Size-Dependent Electrical Transport in CdSe Nanocrystal Thin Films

Aug 23, 2010 - Third, the charge transport could be described in terms of the nearest-neighbor-hopping mechanism with a size-dependent activation ener...
1 downloads 8 Views 382KB Size
Download PDF
pubs.acs.org/NanoLett

Size-Dependent Electrical Transport in CdSe Nanocrystal Thin Films Moon Sung Kang, Ayaskanta Sahu, David J. Norris,* and C. Daniel Frisbie* Department of Chemical Engineering and Materials Science, University of Minnesota, 421 Washington Avenue SE, Minneapolis, Minnesota 55455 ABSTRACT Electrical transport in films of CdSe nanocrystals with diameters varying from 2.9 to 5.1 nm was examined over 233-300 K by employing electrolyte gating to control the electron density. The transport parameters varied strongly and systematically with nanocrystal diameter. First, a strong correlation was observed between the device turn-on voltage and the size-dependent position of the lowest unoccupied electronic states of the nanocrystals. Second, the electron mobility increased with increasing particle diameter and reached a high value of 0.6 cm2/(V s) for films with 5.1 nm nanocrystals. Third, the charge transport could be described in terms of the nearest-neighbor-hopping mechanism with a size-dependent activation energy and a pre-exponential factor for mobility. The activation energy can be viewed as a size-dependent charging energy of an individual nanocrystal. Collectively, the combination of size- and temperature-dependent measurements provides a powerful approach to understanding electrical transport in nanocrystal films. KEYWORDS CdSe nanocrystals, electrical transport, size dependence, thin-film transistors, ion-gel gate dielectric

T

he size-dependent tunability of electronic energy levels in semiconductor nanocrystals (NCs) has driven the long-term focus on optical properties of these materials.1-5 More recently, researchers have attempted to exploit these properties and incorporate NCs into useful optoelectronic devices. In particular, the tunable absorption and emission of NCs can potentially allow solar cells,6,7 lightemitting diodes,8,9 and photodetectors10 to be optimized systematically. However, for this goal, not only the optical properties of NCs but also the charge transport through NC assemblies must be understood. Moreover, because devices typically involve electrical conduction through films of NCs, improved charge transport can immediately lead to better device performance. Increases in the electrical coupling between NCs, for example, have led to higher power conversion efficiencies in NC-based solar cells.7,11 Unfortunately, compared to the extensive studies on optical properties, electrical conduction in NC assemblies has been much less explored.12-19 Most studies on the charge transport mechanism have focused on the low-temperature regime,12-15 and other transport studies performed near room temperature were mostly based on poorly conductive films.16,17 More surprisingly, the effect of particle size, which is the most characteristic parameter of NCs, on electricaltransport properties of NC films has not been addressed systematically18,19 with one recent exception.20 Liu et al.20 observed a clear increase in carrier mobility with PbSe NC size and ascribed this behavior to (i) the variation in total hops needed for carriers to transport through NC films with

different particle sizes and (ii) the different depth of trap states for different-sized NCs. However, a complete assessment of charge transport also requires an examination of the temperature dependence of the conduction. So far, the combination of size and temperature dependence of charge transport has not been reported for semiconductor NC films. Here we address the size- and temperature-dependence of electrical transport in films of CdSe NCs, which are the benchmark NC system.16,21,22 Thin-film transistors (TFTs) were used as the test structure to probe the fundamental transport metrics, including carrier type, turn-on voltage, and mobility. Moreover, the current-voltage (I-V) characteristics of TFTs measured as a function of temperature allowed us to investigate the mechanism of carrier transport through NC films. We used six different particle diameters and observed systematic reductions in device turn-on voltage and increases in mobility with increasing NC size. The electron conduction is consistent with the nearest-neighborhopping mechanism in which particle size impacts both the activation energy and the pre-exponential factor for the carrier mobility. Because these results focus on the role of particle size, which is the most characteristic parameter of NCs, they provide fundamental information for understanding charge transport in NC films. To control the carrier concentration in the films, we employed a high-capacitance gel electrolyte as the gatedielectric material for the TFTs.23 The gel is a self-assembled network of triblock copolymer infused with an ionic liquid, and it has several advantages for charge transport experiments. First, as an electrolyte, it enables extremely large carrier densities to be obtained, on the order of ∼1 electron per particle or ∼1014 electrons/cm2, through electrochemical charging.24 At such high carrier densities charge traps are

* To whom correspondence should be addressed, [email protected] and [email protected]. Received for review: 07/6/2010 Published on Web: 08/23/2010 © 2010 American Chemical Society

3727

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727–3732

FIGURE 1. (a) Optical absorbance spectra of six different sizes of CdSe NCs dispersed in hexane. (b) Schematic cross-sectional diagram of an ion-gel-gated CdSe NC TFT (not to scale). The length and the width of the channel varied from 100 to 200 µm and 1 to 2 mm, respectively. (c) Schematic diagram of a gate electrode/ion gel/NC film in cross section (not to scale). For a given gate voltage, VG, the potential drop mostly occurs at the gate electrode/ion gel and ion gel/NC film interfaces. Therefore, by measuring the reference potential, Vref, relative to the grounded source electrode, the potential drop at the ion gel/semiconductor interface can be accurately monitored.

easily compensated, meaning the measured transport can better reflect the inherent properties of the NC film. Second, the ion gel is readily compatible with variable temperature measurements because it does not freeze until ∼220 K. Finally, the potential drop at the gel/NC-film interface can be monitored by means of a reference electrode which allows very small shifts in the conduction onset voltage (referred to as the turn-on voltage) to be detected.25 Specifically, we used ion gels comprising the triblock copolymer poly(styrene-block-methyl methacrylate-blockstyrene) (PS-PMMA-PS) (10 wt %) and the ionic liquid 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI]) (90 wt %). The gels were prepared by a procedure described previously (see the Supporting Information).24 The polymer network in the ion gel provides solid integrity to the electrolyte, while not appreciably changing the ionic conductivity. Separately, six different sizes of CdSe NCs, passivated primarily with trioctylphosphine, trioctylphosphine oxide, hexadecylamine, and dodecylphosphonic acid as surface ligands, were prepared by modified literature methods (see Supporting Information).26 The optical absorbance spectra of these CdSe NCs are shown in Figure 1a along with diameters determined from a published calibration.22 Figure 1b shows a schematic cross section of an ion-gelgated CdSe NC TFT that was fabricated by the following procedure. In a nitrogen glovebox, spin-coated films of NCs (∼50 nm thick) were prepared from a 20 mg/mL dispersion of CdSe NCs in anhydrous octane on a Si/SiO2 substrate that was prepatterned with source and drain Cr/Au (2.5 nm/37.5 nm) electrodes. The length (L) and the width (W) of the channel varied from 100 to 200 µm and from 1 to 2 mm, respectively. To improve the conduction of the as-deposited insulating films, they were chemically treated with 0.08 M © 2010 American Chemical Society

FIGURE 2. I-V and C-V characteristics of an ion-gel-gated CdSe NC TFT based on 4.2 nm CdSe NCs. (a) ID-VG characteristic at VD of 0.1 V. VG was swept at 100 mV/s. F and R represent the forward and the reverse sweep directions. (b) ID-Vref characteristic at VD of 0.1 V. Vref was measured simultaneously during the VG sweep. The refert enced voltage at the onset of conduction is named VRef and is indicated with a red arrow. (c) ID-VD characteristic at different VG values. VD was swept at 100 mV/s. (d) C-V characteristic of a gate electrode/ion gel/CdSe NC film test structure embedded in the TFT. The measurement was performed at a frequency of 10 Hz.

NaOH in anhydrous methanol for 10 min. The role of NaOH was to remove the original bulky ligands and form cadmium hydroxide complexes.27,28 The treatment resulted in a reduction of the interparticle spacing to ∼0.2 nm, which was obtained from X-ray scattering, while retaining the excitonic features in the absorption spectra. Due to the reduced interparticle spacing, micrometer-wide cracks were formed in the NC films. These were filled with a second round of spin-coating from the same CdSe NC dispersion, followed by another cycle of NaOH treatment. The resulting films were then continuous without noticeable cracks. On top of the chemically treated film, the ion gel was spread over the channel region (∼120 µm thick). Finally, a platinum electrode (area, A ) (3-6) × 10-3 cm2) was laminated to the top of the ion gel to serve as the gate electrode. The entire device fabrication was performed inside a nitrogen glovebox. Figure 2 displays typical I-V characteristics and the capacitance-voltage (C-V) characteristic obtained from a film of 4.2 nm CdSe NCs measured at room temperature. All electrical characterization was performed in a Desert Cryogenics (Lakeshore) probe station under vacuum (∼10-6 Torr). An oxidized Ag wire was used as the reference electrode.25 Figure 2a displays the drain current-gate volt3728

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727-–3732

sizes of CdSe NCs. These values were plotted versus the absolute energies relative to the vacuum level of the lowest conduction-band state in each of the samples. These energies were estimated by proportioning the energy shifts of the first absorption feature into electron and hole levels using the simple effective mass approximation and then combining the electron offsets with the position of the conductionband edge in CdSe.29 The data show a clear correlation t . Smaller between the position of the electron levels and VRef applied voltages were necessary to inject electrons into films of larger NCs which have lower electron levels.30 Moreover, the slope of the trend line is close to 1, which demonstrates a nearly 1-to-1 correspondence between the turn-on voltage and the lowest electron level. This indicates that the position of this level, rather than trap states, determines the potential required for carrier injection in our devices. This conclusion is further supported by a sharp turn-on in the devices with subthreshold swings of ∼60 mV/(decade of drain current), which indicates that the charge transport is not significantly suppressed by trap states. Note that the theoretical lower limit for the subthreshold swing in a TFT is 59 mV/(decade of drain current).24,31 Electron mobilities (µ) in these NC TFTs were computed using the equation µ ) (L/W)(ID/enVD) which is derived from Ohm’s law. Here e and n are the elemental charge and sheet carrier density, respectively. Two independent methods were employed to obtain n. The first method involved measuring the C-V characteristics of the ion gel in the TFTs using an HP 4192A LF impedance analyzer and then applying the equation n ) ∫C dV/e to determine the carrier density (cm-2).24 For the measurement, the source and drain electrodes were grounded and a bias was applied to the gate electrode. Figure 2d displays the C-V characteristic at a frequency of 10 Hz obtained for the same film from which the I-V characteristics were taken. Capacitance increased with larger applied voltage showing the electron accumulation during the voltage sweep. An electron density of 1.3 × 1014 carriers/cm2 (∼1 carrier/particle)24 was obtained from the integral between 1.15 V (the corresponding VG value at t ) and 2.5 V. The large carrier densities (greater than 1014 VRef carriers/cm2) are a result of the high capacitance of the gel electrolyte and the penetration of the electrolyte into the NC lattice.23 Using this n, an electron mobility of 0.3 cm2/(V s) was obtained. Note that penetration of electrolyte into NC films does not affect the average mobility calculation using the equation above. The calculation of µ is based on Ohm’s law and requires only that the sheet carrier density n is known. An effective two-dimensional sheet density can be accurately defined even if the transport in the NC film is distributed through the film thickness. The second method employed to obtain n involved measuring the gate current-gate voltage (IG-VG) characteristics at different gate voltage sweep rates (rV). The details of this method are described in the Supporting Information. From this method, an electron density of 1.1 × 1014 carriers/

t FIGURE 3. (a) Size dependence of VRef . The values were averaged from three to five devices and the error bar represents 1 standard deviation. The position of the lowest unoccupied electronic state for a given sized NC was estimated as described in the text. The slope of the trend line is -0.88 and the R2 value of the linear regression t is 0.933, implying a strong correlation between VRef and the position of the electron level. (b) Size dependence of mobility. Mobilities were computed at a gate voltage of 2.5 V. The values were averaged from three to five devices and the error bar represents 1 standard deviation.

age (ID-VG) characteristic of the film obtained at a constant drain voltage (VD) of 0.1 V. VG was swept from 0 to 2.5 V at a rate of 100 mV/s. The curve reveals characteristic nchannel conduction in which ID increases more than 3 orders of magnitude with positive VG. Also, a pronounced hysteresis was observed such that ID for the reverse sweep (R) was higher than for the forward sweep (F). By recording the reference potential (Vref) relative to the grounded source using the Ag reference wire embedded in the ion gel during the VG sweep, the potential drop across the ion gel/NC film interface was measured (see Figure 1c). Note that the measured reference potential did not depend on the location of the reference wire within the gel. Figure 2b displays ID as a function of Vref which shows negligible hysteresis compared to that shown in Figure 2a. Because the reference potential measures the potential drop across the ion gel/NC film interface, the disappearance of the hysteresis in ID-Vref plot indicates that the hysteresis of the original ID-VG trace is likely the result of sluggish motion of ions at the ion gel/ gate electrode interface, while the penetration of the electrolyte into the NC films is reversible. Finally, Figure 2c shows an ID-VD characteristic of the film measured at five different VG values, which also demonstrates typical nchannel conduction through the film. ID-VD characteristics of NC films with other NC sizes are shown in Figure S1 in the Supporting Information. As mentioned above, accurate measurement of the potential drop across the ion gel/NC film interface by the reference electrode allows determination of the turn-on t . This parameter denotes the interfacial potential voltage, VRef at which free carriers are injected into the NC films. Because the injection of free carriers leads to onset of conduction, t can be determined from the voltage at which ID sharply VRef rises, as indicated with a red arrow in the inset of Figure 2b. t that were obFigure 3a shows the average values of VRef tained from three to five devices for each of our six different © 2010 American Chemical Society

3729

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727-–3732

cm2 and an electron mobility of 0.4 cm2/(V s) were obtained. These values are reasonably consistent with those obtained from the first method above. To our knowledge, these mobilities are the highest values that have been achieved to date in CdSe NC films. We attribute such high mobilities to (i) the high induced carrier densities obtained from electrolyte gating,21,32 which results in substantial trap filling, and (ii) the short interparticle distance between NCs resulting from NaOH treatment. (It has been suggested that OHgroups are the shortest ligands for interparticle coupling.27,28) Also, note that by using the ion-gel gate dielectric, high electron mobilities in our CdSe NC films were achieved while retaining the integrity of the solid state device, which is an important practical requirement for applications of CdSe NC devices. Figure 3b depicts the average mobility values that were obtained from three to five devices for each of our six different sizes of CdSe NCs at a fixed gate voltage of 2.5 V. Clearly, we observe an increase in electron mobility with increasing particle size, similar to the trend observed for PbSe NC films.20 A mobility as high as 0.6 cm2/(V s) was obtained for the largest particles (5.1 nm). In general, the mobility in NC films depends on the carrier concentration and the mobility values above, calculated at a fixed gate voltage, were not necessarily obtained at a constant carrier concentration due to threshold voltage shift. For a more rigorous comparison, therefore, we computed mobilities for the NC films with different particle sizes calculated at a fixed carrier concentration of 1 × 1014 carriers/cm2. However, a monotonic increase in mobility was still observed with increasing particle size, as shown in Figure S3 in the Supporting Information. To elucidate the origin of the size-dependent mobility, we examined the temperature dependence of the mobility for NC films with different particle diameters within a temperature range of 233-300 K. Mobilities at a given temperature were calculated from ID-VG and C-V characteristics obtained at that temperature (see Figure S4 and a description of our method in the Supporting Information). Figure 4a displays the temperature-dependent mobility, computed at a fixed carrier concentration of 1 × 1014 carriers/cm2, for our CdSe NC films with four different particle sizes.33 Over 233-300 K, the mobilities followed an Arrhenius-type behavior, µ ) µ0 exp(-EA/kBT), where µ0, EA, kB, and T, are the pre-exponential factor, the charge-transport activation energy, the Boltzmann constant, and the temperature in Kelvin, respectively. Arrhenius behavior in the mobility can be explained by either the multiple-trapping-and-release (MTR) model34 or the nearest-neighbor-hopping (NNH) model.35 The MTR model assumes that most of the carriers are trapped in localized trap states, and the charge transport occurs through extended states when the carriers are thermally activated (or released) from the trap states. The NNH model describes electron transport as sequential hopping (activated tunneling) between neighboring localized states. © 2010 American Chemical Society

FIGURE 4. (a) Temperature dependence of the mobility, which was computed at a carrier concentration of 1 × 1014 carriers/cm2. (b) Size dependence of EA (blue) and µ0 (red), where d is the sum of the particle size and the interparticle spacing. The error bar represents 1 standard error computed from the linear regression on the fitting of mobility versus temperature.

Collectively, our data suggest that it is likely that the electron transport in the CdSe NC films follows the NNH mechanism rather than the MTR model. First, the low values of the subthreshold swing evident in the ID-VG characteristics suggest that charge transport is nearly free from the influence of trapping. We note that trap-free transport has also been achieved in electrolyte-gated ZnO NC assemblies.36 More conclusive evidence can be found by examining EA as a function of n. If the charge transport follows the MTR model, EA should decrease with increasing n. This is because as n increases, deeper traps are filled and carriers occupy shallower traps on average, which results in a smaller mean value of EA. However, we observe that the correlation between n and EA is rather weak, and in fact, EA increases with n slightly (see Figure S4 in the Supporting Information). A similar trend between n and EA was reported for electrolytegated ZnO NC assemblies.36 Therefore, it seems unlikely that the MTR model explains the Arrhenius behavior observed in our CdSe NC films. Instead, the electron transport is more consistent with the NNH transport mechanism, where electrons hop between the lowest electron levels. From the slope and the y-intercept of the Arrhenius plots, we extracted EA for the nearest neighbor hopping and µo, respectively, for four different diameter particles, as shown in Figure 4b. These values are summarized in Table 1. A monotonic decrease in EA and an increase in µo are observed. Thus, the increase in electron mobility with particle size originates from favorable changes in both EA and µo. 3730

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727-–3732

TABLE 1. Estimated Activation Energies (EA), Pre-exponential Factors (µ0), and the R2 Values for the Linear Regression Fitting Obtained from Figure 4a, for NC Films with Four Different Sizes (D) D (nm)

EA (meV)

µ0 (cm2/(V s))

R2

5.1 4.5 4.2 3.6

30 35 40 56

1.6 1.3 1.2 1.2

0.984 0.996 0.987 0.973

Because variations in the total number of hops are unrelated to the activated process, this effect should influence µo instead of EA in the Arrhenius relation. Thus, one would expect that NC films made from larger particles will exhibit higher µo. A second explanation for the size dependence of µo involves inhomogeneities in the positions of the electron levels. Because the electrons hop through electronic states that have a distribution in energy due to variations in NC size, films made from NCs that have a narrower energy distribution will have stronger interparticle coupling than those that have a broader distribution. Stronger coupling leads to higher values of µo. From the simple effective mass approximation, one can estimate the width of the lowest electron state (∆E1S) by

Determining the origin of the size dependence of EA is important for understanding the underlying charge transport mechanism through the CdSe NC films. It has been shown in the literature that electrolyte-gated CdSe NC films, treated with NaOH, follow the Efros-Shklovskii variable-rangehopping (ES-VRH) model at lower temperatures.11,28 This model takes into account the probability of hopping between states distributed spatially and energetically and a gap in the electronic density of states arising from Coulombic correlations, which results in an optimal hopping distance at a given temperature.37 However, at higher temperatures, where enough thermal energy is provided, consideration of the energy separation in finding the optimum hopping distance becomes less important and the transport occurs via hopping between the nearest neighboring states, i.e., the NNH mechanism.14 In this regard, a NC system that follows the ES-VRH model at lower temperatures eventually transitions to follow the NNH model at higher temperatures. For such a NC system, one can propose that EA for NNH is proportional to the Coulombic charging energy of an individual NC, such that EA can be given by βe2/(4πεε0d),14 where β is a proportionality constant and e2/(4πεε0d) is the Coulombic charging energy of an individual NC. ε0 is the vacuum permittivity, ε is the dielectric constant, and d is the nearestneighbor distance between NCs, i.e., the sum of the particle diameter and the interparticle spacing. (A derivation for EA is shown in the Supporting Information.) Moreover, because the change of ε is small over the range of particle diameters from 3.6 to 5.1 nm,38 it is reasonable to attribute the change in EA to d. In agreement with this model, the activation energies extracted for our samples scale as 1/d, as seen in Figure 4b. Note that the bulk dielectric constant of bulk CdSe (∼9)38 is much smaller than that of bulk PbSe (∼210),39 so that the charging energy for CdSe NCs cannot be neglected, as suggested previously for PbSe NCs.20 The role of the charging energy has also been considered in an earlier publication that is primarily focused on the ES-VRH transport at low temperature (T < 120 K) for a single-size 5.4 nm CdSe NC system.14 Overall, the importance of Figure 4b is that it demonstrates scaling of EA with 1/d and thus supports the physical picture that Coulomb charging is a key factor in transport through NC films. To explain the size dependence of µo, we must consider two possibilities. First, it is reasonable that electrons traveling via NNH through NC assemblies will require fewer hops to cross a channel of fixed length as the NC size increases.20 © 2010 American Chemical Society

∆E1S ) h2 /(4mR2) ∆R/R where h, m, R, and ∆R/R are the Planck constant, the effective mass of the electron, the radius of the particle, and the size distribution of the particles, respectively. Therefore, for a given particle size distribution, larger particles will have smaller ∆E1S and thus higher µo values. To quantify the exact contributions of each of the explanations, however, more detailed modeling will be required. In summary, we report here the effect of NC size on the electrical transport through thin films of CdSe NCs by using TFTs gated with a gel electrolyte. First, we observed that changes in the particle size, which shifts the position of the lowest unoccupied electronic states of NCs, directly influence the device turn-on voltage of our TFTs. Second, the electron mobility increases with particle size, as high as 0.6 cm2/(V s), and follows the Arrhenius relation within a temperature range of 233-300 K, which is consistent with the nearestneighbor-hopping transport model. The size dependence of the mobility can be explained by both the size dependence of EA, which scales with the Coulombic charging energy of an individual particle, and the size dependence of µ0, which is associated with the total number of interparticle hops and the energy distribution of the electronic states. We believe that fundamental understanding of the size- and temperature-dependent transport properties of NC films will advance the applications of these materials in useful electronic/ optoelectronic devices. Acknowledgment. We thank Professor Timothy Lodge and Keun Hyung Lee for the synthesis of the PS-PMMA-PS triblock copolymer. We also thank Professor Dong Yu, University of California, Davis, for helpful discussion. This work was supported primarily by the MRSEC Program of the National Science Foundation (NSF) under Award Number DMR-0819885. Additional support was provided by the NSF Materials World Network under Award Number DMR0908629 and by the Center for Nanostructure Applications at the University of Minnesota (UMN). A.S. acknowledges funding from the Industrial Partnership for Research in Interfacial and Materials Engineering (IPRIME). We utilized resources at the UMN Characterization Facility and the UMN 3731

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727-–3732

Nanofabrication Center, which receive partial support from the NSF under the NNIN program.

(20) Liu, Y.; Gibbs, M.; Puthussery, J.; Gaik, S.; Ihly, R.; Hillhouse, H. W.; Law, M. Nano Lett. 2010, 10, 1960. (21) Yu, D.; Wang, C.; Guyot-Sionnest, P. Science 2003, 300, 1277. (22) Yu, W. W.; Qu, L.; Guo, W.; Peng, X. Chem. Mater. 2003, 15, 2854. (23) Cho, J. H.; Lee, J.; Xia, Y.; Kim, B.; He, Y.; Renn, M. J.; Lodge, T. P.; Frisbie, C. D. Nat. Mater. 2008, 7, 900. (24) Kang, M. S.; Lee, J.; Norris, D. J.; Frisbie, C. D. Nano Lett. 2009, 9, 3848. (25) Xia, Y.; Cho, J.; Paulsen, B.; Frisbie, C. D.; Renn, M. J. Appl. Phys. Lett. 2009, 94, No. 013304. (26) Reiss, P.; Bleuse, J.; Pron, A. Nano Lett. 2002, 2, 781. (27) Jarosz, M. V.; Porter, V. J.; Fisher, B. R.; Kastner, M. A.; Bawendi, M. G. Phys. Rev. B 2004, 70, 195327. (28) Yu, D.; Wehrenberg, B. L.; Jha, P.; Ma, J.; Guyot-Sionnest, P. J. Appl. Phys. 2006, 99, 104315. (29) Jasieniak, J.; Pacifico, J.; Signorini, R.; Chiasera, A.; Ferrari, M.; Martucci, A.; Mulvaney, P. Adv. Func. Mater. 2007, 17, 1654. (30) Wang, C.; Shim, M.; Guyot-Sionnest, P. Science 2001, 291, 2390. (31) Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices, 3rd ed.; Wiley: Hoboken, NJ, 2006. (32) Vanmaekelbergh, D.; Liljeroth, P. Chem. Soc. Rev. 2005, 34, 299. (33) Because the capacitance of an ion gel decreases with decreasing temperature, computing mobility at the same carrier concentration for different temperatures is important. This is because a decrease in the capacitance itself can also contribute to the lowering of the mobility, in addition to the inherent decrease in the mobility. Note that the capacitance of ion-gel-gated devices made from smaller NCs decreased more rapidly with decreasing temperature than those made from larger NCs. For devices made from the smallest NCs, it was not possible to induce a carrier concentration of 1 × 1014 carriers/cm2 below a certain temperature. As a result, measurements were carried out within a narrower temperature range for the films made from 4.2 and 3.6 nm NCs and an even narrower range for the smallest NCs. (34) Gilles, H.; Mohsen, E. H.; Riadh, H. J. Appl. Phys. 2000, 87, 4456. (35) Baranovski, S. Charge Transport in Disordered Solids with Applications in Electronics; John Wiley & Sons, Ltd.: Chichester, 2006. (36) Roest, A. L.; Kelly, J. J.; Vanmaekelbergh, D. Appl. Phys. Lett. 2003, 83, 5530. (37) Shklovskii, B. I.; Efros, A. L. Electronic Properties of Doped Semiconductors; Springer-Verlag: Berlin, 1984. (38) Wang, L.-W.; Zunger, A. Phys. Rev. B 1996, 53, 9579. (39) Luther, J. M.; Law, M.; Beard, M. C.; Song, Q.; Reese, M. O.; Ellingson, R. J.; Nozik, A. J. Nano Lett. 2008, 8, 3488.

Supporting Information Available. Materials, detailed experimental methods, derivations of equations, and supplementary figures. This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19)

Klimov, V. Semiconductor and Metal Nanocrystals; Marcel Dekker, Inc.: New York, 2004. Du, H.; Chen, C.; Krishnan, R.; Krauss, T. D.; Harbold, J. M.; Wise, F. W.; Thomas, M. G.; Silcox, J. Nano Lett. 2002, 2, 1321. Wehrenberg, B. L.; Wang, C.; Guyot-Sionnest, P. J. Phys. Chem. B 2002, 106, 10634. Nirmal, M.; Brus, L. Acc. Chem. Res. 1998, 32, 407. Efros, A. L.; Rosen, M. Annu. Rev. Mater. Sci. 2000, 30, 475. Nozik, A. J. Physica E 2002, 14, 115. Kamat, P. V. J. Phys. Chem. C 2008, 112, 18737. Colvin, V. L.; Schlamp, M. C.; Alivisatos, A. P. Nature 1994, 370, 354. Coe, S.; Woo, W.-K.; Bawendi, M.; Bulovic, V. Nature 2002, 420, 800. Konstantatos, G.; Howard, I.; Fischer, A.; Hoogland, S.; Clifford, J.; Klem, E.; Levina, L.; Sargent, E. H. Nature 2006, 442, 180. Luther, J. M.; Beard, M. C.; Song, Q.; Law, M.; Ellingson, R. J.; Nozik, A. J. Nano Lett. 2007, 7, 1779. Houtepen, A. J.; Kockmann, D.; Vanmaekelbergh, D. l. Nano Lett. 2008, 8, 3516. Talapin, D. V.; Murray, C. B. Science 2005, 310, 86. Yu, D.; Wang, C.; Wehrenberg, B. L.; Guyot-Sionnest, P. Phys. Rev. Lett. 2004, 92, 216802. Wehrenberg, B. L.; Yu, D.; Ma, J.; Guyot-Sionnest, P. J. Phys. Chem. B 2005, 109, 20192. Ginger, D. S.; Greenham, N. C. J. Appl. Phys. 2000, 87, 1361. Drndic, M.; Jarosz, M. V.; Morgan, N. Y.; Kastner, M. A.; Bawendi, M. G. J. Appl. Phys. 2002, 92, 7498. Mentzel, T. S.; Porter, V. J.; Geyer, S.; MacLean, K.; Bawendi, M. G.; Kastner, M. A. Phys. Rev. B 2008, 77, No. 075316. Romero, H. E.; Drndic, M. Phys. Rev. Lett. 2005, 95, 156801.

© 2010 American Chemical Society

3732

DOI: 10.1021/nl102356x | Nano Lett. 2010, 10, 3727-–3732