Isabella R.
Franchini
a,
Adriano
Cola
b,
Aurora
Rizzo
c,
Rosanna
Mastria
c,
Anna
Persano
b,
Roman
Krahne
a,
Alessandro
Genovese
a,
Andrea
Falqui
a,
Dmitry
Baranov‡
a,
Giuseppe
Gigli
c and
Liberato
Manna
*a
aIstituto Italiano di Tecnologia, Via Morego 30, 16163, Genova, Italy. E-mail: liberato.manna@iit.it
bIMM-CNR, Institute for Microelectronics and Microsystems – Unit of Lecce, Via Monteroni, I-73100, Lecce, Italy
cNational Nanotechnology Laboratory (NNL), CNR-Nanoscienze, Via per Arnesano km 5, I-73100, Lecce, Italy
First published on 3rd August 2010
Tetrapod-shaped CdSe(core)/CdTe(arms) colloidal nanocrystals, capped with alkylphosphonic acids or pyridine, were reacted with various small molecules (acetic acid, hydrazine and chlorosilane) which induced their tip-to-tip assembly into soluble networks. These networks were subsequently processed into films by drop casting and their photoconductive properties were studied. We observed that films prepared from tetrapods coated with phosphonic acids were not photoconductive, but tip-to-tip networks of the same tetrapods exhibited appreciable photocurrents. On the other hand, films prepared from tetrapods coated with pyridine instead of phosphonic acids were already highly photoconductive even if the nanocrystals were not joined tip-to-tip. Based on the current–voltage behavior under light we infer that the tunneling between tetrapods is the dominant charge transport mechanism. In all the samples, chemically-induced assembly into networks tended to reduce the average tunneling barrier. Additionally, pyridine-coated tetrapods and the tip-to-tip networks made out of them were tested as active materials in hybrid photovoltaic devices. Overall, we introduce an approach to chemically-induced tip-to-tip assembly of tetrapods into solution processable networks and demonstrate the enhancement of electronic coupling of tetrapods by various ligand exchange procedures.
Assembly of shape controlled nanocrystals like nanorods2 (and the study of their photoconductive properties),3 or of branched nanocrystals4 has also been investigated, although to a lower extent. Tip-to-tip assembly of these types of nanocrystals, to form one dimensional arrays (for nanorods), or 3D network structures (for branched nanocrystals) is hard to tackle but nonetheless is a technologically important research field. For example, it could lead to the fabrication of nanocomposite fibers with controllable and combined mechanical, electric, thermoelectric, or photo-conductive behavior, to cite a few, or to lightweight, high performing materials useful in catalysis or in tissue engineering. Some reports of tip-to-tip assembly of shape-controlled nanocrystals, mainly nanorods, have appeared so far, which exploited the higher reactivity of nanorod tips towards the binding of organic molecules or the growth of domains of other materials that could mediate tip-to-tip nanocrystal organization.5
Tetrapod-shaped colloidal nanocrystals of semiconductors are a popular class of branched nanocrystals and have been exploited in various applications, most notably in thin-film photovoltaic devices based mainly on blends with organic molecules and/or polymers.6,7 The efficiency of these types of devices is governed by many parameters, one of which is electron transport in the active layer, which proceeds via electrons moving from nanocrystal to nanocrystal. Efficient electron transport requires the presence of “percolating” pathways for electrons, hence arrays made of nanocrystals that are sufficiently close to each other and which connect the anode and the cathode of the device. Attempts to improve transport through films of nanocrystals have involved high-temperature annealing in order to fabricate network structures of interlinked nanocrystals, such that direct electrical contact of a large number of nanocrystals was guaranteed, hence more pathways for electrons throughout the whole layer were created.8
We propose here an alternative approach for the fabrication of network structures of tetrapods, based on the pre-organization of semiconductor CdSe(core)/CdTe(arms) tetrapods into stable and soluble aggregates already in the solution phase, which can be potentially useful in photovoltaic applications, e.g. as inks for printable solar cells. To this aim we exploited the higher reactivity of the tips of the tetrapod arms, compared to other regions of the tetrapod surface, under various chemical treatments. The preferential removal of surfactants from these regions, or their exchange with short molecules, led to destabilization of the tips, which caused aggregation of the tetrapods in solution, such that they started organizing in tip-to-tip connected three-dimensional network structures that were still soluble. Solutions of these networks were drop-cast on substrates, yielding nanocrystal films upon solvent evaporation.
We found that nanocrystal films prepared from as-synthesized tetrapods, coated at their surface with alkylphosphonic acids (hence as they were obtained from the synthesis), exhibited no photocurrent, while their organization into networks in solution (upon reaction with various molecules) yielded films with appreciable photocurrent (few pA at 0.5 V μm−1). For comparison, films prepared from non-assembled tetrapods, but capped with pyridine instead of phosphonic acids, exhibited already much higher photoconductivity (few nA at 0.5 V μm−1). Notably, there was no substantial variation in phototransport if pyridine-capped tetrapods were assembled into networks.
In order to investigate the possibility to integrate the network structures as photoactive materials in hybrid solar cells, the tetrapods coated with pyridine were chosen because of their higher photoconductivity. Compared to the devices prepared with non-assembled tetrapods, those prepared with networks assembled via chlorosilane and hydrazine showed a slight reduction in their performances (as discussed herein), while those prepared with networks assembled via acetic acid showed a slight improvement. Note that hereinafter throughout the text terms “assembled” and “assembly” refer to tip-to-tip joined tetrapods forming a network.
Fig. 1 (a) Wide-field low magnification TEM image of CdSe@CdTe tetrapods prepared by the seeded-growth approach and coated with alkylphosphonic acids. The tetrapods consist of CdSe ‘sphalerite’ seeds, onto which four wurtzite CdTe arms were grown; (b–c) the same tetrapods assembled by addition of acetic acid, followed by sonication for 30 min; (d) by addition of trimethylchlorosilane; and (e) by addition of hydrazine. The solvent in all cases was toluene. The samples were freshly prepared in solution, a drop of solution was deposited on a carbon coated copper grid, and the solvent was allowed to evaporate. The networks imaged under TEM represent what is left under solvent evaporation. It is possible that they differ from the networks in solution by many aspects, because several processes might occur during evaporation: bigger networks might break into smaller networks as a result of convective flow; separate smaller networks might aggregate at the later stages of solvent evaporation due to strong van der Waals interactions; a network could partly collapse as a result of escape of solvent that was previously entrapped in between the tetrapods. |
For both types of ligand-coated tetrapods we followed three different procedures in order to destabilize the tip regions of the tetrapods. These were: (i) addition of acetic acid, followed by sonication for 30 min at room temperature. A similar procedure, although involving an amine, was found by O'Sullivan et al. to induce oriented attachment in CdS and Ag2S nanorods;11 (ii) addition of trimethylchlorosilane, which is known to strip off alkylphosphonates from the surface of nanocrystals efficiently;12 (iii) addition of hydrazine, a strongly coordinating bi-dentate ligand which displaces surfactants from the surface of the nanocrystals and promotes nanocrystal-to-nanocrystal bridging, and like pyridine can be removed by mild thermal treatment.13 The influence of hydrazine on the electronic and transport properties of nanocrystal films has been studied in recent years.13,14
All the procedures above were refined in order to identify the amount of reagents that had to be added such that the networks of tetrapods formed were still stable in solution (see the ESI†). Optical absorption spectra were recorded from solutions containing the various samples. For transmission electron microscopy (TEM) analysis, a drop of solution was deposited on a carbon coated copper grid, and the solvent was allowed to evaporate. Additionally, TEM analysis was carried out on representative grids after they had been annealed at 450 K for 30 min under vacuum, to check if any structural transformation occurred upon annealing.
Ti/Au (3/30 nm) interdigitated electrodes, 2 μm spaced, covering a total active area of about 25 × 25 μm2, were deposited by standard photolithography and lift-off on Si/SiO2 substrates. Solutions of the various nanocrystal samples were deposited by drop casting onto these interdigitated electrodes. Then samples were annealed at 450 K for 30 min under low vacuum (p ∼ 10−3 torr).
Current–voltage scans, both under white light illumination and in the dark, were carried on all the films in air at room temperature in an electromagnetically shielded K. Suss PM5 probe station, using a HP 4140B pA meter/DC voltage source, in order to apply the bias voltage at one electrode and measure the current at the other electrode. In photocurrent spectral experiments, a 100 W tungsten halogen lamp in combination with a 0.11 m monochromator was used as the excitation source. Light intensity was monitored with a calibrated Si photodiode in order to evaluate the total incident optical power and to normalize the device photocurrent spectra. After all transport measurements were carried out, the samples were inspected by scanning electron microscopy (SEM).
Photovoltaic (PV) devices were fabricated in a bilayer configuration. Concentrated solutions (≈ 60 mg ml−1) of different pyridine-coated tetrapods (non-assembled, assembled via acetic acid, via hydrazine and via chlorosilane) were prepared (see the ESI for details†) and deposited by spin-coating at 1500 rpm onto indium tin oxide (ITO) substrates coated with PEDOT:PSS.15 The samples were annealed on a hot plate at 150 °C for 20 min and held at 6 × 10−6 mbar overnight, after which a 30 nm thick layer of C60 was deposited by thermal evaporation and a 130 nm thick aluminium cathode was evaporated on it through a shadow mask. Finally the samples underwent a second annealing step at 150 °C for 30 min. All manipulations, except for the PEDOT:PSS deposition, were performed in a glove box under nitrogen atmosphere. Before their characterization, all devices were sealed using a glass lid attached by a bead of UV-cured epoxy resin. Current–voltage characteristics of the PV devices were studied using a Keithley 2400 Source Measure Unit and a solar simulator Spectra Physics Oriel 150 W with AM 1.5G filter set. The integrated intensity was set to 100 mW cm−2 using a thermopile radiant power meter with fused-silica window (Spectra Physics Oriel, model 70260).
In the case of pyridine-coated nanocrystals, large aggregates of tetrapods were observed already on grids prepared by deposition of the solution containing non-assembled nanocrystals, hence before any assembly treatment (Fig. 2a). In these aggregates, tetrapods appeared however to prefer side-to-side interactions, most likely due to the strong van der Waals interactions between these nanocrystals due to the short pyridine ligands coating their surface. As the starting non-assembled samples in pyridine were stable in solution, it is likely that the aggregates seen under TEM were not formed in solution, but rather during the last stages of solvent evaporation on the grid. Samples of pyridine-coated nanocrystals assembled into networks showed both side-to-side and tip-to-tip assembly (Fig. 2b-e).
Fig. 2 (a) Wide-field low magnification TEM image of the same CdSe@CdTe tetrapods as shown in Fig. 1a, after their phosphonic acid surfactants were exchanged with pyridine; TEM images of these pyridine coated tetrapods organized into networks upon reaction with: (b–c) acetic acid, followed by sonication for 30 min; (d) trimethylchlorosilane; (e–f) hydrazine. In (f) several junctions can be seen between tetrapods. |
High-resolution TEM (HRTEM) analysis on the various networks indicated that in general there were no preferential epitaxial relationships between touching arms (Fig. 3). Occasionally, two arms belonging to different tetrapods shared a very small (1–2 nm wide) sphalerite domain, but most often the interfacial region between two touching arms was amorphous. Oriented attachment in these nanocrystals is hard to achieve, because all the tips of tetrapod arms have the same polarity, as they have wurtzite structure. This is different from the case of wurtzite nanorods, which indeed expose both opposite tip regions along their elongation direction to the external environment. Hence oriented attachment in a head-to-tail arrangement of the respective lattice polarities is possible for wurtzite nanorods and has been observed.11 By contrast, no attachment between arms belonging to two tetrapods could result in a straight junction made of a single crystalline domain (see Fig. 3 for example). On the other hand a small sphalerite domain can bridge the two arms by linking each arm via one of its (111) facets, and the arms bind to this domain via their (001) facets (see Fig. 3d). Sphalerite domains could be located occasionally on the tip of an arm as a consequence of the defective growth of the arms during the tetrapod synthesis, so that wurtzite regions are interrupted by a few layers of sphalerite structure, especially in tetrapods with CdTe arms.9 In other cases, the linking between the arms occurred either by means of amorphous regions (Fig. 3e), or by bridging molecules which brought tips close to each other, but without direct touching. Even though all solutions contained the same concentration of tetrapods, when they were deposited on substrates with interdigitated electrodes the resulting film thickness varied from roughly one monolayer of tetrapods to approximately three monolayers (as seen by SEM). A representative SEM image of tetrapod film in between electrode structures is shown in Fig. 4.
Fig. 3 (a–c) Low-magnification HRTEM images of the CdSe@CdTe tetrapods networks showing several zones where attachment between tetrapod arms had occurred. (d) High magnification HRTEM of tips of wurtzite CdTe arms belonging to two different tetrapods and connected by a small sphalerite CdTe crystalline domain or (e) by an amorphous region. |
Fig. 4 Scanning electron microscopy (SEM) image of a typical device showing pyridine coated tetrapods, linked via reaction with hydrazine into network structures, in the region between two electrodes. More SEM images are reported in the ESI.† |
After annealing, in the sample of pyridine-coated tetrapods, the I–V characteristics became stable and reproducible with negligible hysteresis, but both dark current and photocurrent of the films decreased (Fig. 5a). The decrease in photocurrent might be correlated with the increased tendency of the samples towards oxidation, once a film of organic contaminants had been removed. The same annealing treatment on the sample coated with phosphonic acids caused the dark current to fall below the instrument detection limit, while still no photocurrent could be detected (Fig. 5e). As a control experiment, inspection of TEM grids on which tetrapod networks had been deposited, evidenced that no structural changes on the networks had occurred due to annealing (see the ESI†).
Fig. 5 I–V curves in the dark (black open squares) and under white light (red circles) for pyridine-coated and annealed and samples that were not assembled (a) and assembled with acetic acid (b), hydrazine (c), and chlorosilane (d). Panels (e–h) refer to phosphonic acid-coated and annealed samples, respectively non-assembled (e) and after the same assembly treatments as for pyridine-coated samples (f–h). Solid lines are least-squared fits to I–V curves in the dark and under white light with eqn (1). |
Despite the complexity of the carrier transport mechanisms, involving exciton ionization, transport across tetrapods, injection from contacts, the I–V behaviour of the films could be qualitatively explained in terms of tunneling. This is in agreement with the work of Porter et al. for films of CdTe nanocrystals, where similar transport studies using gold electrodes in planar configuration were carried out.16 Following Drndić and Porter's approach,16,17 we assume an exponential dependence of the current on the voltage that can be modeled as:
(1) |
Here V is the applied voltage, R0light is the equivalent zero-field resistance which depends on the number of conducting channels and in thus also scales inversely with the film thickness, and V0 is a characteristic voltage that gives a measure for the tunnel barriers which have to be overcome in a conductive path connecting the electrodes. V0 therefore scales linearly with the electrode distance, and by dividing V0 by the average number of nanocrystals that form a conductive path (in our case around 50 tetrapods are needed to connect the 2 μm spaced electrodes) one can obtain an averaged value for the potential barrier in between individual nanoparticles.16 Using this fitting model we obtain V0 = 51.4 V and R0light = 1.9 GΩ (Table 1, second row) for films of pyridine-coated tetrapods. These values are comparable to those obtained by Drndić et al. and Porter et al.16,17 for densely packed spherical CdTe and CdSe nanocrystal films. Dividing V0 by 50, i.e. the number of tetrapods that are needed to bridge the electrode gap of 2 μm, yields 1 V for the average potential barrier between two adjacent nanocrystals. However, we note that for films consisting of tetrapods, due to the intrinsic complexity of the tetrapods assembly in three dimensions, involving a multitude of mutual spatial configurations for neighboring nanocrystals, it is not straightforward to correlate V0 quantitavely to a specific and unique potential barrier between two adjacent nanocrystals. Furthermore, the fact that no photocurrent was measured on the tetrapod samples coated with phosphonic acids suggests that these large molecules represent insuperably high potential barriers for the charge carriers.
V 0/V | R 0light/GΩ | R 0dark/R0light | |
---|---|---|---|
Pyridine-coated “non-assembled” | 51.4 | 1.9 | 60.3 |
Pyridine-coated “networks” (hydrazine) | 13.9 | 2.0 | 43.8 |
Pyridine-coated “networks” (acetic acid) | 17.9 | 2.2 | 72.5 |
Pyridine-coated “networks” (chlorosilane) | 7.2 | 25.4 | 42.8 |
Phosphonic acids-coated “non-assembled” | N A | NA | NA |
Phosphonic acids-coated “networks” (hydrazine) | 27.5 | 226 | 58.3 |
Phosphonic acids-coated “networks” (acetic acid) | 11.3 | 2.2 × 104 | 2.9 |
Phosphonic acids-coated “networks” (chlorosilane) | 15.5 | 253 | 85.9 |
Porter additionally points out that the dark I–Vs can be explained by the same functional expression as of eqn (1), albeith with a different set of parameters.16 Indeed, the I–V curve under dark is also reasonably fitted by eqn (1) (Fig. 5). In this new fit the R0light parameter is replaced with R0dark. The R0dark/R0light can be considered as a valuable figure of merit for the photoresponse, since a material with a high R0dark/R0light ratio is suitable for photovoltaics. The R0dark/R0light ratios calculated for the various samples studied here are reported in Table 1 (last column).
All the treatments of phosphonic-acid coated tetrapods to form networks gave rise to a notable photocurrent signal. The signal was much lower than that of the corresponding pyridine samples. One minor reason for this could be the variation in film thickness, but we remind that films from non-assembled tetrapods in this case were not conductive. Hence, considering that transport though phosphonic acid coated samples is much more difficult than for pyridine-coated nanocrystals, the assembly here has the effect of opening at least one transport channel in between tetrapods, namely tunneling from tip to tip. The dark currents in these sets of samples became measurable too, although they were remarkably low (below 10−13 A at 0.5 V μm−1).
The V0 values of the phosphonic acid-coated samples were in the same range as those of the films made from pyridine coated tetrapods, but the R0light were much higher, which is expected since transport in these networks should occur only from tip to tip, hence the films overall are more resistive. One has to keep in mind however that the comparisons among the various R0light and among the R0dark/R0light ratios are to be taken with care, since they also depend on film thickness.
Fig. 6 (a) Optical absorption spectra of tetrapods coated with pyridine. Exchange of surfactant with pyridine led to a blue shift of about 3–4 nm of the lowest exciton absorption peak (which is basically due to absorption from the CdTe arms in these samples), with respect to the as-synthesized tetrapods coated with phosphonic acids. This indicates a moderate etching of the surface of the nanocrystals, hence contraction in diameter. The various treatments aiming at network formation had instead little effect on the optical spectra of the nanocrystals, with respect to the pyridine-coated, non-assembled sample, except for hydrazine, which caused a substantial blue shift (around 10 nm). This can be explained by the stronger etching behavior of this type of molecule. (b) Photocurrent spectra from annealed films made from the same samples. |
Both films prepared from networks made with hydrazine and with chlorosilane exhibited photocurrents lower than expected in the region below 500 nm. While the origin of this behavior is not entirely clear at present, it is worth pointing out that both hydrazine and chlorosilane treatments are considerably more aggressive for the tetrapods than the acetic acid treatment. Exposure of tetrapods at increasing concentrations of hydrazine or chlorosilane, apart from inducing aggregation of the nanocrystals, were found to lead to severe etching of their surface, along with the collateral nucleation of very small nanocrystals (1–3 nm diameter), most likely arising from re-crystallization of atomic species stripped from the surface of the tetrapods. Eventually, exposure to large concentrations of these aggressive chemicals was found to completely reshape the tetrapods. Much less pronounced effects were seen on the other hand for exposure to increasing concentrations of acetic acid molecules. Hence low photocurrents at high irradiation energies for hydrazine and chlorosilane-treated samples might be related to the presence of small nanocrystals and/or organometallic clusters which might absorb light in this range but might not contribute to transport.
PCE (%) | FF | V oc/V | J sc/mA cm−2 | |
---|---|---|---|---|
Non-assembled | 0.14 | 0.32 | 0.38 | 1.15 |
Networks acetic acid | 0.21 | 0.46 | 0.38 | 1.19 |
Networks hydrazine | 0.07 | 0.43 | 0.40 | 0.38 |
Networks chlorosilane | 0.10 | 0.41 | 0.42 | 0.60 |
Fig. 7 (a) Schematic of the photovoltaic device structure. The active layer consists of a bilayer of either tetrapods or networks, and fullerene (C60). (b) Current–voltage characteristics for photovoltaic cells made with non-assembled pyridine-coated tetrapods, and with networks prepared with acetic acid, hydrazine and chlorosilane, respectively, under illumination of one-sun (AM 1.5 G). |
The device made with networks assembled via acetic acid had a power conversion efficiency (PCE) that was slightly higher than that of the device made with non-assembled tetrapods. The short circuit current (Jsc), which represents the maximum photogenerated current, is similar for these two devices. Also, both devices have the same open circuit voltage. As a consequence, the lower PCE for the device made with non-assembled tetrapods is due to the reduced field factor (FF). The lower FF is related to a higher series resistance of the cell, which arises from the resistance of the materials to current flow.18 On the other hand, the lower PCE for the devices made with networks prepared via hydrazine and chlorosilane results from a lower Jsc rather than a lower FF. This preliminary comparative study on the device performances suggests that the use of networks of tetrapods might help to improve the hole conductivity, i.e. a reduction of the resistance to the holes current flow, which results in higher FF values.
With respect to the device made with non-assembled nanocrystals, the photocurrent was improved slightly only for the device made with networks of tetrapods assembled with acetic acid, while it was actually reduced for devices made with networks assembled with hydrazine or with chlorosilane. The two latter treatments appear to have a negative impact on the charge separation mechanism at the interface with the C60 layer. The better performance of the devices based on non-assembled tetrapods and on the networks prepared with acetic acid was also confirmed in terms of the R0dark/R0light values of the corresponding films (Table 1). These findings could be explained by considering two possible implications connected with our procedures: (1) different effects of the chemical treatment on the tetrapods in the case of acetic acid with respect to chlorosilane and hydrazine; (2) morphology differences (grain size) in the resulting films, which influence charge separation between the tetrapods and C60. Based on previous reports,19 a hydrazine treatment of CdTe films for photovoltaic applications can lead to detrimental results. It can promote for example the formation of a thin insulating layer of CdTeO3, which can affect negatively device efficiency. The chlorosilane treatment, as was originally reported,15 inevitably leaves chlorine atoms on the surface of nanocrystals. In analogy with studies on bulk CdTe,20 chlorine doping (either by Cl2(g) treatment or by CdCl2 deposition) may result in reduced device performance. Preliminary atomic force microscopy (AFM) studies on the morphology of tetrapod films prepared by us have revealed that the average grain size in films of acetic acid treated tetrapods is smaller than in case of hydrazine or chlorosilane treatments (see the ESI for additional details†). In this scenario, acetic acid behaves rather as a short ligand, which replaces in part the molecules bound to the surface of nanocrystals (with preference at the tips), without affecting drastically the local chemical environment of tetrapods.
The advantage of tetrapod networks, for example for hybrid organic–inorganic photovoltaic devices, is on one hand their porous structure that provides large interfacial areas with organic molecules/polymers in order to collect one type of photo-excited charge carriers, while on the other hand the barriers for the collection of the second carrier type are reduced in the all-inorganic networks formed by the tip–tip assembly of the tetrapods, as is demonstrated by the results of our fitting model. Another advantage might be that tetrapods are 3D structures, therefore exploiting tip-to-tip assembly could lead to the highly porous 3D conductive networks or aerogels accessible through solution-based chemistry, while neither dots nor rods would provide this opportunity.
Footnotes |
† Electronic supplementary information (ESI) available: Details on chemicals, syntheses and assembly procedures; estimate of nanocrystal concentrations, average number of surfactants bound to their surface, and ratio of number of additive molecules to bound surfactants; transport characteristics of “as-deposited” non-annealed films; additional SEMs of deposited films; TEMs of networks before and after annealing; AFM characterization of the nanocrystal films used as active layers for the photovoltaic devices. See DOI: 10.1039/c0nr00308e |
‡ Present address: Department of Chemistry, University of Chicago, Chicago, IL 60637, USA |
This journal is © The Royal Society of Chemistry 2010 |