1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Voltage polarity effects in Ge2Sb2Te5-based phase change memory devices
Rent:
Rent this article for
USD
10.1063/1.3626047
/content/aip/journal/jap/110/5/10.1063/1.3626047
http://aip.metastore.ingenta.com/content/aip/journal/jap/110/5/10.1063/1.3626047
View: Figures

Figures

Image of FIG. 1.
FIG. 1.

(Color online) Schematic cross-section of a PCM pore cell, indicating the directions of current flow for “good” and “bad” polarity behavior. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 2.
FIG. 2.

(Color online) Electrical test setup for the characterization of the switching characteristics of prototype pore and bridge-cell PCM devices.

Image of FIG. 3.
FIG. 3.

(Color online) RESET programming characteristics of PCM pore devices, for “good” polarity behavior, for nominal critical dimensions varying from 10 nm to 70 nm. Actual CD values (shown in parentheses) tended to be 7 to 10 nm larger than the nominal CDs.

Image of FIG. 4.
FIG. 4.

(Color online) Measured RESET current of PCM pore devices as a function of the actual CD and technology node (Ref. 21), compared to the predictions of electrothermal modeling (Ref. 22) and of the ITRS (Ref. 21). Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 5.
FIG. 5.

(Color online) Constant-amplitude negative (on the BEC) pulses provide a modest window for the SET operation, unless the pulse is so long that crystallization occurs during the 0.5% ramp-down. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 6.
FIG. 6.

(Color online) Ramped-down pulses (over the final 80%) provide a large amplitude-insensitive window for the SET operation for negative (on the BEC) voltages. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 7.
FIG. 7.

(Color online) For positive bias on the BEC, constant-amplitude pulses show only a narrow window, with many voltages leading to resistances higher than the initial RESET condition, and some that are up to higher. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 8.
FIG. 8.

(Color online) For positive bias (on the BEC) ramped-down pulses, the SET window is slightly wider, but excessively high resistances can be obtained for even modest (10 to ) durations. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 9.
FIG. 9.

(Color online) (a) Top view SEM image (width = 250 nm, length = 125 nm) and (b) lateral cross-section (along the center of the GST line) of a laterally symmetric phase-change bridge cell fabricated over tungsten electrodes. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 10.
FIG. 10.

(Color online) Whereas application of (a) an alternating-polarity SET pulse immediately following a RESET pulse produces low resistances in a symmetric bridge device (W = 20 nm, L = 200 nm), (b) a same-polarity SET pulse often produces high resistance. The formation of this high resistance state is always associated with a sharp cutoff of I DUT late within the ramp-down. Note that unlike in part (b), for which the device began in the SET state, the device in part (a), having just been RESET, first underwent electrical breakdown. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 11.
FIG. 11.

(Color online) TEM image (top) and EELS scan (bottom) of a PCB device (40 nm width, 200 nm length) after a single 5 ms, 3.5 V, constant-amplitude pulse was applied, showing significant motion of Ge and Sb to the negative electrode and Te to the positive electrode. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 12.
FIG. 12.

(Color online) High-resolution Auger analysis shows that whereas (a),(b) Te continuously leaves an uncapped Te-rich GST film during optical exposure, after short pulses (c) Sb is only slightly affected. However, for long pulses in the power regime in which both significant material motion (as measured via AFM) and the significant reflectivity increase associated with crystallization are observed, (d) Sb is consistently observed to agglomerate within the optically heated region. We attribute this phenomenon to the crystallization-induced segregation of Sb within Te-rich GST (Ref. 19). Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 13.
FIG. 13.

(Color online) Cross-sectional STEM and EELS analysis of an untested large (80 nm nominal CD) PCM pore device. Although surfaces exposed to air during processing are slightly oxidized, each element is uniformly distributed in the as-fabricated device. These larger pore devices needed for TEM show the same polarity behavior as smaller pores, except that ultrahigh resistance states could be created only with millisecond-long pulses. Colorbars are normalized by pixel regions in the upper corners (outside the pore region). Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 14.
FIG. 14.

(Color online) STEM and EELS for an 80 nm (nominal CD) pore device tested with a single negative-on-BEC 5 ms ramp-down pulse. The device resistance is low, and Ge has been driven into the lower corners, displacing Sb, but Te is relatively unchanged. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 15.
FIG. 15.

(Color online) STEM and EELS for an 80 nm (nominal CD) pore device tested with a single positive-on-BEC 5 ms ramp-down pulse. The device resistance is high, Ge has been driven up out of the lower corners, and Sb has coalesced locally away from Te. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 16.
FIG. 16.

(Color online) STEM and EELS for an 80 nm (nominal CD) pore device after a single high-amplitude positive-on-BEC 5 ms ramp-down pulse. The device resistance went ultrahigh before the end of the ramp-down, with Ge and Sb driven up, leaving only Te at the bottom. The formation of the void in the lower right-hand corner of the pore was violent enough to leave tungsten at the BEC. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 17.
FIG. 17.

(Color online) STEM and EELS for a 70 nm (nominal CD) pore device after two 5 ms ramp-down pulses. The first, positive-on-BEC pulse created a high resistance state; the second, negative-on-BEC pulse recovered the device, leaving an internal void but bringing Ge into the lower corners, balance to the Sb and Te distributions, and a low final resistance. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 18.
FIG. 18.

(Color online) Electrothermal modeling shows that even toward the end of the ramp-down portion of the applied pulse, temperatures should be suitable for crystallization where Sb is present and for melting where Te is enriched. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Image of FIG. 19.
FIG. 19.

(Color online) Our explanation of the polarity-dependent effects in GST-based PCM devices. (a) During a “good” polarity pulse, Ge and Sb are pulled by bias down through the melt toward the negatively biased BEC. Sb might aggregate, but all local stoichiometries support rapid crystallization. (b) Early in a “bad” polarity ramp-down pulse, however, Ge leaves the bottom of the pore, setting up conditions (c) for Sb to further segregate via crystallization at temperatures at which Te-rich material remains fluid. If the quench of Te-rich material to amorphous focuses current into a portion of the BEC, a void can be created. Reprinted with permission from A. Padilla, G. W. Burr, K. Virwani, A. Debunne, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, D. Dupouy, A. J. Kellock, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, Tech. Dig. - Int. Electron Devices Meet. 29.4 (2010). © 2010, IEEE.

Loading

Article metrics loading...

/content/aip/journal/jap/110/5/10.1063/1.3626047
2011-09-01
2014-04-24
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Voltage polarity effects in Ge2Sb2Te5-based phase change memory devices
http://aip.metastore.ingenta.com/content/aip/journal/jap/110/5/10.1063/1.3626047
10.1063/1.3626047
SEARCH_EXPAND_ITEM