CIRP Annals - Manufacturing Technology 67 (2018) 29–32

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology

jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp

Demanufacturing photovoltaic panels: Comparison of end-of-life

treatment strategies for improved resource recovery
Joost R. Duflou (1)*, Jef R. Peeters, Diego Altamirano, Ellen Bracquene, Wim Dewulf (2)
KU Leuven, Department of Mechanical Engineering, Celestijnenlaan 300A, B-3001 Heverlee, Belgium

A R T I C L E I N F O A B S T R A C T

Article history: Predictive models to forecast the volume and material composition of end-of-life photovoltaic (PV) panels
Available online 1 May 2018 indicate that substantial material resources can potentially be recovered from silicon based PV panels in
the next decades. The technical feasibility of selective mechanical delamination through milling and
Keywords: cleaving was experimentally studied. The achievable material recovery results are compared to current
Recycling practices in end-of-life treatment, demonstrating a substantial potential to improve resource
Optimization preservation. A comparative LCA study allows to conclude that a well-designed demanufacturing
Photovoltaic panels strategy based on selective delamination can substantially reduce the environmental impact associated
with end-of-life processing of PV panels. The improved silver recovery offers perspectives for the
economic viability of the described demanufacturing strategy.
© 2018 Published by Elsevier Ltd on behalf of CIRP.

1. Introduction the strategic importance of the respective material fractions and

1.1. End-of-life treatment needs for photovoltaic systems
As part of the trend towards renewable energy generation,
considerable investments have been made in photovoltaic (PV)
systems over the past decades, both by enterprises and private
households. Based on the expected functional lifetime of such
installations, several authors have predicted the volumes of PV
panels that will require end-of-life (EOL) treatment in the years to
come [1–3]. Using the distribution delay forecasting method, and
taking into account the technological evolution of PV panels in
terms of material composition, Peeters et al. [4] developed a
predictive model that allows to anticipate the different material
mass flows in EOL PV streams. Using the statistics of recently
collected EOL PV panels for the example of the Flemish region in
Belgium to calibrate this model, an EOL mass stream of crystalline
silicon (c-Si) based PV panels increasing to 23,000 ton per annum,
equal to 3.4 kg per capita, is anticipated.
While other technologies are visible in the PV market today, c-Si
based panels have been dominant and are expected to dominate
the EOL treatment scenarios for the coming decades. The results of
the predictive model of Peeters et al. [4], as summarized in Fig. 1,
are prone to some uncertainty due to the high sensitivity of the
predictions to the material composition, the anticipated failure
rates and the future material prices. However, some major
conclusions can still be drawn from this study with respect to
* Corresponding author.
E-mail address: joost.duflou@kuleuven.be (R. Duflou).
their embodied economic and environmental importance. As
visible in Fig. 1b and c, the relevance of the silver, aluminium and
glass fractions must be taken into account when determining well-
designed methods for EOL PV panel treatment.
Looking at the c-Si PV panel architecture that will dominate the
EOL treatment activities in the coming two decades, the layer
structure depicted in Fig. 2 is rather constant: embedded between
two layers of Ethylene Vinyl Acetate (EVA), the silicon solar cells
are coated with metallization paste containing silver and
aluminium, and interconnected by silver coated copper bus bars.
A layer of antimony containing, low Fe hardened glass guarantees
long term physical protection and optimal light transmission. To
the non-illuminated side a polymer layer, typically PET based,
completes the sandwich structure. The PV panels are often finished
and mounted by means of an aluminium frame, and a junction box
glued to the backside contains the electronic components to
connect the solar cell chains to the PV network.
1.2. State-of-the art in EOL treatment of silicon based PV panels
Driven by the recast Waste of Electric and Electronic Equipment
(WEEE) Directive 2012/19, in the EU EOL c-Si PV panels are
currently typically processed in general recycling plants, together
with other WEEE or laminated glass [5,6]. The materials are
separated by means of size-reduction with a hammer mill or
shredder, followed by magnetic and eddy-current separation. Only
the glass, aluminium, copper and steel fractions are recovered,
while the solar cells and plastics containing residue is incinerated
or landfilled. The resulting secondary material quality is rather low
due to impurities, limited separation efficiencies and presence of

https://doi.org/10.1016/j.cirp.2018.04.053
0007-8506/© 2018 Published by Elsevier Ltd on behalf of CIRP.
30 J.R. Duflou et al. / CIRP Annals - Manufacturing Technology 67 (2018) 29–32

Fig. 1. Predicted mass volumes for EOL silicon based PV panels for the example of the Flemish region of Belgium (a), potential economic value at 2017 material prices for
assumed cost free recovery without quality losses (b) and corresponding maximum potentially avoidable environmental impact by closed loop material retrieval for assumed
impact free recovery replacing virgin material production (c).

Fig. 2. Typical layer structure of silicon based PV panel with indication of common Fig. 3. Baseline scenario: current industrial practice PV panel processing in WEEE
dimensions (varying between manufacturers and types). treatment plant.

hazardous compounds [7]. In terms of mass recovery rate this The glass fraction recovered from this baseline scenario is
treatment method allows to meet the WEEE directive require- highly contaminated with uncontrolled impurities (almost 10% of
ments, but results in low value retrieval, and critical materials, the glass fraction mass), limiting the useful applications for this
such as silver and antimony, as well as the silicon metal are fraction and preventing valorisation of the embedded antimony.
typically not valorised [8]. The glass recovered from c-Si panels is The copper recovery in the baseline scenario is limited (34,7%) due
still heavily contaminated with silicon, polymers and metals. to imperfect separation and liberation. The silver resources are lost
However, it can be blended with other recycled glass and used as since the recovery is not considered economically viable for the
thermal insulating material in the glass-foam or glass-fibre expected concentration present in this fraction (see Table 1).
industries [3].
Table 1
Besides the conventional mechanical recycling method, various PV waste material composition.
strategies have been explored to achieve the delamination of the
so-called ‘PV-sandwich’, including pyrolysis [9,10] and the use of Panel part Material fraction % kg/tonne of
PV waste
organic solvents [11,12] to respectively thermally break down or
chemically dissolve the EVA encapsulant. A number of patents Frame Aluminium 14.7% 147.0
Steel 8.65% 86.5
have also been filed with respect to chemical end-processing of the
solar cells to recover the silver [13]. Komoto et al. [13] also Junction box & cable Copper 1.90% 19.0
identified a number of controlled mechanical separation techni- Plastic 2.85% 28.5
ques that are comparable to the processing methods described in Encapsulant Ethylene-vinyl 4.52% 45.2
Section 2.3 of this article. acetate (EVA)

Backsheet Polyethylene 1.91% 19.1

1.3. Objectives and scope terephthalate (PET)

Front glass Glass 59.51% 595.1

The diversity of proposed EOL processing methods for c-Si PV
panels, resulting in a wide range of material recovery rates and Solar cell Silicon 1.82% 18.2
Aluminium 2.01% 20.1
purity levels, requires a detailed comparative assessment in order
Copper 1.99% 19.9
to determine the most advisable EOL treatment procedure. In this Silver 0.12% 1.24
study both data as reported in literature and experimentally
determined material recovery results have been used for this
purpose. 2.2. Thermal and chemical treatment scenario

2. Comparative study: scenario specifications
Three principal treatment methods have been compared: the
process steps included in the respective procedures are summa-
rized in the following subsections.
This scenario, as described in detail by Park et al. [14] and
summarized in Fig. 4, consists of a pyrolysis treatment at around
400  C to break down the polymers (EVA and back sheet) and

2.1. Baseline scenario: destructive separation

The commonly used separation procedure, as applied in most

recycling plants today, is summarized in Fig. 3. No preliminary
removal of the junction box or aluminium frame is considered. A
material flow analysis was conducted for this scenario by sampling
experiments at a WEEE treatment plant in the Flemish region of
Belgium. Fig. 4. Thermal and chemical treatment scenario [14]: overview.
 J.R. Duflou et al. / CIRP Annals - Manufacturing Technology 67 (2018) 29–32 31

release the PV cells and glass layer. Further chemical treatment by
means of HNO3 allows to recover the metals from the PV cells (Cu
and Ag). The applied recovery rates for Ag and Cu in this process are
according to Huang et al. [7].
2.3. Delamination scenario: selective mechanical separation
In this scenario, after manual removal of the junction boxes, the
PV panel is treated by means of a cutting process that separates the
glass fraction from the remainder of the sandwich structure. The
aluminium frame is removed after this delamination step. The
solar cells and polymer containing fraction serves as input for an
integrated precious metal smelter-refinery, in which the polymer
content is thermally valorised. While assuring a yield for Ag and Cu
of more than 95%, the aluminium used in the internal connections
(metallization paste) is not recovered in this process.
For the delamination purpose several strategies were experi-
mentally tested. Milling proved rather impractical: the limited
productivity, high tool wear and noise levels, and the difficulty to
systematically collect the resulting polymer and metal containing
swarf formed reasons not to withhold this solution. Cleaving by
means of a chisel, clamped in a flexibly mounted support structure
and guided by the glass panel (Fig. 5), proved to allow systematic
removal of the EVA, solar cells and back sheet layers with limited
EVA residues left on the glass surface. Operating at elevated
temperatures allows to reduce the cleaving force and increase the
removal rate for a given force level (Fig. 6). The experimental setup
was tested to achieve a throughput of 327 kg/h, whereas
opportunities to double the throughput were identified, opening
perspectives for an industrial scale treatment line.
using system expansion as prescribed by ISO 14040-2006 in order
to assure comparability with respect to the achieved material
output. Each scenario thus provides the same amount of materials
or ‘basket of products’. If not recovered from secondary resources,
primary production is assumed for this purpose [16].
The functional unit used is 1 kg of PV waste treatment. The
ReCiPe midpoint method with European dataset was used to
convert the LCI data to environmental effects.
3.2. Material recovery data
The material recovery efficiencies for the respective treatment
scenarios were either obtained from literature (thermal and
chemical treatment) or experimentally determined by analysing
the material composition of the different obtained fractions. The
efficiencies of the processes, as used in the different scenarios
described above, allow to determine the expected yields for a given
waste composition.
The assumed waste composition, as shown in Table 1, was
experimentally determined by performing a material flow analysis
of a shredder based process in which 12 tonne of PV waste was
processed. The silver content is based on the average waste
forecast prediction for 2020 [4].
Table 2 contains the expected output quantities per tonne of
EOL PV panels processed according to the different scenarios.
Taking the data of Table 2 and the quality difference between
the glass fractions obtained in the baseline scenario and the other
strategies into account, the basket of output products used for the
LCA study was set to: solar glass: 583.23 kg; low grade glass:
532.94 kg; aluminium: 144.10 kg; steel: 84.77 kg; copper: 36.95 kg
and silver: 1.18 kg per tonne of processed PV waste.

Table 2
Material recovery yields for the considered treatment scenarios based on
observations at an industrial treatment plant (Baseline), Ref. [7] (Thermal and
chemical) and industrial data (Delamination).

Material Baseline scenario Thermal and Delamination

Fig. 5. Experimental, robot mountable, self-aligning cleaving head. fraction chemical scenario scenario

kg recov./ % recov. kg recov./ % recov. kg recov./ % recov.

Tonne PV (kg recov./ Tonne PV (kg recov./ Tonne PV (kg recov./
waste kg input) waste kg input) waste kg input)

Glass 532.94 89.6% 583.23 98% 583.23 98%

Aluminium 130.45 78.1% 144.10 86% 144.10 86%
Steel 80.14 92.7% 84.77 98% 84.77 98%
Copper 13.49 34.7% 33.06 85% 36.95 95%
Silver 0.917 74% 1.18 95%

3.3. LCA results and analysis

For the process combinations and material flows as summa-

Fig. 6. Observed cleaving force levels in the feed direction at different temperatures
for a 75 mm wide chisel mounted in the setup described in Fig. 5.
rized in the previous section, the comparative assessment results
for the different midpoint impact categories are shown in Fig. 7.

The purity of the glass fraction recovered in this delamination

scenario was found to be suitable for closed loop recycling, similar
to the thermal and chemical treatment scenario. An optical sorting
step has been anticipated in the delamination scenario as a quality
assurance measure for contamination avoidance. LCI data for this
process were obtained from Ref. [15].

3. Comparative assessment

3.1. Methodology

The comparative environmental impact assessment of the EOL

treatment procedures was conducted using the Simapro software
with the Ecoinvent 3.3 database. The different considered
scenarios were modelled following an attributional approach, Fig. 7. Relative impact comparison for the different EOL treatments.
32 J.R. Duflou et al. / CIRP Annals - Manufacturing Technology 67 (2018) 29–32
While for many impact categories the delamination scenario
proves to be a promising alternative, especially in terms of avoided
metal depletion its effectiveness is apparent.
As can be seen in Fig. 8, the main impact contribution for the
delamination scenario is linked to the metal recovery process,
which is energy intensive and in which the polymer fraction is
sacrificed for caloric recovery. The small amount of aluminium
used for the internal connection/metallization paste is not
recovered in this process. The slags can be used as secondary
resource for building materials, but this is not considered in the
current study.
Fig. 8. Contributions by processing steps to the different impact categories for the
delamination EOL treatment scenario.
While in the studied delamination scenario the impact
associated with material substitution (avoided material losses or
material degradation) is further reduced compared to the thermal
and chemical treatment scenario, the impact of the separation
process is also substantially lowered and is of the same order as the
currently used baseline procedure (Fig. 9). The avoided pyrolysis is
the major cause for this reduction. In the delamination scenario the
main cause for impact consists of the metal recovery procedure
following the material separation stage.
Fig. 9. Endpoint comparison per kg of PV panel treated for the different EOL
scenarios according to ReCiPe H/A(EU).
3.4. Economic considerations
While exact financial returns of the delamination scenario are
highly sensitive to the variable material prices and would require a
detailed investment and operational cost analysis, a comparison of
the material yield allows a crude assessment of the economic
relevance. Since the silver content represents the major economic
potential (see Fig. 1B), the expected silver yield improvement of
1.18 kg/tonne of processed EOL PV panels compared to the baseline
scenario, which at current silver prices corresponds to 543.7s/
tonne, forms a solid basis for an investment in an automated line
for selective cleaving of the PV panels, provided a sufficiently large
logistic supply of EOL panels can be organised. In the long term the
reduced silver and copper content in more recently produced
panels [4] will, however, negatively affect the profitability of such
an investment.
4. Conclusions
Maximizing the silver and copper recovery and preserving the
purity of the low Fe glass fraction form strategies that lead to
substantial impact avoidance. Mechanical cleaving, possibly
assisted by a temperature increase of the EVA layer, allows to
separate fractions that can be post-processed for metal retrieval
and low Fe glass recuperation respectively. The aluminium frame
recycling represents a third important material stream that
contributes to the profitability of EOL demanufacturing of PV
panels. While currently the volumes of PV panels offered for EOL
treatment are still limited, it can be expected that within the next
decade EOL processing of PV panels will become an industrially
relevant activity for which advisable treatment methods have been
identified and validated in terms of technical feasibility.
Acknowledgements
The authors acknowledge the Flemish Agency for Innovation
and Entrepreneurship (VLAIO) and the Flemish Environmental
Technology Platform (MIP) for funding this research as part of the
ReSolar project. Galloo, Umicore, Novolpolymers, Soltech, PV
Vlaanderen, OVAM and PV Cycle are acknowledged for their
contributions as partners in this project.
References
[1] Wambach K (2004) Recycling of Solar Cells and Photovoltaic Modules. Pro-
ceedings. 19th European Photovoltaic Solar Energy Conference, Munich.
[2] Paiano A (2015) Photovoltaic Waste Assessment in Italy. Renewable and Sus-
tainable Energy Reviews 41:99–112.
[3] Weckend S, Wade A, Heath G (2016) End-of-Life Management: Solar Photovoltaic
Panels, International Renewable Energy Agency and International Energy
Agency Photovoltaic Power Systems.
[4] Peeters JR, Altamirano D, Dewulf W, Duflou JR (2017) Forecasting the Compo-
sition of Emerging Waste Streams with Sensitivity Analysis: A Case Study
for Photovoltaic Panels in Flanders. Resources Conservation and Recycling
120:14–26.
[5] Pennington D, Latunussa C, Mancini L, Blengini G, Ardente F (2016) Analysis of
Material Recovery from Silicon Photovoltaic Panels, Publications Office of the
European Union EUR 27797, Luxembourg. http://dx.doi.org/10.2788/786252.
[6] Wambach K, Sander K (2015) Perspectives on Management of End-of-Life
Photovoltaic Modules. 31st European Photovoltaic Solar Energy Conference
3073–3078.
[7] Huang B, Juan Z, Jingyang C, Bing X, Feng Z, Xiangyu W (2017) Environmental
Influence Assessment of China’s Multi-Crystalline Silicon Photovoltaic Mod-
ules Considering Recycling Process. Solar Energy 143:132–141.
[8] Stolz P, Frischknecht R (2016) Life Cycle Assessment of Photovoltaic Module
Recycling, Swiss Federal Office of Energy SFOE report 174-LCA-Recycling-PV-
Modules-v1.1.
[9] Zeng DW, Born M, Wambach K (2004) Pyrolysis of EVA and its Application in
Recycling of Photovoltaic Modules. Journal of Environmental Sciences 16
(6):889–893.
[10] Wang T-Y, Hsiao J-C, Du C-H (2012) Recycling of Materials from Silicon Base
Solar Cell Module. Proceedings 38th IEEE Photovoltaic Specialists Conference
(PVSC).
[11] Kang S, Yoo S, Lee J, Boo B, Ryu H (2012) Experimental Investigations for
Recycling of Silicon and Glass from Waste Photovoltaic Modules. Renewable
Energy 47(November):152–159.
[12] Kim Y, Lee J (2012) Dissolution of Ethylene Vinyl Acetate in Crystalline Silicon
PV Modules Using Ultrasonic Irradiation and Organic Solvent. Solar Energy
Materials and Solar Cells 98(March):317–322.
[13] Komoto K, Lee JS, Wade A, Heath G (2017) Technology Trends in PV Module
Recycling from Viewpoints of Patents and R&D Projects. Proceedings 33rd
European Photovoltaic Solar Energy Conference 1515–1518.
[14] Park J, Kim W, Cho N, Lee H, Park N (2016) An Eco-Friendly Method for
Reclaimed Silicon Wafers from a Photovoltaic Module: From Separation to
Cell Fabrication. Green Chemistry 18(6):1706–1714.
[15] Latunussa CEL, Ardente F, Blengini GA, Mancini L (2016) Life Cycle Assessment
of an Innovative Recycling Process for Crystalline Silicon Photovoltaic Panels.
Life Cycle Environmental Ecology and Impact Analysis of Solar Technology 156
(November):101–111.
[16] Wäger PA, Hischier R (2015) Life Cycle Assessment of Post-Consumer Plastics
Production from Waste Electrical and Electronic Equipment Treatment Resi-
dues in a Central European Plastics Recycling Plant. Science of The Total
Environment 529:158–167.