LAKE WALES Fla.—The mystery of why silicon resonators can never get close to their theoretical minimum frequency fluctuation has finally been solved by Leti—Laboratoire d'électronique des technologies de l'information (Leti, Grenoble, France)—a subsidiary of the Commissariat à l’Energie Atomique et aux Energies Alternatives (CEA).
Since the invention of the silicon resonator for MEMS devices, such as the quartz crystal substitutes from SiTime Corp. (Sunnyvale, Calif.), the source of frequency fluctuations has been a mystery attributed to a half-dozen or more causes by the same number of research organizations. Phase noise, jitter, has long been known to be caused by thermal noise, and thus is controllable by stabilizing the temperature, but no one has identified the source of frequency variations, thus limiting the accuracy of the devices.
Frequency stability monocrystalline silicon nanoelectromechanical system (NEMS) resonator used to perform measurements using resonating bar (red arrow) 3.2 microns-by-300 nm-by-160 nm (thick).
The problem has become even more acute since non-timing devices—such as gas sensors—have descended to nanoscale sizes (NEMS) for the purpose of measuring mass and force by how much the frequencies vary. Noise fluctuations for these devices limit their accuracy and accordingly prompted Leti lead a search to solve the mystery of their source.
[ Understanding the IoT Platform Wars: Technical Trade Offs at ESC Boston]
Leti and colleagues used the fictional Sherlock Holmes scientific method of solving the problem—namely by eliminating all the other possible causes, leaving only a single source as the declared solution to the mystery. By surveying the literature and setting up experiments to discount all the other proposed sources of frequency fluctuation of noise, Leti and colleagues have concluded that the only remaining possibility must be the answer—namely mechanical noise.
"We have found that none of the frequency fluctuation explanations suggested so far have explained it," Leti engineer researcher, Sebastien Hentz, told EE Times.
Leti then verified that mechanical noise was the cause by measuring the frequency fluctuations on both sides of the resonant frequency, at which all these devices normally operate, finding that the noise amplitude was exactly the same, independent of frequency.
"Usually you perform your measurements exactly at resonant frequency, but we also used slightly shifted frequencies on both sides of its resonant frequency, concluding that the noise in those signals was exactly the same, which means that all the spectrum's frequency variations, including the resonant frequency, is due to mechanical noise," Hertz told EE Times.
Now the Leti engineers are rolling up their sleeves to find the mechanical source of the frequency fluctuations and how to eliminate, or at least mitigate them. Hentz gave no timeline on when they expect to have an answer, but indicated that they had a number of obvious candidates to test first.
"Our next step is to explore potential sources of mechanical noise," Hentz told us. "One of the first candidates is defects in the crystalline lattice that may move over time."
The experimentally measured frequency stability (green) and the analytically calculated thermo-mechanical limit at a temperature of 300 degrees Kelvin for the frequency determination (orange) are plotted.
If its moving defected is not the cause, and its brainstorming sessions does not identify the source, Leti may have to call in physicists or material scientists to inject new ideas to test. And once identified, will move on to finding a cure, or at least a way of mitigating the mechanical noise source.
According to the mechanical source(s) and possible solutions, it may at that time be possible to raise the frequency of resonators to the gigahertz (GHz) range from the megahertz (MHz) range it is restricted to today, which would be a boon to the electronics industry. Of course, sensors based on resonators would also become more sensitive with a lower noise floor.
Other organizations that helped with the research are the University of Grenoble (France), the California Institute of Technology (CalTech, Pasadena, California), the Indian Institute of Science (Bangalore) and the École Polytechnique Fédérale de Lausanne (Switzerland).
Get all the details in Frequency fluctuations in silicon nanoresonators.
— R. Colin Johnson, Advanced Technology Editor, EE Times