Franziska Niederdraenk, Christian Kumpf
The study of semiconductor nanoparticles is of broad interest in many different fields of modern physics, chemistry and bio-technology. This is - on the one hand - due to their enormous potential for applications: Just two out of many examples are quantum dots in electronic devices (single-electron transistors, in light emitting diodes, solar cells, lasers, etc.) and biomarkers for cell tracking in biochemistryor biomedical research. Most applications are based on size-induced quantum confinement effects which are, even though frequently used, often not yet understood in detail. This aspect makes (very small) nanoparticles also interesting for fundamental research. The understanding of their optical, electronic and structural properties could be significantly improved if better structural information was available.
Therefore our project aims on a better understanding of the detailed structure of ultra-small nanpoparticles (1-5 nm) and the influence on electronic and optical properties. We developed a novel fit approach for powder x-ray diffraction data which is based on modeling at an atomic level. The method yields not only precise information on basic parameters like size, shape and structure, which could in principle be determined by many different techniques (TEM, AFM, simple XRD techniques like Rietveld refinement, UV-VIS absorption spectroscopy, etc.). Our method is can also detect details like impurities, stacking faults, strain and relaxation effects. Furthermore it became possible to consider distributions of the basic parameters (e.g., size distribution) by implementing an ensemble averaging routine. Therefore a much more realistic model for the ensemble of nanoparticles can be found by fitting the diffraction data.
The experiments are mostly performed at the Hamburger Synchrotronstrahlungslabor. As examples we show a refinement of CdS, ZnS and ZnO in the following.
In Fig. 2 three different methods of data modeling are compared. The main figure shows the best fit which could be obtained using our ensemble modeling method. Data stems from ZnO nanoparticles which were wet-chemically synthesized and are stabilized by an organic shell around the ZnO core. The shell is not visible in the XRD data. The measured data is very well reproduced by our model calculation. This fit was obtained by using an ensemble of 50 cylindrically shaped particle which are (on average) 3.6 nm in diameter and 3.8 nm in height. The key-finding is an structural disorder in the particles: They have a high stacking fault probability of 14%, i.e. an average number of stacking faults of 2 per particle while the average number of stacking layers in the particles is 14. A stacking fault corresponds to a Zincblende-like stacking sequence (ABCABC) whereas the "normal" stacking is Wurtzite-like (ABAB). Hence, on average there are 2 Zincblende-like stacked layers per particle embedded in 14 Wutzite layers.
Compared to other fitting methods the ensemble modeling shows clear advantages: A Rietveld fit tries to find the best fitting individual particle (no ensemble) with a homogenious crystal structure (no stacking faults). The red line in Fig. 2 shows that the fit to the data is much worse. The single line fit procedure fits a sum of individual peaks to the measured date , each having its individual position, width and height (i.e., at first, no atomic model is taken as a basis for the fit). Due to the high number of fit-parameters a "perfect" fit can be obtained (see blue line in Fig. 2). However, the problem comes with interpreting the fit results in terms of finding a structural model. From the width of each and every Bragg reflection a specific size of the particle can be calculated, or, to be more precise, a "length" of the particle in a certain spatial direction which corresponds to the direction perpendicular to the diffraction planes of the Bragg reflection used. Hence one ends up with a number of "sizes" of the same particle in certain directions. In most of the cases these numbers cannot be put together to a plausible model since all other effects (like, e.g. stacking faults) are neglected.
F. Niederdraenk, K. Seufert, A. Stahl, R.S. Bhalerao-Panajkar, S. Marathe, S.K. Kulkarni, R.B. Neder, C. Kumpf, Ensemble modeling of very small ZnO nanoparticles, Phys. Chem. Chem. Phys. 13 (2), 498 (2011).
F. Niederdraenk, K. Seufert, P. Luczak, S.K. Kulkarni, C. Chory, R.B. Neder, and C. Kumpf, Structure of small II-VI semiconductor nanoparticles: A new approach based on powder diffraction, phys. stat. sol. (c) 4(9), 3234-3243 (2007);
C. Barglik-Chory, R. Neder, V. Korsunskiy, F. Niederdraenk, C. Kumpf, E. Umbach, M. Schumm, M. Lentze, J. Geurts, G. Astakhov, W. Ossau, and G. Müller, Influence of liquid-phase synthesis parameters on particle sizes and structural properties of nanocrystalline ZnO powders, phys. stat. sol. (c) 4(9), 3260-3269 (2007);
C. Kumpf, Structure determination of very small (1-5~nm) nano-particles, Appl. Phys. A 85 (4), 337-343 (2006);
C. Kumpf, R.B. Neder, F. Niederdraenk, P. Luczak, A. Stahl, M. Scheuermann, S. Joshi, S.K. Kulkarni, C. Barglik-Chory, C. Heske, and E. Umbach, Structure determination of CdS and ZnS nanoparticles: Direct modelling of synchrotron radiation diffraction data, J. Chem. Phys. 123, 224707 (2005);
C. Barglik-Chory, D. Buchold, M. Schmitt, W. Kiefer, C. Heske, C. Kumpf, O. Fuchs, L. Weinhardt, A. Stahl, E. Umbach, M. Lentze, J. Geurts, G. Müller, Synthesis, structure, and spectroscopic characterization of water-soluble CdS nanoparticles, Chem. Phys. Lett. 379, 443 (2003);
A.S. Ethiraj, N. Hebalkar, S.K. Kulkarni, R. Pasricha, J. Urban, C. Dem, M. Schmitt, W. Kiefer, L. Weinhardt, S. Joshi, R. Fink, C. Heske, C. Kumpf, E. Umbach, Enhancement of photoluminescence in manganese-doped ZnS nanoparticles due to a silica shell, J. Chem. Phys. 118 (19), 8945 (2003).
R. Neder, Univ. Erlangen-Nürnberg,
S.K.K. Kulkarni, Univ. of Pune, India
C. Graf, E. Rühl, Freie Univ. Berlin,
F. Reinert, G. Müller, Univ. Würzburg
C. Chory, Univ. Oldenburg, ehemals Univ. Würzburg