UNE Uses AI to Target Hard-to-Recycle Silicon in Solar Panels

Researchers at the University of New England (UNE) are testing whether artificial intelligence can help solve one of the most difficult challenges in solar panel recycling: cleanly separating silicon wafers from the rest of a photovoltaic module.

The project, led through UNE’s Australian Institute for Strategic Artificial Intelligence, is using AI and supercomputing to screen potential solvents that may allow silicon wafers to be separated from their bonded substrate with minimal contamination. The research is at an early stage, but it targets a practical gap in current recycling processes. Standard mechanical recycling can already recover much of a solar panel’s mass, including glass and aluminium, but the silicon wafer is harder to recover in a clean and reusable form because it is tightly bonded within the module to withstand decades of sunlight, heat, moisture and weather exposure.

Why Silicon Recovery Matters

Silicon wafers are central to the operation of most solar panels. They absorb photons from sunlight, release electrons and generate electric current. However, the same durability that allows panels to operate for 25 years or more also makes them difficult to disassemble at the end of life.

Encapsulants and bonding materials protect the cell during operation, but they complicate recycling because they are designed not to break down easily. If the wafer cannot be separated cleanly, the recovered silicon may be contaminated or downgraded, limiting its value and reducing the economic case for recycling.

Recovering silicon more effectively could support a more circular solar supply chain. It could also help reduce dependence on virgin material inputs and improve the lifecycle performance of photovoltaic technology, which is increasingly central to national and corporate net-zero strategies.

How AI is Being Used

UNE’s approach is to replace slow, trial-and-error chemical testing with computational screening. Instead of preparing and testing large numbers of candidate compounds in the laboratory, researchers are using quantum chemical simulations and AI models to predict how molecules might interact with the materials inside a panel.

Promising solvent candidates can then be passed into experimental testing, creating a feedback loop between digital prediction and laboratory observation. This could reduce the time and cost involved in identifying chemicals that are effective enough for recycling but still suitable for safe and scalable use.

The work is being supported by a pairing of UNE’s AI platform and a $2.7 million Australian Research Council-funded automated robotic laboratory shared by several institutions. According to UNE, this combination is intended to accelerate the discovery of recycling pathways by allowing researchers to test computational predictions more efficiently in real-world conditions.

Australia’s Looming Panel Waste Problem

The research comes as Australia faces a growing solar waste issue. The federal Department of Climate Change, Energy, the Environment and Water says about one in three Australian households now has rooftop solar, while businesses are also increasing adoption.

Only a small number of end-of-life panels are currently recycled, with high logistics costs, recycling costs and limited processing capacity holding back wider recovery. By 2035, Australia is expected to generate around one million tonnes of solar panel waste, equivalent to about 50 million panels.

The Australian Government announced $24.7 million in January 2026 for a national solar panel recycling pilot. The pilot aims to collect up to 250,000 panels from around 100 sites and gather data on collection, transport and recycling options, including the cost of moving panels from where they are removed to where they can be processed.

That information is expected to support future policy work on product stewardship for solar panels. A product stewardship approach would place more responsibility on those involved in the manufacture, sale, installation and management of panels to plan for reuse, recycling and disposal at the end of a panel’s life.

A Policy Issue as Well as a Technology Issue

Internationally, solar panel end-of-life management is moving up the policy agenda. Several major markets are developing or strengthening rules for photovoltaic waste, with approaches ranging from extended producer responsibility schemes to incentives for recycling infrastructure.

For the solar industry, the importance of better recycling is both environmental and economic. Panels contain materials such as aluminium, copper, silver and silicon. Recovering these materials can reduce waste, lower demand for virgin resource extraction and improve the circularity of the renewable energy supply chain.

However, commercial viability depends on more than technical recovery rates. Transport, sorting, contamination, labour, energy use and regional processing capacity all affect whether recycling can compete with disposal. Even where valuable materials are present, the business case can be weakened if panels must be moved long distances or processed through expensive, low-volume facilities.

Regional Implications for Renewable Energy Zones

UNE’s location gives the research an added regional dimension. The university is based in northern New South Wales, close to the New England Renewable Energy Zone and major solar development.

ACEN Australia is providing panels from its New England Solar project near Uralla to support the research. The project is being built in stages and is planned as a 720 MW solar and battery development across about 2,000 hectares, with more than 1.5 million solar panels expected to be installed when complete.

That proximity matters. Large-scale solar farms and high rooftop solar penetration will create distributed waste streams over time. Shipping thousands of tonnes of panels over long distances adds cost and emissions, weakening the case for recycling.

Regional recycling capacity, supported by better disassembly technologies, could therefore become important for renewable energy zones, councils, asset owners, waste companies and manufacturers. If panels can be collected, assessed and processed closer to where they are installed, recycling could become more practical and lower-emission.

Limits and Next Steps

The UNE research should not be read as a complete recycling solution on its own. The use of AI to identify solvents is one step in a wider chain that includes collection, testing for reuse, safe handling, mechanical separation, materials purification and market development for recovered outputs.

Any solvent-based process would also need to be assessed for cost, safety, emissions, waste chemistry and scalability. A chemical that works well in a simulation or small laboratory test may not automatically be suitable for large-scale industrial use. It must also meet workplace safety, environmental and waste-management requirements.

However, the project points to a practical direction for the sector. As solar deployment expands, end-of-life planning is becoming part of the net-zero transition rather than a separate waste issue.

Technologies that improve material recovery, especially for valuable and difficult-to-access components such as silicon, could help reduce landfill, strengthen domestic recycling capability and improve the lifecycle performance of solar power. For asset owners, policymakers and recyclers, the research underlines a broader message: clean energy infrastructure also needs a credible end-of-life strategy.

Source: www.sustainabilitymatters.net.au