• 沒有找到結果。

The fabrication of a programmable via using phase-change material in CMOS-compatible technology

N/A
N/A
Protected

Academic year: 2021

Share "The fabrication of a programmable via using phase-change material in CMOS-compatible technology"

Copied!
6
0
0

加載中.... (立即查看全文)

全文

(1)

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 04:09

Please note that terms and conditions apply.

The fabrication of a programmable via using phase-change material in CMOS-compatible

technology

View the table of contents for this issue, or go to the journal homepage for more 2010 Nanotechnology 21 134001

(http://iopscience.iop.org/0957-4484/21/13/134001)

(2)

CMOS-compatible technology

Kuan-Neng Chen

1

and Lia Krusin-Elbaum

2

1Department of Electronics Engineering, National Chiao Tung University, Hsinchu, Taiwan 2IBM T J Watson Research Center, Yorktown Heights, NY 10598, USA

E-mail:[email protected]

Received 3 August 2009, in final form 27 August 2009 Published 8 March 2010

Online atstacks.iop.org/Nano/21/134001 Abstract

We demonstrate an energy-efficient programmable via concept using indirectly heated phase-change material. This via structure has maximum phase-change volume to achieve a minimum on resistance for high performance logic applications. Process development and material investigations for this device structure are reported. The device concept is successfully demonstrated in a standard CMOS-compatible technology capable of multiple cycles between on/off states for reconfigurable applications.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

A programmable via using phase-change materials is an enabling technology for high performance reconfigurable logic applications with disruptive improvements in area and power efficiencies. Reconfigurable circuits play important roles in a wide range of microelectronic applications, such as repair of defective memory elements [1, 2] and field-programmable gate arrays (FPGAs) [3]. For repair of defective memory elements, the anti-fuse approach, often applied to DRAM repair, and the electrical fuse (eFUSE) are utilized for programmable circuits. However, both structures can be only used once and require high programming voltages [1,2] or currents [4–6]. Regarding the FPGA concept, an array of configurable logic blocks is arranged with switches to rearrange the interconnections between the logic blocks. However, current FPGAs use flash memory and require extra space for the flash transistor. In particular, the process is not standard CMOS-compatible. Therefore, it can be seen that the current challenges for programmable circuits include ‘multiple operation’ and ‘CMOS-compatible fabrication’.

To date, phase-change materials have drawn the most attention from semiconductor memory developers as a possible replacement for flash memory [7–10]. In contrast to phase-change memory, in which the phase-phase-change volume is to be minimized for the smallest possible memory cell size, the

volume of the phase-change material in a programmable via is to be maximized for the lowest contact resistance. In this paper, we introduce and review an indirect-heating concept to fabricate a programmable via with minimal consuming area [10]. Logic circuits associated with the programmable via are decoupled from the configuration circuits by using an independently contacted heater electrode. Since this device structure can be fabricated with CMOS-compatible processes and can be operated for multiple cycles between OFF/ON (RESET/SET) states, it shows the great potential in future reconfigurable applications. In addition, process development and material investigations of this device are reported.

2. Concept of the device structure

The schematic diagram of this programmable via device structure is shown in figure 1. The via is filled with phase-change material (Ge2Sb2Te5), which can be switched

between resistive (OFF–amorphous–RESET) and conductive (ON–crystalline–SET) states. An external heater using doped TaN material is integrated with the programmable via. The switching process is based upon the ‘programming’ current pulse passed through the heater. During the OFF switching operation, an abrupt high-current pulse is used to melt and quench/amorphize a thin region of phase-change via adjacent to the heater. The ON switching operation is accomplished by

(3)

Nanotechnology 21 (2010) 134001 K-N Chen and L Krusin-Elbaum

Figure 1.Schematic diagram of the programmable via concept using phase-change material (Ge2Sb2Te5) and indirect-heating material

(doped TaN). Tungsten (W) is used for electrical connection.

Figure 2.Sheet resistance behaviors of pure and doped TaN materials during temperature operation.

a relatively low but long current pulse applying through the heater. The heat generating from the current on the heater is capable of annealing the amorphous phase-change materials to the crystalline state. It should be noted that, in this concept demonstration, the via resistance has been measured between the top contact of the via and one of the heater contacts by a ‘read’ current, as shown in figure1.

For the heater design, in order to achieve the best efficiency of electrical–thermal transformation from heater to via, the heater should be a thin layer and its material should be a refractory metal with relatively high resistivity and low thermal conductivity. Therefore, a doped TaN material is used as the heater because of its low thermal conductivity and high sheet resistance [11]. Figure2shows two examples of doped TaN materials with different doping concentrations.

3. Device fabrication

The device fabrication includes three sections using a standard CMOS technology: (a) heater, (b) programmable via and (c) electrical connection. First, the heater fabrication started from deposition of the doped TaN film above the silicon oxide film on the Si wafer. After the pattern of the heater was defined on the doped TaN film by RIE (reactive ion etching), another silicon oxide film was deposited to cover the heater.

The programmable via fabrication was followed by the heater fabrication. The via was defined by lithographic process

Figure 3.SEM image showing the via hole clearly defined on the doped TaN heater region.

Figure 4.Cross-sectional SEM image showing that the bottom of the via hole lands well on top of the doped TaN heater.

then etched by RIE. The etched process required a good etching selectivity between oxide and doped TaN. Figures3 and4show the via hole is clearly defined and its bottom lands well on top of the doped TaN heater. The aspect ratio of the via is better than 1:1, which is helpful for the later phase-change material (Ge2Sb2Te5) deposition. After depositing TiN

(diffusion barrier), Ti (adhesion layer) and Ge2Sb2Te5 onto

the via, a CMP (chemical–mechanical polishing) process was performed to removed Ge2Sb2Te5/Ti/TiN materials outside the

via. In order to prevent any contamination into the Ge2Sb2Te5

via, a TiN cap layer, which was designed to be larger than via size, was fabricated to fully coverage the Ge2Sb2Te5 via.

Figures5and6show SEM images of device structures after the cap layer was fabricated. During the Ge2Sb2Te5/Ti/TiN CMP

process, when the CMP polishing pad was reaching the oxide surface, the polishing rate of Ge2Sb2Te5in the via was higher

than that of oxide. This rate difference led to the via level lower than the adjacent oxide level after the CMP process. Therefore, it can be seen that the central region of the cap layer, which is corresponding to the via area, is lower than the adjacent region, as shown in figure6.

The final section of device fabrication is electrical connection. Two W vias were fabricated to connect the 2

(4)

Figure 5.Top-down SEM image of the device structure after the cap layer was fabricated.

Figure 6.SEM image of the device structure after the cap layer was fabricated. The central region of the cap layer, which is

corresponding to the via area, is lower than the adjacent region.

heater for electrical current delivery in the heater. The device structure was completed after the testing W pads were fabricated. Figure 7 shows the SEM image of the programmable via region of the final device. Vias with diameters of 200–280 nm were fabricated atop 20 nm thick TaN heaters that had the same width as the via diameter, and a length-to-width ratio of 6. The resistance of the doped TaN heater is ∼1700  and the resistivity of the doped TaN is 6× 10−4 cm.

4. Electrical characterization

Figure 8 shows a TEM image of the via/heater interfacial region after ON switch operation. Clear crystalline and grain structures of the phase-change material are observed near the heater interface after the ON operation, indicating a crystalline state. Figure9shows a TEM image of the via/heater interfacial region after OFF switch operation. No crystalline and grain structure of the phase-change material are observed near the heater interface after the OFF operation, indicating an amorphous state.

Figure 7.SEM image of the programmable via region of the final device.

Figure 8.Cross-sectional TEM image of the via–heater interface region after ON switching operation.

Figure 9.Cross-sectional TEM image of the via–heater interface region after OFF switching operation.

Figures10and11show the R–I switching characteristics of the programmable via without device history. In figure10, starting from the OFF state, 1 μs pulses with gradually increased power were applied to the heater. When the pulse current reached around 0.9 mA, the via resistance started to

(5)

Nanotechnology 21 (2010) 134001 K-N Chen and L Krusin-Elbaum

Figure 10.R–I characteristics for switching to the ON state. Via

size= heater width = 280 nm; heater length = 1680 nm; heater thickness= 20 nm. ON pulse = 200 ns ramp up + 1000 ns plateau + 200 ns ramp down.

Figure 11.R–I characteristics for switching to the OFF state.

Device geometry is as listed in the caption of figure10. OFF pulses are 19 ns rise time and 2 ns fall time.

decrease and finally implemented switching of the device to the ON state. In figure11, 50 ns pulses with gradually increased power were applied to the heater from the ON state. After each pulse, the via was switched back to the ON state. When the pulse current reached around 2 mA, the via resistance started to increase and finally reached the OFF state. The endurance test results in figure12show stable OFF/ON ratios around 400 without obvious degradation. An estimated OFF/ON ratio of an ideal four-terminal device can reach 1000 when resistance contribution from the heater to ON resistance becomes small. The temperature dependence of resistance in the OFF state is shown in figure13.

5. Simulation

In order to further understand the behavior and predict the performance of this energy-efficient programmable via structure, several simulations were performed and compared

Figure 12.Cycling data from endurance tests at room temperature. Device geometry and pulse conditions are as listed in the captions of figures10and11.

Figure 13.Measured temperature dependence of resistance in OFF state.

with the experimental results of the prototype device. Figure14 shows simulated required OFF current and programmable via dimension. Here via size = heater width = 1/6 × heater length; heater thickness = 20 nm; 50 ns width OFF pulses with 19 ns rise time and 2 ns fall time. The experimental data of this device structure is also marked in the figure, showing good prediction. Figure15 shows the simulated OFF current and pulse time at plateau. Here, via size= heater width = 280 nm; heater length= 1680 nm; heater thickness = 20 nm. OFF pulses were 19 ns rise time and 2 ns fall time. Again, the experimental data, as marked in the figure, showing good prediction within reasonable range from the simulation. These simulation results provide references for future device performance predictions.

6. Conclusions

A programmable via using an indirectly heated phase-change switch is proposed and demonstrated in this paper. In addition 4

(6)

Figure 14.Simulated required OFF current and programmable via dimension. Via size= heater width = 1/6 × heater length; heater thickness= 20 nm; 50 ns width OFF pulses with 19 ns rise time and 2 ns fall time.

Figure 15.Simulated OFF current and pulse time at plateau. Via size= heater width = 280 nm; heater length = 1680 nm; heater thickness= 20 nm; OFF pulses with 19 ns rise time and 2 ns fall time.

to excellent electrical characterization, the device fabrication is processed in a standard CMOS technology and the device is capable of multiple operations. Therefore, this structure can be utilized for reconfigurable logic applications. Finally the good

National Science Grant Council grant no. NSC 98-2218-E-009-013-MY2. The authors acknowledge help of IBM fabrication facilities from the Material Research Laboratory (MRL) and Central Scientific Services (CSS) at T J Watson Research Center, Yorktown Heights, NY.

References

[1] Mazumder P and Yih J S 1993 A new built-in self-repair approach to VLSI memory yield enhancement by using neural-type circuits IEEE Trans. Computer-Aided Des.

Integr. Circuits Syst.12124–36

[2] Tanabe A et al 1992 A 30 ns 64-Mb DRAM with built-in self-test and self-repair function IEEE J. Solid-State Circuits

271525–33

[3] Borgatti M, Lertora F, Foret B and Cali L 2003 A reconfigurable system featuring dynamically extensible embedded microprocessor, FPGA, and customizeable I/O

IEEE J. Solid-State Circuits38521–9

[4] Kothandaraman C, Iyer S K and Iyer S S 2002 Electrically programmable fuse (eFUSE) using electromigration in silicides IEEE Electron Device Lett.23523–5

[5] Geppert L 2004 Electrical fuses lets chips heal themselves

IEEE Spectr. 41 16–20

[6] Tian C, Park B, Kothandaraman C, Safran J, Kim D, Robson N and Iyer S S 2006 Reliability qualification of CoSi2electrical fuse for 90 nm technology IEEE Int.

Reliability Physics Symp. Proc. (Mar. 2006) pp 392–7

[7] Lai S and Lowrey T 2001 OUM-A 180 nm nonvolatile memory cell element technology for stand alone and embedded applications IEDM Tech. Dig. pp 803–6

[8] Redaelli A, Pirovano A, Pellizzer F, Lacaita A L, Ielmini D and Bez R 2004 Electronic switching effect and phase-change transition in chalcogenide materials IEEE Electron Device

Lett.25684–6

[9] Happ T D et al 2006 Novel one-mask self-heating pillar phase change memory Proc. VLSI Tech. pp 120–1

[10] Chen K N et al 2008 Programmable via using indirectly heated phase-change switch for reconfiguration applications IEEE

Electron Device Lett.29131–3

[11] Suh Y-S, Heuss G P, Misra V, Park D-G and Lim K-Y 2003 Thermal stability of TaSixNyfilms deposited by reactive

數據

Figure 1. Schematic diagram of the programmable via concept using phase-change material (Ge 2 Sb 2 Te 5 ) and indirect-heating material
Figure 7. SEM image of the programmable via region of the final device.
Figure 13. Measured temperature dependence of resistance in OFF state.
Figure 15. Simulated OFF current and pulse time at plateau. Via size = heater width = 280 nm; heater length = 1680 nm; heater thickness = 20 nm; OFF pulses with 19 ns rise time and 2 ns fall time.

參考文獻

相關文件

In JSDZ, a model process in the modeling phase is treated as an active entity that requires an operation on its data store to add a new instance to the collection of

Students are asked to collect information (including materials from books, pamphlet from Environmental Protection Department...etc.) of the possible effects of pollution on our

We explicitly saw the dimensional reason for the occurrence of the magnetic catalysis on the basis of the scaling argument. However, the precise form of gap depends

Let T ⇤ be the temperature at which the GWs are produced from the cosmological phase transition. Without significant reheating, this temperature can be approximated by the

(A) The drought has authorities worried that the effects of climate change are larger than expected as Australia experiences half of its warmest days on record lately.. (B)

[This function is named after the electrical engineer Oliver Heaviside (1850–1925) and can be used to describe an electric current that is switched on at time t = 0.] Its graph

Figure 6 shows the relationship between the increment ∆y and the differential dy: ∆y represents the change in height of the curve y = f(x) and dy represents the change in height

When the relative phases of the state of a quantum system are known, the system can be represented as a coherent superposition (as in (1.2)), called a pure state; when the sys-