Growth and Characterization of Strained and Alloyed Type-II ZnTe

Nov 18, 2012 - Andrew A. R. Watt,. † and Jason M. Smith*. ,†. †. Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, Unit...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCC

Growth and Characterization of Strained and Alloyed Type-II ZnTe/ZnSe Core−Shell Nanocrystals Simon M. Fairclough,† Edward J. Tyrrell,† Darren M. Graham,‡ Patrick J. B. Lunt,‡ Samantha J. O. Hardman,‡ Annette Pietzsch,§ Franz Hennies,§ Jonathan Moghal,† Wendy R. Flavell,‡ Andrew A. R. Watt,† and Jason M. Smith*,† †

Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, United Kingdom The Photon Science Institute and School of Physics and Astronomy, The University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom § MAX-lab, Lund University, P.O. Box 118, SE-221 00 Lund, Sweden ‡

S Supporting Information *

ABSTRACT: We investigate the growth and the physical and optical properties of type-II heterostructured ZnTe/ZnSe colloidal nanocrystals, focusing on the role of the 7% lattice mismatch between the two materials in determining growth homogeneity and band structure. We find that the lattice mismatch between the two materials places limitations on the range of structures that can be grown, and for those in which coherent growth is achieved we present clear evidence that the low bulk modulus ZnTe cores are compressed by the higher modulus ZnSe shells, accentuating the red-shift of the excitonic state with increasing shell thickness. By employing a variety of characterization tools we build a clear picture of the core−shell architecture. We show how strain is manifested in structures with sharp core−shell interfaces and how intentional alloying of the interface can influence the growth and exciton energies. We show that a (2,6)-band effective mass model is able to distinguish between the as-grown “sharp” and “alloyed” interfaces, indicating that the alloyed structures incorporate reduced strain.



INTRODUCTION

tuning of the effective type-II band gap across a wider range, from 500 nm into the NIR with a lattice matched CdSe shell.4 ZnTe/ZnSe NCs have also been produced,8 offering the first cadmium-free (and therefore lower toxicity) type-II nanocrystals. However, this material combination contains a large lattice mismatch of ∼7%, potentially limiting the range of structures that can be made defect-free, and introducing a high degree of strain where coherent growth is achieved. Such built-in strain is known to substantially modify exciton energies in core−shell structures, and compression of the core by a smaller lattice constant shell can result in more pronounced type-II behavior,2,10 particularly where the core has a small bulk modulus relative to the shell and is therefore easily compressed. In this paper we extend these studies by investigating in detail a size series of zinc blende ZnTe/ZnSe core−shell NCs. We observe strain effects due the differences in lattice constants and bulk moduli of the constituent materials (ZnTe 6.10 Å, 50.5 GPa; ZnSe 5.67 Å, 62.4 GPa11,12). We find that the growth of a uniform ZnSe shell occurs for core sizes less than 3.6 nm in diameter. The existence of about 0.45 nm of alloying at the core−shell interface is established using X-ray photoelectron spectroscopy (XPS). Evidence of strain in these structures is identified using X-ray diffraction (XRD), XRD simulation, and

Type-II heterostructured nanocrystals (NCs) have attracted great interest in recent years as they provide a unique means to spatially separate charge carriers within a quantum dot.1,2 In contrast to commonly available type-I core−shell heterostructures, in which both charge carriers are confined to the core and the band gap is determined by the core size alone, a type-II heterostructure has a staggered band alignment that can localize the electron and hole in separate regions. Such charge separation brings a number of practical benefits: It allows the effective band gap of a NC to be made smaller than the constituent bulk materials.3−5 It also facilitates control over the degree of wave function overlap between the electron and hole, which determines the photoluminescence (PL) and Auger recombination lifetimes,1,6 and the electron−hole exchange interaction.7 This ability to engineer the excitonic properties leads to a number of potential applications in fields such as photovoltaics8 and solution processed laser materials.9 Following the seminal work of Kim et al.1 on the growth of CdTe/CdSe type-II NCs, significant research effort has been directed toward cadmium chalcogenide type-II systems which show good optical performance over the red and NIR regions of the spectrum. It has since been demonstrated that ZnTe offers an attractive alternative to CdTe as a core material, as it retains the low ionization potential (high valence band edge) for effective hole localization, but its wider bulk band gap offers © XXXX American Chemical Society

Received: September 4, 2012 Revised: November 17, 2012

A

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

The core NCs were cleaned by the addition of excess methanol and subsequent centrifugation for 10 min at 4000 rpm, followed by extraction of the supernatant and redispersion of the NCs in hexane. All subsequent cleaning steps again used methanol as the nonsolvent and redispersion in hexane. ZnSe Shell Growth. The lattice mismatch of ∼7% between the core and shell materials introduces the potential for defect creation through inhomogeneous growth. Many techniques have been employed for shell growth; however, recently “successive ionic layer adsorption and reaction” (SILAR) has been shown to retain high quantum yields by promoting homogeneous defectfree growth while maintaining or enhancing the monodispersity of the system.3,16 The separate injection and growth of stoichiometric amounts of the anion and cation precursors means that the two do not coexist in solution; therefore, the likelihood of separate nucleation of the shell precursors is negligible. Here we use zinc oleate and Se-TOP precursors to enable a common ligand combination throughout the synthesis and characterization. To grow a specific shell thickness, typically 2 mL of the original ZnTe dispersion was cleaned as above and mixed with 2 mL of the BE/OA/TP solution. This was subsequently placed under vacuum for 20 min at 100 °C to remove low boiling point materials from the solution. The core reaction was empirically found to be quantitative, and we therefore use the average diameter of the core, as determined through TEM, to calculate the concentration of the NCs in the reaction solution using the technique described previously in ref 4. For these reactions, typically 0.708 μmol, 0.210 μmol, and 0.101 μmol of ZnTe NC were used for the growth on the 2.4, 3.6, and 4.6 nm diameter cores, respectively. Using the SILAR technique, aliquots of 0.1 M zinc oleate or 0.1 M Se-TOP were subsequently injected and heated to the growth temperature of 235−240 °C. The temperature for each injection and growth was optimized to maximize the reactivity of the precursor while minimizing the risk of Ostwald ripening. The smallest cores, for example, required the injection of the first precursor at a temperature of less than 180 °C which was then raised to 235 °C for the first monolayer (ML) growth, while subsequent MLs were injected and grown at 240 °C. An Anglia Instruments high temperature in situ absorption dip probe was used to observe the evolution of the absorption peak to monitor shell growth. Shifts of the absorption peak could only be observed for the Se injection; similar observations have been seen in other SILAR reactions.3 For a typical reaction up to 25 min was required for Se growth and 30 min for zinc growth. The core−shell NCs were initially cleaned with an excess of a premade 3:1 methanol/butanol solution, centrifuged to enable the NCs to precipitate out of the solution, and the NCs redispersed in hexane. Subsequent cleaning steps used only methanol as the nonsolvent and redispersed with small amounts of hexane. Growth of Alloyed Interface NCs. In order to grow alloyed interfaces we modified the shelling procedure described above to make use of the growth kinetics with mixed precursors as established by ref 5. Starting with the solution of ZnTe cores, Te-TOP and Se-TOP were injected, each in quantities equivalent to one monolayer of growth, at 180 °C. The solution was then heated to 240 °C for a reaction time of 25 min. We then introduced 4/3 monolayers quantity of Zn oleate and allowed the reaction to proceed for a further 30 min. This sequence was repeated twice, omitting the Te-TOP, thus producing four monolayers of fully reacted growth with a composition graded from Te-rich at the core interface to Se-rich at the surface.

high resolution transmission electron microscopy (HR-TEM). The dependence of exciton energy on shell thickness is consistent with a high degree of strain, and is modeled effectively using the continuum elasticity model developed by Balasubramanian et al.13 and Smith et al.2 This result contrasts with that of NCs where alloying at the core−shell interface is intentionally increased and a strain-relaxed model is found to be more appropriate.



METHODS The synthetic routes to producing ZnTe cores are limited and have predominantly used the highly pyrophoric diethyl zinc.4,8,14 In order to develop a “greener” synthetic technique, we instead modified the procedure of Zhang et al.,15 using the air stable zinc oleate as the zinc precursor. This route does however require the use of a superhydride as a strong reducing agent which reacts with the Te precursor before injection. The reduced Te enables a lower nucleation temperature and produces a narrow size distribution while not contributing measurably to the composition of the ZnTe cores (Figure S1, Supporting Information). This synthetic route enables large quantities of NCs to be synthesized while keeping the size distribution to 7−9% which produces strong excitonic features in the absorption spectra and a good base to produce highly luminescent core−shell NCs. Chemicals. All chemicals were purchased from Sigma Aldrich and used without further purification or modification unless stated: benzyl ether (BE) 98%, butylamine 98%, octyl ether (OE) 99%, oleic acid (OA) 99%, selenium (Se) pellets 99.9%, superhydride in THF (1 M), tellurium (Te) pellets 99.999% mesh, trioctylphosphine (TOP) 90%, zinc acetate dihydrate 98% (Acros) and butanol (anhydrous), hexane (anhydrous) and methanol (anhydrous). All reactions and optical characterization were conducted using standard air-free techniques. A 1 M superhydride solution was prepared by solvent exchange with OE. Here equal volumes of OE and superhydride THF were mixed and placed under vacuum. This mixture was then heated slowly to 110 °C for 30 min until the solution was clear. This solution was allowed to cool and placed under nitrogen. 0.8 M Te-TOP, 0.1 M Te-TOP, and 0.1 M Se-TOP precursors were prepared by dissolving Te pellets and Se pellets in TOP in a nitrogen-filled glovebox and stirred overnight. A 0.1 M zinc oleate solution was prepared using zinc acetate dihydrate in OA and BE solution with (9:1) OA/Zn molar ratio for the 2.4 nm cores and (2.5:1) OA/Zn molar ratio for the large cores. This solution was heated to 160 °C under vacuum for 40 min and placed under nitrogen flow, and the temperature was then raised to 180 °C until the solution was clear. A solution of vacuum degassed BE OA was mixed with TOP in a volume ratio of (15:1:1) (BE/OA/TP). ZnTe Cores. In a typical reaction, 0.8 mmol (0.1756 g) of zinc acetate dihydrate was placed in a three neck vessel with 15 mL of BE and 1 mL of OA. The solution was heated slowly to 150 °C under vacuum for 40 min until the solution turned clear, and was then placed under nitrogen flow at 160 °C for a further 10 min before being heated to the reaction temperature at 245 °C. Separately, the Te precursor was prepared by reacting 1 mL of 0.8 M Te-TOP with 0.8 mL of 1 M superhydride OE in a nitrogen-filled glovebox, where it was allowed to react for 2−3 min at room temperature. The reacted turbid pink precursor solution was transferred from the glovebox via syringe into the hot reaction solution. The solution temperature was allowed to recover and kept at 245 °C until the desired NC size was obtained, whereupon the solution was quenched using a water bath. B

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Further ZnSe shell growth was then carried out as described previously. Characterization. Samples for transmission electron microscopy (TEM) were cleaned four times and drop cast onto a 400 mesh Cu grid with an ultrathin carbon film. TEM and electron dispersive X-ray spectroscopy (EDX) were taken using JEOL 2010 TEM operating at 200 kV with Oxford Instruments EDX. HR-TEM images were taken using a JEOL 4000EX TEM operating at 400 kV. A gold nanoparticle reference was

used to calibrate the microscope before every session. XRD patterns were recorded using a Philips PW 1830 with a Cu Kα source. Samples for XPS required a ligand exchange to the shorter chain butylamine to prevent excess charging during the measurements. These samples were prepared by adding excess butylamine to a cleaned NC solution and left overnight at 50 °C under nitrogen. The samples were cleaned and redispersed with excess methanol and hexane and stored under nitrogen. The samples were rapidly drop-cast onto ITO-coated glass substrates in air, and immediately inserted into the vacuum system of the spectrometer. The total air exposure was less than 15 min, including the pumping time. These measures are effective in suppressing surface oxidation of the NC,17 and no significant surface oxidation was observed in XPS. The XPS experiments were carried out on the I511-1 beamline at MAXlab, Sweden, using a Scienta R4000 electron energy analyzer. Photoemission spectra were recorded at room temperature at a total instrumental resolution of 170 meV (at 250 eV photon energy) to 450 meV (at 750 eV photon energy). Optical absorption spectra were recorded using a Varian Cary 5000 UV−vis−NIR spectrometer. Ensemble PL spectra were taken on Varian Cary Eclipse fluorescence spectrophotometer. The absolute quantum yields were measured using 365 nm excitation from a diode laser, utilizing a Labsphere integrating sphere and calculated using the method of de Mello et al.18 PL lifetimes were

Figure 1. Evolution of the absorption spectra of ZnTe cores with size, showing good quantum confinement and monodisperse behavior up to 4.6 nm in core diameter.

Figure 2. Low resolution TEM images showing spherical (a) 2.4 nm and (d) 3.6 nm ZnTe cores. Low resolution images of ZnTe/ZnSe core−shell NCs with a 2.4 nm core showing spherical particles with (b) 2 ML shell and (c) 4 ML shell. Low resolution images of ZnTe/ZnSe core−shell NCs with a 3.6 nm core showing spherical structures for (e) 2 ML shells, and (f) anisotropic tetrapod or cubic structures observed with 5 ML shells. The insets show high magnification images, including some high resolution images and the corresponding size distributions. C

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Figure 3. (a) Comparison of the experimental and simulated XRD powder diffraction patterns for 3.6 nm diameter ZnTe cores (black), 2.4 nm diameter ZnTe cores with 4.5 ML ZnSe shell (red), 3.6 nm diameter ZnTe cores with 5 ML ZnSe shell (green), and 4.6 nm diameter ZnTe cores with 4.5 ML ZnSe shell (blue). The bulk diffraction peaks of ZnTe and ZnSe are highlighted at the bottom and top of the plot, respectively. (b) The radial and tangential strain coefficients for a 4.6 nm diameter NC with 5 ML ZnSe as a function of radial position. (c) The corresponding radial and tangential lattice constants of part b as a function of radial position.

two-band electron and six-band hole Hamiltonian to take account of the complex nature of the valence band and its coupling to the conduction band. Electron and hole energy (h) respectively where Q = S, P, levels are labeled nQ(e) j and nQj D,... denotes the lowest value of angular momentum for the envelope function, n is the level number, and j is total angular momentum.22 We therefore describe the lowest exciton (h) transition as 1S(e) 1/21S3/2. The Coulomb interaction between the charge carriers and the dielectric polarization energies are treated as first order perturbations.24 To include the effect of lattice strain on the electronic structure we adopt the continuum elasticity model of Balasubramanian et al.13 as above. The resulting changes in unit cell volume are translated to shifts in the conduction and valence band edges, thus modifying the single particle energies in the calculation described above. The PL lifetimes were calculated using the procedure of de Mello Donegà et al.25 Further details of the model and the material parameters used are given in the Supporting Information.

obtained by exciting the NC using a Picoquant 470 nm diode laser and measuring the resulting signal using a Perkin-Elmer SPCM-AQR-14 single photon avalanche detector coupled to a Edinburgh Instruments TCC-900 time correlated single photon counting card. XRD Modeling. XRD patterns were simulated by extending previously described techniques2,19,20 to build a strained spherical core−shell crystal. To include the effect of lattice strain we adopt the continuum elasticity model of Balasubramanian et al.13 to find the radial and tangential lattice constant as a function of radial position. This simple model assumes isotropic and spherically symmetric material regions and gives purely hydrostatic strain in the core, with tangential tensile and radial compressive strain in the shell. Each octant of the spherical NC was built individually using a zinc blende structure and common boundary atoms. The NC was built radially from the core taking into consideration previous NC positions and the radial position of the new atom. The Debye equation was then solved for these structures using the “DISCUS” software package.21 Further details of the model and the material parameters used are given in the Supporting Information. Exciton Energy Modeling. Exciton energies were calculated using a multiband effective mass theory developed by Pokatilov et al.22 for quantum-dot heterostructures and recently applied to type-II CdTe/CdSe NCs.23 The theory utilizes a



RESULTS AND DISCUSSION The ZnTe cores exhibit good quantum confinement for NCs up to 4.6 nm in diameter (Figure 1) with the 3.6 nm diameter NCs highlighting higher order excitonic features indicative of a small size distribution. TEM confirms that the NCs are spherical D

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

that the position of this peak is not consistent with that which would be expected on the basis of the composition alone and the application of Vegard’s law for homogeneously alloyed materials. This structure comprises 10% ZnTe and 90% ZnSe which would give a lattice constant of 5.71(±0.01) Å through Vegard’s law, smaller than the lattice constant obtained through XRD of 5.77(±0.03) Å. Larger core diameters have larger variations between XRD derived lattice constant and the lattice constant obtained through Vegard’s law. Figure 4a,b shows the typical HR-TEM images and their FFTs observed for NCs with large shells. The smallest core diameter displays an almost constant interplane distance over the whole NC, corresponding to a zinc blende lattice constant of 5.83 Å. For the largest core diameters and shell thickness, variations in lattice constant are more pronounced and indicate lattice bowing from the core to the shell (Figure 4b). An inspection of the FFT over the whole NC shows an average lattice constant of 5.92 Å; however, a closer inspection of the center and edge lattice constants shows a variation between 5.97 and 5.85 Å, respectively. This result is consistent with other strained systems reports.10,19 To account for this change in lattice constant and lack of agreement with Vegard’s Law, we simulate the XRD patterns by incorporating stress and strain into a core-shell NC with significant lattice mismatch. Using the continuum elasticity model in ref 13, we a build a core−shell NC model that accounts for the hydrostatic pressure of the shell compressing the core and biaxial strain in the shell. This elastic model predicts significant strain within the core and at the core−shell interface which then diminishes through the shell toward the surface (Figure 3b). The degree of strain is up to 1.5% and 4% in the core and shell, respectively (Figure 3c). Simulating each diffraction pattern using this elastic model shows good agreement with experimental peaks and accounts for some of the broadening and anisotropic peaks of the ⟨111⟩ peak in large diameter cores (Figure 3a). Further simulation of unstrained and homogeneously alloyed system using Vegard’s law showed significant disagreement with experiment (see Supporting Information). Although there is a variation of lattice constant in the model for the smallest core diameter, the simulated and experimental diffraction patterns show a single zinc blende phase. This can be accounted for by the relative accumulated shifts in atomic positions. For the smallest core diameters these shifts will not be as pronounced when compared to larger core diameter NCs. The accumulated shifts in atomic positions will be significant and therefore lead to broader anisotropic diffraction peaks. This relative accumulated shift for the smallest cores explains the apparent single lattice constant observed in the HR-TEM image of Figure 4a. These results further suggest that the anisotropic growth seen in Figure 2 is strain related and is supported by similar anisotropic growth observed in CdS/ZnSe core−shell systems.5 To test whether strain could be a contributing factor to shell growth dynamics we investigated whether the synthesis of an alloyed interface would mitigate strain effects at the core−shell interface. By deliberately alloying 4 ML between the core and shell as described previously, we see a marked improvement in the spherical nature of large NCs with thick shells (Figure 5). We now investigate the quality of the core−shell interfaces in our smallest NCs using X-ray photoemission spectroscopy (XPS). Synchrotron-XPS makes use of the tunability of the source which enables the kinetic energy of the photoemitted

or quasi-spherical in shape with a size distribution of 7−9%, providing a good seed for epitaxial shell growth (Figure 2a,d). The addition of ZnSe maintains the size distribution of 7−9% and produces spherical or quasi-spherical particles for all core sizes except for core diameters above 3.6 nm with shell thicknesses over 4 ML (Figures 2, 4, and 5a). For the largest core

Figure 4. (a) HR-TEM of a spherical 2.4 nm ZnTe NC with 4 ML ZnSe shell showing a homogeneous crystal structure. The corresponding fast Fourier transform (FFT) shows the ⟨111⟩ reciprocal lattice plane to be 0.297 nm−1 corresponding to a zinc blende lattice constant of 5.83 Å. (b) HR-TEM of anisotropic cubic shaped 4.6 nm ZnTe NC with 5 ML ZnSe shell. The corresponding full, center, and edge FFT shows the reciprocal ⟨111⟩ lattice distance to be 0.292, 0.290, and 0.296 nm−1 corresponding to a zinc blende lattice constant of 5.93, 5.97, and 5.85 Å, respectively.

Figure 5. (a) 4.6 nm ZnTe core with 5 ML of ZnSe showing anisotropic growth. (b) 4.6 nm ZnTe core with a 10% alloyed Te interface with shell thickness equivalent to 5 ML ZnSe growth.

diameters and largest shells, anisotropic cubic or quasi-tetrapod structures are seen in Figures 2f, 4b, and 5a. To test the robustness of these growth behaviors, we attempted several shelling techniques including SILAR-Thermal cycling26 and with various zinc precursors27 which have been shown by others to produce spherical NC growth, and observed no change in the resulting NCs. Inspection of the XRD patterns (Figure 3) and HR-TEM fast Fourier transforms (FFT) (Figure 4) indicates predominately zinc blende crystal structures, independent of core diameter and shell thickness. The XRD ⟨111⟩ peak for the 2.4 nm core with 4.5 ML shell structure reveals a lattice constant that is between that of ZnTe and ZnSe. Quantitative analysis shows E

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Figure 6. (a) Core-level XPS showing the Te 4d and Se 3d signals as a function of X-ray photon energy for ZnTe/ZnSe NC with 2.4 nm diameter core and 4 ML shell. Estimated sampling depths are given in brackets. The signal intensity has been normalized to the Se 3d elemental peak. (b, c) The elemental ratio N(Se)/N(Te) as a function of X-ray photon energy (■) compared with the trends predicted by the spherical core−shell model of Shard et al.,34 for a core diameter of (b) 2.4 nm and (c) 3.3 nm with shell thicknesses ranging from 0.2 to 0.8 nm. (d) A schematic diagram showing a core−shell structure (i) assuming no alloying at the core−shell interface and (ii) the proposed alloying at the core−shell interface as deduced using the results of part c.

NCs) and shell thicknesses ranging from 0.2 to 0.8 nm. As can be seen, the photon energy dependence can be accounted for by a ZnSe shell thickness of 0.5 nm. This would therefore suggest an overall core−shell particle diameter of 3.4 nm, in contrast to the 4.5 nm determined from TEM images of the core−shell NC. In order to account for an overall diameter of 4.5 nm it is necessary to increase the core diameter value used in the model to 3.3 nm, which in turn increases the estimated shell thickness to 0.6 nm (Figure 6c). The observation of a larger core size in XPS than in TEM suggests that some alloying is taking place. The XPS results therefore provide strong evidence for the existence of a “core−shell” structure, although the difference between the core size estimated and the value determined from TEM images suggests an alloyed region of around 0.45 nm thickness (i.e., below 2 ML) between the core and shell materials, as shown schematically in Figure 6d. From these results we can relate the alloying at the core− shell interface to the interdiffusion of Te and Se atoms and note similar observations and synthesis methodology in the case of CdSe/CdS and CdTe/CdSe NCs.10,35 The thickness of the alloyed layer is consistent with that observed in CdSe/CdS NCs.35 Optical Characterization. The ZnTe cores show strong excitonic behavior in the absorption spectra, and the addition of ZnSe shells to the ZnTe cores produces a red-shift of the first excitonic absorption features of up to 100 nm with 5 ML shell thickness (Figure 7a). This red-shift is accompanied by a weakening of the excitonic absorption features consistent with the reduced oscillator strength within type-II structures and comparable with other ZnTe/ZnSe structures.8

electrons to be varied. Thus, tuning the X-ray energy enables depth-resolved studies to be conducted due to the dependence of the photoelectron inelastic mean free path on kinetic energy.17,28−30 The sampling depth from which 95% of the detected electrons originate is approximately 3 times the inelastic mean free path. Figure 6a shows the normalized XPS intensities of the Te 4d and Se 3d core levels as a function of photon energy and hence sampling depth, for a 2.4 nm core NC with a 4 ML shell. After correcting for variations in the relative photoionization cross sections with photon energy,31 the resulting elemental ratios, N(Se)/N(Te) sampled by XPS at each photon energy, are shown in Figure 6b,c. The spectrum taken at 255 eV photon energy, with an estimated sampling depth of 1.9 nm (significantly smaller than the NC diameter), is expected to probe mainly the ZnSe shell, as confirmed by the high Se/Te ratio. On increasing the sampling depth to 4.3 nm, this ratio decreases markedly, providing unambiguous confirmation that the core of the NC is Te-rich; i.e., the system is not homogeneously alloyed. As the flux of photoelectrons is attenuated according to the Beer−Lambert law as it emerges from the sample, it is possible to obtain an estimate of the thickness of the shell by a simple model incorporating the inelastic mean free paths.32 We adopt a “two-layer model”33 that assumes a ZnSe shell overlies a ZnTe core, and use the analytical model developed by Shard et al.34 which accounts for a surface made up of spherical particles with radii of the same order as the inelastic mean free path length. Figure 6b shows the trends predicted in the N(Se)/N(Te) elemental ratios for NCs with a core diameter of 2.4 nm (determined from TEM images of core-only F

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Figure 7. (a) UV−vis absorption and (b) PL spectra for 3.6 nm diameter ZnTe cores with 1, 2, 3, 4, and 5 ML of ZnSe. (c) A contour plot of the (h) theoretical absorption wavelength (nm) in the strained regime of the 1S(e) 1/21S3/2 transition showing the type-I, type-II, and quasi type-II regions. (d) The PL lifetime decays for a 3.9 nm diameter ZnTe core with 3, 4, and 5 ML of ZnSe. Inset are the extracted triexponential lifetime components as a function of shell thickness with a comparison to the proposed theoretical lifetime (- - -).

The ZnTe core-only structures show little to no PL while the addition of a small amount of the zinc precursor enables weak excitonic PL to be observed implying that Te surface trap sites are responsible for rapid nonradiative recombination of the exciton. The PL intensity increases markedly as the ZnSe shell is made thicker with the highest quantum yields of 12% occurring with the addition of 1.5−3 ML ZnSe. This result supports that of Bang et al. who observed the highest quantum yield at 1.4 ML,8 albeit with lower absolute quantum yields. The full width at half-maximum of the ensemble emission spectrum is 25−35 nm for up to 5 ML shell thickness, consistent with a narrow size distribution. Figure 7d shows the PL decay of these NCs with increasing shell thickness and is fitted with triexponential decays. An inspection of inset of Figure 7d shows that all lifetime components increase with increasing shell thickness and show some agreement with the trend predicted by theory (vide infra). The figure also highlights the longest lifetime component for 5 ML growth to be as high as 65 ns, indicative of a type-II system and a significant reduction in the wave function overlap.1,8,36 While the lack of a clear single exponential tail in the decay could be due to size inhomogeneity, this observation alongside the shorter lifetime components could also suggest that other mechanisms are contributing to the decay such as charging or charge trapping effects.37−39 It must be noted that while we have used OA and TOP to keep a common ligand throughout the synthesis and characterization, once cleaned and redispersed in inert conditions and with anhydrous degassed solvents the PL diminishes within 24 h. We can further understand the electronic structure by observing the dependence on shell thickness of the measured band gap.

The approach used has previously been successful in modeling the transitions of CdTe/CdSe NCs.23 The inclusion of lattice strain, as described in the XRD model, in the effective mass calculations gives rise to significant shifts in the conduction band edges of both the core and shell. Compressive strain in the core increases the band gap which manifests predominantly as an increase in conduction band energy, while tensile strain in the shell reduces the conduction band energy in this region, most notably near to the core−shell interface (as discussed below, Figure 8d). The strain-induced shifts in the conduction band potential can be as high as several hundred meV for the core and are considerably smaller in the shell. Shifts in the valence band are much smaller than the conduction band, typically being less than 100 meV in the core and even less in the shell. These shifts in confinement potential lead to a more pronounced type-II band alignment (“double straining”2) which tends to red-shift the exciton energy. The size of the shift in the band edge exciton energy with increasing shell thickness is also found to depend on the valence band offset Ev and the bandgap Eg3 in the external matrix due the large intrinsic bandgap of the core−shell materials. Since these values are not well-known for these materials (see Supporting Information), we use them as fitting parameters when testing the agreement between the strained and unstrained NC models with the experimental data. By varying Eg3 and Ev we find the parameters that minimize the deviation from the experimental data. These parameters are stated in the Supporting Information and are within the range of values reported in the literature. (h) Figure 7c shows a contour plot of the calculated 1S(e) 1/21S3/2 exciton energies as a function of core radius and shell thickness in the strained regime. The figure highlights the type-I, quasi-type-II, G

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

(h) Figure 8. Comparison of the energy deviation, ΔE, of the 1S(e) 1/21S3/2 exciton energy from the core absorption feature calculated with and without strain (lines) and compared to the first absorption feature from the experimental data (square icon) for (a) 2.4, (b) 3.6, and (c) 4.6 nm ZnTe core as a function of shell thickness. Part c also shows the trend for the ZnTe/ZnTe1MLSeXML alloyed interface. (d) A plot of the conduction and valence band profiles calculated with (red) and without (black) strain for a 2.4 nm ZnTe core with a 4 ML ZnSe shell.

inclusion of strain in the model improves the agreement by more than an order of magnitude (see Supporting Information) and highlights that the core is being actively compressed. Experimental data for 3.6 nm diameter NCs shows better agreement to the strained model than to the unstrained model. However, there is better agreement with the unstrained model with decreasing core diameter (Figure 8a,b). From these results we can hypothesize that there is light alloying of the interface that does not significantly modify the core−shell structure, as shown from XPS, which reduces the strain and compression on the core and therefore mitigates the associated shift in exciton energy. To test whether alloying can significantly influence the manifestation of strain within the system, we actively alloy the interface of the largest cores over the first 4 ML with the equivalent of 1 ML growth of ZnTe. A very different dependence on thickness is observed, much closer to that predicted for an unstrained system. This result suggests that alloying mitigates the compressive strain on the core and highlights the significance of compression of the core on the exciton energy. We summarize the results of this study in Figure 9. Figure 9a shows a hypothetical unstrained non-interacting core−shell and the band edge potentials of the bulk materials. Figure 9b shows a fully strained core−shell NC where the strain associated with the epitaxial growth of the shell significantly alters the band gap profile from the unstrained system. This figure serves to describe our larger core diameter NC and explains the slow redshift in band gap energy with increasing shell thickness and the lattice bending in the HR-TEM images. Figure 9c shows a NC

and type-II localization regimes which describe the degree of wave function overlap between the two charge carries. Type-I describes a regime where both carriers are delocalized over the entire volume of the NC, quasi-type-II has one carrier delocalized over the entire NC volume while the other one is primarily located either in the core or in the shell, and type-II describes the situation with the electrons and holes separately localized in the core or shell. These are described in the context of the model previously in ref 23. For comparison we also highlight, with the thick dashed line, the trend line that corresponds to the experimental data in Figure 7a. Inspection of the calculated exciton energies and regimes with the experimental data shows close agreement. To quantify the exciton energy variations for all the core diameters investigated we show the red-shift of the exciton energy with increasing shell thickness from the core exciton energy for the strained and unstrained cases in Figure 8a−c. A comparison of the two models reveals that the addition of strain to the system can cause a red-shift of ∼55 meV for the thickest shells. The significance of the strain can be seen in Figure 8c for the largest core sizes. The strained model shows a plateau feature at small shell thicknesses which accurately describes the experimental data. This plateau feature is due to competing factors; the band gap energy increases due the hydrostatic compression of the shell on the ZnTe core and the increasingly type-II nature of the overall NC. To quantify the improved agreement between experiment and calculation, we calculate the value of the sum of squared residuals between the trend model and the measured band edge absorption features. For the largest core size the H

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C



ACKNOWLEDGMENTS



REFERENCES

Article

S.M.F. is funded by the UK Engineering and Physical Sciences Research Council. J.M.S. acknowledges support from HewlettPackard Ltd. The research leading to the XPS results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under grant agreement 226716, enabling access to MAX-lab. A note of thanks must be given to Ian Sellers of the University of Oklahoma for constructive conversations within this project and to Susannah Speller of University of Oxford for constructive conversations with regards to XRD.

Figure 9. Schematic crystal structure and band structure diagrams for (a) an unstrained NC with no interaction between the core−shell and (b) a strained and (c) an alloyed NC. The unstrained model has an uncompressed core and noninteracting unstrained shell. The strained core has a sharp compositional core−shell interface that compresses the core and strain relaxation within the shell. The alloyed sample shows no compression of the core and illustrates how the lattice mismatch is mitigated by the alloyed interface.

(1) Kim, S.; Fisher, B.; Eisler, H.-J.; Bawendi, M. G. J. Am. Chem. Soc. 2003, 125, 11466−11467. (2) Smith, A. M.; Mohs, A. M.; Nie, S. Nat. Nanotechnol. 2009, 4, 56−63. (3) Nemchinov, A.; Kirsanova, M.; Hewa-Kasakarage, N. N.; Zamkov, M. J. Phys. Chem. C 2008, 112, 9301−9307. (4) Xie, R.; Zhong, X.; Basché, T. Adv. Mater. 2005, 17, 2741−2745. (5) Ivanov, S.; Piryatinski, A.; Nanda, J.; Tretiak, S.; Zavadil, K.; Wallace, W.; Werder, D.; Klimov, V. J. Am. Chem. Soc. 2007, 129, 11708−11719. (6) Oron, D.; Kazes, M.; Banin, U. Phys. Rev. B 2007, 75, 1−7. (7) McDonald, P. G.; Tyrrell, E. J.; Shumway, J.; Smith, J. M.; Galbraith, I. Phys. Rev. B 2012, 86, 125310−125318. (8) Bang, J.; Park, J.; Lee, J. H.; Won, N.; Nam, J.; Lim, J.; Chang, B. Y.; Lee, H. J.; Chon, B.; Shin, J.; Park, J. B.; Choi, J. H.; Cho, K.; Park, S. M.; Joo, T.; Kim, S. Chem. Mater. 2010, 22, 233−240. (9) Klimov, V. I.; Ivanov, S. A.; Nanda, J.; Achermann, M.; Bezel, I.; McGuire, J. A.; Piryatinski, A. Nature 2007, 447, 441−446. (10) Cai, X.; Mirafzal, H.; Nguyen, K.; Leppert, V.; Kelley, D. F. J. Phys. Chem. C 2012, 116, 8118−8127. (11) Data in Science and Technology: Semiconductors; Madelung, O., Ed.; Springer-Verlag: New York, 1991. (12) Wei, S.-H.; Zunger, A. Phys. Rev. B 1999, 60, 5404−5411. (13) Balasubramanian, S.; Ceder, G.; Kolenbrander, K. D. J. Appl. Phys. 1996, 79, 4132−4136. (14) Jiang, F.; Li, Y.; Ye, M.; Fan, L.; Ding, Y.; Li, Y. Chem. Mater. 2010, 22, 4632−4641. (15) Zhang, J.; Sun, K.; Kumbhar, A.; Fang, J. J. Phys. Chem. C 2008, 112, 5454−5458. (16) Li, J. J.; Wang, Y. A.; Guo, W.; Keay, J. C.; Mishima, T. D.; Johnson, M. B.; Peng, X. J. Am. Chem. Soc. 2003, 125, 12567−12575. (17) Hardman, S. J. O.; Graham, D. M.; Stubbs, S. K.; Spencer, B. F.; Seddon, E. A.; Fung, H.-T.; Gardonio, S.; Sirotti, F.; Silly, M. G.; Akhtar, J.; O’Brien, P.; Binks, D. J.; Flavell, W. R. Phys. Chem. Chem. Phys. 2011, 13 (45), 20275−20283. (18) de Mello, J. C.; Wittmann, H. F.; Friend, R. H. Adv. Mater. 1997, 9, 230−232. (19) Dabbousi, B. O.; Rodriguez-Viejo, J.; Mikulec, F. V.; Heine, J. R.; Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G. J. Phys. Chem. B 1997, 101, 9463−9475. (20) Bawendi, M. G.; Kortan, A. R.; Steigerwald, M. L.; Brus, L. E. J. Chem. Phys. 1989, 91, 7282−7290. (21) Proffen, T.; Neder, R. B. J. Appl. Crystallogr. 1997, 30, 171−175. (22) Pokatilov, E. P.; Fonoberov, V. A.; Fomin, V. M.; Devreese, J. T. Phys. Rev. B 2001, 64, 245329. (23) Tyrrell, E. J.; Smith, J. M. Phys. Rev. B 2011, 84, 165328− 165340. (24) Bolcatto, P. G.; Proetto, C. R. J. Phys.: Condens. Matter 2001, 13, 319. (25) de Mello Donegà, C.; Koole, R. J. Phys. Chem. C 2009, 113, 6511−6520. (26) Blackman, B.; Battaglia, D.; Mishima, T.; Johnson, M.; Peng, X. Chem. Mater. 2007, 19, 3815.

with an alloyed interface and highlights the alloyed band gap profile potential which shows similarity to the band gap potential of the unstrained model. This figure serves to illustrate the alloying observed through atom interdiffusion in the smallest core diameter NCs or through design for the larger cores diameter NCs. The similarity of the band gap profile with the unstrained model reflects the lack of red-shift in band gap energy when compared to a strained model.



CONCLUSIONS We present a study of ZnTe/ZnSe core−shell NCs grown by the SILAR technique, which offer a cadmium-free route to type-II NCs with band gaps tuned across the visible spectrum. XRD, XRD simulation, and HR-TEM characterization show that strain at sharp core−shell interfaces can lead to anisotropic growth and show a variation of lattice constant over NC for the largest core diameters and shell thicknesses. Using a (2,6)-band effective mass model we are able to effectively simulate the exciton energy shift with shell thickness and thus identify the effect of strain on the exciton energy. Using XPS we observe evidence of interdiffusion of the anion atoms which produces 0.45 nm of alloying at the core−shell interface for the smallest core diameters. Intentionally alloying 4 ML at the core−shell interface reduces the anisotropic growth but also mitigates the strain-induced excitonic shifts.



ASSOCIATED CONTENT

S Supporting Information *

Further details of the synthesis procedure and an EDX spectrum showing the compositional analysis of the ZnTe cores. Further details and parameters used for the XRD simulation, compositional analysis used to obtained the lattice constant using Vegard’s law, and the effective mass approximation model. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. I

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

(27) Pradhan, N.; Goorskey, D.; Thessing, J.; Peng, X. J. Am. Chem. Soc. 2005, 127, 17586−17587. (28) Santra, P. K.; Viswanatha, R.; Daniels, S. M.; Pickett, N. L.; Smith, J. M.; O’Brien, P.; Sarma, D. D. J. Am. Chem. Soc. 2009, 131, 470−477. (29) Borchert, H.; Haubold, S.; Haase, M.; Weller, H.; McGinley, C.; Riedler, M.; Müller, T. Nano Lett. 2002, 2, 151−154. (30) Huang, K.; Demadrille, R.; Silly, M. G.; Sirotti, F.; Reiss, P.; Renault, O. ACS Nano 2010, 4, 4799−4805. (31) Fujimori, A.; Minami, F.; Sugano, S. Phys. Rev. B 1984, 9, 5225− 5227. (32) Tanuma, S.; Powell, C. J.; Penn, D. R. Surf. Interface Anal. 1991, 17, 911−926. (33) Martin, J. E.; Herzing, A. A.; Yan, W.; Li, X.-q.; Koel, B. E.; Kiely, C. J.; Zhang, W.-x. Langmuir 2008, 24, 4329−4334. (34) Shard, A. G.; J., W.; Spencer, S. J. Surf. Interface Anal. 2009, 41, 541−548. (35) García-Santamaría, F.; Brovelli, S.; Viswanatha, R.; Hollingsworth, J. A.; Htoon, H.; Crooker, S. A.; Klimov, V. I. Nano Lett. 2011, 11, 687−693. (36) Chin, P. T. K.; de Mello Donega, C.; van Bavel, S. S.; Meskers, S. C. J.; Sommerdijk, N. A. J. M.; Janssen, R. A. J. J. Am. Chem. Soc. 2007, 129, 14880−14886. (37) Cadirci, M.; Stubbs, S. K.; Fairclough, S. M.; Tyrrell, E. J.; Watt, A. A. R.; Smith, J. M.; Binks, D. J. Phys. Chem. Chem. Phys. 2012, 14, 13638−13645. (38) Galland, C.; Ghosh, Y.; Steinbrück, A.; Hollingsworth, J. A.; Htoon, H.; Klimov, V. I. Nat. Commun. 2012, 3, 908. (39) Sher, P.; Smith, J.; Dalgarno, P.; Warburton, R.; Chen, X.; Dobson, P.; Daniels, S.; Pickett, N.; O’Brien, P. Appl. Phys. Lett. 2008, 92, 101111.

J

dx.doi.org/10.1021/jp3087804 | J. Phys. Chem. C XXXX, XXX, XXX−XXX