YYS01b-22 Microfabrication of Radio Frequency Filter Chips

Notify of
Inline Feedbacks
View all comments
ZHENG, Zheyang
January 18, 2023 1:56 pm
  1. What is the inductance value of the inductor you are going to make in our NFF?
  2. How many lithography steps you need to form the inductor?
  3. Why “thicker metal” and “suspending structure” lead to high-Q?
SO, Huen Ling
January 18, 2023 4:48 pm
Reply to  ZHENG, Zheyang

Thanks for your question! 

In response, I would like to answer the third question first. First, we learned in our studies that a higher Q tends to mean better filter performance, so we want to achieve a higher Q., And we learned in our references that thicker metals lead to higher Q. We can achieve our desired results by using thick metal plating and suspension structure technology.

We also learned in our references that thicker metals lead to higher Q values. For low or medium frequencies, lowering the Rs (Rs represents series resistance of the planar inductor) increases Q to a large extent. By reducing the Rs, thicker metal plating techniques increase Q to meet the criteria for high Q inductors. For inductors, a lower Rs leads to a higher Q factor in the appropriate frequency range (excluding reactance and resonant frequencies). According to the study’s results, to obtain the lowest RF resistance, the thickness of the metal needs to be at least twice the skin depth (which can be calculated for different frequencies based on the free-space permeability, frequency, and the metal resistivity of different materials).

In the case of reactance or resonant frequency, the effect of lowering the series resistance on increasing Q is not significant. In this case, higher Q factors, higher resonant frequencies, and greater reactance are generally obtained by lowering the parasitic capacitance in the substrate. A suspension structure can achieve the purpose of reducing parasitic capacitance.

For higher operating frequencies above 2 GHz, having lower capacitance can significantly improve the Q-factor and increase resonant frequency. When we read the reference paper, we found that the Q of the case with 0.2Cp, 0.4Cs is higher than that of 1Cp, 1Cs at the same frequency, while the resonant frequency increases from 8GHz to 17GHz (Cs and Cp represent the capacitance between the different windings, and the capacitance between the planar inductors and the ground, respectively). Therefore, obtaining lower parasitic capacitance at high operating frequencies leads to high Q inductance.

Moreover, there are examples of successful resonant frequency increases of 20-turn 125-nH inductors from 800MHz to 3GHz with Q of 4 using the substrate etching technique to make planar inductors. This technique increases the performance of 3.9nH inductors from Q of only 3.9 at 2.1GHz to Q of 17 at 8.6GHz. While recently developed inductors utilize polysilicon material with relatively low resistance. The copper-lined cavity techniques; perform very well and can increase the Q of a 2.7nH inductor to 36 at 5 GHz. New post-processing GaAs wet etching techniques with a series of 8-um GaAs can levitate the entire inductor and use deeper cavities to reduce parasitic capacitance. The Q increases from 11.5 at 6.1GHz to 19.5 at 10GHz.

Furthermore, regarding the second question, our group members have just recently finished taking the NFF exam. Although our process is slow, in the coming days, we will discuss how to sketch the corresponding process flow and discuss this issue in more depth to finish the project faster.

As for the first question, we still need to explore in depth a value of inductance that would be appropriate for our project. We will go into more depth soon. Thank you, Professor, for your questions and for letting us know that we still have many shortcomings and can use this time to optimize our research process.