Letter pubs.acs.org/NanoLett
Thermal Doping by Vacancy Formation in Copper Sulfide Nanocrystal Arrays Y. Bekenstein,†,‡,§ K. Vinokurov,†,§ S. Keren-Zur,‡,§ I. Hadar,†,§ Y. Schilt,†,§ U. Raviv,†,§ O. Millo,*,‡,§ and U. Banin*,†,§ †
Institute of Chemistry, ‡Racah Institute of Physics, and §Center for Nanoscience and Nanotechnology, the Hebrew University of Jerusalem, Jerusalem 91904, Israel S Supporting Information *
ABSTRACT: A new approach for doping of Cu2S nanocrystal arrays using thermal treatment at moderate temperatures (T < 400 K) is presented. This thermal doping process yields conductance enhancement by 6 orders of magnitude. Local probe measurements prove this doping is an intraparticle effect and, moreover, tunneling spectroscopy data signify p-type doping. The doping mechanism is attributed to Cu vacancy formation, resulting in free holes. Thermal-doping temperature dependence exhibits an Arrhenius-like behavior, providing the vacancy formation energy of 1.6 eV. The moderate temperature conditions for thermal doping unique to these nanocrystals allow patterned doping of nanocrystal films through local heating by a focused laser beam, toward fabrication of nanocrystal-based electronic devices. KEYWORDS: Doping semiconducting nanocrystals, vacancy formation, nanocrystal arrays, patterned doping, scanning tunneling spectroscopy, copper sulfide nanocrystals
S
controlled manner. p-type doping was identified by scanning tunneling spectroscopy (STS) of single Cu2S NCs and is attributed to formation of Cu lattice vacancies within the NCs. Unlike the previous methods for aliovalent doping, which introduce either substitutional or interstitial impurities, in the thermal doping approach no additional impurities are introduced to the NC; rather the opposite process takes place, the expulsion of atoms from the intrinsic material. Furthermore, laser-induced heating was used to conduct the thermal doping process, opening the path to patterning via use of local heating by a focused laser beam in the far- or near-field. Highly monodisperse, faceted 14 nm Cu2S NC passivated by dodecanethiol were synthesized by a high-temperature reaction (details in section S1 of the Supporting Information, absorption spectra, X-ray diffraction, thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) curves are shown in Supporting Information Figures S1−S3 respectively).14 The NCs were deposited from solution using a controlled slow solvent evaporation method on Si/SiO2 substrates with prepatterned electrodes (Cr/Au, 2 μm spacing) forming ∼150 nm thick films (Figure 1a). Tuning the NCs organic ligands layer concentration enables us, through self-assembly, to achieve also highly ordered superlattices as depicted in Figure 1B, and verified by distinct peaks in the small-angle X-ray scattering (SAXS) data (Figure 1c). NCs were kept under inert
emiconductor nanocrystals (NCs) manifest size dependent tunable properties and can be deposited on a variety of substrates by facile wet-chemical methods. Consequently, NCs are prospective materials in diverse printed electronics applications including transistors,1 functional circuits,2 photovoltaics,3 and light-emitting diodes.4 These applications require tuning of the electrical properties of the NC film to achieve efficient charge transport, along with patterning capabilities to construct functional circuits. Commonly, in bulk semiconductors lithographically patterned doping is ubiquitously used to fabricate complex microelectronic circuitry. In the case of semiconductor NC films, such control is still lacking. There is still a gap in methods of effective aliovalent doping of NCs. A first approach was to introduce excess charge carriers into the NCs via chemically altering their surfaces, through the use of suitable electron donating coating ligands that yield n-type doping characteristics.1,5,6 Another approach is to introduce impurities into the NC during synthesis, or postsynthesis, achieving both n and p-type characteristics.7−11 NC films were also doped by deposition of In electrodes followed by heating to induce In impurity diffusion,12 as well as by control over the NCs stoichiometry via evaporation of thin layers.13 However, additional approaches to NC doping are required especially because patterned doping is not obvious by the previous methods. Here we present a novel approach to doping of semiconductor NC films termed “thermal doping”. This approach is demonstrated for Cu2S NC arrays where moderate temperature treatment leads to significant conductance enhancement in a © 2014 American Chemical Society
Received: November 24, 2013 Revised: February 18, 2014 Published: February 24, 2014 1349
dx.doi.org/10.1021/nl4043642 | Nano Lett. 2014, 14, 1349−1353
Nano Letters
Letter
Figure 1. Cu2S NC array thermal doping effect. (a) High-resolution scanning electron micrograph (HRSEM) of NCs array between electrodes (arrows) (scheme in d inset); scale bar 200 nm. (b) HRSEM of self-assembled highly ordered superlattices a few micrometers in size; scale bar 150 nm. SAXS raw (b inset) and azimuthally integrated data (c) measured on NC arrays showing peaks manifesting superlattice ordering. The oriented 2D pattern presents features of hexagonal order, which is confirmed by the relative positions of the harmonics. Scattering curves prior (blue curve) and after (red curve) thermal doping are compared, indicating a slight decrease, by 0.14 nm, in the inter-NC distance. (d) Film conductance at 1 V versus temperature exhibits minor temperature dependence below 300 K (blue curve). At elevated temperatures (350−400 K), film conductance increases significantly. Error bars depict the conductance changes at these temperatures with annealing time (Supporting Information Figure S9). (e) Plots of Id versus Vsd at 20 K (blue curve) and after the thermal doping process at 380 K (red curve) showing increase in conductance by ∼6 orders of magnitude. Both data sets (d,e) were acquired on the disordered arrays. Cu2S NCs were 14 nm in diameter with hexagonal truncated biprism geometry, as depicted in the HRTEM image and cartoon presented in (d inset).
organic ligands, thus reducing the inter-NC spacing.17−19 This annealing, however, was achieved at significantly higher temperatures, ∼500 K, typically leading to removal of capping organic ligands and reduction of interparticle spacing by ∼0.5 nm and in some cases even to sintering.20 Here, the irreversible remarkable increase in conductance occurs at moderate treatment temperatures and may originate both from an inter-NC effect, of reducing the NC separation upon heating, and an intra-NC effect leading to enhanced number of charge carriers, namely NC doping. We will first estimate the former effect, showing that this effect is the minor one. To directly assess the change in inter-NC spacing, SAXS was used. Figure 1c depicts scattering intensity versus the magnitude of the reciprocal lattice vector, q, before and after the thermal treatment. The SAXS data indicate only a slight reduction by 0.14 nm in inter-NC spacing upon thermal treatment at 380 K. Therefore, unlike the prior work, here the increased conductance after heating cannot be explained solely nor mainly by the decrease in interparticle separation. This conclusion is supported by an estimate of the tunneling probability of charge carriers between adjacent NCs. When the tunneling barrier width is decreased by Δd, the inter-NC tunneling probability increases by ∝ e2·κ·Δd, where in our case κ−1 is 0.97 Å, similar to the tunneling decay value through saturated alkane chains21 (section S2 in the Supporting Information). For the measured Δd = 0.14 nm, the tunneling
environment, as exposure to oxygen may also induce copper vacancies in a competing process of oxidation where a plasmon band was observed due to free carriers.15,16 Drain-current (Id) measurements were conducted under high vacuum in a variable temperature probe-station. Source-drain bias voltages (Vsd) of 1−10 V were used, yielding initially very low conductance values of ∼1−10 pS at room temperature (RT). The film was heated and conductance (G) measurements between 340 and 400 K were preformed (Figure 1d). Remarkably, films treated at 380−400 K showed 4−6 orders of magnitude increase in G. Figure 1e depicts I−V curves demonstrating current enhancement at 10 V from 6.4 pA at 20 K (blue curve) to 30 μA at 380 K (red curve). The rise in film conductance is irreversible; cooling the film does not reduce G back to its original values, as depicted in Figure 2a showing G(T) curves measured for the NC films after treatment at increasing temperatures between 350 and 390 K, exhibiting exponential increase with temperature. Figure 2b portrays the logarithm of conductance versus the inverse of the treatment temperature for three different samples measured at RT and also for one sample at various measurement temperatures. All data sets exhibit Arrhenius-like dependence for the thermal process with a common slope corresponding to an energy of ∼1.6 eV. High-temperature annealing was previously applied to NC films for increasing their conductance, by removing capping 1350
dx.doi.org/10.1021/nl4043642 | Nano Lett. 2014, 14, 1349−1353
Nano Letters
Letter
reveals approximately 2 orders of magnitude increase in the single NC conductance before and after the thermal treatment (Figure 3b). Statistical analyses for hundreds of thermally treated single NCs corroborate these results (Supporting Information Figure S8). The significant increase of conductance at the single NC level shows that the thermal treatment process induces an intraparticle modification leading to the introduction of charge carriers. For further understanding of these intraparticle changes, STM and STS were used. Dilute samples of Cu2S NCs were deposited on atomically flat Au surfaces, where isolated NC can be measured. The STM tip was positioned above a single Cu2S NC realizing a double barrier tunnel junction configuration, and I−V spectra were acquired. Figure 3c portrays typical dI/ dV−V tunneling spectra, measured at 4.2 K before (blue line) and after (red and black lines) thermal treatment. The dI/dV− V spectra are proportional to the local density of states. The spectrum for prethermal treated NCs depicts typical semiconductor density of states with a clear band gap.22 After the thermal treatment, a finite density of states inside the band gap is identified. Most notably, a distinct offset of the band-edges toward higher energies is observed. This offset is demonstrated by statistical analysis of the band-edge shifts (Figure 3d; Supporting Information Figure S4 shows additional STS curves), where in the thermal treated NCs (red bars), a statistically significant increase of the band offsets toward positive values is evident, compared with the pretreated NCs (blue bars). This analysis indicates a shift of the Fermi energy toward the valence band, akin to p-doped semiconductors.7,23 p-type behavior of Cu2S is known in the bulk and attributed to Cu vacancies, which form due to the low chemical potential of Cu(0).24 This behavior results in a series of Cu depleted substances with a rich phase diagram, for example, the near stochiometric Djurleite phase (Cu1.96S), resulting in free hole charge carriers.25 In bulk Cu2S thin films, thermal annealing at 430 K increased the film conductance only by a factor of 2.26 The vacancy formation is therefore significantly enhanced in Cu2S NCs due to facile diffusion of Cu atoms to the surface, directly related to a size effect. Given the significant change in conductance and the related p-type behavior induced after the thermal treatment, we coin the term “thermal doping” for this process. We conjecture a simplified model for the hole conductance G in the Cu2S NCs array and its dependence on the thermal doping temperature Ttd
Figure 2. Temperature dependence of conductance in the thermal doping process. (a) G(T) curves after the thermal treatment at various measurement temperatures. Repeated cooling (triangles) and heating (circles) indicate no hysteresis effect below ∼350 K, demonstrating the irreversibility of the conductance increase. (b) Arrhenius-like plot of the natural logarithms of the conductance versus the inverse of the thermal doping temperature for three different samples measured at room temperature (squares, circles, triangles) and for one sample at various measurement temperatures (blue, 300 K, magenta, 200 K; red, 100 K). The slope of the fitted lines yields the Cu vacancy formation energy ενf ≈ 1.6 eV.
probability increases only by a factor of ∼20. Moreover, the use of relatively low temperatures does not lead to chemical changes of the ligands layer (Supporting Information Figure S3). The decrease in interparticle separation is therefore assigned to residual solvent removal accompanied by enhanced packing and reordering of the interpenetrating alkane chains on the NC surfaces. High-resolution transmission electron microscopy (HRTEM) and SAXS measurements of the film prior and post to the thermal doping process showed no evidence for fusing between neighboring NCs, which in the present case are coated by strongly bound dodecanethiol ligands thus NCs sintering is unlikely (Supporting Information Figure S6). Therefore, the inter-NC changes upon thermal treatment can account only for a small fraction of the observed 4−6 orders of magnitude conductance enhancement. We next consider the contribution of intraparticle effects to the remarkable increase in conductance upon thermal treatment of the NC films. To this end, a study of the electronic properties at the single particle level was conducted by local probe measurements using both conductance atomic force microscopy (C-AFM) and scanning tunneling microscopy (STM). Local conductance of single NCs were measured via C-AFM while applying 3 V bias between the conducting tip and a gold surface on which NCs were deposited, yielding selfassembled NC monolayers (Figures 3, Supporting Information Figure S7). Figure 3a exhibits a topography and current map of a typical array. Cross-sectional analysis of the current map
G(Ttd) ∝ e−εvf / kBTtd ·e−2κd(Ttd)
(1)
where the first exponential expression represents the intraparticle term of copper vacancies formation with a “formation energy” ενf (kB is the Boltzmann constant).27 The second term is the aforementioned interparticle probability for electron tunneling between neighboring NCs, which contributes to increased conductance upon the thermal treatment by only a factor of ∼20, and therefore can be neglected leading to the approximate dependence d ln G d
( ) 1 Ttd
≅−
εvf kB
(2)
This relation is indeed fully consistent with the observed Arrhenius-like activation in Figure 2b, allowing to derive the Cu vacancy formation energy, ενf ∼ 1.6 eV. Using this value, we 1351
dx.doi.org/10.1021/nl4043642 | Nano Lett. 2014, 14, 1349−1353
Nano Letters
Letter
Figure 3. Single NC thermal doping effect. (a,b) C-AFM measurements (scheme in b inset) performed on an ordered monolayer of Cu2S NCs deposited on annealed Au substrate. (a inset) AFM topographic (left) and current (right) micrographs before thermal doping, measured with 3 V tip−sample bias voltage, scale bar 50 nm. Cross-section along the line drawn in the current image is plotted before (a and b blue curve), and after thermal doping at 380 K (b red curve), showing 2 orders of magnitude local enhancement. (c inset) STM current map of a single Cu2S NC, scale bar 4 nm. (c) dI/dV−V tunneling spectra, measured at 4.2 K before (blue line) and after (red) thermal doping (more curves in Supporting Information Figure S4). (d) Histograms of relative band-offset values measured on 24 single NCs, comparing the thermally doped (red) and predoped (blue) populations. An increase in the relative band offset toward positive values is observed after thermal doping akin to p-type doping.
estimate an increase by a factor of ∼5 × 104 in the hole density following the thermal doping process at 380 K, compared with the as-prepared NC array. This increase translates to a ∼280 meV shift of the Fermi energy toward the valence band, similar to the values observed in the STS measurements on single NCs (see Section S4 in Supporting Information for details of this derivation). Cu2S NCs are synthesized in a method that yields minimal copper vacancy content and are highly stoichiometric, as supported by the lack of optical surface plasmon (Supporting Information Figure S2) and by variable temperature powder Xray diffraction (XRD) measurements (Supporting Information Figure S1). The XRD data is consistent with Cu depletion from the initial stoichiometric Cu2S low chalcocite phase, toward a Cu depleted phase. In a conservative estimate of a single vacancy (hole) per 1000 as synthesized NCs, and assuming a free electron model with linear relation between conductance and vacancy (charge carrier) concentration, we extract ∼50 vacancies (holes) per a 14 nm thermally doped NC. Unfortunately, XPS measurements cannot distinguish between the relevant oxidation states, Cu(0) and Cu(1), due to nearly identical corresponding peak positions (Supporting Information Figure S5). Thermal doping of semiconductor NC films has diverse applications in bottom-up solution based preparation of electronic devices. Moreover, the ability to locally dope specific regions of the film while maintaining different conductance nearby, may serve as a patterning method for profile doping. We implemented local Cu vacancy formation via thermal doping by using a focused laser as the heat source (Figure 4a, Supporting Information Figure S10). After exposure to laser intensity of ∼2 × 105 W/cm2, the film conductance irreversibly
increased by 6 orders of magnitude, similar to the thermal doping process discussed above. To further characterize the laser-induced thermal doping process, NC arrays were assembled on an indium-tin-oxide (ITO) film. Our AFM is equipped with inline optics which enables us to correlate the AFM scan with the optical axis used for laser illumination. The samples were thermally heated via laser illumination (10 mW, 532 nm) after which the illuminated area was characterized with the AFM. Both topography and Kelvin probe microscopy (KPM) data were measured using a dual pass technique (more details are provided in the methods section of the Supporting Information). Figure 4b,d depicts 3D maps of the potential and topography of the thermally doped area. The line cross-section taken along the potential map (Figure 4b) presented in Figure 4c depicts a ∼80 mV change in the potential of the heated NC array (blue curve), while no change can be detected in the corresponding topography scan (black curve). The increase of the surface potential in the illuminated region is consistent with p-type doping and the concomitant increase of hole concentration in the NCs at this region. Our KPM data thus indicate that also in the case of laser annealing the conductance enhancement is predominantly due to intraparticle p-doping. Moreover, using a controlled movement of the sample stage we were able to generate patterned doping, which was characterized with the KPM technique, as portrayed in Figure 4e. This demonstrates the feasibility of local thermal doping via laser irradiation. Cu2S is a binary copper chalcogenide and is a mother compound for a family of other semiconductors, for example, CuInS2. Our findings open the path to fabrication of diverse electronic, optoelectronic, and solar cell devices using the thermal doping approach, which is also applicable for these materials. 1352
dx.doi.org/10.1021/nl4043642 | Nano Lett. 2014, 14, 1349−1353
Nano Letters
Letter
Memorial Chair. K.V. thanks Israel’s ministry of science and technology. We acknowledge the use of the facilities of the unit for nanocharacterization at the Hebrew University and thank Dr. Inna Popov for help in sample characterization, and Dr. Vitaly Gutkin for assistance in the XPS measurements.
■
Figure 4. Optical local thermal doping. (a) Normalized film conductance versus temperature before (blue) and after irradiation by laser powers of 1, 5, and 7 mW. G was normalized by the respective G0 at 300 K of the as-prepared film. Irradiation by 7 mW increases the film conductance by 6 orders of magnitude. (a inset) Optical micrograph showing superlattices deposited on microelectrodes; red circle marks the local laser irradiated thermally doped region, scale bar is 8 μm. (b) KPM potential and (d) topography maps of laser thermally doped NC array. Scan area is 6 μm2. (c) Line cross sections taken along the lines marked in panels b and d. The potential crosssection (blue curve) depicts a ∼80 mV increase with respect to the background, whereas the topography cross-section (black curve) shows that no change was inflicted to the array morphology due to laser irradiation. (e) KPM potential map depicting pattered doping achieved via optical thermal doping and controlled movement of the substrate stage, scale bar is 5 μm.
■
ASSOCIATED CONTENT
S Supporting Information *
Sections S1−S5 include Materials and Methods, additional calculations, and discussions. Figures S1−S10 include additional absorbance spectra, TGA, DSC, XRD, XPS, TEM, AFM, STS, optical thermal doping setup scheme and additional measurements. Additional references 1−7. This material is available free of charge via the Internet at http://pubs.acs.org.
■
REFERENCES
(1) Talapin, D. V; Murray, C. B. Science 2005, 310, 86−9. (2) Kim, D. K.; Lai, Y.; Diroll, B. T.; Murray, C. B.; Kagan, C. R.; Cherie, R. Nat. Commun. 2012, 3, 1216. (3) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462−465. (4) Bawendi, M.; Coe, S.; Woo, W.-K.; Bulović, V. Nature 2002, 420, 3−6. (5) Wang, C.; Shim, M.; Guyot-Sionnest, P. Science 2001, 291, 2390−2. (6) Yu, D.; Wang, C.; Guyot-Sionnest, P. Science 2003, 300, 1277− 80. (7) Mocatta, D.; Cohen, G.; Schattner, J.; Millo, O.; Rabani, E.; Banin, U. Science 2011, 332, 77−81. (8) Xie, R.; Peng, X. J. Am. Chem. Soc. 2009, 131, 10645−51. (9) Tuinenga, C.; Jasinski, J.; Iwamoto, T.; Chikan, V. ACS Nano 2008, 2, 1411−21. (10) Sahu, A.; Kang, M. S.; Kompch, A.; Notthoff, C.; Wills, A. W.; Deng, D.; Winterer, M.; Frisbie, C. D.; Norris, D. J. Nano Lett. 2012, 12, 2587−94. (11) Roy, S.; Tuinenga, C.; Fungura, F.; Dagtepe, P.; Chikan, V.; Jasinski, J. J. Phys. Chem. C 2009, 113, 13008−13015. (12) Choi, J.-H.; Fafarman, A. T.; Oh, S. J.; Ko, D.-K.; Kim, D. K.; Diroll, B. T.; Muramoto, S.; Gillen, J. G.; Murray, C. B.; Kagan, C. R. Nano Lett. 2012, 12, 2631−8. (13) Oh, S. J.; Berry, N. E.; Choi, J.-H.; Gaulding, E. A.; Paik, T.; Hong, S.-H.; Murray, C. B.; Kagan, C. R. ACS Nano 2013, 7, 2413−21. (14) Han, W.; Yi, L. X.; Zhao, N.; Tang, A. W.; Gao, M. Y.; Tang, Z. Y. J. Am. Chem. Soc. 2008, 130, 13152−13161. (15) Luther, J. M.; Jain, P. K.; Ewers, T.; Alivisatos, A. P. Nat. Mater. 2011, 10, 361−366. (16) Scotognella, F.; Della Valle, G.; Srimath Kandada, A. R.; Dorfs, D.; Zavelani-Rossi, M.; Conforti, M.; Miszta, K.; Comin, A.; Korobchevskaya, K.; Lanzani, G.; Manna, L.; Tassone, F. Nano Lett. 2011, 11, 4711−7. (17) Drndic, M.; Jarosz, M. V.; Morgan, N. Y.; Kastner, M. a.; Bawendi, M. G.; Introduction, I.; Drndić, M. J. Appl. Phys. 2002, 92, 7498. (18) Lee, J.-S.; Kovalenko, M. V; Huang, J.; Chung, D. S.; Talapin, D. V Nat. Nanotechnol. 2011, 6, 348−52. (19) Fischbein, M. D.; Puster, M.; Drndic, M. Nano Lett. 2010, 10, 2155−61. (20) Law, M.; Luther, J. M.; Song, Q.; Hughes, B. K.; Perkins, C. L.; Nozik, A. J. J. Am. Chem. Soc. 2008, 130, 5974−85. (21) Salomon, A.; Cahen, D.; Lindsay, S.; Tomfohr, J.; Engelkes, V. B.; Frisbie, C. D. Adv. Mater. 2003, 15, 1881−1890. (22) Bekenstein, Y.; Vinokurov, K.; Levy, T. J.; Rabani, E.; Banin, U.; Millo, O. Phys. Rev. B 2012, 86, 85431. (23) Wolf, O.; Dasog, M.; Yang, Z.; Balberg, I.; Veinot, J. G. C.; Millo, O. Nano Lett. 2013, 13, 2516−21. (24) Sands, T. D.; Washburn, J.; Gronsky, R. Phys. Status Solidi A 1982, 72, 551−559. (25) Evans, H. T. Science 1979, 203, 356−358. (26) Grozdanov, I.; Najdoski, M. J. Solid State Chem. 1995, 114, 469−475. (27) Ashcroft, N.; Mermin, D. Solid State physics; Thomson Learning: Boston, MA, 1976.
AUTHOR INFORMATION
Corresponding Authors
*E-mail: (U.B.)
[email protected]. *E-mail: (O.M.)
[email protected]. Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC Grant Agreement [246841] (U.B.), and the ISF [478/09] (O.M.). O.M. acknowledges support from the Harry de Jur Chair in Applied Science. U.B. thanks the Alfred and Erica Larisch 1353
dx.doi.org/10.1021/nl4043642 | Nano Lett. 2014, 14, 1349−1353
