Life Cycle Aspects of Nanoproducts, Nanostructured Materials, and Nanomanufacturing: Problem Definitions, Data Gaps, and Research Needs

Chicago, Illinois: November 5-6, 2009

Comparative LCA of Nanotechnologies in Thin-Film PV Devices

Hyung Chul Kim, Ph.D., Associate Research Scientist, Center for Life Cycle Analysis Dept. of Earth and Environmental Engineering, Columbia University

Introduction
Increasing attention to the potential effects of nanotechnology on human- and ecological-health triggered the recent surge of life-cycle assessment (LCA) studies of, for example, carbon nanofibres (CNFs) and polymer nanocomposites. Producing nanostructures tends to be more energy intensive per unit mass than fabricating conventional structures.  For instance, from cradle- to- grave, CNFs require 13-50 times more primary energy per mass basis than does primary aluminum (Khanna and Bakshi 2009).  However, if their usage phase is accounted for, the nanoparticle-based composites may achieve a net energy savings over their entire life cycle compared with conventional materials since lesser amounts can be used to meet their materials properties, like stiffness and strength (Roes et al 2007).  Overall, our current understanding of the energy- and emissions-implications of these products carries a large degree of uncertainty due to the limited data available.  There will be paramount benefits in proactively assessing the potential environmental risks during the life cycle of a nanotechnology before it fully matures.  We present here the projected energy use and greenhouse gas emissions from the life cycles of two nanotechnology-based thin film Photovoltaic (PV) technologies under R&D, nanoparticle-based CdTe and a-Si with nanocrystalline Si bottom layer.       

Cadmium Telluride Solar Cells
Currently, thin film CdTe is a recognized leader of contemporary PVs in terms of cost and energy payback.  A module efficiency of 10.8% and manufacturing cost of less than $1/Wp was recently reported (First Solar 2009).  The life cycle of the current CdTe PV largely consists of material synthesis, film deposition, and device fabrication.  Fthenakis and Kim (2006) evaluated the life-cycle of CdTe PV manufactured using micro-particle-based CdTe film which laid out by vapor transport deposition, reporting a primary energy demand of 1200 MJ per m2 of module, which corresponds to an energy payback-time (EPBT) of 0.75 years under the average US solar-irradiation of 1800 kWh/m2/yr.  A possible spin-off of the current CdTe PV is manufactured from colloidal semiconductor nanocrystals prepared in a solution of phosphorus compounds.  Several research groups (Gur et al 2005) employ such rod-shape CdTe nanocrystals and conducting polymers to produce solar cells by inexpensive processing techniques expectedly ink-jet printing.  The electricity-conversion efficiency of this device under R&D remains still very low, ~2.9%.  

We compared the materials- and primary energy-demand of micro- and nano-particle-based CdTe PVs from cradle to gate, i.e., from raw material production to module manufacturing.  Under the current laboratory process conditions, the solvents (i.e. toluene, isopropanol, hexane, and pyridine) and phosphorus compounds used in growing and purifying the nanoparticles dominate the primary energy demands, accounting for 99.9% of the total of 41,000 MJp/m2.  However, the full-scale commercial production of nanopowder will undoubtedly be more efficient than the laboratory process.  A prospective analysis has been conducted based on literature and industry experience.  With significantly less solvent usages as observed in the chemical- and pharmaceutical-industries (Bisio and Kabel 1985), and with highly efficient material usages of inkjet printing in comparison to the spin coating used in R&D (i.e.,99% vs. 1%), the primary energy to manufacture nanoparticle-based CdTe could potentially drop to ~500 MJp/m2 excluding encapsulation (e.g. glass).  Commercial producers might adopt recycling or energy recovery of the used solvents, which could further lower energy use. Future follow-up studies could identify options for solvent use and recycling during the synthesis of nanopowders, thus yielding better estimates.

Multi-junction Amorphous Silicon Solar Cells
Since a-Si PV can be deposited on stainless steel as well as glass, it is perfect for building integrated applications.  Compared to the bulk c-Si PVs, while fabricating thin film a-Si is simple and inexpensive, the conversion efficiency of single layer a-Si is relatively low.  Commercial modules stack multiple layers of such films to form multi-junction PVs wherein each layer coverts a different spectrum of sunlight.  The current a-Si is in the form of triple junction a-Si/a-SiGe/a-SiGe, in which Ge is alloyed with Si during deposition to control the bandgap of the layers; the efficiencies of commercial triple-junction cells are about 6.3-7% (United Solar 2009).  Intensive development is underway on nanocrystalline-Si (nc-Si) films that might replace the a-SiGe bottom cell, as the former has a better stability upon exposure to sunlight than the latter, thus increasing the conversion efficiency.  However, since nc-Si film is relatively thick (i.e. 1-3 μm) due to the low absorption coefficient (Wronski et al 2008), laying out such a layer at the current deposition rate, i.e., 1-3 Å/s requires a long deposition time, limiting the throughput of equipment and consuming a large amount of energy.  To increase the rate, nc-Si thin film is often laid out by Very High Frequency (VHF) plasma-enhanced chemical vapor deposition (PECVD), instead of the common PECVD.  The deposition rate of nc-Si layer in R&D modules currently is at 5-8 Å/s.

We assessed three prospective combinations of hybrid amorphous- and crystalline-silicon cells.  The first configuration is a tandem junction that consists of a-Si top layer and nc-Si bottom layer, i.e., a-Si/nc-Si the second configuration is a triple junction with nc-Si as the bottom layer, i.e., a-Si/a-SiGe/nc-Si; and, lastly, a triple junction with two layers of nc-Si, i.e., a-Si/nc-Si/nc-Si.  The life cycle energy demand to deposit nc-Si was estimated from parametric analyses of film thickness, deposition rate, precursor gas usage and power for generating gas plasma. 

We found that extended deposition time and increased gas usages associated to the relatively high thickness of nc-Si lead to a comparable or larger primary energy demand for the nc-Si bottom layer designs compared with the current triple junction a-Si.  a-Si/nc-Si module requires a comparable amount of energy, ranging from 750-1270 MJp/m2, as the current triple-junction module that requires 860 MJp/m2.  In contrast, the deposition of a-Si/a-SiGe/nc-Si and a-Si/nc-Si/nc-Si-modules takes significantly more energy, ranging from 800-1390, and 950-1510 MJp/m2, respectively. In addition, our review shows that the conversion efficiency of nc-Si bottom layer designs has not yet surpassed that of the current triple junction a-Si design.  The combined effect will cause the former to have a longer by 20-30% energy payback time (EPBT) than the currently commercial option.  However, future scenario analyses show that if nc-Si film is deposited at a higher rate, (i.e. 2-3 nm/s), and at the same time the conversion efficiency reaches 10%, the EPBT could drop by 50%. Besides PECVD, other film-deposition technologies might produce nc-Si at an even higher rate, although the deposition quality is not satisfactory yet.  A timely update of this analysis will be needed in the future if these new technologies dominate.