I am using an FGMOSFET with tunneling gate and control gate as an analog memory for simulation in SPICE. I use -25V to inject electrons into the floating gate and 25V to remove electrons. Everything was going well until I read this paper:http://citeseerx.ist.psu.edu/viewdoc/download?doi=

and it says

Bidirectional tunneling can be used to add or remove electrons from a floating gate. This solution, however, requires either dual polarity high voltages, or a single polarity high voltage and a means for pulling the floating gate to this voltage when adding electrons, and pulling it near ground when removing them. Both approaches are unattractive. The dual polarity solution has a negative voltage much lower than the substrate potential; the single polarity solution does not support simultaneous memory reading and writing.

With the current technology we have, is this still valid. Are there technologies that allow a higher negative voltage than the substrate. I would really appreciate any help to point me in the right direction. Thanks

  • \$\begingroup\$ You could always tie the N-WELL to GND and then use a MOSCAP to pull negative. You only need 10V to tunnel on 350nm. \$\endgroup\$
    – b degnan
    Dec 13, 2018 at 18:48

1 Answer 1


The basic answer is yes. This is an intrinsic problem of IC design, your substrate has to be biased in a way that reverse-biases all parasitic PN junctions in the circuit. For most processes you have a P-substrate which means that the substrate has to be at the lowest potential in the circuit (generally -Vss). For a flash memory, where you also want a high density of elements, this limitation remains.


  • In modern processes, where oxide thickness is just a few atoms, you can get tunneling going with less than 6V across a gate.
  • There are processes (e.g., trench-isolated) in which you can have isolated islands of substrate that can tolerate voltages in some cases in excess of ±300V.
  • High-voltage options on some processes can tolerate N-wells and P-wells at voltages in excess of 30V above the substrate.

With those considerations, you could segment your flash memory into trench-isolated sections, and drive those sections to a high enough voltage to tolerate positive and negative excursions.

But in most cases you might be better off by using injection processes (high-energy electron scattering at a junction edge) to place the electrons in the floating gate, not just a potential difference to drive them through tunneling.

  • \$\begingroup\$ Thanks for the reply. Can you recommend any books, or articles about this topics. i have also read about triple well technology. Can I use that to isolate a substrate for the FGMOSFET. I'm using an nchannel fgmosfet \$\endgroup\$
    – ElecNoob
    Nov 28, 2018 at 5:26
  • 1
    \$\begingroup\$ @ElecNoob most of what I know about these technologies comes directly from fab datasheets. For an example, look at the XU035 CMOS deep N-well technology from XFAB (xfab.com/technology). Or at their SOI technologies. XU035 does have three well types. Yes, the deep N-well can be driven to a high voltage, while the enclosed P-well can go negative. \$\endgroup\$ Nov 28, 2018 at 5:45
  • \$\begingroup\$ Thanks for the awesome replies and the link. I really appreciate it. :))) \$\endgroup\$
    – ElecNoob
    Nov 28, 2018 at 6:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.