Will printed electronics deliver devices at the price point needed for the IoT?

Does everything being connected mean that every device talks to everything else, or do some things just listen? How much intelligence is distributed and what is controlled centrally? One thing is for sure, it will only become a reality if the cost of the electronics at the extremities can become affordable. The sensors, switches, displays and I/O all need to cost pennies rather than pounds – so silicon may not always fit the bill. New technologies are emerging that could provide the answer. Headline grabbing displays, for example, created with printed/organic electronics are not new, but the technology behind them can be scaled to provide the right price points. Thinfilm describes itself as a technology company that is reinventing itself as a product company. It had developed a polymer technology that could create dense rewritable memories and back in 2005 it partnered with Intel to try and find an alternative to NAND flash by combining its technology with silicon. However, what silicon does best is create dense circuits and the new technology could not improve on silicon performance, but the company has looked since looked at creating other electronic components using polymer technology, not just memories. Davor Sutija, Thinfilm's CEO, commented: "Over the past four or five years, Thinfilm has looked at printed electronics, the idea of taking its memory technology and combining it with printed logic, printed sensors, printed displays to create ultra scalable, in fact disposable electronics that is complementary to what you can achieve with silicon circuitry." The first fruits of these labours have recently been announced. The Smart Sensor Label is a demonstration of closed system created in printed and organic electronics and is a temperature-tracking label, designed for monitoring of perishable goods. Thinfilm has chosen the Smart Sensor label to demonstrate one of the key application areas it sees for this technology. Sutija explained: "Over the last year we have signed four $1billion customers to help us develop and commercialise this technology into distinct product in given markets. One is in toys and games and we have an exclusive consulting agreement with Hasbro. And one is a partnership with Bemis, an innovation leader in packaging. Their ceo believes that one day printed electronics will be on every package that they make – and they make 200 billion. These two show the distinct possibilities of printed electronics – it is not that they are a direct replacement for silicon. They can achieve a unique functionality per cost ratio and therefore make electronic systems ubiquitous and therefore use them in applications where silicon is not well suited, for example disposable packaging, toy and game applications of limited lifetime." It is the cost model that Sutija believes is compelling and opens up these opportunities. If a new silicon fab costs $8 – 10billion it could typically result in a chip costing $3. The nature of the process is to deposit materials at high temperatures and patterning them subtractively - taking away 99% of the material by etching and leaving the pattern behind. Thinfilm uses an additive process. Because it uses organic materials it is done in air at low temperature (typically under 140oC). Capital cost, claims Sutija, are about 1% or 2% of a silicon fab. Although the polymers can be printed using traditional screen or direct printing, the process Thinfilm uses to make its memory films is called slot die coating where an ink is pushed through a very thin slot and then cured. These films have a thickness of about 130nm when cured, or about 1000 atomic layers. The print and cure cycle can be repeated several times. Starting with a substrate, usually the same PET (Polyethylene terephthalate) that is used in Coke bottles, Thinfilm uses a roll-to-roll process up to build up the layers, typically 5 - 7 for memories and 10 for logic. To get to this stage of commercialisation, Sutija believes another industry must take some credit: "You stand on the shoulders of giants. The photovoltaic industry invested $2bn in Thinfilm PV. Machine companies and materials companies have got such performance over the past four years that we now believe there is going to be an inflection point in printed electronics. If it hadn't been for those $2bn of mostly mislaid investment then we wouldn't be in the position we are today to create disposable electronics." Once printed the memories can be rewritten around 100,000 times which Sutija claims: "will open up myriad of applications where instead of using silicon chips you can now use a number of bits of memory - at present the limitation is that we have 20bit tags available and we are now working on passive arrays with 36 and maybe higher bit counts. These are obviously very small number of bits compared to silicon memory, but it is enough to store local information from sensors, store unique IDs that change, for example in game pieces or in storing value, but it is not enough to store a song or photograph." Essentially each bit of information is a ferroelectric capacitor which can either be charged in a high energy state or discharged in the low energy state. It will stay this way unless rewritten. "The memory stays there archiveally, in fact it is more archiveal than NAND flash," said Sutija. "15 years later the information is still there - it is a remarkable material." For standalone memories the power will be supplied by an external reader, but in closed systems, like the Smart Sensor Label, a zinc based battery can also be printed. Thinfilm worked with ACREO for the printable electrochromic displays used in the labels. The cost benefits of this completely printable closed system are stark according to Sutija: "The cheapest silicon based temperature sensors cost $5 – 15, and those are considered disposable. Our initial price point is in the 30 – 50 cent range for temperature sensor systems and we expect that it is going to be between 10 and 25 cents in the next few years. The only other technology available is to use chemistry – that is a colour change to indicate temperature exposure. Those colour change tags also cost 10 to 30 cents but they are analogue – they require someone to read them and make a judgement as to whether the colour has changed. "Because it is diffusion chemistry the colour keeps changing. You can't tell a week later if the person made the right decision – if the vaccine was safe when they gave it to the child - because two weeks later the colour will definitely have changed. In our systems you can measure an electronic value of the temperature exposure - you can read that out and store it. It is stored also in our permanent memory and so you also have a lineage of why decisions were made at any time." Sutija concluded: "With our customers, we are delivering the ubiquitous intelligence necessary to turn the Internet of Things into the Internet of Everything."