The amazing non-living materials that have memory, learn and react to the environment

Advanced Materials

Editor of the Technological Innovation Website – 12/06/2022

the incrs

The system based on nickel oxide, which demonstrates learning behaviors and is easily implemented in RRAM memories, has already been used to control a small screen (below).
[Imagem: Sandip Mondal et al. – 10.1002/aisy.202200069]

Materials that imitate life

A real revolution is taking place in a little-known area of ​​research, involving what researchers call non-living materials, ranging from materials that mimic how the brain stores memories to mechanical neural networks that learn and actively react to the environment.

Although they are inspired by the way living beings work, these materials have nothing biological – and, often, not even organic -, but they still behave as if they were capable of learning and repeating behaviors, sometimes in a surprising way. .

See, below, three surveys showing more news in the area, each using a different approach.

habituation and awareness

Sandip Mondal and colleagues at Purdue University, USA, believe it is possible to create a new generation of supercomputers using materials that exhibit behaviors similar to learning.

Mondal applied very short electrical pulses to oxygen-deficient nickel oxide and observed two different electrical responses seen in living beings.

The first response, a kind of habituation, occurs when the material “gets used” to being lightly electrocuted: Although the electrical resistance of the material increases after an initial shock, it soon gets used to the electrical stimulus, and starts to no longer react to he. The other response, a sensitization, occurs when a larger dose of electricity is administered: When the shock is intense, the material’s response grows rather than diminishes over time.

“Virtually all living organisms demonstrate these two characteristics. They really are a fundamental aspect of intelligence,” said Professor Shriram Ramanathan, whose team has developed several types of neuromorphic components, which mimic the behavior of the brain.

And, as the behavior of the material is electrically controlled, it becomes a new alternative for building brain-inspired circuits.

“Being able to manipulate materials in this way allows the hardware to take on some of the responsibility for intelligence,” proposes Ramanathan. “Use quantum properties [o comportamento do material no pode ser explicado pela fsica clssica] to achieve intelligence in hardware represents a fundamental step towards energy-efficient computing.”

the incrs

This “mechanical homeostasis” was used to control a spring that could do useful work.
[Imagem: Hang Zhang et al. – 10.1038/s41565-022-01241-x]

In vivo material with homeostasis

Hang Zhang and colleagues at Aalto University in Finland, meanwhile, have developed a synthetic system that responds to changes in environmental conditions in the same way that living things do.

The system uses a feedback loop to maintain its internal conditions, just as our body uses sweat to maintain our temperature, a process known as homeostasis and which, in nature, is unique to living beings – although the feedback Although important in some artificial systems, such as thermostats, they lack the dynamic adaptability or robustness of homeostatic living systems.

The new system consists of two gels, with different properties, placed side by side. Interactions between the two gels cause the system to respond homeostatically to environmental changes, maintaining its temperature within a narrow range when stimulated by a laser.

The laser is fired through the first gel and then bounces off a mirror onto the second gel, where it heats up the suspended gold nanoparticles. Heat moves from the second gel to the first, raising its temperature. The first gel is only transparent when it is below a specific temperature; when it gets hotter, it becomes opaque. This change prevents the laser from hitting the mirror and heating the second gel. The two gels then cool until the first one is clear again, at which point the laser is again able to pass through and the heating process begins again.

In other words, this laser, gel, and mirror arrangement creates a feedback loop that keeps the gels at a specific temperature.

“The tissues of living organisms are typically soft, elastic and deformable,” said Zhang. “The gels used in our system are similar. They are soft polymers that are swollen in water and can provide a fascinating variety of responses to environmental stimuli.”

“Life-inspired materials offer a new paradigm for dynamic and adaptive materials that are likely to attract researchers in the coming years,” added Professor Olli Ikkala. “Carefully designed systems that mimic some of the basic behaviors of living systems will pave the way for truly intelligent materials and soft interactive robotics.”

the incrs

Memory can be written and erased, just like a computer.
[Imagem: Keim/Medina – 10.1126/sciadv.abo1614]

material with memory

Shape memory materials, which transition between different solid phases, are already well known and have been explored for various uses, including as memories in computer systems.

Nathan Keim and Dani Medina of Pennsylvania State University in the USA have now discovered how to use and erase memories in soft and messy materials, including emulsions, materials with consistencies such as ice cream, mayonnaise, beauty creams, etc.

The duo created a paste by pouring oil over water in a dish, then spreading a tightly packed layer of 25,000 microscopic plastic particles across the boundary between the liquids. The particles are electrostatically charged and therefore repel each other, which allows them to form a smooth, mayonnaise-like solid. This soft solid can be deformed in a controlled manner, and the movement of the particles tracked using a microscope.

“We deform our material by shear, which involves moving one side of the material relative to the other, like pulling the corner of a rectangle to the side so that it becomes a parallelogram,” explained Keim. This type of deformation is known as mechanical annealing, and performing it reduces the overall energy of the structure. Repeating this annealing to the same magnitude many times creates a memory of the strain in the material, which subtly affects how the material responds to strain of other magnitudes in the future.

And you can also erase the memory, just by applying distortions of smaller and smaller magnitudes, which is somewhat reminiscent of the erasure method in ferromagnetic memories, where a strong magnetic field is applied and its direction is alternated, gradually making the field weaker.

According to the team, this could be useful to give neuromorphic capabilities to materials and structures, making it possible to monitor force intensities, remember the stresses to which the structure was subjected or perform failure analyses.

Bibliography:

Article: All-Electric Nonassociative Learning in Nickel Oxide
Authors: Sandip Mondal, Zhen Zhang, ANM Nafiul Islam, Robert Andrawis, Sampath Gamage, Neda Alsadat Aghamiri, Qi Wang, Hua Zhou, Fanny Rodolakis, Richard Tran, Jasleen Kaur, Chi Chen, Shyue Ping Ong, Abhronil Sengupta, Yohannes Abate, Kaushik Roy, Shriram Ramanathan
Magazine: Advanced Intelligent Systems
Vol.: 8, Issue 40
DOI: 10.1002/aisy.202200069

Article: Feedback-controlled hydrogels with homeostatic oscillations and dissipative signal transduction
Authors: Hang Zhang, Hao Zeng, Amanda Eklund, Hongshuang Guo, Arri Priimagi, Olli Ikkala
Magazine: Nature Nanotechnology
DOI: 10.1038/s41565-022-01241-x

Article: Mechanical annealing and memories in a disordered solid
Authors: Nathan C. Keim, Dani Medina
Magazine: Science Advances
DOI: 10.1126/sciadv.abo1614

Follow Technological Innovation Site on Google News

Other news about:

More topics

Source link

About Admin

Check Also

InDriver competes with Uber and 99 as the cheapest transport app

Have you ever needed to call an Uber or 99 driver and the ride price …

Leave a Reply

Your email address will not be published. Required fields are marked *