Liquid Metal making Smart Objects, New approach to Computer Memory and Predicting Hit records

June 29

This week we examine a new method for stamping liquid metal on paper as a way to create smart objects. We investigate a new approach to computer memory that is not only faster but much less energy intensive. We look at a project that is growing naturally decaffeinated coffee beans and finally we discover a new way to predict hit records.

Liquid Metal making Objects Smart

Liquid metal is used in a range of products including circuits and sensors however it has not been used as a coating. A team from Tsinghua University in China have developed a method of stamping liquid metal onto objects such as paper and making it stick without the use of any adhesive.

Particles in liquid metal stick closely together. This makes it difficult to adhere the metal to any surface without using a glue. The glue usually interferes with the conductivity of the metal. The team used an alloy of bismuth, indium and tin oxide to form this new liquid metal. The BiLnSn compound does not oxidize when exposed to air and is solid at room temperature (melting point of 62C).

To determine the optimum adhesion to paper the team used a trial and error method of varying the amount of pressure applied via the stamp. It turned out that a low amount of pressure delivered the optimal results. Once the metal was stamped and adhered to the paper, the team folded an origami cube out of the metal covered paper. This required the edges to be able to adhere to each other without the use of a glueing agent. When the cube was unfolded, the coated paper could fold itself back into its’ original cubic shape.

The team created other 3D structures out of individual flat pieces of metal covered paper. The structures not only kept their shape, the coating was able to be peeled off without affecting the paper at all. The coating could be recycled and reused repeatedly.

The team is now testing the process with other materials. The hope is to create smart objects and soft robots. Soft robots can fit into tight spaces that other robots are not able to navigate. The team is also investigating coatings that will not peel off once solidified. They envisage packaging boxes that will open themselves, bandages for use on human skin and underwater applications. Soft robots are used to explore the ocean depths where the pressure is too high for humans. The electrical conductivity and thermal capabilities of the materials are also likely to make it a great candidate for the development of robots for space exploration.

New Type of Computer Memory

A team from Cambridge University have developed a new design for computer memory that could greatly improve performance and reduce the energy demands of computing and communications. It has been predicted that within ten years, one third of global energy consumption will be in computing and communications.

The new device processes data in a similar way to the synapses in the human brain. The device uses hafnium oxide and a tiny self assembled barrier which raises and lowers itself to allow electrons to pass. This method of changing the electrical resistance in computer memory and processing information in the same place could lead to memory devices with far greater density, higher performance and lower energy consumption.

The new memory is known as resistive switching memory. Current memory is capable of two states, one or zero (i.e. on or off). Resistive switching memory would be capable of a continuous range of states giving greater density and speed. It is estimated that a standard USB stick would be capable of holding between 10 and 100 times the data.

The barrier that the team had to overcome is the fact that at the atomic level, hafnium oxide has no structure. The oxygen and hafnium atoms mix randomly. By adding barium to thin films of hafnium oxide structures the barium started to form bridges perpendicular to the hafnium oxide plane. The vertical barium rich bridges are highly structured and allow electrons to pass through. At the point where the bridge meets the device an energy barrier is created which allows electrons to cross. By controlling the height of the barrier it changes the electrical resistance of the composite material. This allows multiple states to exist in the material (as opposed to the off and on in conventional memory).

The hafnium oxide composites self assemble at low temperatures (removing the need for increasingly expensive fabrication factories, US$10B+++). The materials show high levels of performance and uniformity making them promising for future memory applications. Hafnium oxide is already used in the semiconductor industry so integration into existing manufacturing processes is not difficult.

Naturally Decaffeinated Coffee Beans

This story fills me with sorrow. As a dedicated coffee drinker I don’t think I can support the removal of caffeine from coffee however we all make our own choices. There have always been concerns with the way that caffeine is removed from coffee to created decaffeinated coffee however for those with caffeine sensitivity it is vital.

Researchers at Instituto Agronomico de Campinas (IAC) in Brazil have been working on a naturally grown decaffeinated coffee bean. Arabica coffee beans have roughly 1.2% caffeine. The main effect of which is to stimulate the human nervous system (on a side note it was allegedly a goat herder in Africa that noticed the increased energy of his goats after eating coffee beans that first tried the coffee bean to see what affect it had on humans, and thus an industry and lifestyle was born, thank you sir!).

For a coffee bean to be decaffeinated it must contain less than 0.1% caffeine. Current common methods of decaffeinating coffee are a mixture of water and methylene chloride or ethyl acetate. Since 2004 the IAC has been carrying out artificial hybridizations where two genetically diverse plants are cross bread. They initially crossed Ethiopian plants with a natural 0.1% caffeine content and elite coffee plants with normal levels of caffeine. Plants in the cross had between 0.1% and 0.3% caffeine.

After a further almost 20 years of breeding and experimentation, the team is able to select breeds with caffeine contents of up to 0.1% caffeine and with commercially acceptable productivity. The goal is to make a new caffeine free breeds available to growers in the State of Sao Paulo in Brazil with the hope of developing a new thriving industry.

Machine Learning identifying Hit Records

There are tens of thousands of songs released everyday. It is so hard to identify what will be a hit and what won’t. Every successful artist has a long story of rejection and being told that their music will never be a hit. Radio stations and streaming services face a similar dilemma. What if we could use machine learning to determine what will be a hit and what won’t?

A team at Claremont Graduate University in California have done exactly that. They measured the neural activity of 33 people to predict if millions would react in the same way to music. Amazingly their algorithm provided a 97% hit rate.

The participants were asked about their preferences and some demographic data before being played a set of 24 songs. The team measured the nerophysiologic responses to the music. These responses are connected to mood and energy levels. This allowed them to predict market outcomes including the number of streams of a song.

The approach is called “neuroforecasting”. Basically it is capturing the neural activity of a small group of people and using it to predict population level effects. After testing a range of algorithms they found that a linear statistical model identified hit songs with a success rate of 69%. When they applied some machine learning to the data the success rate jumped to 97%.

If future wearable neuroscience devices become commonplace, the right entertainment could be sent to audiences based upon their individual neurophysiology. Be careful though, your partner will know exactly what you are thinking.

My main concern with this type of approach is that it is likely to only recommend types of music or entertainment that you have seen or enjoyed before. In a similar way to the Amazon and Netflix algorithms that simply pattern match (really badly in my opinion) what people have watched before. We need to ensure that we don’t destroy the joy of discovery of and access to new favorites and the development of new types of music.

Paying it Forward

If you have a start-up or know of a start-up that has a product ready for market please let me know. I would be happy to have a look and feature the startup in this newsletter. Also if any startups need introductions please get in touch and I will help where I can.

If you have any questions or comments please comment below.

I would also appreciate it if you could forward this newsletter to anyone that you think might be interested.

Till next week.