Bismuth Doping of Germanium Nanocrystals ... - ACS Publications

4 downloads 0 Views 2MB Size Report
Bismuth Doping of Germanium Nanocrystals through Colloi- dal Chemistry. Katayoun Tabatabaei,a Haipeng Lu,b Bradley M. Nolan,a Xi Cen,c Cliff E. McCold,c ...
Article pubs.acs.org/cm

Bismuth Doping of Germanium Nanocrystals through Colloidal Chemistry Katayoun Tabatabaei,† Haipeng Lu,‡ Bradley M. Nolan,† Xi Cen,§ Cliff E. McCold,§ Xinming Zhang,§ Richard L. Brutchey,‡ Klaus van Benthem,§ Joshua Hihath,∥ and Susan M. Kauzlarich*,† †

Department of Chemistry, University of California, Davis, One Shields Avenue, Davis, California 95616, United States Department of Chemistry, University of Southern California, Los Angeles, California 90089, United States § Department of Materials Science and Engineering, University of California, Davis, One Shields Avenue, Davis, California 95616, United States ∥ Department of Electrical and Computer Engineering, University of California, Davis, One Shields Avenue, Davis, California 95616, United States ‡

S Supporting Information *

ABSTRACT: Nanogermanium is a material that has great potential for technological applications, and doped and alloyed Ge nanocrystals (NCs) are actively being considered. New alloys and compositions are possible in colloidal synthesis because the reactions are kinetically rather than thermodynamically controlled. Most of the Group V elements have been shown to be n-type dopants in Ge to increase carrier concentration; however, thermodynamically, Bi shows no solubility in crystalline Ge. Bi-doped Ge NCs were synthesized for the first time in a microwave-assisted solution route. The oleylamine capping ligand can be replaced by dodecanethiol without loss of Bi. A positive correlation between the lattice parameter and the concentration of Bi content (0.5−2.0 mol %) is shown via powder X-ray diffraction and selected area electron diffraction. X-ray photoelectron spectroscopy, transmission electron microscopy (TEM), scanning TEM, and inductively coupled plasma−mass spectroscopy are consistent with the Bi solubility up to 2 mol %. The NC size increases with increasing amount of bismuth iodide employed in the reaction. Absorption data show that the band gap of the Bidoped Ge NCs is consistent with the NC size. This work shows that a new element can be doped into Ge NCs via a microwaveassisted route in amounts as high as 1−2 mol %, which leads to increased carriers. Colloidal chemistry provides an inroad to new materials not accessible via other means.

1. INTRODUCTION Ge as a nanomaterial is being investigated extensively because of its suitable band gap (0.67 eV) and high carrier mobilities (μe = 3900 cm−2 V−1 S−1, μh = 1900 cm−2 V−1 S−1), leading to good transport characteristics.1 In addition, its high absorption coefficient (ca. 2 × 105 cm−1) and large exciton Bohr radius (24 nm)2 make it a potential candidate in nanoform for optoelectronic,1,3−5 energy conversion,6,7 bioimaging,8 photodetectors,9 chemical sensing,10 and lithium-ion batteries.11 Optical and electronic properties of Ge semiconductor nanocrystals (NCs) can be well-tuned by versatile approaches such as controlling their size, morphology, surface chemistry, and composition.12 Incorporation of a very small number of dopant atoms in the semiconductor NCs can be more complicated than theoretically anticipated based on their bulk counterparts due to a finite lattice within the host NC13 and can dramatically alter their electronic band structure providing them with great potential for applications in microelectronics, integrated optoelectronics, photovoltaics and bioimaging.14−20 Electrical control of Ge NCs by chemical doping is a less© XXXX American Chemical Society

explored area, and incorporation of low-level quantities (∼1 mol %) of the main group elements Sn, Sb, P, Al, Ga, and In to yield alloyed Ge NCs has been reported. The authors referred to them as alloyed rather than doped due to the absence of free charge carriers via electron paramagnetic resonance (EPR) spectroscopy.21 In this work, we focus on Bi-doped Ge NCs because it has been shown that Ge1−xBix can be prepared as an amorphous film with xmax ∼ 0.3222,23 and that Bi doping increases the ntype conductivity, resulting in potential IR and THz broadband applications.24 Bi is used as a nanoparticle seed and catalyst for growing Ge nanowires in a vapor−liquid−solid (VLS) mechanism.25,26 In these examples, Bi is appropriate as a seed because it should not be incorporated into Ge. According to the phase diagram for Bi−Ge, there is no evidence for solubility at any temperature or composition.27 However, nanomaterials do Received: May 31, 2017 Revised: July 29, 2017 Published: July 30, 2017 A

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials

1 h at 150 °C prior to use. 1-Dodecanethiol (98%) was purchased from Sigma-Aldrich and degassed under vacuum at 150 °C for a minimum of 1 h prior to use. Hydrazine (anhydrous, 98%) was purchased from Sigma-Aldrich and used as received. Methanol, toluene, hexane, and acetonitrile were purchased from Fisher Scientific, purified using a solvent purification system, and stored in a glovebox under argon. 2.2. Synthesis. All pristine and doped-germanium NCs were prepared using a CEM microwave reactor (Discover SP) in dynamic mode. To ensure temperature accuracy, regular calibrations were performed using ethylene glycol as a standard. In a typical synthesis to prepare Ge NCs capped with oleylamine on the surface, 0.4000 mmol (130.6 mg) of GeI2 was weighed and loaded into a 35 mL microwave tube (purchased from CEM), and 8 mL of degassed oleylamine was added using a calibrated pipet in an argon-filled glovebox. To obtain larger NCs, calculated amounts of GeI4 were weighed and loaded into the tube at the same time as GeI2 to obtain a 0.4 mmol solution, as reported previously.38 The microwave reaction vessel was sealed with a cap obtained from CEM. The tube was sonicated in a water bath at room temperature for ∼10−20 min until thorough dissolution of the starting material (GeI2 or GeI2/GeI4 mixture). Complete dissolution of GeI2 in oleylamine leads to a yellowish solution. In the case of addition of calculated amounts of GeI4 to GeI2, the color of the solution after sonication was observed as pale yellow to colorless depending the GeI2/GeI4 ratio. The contents of the microwave tube were heated for 60 min at 250 °C in the microwave reactor. The color of the final product was dark brown to black depending on the precursor (GeI2/GeI4) ratio. In a typical isolation procedure, the microwave tube with the final product was transferred to an argonfilled drybox. The contents of the microwave tube were transferred to a 50 mL centrifuge tube, and 2−3 mL of anhydrous toluene and 25− 30 mL of anhydrous methanol as an antisolvent were used for isolation. The centrifuge tube was sealed with parafilm and centrifuged (8500 rpm) at room temperature for 10−20 min. The centrifuge tube was transferred to the drybox; the colorless supernatant was discarded, and the dark-brown precipitate collected at the bottom of centrifuge tube was redispersed in 5−6 mL of toluene or hexane. The precipitation procedure was carried out twice without exposure to ambient atmosphere, and the final toluene/hexane solution was stored under inert atmosphere for further use and characterization. To ensure that bismuth(III) iodide could be reduced by oleylamine, 0.4000 mmol of Bi precursor (235.9 mg of BiI3) was heated along with 8 mL of oleylamine after sonication and complete dissolution in the solvent for 60 min at 250 °C in the CEM microwave reactor. The isolation procedure is as described above (Supporting Information, Figure S2). To incorporate Bi as a dopant, first, a stock solution of 0.04 M BiI3 in oleylamine was prepared. To prepare the 0.04 M stock solution of BiI3 in oleylamine, 0.200 mmol of BiI3 (120 mg) was dissolved in 5 mL of oleylamine and sonicated until complete dissolution of Bi precursor in oleylamine was achieved. At the same time, 0.4000 mmol (130.6 mg) of GeI2 was weighed and loaded into a 35 mL microwave tube in an argon-filled drybox. The tube was sonicated as described above and transferred to a glovebox. Using a calibrated pipet, calculated volumes of BiI3 stock solution were added to GeI2 solution in oleylamine so that the total volume of the mixture was 8 mL. The volumes added from the Bi stock solution were 25 μL (0.25 mol %), 50 μL (0.5 mol %), 100 μL (1.0 mol %), 150 μL (1.5 mol %), 200 μL (2.0 mol %), 250 μL (2.5 mol %), 300 μL (3.0 mol %), 400 μL (4.0 mol %) and 500 μL (5.0 mol %). The synthesis and isolation were carried out as described above for Ge NCs. As synthesized, the surface of the Ge NCs is capped with oleylamine ligands. To remove the oleylamine ligands from the surface of Ge NCs, 5 mL of 5 M hydrazine solution freshly made by mixing 0.80 mL of anhydrous hydrazine in 4.2 mL dry acetonitrile was added to 5 mL of oleylamine-capped Ge NCs dispersed in toluene. The mixture was stirred at room temperature for at least an hour under inert atmosphere. The ligand-free Ge NCs were precipitated by centrifugation (8500 rpm) for 30 min at room temperature using 6 mL of toluene and 4−5 mL of methanol. To ensure the complete

not necessarily follow the miscibility behavior anticipated based on the thermodynamic phase diagrams.28−31 Mirkin et al. recently published a combinatorial polyelemental nanoparticle library using different metals (Ag, Au, Co, Ni, and Cu) via polymer nanoreactor-mediated synthesis, providing insightful information regarding the parameters leading to alloy formation and phase segregation at the nanoscale; while their binary NPs followed the mixture behavior trend expected from their bulk phase diagram for these specific metals, the higher-order nanostructures showed complexities beyond the miscibility of the metals associated with their bulk phase diagram.30 This deviation from bulk phase diagrams is also seen in recently reported Ge1−xSnx alloy NCs.6,32 Among cost-effective methods, solution-based syntheses of Ge NCs represent an efficient method to produce crystalline Ge in a robust and size- and/or shape-controlled manner and result in higher yields compared to instrumentation-intensive top-down methods.33,34 A comprehensive list of known colloidal methods for the synthesis of Ge NCs was published recently by Vaughn et al.5,35−37 Among different colloidal synthetic approaches, one pot microwave-assisted reduction of Ge halide salts (GeX2/GeX4, X = Cl, Br, or I) promises a scalable and low-cost production of highly crystalline Ge by overcoming the traditional disadvantages of long reaction times, high temperature and pressure, complicated precursor preparation, and exhaustive postsynthetic purification steps.5,38,39 In conventional heating routes, a minimum temperature of 250 °C is required to obtain crystalline Ge, as lower temperatures result in amorphous materials. Additionally, high energy barriers stemming primarily from the strong covalent character of Ge− Ge bonds also lead to challenges in precise control over nucleation and growth stages of their synthetic routes.40,41 Tuning the electronic properties of Ge NCs through composition manipulation using colloidal chemistry has only recently been demonstrated in Ge−Sn alloy formation and a few dopants.6,32,21 GeSn thin films and NCs have attracted intense research attention because of the potential for inducing strain-induced direct band transition in Ge, providing the incentive to investigate Bi doping of Ge NCs.24 GeBi amorphous thin films have been studied to investigate structure and IR properties.22−24 Similar to GeSn, GeBi films with high Bi concentration have the potential to become direct gap, and additionally, Bi should provide n-type carriers. Herein, we present the synthesis and characterization of Bidoped Ge NCs via a microwave-assisted colloidal route. The use of halide precursors and their effect on NC growth was investigated. The NCs were characterized by powder X-ray diffraction (PXRD), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), inductively coupled plasma−mass spectroscopy (ICP−MS), and UV−vis−NIR spectroscopy. Electrical conductance was measured on films. This work demonstrates the possibility of new compositions of Group IV NCs from colloidal routes.

2. EXPERIMENTAL PROCEDURES 2.1. Chemicals. Germanium(II) iodide, GeI2, was purchased from Prof. Richard Blair’s laboratory (University of Central Florida) and was characterized by PXRD to be phase-pure (Supporting Information, Figure S1). Germanium(IV) iodide, GeI4, (99.999%) was purchased from STREM chemicals. Bismuth(III) iodide, BiI3, (99.999% trace metal basis) was purchased from Sigma-Aldrich. Oleylamine ((Z)octadec-9-enylamine, CH3 (CH2)7CHCH−(CH2)7CH2NH2) was purchased from TCI America (technical grade, >40%) or SigmaAldrich (>98% primary amine) and degassed under vacuum for at least B

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials removal of free-standing oleylamine ligands and hydrazine, several cycles of washing with toluene and subsequent centrifugation followed. Lack of dispersibility of the dark-brown precipitate in toluene or hexane qualitatively demonstrated the successful removal of oleylamine ligands from the surface. The ligand-free Ge NCs were subsequently passivated with dodecanethiol. A solution composed of 10 mL of the thiol added to the dark-brown precipitate was heated for 60 min at 150 °C in the microwave (dynamic mode). The complete dispersion of Ge NCs in dodecanethiol was a qualitative indication of coordination of thiol ligands to the Ge NC surface. To isolate thiol-capped Ge NCs, 10 mL of toluene (or hexane) along with 20 mL of antisolvent (methanol) was added to the reaction mixture, and after 10−20 min of centrifugation (8500 rpm), the dark-brown thiol capped-Ge NCs (pristine or doped) were obtained. Particles were washed several times with toluene and methanol. The Ge NCs redispersed in 5−6 mL nonpolar solvent were stored for further characterization. These thiolcapped Ge NCs can also be stored under ambient conditions for several months.

electron probe convergence semiangle was approximately 23 mrad and the ADF inner detector semiangle was 33 mrad, hence resulting in medium angle annular dark field contrast for which some diffraction contrast cannot be neglected. Energy dispersive X-ray spectroscopy mapping (EDS) of the samples was carried out using a Thermo Corporation EDS spectrometer attached to the JEOL JEM 2500SE. Elemental maps were also collected using an FEI TitanX 60−300 microscope operated at 200 keV at the National Center for Electron Microscopy (NCEM), which is part of the Molecular Foundry at Lawrence Berkley National Laboratory. Images were acquired with a high-angle annular dark-field (HAADF) detector for STEM mode. A Bruker windowless EDS detector with a solid angle of 0.7 sr enables high count rates with minimal dead time. High-resolution TEM micrographs (HRTEM) were recorded at 300 keV with a CM 300 at NCEM. 3.3. X-ray Photoelectron Spectroscopy. XPS spectra were obtained using a Kratos Axis Ultra X-ray photoelectron spectrometer with an analyzer lens in hybrid mode. High-resolution scans were performed using a monochromatic aluminum anode with an operating current of 6 mA and voltage of 10 kV using a step size of 0.1 eV, a pass energy of 40 eV, and a pressure range between 1−4 × 10−8 Torr. The binding energies for all spectra were referenced to the C 1s core level at 284.8 eV. All XPS samples were prepared by drop-casting suspensions of NCs onto cleaned copper substrates under anaerobic conditions followed by drying at 90 °C. 3.4. Inductively Coupled Plasma−Mass Spectrometry. Quantitative Ge and Bi concentrations were acquired by ICP−MS. To digest the samples in 3% trace element HNO3, 1 mL of Bi-doped Ge NCs dispersed in toluene was stirred under ambient atmosphere overnight. Two sets of dilutions for Ge and Bi were prepared for Bidoped Ge samples to provide Ge and Bi concentrations appropriate for ICP−MS detection range. This was necessary to provide the appropriate concentrations in ppb for the standard range in ICP measurements. Prepared acidified aqueous samples were analyzed by the Interdisciplinary Center for Plasma Mass Spectrometry at the University of California at Davis using an Agilent 7500CE ICP−MS (Agilent Technologies, Palo Alto, CA). Concentrations of Ge and Bi were quantified using ICP−MS. The samples were introduced using a MicroMist Nebulizer (Glass Expansion 4 Barlow’s Landing Rd., Unit 2A Pocasset, MA 02559) into a temperature controlled spray chamber with He as the Collision Cell gas. Instrument standards were diluted from Inorganic Ventures Ge and Bi standards (Inorganic Ventures, Inc. 300 Technology Drive, Christiansburg, VA 24073) to 0.5, 1, 5, 10, 100, 500, 1000, and 2000 ppb with 3% Trace Element HNO3 (Fisher Scientific) in 18.2 MΩ cm water. Independent sources of Ge and Bi were analyzed initially along with quality control standards with Ge and Bi at 100 ppb analyzed every 10th sample (or less) as quality controls. Sc, Ga, Y, In, and Tm Inorganic Ventures standards were diluted to 15 ppm in 3% HNO3 and introduced by peripump as internal standards in a 1:30 ratio to the sample. 3.5. UV−vis−NIR Spectroscopy. Diluted dispersions of the oleylamine-capped Ge NCs were used to obtain spectra at room temperature on a UV-3600 Plus Shimadzu UV−vis−NIR spectrophotometer by using optically transparent quartz cuvette with optical path length of 1 cm. The spectra were recorded in the range of 300−1600 nm at a medium scan and 0.5 nm wavelength interval. 3.6. Conductivity Measurements. Chips were patterned using standard photolithography to define electrode areas on 100 nm silicon nitride on silicon wafers. Etching 30 nm of surface nitride followed by electron-beam evaporation of 30 nm of Au into the patterned areas resulted in planar chip surfaces for electrical measurements. Ten microliters of 66 mM NCs solution was drop cast onto patterned chips in a nitrogen atmosphere. Thin films of NCs formed as the solvent evaporated. A probe station was used to contact the patterned electrodes and measure current−voltage characteristics of six thin-film devices per chip. Film thickness was measured using a Dektak XT profilometer. NCs with three levels of bismuth dopant (0, 0.5, and 1 mol %) were measured, and the average conductivity was calculated using the average of 6 resistance measurements for each chip and the

3. MATERIALS CHARACTERIZATION Pristine and doped Ge NCs were characterized by powder X-ray diffraction (PXRD), transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), energy-dispersive X-ray spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS), UV−vis−NIR spectroscopy and inductively coupled plasma-mass spectrometry (ICP−MS). 3.1. Powder X-ray Diffraction (PXRD). PXRD patterns were obtained from 0.4 mL (∼2 mg) Ge NCs deposited by drop casting the toluene dispersion of Ge NCs onto a quartz substrate or silicon (Si510) single-crystal zero-background holder. The dark brown film of Ge NCs obtained after solvent evaporation was scanned for 2 h using a Bruker D8 Advance diffractometer (Cu Kα, 40 kV, 40 mA, λ = 1.5418 Å) in a 2θ range of 20−75° with a 0.02°/step size or on a Rigaku Miniflex 600 diffractometer dTex Ultrahigh-speed silicon strip detector which uses a standard copper X-ray tube (Cu Kα, λ = 1.5418 Å) with a nickel Kβ filter on the diffracted beam. The step resolution of the analysis scan in the range of 20−80° 2θ was 0.02°. The obtained patterns were compared to diamond cubic Ge (04-0545) powder diffraction file from the International Center of Diffraction Data (ICDD) database. To estimate the crystallite size of the NCs, the Scherrer method was performed by fitting the 220 reflection (Pseudo-Voigt) using Jade 6.0 software.42,43 Rietveld refinement was performed using JANA 2006 package. The standard deviations corresponding the lattice parameters were calculated considering the Berar’s correction factor. 3.2. Transmission and Scanning Transmission Electron Microscopy. Electron-transparent specimens for both TEM and STEM were prepared by drop casting dilute dispersions of Ge NCs in toluene onto either a holey carbon film supported by a 300 mesh copper specimen grid (SPI) or lacy carbon supported by a 400 mesh copper grid (Ted Pella). The grids were dried overnight under an incandescent lamp followed by oven drying at 80 °C to minimize any contamination during electron beam irradiation. The transmission electron microscopy imaging and selected area electron diffraction (SAED) of the samples were performed using a JEOL-JEM 2500SE transmission electron microscope (JEOL Ltd. Tokyo, Japan) at the Advanced Materials Characterizations and Testing Laboratory (AMCaT) at the University of California, Davis. This instrument is operated at 200 keV and is equipped with a Schottky field−emission electron gun (FEG) and a retractable 1k × 1k Gatan Multiscan CCD camera (model 794). Digital Micrograph software provided by Gatan Inc. was used to capture images. To determine the average particle size and respective standard deviation, 200−250 individual NCs were imaged from different sample areas and multiple specimen grids. Particle sizes were determined from intensity line profiles across individual particles in one consistent direction using the ImageJ software package. NCs were also imaged at 200 keV in STEM mode with an aberration-corrected JEOL JEM-2100F/Cs STEM equipped with a Gatan annular dark field (ADF) detector. For STEM imaging, the C

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials average film thickness on the chip. Resistive NCs devices were 70 × 4 μm (W × L) with thicknesses of approximately 100 nm. 3.7. FTIR Spectroscopy. FTIR measurements were carried out with a Bruker Alpha spectrophotometer.

was incorporated into the NC and not removed via the hydrazine treatment. The PXRD patterns of both oleylamine and dodecanethiol capped Ge NCs prepared from GeI2 and GeI2/BiI3 with 0.5−2 mol % Bi precursor are shown in Figures 1a and b, confirming

4. RESULTS AND DISCUSSION All pristine and doped Ge NCs were synthesized by the reduction of germanium iodides with oleylamine assisted by microwave heating under the air-free environment previously developed.44 The reaction proceeds by initial nucleation of elemental Ge seeds through reduction of GeI2 at low temperature and completes by further growth of Ge NCs at higher temperature.9 Oleylamine provides stability and solubility of the NCs in addition to reducing the Ge precursor. In this work, the reaction temperature is increased from 210 °C used in prior publications by 40 degrees to 250 °C as we observed enhanced crystallinity and stability of the resulting NCs at ambient atmospheric conditions.38,45 Increasing the reaction temperature to 270 °C did not affect the product yield, crystallinity, or stability further. Therefore, all reactions described herein are performed at a maximum reaction temperature of 250 °C. Reaction times shorter than 30 min led to reddish colored solution and presumably very small particles, which could not be precipitated using antisolvent. Therefore, the reaction time was always kept to 60 min once 250 °C was reached. Oleylamine is a poor microwave absorber, and as pure solvent, it takes almost 1 h to reach 250 °C. In this reaction, it is the presence of the halide containing precursors that allow the reaction to proceed efficiently. Temperature ramping time for Ge NCs in oleylamine using GeI2 as the precursor is 10−12 min. For reactions where GeI4 is added to GeI2 to obtain larger Ge particles, the ramping rate is even faster due to better microwave absorbance of GeI4. It was shown in a previous study that using a mixture of GeI2/GeI4 provides size control.38 Therefore, GeI2/GeI4/BiI3 mixtures were employed to obtain larger NC diameters, and the addition of BiI3 leads to even faster times to reach 250 °C. BiI3 was utilized as the Bi source to produce Bi-doped Ge NCs. It has been shown that Bi3+ can be reduced in oleylamine with n-butyllithium.46 However, it was important to determine whether or not oleylamine could reduce Bi3+ without an additional reducing agent to keep the reaction as simple as possible. The standard reduction potential of Bi(III) is +0.308 V compared with that of Ge(II) of +0.24 V.47 The PXRD pattern of the product from the reaction of BiI3 in oleylamine at 250 °C for 1 h is shown in the Supporting Information (Figure S2) and is consistent with elemental bismuth. To prepare the doped Ge NCs, the same reaction parameters (maximum temperature and time) were followed and, initially, a mixture of GeI2 and BiI3 was investigated.9 To determine that the dopants were not simply on the surface or easily removed, a ligand exchange study was performed. Ligand exchange of oleylamine-capped doped Ge NCs by dodecanethiol, an alkane monothiol, was carried out as previously reported by first removing all surface ligands with hydrazine and heating the NCs in the presence of dodecanethiol.45 Both oleylamine and dodecanethiol-capped Ge NCs were characterized by PXRD and TEM to compare the crystallite and particle size. FTIR spectra of oleylamine and dodecanethiol capped Ge NCs to reveal the surface chemistry of the NCs are provided in Figure S3. Dodecanethiol-capped samples were investigated via XPS analysis to ensure that the Bi

Figure 1. PXRD patterns of (a) oleylamine-capped and (b) dodecanethiol-capped Bi-doped Ge NCs prepared with varying amounts (0.0−2.0 mol %) BiI3 compared to the reference pattern (PDF no. 04-0545). Crystallite sizes are provided in Tables S1 and S2. (c) Lattice parameter data vs bismuth content (0.0−2.0 mol %) for Bidoped Ge NCs (small peak at 37.8° indicates the peak due to the sample holder; the X-ray pattern of the blank substrate is shown in the Supporting Information (Figure S10)).

the presence of Ge NCs. The summary of crystallite size obtained from Scherrer analysis using BiI3 with constant amount of GeI2 (0.40 mmol) and oleylamine as passivating ligand are shown in the Supporting Information (Table S1) in addition to the dodecanethiol exchanged NCs (Table S2). The crystallite size of the NCs increases consistently as the Bi iodide concentration increases. The diffraction peaks at 27.3, 45.4, and 54.5° are attributed to the [111], [220], and [311] lattice planes of diamond-cubic Ge, respectively. The crystallite sizes are 4.2 (1) and 4.6 (1) nm for oleylamine and dodecanethiolcapped Bi-doped Ge NCs (1 mol %), respectively. The crystallite sizes of dodecanethiol-capped Ge NCs are slightly larger compared to NCs capped by oleylamine. This is consistent with pristine Ge where the Ge crystallite sizes are 3.2 (1) and 3.4 (1) nm for oleylamine and dodecanethiolcapped NCs, respectively. The addition of BiI3 to GeI2 results D

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials

aggregation. The statistical analysis of Ge NCs diameters and their fitting to a Gaussian model are shown in the Supporting Information (Figure S5). Incorporating Bi as n-type dopant using BiI3 as the Bi precursor resulted in an increase in Ge NCs mean diameter to 5.5−18 nm depending on the BiI 3 concentration. The influence of halide reagents for doping such as InCl3 to incorporate In on the growth rates of CdSe NCs has been studied and also shows size increase with increasing dopant concentration.48 Besides particle diameter enhancement, a broader size distribution (0.8 to ≥2.2 nm) is observed using higher amounts of dopant precursor, although the relative size distribution (rsd) % are fairly similar to results from GeI2/GeI4 reactions, suggesting that the inhomogeneity in NC diameter is an inherent constraint of solution-based Ge NC synthesis with these halide reagents.38 Doping can affect the growth kinetics of the NCs through altering surface energy of the NCs and chemical potential of both dopant and host components and is even more pronounced in the solid phase (NCs) compared to solution phase as a result of strong interactions between dopants and host.13,49 The increase in size distribution stemming from the influence of the dopants by increasing enthalpy of mixing due to different bond energies of the host−host, host−dopant, and dopant−dopant has been documented in the literature.13,50 The combination of surfactants/stabilizing agents can reduce the aggregation rate by balancing the degree of repulsive-attractive forces among the particles, allowing the particles to grow steadily. Additional ligands/surfactants binding more strongly to the Ge surface may be important for controlling growth and preventing aggregation of NCs. Coalescence of smaller Ge NCs is observed using dopant concentration over 2 mol %, which makes it challenging to calculate particle average diameter. The agglomeration of NCs is clearly observed in Figure 2e using 5 mol % BiI3. The increases in average diameter of Ge NCs along with increasing BiI3 concentration can be attributed to the presence of more iodine that may also contribute to the oxidation of Ge (II) to Ge (IV). Formation of more Ge (IV) results in the increased growth kinetics of NCs immediately after the initial burst nucleation stage. In the absence of dopant precursor (BiI3), the fast nucleation results in the formation of monodispersed small particles. Higher iodide concentration impacts the growth kinetics, yielding larger and more polydisperse NCs. Elemental mapping and HRTEM images were obtained on larger Ge NCs that were prepared using multivalent Ge precursors (GeI2/GeI4): GeI2/GeI4/BiI3 provided Bi-doped Ge NCs where size depends both on the relative amounts of all three reagents. Figure 3 shows two micrographs of oleylamine and dodecanethiol-capped Bi-doped Ge NCs (2 mol % BiI3 and GeI2: 0.35 mmol, GeI4: 0.05 mmol). The histograms of Ge NCs diameters and their fit to a Gaussian model are shown in the Supporting Information (Figures S6, S7). In addition to the inherent particle size distribution observed in Ge NCs prepared by colloidal based methods, using three different precursors (GeI2/GeI4/BiI3) with different reduction potentials contributes to the complexity of controlling the growth kinetics of NC formation. This complexity complicates mapping the size dependent properties of these NCs and will inhibit the formation of well-ordered arrays.51,52 High-angle annular dark-field and HRTEM images of Bidoped Ge NCs are shown in Figures 4 and 5. HRSTEM and HRTEM micrographs show lattice fringes consistent with the crystallinity of the particles. Lattice spacing obtained from a fast

in increased crystallite size from 3.2 (1) to >17 nm (0.0 to 5 mol % BiI3). No reflections characteristic of elemental Bi were detected, suggesting that the Bi is incorporated within the lattice. In addition, the lattice parameter of the oleylaminecapped Bi-doped Ge NCs increases from 5.655 (3) Å (Ge NCs) to 5.683 (3) Å (2 mol % Bi-doped Ge NCs), as shown in Figure 1c (refinement patterns are shown in the Supporting Information, Figure S4). Using 3 mol % Bi precursor leads to a decrease in lattice parameter (5.664 (3) Å), which suggests that maximum incorporation of Bi in Ge lattice is limited to 2 mol %. In typical bulk material solid solutions, the lattice parameter maximum might remain unchanged. However, in this example, the size control of the NCs is also lost (see Figure 2), and the

Figure 2. Representative TEM images (a−e) of oleylamine-capped pristine and doped Ge NCs with 0.4 mmol GeI2 and varying amounts of BiI3 (0.0−5.0 mol %). (b) The inset provides the HRTEM micrograph showing the lattice fringes corresponding to the (220) plane of the diamond-cubic Ge for the Ge NCs using 1 mol % BiI3 (the scale bar indicates 5 nm). (f) TEM micrograph provided as an example for dodecanethiol-capped pristine Ge NCs showing an increased particle diameter relative to that of oleylamine-capped pristine Ge NCs shown in panel a. At the top of each image, the quantity of dopant precursor, average diameter, and size distribution are provided (with the exception of panel e due to agglomeration of NCs).

average amount of Bi incorporated appears to be less than that observed for the 2 mol % sample. The presence of free Bi is not observed in any of the PXRD patterns. This is attributed to either the small amounts (below detection) or small crystallite size that is well-dispersed and not observable via conventional X-ray diffraction. Figure 2 shows the representative bright field TEM micrographs of oleylamine-capped pristine and doped Ge NCs synthesized with different amounts of BiI3. The TEM image of pristine Ge NCs shows quasi-spherical particles with an average diameter of 4.6 (6) nm and no evidence of E

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials

Figure 5. HRTEM of these oleylamine-capped Bi-doped Ge NCs recorded with a CM 300 at NCEM operated at 300 kV (0.35 mmol GeI2, 0.05 mmol GeI4, and 1.5 mol % BiI3; the inset shows the fast Fourier transformation calculated from the area marked by the red square).

Figure 3. TEM images of (a, b) oleylamine-capped and (c, d) dodecanethiol-capped pristine and doped Ge NCs (0.35 mmol GeI2, 0.05 mmol of GeI4, and varying amounts of BiI3). At the top of each image, the quantity of dopant precursor, average diameter, and size distribution are provided.

EDS spectroscopy was performed to determine the elemental composition of the pristine and doped Ge NCs. The EDS spectra obtained from samples prepared by using BiI3 are shown in Supporting Information, and Ge, Bi, C, O, and Cu are clearly distinguished. The C and Cu peaks are due to the holeycarbon TEM grids. The oxygen signal is attributed to the oxidation of Ge NCs in the process of TEM grid preparation under atmosphere conditions. The Bi Mα peak at 2.44 eV confirms the incorporation of Bi in the Ge NCs core or on its surface (Supporting Information, Figure S8). The quantitative elemental mapping with energy dispersive X-ray spectroscopy in STEM mode (EDS-STEM) of the 2 mol % Bi-doped Ge NCs was acquired to determine the distribution of dopants. Figures 6a−c show STEM-EDS elemental mappings corresponding to Ge, Bi, and O for an agglomeration of particles. The Ge and Bi mapping (Figures 6a and b) show both elements distributed throughout the particles. The oxygen map shows the highest intensity color at the perimeters, suggesting surface oxidation. Two regions are circled in an overlay of the Ge and Bi elemental mapping (Figure 6d) and the atomic percentages of Ge and Bi and standard deviation are summarized in Table 1. Analyses with the ESPRIT software package by K-factor calculations that include ZAF correction, i.e., considering the effects of atomic number, absorption, and fluorescence of X-rays within the specimen provided chemical distribution maps for Ge and Bi in the selected regions of interest.53 The spectral intensities in region 2 reveal 1.0 ± 0.8% of Bi; while region 1 shows a significantly larger amount of Bi (7 ± 3%), hence representing a region slightly enriched with Bi. While region 1 shows a high standard deviation, the signal is clearly distinguished. Additional images and maps can be found in the Supporting Information (Figure S9, Table S3). Because no peaks in X-ray patterns of the Bi-doped Ge NCs can be attributed to segregation of Bi in reactions using 2−5 mol % BiI3, some control reactions were performed to

Figure 4. HRSTEM images of oleylamine-capped (a) pristine (0.35 mmol GeI2, 0.05 mmol GeI4) and (b) doped Ge NCs (0.35 mmol GeI2, 0.05 mmol GeI4, 2 mol % BiI3). (c, d) Small angle annular dark field diffraction and Z-contrast micrographs, respectively, of oleylamine-capped doped Ge NCs (0.35 mmolGeI2, 0.05 mmol GeI4, and 1 mol % BiI3) acquired in STEM mode.

Fourier transform of the particle enclosed in the red square resulted in 0.320 ± 0.003 nm, which is in agreement with XRD refinement data of the 1.5 mol % BiI3 product. Composition heterogeneity observed in HRSTEM and Z-contrast images (Figures 4b and d) is attributed to the presence of heavier Bi dopants, resulting in brighter spots compared to Ge NCs. F

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials

with 7 mol % BiI3. The Bi rich region observed in elemental mapping in Figure 6 for Ge NCs prepared by using 2 mol % BiI3 can be a combination of some Bi NC formation in assynthesized Ge NCs residing on the Ge NCs surface and some Bi atoms leaving the host NCs under the electron beam after an extended time. The absence of a Bi phase in X-ray patterns of 2−6 mol % BiI3 may be due to insufficient diffraction planes on the surface to produce detectable signals attributed to Bi atoms. By increasing the Bi concentration to 7 mol %, the diffraction pattern of Bi is discernible, suggesting that the solubility of Bi in Ge decreases, leading to segregation of Bi. SAED patterns collected on oleylamine-capped Bi-doped Ge NCs (Supporting Information, Figure S10) are in reasonable agreement with the lattice spacing determined from XRD analysis, leading to increase in interplanar spacing of lattice planes (0.322 ± 0.003, 0.326 ± 0.004, 0.327 ± 0.003, and 0.326 ± 0.004 nm for 0.50, 1.0, 2.0, and 3.0 mol % BiI3, respectively). Additional diffraction spots (indicated in the Supporting Information) are apparent in the SAED patterns for samples with BiI3 concentration higher than 5 mol %; the diffraction spots can be indexed as Bi, but they are few and weak in intensity, making the identification inconclusive. Selected area electron diffraction patterns obtained from oleylamine-capped Bi-doped Ge NCs show increased d-spacing with increasing Bi concentrations up to 2 mol % and a decrease in d-spacing for Bi concentrations at and above 3 mol %, consistent with the conclusion drawn from XRD refinement patterns. The corresponding ring-like SAED patterns confirm the polycrystalline nature of the Bi-doped Ge NCs. In samples prepared with higher Bi concentrations, the diffractions spots for Bi phase are detectable starting from 6 mol % BiI3 (Supporting Information, Figure S11), a slightly lower concentration than what can be detected by PXRD. High-resolution XPS for both 1 and 2% Bi-doped dodecanthiol-capped Ge NCs provided in Figure 8 shows the

Figure 6. (a−c) STEM-EDS elemental mappings corresponding to Ge, Bi, and oxygen of oleylamine-capped Bi-doped Ge NCs (0.35 mmol GeI2, 0.05 mmol GeI4, 2 mol % BiI3), respectively. (d) Integrated EDS spectra of the agglomerated oleylamine-capped Bidoped Ge NCs extracted from the regions highlighted in the inset.

Table 1. Summary of Chemical Composition Analysis for Ge and Bi from the Indicated Regions 1 and 2 of Figure 6 of the Agglomerated Oleylamine-Capped Bi-Doped Ge NCsa

a

region in map 1

atom % (Ge)

atom % (Bi)

1 2

93 ± 4 99 ± 3

7±3 1.0 ± 0.8

0.35 mmol GeI2, 0.05 mmol GeI4, 2 mol % BiI3.

determine the maximum level of Bi incorporation with the goal of identifying excess elemental Bi by X-ray diffraction. Figure 7 shows the X-ray patterns of oleylamine-capped Ge NCs prepared by using 5, 6, and 7 mol % BiI3. The minimum Bi concentration at which the Bi phase emerged was for Ge NCs

Figure 8. High-resolution XPS spectra of S 2p and Bi 4f peaks for dodecanethiol-capped Bi-doped Ge NCs (pristine and 1 and 2 mol % BiI3). S 2p peaks were deconvoluted into three Gaussian peaks (S 2p1/2, Bi 4f5/2, and S 2p3/2). * indicates the presence of Bi3+.

presence of Bi. The S 2p peaks observed in the XPS spectra are clearly broadened in both the 1 and 2% Bi-doped Ge NCs and can be deconvoluted into three Gaussian peaks with binding energies of 163.84, 163.0, and 162.7 eV being assigned to S 2p1/2, Bi 4f5/2, and S 2p3/2 peaks, respectively. The S 2p doublet shows a 1:2 branching ratio and 1.14 eV of spin−orbit splitting, which is consistent with previously reported studies.54 The presence of S 2p peaks confirms the successful effective

Figure 7. PXRD patterns of oleylamine-capped Bi-doped Ge NCs prepared with x (x = 3.0, 5.0, 6.0, and 7.0) mol % BiI3 compared to those of the Ge reference pattern (PDF no 04-0545) and Bi reference pattern (PDF no. 44-1246). * indicates the peak due to the sample holder. The X-ray pattern of the blank substrate is shown in the Supporting Information (Figure S10). G

DOI: 10.1021/acs.chemmater.7b02241 Chem. Mater. XXXX, XXX, XXX−XXX

Article

Chemistry of Materials

gap (Eg). Bulk Ge is an indirect band gap material, and Tauc plot transitions demonstrate that both pristine and doped Ge NCs are indirect band gap materials. As shown from the Tauc plots, with increasing average size of the Ge NCs by introducing higher concentrations of BiI3 in the reaction solution, the optical band gaps decrease, as expected. The calculated band gaps are similar to the band gap values reported in previous work.69,70 No additional absorption is observed, which excludes any plasmon resonance or other electronic transitions.71−73 To investigate the electrical conductivity of the doped Ge NCs compared to that of pristine Ge NCs, samples doped with 0.5 and 1 mol % BiI3 were measured as described above, and the conductivities were found to be of the order of nS/m. Thin films consisting of nanoparticles separated by ligand capping layers have been often explored for their electrical properties. A SEM micrograph of the platform is shown in the Supporting Information, Figure S 14. In these cases, the ligands act as tunneling barriers between the nanoparticles, and the transport through the film is typically dominated by a hopping mechanism.74−77 The electrical conductivity of the Ge NCs films is low and can be understood using a hopping-based model of electrical conductivity. The general expression for conductivity is σ = nqμ, where n is the charge carrier density, q is the elementary charge, and μ is the mobility. The mobility of a hopping system is given by μ ∝ exp(−EA/kBT)exp(−βl), where EA is the activation energy, kB is the Boltzmann constant, T is the temperature, β is the tunneling decay constant of the ligand, and l is the interparticle separation.78 This expression shows that two components are involved in mobility of charges in a hopping process. The Arrhenius component represents the thermally induced alignment of energy levels, and the second term represents the probability that a charge carrier will tunnel from one particle to another through the molecular barrier. In this case, the activation energy includes the intraparticle energy barrier associated with the band gap of the Ge NCs core. The film conductivity increases with increasing level of Bi dopant in the NCs (Figure 10). This result is likely due to two effects: first, the increase in dopants corresponds to both a higher charge carrier density n, and second, because when the particle size increases and particle band gap decreases (Figure 9) with

displacement of oleylamine by dodecanethiol on the Ge surface. The S 2p3/2 with binding energy of 162.7 eV can be assigned to bond thiolate group,55 and S 2p1/2 with binding energy of 163.84 eV can be attributed to free thiol.56 Additionally, the corresponding Bi 4f7/2 peak is observed with a binding energy of 157.4 eV (ΔE = ∼5.6 eV from 4f5/2). The value is consistent with a reduced form of Bi (4f7/2 binding energy for elemental Bi = 157 eV).57−59 The extra peaks observed at ca. 159 and 164.5 eV in the XPS spectrum of the 2% Bi-doped Ge NCs can be attributed to either Ge−Bi bonding24 or to Bi−O bonding59,60 because both of these peaks have the same binding energy. The absence of broad vibrational modes in FTIR spectra (Figure S3) of Bi-doped Ge NCs related to Bi−O, however, suggests the assignment of this peak mostly to Ge−Bi bonding.61,62 The increase in relative intensities of Bi 4f peaks from 1 to 2% Bidoped Ge NCs is consistent with the increasing dopant concentration, which further confirms the assignment of the Bi 4f peaks. XPS spectra obtained for Bi concentrations