Wednesday of the 7th World Conference on Photovoltaic Energy Conversion saw the last of the technical plenary sessions in Areas 2, 3, 4, and 8.
In Area 3 Aaron Ptak from NREL gave an inspiring plenary presentation of hydride vapor phase epitaxy (HVPE) for the growth of high-efficiency III-V solar cells. Dr. Ptak's team has been able to produce HVPE-grown GaAs cells with nearly the same performance as conventional MOVPE growth, but at up to 30x higher growth rates. Detailed cost modeling suggests that, with scale, HVPE has the potential to make III-V photovoltaics competitive for a broad range of terrestrial applications, as well as making (aero)space use considerably more affordable.
For Area 2, Atiye Bayman, CTO of Miasole Hi-Tech gave a nice overview of the status of CIGSe technology over the world, and especially the rapid growth of production capacities which should achieve over 10 GW before 2020. Then, she more precisely described the technologies developed at Miasole and their perspectives for the next gen (light weight, wider band gap absorber materials by introducing silver in place of Cu, improved TCO). Finally, she provided a highly positive message about the future of CIGS technology.
Dr Anthony Burrell of NREL presented a very instructive lecture on the different options for battery storage technologies in the Area 8 plenary. In 2016, a total capacity grid-scale battery capacity of about 6,500MWh were installed, of which 63% were with a Li-ion technology. The estimated current cost of manufacturing of such Li-ion batteries is $220/kWh (in 2018) and exhibit a learning rate (LR) of 21.6%, similar to the learning rate of crystalline-Si PV modules. The largest grid-scale battery storage system is installed in Hornsdale, South Australia, has a total power/capacity of 100MW/129MWh and was installed in just 63 days at a cost of $325/kWh. The current market for battery is driven by the EV market. It is estimated that the global sales of EV's will reach 20M units by 2030. It is projected that Li-ion technology will dominate the EV and the grid-scale market at least for the next 20 years.
Finally, in Area 4, Qi Wang of JinkoSolar covered the advances in PERC manufacturing since the technology inception in 1983. He shared the road map for capacity and efficiency of Jinko Solar and touched on the main silicon issues that have led to those improvements. He pointed to advanced metallization as the key moving forward and commented on the effect that the new regulations will pose to the Chinese installed capacity. He reported a world record for PERC cell with 5 bus bars from May 2018 23.98% and an outstanding electricity cost of 0.0242$/kWh in their latest PERC power plant installation.

Following this extended plenary session and after a well-earned coffee break it was time for another well attended poster session across Areas 1, 2, 4, 5, 9, and 10. The following submissions were recipients of poster awards in their respective Areas:
Area 5
H Wu, H Nguyen, D Macdonald        
Area 10
R Sharma   

Area 9
F Vignola, J Peterson, R Kessler,  et al      

Area 1
M Giteau, S Collin, Z Jehl,  et al         
Area 2
A Ren, H Xu, A Delamarre,  et al         
Area 4-I
K Fong, T Kho, W Liang,  et al         
Area 4-II
N Chang, B Hallam, D Chen,  et al  
After taking a slightly later lunch, attendees returned for the two afternoon technical sessions. In the first set, from 2.00 to 3.30 pm the following highlights were provided by the session chairs:

Area 10 – Sustainability
Keiichi Komoto from Mizuho Institute (Japan) described a NEDO-funded R and D project developing a module recycling process that can treat multiple module types, and its beneficial environmental impact evaluated through life cycle assessment. One key aspect is the limited cost of recycling that must be put in perspective with its expected added value. Michael Bolen of EPRI presented a survey of existing PV module recyclers in Europe in terms of processes and recovery rate. He highlighted that the energy consumed by recycling is a factor of 1000 below the lifetime PV electricity production of a module. Parikhit Sinha of First Solar described ecological toxicity modeling for CdTe modules showing that the current concentration of CdTe in the environment is 1000 times lower than pure Cd coming from other sources including human activities. Ilke Celik from University of Toledo described a study of the energy-water-food-land nexus for agrivoltaic systems (agriculture-PV) systems through life cycle assessment and ecological network analysis. Karen Drozniak of First Solar reviewed multiple mandatory and voluntary efforts to define sustainability criteria internationally and enable fair comparisons across products for greater recognition and improved identification of sustainability in the marketplace. Joswin Leslie from the Arizona State University presented the TCLP method for assessing the toxicity of some key components of PV modules and the necessity to use proper methods. The entire session shows the challenges to ensure the complete sustainability of the PV industry and development in the light of its expected massive development.

Area 2 - Defects Characterization
Aaron Arehart presented interesting results on the light-induced instabilities on the Voc and the effect of GBs. Michael Lloyd used a single crystal CZTSe as a platform to study fundamental device properties. Giovanna talk shed some light on mechanism understanding of CIGS interface grain boundary defect states through AS and modeling. Darius Kuciauskas presented in his talk that the key leading to higher CdTe efficiency is by doping in Cd rich condition and avoid As dimer to improve hole concentration and not lose lifetime.

Area 1 - Novel Material Systems
Adele Tamboli, who was awarded the WCPEC young researcher award presented an exciting perspective on new materials for multijunction solar cells going beyond III/Vs. As a novel route II-IV-V2 materials of low costs multijunction devices are suggested. Exciting high voltage of 1.3 V for ZnSiP2 as a novel PV material. Ah-Jin Cho presented achievements on WSe2/MoS2 p-n heterojunction based transparent PV, promising light weight, transparency and flexibility for BIPV. Ryoji Katsube of Kyoto University presented results regarding the phase diagram of Mg-P-Zn. He found that by using Sn as a "reaction field" he could achieve homogeneous MgZnP2. You-Lan Li of National Chiao-Tung University found that by using metal assisted chemical etching, they could produce multi-scaled textured silicon to enhance light absorption in pyramidal Si/PEDOT:PSS hybrid cells.

Area 9 - Solar Resource
The weather was fine in the Solar Resource session. Greg Kimball used several weather databases to aggregate weather data spatially and concluded that P99 downside estimates of inter-annual variability for GHI were 4%-8% throughout the continental US. Dr. Gueymard described the need for atmospheric aerosol estimates at hourly time intervals for irradiance orientation transposition modeling. Ian Riedel discussed preliminary work on site-to-site variations in average photon energy, and called for collaborators to help expand this pool of spectral distribution characterizations for more sites. Ian Peters eloquently discussed atmospheric "Haze" and its impact on photovoltaic performance in various locations, estimating an 11.5% annual reduction in PV system performance due to aerosols in New Delhi, India. Mauro Pravettoni reviewed improvements in reproducibility observed in ongoing inter-laboratory calibration of spectro-radiometric reference instrumentation since 2011. Winding the session up, Aron Habte presented on active standard development to formalize the correlation between broadband global horizontal irradiance and two overlapping UV bands to simplify modeling of site-specific equipment UV degradation.

Area 4 - Silicon Material, Feedstock and Wafers: Analysis
Fiacre Rougieux showed in a very good talk the consequences of multiple defect levels of impurities on the correct defect parameter determination. Robert Lee Chin showed how to extract defect parameters in FZ silicon from T-dependent micro-PL spectroscopy. Mallory Jensen revealed the impact of different injection conditions on the LeTID behaviour in Al BSF vs PERC cells. H.S. Rad presented how to process thin silicon foils to solar cells from a technological point of view demonstrated different approaches. Martin Schubert showed how to separate and quantify different losses in epitaxially grown silicon wafers by means of lifetime analyses. Pablo Guimea Coll demonstrated how to spall wafers from an ingot with small roughness via physical understanding of the underlying effects and an effective damping.

Area 8 - Building Integrated PV and PV System Modeling
King presented limitations in current methods of establishing module series resistance and developed a new method to improve accuracy. Stein presented a review of open source PV modeling tools including languages, developing institutions and best practices. Ana Martins presented glass free light weight module with a target of 6kg/m2 and demonstrated meeting the challenges of cell breakage and mechanical stability. Bennet Meyers presented a novel method to extract clear sky signal from measured data. This addresses the issue of determining such a model by other means that has high degree of error. Jorg Palm identified key requirements for BIPV (e.g., shading tolerance, aesthetics, mounting structures, size, color) that go beyond normal PV product and presented new products that meet those requirements. Muhammad Alam presented inputs and constraints of two new tools, one for meteorological data and the other for determining performance gains for bifacial modules.

Area 5 - Characterization of Perovskites
Adrien Bercogel presented perovskite material MA0.17FA0.83Pb(I0.83Br0.17)3 investigated with TRPL mapping and the lifetime fit to a diffusion model to extract the radiative recombination rate, the electron diffusion coefficient, and a trap concentration. The lifetime decreased with the intensity of the photon flux. The addition of 5% Cs at the A site decreased the trap density and increased the open circuit voltage and efficiency. A photograph of the surface showed ridge like features, also seen in a PL map, which were determined by EDS mapping to be rich in Cs with a longer lifetime from TRPL mapping. Steven P. Harvey discussed TOF-SIMS measurements on perovskites with two different A-site compositions: Cs0.25FA0.75PbI3 and Cs0.75FA0.25PbI3. The depth profile for FA was very non-uniform and strongly affected by the A-site admixture of Cs. Harvey cautioned that if the beam energy was too high (>1keV) it could badly distort the depth profile. By rastering the SIMS depth profile it was possible to reconstruct a 3d map of the device with 100 nm lateral resolution. Xueying Li presented nanoprobe X-ray diffraction to investigate several perovskite compounds where the x-ray beam was rastered and a diffraction pattern was imaged at each position. The degradation of the sample due to the accumulated X-ray dose during the measurement was determined by comparing a sequence of diffraction patterns to the initial frame; it was found that the same region could be scanned about 10x before significant X-ray damage occurred; this allows a sufficiently long window to use this method to measure degradation from other environmental stresses before the sample is damaged. Laura E. Mundt showed both dark lock-in thermography imaging (DLIT) and LBIC imaging on two blade-coated perovskite samples (MAPbI3) with different hole transport layers: PEDOT:PSS and NiOx. The NiOx sample had significantly better performance, allowing the characterization methods to be compared. The DLIT images were taken for different voltages and analyzed to obtain a map of JV curves for each image pixel, based on a fitting procedure originally introduced by Breitenstein. The PEDOT:PSS DLIT and LBIC images clearly show more defects than for the NiOx sample, especially along the sample edges. This edge loss is also clear in from the JV parameters reconstructed from the DLIT images. Shijing Sun demonstrated powder X-ray diffraction studies on both MAPbBr3 and FAPbBr3 to determine the increase in crystal strain caused by replacing MA with FA. FA shows higher strain, especially evident from the increase of the FWHM of the peaks as a function of Bragg angle. This increase in strain is also verified by ab initio calculations. Finally in this session, Sarah Wieghold presented mixed perovskite films ( MA0.17FA0.83Pb(I0.83Br0.17)3 ) prepared by spin coating using 3 different precursor solutions: 0.4 M, 0.8 M and 1.2 M. The grain size and film thickness increased with precursor concentration, allowing films with different grain sizes (100 nm, 250 nm, and 500 nm) to be compared. TRPL plots showed that the lifetime increased from 0.29 us to 0.83 us with precursor concentration, and X-ray pole figures show much greater crystal orientation for the larger grains, suggesting that large and uniformly oriented grains improve performance of perovskite films.

Following a re-charge it was on to a further set of oral technical sessions covering work in Areas 2, 3, 4, 6, and 9. Some of the highlights provided by the session chairs are given below:

Area 6 - Advances in Organic Solar Cells
E. Booker used magnetic fields to modulate the recombination processes in OPV via hyperfine field effects that remove degeneracy of CT states. Also, he presented some new results on optical properties dodecylammonium lead iodide perovskites. The PL room-temperature spectra are consistent with bi-exciton model. This is a promising material for optical micro-cavity lasers. Ning Li examined the stability of bulk hetero junction organic PV for various solution processed materials. Degradation mechanism was ascribed to microstructure instabilities in the absorber even in the dark at room temperature for some systems. Lorenzo Torto presented result on drift-diffusion modeling of recombination mechanisms in OPV. The effects of non-constant charge distribution inside the absorber are revealed. Also, novel compact model is presented to do fast simulations.

Area 3 - Hybrid Tandems: Components and Subcells
Bernd Stannowski from HZB (in collaboration with Oxford PV) presented multiple >25% efficient monolithic, two-terminal perovskite/Si tandem cells. These cells utilized a nanocrystalline SiOx layer to both optimize optical coupling and reduce shunting .For GaAsP/Si monolithic cells, Daniel Chmielewski from Ohio State reported on impressive progress in the development of metamorphic tunnel junctions and Daniel Lepkowski (also OSU) presented a rear emitter design for reduced sensitivity to dislocation density. The tunnel junctions work well enough one sun and concentration, and full tandem devices using these designs will be presented by Tyler Grassman tomorrow. Robby Peibst from ISFH presented a silicon bottom cell optimized for two-terminal thin film/Si tandems using a PERC rear structure and a POLO (passivated contact) front tunnel junction and passivation layer. We look forward to seeing these bottom cells implemented in tandems! Tushar Shimpi from Colorado State presented band gap tuning of CdTe using Zn alloying by using a co-sublimation process followed by a CdCl2 passivation treatment. By adding a protection layer to prevent loss of ZnCl, they were able to make a 1.7 eV band gap device for implementation as a top cell. Emily Warren from NREL finished the session with an interesting discussion of the operating principles of 3 terminal cells in tandem arrangements. After taking the audience through the novel way of envisioning performance and some of the possible designs it was shown there is the promise of greater robustness to spectrum changes in this approach.

Area 4 - Device Physics and Simulation
Otto Breitenstein from Max Planck Institute in Halle gave us a sweeping review of spatially resolved methods to map recombination currents in mc Si solar cells and also their relation to structural defects and impurities, as well as their influence on the device IV characteristics. David Payne representing UNSW described a method to use AFM height distributions and haze measurements to describe the reflection and scattering characteristics of black silicon, including detailed modeling. Divya from SERIS introduced a combined workflow using PC1D and Griddler to evaluate the effectiveness of different phosphorous emitters for solar cells, taking into account aspects such as passivation, sheet resistance and metal induced recombination. Po-Sheng Huang from Motech shared with us some industry data on the relationship between regeneration at different temperatures and light intensities. Sven Wasmer from Fraunhofer ISE gave a thorough systematic uncertainty analysis of solar cell simulations and the impact of uncertainties in material parameters from the literature.

Area 4 - Metallization, Contact Formation and Module Integration
S. Jonai presented on the possibility that paste additives can contribute to PID. M. Hilali showed the role of silver particle surface texture on the aspect ratio of screenprinted lines with aspect ratios of 38% being achieved. V. Mihailetchi showed that the thickness of SiN influences the contact resistivity and contact recombinaton of screenprinted contacts. J. Schube demonstrated the contacting of ITO layers of cells with passivating contacts with low temperature silver pastes and phototonic sintering.

Area 8 - PV System Design, Optimization and Performance
Olindo Isabella presented a PV performance modeling approach (Dutch PV Portal 2.0) for BIPV systems that includes curved and custom-shaped arrays, complex shading scenarios. Riccardo Leto presented on work to develop a CPV module with an integrated, internal microtracker that allows the module to be placed in a fixed tilt orientation, and presented the performance of the module at different orientations. Eneko Ortega presented a low cost module-scale monitoring system concept based on using low cost capacitors and switches to measure module IV characteristics that showed good promise for being able to detect multiple types of failures. Kendra Passow presented on First Solar's methods and results for validating the performance of different AR coatings. They found that AR coated modules produced about 2% more energy than uncoated modules in Australia. Steve Ransome presented a detailed performance analysis of PV module characteristics using IV data collected and made available by NREL. He showed that the data was well characterized by the Loss Factors Model and the Mechanistic PV Performance Model. Lin Simpson of NREL presented results of a study that used various desiccant materials to reduce the operating temperature of PV modules. They showed that temperature reductions caused energy increases as much as 5%.

Area 9 – Forecasting
Marc Perez: New SUNY forecast model, blend of various NWP models; validation discussing super smart persistence; blend changes depending on time horizon and sun geometry; EPRI forecast trial. Barry Mather: quantification of temporal solar power variability over large areas; stochastic cloud cover scenarios (downscaling) with many PV interconnections; non-Gaussian mixture model trained with measurements in Oregon; how to model realistic noise for short periods? Dazhi Yang: Review of solar forecasting using text mining. Which problems have not been addressed yet? Forecasts for single systems vs. aggregated systems. Which forecast method to use where? Geographical 2-level hierarchy reconciliation. Temporal hierarchy reconciliation. Sven Killinger: Spatiotemporal aggregation levels, sensitivity analysis using 4 scenarios. 728 PV systems in Germany with data from distribution system operators and TSO. Improvements with increasing portfolio size. Dennis van der Meer: Probabilistic forecasting using probability density function or prediction intervals; using ensembles improves; clear-sky index has a peak probability density around 1. Data from Oahu, HI. Training by separating clear and overcast conditions, can be optimized. Correlation between different sites using spline interpolation. Persistence ensemble as benchmark. Continuous ranked probability skill score for validation against measurements, as a function of temporal resolution and temporal variability. Zibo Dong: Solar forecasting is very challenging in southeast Asia; need for decision tree-based machine learning approach: extremely randomized trees, adaptive boosted trees, gradient boosting trees, deep learning neural networks; all approaches have similar accuracy, but combination (stacked generalization) improves results in Singapore.

After a short break the day finished with the PV Jobs Fair and for others it was off to explore the surrounds to find somewhere to unwind after a long day and talk about all the great work in PV they had seen through the day.