oc of the inverted perovskite solar cells from an initial 1.12 V to 1.24 V. With the optimization of the device fabrication process, the champion inverted wide-bandgap cell delivers a power conversion efficiency of ~19% and sustains more than 72% of its initial efficiency after continuous illumination for 70 h without encapsulation. The improvement on performance of wide-bandgap perovskite subcells enables us to fabricate efficient and stable perovskite tandem solar cells.

The record power conversion efficiency is 28 % for perovskite-silicon tandems and 23.2 % for perovskite-perovskite tandems. One of the challenges that must be overcome to achieve efficiency greater than 30% is to reduce the voltage loss in high bandgap perovskite cells and prevent light-induced phase separation. We have found that treating the surface of perovskites can dramatically reduce the extent of light -induced phase separation, which has interesting implications for how the process occurs. We have also developed new strategies for increasing the bandgap in perovskite compounds that have modest amounts of bromine. We have been able to make semitransparent high bandgap solar cells with greater than 20% power conversion efficiency that do not suffer from light-induced phase separation. These advances in combination with improvements in the atomic layer deposition of highly impermeable metal oxide contacts enable the fabrication of highly efficient and stable tandems.

Tandem and multijunction solar cells offer the most straightforward path to solar cell efficiencies over 30%. Three terminal (3T) tandem solar cells can overcome some of the limitations of two terminal (current matched) and four terminal (independently operated) solar cell designs. Simulations show that three terminal devices based in IBC-Si bottom cells enable the same robust performance of independently operated subcells under varying illumination conditions but does not require lateral current extraction between the cells, which becomes challenging when scaling devices to large areas.

In this talk I will give an overview of the history of three terminal devices and discuss the wide range of possibilities to create high efficiency tandem devices from combining various top cells with Si bottom cells. I will show how TCAD models agree with simple physical models and experimental results to explain the trends in the behavior of 3T tandems.

Multi-junction solar cells are a key pathway towards achieving higher photovoltaic efficiencies.The theoretical efficiency limit of a single-junction (1J) Si solar cell is 29.6%¹, whereas efficiencies >32% have already been achieved for 1J III-V top cells stacked on Si bottom cells in both the two terminal (2T) and four terminal (4T) configurations.^2,3 We will present a third path towards achieving efficiencies >32% with mechanically-stacked III-V-on-Si (III-V//Si) tandem solar cells using a three terminal (3T) configuration.

The typical tandem device architectures either connect the sub-cells in series in a 2T configuration, or operate the stacked sub-cells independently, which requires four terminals (4T). Both configurations, however, have considerable drawbacks. The 2T configuration requires that the sub-cells are current matched to operate efficiently and so this narrowly constrains the choice of the sub-cell materials. The 4T configuration does not require sub-cell current matching but this design prohibits the possibility of monolithic growth and necessitates the inclusion of gridlines or a lateral conduction layer at the back of the top cell which reduces the transmission of light to the bottom cell.⁴

The 3T configuration is a hybrid approach devised to address the constraints of the other two. The additional contact associated with the interdigitated back contact (IBC) Si bottom cell enables extraction or injection of current which circumvents the need for current matching between the sub-cells. The 3T design does not require an intermediate grid and is potentially compatible with both mechanical stacking, if a transparent conductive adhesive (TCA) is used, or monolithic growth. Moreover, simulations predict that 3T tandem cells could achieve efficiencies over 32%, comparable to record 4T tandem cell efficiencies.^3,4

We have fabricated and measured 3T mechanically-stacked III-V-on-Si (III-V//Si) tandem solar cells and will present an overview of how a 3T tandem solar cell operates. We will compare the JV and QE characteristics of a GaInP//Si tandem cell to a GaAs//Si tandem cell and analyze how the performance between these two 3T tandem solar cells differ, depending on which sub-cell is current limiting.

(1) Schafer, S. and Brendel, R. Accurate Calculation of the Absorptance Enhances Efficiency Limit of Crystalline Silicon Solar Cells with Lambertian Light Trapping. IEEE J. Photovoltaics. 2018, 8, 1–3.
(2) R. Cariou et al, III-V-on-silicon solar cells reaching 33% photoconversion efficiency in two-terminal configuration. Nat. Energy,2018, 3, 326 – 333.
(3) Warren, E.L.; Deceglie, M.G.; Reinäcker, M.; Peibst, R.; Tamboli, A.C.; Stradins, P. Maximizing tandem solar cell power extraction using a three-terminal design. Sustainable Energy & Fuels. 2018,2, 1141 – 1147.
(4) Essig, S; Allebé, C; Remo, T; Geisz, J.F., Steiner, M.A.; Horowitz, K; Barraud, L.; Ward, J.S.; Schnabel, M.; Descoeudres, A.; Young, D.L.; Woodhouse, M.; Despeisse, M.; Ballif, C.; Tamboli, A.C. Raising the One-Sun Conversion Efficiency of III–V/Si Solar Cells to 32.8% for Two Junctions and 35.9% for Three Junctions. Nat. Energy. 2017, 2, 17144.

The interfacial layer between semiconductor and metal cathodes in photovoltaics play a critical role in enhancing the transport of charges carriers or preventing the recombination at the interface. Here, high crystalline and self-assembled poly(vinylidene fluoride-co-trifluoroethylene) (P(VDF-TrFE)) thin films are shown to function as a passivation layer and also acts as back surface field layer for photovoltaic devices, simultaneously. The P(VDF-TrFE) thin films are spincoated on the bottom of n-Si surface by modified breath figure methods. After that, poly(3,4-ethylene dioxy thiophene):poly(styrene sulfonate) are formed on the top of nano-textured Si as the junction. Comparing previously reported organic / Si hybrid solar cells with this study, such devices demonstrate record-breaking photovoltaic conversion efficiency of 18.1% with electrostatic passivation and reflection of minority carriers by the induced built-in electric field. Furthermore, due to its switchable ferroelectric property, the P(VDF-TrFE) thin films can be utilized to the p-Si solar cells as passivated back surface field layers. Finite-difference time-domain simulation reveals that the electric field due to the below spontaneous polarization causes band bending of Si, reflecting minority carriers and reducing the surface recombinations of the devices.

High-efficiency solar cell devices characterized by extremely high open-circuit voltage (V_OC) values have shown that the traditional constraints imposed by extrinsic recombination processes will be eventually surpassed at some time in the foreseeable future. The accurate evaluation of Auger lifetime and its temperature-dependence are thus fundamental not only for the correct interpretation of effective carrier lifetime data, but also for the simulation of device performance, especially when these are deployed in the field, where the operating conditions can greatly vary from the standard testing conditions. In this work, we present the Auger lifetime across a range of temperatures from 300 to 500 K and a range of injection level from 5 x 10¹⁴ to 1 x 10¹⁶ cm^-3 showing that, in stark opposition with what generally accepted, a strong increment of the lifetime values happens at high temperatures. Based on these results, we discuss the ambipolar Auger coefficient in the high injection range and propose a parameterization for its temperature dependence in agreement with a model previously presented in literature. Finally, we evaluate the intrinsic-limited implied voltage (iV) within the same range of injection level and temperature, and show that the evaluated strong increment of Auger lifetime counteracts the typical drop of high-efficiency solar cells performance with high temperature.

Heterojunction back-contact (HBC) crystalline silicon (c-Si) solar cells with energy conversion efficiency > 26%, which reaches nearly the theoretical limit of ∼30%, have been achieved, hence reducing fabrication cost becomes more and more valuable. The decreasing thickness of c-Si to < 100 µm is one of the keys since c-Si typically takes half of the total cell cost. Usually, the surface of c-Si is textured to improve sunlight confinement or efficiency. For such thin c-Si, small texture size and low amount of Si etching loss are strictly required to maintain its mechanical strength during fabrication processes. In mass-production, wet processes are preferable for lowering the cost. We have already reported a novel method, so-called microparticle-assisted texturing (MPAT) process, in which glass microparticles were mixed with conventional alkaline-based texturing chemical solutions to reduce the texture size, etching duration, and etched c-Si thickness by almost one order of magnitude. The MPAT process was applicable to c-Si with a thickness down to 50 µm. The superiorities were attributed to that the glass microparticles with certain kinetic energy can sweep out reaction-generated hydrogen bubbles from the c-Si surface to speed up the texture formation. However, the previous work focused on only mirror-polished c-Si wafers. In this work, we studied the MPAT process on as-cut c-Si used for actual solar cell mass-production. The similar superiorities such as small texture size (< 3 µm), low optical reflectivity (R~7%) quick formation (3–5 min), and extremely low c-Si loss [< 2 µm, even smaller than the depth of the saw-damaged layer (SDL) ~ 6 µm] were revealed. In addition, we also developed a suitable wet chemical cleaning prior to surface passivation using catalytic chemical vapor deposition (Cat-CVD) silicon nitride (SiN_x)/amorphous silicon (a-Si) stacked layers. A world-record low surface recombination velocity (SRV) ~ 0.38 cm/s was achieved. Owing to the high-quality surface passivation, both SDL removal and texturing can be done by only the MPAT process within a few minutes. After coating SiN_x/a-Si, the reflectivity reaches ~ 0.4 % at minimum, and < 2% in wide wavelength 450–950 nm. Toward mass-production using the MPAT process, etching of multiple c-Si full-size wafers with a pitch of 5 mm was proved to be possible to obtain uniform textures with a yield almost 100%. Therefore, the MPAT process is almost close to the mass-production of low-cost and high-performance thin c-Si-based solar cells.

Polycrystalline silicon (poly-Si) has proven to be a game-changing material in the field of high thermal budget carrier-selective passivating contacts (CSPCs) for c-Si solar cells beyond PERC architecture.^1-3 However, doped poly-Si exhibits a very high free carrier absorption (FCA), which has turned the attention of researchers towards wide bandgap materials, such as poly-SiO_x and poly-SiC_x. In these materials, the opto-electronic properties depend on oxygen⁴ and carbon⁵ alloying, respectively. To properly model the generation profile in solar cells based on poly-Si, poly-SiO_x or poly-SiC_x CSPCs, a thorough understanding of their optical properties, especially the absorption coefficient, is essential.
As weak FCA at long wavelengths is difficult to detect using ellipsometry, we have used in this work two techniques, the absolute photothermal deflection spectroscopy (PDS) and the reflection / transmission (RT), that are sensitive enough to measure absorption values below 1%. In this way, we could obtain the absorption coefficient of n- and p-type doped poly-SiO_x and poly-SiC_x CSPCs on quartz and on crystalline silicon (c-Si) wafer substrates, respectively, in the wavelength range between 300 nm to 2000 nm. The absorption coefficients of doped CSPCs obtained from these measurement techniques are in the same order to magnitude, albeit we observed small differences owing to the different substrates and/or the measurement techniques.
By changing the oxygen-to-silane gas ratio, R_{CO2 =} [CO₂] / ([CO₂] + [SiH₄]), and carbon-to-silane gas ratio, R_CH4 = [CH₄] / ([CH₄] + [SiH₄]), the optical properties of our in-house developed poly-SiO_x and poly-SiC_x layers can be obviously altered. We note a decrease in the FCA for n- and p-type doped poly-SiO_x layers in the infrared region with respect to the reference poly-Si (R_CO2 = 0). Among n- and p-type doped poly-SiO_x and poly-SiC_x layers, n-type doped poly-SiO_x (R_CO2= 0.83) layer with a doping concentration of 1e20 cm^-3 has the least FCA in the wavelength range between 800 nm and 1200 nm. In this respect, for the same thickness, it would be better to use n-type doped poly-SiO_x rather than n-type doped poly-SiC_x at the back side of front/back-contacted or interdigitated back-contacted c-Si solar cell architectures.
The obtained absorption coefficient values can now be used as input to study the optical behaviour of single- and multi-junction solar cells endowed with such CSPCs. Simulations of two, three and four terminal perovskite / c-Si solar cells are ongoing and will be presented at the conference.


¹ G. Yang, et al., APL, 2016.
² G. Yang, et al., SOLMAT, 2016.
³ F. Feldmann, et al., SOLMAT, 2017.
⁴ G. Yang, et al., APL, 2018.
⁵ A. Ingenito, et al., IEEE J. Photovolt., 2019.

Transparent electrodes are used in many optoelectronic devices, such as solar cells, high speed photodetectors, imaging arrays, and displays. Common materials used for transparent electrodes include conductive oxides, carbon nanotubes, graphene, and metal nanowire networks. In all current approaches a balance needs to be struck between good electrical conductivity and high optical transmission: thinner films lead to higher transmission, but lower electrical conductivity. Among the many approaches, metal nanowire electrodes provide extremely high conductivity, but introduce significant shadowing losses due to reflection and absorption.

In this presentation we numerically investigate the optical and electrical performance of silver nanowire electrodes that can virtually eliminate all shadowing losses by the metallic contacts. This is achieved through light trapping via total internal reflection at the surface of a thin dielectric cover layer. Two kinds of interdigitated electrodes are considered, using either cylindrical silver wires or triangular silver wires. It is shown that both designs can simultaneously provide high optical transparency and high electrical conductivity when using micron sized silver wires. The relative contribution of radiative and non-radiative loss is evaluated as a function of wire size, including the effect of surface plasmon polariton mediated dissipation. The performance is analyzed at high metal coverage to highlight the importance of light trapping. It is shown that 2 μm wide triangular silver wires with 25% metal coverage, embedded in a Si₃N₄ cover layer can provide a peak transparency of 98% and an average optical transmission of 93% across a broad wavelength range spanning from 400nm to 1.1μm, while offering sheet resistivity as low as 0.35Ω/sq. Reducing the wire width to 300nm reduces the light trapping efficiency of triangular electrodes by ~40%, which is quantitatively explained in terms of the angular distribution of the reflected light from isolated wires. A new figure of merit is proposed to evaluate the overall performance of light trapping transparent metallic electrodes, and it is shown that triangular electrodes progressively outperform cylindrical electrodes as the wire size increases. Methods for producing large area light trapping electrodes will be discussed.

Parasitic absorption in transparent electrodes is one of the main roadblocks to enable power conversion efficiencies (PCEs) for perovskite-based tandem solar cells beyond 30%. To reduce such losses and maximize light coupling, the broadband transparency of such electrodes should be improved, especially at the front of the device. Here, we show the excellent properties of Zr-doped indium oxide (IZRO) transparent electrodes for such applications, with improved near-infrared (NIR) response, compared to conventional In-doped tin oxide (ITO) electrodes. Optimized IZRO films feature a very high electron mobility (up to ∼77 cm²/V.s), enabling highly infrared transparent films with very low sheet resistance (∼18 Ω/sq for annealed 100 nm films). For devices, this translates in a parasitic absorption of only ∼5% for IZRO within the solar spectrum (250-2500 nm range), to be compared with ∼10% for commercial ITO. Fundamentally, we find that the high conductivity of annealed IZRO films is directly linked to promoted crystallinity of the indium oxide (In₂O₃) films due to Zr-doping. Overall, on four-terminal perovskite/silicon tandem device level, we obtained an absolute 3.5 mA/cm² short-circuit current improvement in silicon bottom cells by replacing commercial ITO electrodes with IZRO, resulting in improving the PCE from 23.3 to 26.2%. [1]

References
[1] E. Aydin et al., Advanced Functional Materials (2019), 1901741.

Thin films are ubiquitous in the preparation of crystalline silicon solar cells. With the introduction of the PERC technology also the method of atomic layer deposition (ALD) has been introduced in high volume manufacturing in photovoltaics. Currently, the technique is quickly gaining market share for the deposition of ultrathin Al₂O₃ nanolayers for rear side surface passivation. The advantages of ALD are that it is scalable, that it is well suited to prepare high quality and uniform nanolayers, and that it is a "soft" deposition technique preventing interface damage. In this presentation the state of the art of ALD for silicon photovoltaics will be discussed as well as some ongoing developments and potential new applications. This includes new materials for surface passivation, nanolayers for passivating contacts, transparent conductive oxides as well as applications of ALD for polysilicon passivating contact solar cells.

One high efficiency crystalline silicon (c-Si) solar cell that has potential to be cost-competitive with Al-BSF devices while maintaining the passivation quality of PERC structures is the silicon heterojunction. Most noteably is the Heterojunction with Intrinsic Thin-layers (HIT) solar cell, which chemically passivates the surface of the c-Si with intrinsic hydrogenated amorphous silicon (a-Si:H) while allowing charge carriers to conduct through the contact. Doped a-Si:H layers then create the potential gradient necessary for carrier diffusion and charge collection to occur.

Even with it's high performance and simplified fabrication process, the HIT solar cell has potential to be improved upon. The a-Si:H layers produce parasitic absorption of high energy light. Many groups have also found the a-Si:H contacts to be sensitive to annealing treatments above 200°C. This limits the processing space for the metallization step, which typically occurs at much higher temperatures.

In this work we use Atomic-Layer-Deposited (ALD) MoOx as a hole-selective contact to c-Si in combination with thin SiOx and Al2O3 passivation layers to study the solar cell parameters of carrier selective contacts without the use of a-Si:H. While the optical properties of MoOx present a significant reduction in parasitic absorption compared to a-Si:H, the temperature stability of the material is still in question. Many groups have been unable to anneal MoOx based contacts above 120°C without degrading the electrical properties of the contact. We show that by using a thin Ni capping layer before Al metallization, the contact remains stable up to 300°C with contact resistivities below 10 mΩ-cm2. This presents significant improvements on the thermal budget of the MoOx processing sequence and will allow for more appropriate contact formation steps.

Using Ultraviolet Photoelectron Spectroscopy we measure a work function of 6.2 eV in our 5nm MoOx contact. We simulate the band-bending and hole concentration at the c-Si surface as a function of the MoOx contact work function and show that our films exhibit sufficient hole-selective properties. Additionally, High Resolution Transmission Electron Microscopy images show that when a thin Ni capping layer is not used prior to Al metallization, a 2-3 nm Al2O3 layer forms at the Al/MoOx interface. This insulating interlayer contributes to a large barrier to hole transport, making the Al/MoOx contact incompatible with high efficiency heterojunction solar cells. The Al/Ni/MoOx contact, however, exhibits no such interlayer. This suggests that Ni may act as a diffusion barrier to O species during solar cell fabrication.

Finally, we show that by using a thin (∼1 nm) Al2O3 passivation layer at the MoOx/c-Si interface, we are able to achieve a minority carrier lifetime of over 1 ms on n-type c-Si.

Recently, Yang et al. have reported 22.1% efficient c-Si solar cell by applying an atomic-layer-deposited (ALD) TiO_x thin-layer as electron contact to n-type base [1]. The origin of the electron selectivity of TiO_x has been ascribed to the asymmetric current flow at the (n) c-Si/TiO_x interface where the conduction band offset is much lower than the valence band offset. On the other hand, we recently found that TiO_x can be tuned from electron to hole selectivity by controlling the ALD condition etc. [2]. This offers an interesting possibility that TiO_x can be used as a hole selective contact alternative to the widely-used p-type a-Si:H and transition-metal oxides such as MoO_x, WO_x and V₂O_x. In this contribution, we show for the first time that TiO_x thin layers can act as efficient passivating hole selective contacts.
TiO_x thin layers were deposited by thermal-ALD on c-Si (FZ, 1 Ωcm, (100), planer, n-type). Firstly, carrier selectivity of the deposited TiO_x was studied by measuring V_oc of solar cells. To decouple the carrier selectivity from its surface passivation an intrinsic a-Si:H buffer layer was inserted between c-Si and TiO_x. A standard SHJ structure of either (i-n) a-Si:H/ITO or (i-p) a-Si:H/ITO stack was formed as a counter electron or hole contact, respectively. It is found that V_oc of the solar cell is 200-400 mV higher when using our TiO_x as hole contact than using it as electron contact. The SPV measurement showed that the TiO_x induces large band bending (~900 mV) with respect to (n) c-Si while almost no band bending is created when deposited on (p) c-Si. This implies the presence of the negative fix charge in the TiO_x, which is considered as one of the origins of the observed hole selectivity of the TiO_x. Furthermore, we found that the carrier selectivity of TiO_x depends significantly on the work function (WF) of the capping metal (or TCO) contact. The V_oc is increased monotonically with increasing the WF of the capping layer when using TiO_x as hole contacts. This indicates that band bending in (n) c-Si is significantly influenced by the WF of the capping layer, as it is well-known in the MIS contact system. By using an ITO/TiO_x/(i) a-Si:H stack as an emitter layer on (n) c-Si, we obtained a relatively high V_oc of 650 mV. Furthermore, the TiO_x is also found to act as a good passivation layer with respect to c-Si. An effective lifetime of >1 ms was obtained by depositing TiO_x on (n) c-Si without a-Si:H buffer layer. By optimizing both the surface passivation and the hole selectivity of the TiO_x layer, we attained solar cell efficiencies of >18%, demonstrating that TiO_x has potential of working as an efficient passivating hole selective contact. We discuss the origin of the hole transport in the TiO_x which contradicts to the previous transport model based on the band alignment at the TiO_x/Si interface.
[1] X. Yang et al. Prog. Photovolt: Res. Appl. 25, 896 (2017). [2] T. Matsui et al. Energy Procedia 124, 628 (2017).

The excellent electrical, optical and structural tunability of doped zinc oxide (ZnO) makes it a very good candidate for the replacement of indium-based material in the manufacturing of transparent conductive oxides. In this work, we present a comprehensive investigation of ALD grown hafnium doped ZnO within the context of its integration in different solar cells architectures. Specifically, we focus on the low range doping region, where Hf substitution is believed to be the key for band gap tunability without negatively effecting the carrier transport behavior. Scanning and transmission electron microscopy (SEM and TEM), x-ray diffraction, Kelvin probe force microscopy, Hall-effect measurements, spectrophotometry and ellipsometry were utilized to provide conclusive evidence of the suitability of Hf doped ZnO in different solar cells architectures. A band gap increase is being observed, as well as an increase in transmittance with doping. Electrically, doping is causing a decrease in resistivity and in work function. These results are interpreted in light of first-principles density functional theory simulation (DFT) to elucidate the mechanisms responsible for the electronic and electrical properties of Hf doped ZnO. DFT calculations predict a modification in the band structure of ZnO when Hf is substituted and/or embedded in the ZnO matrix as HfO₂ phases. The experimentally measured and theoretically calculated modifications in the properties of the ZnO with Hf doping, validates its compatibility with different solar cells architectures.

Antimony sulphide (Sb₂S₃) is another attractive photovoltaic material in the chalcogenide group and has drawn a great attention worldwide in the last decade. In contrast to the widely investigated and commercially competitive thin film solar cells such as CuIn_xGa_1-xSe₂ and CdTe, Sb₂S₃ is non-toxic and exists naturally as stibnite minerals, with both while Sb and S are bothas earth abundant elements . Sb₂S₃ is a binary compound with a single phase, consisting of linked one-dimensional ribbons. Such a ribbon structure provides a preferential pathway for electron transfer if withalong the desired orientation<span style="font-size:10.8333px">.</span> Antimony sulphide has high absorption coefficient of α>10⁵ and a bandgap of ~1.7eV, making it a suitable top cell candidate for tandem solar cells with silicon to overcome the single-junction Shockley-Queisser efficiency limitation. In the last decade Sb₂S₃ has been widely utilized as an efficient sensitizer in dye sensitized solar cells fabricated by chemical bath deposition . However, the CBD method is time-consuming with many undesirable oxide by-products, which usually require complex post treatment for the removal of these residuals.Moreover, reported high efficiency Sb₂S₃ solar cells have utilized unstable and expensive organic hole transport materials such as P3HT Spiro-OMeTAD, and PEDOT:PSS<span style="font-size:10.8333px">.</span> While inorganic hole transport materials such as NiO and CuSCN have been investigated, the results have not been promising. Recently, Lijian et al. used V₂O₅ as the hole transport layer (HTL) and obtained a PCE of 4.8%, demonstrating the highest efficiency of a fully inorganic planar Sb₂S₃ solar cells up to date. However, the open circuit voltage of such the record cell is 550 mV, which indicates a large Voc deficit implying there is still much work to be done to improve the Sb₂S₃ absorber quality and interface engineering. Recently, a push towards dry vacuum-based methods, such as the thermal evaporation and atom layer deposition, have been employed to grow high quality Sb₂S₃ and Sb₂Se₃ in a relatively clean environment. Additionally, the rapid thermal evaporation (RTE) method has been recognized as an effective and reliable method to grow Sb₂S₃ thin films, achieving an efficiency of 3.5% with a high Voc of 710mV when using CdS as electron transport layer (ETL)However, the reported orientation of the Sb₂S₃ is not well controlled when deposited by the RTE method. In this work, we report the first fabricated Sb₂S₃ thin films with vertical orientation by VTD method . To better understand the key factor that enables the vertical growth of Sb₂S₃, we use the RTE method as a reference, which does not create vertically aligned Sb₂S₃ crystals on a CdS buffer layer. We achieved the an efficiency of 4.73% with a high Voc of 710 mV by using the iTO/CdS/Sb₂S₃/Gold configuration via VTD method compared to 370 mV using RTE method. We propose a simple model to describe the growth process.

The field of c-Si photovoltaics has strongly diversified in recent years with the advent of a wide variety of novel passivation and passivating contact materials. Recently, we have demonstrated excellent surface passivation using stacks of ultrathin (~1.5 nm) RCA SiO_x capped with ALD ZnO/Al₂O₃, with an implied open-circuit voltage (iV_oc) of 725 mV on planar c-Si(n) wafers.[1] Within this SiO_x/ZnO/Al₂O₃ stack, the RCA SiO_x enables chemical passivation, similar as in poly-Si passivating contacts. The Al₂O₃ layer on top serves as a dense capping layer: It prevents effusion of H from the ZnO upon annealing, which is needed to hydrogenate the SiO_x.
The unique aspect of the (doped) ZnO is that it is suited as an antireflection coating (simulated J_sc of 41.6 mA/cm²) which is also conductive (< 1 mΩcm). Therefore, if a proper tunnel contact between c-Si/SiO₂/ZnO can be made, the ZnO could serve as a full-area passivating, antireflective and lateral transport layer on the front side of a c-Si solar cell. In this work, we demonstrate several crucial steps to enable this application in industrial cells by looking into the influence of texture and doping level, oxide preparation method, metal contacting and thermal stability.
Firstly, we verified that the stack also passivates on textured c-Si(n) wafers (iV_oc = 728 mV) and that its passivation on n⁺ diffused surfaces (100 Ω/sq) is on par with industrial SiN_x. The stack is thermally stable up to ~550 ^oC, which is not firing-compatible, but allows for a much higher paste curing temperature than for HIT-type cells.
Secondly, we show that the SiO_x can be prepared in many ways (RCA, LTO, UV/O₃, NAOS) which all yield good passivation. UV/O₃ however yields the best passivation and is a room-temperature, single-sided treatment which allows for accurate control over the oxide thickness (< 1.65 nm).
Thirdly, in order to be able to contact the ZnO by metal, we can selectively remove the insulating Al₂O₃ capping layer from the ZnO after hydrogenation by a wet-etch. Interestingly, no proper tunnel contact (>10 Ωcm²) can be made on 3 Ωcm c-Si(n) wafers, whereas a “first-try” value of ~0.1 Ωcm² was obtained on n⁺ diffused surfaces (100 Ω/sq). This contact resistivity is sufficiently low to use ZnO as a “hybrid” homo/heterojunction contact on n⁺ surfaces: the n⁺ doped Si surface provides electron-selectivity and facilitates tunneling, whereas the ZnO provides full-area passivation and aids in lateral transport, potentially allowing for higher Ohmic FSFs.
Ongoing work focuses on the effect of the doping levels of both the c-Si and ZnO and the integration of the ZnO on the front of a PERC-type cell.

[1] B. W. H. van de Loo, B. Macco, J. Melskens, W. Beyer, and W. M. M. Kessels, “Silicon surface passivation by transparent conductive zinc oxide,” J. Appl. Phys., vol. 125, no. 10, p. 105305, 2019.

Surface passivation of crystalline Si (c-Si) is a key enabler for achieving high efficiency c-Si solar cells. While an Al₂O₃ surface passivation layer grown by atomic layer deposition (ALD) is known for excellent surface passivation for p-type Si from the negative fixed charges in the Al₂O₃ layer, low refractive index of Al₂O₃ demands an additional anti-reflection coating to suppress optical reflection, which leads to extra capital cost in c-Si solar cell manufacturing. Therefore, the multi-functional passivation layer that can provide both the high level of passivation quality and optimum optical property is necessary for high efficiency and cost-effective silicon solar cells.
Here, we designed the multifunctional Al-doped TiO₂ passivation layer using ALD single process for low cost high efficiency silicon solar cells. TiO₂ film grown by ALD is a promising candidate for front multifunctional passivation layer of p⁺-emitter/n-base structure due to negative fixed charge densities and appropriate refractive index for anti-reflection coating of silicon solar cell. ALD process allows to control the composition accurately and provide excellent passivation quality with thin films. In this work, we controlled the Al concentration in TiO₂ film with amorphous phase from 0 to 15.5 % by adjusting the cycle ratio of Al₂O₃ to TiO₂ in ALD process. We then successfully demonstrated that the fixed charge densities of Al-doped TiO₂ passivation layer can be controlled from -8×10¹¹ cm^-2 to -3×10¹² cm^-2 by varying the amount of Al concentration. As a result, we achieved implied V_oc up to 709 mV with 15nm thick Al-doped TiO₂ on n-Si by maximizing field-effect passivation. We also investigated the effect of film thickness on surface passivation quality. By Al doping in TiO₂ passivation layer_, it showed significant enhancement of passivation performance compared to TiO₂ film from 5 to 55 nm thickness. Al-doped TiO₂ film provided high level of passivation quality leading to implied V_oc of 700 mV from 15 to 55 nm. Finally, we have conducted research to reduce the total reflectance of silicon by applying Al-doped TiO₂ passivation layer of which the refractive index became close to 2.3, ideal refractive index for silicon solar cell. Therefore, we successfully demonstrated that the ALD Al-doped TiO₂ multifunctional passivation layer is a suitable candidate for high efficient Si solar cells based on the p⁺-emitter/n-base structure with excellent optical property and outstanding passivation characteristic.

The growth of oxides by atomic layer deposition (ALD) over lead-halide perovskites in solar cells is attracting increasing attention for improving environmental and mechanical stablity. A wide range of materials have now been investigated, including SnO₂, TiO₂, Al-doped ZnO and zinc tin oxide. These oxide overlayers have led to unencapsulated perovskite solar cells achieving stable performance for 4500 h in ambient air. However, the range of growth temperatures that can be used to grow the oxide overlayers over the perovskite films is restricted to typical values of only 60 - 100 ^○C due to the low stability of the perovskites. This limits the mobility and density of the oxide films achievable. In this work, we show that we can open up the processing window of oxides grown over lead-halide perovskites by using atmospheric pressure chemical vapor deposition (AP-CVD). This technique yields oxide films with similar uniformity, density and electronic properties as ALD films at similar growth temperatures, but with orders of magnitude higher growth rates [1]. We investigate the growth of TiO_x over thermally-sensitive CH₃NH₃PbI₃ films. We achieve a growth rate of 1.19 ± 0.04 nm s^-1 at a deposition temperature of 150 ^○C, which allows 7 nm TiO_x films to be grown in 6 s (compared to >30 min for ALD). We show that this rapid deposition enables TiO_x to be directly grown on CH₃NH₃PbI₃ films without damage to the bulk or surface, as shown by our X-ray diffraction, X-ray photoemission spectroscopy and time-resolved photoluminescence measurements. Indeed, we show that the TiO_x overlayers can be grown at temperatures exceeding 180 ^○C without a significant drop in efficiency in CH₃NH₃PbI₃ solar cells. These results can be generalised to triple-cation provskite devices, as well as to AP-CVD SnO_x overlayers. In particular, we show that the conformal nature of the oxide overlayers lead to perovskite devices with improved performance (reaching 19.7% for triple-cation perovskite devices using a 60 nm SnO_x overlayer). Our work demonstrates AP-CVD to be a versatile technique for growing high-quality oxides over a wide range of processing conditions.

Reference
[1] R. L. Z. Hoye, et al., ACS Appl. Mater. Interfaces, 2015, 7, 10684

Surface-passivating, carrier-selective contacts formed by heavily doped silicon films and thin interfacial oxide layers are capable of achieving high performance, leading to record efficiencies for silicon solar cells at both laboratory (25.8% TOPCon on n-type wafers and 26.1% POLO on p-type wafers) and industrial scale (24.6% by Trina Solar on 245 cm² n-type wafers). This type of passivating contact is increasingly regarded as the basis for the next generation of high performance technology for silicon solar cells. It can also constitute a baseline technology for PV research labs due to its simple implementation and excellent electrical performance. Different approaches have been explored for growing the thin interfacial oxide, depositing the silicon film, and incorporating dopants into it, most of which are based on various forms of chemical vapour deposition. In this contribution, we will present a novel approach based on physical vapour deposition. We have formed p-type passivating contacts based on PVD deposited silicon films doped in-situ with boron atoms. A low recombination current density J_oc = 20 fA/cm² and a low contact resistivity ρ_c = 10 mΩ-cm² have been achieved. Such a PVD-formed p-type passivating contact has been successfully implemented as full area rear hole selective contacts in p-type silicon solar cells. The best devices have reached a conversion efficiency of 23% with an open-circuit voltage of 701mV. At the same time, we are working on the development of n-type passivating contacts using a similar approach. In this presentation, we will compare the PVD approach with chemical vapour deposition approaches in terms of their fabrication processes and electrical performance. We will discuss results for PVD p-type passivating contacts and show work in progress for PVD n-type passivating contacts.

The first part of the talk will deal with (sub-)monolayers of ALD metal oxides, mainly Al2O3, deposited onto tunnel-SiO2. Wet-chemically grown and dry-thermally oxidized tunnel-SiO2 have a different surface termination, which leads to a different initial deposition during the first ALD-cycles. Despite the deposition of just sub-monolayers the Si-surface passivation can reach very good levels, even before a forming gas annealing, while maintaining a low contact resistivity due to tunneling through the ultra-thin layer stack. The surface passivation is explained via the formation of induced acceptor states in SiO2, which capture electrons from the dangling bonds at the Si/SiO2 interface so that these defects are electronically deactivated, while a negative fixed charge in the dielectric layer enables field effect passivation [1,2,3].

In the second part, the kinetics of the thermal activation as well as the annihilation of the grown-in defects in float-zone (FZ) Si wafers are studied. In the critical temperature window between approx. 400-800°C the bulk-lifetime of FZ-Si is decreased by more than 2 orders of magnitude and this degradation takes place on very short timescales. In addition to conventional furnace annealing, rapid thermal annealing (RTA) and millisecond flash lamp annealing (FLA) were studied. It will be demonstrated that also the recovery of the bulk lifetime can be achieved by thermal treatments that are much shorter and at lower temperatures than previously reported. Finally, the role of impurities detected by SIMS and their possible defect configurations (modelled by density functional theory) are discussed.

[1] D. König & D. Hiller et al., Sci. Rep. 7, 46703 (2017)
[2] D. Hiller et al., ACS Appl. Mater. Interfaces 10, 30495 (2018)
[3] D. Hiller et al., J. Appl. Phys. 125, 015301 (2019)