This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 140.113.38.11
This content was downloaded on 25/04/2014 at 15:46
Please note that terms and conditions apply.
SONOS memories with embedded silicon nanocrystals in nitride
View the table of contents for this issue, or go to the journal homepage for more 2008 Semicond. Sci. Technol. 23 075033
(http://iopscience.iop.org/0268-1242/23/7/075033)
Semicond. Sci. Technol. 23 (2008) 075033 (4pp) doi:10.1088/0268-1242/23/7/075033
SONOS memories with embedded silicon
nanocrystals in nitride
Mei-Chun Liu
1, Tsung-Yu Chiang
1, Po-Yi Kuo
2, Ming-Hong Chou
3,
Yi-Hong Wu
3, Hsin-Chiang You
4, Ching-Hwa Cheng
3, Sheng-Hsien Liu
3,
Wen-Luh Yang
3, Tan-Fu Lei
2and Tien-Sheng Chao
11Department of Electrophysics, National Chiao Tung University, Hsinchu, Taiwan 2Institute of Electronics, National Chiao Tung University, Hsinchu, Taiwan 3Department of Electronic Engineering, Feng Chia University, Taichung, Taiwan
4Department of Computer Science and Information Engineering, Asia University, Taichung, Taiwan E-mail:[email protected]
Received 5 March 2008, in final form 28 April 2008 Published 28 May 2008
Online atstacks.iop.org/SST/23/075033
Abstract
We have successfully demonstrated SONOS memories with embedded Si-NCs in silicon nitride. This new structure exhibits excellent characteristics in terms of larger memory windows and longer retention time compared to control devices. Using the same thickness 2.5 nm of the bottom tunneling oxide, we found that N2O is better than O2oxide. Retention
property is improved when the thickness of N2O is increased to 3.0 nm.
(Some figures in this article are in colour only in the electronic version)
Silicon–oxide–nitride–oxide–silicon (SONOS) non-volatile memories have been proposed to overcome the oxide thickness limit of a conventional floating gate structure [1, 2]. In SONOS, charges can be stored in the nitride which offers several advantages over the traditional floating gate flash memory: simple process, higher density, no floating gate coupling effect, multi-bit operation and elimination of the drain-induced turn-on effect [3–5]. Although scaled SONOS can be operated at a low bias, endurance and data retention are still challenging [6]. Recently, metal-oxide-semiconductor (MOS) memories with embedded silicon nanocrystals have attracted a great interest to mitigate the problem of retention and endurance [7–11]. In this paper, SONOS memories with embedded silicon nanocrystals (NCs) are proposed. Si-NCs are introduced inside the silicon nitride film of SONOS structure [12]. We found that the memory window of SONOS with Si-NCs can be increased 2.5 times and the endurance can also be increased significantly.
Figure 1 shows the device’s structure. P-type silicon (1 0 0) wafers were used. After the local-oxidation-of-silicon, or LOCOS, isolation step, a tunneling oxide was first thermally grown in N2O (2.5 and 3.0 nm) or O2 (2.5 nm). Then,
a 3 nm thick silicon nitride film was deposited in a low-pressure chemical vapor deposition (LPCVD) system using SiH2Cl2 and NH3 at 780 ◦C. Wafers were deposited on a
thin amorphous silicon layer which was grown in LPCVD by using SiH4(85 sccm, pressure∼100 mTorr, at 550◦C).
This amorphous silicon layer was crystallized into Si-NCs in the following elevated temperature step of nitride deposition. To sandwich Si-NCs, a top 4 nm silicon nitride was capped on the amorphous silicon nucleation. A blocking oxide about 20 nm was then deposited using high-density plasma chemical-vapor-deposition (HDPCVD). A 200 nm thick poly-Si film was deposited as the control gate. Standard MOSFET fabricating steps were followed to complete final devices.
The formation of Si-NCs was confirmed by atomic force microscopy (AFM) as shown in figures2(a)–(c). Compared to the control one (figure2(a)), the roughness was increased as the deposition time of SiH4was increased from 1.5 min to
2 min. The average size was estimated to be around 8–10 nm and the density can be as high as 3–7× 1011cm−2. Si-NCs
distribution and sizes obtained with AFM were verified using HR-TEM (results to be published elsewhere). These Si-NCs are well separated with an average distance of >6 nm, which ensures electrical isolation between two NCs. The C–V was measured from the transistor with a W/L of 100× 100 µm2
when source/drain were grounded. Figure3shows program window of all samples. The programmed state was achieved by a constant VG= 25 V for 10 s. The program windows of SONOS, SONOS with Si-NCs deposited for 1.5 min and
Semicond. Sci. Technol. 23 (2008) 075033 M-C Liu et al
Figure 1. The cross-sectional scheme of a Si-NCs SONOS memory structure with the nitride film embedded with the silicon
nanocrystals.
(a) (b)
(c)
Figure 2. AFM pictures of Si nanocrystals deposited on Si3N4(a) control sample, (b) Si-NCs 1m30s sample and (c) Si-NCs 2 min sample, with the same growth conditions. The densities are, respectively, 6.7× 1011and 3× 1011cm2. The diameters are about, respectively, 8 and 10 nm.
2 min are 3.85 V, 6.25 V and 8.98 V, respectively. It is noted that the program window of Si-NCs is larger than that of the control sample. And large Si-NCs size (2 min) results in a large memory window due to the increased trapping site. The Si-NCs memory window of 8.98 V is large enough for multi-level operation. For this sandwiched structure with embedded Si-NCs, the electrons could be stored in Si-NCs, or at the interface of silicon nitride and Si-NCs. However, from a rough calculation, the charge in each dot would be around 10, which could not result in such a memory window. An alternative explanation is trapping by nitride traps, while Si-nanodots facilitate injection of electrons from the substrate. Therefore the relatively large resultant memory window can be obtained for SONOS with embedded Si-NCs.
The threshold voltage shown in figure 4 is extracted from the constant drain current at 10−7A. The data retention characteristics of the different Si-NC memories at a high temperature of 250 ◦C are shown in figure 4(a). SONOS with Si-NCs exhibits a good performance than the control one.
Program Window@tunnel oxide N2O 3nm VG=25V , t=10sec VG (V) -2 0 2 4 6 8 C ( p F ) 2 4 6 8 10 12 14 16 control_fresh control_program Si-NCs_2min_fresh Si-NCs_2min_program Si-NCs_1m30s_fresh Si-NCs_1m30s_program 3.58V 6.25V 8.98V
Program Window@tunnel oxide N2O 3nm
VG=25V , t=10sec G -2 0 2 4 6 8 2 4 6 8 10 12 14 16 control_fresh control_program Si-NCs_2min_fresh Si-NCs_2min_program Si-NCs_1m30s_fresh Si-NCs_1m30s_program 3.58V 6.25V 8.98V
Figure 3. Program window characteristic was a different sample. The program windows of control, Si-NCs 1m30s and Si-NCs 2 min sample are about 3.58 V, 6.25 V and 8.98 V, respectively.
Time (sec ) Time (sec ) 100 101 102 103 104 -0.8 -0.6 -0.4 -0.2 0.0 N2O 3nm N2O 2.5nm O2 2.5nm Retention@250oC Different Tunnel Oxide Thickness Si-NCs_1m30s Program ∆Vt : 2V Retention@250oC Tunnel Oxide_N2O 3nm Program ∆Vt : 2V 100 101 102 103 104 105 106 107 108 109 ∆ Vt shift (V ) ∆ Vt shift (V ) -2.5 -2.0 -1.5 -1.0 -0.5 0.0 control Si-NCs_2min Si-NCs_1m30s (a) (b)
Figure 4. (a) Data retention characteristics of different Si-NCs sizes when programming Vt= 2 V at T = 250◦C. (b) Data retention
characteristics of different tunnel oxide films when programming Vt= 2V at T = 250◦C.
SONOS with Si-NCs shows only about 0.08 V degradation for 104s. Only 14% charge loss when extrapolates to 10 years. It
is known that data loss of SONOS is mainly due to thermionic emission and direct tunneling of charges [13]. The retention performance was not degraded though the trapping efficiency of the memory media with Si-NCs strongly improved. Figure4(b) shows the data retention characteristics of different tunnel oxides with an initial programming Vt= 2 V at T = 250◦C. It is found that, with the same thickness of 2.5 nm, N2O
is better than 2.5 nm O2oxides due to better quality [14,15].
On the other hand, retention property is improved when thickness of N2O is increased to 3.0 nm, resulting from the
reduction of direct tunneling probability.
In conclusion, we have successfully demonstrated SONOS memories with embedded Si-NCs in silicon nitride. Based on the above result, embedded Si-NCs in silicon nitride of SONOS memories exhibit excellent characteristics in terms of larger memory windows and long retention time.
Acknowledgments
This work is supported by the National Science Council, Taiwan, under contract no: NSC-95-2221-E-009-272. The authors would like to thank the processes support from National Nano Device Labs (NDL) and the Nano Facility Center of the National Chiao Tung University.
References
[1] Cappelletti P, Golla C, Olivo P and Zanoni E 2000 FLASH
Memories (Dordrecht: Kluwer Academic) p 165
[2] Semiconductors Industry Association 2005 International Technology Roadmap for Semiconductors
[3] White M H, Adams D A and Bu J 2000 IEEE Circuits Device
Mag.16 22
[4] Chan T Y, Young K K and Hu C 1987 IEEE Electron Device
Semicond. Sci. Technol. 23 (2008) 075033 M-C Liu et al [5] Cho M K and Kim D M 2000 IEEE Electron Device Lett.
21 399
[6] Yeh C C, Tsai W J, Lu T C, Liao Y Y, Zous N K, Chen H Y, Wang T, Ting W, Ku J and Lu C Y 2004 Physical and failure analysis of integrated circuits, IPFA Proc. 11th Int.
Symp. p 247
[7] Tiwari S, Rana F, Chan K, Hanafi H, Chan W and Buchanan D 1995 IEDM Tech. Dig. p 521
[8] Tiwari S, Rana F, Hanafi H, Hartstein A, Crabbe E F and Chan K 1996 Appl. Phys. Lett.68 1377
[9] Chang T C, Yan S T, Liu P T, Chen C W, Lin S H and Sze S M 2004 Electrochem. Solid-State Lett.7 G17
[10] Chen J H, Lei T F, Landheer D, Wu X, Liu J and Chao T S 2007 Electrochem. Solid-State Lett.10 H302
[11] Wu J H and Li P W 2006 Semicond. Sci. Technol. 21 507
[12] Rao R A et al 2004 Solid-State Electron.48 1463
[13] Chen C H, Chiang P Y, Chung S S, Chen T, Chou G C W and Chu C H 2006 VLSI Technology, Systems, and Applications p 1
[14] Sharma U, Moazzami R, Tobin P, Okada Y, Cheng S K and Yeargain J 1992 IEDM Tech. Dig. p 461
[15] Joshi A B, Yoon G, Kim J, Lo G Q and Kwong D L 1993
IEEE Trans. Electron Devices40 1437
