J. Phys. Chem. B 1999, 103, 7831-7838
7831
Electron Transport in Nanoparticulate ZnO Films Eric A. Meulenkamp* Philips Research Laboratories EindhoVen (WA13), Prof. Holstlaan 4, 5656 AA EindhoVen, The Netherlands ReceiVed: May 5, 1999; In Final Form: July 13, 1999
Electron transport in a nanocrystalline thin film was studied using an electrode geometry in which the resistance was measured between two contacts which are separated by a narrow gap bridged by the material of interest. This setup, which allows the direct study of electron transport in a wet nanocrystalline thin film, was used to investigate in situ the electron mobility in ZnO films in contact with liquid electrolyte solutions under accumulation conditions. The setup is also suitable for widely different experimental conditions and materials. Nanocrystalline ZnO films with 5 nm particle size were prepared from a washed and concentrated ethanolic dispersion. Characterization by optical transmission and scanning electron microscopy revealed a porous homogeneous structure with the same particle size as in the original sol. The mobility ranged from 10-3 to 10-1 cm2/Vs for an electron density between 5 × 1018 and 1 × 1020 cm-3. The mobility depended on the electron density and the solvent (aqueous or nonaqueous), but was not affected by the type or the concentration of the electrolyte ions, or the dielectric constant of the solvent. Interparticle electron transfer is very likely the slow step in electron transport. The importance of the surface chemistry of the particles and the film porosity and heterogeneity is outlined.
1. Introduction Nanocrystalline films have attracted considerable attention recently because of their possible use in solar cells,1 lightemitting diodes,2 electrochromic windows, and batteries.3,4 Transparent conducting films from nanosized powders and sols (nanoparticulate films) are also the focus of interest.5-7 Electron transport in such nanoparticulate films is obviously an important issue. Transport is generally found to be much slower than in (compact) polycrystalline or monocrystalline materials. This has been discussed in terms of poor interparticle electron transfer. For wet porous devices the influence of the penetrating electrolyte has been considered.8-11 The conductivity of dry (doped) thin films has been studied by sheet resistance and Hall mobility, but these methods have not found application on wet, poorly conducting films. Electron transport in such systems has been investigated under photoanodic conditions. Data analysis, on the basis of several assumptions, led to values for the electron diffusion coefficient. Only a limited number of studies are available and several questions remain,8-13 regarding, for instance, the role of the electrolyte, the magnitude of electron concentration gradients, and the electron transfer to the back contact. In this paper, a versatile in situ technique is used that can yield the conductivity of wet films and its dependence on the charge-carrier concentration. The experiment also allows easy comparison with dry films. The method was originally developed by Wrighton and co-workers,14 and has been used to study oxide and polymer materials.15-18 A transistor-like electrode geometry is used, which is depicted schematically in Figure 1. Two inert electrodes, similar to source and drain in a fieldeffect transistor (FET19) are electrically connected by a “bridge” of the material studied. The setup is placed in an electrolyte solution with a reference and counter electrode. Source and drain are connected to the working electrode inputs of a bipotentiostat. * E-mail:
[email protected].
Figure 1. Schematic picture of the electrode geometry used for the determination of the resistance. The source-drain current iSD is measured as a function of the electron concentration in the ZnO layer. When no charge is built up, the bridge material is insulating and iSD ≈ 0. The situation shown represents the conductive state, with electrons in the ZnO particles, and charge-balancing cations in the surrounding electrolyte.
This allows control of their electrochemical potential and, hence, of the electron concentration and resistance of the bridge. The resistance is obtained from measurement of the source-drain current iSD for a small source-drain potential difference. Porous nanocrystalline ZnO was chosen for the investigations. We have recently reported on the synthesis and dissolution of ZnO sols20,21 and obtained a method to deposit transparent, homogeneous films. The electron transport is studied under accumulation. The optical properties of ZnO films under these conditions show remarkable variations,22-25 not all of which are understood. Electrical data can contribute to a better understanding of such films. A characterization of the same films under photoanodic conditions by intensity-modulated photocurrent spectroscopy (IMPS) is reported elsewhere.26 In this
10.1021/jp9914673 CCC: $18.00 © 1999 American Chemical Society Published on Web 08/27/1999
7832 J. Phys. Chem. B, Vol. 103, No. 37, 1999
Meulenkamp
paper we describe conditions for the successful use of the electrode geometry in Figure 1. Then, the morphology of the ZnO films and the characteristics of electron accumulation in these films are outlined. Results are presented for the potentialdependent electron mobility in aqueous and nonaqueous electrolytes. 2. Experimental Section Preparation of ZnO Films. The synthesis of the ZnO sol was described in detail elsewhere.20 A volume of 50 mL of 0.14 M LiOH‚H2O (Aldrich) in ethanol (Merck, Selectipur) was added to 50 mL of 0.10 M Zn(Ac)2‚2H2O (Aldrich) in ethanol at 0 °C under vigorous stirring. After a few hours crystalline wurtzite ZnO particles are formed, which continue to grow at a rate depending on the solution composition and temperature. When the desired particle size was reached, typically 5 nm, the sol was thoroughly washed by repeated ZnO precipitation by heptane (Merck, p.a.), removal of the supernatant liquid, and redispersion in ethanol.20 Washing removed Li+ and Ac- ions and water. The resulting sol had approximately a 4× higher ZnO content and a 50-100× lower conductivity than the original sol. Films (0.05-0.1 µm thickness) were deposited by spin-coating on ITO/glass or Au/glass substrates, followed by a 15 min, 150 °C anneal in air. Thicker films were obtained by repeating this procedure. Electrochemical Experiments. Conventional electrochemical cells were used. The equipment consisted of an Autolab PG20 potentiostat or a Schlumberger SI1286 Electrochemical Interface for three-electrode experiments, an EG&G Model 366A bipotentiostat for resistance measurements, and a Schlumberger SI1255 HF Frequency Response Analyzer for impedance measurements. All chemicals were from Merck (Selectipur or p.a.) unless otherwise indicated. A 0.1 M phosphate buffer (pH ) 8) was used as an aqueous electrolyte solution. The pH was adjusted with H3PO4/NaOH. Pt sheet and Ag/AgCl served as counter and reference electrodes, respectively. Solutions were purged with nitrogen. All experiments involving nonaqueous electrolytes were carried out in an argon-filled glovebox ( 400 nm (the sub-bandgap region). A λ-4 dependence, characteristic of light scattering,27 is not seen in this wavelength range. The transmission at about 400 nm and for λ > 650 nm is larger than 100%. This and the complex undulations are due to interference, and point to a uniform thickness over larger areas (spot size ≈ 5 mm2) and a refractive index smaller than that of the underlying ITO, in agreement with the porous nature of the film. The steep decrease at λ < 370 nm is due to ZnO band gap absorption. Size quantization plays a role as the onset of the absorption is shifted by about 10 nm as compared to bulk ZnO. The optical band gap of the film and the washed sol were identical within experimental error. This indicates that extensive particle growth or neck formation do not take place. This is due to the low anneal temperature (150 °C) and was also reported by Hoyer et al.22,23 There is only limited electronic interaction between particles as this also would lower the optical band gap.28 Chemical analysis showed that Li and Ac were not present above their detection limits (0.5 and 0.2 at. % relative to Zn,
J. Phys. Chem. B, Vol. 103, No. 37, 1999 7833
Figure 4. Optical transmission of a ZnO/ITO/glass sample, obtained with an ITO/glass reference.
respectively). The Zn content was used to calculate the relative density. The average value was 55%. Variation was found between ZnO batches (35-75%). The absorption coefficient R at ≈350 nm, i.e., at the excitonic peak or shoulder, was about 6 × 104 cm-1. For the sol, a value of 8 × 104 cm-1 was found. Although the former is slightly higher (taking into account the porosity), this indicates again that little interaction occurs between particles. The film preparation and properties are slightly different from much of the work that has been published,24,25,29-31 where a high-temperature anneal (400 °C) was used to convert the asdeposited film containing impurities to a porous ZnO film. This always resulted in a particle size increase and a coarser structure and sometimes substantial light scattering could not be avoided. To summarize, the porous ZnO films studied here are chemically pure and consist of wurtzite ZnO with a volumeaveraged particle size of about 5.0 nm.20 The films are very homogeneous down to a length scale of 50 nm. They retain the optical properties of the starting ZnO sol, indicating that little electronic interaction occurs between particles. It should be emphasized that these films could only be obtained after thorough washing of the sol and low-temperature annealing. Otherwise, scattering, somewhat opaque films with low ZnO content were deposited. 3.3. Electron Accumulation. The characteristics of electron accumulation were studied with ordinary electrodes without gap, and with gap electrodes with source and drain short-circuited. Figure 5 shows the current-potential (i-E) curve in 1.0 M LiClO4/PC for three electrodes: ZnO/ITO, ZnO/Au, and bare ITO. At E > 2.7 V the curves are virtually identical: the electron concentration in ZnO is low if the Fermi level is considerably below the conduction band edge. The film behaves as a dielectric and the electrochemical reactions and electrochemical impedance are typical of the underlying ITO or Au. We define a characteristic potential Eonset below which substantial electron flow into ZnO becomes possible. The values found here are in good agreement with earlier work.22-25 For E < Eonset a large current is observed, which changes sign as the scan direction is reversed. At a sufficiently low scan rate the current is almost entirely capacitive and is proportional to the scan rate and the ZnO film thickness. Since the current is due to (dis-)charging of the ZnO nanoparticles, the electron density in ZnO can be calculated. The counter charge resides in the electrolyte.
7834 J. Phys. Chem. B, Vol. 103, No. 37, 1999
Figure 5. Current-potential (i-E) curves of a bare ITO electrode (“reference”), and a ZnO/ITO and a ZnO/Au electrode. The scan direction and the position of Eonset are indicated by arrows. The scan rate was 20 mV s-1.
Figure 6. Electron accumulation in ZnO. The left y-axis is the cumulative charge, which was recalculated into electrons/particle (right axis) using the particle size (5.0 nm), and the ZnO content of the film. The potential axis is defined with respect to Eonset. Data shown were obtained in H2O (closed symbols) and PC (open symbols). Squares indicate ZnO and circles ITO.
The extent of electron accumulation was determined by potential step measurements (current integration) and by measurement of the differential capacitance at a constant potential. Some scatter in the absolute values between experiments was observed. The potential dependence was typically that shown in Figure 6. These results were obtained using electrodes from a single batch. To compare results obtained in aqueous electrolyte (closed symbols) and nonaqueous electrolyte (open symbols), the potential axis is defined with respect to Eonset (-0.5 V in the aqueous phosphate electrolyte). Results for bare ITO are included. It is clear that the charge is larger by a factor of 2-2.5 in the aqueous system. It increases more than linearly with Eacc in the case of ZnO. The electron density in ZnO could not be calculated accurately for Eacc e 0.05-0.1 V. The contribution of ITO to the total measured charge is relatively large at small Eacc. It was difficult to determine what part of the underlying ITO substrate was in contact with the electrolyte. Hence, the correction introduced a noticeable error for an electron density smaller than about 5 ×
Meulenkamp 1018 cm-3. The right-hand axis in Figure 6 shows the calculated electron density in ZnO. The highest value shown of about 10 electrons/particle is in good agreement with reports by Weller and co-workers.22,23 It corresponds to 1.5 × 1020 cm-3, typical for a degenerate semiconductor. The experimentally accessible range of the accumulation potential Eacc was limited by reduction of the ZnO (and ITO) film and by the increasing rate of sidereactions, as evidenced by irreversible coloration of the samples and the (i-E) curves of bare ITO, respectively. Thus, the electrical conductivity could be studied accurately for an electron density between about 5 × 1018 and 1.5 × 1020 cm-3. Finally, the absolute value of the charge is briefly discussed. The slope of the curves in Figure 6 is equal to the capacitance C. The interpretation is straightforward for ITO, where C represents the Helmholtz capacitance CH. It is more complicated for ZnO. A maximum uptake of three electrons in a 0.1 V range is clear from Figure 6. This corresponds to 4.8 × 10-18 F, or 6 µF cm-2, lower than may be expected for an aqueous zinc oxide/ solution interface (CH ≈ 50 µF cm-2).32 This difference is partly due to incomplete wetting of the ZnO nanoparticle: if, for example, the available surface area is only 50%, 12 µF cm-2 would have been found. A detailed comparison is precluded by the fact that the nanoparticles may also show different surface chemistry and, hence, CH than the ZnO used for the colloid chemical studies.32 The capacitance found for low Eacc, however, is so much smaller than the literature value, that an additional effect must be at work. This is assumed to be electrostatic electron-electron repulsion in the ZnO particles owing to the small particle size.33 Coulomb blockade,34 which is an example of the consequences of this phenomenon, leads to a lowering of the capacitance compared to values for a large ZnO particle, as observed here. 3.4. Conductivity Measurements. For each electrode, (i-E) curves were recorded and the cumulative charge was determined as a function of potential as outlined in section 3.3. Then, the fulfilment of the conditions discussed above, in particular that RCT is large, was checked. The resistance across the gap was determined, at a particular potential, for small potential differences ∆E between source and drain (e 10 mV). This ensured that the actual electron density in the ZnO bridge was close to that in the remainder of the film. An example is shown in Figure 7. Here, ileft and iright are the currents through the ITO electrodes (see Figure 2). They are equal to iFL + iZnO and iFR - iZnO, respectively. Analysis showed that ileft + iright was very close to zero in all experiments. In other words, the Faradaic currents were negligibly small. The slopes ∆E/∆i were identical (except for the sign). The nonzero current at ∆E ) 0 is due to a small error in the bipotentiostat output: although the applied potential was nominally the same, a 0.8 mV offset was found. For the remainder of the work presented, the slope of the ∆E-i curve was always determined. Figure 8 shows an example of a typical result for an electrode investigated in two electrolytes, propylene carbonate (PC) and H2O. The dependence of the conductance G on the amount of charge is shown. The conductance at zero charge was always 1019 cm-3. In this respect
Electron Transport in Nanoparticulate ZnO Films some older work on polycrystalline ZnO is also of importance. It was shown that oxygen chemisorption induced a modulation of the grain boundary barrier, which in turn led to a variation in both the mobility and the carrier concentration.37 Oxygen chemisorption and its effects on (surface) conductivity were also discussed by other workers.38 Hilgendorff et al. mention the possible role of surface traps introduced by water and oxygen in their dense nanocrystalline films.5 Finally, the spatial distribution of the current within a ZnO nanoparticle, or the time-averaged location of mobile electrons is briefly discussed. The possibility of surface conductivity is of particular interest. It is well-known for single-crystal and polycrystalline ZnO38 and is related to a spatial distribution of defects and to the presence of a space charge. The nanoparticles are too small to develop a significant space charge and a distinction between surface and bulk conductivity is rather arbitrary in such small particles. Therefore, the concept of surface conductivity as it is used in older work38 does not seem useful here. On the other hand, surface conductivity may be due to the presence of a mobile surface species. One can imagine, for example, that a surface-bonded proton and an electron are transported in a concerted manner. This implies that the electron is actually trapped at the surface. Although this type of surface conductivity cannot be ruled out entirely, it is not supported by some other observations. Such a mechanism is anticipated to result in a much larger effect of the solvent (notably protonic vs nonprotonic and the value of the dielectric constant) than found here. An effect of the type and concentration of the electrolyte ions could also be expected. Moreover, optical absorbance data in the UV25 and IR45 region and luminescence studies46 are consistent with the presence of (quasi-)free electrons and do not provide evidence for surface electron trapping. Hence, the present data can be described as bulk conductivity, as was tacitly assumed in the above discussions. The possible role of surface conductivity in nanoporous systems remains, however, an interesting topic for further investigations. It may be of particular importance when the metal ion is of variable valency, as is found in intercalation materials. Note, in this respect, that ZnO is not listed among these.47 5. Conclusions It was shown that the electron mobility in a nanoparticulate thin film can be determined in a direct manner in a transistorlike electrode geometry. The dimensions of the probe electrodes can be varied, which makes it possible to study materials under a wide range of experimental conditions. The in situ nature of the technique enables investigations on systems which exist only in solution. Electron accumulation in well-defined ZnO films was studied as an example. The mobility in these nanoparticulate ZnO films was measured accurately for an electron density between about 5 × 1018 and 1 × 1020 cm-3. It varied between 10-3 and 10-1 cm2/Vs. The maximum value is markedly higher than in TiO2 nanoporous films, and similar to the lower limit found for polycrystalline dense ZnO films. The mobility depended on the solvent and the electron concentration, with the higher values found in nonaqueous electrolyte at high electron density. The mobility did not depend on the type and concentration of the indifferent electrolyte ions, and the dielectric constant of the solvent. These effects were interpreted in terms of slow interparticle transfer. The surface chemistry of the particles, through its
J. Phys. Chem. B, Vol. 103, No. 37, 1999 7837 influence on the presence of traps and the nature of the energy barrier between particles, plays a decisive role. The film porosity and heterogeneity also have a strong impact on the electron transport as they determine what part of the ZnO particles contributes to the conductivity. Their role was clarified with concepts well-known in percolation theory. Acknowledgment. I thank J. H. T. Hengst, C. J. Geenen, F. G. Holthuysen, S. Aarts, and D. H. G. Haex (Philips Research Eindhoven) for chemical and structural analysis of the ZnO films. References and Notes (1) O’Regan, B.; Gra¨tzel, M. Nature 1991, 353, 737. (2) Colvin, V. L.; Schlamp, M. C.; Alivisatos, A. P. Nature 1994, 370, 354. (3) Hagfeldt, A.; Vlachopoulos, N.; Gra¨tzel, M. J. Electrochem. Soc. 1994, 141, L82. (4) Kang, T.-S.; Kim, D.; Kim, K.-J. J. Electrochem. Soc. 1998, 145, 1982. (5) Hilgendorff, M.; Spanhel, L.; Rothenha¨usler, Ch.; Mu¨ller, G. J. Electrochem. Soc. 1998, 145, 3632. (6) Behr, G.; Werner, J.; Oswald, S.; Krabbes, G.; Dordor, P.; Elefant, D.; Pitschke, W. Solid State Ionics 1997, 101, 1183. (7) Bommel, M. J. van; Groen, W. A.; Hal, H. A. M. van; Keur, W. C.; Bernards, T. N. M. J. Mater. Sci., submitted. (8) Solbrand, A.; Lindstro¨m, H.; Rensmo, H.; Hagfeldt, A.; Lindquist, S.-E. J. Phys. Chem. B 1997, 101, 2514. (9) Zaban, A.; Meier, A.; Gregg, B. A. J. Phys. Chem. 1997 101, 7985. (10) Dloczik, L.; Ileperuma, O.; Lauermann, I.; Peter, L. M.; Ponomarev, E. A.; Redmond, G.; Shaw, N. J.; Uhlendorf, I. J. Phys. Chem. B 1997, 101, 10281. (11) Wahl, A.; Augustynski, J. J. Phys. Chem. B 1998, 102, 7820. (12) Ko¨nenkamp, R.; Henninger, R.; Hoyer, P. J. Phys. Chem. 1993, 97, 7328. (13) de Jongh, P. E.; Vanmaekelbergh, D. Phys. ReV. Lett. 1996, 77, 3427; J. Phys. Chem. B 1997, 101, 2716. (14) Kittlesen, G. P.; White, H. S.; Wrighton, M. S. J. Am. Chem. Soc. 1984, 106, 7389. (15) Paul, E. W.; Ricco, A. J.; Wrighton, M. S. J. Phys. Chem. 1985, 89, 1441. (16) Natan, M. J.; Mallouk, T. E.; Wrighton, M. S. J. Phys. Chem. 1987, 91, 648. (17) Shibuya, M.; Nishina, T.; Matsue, T.; Uchida, I. J. Electrochem. Soc. 1996, 143, 3157. (18) Shibuya, M.; Yamamure, S.; Matsue, T.; Uchida, I. Chem. Lett. 1995, 749. (19) Sze, S. M. Semiconductor DeVices, Physics and Technology; John Wiley & Sons: New York, 1985. (20) Meulenkamp, E. A. J. Phys. Chem. B 1998, 102, 5566. (21) Meulenkamp, E. A. J. Phys. Chem. B 1998, 102, 7764. (22) Hoyer, P.; Eichberger, R.; Weller, H. Ber. Bunsen-Ges. Phys. Chem. 1993, 97, 630. (23) Hoyer, P.; Weller, H. J. Phys. Chem. 1995, 99, 14096. (24) Redmond, G.; O′Keeffe, A.; Burgess, C.; MacHale, C.; Fitzmaurice, D. J. Phys. Chem. 1993, 97, 11081. (25) Enright, B.; Fitzmaurice, D. J. Phys. Chem. 1996, 100, 1027. (26) de Jongh, P. E.; Meulenkamp, E. A.; Vanmaekelbergh, D.; Kelly, J. J. To be submitted. (27) van de Hulst, H. C. Light Scattering by Small Particles; Dover Publications: New York, 1981. (28) Vossmeyer, T.; Katsikas, L.; Giersig, M.; Popovic, I. G.; Diesner, K.; Chemseddine, A.; Eychmu¨ller, A.; Weller, H. J. Phys. Chem. 1994, 98, 7665. (29) Lemon, B. I.; Hupp, J. T. J. Phys. Chem. B 1997, 101, 2426. (30) Rensmo, H.; Keis, K.; Lindstro¨m, H.; So¨dergren, S.; Solbrand, A.; Hagfeldt, A.; Lindquist, S.-E. J. Phys. Chem. B 1997, 101, 2598. (31) Bedja, I.; Kamat, P. V.; Hua, X.; Lappin, A. G.; Hotchandani, S. Langmuir 1997, 13, 2398. (32) Blok, L. The ionic double layer on zinc oxide in aqueous electrolyte solutions. Ph.D. Thesis, Utrecht University, 1968. (33) Brus, L. E. J. Chem. Phys. 1984, 80, 4403. (34) See, e.g., Simon, U. AdV. Mater. 1998, 10, 1487, and references therein. (35) Janz, G. J.; Tomkins, R. P. T. Nonaqueous Electrolytes Handbook, Vol. I; Academic Press: New York, 1972.
7838 J. Phys. Chem. B, Vol. 103, No. 37, 1999 (36) Riddick, J. A.; Bunger, W. B. in Techniques of Chemistry, Vol. II, Organic SolVents - physical properties and methods of purification, 3rd ed.; John Wiley & Sons: New York, 1971. (37) Roth, A. P.; Williams, D. F. J. Appl. Phys. 1981, 52, 6685. (38) Heiland, G.; Mollwo, E.; Sto¨ckmann, F. In Solid State Physics, Vol. 8; Academic Press: New York, 1959; p 191. (39) Chopra, K. L.; Major, S.; Pandya, D. K. Thin Solid Films 1983, 102, 1. (40) Rupprecht, H. J. Phys. Chem. Solids 1958, 6, 144. (41) Sunde, S. J. Electrochem. Soc. 1996, 143, 1123. (42) Stauffer, D.; Aharony, A. Introduction to Percolation Theory, 2nd ed.; Taylor & Francis: London, 1994.
Meulenkamp (43) About 30% of the particles have a size more than 10-15% different from the average (5.0 nm).21 A 4 nm particle has an approximately 67 meV wider band gap,20 which results in shift of the conduction band edge and Eacc of about 50 meV. (44) See, e.g., Orton, J. W.; Powell, M. J. Rep. Prog. Phys. 1980, 43, 1263. (45) Meulenkamp, E. A. Unpublished results. (46) van Dijken, A.; Meulenkamp, E. A.; Vanmaekelbergh, D.; Meijerink, A. Manuscript in preparation. (47) Granqvist, C. G. Handbook of Inorganic Electrochromic Materials; Elsevier: Amsterdam, 1995.
