Fraunhofer Institute for Silicon TechnologyAuthor: Björn Gojdka
The history of microelectronics is largely a history of manufacturing processes. The planar process, photolithography, chemical-mechanical polishing, deep reactive ion etching, wafer bonding — each of these steps made device concepts possible that had been out of reach before. Progress in semiconductor technology has rarely come from new physics alone; it has come from new ways of building things. PowderMEMS, a micromanufacturing technique invented at the Fraunhofer Institute for Silicon Technology ISIT, follows that pattern. It adds the ability to place three-dimensional functional structures inside a substrate, in geometries and made from materials which are challenging for conventional deposition methods.
New freedom for designing micro components
The process sequence is short. Cavities are patterned into a silicon, glass or ceramic substrate with depths of up to several hundred micrometres and placement accuracy in the low micrometre range. These cavities are then filled with dry, micrometre-sized powder — no solvents and no organic binders, so neither viscosity nor trapped air constrain the geometry. Fraunhofer ISIT has integrated the technique into its 200 mm line for MEMS and microelectronic devices with Fraunhofer IFAM contributing its powder-metallurgical expertise to an automated wafer-level filling method. The loose powder bed is subsequently consolidated by atomic layer deposition: Al₂O₃ grows conformally on every individual particle until neighbouring grains connect into a mechanically stable, porous body. Unlike sintering, this happens at temperatures as low as 80 °C. Decisive for many applications, the process temperature is low enough to work on substrates that already carry finished components. PowderMEMS can therefore be inserted before, within or after a MEMS or post CMOS process flow
Why micromagnets are the difficult case
Permanent micromagnets are currently the main focus of development, for a simple reason: they are hard to integrate. Sputtered hard-magnetic films are thin and consequently provide only a small magnetic moment; electroplating is limited in thickness and material choice; sintering would destroy any device already present on the wafer. As a result, permanent magnets in sensor modules are still assembled discretely, adding process steps and package volume. NdFeB micromagnets made by PowderMEMS reach a remanence of about 0.55 T, with an intrinsic coercivity corresponding to the bulk value of the powder. The magnets can be operated between 150 °C and 4.3 K, making them applicable in automotive and cryogenic applications.
Applications from established sensors to quantum hardware
PowderMEMS NdFeB micromagnets were demonstrated to replace the discrete bias magnets of a commercial AMR current sensor. Integrated directly into the sensor chip, the modified devices met the specifications of the discretely biased version, including step response time and offset after overcurrent pulses. Also, back-biased 3D Hall sensors with a total height below 1 mm were demonstrated on contactless gear wheel speed measurement together with Fraunhofer IIS. Compared to discrete assembly, the integrated approach replaces pick-and-place by lithographic alignment, removes the need for magnet assembly and allows smaller packages. Quantum hardware poses a related problem. Nitrogen-vacancy (NV) centres in diamond require a defined static bias field or field gradient, both for magnetometry and for spin-based quantum processors. Because the magnets work at cryogenic temperatures without dissipating power, they are also relevant for cooled architectures. In trapped-ion computing, where qubits can be addressed by magnetic gradients, ion traps on structured glass wafers with integrated micromagnets have been fabricated and handed over to an industrial partner for evaluation
From idea to pilot series
Fraunhofer ISIT continuously advances the PowderMEMS process together with IFAM. Novel micro components are realized with other research organizations and companies, covering simulation and design, prototypes and pilot series on ISIT’s 200 mm line. For higher volumes, the technology can be transferred to the partner’s own fabrication environment under license. Not every technological step forward needs a new material or a new physical effect - sometimes all it takes is finding a better way to manufacture.
