Researchers report a memristor that both senses water vapor and stores that information as a change in electrical resistance. A memristor is an electronic element whose resistance depends on its past electrical history, in a way often compared with a biological synapse, the adjustable junction between nerve cells. The device uses a sandwich of gold, porous nickel pyrophosphate (Ni2P2O7), and fluorine-doped tin oxide, a common transparent electrode. It is described in a paper published in eScience.
Most artificial synapses need a wired pulse or a physical touch. Here, moisture in the air is enough. In one test, a moist finger held about one millimeter above the surface shifted the device’s conductance. Closer placement produced stronger and longer-lasting changes, which the researchers treat as a simple analog of short-term and long-term memory.
How moisture alters the material
Porous Ni2P2O7 microsheets were grown by a hydrothermal method, a sealed, heated-water synthesis, and checked with X-ray diffraction and electron microscopy. When relative humidity, the amount of water vapor in air compared with the maximum at that temperature, rose from 42 percent to 82 percent, the device could move more charge. High- and low-resistance states stayed distinct over hundreds of cycles and did not fade quickly, which the researchers call non-volatile memory.
Density functional theory, a standard quantum method for calculating how electrons sit in a solid, indicates that adsorbed water molecules narrow the bandgap, the energy gap electrons must cross to conduct, and add hybrid electronic states near the conduction band. That accounts for the rise in conductivity with humidity. Support vector machine and multilayer perceptron classifiers, two routine pattern-recognition algorithms, assigned proximity and humidity states with more than 97 percent accuracy in the reported tests.
The researchers present the device as a single component that perceives humidity and computes on that signal, and they point to possible uses in touch-free interfaces, sterile settings, environmental nodes, and soft robots.