Atomic layer deposition of thin oxide films for

ECS Transactions, 58 (10) 163-170 (2013) 10.1149/05810.0163ecst ©The Electrochemical Society

Atomic layer deposition of thin oxide films for resistive switching K. Fröhlicha, P. Jančovič, B. Hudeca, J. Dérer, A. Paskalevab, T.Bertaudc, T. Schroederc, d a

Institute of Electrical Engineering, SAS, Dúbravská cesta 9, 841 04 Bratislava, Slovakia Institute of Solid State Physics, BAS, 72 Tzarigradsko Chaussee, 1784 Sofia, Bulgaria c IHP, Im Technologiepark 25, 15236 Frankfurt (Oder), Germany d BTU, Konrad Zuse Strasse 1, 03046 Cottbus, Germany

b

Atomic layer deposition was used for preparation of TiO2 and HfO2 thin films for resistive switching in metal-insulator-metal structures with Pt and TiN top and bottom electrodes, respectively. To obtain stable bipolar resistive switching loops in TiO2-based structures Al2O3 barrier with the thickness of 3 - 5 nm was necessary. HfO2based structures with the insulator thickness less than 10 nm exhibited stable bipolar resistive switching. Ratio between high resistivity and low resistivity state varied between 20 and 100 depending on structure preparation and composition as well as on parameters of DC current – voltage measurement. Resistive switching effect was demonstrated in metal-insulator-metal structures with HfO2 layers thickness below 3 nm.

Introduction Resistive switching (RS) in transition metal oxide based metal-insulator-metal (MIM) structures have recently attracted considerable interest. The RS effect has been proposed as a basis for future non-volatile memories. Memory cell based on RS combines advantages of a Flash and DRAM type memories i.e. nonvolatility and fast response. Unfortunately, physical mechanism responsible for RS is not clear up to now, albeit a lot of investigations aiming at its elucidation have been performed in the last period [1-5]. RS memory cell is based on a capacitor-like metal-insulator-metal (MIM) structure with transition metal oxide sandwiched between two electrodes. The RS cell structure is simple; it can be prepared at low temperatures by CMOS compatible back end of line process and easily scaled down. Random access memory based on RS has been therefore identified by the International Technology Roadmap for Semiconductors as one of the promising emerging memories. It is envisioned that high speed, potentially low power nonvolatile RS based RAM can combine properties of today Flash and DRAM memories and outperform their performance. Operation of the RS based memory cell is based on the reversible change between two resistivity states induced by applied voltage. As a first step “soft” breakdown termed as electroforming should be performed. Transition from the low resistivity state (LRS) to the high resistivity state (HRS) is realized by a RESET process while switching from the high resistivity state back to the low resistivity state proceeds by SET process. If both SET and RESET occur at the same polarity of the applied voltage, the switching mode is defined as unipolar RS. Alternatively, bipolar RS requires applying of the opposite polarity bias voltage to reset the cell to from the LRS to the HRS. In the case of bipolar switching

Downloaded on 2013-10-20 to IP 194.95.141.200 address. Redistribution 163 subject to ECS license or copyright; see ecsdl.org/site/terms_use

ECS Transactions, 58 (10) 163-170 (2013)

negative differential resistance can be observed at RESET transition. Bipolar RS attracts interest in the recent period due to its lower reset current and improved operation stability. Several transition metal oxides were examined as insulating layers in RS devices. TiO2 is one of the earliest materials studied for RS applications. Bipolar RS in TiO2-based RS cells was observed in several studies [6-10]. MIM structures with HfO2 as insulating film were intensively studied in the last period due to CMOS compatibility and excellent switching properties [11-14]. TiN and Pt electrodes were often employed in both types of RS cells. TiN, which can be easily oxidized during the dielectric growth, acts as an oxygen reservoir due to its high affinity to oxygen while Pt serves as inert electrode that exhibits high Schottky barrier with TiO2 and HfO2 films. In our contribution we present the preparation and properties of RS MIM structures comprising TiN bottom electrode, TiO2 or HfO2 dielectrics, and Pt top electrode. The TiO2 and HfO2 thin films were prepared by atomic layer deposition (ALD). We analyze the effect of electrical parameters on DC RS loops and we present RS properties of the TiO2- and HfO2-based structures. We show that ALD is a promising technique for preparation of RS cell because structures prepared using this technique exhibit very good characteristics.

Experimental TiN bottom electrode was reactively sputtered in Ar/N2 plasma at temperature of 200 °C. Thickness of the layer was 70 nm and resistivity about 200 cm. Pt top electrode was evaporated using electron gun at room temperature through the shadow mask. Thickness of the Pt electrode was 30 nm and was capped by 30 nm of Au. TiO2, HfO2 and Al2O3 thin films were prepared by ALD at 300 °C in Beneq TFS 200 equipment. Titanium isopropoxide (TTIP) and trimethylaluminium (TMA) were used as precursors for ozone assisted ALD mode to grow the TiO2 and Al2O3 films, respectively. TTIP precursor was kept at temperature of 60 °C to ensure sufficient vapor pressure. Growth per cycle (GPC) of 0.04 nm was attained for the TiO2 deposition using purging times of 10 s after both precursor dosing and ozone pulses. Similarly, GPC of 0.1 nm was observed for the Al2O3 growth with the same purging pulses duration. For the preparation of the HfO2 films tetrakis (ethylmethylamino) hafnium (TEMAH), was employed as a precursor. TEMAH precursor was heated to 70 °C to provide sufficiently high vapor pressure. Thermal, plasma and ozone ALD modes were used to deposit HfO2 films. GPC of 0.08 nm was obtained for thermal mode using purging times of 10 s. Plasma mode gave the highest GPC of 0.1 nm for the plasma mode with purging time of 10 s. Ozone mode resulted in GPC of 0.07 nm for the same purging time periods. Phase composition of the films was analyzed using grazing incidence X-ray diffraction on Bruker AXS-D8 Discover. Thickness of the films was determined using X-ray reflectivity using the same equipment. Electrical characterization of the prepared MIM structures was performed using Keithley 4200 Semiconductor Characterization System. The top electrodes prepared by evaporation using shadow mask had dimensions of 100 x 100 and 300 x 300 m2. After an initial electroforming using a current compliance DC RS loops were acquired by sweeping with DC bias voltage between Vstop(SET) and Vstop(RESET). During the measurement the top electrode was always biased and the bottom electrode was grounded.

164


Results and discussion Structure of the ALD grown films Figures 1a and 2a display X-ray diffraction patterns of the TiO2 and HfO2 films deposited by ALD at 300 °C. It can be seen, that while TiO2 films crystallize in anatase phase, HfO2 films grown at the same temperature exhibit amorphous character. Fig. 1b shows X-ray reflectivity of the TiO2 film with thin Al2O3 layer on the top. Fitting of data resulted in determination of 32 nm for the TiO2 and 3 nm for the Al2O3 layers. X-ray reflectivity for amorphous HfO2 film with the thickness 13 nm is displayed in the Fig. 2b. Effect of measurement parameters of RS loops It is widely accepted that RS is caused by a conductive filament (CF) typically tens or hundreds of nanometers in diameter which is formed during the electroforming process and which can change its resistive state in dependence on the polarity and value of the

10

10

215/301

116 220

213

105 211

200

103 112

TiO2 anatase

counts (a.u.)

4

101

counts (a.u.)

10

3

0

10

3.5 nm Al2O3 + 32 nm TiO2 -2

10

-4

10

-6

2

10

-8

a)

20

30

40

50

60

70

10

80

0

1

2

3

4

b)

2 theta (deg)

5

6

7

8

2 theta (deg)

10

10

3

counts (a.u.)

counts (a.u.)

Fig. 1. Grazing incidence X-ray diffraction pattern of the 32 nm thin TiO2 film, a), and X-ray reflectivity of the 32 nm TiO2 + 4 nm Al2O3 structure, b). Indexes in the X-ray diffraction pattern denote anatase phase.

2

0

10

13 nm HfO2 -2

10

-4

10

10

a)

1

20

30

40

50

60

70

-6

10

80

2 theta (deg)

b)

0

1

2

3

4

5

2 theta (deg)

Fig. 2. Grazing incidence X-ray diffraction pattern of the 13 nm thin HfO2 film a), and X-ray reflectivity of the 13 nm HfO2 layer.

165

6


Current (A)

Current (A)

applied voltage. Although RS phenomenon is still a subject of intense studies, the acquired knowledge is far from enough to allow fabrication of reliable RS devices with predictable and controllable behavior. The reason for that is the strong dependence of RS effect on the stack parameters, e.g. switching material, top and bottom electrodes materials, thickness of layers, deposition techniques, etc. This is further complicated by the strong dependence on the measurement parameters like applied voltage in SET and RESET direction, compliance current, temperature. The existence of other processes which proceeds in parallel with the RS could also significantly affect the phenomena observed. In Fig. 3a-d the RS in Pt/HfO2/TiN structures in dependence on the stop voltage to which SET curve is swept, Vstop(SET), is presented. In this stack it was possible to perform SET process without any current compliance applied. Transition to LRS (SET) is observed at 0.7 V, while RESET (decrease of the current due to negative differential resistance in positive polarity) takes place below +1 V, Fig. 3a. There are two observations to be pointed out. First, with increasing Vstop(SET), the RESET voltage also increases, e.g. for Vstop(SET)=-1 V the RESET is less than 1 V; for Vstop(SET)=-1.5 V, it is at about 1.3 V. This means that the RESET voltage could be controlled at least within some voltage range by changing the applied voltage in SET direction. The second point to be mentioned is the decrease of HRS/LRS ratio from ~90 to ~20, when changing Vstop(SET) from -1 to -1.5 V. This result implies that the 0.01

1E-3

0.01

1E-3

1E-4

1E-4

1E-5

1E-5

Vstop(SET) = -1 V

Vstop(SET) = -1.1 V

Vstop(RESET) = +1.8 V

1E-6


1E-6

HRS/LRS = 87 1E-7 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

-1.0

-0.5

0.0

0.01

1E-3

1E-4

1.5

2.0

0.01

1E-3

1E-5

Vstop(SET) = -1.2 V

Vstop(SET) = -1.5 V


1E-6


1E-6

HRS/LRS = 33

c)

1.0

Voltage (V)

1E-4

1E-5

1E-7 -1.5

0.5

b)

Voltage (V)

Current (A)

Current (A)

a)

HRS/LRS = 56 1E-7 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

HRS/LRS = 20 1E-7 -1.5

2.0

d)

Voltage (V)

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

Voltage (V)

Fig. 3. Resistive switching loops of the HfO2-based MIM structures for various Vstop(SET) voltages ranging from -1V to -1,5 V. Vstop(RESET) voltage was kept constant 1.8 V.

166

Current (A)

Current (A)


0.01

1E-3

0.01

1E-3

1E-4

1E-4

1E-5

1E-5

Vstop(SET) = -1.5 V

Vstop(SET) = -1.5 V Vstop(RESET) = 1.3 V

1E-6

Vstop(RESET) = 0.8 V

1E-6

HRS/LRS = 30

HRS/LRS = 16

a)

1E-7 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

1E-7 -1.5

2.0

-1.0

b)

Voltage (V)

-0.5

0.0

0.5

1.0

1.5

2.0

Voltage (V)

Fig. 4. Resistive switching loops of the HfO2-based MIM structures for RESET voltages 1.3 and 1.5 V. Current compliance was set to 5 mA during SET operation.

measurement conditions should be carefully optimized in order to maximize RS effect. For some Pt/HfO2/TiN samples it is observed that the RS ratio could be changed also by changing the applied voltage in RESET direction. Fig. 4 demonstrates that by changing Vstop(RESET) from 1.3 to 1.5 V it is possible to increase HRS/LRS ratio from ~16 to ~30. The RESET position remains unchanged. This opens up the way to fabricate multilevel RS devices. Very similar behavior of the RS loops characteristics on Vstop(SET) and Vstop(RESET) parameters was observed also for TiO2-based MIM structures. Resistive switching in TiO2-based MIM structures

1E-2

SET

Pt/TiO2/TiN

RESET

1E-3 1E-4

1 000

1E-3

LRS current 100

1E-4

HRS/LRS

1E-5

10

HRS current

1E-6

1E-5

1E-7 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

a)

HRS/LRS

1E-1

Current @ -0.2 V (A)

Current (A)

TiO2 films with the thickness 30 nm on TiN bottom electrode exhibited rather high leakage currents of the order of 10-2 A/cm2 at 1 V. It is very likely connected with the polycrystalline anatese character of the film and high amount of defects. High leakage currents prevented to perform electroforming and RS. To suppress leakage currents thin Al2O3 layers were grown on top of the TiO2 films. Al2O3 films with the thickness of 3 to 5 nm were sufficient to decrease the leakage current density by several orders of magnitude.

0

20

40

60

80

1 100

Number of RS loops

Voltage (V)

b)

Fig. 5. Resistive switching loops of the MIM structure with 32 nm TiO2 + 3 nm Al2O3, a), and corresponding currents in HRS and LRS at reading voltage of -0.2 V and their ratio b).

167


Electroforming of the Al2O3/TiO2 was a two step process. First, positive forming step with certain current compliance (at least 100 A) was necessary in order to achieve proper electroforming in the subsequent negative forming step. Forming was not possible without the first (positive) forming step, during which the electron injection from TiN to TiO2 took place filling the traps in TiO2. After electroforming, the TiO2 films with 3 - 5 nm of Al2O3 barrier exhibited stable bipolar DC RS loops, Fig. 5. Operating currents were of the order of 10-3 A/cm2 with SET and RESET transitions below 1 V. High resistivity state to low resistivity state ratio at the reading voltage +0.2 V was between 50 and 90. Very similar resistive loops were obtained for the TiO2 films with the Al2O3 barrier 4 and 5 nm. We suppose that in these stacks the TiO2 layer is acting as an oxygen sink which helps to stabilize the filamentary RS which is probably localized in the Al2O3 layer. Independence of the operating currents on the thickness of Al2O3 (3-5 nm) supports this hypothesis. Resistive switching in HfO2-based MIM structures

Pt/HfO2/TiN SET

0.01

RESET

1E-3 1E-4

1E-3

1000

LRS current 1E-4

100

HRS/LRS LRS current

1E-5

1E-5

HRS/LRS

0.1

Current @ +0.2 V

Current (A)

Leakage current density in the HfO2 layers is several orders of magnitude lower

10

1E-6 1E-7 -2.0

1E-6 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

1

0

20

40

60

80

100

Number of loops

Voltage (V)

a)

b)

SET

0.01

Pt/HfO2/TiN

1000

RESET

LRS current

1E-3

1E-3 1E-4

100

HRS/LRS

1E-4

1E-5

HRS/LRS

0.1

Current @ +0.2 V

Current (A)

Fig. 6. Resistive switching loops of the MIM structure with 5.4 nm thin HfO2 film, a), and corresponding LRS and HRS currents and their ratio at the reading voltage 0.2 V, b).

10

HRS current

1E-6 1E-7 -2.0

a)

1E-5 -1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

Voltage (V)

0

b)

5

10

15

1 20

Number of loops

Fig. 7. Resistive switching loops of the MIM structure with 3 nm thin HfO2 film, a), and corresponding LRS and HRS currents and their ratio at the reading voltage 0.2 V, b).

168


comparing to TiO2 films. Consequently, we reduced thickness of the HfO2 films below 10 nm. Corresponding electroforming voltage was therefore reduced in comparison to TiO2 films. Fig. 6 shows RS loops of the MIM structure with 5.4 nm thin HfO2 film grown by plasma ALD. The structure exhibits stable RS with HRS/LRS ratio of 100 at the reading voltage +0.2 V. Similar RS loops were observed in MIM structures with HfO2 films grown by ozone and thermal ALD. Low leakage current density of the HfO2 films makes it possible to further reduce its thickness. A number of samples with HfO2 with thicknesses of about 2.3 to 3.5 nm have been investigated. In all the samples stable RS has been observed. Fig. 7 displays RS loops of the MIM structure with 3 nm thin HfO2 film. The MIM structure exhibits stable RS loops with HRS/LRS of 20. Even the thinnest sample (2.3 nm) has shown RS with HRS/LRS ratio of about 8-9. Electroforming of these structures occurs at the voltage close to -1 V. Therefore, the results reveal that RS phenomenon occurs in a very confined region and can exist even in an ultrathin films with a thickness below 3 nm. Interestingly, similarly as for Al2O3/TiO2 dielectrics, operation current of all MIM structures was in the order of mA.

Summary In summary, we have demonstrated that ALD is suitable deposition technique for preparation of dielectric films for RS. TiO2- and HfO2-based MIM structures with TiN bottom, Pt top electrodes exhibit stable RS with HRS/LRS ratio up to 100. It was shown that for TiO2 films Al2O3 barrier with the thickness 3-5 nm is necessary to obtain stable bipolar RS. HfO2 films exhibit lower leakage current density and the thickness of the dielectric layer can be decreased to several nm. The RS effect in MIM structures with HfO2 films can be scaled-down to thicknesses below 3 nm. MIM structures with ALD grown dielectrics are promising as future memory cells.

Acknowledgments This work was supported by the VEGA project 2/0147/11, project APVV-0509-10, Centre of Competence project ITMS 26240220073 (0.5). IHP authors gratefully acknowledge the financial support from the Deutsche Forschungsgemeinschaft (DFG) for the RRAM project under contract SCHR 1123/7-1.

References 1. 2. 3. 4. 5.

S. Yu, X. Guan, H.-S. Philip Wong, Appl. Phys. Lett., 99, 063507 (2011). F. Nardi, S. Laurentis, S. Balatti, D. C. Gilmer, D. Ielmini, IEEE Trans. Electron Dev., 59, 2461 (2012). J. J. Yang, I. H. Inoue, T. Mikolajick, C. S. Hwang, MRS Bulletin, 37, 131 (2012). H.-S. Philip Wong, H-Y. Lee, S. Yu, Y-S. Chen, Y. W. P-S. Chen. B. Lee. F. T. Chen, M-J. Tsai, Proc. IEEE, 100, 1951 (2012). D. S. Jeong, R. Thomas, R. S. Katiyar, J. F. Scott, H. Kohlstedt, A. Petraru, C. S. Hwang, Rep. Prog. Phys., 75, 076502 (2012).

169


6.

7. 8. 9. 10. 11.

12.

13. 14.

C. Yoshida, K. Tsunoda, H. Noshiro, Y. Sugiyama, Appl. Phys. Lett., 91, 2235510 (2007). J.J. Yang, M.D. Pickett, X. Li, D.A.A. Ohlberg, D.R. Stewart, R.S. Williams, Nature Nanotechnology, 3, 429 (2008). Y.C. Bae, A.R. Lee, J.S. Kwak, H. Im, Y.H.Do, J.P. Hong, Appl. Phys. A, 102,1009 (2011). K.M. Kim, B.J. Choi, M.H. Lee, G.H. Kim, S.J. Song, J.Y. Seok, J.H. Yoon, S. Han, C.S. Hwang, Nanotechnology 22, 254010 (2011). F. Zhang, X.M. Li, X.D. Gao, L. Wu, X. Cao, X.J. Liu, R. Yang, J. Appl. Phys. 109, 104504 (2011). L. Goux, P. Czarnecki, Y. Y. Chen, L. Pantisano, X. P. Wang, R. Degraeve, B. Govoreanu, M. Jurczak, D. J. Wouters, L. Altimime, Appl. Phys. Lett., 97, 243509 (2010). C. Walczyk, D. Walczyk, T. Schroeder, T. Bertaud, M. Sowinska, M. Lukosius, M. Fraschke, D. Wolansky, B. Tillack, E. Miranda, C. Wenger, IEEE Trans. Electron Devices, 58, 3124 (2011). K-L. Lin, T-H. Hou, J. Shieh, J-H. Lin, C-T. Chou, Y-J. Lee, J. Appl. Phys. 109 084104 (2011). Y. S. Lin, F. Zheng, S. G. Tang, H. Y. Liu, C. Chen, S. Gao, Y. G. Wang, F. Pan, J. Appl. Phys., 113, 064510 (2013).

170