A team led by researchers at the University of Oxford has demonstrated that the Pb in solar cells based on Pb halide perovskites can be replaced with tin.
Although the amount of lead in each cell is tiny, the presence of a toxic metal could still be a barrier to commercialisation.
“We wanted to try and replace the lead with something similar but non-toxic. Tin has been reported in perovskites before, but not in a solar cell, so we decided to see if it would work,” said Oxford physicist Nakita Noel. “We found that by using tin we managed to keep everything that is good about lead in a solar cell but use a metal that is safe, cheap, and abundant.”
The prototype achieved 6% efficiency, with 20% theoretically possible.
Now tin has replaced Pb, other metals could be found to work in the perovskite structure.
One large disadvantage is that the tin perovskite degrades in the presence of oxygen and moisture. The researchers are confident that further work will enable them to create cells that are stable and can operate in air for long periods of time.
The work is published in the journal Energy & Environmental Science in a paper ‘Lead-free organic-inorganic tin halide perovskites for photovoltaic applications’.
It uses random feedback matrices to process errors update network parameters.
Multi-layer neural networks, inspired by the brain, are used for speech and image recognition within data sets.
Conventionally, these are trained using a ‘training set’ of inputs and expected outputs.
The difference between the expected and actual outputs is fed-back to adjust, and hopefully improve, the connection weightings of each layer.
However, “it would be impossible for the brain to implement the highly complex algorithms currently used to train these deep neural networks on a computer”, said Isis Innovation – the University’s intellectual property licencing company.
Understanding this has led to the feedback alignment algorithm, based on simpler circuitry requirements, and it is said to have had a number of unexpected benefits. For example, networks are trained quicker than through techniques such as ‘back propagation of errors’.
“Feedback alignment is often quicker than existing methods. Novel network dynamics allow learning steps which approximate second order techniques, with no more computation than that required for a first order technique,” said Isis. “The technique is more robust to network initialisations, and is successful even when other algorithms struggle to learn at all. Decoupling the feedback function avoids the central difficulty with training deep neural networks: the ‘vanishing gradient’ problem.”
It is applicable to both feed-forward and recurrent network architectures, on regression and classification problems. Existing neural network tools can be modified to use it.
For hardware-based networks, in digital cameras for example, the Oxford algorithm reduces the error-feedback precision required, said Isis.
The algorithm is the subject of US patent application 61858928.
For more information, contact Isis Innovation.
