Piezo-nanowires boost fingerprint resolution
This stems from earlier research at CEA-Leti, demonstrating the potential of such nanowires to form very sensitive sensors or to collect piezoelectric charges for energy harvesters. In a paper published in 2012, researchers from CEA-Leti were characterizing the sensitivity of vertically grown ZnO nanowires on a silicon substrate, distributed in a pixel configuration through appropriate seeding layer patterns and metallic electrodes.
Fig. 3: Finite element calculations for the PiezoMAT concept using COMSOL Multiphysics: (a) Side view of a microfabricated pixel; (b) 3D representation of the simulated pixel; (c) Meshing and boundary conditions; (d) Profile of the piezoelectric potential generated within the bent nanowire (A) and the collecting electrodes (B) for a nanowire 600nm long with a 25nm radius.
A research engineer at CEA-Leti and one of the co-authors, Emmanuelle Pauliac-Vaujour told us more about the PiezoMAT project.
“We have already proven the high sensitivity of ZnO nanowires for pressure detection, and with this new project, we aim to harness each nanowire as a pressure point (a pixel) for what would be a very high resolution fingerprint sensor”, said Pauliac-Vaujour in an interview with eeNews Europe.
Nowadays, state-of-the-art commercial fingerprint sensors don’t go beyond 1000DPI in resolution (dot per inch), and a resolution of 500DPI is the current international standard for compliance with the U.S. Federal Bureau of Investigation automated fingerprint identification systems. As an example, a 12.8x15mm fingerprint sensor with a resolution of 508DPI translates into 256×300 pixels that are 50µm by 50µm each.
By comparison, using a single ZnO nanowire for each pixel (with a diameter of 0.5µm and 6µm long) could yield fingerprint sensors of pixel sizes well under one micron square, translating into resolutions up to hundreds of times higher, featuring up to a hundred million pixels. Those fairly “large” nanowires would be used because they would be easier to connect to a chip. But ZnO nanowires can be grown with much smaller diameters, it is just a matter of patterning finer nucleation windows in poly(methylmethacrylate)-coated ZnO surfaces.
Fig. 1: SEM image of a highly ordered ZnO nanowire array grown on poly(methylmethacrylate) (PMMA) templates. R. Erdelyi et al., Crystal Growth and Design vol. 11, pp.2515-2519 (2011).
“If we can demonstrate fingerprint sensors based on pixels under a micron square, this will already be a huge competitive advantage” said Pauliac-Vaujour, adding that having French industrial partner Safran Morpho onboard the project had been key to secure European funding. A supplier of authentication solutions and biometrics, the company will help define the specs of the demonstrator for commercial applications. It will also be able to validate the sensor concepts with its own fingerprint capture and analysis algorithms.
“We’ll first explore this new concept on a silicon base, because this is where the read-out electronics will be integrated, and we figured out that if we use a die of similar size to existing CMOS image sensors or capacitive fingerprint sensors, then we’ll be cost-competitive with existing solutions”, explained Pauliac-Vaujour. “Once the concept validated, we should be able to shrink the electronics and the pixel size too” she added.
Fig. 2: The PiezoMAT fingerprint sensor concept showing vertically aligned interconnected ZnO nanowires, each constituting a single pixel of the sensor.
Towards thinner sensors
“Beyond mere high-resolution purposes, the PiezoMAT acquisition technology presents an interest in terms of miniaturisation and in particular of sensor thickness in the case of larger area 4-finger sensors (which are also one of Morpho’s fields of interest)” said Pauliac-Vaujour in a complimentary email exchange. “Moreover, because it is mechanical, it may also be advantageous for imaging very dry or moist fingers, which is problematic with capacitive and optical sensors. This remains to be demonstrated based on initial PiezoMAT’s results”.
The main challenges to be solved with the PIEZOMAT project are encapsulation and robustness. Encapsulation can be tricky since it can directly impact the sensor’s sensitivity and resolution if it averages out the pressure across adjacent pixels.
French company Specific Polymers will contribute to this aspect of the project. While CEA-Leti will provide the silicon substrates, both the Research Centre for Natural Sciences, Hungarian Academy of Science (Budapest, Hungary) and Universität Leipzig (Germany) will help with the growth of ZnO nanowires, following two different approaches.
Both the Kaunas University of Technology (Lithuania) and the Tyndall National Institute (Cork, Ireland) will contribute with multi-physics models at different scales. Another research partner, Fraunhofer IAF (Freiburg, Germany) will characterize each steps of the fabrication process.
What is there more to see above 500DPI?
Surely Safran and other manufacturers are racing for higher pixel densities, as a sales argument. But does this imply that existing 500dpi solutions are prone to errors in fingerprint matching and to what extent would this error rate be reduced?
In other words, what is the relationship between fingerprint matching error rate and pixel density in such sensors?
To that question, Pauliac-Vaujour had a three-fold answer. “The 500dpi international standard was imposed by the FBI and is sufficiently reliable for fingerprint recognition. The question of a higher resolution addresses more specific issues: it increases image quality, which, from a purely visual point of view, will considerably facilitate the work of experts”.
“Obtaining level 3 characteristics of fingerprints (pores, ridge shape, etc.) is only achievable with resolutions beyond 1000dpi and will provide additional information against ID fraud. Indeed, up to date, it is impossible to reproduce pores on a fake finger for example. It is already very difficult to cheat a fingerprint sensor nowadays, but acquiring additional information would increase the level of resistance even further”.
“Finally, some fast-growing populations appear to have fingerprints with similar looking characteristics (one of our experts gave me the example of the Indian population). Additional points of comparison facilitate the differentiation of the individuals in these populations, which today requires several-finger acquisition typically”, she concluded.
Visit CEA-Leti at www.leti.fr/en