Nitrogen-implanted Ge₂Sb₂Te₅ film used as multilevel storage media for phase change random access memory

Bo Liu¹, Ting Zhang¹, Jilin Xia¹, Zhitang Song^1,2, Songlin Feng² and  Bomy Chen³

¹Research Center of Functional Semiconductor Film Engineering and Technology, Shanghai Institute of Micro-system and Information Technology, CAS, Shanghai 200050, People's Republic of China
²State Key Laboratory of Functional Materials for Informatics, Shanghai Institute of Micro-system and Information Technology, CAS, Shanghai 200050, People's Republic of China
³Silicon Storage Technology, Inc., 1171 Sonora Court, Sunnyvale, CA 94086, USA

Email: lb73sd@yahoo.com.cn

Received 3 February 2004
Published 21 April 2004

Contents

• 1. Introduction
• 2. Experiments
• 3. Results and discussions
• 4. Conclusions
• Acknowledgments
• References

1. Introduction

In 1968, Ovshinsky discovered a new order-disorder memory phenomenon in chalcogenide film materials, later termed `ovonic memory' and subsequently found a laser optical memory effect [1-4]. As phase changes occur in these materials, the change in optical constants is accompanied by a much larger change in electrical conductivity. So chalcogenide semiconductor films can also be applied to electrical write and erase nonvolatile memory devices, which are named `chalcogenide random access memory' (C-RAM) or `ovonic unified memory' (OUM). This electrical phase change technology has been developed for commercial and space applications by Ovonyx, Inc. at Intel and Lockheed Martin, respectively, since 1999 [5]. C-RAM is a possible substitute for all kinds of current memory devices such as dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, and others. C-RAM has a simple cell structure with high scalability; it is nonvolatile, has a relatively high read/write operation speed and a long cycle life [6, 7]. Furthermore, superior radiation tolerance makes it attractive for space-based applications [8].

Many studies on C-RAM using chalcogenide semiconductors such as GeTeAsSi [1, 9], GeTe [10], GeSbTe [6, 8, 11], GeTeBi [12, 13], GeSb (Cu, Ag) [14], GeTeAs [15], In-Te [16], AsSbTe [17, 18], SeSbTe [19], PbGeSb [20], GeSbTe-N [21], etc have been reported. But all of the above chalcogenide semiconductors were used only as single-level storage media for C-RAM. In this letter, we report for the first time that the nitrogen-implanted Ge₂Sb₂Te₅ film can be used as multilevel storage media for C-RAM.

2. Experiments

In our experiments, Ge₂Sb₂Te₅ thin film was deposited by RF sputtering on a Si(100)/SiO₂ substrate and the temperature of the substrate was controlled at room temperature. The background pressure is below 2 × 10 ^- 4 Pa. The sputtering parameters of the film are sputtering power of 300 W and Ar sputtering pressure of 0.15 Pa. The thickness of the film before implantation is 200 nm, which is determined using a Tencor Alpha-Step 500 Profiler. The film growth rate is 0.7 nm s ^- 1. N ⁺  ion was implanted into Ge₂Sb₂Te₅ films with the implant dose ranging from 6.44 × 10¹⁵ to 4.51 × 10¹⁶ cm ^- 2. The nitrogen implant dose was determined by inductively coupled plasma atomic emission spectrometry (ICP-AES). The implantation energy was 65 keV. The samples were crystallized by rapid thermal annealing under Ar atmosphere for 1 min. The film resistance is measured with a four-point probe. The structure of the films was examined by x-ray diffraction (XRD) analyses using a Rigaku D/MAX 2550 V diffractometer. For XRD experiments, Cu Kα (λ  =  0.154 18 nm) radiation was used.

3. Results and discussions

Figure 1 shows the dependence of the sheet resistance of the nitrogen-implanted Ge₂Sb₂Te₅ films on annealing temperature. The sheet resistance of the amorphous (R_amo) Ge₂Sb₂Te₅-N film is larger than 1 × 10⁷ Ω. For the Ge₂Sb₂Te₅ film without nitrogen implantation, the resistance decreases with increasing temperature and reaches a steady value when the temperature is higher than 500 °C. But for the nitrogen-implanted Ge₂Sb₂Te₅ film, the dependence of resistance on temperature is different from that for the Ge₂Sb₂Te₅ film without nitrogen implantation. When the annealing temperature increases, the resistance of the Ge₂Sb₂Te₅-N film with a nitrogen implant dose of 6.44 × 10¹⁵ cm ^- 2 first decreases quickly and then reaches a steady value after 180 °C; the resistance further decreases quickly at about 400 °C and then reaches another steady value again at temperature higher than 420 °C. The change of resistance with temperature for the Ge₂Sb₂Te₅-N film with a nitrogen implant dose of 1.92 × 10¹⁶ cm ^- 2 is similar to that for the Ge₂Sb₂Te₅-N film with a nitrogen implant dose of 6.44 × 10¹⁵ cm ^- 2. However, when the nitrogen implant dose reaches 4.51 × 10¹⁶ cm ^- 2, the second step of resistance decreasing disappears. The resistance changes very slightly and is almost the same at temperature higher than 180 °C.

Figure 2 shows the XRD patterns for the nitrogen-implanted crystalline Ge₂Sb₂Te₅ films. From figure 2(a), the structure of the crystalline Ge₂Sb₂Te₅ film is identified as a face-centred cubic (FCC) structure at low annealing temperature [22] and the lattice parameter is calculated as a  =  0.599 06 nm. However, the structure changes into a hexagonal structure with lattice parameters a  =  0.4216 nm and c  =  1.7174 nm when the annealing temperature is high. The crystal structure of the Ge₂Sb₂Te₅-N film is also identified as a FCC structure at low annealing temperature. The lattice parameter increases slightly with increasing nitrogen implant dose. The increase of the lattice parameter is due to the fact that a doped nitrogen atom occupying the tetrahedral interstitial site distorts the unit cell and the volume of the tetrahedral site is not large enough for a nitrogen atom to occupy [23]. From figures 2(b) and (c), when the nitrogen implant dose is below 1.92 × 10¹⁶ cm ^- 2, the crystal structure of the Ge₂Sb₂Te₅-N film is still identified as a hexagonal structure at high annealing temperature. But the intensity of the peaks for the hexagonal structure decreased greatly with increasing nitrogen implant dose. However, the peaks for the hexagonal structure totally disappeared and the crystal structure of the Ge₂Sb₂Te₅-N film is identified as a FCC structure at high annealing temperature when the nitrogen implant dose reaches 4.51 × 10¹⁶ cm ^- 2 (see figure 2(d)). The phase transition from FCC to hexagonal structure was suppressed by nitrogen implantation at high implant dose. We first report this result.

Ovshinsky [5] has supposed that multi-state phase change memory can be realized using Ge₂Sb₂Te₅ film because the resistance difference between the amorphous and crystalline state is very large. But there is no resistance steady state with large temperature tolerance in the resistance-temperature curve for the Ge₂Sb₂Te₅ film. This is not beneficial for realizing multilevel storage. However, if there are obvious steps in the resistance-temperature curve and the resistance in each step almost remains constant in a wide temperature range, the multilevel storage will be realized easily. From the above we can see that there are two obvious steps in the resistance-temperature curve of the Ge₂Sb₂Te₅-N film with a minor nitrogen implant dose, for example, about 6.44 × 10¹⁵ to 1.92 × 10¹⁶ cm ^- 2. This phenomenon is very important for multilevel storage. We can define the amorphous state with high resistance as the `0' state, the crystalline state with medium resistance as the `1' state and the crystalline state with low resistance as the `2' state. The resistance ratio of `0' to `1' or `0' to `2' state is about 10<sup>4</sup>-10<sup>6</sup> and the resistance difference between `1' state and `2' state is at least of order 1 (see figure 1). The ratio is sufficiently large for application to the memory devices. Then multilevel storage with Ge₂Sb₂Te₅-N media for C-RAM can be performed easily. Three-level storage can be realized: level 1 (0 to 1), level 2 (0 to 2) and level 3 (1 to 2), just as shown in figure 3. However, the nitrogen implant dose must be selected correctly because the two steps may change into one step if the nitrogen implant dose is higher than 4.51 × 10¹⁶ cm ^- 2. This phenomenon is also very important for single-level storage because the resistance of the crystalline Ge₂Sb₂Te₅-N film is almost constant, that is, the recording voltage margin is large.

The Ge₂Sb₂Te₅ films in the amorphous, the FCC and the HCP structures show a semiconductor, semi-metallic and metallic behaviour, respectively [24]. Obviously, the metal has much higher conductivity than the semi-metal and the semiconductor. The results from resistance measurements compared well to those from XRD. The detailed reason of nitrogen implantation affecting the resistance-temperature curve for the Ge₂Sb₂Te₅ film is not very clear now. This may be due to the structure change of Ge₂Sb₂Te₅ film caused by nitrogen implantation because we have found the existence of Ge₃N₄ in Ge₂Sb₂Te₅-N film. Further research about the effect of nitrogen implantation on the structure or resistance and the multilevel storage behaviour of memory cell using Ge₂Sb₂Te₅-N film is underway.

4. Conclusions

In summary, N ⁺  ion was implanted into Ge₂Sb₂Te₅ films and two obvious steps were observed in the resistance-temperature curve of the Ge₂Sb₂Te₅-N film with a minor nitrogen implant dose. The two steps may change into one step because the phase transition from FCC to hexagonal structure is suppressed by nitrogen implantation if the nitrogen implant dose is higher than 4.51 × 10¹⁶ cm ^- 2. The favourite nitrogen implant dose is about 6.44 × 10¹⁵ to 1.92 × 10¹⁶ cm ^- 2 in our study. This phenomenon is very important for multilevel storage. Three-level storage with Ge₂Sb₂Te₅-N media for C-RAM can be performed easily. This multilevel storage technology will dramatically increase the capacity of C-RAM. In addition, the resistance of the crystalline Ge₂Sb₂Te₅-N film with a nitrogen implant dose higher than 4.51 × 10¹⁶ cm ^- 2 is almost constant, which is also very important for single-level storage because the recording voltage margin is greatly improved.

Acknowledgments

This work is supported by National 863 Project under contract of 2003AA32720, Shanghai Nanotechnology Promotion Center (0352nm016, 0359nm004, 0252nm084), China Postdoctoral Foundation (2003034308), K C Wong Education Foundation (Hong Kong), the special Funds for Major State Basic Research Project (no 001CB610408), Fore-research of basic research project (2001CCA02800) and Science and Technology Council of Shanghai (03dz11009). The authors thank Miss Jing Chen at Shanghai Institute of Technical Physics, CAS, for help in the experiment of resistance measurement.

References

[1] 

Ovshinsky S R 1968 Phys. Rev. Lett. 21 1450 
CrossRef

[2] 

deNeufville Feinleib J, Moss S C and Ovshinsky S R 1971 Appl. Phys. Lett. 18 254 
CrossRef

[3] 

Ohta T, Nishiuchi K, Narumi K, Kitaoka Y, Ishibashi H, Yamada N and Kozaki T 2000 Japan. J. Appl. Phys. 39 770 
CrossRef

[4] 

Maeda T, Terao M and Shimano T 2003 Japan. J. Appl. Phys. 42 1044 
CrossRef

[5] 

Ovshinsky S R and Czubatyj W 2001 Proc. SPIE 4085 15 
CrossRef

[6] 

Wicker G 1999 Proc. SPIE 3891 2 
CrossRef

[7] 

Lai S and Lowrey T 2001 IEEE Conf. Proc. Int. Electron Devices Meeting 36.5.1  

[8] 

Bernacki S, Hunt K and Tyson S 2000 IEEE Trans. Nucl. Sci. 47 2528 
CrossRef

[9] 

Bunton G V and Quilliam R M 1973 IEEE Trans. Electron Devices 20 140 
CrossRef

[10] 

Gosain D P, Nakamura M and Shimizu T 1989 Japan. J. Appl. Phys. 28 1013 
CrossRef

[11] 

Kang D H, Ahn D H and Kim K B 2003 J. Appl. Phys. 94 3536 
CrossRef

[12] 

Bhatia K L, Singh M and Katagawa T 1995 Semicond. Sci. Technol. 10 65 
IOPscience

[13] 

Bhatia K L, Kishore N and Malik J 2002 Semicond. Sci. Technol. 17 189 
IOPscience

[14] 

Ramesh K, Asokan S and Sangunni K S 1999 Appl. Phys. A 69 421 
CrossRef

[15] 

Tsendin K D, Lebedev E A and Kim Y H 2001 Semicond. Sci. Technol. 16 394 
IOPscience

[16] 

Rajesh R and Philip J 2003 Semicond. Sci. Technol. 18 133 
IOPscience

[17] 

Nakayama K, Kitagawa T and Ohmura M 1993 Japan. J. Appl. Phys. 32 564 
CrossRef

[18] 

Nakayama K, Kojima K and Hayakawa F 2000 Japan. J. Appl. Phys. 39 6157 
CrossRef

[19] 

Nakayama K, Kojima K and Imai Y 2003 Japan. J. Appl. Phys. 42 404 
CrossRef

[20] 

Saheb P Z, Asokan S and Gowda K A 2003 Appl. Phys. A 76 665 
CrossRef

[21] 

Horii H, Yi J H and Park J H 2003 Symposium on VLSI Technology Digest of Technical Papers 177 

[22] 

Yamada N, Ohno E, Nishiuchi K, Akahira N and Takao M 1991 J. Appl. Phys. 69 2849 
CrossRef

[23] 

Jeong T H, Kim M R and Seo H 2000 Japan. J. Appl. Phys. 39 2775 
CrossRef

[24] 

Galvan A M and Hernandez J G 2000 J. Appl. Phys. 87 760 
CrossRef