Letter pubs.acs.org/NanoLett
Direct Observation of Band Structure Modifications in Nanocrystals of CsPbBr3 Perovskite Junhao Lin,*,†,# Leyre Gomez,‡,# Chris de Weerd,‡ Yasufumi Fujiwara,§ Tom Gregorkiewicz,*,‡,§ and Kazutomo Suenaga† †
National Institute of Advanced Industrial Science and Technology (AIST), AIST Central 5, Tsukuba 305-8565, Japan Institute of Physics, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands § Division of Materials and Manufacturing Science, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan ‡
S Supporting Information *
ABSTRACT: We investigate the variation of the bandgap energy of single quantum dots of CsPbBr3 inorganic halide perovskite as a function of size and shape and upon embedding within an ensemble. For that purpose, we make use of valenceloss electron spectroscopy with Z-contrast annular dark-field (ADF) imaging in a state-of-the-art low-voltage monochromatic scanning transmission electron microscope. In the experiment, energy absorption is directly mapped onto individual quantum dots, whose dimensions and location are simultaneously measured to the highest precision. In that way, we establish an intimate relation between quantum dot size and even shape and its bandgap energy on a single object level. We explicitly follow the bandgap increase in smaller quantum dots due to quantum confinement and demonstrate that it is predominantly governed by the smallest of the three edges of the cuboidal perovskite dot. We also show the presence of an effective coupling between proximal dots in an ensemble, leading to band structure modification. These unique insights are directly relevant to the development of custom-designed quantum structures and solids which will be realized by purposeful assemblage of individually characterized and selected quantum dots, serving as building blocks. KEYWORDS: Valence-loss EELS, CsPbBr3 nanocrystal, quantum confinement effect, bandgap modification, monochromatic STEM
P
The correlation between the NC size and the bandgap energy is commonly obtained by two separate experiments performed on NC ensembles. For instance, (high-resolution) transmission electron microscopy (HR-TEM) or dynamic light scattering (DLS) determines the average NC size, while optical spectroscopy techniques are used to measure the bandgap. Since the properties of NCs in an ensemble are not typically different from those of single NCs,32 accurate correlation is highly challenging. Although photoluminescence (PL) and absorption spectroscopy on single NCs is readily possible, it can only be carried out upon extreme dilution.33−35 This is fundamentally restricted by the diffraction limit, as the wavelength of the photon source used to excite a NC is significantly larger than its dimensions. Hence, obtaining information on the structural geometry simultaneously with its optical characteristics for a single NC is challenging, and the behavior of excitons in a single NC, that may be subject to the influence of its neighbors, remains elusive. Replacing the
erovskite materials have attracted much attention recently due to their advantageous optical properties1 and potential application for highly efficient and low-cost photovoltaic devices.2−6 Cesium lead halide (CsPbX3, X = Cl, Br, I, Cl/Br, and Br/I) nanocrystals (NCs, nanometer sized quantum dots) are of interest because of their high photoluminescence quantum yields (PL QYs, 50−90%), narrow emission bands, and their emission color can be easily varied by changing the composition.7−16 Since they combine the advantageous properties of perovskites and NCs,17 they are an interesting material for various optoelectronic applications.18−23 Semiconductor nanostructures and specifically NCs are widely investigated due to the attractive possibility of energy structure tuning by size.24−26 As the NC diameter decreases and approaches the Bohr radius of the particular material, the quantum confinement (QC) sets in, modifying the wave function of the free electron and hole.27−30 Hence, the energy band structure is affected, and in the particular case of semiconductor NCs, the bandgap energy increases. In optical spectroscopy this effect manifests itself most profoundly by the blue-shift of both the absorption edge and the emission, as the NC size decreases.31 © 2016 American Chemical Society
Received: August 23, 2016 Revised: October 8, 2016 Published: October 13, 2016 7198
DOI: 10.1021/acs.nanolett.6b03552 Nano Lett. 2016, 16, 7198−7202
Letter
Nano Letters excitation source with electrons has been shown as a feasible solution, since the low-loss electron energy loss spectroscopy (EELS) arising from the low-energy excitations, i.e., valence electron excitations, is an analogue of optical absorption spectroscopy,36 which provides the possibility to investigate the optical properties on nanometer-sized objects in parallel with their structural parameters. In the past, monochromatic EELS has been used to study CdSe NCs, but the resolution, both in real space and energy-wise, were insufficient in order to yield truly microscopic insights.37 However, recent developments in the low-energy monochromatic transmission electron microscope with advanced aberration correctors have enabled the combination of electron spectroscopy with ultrahigh spatial and energy resolutions.38,39 Using this technique, here we study the effects of QC on single CsPbBr3 NCs that are either isolated or in the ensemble. This provides comprehensive information which is not accessible in an ensemble experiment. Our findings have a general character, being relevant to other nanomaterials as well, providing unique insights on cooperative effects between NCs. The CsPbBr3 NCs have been synthesized by a wet-chemistry method, yielding colloidal near-monodisperse NCs with a cubic perovskite structure and a very high PL quantum yield of ∼70− 80%. The NC size can be tuned by controlling the synthesis temperature and isolated NCs and clusters can be easily created by diluting the sample. The NCs are supported on a graphene substrate for investigation by electron microscopy. The microscope is operated at 60 kV, and the probe size used for the low-loss EELS measurements is about 1.6 Å, providing a high spatial resolution during the collection of the EEL spectra. Figure 1a shows two low magnified Z-contrast annual dark-field
(ADF) images, indicating the presence of single NCs as well as their larger clusters and ensembles. Since in the perovskite structure all faces of a cubic NC are equivalent, there will be no preferential ordering during deposition and aggregation. Therefore, the fact that most of the NCs appear to be square in the projection view, implies that they are actually cubic; this notion is further confirmed by the statistical analysis of the averaged ADF image intensity from many NCs (see Figure S1 and the relevant text in Supporting Information for further details). We found that most NCs were oriented in the [001] direction, suggesting a form of self-organization similar to the formation of so-called quantum dot solids as observed for CdSe and Ag NCs.40,41 An individual CsPbBr3 NC of approximately 6 nm in size, is presented in more detail in Figure 1b. Spots with different brightness within the NC alternate in a periodically quadratic manner. Each spot corresponds to an atomic column consisting of different atoms. The intensity of the Z-contrast image is directly related to the atomic number of the imaged species and the number of atoms inside each column,42 thus revealing the cubic structure of CsPbBr3 perovskite projected along the [001] direction. For illustration, a schematic of the cubic perovskite structural model is given to the right. A noncrystalline surface layer (highlighted by the white lines, of ∼5 Å width) around the crystalline core is visible; it is an intrinsic feature of the investigated NCs and is not caused by electron irradiation (see detailed discussion in Supporting Information, Figure S2). This is consistently observed for all the examined NCs, also for those in an ensemble (see Figure S1). This noncrystalline layer can arise due to strain on the surface driven by the high surface energy, or due to the chemical bonding with surface ligands.43 Figure 1c and d shows the simultaneously acquired energy-dispersive X-ray spectrum (EDS) and core-loss EELS of the same NC as shown in Figure 1b, confirming the chemical composition of the NC core, without the presence of any other impurities. The atomic ratio of Cs and Pb is close to 1:1, as determined from the quantification of the EDS on the L-lines of the two elements, again consistent with the CsPbBr3 perovskite. To investigate the correlation between the optical absorption and the dimensions of a single NC, we performed a systematic study on many individual NCs with different sizes that are either isolated or located within an ensemble. Figures 2a and b show ADF images of NCs, isolated and in an ensemble for EELS collection, respectively. For both environments, typical low-loss EEL spectra of NCs with sizes of approximately 6 and 8 nm are depicted in Figure 2cthe raw dual EELS data and a full set of EEL spectra of NCs with different sizes, isolated and within an ensemble, are provided in Figures S3−S5 in the Supporting Information. When an electron in a NC is excited from the top of the valence band to the conduction band, a characteristic step will appear in the low-loss EEL spectrum, followed by a steady increase of the signal. This corresponds to the band-to-band absorption which starts at the bandgap energy and whose magnitude grows as the density of states increases with energy, in complete analogy with optical absorption of NCs. Such an abrupt onset enables the unambiguous (single) NC bandgap energy determination (see the detailed process that we use to estimate the bandgap energy in the Supporting Information), where as expected, the bandgap energy increases with decreasing size, both for isolated and ensemble-embedded NCs. Moreover, a clear difference in the step-like absorption onset, and hence the bandgap energy, can be seen for the isolated and ensemble-embedded NCs of similar size,
Figure 1. Atomic structure and chemical composition of the CsPbBr3 NCs. (a) Low magnified ADF images showing single NCs and clusters of NCs that are obtained by diluting the as-prepared sample. (b) Atomic resolution Z-contrast image of an individual cubic CsPbBr3 NC with an edge size of ∼6 nm. The atomic structure model is provided on the right. The white arrows indicate the width of an amorphous layer (∼0.5 nm) surrounding the crystalline core. (c,d) Energy-dispersive X-ray spectrum (c) and core-loss EELS (d) of the same region shown in panel b to determine the elements that are present. The Si signal in (c) comes from the contamination in the underlying graphene substrate. 7199
DOI: 10.1021/acs.nanolett.6b03552 Nano Lett. 2016, 16, 7198−7202
Letter
Nano Letters
between the amorphous and crystalline structure is not sharp, which makes it hard to completely exclude the contribution of the amorphous layer, we measured the size of the NCs from the crystalline core, taking the width of the amorphous layer as the uncertainty of the size determination. The data have been fitted using the effective mass approximation (EMA) theory, which is described by a 1/r2 function, where r is the size of NC (see Supporting Information for further details). Surprisingly, we find that small isolated NCs appear to have higher bandgap energy as compared to a NC of the same size in an ensemble. Reversely, a large NC apparently has a lower bandgap energy if isolated than when in an ensemble (see also Figure S6 in the Supporting Information for more details on ensemble spectroscopy measurements). This result shows that two adjacent NC do not simply “merge” upon interaction into a single large one, but rather “average” their bandgaps. Statistically, large NCs that are situated in an ensemble are most likely surrounded by equally sized or smaller NCs, whereas small NCs are surrounded by larger ones. These somewhat surprising results provide direct evidence of effective coupling between proximal NCs where their bandgap energy, and therefore energy structure, is influenced by the neighbors. Similar band structure modification due to NC interaction, such as bandgap reduction in InAs QD ensembles, has been reported previously.44 The bandgap “averaging” effect observed in our experiment indicates the “collective” dielectric response of the NCs ensemble, presumably through the surface ligands/substrate which couple the NCs in the ensemble. It is important to note, however, that the observation of a clear step at the bandgap energy excludes a possibility that the observed NC bandgap modification could arise due to small delocalization of the probing electron beam, typical for EELS (see Figure S7 in Supporting Information for a more detailed discussion on this matter). The characteristics discussed this far are derived from measurements on cubic NCs and provide no information on the shape-dependent profile of the bandgap energy. In Zcontrast ADF images, the thickness of the sample is also proportional to the image intensity (or contrast profile).45 A brighter intensity indicates large(r) NC sizes. As the ADF images provide the surface dimensions, the thickness can be qualitatively extracted by integrating the ADF signal over the NC (see also Figure S1). The dependence of the bandgap energy on the shape of a NC can be resolved as we occasionally found some NCs with a shape other than cubic, as can be seen in Figure 4a. The three NCs, from left to right, can be considered as a cube, plate, and rod; their corresponding normalized ADF image intensities are given in the panel below. We conclude that the plate- and rod-like NCs differ in thickness with a factor of 3 and 2, respectively, as compared to the cubic NC. Figure 4b shows their corresponding low-loss EEL spectra. Although the plate- and rod-shaped NCs clearly have a different volume, we observe similar bandgap energy. The dependence of the bandgap energy on the NC shape is qualitatively shown in Figure 4c. The red solid dots indicate the bandgap energy dependence on the size, for isolated (or individual) cubic NCs. The data has been fitted using the EMA theory. The green and blue solid dots represent the average edge size as obtained from the ADF image (where only two edges can be distinguished, and are averaged) for the plate- and rod-shaped NCs, respectively. There is a clear discrepancy for the plate and rod, as the size appears to be too large. However, if the thickness is taken into account, and when we now correlate the
Figure 2. Bandgap energy of individual near-cubic CsPbBr3 nanocrystals. (a) ADF images showing isolated near-cubic NCs with a size of ∼6 nm (green) and ∼8 nm (purple) respectively. (b) ADF image of an ensemble where two NCs with similar sizes as in panel a are indicated (red, blue). (c) The corresponding low-loss EEL spectra of the isolated (top) and in ensemble (bottom) NCs as shown in panels a and b. The semitransparent black solid lines superimposed on the spectra are obtained by smoothing. The bandgap energy is determined from the peak on the first derivative of the spectra appearing at the abrupt onset of absorption, as illustrated by the dashed lines under the spectra for all four NCs (see Supporting Information for detailed processing of the data). The decaying background in the first two curves comes from the tail of the zero loss peak.
suggesting a NC-NC interaction which influences their energy structure and therefore absorption. The size dependences of the bandgap energy for single and ensemble-embedded NCs are compared in Figure 3. The bandgap energies of isolated NCs (red circles) and single NCs in an ensemble (blue triangles) have been determined for edge sizes in the range of 5−15 nm. Here, since the boundary
Figure 3. Influence of the presence of neighboring NCs. The bandgap energies of single NCs that are either isolated (red dots) or in an ensemble (blue triangles) as determined from low-loss EELS measurements on single NCs, plotted as a function of their sizes. Both data sets have been fitted using the effective mass approximation (see Supporting Information for further details). Small isolated NCs appear to have higher bandgap energy as compared to a NC in an ensemble with the same size. Reversely, a large NC has lower bandgap energy if isolated than present in an ensemble. This indicates that their energy structure is adjusted according to surrounding objects due to an effective coupling between them. The accuracy of the bandgap energy determination is further discussed in Figure S4 in Supporting Information). 7200
DOI: 10.1021/acs.nanolett.6b03552 Nano Lett. 2016, 16, 7198−7202
Letter
Nano Letters
■
NCs in the ensemble, optical absorption properties, EELS delocalization of a single NC and additional reference (PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected]. *E-mail:
[email protected]. Author Contributions #
J.L. and L.G. contributed equally to this work. K.S., T.G., Y.F., and J.L. conceived the project and designed the experiments. L.G. synthesized the perovskite nanocrystals. J.L. performed the LL-EELS measurements and analyzed the data. C.W. performed the optical spectroscopy measurements and analyzed the data. All authors discussed the results and their interpretation. J.L and C.W. cowrote the manuscript with contributions from K.S. and T.G. who also cosupervised the project.
Figure 4. Shape-dependent energy bandgap of isolated single NCs. (a) Z-contrast ADF images of three NCs that have different shapes. Two edge lengths can be derived from the ADF image, while the thickness is estimated based on the normalized ADF image intensity as shown in the panel underneath. The thickness of each NC is estimated qualitatively to be 13−14 nm (cube, red), 6−7 nm (platelet, green), and 8−9 nm (rod, blue). (b) The low-loss EEL spectra of the corresponding NCs with their bandgap energies given. The semitransparent black solid lines are the smoothed spectra and the dashed lines their first derivatives. (c) A schematic comparison of the “average” edge size, as solely determined from the ADF image, vs bandgap energy for these three shapes (solid dots) showing a clear discrepancy for the plate and rod. The average edge size of a (noncubic) NC is determined as the square root of the surface. Taking now the thickness into account, the actual shortest edge of the NCs can be considered where subsequently the bandgap values of the plate and rod-shaped NCs agree well with our earlier observations (open squares).
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was financially supported by Stichting voor Fundamenteel Onderzoek der Materie (FOM) and by Technologiestichting STW. J.L. and K.S. acknowledge JSTACCEL and JSPS KAKENHI (JP16H06333 and JP25107003) for financial support. Y.F. and T.G. thank Osaka University for International Joint Research Promotion Program.
■
(1) Ishihara, T. Optical Properties of Pb-based inorganic-organic perovskites. Optical Properties of Low−Dimensional Materials; World Scientific Publishing Co Pte Ltd., 1995, p 1. (2) Kazim, S.; Nazeeruddin, M. K.; Grätzel, M.; Ahmad, S. Angew. Chem., Int. Ed. 2014, 53 (11), 2812−2824. (3) Sum, T. C.; Mathews, N. Energy Environ. Sci. 2014, 7 (8), 2518− 2534. (4) Green, M. A.; Ho-Baillie, A.; Snaith, H. J. Nat. Photonics 2014, 8, 506−514. (5) Gratzel, M. Nat. Mater. 2014, 13, 838−842. (6) Park, N.-G. J. Phys. Chem. Lett. 2013, 4 (15), 2423−2429. (7) Protesescu, L.; et al. Nano Lett. 2015, 15 (6), 3692−3696. (8) Lignos, I.; et al. Nano Lett. 2016, 16 (3), 1869−1877. (9) Nedelcu, G.; et al. Nano Lett. 2015, 15 (8), 5635−5640. (10) Sun, S.; Yuan, D.; Xu, Y.; Wang, A.; Deng, Z. ACS Nano 2016, 10 (3), 3648−3657. (11) Swarnkar, A.; et al. Angew. Chem. 2015, 127, 15644−15648. (12) Akkerman, Q. A.; et al. J. Am. Chem. Soc. 2015, 137 (32), 10276−10281. (13) Huang, H.; Susha, A. S.; Kershaw, S. V.; Hung, T. F.; Rogach, A. L. Adv. Sci. 2015, 2, 1500194. (14) Raino, G.; et al. ACS Nano 2016, 10 (2), 2485−2490. (15) Schmidt, L. C.; et al. J. Am. Chem. Soc. 2014, 136 (3), 850−853. (16) Zhang, F.; et al. ACS Nano 2015, 9 (4), 4533−4542. (17) Bawendi, M. G.; Steigerwald, M. L.; Brus, L. E. Annu. Rev. Phys. Chem. 1990, 41 (1), 477−496. (18) Wang, Y.; et al. Nano Lett. 2016, 16 (1), 448−453. (19) Wang, Y.; Li, X.; Song, J.; Xiao, L.; Zeng, H.; Sun, H. Adv. Mater. 2015, 27 (44), 7101−7108. (20) Hu, F.; et al. ACS Nano 2015, 9 (12), 12410−12416. (21) Park, Y. S.; Guo, S.; Makarov, N. S.; Klimov, V. I. ACS Nano 2015, 9 (10), 10386−10393. (22) Song, J.; et al. Adv. Mater. 2015, 27 (44), 7162−7167. (23) Yakunin, S.; et al. Nat. Commun. 2015, 6, 8056. (24) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226−13239.
shortest edge with the bandgap energy, the data agrees well with the results of the cubes (open squares). This profoundly indicates that the bandgap energy of a NC is determined by its shortest dimension rather than the volume, which also suggests the isotropic properties of the excitons when under confinement. In conclusion, taking advantage of the recent advances in the aberration-corrected monochromatic EELS technique, we were able to determine absorption spectra of single cubic CsPbBr3 perovskite NCs, whose individual dimensions were measured simultaneously. In that way, an intimate correlation between the bandgap energy and size and shape of a CsPbBr3 NC has been obtained. We have also demonstrated how properties of individual NCs change due to their mutual interactions in an ensemble, thus paving the way toward purposeful development of larger quantum structures and QD solids with a-la-carte properties.
■
REFERENCES
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b03552. Information on methods and chemicals, Figures S1−S7 including detailed discussion on the amorphous layer, raw EELS data collection, static analysis of NC ensemble, valence-loss EELS characterizations of single NCs and 7201
DOI: 10.1021/acs.nanolett.6b03552 Nano Lett. 2016, 16, 7198−7202
Letter
Nano Letters (25) Lim, X. Nature 2016, 531, 26−28. (26) Murray, C. B.; Kagan, C. R.; Bawendi, M. G. Annu. Rev. Mater. Sci. 2000, 30 (1), 545−610. (27) Alivisatos, A. P. Science 1996, 271, 933−937. (28) Efros, A. L.; Efros, A. L. Sov. Phys. Semicond. 1982, 16 (7), 772− 775. (29) Nirmal, M.; Brus, L. Acc. Chem. Res. 1999, 32 (5), 407−414. (30) Gurevich, L.; Canali, L.; Kouwenhoven, L. P., Quantum confinement VI: Nanostructured materials and devices. Proceedings of the International Symposium, The Electrochemical Society Inc., 2001, p 19. (31) Haug, H.; Koch, S. W. Quantum theory of the optical and electronic properties of semiconductors; World Scientific Publishing Co Pte Ltd., 1990, 5. (32) Norris, D. J.; Bawendi, M. G. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 53 (24), 16338. (33) Empedocles, S. A.; Norris, D. J.; Bawendi, M. G. Phys. Rev. Lett. 1996, 77 (18), 3873. (34) Grundmann, M.; et al. Phys. Rev. Lett. 1995, 74 (20), 4043. (35) Valenta, J.; Juhasz, R.; Linnros, J. Appl. Phys. Lett. 2002, 80 (6), 1070−1072. (36) Kapetanakis, M. D.; et al. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 92 (12), 125147. (37) Erni, R.; Browning, N. D. Ultramicroscopy 2007, 107 (2), 267− 273. (38) Tizei, L. H.; et al. Phys. Rev. Lett. 2015, 114 (10), 107601. (39) Sasaki, T.; Sawada, H.; Hosokawa, F.; Sato, Y.; Suenaga, K. Ultramicroscopy 2014, 145, 50−55. (40) Murray, C. B.; Kagan, C. R.; Bawendi, M. G. Science 1995, 270, 1335. (41) Markovich, G.; et al. Acc. Chem. Res. 1999, 32 (5), 415−423. (42) Krivanek, O. L.; et al. Nature 2010, 464, 571−574. (43) De Roo, J.; et al. ACS Nano 2016, 10 (2), 2071−2081. (44) Steiner, D.; Aharoni, A.; Banin, U.; Millo, O. Nano Lett. 2006, 6 (10), 2201. (45) Pennycook, S. J.; Nellist, P. D. Scanning Transmission Electron Microscopy: Imaging and Analysis; Springer Science & Business Media, 2011.
7202
DOI: 10.1021/acs.nanolett.6b03552 Nano Lett. 2016, 16, 7198−7202
