Article pubs.acs.org/cm
Nature of the Infrared Transition of Colloidal Indium Nitride Nanocrystals: Nonparabolicity Effects on the Plasmonic Behavior of Doped Semiconductor Nanomaterials Zhihui Liu and Rémi Beaulac* Department of Chemistry, Michigan State University, East Lansing, Michigan 48824-1322, United States S Supporting Information *
ABSTRACT: As-synthesized colloidal indium nitride (InN) nanocrystals are degenerately doped with carrier densities large enough to lead to strong localized surface plasmon resonances (LSPR) in the infrared. Intriguingly, the LSPR energy is almost independent of carrier density, which premises that simple classical models that are often used to describe metallic systems inadequately describe the plasmonic response of InN nanoparticles. Here, an oxidative titration approach is used to directly quantify carrier densities in colloidal InN nanocrystals, eliminating the need to rely on any specific model. A sizeindependent carrier density value of (7.4 ± 0.4) × 1020 cm−3 is obtained for diameters varying between 4 and 9 nm, corresponding to about 30 to 300 electrons per nanocrystal, depending on size. Upon oxidation with nitrosonium salts, the carrier density in InN nanocrystals can be reduced to (3.9 ± 0.3) × 1020 cm−3, also independent of size. The unusual plasmonic signatures of colloidal InN nanocrystals are shown to arise from the nonparabolicity of the conduction band dispersion, which explains the nearly invariant LSPR energy as a function of carrier density, as well as the size dependence of the LSPR energy.
■
INTRODUCTION Metals and doped semiconductors are characterized by strongly delocalized electronic wave functions that allow charge carriers to be easily transported over large distances in response to external electric fields. At the high charge carrier densities that characterize the metallic state (Ne ∼ 1022 cm−3),1,2 the motion of these delocalized charge carriers becomes sufficiently correlated to induce the formation of plasmons, collective (and quantized) oscillations of the charge density.1,3 Although metals cannot sustain electric fields in the bulk, strong plasmon−photon coupling effects, known as surface plasmon resonances, arise when charge carriers close to the surface interact resonantly with impinging electromagnetic fields of the proper frequency. In metallic nanostructures, the motion of charge carriers is necessarily restricted in space, giving rise to localized surface plasmon resonances (LSPR).3−5 In practice, the energy of LSPR can be concentrated locally into large electric fields that allow surface-enhanced Raman spectroscopy (SERS),6,7 delivered locally as heat,8 or used to generate electrochemical work, effects that are generally collectively combined under the umbrella term of plasmonics.3,9 While metallic nanomaterials largely dominate plasmonics studies, doped semiconductor nanomaterials are now recognized as legitimate candidates for similar applications,10−21 with the consequential advantage that many of the fundamental parameters that modulate plasmonic properties (such as carrier density, dielectric constant, and effective mass) can be directly and, at least in principle, predictably tuned by straightforward © 2017 American Chemical Society
chemical modifications (“doping”) of the materials. Whereas any semiconductor material can in principle serve such a purpose, in practice most studies in this area have involved stable transparent oxide nanomaterials10,14,18,19,21 such as indium oxide (In2O3) and zinc oxide (ZnO), which yield airstable n-type nanomaterials when doped with specific donor impurities (for instance Sn4+ for In2O3 and Al3+ for ZnO).22 Group-III nitride nanomaterials are also interesting alternatives for plasmonics applications but have so far remained unexploited in that respect, likely due to challenges associated with their synthesis in the colloidal form.23 Indium nitride (InN) is a particularly interesting candidate in regard to its strong tendency to accumulate electrons due to its unusually low-energy conduction band (CB).24−26 Furthermore, whereas the dielectric constant of InN matches closely those found typically in oxides,26,27 stemming from similar lattice ionicities, InN is a narrow, direct bandgap material (Eg ∼ 0.7 eV),26,28−30 starkly contrasting with most oxides, which generally possess very large bandgaps. As shown in Figure 1 and previously discussed in the literature,31,32 colloidal InN nanocrystals (NCs) are indeed characterized by strong infrared (IR) transitions and Burstein− Moss shifts of the interband transitions, two archetypical signatures of degenerately doped semiconductors. The exact Received: June 20, 2017 Revised: July 25, 2017 Published: August 2, 2017 7507
DOI: 10.1021/acs.chemmater.7b02545 Chem. Mater. 2017, 29, 7507−7514
Article
Chemistry of Materials
octadecene in a 50 mL three-neck round-bottom flask with a condenser, rubber-sealed caps, and a stir bar. The mixture was stirred and heated on a heating mantle to the desired temperature for a certain amount of time to control the size of InN NCs as listed in Table S1 and cooled to room temperature. Postsynthesis processing began with separating ODE from the resulting black solid by centrifugation at 4000 rpm for 10 min. The black solid was then sonicated with 20 mL of toluene for 3 min and precipitated by centrifugation (4000 rpm, 10 min) followed by sonicating in 20 mL of ethanol and precipitating by centrifugation, twice. The resulting blackbrown solid was sonicated with nitric acid (∼3.5 vol % aqueous, 12 mL) for 5 min before centrifugation (4000 rpm, 10 min) followed by washing with 20 mL of ethanol to wash away residual acid. Twelve milliliters of OLA was added to the mixture, sonicated for 30 min, and precipitated with 12 mL of ethanol (4000 rpm, 10 min), and then the precipitate was suspended in 16 mL of toluene. The suspension was centrifuged (1000 rpm, 5 min) to separate any unreacted In2S3, followed by the addition of 16 ml of ethanol to the suspension and another centrifugation cycle (4000 rpm, 10 min). The majority of the resulting black solid was suspended in 15 mL of hexane, while a small amount was kept in 2 mL of toluene for TEM. The ligand exchange process was performed following a published procedure.33 A solution of Et3OBF4 was prepared by dissolving ∼1 g of Et3OBF4 in 15 mL of ACN. The Et3OBF4 solution was added to the suspension of InN NCs in hexane. The mixture was shaken for ∼1 min before adding 2 mL of toluene. The mixture was centrifuged at 4000 rpm for 10 min and the resulting solid was suspended in 16 mL of ACN and precipitated again with 4 mL of toluene. Then the solid was suspended in 15−20 mL of ACN and stored in the glovebox for characterization. Transmission electron microscopy (TEM) images were recorded on a JEOL 2200 FS microscope operating at 200 keV. Formvar-coated copper grids (Ted Pella, 01824) were used as nanocrystal supports for TEM. InN−OLA toluene suspensions were drop-casted on the TEM grid prior to the measurements. Images were analyzed using ImageJ. Powder X-ray diffractometry (XRD) was performed on a Bruker D8 DaVinci diffractometer equipped with a Cu Kα radiation (0.15406 nm) line source operating at 40 kV and 40 mA. Samples were placed in a PVMA sample holder with zero-background silica plates by dropcasting concentrated nanocrystal suspensions. The sample was scanned from 25° to 70° using 10 s/scan. Diffraction lines are compared to the JCPDS card number: 00-050-1239. Absorption spectra of NOBF4-treated NCs were obtained by adding a solution of NOBF4 in ACN to a solution of Et3OBF4-treated NCs in ACN in a 1 cm quartz cuvette with a septum-sealed screw-cap (Starna Cells, 1-Q-10-GL14-C) in the glovebox and allowed to stand for 1 h before measurement. Ultraviolet/visible/near-infrared (UV−vis-NIR) absorption spectra were measured on a OLIS 17 spectrometer from 400 to 2700 nm. Fourier transform infrared (FTIR) spectra were obtained from a Mattson Galaxy 3020 from 4000 to 400 cm−1 at a resolution of 4 cm−1 for 16 scans. Background spectra were measured on a clean silicon window (Edmund Optics, no. 68-523). Films of NOBF4-treated NCs were obtained from the same suspensions used for the UV−vis−NIR measurements. The NOBF4/NCs mixtures were shaken, allowed to stand for 1 h in the glovebox, and then drop-cast on the same silicon window. The UV−vis−NIR and FTIR spectra for the same sample were connected by matching the absorbance in the overlapping region (4000−3704 cm−1). Inductively coupled plasma−atomic emission spectroscopy (ICPAES) was used to calculate the concentration of the suspension of InN NCs. A 50 μL amount of InN NCs in ACN was dried under vacuum, and the sample was calcined in air at 500 °C for 2 h, digested in 2 mL of aqua regia (1.5 mL of HCl and 0.5 mL of HNO3), and then diluted to 100 mL in volumetric flask with deionized water. The concentration of In was determined by the method of standard additions. A 10 mL sample solution was added to each 25 mL volumetric flask to make five standard solutions of 0.1, 0.5, 1, 2, and 5 ppm. X-ray photoelectron spectroscopy (XPS) was performed on a PerkinElmer Phi 5600 ESCA system, with a Mg Kα X-ray source at a
Figure 1. Spectroscopic signatures of colloidal InN NCs at various oxidation levels. (a) Absorbance spectra of InN NCs (d = 4.2 nm, [NC] = 1.8 μM), with various amounts of NOBF4 oxidant added (0− 30 equiv per NC). The arrows indicate the direction of the shifts upon oxidation. The sharp features between 3800 and 4500 cm−1 are due to solvent overtone modes. All absorption spectra have been corrected for dilution effects (see Supporting Information (SI) for further details). (b) Differential absorption spectra, ΔA = Aoxidized − A0.
origin of the donor defects responsible for the native doping of InN NCs is still unknown but reflects the instability of intrinsic InN and the strong tendency of native defects to be donors, consistent with the behavior of the bulk material.24,29 Palomaki et al. have previously shown that the optical features shown in Figure 1 can be directly and reversibly modulated by chemical redox processes, convincingly demonstrating the presence of a large number of excess carriers in the conduction band (CB) of InN NCs.32 Although no direct measurement of the carrier densities could be made at the time, indirect estimates using a simple classical model (further described below) suggested densities on the order of 1020 cm−3, consistent with the assignment of the IR band as a LSPR.32 The purpose of this contribution is 3-fold: (1) demonstrate the inability of the simplest classical (Drude) model to properly describe the plasmonic behavior of InN and consequently to correctly quantify carrier densities; (2) provide a direct (i.e., model-free) quantification of the carrier density of InN NCs and describe the size dependence of their plasmonic behavior; (3) show that the small bandgap of InN directly impacts the behavior of the delocalized, but confined, charge carriers, suggesting that the plasmonic behavior of InN NCs directly reflects the nonparabolic dispersion of the CB.
■
EXPERIMENTAL SECTION
Indium sulfide (In2S3, Alfa Aesar, stored in N2 glovebox), sodium amide (NaNH2, Alfa Aesar, stored in glovebox), Et3OBF4 (Fluka, ≥97%, stored in glovebox), NOBF4 (Fluka, ≥98%, stored in glovebox), oleylamine (OLA, Sigma-Aldrich, 98%), anhydrous acetonitrile (ACN, purified by running through alumina drying column, stored in a glass bottle with 3 Å molecular sieve in glovebox). Octadecene (ODE, Sigma-Aldrich, technical grade, 90%) was heated overnight at 100 °C under vacuum, ∼50 mTorr, and stored in the glovebox. Toluene was purified by an alumina drying column. Ethanol (Macron, absolute) and hexane (Macron) were used as received. Indium nitride NCs were prepared from In2S3 and NaNH2 at ambient pressure by adapting previously published methods.31,32 The method was modified as follows. Under nitrogen atmosphere, 0.5 mmol of In2S3 was combined with 5 mmol of NaNH2 in 10 mL of 7508
DOI: 10.1021/acs.chemmater.7b02545 Chem. Mater. 2017, 29, 7507−7514
Article
Chemistry of Materials take-off angle of 45°, under ultrahigh vacuum conditions (
