(Invited) Visualization of Conductive Filament of ReRAM during Resistive Switching by In-Situ TEM

Wednesday, October 14, 2015: 14:40
103-B (Phoenix Convention Center)
Y. Takahashi, M. Kudo (Kyusyu University), and M. Arita (Hokkaido University)
Resistive random access memories (RRAMs) have great potential as a candidate for next-generation nonvolatile memories. A 16Gb Cu-based ReRAM chip with 180MB/s write and 900MB/s read performance fabricated at the 27nm node has been demonstrated [1-3]. The remaining issue is to clarify the switching mechanism to understand the degradation mechanisms of switching and to guarantee the reliability. To overcome the problem, we employed in-situ transmission electron microscopy (TEM) and attained reproducible resistive switching. Direct observation of formation and rupture of a Cu filament were also achieved corresponding to repeatable switching of SET (high resistance to low) and RESET (low resistance to high) characteristics, respectively. The most important point is that the achieved electrical characteristics in the in-situ TEM are almost the same with those measured in the real ReRAM devices.

In this work, we used two types of Cu-based ReRAM (CBRAM: Conductive Bridge RAM) memories for in-situ TEM observation. One has Cu/MoOx structure, where MoOx is used as a solid electrolyte. The TEM sample was prepared using the ion-shadow method [4], which is an ion milling technique with carbon mask particles. This method is simple and effective to make thin samples for TEM observation because many cone-shaped needles with miniaturized ReRAM devices were formed. The other is Cu-Te based ReRAM that has a structure similar to previously reported cells [1-3]. The 30- or 70-nm Cu-Te ReRAM cell was thinned by conventional FIB techniques to 100 nm thick for TEM observation [5]. The method is suitable to cut up the small ReRAM devices from a patterned memory chip.

In-situ TEM observation was performed with a homemade TEM sample holder with a piezo actuator that was used to control a Pt-Ir probe needle so as to contact the needle to the top electrode of the ReRAM to be measured. Near the TEM specimen, a MOSFET was serially connected to control the current compliance. Together with the I-V characteristics, real-time TEM videos were also collected by the CCD Camera.

The clear formation and rupture of Cu filament were observed in TEM corresponding to SET and RESET resistive switching for the Cu/MoOx samples. The size of the filament increased as the increase of compliance current. An interesting results is that the positions of the filament changed. The phenomena was thought to be caused by the over-reset where the filament seeds were completely removed.

In the Cu-Te based sample, filament formation was not detected at low current compliance though the SET and RESET resistive switching were clearly observed. When the current compliance increased higher than 100 mA, filament formation corresponding to the SET process was detected in the TEM images by enhancing the contrast. In this device, the filament formed almost the same position. In addition, we achieved 100k pulse switching cycles of SET and RESET inside the TEM without any damage.

These results clearly show that the in-situ TEM will be a powerful tool to analyze the mechanisms and the reliability of ReRAM.


The authors wish to thank Drs. K. Ohba, M. Shimuta, I. Fujiwara, S. Yasuda, A. Maesaka, S. Kusanagi, and K. Aratani from SONY Corporation, and Drs. T. Fujii, K. Hamada, and N. Sakaguchi from Hokkaido University for their technical supports and meaningful discussions. This work was partly supported by Grants-in-Aid from the Japan Society for the Promotion of Science (JSPS KAKENHI 24360128, 25420279, and 2663014104) and the Nanotechnology Platform Program organized by MEXT.


[1] W. Fackenthal, et al.: “A 16Gb ReRAM with 200MB/s Write and 1GB/s Read in 27nm Technology,” ISSCC, p.338 (2014).

[2] S. Sills, et al.: “A Copper ReRAM Cell for Storage Class Memory Applications,” VLSI Symposium, p. 80 (2014).

[3] J. Zahurak, et al.: “Process Integration of a 27nm, 16Gb Cu ReRAM,” IEDM, p.140 (2014).

[4] M. Kudo, M. Arita, Y. Ohno, and Y. Takahashi: “Filament formation and erasure in molybdenum oxide during resistive switching cycles,” Appl. Phys. Lett. 105, 173504 (2014).

[5] M. Kudo, M. Arita, Y. Takahashi, K. Ohba, M. Shimuta, and I. Fujiwara: “Visualization of conductive filament during write and erase cycles on nanometer-scale ReRAM achieved by in-situ TEM,” IMW (2015).