There is a trend towards roll-to-roll coating for many electronic devices such as displays and photovoltaics. The reasoning is that moving to roll-to-roll manufacturing will be a route to reducing costs. The worry is that not everybody making the decisions fully understands what they are asking and how it does not necessarily match their expectations.
Take photovoltaics for an example and the copper indium gallium diselenide (CIGS) type devices. I have seen schematics of a vision of the process where there is a single unwind roll and rewind roll with all of the different coating steps included between the two rolls. Whilst this looks good on paper it is very difficult to achieve in practice. Firstly there is the mismatch in the coating speeds for each of the process steps. This can be to some extent compensated for by using multiple sources for the slowest of the processes to speed them up to better match the fastest of the processes. However there will be some compromise on the speed. Having multiple sources for the slowest process takes up space and it will add to the complexity of the winding system. For example it may require additional deposition drums, depending on system design, or additional heating or cooling of the web. Whenever there is additional complexity there is the increased possibility of things going wrong and, at least initially, the yield will be reduced. One way to look at the possibility of making a single pass roll-to-roll process work is to look at the yields of each of the separate processes. If you have 5 processes and each one has a 90% yield the cumulative yield after the 5th process will be down to less than 60%.
If we look at the CIGS devices they often have a mixture of processes such as a reactive deposition for an oxide barrier layer, a more precise reactive deposition process for the transparent conductors, various evaporation depositions for the CIGS layers followed by further conducting layers. One of the popular methods of depositing one of the layers on top of the CIGS layer is by a wet chemical process if this were to be included in the single pass single roll-to-roll process it would require the web to be brought out of vacuum to go through the chemical bath before entering another vacuum system for the intrinsic ITO and conducting ITO layers. Bringing the web out of vacuum, passing it through a chemical bath and taking it into another vacuum system is not a trivial exercise and significantly adds to the pumping costs of the vacuum systems. Also often neglected from the description of the process is that many device designs requires laser scribing on up to 3 occasions. Again this process is currently done as a separate process on a separate rewinder and this is usually done as a step and repeat process in order to get the laser scribing to be accurately aligned and each scribed line to be precisely spaced from the other scribed lines. This process would need to be developed to take place on a moving web and at a speed that matches the deposition processes.
In my view too many people who have never run any vacuum deposition process get carried away by the idea of a single pass single roll-to-roll process. They are blind to the reality and see only the economics of the process when it all works well. It is fine to have this vision so long as there is a practical plan to achieve it. Until the laser scribing can be done on a moving web at the same rate as the vacuum deposition processes then it can never be integrated into the process and so it forces the manufacturing to be broken down into separate batch processes. This also allows the manufacturing to be de-bottlenecked by either having multiple systems for the slowest of the deposition processes.
To maintain high yields each process step needs to be improved to maximise the yield for each step. When the yield has been maximised, to some predetermined target yield, then consideration can be given to trying to combine processes into single system and then developing the integrated processes back up to the same maximised yield. This sequential development of integrating different steps together will take time but this planned development takes out the high risk of integrating too many steps together in a single step. This type of development looks to be slow and for investors that are looking for a quick return can appear to be too slow. Anybody who has had to scale up vacuum deposition processes will know that, even for a process that is well developed, it is all too often the case that once scaled up the process needs to be redeveloped or re-optimised. Few processes will be ‘turn-key’ and as more processes are integrated together the complexity and time for optimisation and yield increases hugely. All of this process optimisation will have to be done on a production system that will not be making money during the bulk of this process development and optimisation. Doing the small step development reduces this cost by minimising the final full size production machine optimisation time.
Hence my advice would be not to rush into roll-to-roll processing but to work towards batch processing followed by planned step by step integration and yield maximisation.