Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites

Copyright © 2007 American Scientific Publishers All rights reserved Printed in the United States of America

Journal of Nanoscience and Nanotechnology Vol. 7, 4534–4539, 2007

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites Basudev Pradhan, Swarup K. Majee, Sudip K. Batabyal, and Amlan J. Pal∗ Indian Association for the Cultivation of Science, Department of Solid State Physics and Centre for Advanced Materials, Kolkata 700032, India The article reports observed electrical bistability in thin films of ZnO nanoparticles embedded in an insulating polymer matrix. From the current–voltage and impedance characteristics, they studied transport mechanisms involved in the two conducting-states. The electrical bistability in such films has been associated with a memory phenomenon. The bistability, which is reversible in nature, led to read-only and random-access memory applications in the to: devices based on such nanoparticles. Delivered by Ingenta

Indian Assoc for theMemory Cultiv Phenomenon, Sc. Keywords: ZnO Nanoparticles, Electrical Bistability, Dielectric Properties, Conduction Mechanism.IP

: 202.54.54.240 Wed, 12 Nov 2008 18:30:41

RESEARCH ARTICLE

1. INTRODUCTION To achieve high-density memory elements, molecular and nanomaterials are being characterized in recent years. As molecular materials, different organic molecules1–4 and composites5
6 have demonstrated some remarkable improvement in organic memory devices. In the class of nanomaterials, metals,7
8 and semiconducting nanoparticles in polymer matrices are being used.9–13 The memory phenomenon in such materials arises due to electrical bistability, which results in (at least) two reproducible and stable current–voltage characteristics. A suitable voltage pulse generally triggers the bistability or electrical switching. In effect, device current at a voltage remains dependent on the preceding bias sequence resulting in memory phenomenon. Semiconducting nanoparticles such as CdSe, ZnO, etc., exhibited electrical bistability and associated memory phenomenon.10–13 Electrical bistability in such systems has been explained in terms of charge trapping or carrier confinement in the nanoparticles. In the light of two conducting states at a bias, it is worthwhile to study the conduction mechanisms involved in the high- and low-conducting states. Dielectric spectroscopy can be a useful tool in this direction. In view of the fact that a detailed understanding of the mechanisms will guide the research for improved switching and memory applications, we chose ZnO nanoparticles in an inert polymer matrix. ZnO has some unique properties, such as, high exciton binding ∗

Author to whom correspondence should be addressed.

4534

J. Nanosci. Nanotechnol. 2007, Vol. 7, No. 12

energy (60 meV), wide band-gap (3.37 eV) at room temperature, high breakdown strength, cohesion, and exciton stability. Moreover, ZnO is one of the hardest materials in the family of II–VI semiconductors.14
15 In the present work, from the current–voltage characteristics of the device and dielectric spectroscopy of the active material in the low- and the high-conducting states, we aim to study the detailed conduction mechanism in the two-states.

2. EXPERIMENTAL DETAILS ZnO nanoparticles, capped with poly(methylmethacrylate) (PMMA, MW = 6500) were grown from zinc acetate.16 Solutions of zinc acetate (3 mM, 25 ml) and PMMA (0.5 mM, 50 ml) were mixed together. The temperature was set to 60  C. During vigorous stirring of the mixed solution, aqueous solution of NaOH (3.7 mM) was added in drops. After 30 minutes, a bluish-white solution formed which was cooled down to room temperature and centrifuged for several occasions. Here the carboxylic ion groups interact with positively charged zinc atoms to yield chemisorbed polymer on the surface of ZnO nanoparticles. The PMMA restricts the growth of ZnO nanoparticles and regulates the size of them. The product was washed in chloroform and ethanol to separate out powder of ZnO nanoparticles. The powder was dried in vacuum for 10 h. The nanoparticles were characterized by high-resolution transmission electron microscope (HRTEM, JEOL-JSM 2010 operated at 200 kV) and field emission scanning electron microscope (FE-SEM, JEOL JSM-6700F). Absorption and photoluminescence (PL) spectra of the particles 1533-4880/2007/7/4534/006

doi:10.1166/jnn.2007.896

Pradhan et al.

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites

J. Nanosci. Nanotechnol. 7, 4534–4539, 2007

Photoluminescence intensity (a.u.)

HRTEM images of the nanoparticles showed lattice spacing corresponding to ZnO crystals confirming formation of ZnO nanoparticles. FE-SEM image of a spun-cast film shows homogeneous dispersion of the nanoparticles in the insulating polymer matrix (Fig. 1). The size of the ZnO nanoparticles, as identified from the HRTEM and SEM images, was around 6 nm. Electronic absorption and PL spectra of the nanoparticles (in dispersed solution) are shown in Figure 2.

Absorbance (a.u.)

3. RESULTS AND DISCUSSION

0.4

0.2

0.0 300

400

500

Wavelength (nm) Fig. 2. Electronic absorption and photoluminescence spectra of ZnO nanoparticles in dispersed solution.

4535

RESEARCH ARTICLE

dispersed in water were recorded with Hewlett Packard UV-Vis Spectrophotometer (HP-8453) and PerkinElmer Fluorescence Spectrometer (LS55), respectively. We characterized the nanoparticles by X-ray diffraction (XRD) analysis, which shows patterns corresponding to the standard spectrum of ZnO crystals. Devices were fabricated on indium tin oxide (ITO) coated glass substrates, which were cleaned and processed following standard protocol. In brief, stripped ITO substrates were repeatedly sonicated in soap:water mixtures. After thorough wash and sonication in deionized water, the slides were treated with methanol and acetone in sequence. Finally, they were stored in water till used for film deposition. To deposit films of ZnO nanoparticles in PMMA matrix, the ZnO nanoparticles were dispersed in chloroform solution of PMMA. By keeping the concentration of the solution that was spun fixed at 2 mg/ml, concenFig. 1. SEM image of PMMA:ZnO (50%) thin films. tration of ZnO nanoparticles in the PMMA solution was varied up to 60 wt%. The mixed solutions were spun at Delivered by Ingenta to: a speed of 1000 rpm for 30 s on the ITO substrates. For The particles show absorption edge at 370 nm, which corIndian Assoc for the Cultiv Sc. comparison, a solution containing only PMMA was also responds to a band gap of 3.35 eV. The PL spectrum, with IP : 202.54.54.240 spun on a separate substrate. The thin films, which had an excitation wavelength of 290 nm, shows a characteristic Wed, 12 Nov 2008 18:30:41 a thickness of around 70 nm, were annealed at 60  C in emission at 360 nm along with a broad band in the visible −3 vacuum (10 Torr) for 2 h. Aluminum (Al) was vacuumregion. The PL in the UV region appears due to emission evaporated on top of the annealed films from a tungsten from the absorption edge. The observed emission in the filament basket at a pressure below 10−5 Torr. Active area visible domain (∼550 nm) has been reported by several of the devices, containing PMMA:ZnO films with different workers.17
18 ZnO concentration and only PMMA films, was 6 mm2 . Current–voltage I–V  characteristics of the devices The devices were characterized in a shielded vacuum were recorded by scanning applied voltage from 0 to chamber with a Yokogawa 7651 dc source and a Keithley +VMax and then to −VMax followed by a scan from −VMax 486 picoammeter. Bias was applied in loops with respect to +VMax . We have varied the amplitude of VMax to gento the Al electrode. It was swept in steps of 0.05 V erate a range of I–V characteristics. The characteristics with a sweep-speed of 2.5 mV/s. Amplitude of maximum were reproducible during repeated voltage cycles. A typvoltage VMax  was also varied. The dc source, coupled ical I–V curve between ±20 V in devices based on with fast switching-transistors, was used to generate voltPMMA:ZnO (30%) is presented in Figure 3. The characage pulse of different widths and amplitudes, which were teristics in the reverse bias show electrical bistability—that used to “write,” “read,” or “erase” a state of the devices. is, the amplitude of device current at any voltage during Impedance spectroscopy of the device was carried out the sweep from −VMax was several orders higher in magniwith a Solartron 1260 Impedance Analyzer in the 1 Hz to tude as compared to that during the sweep towards −VMax . 12 MHz frequency range. A 100 mV rms signal was used as the test voltage. Measurements were carried out with0.6 out any dc bias in a parallel mode configuration (CP − RP network). The instruments were controlled with a personal computer via a general-purpose interface bus (GPIB).

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites

Pradhan et al. –3 (a) 10

0.0

Current (A)

Current (mA)

10 –5

– 0.1

10 –7 PMMA only ZnO:PMMA (25%) ZnO:PMMA (50%)

10 –9

10 –11

– 0.2 –2

–1

0

1

2 10 –13

Voltage (V)

–2

Fig. 3. Current–voltage characteristics of a device based on thin films of PMMA:ZnO (30%) in a voltage loop. Arrows show the direction of voltage sweep.

–1

0

1

2

Voltage (V) 3 (b) 10

RESEARCH ARTICLE

On/Off ratio

The increase in conductivity at a suitable negative voltage, or conductance switching, is associated with a memory Delivered phenomenon. In other words, no bias is required to sustain by Ingenta to: Indian Sc. the higher conducting state during the scan from Assoc −VMax . for the Cultiv 102 IP : 202.54.54.240 Only a matching positive voltage +VMax  could erase the Wed, 12 Nov 2008 18:30:41 high-conducting state to a low one. To study the role of ZnO nanoparticles in polymer/ nanoparticle blends on the electrical bistability, we have compared the I–V characteristics of devices having different concentrations of ZnO nanoparticles. Figure 4(a) 101 0 10 20 30 40 50 60 shows that in the voltage range of our measurement, the ZnO concentration (%) devices based on only PMMA do not show any change in conductivity during the two sweep directions. The devices Fig. 4. (a) Current–voltage characteristics of three devices in two sweep with ZnO, on the other hand, exhibit electrical bistability directions. Amplitude of device current has been plotted. Arrows reprewith a reproducible transition between a high- and a lowsent the direction of voltage sweep. In (b), On/Off ratio at −10 V is shown as a function of ZnO nanoparticle concentration. The broken line conducting state. Typical cases with 25% and 50% ZnO, is to guide the eye. referred to as PMMA:ZnO (25%), PMMA:ZnO (50%), respectively, are shown in the figure. With increase in for electron injection to conduction band of ZnO nanoparZnO concentration in the PMMA matrix, amplitude of ticles is lower from the Al electrode as compared to that current for both the conducting-states increases. The ratio from ITO electrode. (Work-function of ITO19 and Al are between the two currents at the same voltage, the On/Off 4.7 and 4.3 eV, respectively). Since ZnO is usually an ratio, however depends on ZnO concentration. A plot of electron conductor, amplitude of voltage (to switch) is the ratio as a function of ZnO concentration is shown in determined by the difference in fermi energy of the metal Figure 4(b). The On/Off ratio is highest at an optimum and the conduction band of the semiconductor. The thresZnO concentration. The ratio, which also depends on the hold voltage required for the transition to the high convoltage at which it is measured, reaches more than 700 ducting state (1.4–1.6 V) is very close to the difference (at −10 V) for a device with 30% ZnO nanoparticles. As between the work function of ITO (4.7 eV) and band edge compared to organic systems,1–4 the On/Off ratio is low. of the conduction band of ZnO nanoparticles (3.37 eV). This could be due to high off-state current in ZnO (as We have further attempted to fit the high- and low-state compared to organic materials, which has intrinsically low currents of the devices based on PMMA:ZnO composites conductivity). with the existing conduction models (Fig. 5). A plot of Electrical bistability and memory phenomenon in ln(I) versus V 1/2 shows linear fit. For both the conductPMMA:ZnO matrix can be explained in terms of charge ing states, i.e., for voltage sweeps from 0 to −20 V and confinement in the nanomaterials. Under a suitable negfrom −20 to 0 V representing low- and high-conducting ative voltage, the ZnO nanoparticles with higher surface states, respectively, the plots fit to straight lines suggesting charge density form percolative networks. Such networks that the current is limited by injection from the electrode finally give rise to channels or filaments across the device with conduction mechanism following thermionic emisfor conduction resulting in a high-conducting state. The high-state is reached at a negative voltage; energy required sion model.20 According to this model, current density (J ) 4536

J. Nanosci. Nanotechnol. 7, 4534–4539, 2007

Pradhan et al.

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites (a) –2.0

10 –3 10 –4

–1.5

–5

Z′′(MΩ)

Current (A)

10

High-state (– 2.0 to 0 V) Low-state (0 to –2.0 V)

10 –6 10 –7

–1.0

High-state of PMMA:ZnO (50%) Low-state of PMMA:ZnO (50%) High-state of PMMA:ZnO (25%) Low-state of PMMA:ZnO (25%)

–0.5

10 –8 0.0

10 –9 0.0

0.0 0.5

1.0

0.5

1.5

1.5

2.0

2.5

104

105

106

Z′(MΩ)

Voltage1/2 (V1/2) (b)

Dielectric constant, ε′

Fig. 5. Current versus square root of voltage amplitude of a device based on PMMA:ZnO (30%) film in low- and high-conducting states. Results from the negative bias section of I–V characteristics are plotted here.

1.0

18

15

Delivered by Ingenta to: Indian Assoc for the Cultiv Sc. 12 at a field (F ) is given by: IP : 202.54.54.240 12 Nov 2008 18:30:41 J = AT 2 exp− − F 1/2 /kTWed,

J. Nanosci. Nanotechnol. 7, 4534–4539, 2007

9

6 101

102

103

Frequency (Hz) Fig. 6. (a) Cole–Cole plots from 1 Hz to 1 MHz (eight points/decade) for devices based on PMMA:ZnO (25%) and PMMA:ZnO (50%) films at 0 V dc bias after switching to a high-conducting and a low conducting state. (b) Frequency response of real part of dielectric constant as function at zero dc bias for the two states of the devices.

The diameter of the semicircular arc, which represents the bulk resistance of the device, decreases with increase in concentration of ZnO nanoparticles in the device. The diameter also decreases with a transition or switching of the device to a high-conducting state. This shows that the bulk resistance of the devices decreases during switching, presumably due to charge confinement in the nanoparticles. This is in support of the results obtained from the I–V characteristics of the two conducting-states. From the Cole–Cole plots, we have calculated dielectric constant of the active material of the devices as a function of frequency. Figure 6(b) shows such a plot for the two states of the devices. At higher frequencies, the dielectric constant decreases due to phase lag between the electric field and molecular polarization. For all the devices, the dielectric constant of the active material did not change with conductance switching. This supports the results obtained from the fitting of I–V characteristics (Fig. 5). The results therefore show that the change in resistive properties, as dictated by the band gap of the material and correspondingly barrier height at the interfaces (induced by carrier confinement in the nanoparticles), controls 4537

RESEARCH ARTICLE

where, A is a constant, T the temperature,  the energy barrier with the electrode, and  relates the relative dielectric constant (r ) of the material as  = e3 /40 r 1/2 . Here 0 is permittivity of free space and e is electronic charge. The slopes of the plots in Figure 5 do not change for the two states suggesting that the dielectric constant of the material does not change during the transition. Here we did not attempt to fit the off-state current in the higher bias region, where a transition from the low-state to the highone occurred. The intercept with the abscissa however differed in the two-states. For the high-conducting state, the extrapolated intercept at 0 V was higher as compared to that for the low-state. This suggests a lower barrier () with the electrode for electron injection in the high-state. This would require the conduction band to decrease, which might occur due to charge confinement in the nanoparticles. The analysis hence supports the observed switching to a high-state (relating barrier heights) in the negative bias direction. We have employed impedance spectroscopy to study dielectric properties of the devices. We have measured real and imaginary components of complex impedance (Z and Z

, respectively) as a function of frequency. The components were measured before and after inducing the high- and the low-conducting state. Since no dc bias was applied during the measurements, the impedance spectrum in effect probed the two states. Cole–Cole plots (Z versus Z

) for two typical devices in their high- and lowconducting states are presented in Figure 6(a). Each of the Cole–Cole plots fits to a semicircle or shows a trend of one. The devices can hence be modeled to a parallel combination of a resistor RP  and a capacitor CP .

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites

Pradhan et al.

4. CONCLUSIONS

10–4 High-state of ZnO:PMMA (50%) Low-state of ZnO:PMMA (50%) High-state of ZnO:PMMA (25%) Low-state of ZnO:PMMA (25%)

10–5

Current (A)

RESEARCH ARTICLE

Current (µA)

Voltage (V)

electrical bistability in these materials. At any frequency, 2 the value of dielectric constant of the polymer:ZnO com1 posite is somewhat higher than that of the respective com0 ponents (dielectric constant of PMMA and ZnO are about –1 3.9 and 8.5, respectively). This could be due to the stabiliz–2 ers used during synthesis of nanoparticles. The dielectric constant of the active layer also increases with increase 0 in concentration of ZnO nanoparticles. There have been some reports available in the literature showing increase in –10 dielectric constant in polymer matrices.21
22 Further work is required by varying temperature of the device to explain the change in dielectric constant in such matrices. –20 The retention time of the states, specially required for 0 500 1000 1500 Time (s) memory applications, was measured under a probe voltage (Fig. 7). Here, we probe the high- and the low-conducting Fig. 8. Switching performance of a device based on PMMA:ZnO (50%) states that have been induced by −20 and +20 V pulse during a continuous “write-read-erase-read” sequence. The voltage pulse (width = 15 s), respectively. The results show that current is shown in the upper trace, while the corresponding current response is shown in the lower one. The high-conducting state has been induced under the same probe voltage depends on the direction (“write”) by −2 V pulse, while the low-one has been reinstated (“erase”) of preceding pump voltage pulse. When a high-state was Delivered by Ingenta to: by +2 V. In between switching, the states were probed (“read”) by meainduced by a negative voltage (−20 V)Indian pulse, the curAssoc for the Cultiv Sc. −10 V pulse. suring current under rent under probe voltage was much higher as compared IP : 202.54.54.240 sequence for up to 2 hours. A section of the voltage to that after inducing a low-conducting state. For 12 devices Wed, Nov 2008 18:30:41 sequence and corresponding current from a device with with higher concentration of ZnO particles, probe current PMMA:ZnO (50%) is shown in Figure 8. Here also, for both the states were expectedly higher. In other words, −20 and +20 V pulse (width = 15 s) were applied to the high- and low-states of both the devices are distin“write” the high-state and “erase” to a low-conducting guishable for several hours for read-only memory (ROM) state, respectively. The states were probed by measuring applications. device current under −10 V. The figure shows that the Reversibility of electrical bistability in the device magnitude of current under “read” voltage is much higher can best be studied under a voltage pulse sequence, during probing the high-conducting state as compared to namely “write-read-erase-read” cycle. In such a cycle, the that while addressing the low-one. The results hence show high- and low-conducting states are induced (“write” and that one can flip-flop the two states of the devices and “erase,” respectively) in cycles and the states are moniprobe them successfully for rewritable or random-access tored or “read” in between the “write” and “erase” pulses. memory (RAM) applications. We have characterized the devices under such voltage

10–6

10–7

0

2000

4000

6000

Time (s) Fig. 7. Long time response of the high- and the low-conducting state of the devices based on PMMA:ZnO (25%), PMMA:ZnO (50%) films under −10 V pulse (width = 10 s, duty cycle = 50%). The high- and the low-states have been induced by −20 V and 2.0 V, respectively. The absolute value of current has been plotted in the figure.

4538

In conclusion, we have observed electrical bistability in thin films of ZnO nanoparticles embedded in an inert polymer matrix. Concentration of the nanoparticles has been varied to obtain a wide range of films that have exhibited memory-switching. The transition to the high-conducting state has been explained in terms of charge confinement in the nanoparticles. The conduction mechanisms in the two conducting states have been modeled by an injectionlimited process. From the current–voltage and impedance characteristics, we have attributed the bistability to be associated with a change in bulk resistance. Dielectric constant of the active material however has not shown a change during switching. The devices have exhibited random-access memory applications under “write-readerase-read” voltage sequence and also read-only memory for several hours. Acknowledgments: The authors acknowledge financial support from the Department of Science and Technology, J. Nanosci. Nanotechnol. 7, 4534–4539, 2007

Pradhan et al.

Electrical Bistability in Zinc Oxide Nanoparticle-Polymer Composites

Government of India through projects SP/S2/M-44/99 and Ramanna Fellowship (SR/S2/RFCMP-02/2005).

10. M. D. Fischbein and M. Drndic, Appl. Phys. Lett. 86, 193106 (2005). 11. K. Mohanta, S. K. Majee, S. K. Batabyal, and A. J. Pal, J. Phys. Chem. B 110, 18231 (2006). 12. E. K. Kim, J.-H. Kim, D. U. Lee, G. H. Kim, and Y.-H. Kim, Jpn. References and Notes J. Appl. Phys. 45, 7209 (2006). 1. R. McCreery, J. Dieringer, A. O. Solak, B. Snyder, A. M. Nowak, 13. F. Verbakel, S. C. J. Meskers, and R. A. J. Janssen, Appl. Phys. Lett. W. R. McGovern, and S. DuVall, J. Am. Chem. Soc. 125, 10748 89, 102103 (2006). (2003). 14. Ü. Özgür, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov, S. Do˘gan, 2. Z. J. Donhauser, B. A. Mantooth, K. F. Kelly, L. A. Bumm, J. D. V. Avrutin, S. J. Cho, and H. Morkoç, J. Appl. Phys. 98, 041301 Monnell, J. J. Stapleton, D. W. Price Jr., A. M. Rawlett, D. L. Allara, (2005). J. M. Tour, and P. S. Weiss, Science 292, 2303 (2001). 15. S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo, and T. Steiner, J. Vac. 3. D. Ma, M. Aguiar, J. A. Freire, and I. A. Hümmelgen, Adv. Mater. Sci. Tech. 22, 932 (2004). 12, 1063 (2000). 16. X.-W. Du, Y.-S. Fu, J. Sun, X. Han, and J. Liu, Semicond. Sci. 4. A. Bandhopadhyay and A. J. Pal, J. Phys. Chem. B 107, 2531 (2003). Technol. 21, 1202 (2006). 5. S. Paul, A. Kanwal, and M. Chhowalla, Nanotechnology 17, 145 17. J. H. He, C. S. Lao, L. H. Chen, D. Davidovic, and Z. L. Wang, (2006). J. Am. Chem. Soc. 127, 16376 (2005). 6. B. Mukherjee, S. K. Batabyal, and A. J. Pal, Adv. Mater. 19, 717 18. Y. H. Tong, Y. C. Liu, S. X. Lu, L. Dong, S. J. Chen, and Z. Y. (2007). Xiao, J. Sol-Gel Sci. Technol. 30, 157 (2004). 7. S. Paul, C. Pearson, A. Molloy, M. A. Cousins, M. Green, 19. Various surface treatments directly affect the work function of ITO. S. Kolliopoulou, P. Dimitrakis, P. Normand, D. Tsoukalas, and M. C. 20. H. Vestweber, J. Pommerehne, R. Sander, R. F. Mahrt, A. Greiner, Petty, Nano Lett. 3, 533 (2003). W. Heitz, and H. Bässler, Synth. Met. 68, 263 (1995). 8. J. Y. Ouyang, C. W. Chu, C. R. Szmanda, L. P. Ma,Delivered and Y. Yang, by Ingenta 21. P. Murugaraj, to: D. Mainwaring, and N. Mora-Huertas, J. Appl. Phys. Nature Mater. 3, 918 (2004). 054304Sc. (2005). Indian Assoc for the 98, Cultiv 9. B. Pradhan, S. K. Batabyal, and A. J. Pal, J. Phys. Chem. B 110, 22. J. X. Lu, K. S. Moon, J. W. Xu, and C. P. Wong, J. Mater. Chem. IP : 202.54.54.240 8274 (2006). 16, 1543 (2006).

Wed, 12 Nov 2008 18:30:41

Received: 13 March 2007. Revised/Accepted: 12 April 2007.

RESEARCH ARTICLE

J. Nanosci. Nanotechnol. 7, 4534–4539, 2007

4539