Solving The Production Problems Of MEMS

Sensors are everywhere and MEMS (Micro Electro Mechanical Systems) are made in their billions with the market being worth billions of dollars. However, there is a huge problem for this industry that stops it growing fast to the trillions. Every type of MEMS sensor must be designed and made using a dedicated, production process that can take five to seven years to bring to market. It is also hard to ramp up production apart from constructing a second identical manufacturing line. There are no economies of scale available as every process is unique.

 

This bottleneck is stopping the vision of IoT of having smart sensor nodes everywhere because ramping up to the trillions of nodes that this could have, would be impossible to produce with current MEMS manufacturing techniques. A British electronics company, Nanusens, has solved this problem with a novel way to make MEMS sensors that uses the same standard production technology used to produce virtually all electronic chips that is called CMOS (Complementary Metal-Oxide Semiconductor). It is used in all the major fabs in the world so that there is no limit to the number of sensors that can be produced.

 

Current MEMS devices consist a chip with the moving parts that form the sensor built on the surface of a silicon wafer, and a second chip with all the control electronics. The second chip is easy to make using standard CMOS processes but the first chip, with the tiny moving mechanical parts, is the part that needs proprietary processing and therefore is the limiting factor that prevents rapid increases in production volumes.

 

The Nanusens solution is to make both parts in one chip by shrinking the MEMS part so that it can be formed within the layers of the control chip. The MEMS structure is created with metal layers in the same way that a normal CMOS chip is made. The structure is then released by etching away the surrounding silicon dioxide so that it can move freely.

Nanusens shrink MEMS to create nanosensors within the CMOS layers

 

This sounds simple in theory but, in practice, the released metal MEMS structures distort due to stresses in the metal that are only an issue when the metal is released. This is because the CMOS process was never designed to release metal layers in this fashion. The team at Nanusens has spent years researching and perfecting MEMS structures that are stable when released. Patenting these designs will prevent other from copying Nanusens’ breakthrough of MEMS-within-CMOS.

 

The MEMS structures that Nanusens create are ten times smaller than current MEMS structures effectively taking them from the microscopic realm into the nanoscale making them NEMS (Nano Electro Mechanical Systems). One of these NEMS motion sensor structures measures only 100 by 150 microns and takes up less than ten percent of the chip area with the rest being the control electronics.

 

Initially, Nanusens will be making devices with just one type of sensor on them with the first being a 2D motion sensor. However, as the sensor structures are so small, it is easy to include several different types of sensor structures on the same chip as the way to make them is identical. As a bonus, the control electronics can be reused so that the overall size of the chip only increases slightly as more sensor structures are added.

 

Nanusens thus solves the bottleneck of MEMS growth by using standard CMOS production techniques simultaneously solving the limiting factors of cost (now vast economies of scale are available to reduce prices) and volume production (now they can be made in any fab).

 

In addition, there are other benefits to this new technology of nanosensors. The final packaged product is much smaller than current equivalents. Current MEMS designs require two chips and, when packaged, the resulting volume is 4mm3. The Nanusens NEMS solution is only a single chip so that the package is much smaller at just 1mm3. A saving of 3mm3 can be tremendously significant in applications where space is at a premium which is why the company’s first target market is earbuds.

 

Earbuds have two problems that using nanosensors solves. The first is operational life because the small form factor means that batteries are small. The 3mm3 saving per sensor replaced enables larger batteries to be used for a longer listening experience. Savings are even greater when several sensors are combined into one Nanusens device.

X ray of an earbud show how densely the components are packed within.

 

The other problem is that designers want to put more and more sensors into earbuds but hit the problem of packing them into the already tiny space. Nanosensors remove this barrier so that sensors can be added to improve the operational life of the batteries by enabling the earbuds to detect when they can turn themselves off to save power. For example, a motion sensor detects if they are in the ears whilst a temperature sensor check that they are in the ears and not loose in a pocket.

 

“If MEMS isn't small enough for you then NEMS should be,” says Stuart Robinson, Director, Handset Component Technologies service at Strategy Analytics. “Nanusens has taken MEMS technology to the next smaller level, developing nanoscale sensors inside the CMOS that can be integrated in the same silicon as the sensor controller. This is not a big deal for some applications but it's a game-changer for small devices that have multiple sensors such as accelerometers, gyroscopes, pressure and humidity sensors. Space-saving sensors leaves more room for other components such as a bigger battery. Every cubic millimeter counts in some devices so Nanusens is onto a winner here.” 

Using Nanusens nanosensors can free up space inside earbuds for larger batteries and more sensors.

 

There is another benefit to shrinking MEMS. Current MEMS structures have feature sizes of larger than one micron while NEMS structures are 0.3 micron as they are made with a 0.18-micron process node. This means NEMS structures are better able to withstand impacts, which is important for use in wearable devices that are often dropped. This greater reliability is because the load mass is much smaller in a NEMS than in a MEMS and a smaller mass does less damage under impact condition. The load mass is suspended on springs and any movement of the load mass is detected using changes in capacitance. In addition, smaller structures are less susceptible to short ranges forces such as Van der Vaal’s forces that can hold surfaces together and thus prevent the movement of parts.

 

Nanusens has tested the manufacturability of its technology with many prototype wafers that have been deliberately done in a variety of fabs to prove the fab independence of the process. A nanoswitch has been successfully tested for over 40 million movements to prove that the nanosprings that feature in most NEMS designs are reliable.

 

The first product samples will be for 2D motion sensors with availability in September 2019 with bone conduction sensors later in the year. Because the fundamental structure of the various types of sensors is very similar and the manufacturing process is the same, Nanusens will be able to launch several different sensors in 2020 along with sensor combination in single chip solutions.

 

Founded in 2014 by Dr. Josep Montanyà, Dr. Marc Llamas, and Dr Daniel Fernandez, Nanusens is headquartered in London, England with Research and Development offices in Barcelona, Spain and Shenzen, China. Tel +34 935824466  [email protected]