Multilevel Nonvolatile Small-Molecule Memory Cell Embedded with Ni

Mar 24, 2009 - Four-level nonvolatile small-molecule 4F2 memory cells were developed with a sandwiched device structure consisting of an upper Al elec...
0 downloads 9 Views 3MB Size
Download PDF
NANO LETTERS

Multilevel Nonvolatile Small-Molecule Memory Cell Embedded with Ni Nanocrystals Surrounded by a NiO Tunneling Barrier

2009 Vol. 9, No. 4 1713-1719

Jea-Gun Park,*,† Woo-Sik Nam,† Sung-Ho Seo,† Yool-Guk Kim,† Young-Hwan Oh,† Gon-Sub Lee,† and Un-Gyu Paik‡ National Program Center for Tera-bit-leVel NonVolatile Memory DeVelopment, Department of Electrical and Computer Engineering, and Department of AdVanced Material Engineering, Hanyang UniVersity, 17 Haengdang-dong, Seongdong-gu Seoul 133-791, Korea Received February 10, 2009; Revised Manuscript Received February 26, 2009

ABSTRACT 2

Four-level nonvolatile small-molecule 4F memory cells were developed with a sandwiched device structure consisting of an upper Al electrode, upper small-molecule layer (Alq3, aluminum tris(8-hydroxyquinoline)), Ni nanocrystals surrounded by NiO tunneling barrier, lower smallmolecule layer, and bottom Al electrode. In particular, an in situ O2-plasma oxidation process following Ni evaporation was developed to produce uniformly stable 10 nm Ni nanocrystals surrounded by a NiO tunneling barrier embedded in the small-molecule layer. They presented a memory margin (Ion/Ioff ratio) of ≈1 × 103, a retention time of more than 105 s, an endurance of more than 5 × 102 erase-and-program cycles, and multilevel cell (MLC) operation, being a terabit nonvolatile memory-cell. A vertically double-stacked 4F2 multilevel nonvolatile memory cell was also developed, showing a memory margin of ≈1 × 103 in both the top and bottom memory cells and eight-level cell operation.

Recently, organic devices have been proposed as nonvolatile memory cells with terabit-level integration density because they have demonstrated the possibility of bistable switching characteristics (memory behavior), minimal device feature size of 4F2, and fast access and store times of ∼10 ns, and the lowest production cost due to their simple device structure (only one bistable resistance between electrodes).1 These memory cells are fabricated with a simple sandwiched device structure of top metal layer, conductive organic layer with1-12 or without11-31 embedded metal nanocrystals, and bottom metal layer. The conductive organic layers contained conductive small molecules1-6,11-23 or conductive polymer.7-12,24-31 However, the reported organic memory cells do not present sufficient and reproducible nonvolatile memory-cell characteristics, such as a data retention time of ∼10 years, endurance of erase and program cycles of greater than ∼104, and multilevel cell (MLC) operation. In particular, they do not delineate the detailed physical and chemical structure and nonvolatile-memory operation mechanism of organic devices, resulting in very poor nonvolatile memory cell characteristics reliability. * Corresponding author, [email protected]. † National Program Center for Tera-bit-level Nonvolatile Memory Development, Department of Electrical and Computer Engineering. ‡ Department of Advanced Material Engineering. 10.1021/nl900429h CCC: $40.75 Published on Web 03/24/2009

 2009 American Chemical Society

Device Fabrication and Chracterization. In our study, we developed in situ O2-plasma oxidation, a new process to produce stable and reproducible metal nanocrystals (Ni) surrounded by a tunneling insulator layer (NiO) and embedded in a conductive organic layer (aluminum tris(8-hydroxyquinoline), Alq3) (Figure 1a). Ni was selected as a metal nanocrystal because it has a deep work function among pure metals (5.15 eV), and NiO surrounding Ni nanocrystals was expected to adhere well to the small-molecule layer (Alq3). Thus, new 4F2 small-molecule memory cells were fabricated with the sandwiched device structure of an upper aluminum (Al) electrode, a small-molecule layer (Alq3) embedded with Ni nanocrystals surrounded by NiO tunneling barrier, and lower Al electrode. The Ni nanocrystals surrounded by aNiO tunneling barrier were produced by Ni evaporation at rates below 0.1 Å/s followed by in situ O2-plasma oxidation for 300 s at 200 W without breaking the vacuum in a multichamber evaporator (Figure 1c). The 4F2 crossover memory cells were fabricated with a sandwiched device structure of a lower Al electrode, lower conductive small-molecule (Alq3) layer, Ni nanocrystals surrounded by NiO, upper conductive small-molecule (Alq3) layer, and upper Al electrode, where Alq3 was aluminum tris(8-hydroxyquinoline). Alq3 was used as a conductive small-molecule material because it chemi-

Figure 1. Small-molecule nonvolatile 4F2 memory cell with Al/ Alq3/(Ni nanocrystals surrounded by NiO)/Alq3/Al device structure: (a) chemical structure of Alq3, (b) perspective view of memory cell, (c) multichamber evaporator, (d) x-TEM image (at 200 kV acceleration voltage), (e) magnified x-TEM image (at 1.2 MV acceleration voltage), (f) AES profile, and (g) XPS profile.

cally reacts well with NiO surfaces. To isolate the memory cells from the substrate, the 4F2 crossover memory cells were fabricated on thermally grown SiO2 on Si wafers. To avoid contamination (e.g., H, O, N, Cl, and F atoms), all fabrication processes were carried out in a multichamber evaporator without breaking the vacuum; see Figure 1c. The 80-nmthick bottom Al electrode was thermally evaporated at 5.0Å/s at a 10-5 Pa chamber pressure by using a first shadow mask. The 35-nm-thick lower conductive small-molecule (Alq3) layer was thermally evaporated at 1.0 Å/s using a second shadow mask. Next, the middle 10-nm-thick Ni layer was thermally evaporated at 0.1 Å/s using a third shadow mask, and the wafer was transferred to the O2 plasma chamber (5 × 10-4 Pa, 200 W, 40 V, and 300 s) to oxidize the surface of Ni nanocrystals via radical oxygen diffusion through the grain boundaries of the Ni layer, thereby producing a NiO tunneling barrier surrounding the Ni nanocrystals. Next, the lower 30-nm-thick conductive small-molecule (Alq3) layer was thermally evaporated at 1.0 Å/s using a fourth shadow mask. Finally, the 80-nm-thick upper Al electrode was thermally evaporated at 5.0 Å/s using a fifth mask, which was cross patterned against the lower Al electrode. The sample for cross-sectional transmitted electron microscopy (x-TEM) observation was prepared using rough and fine focused-ion-beam etching. The crystal structure of the Ni nanocrystals was characterized using ultra-high-voltage TEM (1.2 MV). To investigate the chemical composition of the interface between the small-molecule (Alq3) layer and NiO tunneling barrier, three samples were prepared: 10 nm Ni layer evaporation without in situ O2-plasma oxidation, 10 1714

nm Ni layer evaporation followed by in situ O2-plasma oxidation, and upper 2 nm Alq3 layer evaporation on a 10 nm Ni layer evaporation followed by in situ O2-plasma oxidation on a 35 nm lower Alq3 layer and Al electrode. The chemical composition was characterized by X-ray photoelectron spectroscopy with a 0.1 × 0.4 mm beam diameter and 2-3 nm depth resolution. In addition, the chemical composition profile from the upper small-molecule (Alq3) layer to lower small-molecule layer was characterized by Auger electron spectroscopy with a 10 nm beam diameter. The dc I-V characteristics and ac characteristics for the memory cells were measured using Agilent 4155C semiconductor parameter analyzer. 4F2 Memory Cell Structural and Chemical Characteristics. A cross-sectional transmitted electron microscopy (x-TEM) image of one of the 4F2 memory cells we fabricated is shown in Figure 1d. The thicknesses of the upper Alq3 layer, the layer of Ni nanocrystals surrounded by NiO tunneling barrier, and the lower Alq3 layer were about 30, 10, and 35 nm, respectively. A magnified x-TEM image (obtained at a 1.2 MV acceleration voltage) of the memory cell shown in Figure 1d is shown in Figure 1e, where the height and width of the uniformly distributed Ni nanocrystals are seen to be about 10 nm. The nanocrystals were well isolated from one another by the surrounding insulating NiO, and the isolation distance was 4-5 nm. The crystal structure of the metal nanocrystals was a face-centered cubic of pure poly nickel, dominantly composed of (111), (200), and (220) (see the µ-diffraction pattern in Figure 1e). The crosssectional profile of the chemical composition for the 4F2 memory cell shown in Figure 1d was determined using Auger electron spectroscopy (AES) and is shown in Figure 1f. The middle nanocrystal layer is seen to have been mainly composed of Ni, and the oxygen concentration at the interface between the small-molecule layer (Alq3) and Ni nanocrystal layer was higher than that at the center of the Ni nanocrystal layer. From parts e and f of Figure 1, the surface and bottom of the Ni nanocrystal layer were obviously well oxidized and the Ni nanocrystals were well isolated by NiO tunneling barrier. To confirm the chemical composition profile of the metal nanocrystal surface, three samples were prepared: 10 nm Ni layer evaporation without in situ O2-plasma oxidation, 10 nm Ni layer evaporation followed by in situ O2-plasma oxidation, and upper 2 nm Alq3 layer evaporation on 10 nm Ni layer evaporation followed by in situ O2-plasma oxidation on 35 nm lower Alq3 layer and Al electrode. The chemical composition (Figure 1g) was obtained using X-ray photoelectron spectroscopy (XPS). The Ni layer surface without in situ O2plasma oxidation was composed of Ni (851.7 eV binding energy) while the Ni layer surface with in situ O2-plasma oxidation was composed of NiO (854.6 eV binding energy). In addition, the surface of the upper 2 nm Alq3 layer evaporation on 10 nm Ni layer evaporation followed by in situ O2-plasma oxidation was mainly composed of NiCO3 (855.4 eV binding energy) and Ni2O3 (855.6 eV binding energy). These results indicate that in situ O2-plasma oxidation oxidizes the surrounding surface of Ni nanocrystals Nano Lett., Vol. 9, No. 4, 2009

Figure 2. Electrical characteristics of a small-molecule nonvolatile 4F2 memory cell: dc I-V curves.

well, and following Alq3 evaporation chemically reacts well with NiO, probably improving the endurance characteristics of the nonvolatile memory cell because of a good interface characteristic between the tunneling barrier (NiO) and smallmolecule layer (Alq3). From parts d-g of Figure 1, one can infer that in situ O2-plasma oxidation caused oxygen radicals to diffuse through the Ni grain boundaries and oxidize the Ni nanocrystals, producing approximately 10 nm Ni nanocrystals surrounded by a NiO tunneling barrier of 4-5 nm. Single-Layer 4F2 Memory Cell Electrical Characteristics. The dc current versus voltage characteristics for the small-molecule memory cell shown in Figure 1 are shown in Figure 2. The current increased slightly with the applied bias from 0 V up to Vth (2.2 V). This current is called Ioff, and the state of the memory cell when Ioff flows is called either the low-current state or the high-resistance state. The current abruptly increased with the applied bias from Vth to Vp (4.0 V program voltage) and then decreased with increasing applied bias from Vp to Ve (7.3 V erase voltage), showing a negative differential resistance (NDR) region (see the inset in Figure 2). Finally, the current increased slightly with the applied bias above Ve (7.3 V up to 10.0 V), which followed another low current state (called the second low current state). Note that the current path of Ioff is extended to the second low current state if the memory cell is fabricated with only a conductive small-molecule layer (Alq3) without embedded Ni nanocrystals surrounded by a NiO tunneling barrier. After the first applied voltage was swept from 0 to 10.0 V (called erase), shown in (1) in Figure 2, the second applied bias was swept from 0 to Vp (4.0 V program voltage), called program, and then the current followed the low current state (Ioff). After the second applied bias 0 to Vp was swept, shown in (2) in Figure 2, the third applied bias (reading after program) was swept from 0 to Vp (4.0 V) again, shown in (3) in Figure 2, and then the current followed Ion (called either high current state or low resistance state). Therefore, the memory cell can read Ioff (the low current state) at 1.0 V after erase (biasing above Ve). Otherwise, the memory cell reads Ion (the high current state) at 1.0 V after program (biasing Vp), which can behave as a nonvolatile memory cell. Surprisingly, the ratio of Ion to Ioff (memory margin) was ≈1.2 × 103, which is a sufficient Nano Lett., Vol. 9, No. 4, 2009

current difference (bistable resistance difference) for MLC nonvolatile memory-cell behavior. After the third applied bias (0 to Vp) was swept, the fourth applied bias was swept from 0 to Vint1 (5.0 V), shown in (4) in Figure 2, and called a program of intermediate current state 1. The current then followed Ion. After the fourth applied bias (0 to Vint1) was swept, the fifth applied bias (reading after the program of intermediate current state 1) was swept from 0 to Vint1 (5.0 V) again, shown in (5) in Figure 2, and then the current followed Iint1 (intermediate current state 1). Thus, the memory cell can read Iint1 (intermediate current state 1) at 1.0 V after the program of intermediate current state 1 (biasing Vint1). After the fifth applied bias (0 to Vint1), the sixth applied bias was swept from 0 to 6.0 V, shown in (6) in Figure 2, and called a program of intermediate current state 2. The current then followed Iint1. After the sixth applied bias (0 to 6.0 V), the seventh applied bias (reading after the program of intermediate current state 2) was swept from 0 to 6.0 V again, shown in (7) in Figure 2, and then the current followed Iint2 (intermediate current state 2). Thus, the memory cell can read Iint2 (intermediate current state 2) at 1.0 V after the program of intermediate current state 2 (biasing Vint2). After the seventh applied bias (0 to Vint2), the eighth applied bias (erase after the program of intermediate current state 2) was swept from 0 to 10.0 V, shown in (8) in Figure 2, and then the current followed Iint2, Vth, Vp, NDR region, Ve, and the second low current state. After the eighth applied bias (0 to 10.0 V), the ninth applied bias (reading after erase) was swept from 0 to 10.0 V again, shown in (9) in Figure 2, and then the current followed Ioff, Vth, Vp, NDR region, Ve, and the second low current state. Finally, after the ninth applied bias (0 to 10.0 V), the tenth applied bias was swept from 10.0 to 0 V, and then the current followed the second low current state, Ve, NDR, Vp, and Ion. In summary, this memory cell could operate as a multilevel (four levels) nonvolatile memory cell by reading the current at 1.0 V after biasing Ve, Vint1, Vint2, and Vp. Multistable dc I-V characteristics that were nearly mirror images of those observed when the positive bias was applied were observed when negative bias was applied. This memory cell showed reproducible current paths for three program states and erase, indicating that it could be used as a commercially available nonvolatile memory. In particular, the current path of this memory cell followed the low current state, high current state, low current state, and high current state when the applied bias was swept from 0 to 10, 0, -10, and 0 V, respectively (see (1) and (10) for both positive and negative bias regions in Figure 2). This hysteresis of the current path means that electrons are charged on Ni nanocrystals during program and are discharged on the nanocrystals during erase, which will be explained later. The dc retention time characteristic obtained from another 4F2 memory cell is presented in Figure 3a. Four reading current states at 1 V, Ioff, Iint2, Iint1, and Ion are presented at ≈1.0 × 10-6 A after the erase of 10.0 V, ≈1.0 × 10-5 A after the program of intermediate current state 2 of 4.5 V, ≈1.0 × 10-4 A after the program of intermediate current state 1 of 4.0 V, and ≈1.0 × 10-3 A after the program of 1715

Figure 3. Electrical characteristics of small-molecule nonvolatile 4F2 memory cell: (a) dc retention, and (b) dc endurance.

which is a typical current conduction mechanism of conductive small-molecule material32 (Figure 5a). Note that the SCLC is proportional to

()

J∝d

V d2

n

where d is a conductive small-molecule layer thickness, V is the applied bias between the top and bottom electrode, and n is an index. The index of the SCLC was ≈2.41 for the region from 0 to Vth. The current conduction mechanism for the region from Vth to Vp correlated well with FowlerNordeim (F-N) tunneling (Figure 5b). Note that the F-N tunneling current was proportional to Figure 4. Electrical characteristics of small-molecule nonvolatile 4F2 memory cell: temperature dependence of dc I-V curves.

3.5 V, respectively. This memory cell showed a retention time of 105 s for four current states, which is probably extended to 10 years (∼1 × 108 s). The dc endurance characteristic of another 4F2 memory cell is shown in Figure 3b. A memory margin (Ion/Ioff ratio) of about 2.1 × 103 was sustained for up to about 5 × 102 erase-and-program cycles, which is probably extended to 1.1 × 103 after erase-andprogram cycles of 1 × 105. Nonvolatile Memory Cell Operation Mechanism. To investigate the mechanism of nonvolatile memory-cell behavior, we investigated the temperature dependence of the dc I-V characteristics for another 4F2 memory cell (Figure 4). The nonvolatile memory-cell behavior (the difference between Ion and Ioff) disappeared at ≈220 K, indicating the thermal emission of electrons trapped on the Ni nanocrystals. The activation energies for thermal emission of electron charge (Vth-Vp) and discharge (Vp-VNDR) on the Ni nanocrystals were, respectively, 0.213 and 0.128 eV. In addition, the current of the memory cell increased with a decreased memory-cell operation temperature of less than 220 K, indicating that the Ni nanocrystal layer behaved as a metal layer. Furthermore, the detailed current conduction mechanism for each of the memory-cell operation regions was correlated (Figure 5). The current conduction mechanism for the regions from 0 V to Vp (Ioff, low current state) correlated well with space charge limited current (SCLC), 1716

( Vb )

J ∝ V2 exp -

where b is a constant and V is the applied voltage.33 b is proportional to ΦB2/3, where ΦB is the tunneling barrier height. In particular, the current at Vp decreased with the memory cell operation temperatures up to 220 K and was saturated, indicating that the current conduction mechanism at Vp follows thermionic emission,33 as shown in Figure 5c. From parts b and c of Figure 5, the current conduction mechanism for Vth to Vp obviously follows thermionic field emission, which is the F-N tunneling of thermally excited electrons. A comparison of Figures 1e, 5b, and 5c confirms that electrons are charged on the Ni nanocrystals through tunneling the NiO barrier. The current conduction mechanism for the region from Vp to Ve (NDR region) also correlated well with F-N tunneling (Figure 5d). From parts c and d of Figure 5, the current conduction mechanism for the NDR region is also evidently associated with thermionic field emission. This result also indicates that electrons on Ni nanocrystals are discharged by tunneling the NiO barrier, and the mechanism of this discharge is probably associated with the Simmon-Verderber theory.34 Furthermore, the region above Ve followed the conduction mechanism of SCLC where the index of the SCLC was ≈3.84. The regions for Ion, Iint1, and Iint2 followed the current conduction mechanism of SCLC (see parts f-h of Figure 5), where the indexes of SCLC were 1.78, 1.77, and 1.76, respectively. In addition, the regions for Ion, Iint1, Nano Lett., Vol. 9, No. 4, 2009

Figure 5. Current conduction mechanism for small-molecule nonvolatile memory behavior: (a) Ioff state (low-current state), (b) Vth Vp, (c) Vp, (d) VNDR - Ve, (e) Ve - 10 V, (f) Ion state (high-current state), (g) Iint1 state (intermediate state 1), and (h) Iint2 state (intermediate state 2).

and Iint2 depended on the operation temperature (Figure 4). Thus, the current conduction mechanism for the regions of Ion, Iint1, and Iint2 were associated with the conduction mechanism of SCLC combined with thermionic emission. Double-Stacked 4F2 Memory Cell Electrical Characteristics. A vertically double-stacked 4F2 memory cell was developed with a sandwiched device structure of an upper 80 nm Al electrode, upper small-molecule layer (Alq3) embedded Nano Lett., Vol. 9, No. 4, 2009

with Ni nanocrystals surrounded by NiO tunneling barrier, middle 80 nm Al electrode, lower small-molecule layer (Alq3) embedded with Ni nanocrystals surrounded by NiO tunneling barrier, and lower 80 nm Al electrode. The dc current vs voltage characteristics obtained from the lower 4F2 memory cell is shown in Figure 6b. The Vth, Vp, NDR region, Ve, and memory margin (Ion/Ioff) for the lower memory cell were 2.1, 3.9, 3.9-6.8, 6.8, and 1.8 × 103, respectively. This lower memory 1717

Figure 6. Electrical characteristics of vertically double stacked small-molecule nonvolatile 4F2 memory cell: (a) perspective view of memory cell. Lower memory cell: (b) dc I-V curves, (d) dc retention, and (f) dc endurance. Upper memory cell: (c) dc I-V curves, (e) dc retention, and (g) dc endurance.

cell demonstrated good MLC operation, reading reproducible four current states after erase, intermediate state 1 program, intermediate state 2 program, and program. The dc retention 1718

time characteristic obtained from the lower memory cell is shown in Figure 6d. Four reading current states at 1.0 V, Ioff, Iint2, Iint1, and Ion are presented at ≈2.0 × 10-6 A after Nano Lett., Vol. 9, No. 4, 2009

the erase of 10.0 V, ≈2.0 × 10-5 A after the program of intermediate current state 2 of 6.1 V, ≈2.0 × 10-4 A after the program of intermediate current state 1 of 5.0 V, and ≈2.0 × 10-3 A after the program of 4.0 V, respectively. The cell showed the retention time of 105 s for four current states, which probably extended to 10 years (∼1 × 108 s). The dc endurance characteristic of the lower 4F2 memory cell is shown in Figure 6f. A memory margin (Ion/Ioff ratio) of about 4.4 × 102 was sustained for up to about 5 × 102 erase-and-program cycles, probably extending to 1.9 × 102 after erase-and-program cycles of 1 × 105. The dc current versus voltage characteristics obtained from the upper 4F2 memory cell are shown in Figure 6c. The current level for the upper cell was approximately 2 orders lower than that for the lower memory cell because the thickness of the smallmolecule (Alq3) layer (∼41 nm) for the upper memory cell was ∼41% thicker than that for the lower memory cell (∼29 nm). The Vth, Vp, NDR region, Ve, and memory margin (Ion/ Ioff) for the lower memory cell were 2.0, 3.4, 3.4-6.5, 6.5, and 3.8 × 103, respectively. The upper memory cell also demonstrated good MLC operation, reading reproducible four current states after erase, intermediate state 1 program, intermediate state 2 program, and program. The dc retention time characteristic obtained from the upper memory cell is shown in Figure 6e. Four reading current states at 1 V, Ioff, Iint2, Iint1, and Ion, are presented at ≈1.0 × 10-8 A after the erase of 10 V, ≈2.0 × 10-7 A after the program of intermediate current state 2 of 6.5 V, ≈2.0 × 10-6 A after the program of intermediate current state 1 of 5.5 V, and ≈3.0 × 10-5 A after the program of 4.0 V, respectively. The upper memory cell showed the retention time of 105 s for four current states, which was probably extended to 10 years (∼1 × 108 s). The dc endurance characteristic of the upper memory cell is shown in Figure 6g. A memory margin (Ion/Ioff ratio) of about 2.1 × 102 was sustained for up to about 5 × 102 erase-and-program cycles, probably extending to 1.9 × 102 after erase-and-program cycles of 1 × 105. Parts b-g of Figure 6 show that the vertically double-stacked 4F2 memory cell in Figure 6a could operate eight current levels at reading after erase, intermediate state 1 program, and intermediate state 2 program that corresponds to doublestacked MLC (four-level cell) or three-bit nonvolatile memory cell behavior in a memory cell feature size of 4F2. Summary and Outlook. This is the first demonstration of the feasibility of terabit-level nonvolatile small-molecule memory. In summary, the small-molecule (Alq3) memory cell embedded with Ni nanocrystals surrounded by a NiO tunneling barrier between the top and bottom electrodes showed excellent reproducible nonvolatile memory-cell characteristics, such as retention time and endurance of erase-and-program cycles. In particular, in situ O2-plasma oxidation following Ni evaporation effectively produced a NiO tunneling barrier surrounding Ni nanocrystals embedded in the small-molecule (Alq3) layer, resulting in a NDR region in the current versus voltage characteristics and providing multilevel cell operation of nonvolatile memory. In addition, the memory characteristics, such as Vth, Vp, Ve, Ioff, Iint1, Iint2, and Ion, were very sensitive to small-molecule layer thickness, Ni nanocrystal size, and NiO Nano Lett., Vol. 9, No. 4, 2009

tunneling barrier thickness. Furthermore, the retention time characteristic could be improved by choosing a metal nanocrystal with deeper work function (e.g., Ni instead of Al) and the endurance characteristic of erase-and-program cycles could be improved by introducing a chemically reactive interface between the smallmolecule (Alq3) layer and NiO tunneling barrier. Acknowledgment. This research was supported by “The National Research Program for Terabit Nonvolatile Memory Development” sponsored by the Korean Ministry of Knowledge Economy. References (1) Ma, L. P.; Liu, J.; Yang, Y. Appl. Phys. Lett. 2002, 80, 2997–2999. (2) Bozano, L. D.; Kean, B. W.; Deline, V. R.; Salem, J. R.; Scott, J. C. Appl. Phys. Lett. 2004, 84, 607–609. (3) He, J.; Ma, L.; Wu, J.; Yang, Y. J. Appl. Phys. 2005, 97, 064507. (4) Park, J. G.; Lee, G. S.; Chae, K. S.; Kim, Y. J.; Miyata, T. J. Korean Phys. Soc. 2006, 48, 1505–1508. (5) Ma, L. P.; Pyo, S.; Ouyang, J.; Xu, Q.; Yang, Y. Appl. Phys. Lett. 2003, 82, 1419–1421. (6) Wu, J.; Ma, L.; Yang, Y. Phys. ReV. B 2004, 69, 115321. (7) Tseng, R. J.,; Huang, J.; Ouyang, J.; Kaner, R. B.; Yang, Y. Nano Lett. 2005, 5, 1077–1080. (8) Bozano, L. D.; Kean, B. Y.; Beinhoff, M.; Carter, K. R.; Rice, P. M. AdV. Funct. Mater. 2005, 15, 1933–1939. (9) Ouyang, J. Y.; Chu, C. W.; Szmanda, C. R.; Ma, L. P.; Yang, Y. Nat. Mater. 2004, 3, 918–922. (10) Tseng, R. J.; Tsai, C.; Ma, L. P.; Ouyang, J.; Ozkan, C. S.; Yang, Y. Nat. Nanotechnol. 2006, 1, 72–77. (11) Yang, Y.; Ouyang, J. Y.; Ma, L. P.; Chu, C. W. AdV. Funct. Mater. 2006, 16, 1001. (12) Scott, J. C.; Bozano, L. D. AdV. Mater. 2007, 19, 1452–1463. (13) Kang, S. H.; Crisp, T.; Kymissis, I.; Bulovic´, V. Appl. Phys. Lett. 2004, 85, 4666–4668. (14) Chen, J.; Xu, L.; Lin, J.; Geng, Y.; Wang, L.; Ma, D. Appl. Phys. Lett. 2006, 89, 83514. (15) Chen, J.; Ma, D. J. Appl. Phys. 2004, 100, 34512. (16) Lauters, M.; McCarthy, B.; Sarid, D.; Jabbour, G. E. Appl. Phys. Lett. 2005, 87, 231105. (17) Tu, C. H.; Lai, Y. S.; Kwong, D. L. Appl. Phys. Lett. 2006, 89, 62105. (18) Mu¨ller, R.; Genoe, J.; Heremansb, P. Appl. Phys. Lett. 2006, 88, 242105. (19) Guo, P.; Dong, Y. W.; Ji, X.; Lu, Y. X.; Xu, W. IEEE Electron DeVice Lett. 2007, 28, 572–574. (20) Kever, T.; Bo¨ttger, U.; Schindler, C.; Waser, R. Appl. Phys. Lett. 2007, 91, 83506. (21) Caironi, M.; Natali, D.; Sampietro, M. Appl. Phys. Lett. 2006, 89, 243519. (22) Tu, D.; Shang, L.; Liu, M.; Wang, C.; Jiang, G.; Song, Y. Appl. Phys. Lett. 2007, 90, 052111. (23) Chen, L.; Chen, L.; Xia, Y.; Liang, X.; Yin, K.; Yin, J.; Liu, Z. Appl. Phys. Lett. 2007, 91, 073511. (24) Chu, C. W.; Ouyang, J.; Tseng, J. H.; Yang, Y. AdV, Mater. 2005, 17, 1440–1443. (25) Lauters, M.; McCarthy, B.; Sarid, D. Appl. Phys. Lett. 2006, 89, 013507. (26) Paul, S. IEEE Trans. Nanotechnol. 2007, 6, 191. (27) Lai, Y. S.; Tu, C. H.; Kwong, D. L. Appl. Phys. Lett. 2005, 87, 122101. (28) Ling, Q. D.; Lim, S. L.; Song, Y.; Zhu, C. X.; Chan, D. S.; Kang, E. T.; Neoh, K. G. Langmuir 2007, 23, 312–319. (29) Baek, S. S.; Lee, D. J.; Kim, J. Y.; Hong, S. H.; Kim, O. H.; Ree, M. AdV. Func. Mater. 2007, 17, 2637–2644. (30) Ling, Q. D.; Song, Y.; Ding, S. J.; Zhu, C.; Chan, D. S. H.; Kwong, D. L.; Kang, E. T.; Neoh, K. G. AdV. Mater. 2005, 17, 455–459. (31) Mu¨ller, S.; Perlov, C.; Jackson, W.; Taussig, C.; Forrest, S. R. Nature (London) 2003, 426, 166–169. (32) Molecular Electronics from principles to practice, 1st ed.; Petty, M. C., Ed.;John Wiley & Sons Ltd.: Chichester, 2007. (33) Sze, S. M.; Ng, K. K. Physics of Semiconductor DeVices, 3rd ed.; John Wiley & Sons: Hoboken, NJ, 2007. (34) Simmons, J. G.; Verderber, R. R. Proc. R. Soc. London, Ser. A 1967, 301, 77–102.

NL900429H 1719